# SPERM DIFFERENTIATION AND SPERMATOZOA FUNCTION: MECHANISMS, DIAGNOSTICS, AND TREATMENT

EDITED BY : Tomer Avidor-Reiss, Zhibing Zhang and Xin Zhiguo Li PUBLISHED IN : Frontiers in Cell and Developmental Biology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-747-8 DOI 10.3389/978-2-88963-747-8

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

## SPERM DIFFERENTIATION AND SPERMATOZOA FUNCTION: MECHANISMS, DIAGNOSTICS, AND TREATMENT

Topic Editors:

Tomer Avidor-Reiss, University of Toledo, United States Zhibing Zhang, Virginia Commonwealth University, United States Xin Zhiguo Li, University of Rochester, United States

Citation: Avidor-Reiss, T., Zhang, Z., Li, X. Z., eds. (2020). Sperm Differentiation and Spermatozoa Function: Mechanisms, Diagnostics, and Treatment. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-747-8

# Table of Contents

*05 Editorial: Sperm Differentiation and Spermatozoa Function: Mechanisms, Diagnostics, and Treatment*

Tomer Avidor-Reiss, Zhibing Zhang and Xin Zhiguo Li

*09 Determination of a Robust Assay for Human Sperm Membrane Potential Analysis*

Carolina Baro Graf, Carla Ritagliati, Cintia Stival, Paula A. Balestrini, Mariano G. Buffone and Darío Krapf


Tomer Avidor-Reiss, Matthew Mazur, Emily L. Fishman and Puneet Sindhwani


Felipe A. Navarrete, Luis Aguila, David Martin-Hidalgo, Darya A. Tourzani, Guillermina M. Luque, Goli Ardestani, Francisco A. Garcia-Vazquez, Lonny R. Levin, Jochen Buck, Alberto Darszon, Mariano G. Buffone, Jesse Mager, Rafael A. Fissore, Ana M. Salicioni, María G. Gervasi and Pablo E. Visconti

*137 Proteomic Changes in Human Sperm During Sequential* in vitro *Capacitation and Acrosome Reaction*

Judit Castillo, Orleigh Adeleccia Bogle, Meritxell Jodar, Forough Torabi, David Delgado-Dueñas, Josep Maria Estanyol, Josep Lluís Ballescà, David Miller and Rafael Oliva

*153 Novel Techniques of Sperm Selection for Improving IVF and ICSI Outcomes*

Iván Oseguera-López, Sara Ruiz-Díaz, Priscila Ramos-Ibeas and Serafín Pérez-Cerezales

#### *176 Successful ICSI in Mice Using Caput Epididymal Spermatozoa*

Raúl Fernández-González, Ricardo Laguna, Priscila Ramos-Ibeas, Eva Pericuesta, Víctor Alcalde-Lopez, Serafín Perez-Cerezales and Alfonso Gutierrez-Adan

#### *181 Signaling Enzymes Required for Sperm Maturation and Fertilization in Mammals*

Souvik Dey, Cameron Brothag and Srinivasan Vijayaraghavan

*196 A Kinase Anchor Protein 4 is Vulnerable to Oxidative Adduction in Male Germ Cells*

Brett Nixon, Ilana R. Bernstein, Shenae L. Cafe, Maryse Delehedde, Nicolas Sergeant, Amanda L. Anderson, Natalie A. Trigg, Andrew L. Eamens, Tessa Lord, Matthew D. Dun, Geoffry N. De Iuliis and Elizabeth G. Bromfield


Carolina Baro Graf, Carla Ritagliati, Valentina Torres-Monserrat, Cintia Stival, Carlos Carizza, Mariano G. Buffone and Dario Krapf

*238 Quantitative Intracellular pH Determinations in Single Live Mammalian Spermatozoa Using the Ratiometric Dye SNARF-5F*

Julio C. Chávez, Alberto Darszon, Claudia L. Treviño and Takuya Nishigaki

*251 Membrane Potential Determined by Flow Cytometry Predicts Fertilizing Ability of Human Sperm*

Lis C. Puga Molina, Stephanie Gunderson, Joan Riley, Pascal Lybaert, Aluet Borrego-Alvarez, Emily S. Jungheim and Celia M. Santi

*263 Essential Role of Sperm-Specific PLC-Zeta in Egg Activation and Male Factor Infertility: An Update*

Alaaeldin Saleh, Junaid Kashir, Angelos Thanassoulas, Bared Safieh-Garabedian, F. Anthony Lai and Michail Nomikos

*272 Toward Development of the Male Pill: A Decade of Potential Non-hormonal Contraceptive Targets*

Katarzyna Kent, Madelaine Johnston, Natasha Strump and Thomas X. Garcia

# Editorial: Sperm Differentiation and Spermatozoa Function: Mechanisms, Diagnostics, and Treatment

Tomer Avidor-Reiss 1,2 \*, Zhibing Zhang3,4 and Xin Zhiguo Li 5,6

*<sup>1</sup> Department of Biological Sciences, University of Toledo, Toledo, OH, United States, <sup>2</sup> Department of Urology, College of Medicine and Life Sciences, University of Toledo, Toledo, OH, United States, <sup>3</sup> Department of Physiology, Wayne State University, Detroit, MI, United States, <sup>4</sup> Department of Obstetrics & Gynecology, Wayne State University, Detroit, MI, United States, <sup>5</sup> Center for RNA Biology: From Genome to Therapeutics, Department of Biochemistry and Biophysics, University of Rochester Medical Center, Rochester, NY, United States, <sup>6</sup> Department of Urology, University of Rochester Medical Center, Rochester, NY, United States*

Keywords: spermatogenesis, sperm, capacitation, manchette, centriole, cilia, contraceptive pill, piRNA

**Editorial on the Research Topic**

**Sperm Differentiation and Spermatozoa Function: Mechanisms, Diagnostics, and Treatment**

#### INTRODUCTION

Edited and reviewed by: *Philipp Kaldis, Lund University, Sweden*

\*Correspondence: *Tomer Avidor-Reiss tomer.AvidorReiss@utoledo.edu*

#### Specialty section:

*This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology*

> Received: *05 March 2020* Accepted: *16 March 2020* Published: *07 April 2020*

#### Citation:

*Avidor-Reiss T, Zhang Z and Li XZ (2020) Editorial: Sperm Differentiation and Spermatozoa Function: Mechanisms, Diagnostics, and Treatment. Front. Cell Dev. Biol. 8:219. doi: 10.3389/fcell.2020.00219* One of the most remarkable processes in nature is the transformation of a generic round stem cell into a streamlined spermatozoon that fertilizes the egg, complements it, and activates the program that starts a new life. Sperm are unique because they undergo hyper evolution due to direct selection by sperm competition, resulting in many molecular, and structural invocations. This extraordinary biology affects virtually every process and structure of the sperm during spermatogenesis. The manchette reshapes the nucleus. The protamine repacks the DNA. New RNA granules form. The protein-based centrioles are remodeled. A flagellum with a unique configuration is formed. Membrane-bound organelles such as Golgi and ER are converted and reduced to acrosome and residual bodies. 80S ribosomes along with most of the cytoplasm are eliminated.

The spermatozoon continues to mature during its transport through the epididymis and female reproductive system. There, the spermatozoon gains motility, undergoes capacitation, and obtains epigenetic information. Hyperactivation and acrosome reaction allow it to fertilize eggs. Post-fertilization, the spermatozoon components complement the egg and activate embryo development.

All these unique properties of sperm are mediated by many sperm-specific proteins, making spermiogenesis an ideal target for male contraceptive pills. Abnormalities in sperm differentiation result in infertility, miscarriage, and congenital diseases, with a recent advance in transgenerational inheritance indicating that alterations in the sperm "epigenome" can impact the life-long wellness of offspring. Therefore, sperm dysfunction is a significant factor in men's infertility and farm animals' subfertility. Resolving this condition requires novel ideas at multiple levels to improve the diagnosis of sperm dysfunction and development of new treatments to complement in vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI). These also warrant the advancement of reproductive technologies such as sperm cryopreservation.

Below we present the 20 papers in this collection that address many of the critical aspects of sperm biology, medicine, and technology. We divided them into four categories: Unique Sperm Organelles, Spermatogenesis, Spermatozoon Maturation, and Sperm Contribution to the Zygote.

#### UNIQUE SPERM ORGANELLES: ACROSOME, MANCHETTE, ATYPICAL CENTRIOLES, MIGRATING CILIA TRANSITION ZONE, AND FIBROUS SHEATH

**The acrosome** is a specialized sperm organelle that contains digestive enzymes and covers the sperm head, which continues to surprise us with new findings (Zhang et al., 2020; Wang et al., 2020). It was proposed to be derived from the lysosome, or related organelle, or be a direct Golgi derivative (Hartree and Srivastava, 1965; Aguas and Pinto da Silva, 1985; Berruti et al., 2010). The review by Khawar in this collection provides a brief historical overview and highlights recent findings on acrosome biogenesis in mammals.

**The manchette** is a unique transient microtubular structure that shapes the sperm head and facilitates the development of the sperm neck and tail (Mendoza-Lujambio et al., 2002; Yang et al., 2018). A research paper in this collection by Tapia Contreras and Hoyer-Fender shows that the protein coiled-coil domain containing 42 (CCDC42) is a new centrosome protein that functions in the head-to-tail coupling apparatus (HTCA) and sperm tail formation.

**The centrioles** are an evolutionarily-conserved sub-cellular organelle that evolved to have unique properties in sperm cells (Avidor-Reiss, 2018; Avidor-Reiss and Turner, 2019). Each human spermatozoon contains two remodeled centrioles that it contributes to the zygote (Fishman et al., 2018). Most previous investigations into the role of mammalian centrioles during fertilization were completed in murine models. However, because mouse sperm and zygotes appear to lack centrioles, these studies provide information that is limited in its applicability to humans (Avidor-Reiss, 2018; Avidor-Reiss and Fishman, 2018). The review by Avidor-Reiss et al. in this collection comprehensively summarizes and updates the role of centrioles in human reproduction.

**The ciliary transition zone** is a gate separating the cilium from the cytoplasm via (Malicki and Avidor-Reiss, 2014). However, uniquely in the sperm, the transition zone migrates and separates the axoneme into two distinct compartments (Basiri et al., 2014; Avidor-Reiss and Leroux, 2015; Avidor-Reiss et al., 2017). Research papers in this collection by Persico et al. show that the microtubule-depolymerizing protein Kinesin-13 Klp10A is enriched in the ciliary transition zone and the base of the migrating transition zone of Drosophila melanogaster sperm. They suggest that Klp10A may be a core component of the ciliary transition zone.

**The fibrous sheath** is a unique sperm flagellum structure that is responsible for regulating signal transduction, metabolic pathways, and mechanical rigidity of the flagellum (Eddy, 2007; Mukai and Travis, 2012; Lindemann and Lesich, 2016). A-Kinase Anchor Protein 4 (AKAP4) is a major component of the fibrous sheath (Fang et al., 2019). A research paper in this collection by Nixon et al. found that AKAP4 is a target of oxidative stress caused by electrophilic aldehyde, 4-hydroxynonenal (4HNE) in sperm cells. These findings suggest that AKAP4 may be a biomarker of sperm quality, warranting the design of an antioxidant treatment for infertility.

### SPERMATOGENESIS: NON-HORMONAL MALE CONTRACEPTIVE PILL AND SPERM COMPETITION

**The non-hormonal male contraceptive pill** can be achieved by interfering with the many remarkable processes of spermatogenesis (Thirumalai and Page, 2019). The review by Kent et al. in this collection comprehensively summarizes recently discovered potential target genes for such a pill. They discuss novel contraceptive targets, new data for potential druggability, and possible effects from paralog proteins.

**Sperm competition** is a unique form of evolutionary selection that drives sperm innovation (van der Horst and Maree, 2014). Human sperm properties suggest that sperm have low-risk sperm competition as expected from a long history of male sexual dominance or monogamy. In naked mole-rats, only one male is reproductively active, and spermatogenesis is suppressed in the other males of the colony (O'Riain et al., 2000). This single male dominance example is a classic case of reducing sperm competition, leading to simplified, polymorphic, and slow-swimming spermatozoa (Van Der Horst et al., 2011). The research paper in this collection by van der Horst et al. show that lack of sperm competition also results in testicular structure and spermatogenesis degeneration.

### SPERMATOZOON MATURATION IN THE EPIDIDYMIS AND FEMALE REPRODUCTION: SIGNALING ENZYMES, SMALL RNAs, AND INTRACELLULAR, AND EXTRACELLULAR SPERM pH

**Signaling enzymes** are central to the mechanism that regulates spermatozoon maturation since sperm is a transcriptionally silent cell (Freitas et al., 2017). Two papers address this subject in this collection. The review by Dey et al. recaps the contribution of four signaling enzymes that are present as specific isoforms only in placental mammals. Protein phosphatase PP1γ2, glycogen synthase kinase 3 (GSK3), calcium-regulated phosphatase calcineurin (PP2B), and protein kinase A (PKA) have critical roles in sperm maturation and hyperactivation. A research paper by Castillo et al. describes the quantitative proteomic profiling of capacitated and calcium activated spermatozoa. They found 36 proteins with significant changes in their relative abundance within these conditions and many that have post-translational modifications. These results contribute to our knowledge of the molecular basis of human fertilization.

**Small RNAs** such as PIWI-interacting RNAs (piRNAs), transfer RNA fragments (tRFs), microRNAs (miRNAs), and other non-coding RNAs are abundantly present in sperm and function in spermatogenesis and intergenerational epigenetic inheritance (Krawetz et al., 2011; Sharma et al., 2016; Sun et al., 2018; Perez and Lehner, 2019). The review by Sharma in this collection explores the mechanism of sperm small RNA remodeling during post-testicular maturation in the epididymis, and the potential role of this remodeling in intergenerational epigenetic inheritance.

**Membrane potential** is an essential feature during capacitation for sperm fertilization capability (López-González et al., 2014; Ritagliati et al., 2018). Two research papers in this collection develop methods to determine membrane potential. Molina et al. developed a technique based on flow cytometry and showed it can predict the fertilizing ability of human sperm. Baro Graf et al. developed a method based on potentiometric dye in a fluorometric assay. They both showed that the plasma membrane potential of capacitated sperm correlates with the sperm acrosome reaction.

**Intracellular and extracellular sperm pH** are important factors in sperm function as sperm encounter opposite PH conditions during its transport. Therefore, controlling sperm pH plays a crucial role in mammalian sperm physiology (Nishigaki et al., 2014). Two papers in this collection address this subject. A research paper by Chávez et al. established a robust imaging method that allows for the determination of absolute intracellular pH values in a single spermatozoon. The review by Touré specifies the current knowledge regarding PH regulation in sperm and its environment that is controlled by the distinct epithelia. They specifically discuss the role of solute carrier 26 (SLC26) proteins and their interaction with cystic fibrosis transmembrane conductance regulator channel (CFTR).

#### SPERM CONTRIBUTION TO THE ZYGOTE: EGG ACTIVATION, SPERM CRYOPRESERVATION, TRANSIENT SPERM STARVATION, AND INTRACYTOPLASMIC SPERM INJECTION

**Egg activation** by sperm factor is essential to prevent polyspermy **(**Swann and Lai, 2016**)**. PLC-Zeta is a novel, testis-specific phospholipase C isoform that activates the egg post-fertilization (Swann et al., 2012). The review by Saleh et al. in this collection summarizes and updates the essential role of sperm-specific PLC-Zeta. However, PLC-Zeta may not be the only activation factor, and a potential "alternative" sperm factor may also be present.

**Sperm cryopreservation** is an essential technique for fertility management (Ezzati et al., 2020). However, the post-thaw viability of sperm differs among bulls—a research paper in this collection by Ugur et al. describes a multivariate and univariate analysis to identify potential freezability biomarkers. Their findings suggest that amino acids may have important roles in seminal plasma, although differences in amino acid concentration do not mediate this process.

**Transient sperm starvation** is a new and novel method to improve sperm performance, as described by the research

#### REFERENCES

paper in this collection by Navarrete et al.. This study discusses starvation increased hyperactivated motility, the ability to fertilize in vitro, and the production of pups in mice. Starvation also increased the fertility of a sub-fertile mouse and enhanced ICSI success in bovine. These findings raise the possibility that starvation may be used to improve assisted reproductive technologies in other mammalian species, including humans.

**Intracytoplasmic sperm injection (ICSI)** is a clinical treatment that introduces sperm to the oocyte independent of the sperm's ability to move and the two gametes' ability to fuse (Sánchez-Calabuig et al., 2014). Two papers in this collection address this subject. A research paper in this collection by Fernández-González et al. compared the ICSI success rate of spermatids from different regions of epididymis. They found that mice caput spermatozoa and caudal spermatozoa have similar potential to produce embryos and offspring by ICSI. A review in this collection by Oseguera-López et al. discusses novel techniques of sperm selection for improving IVF and ICSI outcomes. They describe the latest technologies focusing on those proven to improve sperm genetic integrity, fertilization capacity, embryo production in vitro survival, pregnancy, or delivery rates.

#### CONCLUSION

Significant progress was made over the years in understanding the fundamental biology of the sperm. However, as evident from the studies described above, many aspects of this biology remain poorly understood. More studies are required to identify the molecular mechanisms that mediate and regulates spermatogenesis as well as spermatozoon maturation and capacitation. We need to understand better what controls sperm movement and its infraction with the female reproductive tract and the egg. A crucial future endeavor is identifying other essential contributions the sperm makes to the zygote in addition to the DNA. Gaining this information will guide future translational research towered the diagnosis and treatment of infertility, assisting the domesticated animal industry, the development of a male contraceptive pill, and improve the wellness of the next generation.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

TA-R is supported by Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) grant number R03 HD098314 and R21 HD092700.

Avidor-Reiss, T. (2018). Rapid evolution of sperm produces diverse centriole structures that reveal the most rudimentary structure needed for function. Cells. 7:67. doi: 10.3390/cells7070067

Avidor-Reiss, T., and Fishman, E. L. (2018). It takes two (Centrioles) to Tango. Reproduction 157, R33–R51. doi: 10.1530/REP-18-0350

Aguas, A. P., and Pinto da Silva, P. (1985). The acrosomal membrane of boar sperm: a Golgi-derived membrane poor in glycoconjugates. J. Cell Biol. 100, 528–534. doi: 10.1083/jcb.100.2.528


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Avidor-Reiss, Zhang and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Determination of a Robust Assay for Human Sperm Membrane Potential Analysis

Carolina Baro Graf1,2, Carla Ritagliati<sup>1</sup> , Cintia Stival<sup>1</sup> , Paula A. Balestrini<sup>3</sup> , Mariano G. Buffone<sup>3</sup> and Darío Krapf1,2 \*

<sup>1</sup> CONICET-UNR, Laboratoty of Cell Signal Transduction Networks, Instituto de Biología Molecular y Celular de Rosario, Rosario, Argentina, <sup>2</sup> UNR, Laboratorio de Medicina Reproductiva, Facultad de Ciencias Bioquímicas y Farmacéuticas, Rosario, Argentina, <sup>3</sup> Consejo Nacional de Investigaciones Científicas y Tecnológicas, Instituto de Biología y Medicina Experimental, Buenos Aires, Argentina

#### Edited by:

Tomer Avidor-Reiss, University of Toledo, United States

#### Reviewed by:

Jormay Lim, National Taiwan University, Taiwan Harvey Florman, University of Massachusetts Medical School, United States

#### \*Correspondence:

Darío Krapf krapf@ibr-conicet.gov.ar

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

> Received: 13 March 2019 Accepted: 23 May 2019 Published: 11 June 2019

#### Citation:

Baro Graf C, Ritagliati C, Stival C, Balestrini PA, Buffone MG and Krapf D (2019) Determination of a Robust Assay for Human Sperm Membrane Potential Analysis. Front. Cell Dev. Biol. 7:101. doi: 10.3389/fcell.2019.00101 Mammalian sperm must undergo a complex process called capacitation in order to fertilize the egg. During this process, hyperpolarization of the sperm plasma membrane has been mostly studied in mouse, and associated to its importance in the preparation to undergo the acrosome reaction (AR). However, despite the increasing evidence of membrane hyperpolarization in human sperm capacitation, no reliable techniques have been set up for its determination. In this report we describe human sperm membrane potential (Em) measurements by a fluorimetric population assay, establishing optimal conditions for Em determination. In addition, we have conducted parallel measurements of the same human sperm samples by flow cytometry and population fluorimetry, before and after capacitation, to conclusively address their reliability. This integrative analysis sets the basis for the study of Em in human sperm allowing future work aiming to understand its role in human sperm capacitation.

Keywords: membrane potential, population fluorimetry, sperm capacitation, human sperm, carbocyanine dye, diSC3(5)

## INTRODUCTION

During fertilization, the sperm cell has a simple and fundamental goal: to merge with the oocyte and deliver its genetic material. However, mammalian sperm are not able to fertilize an oocyte shortly after ejaculation. To gain fertilizing competence, they must go through a process called sperm capacitation. Capacitation includes a complex series of molecular events that normally take place in the female genital tract but can be mimicked in vitro in defined media. This process prepares sperm to acquire hyperactivated motility (HA) and to undergo the acrosome reaction (AR) upon stimulation (Chang, 1951; Austin, 1952).

At the molecular level, sperm capacitation is associated with increased membrane fluidity, changes in intracellular ion concentrations (Visconti et al., 2011), hyperpolarization of the sperm plasma membrane (Zeng et al., 1995; Hernández-González et al., 2006), increased activity of protein kinase A (PKA) (Krapf et al., 2010), and protein tyrosine phosphorylation (Arcelay et al., 2008). Data regarding capacitation has been acquired in different mammalian species. Moreover, due to the complexity of the capacitation process, each of these events has been studied independently.

Thus, knowledge regarding how these events interconnect to regulate capacitation is mostly unavailable.

One important molecular process leading to the capacitated state is hyperpolarization of the sperm plasma potential, which has been mostly studied in mouse (Zeng et al., 1995; Hernández-González et al., 2006; De La Vega-Beltran et al., 2012). In this species, Em hyperpolarization has been shown to be both necessary and sufficient for sperm to undergo stimulated AR (De La Vega-Beltran et al., 2012). The mechanisms that drive Em hyperpolarization in mouse sperm involve the opening of the potassium channel Slo3 (Chávez et al., 2013), since sperm from Slo3 KO mice fail to hyperpolarize during capacitation. Most importantly, these sperm do not undergo acrosomal reaction upon stimulation (Santi et al., 2010; Yang et al., 2011), and, as expected from these data, Slo3 KO male mice are sterile. Despite the clear importance of membrane hyperpolarization during mouse sperm capacitation, this Em change has not been studied in detail in human sperm. A few reports have independently analyzed Em of sperm samples: while Linares-Hernández et al. (1998) reported an Em for non-capacitated sperm of around −40 mV, Patrat et al. (2005) reported that capacitated sperm exhibit an Em of about −58 mV. Both studies used fluorimetric population assays using the carbocyanine DiSC3(5). In addition, a qualitative study suggested by flow cytometry that human sperm undergo hyperpolarization during capacitation (Brewis et al., 2000). It is worth noticing that the consortium of Drs. Treviño and Darszon, further substantiated these qualitative findings, showing by flow cytometry that a subpopulation of human sperm hyperpolarizes during capacitation (López-González et al., 2014). However, these studies did not measure Em changes quantitatively on the same sample, i.e., during capacitation, which is crucial for proper human sperm analysis considering the biological variabilities found in these cells.

Despite the clear importance of hyperpolarization in mouse sperm cells, data regarding human sperm is scarce. Our manuscript presents an integrative study, since the same samples have been evaluated through flow cytometry analysis and population Em measurements, both before and after capacitation. We have established a solid reproducible methodology. Moreover, our data show that hyperpolarization is observed in human sperm along incubation in capacitating conditions, although alternative behaviors are also reported here. Finally, the methodology and implications of Em hyperpolarization in human sperm are herein discussed.

#### MATERIALS AND METHODS

#### Reagents

Chemicals were obtained from the following sources: bovine serum albumin (BSA) and HEPES were purchased from Sigma (St. Louis, MO, United States). Propidium iodide (PI) from Santa Cruz (Santa Cruz, CA, United States), 3,3-dipropylthiadicarbocyanine iodide [DiSC3(5)] from Invitrogen (Carlsbad, United States). All other chemicals were of analytical grade.

#### Ethical Statement

The study protocol was approved by the Bioethics Committee of the Instituto de Biología y Medicina Experimental (CONICET). All subjects gave written informed consent in accordance with the Declaration of Helsinki.

#### Culture Media

HEPES-buffered human tubal fluid (HTF) was used throughout the study, containing (in mM) 4.7 KCl, 0.3 KH2PO2, 90.7 NaCl, 41.2 MgSO4, 2.8 Glucose, 1.6 CaCl2, 3.4 Sodium Piruvate, 60 Sodium Lactate, 23.8 HEPES, and was termed non-capacitating media. For capacitating media, 15 mM NaHCO3, and 0.5% w/v BSA were added. In all cases, pH was adjusted to 7.4 with NaOH.

#### Human Sperm Preparation

Semen samples were obtained by masturbation from 10 healthy donors after 3–5 days of abstinence and analyzed following WHO recommendations (World Health Organization, 2010). All samples fulfilled semen parameters (total fluid volume, sperm concentration, motility, viability, and morphology) according to WHO normality criteria. Samples were allowed to liquefy for 1 h at 37◦C in water bath. Then, sperm ejaculates were allowed to swim-up in non-capacitating HTF media (see above) at 37◦C for 1 h. Motile selected spermatozoa were washed 5 min 400 × g. Conditions were performed in 400 µl of non-capacitating or capacitating media to a final cell concentration of 7 × 10<sup>6</sup> cells/ml for fluorimetric population assays and 1.25 × 10<sup>6</sup> cells/ml for flow cytometry measurements. Sperm were capacitated for 3 h at 37◦C.

#### Determination of Membrane Potential by Flow Cytometry

Sperm plasma membrane potential (Em) was assessed using different concentrations of DiSC3(5) for 10 min, for proper determination of dye concentration. Unless specified, dye concentration was then of 50 nM. In addition, 2 µM of PI was added 30 s before collecting data to monitor viability. Data were recorded as individual cellular events using a BD FACSAriaTM II cell sorter flow cytometer (BD Biosciences). Forward scatter (FSC) and side scatter (SSC) fluorescence data were collected from 20,000 events per sample. Debris and cell aggregates were excluded from analysis by gating side vs. FSC cytogram. Negative stain for PI was selected for living cells and positive cells for DiSC3(5) were detected using the filter for allophycocyanine (APC) (660/20). Normalization was performed by adding 1 µM valinomycin and 39.6 mM KCl. Data were analyzed using FACS Diva and FlowJo software (Tree Star 10.0.7r2).

#### Determination of Membrane Potential by Fluorimetric Population Assay

After incubation in each condition, human sperm, at a concentration of 3 × 10<sup>6</sup> sperm in 1.7 ml, were loaded with DiSC3(5) at different concentrations to determine the ideal dye concentration. Unless specified, dye concentration was determined at 1 µM (dissolved in DMSO at 5 mM). Sperm were transferred to a gently stirred cuvette at 37◦C, and the fluorescence was monitored with a Varian Cary

(B) Representative fluorescence traces using different human sperm concentrations at 1 µM DiSC3(5). Optimal number of sperm is indicated in red. (C) Representative fluorescence traces using different dye concentrations, at constant sperm number (3 × 10<sup>6</sup> /1.7 ml). Optimal dye concentration is indicated in red.

Eclipse fluorescence spectrophotometer at 620/670 nm excitation/emission wavelengths. Recordings were initiated when steady-state fluorescence was reached (approximately 10 min). Calibration was performed by adding 1 µM valinomycin and sequential additions of KCl (in µL): 4, 4, 7, and 15 for calibration curve 1; 8, 13.34, 13.1, and 13.3 for calibration curve 2; 1.67, 3.4, 5.95, and 10.2 for calibration curve 3. Unless specified, calibration curve 1 was used. Final sperm membrane potentials were obtained by linearly interpolating the theoretical Em values against arbitrary fluorescence units of each trace. The theoretical Em values were obtained using Nernst equation, considering 120 mM the internal K<sup>+</sup> concentration in sperm. This internal calibration for each determination compensates for variables that influence the absolute fluorescence values.

## Statistical Analysis

One-way repeated measures analysis of variance was used to compare between Em values obtained with different calibration curves. Statistical significances are indicated in the figure legends.

## RESULTS

#### Em Measurement in Human Sperm

Em measurements were performed with a positively charged carbocyanine probe, DiSC3(5). This dye partitions into sperm cells according to their membrane potential but independently on the nature of ionic fluxes, rendering it suitable for potential measurements of the plasma membrane. The technique has

long been used in mouse sperm, in a robust and reproducible way, as shown in **Figure 1A** (Chou et al., 1989; Zeng et al., 1995; Demarco et al., 2003). The dye is added to the sperm suspension, where it partitions into the cells originating a decrease in fluorescence due to accumulation of the dye in the lipid bilayer of the cells, in a less-fluorescent form (Plášek and Hrouda, 1991). Em hyperpolarization favors the influx of more dye to the cell, resulting in decreased extracellular fluorescence. Once the dye gives a stable signal, reached after a 10 min incubation in the sperm suspension, calibration starts by adding valinomycin (a K<sup>+</sup> ionophore) followed by sequential additions of KCl, increasing extracellular K<sup>+</sup> concentration to different known values. The original resting potential can be obtained following the Nernst equation, provided that the membrane behaves as a K<sup>+</sup> electrode where the internal K<sup>+</sup> concentration is 120 mM (Darszon et al., 1999; **Figures 1A,B**).

At appropriate concentrations of sperm and probe, this method provides a highly reproducible value of plasma membrane potential. In order to analyze optimal conditions for human sperm, both sperm and probe concentrations were assayed. When the probe was fixed at 1 µM, shifting sperm cell number from 1 × 10<sup>6</sup> to 1 × 10<sup>7</sup> (in a cuvette containing 1.7 ml) gave different responses. The best dynamic range was observed when 3 × 10<sup>6</sup> sperm were used (1.76 × 10<sup>6</sup> sperm/ml) (**Figure 1C**). Once the optimal sperm number was determined, different DiSC3(5) concentrations were tested. As seen in **Figure 1C**, no other concentration improved signal to noise ratio when compared to 1 µM DiSC3(5). Accordingly, 1 µM dye and 3 × 10<sup>6</sup> sperm/1.7 ml where taken as optimal conditions for the subsequent analysis of human sperm Em. Even though DiSC3(5) bears a short alkyl group that renders it poorly hydrophobic, migration to the mitochondria cannot be totally

excluded, which would originate minor contributions to the observed Em. However, as showed in **Supplementary Figure 1**, the mitochondrial electron transport inhibitor rotenone did not affect DiSC3(5) signal. Thus, Em measurements of both non capacitated and capacitated sperm were unaffected by mitochondrial dysfunction, further substatntiating that mitochondrial contribution is negligible.

## Calibration of Em Measurements

In order to avoid errors due to dye loading or sperm concentration differences, every measurement carries an internal calibration curve after fluorescence stabilization (Ritagliati et al., 2018). As shown in **Figure 2A**, calibration was performed by adding 1 µM valinomycin and sequential additions of KCl. Initial potassium concentration equals 5.04 mM KCl; then, additions of a 2 M KCl solution, sequentially rose KCl concentration to 9.72, 14.38, 22.49, and 39.63 mM, corresponding to Em of −84.7, −67.1, −56.7, −44.7, and −29.6 mV, respectively. With the aim of assessing the strength of this method applied to human sperm samples, three different calibration curves were performed (**Figure 2A**, curves 1–3), varying the concentration of KCl added. In the examples shown in **Figure 2A**, the resting membrane potentials were of −63.5, −64.5, and −66.9 mV for calibration curves 1–3, respectively. This procedure was performed on 5 different sperm samples. As shown in **Figure 2B**, the technique showed no significant variations among different calibration curves, supporting the strength of the methodology.

## Comparative Analysis Between Flow Cytometry and Fluorometric Population Assays

Up to now, the few studies that have reported Em measurements in human sperm have mostly used flow cytometry, while only a couple of them have used the fluorescent population assay, in either capacitated or non-capacitated sperm. However, quantitative measurements of Em during capacitation is clearly missing. Thus, we decided to analyze human sperm Em during capacitation, using both flow cytometry and population fluorimetry approaches.

Firstly, in order to establish appropriate working conditions, sperm were loaded with 50 nM DiSC3(5) and run on a flow cytometer. A plot of cell complexity (SSC-A) vs. cell size (FSC-A) is shown in **Figure 3A**, detailing the gating used to eliminate cell debris and aggregates. The selection of individual cells was confirmed using the FSC-H vs. FSC-A plot (**Figure 3B**). Live cells were selected after co-staining sperm with PI, normally obtaining a cellular viability >90% (**Figure 3C**). This representative set up was performed after loading cells with

FIGURE 4 | Human sperm membrane potential determination by flow cytometry and population fluorimetry. (A) Flow cytometry analysis of human sperm in non-capacitating (left panel) and capacitating conditions (right panel). Valinomycin addition causes an increase of fluorescence, in accordance to a hyperpolarized state; KCl addition causes a decrease of fluorescence, as expected for a depolarized state. Each histogram plots percentage of maximum (% of Max) vs. DiSC3(5) fluorescence, obtained by normalizing to the peak height at the mode of the distribution – so the maximum Y-axis value in the absolute-count histogram becomes 100% of total. Lower panel shows median values obtained before and after valinomycin addition. Hyperpolarization response is normalized against valinomycin addition. (B) Population fluorimetric assay of human sperm in non-capacitating (left panel) and capacitating conditions (right panel). In this case valinomycin addition causes a decrease in fluorescence, as expected for a hyperpolarized state; KCl sequential additions increase extracellular fluorescence, depolarizing the sample. Lower panel shows fluorescence values for initial and hyperpolarized states, with their corresponding membrane potential value (in mV). These results are representative of one experiment (n = 5).

different DiSC3(5) concentrations (0.5–100 nM), for both noncapacitated (**Figure 3D**) and capacitated sperm (**Figure 3E**). From these data, 50 nM DiSC3(5) resulted the concentration of choice, being the lowest concentration tested with highest intracellular fluorescence. Final cell concentration was fixed to 1.25 × 10<sup>6</sup> sperm/ml.

Secondly, these conditions were used to analyze Em by flow cytometry, as shown in **Figure 4A**. In flow cytometry using DiSC3(5) (a cationic dye), increased fluorescence correlates to Em hyperpolarization since more dye enters the cell. Addition of valinomycin further increases fluorescence, as the sperm cells hyperpolarize (**Figure 4A**). Subsequent addition of KCl caused the expected depolarization of the sperm cells, rendering lower values of fluorescence. When a capacitated sperm sample was analyzed, valinomycin only caused a slight increase in fluorescence (**Figure 4A**). However, KCl highly depolarized these hyperpolarized cells. Thus, it could be inferred from these data, that this sample suffered hyperpolarization during capacitation.

In order to compare Em measurements of two samples by flow cytometry, the median fluorescence was obtained after valinomycin addition, corresponding to the maximum hyperpolarization state. The median value from the initial state (before valinomycin addition) of each sample can then be used to calculate a normalized degree of hyperpolarization (**Figure 4A**). Accordingly, in non-capacitating conditions the hyperpolarization degree was 49% shifting to 88.3% in capacitating conditions. This normalization is necessary not only for a semi-quantitative analysis, but also when performing a qualitative comparison between conditions (i.e., NC and CAP), since dye concentrations could differ among them.

When the same sperm sample incubated under either non-capacitating or capacitating conditions were analyzed by fluorimetry, values of −63.1 and −74.5 mV were obtained, supporting the data obtained by flow cytometry (**Figure 4B**). It is worth noticing that occasionally, approximately 10% of samples, exhibited unexpected behaviors when analyzed by flow cytometry: upon addition of valinomycin, instead of causing hyperpolarization, a shift to a depolarized state was observed (**Supplementary Figure 2**). When these samples were analyzed by fluorimetry, it was observed that, as expected, valinomycin caused an initial hyperpolarization that is, however, immediately followed by a sustained depolarization (**Supplementary Figure 2**). This result indicates that what was observed as a valinomycin-triggered depolarization, was in fact a later compensation by still unknown ion fluxes, and further supporting the strength of fluorimetric measurements.

#### DISCUSSION

In mouse sperm, changes in the sperm plasma membrane potential (Em) associated with sperm capacitation have been studied for more than 20 years (Zeng et al., 1995). Cauda epidydimal mouse sperm have a depolarized Em of ∼ −40 mV, which hyperpolarizes to ∼ −60 mV when capacitation takes place reviewed by Stival et al. (2016). These values of mouse sperm Em have been conducted using fluorimetric measurements and constitute an average value of a population known to be far from homogenous. Thus, when other techniques as single cell microscopy (Arnoult et al., 1999) or flow cytometry (López-González et al., 2014; Escoffier et al., 2015) were used, two clear sperm sub-populations displaying different Em were observed. One of these sub-populations remains at a depolarized state, while the other shifts to a relatively hyperpolarized value of Em (∼ −80 mV) (Arnoult et al., 1999). These findings are consistent with the observation that only a fraction of the sperm cells are capable to undergo the AR (Salicioni et al., 2007).

In human sperm, it has been recently proposed that certain Em shift is associated with normal sperm function, as assessed by electrophysiology and IVF outcome (Brown et al., 2016). However, this study involved laborious techniques that hampered the analysis of many cells per patient. Other studies involved the qualitative analysis of Em in human sperm by flow cytometry (Brewis et al., 2000) and analysis of either non-capacitated or capacitated sperm by fluorimetry (Linares-Hernández et al., 1998; Patrat et al., 2005). Considering the importance of the study of Em in human sperm using straightforward techniques, we aimed to standardize a reliable method for the analysis of Em during human sperm capacitation.

We have satisfactorily established the optimal conditions to quantitatively measure human sperm Em using the carbocyanine DiSC3(5) with a fluorimeter in a population assay. But most importantly, we have compared two different methods, i.e., flow cytometry and population fluorimetry, by using the same sperm sample. Our data indicate that Em analysis by flow cytometry is a robust and reproducible methodology, when proper normalization for loading control is performed. However, it is important to highlight results that, although very reproducible, are to date difficult to understand. After addition of valinomycin, some samples experience an initial hyperpolarization that is followed by a depolarization. These behaviors are easily followed by fluorimetry, and can then be analyzed accordingly. However, when cytometry is used, due to technical time limitations, depolarization is observed instead of the expected valinomycin driven hyperpolarization. Thus, even though both techniques give the same end point result, cytometry might omit this particular behavior, and give a confusing result.

The mechanisms that regulate the hyperpolarization of human sperm plasma membrane during capacitation are poorly understood. Plasma membrane permeability to the ionic media at any given time defines the cell Em. Thus, Em changes during capacitation reflect changes of ion permeability. The most relevant ions are Na+, K+, and Cl−, which have an equilibrium potential of +40, −80, and −40 mV, respectively. Considering these ion's equilibrium, closure of Na<sup>+</sup> transport as well as opening of a K<sup>+</sup> channel could drive hyperpolarization. In mouse, the current knowledge points toward a high contribution of K<sup>+</sup> to this phenomenon. However, the situation is far from clear in human sperm. Redepolarization might in turn be associated to opening of Na<sup>+</sup> channels and/or Cl<sup>−</sup> channels, shifting the equilibrium to more positive values. The methodology herein described is ideal for standardizing a technique that could aid in the study of human Em physiology.

#### DATA AVAILABILITY

fcell-07-00101 June 7, 2019 Time: 18:30 # 8

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

#### ETHICS STATEMENT

The study protocol was approved by the Bioethics Committee of the Instituto de Biologa y Medicina Experimental (CONICET). All subjects gave written informed consent in accordance with the Declaration of Helsinki.

## AUTHOR CONTRIBUTIONS

CBG, CR, CS, and PB conducted the experiments. All authors analyzed the data and revised final version of the manuscript.

#### REFERENCES


CBG, CR, MGB, and DK conceived the study. CBG, CR, and DK wrote the manuscript.

## FUNDING

This work was supported by grants from the Agencia Nacional de Promoción Científica y Tecnológica de Argentina, PICT 2015- 2294 (awarded to MGB), and PICT 2015-3164 and PICT 2017- 3217 (awarded to DK). CBG is recipient of a scholarship from the National Research Council of Argentina.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00101/ full#supplementary-material

sperm undergoes hyperpolarization during capacitation. Biol. Reprod. 92:121. doi: 10.1095/biolreprod.114.127266


Ritagliati, C., Baro Graf, C., Stival, C., and Krapf, D. (2018). Regulation mechanisms and implications of sperm membrane hyperpolarization. Mech. Dev. 154, 33– 43. doi: 10.1016/j.mod.2018.04.004



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Baro Graf, Ritagliati, Stival, Balestrini, Buffone and Krapf. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# CCDC42 Localizes to Manchette, HTCA and Tail and Interacts With ODF1 and ODF2 in the Formation of the Male Germ Cell Cytoskeleton

#### Constanza Tapia Contreras and Sigrid Hoyer-Fender\*

Johann-Friedrich-Blumenbach-Institute of Zoology and Anthropology – Developmental Biology, Göttingen Center for Molecular Biosciences (GZMB), Georg-August-University of Göttingen, Göttingen, Germany

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Maria Eugenia Teves, Virginia Commonwealth University, United States Bénédicte Durand, Université Claude Bernard Lyon 1, France

> \*Correspondence: Sigrid Hoyer-Fender shoyer@gwdg.de

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 31 May 2019 Accepted: 18 July 2019 Published: 14 August 2019 Citation:

Tapia Contreras C and Hoyer-Fender S (2019) CCDC42 Localizes to Manchette, HTCA and Tail and Interacts With ODF1 and ODF2 in the Formation of the Male Germ Cell Cytoskeleton. Front. Cell Dev. Biol. 7:151. doi: 10.3389/fcell.2019.00151 Terminal differentiation of male germ cells into functional spermatozoa requires shaping and condensation of the nucleus as well as the formation of sperm-specific structures. A transient microtubular structure, the manchette, is mandatory for sperm head shaping and the development of the connecting piece and the sperm tail. The connecting piece or head-to-tail coupling apparatus (HTCA) mediates the tight linkage of sperm head and tail causing decapitation and infertility when faulty. Using mice as the experimental model, several proteins have already been identified affecting the linkage complex, manchette or tail formation when missing. However, our current knowledge is far too rudimentary to even draft an interacting protein network. Depletion of the major outer dense fiber protein 1 (ODF1) mainly caused decapitation and male infertility but validated binding partners collaborating in the formation of sperm-specific structures are largely unknown. Amongst all candidate proteins affecting the HTCA when missing, the structural protein CCDC42 attracted our attention. The coiled-coil domain containing 42 (CCDC42) is important for HTCA and sperm tail formation but is otherwise largely uncharacterized. We show here that CCDC42 is expressed in spermatids and localizes to the manchette, the connecting piece and the tail. Beyond that, we show that CCDC42 is not restricted to male germ cells but is also expressed in somatic cells in which it localizes to the centrosome. Although centrosomal and sperm tail location seems to be irrespective of ODF1 we asked whether both proteins may form an interacting network in the male germ cell. We additionally considered ODF2, a prevalent protein involved in the formation of spermatid-specific cytoskeletal structures, as a putative binding partner. Our data depict for the first time the subcellular location of CCDC42 in spermatids and deepen our knowledge about the composition of the spermatid/sperm-specific structures. The presence of CCDC42 in the centrosome of somatic cells together with the obvious restricted male-specific phenotype when missing strongly argues for a compensatory function by other still unknown proteins most likely of the same family.

Keywords: spermiogenesis, HTCA, tail, CCDC42, centrosome, ODF2, ODF1

## INTRODUCTION

The transformation of spermatids into terminally differentiated sperm is a key event in spermatogenesis. Meiosis II generates spermatids of spherical shape that are then gradually transformed by shaping and condensation of the nucleus and the formation of acrosome and sperm tail, finally resulting in the mature spermatozoon (Fawcett, 1975; Russell et al., 1990; Jan et al., 2012). There is circumstantial evidence indicating that the shaping of the nucleus and the assembly of the sperm tail is provoked by a transient microtubular structure, the manchette (Clermont et al., 1993; Kierszenbaum, 2002; Kierszenbaum et al., 2011). The manchette forms during the acrosomal phase of spermiogenesis and disassembles during the maturation phase (Clermont et al., 1993). In mice, it is first seen in step 8 spermatids and disassembles prior to the formation of the sperm mid-piece around steps 13–14 (O'Donnell and O'Bryan, 2014; Lehti and Sironen, 2016). The manchette consists of a perinuclear mantle of microtubules emanating from the perinuclear ring (Fawcett et al., 1971; Rattner and Brinkley, 1972; Dooher and Bennett, 1973; Rattner and Olson, 1973; Wolosewick and Bryan, 1977; Clermont et al., 1993). However, detection of plus-end tracking proteins as EB3 and CLIP-170 at the perinuclear ring together with the absence of the minus-end binding protein γ-tubulin strongly argues against the perinuclear ring as the microtubule nucleation site (Akhmanova et al., 2005; Kierszenbaum et al., 2011; Lehti and Sironen, 2016). Instead, supporting evidence indicates that the centriolar adjunct serves as a nucleator of manchette microtubules with their plus ends reaching toward the perinuclear ring (Fawcett and Phillips, 1969; Lehti and Sironen, 2016; Fishman et al., 2018). The manchette is linked to the nuclear membrane and this is essential for nuclear shaping. The presence of rod-like elements that link the manchette to the nuclear envelope has first been demonstrated by electron microscopy studies (Russell et al., 1991). Supportively, deletion of SUN4, a testis-specific nuclear membrane protein and component of the linker of nucleoskeleton and cytoskeleton complex (LINC) caused detachment of the manchette and consequently roundheaded sperm (Calvi et al., 2015; Pasch et al., 2015; Yang et al., 2018a). The importance of the manchette for nuclear shaping and male fertility is furthermore exemplified by the seminal discovery of the genetic cause underlying the azh phenotype in mice. Male azh mice are infertile due to a malformed manchette, abnormal spermatozoon head morphology, tail abnormalities and decapitation all caused by a deletion in the Hook1 gene (Mendoza-Lujambio et al., 2002). (Review in: Chen et al., 2016). However, Hook1 is a microtubule-binding protein and most likely responsible for the cross-linking of the manchette microtubules, whereas SUN4 is expected to be an inner nuclear membrane protein. Thus, the true nature of the rod-like elements that link the manchette to the nucleus is still unknown.

The observation of sperm decapitation indicated that the manchette is involved in sperm head to tail coupling and/or development of the sperm tail. Consequently, it was suggested that molecules required for the developing basal body/connecting piece and the sperm tail were delivered via intra-manchette transport meaning that the manchette functions as a track in supporting the delivery of molecules (Kierszenbaum, 2001, 2002; Kierszenbaum et al., 2011). Contradictory, however, are the observations that the manchette is assembled when the axoneme is already developed and that the sperm tail develops irrespective of the detachment of the manchette in SUN4-deficient spermatids (Lehti and Sironen, 2017; Yang et al., 2018a).

The sperm tail develops from the basal body that itself is a derivative of the former centrosome. In spermatids, the daughter centriole of the centrosome is transformed into the proximal centriole, which acts as a seed for the formation of the connecting piece, and inserts into the nuclear indentation (Fawcett and Phillips, 1969). The perpendicular positioned mother centriole is transformed into the distal centriole, which acts as the basal body to initiate sperm tail development. Later on, the distal centriole disintegrates leaving the centriolar vault. The axoneme, the microtubule-based core structure, is the prolongation of the distal centriole that is surrounded by accessory structures as the nine prominent outer dense fibers (ODFs) and the fibrous sheath (FS) in the sperm tail. The ODFs are descending from the segmented columns formed at the proximal centriole of the head-to-tail coupling apparatus (HTCA). They accompany the microtubule doublets of the axoneme throughout the length of the tail whereas the FS is present only in the principal piece. The accessory fibers are important for stiffening the sperm tail thus supporting the elastic recoil of the sperm tail and protecting against shearing forces (Baltz et al., 1990; Lindemann, 1996). At the proximal region of the sperm tail, at the mid-piece, the mitochondrial sheath surrounds axoneme and ODFs.

The HTCA or connecting piece develops from the centrosome. It is an articular structure at the neck region mediating the tight connection between the sperm tail and the nucleus. Although the protein composition of the HTCA is far from being known, a couple of proteins have already been identified that are essential for the formation of the HTCA and/or the sperm tail. One protein essential for the tight connection of sperm head and tail is the outer dense fiber protein 1 (ODF1; also named HSPB10) (Burfeind and Hoyer-Fender, 1991; Schalles et al., 1998; Fontaine et al., 2003). Depletion of ODF1 caused sperm decapitation and male infertility in mice (Yang et al., 2012, 2014). A few interacting proteins have been identified, e.g., the outer dense fiber protein 2 (ODF2), which is a major protein of the sperm tail accessory fibers (Shao et al., 1997). Beyond that, validated ODF1 interacting proteins that are supposed to collaborate in the formation of sperm-specific structures are currently unknown. Proteins known to affect the HTCA or the sperm tail when missing are ideal candidates as putative interaction partners. We, therefore, focused on structural proteins with a reported effect on HTCA and sperm tail formation as putative interaction partners of ODF1. We asked here, whether the coiled-coil domain containing 42 (CCDC42) protein acts as a node in the ODF1 network. Ccdc42 is specifically expressed in testis and brain and its deletion causes male sterility in mice with malformation of the HTCA and the sperm tail. Beyond that, no further phenotypes are evident (Pasek et al., 2016). CCDC42 (coiled-coil domain containing 42) belongs to the CFAP73 family and is a paralog of CFAP73. It contains the DUF4200, the domain of unknown function

that is shared by a couple of coiled-coil domain proteins and cilia-and flagella-associated proteins as CFAP73. The phenotype of Ccdc42-deficient mice suggested a male germ cell-specific function, but its expression and sub-cellular location is so far unknown. We explored here putative interacting proteins of CCDC42 in male germ cells and analyzed its expression and sub-cellular location. We show that CCDC42 is recruited to the manchette and the sperm tail and is specifically enriched in the perinuclear ring of the manchette and the HTCA. Pull down and co-IP experiments both indicated binding to ODF1 and ODF2. Furthermore, CCDC42 localizes to the centrosome/basal body not only in male germ cells but also in somatic cells. Revision of Ccdc42 expression by RT-PCR demonstrated wide-spread expression in somatic tissues. The co-localization of CCDC42 with microtubule-based structures as the manchette and the centrosome/HTCA suggests that CCDC42 is involved in their formation by generating a rigid scaffold. However, as no further phenotypes are evident when Ccdc42 is missing its function in somatic cells most likely might be taken over by other members of the family.

## MATERIALS AND METHODS

#### Ethics Statement

All mouse experiments were reviewed and approved by the local ethic commission. License for animal experiments has been obtained by the Institute of Human Genetics and the Max-Planck-Institute for Experimental Medicine, Göttingen. The guidelines of the German Animal Welfare Act (German Ministry of Agriculture, Health and Economic Cooperation) were strictly followed in all aspects of mouse work.

#### cDNA Synthesis and RT-PCR

Total RNA was prepared form adult mouse tissues as well as from NIH3T3 mouse fibroblasts using peqGOLD RNApureTM (PeqLab, Erlangen, Germany) following the recommendations of the manufacturer. Total RNA was digested with Ambion <sup>R</sup> TURBO DNA-freeTM DNase (Life Technologies) followed by cDNA synthesis using Maxima First Strand cDNA Synthesis (Thermo Fisher Scientific). RT-PCR for detection of transcribed sequences was performed using the following primer combinations: Ccdc42-Nterm\_For (GTGGCACTGTCACTCACC) and C-terminal-Ccdc42-Rev (GGCTCACCAGGAACCTTCTC) generating the full-length product of 1093 bp, Ccdc42-For2 (GGAGACCGAGAATCCA GCC) and Ccdc42-Rev2 (CCGTTGGAATGCCTCCTTCT) for amplification of 305 bp of the 5<sup>0</sup> region (exons 1 + 2), C-terminal-Ccdc42-For (GGAATCCACCCAAGTGTCCC) and C-terminal-Ccdc42-Rev (GGCTCACCAGGAACCTTCTC) generating a fragment of 203 bp of the conserved 3<sup>0</sup> region (exons 6 + 7), Ccdc42-Exon5-For (GAAGAGATCCACGAGGTG) and C-terminal-Ccdc42-Rev for amplification of exons 5-7 (expected fragment size 535 bp), Gapdh-For (GTATGA CTCCACTCACGGCA) and Gapdh-Rev (GTCAGATCCACGA CGGACAC) generating a fragment of 594 bp.

### Plasmid Constructs

PCR amplification based on the Ensembl reference sequence NM\_177779 by using the following primers: Ccdc42-NheI-For (5<sup>0</sup> -GGCTGTTAGGTAGCTAGCGCAAC CATGAGTTTGGG-3<sup>0</sup> ) and Ccdc42-HindIII-Rev (5<sup>0</sup> -GTTA CTTCCTTAAGCTTGCCATCCGGACTTGCTGTbTG-3<sup>0</sup> ), each primer containing restriction enzyme recognition sites. The full-length coding sequence of the Ccdc42 isoform 203 was first cloned into pJET1.2/blunt (Thermo Fisher Scientific) followed by NheI/HindIII digestion and sub-cloning into pCR3.1-Cherry resulting in an in-frame fusion with the C-terminal Cherrytag (pCR3.1, Invitrogen). The full-length coding region of Odf1 was N-terminally fused to ECFP in pECFP-C1 (Clontech Lab.). Odf2 was C-terminally fused to EGFP in pEGFP-N1 (Clontech Lab., #U55762) generating the full length construct 13.8NC-EGFP (Donkor et al., 2004). Sequencing always revealed correct reading frames.

#### Cell Culture and Immunocytochemistry

NIH3T3 (ATCC CRL-1658) or HEK-293 cells (ATCC CRL-1573) were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS), 1000 U/ml penicillin, 1000 µg/ml streptomycin, and 20 mM L-Glutamine (all Gibco) at 37◦C and 5% CO2. NIH3T3 cells were grown on coverslips in 6-well plates and transfected using EndoFectinTM Max Transfection Reagent, following the recommendations of the manufacturer (GeneCopoeia). Twentyfour hours after transfection, cells were washed in phosphatebuffered saline (PBS) and fixed in methanol for 10 min at −20◦C. Specimens were then permeabilized in 0.3% TritonX-100 in PBS for 10 min at room temperature and blocked for 1 h using blocking solution (PBS containing 1% BSA and 0.3% TritonX-100). Samples were incubated with primary antibodies toward CCDC42 (ARP52735\_P050, antibodies-online ABIN2785068), Pericentrin (PRB432C, Covance), acetylated tubulin (6-11B-1; Sigma-Aldrich), gamma-tubulin (GTU-88, Sigma-Aldrich), ODF1 (ABIN4341345, antibodies-online), and GFP (raised in rabbit, self-made) at 37◦C for 1 h. Secondary antibodies used are goat anti-mouse-IgG DyLight 488 (#35503, Thermo Fisher Scientific), and goat anti-rabbit MFP590 (#MFP-A1037, Mobitec). DNA was counterstained with DAPI. Images were taken by confocal microscopy (LSM 780, Zeiss) and processed using Adobe Photoshop 7.0.

#### Immunocytology on Testicular Cell Suspensions

Fresh testes from laboratory mice of strain C57/Bl6, or frozen epididymides from wild-type mice or Odf1-ko mice (Yang et al., 2012) were minced in PBS, transferred onto superfrost slides, and fixed either in 2% or 3.7% paraformaldehyde in PBS for 20 min. Cells were permeabilized afterward in 0.3% Triton X-100 in PBS for 10 min, followed by blocking for 1 hr in blocking solution (PBS containing 1% BSA and 0.3% Triton X-100). The following antibodies were used for immunocytology: anti-α-tubulin (mouse monoclonal, DM1A, Calbiochem), antiacetylated tubulin (6-11B-1, Sigma-Aldrich), anti-CCDC42

FIGURE 1 | CCDC42 localizes in the manchette, the perinuclear ring and the connecting piece during spermiogenesis. Suspension preparations of adult mouse testis were incubated with antibodies against acetylated tubulin (green) and CCDC42 (red). Weak staining for CCDC42 was observed in the cytoplasm in round spermatids (A–D). In elongating spermatids (eS) CCDC42 decorated the manchette and more strongly the perinuclear ring (E–L). Additionally, the HTCA or connecting piece harbored CCDC42 as well (I–L, arrow in J). CCDC42 also located to the connecting piece in sperm (M–P) and to the tail with highest expression in the principal piece (M–T). In detached tails, the anterior region, which corresponds to the attachment site to the nucleus, stained for CCDC42 (Q–T; framed in R and S and inset in R showing the enlarged region). Secondary antibodies used are anti-mouse IgG-Dylight488 and anti-rabbit IgG-MFP590 (A–H,M–P) or anti-rabbit IgG-Dylight488 and anti-mouse IgG-Alexa Fluor R 555 (I–L,Q–T). Nuclear staining with DAPI (blue). Bars are of 10 µm except for M–P in which 5 µm scales are used.

(ARP52735\_P050; antibodies-online ABIN2785068), anti-ODF1 (antibodies-online, ABIN4341345), guinea pig anti-SUN4 (selfmade, Manfred Alsheimer, Würzburg). Primary antibodies were detected using different combinations of secondary antibodies as goat anti-mouse-IgG Alexa Fluor 555 IgG (H + L) (A21422, Molecular Probes) and goat anti-rabbit-IgG DyLight 488 (#35553, Thermo Fisher Scientific), goat anti-mouse-IgG DyLight 488 (#35503, Thermo Fisher Scientific) and goat anti-rabbit-MFP590 (#MFP-A1037, Mobitec), goat anti-mouse-IgG DyLight 488 (#35503, Thermo Fisher Scientific) and goat anti-rabbit-IgG (H + L) Alexa Fluor R 555 (F[ab]2 fragment; #A21430, Life Technologies), and goat anti-guinea pig-IgG Cy3 (Dianova #106-166-003). DNA was counterstained with DAPI (4<sup>0</sup> , 6-Diamidino-2-phenylindole; Sigma D-9542), and the acrosome was decorated with FITC-labeled peanut lectin (PL-FITC). Images were taken by confocal microscopy (LSM 510, Zeiss) and processed using Adobe Photoshop 7.0. In some pictures, the fluorescent colors are replaced by pseudo-colors.

#### Co-immunoprecipitation

HEK-293 cells were transfected using EndoFectinTM Max Transfection Reagent (GeneCopoeia), and 24 h post-transfection harvested by trypsinization followed by two rinses in PBS. The cell pellet was resuspended in 1 ml lysis buffer (150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 7.6, containing protease inhibitor cocktail (Halt Protease Inhibitor Cocktail 100x, Thermo Fisher Scientific, #78438). Subsequently, cell lysates were passed 10x through a syringe with a 21-gauge needle, followed by 3-times sonication for 45 s each. The soluble fraction was obtained by centrifugation at 15,000 × g for 15 min at 4◦C. The supernatant was split into three parts and incubated with different antibodies for co-IP, either with rabbit anti-GFP (self-made), rat anti–RedFP (5F8, Chromotek), or rabbit anti-GAL4 (DBD) (sc-577, Santa Cruz Biotechnology). Protein G agarose beads (Thermo Fisher Scientific) were washed three times with lysis buffer and the antibody/supernatant mixture was then added. Samples were incubated overnight at 4◦C on a rotating wheel. Afterward, beads were washed four times with lysis buffer, and bound proteins obtained by suspension of beads in SDS sample buffer and heating at 95◦C for 10 min. Proteins were fractionated on SDS-PAGE and transferred onto nitrocellulose membrane (Amersham Hybond-ECL, GE Healthcare) (Laemmli, 1970; Towbin et al., 1979). The membrane was incubated in blocking solution [5% dry milk in TBST (10 mM Tris-HCl, pH 7.6, 150 mM NaCl 0.05%

Tween 20)] for 1 h. Membranes were incubated overnight at 4 ◦C with primary antibodies either rabbit anti-GFP (self-made), mouse anti-GFP (MAB3580, Chemicon), or rat anti-RedFP (5F8, Chromotek) in blocking solution. Following washing of membranes in TBST, they were incubated with horseradish peroxidase-conjugated secondary antibodies either anti-rabbit IgG, anti-mouse IgG, or anti-rat IgG (Jackson ImmunoResearch, WestGrove, PA, United States or Sigma Biosciences, St Louis), respectively. Chemiluminescence detection was performed using ClarityMax Western ECL Substrate (Bio-Rad, #1705062) and images captured with Chemdoc (Bio-Rad).

#### Pull Down Assay

Preparation of bacterially expressed and refolded His-tagged ODF2-fusion protein (6xHis-13.8NC) was essentially as described in Yang et al. (2018b). Ni-NTA agarose (Qiagen GmbH, Hilden) was washed in wash buffer (50 mM NaPi, 500 mM NaCl, 30 mM imidazole, pH 7.6, containing protease

inhibitors (Halt Protease Inhibitor Cocktail 100x, Thermo Fisher Scientific, #78438) and 0.2 mM PMSF). For in vitro interaction, 6xHis-ODF2 proteins were added to the washed Ni-NTA agarose and incubated for 1 h at 4◦C in constant agitation. NIH3T3 cell lysate containing CCDC42-Cherry proteins was split into halves and one half added to the resin followed by incubation for 2 h at 4◦C and constant agitation. The second half was used for the negative control. The beads were washed four times in wash buffer, followed by a final overnight washing step. To elute bond proteins, Ni-NTA agarose was incubated for 10 min at 4 ◦C in elution buffer (50 mM NaPi, 500 mM NaCl, 500 mM imidazole, 0.2 mM PMSF). Eluates were boiled in SDS-sample buffer and analyzed by Western-blotting using rabbit anti-ODF2 antibodies (ESAP 15572, ABIN2430582, antibodies-online) and rat anti-RedFP antibodies (5F8, Chromotek). As negative control, the resin was incubated with one half of the NIH3T3 cell lysate containing CCDC42-Cherry proteins but without 6xHis-ODF2 proteins and processed as described.

#### RESULTS

## CCDC42 Localizes to Manchette, Perinuclear Ring, Connecting Piece and Sperm Tail

The coiled-coil domain containing protein of 42 kDa (CCDC42) belongs to the DUF4200 family of proteins containing the domain of unknown function 4200. Important paralogs of CCDC42 are CFAP73, CCDC38, and CFAP100, also known as CCDC37/MIA1. CCDC42 orthologs are present in most organisms from vertebrates to choanoflagellates (Ruan et al., 2008; Guindon et al., 2010). Despite their widespread occurrence, information is currently scarce. Ccdc42 is specifically expressed in testis and brain but deletion of Ccdc42 seems to affect exclusively male germ cells resulting in infertility (Pasek et al., 2016). Ccdc42-deficient spermatids developed malformed HTCA and sperm tail that are functionally insufficient. However, beyond that no further information about CCDC42 is available. We first studied the subcellular localization of CCDC42 during spermatogenesis in the mouse using a commercially available antibody. The antibody was first validated by immunocytology and Western blotting using the Ccdc42-expression construct (validation results are online at antibodies-online and are shown in **Figure 9**<sup>1</sup> ).

Weak expression of CCDC42 was first observed in the cytoplasm of round spermatids (**Figures 1A–D**). In elongating spermatids CCDC42 co-localized with the manchette microtubules decorated by acetylated tubulin and more strongly with the perinuclear ring which marks the anterior border of the manchette (**Figures 1E–L**) Additionally, CCDC42 localized to the connecting piece detectable as two adjacent spots at the posterior end of the nucleus (**Figures 1I–L**, arrow in J). In sperm, CCDC42, again, was found at the connecting piece region and located to the sperm tail. CCDC42 strongly decorated the principal piece but only weakly the middle piece whereas acetylated tubulin marked the whole tail (**Figures 1M–P**). Labeling of detached sperm tails confirmed prevalent localization of CCDC42 in the principal piece whereas acetylated tubulin marked the whole tail. However, the anterior region of detached sperm tails, which corresponds to the former attachment site of the tail to the head, showed presence of CCDC42 visible by

<sup>1</sup>https://www.antibodies-online.com

FIGURE 5 | Co-localization of CCDC42-Cherry and ODF1-ECFP in NIH3T3 cells. NIH3T3 cells were transfected with expression plasmids encoding either CCDC42-Cherry or ODF1-ECFP as marked on the left side. Proteins were detected either by their fluorescent tag (CCDC42-Cherry or ODF1-ECFP) or by immunostaining (anti-ODF1, anti-gamma tubulin, or anti-Pericentrin). (A–D) CCDC42-Cherry co-localizes with ODF1-ECFP when ectopically expressed. Ectopically expressed ODF1-ECFP localizes to the centrosome as verified by immunostaining using the centrosomal marker proteins γ-tubulin (E–H) or Pericentrin (I–L). Immunostaining for γ-tubulin (M–P) or Pericentrin (Q–T) demonstrates centrosomal location of CCDC42-Cherry as well. Nuclear counterstain with DAPI (blue). Bars are of 5 µm (A–D) or 2 µm (E–T).

(B) as first antibodies followed by detection of first antibodies using the secondary antibodies anti-rat IgG (for detection of CCDC42-Cherry) or anti-mouse IgGs (for detection of ODF1-ECFP). ODF1-ECFP, captured by rabbit anti-GFP antibody, co-precipitated CCDC42-Cherry with an expected molecular mass of ∼70 kDa. (A) When CCDC42-Cherry was captured using the rat anti-RedFP-antibody (B), ODF1-ECFP was co-purified as well (B). Immunoblotting using the first antibody against GFP raised in mouse (B), followed by the secondary anti-mouse IgG antibody, detected the expected ODF1-ECFP protein of ∼54 kDa but only weakly reacted with the control IgGs, which were raised in rabbit (control) and which show a slightly higher molecular mass.

two adjacent spots (**Figures 1Q–T**; framed in R, S and enlarged inset in R). A stronger staining of the principal piece of the sperm tail than of the mid-piece might either reflect an unequal distribution of CCDC42 along the sperm tail or is caused by different accessibilities of the antibodies due to the presence of the mitochondrial sheath in the mid-piece.

Decoration of the manchette, the perinuclear ring, and the connecting piece was also demonstrated by using antiα-tubulin antibody staining in conjunction with anti-CCDC42, both antibodies subsequently detected by varying secondary antibodies (**Figure 2**). The control, with anti-α-tubulin antibody incubation but omitting anti-CCDC42 antibody, followed by both secondary antibodies subsequently, revealed only α-tubulin staining thus supporting specificity of anti-CCDC42 staining (**Figures 2I–L**).

When concurrently stained for the nuclear envelope protein SUN4, the weak CCDC42 staining of the cytoplasm in round and early elongating spermatids was confirmed (**Figures 3A– H**). The SUN4 positive domain partially overlapped with the CCDC42 positive region as this is the region where the manchette develops. In early elongating spermatids, the SUN4 and CCDC42 localization domains seemingly overlapped corresponding most likely to the region where the manchette has formed and to which SUN4 locates (I-L) (Yang et al., 2018a). Again, CCDC42 more strongly decorated the perinuclear ring (I–L). The weak cytoplasmic staining for CCDC42 in the round spermatid seemed to be beyond background staining as demonstrated by the control experiment (**Figures 3M–O**).

## CCDC42 Interacts With ODF1 and ODF2

The small heat shock protein ODF1/HSPB10 is a main protein component of the sperm tail ODFs. Beyond that, it locates to the connecting piece and is essential for the tight connection between head and tail (Schalles et al., 1998; Yang et al., 2012, 2014). Immunocytological inspection confirmed sperm tail location of ODF1 and, additionally, showed expression in the manchette of elongating spermatids (**Figure 4**). Location of ODF1 thus resembled that of CCDC42 raising the question whether both proteins also physically interact.

We first investigated whether the location of ODF1 and CCDC42 is interdependent when ectopically expressed in NIH3T3 mouse fibroblasts. Cells were transfected with expression plasmids encoding either CCDC42 fused to Cherry (CCDC42-Cherry) or ODF1 fused to ECFP (ODF1-ECFP) and the proteins detected either by their fluorescent tags or by immunostaining (**Figure 5**). Ectopic expression of CCDC42-Cherry revealed bright staining of one or two dots close to the nucleus that overlap with ODF1-ECFP expression (**Figures 5A–D**). Since twin-dots are a typical signature of the centrosome, we verified centrosomal location of ODF1-ECFP (**Figures 5E–L**) as well as of CCDC42- Cherry (**Figures 5M–T**) using immunostaining for the

for acetylated tubulin (green) and CCDC42 (red). Nuclear staining with DAPI. Loss of ODF1 caused head detachment. The sperm head is, therefore, missing in Odf1-ko sperm (E–H). No red staining was evident when anti-CCDC42 antibody incubation was omitted (I–L, control). A–H: 2 µm bars, I–L: 5 µm bars.

FIGURE 8 | CCDC42 interacts with ODF2. (A–D) Co-transfection of expression plasmids encoding ODF2-EGFP (green) or CCDC42-Cherry (red) in NIH3T3 cells and detection by their fluorescent tags. Fusion proteins co-localize in the cytoplasm. Nuclear staining with DAPI (blue). Bars of 10 µm. (E) In vitro interaction of His-tagged ODF2 (6xHis-ODF2/6xHis-13.8NC) and Cherry-tagged CCDC42. Co-purification of bacterially expressed 6xHis-ODF2 and CCDC42-Cherry, ectopically expressed in NIH3T3 cells, by affinity purification of 6xHis-ODF2 using Ni-NTA agarose (pull down). Low binding of CCDC42-Cherry to the agarose beads in the absence of 6xHis-ODF2 (control). Proteins were eluted from the beads, and analyzed by immunoblotting using antibodies against ODF2 (6xHis-ODF2) or RedFP (CCDC42-Cherry).

centrosomal marker proteins γ-tubulin (**Figures 5E–H,M– P**) or Pericentrin (**Figures 5I–L,Q–T**) (Doxsey et al., 1994). Our results show that CCDC42-Cherry co-localizes with ODF1-ECFP, when both proteins are ectopically expressed in NIH3T3 cells, notably at the centrosome. However, recruitment of CCDC42-Cherry to the centrosome is independent of ODF1 due to the fact that ODF1 is not at all expressed in somatic cells and CCDC42 locates to the centrosome despite absence of ODF1-ECFP (**Figures 5M–T**) (Yang et al., 2012).

Nevertheless, the physical interaction between CCDC42-Cherry and ODF1-ECFP was proven by coimmunoprecipitation. Both proteins were ectopically expressed in cultured cells by transient transfection of expression plasmids, and one of either protein immunoprecipitated out of the cell lysate. The immunoprecipitate was analyzed by immunoblotting detecting the protein that co-precipitated with the fished protein (**Figure 6**). Co-immunoprecipitation of CCDC42-Cherry and ODF1-ECFP was verified in either direction, capturing either CCDC42-Cherry or ODF1-ECFP. Our data thus indicate physical interaction between CCDC42 and ODF1. However, ODF1 is not only dispensable for recruitment of CCDC42 to the centrosome but also to the sperm tail. Immunocytology on epididymal sperm of Odf1-ko mice showed decoration for CCDC42 similar as in wild-type sperm (**Figure 7**, wild-type sperm in A-D, Odf1 ko sperm in E-H). Acetylated tubulin staining identified the sperm tail. As ODF1 is essential for the anchorage of the sperm head to the tail causing head detachment when missing, the sperm head is absent in ODF1-ko sperm (**Figures 7E–H**) (Yang et al., 2012). In the control staining, when omitting anti-CCDC42 incubation but subsequent incubation with both secondary antibodies, the sperm tail is clearly visible by acetylated tubulin decoration but missed any red staining (**Figures 7I–L**).

Another important protein of the sperm tail and the connecting piece is ODF2. ODF2, furthermore, is an essential component of the centrosome and the basal body in somatic cells (Brohmann et al., 1997; Schalles et al., 1998; Nakagawa et al., 2001; Hüber et al., 2008). We, therefore, asked whether ODF2 is another binding partner of CCDC42. Co-transfection assays of expression plasmids in NIH3T3 cells revealed similar location of ODF2-EGFP and CCDC42-Cherry. ODF2 fused to EGFP (13.8NC-EGFP) often generated fibrous structures to which CCDC42-Cherry proteins localize (**Figures 8A–D**). Additionally, a physical interaction between CCDC42-Cherry and ODF2 was proven by pull-down assays (**Figure 8E**). Bacterially expressed and refolded 6xHis-tagged ODF2 was affinity purified using Ni-NTA agarose in the presence of CCDC42-Cherry, ectopically expressed in cell culture. Thereafter, eluates were immunoblotted for detection of the target protein 6xHis-ODF2 and its putative binding partner CCDC42-Cherry. CCDC42-Cherry co-purified with 6xHis-ODF2 (**Figure 8E**, pull down), whereas almost no binding of CCDC42-Cherry to the beads was observed in the absence of 6xHis-ODF2 (**Figure 8E** control). Our data, therefore, indicate ODF2 as another binding partner of CCDC42.

marker protein γ-tubulin (anti-gamma-tubulin, green). (I–L) Omitting anti-CCDC42 antibody incubation showed no red decoration of the centrosome, which was otherwise detected by anti-γ-tubulin staining (anti-gamma-tubulin, green), demonstrating anti-CCDC42 antibody specificity. Nuclear counterstain with DAPI (blue). Bars are of 2 µm (A–D,I–L) or of 5 µm (E–H).

## CCDC42 Is a Centrosomal Protein in Somatic Cells

By ectopic expression of CCDC42-Cherry in NIH3T3 cells we have observed a predominant centrosomal location (**Figure 5**). This prompted us to investigate whether CCDC42 is endogenously expressed in somatic cells being a novel component of the centrosome. We first transfected cells with the CCDC42-Cherry expression plasmid for validation of the anti-CCDC42 antibody (**Figures 9A–D**). The antibody specifically decorated only transfected cells and co-localized with the fusion protein CCDC42-Cherry thus demonstrating its validity (**Figures 9A–D**). We next incubated untransfected NIH3T3 cells with the anti-CCDC42 antibody. Immunostaining of the endogenous CCDC42 decorated a twin-spot near the nucleus that additionally stained for the centrosomal marker γ-tubulin (**Figures 9E–H**). The centrosome was exclusively decorated by γ-tubulin staining but did not show a red fluorescence when omitting anti-CCDC42 antibody incubation (**Figures 9I–L**, control). Immunocytological data thus indicate expression of CCDC42 in somatic cells, which is contradictory to its reported restricted expression pattern.

## Expression of Ccdc42 Isoforms

According to Pasek et al. (2016), Ccdc42 is expressed in testis and brain. In mouse testes, weak expression was first observed at 10-days of age that raised in 15-days old testis and maintained into adulthood. Testicular expression thus corresponds with the onset of meiosis around day 10 and its increase roughly coincides with the progression of spermatid differentiation during spermiogenesis (Nebel et al., 1961). However, three putative CCDC42 isoforms have been reported in mice (UniProtKB – Q5SV66) produced by alternative splicing (**Figure 10**). The longest isoform 203 (Q5SV66) consists of 316 amino acids (aa). In isoform 201 (Q5SV65) the sequence encoded by exon 5 is missing resulting in a putative protein of 238 aa. Isoform 202 (Q5SV66-2) has a postulated length of 169 aa since the N-terminal end encoded by exons 1-4 is completely missing. Instead, translation of the protein starts at the 3<sup>0</sup> end of

intron 4 encoding sequence MALGSQLFSDPSPLIPQ upstream of exon 5 encoded sequences. According to Interpro, the domain of unknown function 4200 (DUF4200) comprises aa 44-161 at the N-terminal half of isoform 203 and is therefore present in both, isoform 203 as well as in isoform 201, albeit shortened in the latter, but is largely missing in isoform 202 (**Figure 10**, depicted in violet). The coiled-coil region at the C-terminal end is present in all three isoforms (according to SMART; in **Figure 10** highlighted in yellow). The antibody ABIN2785068, used for immunocytology, detects the epitope of 50 aa in the DUF4200, which is present in isoforms 203 and 202 but not in 201 (**Figure 10**, enframed in red).

The epitope, to which antibody ABIN2785068 is directed, is framed in red.

Expression of Ccdc42 was reported to be restricted to testis. In somatic tissues, with the only exception being the brain, Ccdc42 seems to be not expressed. We confirmed testicular expression by RT-PCR using different primer combinations (**Figure 11**). The full-length product of isoform Ccdc42-203 was expected to be of 1093 bp, which was confirmed by RT-PCR (**Figure 11A**, exons 1–7). However, for the putative isoform Ccdc42-201, a length of 729 bp was expected since exon 5 was skipped but we could not amplify the expected fragment (**Figure 11A**). Amplification of 305 bp of the 5<sup>0</sup> region (exons 1 + 2) again confirmed testicular expression of Ccdc42 (**Figure 11A**, exons 1 + 2). When amplifying 203 bp of the conserved 3<sup>0</sup> region, Ccdc42 expression was also demonstrated in NIH3T3 mouse fibroblasts (**Figure 11A**, exons 6 + 7).

Our results thus indicate that expression of Ccdc42 isoforms is not restricted to testis. This prompted us to revise Ccdc42 expression in other tissues (**Figure 11B**). We performed a nested PCR in order to detect even low expression levels. The first RT-PCR was performed to amplify exons 5–7 encoding part of the DUF domain with the epitope detected by the antibody, and the conserved coiled-coil region. We found strong expression in testis and epididymides and weak expression in the brain.

Sequencing of the RT-PCR products generated form testis and epididymides confirmed Ccdc42. Gapdh amplification, albeit demonstrating successful cDNA synthesis in all tissues, also indicated low amounts of cDNA in spleen and brain, which most likely accounts for the weak RT-PCR band found in brain cDNA. When performing a nested PCR on the first PCR products by amplifying exons 6 + 7 a fragment of the expected size was found in all tissues, with the exception of kidney. The RT-PCR fragments generated in ovary and NIH3T3 cDNA were sequenced confirming Ccdc42 amplification. Our results,

side by side on one gel but uninformative lanes were skipped from the picture

as indicated by white lines.

therefore, indicate that Ccdc42 expression is not restricted to testis. However, since the full-length product and the 5<sup>0</sup> region (exons 1 + 2) were found only in testis but not in NIH3T3 cells it is most likely that in somatic cells another isoform than the full-length CCDC42 isoform 203 is predominantly expressed, and that this isoform was detected by immunocytology in the centrosome.

### DISCUSSION

The HTCA is a complex structure present in the neck region of the sperm interconnecting the head and the tail. It develops from the centrosome that itself is composed of a pair of centrioles and associated components. During spermiogenesis, the proximal centriole inserts into a nuclear indentation, known as the implantation fossa, opposed to the acrosomal cap, and the linkage complex and the longitudinal columns of the connecting piece are formed (Fawcett and Phillips, 1969). The sperm tail starts outgrowing from the distal centriole of the former centriole that is now the basal body. In order to transmit only one centriole during fertilization, sperm of most vertebrates have disintegrated the distal centriole leaving only the proximal centriole. In contrast to centriole reduction in most vertebrates, in mice and other rodents, both distal and proximal centrioles degenerate during spermiogenesis leaving the centriolar vaults (Schatten, 1994; Hoyer-Fender, 2011). The current dogma of centrosome reduction in sperm was recently revisited by investigating the centrosomal protein inventory in human sperm. These data showed that the distal centriole is remodeled into an atypical centriole surrounded by a pericentriolar matrix instead of being completely vanished (Fishman et al., 2018). It is, therefore, feasible to view the neck structure as a specialized form of the pericentriolar matrix, and the centrioles as nucleation site for both the sperm tail and the manchette MTs. Few proteins are currently known that affect the head to tail linkage when missing, in between ODF1 and CCDC42 (Yang et al., 2012; Pasek et al., 2016). ODF1 is located in the sperm tail ODFs and in the connecting piece (Schalles et al., 1998). It interacts with ODF2, the major outer dense fiber protein, but no further interacting proteins have been confirmed (Shao et al., 1997). To figure out the interrelationship of HTCA proteins, and how they function in the formation of the HTCA, ODF1 interacting proteins are of utmost importance.

The coiled-coil domain containing 42 protein CCDC42 is highly conserved in evolution with orthologs existing in most organisms from vertebrates to choanoflagellates (Ruan et al., 2008; Guindon et al., 2010). Additionally, important paralogs of CCDC42 (also named CCDC42A) exist as CFAP73 (also named CCDC42B and MIA2), CCDC38 and CFAP100 (also named CCDC37 and MIA1). They altogether constitute the CFAP73 protein family. These proteins are in essential coiled-coil domain proteins and share the domain of unknown function DUF4200. Coiled-coil domain containing proteins are often involved in ciliary motility (Inaba and Mizuno, 2016; Zur Lage et al., 2019). In C. reinhardtii the gene product of the CCDC42 homolog MIA2 is a dynein regulator and necessary for ciliary motility

(Yamamoto et al., 2013). Further coiled-coil domain proteins as CCDC39 and CCDC40 are essential for ciliary motility by assembly of the dynein regulatory complex (Becker-Heck et al., 2011; Merveille et al., 2011; Blanchon et al., 2012). Ccdc42-ko mice are phenotypically normal but males are sterile. Sterility of Ccdc42-deficient male mice is most likely caused by the malformation of the HTCA and the sperm tail resulting in functional insufficiency (Pasek et al., 2016). These data suggest that CCDC42 has an important function in male germ cells but is otherwise dispensable (Pasek et al., 2016). Since CCDC42-deficiency affected exclusively the male germ cell, CCDC42 seems not to be involved in ciliary motility because otherwise, a more generalized phenotype has to expected. CCDC42-deficient spermatids are characterized by a multiplicity of the HTCA, defective nuclear shaping despite presence of a manchette, dislocation of the HTCA from its implantation site and a loss of flagellar outgrowth from the HTCA (Pasek et al., 2016).

We have demonstrated by immunocytology that CCDC42 colocalizes with ODF1- and ODF2-fusion proteins when ectopically expressed, and with structures comprising these proteins endogenously. Furthermore, co-precipitation by pull-down and co-immunoprecipitation experiments indicated binding of CCDC42 to ODF1 as well as to ODF2. However, recruitment of CCDC42 to the sperm tail and the centrosome in somatic cells does not require ODF1. CCDC42 consists in essential of coiled-coil domains, which is also the main feature of ODF2. Coiled-coil domains are important oligomerization domains mediating homodimerization as well as heterodimerization (Mason and Arndt, 2004). It is, therefore, most likely that ODF2 and CCDC42 interact by means of their coiled-coil domains. This mutual interaction might contribute to the stabilization of important cytoskeletal structures potentially mediated by ODF1. Although the true molecular function of ODF1 is not known, it belongs to the small heat shock protein family and might, therefore, act as a chaperone in protein folding (Fontaine et al., 2003). The protein complex consisting of the core proteins ODF2/ODF1/CCDC42 may thus build the rigid scaffold essential for the formation of the connecting piece and the sperm tail. When missing any one of these proteins the rigid scaffold is damaged causing failure of the linkage complex and the sperm tail. Whether CCDC42 interacts with any of those proteins that have a reported function in HTCA formation, as Centrin 1, Centrobin, Spata6, or Azi1/Cep131 awaits further investigation (Liska et al., 2009; Avashti et al., 2013; Hall et al., 2013; Yuan et al., 2015). However, since CCDC42-deficient sperm often show two instead of one basal body inserted into the nuclear membrane, centriole duplication, as well as correct attachment of the centrioles to the implantation fossa seem to be affected (Pasek et al., 2016). As similar phenotypes have been observed concerning mutations in Centrobin, Centrin 1, and Azi/Cep131 a functional interaction with CCDC42 is likely (Liska et al., 2009; Avashti et al., 2013; Hall et al., 2013). Our observation that CCDC42 is expressed in the manchette and in particular in the perinuclear ring illuminates its involvement in the acrosome-acroplaxome complex formation and nuclear shaping that are both affected when CCDC42 is missing. Since loss of CCDC42 did not prevent manchette formation, it is most likely involved in stabilizing the manchette or is a passenger protein transported via the manchette. Its accumulation in the perinuclear ring, however, points toward a stabilizing function and its involvement in the attachment of the manchette to the nuclear membrane.

Our data show that CCDC42 expression is not restricted to testis and brain but instead is found also in somatic tissues. We detected the endogenous protein in the centrosome of somatic NIH3T3 cells and hence identified CCDC42 as a novel component of the centrosome and the sperm tail not found before by large scale proteomics screens (Amaral et al., 2013; centrosome database Centrosome:DB). However, albeit RT-PCR experiments confirmed expression of Ccdc42 in somatic tissues, the full-length sequence could only by amplified from testis cDNA. It is therefore probable that the full-length isoform 203 is restricted to testis or more specifically to male germ cells whereas another isoform is expressed in somatic tissues. We could not verify expression of isoform 202, which starts with translated sequences encoded by intron 4 since a primer that binds to these 5<sup>0</sup> sequences has amplified an unrelated almost unknown sequence (C6H1orf158). Furthermore, we got no indications by RT-PCR of isoform 201, which was expected to be encoded by a smaller cDNA due to skipping of exon 5. Our data additionally show that CCDC42 is a component of the centrosome in somatic cells and most likely functions in scaffolding the centrosome via interaction with ODF2/Cenexin. However, since the only obvious phenotype of CCDC42-deficient mice is male infertility, CCDC42 is either dispensable for the somatic centrosome or its function has been taken over by other members of the CFAP73 family.

## DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

## AUTHOR CONTRIBUTIONS

CTC did the experiments and prepared the figures. SH-F was the project leader and wrote the manuscript. Both authors read and approved the final manuscript.

## FUNDING

CTC got a Ph.D. grant of the DAAD/CONICYT BECAS Chile program.

## ACKNOWLEDGMENTS

We thank Manfred Alsheimer (Würzburg) for the kind gift of α-SUN4 antibody.

## REFERENCES



essential for normal ciliary motility. J. Cell Biol. 201, 263–278. doi: 10.1083/jcb. 201211048


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Tapia Contreras and Hoyer-Fender. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Microtubule-Depolymerizing Kinesin-13 Klp10A Is Enriched in the Transition Zone of the Ciliary Structures of Drosophila melanogaster

#### Veronica Persico<sup>1</sup> , Giuliano Callaini<sup>2</sup> \* and Maria Giovanna Riparbelli<sup>1</sup>

<sup>1</sup> Department of Life Sciences, University of Siena, Siena, Italy, <sup>2</sup> Department of Medical Biotechnologies, University of Siena, Siena, Italy

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Bénédicte Durand, Université Claude Bernard Lyon 1, France Maurice Kernan, Stony Brook University, United States

> \*Correspondence: Giuliano Callaini callaini@unisi.it

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

> Received: 31 May 2019 Accepted: 07 August 2019 Published: 21 August 2019

#### Citation:

Persico V, Callaini G and Riparbelli MG (2019) The Microtubule-Depolymerizing Kinesin-13 Klp10A Is Enriched in the Transition Zone of the Ciliary Structures of Drosophila melanogaster. Front. Cell Dev. Biol. 7:173. doi: 10.3389/fcell.2019.00173 The precursor of the flagellar axoneme is already present in the primary spermatocytes of Drosophila melanogaster. During spermatogenesis each primary spermatocyte shows a centriole pair that moves to the cell membrane and organizes an axoneme-based structure, the cilium-like region (CLR). The CLRs persist through the meiotic divisions and are inherited by young spermatids. During spermatid differentiation the ciliary caps elongate giving rise to the sperm axoneme. Mutations in Klp10A, a kinesin-13 of Drosophila, results in defects of centriole/CLR organization in spermatocytes and of ciliary cap assembly in elongating spermatids. Reduced Klp10A expression also results in strong structural defects of sensory type I neurons. We show, here, that this protein displays a peculiar localization during male gametogenesis. The Klp10A signal is first detected at the distal ends of the centrioles when they dock to the plasma membrane of young primary spermatocytes. At the onset of the first meiotic prometaphase, when the CLRs reach their full size, Klp10A is enriched in a distinct narrow area at the distal end of the centrioles and persists in elongating spermatids at the base of the ciliary cap. We conclude that Klp10A could be a core component of the ciliary transition zone in Drosophila.

Keywords: male gametogenesis, Klp10A, ciliary structures, transition zone, centrioles, Drosophila

## INTRODUCTION

Gametes are specialized cells that give rise to the offspring. Male and female gametes are produced by spermatogenesis and oogenesis occurring in testis and ovary, respectively. During spermatogenesis, the spermatogonial cells undergo mitotic divisions from which originate the spermatocytes that give rise to the haploid spermatids. Spermatids are round cells that differentiate during the final phase of spermatogenesis, called spermiogenesis, undergoing morphological changes to become mature sperm.

Several sperm dysfunctions cause infertility due to axoneme anomalies. Axonemal defects in primary cilia are often observed in human diseases, ciliopathies, in which genetic mutations affect ciliary assembly (Badano et al., 2006; Bisgrove and Yost, 2006; Baker and Beales, 2009;

Nigg and Raff, 2009; Bettencourt-Dias et al., 2011; Waters and Beales, 2011; Inaba and Mizuno, 2015). In light of these observations, it appears of particular interest to study the mechanisms and the proteins involved in axoneme assembly.

The process of spermatogenesis shows fundamental similarities across the various phyla. For example, several cellular and regulatory mechanisms of spermatogenesis found in Drosophila melanogaster are conserved in other organisms, including humans. Drosophila melanogaster is a powerful system for studying the spermatogenesis and spermiogenesis processes, since it is relatively easy to examine the different stages involved in sperm development. The Drosophila testes consist of two close-ended tubules where the different stages of spermatogenesis are arranged in a chronological order. At the apical tip of the testis there is a cluster of somatic cells, the hub, that together with the Germline Stem Cells (GSCs) and the Somatic Stem Cells (SSCs) form the stem cell niche (Hardy et al., 1979). The hub is surrounded by 6–9 GSCs each associated with two SSCs. Each GSCs undergoes an asymmetrical division that originates two daughter cells: one cell remains in contact with the hub and maintains stemness, the other cell, called gonioblast moves away and begins differentiation (Yamashita et al., 2005). The gonioblast, enclosed by two non-dividing SCCs undergoes four mitotic divisions with incomplete cytokinesis leading to 16 spermatogones. These cells grow and become primary spermatocytes that produce 64 haploid spermatids at the end of meiosis (Fuller, 1993). The spermatids undergo spermiogenesis during which several cytological events transform the round spermatids into mature spermatozoa (Fabian and Brill, 2012).

Early spermatocytes inherit a centrosome that soon duplicates, so that at the beginning of the first meiotic prophase the germ cells hold two pairs of short centrioles composed of nine triplet microtubules and a central cartwheel (Riparbelli et al., 2009, 2012). The centrioles of each pair migrate toward the cell surface where each of them organizes the axoneme of a cilium like region (CLR) that protrudes from the cell membrane (Tates, 1971; Fritz-Niggli and Suda, 1972; Riparbelli et al., 2012). The centriole/CLR complexes increase in length during prophase progression; therefore, elongated centrioles and extended axonemes are found in mature spermatocytes (Gottardo et al., 2013). The CLRs remain throughout meiosis and are internalized with the centrioles to organize the spindle microtubules. The second meiotic division in Drosophila and most insects is not preceded by centriole duplication. Consequently, the secondary spermatocytes have only one centriole at each spindle pole. At the end of meiosis, the complex centriole/CLR is inherited by the spermatid and will be the precursor of the sperm flagellum (Gottardo et al., 2013). Since centrioles and CLRs are easily visible, they represent a very useful model for studying the localization of proteins involved in centriole and axoneme organization.

In addition to the CLRs found in male germ cells, Drosophila displays another type of ciliary structure, the sensory cilia that are associated with type-I sensory neurons (Gogendeau and Basto, 2010; Jana et al., 2016). Although, sensory cilia and CLRs are similar structures, their assembly relies on different mechanisms. Sensory cilia formation requires intraflagellar transport (IFT) mechanisms that are dispensable for CLR growth and sperm flagella elongation (Han et al., 2003; Sarpal et al., 2003; Avidor-Reiss and Leroux, 2015). The IFT-mediated process of axoneme assembly in canonical primary cilia depends on the transition zone (TZ), a specialized region at the junction between the centriole and the axoneme that is involved in cilia maintenance and compartmentalization (Reiter et al., 2012; Gonçalves and Pelletier, 2017) and is characterized by the Ylinks, evolutionarily conserved structures that span the space between the doublet microtubules and the plasma membrane (Fisch and Dupuis-Williams, 2011; Czarnecki and Shah, 2012; Garcia-Gonzalo and Reiter, 2012).

The base of each sensory cilia in Drosophila displays Ylike structures (Vieillard et al., 2016; Jana et al., 2018) and contains some conserved TZ module proteins (Basiri et al., 2014; Pratt et al., 2016; Vieillard et al., 2016; Jana et al., 2018), suggesting that this specialized region can be regarded as a typical TZ. Remarkably, the CLRs of Drosophila spermatocytes contain the same TZ module proteins reported in sensory neurons (Basiri et al., 2014; Vieillard et al., 2016; Jana et al., 2018), but serial section analysis failed to reveal the presence of the typical Y-links (Gottardo et al., 2018). Therefore, the lack in Drosophila spermatocytes of a structured TZ, which in primary cilia represents a size-dependent diffusion barrier (Takao and Verhey, 2016; Garcia-Gonzalo and Reiter, 2017; Jensen and Leroux, 2017), points to the direct recruitment from the cytoplasm of the proteins required for CLR elongation, namely the cytosolic pathway of assembly (Avidor-Reiss and Leroux, 2015). However, despite the different assembly mechanisms, the growth of both CLRs and sensory neurons relies on the proper dynamics of the axonemal microtubules.

To gain insights in the organization of the microtubule scaffold during ciliogenesis we analyze the distribution of the kinesin-like protein Klp10A in type-1 sensory neurons and during Drosophila spermatogenesis, focusing our attention on the CLRs that form in male germ cells in the absence of IFT. Drosophila Klp10A is a microtubule-depolymerizing kinesin of the Kinesin 13 family. Kinesin 13 motors differs from the other kinesins in that they do not move along microtubules, but promote tubulin dimer disassembly, playing a key role in microtubule dynamics (Walczak et al., 1996; Desai et al., 1999). It has been proposed that the movement of the kinesin-13 specific loop-2 relative to the other areas of the kinesin-13– tubulin interface determines key conformational changes leading to tubulin bending and microtubule depolymerization (Benoit et al., 2018). Previous data reported that Drosophila Klp10A is involved in microtubule dynamics throughout interphase and cell division (Rogers et al., 2004; Goshima and Vale, 2005; Mennella et al., 2005) by affecting EB1 turnover (Do et al., 2014). Mutations in Klp10A lead to overly long centrioles in germ line and somatic Drosophila cells (Delgehyr et al., 2012; Franz et al., 2013; Chen et al., 2016). Moreover, the distribution of proteins involved in centriole assembly and function, such as Drosophila pericentrinlike protein (Dplp), Spd2, Sas4, and Sak/Plk4, is affected in the absence of Klp10A (Gottardo et al., 2016). We show here that in young spermatocytes the Klp10A protein is localized to the distal ends of centrioles that dock to the plasma membrane and

is concentrated at the zone between the centriole and the CLR in mature spermatocytes. The Klp10A signal persists at the base of the ciliary cap in elongating spermatids. Moreover, this protein is enriched just above the distal centriole in sensory type-1 neurons. Our observations suggest that Klp10A could be a core component of the TZ of the ciliary structures in Drosophila.

#### MATERIALS AND METHODS

#### Fly Stocks

The Klp10A mutant line was described in Peter et al. (2002) and the stock containing the Unc–GFP transgene in Baker et al. (2004). Oregon R flies were used as controls. Flies were raised on standard Drosophila medium at 24◦C.

#### Antibodies and Reagents

We used the following antibodies: mouse anti-acetylated tubulin (1:100; Sigma-Aldrich); rabbit anti-Spd2 (1:500; Rodrigues-Martins et al., 2007); mouse anti-Sas4 (1:200; Gopalakrishnan et al., 2011); and rabbit anti-Klp10A (1:300; Laycock et al., 2006). Staining of mutant Drosophila male germ cells with the antibody against Klp10A did not reveal appreciable signal (Delgehyr et al., 2012). The secondary antibodies used (1:800) Alexa Fluor-488- and Alexa Fluor-555-conjugated anti-mouse-IgG and anti-rabbit-IgG, were obtained from Invitrogen. Hoechst 33258, Paclitaxel (Taxol, from Taxus brevifolia), Dimethyl sulfoxide (DMSO), and Shields and Sang M3 Insect Medium were purchased from Sigma. Taxol was dissolved at 1 mg/ml in DMSO and stored frozen at −20◦C.

#### Immunofluorescence Preparations

Testes were dissected in phosphate buffered saline (PBS), squashed under a small cover glass and frozen in liquid nitrogen. After removal of the coverslip the samples were fixed in methanol for 10 min at −20◦C. For antigen localization, the samples were washed 20 min in PBS and incubated for 1 h in PBS containing 0.1% bovine serum albumin (PBS-BSA, Sigma-Aldrich). The samples were then incubated overnight at 4◦C with the specific antisera in a humid chamber. After washing in PBS-BSA the samples were incubated for 1 h at room temperature with the appropriate secondary antibodies. DNA was visualized after incubation of 3 min in Hoechst 33258 (1 µg/ml, Sigma-Aldrich). After rinsing in PBS the samples were mounted in 90% glycerol in PBS. Images were taken by an Axio Imager Z1 microscope (Carl Zeiss) equipped with an AxioCam HR cooled charge-coupled camera (Carl Zeiss). Gray-scale digital images were collected separately and then pseudocolored and merged using Adobe Photoshop 5.5 software (Adobe Systems).

#### Drug Treatment

Testes were dissected in M3 medium from young pupae that contain cysts at the spermatogonial and spermatocyte stages. Testes were transferred in a 200 ml of M3 medium into a sterile 24-well plate. To assess the effect of microtubule stabilization on cilia length the dissected testes were incubated 24 h in M3 medium containing Taxol 5 mM. Controls testes were incubated in M3 medium containing DMSO alone. Specimens were fixed and stained following 24 h drug or DMSO treatments.

#### Transmission Electron Microscopy

Testes and antennae from control and Klp10A pupae were dissected in PBS, and fixed in 2.5% glutaraldehyde in PBS overnight at 4◦C. After washing in PBS, the samples were postfixed in 1% osmium tetroxide in PBS for 1–2 h at 4◦C. The material was then dehydrated through a graded series of ethanol, infiltrated with a mixture of Epon-Araldite resin and polymerized at 60◦C for 48 h. Ultrathin sections were cut with a Reichert ultramicrotome, collected with formvar-coated copper grids, and stained with uranyl acetate and lead citrate. TEM preparations were observed with a Tecnai G2 Spirit EM (FEI) equipped with a Morada CCD camera (Olympus).

## RESULTS

Staining of control testes with an antibody against Spd2, a widely conserved centriole associated protein involved in centrosome organization (Varadarajan and Rusan, 2018), shows that dotlike centrioles were typically found in stem cells (**Figure 1A**) and spermatogones (**Figure 1A'**), whereas elongating rodlike centrioles were observed during spermatocyte maturation (**Figure 1A"**). Unusually long and short centrioles were observed in germ line stem cells (**Figure 1B**), spermatogones (**Figure 1B'**) and spermatocytes (**Figure 1B"**) of Klp10A testes. Abnormal centrioles were often associated with irregular mitotic (**Figure 1B'**) or meiotic (not shown) spindles. The centrioles of control early prophase spermatocytes moved to the cell surface to organize a CLR. Then, the CLRs elongated concurrently with the centrioles and reached their full dimensions in mature primary spermatocytes (**Figure 1A"'**). The abnormal shape of the Klp10A centrioles points to defects in CLR assembly. We find, indeed, that 64.4% of the centrioles (56/87) examined in mature primary spermatocytes at the ultrastructural level were elongated with reduced CLRs (**Figure 1B"'**). These CLRs were abnormal in shape reflecting the incomplete wall of the centrioles. 35.6% of the centrioles (31/87) found in late primary Klp10A spermatocytes undocked to the plasma membrane and were unable to organize distinct CLRs (**Figure 1B""**).

Surprisingly, the Klp10A antigen was not detected on centrioles of the stem cell niche and spermatogones (**Figure 2A**) although abnormal centrioles were found at the beginning of spermatogenesis in mutant testes. Klp10A became apparent in young primary spermatocytes (**Figure 2A'**) but its distribution did not overlap the localization of the Sas4 antigen that was restricted to the basal region of the centriole. The gap between the localization of the two proteins increased in mature primary spermatocytes (**Figure 2A"**). The increased distance between the basal region of the centriole, as evidenced by the Sas4 labeling, and the distal localization of Klp10A might be correlated to the centriole elongation occurring during spermatocyte maturation. Double labeling with the centriole-associated Spd2 protein showed that the Klp10A antigen was mostly localized at the

distal ends of the centrioles, the region where the CLR organized, although a faint distribution was also seen on the whole centriole (Delgehyr et al., 2012; **Figures 2B–B"**). Therefore, to uncover an eventual relationship between the spatial expression of Klp10A and the CLR, we performed a double labeling of control testes with the anti-Klp10A antibody and an antibody against acetylated-tubulin. This antibody mainly recognizes the stable microtubules associated with the axonemal structures of the flies, allowing us to easily detect the CLRs in primary spermatocytes (**Figures 2C–C"**). Immunofluorescence analysis revealed a strong Klp10A signal coincident with the CLRs (**Figures 2D–D"**).

### Klp10A Is Enriched on the Spermatocyte CLRs

To better delineate the localization of the Klp10A signal within the ciliary structures of the primary spermatocytes, we performed immunolocalization experiments on Drosophila spermatocytes expressing the Uncoordinated-GFP (Unc-GFP) fusion protein. The recruitment of Unc-GFP represents a useful tool to follow the assembly of the ciliary projections in the Drosophila spermatocytes. According to previous studies the Unc-GFP signal first appears in young spermatocytes as distinct dots when the centrioles reach the cell surface and start to organize the CLRs (Baker et al., 2004; Riparbelli et al., 2012). In mature primary spermatocytes the Unc-GFP signal became more complex and was found in three distinct regions of the centriole/CLR complexes (**Figure 3A**): one proximal region corresponding to the middle and the distal end of the centriole, one region overlapping the CLR, and a middle region between the centriole and the ciliary axoneme. Double labeling with the anti-Spd2 antibody shows that the Unc-GFP signal was not detected in the basal region of the centriole (**Figure 3B**), whereas it was very strong in the middle of the centriole/CLR complex (**Figure 3A**) a region that likely corresponds to the transition region between the centriole and the CLR (Vieillard et al., 2016). The distal Unc-GFP signal overlapped the ciliary axoneme recognized by the anti-acetylatedtubulin antibody (**Figure 3C**). A distinct Klp10A localization was first observed in young primary spermatocytes when the Unc-GFP signal appeared at the time of centriole-to-basal body conversion (**Figure 3D**). The Klp10A antibody also recognized filamentous structures that were closely associated with one of the sister centriole pairs (**Figure 3D**). It has been shown that these structures correspond to distinct microtubule bundles that extend into the peripheral cytoplasm of the polar spermatocytes from one of the mother centrioles of each pairs (Riparbelli et al., 2018). The spatial localization of Klp10A became distinct during prophase progression (**Figures 3E,F**) and was obvious in mature primary spermatocytes when the centriole/CLR complexes reached their full dimensions (**Figure 3G**). A feeble Klp10A signal was detected along the whole centriole, whereas a stronger labeling was observed just above the intermediate Unc-GFP dot (**Figure 3G**). This staining persisted during the further meiotic divisions (**Figures 3H,I**). A small cluster of Klp10A was observed between the centriole pairs during prophase progression (**Figures 3D–G**). This cluster increased in size at the poles of the first metaphase spindles (**Figure 3H**), but strikingly reduced during the second meiosis when the parent centrioles moved away (**Figure 3I**).

It was previously showed that treatment of young primary spermatocytes with taxol leads to the dramatic elongation of the ciliary axoneme with the ensuing extension of the distal Unc-GFP domain (Riparbelli et al., 2013). However, the Unc-GFP localization at the centriole and at the TZ remains unmodified. The Klp10A distribution was also unchanged in taxol treated primary spermatocytes and the stronger signal was still localized close to the intermediate Unc-GFP dot, despite the CLR was unusually elongated (**Figures 4A–C**).

Each young spermatid inherits at the end of meiosis one centriole/CLR complex that represents the basis for the sperm axoneme formation. This structure look like the centriole/CLR complexes found during meiotic progression and also display a short axoneme (**Figure 5A**) and an intermediate Unc-GFP dot (**Figure 5A**). Klp10A was still observed next the Unc dot (**Figure 5A'**). At the onset of spermatid elongation, some Unc-GFP labeling was still associated with the centriole, whereas the Unc-GFP dot moved away from the distal end of the centriole concurrently with the elongation of the axoneme (**Figure 5B**; Gottardo et al., 2013). The Klp10A staining persisted close to the Unc-GFP dot (**Figure 5B'**).

EM analysis confirmed that the CLRs, or ciliary caps, inherited by the young control spermatids (**Figure 5C**) persisted at the apical end of the elongating sperm axoneme increasing two–three times their length (**Figure 5D**). Ciliary caps were also found in young (**Figure 5E**) and elongating (**Figure 5F**) mutant spermatids but their length was very reduced compared to controls. By contrast, the centriole was more prominent (**Figure 5E**). The basis of the invaginating cell membrane that surrounded the ciliary cap was associated with a small ring of dense material, the

FIGURE 3 | Klp10A signal is stronger in the transition zone of the CLRs. (A) Unc-GFP recognizes three distinct regions on the centriole/CLR of mature primary spermatocytes: a proximal region (double arrowhead), one intermediate region (arrow) and a distal region (arrowhead). Counterstain with the anti-Spd2 (B) and anti-acetylated tubulin (C) antibodies shows that Unc-GFP is localized along the centriole and the CLR, but neither of these antibodies recognize the Unc-GFP intermediate dot. (D) The Klp10A signal is first detected at the tip of the parent centrioles in young primary spermatocytes when a Klp10A filamentous structure is also observed at the base of one of the sister centriole pairs. The localization of Klp10A becomes restricted just above the intermediate Unc-GFP dot as the first prophase progresses (E,F) and is very distinct in mature primary spermatocytes (G); insets in G are Unc-GFP and Klp10A separate channels showing a remnant Klp10A staining along the centriole. A cytoplasmic cluster of Klp10A is present between the parent centrioles during prophase (G) and increases in size during metaphase of the first meiotic division (H); this cluster is very reduced during prophase of the second meiosis (I). Bar: 2 mm.

so-called "ring centriole" (Phillips, 1970) that in control spermatids was orthogonal to the axoneme (**Figures 5C,D**), whereas it was obliquely oriented in Klp10A mutant spermatids (**Figures 5E,F**).

### Reduced Klp10A Expression Results in Strong Structural Defects of Sensory Type I Neurons

For a better understanding of the Klp10A role in centriole/axoneme assembly, we examined the ultrastructure of the Johnston's organ, a large array of sensory type I neurons in the antennae. Each unit, or scolopidium, of the Johnston's organ usually displays two ciliary processes that assemble throughout a compartmentalized process of ciliogenesis (Avidor-Reiss and Leroux, 2015) and house at their base two short linearly arranged centrioles, one distal to the other. Unlike, primary spermatocytes, in which both the parent centrioles assemble a ciliary projection, only the distal centriole of the scolopidium templates the sensory ciliary axoneme (**Figure 6A**). The proximal centriole is shorter and enclosed within the rootlets emerging from the base of the distal centriole (**Figure 6A**). The distal centriole lacks a cartwheel and consists of nine doublet microtubules (**Figure 6A'**) that extend in a distinct TZ (**Figure 6A"**) and then in the ciliary axoneme (**Figure 6A"'**). Double labeling of transgenic flies expressing Unc-GFP, a protein specifically associated with the tips of the sensory dendrites where the distal centrioles are found (Baker et al., 2004; Enjolras et al., 2012), revealed that Klp10A was localized just above the Unc-GFP dot (**Figure 6A**, inset).

Seventy two percent (31; n = 43) of the ciliary structures of the mutant Johnston's organs examined in longitudinal section were short with disorganized axonemal microtubules (**Figure 6B**). The rootlets that emerged from the base of the distal centriole were often disorganized and the proximal centriole was not positioned properly at the base of the cilia in several neurons (**Figure 6B**). Thus, proximal and distal centrioles lost their coaxial orientation. Distinct microtubules emerged from the proximal centriole and extended toward the basal cell cytoplasm to form elongated cylindrical structures often frayed at their extremities (**Figures 6C,D**). 66.6% of the distal centrioles examined in cross section (16; n = 24) look like control centrioles with nine microtubule doublets (**Figure 6C'**) whereas the remnant 34.4% (8; n = 24) had structural defects. However, 86% of the putative TZ examined (37; n = 43) showed discontinuities in the electron dense material associated with the axonemal microtubules (**Figure 6C"**). These discontinuities could also be observed in longitudinal sections (**Figure 6C**). Doublets of the ciliary axonemes were missing or misplaced (**Figure 6C"'**). Cross sections through the extension of the proximal centrioles often showed a disorganized wall in which we find naked doublets and doublets immersed in an electron dense material (**Figure 6D'**) reminiscent of the material observed through the TZ at the tip of the distal centrioles. Short lateral projections emerged from the A-tubules (**Figure 6D'**). Such lateral projections were also observed within the TZ of controls (**Figure 6A"**) and mutant (**Figure 6C"**) axonemes. Sections through the frayed extremities of the proximal centriole extensions revealed disorganized doublets with opened B-tubules (**Figure 6D"**).

centrioles (c), as observed in control testes (C). Bar: 2 mm.

#### DISCUSSION

The growth of the primary cilia in vertebrate cells is mediated by an IFT mechanism that requires specific carriers to move axonemal and ciliary membrane components from the ciliary base to its tip and back (Ishikawa and Marshall, 2017). Essential for the proper execution of this process is a specific region, the TZ that is restricted at the distal end of the basal body (Gonçalves and Pelletier, 2017).

Drosophila spermatocytes are characterized by surface protrusions, the CLRs, that look like primary cilia, but grow by IFT independent mechanisms (Han et al., 2003; Sarpal et al., 2003) and are not reabsorbed during cell division (Tates, 1971; Riparbelli et al., 2012). In contrast to basal bodies of conventional primary cilia that do not change length during ciliogenesis, the short centrioles of the young Drosophila spermatocytes elongate concurrently with the extension of the CLR axoneme (Fritz-Niggli and Suda, 1972; Gottardo et al., 2013). Because the centrioles of the Drosophila spermatocytes are continuous with the ciliary axoneme, they should not elongate. This aspect raises intriguing questions concerning the mechanisms of centriole elongation and the functional roles of the CLR.

The finding of conserved TZ proteins within the CLRs, namely Cep290, Chibby and components of the MKS complex (Enjolras et al., 2012: Basiri et al., 2014; Pratt et al., 2016; Vieillard et al., 2016) had led to the hypothesis that the whole CLR could be a modified TZ. However, TZ proteins are classically restricted to the basal region of the ciliary axoneme where Y-shaped links connect the axonemal doublets to the ciliary membrane. In contrast, Drosophila CLRs lack Y-links, consistent with the

absence from Drosophila of the NPHP module proteins (Barker et al., 2014; Basiri et al., 2014) which are required to assemble the Y-links in primary cilia (Williams et al., 2011). We find Y-links like structures at the base of the sensory cilia confirming recent observations (Vieillard et al., 2016; Jana et al., 2018) that such links can also form in the absence of NPHP module proteins. However, typical Y-links usually arose from the space between the A and the B tubules (Szymanska and Johnson, 2012), whereas we find that in sensory Drosophila neurons the Y-links extend from the anterior margin of the A-tubule, raising questions on the actual similarity between these structures.

Unlike vertebrate cells in which the MKS module is crucial for cilia assembly and maintenance, sensory cilia lacking MKS proteins exhibit only subtle defects in the adult Drosophila flies (Pratt et al., 2016; Vieillard et al., 2016). Since the TZ of vertebrate primary cilia represents a selective gate that limits the transit of molecules within the ciliary compartment (Garcia-Gonzalo and Reiter, 2017; Jensen and Leroux, 2017), the

emerging from the A-tubule; arrows, Y-links. Bars: 0.4 mm (A–D); 1 mm (inset A); 50 nm (A'–A"'C'–C"'D',D").

absence of structured Y-links within the Drosophila CLRs points to a simple diffusion process of cytoplasmic proteins into the ciliary compartment, the so-called cytosolic pathway of assembly (Avidor-Reiss and Leroux, 2015). However, it is unclear how Drosophila TZ proteins are involved in modulating the axoneme assembly. The reduced expression of Cep290 led to abnormal and incomplete ciliary structures suggesting that this protein could be involved in the organization of the ciliary cap by enhancing the formation of a compartmentalized domain in which the axoneme tip would be organized (Basiri et al., 2014). Both the growth of the CLRs during spermatocyte maturation and the elongation of the sperm axoneme relies on microtubule assembly dynamics within the ciliary cap. It has been shown that mutations in the microtubule-depolymerizing Kinesin-13 Klp10A, lead to centriole and CLR defects during Drosophila male gametogenesis (Delgehyr et al., 2012; Gottardo et al., 2013). We show here that Klp10A is mainly localized to the basal region of the CLRs in the Drosophila spermatocytes just above the intermediate Unc-GFP dot. The distribution of Klp10A during male meiosis is strikingly similar to that of the TZ proteins

Chibby and Cep290 (Enjolras et al., 2012; Basiri et al., 2014). During spermiogenesis Klp10A moves away from the apical tip of the centriole acquiring a localization like that of Cep290 and Chibby at the base of the ciliary cap. As in Cep290 mutants (Basiri et al., 2014), the centrioles of Klp10A primary spermatocytes are very elongated whereas the CLRs are reduced in length. These findings suggest that Klp10A may be a component of the TZ that ensures the proper organization of the axonemal doublets during CLR assembly. In the absence of Klp10A the behavior of the C-tubule is altered and the ciliary axoneme does not assemble properly, but very elongated centrioles are found (Gottardo et al., 2013). This suggest a balance between centriole and ciliary axoneme elongation, likely mediated by the direct action of Klp10A in controlling microtubule dynamics. The localization of Klp10a does not change with the extension of the ciliary axoneme after taxol treatment, suggesting that this protein performed its function in a stable and restricted defined region at the ciliary base. Klp59D, another kinesin-13 family member, is enriched within the whole CLR and along the ciliary cap of the elongating spermatids (Vieillard et al., 2016). However, unlike the short CLRs observed in Klp10A mutants, Klp59D spermatocytes display very elongated CLRs, suggesting that these kinesins perform opposite roles during the assembly of the spermatocyte TZ.

The early Klp10A signal has been found in young primary spermatocytes when the centrioles dock to the cell membrane and start to assemble the ciliary axoneme, although defects in centriole structure have been observed in GSCs and spermatogones. Chibby and Cep290 signals were also found at the time of centriole docking to the cell surface in primary spermatocytes (Enjolras et al., 2012: Basiri et al., 2014), but the structure of the centrioles during early stages of spermatogenesis was not investigated in these mutants. The discrepancy between the spatiotemporal appearance of the defects in centriole structure and the early detection of the Klp10A signal is still unclear. A cytoplasmic localization of Klp10A to the centrosomes has been observed in male GSCs (Chen et al., 2016), but a specific Klp10A signal on these centrioles has not been reported.

In control Johnston's organs, two centrioles are typically aligned above the end of each dendrite. The distal centriole templates the ciliary axoneme that displays an elongated TZ. The proximal centriole is shorter and unable to nucleate a ciliary axoneme. Klp10A is localized at the distal ends of the sensory dendrites just above the Unc-GFP signal, a region that corresponds to the TZ (Enjolras et al., 2012). It has been shown that the TZ in auditory neurons consists of two regions and that the Unc-GFP signal is restricted to the proximal part of the TZ (Jana et al., 2018). Thus, our observations suggest that Klp10A could be specifically associated with the distal region of the TZ. Remarkably, the loss of Klp10A function leads to the extension of both the centrioles in opposite directions. This suggests that the coaxial centrioles in the Johnston's organ could be disposed base to base. Alternatively, the unusual orientation of the centrioles could be due to the disruption of the rootletin cage thus enhancing the elongation of the proximal centriole toward the basal cytoplasm of the cell. It has been shown that reduced Rootletin function in chordotonal Johnston's organs impairs ciliary rootlet assembly and the proximal centriole is displaced or missing, suggesting that rootlets may ensure the proper positioning of the proximal centriole within the base of the sensory cilium (Chen et al., 2015; Styczynska-Soczka and Jarman, 2015). Remarkably, the extensions of both the proximal and the distal centrioles display doublets surrounded by electron dense material as found in control TZ. We also observed short lateral projections associated with the A-tubules. Such projections are found together with the Y-links in the TZ of control Johnston's organs. Therefore, the distal regions of the elongated centrioles in Klp10A Johnston's organs may be equivalent to abnormally shaped TZs, even if they lack distinct Y-links. Our findings also suggest that the proximal centriole has the potential to assemble a TZ-like structure, a special skill until now ascribed to the distal centriole alone, unless the proximal centriole was converted in a distal one by the loss of centrobin (Gottardo et al., 2015). The observation that the proximal centriole starts ciliogenesis in Klp10A depleted Johnston's organs is consistent with a critical role of this protein in controlling both proximal and distal centriole elongation. However, the proximal centrioles elongate to form TZ-like regions, but never ciliary structures are properly assembled. The phenotype we observed in mutant Klp10A Johnston's organs is strikingly different from what has been found in mutant of other TZ proteins, such as Unc, Chibby and Cep290, in which both distal and proximal centrioles do not elongate (Baker et al., 2004; Enjolras et al., 2012; Basiri et al., 2014; Vieillard et al., 2016). Remarkably, all these TZ proteins are restricted or enriched to the proximal part of the TZ (Jana et al., 2018), whereas Klp10A is apparently enriched to the distal part alone.

We suggest that Klp10A can be regarded as a core component of the TZ in Drosophila, involved in the assembly and maintenance of the ciliary axoneme in male germ cells and chordotonal organs irrespective of the compartmentalized or cytosolic mechanisms of ciliogenesis.

#### DATA AVAILABILITY

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

#### AUTHOR CONTRIBUTIONS

VP and MR performed all the experiments. GC and MR designed the experiments and wrote the manuscript.

#### ACKNOWLEDGMENTS

We would like to thank H. Rangone and M. Kernan for providing the Klp10A and the Unc–GFP stocks. We also thank to J. Gopalakrishnan, M. Bettencourt-Dias, and A. Rodrigues-Martins, for generously providing the antibodies used in this study.

#### REFERENCES

fcell-07-00173 August 21, 2019 Time: 14:32 # 11


and compartmentalization. EMBO Rep. 13, 608–618. doi: 10.1038/embor. 2012.73


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a past co-authorship with one of the authors GC.

Copyright © 2019 Persico, Callaini and Riparbelli. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Mechanism of Acrosome Biogenesis in Mammals

#### Muhammad Babar Khawar 1,2, Hui Gao<sup>1</sup> and Wei Li 1,2 \*

*<sup>1</sup> State Key Laboratory of Stem Cell and Reproductive Biology, Institute of Zoology, Chinese Academy of Sciences, Beijing, China, <sup>2</sup> University of Chinese Academy of Sciences, Beijing, China*

During sexual reproduction, two haploid gametes fuse to form the zygote, and the acrosome is essential to this fusion process (fertilization) in animals. The acrosome is a special kind of organelle with a cap-like structure that covers the anterior portion of the head of the spermatozoon. The acrosome is derived from the Golgi apparatus and contains digestive enzymes. With the progress of our understanding of acrosome biogenesis, a number of models have been proposed to address the origin of the acrosome. The acrosome has been regarded as a lysosome-related organelle, and it has been proposed to have originated from the lysosome or the autolysosome. Our review will provide a brief historical overview and highlight recent findings on acrosome biogenesis in mammals.

Keywords: acrosome biogenesis, autolysosome, lysosomes, globozoospermia, spermiogenesis

#### Edited by:

*Tomer Avidor-Reiss, University of Toledo, United States*

#### Reviewed by:

*Rajprasad Loganathan, Johns Hopkins University, United States Giuliano Callaini, University of Siena, Italy*

> \*Correspondence: *Wei Li leways@ioz.ac.cn*

#### Specialty section:

*This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology*

Received: *07 May 2019* Accepted: *29 August 2019* Published: *18 September 2019*

#### Citation:

*Khawar MB, Gao H and Li W (2019) Mechanism of Acrosome Biogenesis in Mammals. Front. Cell Dev. Biol. 7:195. doi: 10.3389/fcell.2019.00195*

## INTRODUCTION

Sexual reproduction requires the fusion of two gametes in a critical multistep process termed fertilization. To ensure the success of fertilization, each step needs to proceed in a very precise manner. One of the key steps that ensures successful fertilization is acrosome reaction (AR). Sperm-egg fusion is a carbohydrate-dependent event that takes place via interaction between several glycan-binding molecules (receptors) present on the sperm plasma membrane with their corresponding glycans (ligands) localized to the zona pellucida (oocyte) (Yanagimachi, 1994; Tulsiani et al., 1997; Shur, 1998; Töpfer-Petersen, 1999; Wassarman, 1999). This irreversible interaction of the gametes leads to a calcium-mediated signal transduction cascade of events that results in the release of the acrosomal contents via exocytosis, which is termed AR. Several hydrolytic and proteolytic acrosomal enzymes are released in order to facilitate sperm fusion with the oocyte (Ikawa et al., 2010). Any structural or functional acrosomal abnormality could impair sperm fusion, and ultimately result in infertility. Moreover, studies have shown that intra-cytoplasmic insemination with sperm containing acrosomal abnormalities did not lead to successful fertilization, even in the absence of fertilization barriers, because the oocyte was unable to be efficiently activated (Nasr-Esfahani et al., 2008, 2010a,b). Only intracytoplasmic sperm injection (ICSI) followed by assisted oocyte activation with calcium ionophore was found to achieve high live birth-rates (Shang et al., 2019). Thus, the acrosome is indispensable for fertilization.

Acrosome biogenesis in mammals is accompanied with spermatid differentiation during spermiogenesis, which is characterized by the transformation of a spermatid into a spermatozoon. The process continues throughout the reproductive lifespan of the male. Although the acrosome morphology varies from species to species, two basic parts make up the acrosome in all mammals: (1) a large anterior part that varies in shape (paddle, hook, and spatula-like) and in size (Bedford, 2014) and (2) an equatorial segment (ES), which is the smaller and thinner part of the acrosome found in the middle of the sperm head. The acrosomal contents are enclosed in a single membrane that is generally divided into an outer acrosomal membrane (OAM) and an inner acrosomal membrane (IAM). The OAM lies immediately beneath the plasma membrane of the spermatid and both of these membranes fuse at the time of AR (Yanagimachi, 2011). The IAM lies above the nuclear envelope as a cap and does not fuse during AR. The luminal contents are heterogeneous and are usually categorized as soluble and particulate material. The soluble material is comprised of hydrolytic enzymes that take part in AR and help to disperse the oocyte coverings. The particulate material is the acrosomal matrix that facilitates the sperm-oocyte interaction during fertilization by providing a stable protein scaffold (Buffone et al., 2004; Suryavathi et al., 2015).

Researchers have characterized the process of spermiogenesis in mice, and they have identified 16 steps, with acrosome biogenesis being a key event. Acrosome biogenesis is classically divided into four major phases: Golgi (1–3 steps), cap (4– 7 steps), acrosome (elongation) (8–12 steps), and maturation (final) phases (13–16 steps) (**Figure 1**). This four-phase division of acrosome biogenesis was proposed more than half-a-century ago, based mainly on light microscopic analysis of testicular sections stained with periodic acid–Schiff (PAS) (Clermont and Leblond, 1955).

The 1st phase is called the Golgi phase because the Golgi apparatus is an essential organelle that supports early spermiogenesis (Leblond and Clermont, 1952; Hess, 1990; Russell et al., 1993). During the 1st phase, the Golgi apparatus is very active in producing several glycoproteins, and the trans-Golgi network gives rise to several small proacrosomal vesicles that are required for the formation of a mature acrosome. These proacrosomal granules fuse to form a large solitary acrosomal granule near the concave region of the nuclear surface. The central part of the acrosomal granule is bound to the nuclear envelope while the peripheral part is associated with the perinuclear theca (**BOX 1**; **Figure 2**). In the 2nd (Cap) phase, the acrosomal granule becomes enlarged with glycoproteinrich contents. Moreover, it begins to flatten upon touching the nuclear envelope, and spreads over the nucleus to form a cap. At the same time, the Golgi complex moves to the nascent neck region located in the distal end. The acrosomal granule gradually covers 1/3 of the nuclear surface and spreads, transforming into a very thin layer (**Figure 1**). Near the distal end of the developing acrosome, the "acroplaxome" can be found, which is an important structure that supports spermiogenesis. It consists of a marginal ring, which consists of an acrosomal plate that is made up of keratin and F-actin (**BOX 1**; **Figure 2**). At the time of elongation of the spermatid head, the marginal ring is associated with the growing edge of the acrosome and the nuclear surface (nuclear plate). Thus, acroplaxome not only binds the acrosome with the nucleus but also ensures the developing acrosomal cap remains anchored to the nuclear envelope undergoing elongation (Kierszenbaum et al., 2003). In the 3rd (acrosomal) phase, the acrosomal system begins to migrate over the ventral surface of the elongating spermatid nucleus and this migration ends in step 14 spermatid (**BOX 1**; **Figure 1**; Hess and de Franca, 2008). At this stage, the acrosome undergoes condensation and attaches itself to the IAM, while chromatin also undergoes intense condensation. Several cytoskeletal proteins including calmodulin (Camatini et al., 1992), actin (Talbot and Kleve, 1978), and α-spectrin-like antigens (Virtanen et al., 1984), play very important roles in the acrosomal organization. The elongating spermatids show initiation of manchette microtubule formation near the nuclear ring region (perinuclear ring), thinning of the cytoplasm and a gradual orientation of the acrosome toward the overlying plasma membrane (**BOX 1**; **Figure 2**). The 4th (maturation) phase involves a few changes in nuclear morphology and acrosomal migration. Condensation of the nucleus continues and acrosomal granule spreads over the entire acrosomal membrane and the acrosome differentiates into anterior and posterior regions. The anterior portion becomes the acrosome apex, while the rest of the acrosome covers nearly all the nuclear surface, except the part attached to the sperm tail (Russell et al., 1993). Moreover, excess cytoplasm, cytoplasmic components (lipids) and multiple unwanted organelles including mitochondria, vesicles, and ribosomes are disposed of in the form of cytoplasmic droplets prior to spermiation (**Figure 1**; Hess et al., 1993; Russell et al., 1993; De Franca et al., 1995). The resulting residual bodies released from spermatids are taken up and digested by Sertoli cells. Although, the morphogenetic process of the acrosome is well studied, the precise molecular mechanism of sperm acrosome biogenesis is not yet fully understood. Here, we describe the most recent progress that has been made in understanding the molecular mechanism underlying acrosome biogenesis.

#### THE MOLECULAR MECHANISM UNDERLYING ACROSOME BIOGENESIS

Several ER and Golgi-associated proteins actively participate in acrosome biogenesis. The endoplasmic reticulum (ER) is the main site of protein synthesis and folding (Vitale et al., 1993), while the Golgi apparatus directs glycosylation, processing, and sorting of newly synthesized proteins by the ER. The biosynthesis of some acrosome-specific proteins, such as acrosin, starts during the meiotic pachytene stage and continues as round spermatids enter the elongation stage (Anakwe and Gerton, 1990; Kashiwabara et al., 1990; Escalier et al., 1991). These proteins then enter the exocytic route and are transported to their respective targeted area in the form of proacrosomal granules that originate from the Golgi apparatus. The presence of these proacrosomal granules in the pachytene stage has been confirmed by numerous studies (Nicander and Ploen, 1969; Fawcett, 1975; Anakwe and Gerton, 1990; Suarezquian et al., 1991; Ramalho-Santos et al., 2002). Moreover, the presence of several ERassociated proteins such as protein O-mannosyltransferase 1 (POMT1) (Prados et al., 2007), POMT2 (Willer et al., 2002) and Calreticulin (Nakamura et al., 1993), have also been detected in the acrosome. Another ER-associated protein that has been detected in mouse testes is HSP90B1 (gp96/Grp94; glucoserelated protein 94) (Asquith et al., 2005; Yang and Li, 2005). Germ cell-specific Hsp90b1 knockout resulted in spermatozoa characterized by globular/abnormal heads, similar to those in

globozoospermia syndrome (Audouard and Christians, 2011). Therefore, HSP90B1 has been suggested to be a testis-specific chaperone and to be required for the proper folding of acrosomal proteins. The impairment of protein folding during HSP90B1 deficiency suggests an important role of ER protein folding in acrosome biogenesis. β-Glucosidase 2 (GBA2) is another ER-associated protein, and is involved in the metabolism of bile acid–glucose conjugates (Matern et al., 1997). GBA2 disruption results in the formation of abnormal spermatozoa characterized by enlarged heads and the absence of acrosome (Yildiz et al., 2006). In short, these proteins are indispensable for acrosome biogenesis.

In addition, the highly dynamic trafficking of the Golgiderived vesicles is also involved in acrosome biogenesis. For instance, several Golgi proteins, including Golgin-95/GM130, Golgin-97, Giantin, and β-COP have been found in acrosomal associated membranes (Moreno and Schatten, 2000; Moreno et al., 2000; Ramalho-Santos et al., 2001; Hermo et al., 2010). Among these proteins, β-COP and Clathrin have been described to participate in anterograde and retrograde transport of vesicles during acrosome formation (Martínez-Menárguez et al., 1996b; Moreno et al., 2000; Ramalho-Santos et al., 2001). Golgi-derived vesicle trafficking during acrosome biogenesis can generally be divided into three steps: vesicle formation, trafficking, and fusion. Some of the important proteins that contribute to these three events are described below.

Vesicle formation and trafficking are key events of acrosome biogenesis. Stromal membrane-associated protein 2 (SMAP2) regulates the production of clathrin-coated vesicles from the trans-Golgi network (TGN) by interacting with CALM (Clathrin and the Clathrin assembly protein) and Syntaxin 2 (a component of SNARE complex that helps in membrane fusion), contributing to acrosome biogenesis (Funaki et al., 2013). GOPC (Golgiassociated PDZ- and coiled-coil motif-containing protein) was identified as a frizzled-interacting protein that is involved in vesicular trafficking from the Golgi apparatus (Yao et al., 2001). GOPC is predominantly localized in the trans-Golgi region in round spermatids and plays an important role in the transport and fusion of the proacrosomal vesicles with the growing acrosome (Yao et al., 2002). Golgin subfamily A member 3 (GOLGA3) is another Golgi-associated protein that is highly expressed during the round spermatid stage and is believed to contribute to acrosome biogenesis by interacting with GOPC (Banu et al., 2002; Hicks and Machamer, 2005; Bentson et al., 2013). Another important protein that is involved in protein transport is protein interacting with C kinase 1 (PICK1), which is primarily localized around the Golgi apparatus in spermatids (Arvan and Castle, 2013). PICK1 regulates vesicle trafficking from the Golgi apparatus to the developing acrosome by interacting with GOPC (Xiao et al., 2009). Hence, Golgi-associated proteins are of great significance in vesicle formation and trafficking during acrosome biogenesis. Autophagic machinery also participates in acrosome biogenesis by regulating vesicular trafficking. Autophagy refers to the intracellular catabolic pathway which is responsible for the degradation and recycling of organelles and cytosolic proteins via autophagosomes (double-membrane vesicle) (Yang and Klionsky, 2009; Mizushima and Levine, 2010). Microtubuleassociated protein 1A/1B-light chain 3 (LC3) activated by autophagy related protein 7 (ATG7) is delivered to Golgi-derived vacuoles either directly or indirectly (via phagophores) where the activated protein either facilitates the fusion of vesicles or guides them toward the nucleus (Wang et al., 2014). These investigations suggest that the autophagic pathway might be involved in acrosome biogenesis.

Proacrosomal granules undergo fusion with each other to form a single large acrosomal granule at the nuclear surface. Vesicle fusion requires SNARE complexes that help in the fusion of opposing membranes (Rizo and Südhof, 2002; Ungermann and Langosch, 2005; Zhao and Brunger, 2016). Moreover, the disruption of fatty acid desaturase 2 (FADS2) results in Syntaxin 2 scattering, which eventually impairs acrosome formation (Roqueta-Rivera et al., 2011). TATA element Modulatory Factor (TMF/ARA160) is another Golgiassociated protein required for the fusion of vesicles to the targeted membrane (Bel et al., 2012; Miller et al., 2013), and interruption of its expression leads to a complete absence of the acrosomes, suggesting that TMF/ARA160 probably supports the transport and docking of proacrosomal vesicles to the nucleus (Lerer-Goldshtein et al., 2010). Human Rev-binding (HRB) is another critical protein required for the docking of Golgiderived proacrosomal vesicles; it binds to the cytosolic side of proacrosomal vesicles and links the Golgi apparatus and the nuclear surface (Kang-Decker et al., 2001). Polypeptide Nacetylgalactosaminyltransferase 3 (GALNT3) is located in the cis-medial region of the Golgi, and its disruption leads to failure of proacrosomal vesicles fusion and transport to nuclear surface suggesting the significance of protein O-glycosylation in acrosome biogenesis (Miyazaki et al., 2013). Besides, many other proteins important for vesicle formation or trafficking such as SMAP2 and GOPC also contribute to the fusion of proacrosomal vesicles (Yao et al., 2002; Funaki et al., 2013).

In addition to the transportation to the concave region of the nuclear surface and fusion to form a single large acrosomal granule, the attachment and spreading of the acrosome over the nucleus is also very important to its function. Sperm acrosome-associated 1 (SPACA1), an acrosomal membrane protein, participates in the process of acrosome attachment to the nucleus and the disruption of SPACA1 leads to the detachment of the acrosome from the nucleus (Fujihara et al., 2012). Zona pellucida-binding protein 1 (ZPBP1) is another important protein localized on the periphery of the acrosomal membrane (Lin et al., 2007; Yu et al., 2009), its absence results in the compaction failure of the acrosome and subsequently leads to acrosome fragmentation (Lin et al., 2007). Fer Testis (FerT) is a member of Fes/Fps (nonreceptor tyrosine kinases family), and it regulates cytoskeletal reorganization, cell adhesion, and vesicular transport (Greer, 2002) by attaching itself to the cytosolic surface of OAM and coexists with phosphorylated cortactin (an F-actin regulator protein) in the acroplaxome (Kierszenbaum, 2006; Kierszenbaum et al., 2008). Developmental pluripotency-associated 19-like 2 (DPY19L2) acts as a bridge between the nucleus and the acroplaxome, and its deficiency leads to a loss of the acrosome due to disruption of the nuclear/acroplaxome junction (Pierre et al., 2012). Similarly, disruption of Lamin A/C (a structural component of the nuclear envelope) (Dittmer and Misteli, 2011) leads to acrosome fragmentation and malformed acrosome (Shen et al., 2014). A vast array of specific proteins is involved in acrosome biogenesis and any defect may result in malformation of the acrosome and eventually lead to infertility (**Table 1**; **Figure 3**).

Although most acrosomal substances are transported to the developing acrosome via the ER-Golgi route (Fawcett and Hollenberg, 1963; Clermont and Tang, 1985), numerous other routes are also suspected to exist for the transfer of acrosomal components to the developing acrosome. Toshimori (1998) has reported the existence of an extra-Golgi tract and a Golgi tract, which includes a Golgi-acrosomal granule tract and a Golgi-head cap tract (Toshimori, 1998).

TABLE 1 | Mouse models related to acrosome biogenesis.


#### POTENTIAL WAYS FOR THE ORIGIN OF ACROSOME

Previously, immunocytochemical investigations of glycoprotein synthesis in the Golgi established the acrosome as a direct Golgi derivate (Friend and Fawcett, 1974; Tang et al., 1982; Aguas and da Silva, 1985; Anakwe and Gerton, 1989; Moreno et al., 2000). Later, the acrosome was proposed to be a specialized lysosome based on an acidic pH, protease activities and the presence of hyaluronidase (Hartree and Srivastava, 1965; Allison and Hartree, 1970). A non-lysosomal origin of the acrosome has also been proposed (Martínez-Menárguez et al., 1996a). Martínez-Menárguez et al. (1996a) reported the absence of two well known lysosomal markers; lysosomal membrane glycoprotein (Igp) 120 and mouse Lamp-1 in acrosomal membranes. Moreover, acrosomal and proacrosomal vesicles both lacked two important endosomal markers, cation-dependent and independent mannose 6-phosphate receptors suggesting some lysosomal features are absent in the acrosomes (Martínez-Menárguez et al., 1996a). In addition, small GTPases were found to be associated with acrosome development (Ramalho-Santos et al., 2001), leading researchers to postulate that the acrosome is a unique cellular organelle, and could be considered a secretory granule (Moreno, 2003). More recently, the acrosome has been suggested as a novel lysosome-related organelle (LRO) (Berruti et al., 2010; Berruti and Paiardi, 2011, 2015). The details of our understanding about acrosome biogenesis are given below:

proteins mentioned in the figure.

## Is the Acrosome a Direct Golgi Derivative?

The release of acrosomal contents, during AR, led to concerns and questions about how acrosomal substances are synthesized and stored. To answer these questions, researchers turned their center of investigation toward biosynthetic pathways, especially those of the trans-Golgi network. In line with this, the synthesis, target, and fate of a number of acrosomal proteins in conjunction with lysosomal (Lamp-1, cathepsin D) and Golgi markers (giantin, β-COP, golgin 97) have been precisely examined using immunocytochemical techniques (Aguas and da Silva, 1985; Anakwe and Gerton, 1989; Martínez-Menárguez et al., 1996b; Moreno et al., 2000; Ramalho-Santos et al., 2001). Subsequently, some of these acrosomal-associated proteins were found to be produced in the Golgi complex of spermatocytes (cells that lack an acrosome) and later transported to the acrosome (Anakwe and Gerton, 1989; Escalier et al., 1991). These results were in complete accordance with the previous findings (Fawcett and Bloom, 1986). Therefore, the acrosome was interpreted to be directly derived from the Golgi complex, which acts as a source for membrane and protein contents. In other words, the TGN could be considered the main player during acrosome development, and no lysosome-related characteristics needed to be attributed to the acrosomal vacuole. Meanwhile, more investigations were carried out to identify the origin of acrosomal proteins, their trafficking, and sorting. Consequently, some novel roles of acrosomal proteins were revealed. For instance, acrosin was found to be stored in an inactive state as proacrosin and only activated by the protease acrolysin (McRorie et al., 1976) at the time of AR. Activated acrosin accelerates the release of acrosomal contents by dispersing the acrosomal matrix (Mao and Yang, 2013). Therefore, acrosin is now believed to help in not only the cleavage and subsequent activation of acrosomespecific proteases but also their release via exocytosis (Mao and Yang, 2013). These results were consistent with previous findings (Baba et al., 1994) that showed disruption of acrosin did not affect fecundity but led to a delayed fertilization rate in mice. However, some data exist that are not concordant with the idea that the acrosome is directly derived from Golgi. For instance, the Golgi apparatus detaches from the acrosomal space and moves in the opposite direction during the late capping phase. Furthermore, differentiating spermatids are characterized by the presence of an atypical micro-tubular organization. For example, a cortical microtubule array, despite the absence of centrosome, has been observed in the Golgi phase and tends to disappear during the formation of the manchette, which is a transient structure over the nuclear envelope at the late capping phase (Cherry and Hsu, 1984; Moreno and Schatten, 2000). In fact, this idea is so deeprooted in the field that a variety of proteins in somatic cells that are involved in endocytosis have also been investigated, and were found to play a role in the biosynthetic/anterograde pathway, supporting acrosome biogenesis in spermatids (Berruti and Paiardi, 2011, 2015). Hence, in spite of contrary reports, a large majority of the reproductive-research community still supports the idea that the acrosome is a direct Golgi derivative.

#### Is the Acrosome a Secretory Granule?

The notion that the acrosome is a direct Golgi derivative prevailed until the emergence of reports strengthened earlier evidence for the presence of a range of hydrolytic enzymes and a low pH maintained by the activity of V-ATPase (Sun-Wada et al., 2002). These characteristics were found to be common between lysosomes and acrosomes. Moreover, two Rab family members were identified to be involved in endocytosis and acrosome development, such as Rab5 (Simonsen et al., 1998) and Rab7 (Ramalho-Santos et al., 2001), respectively. Although acrosomes and lysosomes share several common characteristics, there are also dissimilarities. As mentioned earlier, the acrosome has been suggested to be a modified secretory granule (Moreno and Alvarado, 2006). For instance, serine proteinases (unique to testis), AM67 (a secretory component protein), acrosin acrogranin (Ohmura et al., 1999; Abou-Haila and Tulsiani, 2000) and exocytotic properties support the idea that the acrosome could be considered analogous to a secretory granule. Secretory granules are known to carry luminal protein contents that are directly delivered by the biosynthetic pathway to the targeted organelle and do not traverse to other parts of the endosomal system (Arvan and Castle, 1998). Secretory lysosomes, however, receive proteins through biosynthetic and endocytic pathways and serve as both degradative and secretory compartments (Blott and Griffiths, 2002). Secretory lysosomes are also mostly found in hematopoietic lineage-derived cells. To solve the mystery of whether the acrosome is really a secretory lysosome, Lamp-1 and Lamp-2 (lysosome specific proteins) were studied in detail during spermiogenesis. Results showed that both Lamp-1 and Lamp-2 were found to link with cytoplasmic vesicles only and not to the growing acrosome (Moreno, 2003). Although AR is usually associated with a somatic cell exocytosis, it still has several exclusive characteristics. For instance, AR is an irreversible "allor-nothing" event. Moreover, in contrast to a single large vacuole of sperm, there are several secretory vesicles in cells that show exocytosis. In addition, AR takes place only once in each sperm because once the acrosome has reacted, it cannot be replaced by further biogenesis. Similarly, acrosomal membranes are lost and cannot be recycled at the time of AR (Berruti, 2016). In each sperm, only a single secretory granule exists, in contrast to the numerous secretory vesicles found in most other exocytotic cells. A feature that makes the acrosome unique is that the acrosome remains undocked prior to the required exocytosis stimulus, which is in contrast to other granules that are docked even before the application of the relevant stimulus (**Table 2**; Zanetti and Mayorga, 2009; Tsai et al., 2010; Rodríguez et al., 2011, 2012). Another interpretation has also been made in regards to secretory granule/lysosome nature of acrosome. Sperm traversing through the zona pellucida doesn't show any enzymatic lysis in eutherian (Bedford, 1998). A physical thrust, from the sperm head's structure was implicated in vitelline coat invasion (Bedford, 1998; Berruti, 2016). This investigation pointed out the importance of the "exocytotic" release of the acrosomal contents to penetrate the zona pellucida.

#### Is the Acrosome a Lysosome-Related Organelle?

The acrosome contains a vast array of acidic hydrolytic enzymes that are essential for the AR to take place normally and help the sperm to bind and dissolve the oocyte coverings during the exocytic release at the time of fertilization (Hartree TABLE 2 | Similarities and differences between acrosome and secretory granules of other exocytotic cells.


and Srivastava, 1965; Allison and Hartree, 1970; Jin et al., 2011). Initially, researchers identified the lysosomal enzyme, hyaluronidase, and later several other hydrolytic enzymes such as acid phosphatase (another lysosomal enzyme), glycohydrolases, proteases, esterases, and aryl sulfatases in the acrosomes. These findings led to the suggestion that the acrosome is nothing unique, but a specially modified lysosome that has evolved to facilitate the fertilization process (Hartree and Srivastava, 1965; Allison and Hartree, 1970; Zaneveld and De Jonge, 1991; Tulsiani et al., 1998). Further strengthening this hypothesis was the characterization of analogous histochemical properties, the acidic pH of both the organelles (Allison and Hartree, 1970) and pro-acrosomal vesicle biogenesis in the Golgi apparatus (**Table 3**; Burgos and Fawcett, 1955; Dooher and Bennett, 1973). Among all the acrosomal enzymes, acrosin remains the most well-characterized protease (McRorie and Williams, 1974), but its location in the acrosome remains controversial. Is acrosin associated with the IAM, in the vicinity of the acrosomal membrane, or in the acrosomal matrix (Polakoski and Zaneveld, 1976; Shams-Borhan et al., 1979; Berruti and Martegani, 1982, 1984; Castellani-Ceresa et al., 1983)? In contrast, many believed acrosin existed to help sperm progress through the zona pellucida, and was therefore named the zona pellucida proteolytic enzyme/zonapenetrating enzyme (Chang and Hunter, 1975). Previously, a main research focus was to determine the function of different hydrolytic enzymes, especially acrosomal enzymes involved in the swift focal lysis of the outer coverings of the oocyte for fertilization. Researchers believed that the acrosome was a specialized lysosome, and enzymatic lysis would be required for sperm–oocyte fusion. But no consensus existed to unite both of those ideas (Bedford, 2014). The belief that the "acrosome is a specially modified lysosome" was finally shattered with the advent of gene-knockout technology, which revealed that acrosin is not indispensable for fertilization (Baba et al., 1994). In the

TABLE 3 | Similarities between acrosome and lysosome.


#### BOX 1 | Glossary.

Perinuclear theca: a condensed cytoskeletal structure that encompasses the nucleus of the mammalian spermatozoa, except near the tail implantation region. The structure consists of two distinct regions, a subacrosomal layer/perforatorium, and postacrosomal regions.

Acroplaxome (Greek words akros, topmost; platys, flat; soma, body): a structure located in the sub-acrosomal space that holds the developing acrosome to the spermatid nucleus.

Manchette: is a temporary structure that persists only during elongation and encompasses the elongating spermatid head. The structure lies below the marginal ring of the acroplaxome and consists of a perinuclear ring and inserted microtubular mantle.

Step 14 Spermatid: Spermiogenesis occurs in 16 steps (1-16) in mice. Step 14 spermatids are characterized by chromatin condensation, nucleus elongation, and transformation of the head in sickle a shape.

past few years, the acrosome was suggested to be a lysosomerelated organelle (LRO) based on findings that acrosome biogenesis involves both the biosynthetic and endocytic pathways (Berruti et al., 2010; Berruti and Paiardi, 2011).

LROs are special membrane-bound organelles that received cargo from early endosomal intermediates and link biosynthetic and endosomal systems (Delevoye et al., 2009). LROs show a unique morphology, composition, and physiology and represent the resident cell. It has been suggested that the acrosome is an LRO, and it receives diverse protein cargos from more nonredundant pathways that contribute to acrosome biogenesis. It could therefore be construed that the lytic contents are primarily delivered from the ER-Golgi-TGN route while the acrosomal matrix scaffold and membranous constituents are contributed by the early endosome–endosome intermediates-TGN route. In addition, the complexity of the structure and physiology of the mature LRO is supported by the involvement of several routes. In line with this, several acrosomal features such as high spatial regulation of acrosome biogenesis, highly polarized location, and "modular" organization of its ingredients are consistent with the proposed nature of the acrosome as an LRO (Marks et al., 2013). Proteomic analysis data (Guyonnet et al., 2012) also uncovered unique analogies of biogenesis and protein contents between acrosomes and LROs. For instance, melanosome biogenesis follows four developmental stages (Seiji et al., 1963) that are similar to acrosome biogenesis. Furthermore, both the melanosome matrix and the acrosomal matrix can self-aggregate. Theses matrices both possess a core made up of a firm amyloidogenic structure that is later transformed into a functional matrix by successive attachment of several proteins (Delevoye et al., 2009; Guyonnet et al., 2012). Recently, we found that germ cell-specific atg7-knockout mice produce a globozoospermia-like phenotype due to a malformed acrosome, and autophagy was found to mediate the proacrosomal vesicle transport or fusion in the acrosome (Wang et al., 2014). Thus, the acrosome is proposed to have originated from an autolysosome rather than a lysosome alone. In support of this hypothesis, Sirt1 (sirtuin 1) and Tbc1d20 (TBC1 domain family, member 20) have been found to be involved in acrosome biogenesis and regulate autophagic flux (Sidjanin et al., 2016; Liu et al., 2017).

#### PERSPECTIVE AND CONCLUDING REMARKS

Although the origin and the first appearance of acrosomes are yet to be determined, it could have originated from the simplest eukaryotes, such as yeast. A prerequisite for yeast mating is the formation of shmoo tips/ projections (Merlini et al., 2013), which show several features similar to the typical acrosome, such as the presence of degradative enzymes and the trafficking of some vesicles. Therefore, the primary form of the acrosome could have developed very early in the evolutionary history. Nevertheless, the existence of a true acrosome can be traced back to the evolution of heterogamy, heterozygosity, and crossfertilization. Initially, all organisms were homogametic, and most were self-fertilizing. Later, some organisms evolved crossfertilization. To better adapt the ever-changing circumstances of the earth (Shields, 1982; Schmidt-Rhaesa, 2007; Stearns, 2013), protective coverings around eggs developed to ensure gamete integrity, and sperms needed to arm with a powerful weapon in its arsenal, the acrosome, to evade these protective gamete vestments (Schmidt-Rhaesa, 2007). This review provides a brief historical overview and highlights new break-through on acrosome biogenesis. The acrosome might be generated from a combination of many membrane trafficking systems during gametes fusion, and it might have evolved during the arms race between sperm and ovum. A collective effort to uncover unidentified components, their interactions, and their regulatory mechanism(s) is urgently needed to elucidate a more complete picture of this highly complicated secretory vesicle. Our current era definitely will be a time of deep understanding of acrosome biogenesis.

#### AUTHOR CONTRIBUTIONS

MK collected the data, drew the figures, and wrote the manuscript. HG revised the figures and the manuscript. WL proposed the idea and revised the manuscript.

#### FUNDING

This work was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA16020701), the National Key R&D Program of China (Grant No. 2016YFA0500901), and the National Natural Science Foundation of China (Grant No. 91649202).

## REFERENCES


the spermatid ESCRT-0 complex and microtubules. Biol. Reprod. 82, 930–939. doi: 10.1095/biolreprod.109.081679


Simonsen, A., Lippe, R., Christoforidis, S., Gaullier, J.-M., Brech, A., Callaghan, J., et al. (1998). EEA1 links PI (3) K function to Rab5 regulation of endosome fusion. Nature 394, 494–498. doi: 10.1038/28879


Stearns, S. C. (2013). The Evolution of Sex and Its Consequences. Basel: Birkhäuser.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Khawar, Gao and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Role of Sperm Centrioles in Human Reproduction – The Known and the Unknown

Tomer Avidor-Reiss1,2 \*, Matthew Mazur1,2, Emily L. Fishman<sup>1</sup> and Puneet Sindhwani1,2

<sup>1</sup> Department of Biological Sciences, College of Natural Sciences and Mathematics, The University of Toledo, Toledo, OH, United States, <sup>2</sup> Department of Urology, College of Medicine and Life Sciences, The University of Toledo, Toledo, OH, United States

Each human spermatozoon contains two remodeled centrioles that it contributes to the zygote. There, the centrioles reconstitute a centrosome that assembles the sperm aster and participate in pronuclei migration and cleavage. Thus, centriole abnormalities may be a cause of male factor infertility and failure to carry pregnancy to term. However, the precise mechanisms by which sperm centrioles contribute to embryonic development in humans are still unclear, making the search for a link between centriole abnormalities and impaired male fecundity particularly difficult. Most previous investigations into the role of mammalian centrioles during fertilization have been completed in murine models; however, because mouse sperm and zygotes appear to lack centrioles, these studies provide information that is limited in its applicability to humans. Here, we review studies that examine the role of the sperm centrioles in the early embryo, with particular emphasis on humans. Available literature includes case studies and case-control studies, with a few retrospective studies and no prospective studies reported. This literature has provided some insight into the morphological characteristics of sperm centrioles in the zygote and has allowed identification of some centriole abnormalities in rare cases. Many of these studies suggest centriole involvement in early embryogenesis based on phenotypes of the embryo with only indirect evidence for centriole abnormality. Overall, these studies suggest that centriole abnormalities are present in some cases of sperm with asthenoteratozoospermia and unexplained infertility. Yet, most previously published studies have been restricted by the laborious techniques (like electron microscopy) and the limited availability of centriolar markers, resulting in small-scale studies and the lack of solid causational evidence. With recent progress in sperm centriole biology, such as the identification of the unique composition of sperm centrioles and the discovery of the atypical centriole, it is now possible to begin to fill the gaps in sperm centriole epidemiology and to identify the etiology of sperm centriole dysfunction in humans.

Keywords: centriole, sperm, embrio, cilium, centrosome, infertility, male factor, reproduction

## INTRODUCTION

Centrioles are essential for animal development and physiology, as demonstrated by a variety of experiments that have tested the centriole's role directly (Bettencourt-Dias et al., 2011). Most of these experiments have been performed in vitro, in human immortalized cells, or in animal models, limiting our knowledge of the centriole's role in human reproduction. In general, the centriole's

Edited by:

Karin Lykke-Hartmann, Aarhus University, Denmark

#### Reviewed by:

Paulo Navarro-Costa, Gulbenkian Institute of Science, Portugal Giuliano Callaini, University of Siena, Italy

\*Correspondence: Tomer Avidor-Reiss tomer.avidorreiss@utoledo.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

> Received: 28 June 2019 Accepted: 23 August 2019 Published: 01 October 2019

#### Citation:

Avidor-Reiss T, Mazur M, Fishman EL and Sindhwani P (2019) The Role of Sperm Centrioles in Human Reproduction – The Known and the Unknown. Front. Cell Dev. Biol. 7:188. doi: 10.3389/fcell.2019.00188

role is expected to be similarly essential in humans because of its conservation throughout animal evolution (Carvalho-Santos et al., 2011). However, the centrioles of sperm and the early embryo of murine animals are exceptions to this evolutionary conservation (**Figure 1**). While humans and many other mammals have centrioles in their spermatozoa and early embryos, mice, rats, and hamsters (the most common experimental mammals) do not have recognizable centrioles in their spermatozoa and early embryos (Schatten et al., 1986; Sathananthan et al., 1996; Phillips et al., 2014). These major differences in centriole appearance raise the question: What exactly is the role of the centriole in human fertilization and early embryonic development?

One extreme and unlikely idea is that sperm centrioles are not needed in humans because they are undetectable in mice sperm and appear to be expendable for murine live birth. However, an alternative hypothesis is that mice have evolved a novel biology of sperm and early embryonic centrioles, and, therefore, studying their role in reproduction is not applicable to their role in human reproduction and their clinical implication. Indeed, all studied non-murine mammals do have centrioles in their sperm and early embryo, and, in certain animals, some evidence indicates that these centrioles are necessary. Here, we will explore this evidence and the fact that the sperm of humans and non-murine mammals have two centrioles (one with typical structure and one atypical), not one, as was concluded in the past (Avidor-Reiss and Fishman, 2018). We start with a brief general introduction to centrioles, then progress to discussion of sperm and early embryonic centrioles. Then, we describe clinical studies that implicate the centriole in human reproduction. Finally, we propose future directions to resolve the question of the role of the sperm centriole role in the embryo. This review focuses on the role of the mature sperm (spermatozoon) centrioles in the zygote, and it does not address the role of centrioles in germline (spermatogonia, spermatocyte, and meiosis) development [reviewed in Riparbelli and Callaini (2011)].

#### MOST DIVIDING ANIMAL CELLS HAVE PRECISELY TWO CENTRIOLES WITH A CONSERVED BARREL SHAPE

Centrioles are evolutionarily conserved cellular components essential for fertilization, cell development, and animal physiology through their function in the cell [reviewed in Nigg and Raff (2009)]. The centriole is a cylinder built from nine triplet microtubules arranged in a ring-like formation to form a barrel structure. In the centrosome, the centrioles are surrounded by pericentriolar material (PCM) (Vorobjev and Chentsov Yu, 1982) [reviewed in Lange and Gull (1996), Mennella et al. (2012)]. In cilia and flagella, the centriole triplets are continuous with the doublet microtubules of the axoneme. Centrioles are composed of more than one hundred proteins (Preble et al., 2000) [reviewed in Winey and O'Toole (2014), Nigg and Holland (2018)], and, while new components of centriolar protein structure continue to be discovered, determining their functionality can be a challenge, both because they often have essential roles in early development [reviewed in Schatten and Sun (2010)], and because their function may vary from one cell type to another [previously reviewed in Loncarek and Bettencourt-Dias (2018)].

Most cells have two centrioles during early interphase. Most centrioles form by "duplication," where each of the two-preexisting centrioles direct the formation of one new procentriole, providing a mechanism to control the number of centrioles formed. Centrioles may infrequently form de novo, in the absence of preexisting centrioles or in specialized cell types, but often times, this mechanism results in too many centrioles being formed (Loncarek and Khodjakov, 2009; Shahid and Singh, 2018). Progression from a procentriole to a mature centriole correlates with cell-cycle progression (Lange and Gull, 1996). A pair of centrioles is made of the cell's older mother centriole and a younger daughter centriole. As a result, the number of centrioles in the cell oscillates between two centrioles during early interphase and four centrioles during late interphase and mitosis (Fukasawa, 2007).

Centrioles are integral to the cilia, which confer motility and extra-cellular communication potential to the cell. In most cell types, the cilium is formed by the mature centriole within the cell, which is known as the basal body [reviewed in Avidor-Reiss and Gopalakrishnan (2013)]. However, it was recently proposed that in sperm, the flagellum is formed by the younger daughter centriole, but this proposal requires further support (Garanina et al., 2019). Centrioles are also involved in the cell division process, but their contributions vary in distinct cell types. In general, centrioles recruit PCM to form a centrosome. This centrosome associates with microtubules to assemble an aster that localizes to the spindle poles during mitosis. The asters' main function is to determine the axis of cell division and the number of spindle poles (Bobinnec et al., 1998). Interestingly, some animal cells [oocytes, importantly (Calarco et al., 1972)] and many plant cells (Joshi and Palevitz, 1996) lack centrioles. This implies that centrioles are not necessary in all cell types and that some cells can successfully complete mitosis and cellular organization without centriolar influence. Therefore, the significance of centrioles may vary in the several cell types that normally contain them.

#### HUMAN MATURE OOCYTES HAVE NO CENTRIOLES

Oocytes contain most of the elements necessary for zygotic development (i.e., Golgi, ER, and proteasomes), including centriolar and PCM proteins, but they lack assembled centrioles. Oocyte centrioles are eliminated, either by extrusion or inactivation, during oogenesis in humans and most studied animals (Hertig and Adams, 1967; Sathananthan et al., 1985; Schatten et al., 1986; Pickering et al., 1988; Crozet, 1990; Navara et al., 1994; Crozet et al., 2000) [reviewed in Sathananthan et al. (2006)]. Limited studies have explored the mechanism of oocyte centriole elimination, but they propose a mechanism reliant on some of the same proteins that are involved in centrosome biology, such as PLK1 (Pimenta-Marques et al., 2016). Centriole elimination by the oocyte necessitates contribution of both

spermatozoan centrioles to the zygote to ensure an appropriate number of centrioles in the embryo cell (Connolly et al., 1986; Schatten, 1994; Manandhar et al., 2005; Pimenta-Marques et al., 2016). More information can be found in a recent review on how animal oocytes assemble spindles in the absence of the centriole at Severson et al. (2016).

#### MOUSE SPERMATOZOON AND ZYGOTE HAVE NO RECOGNIZABLE CENTRIOLES

In mice, centriole inheritance to the embryo diverges from other mammalian models. In mice, centrioles appear to completely degenerate during spermiogenesis (Manandhar et al., 1998), and, as a result, sperm centrioles may not be present within the zygote (Schatten, 1994; Hewitson et al., 1997) (**Figure 1C**). It has been proposed that zygotic centrioles are inherited maternally or form de novo (Schatten et al., 1985, 1986, 1991; Gueth-Hallonet et al., 1993; Hewitson et al., 1997). The support for nonpaternal inheritance is based on several observations, including: (1) zygotes do not appear to have a dominant sperm aster and, instead, have maternal mini-asters, (2) the sperm axoneme is present in the oocyte but lacks microtubules and shows no association with mitotic poles, (3) centrioles are observed only starting in the 32/64 cell stage of early embryos (Gueth-Hallonet et al., 1993), and (4) intracytoplasmic sperm injection (ICSI) with disassociated sperm nuclei is sufficient for embryo development (Kuretake et al., 1996; Yan et al., 2008). The question of why the centrioles of the murine zygote are not inherited from the sperm, as is seen with other mammals, remains unanswered.

### PARTHENOGENIC CELLS HAVE AN UNREGULATED NUMBER OF CENTRIOLES

The ability to form a parthenogen (an embryo that is developed from an unfertilized egg that lacks centrioles) and parthenogenic cell lines is often referenced to suggest that sperm centrioles are not essential in mammals. However, it is important to note that mammalian parthenogenesis does not lead to viable offspring (Wininger, 2004). It has been proposed that imprinting defects are the main barrier to viability in mammalian parthenogenesis (Kono, 2006; Miller et al., 2019), but concomitant centriole abnormalities present during mammalian parthenogenesis may also contribute to this barrier.

The importance of sperm centrioles in mammalian parthenogenic embryos is apparent, in that parthenogenic cells have an abnormal number of centrioles, as expected from a de novo mechanism of centriole formation (Brevini et al., 2012). In the absence of preexisting centrioles (as would be expected in parthenogenesis), an unregulated number of centrioles is expected to form de novo in the activated oocytes or subsequent cell types of the embryo (Hinchcliffe, 2011). The presence of too many or too few centrioles, as is observed in mammalian parthenogenic cell lines, can lead to chromosomal instability and, ultimately, cell death (Sir et al., 2013; Godinho and Pellman, 2014). These centriole abnormalities can partially explain the presence of high levels of chromosomal abnormalities in parthenogenic embryos and their ultimate inability to develop viable offspring (Bhak et al., 2006). Therefore, it is possible that non-murine zygotes need two centrioles to produce a viable embryo, and further study of parthenogenic centrioles is

necessary, particularly to determine at what stage they form, how many are formed, and how they function.

### MOST MAMMALIAN SPERMATOZOA AND ZYGOTES HAVE CENTRIOLES

### The Spermatozoon Neck Has One Typical and One Atypical Centriole

The spermatozoon is a streamlined, motile cell that is made of a head and tail and a neck that contains the two centrioles (**Figure 1**). The head carries half of the genetic material of the embryo, along with proteolytic proteins within the acrosome that help in reaching the oocyte (Cox et al., 2002; Yoon et al., 2008). The tail is a specialized cilium that propels the sperm cell to meet the egg (Basiri et al., 2014; Avidor-Reiss and Leroux, 2015). Spermatozoon morphology is variable across mammals due to evolutionary pressure, making it one of the most diverse cells when compared with the corresponding cell type in other species (Gomendio and Roldan, 2008; Breed et al., 2014). Importantly, a major difference is exhibited between human and mouse spermatozoa. While human sperm, like that of most other mammals, is morphologically broad and flat (spatulashaped), mouse sperm is curved, long, and narrow (sickleshaped) (**Figure 1A**). In humans and most other mammals, the tail is attached to the head near its center, but in mice, the attachment is to the side of the head (**Figure 1C**). Also, the tail dimensions are very similar between humans and other mammals, but the mouse tail is about twice the length of that of humans. Mice exhibit substantial evolutionary conservation of many of the critical developmental processes in other mammals and, in general, are a beneficial model. However, the differences described above suggest that some aspects of murine sperm have evolved differently than and away from other mammals, including humans. Therefore, translating information on this topic from mice to humans requires caution.

The human sperm neck contains two centrioles as well as a specialized PCM. The proximal centriole (PC) is found just near the head base, and the distal centriole (DC) is located further from the head, attached to the base of the axoneme (**Figure 1A**). During sperm formation, the PC forms a centriolar adjunct that is thought to organize the development of the neck region and to guide the manchette to form the axis of nuclear flattening (Fawcett and Phillips, 1969; Lehti and Sironen, 2016). At the same time, the DC microtubules extend to form the axonemal microtubules of the tail (Fawcett and Phillips, 1969). The centrioles become embedded in structural material, including the capitulum around the PC, the striated columns filling most of the neck, and the outer dense fibers (ODFs) (**Figure 1A**). The capitulum and the striated columns form a specialized PCM that also contains centrosomal proteins (Fawcett and Phillips, 1969; Fishman et al., 2018).

The mature spermatozoon obtains its unique streamlined morphology and composition during spermiogenesis [reviewed in Avidor-Reiss and Fishman (2018)]. Early spermatids contain both a PC and a DC with typical centriole structure and composition. The fate of the DC and PC during spermatid differentiation into spermatozoa varies according to animal species, but both centrioles are present in mature spermatozoa of humans and most other mammals. The "centriole remodeling" that occurs during differentiation leaves one centriole that maintains a typical structure (the PC) and one that obtains an atypical structure (the DC). This structure consists of splayed microtubules doublets and centrosomal proteins but maintains the ability to function in the zygote (Fishman et al., 2018). What remains unknown is the usefulness, unique function, and fate in the embryo of these remodeled centrioles and specialized PCM.

## The Two Sperm Centrioles Function in Most Mammalian Early Embryos

Most dividing cells require two centrioles, each of which localizes to one of the spindle poles that mediate chromatid separation between daughter cells when the cell divides. The same is expected for the first cell of the embryo (the zygote). Also, immediately after sperm-egg fusion, the sperm centrioles form an aster, and, shortly thereafter, the two sperm centrioles undergo "duplication," forming two new daughter centrioles [reviewed in Avidor-Reiss and Fishman (2018)]. Finally, at some point in embryonic development, centrioles are expected to template a primary cilium. While the timing for this is unknown in humans and non-murine animals, in mice, the first primary cilia have been observed in blastocysts with 64–100 cells, only after implantation on epiblast cells (Bangs et al., 2015). Based on these functions, sperm centrioles are widely expected to be essential in the embryo for development.

In many animals, such as worms and fish, centrosome reduction results in the elimination of PCM proteins, with little or no apparent change to centriole structure. In insects, the two centrioles are modified, one slightly and one so dramatically that it was only recently discovered (Gottardo et al., 2015; Khire et al., 2016; Dallai et al., 2017; Fishman et al., 2017). The oocyte contains appropriate PCM material that joins with the spermatozoan centrioles after they are introduced. The necessity of both sperm centrioles and maternal PCM for embryo development has been demonstrated in flies (Blachon et al., 2014), nematodes [reviewed in Leidel and Gonczy (2005)], and fish (Yabe et al., 2007). These publications suggest that in many animals, remodeled sperm centrioles are essential in the embryo, and, therefore, the lack of an integral role for sperm centrioles in mice is an exception.

The paternal inheritance of the PC in the zygote has been well established in several mammalian models. The presence of sperm centrioles in the zygote has been demonstrated in cows (Navara et al., 1994), sheep (Le Guen and Crozet, 1989), primates (Simerly et al., 1995; Wu et al., 1996), and pigs (Kim et al., 1996). However, the essential functions of zygotic centrioles in these models have not been directly demonstrated. Remodeling of zygotic centrioles has been established through ultrastructural and some immunological studies in cows (Long et al., 1993; Navara et al., 1994; Sutovsky et al., 1996; Sutovsky and Schatten, 1997), sheep (Le Guen and Crozet, 1989; Crozet, 1990; Crozet et al., 2000), rabbits (Longo, 1976; Szollosi and Ozil, 1991;

Yllera-Fernandez et al., 1992; Pinto-Correia et al., 1994; Terada et al., 2000), and cats (Comizzoli et al., 2006). More recently, it was shown that cow zygotes also inherit a second atypical centriole (the DC) (Fishman et al., 2018). This inheritance pattern suggests that centrioles may play an important role in the developing zygote; however, it is unclear exactly what and how these centrioles may be contributing.

### THE ZYGOTE OF HUMANS INHERITS TWO FUNCTIONAL SPERM CENTRIOLES

Though studies in humans are lacking, two sperm centrioles are likely present in the early zygotes and four during mitosis. The human zygote inherits the PC from sperm, which is well established (Sathananthan et al., 1991). Also, the human zygote is likely to inherit the DC because (1) the spermatozoa have a remodeled DC that is attached to the axoneme (Fishman et al., 2018), and (2) the base of the axoneme is located at one of the spindle poles (Asch et al., 1995; Van Blerkom, 1996; Simerly et al., 1999; Kovacic and Vlaisavljevic, 2000).

Kai et al. (2015) showed the presence of two centrosomes in zygotes with two pronuclei (presumably fertilized by a single sperm) by immunofluorescence staining for centrosomes. Sathananthan et al. (1991) identified centrioles within the centrosomes of zygotes with two pronuclei by electron microscopy following in vitro fertilization; two centrosomes with one or two centrioles were observed. Both Kai et al. (2015) and Sathananthan et al. (1991) have observed four centrosomes in early zygotes with three pronuclei, suggesting that dispermic embryos provide extra centrioles that form extra centrosomes. Interestingly, only three and not four spindle poles were observed in these zygotes. The presence of only three poles in dispermic embryos, despite the presence of four centrosomes, raises the question of whether the number of poles is determined by the sperm centrioles or pronuclei. However, these studies are consistent with the idea that human sperm normally contribute two centrioles to the zygote.

After membrane fusion, the centrosome forms one sperm aster near the head, which then enlarges throughout most of the zygote cytoplasm (Van Blerkom, 1996; Simerly et al., 1999). During pronuclear migration, this aster splits into two. Later, each aster localizes to one pole of the forming bipolar mitotic spindle. Interestingly, after initial formation, the asters collapse and are either small or unrecognizable during metaphase. This collapse is transient, as the asters reappear in anaphase. Why asters collapse during metaphase is unknown and needs further investigation.

It is commonly thought that human sperm centrioles are essential for pronuclear migration, based on studies on abnormal sperm asters in the zygote. Immunofluorescence analysis of fertilized zygotes with arrested development has shown disorganized and diminished sperm aster microtubule arrays, along with the lack of pronuclear formation and/or migration (Asch et al., 1995; Van Blerkom, 1996). These concurrent findings suggest that the sperm aster may be responsible for normal pronuclear development. These studies are limited by the fact that microtubules are also nucleated by non-centriolar microtubuleorganizing centers, and that, in many of these studies, a defect in the centriole was not found or studied, making the specific role of the centriole and sperm aster in the zygote unclear.

After entering the egg, the PC is thought to be released from the sperm neck structures before recruiting PCM and forming astral microtubules, a process proposed to be mediated by proteasomes (Wojcik et al., 2000; Rawe et al., 2008; Sutovsky, 2018), but it is unknown if the DC is detached from the neck structures. Spermatozoan proteasomes have been localized to the neck region/midpiece in close association with the centrioles; however, relatively little is known about their function. Spermatozoa from humans and bovine, preloaded with functionblocking anti-proteasome antibody, resulted in disrupted sperm aster formation and pronuclear development/apposition of human oocytes, despite the lack of observable centriole structural deficits (Rawe et al., 2008). This suggests that spermatozoan proteasomes may play an important role in centriole contribution to zygote development. More pharmacological and genetic studies should be done to investigate the mechanism and the precise role of the proteasome in sperm and zygote centrioles.

#### EPIDEMIOLOGICAL EVIDENCE FOR CENTRIOLE CONTRIBUTION TO HUMAN FERTILITY

Sperm with impaired centrioles were long thought to result in embryos with abnormal cleavage and infertility (Sathananthan, 1994; Schatten, 1994). This idea is now widely accepted and is supported by some epidemiological studies using artificial insemination to overcome barriers to fertilization in defective sperm samples. However, this idea is based on few studies, which analyzed 1–10 patients and few fertile donors as a control (**Table 1**). The phenotypes observed in these studies are diverse, providing opportunity for many inferences but making direct correlation with centriole phenotype difficult. Also, many studies are complicated by the presence of other defects in the sperm (in addition to the centriole), which may have a significant impact on the embryo phenotype. In particular, one can expect that some forms of centriolar defect may originate early in germline development and will result in abnormal DNA content due to abnormities in mitosis or meiosis in addition to the centriole defect, as was found for the mutation in the centriolar protein Centrobin (Liška et al., 2009). Unfortunately, these drawbacks negatively impact the confidence that we have in inferring the role of the sperm centriole in embryo development. However, the totality of these studies suggests that sperm centrioles are likely essential for normal embryo development.

Intracytoplasmic sperm injection is a useful treatment for male infertility due to interference with sperm translocation to the egg and fusion with it, but not for infertility due to dysfunction of sperm components that are needed after fusion (Palermo et al., 1992). Indeed, ICSI allows successful fertilization for many couples, but this may require multiple attempts, and, in many other couples, this fails [reviewed in Neri et al. (2014)].

Avidor-Reiss et al.

fcell-07-00188 September 30, 2019 Time: 10:38 # 6

TABLE 1 |Parameters ofvarious studies of direct sperm phenotype dysfunctions.


(Continued) Human Sperm Centrioles



This suggests that in these failed cases, infertility may be caused by processes downstream of gamete fusion that depend on the male contribution. However, the multifactorial nature of reduced semen quality has prevented many of the underlying contributory molecular mechanisms from being fully elucidated.

The only routine clinical assay to assess sperm phenotype is semen analysis, and this assay by itself cannot implicate the sperm centriole in infertility. Semen analysis compares the quality of the patient's semen and sperm against several standardized parameters and helps identify specific defects in the male which are seen in up to one-third of infertile couples (Isidori et al., 2006) (**Table 1**). One group of sperm defects with several phenotypes that could originate from centriolar defect is idiopathic male factors (male infertility due to a sperm defect that has no explanation, i.e., when the male is infertile despite having normal semen analysis, history, and physical examination, and when female factor infertility has been ruled out) (Chemes and Rawe, 2010). For example, it was suggested that abnormal centrioles will form an abnormal sperm tail (which is one phenotype of teratozoospermia, defined as presence of sperm with abnormal morphology), a defective sperm axoneme resulting in poor motility (asthenozoospermia), or a combination of the two (asthenoteratozoospermia). Some of these cases of male factor infertility may be due to a subtle sperm centriolar defect–a hypothesis that will be examined in the following sections.

#### Asthenoteratozoospermia

Extremely abnormal sperm morphology can inhibit spermatozoan movement through the female reproductive tract and prevent fusion with the oocyte. These morphological abnormalities have many causes and take many forms, and their definition is controversial, but they can be present in ∼10% of infertile men (Khan et al., 2011). As described below, some of these cases are suggested to be caused by or accompanied by a centriolar defect, but these defects have not been fully elucidated.

#### Head-Neck Defect

Head-neck defects characterized by separated heads and tails provide the strongest evidence that human centrioles are essential for fertility; however, this evidence does have some limitations. In mice, injection of the sperm nucleus without the sperm tail results in a fertilized egg that produces offspring, suggesting that mouse sperm centrioles are dispensable (Kuretake et al., 1996). In contrast, in humans, injection of the sperm nucleus without the sperm tail is insufficient to produce offspring (Palermo et al., 1997; Emery et al., 2004). This may suggest that centriolar absence is the reason for infertility, but this interpretation is limited by the fact that other components in the tail may be essential for embryo development in humans. For example, it was recently shown that MicroRNA-34c localizes to the sperm neck and is required for the first cleavage division (Liu W.-M. et al., 2012; Fereshteh et al., 2018). Therefore, it is not possible in these types of experiments to disentangle the phenotypical consequences of not providing paternal centrioles from that of not delivering other factors essential in the zygote.

The centriole-containing sperm neck is the mechanically weakest part of the sperm. Externally applied pressure may cause

the sperm neck to break, forming two fragments: a head fragment and a neck-tail fragment (Firat-Karalar et al., 2014). Indeed, decapitated sperm often breaks between the head and tail and only more rarely at the midpiece (Baccetti et al., 1989; Rawe et al., 2002; Porcu et al., 2003; Gambera et al., 2010). Another rare form of break is caused by disassociation between the PC and DC (Holstein et al., 1986). A variety of structural abnormalities have been observed in sperm with head-neck defect, but their causes are still not well understood. Easily decapitated sperm syndrome is one of the mildest subtypes of head-neck defect teratozoospermia, as some heads and tails in the ejaculate remain connected (Perotti et al., 1981; Toyama et al., 2000). Electron microscopy of decapitated sperm often reveals an absent basal plate and/or implantation fossa with observable breaks between the head and tail (Baccetti et al., 1989) (**Figure 2**). Injection of individual or separated sperm parts allows oocyte activation and pronucleus formation but does not facilitate pronuclear migration and fusion, leading to abnormal embryos (Colombero et al., 1996). Embryos injected with dissected isolated sperm tails or separated heads and tails show chromosome mosaicism, suggesting centrosome and centriole dysfunction (Palermo et al., 1997). However, overall, ICSI has mixed outcomes with easily decapitated sperm, sometimes overcoming infertility but many other times failing despite good embryo morphology (Kamal et al., 1999; Saias-Magnan et al., 1999).

Chemes et al. (1999; **Table 1**) used electron microscopy of testicular biopsies from patients with this defect to show that the forming sperm tail developed uncharacteristically independent of the nucleus, a phenomenon which can be caused by a centriolar defect. Rawe et al. (2002; **Table 1**) used immunofluorescence of human spermatozoa with this defect injected into bovine oocytes to show scarce centrosome-associated microtubules and arrested sperm asters, as compared to fertile donor controls that formed sperm asters in a majority of samples. Ejaculated spermatozoa from patients were also injected into human metaphase II oocytes and analyzed by immunofluorescence. No pronuclei fusion or zygote cleavage occurred with the first attempt. The following three attempts selectively injected sperm with near-normal head-neck alignment; they resulted in fusion and cleavage, but the embryos failed to develop into pregnancy. Emery et al. (2004; **Table 1**) performed ICSI with easily decapitated sperm and reported a successful birth when the separated tails were placed immediately next to the sperm head within the oocyte. The authors surmised that careful integration of all parts of the sperm (head and tail) during ICSI ensured that centrioles, which are likely lost in other less delicate techniques, are included and ultimately allow normal development of the oocyte. Indeed (**Table 1**), successful pregnancies can be achieved by ICSI using intact sperm from men with abnormal head-tail junction (Porcu et al., 2003; Gambera et al., 2010).

#### ICSI With Sperm Tail Abnormality (MMAF)

Multiple morphological abnormalities of the flagella (MMAF) is a rare form of asthenoteratozoospermia known to cause male infertility (Wang et al., 2019). MMAF is caused by mutations in proteins of the axoneme or transport the mechanism that forms it [e.g., inner-arm heavy chain DNAH1 (Ben Khelifa et al., 2014), or the intraflagellar transport (IFT)-associated protein TTC21A (Liu et al., 2019)] an estimated 50% or more of the molecular mechanisms behind MMAF are yet to be identified. Recently, a sporadic defect in the centriolar protein CEP135 was found to contribute to MMAF (**Table 1**). Sha et al. (2017) identified an inherited missense homozygous mutation in CEP135 in an infertile male with MMAF. Immunofluorescence analysis showed that CEP135 localized to the PC in normal spermatozoa, while the patient's mutated proteins localized elsewhere in the sperm, forming ectopic aggregates in the sperm neck and flagella (**Figure 2**). Following ICSI, embryos demonstrated pronucleus formation and cleavage, but ultimately the mother failed to become pregnant. This suggests that a centriole abnormality can produce a phenotype that is exhibited later in embryogenesis.

#### Dysplasia of the Fibrous Sheath

The fibrous sheath is a structure surrounding the axoneme in the flagellar principle piece (**Figure 2**). Dysplasia of the fibrous sheath is a rare condition of immotile sperm, which have morphological deformities in the neck, midpiece, and tail and exhibit centriolar dysfunction. Microscopy of dysplastic sperm reveals an increased and disorganized fibrous sheath with disruption of the underlying axoneme (Chemes et al., 1987a). Immunofluorescence with the centriolar protein Centrin-1 shows frequent abnormal positioning of the centrioles, sometimes with duplication, resulting in two nucleus-implantation sites (Moretti et al., 2017). Initial attempts at ICSI with dysplastic sperm were unsuccessful at achieving pregnancy, due to failure of either fertilization or embryo development (Chemes and Rawe, 2003). Nakamura et al. (2005) used bovine oocytes for heterologous ICSI and found that the vast majority of injected oocytes showed no immunocytochemical evidence of sperm aster formation or cytoplasmic microtubule organization. This is consistent with the observation that only 2% of the dysplastic sperm expressed the centriolar protein centrin at the midpiece. Failure to form sperm asters in the setting of severely diminished centrin expression strongly suggests that the sperm centrioles are either abnormal or missing. Therefore, it seems likely that centriole abnormality is contributing to infertility in some of these patients.

#### Globozoospermia

Globozoospermia is a rare condition describing spermatozoa with a round head and disorganized mid-piece that lack both a functional acrosome and acrosomal enzymes (Singh, 1992). An absent acrosome prevents penetration of the oocyte zona pellucida, and was originally believed to be the sole cause of infertility in affected spermatozoa (Rybouchkin et al., 1996). The rate of successful fertilization with ICSI for men with globozoospermia is lower than the general ICSI success rate [reviewed in Dam et al. (2007)], suggesting that there may be additional problems with reproductive machinery that are not overcome by ICSI. Centriole dysfunction may help explain this discrepancy. Indeed, globozoospermic sperm appear to have centriole dysfunction, based on heterologous

ICSI with bovine oocytes [reviewed in Dam et al. (2007)] (**Table 1**) as well as bio-fluorescent staining of Centrin-1 (Moretti et al., 2019). For example, Nakamura et al. (2002) found that globozoospermic sperm demonstrated a significantly lower rate of sperm aster formation compared to control fertile donor sperm, suggesting a centrosome dysfunction. However, these also showed a significantly lower rate of male pronucleus formation and a significantly higher rate of prematurely condensed chromosomes, suggesting a more pleiotropic mechanism. Many papers show that globozoospermic sperm centrioles have normal ultrastructure, but the composition has yet to be fully investigated. However, due to the heterologous nature of globozoospermia, the discovery of the atypical DC, and the functional data showing poor aster formation in globozoospermia embryos, we propose that more work is necessary to test whether or not malformed centrioles may be implicated in this disease.

#### Asthenozoospermia

Poor sperm motility seen in asthenozoospermia impedes proper transport of spermatozoa to the oocyte for penetration and membrane fusion. Theoretically, this infertility should be completely resolved by ICSI since the spermatozoa are artificially transported to within the oocyte. However, the low rate of successful pregnancies with ICSI in the setting of complete asthenozoospermia suggests the presence of additional factors (Nagy et al., 1995; Casper et al., 1996; Kahraman et al., 1996; Nijs et al., 1996; Cayan et al., 2001; Fonttis et al., 2002; Chatzimeletiou et al., 2007; Terada et al., 2009). Immunocytochemical analysis of oocytes which had failed fertilization by an athenozoospermic

donor showed several phenotypes, including two pronuclei without the presence of recognizable sperm aster formation, suggesting centrosomal dysfunction had caused fertilization arrest (Terada et al., 2009). Western blot and ELISA analysis found that the centriolar protein centrin and the centriolar and flagellar protein Tektin 2 are reduced in oligoasthenozoospermic men, suggesting that their lower levels can result in lower pregnancy percentage after ICSI (Hinduja et al., 2008, 2010; Bhilawadikar et al., 2013). Also, recently, the base of the sperm tail was found to function as an atypical centriole in the zygote (Avidor-Reiss and Fishman, 2018). Thus, centriolar dysfunction may not only impair sperm motility as originally believed but may also cause a disruption of zygotic intracellular processes that ICSI cannot remedy.

#### Azoospermia Treated Using Round Spermatids

Azoospermia is defined as the complete absence of sperm from the ejaculate and is present in ∼15% of infertile men (Jarow et al., 1989). It can be treated using spermatids or testicular spermatozoa. However, the use of round spermatids has limited success, and fertilization rates were lower than those obtained using elongating spermatids or spermatozoa [discussed in Liu X.-Y. et al. (2012)]. One possible explanation is that the two sperm centrioles are not yet remodeled in round spermatids. How the status of centriolar remodeling affects embryo development needs further investigation.

#### Unexplained Infertility

Unexplained infertility, the absence of any observable male or female factor, is known to cause infertility in 10–30% of cases (Hart, 2003; Gunn and Bates, 2016). Recently, Garanina et al. (2019) suggested that a longer sperm centriolar adjunct is associated with unexplained infertility. They found through transmission electron microscopy that the average total length of the PC and its centriolar adjunct was significantly longer in the spermatozoa of two unexplained infertility patients than in the spermatozoa of five healthy donors. The two affected patients displayed repeated zygotic arrest after in vitro fertilization. One patient had a centriolar adjunct nearly double the length seen in healthy donors and failed all fertility treatments. The second patient with intermediate findings (∼1.5 longer) finally conceived a healthy baby that was delivered at 40 weeks of gestation. Ultimately, the function of the centriolar adjunct is still unknown. In all mammals except for human, the centriolar adjunct is a transient structure present during spermiogenesis but absent in the mature spermatozoon (**Figure 2**). In human spermatozoa, the centriolar adjunct is present and was postulated to be characteristic of relative immaturity of the human sperm (Zamboni and Stefanini, 1971). Therefore, it is unclear if centriolar adjuncts are essential only in humans, if a long centriolar adjunct causes a problem, or if the long adjunct is a marker for other defects in the centrioles. Further study of the association between centriolar adjuncts and fertility outcome is needed.

## A SPECIFIC AND RAPID ASSESSMENT METHOD OF SPERM CENTRIOLES IS NEEDED

Several assays for studying typical sperm centrioles are available. However, the technology is either insufficiently specific or inappropriately laborious. Sperm cell centrioles has been assessed by electron microscopy, a laborious technique that is inadequate for large-scale studies and is inaccessible in most clinical settings (Chemes et al., 1987b; Moretti et al., 2016). Sperm centriolar protein content has been assessed by Western blot, which studied total protein and is not specific to the centriole and cannot conclusively implicate the centriole in infertility (Hinduja et al., 2010). Sperm centriole function has been assessed by microinjection of human sperm into bovine, rabbit, or hamster oocytes followed by immunofluorescence staining for aster formation, which has been useful in studying abnormal sperm centrioles in some infertility cases (Rawe et al., 2002; Terada et al., 2004; Yoshimoto-Kakoi et al., 2008). However, this method is now illegal according to the Dickey-Wicker amendment (the U.S. federal bill that prohibits funding for the creation of human embryos for research purposes). Because of these limitations, sperm centriole defect in infertile men has only been demonstrated in small case studies (Nanassy and Carrell, 2008; Chemes, 2012; Moretti et al., 2016; Sha et al., 2017). Therefore, there is a need for a specific and high-throughput method for assessing sperm centrioles and determining their association with certain embryonic phenotypes and outcomes.

## FUTURE DIRECTIONS

Previous animal and epidemiological studies of sperm centrioles provide a basis for a correlation between centriole abnormality and embryo development pathology; however, further investigation is needed for conclusiveness on this subject. This investigation should include new model mammals and more conclusive clinical research.

Since mice do not have detectable centrioles in the sperm and early embryo, other mammalian models with these centrioles should be developed. For instance, centriole inheritance in rabbits resembles that of humans, and they are suitable models because they are the smallest non-murine mammal with established genetic engineering (Liu et al., 2004; Terada, 2007). The generation of a model rabbit for studying centriolebased infertility would be very useful; however, to do so, the tools necessary for specifically interfering with the sperm centriole must first be identified. Characterizing the mechanisms of sperm centriole formation and function as well as their unique properties will provide the insights necessary to develop these tools. CRISPR/Cas9 editing can be used to alter spermspecific sequences, which will enable the development of a specific model animal.

Despite the correlation between centriole dysfunction and infertility that has been proposed in the epidemiological literature, the extent and significance of this relationship is still not known. The main reason is that most human studies to

date have not demonstrated a clear case of specific centriole dysfunction. Therefore, a diagnostic test should be developed to better identify sperm with specific centriolar defect. Also, because such defect is likely to affect only a small fraction of all infertility cases, the diagnostic test needs to be easy and rapid. One additional approach for improving epidemiological studies is to include advanced semen analysis that examines DNA and egg activation factors, followed by testing sperm that have only centriolar defect. Ultimately, these efforts should lead to more conclusive clinical research, such as large prospective cohort studies and randomized controlled trials.

Sperm centriole defect may be mediated by genetic, environmental, or infectious factors; however, no directed effort has been made to identify these factors. Therefore, there is a need to identify sperm centriole defect in human, to find their cause, and to characterize its impact on male infertility.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

TA-R conceived, supervised, and wrote the manuscript. MM wrote the manuscript and searched the literature. EF prepared the figures. PS performed the clinical perspective.

## FUNDING

This work was supported by grants R03 HD087429 and R21 HD092700 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD).

### ACKNOWLEDGMENTS

We would like to thank Nahshon Puente and Katerina Turner for assistance in preparing the manuscript.


syndrome associated with impaired fertilization. Fertil. Steril. 47, 310–316. doi: 10.1016/s0015-0282(16)50011-9


parthenogenesis. Biol. Reprod. 54, 1397–1404. doi: 10.1095/biolreprod54.6. 1397



development in rhesus monkeys, and regulation by intracellular calcium ions. Biol. Reprod. 55, 260–270. doi: 10.1095/biolreprod55.2.260


using heterologous ICSI. Syst. Biol. Reprod. Med. 54, 135–142. doi: 10.1080/ 19396360802043091

Zamboni, L., and Stefanini, M. (1971). The fine structure of the neck of mammalian spermatozoa. Anatomical. Record. 169, 155–172. doi: 10.1002/ar.1091690203

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer GC declared a past co-authorship with one of the authors TA-R to the handling Editor.

Copyright © 2019 Avidor-Reiss, Mazur, Fishman and Sindhwani. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Paternal Contributions to Offspring Health: Role of Sperm Small RNAs in Intergenerational Transmission of Epigenetic Information

#### Upasna Sharma\*

Department of Molecular, Cell and Developmental Biology, University of California, Santa Cruz, Santa Cruz, CA, United States

The most fundamental process for the perpetuation of a species is the transfer of information from parent to offspring. Although genomic DNA contributes to the majority of the inheritance, it is now clear that epigenetic information −information beyond the underlying DNA sequence − is also passed on to future generations. However, the mechanism and extent of such inheritance are not well-understood. Here, I review some of the concepts, evidence, and mechanisms of intergenerational epigenetic inheritance via sperm small RNAs. Recent studies provide evidence that mature sperm are highly abundant in small non-coding RNAs. These RNAs are modulated by paternal environmental conditions and potentially delivered to the zygote at fertilization, where they can regulate early embryonic development. Intriguingly, sperm small RNA payload undergoes dramatic changes during testicular and post-testicular maturation, making the mature sperm epigenome highly unique and distinct from testicular germ cells. I explore the mechanism of sperm small RNA remodeling during post-testicular maturation in the epididymis, and the potential role of this reprograming in intergenerational epigenetic inheritance.

Keywords: sperm, RNA, epigenetic, inheritance, embryo development, exosomes, extracellular vesicles, transgeneration effects

## INTRODUCTION

The possibility that our life experiences can influence phenotypes in our descendants has tremendous implications for basic biology and public health and policy (Jirtle and Skinner, 2007). Indeed, there is mounting evidence from worms to mammals, including humans, that parental environment can influence phenotypes in future generations (Rando, 2012; Heard and Martienssen, 2014; Rando and Simmons, 2015). However, the mechanism of such transgenerational inheritance —sometimes referred to as inheritance of acquired traits, or Lamarckian Inheritance remains mysterious. The inheritance of acquired traits was previously refuted, as there was no known mechanism for the environment to alter the genetic material (DNA) transmitted from parents to offspring. With advances in the field of epigenetics, the inheritance that is not based on DNA sequence but how the DNA sequence is utilized, there is a renewed interest in transgenerational inheritance. Epigenetic information carriers (unlike DNA) are highly dynamic and are often modulated by environmental conditions (Sharma and Rando, 2017), suggesting that

#### Edited by:

Tomer Avidor-Reiss, University of Toledo, United States

#### Reviewed by:

Clemence Belleannee, Laval University, Canada Xin Zhiguo Li, University of Rochester, United States Brett Nixon, University of Newcastle, Australia

> \*Correspondence: Upasna Sharma upsharma@ucsc.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 02 June 2019 Accepted: 18 September 2019 Published: 09 October 2019

#### Citation:

Sharma U (2019) Paternal Contributions to Offspring Health: Role of Sperm Small RNAs in Intergenerational Transmission of Epigenetic Information. Front. Cell Dev. Biol. 7:215. doi: 10.3389/fcell.2019.00215

the environment experienced by parents may influence the phenotype of offspring via alterations to the gametic "epigenome." However, demonstrating intergenerational inheritance has been challenging in mammals, as such inheritance is a mechanistically complex process that would require epigenetic information to be maintained throughout the disruptive process of epigenetic reprograming during gametogenesis, carried in gametes, delivered to embryos at fertilization, and then influence embryonic development. Furthermore, the inheritance of acquired traits has been discredited by many, because of the so-called "Weismann barrier" that prevents transmission of information from somatic cells to germ cells (and thus to offspring). Paternal contributions to intergenerational epigenetic inheritance have been considered especially unlikely, because of the sheer difference in size between sperm and oocytes.

Challenging these preexisting notions about inheritance, many recent studies in mammals provide evidence that paternal exposure to different environmental stressors, such as diet, psychological stress, toxicants, etc., can influence offspring phenotypes, and sperm epigenome plays a pivotal role in the transmission of such phenotypes. Here, I review recent studies on intergenerational inheritance via male germline, discuss the potential role of small RNAs as epigenetic information carriers in sperm, and explore the mechanisms of such inheritance. Intergenerational inheritance involves transmission of parental environmental effects via exposure of the developing embryo and/or germ cells to the environmental insult (transmission from parent to offspring or grandoffspring). For instance, if a female (F0) pregnant with male offspring is exposed, both its embryo (F1) and germ cells of the embryo (F2) are directly exposed. On the other hand, transgenerational inheritance affects generations that were not directly exposed to the environmental insult (F3 and onwards) (Heard and Martienssen, 2014). I primarily discuss examples of intergenerational inheritance as it is more widely studied in mammals and few examples of transgenerational inheritance are discussed. In addition, I mainly focus on rodent studies, the readers are directed to some excellent recent reviews on transgenerational inheritance in various model organisms (Rando, 2012; Heard and Martienssen, 2014; Rechavi and Lev, 2017; Boskovic and Rando, 2018; Perez and Lehner, 2019). Understanding the mechanism of intergenerational inheritance has important implications for the etiology of human diseases. Many common metabolic disorders, such as diabetes, have both genetic components and contributions from a patient's lifestyle and environment. Only a fraction of the heritability of such diseases can be explained by genetic variation; instead, it is now increasingly appreciated that epigenetic inheritance likely contributes to such conditions (Rando and Simmons, 2015).

#### INTERGENERATIONAL TRANSMISSION OF PATERNAL ENVIRONMENTAL EFFECTS

Multiple studies in mammals over the past decade have demonstrated that parents pass on information about their environment to their children. In humans, undernutrition in parents has been linked to metabolic diseases in children, with epidemiological studies of the Dutch Hunger Winter of 1944–1945 (Lumey et al., 2007) providing evidence for effects of maternal starvation during pregnancy on the metabolic health of children. In case of the male germline, studies on three generations of family members in Overkalix, Sweden, demonstrated a correlation between food availability to grandfathers and health and mortality rates in their grandchildren (Kaati et al., 2002; Pembrey et al., 2006). Due to the long term interaction of the mother and the developing fetus in the womb, maternal contributions to offspring phenotypes seem plausible. Indeed, alcohol consumption during pregnancy can lead to fetal alcohol syndrome in humans (Jones and Smith, 1973). However, except for a few studies (Huypens et al., 2016), in most of the studies on maternal exposure, it is difficult to separate environmentally induced molecular changes in the oocyte from the direct effect of the environment on the developing fetus (Rando and Simmons, 2015). On the other hand, as fathers mostly contribute just sperm to the developing offspring, mechanistic investigations of paternal effects are more straightforward, and are of great interest (Rando, 2012).

A large number of laboratory studies in rodents have linked paternal treatment regimes with changes in offspring phenotype, such as dietary alterations (Anderson et al., 2006; Carone et al., 2010; Ng et al., 2010; Fullston et al., 2013; Lambrot et al., 2013; Grandjean et al., 2015; Chen et al., 2016; Huypens et al., 2016; Sharma et al., 2016), psychological stress (Dietz et al., 2011; Rodgers et al., 2013; Gapp et al., 2014; Rodgers et al., 2015; Wu et al., 2016), odor fear conditioning (Dias and Ressler, 2014), exposure to endocrine disruptor vinclozolin (Anway et al., 2005) and ethanol exposure (Rompala et al., 2018). In general, to study intergenerational inheritance of paternal phenotypes, male mice are exposed to different environmental conditions (diet, stress, etc.) and mated with control females. The timing and length of exposure vary in different studies and mostly involves either exposure from weaning to sexual maturity or during fetal development. Next, offspring phenotypic changes are monitored, such as altered glucose metabolism in response to a high-fat diet or depressive-like behavior in response to paternal chronic stress.

In rats, paternal consumption of high-fat diet leads to glucose intolerance in F1 female offspring (Ng et al., 2010). Similarly, male mice fed a low protein diet (10 vs. 19% protein, by mass) were found to sire offspring with altered hepatic cholesterol biosynthesis, relative to control offspring (Carone et al., 2010). Additional studies in mice have used interventions ranging from preconception fasting to fetal undernutrition to link paternal nutrition to glucose metabolism in offspring (Anderson et al., 2006; Jimenez-Chillaron et al., 2009). Moreover, multiple studies have investigated the effects of parental psychological stress, such as social defeat stress, early life trauma and chronic variable stress on offspring phenotypes (Dietz et al., 2011; Rodgers et al., 2013; Gapp et al., 2014). These studies report that paternal exposure to psychological stress usually leads to reduced stress sensitivity in offspring, and altered cortisol levels and glucose metabolism often accompany such phenotypes. The earliest studies in transgenerational inheritance examined the effects of exposure of

pregnant female rats to endocrine disruptor Vinclozolin. F1 male offspring of vinclozolin exposed mothers displayed infertility, and this defect was transmitted to males until F4 generation (Anway et al., 2005), suggesting that paternal environmental conditions can influence the health of offspring across multiple generations. In recent years, exposure to other toxicants such as carbon tetrachloride, and drugs such as nicotine and cocaine have also been reported to influence offspring phenotypes (Zeybel et al., 2012; Vassoler et al., 2013; Vallaster et al., 2017).

It is still not clear how paternal exposure is linked to phenotypes observed in offspring, and whether those effects are adaptive. In fact, as discussed above, in most cases transmission of paternal environmental effect increases disease susceptibility in offspring. One explanation for such phenotypes comes from the "thrifty phenotype" hypothesis (Hales and Barker, 2001), wherein a compromised in utero environment programs offspring for a similar environment after birth, and thus, links poor fetal and infant growth with increased risk of metabolic diseases in adults born in an unmatched ex utero environment (Miska and Ferguson-Smith, 2016). For instance, poor nutrient availability during early development programs offspring to withstand a similar environment, leading to diabetes and obesity in subsequent generations in the absence of nutrient deprivation. Furthermore, an adaptive response might only be revealed when offspring are challenged with a specific environmental insult. It was reported that paternal preconception exposure to nicotine induced a higher metabolic tolerance for nicotine as well as other xenobiotics such as cocaine (Vallaster et al., 2017). The offspring showed higher hepatic expression of various genes involved in xenobiotic clearance, suggesting that in the case of paternal exposure to nicotine, the phenotypes observed in offspring are not specific to nicotine but broadly affects offspring tolerance to additional xenobiotics. Interestingly, this protective response was only revealed in offspring that were pre-exposed to nicotine, suggesting that the phenotypes observed in offspring of exposed fathers are influenced by the interaction of paternal environment with the environment of the offspring (Vallaster et al., 2017).

## TRANSMISSION OF EPIGENETIC INFORMATION VIA SPERM

Despite the wealth of knowledge that paternal environment can influence offspring health, the mechanism of this inheritance is not well-understood in most cases. The likeliest scenario is that epigenetic information is delivered to the zygote by sperm, although alternative information carriers such as seminal fluid, microbiome, and female's judgment of male quality can also play a role (Dietz et al., 2011; Rando, 2012). Use of assisted reproduction methods such as in vitro fertilization (IVF) and foster mothers, where the only contribution from parents is their gametes, allows a direct test of whether paternal environmental information is passed via sperm. A few recent studies reported the effects of paternal exposure (such as diet, stress, etc.) on offspring generated via IVF, similar to those seen in offspring produced using natural mating, suggesting that paternal environmental information is indeed transmitted by sperm (Dias and Ressler, 2014; Chen et al., 2016; Huypens et al., 2016; Sharma et al., 2016). Overall, these studies suggest that paternal environment can influence the health of offspring and this information is transmitted via sperm epigenome, and raise the crucial questions: (1) which epigenetic information molecules in sperm carry environmental information? (2) how does the environment influence those epigenetic signaling molecules, and (3) how do those signals influence offspring gene expression and development? Recent studies provide key insights into these mechanistic questions about the process of intergenerational epigenetic inheritance, as discussed below.

#### EPIGENETIC INFORMATION CARRIERS IN SPERM

Epigenetic inheritance, the inheritance of phenotypic changes in the absence of changes in DNA sequence, is essential for the maintenance of cell states through cell division. In addition to epigenetic inheritance of cell states during organismal development, there is a growing body of evidence that epigenetic information can be transmitted from one generation to the next, with famous examples including RNA interference in C. elegans and paramutation in maize (Fire et al., 1998; Alleman et al., 2006; Heard and Martienssen, 2014; Holoch and Moazed, 2015). Studies of cell-state and transgenerational epigenetic inheritance have identified chromatin structure, DNA modifications, small RNAs, and prions as the main molecular carriers of epigenetic information (Sharma and Rando, 2017). The three most well-characterized epigenetic marks in sperm include histone occupancy and histone modifications, cytosine methylation of DNA, and small non-coding RNAs. Here, I focus on the role of small RNAs in intergenerational inheritance, and briefly review the current understanding of the role of chromatin and DNA methylation in such inheritance [for in depth reviews on epigenetic inheritance via chromatin and DNA methylation, please refer to Heard and Martienssen (2014), Miska and Ferguson-Smith (2016), Boskovic and Rando (2018)].

#### Chromatin

In eukaryotes, the genomic DNA is wrapped around a core of histone proteins to form a nucleoprotein complex known as chromatin (Kornberg and Lorch, 1999). This wrapping of DNA around the histone core regulates gene expression by modulating the accessibility of the underlying DNA sequence. Furthermore, post-translational modifications of histone proteins add another layer of regulation by recruiting various chromatin modifying enzymes (activators and repressors) to regulate gene expression. Inheritance of chromatin states from one generation to the next in multicellular organisms is complicated by the fact that chromatin undergoes dramatic changes during gametogenesis, fertilization, and early development (Yadav et al., 2018; Hao et al., 2019). In mammals, during the process of spermatogenesis, the majority of canonical histones are replaced with transition proteins, which in turn are replaced with smaller basic proteins known as protamines (Balhorn, 2007). Importantly, a small fraction of the genome escapes this remodeling. The specific

regions that maintain histone occupancy during spermatogenesis is unclear, some studies show that a subset of developmentally important gene promoters retain histones (Hammoud et al., 2009, 2014; Brykczynska et al., 2010), while other recent studies report histone retention at repeat-rich regions of the genome (Carone et al., 2014; Yamaguchi et al., 2018). These conflicting results are potentially due to differences in the histone retention assays used in these studies and suggest that potentially multiple populations of nucleosomes exist in mature sperm, and depending on the protocol used, different populations are detected (Rando, 2016). It is unclear whether persisting histones in mature sperm play a role in intergenerational inheritance. A study reported that histone retention did not change in obese and lean men (Donkin et al., 2016). On the other hand, a role for histone modifications in such inheritance is suggested —carbon tetrachloride exposure lead to altered histone acetylation at specific promoters in mouse sperm (Zeybel et al., 2012). Importantly, studies from model organisms and rodents demonstrate that paternal mutations in genes coding for chromatin regulators cause altered phenotypes in offspring that do not inherit the mutation per se (Chong et al., 2007; Siklenka et al., 2015). For instance, overexpression of a histone lysine 4 demethylase KDMA1(LSD1) during spermatogenesis lead to impaired development of offspring for three generations, even in offspring lacking KDM1A overexpression (Siklenka et al., 2015). These studies suggest that correct histone modifications and chromatin structure during spermatogenesis is crucial for proper offspring development, and chromatin potentially plays a role in transmission of paternal environmental effects and motivate future studies to elucidate the mechanisms of such inheritance (Rando, 2016).

#### DNA Methylation

Methylation on the fifth carbon of cytosine results in the formation of 5-methylcytosine (5mC), which is one of the most well-characterized epigenetic information carriers. For instance, cytosine methylation is required for germline epigenetic inheritance of imprinted genes — genes that are expressed from only the maternal or paternal alleles and thus, retain the "memory" of the gender of germline they developed from Bartolomei and Ferguson-Smith (2011). As with chromatin, DNA methylation is also globally erased and reset during two rounds of reprograming, one in primordial germ cells during gametogenesis and the other in preimplantation embryos shortly after fertilization (Hackett and Surani, 2013). Certain regions of the genome escape this global erasure, including the imprinting control regions and transposons (Messerschmidt et al., 2014), suggesting that 5mC in sperm could potentially act as a carrier of paternal environmental information in offspring. Indeed, studies have examined the effects of paternal high-fat or low protein diet (Carone et al., 2010; Wei et al., 2014; Shea et al., 2015; de Castro Barbosa et al., 2016; Donkin et al., 2016), paternal folate deficient diet (Lambrot et al., 2013; Ly et al., 2017), and undernutrition (Radford et al., 2014; Holland et al., 2016) on sperm DNA methylation. Although in most of these studies, only a fraction of the differentially methylated regions are maintained in the next generation, these studies provide a strong premise to further investigate the role of DNA methylation in intergenerational inheritance.

#### Small Non-coding RNAs

Small RNAs play key roles in multiple, well-established epigenetic inheritance paradigms in model organisms, such as RNA interference in C. elegans (Fire et al., 1998) and paramutation in maize (Alleman et al., 2006). Since the discovery of micro RNAs in 1993 (Lee et al., 1993), a multitude of different classes of small RNAs (<40 nts) have been discovered which play critical regulatory roles in biology. The well-studied classes of small RNAs include micro RNAs (miRNAs), endogenous silencing RNAs (endo-siRNAs), and the germ-line enriched piwiinteracting RNAs (piRNAs). In addition, recent advances in RNA-sequencing technology has uncovered additional classes of small RNAs derived from the cleavage of tRNAs known as tRNAderived small RNAs or tRNA fragments (tRFs). Fragmentation products of ribosomal RNAs are also detected in some biological contexts (Lambert et al., 2019); whether they are functional remains to be determined. Small RNAs primarily function in gene regulation by binding to Argonaute (Ago) family proteins, and this Ago-small RNA complex can regulate gene silencing at different levels, such as (1) transcriptional regulation by targeting DNA methylation and repressive chromatin formation at target gene (Volpe et al., 2002; Chan et al., 2004), (2) posttranscriptional regulation by degradation or deadenylation of target RNA, and (3) translational repression (Bartel, 2009). It is important to note that a subset of tRNA-fragments (mainly tRNA-halves, see below) do not work through the Ago-dependent pathways, and are involved in translational repression by directly interacting with translation machinery, polyribosomes, stress granules, etc. (Ivanov et al., 2011, 2014; Goodarzi et al., 2015; Gebetsberger et al., 2017; Keam et al., 2017; Kim et al., 2017; Dou et al., 2019).

Small RNAs can be distinguished based on their biogenesis and biological function, which in turn is often determined by their size and specific sequence characteristics (Ghildiyal and Zamore, 2009; Czech and Hannon, 2011). miRNAs are ∼22 nt small non-coding RNAs processed from precursor RNA molecules. The precursor RNAs are transcribed from either intergenic regions or introns of protein-coding or noncoding RNA genes. miRNAs can also be transcribed as a long transcript from two or more miRNA genes placed adjacent to each other, called clusters, which have similar seed sequence and thus belong to the same family (Ha and Kim, 2014). The precursor transcripts are processed by Ribonuclease III enzyme Drosha to form pre-miRNAs (Lee et al., 2003). PremiRNAs are exported to the cytoplasm by Exportin 5 where they are further processed into mature miRNAs by an RNase III endonuclease Dicer and are loaded onto Argonaute (Ago2) to form the miRNA silencing complex (miRISC) (Hutvagner et al., 2001; Lee et al., 2003). miRISC is then recruited to the target mRNAs (mostly at the 3<sup>0</sup> UTR in mammals) through partial basecomplementarity and represses gene expression by inhibiting translation (Ha and Kim, 2014).

Piwi-interacting RNAs or piRNAs, so called because of their association with PIWI (P-element induced wimpy testis) clade

of Argonaute proteins (Aravin et al., 2006; Girard et al., 2006; Grivna et al., 2006; Lau et al., 2006; Watanabe et al., 2006), are highly expressed in germ cells and are required for male fertility —male mice lacking Piwi proteins are sterile (Carmell et al., 2007). In mammals, two major classes of piRNAs have been reported based on their timing of expression and precursor transcripts: (1) pre-pachytene piRNAs: expressed in fetal and newborn mice and are homologous to various retroelements such as LINE elements, and repress transposon expression to maintain germline genomic integrity (Aravin et al., 2007), (2) pachytene piRNAs: derived from intergenic regions, lack repeat sequences, and are proposed to target spermatogenesis-related mRNAs (Li et al., 2013). In flies and vertebrates, piRNAs have preference for 5<sup>0</sup> Uracil and 3<sup>0</sup> 2 0 -O methylation (Vagin et al., 2006; Kirino and Mourelatos, 2007). Importantly, primary piRNAs can be amplified by producing secondary piRNAs via the so-called pingpong cycle of amplification (Iwasaki et al., 2015). Mechanistically, piRNAs repress gene expression at both transcriptional level, by promoting de novo DNA methylation (Aravin et al., 2008), and post-transcriptionally by cleaving target transposon mRNAs (Reuter et al., 2011).

tRNA-fragments (tRFs) have only recently been studied in depth (Lee and Collins, 2005; Sobala and Hutvagner, 2011; Anderson and Ivanov, 2014; Kumar et al., 2016), and are therefore relatively not well-characterized. tRNAs are well-known for their role in translation as an adaptor molecule for converting information encoded in RNA molecules to peptide chains by delivering amino acids (Hoagland et al., 1958). Small RNAs derived from tRNAs include a diverse variety of RNA species, such as fragments generated by cleavage in anticodon loop known as tRNA halves, or smaller 18–22 nts fragments produced by cleavage in the D or T loops (Keam and Hutvagner, 2015). In addition, tRFs can be derived from 5<sup>0</sup> end, 3<sup>0</sup> ends, or middle of pre-tRNAs or mature tRNAs (Lee et al., 2009). tRFs were initially thought to be random degradation products of tRNAs. With the advent of sophisticated deep sequencing methods, recent studies revealed that tRFs are generated in a remarkably site-specific manner and are derived from a subset of tRNA isotypes in a given cell type, suggesting that their biogenesis is a highly regulated process (Keam and Hutvagner, 2015). tRNA cleavage has been characterized in several biological contexts, where it is typically induced in response to stress conditions. In budding yeast and Tetrahymena thermophila, RNase T2 family endonucleases process tRNAs (Andersen and Collins, 2012), whereas in mammalian cells exposed to stress, the RNase A family member Angiogenin (encoded by RNase5) cleaves tRNAs (Fu et al., 2009). tRFs are found in many cell types but are uniquely highly abundant in the germ cells (Peng et al., 2012; Chen et al., 2016; Sharma et al., 2016). For instance, ∼70% of small RNAs in mature mammalian sperm are composed of tRFs (Peng et al., 2012; Sharma et al., 2016). Treatment of sperm small RNAs with T4 polynucleotide kinase (PNK) in the absence of ATP, which removes cyclic phosphates from the 3<sup>0</sup> ends of RNA molecules and, thus, allows cloning of such RNAs for sequencing, revealed additional tRFs and suggested that RNaseA and/or RNase T2 family endonucleases are involved in the biogenesis of tRFs in the male reproductive tract (Sharma et al., 2018). Functions of tRFs are not well-understood, mostly due to their very recent discovery. Different tRFs might act through distinct effector pathways, as evidenced by the fact that tRFs have been reported to regulate gene expression at the transcriptional, posttranscriptional, and translational level. Proposed functions for tRFs include RNA metabolism, global (Sobala and Hutvagner, 2013) and transcript-specific translation inhibition (Ivanov et al., 2011; Goodarzi et al., 2015), ribosome biogenesis (Kim et al., 2017), targeted cleavage of 3<sup>0</sup> UTRs (Elbarbary et al., 2009), regulation of apoptosis (Saikia et al., 2014), and regulation of retroviral elements (Sharma et al., 2016; Schorn et al., 2017).

Since mammalian sperm shed most of their cytoplasm, including RNAs, during development, they were long believed not to carry functional RNAs. However, studies in human sperm reported that sperm carry various RNA species (Ostermeier et al., 2002) and a subset of these RNAs (such as, sperm specific transcripts of Protamine-2 and Clusterin) are delivered to the oocyte (Ostermeier et al., 2004), suggesting their retention in internal structures of sperm head. With recent advances in RNA sequencing it is clear that sperm carry both long (>200 nts) and small RNAs (<200 nts). The long RNAs include mRNAs, long non-coding RNAs, and circular RNAs, and miRNAs, piRNAs, and fragmentation products of tRNAs and rRNAs constitute the major classes of small RNAs in mature mammalian sperm (Godia et al., 2018). A comprehensive database for sperm RNAs has been generated as SpermBase which curates both long and small RNAs in sperm of various species (Schuster et al., 2016b). By utilizing in situ hybridization, quantitative real-time PCR, microarrays, and RNA high-throughput sequencing, studies report that sperm RNAs are localized in specific compartments. Small RNAs such as miRNAs and tRFs have been sequenced in detergent-treatedsperm heads (Yan et al., 2008; Peng et al., 2012), suggesting their presence in sperm nucleus. Some miRNAs and tRFs are also present in sperm tail, and piRNAs are relatively more enriched in sperm tail compared to sperm head (Sharma et al., 2018). piRNA enrichment in tail is consistent with the localization of remnants of chromatoid body, which is involved in piRNA biogenesis, in sperm midpiece during maturation (Meikar et al., 2011; de Mateo and Sassone-Corsi, 2014). On the other hand, majority of long (>200 nt, including mRNAs) RNAs are enriched in the outer membrane of sperm and about one-third of sperm long RNAs are estimated to be present within the nuclear/peri-nuclear theca (Johnson et al., 2015). Moreover, a subset of sperm RNAs are potentially deeply embedded in the nuclear matrix of sperm nucleus, in complex with the nuclear DNA, which might be difficult to isolate during RNA extraction (Lalancette et al., 2008). Together, these studies demonstrate that a subset of sperm RNAs are embedded in the nucleus and can be delivered to the oocyte at fertilization.

The first evidence that sperm RNAs can influence offspring phenotypes comes from paramutation studies in mice, wherein, microinjection of purified sperm RNAs from mutant mice into fertilized wild type oocytes lead to transmission of mutant phenotypes to offspring (Rassoulzadegan et al., 2006; Wagner et al., 2008; Grandjean et al., 2009). While these studies focused on sperm total RNAs, more recent studies are focused on small non-coding RNAs in sperm which have the potential to regulate

gene expression in early embryos. Mature mammalian sperm have a distinct payload of small RNAs, comprising chiefly of tRFs (∼70%), followed by miRNAs as the second most abundant small RNAs and low levels of piRNAs (Peng et al., 2012; Sharma et al., 2016; Hutcheon et al., 2017). Interestingly, abundant tRFs have been reported in sperm of humans, mouse, rat, and bull (Schuster et al., 2016b), suggesting sperm-specific abundance of tRFs is a widespread phenomenon. In addition, fragmentation products of ribosomal RNAs are also present in substantial amount in mature sperm (Chu et al., 2017; Sharma et al., 2018). Some studies have also identified endo-siRNAs and mitochondrial genome enriched small RNAs (mitosRNAs) in mammalian germ cells (Song et al., 2011; Schuster et al., 2016b).

Recent studies provide strong evidence that small RNAs are the environmentally-responsive epigenetic marks in sperm (Fullston et al., 2013; Grandjean et al., 2015; Chen et al., 2016; de Castro Barbosa et al., 2016; Sharma et al., 2016; Zhang et al., 2019). Mice consuming a low protein or high-fat diet have altered levels of small RNAs in mature sperm. A low protein diet leads to upregulation of 5<sup>0</sup> fragment of multiple isoacceptors of tRNA-Glycine (tRF-GlyGCC, tRF-GlyTCC, and tRF-GlyCCC), tRNA-LysCTT and tRNA-HisGTT and down-regulation of a miRNA Let7c (Sharma et al., 2016), while a high-fat diet results in an overall increase in the levels of tRFs (∼11%) (Chen et al., 2016). In rats, high fat diet consumption also showed changes in specific small RNAs including, piRNAs (piR-025883, piR-015935, piR-036085), miRNAs (Let7c, miR293, miR880) and tRFs (GluCTC) (de Castro Barbosa et al., 2016). tRFs and Let7 miRNA thus seem to be common small RNAs modulated by diet. Mice fed a western diet (high fat and high sugar) also have altered levels of specific miRNAs in their sperm (Grandjean et al., 2015). Moreover, exposure to toxicants, such as vinclozolin and chronic ethanol exposure, affected tRF and miRNA levels (Schuster et al., 2016a; Rompala et al., 2018). Similarly, exposure to psychological stress, such as early life trauma or chronic stress, lead to changes in levels of specific miRNAs (Gapp et al., 2014; Rodgers et al., 2015).

The implication of sperm small RNAs in intergenerational inheritance of paternal environmental conditions was demonstrated by studies using direct microinjections of small RNAs purified from the sperm of exposed males into control zygotes. In one of the first such studies, mice were exposed to maternal separation coupled with unpredictable maternal stress (MSUS) to induce early life stress. The MSUS mice showed depressive-life behavior, and such behavior was transmitted to their offspring. Injection of RNAs purified from sperm of MSUS mice into one-cell control embryos lead to the transmission of a subset of the behavioral and metabolic trait of MSUS mice to their offspring. This study provided the first direct evidence that paternal environmental information is transmitted by sperm RNAs, and identified a specific set of miRNAs that change in MSUS sperm in response to stress (Gapp et al., 2014). Another study examined intergenerational inheritance of paternal chronic stress and identified nine specific miRNAs that were significantly up-regulated in sperm of exposed mice (Rodgers et al., 2013). Injection of these nine miRNAs (in combination) in control zygotes lead to the transmission of paternal behavioral traits to the offspring (Rodgers et al., 2015). Similarly, in the case of high-fat diet paradigm, microinjection of purified tRFs (30–40 nt size RNAs) into control zygotes reproduced metabolic phenotypes in offspring (as seen by natural mating), suggesting that tRFs potentially act as epigenetic information carriers of paternal dietary information (Chen et al., 2016).

Overall, multiple recent studies suggest that various environmental exposures direct distinct small RNA changes in sperm —diet mainly affects tRFs, while psychological stress influences miRNAs. Whether such differences arise from variation in the methods used for sperm small RNA profiling by different groups or distinct environmental insults are signaled to different classes of small RNAs is not clear. Irrespective, these studies provide strong evidence that small RNAs are environmentally-responsive epigenetic molecules in sperm. Moreover, a recent study reported that both small and long RNAs influence paternal exposure phenotypes of early life trauma in offspring, with long and small RNAs regulating different behavioral traits (Gapp et al., 2018). Although the identity of the specific long and small RNAs involved in such inheritance and their mechanism of regulating offspring phenotype remains unknown, these studies suggest that both small and long RNAs potentially play a role in transmitting complex paternal phenotypes to offspring.

Whether one epigenetic mark is more potent at responding to a certain environmental insult is still not clear. Depending on the type of environmental exposure and the developmental timing of exposure, one epigenetic mark might be more responsive compared to the other. For instance, sperm of mice fed a low protein diet from birth to weaning displayed changes in DNA methylation at ribosomal DNA (rDNA) (Holland et al., 2016), however, if mice are challenged with a low protein diet starting at weaning, levels of specific small RNAs change while rDNA methylation remains unaltered (Shea et al., 2015; Sharma et al., 2016). Although sperm epigenome could be modulated by the environment throughout the life of an organism, it is likely more vulnerable during early development, including embryogenesis and primordial germ cell (PGC) development, when rapid cell division and global reprograming take place. Finally, there could be a cross-talk between different epigenetic marks to relay paternal environmental information to offspring.

#### ROLE OF SOMA-GERMLINE COMMUNICATION IN SHAPING SPERM RNA PAYLOAD

The ability of environmental conditions, such as diet, to influence phenotypes in future generations requires that environmental exposure induces changes in the epigenome of gametes. The nature and mechanism of shaping and altering the epigenome of mammalian gametes are not well-understood. One possibility is that somatic cells communicate information to the gametes and, thus, to progeny. Although the Weismann barrier has long been thought to be a problem for such communication (Liu, 2011), over the past decade, an increasing number of studies have suggested soma-to-germline shipment of RNAs

in various model organisms (Bourc'his and Voinnet, 2010). Studies in model organisms (e.g., ciliates, flies, and plants), suggest that during gametogenesis RNAs produced in somatic support cells are transferred into developing germ cells where they regulate genomic integrity (Bourc'his and Voinnet, 2010; Martinez et al., 2016). In mammals, there is no direct contact between the somatic support cells and mature gametes, as found in some of the organisms mentioned above —germline and vegetative nuclei are present in the same cytoplasm in ciliates and plants— making soma-germline communication challenging in mammals. Nonetheless, recent studies provide evidence of RNAmediated soma-germline communication in mammals (Sharma et al., 2016, 2018; Morgan et al., 2019; Trigg et al., 2019).

Post-testicular sperm maturation takes place in the epididymis, a long, convoluted tubular organ where sperm enter after exiting the testis. Intriguingly, recent studies report that sperm small RNA payload undergoes dramatic changes during epididymal transit. Deep sequencing of small RNAs from sperm at various stages of development revealed that while testicular sperm populations such as spermatocytes, round spermatids, and mature testicular spermatozoa are highly enriched in piRNAs (Kuramochi-Miyagawa et al., 2004; Aravin et al., 2006, 2007; Li et al., 2013; Sharma et al., 2018), mature sperm from the distal cauda epididymis are chiefly comprised of tRFs (Peng et al., 2012; Sharma et al., 2016, 2018). Moreover, even the proximal caput epididymis sperm are highly abundant in tRFs (Sharma et al., 2016), suggesting that tRFs are specifically gained upon entry in the epididymis. Interestingly, miRNAs also undergo dynamic changes during epididymal transit, wherein miRNAs such as miRNA 17–92 cluster, are present in testicular spermatozoa, lost in proximal caput epididymis sperm and then reappear in mature cauda sperm (Nixon et al., 2015; Sharma et al., 2018). These studies provide evidence of a novel RNA reprograming event during post-testicular maturation in the epididymis, and raise the question of how small RNAs are gained/lost during epididymal transit.

During the process of spermatogenesis, spermatozoa shed most of their cytoplasm as residual body (Sprando and Russell, 1987) and in the cytoplasmic droplet (Bloom and Nicander, 1961), a subset of small RNAs are potentially lost through this mechanism (Sharma et al., 2018). More puzzling is the gain of small RNAs in transcriptionally silent mature sperm, and suggest that small RNA increase during epididymal maturation is likely not mediated by intrinsic pathways. Indeed, recent studies report astonishing observations that small RNAs in mature sperm are shipped from surrounding somatic epididymis epithelial cells (Sharma et al., 2016, 2018). Epididymis epithelium is highly abundant in sperm-specific small RNAs (tRFs and miRNAs) (Sharma et al., 2016), and by spatiotemporal metabolic labeling of RNAs (Gay et al., 2013), it was demonstrated that mature sperm carry RNAs initially synthesized in the somatic epididymis tissue (Sharma et al., 2018). Together, these studies provide the first evidence of RNA-mediated soma-germline communication in mammals and suggest that post-testicular maturation in the epididymis shapes the small RNA payload of mature mammalian sperm. Intriguingly, sperm DNA methylation is also modulated during epididymal maturation —specific genes were unmethylated in the mature testicular spermatozoa, but remethylated during epididymal maturation (Ariel et al., 1994). Whether such changes in DNA methylation occur genome-wide and have functional consequences during early development remains unknown. In addition to epididymis, other somatic accessory cells of the male reproductive tract, such as, Leydig cells and Sertoli cells, which are in close proximity to the developing germ cells and serve important roles in spermatogenesis (Sharpe et al., 1990; Rebourcet et al., 2014), may also play a role in shaping the epigenome of developing germ cells.

What is the mechanism of this soma-germline communication? Sperm acquire a multitude of lipids and proteins during epididymal transit (Gervasi and Visconti, 2017). A subset of these proteins and lipids are delivered from epididymis to sperm via extracellular vesicles (EVs) secreted from epididymis epithelium, known as epididymosomes (Rejraji et al., 2006; Sullivan et al., 2007; Koch et al., 2015; Sullivan, 2015, 2016). Recent studies revealed that epididymosomes are highly abundant in small RNAs and have a similar RNA payload to that of mature sperm. Moreover, in vitro reconstitution studies revealed that epididymosomes can deliver small RNAs to relatively "immature" testicular spermatozoa (Reilly et al., 2016; Sharma et al., 2016), suggesting that epididymosome-mediated delivery of small RNAs is one mechanism of RNA-mediated communication between somatic epididymis cells and sperm.

Deep sequencing of small RNA profiles of epididymis and epididymosomes revealed that only a subset of small RNAs present in the epididymal epithelium are present in epididymosomes (Belleannee et al., 2013; Reilly et al., 2016; Schuster et al., 2016b), suggesting that instead of being passively released by epididymal epithelial cells, subpopulations of epididymal small RNAs are selectively packaged into epididymosomes. The mechanistic basis of selective sorting of small RNAs into epididymosomes remains unknown. Mechanistic studies on RNA sorting into EVs secreted from activated T-lymphocytes revealed that these EVs have a specific miRNA profile which differs from parent cells, with sumoylated hnRNP2AB1 protein being responsible for sorting miRNAs containing a specific tetranucleotide motif into EVs (Villarroya-Beltri et al., 2013). Potentially, similar mechanisms involving specific sequence motifs and/or RNA binding proteins facilitate selective sorting of small RNAs into epididymosomes. Moreover, recent studies in exosomes (a class of 100–150 nm EVs) secreted from HEK293T cells revealed an exosome-specific posttranscriptional modification in tRNAs (Shurtleff et al., 2017), suggesting an additional mechanism of selective sorting of RNAs into epididymosomes. Importantly, recent studies provide insights in the process of delivery of epididymosomal protein cargo to sperm (Zhou et al., 2019). Sperm plasma membrane has unusually high abundance of polyunsaturated phospholipids which compartmentalize its proteins and lipids into specific domains known as lipid rafts (Kawano et al., 2011). Studies suggest a role of these lipid rafts in coordination of initial epididymosome-sperm interaction (Girouard et al., 2009). Since epididymosomal membrane-bound proteins are not detected on sperm (Gatti et al., 2005), it is speculated that epididymosomes adhere transiently to sperm by tethering to receptors at the

post-acrosomal domain of the sperm head, which is followed by formation of a fusion pore (Nixon et al., 2019; Trigg et al., 2019; Zhou et al., 2019).

Intriguingly, epididymosomes are a heterogeneous mix of EVs that can be classified into subcategories based on their size and biogenesis (Sullivan, 2015). The various subpopulations of epididymosomes potentially target different cell types. For instance, a subset of epididymosomes may communicate with maturing sperm (Reilly et al., 2016; Sharma et al., 2016, 2018), another subset between different regions of the epididymis (e.g., lumicrine signaling) (Belleannee et al., 2013), and a third subset between the epididymis and the oviduct following delivery to the female reproductive tract via the seminal fluid. Moreover, epididymosome-mediated small RNA delivery to sperm could potentially be regulated at various levels, for example, by regulating the ability of different EVs to interact with sperm and, possibly, by regulating the subset of sperm that receive cargo from specific classes of EVs.

Other mechanisms of RNA-mediated soma-germline communication potentially involve direct shipment of RNAs to sperm in a complex with RNA binding proteins (RBPs), as has been seen in other mammalian cells (Wang et al., 2010; Arroyo et al., 2011; Sarkies and Miska, 2014). Finally, since ejaculated sperm still has a substantial distance to travel before fertilizing the oocyte in the female reproductive tract, EVs secreted from prostate gland (prostasomes; Sullivan and Saez, 2013) or the female reproductive tract (Al-Dossary et al., 2013), potentially further alter the small RNA payload of fertilization-competent sperm.

The functional significance of epididymis specific small RNA remodeling during post-testicular maturation is not wellunderstood. Recent studies report that small RNA shipment from the epididymis to sperm is essential for embryo implantation (Conine et al., 2018). Embryos generated using intracytoplasmic sperm injections (ICSIs) of immature sperm from the proximal caput epididymis exhibited misregulation of multiple regulatory genes throughout preimplantation development and eventually failed soon after implantation (Conine et al., 2018). Remarkably, microinjections of small RNAs (18–40 nt) purified from cauda epididymosomes completely rescued the post-implantation embryonic lethality phenotype, suggesting an essential role of epididymosomal RNAs in early embryonic development (Conine et al., 2018). However, these studies contradict prior studies in mice reporting successful generation of offspring from ICSI using caput sperm (Suganuma et al., 2005). As discussed in Conine et al. (2018), these discrepancies could arise from differences in the mouse strain background, age of the male, the timing of embryo implantation and the procedure used for the preparation of sperm heads for ICSI. The latter is of particular importance, since prior studies have documented that epididymosomes interact with sperm in a site specific manner (Griffiths et al., 2008; Zhou et al., 2019), and therefore, inclusion/exclusion of certain parts of the sperm could result in different RNA payload being delivered to the oocyte. Future studies, using mice defective in one or more pathways of epididymosome biogenesis and delivery, will shed more light on the functional consequences of shipment of RNA from the epididymis to sperm and its influence on offspring phenotypes.

## ENVIRONMENTAL SIGNALING TO SPERM RNAs

An understanding of how environmental conditions influence epigenetic marks in the gametes is essential to elucidate how paternal environment affects phenotypes in offspring. As discussed above, numerous studies provide evidence that paternal environmental conditions, such as altered diet or stress, influence small RNA levels in sperm, however, the mechanistic understanding of how any of the environmental exposures noted above cause specific changes in sperm small RNA levels is largely lacking. Broadly, paternal environment can affect sperm small RNA levels by (1) regulating the transcription of small RNA precursors, (2) regulating small RNA processing, (3) modulating small RNA decay/stability, (4) influencing sorting of small RNAs into epididymosomes, and (5) affecting delivery of epididymosomal cargo to sperm.

RNA post-transcriptional modifications provide the most compelling potential mechanism of environmental signaling to small RNA levels. RNA modifications can affect RNA basepairing, the secondary structure of the RNA, and RNA–protein interactions and, thus, add a new layer of transcriptional regulation (Roundtree et al., 2017). Indeed, sperm small RNAs from mice consuming a high-fat diet showed increased levels of 5-methylcytosine (m5C) and N-2-methylguanosine (m2G) (Chen et al., 2016) and mice exposed to ethanol were reported to have higher levels of 5<sup>0</sup> -methylaminomethyl-2-thiouridine (mnm<sup>5</sup> s <sup>2</sup>U) and formylcytidine (f5C) (Rompala et al., 2018). Not only do these modifications provide stability to tRFs (Chen et al., 2016), they potentially play a role in intergenerational inheritance of paternal environmental conditions —deletion of a tRNA methyltransferase, Dnmt2, which adds m5C to specific tRNAs, prevents transmission of paternal high-fat phenotype to offspring (Zhang et al., 2018). The modified tRFs potentially remain stable in the oocyte at the time of fertilization and thus allow transmission of paternal epigenetic information. However, in the studies noted above, 30–40 nt small RNA fraction (comprising chiefly of tRFs) were examined to identify modified nucleotides by liquid chromatography with tandem mass spectrometry (LC-MS/MS). It is crucial to identify the precise small RNAs that are modified and the specific location of the modified residue, to better understand the mechanistic basis and functional significance of altered post-transcriptional modifications in response to various environmental stimuli. A combination of mass spectrometry, RNA-sequencing, and affinity pull-down approaches will help resolve these issues. How do changes in environmental conditions lead to changes in tRF modifications? As the levels of tRNA modifications have been shown to respond to nutrient availability (Laxman et al., 2013), dietary changes could affect the levels of critical metabolites, such as S-adenosyl methionine (SAM), resulting in global changes in levels of specific tRNA modifications and alterations in tRNA processing and/or stability (Schaefer et al., 2010;

Tuorto et al., 2012; Sharma and Rando, 2017). However, how non-metabolism related environmental exposures, such as ethanol, affect tRF modifications remains mysterious.

Communication between somatic cells and germ cells is not only crucial for the maintenance of germ cell integrity, as suggested in model organisms (Bourc'his and Voinnet, 2010; Martinez et al., 2016), it also provides a potential mechanism for transmitting environmental information to gametes and thereby influencing phenotypes in future generations (**Figure 1**). For instance, epididymis may serve as an environment sensing organ and transmit this information (in the form of small RNAs) to the maturing sperm, and hence, to offspring. Consistent with such possibility, paternal low protein consumption affects small RNA levels throughout the male reproductive tract, including the epididymis (Sharma et al., 2016). Moreover, paternal exposure to ethanol resulted in similar changes in tRF levels in epididymosomes and sperm, suggesting that paternal environmental information is signaled from the epididymis to sperm via epididymosomes (Rompala et al., 2018). Mechanistically, the authors reported reduced expression of Nsun2 —a 5<sup>0</sup> cytosine methyltransferase known to prevent tRF biogenesis by inhibiting cleavage of specific tRNAs (Tuorto et al., 2012)— in the epididymis of mice exposed to ethanol (Rompala et al., 2018), suggesting that reduced expression of Nsun2 in the epididymis leads to increased levels of tRFs in epididymosomes, which in turn affects levels of tRFs in sperm. Overall, these studies suggest that epididymis-mediated remodeling of sperm small RNA payload is modulated by the environment, potentially leading to the transmission of environmental information to the next generation (**Figure 1**).

#### FUNCTIONAL CONSEQUENCES OF SPERM SMALL RNA PAYLOAD

How do changes in sperm small RNAs in response to various environmental conditions result in the transmission of paternally acquired phenotypes in offspring? Mature sperm deliver their RNAs to the oocyte at the time of fertilization, and hence, potentially regulate early embryonic gene expression to affect offspring phenotypes. Indeed, recent studies provide evidence that sperm small RNAs play important, and in some cases, essential role during early development. Microinjection of a synthetic 5<sup>0</sup> fragment of tRNA-GlyGCC (tRFGlyGCC) or an inhibitor of tRFGlyGCC −among the most abundant small RNAs in sperm, which is up-regulated in low protein sperm− in control zygotes revealed that tRFGlyGCC represses transcription of a specific set of MERVL retroviral element-driven genes in early embryos (Sharma et al., 2016). While tRFGlyGCC was shown to regulate MERVL expression at the transcriptional level (Sharma et al., 2016), the mechanistic basis of tRF-mediated MERVL-driven gene regulation remains unclear. Intriguingly, genes regulated by tRFGlyGCC were previously shown to be activated during zygotic genome activation in two cell embryos and associated with totipotency program of early embryos (Macfarlan et al., 2012). These findings suggest that tRFmediated gene expression changes in early embryos potentially lead to altered placental function, which in turn can lead to metabolic phenotypes in offspring as a secondary effect of altered preimplantation development (Sharma et al., 2016). Additionally, 3 0 tRFs have been shown to inhibit LTR-retrotransposons in preimplantation trophoblast stem cells (Schorn et al., 2017). Northern blot analysis revealed abundant 3<sup>0</sup> tRFs (difficult to sequence with standard small RNA sequencing protocols) in the epididymis and mature sperm (Sharma et al., 2018; Zhang et al., 2018), suggesting that sperm 3<sup>0</sup> tRFs potentially play roles in regulating LTR-retrotransposons in preimplantation embryos.

While these studies examined the functions of specific tRFs, another group examined the broad functions of 30–40 nts sperm small RNAs (comprising chiefly of tRFs). Microinjection of 30–40 nts small RNAs purified from sperm of mice fed high-fat diet in control zygotes resulted in altered expression of multiple genes involved in metabolic regulation in eightcell embryos and blastocysts, suggesting that a cascade of gene expression changes in preimplantation embryos potentially leads to metabolic phenotypes in adult pancreatic islet cells (Chen et al., 2016). Importantly, embryos generated using mice harboring deletion of Dicer, an endonuclease required for miRNAs and endo-siRNAs biogenesis (Ha and Kim, 2014), exhibit reduced developmental potential and altered expression of genes involved in zygotic genome activation (Yuan et al., 2016), further demonstrating that sperm delivered small RNAs play important roles during early development. Overall, these studies point to a role of sperm small RNAs in regulating some crucial steps of early developmental (zygotic genome activation, for instance) and the altered phenotypes in offspring being a secondary downstream effect.

Although, sperm small RNAs are a strong candidate for intergenerational inheritance of paternal environmental effects, many mechanistic questions remain unanswered regarding the delivery and functional consequences of sperm RNAs during early embryonic development. First, the RNAs that change in response to environmental conditions are not particularly abundant in sperm. For instance, the most abundant tRFs in sperm are estimated to be at ∼100–100,000 molecules per sperm (Sharma et al., 2016). Second, the amount of RNA microinjected in the above-mentioned studies is larger than the amount of RNA a single sperm could deliver, and sperm deliver very low amount of RNA at fertilization compared to the amount of RNA in an oocyte. That said, the microinjection studies, along with studies using genetic manipulation of small RNA biogenesis pathways, provide evidence that altering RNA levels or integrity has profound effects on early embryonic development and health of offspring. Moreover, although sperm small RNAs are present in lower abundance, they likely possess some unique features that make them highly efficient in regulating gene expression —potentially due to post-transcriptional modifications (as discussed above) or preloading in some RNA–protein effector complexes. Furthermore, epididymosomes adhering to the surface of the sperm might play important regulatory roles in delivering additional RNAs to the female reproductive tract or the oocyte (Sharma and Rando, 2017; Conine et al., 2018). Three, since mammals do not express a known RNA-dependent RNA polymerase —found in various

sperm. (B) Sperm small RNAs are delivered to the oocyte at fertilization where they potentially affect early embryonic gene expression, resulting in altered early development and health of adult offspring.

model organisms such as plants and worms, where it is involved in amplifying small RNA signals— it is not clear how sperm small RNAs delivered at fertilization will have prolonged effects on phenotypes in adults. Sperm small RNAs likely regulate some crucial developmental events during the first few cell-divisions, which in turn lead to long-lasting effects on offspring phenotypes. Finally, sperm carry multiple additional non-coding RNAs, such as rRNA fragments, piRNAs, snoRNAs, and long non-coding RNAs, future studies should focus on understanding the roles, if any, of these RNAs in shaping offspring phenotypes.

## CONCLUSION AND FUTURE PERSPECTIVES

There is a growing body of evidence, from worms to humans, suggesting that parental environment can influence phenotypes in offspring. However, the mechanistic basis of such inheritance is only starting to be understood. Recent advances in our understanding of intergenerational inheritance revealed that paternal environmental information is transmitted to offspring via sperm and that small RNAs are environmentally-responsive

epigenetic molecules in sperm. Intriguingly, epididymismediated remodeling of sperm small RNA payload may have consequences for early embryonic development and offspring health. As this remodeling can be modulated by paternal environment, it provides a potential mechanism of transmission of environmental information to the next generation. Future studies will focus on elucidating the mechanism of RNAmediated soma-germline communication and its consequences on offspring health.

These studies have important implications for human reproductive health. For assisted reproduction in humans, in some cases spermatozoa from testicular biopsies are used for fertilizing oocytes. Therefore, elucidating the functional consequences of sperm small RNA remodeling during post-testicular maturation in the epididymis has important implications for assisted reproduction in humans. Moreover, the seminal fluid also contains a high abundance of exosomes enriched in non-coding RNAs (Vojtech et al., 2014) and EVs are found in most of the bodily fluids such as serum and cerebrospinal fluid (Raposo and Stoorvogel, 2013; Colombo et al., 2014). These observations suggest the intriguing possibility that small RNA loaded EVs might play a crucial role in long-range communication at an organismal level and transmit epigenetic information from distant organs to the developing germ cells, and, hence, to the offspring.

Many important questions remain unanswered regarding the role of sperm small RNAs in intergenerational inheritance. For example, how many pieces of information is packaged in sperm — do sperm carry information about the general quality of life or transmit more specific information about various environmental exposures? How does the environment influence the small RNA payload in sperm? How do changes in sperm small

#### REFERENCES


RNA levels affect zygote to give rise to altered phenotypes in offspring? Since different paternal exposures, such a high-fat diet or early life trauma, result in a common metabolic phenotypes (for example, glucose intolerance), are some common pathways targeted by distinct classes of small RNAs? Finally, what is the mechanism of transgenerational inheritance? Since mammals lack a known RNA-dependent RNA polymerase, sperm small RNAs potentially influence early embryonic chromatin and/or DNA methylation states to transmit epigenetic information to the future generations. With recent advances in the field of singlecell omics, a more in-depth examination of sperm and embryonic epigenome will help shed light on these and other outstanding questions, and elucidate the mechanism of transgenerational inheritance of acquired traits.

#### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

#### FUNDING

US was supported by the start-up funds from the University of California, Santa Cruz.

#### ACKNOWLEDGMENTS

I would like to thank the members of my lab for their helpful discussions on this topic.





**Conflict of Interest:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Sharma. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Testicular Structure and Spermatogenesis in the Naked Mole-Rat Is Unique (Degenerate) and Atypical Compared to Other Mammals

#### Gerhard van der Horst<sup>1</sup> \*, Sanet H. Kotzé<sup>2</sup> , M. Justin O'Riain<sup>3</sup> and Liana Maree<sup>1</sup>

<sup>1</sup> Department of Medical Biosciences, University of the Western Cape, Bellville, South Africa, <sup>2</sup> Division of Clinical Anatomy, Department of Biomedical Sciences, Stellenbosch University, Stellenbosch, South Africa, <sup>3</sup> Institute for Communities and Wildlife in Africa, Department of Biological Sciences, University of Cape Town, Cape Town, South Africa

#### Edited by:

Zhibing Zhang, Virginia Commonwealth University, United States

#### Reviewed by:

William Godfrey Breed, The University of Adelaide, Australia Sigrid Hoyer-Fender, University of Göttingen, Germany

> \*Correspondence: Gerhard van der Horst gvdhorst7@gmail.com

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 07 June 2019 Accepted: 30 September 2019 Published: 16 October 2019

#### Citation:

van der Horst G, Kotzé SH, O'Riain MJ and Maree L (2019) Testicular Structure and Spermatogenesis in the Naked Mole-Rat Is Unique (Degenerate) and Atypical Compared to Other Mammals. Front. Cell Dev. Biol. 7:234. doi: 10.3389/fcell.2019.00234 The naked mole-rat (NMR) queen controls reproduction in her eusocial colony by usually selecting one male for reproduction and suppressing gametogenesis in all other males and females. Simplified, polymorphic and slow-swimming spermatozoa in the NMR seem to have been shaped by a low risk of sperm competition. We hypothesize that this unique mammalian social organization has had a dramatic influence on testicular features and spermatogenesis in the NMR. The testicular structure as well as spermatogenic cell types and its organization in breeding, subordinate and disperser males were studied using light microscopy and transmission electron microscopy. Even though the basic testicular design in NMRs is similar to most Afrotheria as well as some rodents with intra-abdominal testes, the Sertoli and spermatogenic cells have many atypical mammalian features. Seminiferous tubules are distended and contain large volumes of fluid while interstitial tissue cover about 50% of the testicular surface area and is mainly composed of Leydig cells. The Sertoli cell cytoplasm contains an extensive network of membranes and a variety of fluid-containing vesicles. Furthermore, Sertoli cells form numerous phagosomes that often appear as extensive accumulations of myelin. Another unusual feature of mature NMR Sertoli cells is mitotic division. While the main types of spermatogonia and spermatocytes are clearly identifiable, these cells are poorly organized and many spermatids without typical intercellular bridges are present. Spermatid heads appear to be malformed with disorganized chromatin, nuclear fragmentation and an ill-defined acrosome formed from star-like Golgi bodies. Rudimentary manchette development corresponds with the occurrence of abnormal sperm morphology. We also hypothesize that NMR testicular organization and spermiation are modified to produce spermatozoa on demand in a short period of time and subsequently use a Sertoli cell "pump" to flush the spermatozoa into short tubuli recti and simplified rete testis. Despite the difficulty in finding cellular associations

**89**

during spermatogenesis, six spermatogenic stages could be described in the NMR. These numerous atypical and often simplified features of the NMR further supports the notion of degenerative orthogenesis that was selected for due to the absence of sperm competition.

Keywords: spermatogenesis, spermiogenesis, spermiation, staging, Leydig cells, Sertoli cells

#### INTRODUCTION

Naked mole-rats (NMRs, Heterocephalus glaber) represent one of only two eusocial mammals (Jarvis, 1981, 1985; Faulkes et al., 1991; O'Riain et al., 2000) with reproduction typically restricted to a single female (queen) and 1–3 males within large colonies varying from 40 to 90 individuals. Multiple-paternity has been recorded for this species (Faulkes et al., 1997), but usually the queen selects a male for life and most other males and all other females (subordinates) are reproductively suppressed (Faulkes and Abbott, 1991; Faulkes et al., 1991). This restriction of breeding to a small subset of the population presents a low risk for sperm competition and it is assumed to have shaped the sperm structure (simple and low percentage of normal forms) and motility (slow) in this species (van der Horst et al., 2011). It is not evident at what point during spermatogenesis a path is followed which produces the remarkable heterogeneity of morphologically abnormal spermatozoa in the NMR.

Spermatogenesis is a complex but highly coordinated process which is regulated by paracrine, autocrine, juxtacrine, and endocrine pathways (Chocu et al., 2012). Mammalian spermatogenesis is divided into three phases: mitotic divisions of the spermatogonia and differentiation into spermatocytes (spermatocytogenesis); two consecutive meiotic divisions of the spermatocytes to produce spermatids; and differentiation of the spermatids into spermatozoa (spermiogenesis). During spermiogenesis, structural changes occurring in several cellular compartments include repackaging of the chromatin for transport, development of the acrosome, elongating of the tail, formation of the mitochondrial sheath in the midpiece and reduction of cytoplasmic volume. Here after spermatozoa are released into the seminiferous tubule lumen during the process of spermiation. Seasonality, reproductive lifespan, mode of fertilization and sperm competition are all factors that could influence the organization of spermatogenesis within the testis (Ramm et al., 2014).

Although several papers have appeared on spermatogenesis in the NMR since the early 1990s (Faulkes et al., 1991; Yang et al., 2017), these papers covered a general description of spermatogenesis using wax embedding and hematoxylin-eosin stained sections for brightfield microscopy. The emphasis of these studies was to discern the incidence of spermatogenesis, particularly in relation to the presence/absence of spermatozoa. Their general consensus was that spermatogenesis occurs in breeders and non-breeders, particularly if they were kept separate from the queen, providing evidence of her suppressive role in relation to spermatogenesis specifically.

The first study to pursue greater detail of these processes using transmission electron microscopy was by Onyango et al. (1993) focusing exclusively on non-breeders. Despite this shortcoming, they described most of the cell types of spermatogenesis, albeit without any cellular associations, and related their findings to other mammals. While only few late spermatids or spermatozoa were observed in their investigation, van der Horst et al. (2011) and van der Horst and Maree (2014) described the vastly abnormal spermatozoa in NMR breeders, subordinates and dispersers using both stained sperm smears and electron microscopy. The latter two studies concluded that NMR spermatozoa represent a case of degenerative orthogenesis and it was proposed that this was selected for in the absence of sperm competition.

The aim of the current study is to quantify the testicular structure in the NMR, with special reference to spermatogenesis. We hypothesize that due to the lack of sperm competition there will be degenerate testicular features and "relaxed" spermatogenesis. We investigated testicular ultrastructure and spermatogenesis among breeders, subordinates and dispersers and compared our results to other mammals.

#### MATERIALS AND METHODS

#### Animal Husbandry

The study population originated from various localities in Kenya, and was comprised of 10 colonies housed in separate artificial burrow systems at the University of Cape Town, South Africa. Husbandry details have been described previously by Jarvis (1991). A total of 26 male NMRs (Heterocephalus glaber) were used in this study, including 10 breeders and 10 subordinates from 10 colonies and 6 males that had been isolated from their colonies for a minimum of 2 months. Two of these isolated males had acquired the dispersing male phenotypes (O'Riain et al., 1996). Breeders were adult males confirmed to be in consort with the queen (mutual naso-anal grooming and copulation). Subordinate males were adult males who did not engage in either naso-anal grooming or copulations with their queens. It should be noted that the "subordinate" NMR males referred to in this study is of the same reproductive status as that of "non-breeding" males used in previous NMR studies.

Ethical clearance for the study was obtained from both the University of Cape Town (2005/V7/JOR), the University of the Western Cape (ScR1RC2007/3/30) and Stellenbosch University (P07/09/019). Animals were removed from their custom made burrow systems and anesthetized with halothane by putting a mask over the head. Surgical anesthesia was attained within 5 min and longer exposure to halothane was used to euthanize animals. The entire reproductive system was dissected out and put into

Ham's F10 medium (Invitrogen, Cape Town, South Africa) at 28◦C (NMR body temperature) (van der Horst et al., 2011).

Immediately upon removal of the reproductive systems, small sections of representative parts of the testes were fixed in 2.5% phosphate buffered glutaraldehyde (4◦C) and were subsequently routinely processed for transmission electron microscopy (TEM). During TEM processing, 1 µm epoxy sections were cut for viewing and final trimming using an ultra-microtome. The 1 µm epoxy sections have been used to great effect in this investigation to provide high quality brightfield light microscopy images at low magnifications for measurement of different parts of the testes (see later). The ultrathin sections have been used for TEM.

#### Capturing of Images for Light Microscopy, Measurement, and TEM

For brightfield microscopy, a Nikon E50i microscope was used and images were captured using an Ace ACA 1300-200uc Basler camera and the Morphology Module of the Sperm Class Analyzer (SCA, version 6.4.0.64) computer-aided sperm analysis software (Microptic S.L., Barcelona, Spain). The sections investigated for this purpose represented 1 µm epoxy sections stained with toluidine blue. The measurement tool of the SCA system was used to measure the seminiferous tubule diameters and to calculate surface areas for both the interstitial tissue and the seminiferous tubules. For TEM, a Jeol JEM 1011 transmission electron microscope (Advanced Laboratory Solutions, Johannesburg, South Africa) at 80 kV was used to capture micrographs of different testicular cells and compartments for subsequent description. More than 1000 micrographs from all 26 males were studied for description of spermatogenic cell types, cellular associations, and staging (see below).

#### Statistical Analysis

MedCalc, Version 14 (Mariakerke, Belgium) was used for statistical analyses. Descriptive statistics was used for calculation of averages and standard deviations (SD). Comparisons of testicular morphometry parameters were performed using either ANOVA or Kruskal–Wallis among the different groups and P < 0.05 was considered significant.

#### RESULTS

#### Testicular Compartments

The interstitial component of the testis in the NMR is large and composed mainly of Leydig cells, blood vessels and lymphatics, representing about 50% of the testicular surface area (**Figure 1A**). Analyses of breeders, subordinates and dispersers of the same body weight and stage of testicular development show that seminiferous tubules occupy about 45 to 55% of the testicular surface area. In NMRs the average diameter of the seminiferous tubules varies from 152 to 240 µm and there are no significant differences (P = 0.18) among the three male groups (**Table 1**).

The Leydig cells are arranged in groups (islands) of three or more cells per island and each island is contained within a layer(s) of loose connective tissue (**Figures 1B,C**). The Leydig cells have a spherical or slightly oval nucleus and contain

FIGURE 1 | Testicular compartments of the naked mole-rat (NMR). (A) Seminiferous tubules (ST) and large interstitial compartment (IC) of the testis. (B) Detail of Leydig cells (arrows) arranged in groups of three to four in islands. Lipid droplets (asterisk) are contrasted by osmium tetroxide, appearing black. (C) Interstitial compartment containing numerous Leydig cell islands (circle) with lipid content possibly washed out of vesicles during processing. (D) Wall of the seminiferous tubules, the lamina propria, consisting of three layers, namely a myoid epithelium (MC), a lymphoid epithelium (LE) and a basal lamina (BL) and arrow indicating lymphatic space. (E) Oblique section of arteriole in close association with Leydig cell islands. (F) Enhanced view of arteriole in longitudinal section showing endothelial cells (arrow) and erythrocytes (ER).



At least 10 seminiferous tubules were measured per group (100 seminiferous tubules in total).

mainly euchromatin (genetically active). The cytoplasm of the Leydig cell contains numerous mitochondria as well as vesicles with apparently two types of lipid. One type of vesicle appears transluscent/light or containing no lipid or lipid in a particular form. In the second type of vesicle, the lipid appears dark/blue at light microsope level and as electron dense (black) in TEM. These lipid vesicles originate from an extensive network of smooth endoplasmic reticulum. The lymphatic component is clearly defined as sinusoids containing lymph (**Figure 1D**, black arrow). Arterioles are evident (**Figure 1E**) in the interstitial compartment with three to four endothelial cells observed in transverse sections. These blood vessels appear in close association with several islands of interstitial tissue (**Figure 1E**). **Figure 1F** provides an enlarged view of the endothelial wall of an arteriole.

The wall (lamina propria) of the seminiferous tubules consists of three typical layers, namely an inside basal lamina, a myoid epithelial layer and an outer lymphatic endothelium (**Figure 1D**). A detailed description of all the cells contained within the seminiferous tubules (spermatogenic and nonspermatogenic), as well as their cellular associations and accordingly staging of spermatogenesis follows below. These cellular features and organization were noticed in males from all three reproductive groups.

#### Cell Types of the Seminiferous Epithelium

Sertoli cells in the NMR contain many structural differences compared to other mammals. Their nuclei are usually irregular to ovoid with a prominent nucleolus surrounded by a ring of peri-nucleolar chromatin. In almost all testicular sections of most NMR males, we observed Sertoli cells in division and these cells have an extensive cytoplasm filled with numerous membranelimited vesicles apparently containing fluid (**Figures 2A,B**). Smaller vesicles seem to coalesce to form large fluid-filled vesicles that might be important in the release of spermatids (see below). The Sertoli cell cytoplasm also often contains many other peculiarities including phagosomes that are described in a separate section below.

Dark and light, large A type spermatogonia as well as B type spermatogonia with prominent nucleoli could be identified (**Figure 2C**). All primary spermatocyte types (preleptotene, leptotene, zygotene and pachytene) are present in the NMR testis (**Figures 2D–F**). Dividing primary spermatocytes are clearly designated and some contain intercellular bridges (**Figure 3A**). However, subsequent divisions of these cells do

FIGURE 2 | (A) Sertoli cell with vast number of fluid-filled vesicles in its cytoplasm. Nucleolus of Sertoli cell is surrounded by peri-nucleolar chromatin (arrow). (B) Dividing Sertoli cells (arrow) between two spermatogonia. Similar dividing Sertoli cells are common in all males studied. (C) A type spermatogonium (A Sg) and B type spermatogonium (B Sg). (D)Leptotene primary spermatocyte. (E) Early pachytene primary spematocyte. (F) Mid-pachytene primary spermatocyte showing paired chromosomes.

FIGURE 3 | (A) Pre-leptotene primary spermatocyte showing inter-cellular bridge (arrow). (B) Early round spermatids showing some nuclear condensation just prior to the cap phase of acrosome formation. (C) Star-like Golgi figures (arrow) as part of acrosome formation. Location of Golgi apparatus in Sertoli cell indicated with asterisk in Figure 4D. (D) Cap phase of spermiogenesis showing acrosome formation (AC) and associated endoplasmic reticulum (ER) as well as the peri-nuclear ring (PNR). (E) Start of manchette (M) formation behind peri-nuclear ring (PNR). Endoplasmic reticulum (ER) is associated with the acrosome. (F) Caudal tube (arrow) as end phase of manchette.

not appear to have intercellular bridges (see later on number of spermatids released).

The four main developmental phases of spermatids could be designated in the NMR. Early round spermatids (Golgi phase, **Figure 3B**; also see later **Figures 5A,B** for description of spermatid types) are often associated with a large, star-like arrangement of the Golgi apparatus (**Figure 3C**) that eventually appear to contribute to construction of the acrosome. Round to oval spermatids are showing acrosome formation (cap phase, **Figure 3D**; also see later **Figure 5B**) and the peri-nucleolar ring is also formed. Many microtubules are associated with the Golgi phase and also associate with the acrosome.

During the phase of spermatid elongation, a simple/rudimentary manchette of microtubules is formed below the perinuclear ring (**Figure 3E**). This structure is first observed at the acrosome formation stage as a single row of microtubules and in longitudinal sections it appears as a chain of connected microtubules. Several cisternae of endoplasmic reticulum also appear at this stage in the Sertoli cell cytoplasm (usually associated with cell adhesions) and are arranged in the area of the spermatid acrosomal cap (**Figures 3D,E**). In late spermatids with elongated heads, a clear caudal tube (posterior segment of manchette) emerges (**Figure 3F**) at the posterior part of the head and seems to unwind in a snake-like fashion (**Figure 4A**). Elongated spermatids are also depicted in **Figures 5C,D** in the description of spermatid types. It furthermore seems that microtubules at the posterior part of the sperm head break up into small groups of six to eight microtubules and associate with smaller Sertoli cell vesicles (**Figures 4B,C**).

Finally, the release of mature spermatids/spermatozoa from the Sertoli cells appear to be assisted by this microtubule mechanism of forming larger vesicles and accompanied by vast Sertoli cell fluid formation (**Figures 2A**, **4B–D**). **Figures 4B–D** typically show residual bodies with two vacuoles presumably form which two spermatids that have been released. We refer to this "excessive" fluid formation (compared to any other mammalian species) as the Sertoli cell pump – a novel adaptation in mammals and unique to the NMR, as will be discussed.

Many spermatid morphological abnormalities are evident during spermiogenesis. The major atypical aspects are the heterogeneity of head morphology and the large number of spermatid and sperm heads showing fragmentation (**Figure 4E**). Many residual bodies seem to contain the remains of the vesicles that released the spermatozoa, many of which are presumably phagocytosed by the Sertoli cells forming phagosomes (**Figures 4E,F**). These phagosomes typically exhibit formation of myelin figures which may be partly related to end products of the phagocytosed material (see section below). A further novel and surprising finding in NMR is the presence of phagocytes (unusual) as part of the seminiferous epithelium (**Figure 4F**).

#### Staging of Spermatogenesis

Staging of spermatogenesis has only been performed in sections of seminiferous tubules presenting with most spermatogonia,

FIGURE 5 | (A–E) Morphological details of the four developmental stages of the spermatids. (A) Sa, round spermatid also showing acrosomal granule and part of Golgi phase. (B) Sb, spermatid with acrosome formation and slight elongation. (C) Sc, elongated spermatid showing fragmentation (F). (D) Sc, lobe-shaped elongated spermatid. (E) Sd, elongated mature spermatid/sperm.

spermatocytes, spermatids, and spermatozoa and was more distinct in males sacrificed close to estrous of the queen. Spermatogenesis in the NMR appears to be helical, similar to humans, and more than one stage (spermatogenic cellular association) can be designated in a transverse section of the seminiferous tubules. However, defining these cellular associations in NMRs is complicated due to the heterogeneity in spermatid morphology. It was accordingly important to first distinguish the four spermatid developmental stages (**Figure 5**), namely: Sa, representing round spermatids in Golgi phase (**Figures 3B**, **5A**); Sb, representing acrosome development, peripheral nuclear condensation and start of elongation (**Figures 3D,E**, **5B**); Sc, representing more pronounced elongation and nuclear condensation (**Figures 5C,D**); and Sd, representing mature elongated/mature spermatids (**Figure 5E**).

Despite the limitations posed by spermatid heterogeneity, **Figure 6** illustrates six spermatogenic stages and **Table 2** lists the different cellular associations of spermatogenesis as shown

FIGURE 6 | (A–D) Six spermatogenic stages of the NMR. Note that Stages I and II and V and VI show many overlapping features and have been grouped together as Stage I/II and Stage V/VI. Since all stages show A type spermatogonia, they have not been indicated in every stage. (A) Stage I/II showing A type spermatogonium (A Sg), pachytene primary spermatocyte (P) and spermatid stages Sa and Sd. (B) Stage III, pachytene primary spermatocyte (P), Sb spermatids and Sertoli cell (SC). (C) Stage IV, A type spermatogonium (A Sg), pachytene primary spermatocyte (P) and Sc spermatid. (D) Stage V/VI with zygotene primary spermatocytes (Z) and Sc spermatids. Only zygotene spermatocytes and Sc spermatids are indicated as this combination distinguishes stage V/VI from the other stages.

TABLE 2 | The cellular associations used in defining the spermatogenic stages of the naked mole-rat.


Sa, Sb, Sc, Sd = four developmental stages of spermatids.

in these micrographs. Delineation of these six stages was mainly based on an association of spermatogonia, specific primary spermatocytes and the presence of one or two of the four spermatids types. For example, NMR Stage IV contains, as viewed successively from the basal lamina, A type spermatogonia, B type spermatogonia, leptotene and pachytene primary spermatocytes and mainly type Sc spermatids (**Figure 6C**). For the sake of simplicity and difficulty to consistently distinguish between Stages I and II as well as V and VI, they were respectively grouped together as I/II and V/VI (**Table 2**).

#### Tubuli Recti and the Rete Testis

The transition between a seminiferous tubule and the tubuli recti is shown in **Figure 7A**. The tubuli recti in NMRs are short but have the same basic structure as in other mammals. Their cuboidal type cells contain a very large, irregular nucleus and the cytoplasm forms undulations projecting into the lumen that is packed with glycogen granules (**Figure 7B**). The tubuli recti are confluent with the rete testis which has a large lumen and the cuboidal cells contain short microvilli (**Figure 7C**).

#### Other Features of the Testis

There are numerous structures/organelles in NMRs not commonly observed in the testis of most mammals. **Figure 8A** shows an example of myelin figures as it occurs in different forms inside the Sertoli cytoplasm and in Leydig cells (**Figure 8B**). These myelin figures present in Sertoli cells seem to form part of phagosomes. Annulate lamellae within the Sertoli cells are indicated in **Figure 8C**. These lamellae form many rows of parallel arranged membranes that seem to constrict at specific points and then expand again. There are many examples of these in NMR testes and they have many different types of arrangement. Rows of parallel-oriented and circular membranes are evident, particularly in the distal parts of Sertoli cells and residual bodies (**Figures 8D–F**). Crystals are present inside the Leydig cells but these do not appear to be a regular feature (**Figure 8G**).

## DISCUSSION

It has been well-established in the literature that testicular size is larger in breeders compared to non-breeders in the NMR (Faulkes et al., 1991, 1994; Yang et al., 2017). However, in the

right and arrow showing more myelin figures with tighter arranged membranes inside a phagosome. (B) Typical myelin figure (arrow) in Leydig cell cytoplasm. (C) Annulate lamellae (in between arrows) surrounded by granules (G) in Sertoli cell. (D–F) Membranes possibly of agranular endoplasmic reticulum (arrows) in residual body and distal part of Sertoli cell close the seminiferous tubule lumen. (G) Two presumably para-crystalline structures arranged in groups of two and surrounded by smooth endoplasmic reticulum (SER).

subsequent discussion we concentrated on the events that occur both in the seminiferous tubules as well as the interstitial cell compartment of breeder, subordinate and disperser NMRs. As we did not find any differences in the cell types among the different groups close to estrous in the queen, the descriptions below will mainly relate to breeders.

#### Testicular Compartments and Leydig Cells

The NMR has intra-abdominal testes similar to most Afrotheria (e.g., shrews, moles, aardvark, elephant, manatee, and hyrax), also referred to as testiconds in some rodents (van der Horst, 1972). In many of these testiconds the interstitial tissue is considerably expanded during their breeding season compared to most scrotal mammals (Glover and Sale, 1968; Woodall and Skinner, 1989). In the NMRs used in the present study, the interstitial tissue makes up approximately 50% of the testis surface area/volume, which is similar to the 60% reported for NMRs by Fawcett et al. (1973) and confirmed by others (Onyango et al., 1993; Endo et al., 2002). There is no clear reason for this phenomenon and there are exceptions in testiconds with interstitial tissue only comprising about 9% of testicular volume in some shrews (Kisipan et al., 2014).

The volume of interstitial tissue or the area of the testis occupied by seminiferous tubules seem to be linked to levels of sperm competition in some species, with monogamous species or species with limited male intrasexual competition having a smaller testicular area occupied by tubules (delBarco-Trillo et al., 2013; Peirce et al., 2018). An investigation of various murine rodents by Peirce et al. (2018) revealed that in some, but not all, species with a lower relative testis mass (and thus lower sperm competition level), the seminiferous tubules made up 60– 70% of the testis area. While no difference between breeders, subordinates and dispersers were found in the current study in terms of the percentage testis area occupied by seminiferous tubules, Mulugeta et al. (2017) indicated a larger interstitial (Leydig) cell content for NMR breeders compared to nonbreeders. In another social mole-rat species, Ansell's mole-rat (Fukomys anselli), the relative area occupied by Leydig cells and seminiferous tubules was also not significantly different between breeders and non-breeders (Montero et al., 2016).

A special feature of NMR testes is that the seminiferous tubules appear to be filled with fluid and resultantly considerably distended. Despite this observation, the diameter of NMR seminiferous tubules is in the same range as reported for other mole-rats, shrews and murine rodents (Kisipan et al., 2014; Montero et al., 2016; Peirce et al., 2018). Peirce et al. (2018) found a significantly greater mean diameter of the seminiferous tubules in species with a higher compared to those with lower relative testis mass in two of the three murine tribes investigated. Although de Bruin et al. (2012) reported a larger diameter of seminiferous tubules in breeding compared to non-breeding Ansell's mole-rat males, this was not evident in the study by Montero et al. (2016) on the same species or for NMR as reported by Yang et al. (2017) and in the current study.

Another feature that may be unique to NMR is that Leydig cells form islands (three Leydig cells per island) surrounded by regular/loose connective tissue and closely associated with arterioles. We speculate that the two types of vesicles observed in Leydig cells may reflect different stages of steroidogenesis, eventually producing testosterone. The lipid-rich NMR Leydig cell cytoplasm has been reported before (Fawcett et al., 1973; Onyango et al., 1993; Yang et al., 2017), although no mention was made of different types of lipid droplets (translucent vs. dark) encountered. Yang et al. (2017) observed that the Leydig cell area containing lipid droplets, as well as the density of autophagosomes surrounding these droplets, was larger for NMR breeders compared to non-breeders. The latter study linked the lower plasma testosterone levels in non-breeders to a significant decrease in autophagy observed in these males.

#### Spermatogenesis and Formation of Abnormal Spermatids/Spermatozoa

Spermatogenesis was evident in all males (breeders, subordinates, and dispersers) included in this study, as was reported in numerous previous studies (Faulkes et al., 1991, 1994; Onyango et al., 1993; Mulugeta et al., 2017; Yang et al., 2017). The morphology of the NMR spermatogenic cells, including spermatogonia and spermatocytes, are remarkably similar to that of other mammalian species. Early spermiogenesis in NMR is indeed characterized by the typical main phases such as Golgi phase and cap phase. However, here the similarities end in relation to other mammalian species, with the tail composition and maturation phases being unusual (van der Horst et al., 2011). One exception in the early spermatid phase (Golgi phase) is the formation of bizarre and excessive star-like Golgi bodies which seem to contribute to acrosome formation. These Golgi bodies presumably provide vesicles that will contribute to acrosome formation and its remnants still remain close to spermiation (**Figure 4D**). Furthermore, the elongated spermatids (Sc and Sd) are highly heterogeneous in terms of their head morphology and are often fragmented.

Historically, Fawcett et al. (1971) suggested that manchette formation and function is not related to shaping the form or elongation of the sperm head. Subsequently and recently, several authors have convincingly shown, at both electron microscopic and molecular level, that there are two closely associated developments of cytoskeletal elements linked to protein transport and shaping of the spermatid/sperm head (Lehti and Sironen, 2016; Gunes et al., 2018; Wei and Yang, 2018). The first element is mainly actin-bound (related to ectoplasmic specializations of Sertoli cell) and surrounds both the acrosome area and remainder of the sperm head as part of the acrosome–acroplaxome complex. Secondly, a microtubule-based system that originates below the acrosome and is attached to the peri-nuclear ring is involved in manchette formation and indeed important in elongation of the sperm head as well as determining the morphology of the spermatozoon (Wei and Yang, 2018).

It appears that both the acroplaxome and manchette are temporary structures that, apart from their role in nuclear condensation through elongation, provide an important transport system. The manchette region forms an essential part of intra-microtubule transport (IMT) of a multitude of proteins to the basal body/migrated centrioles in relation to formation of the sperm flagellum. It was suggested that many of these IMT proteins feed into an intra-flagellar transport pathway via several motor proteins (Lehti and Sironen, 2016). Once the sperm tail is formed, the manchette disappears after performing its "zipper" action. This action allows for head elongation due to compression of nuclear proteins that become arginine enriched and converted to smaller protamine molecules and subsequently assist in transport of proteins to the tail (Wei and Yang, 2018). The above authors furthermore provided a large body of evidence that when either manchette formation is faulty/absent or some of the multitude of proteins is lacking, it often results in infertility due to absence of motility.

In the NMR, the manchette appears rudimentary comprising either a single or maximal two layers of microtubules in contrast to other mammals where there are typically multiple layers of microtubules forming a substantial sheath around the posterior part of the sperm head (Fawcett et al., 1971). The high degree of sperm head polymorphisms, fragmentation (illcondensed chromatin) and short tail lacking a fibrous sheath reported here for late spermatids and previously for spermatozoa (van der Horst et al., 2011) are probably a result of this rudimentary/degenerate manchette. It is thus not surprising that

the NMR has many abnormal spermatozoa and poor motility since there is a lack or modification of many key functions in completing spermiogenesis (O'Donnell, 2014).

However, NMR breeders with degenerate spermatozoa are highly successful in impregnating the female, despite van der Horst et al. (2011) and van der Horst and Maree (2014) indicating that mature spermatozoa of NMR breeders are grossly abnormal (only 7% of sperm are morphologically normal) and the percentage motility is less than 14%. These authors explained that in the absence of sperm competition, spermatozoa seem to degenerate and sperm quality decreases dramatically but it is still sufficient to fertilize. Several recent studies on other mammalian and bird species similarly related the presence of a high percentage of abnormal spermatozoa to potentially relaxed sperm competition and a monogamous mating system (Breed et al., 2018; Humann-Guilleminot et al., 2018; Peirce et al., 2018). Dorman et al. (2013) indicated that there seems to be co-evolution of the gametes to aid sperm-zona interactions in the Bandicoot rat (Bandicota indica), where a low level of sperm competition is also apparent.

Apart from the correlation between poor NMR sperm morphology/motility and degenerate manchette organization and formation, there are other modifications of the NMR manchette that does not exist or has not been described in other species, which appear to be special adaptations in NMR for sperm release. It is well-known that the manchette essentially disappears in mammals toward the end of spermiogenesis and after flagellar formation. All that remains of the mammalian manchette is its posterior part, namely the caudal tube. In the NMR, the single row of microtubules (caudal tube) seems to remain as a snake-like configuration in relation to the posterior part of the sperm head inside the residual body being formed by the Sertoli cell. Subsequently, the microtubules become re-arranged around small Sertoli cell vesicles, combine to form larger vesicles and these seem to assist to release late spermatids/spermatozoa into the testicular lumen via a Sertoli pressure pump mechanism. Although similar large vesicles in the Sertoli cells are visible or have been reported before in both NMR and Ansell's mole rat, it was referred to as "lesions" or "vacuoles" and its presence was not discussed (Faulkes et al., 1991; Endo et al., 2002; Montero et al., 2016; Yang et al., 2017). Our hypothesis of the pump mechanism shows some similarities to a simplified spermiation process in a species of toad (Bufo arenarum) where the endoplasmic reticulum of the Sertoli cell becomes fluid-filled after stimulation by luteinizing hormone and elongated spermatids are pushed out from their crypts in the Sertoli cell (Burgos and Vitale-Calpe, 1967a,b). In the NMR, typically two mature spermatids/sperm are jointly released into the lumen as is evident from the now empty vacuoles of the residual body (**Figure 4**). Spermatozoa are usually released in large groups with still intact intercellular bridges as evident in human (Dym, 1977) and other mammals (Hess and Renato de Franca, 2008). This further supports our notion that there are no intercellular bridges in between spermatids close to being released.

Another NMR sperm feature that has not been observed in other mammals relates to the elimination of the spermatid cytoplasm. In mammals, spermatozoa released from the testis possess a cytoplasmic droplet that is eventually shed in the cauda epididymis or vas deferens. However, NMR spermatozoa released in the seminiferous tubule do not contain a cytoplasmic droplet and it appears that all the redundant organelles have been shed before spermatid/sperm release. This apparent modification in the spermiation process is supported by the fact that Sertoli cells lack typical apical processes that push spermatids into the lumen and stalks of cytoplasm that stay in contact with the spermatids up until final release (Beardsley et al., 2006; O'Donnell, 2014). It is possible that in NMRs an alternative mode of cytoplasmic elimination is used as described for non-mammalian vertebrates (Sprando and Russell, 1988; Russel, 1993) and represents yet another feature potentially related to degenerative orthogenesis and the unique NMR spermiation process.

While the above discussion represents a totally new explanation of the mechanism for the release of spermatozoa from the Sertoli cell it still poses the question of how does this degenerative system relates to fertility, as in all the examples we studied the breeders were proven fathers? In our previous papers we have shown the relationship between sperm degenerate features and fertility and lack of sperm competition (van der Horst et al., 2011; van der Horst and Maree, 2014). Accordingly, it becomes apparent at what stage of spermatogenesis sperm morphology becomes abnormal. The early phases of spermatogenesis seem to be normal in the NMR but then during acrosome formation (cap phase), the manchette dictates development of normal sperm morphology and here it is simplified/degenerate. These observations are in agreement with findings of Mulugeta et al. (2017) on differentially expressed genes in the NMR gonads between breeders and non-breeders which indicated upregulation of genes in breeders that are mainly expressed in the post-meiotic stages of spermatogenesis. A possible explanation for the lower sperm numbers and sperm motility that have been reported in NMR non-breeders compared to breeders (Faulkes et al., 1991, 1994) stems from additional molecular data reported by Mulugeta et al. (2017) which indicated that genes involved in the meiotic cell cycle, male genitalia development, gamete generation, spermatogenesis, and sperm motility are upregulated in breeders.

#### Staging of Spermatogenesis

Deciphering the spermatogenic stages in the NMR was difficult for two reasons. Firstly, it needs to be emphasized that the wellorganized arrangement of groups of specific types of primary spermatocytes or spermatids as in most mammals do not exist in NMRs but rather a more diffuse/disorganized arrangement of these cells. Secondly, there are typically many abnormal spermatids/spermatozoa in the NMR and the basis for staging is to relate a particular association of spermatogonia and spermatocytes to a specific spermatid type. The poorly organized arrangement of the various types of spermatogenic cells, also previously reported by Endo et al. (2002) for non-breeders, is possibly due to the fact that intercellular bridges between closely associated cells are scarce.

In the rat there are 19 different types of spermatids, with each of these spermatids having a specific association with spermatogonia and spermatocyte types, resulting in 14 clear

spermatogenic stages (I to XIV). In humans there are only four different spermatid types and accordingly only six spermatogenic stages (I to VI) (Dym, 1977). In the NMR there are also four different types of spermatids that, despite their apparent abnormal and fragmented structure, show some resemblance to those in humans. The spermatid heterogeneity thus posed a challenge to find a coherent system relating to specific spermatid types in defining a specific spermatogenic stage in the NMR. Despite this it was possible to recognize six stages in the NMR that show similarities to human Stages I, II, III, IV, V, and VI. However, in our classification system as indicated in **Table 2** many similarities were evident between Stages I and II as well as V and VI and these stages were respectively grouped together. It may accordingly be then conceived that only four stages are evident in the NMR but due to their closer relationship to human staging (as compared to rodents) we prefer using the human classification rather than delineating a new one for the NMR. We have previously pointed out that in humans whom are largely monogamous there is a low risk of sperm competition (van der Horst and Maree, 2014). Accordingly, it is not surprising to see some similarities between NMR and humans in staging of spermatogenesis. Moreover, complexity in the process of spermatogenesis is in parallel with the evolution of mammals. It appears that the efficiency of spermatogenesis decreased with humans, being less efficient than most mammals such as bulls or rats (Thompson et al., 2018).

While it is accepted that the queen suppresses many aspects of reproduction in the colony, does that mean she suppresses spermatogenesis in all the males all the time? Faulkes et al. (1991) found that while there are differences in testicular weight and size between breeders and non-breeders, all males exhibited spermatogenesis. Our findings agree with Faulkes et al. (1991) and in addition we found that when spermatogenesis is studied at the same period of the estrous cycle in the same colony, breeders, subordinates and dispersers will exhibit similar testicular histology and all may produce and release spermatozoa from the testes. Accordingly, when breeders do not contain spermatozoa in the testis, a non-breeder from the same colony will also not have spermatozoa at that point in time. This observation raises another question: are breeding males also suppressed and do not produce spermatozoa at a certain time in the colony? We have investigated this potential phenomenon which will be deliberated in a separate paper in the near future.

#### Other Features of the NMR Testis

Myelin figures is considered to be non-specific, concentrically layered, osmiophilic and possibly derived from agranular reticulum. In both the Leydig cells as well as the Sertoli cells there appear areas rich in agranular vesicles and in some parts of the cell they may actually form the concentric layers we observed in these cells in NMR (Neaves, 1973). There is good evidence to suggest that these myelin figures may act as important cholesterol storage centers but are also involved in synthesis of various types of steroids such as testosterone and progesterone (Christensen and Fawcett, 1966). Morita et al. (1999) showed that the myelin figures in testis labeled positively for a family of claudine as in myelinated nerves and are very important in formation of tight junctions. It is accordingly not surprising to find that highly concentrated myelin figures as found in NMR Sertoli and Leydig cells are potentially functional in the testis.

Annulate lamellae within the NMR Sertoli cells form different types of arrangements, one of which rows of parallel-arranged membranes appear to constrict at specific points and then expand again just like the outer nuclear membrane. The bulk of information according to older literature (Wischnitzer, 1970) suggests that these organelles originate from the outer nuclear membrane and are important in messaging between nucleus and cytoplasm. Merisko (1989) attributed that, apart from ATP production and membrane biogenesis, there is macro-molecular trafficking via the pores and regulator of genomic expression. They furthermore seem to associate in many instances with ribosomes and become an active source of protein synthesis. The lamellae in this form can either join rough endoplasmic reticulum or be involved in the formation of endoplasmic reticulum. The most important explanation relating to NMR Sertoli cells is that the vast amount of fluid containing vesicles and membranes as well as the extensive agranular reticulum in Leydig cells may actually also be involved in the formation of annulate lamellae and myelin figures. While we can only speculate, it appears that the rows of parallel or circular membranes in the distal Sertoli cells and residual bodies may be associated with the formation of myelin figures and/or annulate lamellae. Christensen and Fawcett (1966) suggested that these membranes can largely be considered smooth endoplasmic reticulum and accordingly have steroid producing functions as suggested above for myelin figures.

Crystals of Reinke are commonly found in Leydig cells as well as the crystalloid of Charcot-Bottcher in Sertoli cells (Dym, 1977). However, we have found a simple crystal-like structure usually occurring in pairs of two in the Leydig cells of the NMR. These are not products of precipitation of processing material as they are clearly embedded in the cytoplasm. There appear to be many types of so called para-crystalline structures both in Leydig cells as well as in the cytoplasm of Sertoli cells (Sohval et al., 1973). Some are in fact almost tubular like in NMR but none bear a close resemblance. Despite all the speculation about the crystalline structures in Leydig cells there seem to be some association with age. These structures appear in the testes of older men as well as in older animals and do not seem to have a specific function (Sohval et al., 1973).

## CONCLUSION

This study used ultrastructural features of the testis to describe both interstitial and testicular components and their associated cell types as well as staging of spermatogenesis in the NMR for the first time. Our results suggest that the main differences in testicular structure and spermiogenesis in NMR compared to other mammals pertains to low levels of sperm competition and supports prior investigations on NMR spermatozoa and investigations on birds and rodents (van der Horst et al., 2011; van der Horst and Maree, 2014). This hypothesis also assists to explain the phenomenon of mature abnormal spermatozoa in

NMR being a function of testicular development/modification, specifically during spermiogenesis.

The testicular compartments of the NMR revealed several aspects that are atypical for a mammalian species and unique for this eusocial species. The NMR interstitial tissue, comprising mostly of Leydig cells, occupies approximately 50% of the surface area/volume of the testis, similar to many other Afrotheria. Seminiferous tubules in NMR show all the typical cell types expected during spermatogenesis with some important exceptions. Mitotic divisions of the Sertoli cells are often found and their cytoplasm is packed with fluidfilled vesicles. These vesicles merge and eventually forms part of our hypothesized Sertoli cell pump which, in addition to other adaptations for a simplified spermiation process, assists to release mature spermatids into the seminiferous tubule lumen. Instead of several spermatids being closely associated not more than two spermatids are found in close vicinity to each other and their release is typically associated with copious amounts of fluid being released in the lumen which appears to assist spermatid release and hence support our Sertoli cell pump hypothesis.

The most prominent feature in which NMR spermiogenesis differs from that of other mammals is in the structure and formation of the manchette which is simplified/degenerate in NMR. Since manchette formation involving many rows of microtubules is associated with producing normal shaped spermatids and normal sperm morphology, it is expected that there will be many diverse and abnormal sperm forms in NMR. Furthermore, the "remnants" of the manchette is modified in NMR and seems to be associated with the release of spermatids from the Sertoli cells. Intercellular bridges are present during early phases of spermatogenesis but then seem to "disappear" and are not present in spermatids. The lack of a cytoplasmic

#### REFERENCES


droplet during release of sperm into the seminiferous tubule lumen emphasize further modification/simplification compared to other mammals and relate to degenerative orthogenesis.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the manuscript/supplementary files.

#### ETHICS STATEMENT

The animal study was reviewed and approved by University of Cape Town (2005/V7/JOR), the University of the Western Cape (ScR1RC2007/3/30), and Stellenbosch University (P07/09/019).

#### AUTHOR CONTRIBUTIONS

GH and LM collated the results and drafted the manuscript. All authors read and approved the final manuscript and designed the study, sampling and analyzed the data.

#### ACKNOWLEDGMENTS

The authors express their gratitude to Prof. J. U. M. Jarvis for her interest in the study and for initially supplying electron micrographs of NMR spermatozoa and rete testis. Ms. N. Muller, Electron Microscopy Unit, Tygerberg Hospital, South Africa, is thanked for her assistance with the preparation of the TEM samples and the electron micrographs.


mole-rat (Heterocephalus glaber) and the damaraland mole-rat (Cryptomys damarensis). J. Reprod. Fertil. 100, 411–416. doi: 10.1530/jrf.0.1000411


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 van der Horst, Kotzé, O'Riain and Maree. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Importance of SLC26 Transmembrane Anion Exchangers in Sperm Post-testicular Maturation and Fertilization Potential

#### Aminata Touré\*

INSERM U1016, Centre National de la Recherche Scientifique, UMR 8104, Institut Cochin, Université de Paris, Paris, France

#### Edited by:

Zhibing Zhang, Virginia Commonwealth University, United States

#### Reviewed by:

Jormay Lim, National Taiwan University, Taiwan Clemence Belleannee, Laval University, Canada

> \*Correspondence: Aminata Touré aminata.toure@inserm.fr

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 16 July 2019 Accepted: 26 September 2019 Published: 18 October 2019

#### Citation:

Touré A (2019) Importance of SLC26 Transmembrane Anion Exchangers in Sperm Post-testicular Maturation and Fertilization Potential. Front. Cell Dev. Biol. 7:230. doi: 10.3389/fcell.2019.00230 In mammals, sperm cells produced within the testis are structurally differentiated but remain immotile and are unable to fertilize the oocyte unless they undergo a series of maturation events during their transit in the male and female genital tracts. This post-testicular functional maturation is known to rely on the micro-environment of both male and female genital tracts, and is tightly controlled by the pH of their luminal milieus. In particular, within the epididymis, the establishment of a low bicarbonate (HCO<sup>3</sup> <sup>−</sup>) concentration contributes to luminal acidification, which is necessary for sperm maturation and subsequent storage in a quiescent state. Following ejaculation, sperm is exposed to the basic pH of the female genital tract and bicarbonate (HCO<sup>3</sup> −), calcium (Ca2+), and chloride (Cl−) influxes induce biochemical and electrophysiological changes to the sperm cells (cytoplasmic alkalinization, increased cAMP concentration, and protein phosphorylation cascades), which are indispensable for the acquisition of fertilization potential, a process called capacitation. Solute carrier 26 (SLC26) members are conserved membranous proteins that mediate the transport of various anions across the plasma membrane of epithelial cells and constitute important regulators of pH and HCO<sup>3</sup> <sup>−</sup> concentration. Most SLC26 members were shown to physically interact and cooperate with the cystic fibrosis transmembrane conductance regulator channel (CFTR) in various epithelia, mainly by stimulating its Cl<sup>−</sup> channel activity. Among SLC26 members, the function of SLC26A3, A6, and A8 were particularly investigated in the male genital tract and the sperm cells. In this review, we will focus on SLC26s contributions to ionic- and pH-dependent processes during sperm post-testicular maturation. We will specify the current knowledge regarding their functions, based on data from the literature generated by means of in vitro and in vivo studies in knock-out mouse models together with genetic studies of infertile patients. We will also discuss the limits of those studies, the current research gaps and identify some key points for potential developments in this field.

Keywords: SLC26, CFTR, pH, sperm, epididymis, motility, capacitation, fertilization

#### INTRODUCTION

fcell-07-00230 October 16, 2019 Time: 17:33 # 2

Spermatozoa constitute one of the most differentiated cell types of the body and are produced within the seminiferous tubules of the testis during spermatogenesis, a complex and tightly regulated process of nearly 2.5 months long in humans and 7 weeks in mice (Heller and Clermont, 1964; Clermont, 1972). At the end of this process, spermatozoa are fully differentiated at the morphological level, and comprise two main compartments: the head and the tail, each fulfilling specific functions that are essential for fertilization (**Figure 1**). The sperm head comprises the nucleus, in which the haploid paternal DNA is highly compacted through replacement of histones by protamines that mediate hyper condensation of the chromatin during spermiogenesis (Miller et al., 2010). In addition, the acrosome, a peculiar vesicle which derives from the Golgi, locates to the anterior half of the sperm head and forms a large cap containing various proteases and membrane receptor that are required to cross the cumulus cell layer and digest the zona pellucida that surround the oocyte (Foster and Gerton, 2016). The tail or flagellum is an organelle of 50 to 100 micrometer long in mammals, which sustains sperm motility and progression within the female genital tract and is thus also indispensable for fertilization. It is composed of an evolutionary conserved microtubule-based structure, called the axoneme, which is also shared with cilia, and contains nine microtubules doublets (MTD) organized around a central pair of microtubules (CP) (Inaba, 2007, 2011). Attached to the MTD, the Inner and Outer Dynein Arms (IDAs and ODAs), which constitute multiprotein complexes with ATPase activity, drive the sliding of the MTD, and orchestrate the sperm flagellum beating. In addition, MTD are connected to each other through the nexin-dynein regulatory complex (NDRC) and to the CP complex via the radial spokes (RS) (Inaba, 2007, 2011; **Figure 1**). This latter multi-protein complex ensures the stability of the axonemal structure and may also function as a scaffold for signaling molecules such as Calmodulin (CaM) and Protein kinase A (Yang et al., 2006).

In spermatozoa from primitive species such as fishes, the flagellum is similar to that of cilia and only comprises the microtubular axoneme surrounded by the plasma membrane. In mammals, the sperm cells harbor peri-axonemal structures that surround the microtubule-based cytoskeleton of the tail beneath the plasma membrane and are required for structural cohesion, energy regulation and cell signaling (Eddy et al., 2003; Eddy, 2007). The structure of mammalian sperm tail can be divided in several compartments, based on their content on periaxonemal structures. Hence, the midpiece (MP) is characterized by the presence of the mitochondrial sheath (MS) and contains outer dense fibers (ODFs), which ensure elasticity and structural integrity (Lehti and Sironen, 2017). The principal piece (PP) is characterized by the fibrous sheath (FS), which comprises two longitudinal columns attached to doublets 3 and 8 that partially displace ODFs and are connected by semi-circular ribs. Proteins from the FS are stabilized by di-sulfide bonds, which suggest that the FS might strengthen the sperm structure and influence its flexibility (Eddy et al., 2003). The FS also behaves as a scaffold for proteins regulating sperm motility and functionality as it harbors glycolytic enzymes and signaling molecules, such as AKAP proteins and cAMP-dependent protein kinase (Eddy, 2007; Lehti and Sironen, 2017). Lastly, the terminal piece encloses the sperm tail and is only composed of the axoneme. In addition, two specific structures of the sperm tail are also distinguishable: the connecting piece, which anchors the tail to the sperm head (Inaba, 2007, 2011) and the annulus, a Septin-ring structure (also called Jensen's ring), which locates at the boundary of the midpiece and the principal piece and acts as a diffusion barrier to ensure the correct localization of proteins along the different compartments of the sperm flagellum (Toure et al., 2011; **Figure 1**).

While spermatozoa released from the testicular seminiferous tubules are morphologically differentiated, they are immotile and unable to recognize and fertilize the oocyte. Importantly, sperm functionality will be conferred by a series of maturation events occurring during their transit through the male and female genital tracts (Fraser, 1992; Yeung and Cooper, 2003). Such post-testicular functional maturation is known to rely on the luminal milieus of the male and female genital tracts, which composition results from specific absorptive and secretory activities of epithelial cells that line the lumen of both tracts. In particular, sperm maturation is tightly controlled by the pH of the luminal milieu. In the first part of this review we will describe the current knowledge on (i) the cellular cross-talks and membrane transporters that are involved in the establishment of the acidic luminal milieu required for epididymal maturation, and (ii) the ionic fluxes and physiological changes occurring in the sperm cells upon capacitation, which are triggered by sperm exposition to high concentration of HCO<sup>3</sup> <sup>−</sup> in the female genital tract. We will next focus on SLC26 proteins, a well-established transmembrane protein family involved in anionic transport and pH regulation in various epithelia. We will provide a comprehensive description of their contributions in the regulation of some hallmarks associated with capacitation and present recent data, which revealed their functions within the epididymal cells. Lastly, we will describe the human disorders related to SLC26 dysfunctions in the male reproductive organs. We will conclude the review by discussing some of the current research gaps in this field and presenting potential perspectives for future research.

#### SPERM EPIDIDYMAL MATURATION

#### Epididymis Structure and Function

When spermatogenesis is completed within the seminiferous tubules of the testis, spermatozoa are released in the lumen and collected through the rete testis and the efferent ducts to join the epididymal tract. The epididymis consists in a unique and highly convoluted tube of nearly 1 and 6 m long, in mice and humans, respectively (Hinton et al., 2011). It is anatomically divided into three principal regions, the caput, corpus, and cauda, following the proximal-distal axis (Hirsh, 1995; Turner, 2008); each of them being further divided in several segments as it is well-documented in rodents (10 and 19 segments in mouse and rats, respectively) (Johnston et al., 2007; **Figure 2**). In some

species, specific morphological features can be observed, like in rodents, where a proximal segment adjoined to the testis, called the initial segment, is distinguishable (Soranzo et al., 1982). The epididymal epithelium is composed of different cell types, which are the principal cells (PCs), the clear cells (CCs), the narrow cells (NCs), the apical/basal cells and the halo cells together with immune cells such as macrophages and dendritic cells (Da Silva and Smith, 2015; Voisin et al., 2018; Breton et al., 2019; **Figure 2**). The proportion of each epididymal cell-type is variable between the different segments of the epididymis, which are associated with specific secretory and absorptive activities providing distinct luminal micro-environments along the tract contributing to sperm maturation (Breton et al., 2019). For instance, PCs, which are the most abundant cell type, are present in all epididymal regions while NCs exclusively locate to the initial segment and CCs are present in the caput and corpus, and highly enriched in the cauda (Hermo et al., 1992; Adamali and Hermo, 1996).

The epithelium of the epididymis constitutes a physical barrier which protects the sperm cells from the immune system and is essential to mediate vectorial transport of ions, solutes, nutrients and water from the blood circulation thus insuring sperm cell survival and protection (Hinton et al., 1995; Gregory and Cyr, 2014). Some of the maturation events occurring during sperm decent along the epididymal tract have been elucidated and it is now well-established that spermatozoa will acquire their motility when transiting through the caput and corpus regions before being stored until ejaculation, in a quiescent state in the caudal region (Bedford, 1967; Orgebin-Crist, 1967). Proteomic and transcriptional analysis of sperm retrieved from the different segments of the epididymis in several species, provided valuable information regarding the changes occurring during epididymal transit. Changes in surface protein content and post-translational modifications have been reported; sperm lipid composition is also described to be modified and influences membrane fluidity in preparation for egg fusion and fertilization (Gervasi and Visconti, 2017). Studies performed on the luminal fluid also highlighted variations in its composition along the epididymal tract, in respect to ions, soluble factors, proteins, non-coding RNA (Johnston et al., 2005; Dacheux et al., 2006, 2009; Yuan et al., 2006; Jelinsky et al., 2007; Dean et al., 2008; Moura et al., 2010; Guyonnet et al., 2011; Liu and Liu, 2015; Liu et al., 2015; Browne et al., 2019; for review see Sullivan and Mieusset, 2016;

FIGURE 2 | Schematic representation of the epididymis structure and ionic exchanges between epithelial cells, which control luminal acidification. (A) Left panel, the testis, efferent ductules and epididymis are schematized. The different regions within the epididymis, in mouse, are indicated: initial segment, caput, corpus and cauda, following the proximal-distal axis. Right panel, the distribution of the different epithelial cell-types within the epididymal tract is also illustrated: principal cells (PCs), clear cells (CCs), narrow cells (NCs), and basal cells; the luminal fluid shows epididymosomes, which are small vesicles transferring material from epithelia cells to the sperm cells. (B) Simplified representation of the main ionic fluxes and cross talks occuring between principal, clear, and basal cells. CCs expressed the V-ATPase pumps, which expression at the plasma membrane is induced by HCO<sup>3</sup> <sup>−</sup> and c-AMP dependent pathway. The HCO<sup>3</sup> <sup>−</sup> influx in CCs is mediated by the NBC sodium- HCO<sup>3</sup> <sup>−</sup> transporter. ATP also induces intracellular rise of Ca2+, which increase V-ATPase translocation at the plasma membrane and proton secretion. PCs express the NHE3 sodium-proton antiporter, which contributes to proton secretion and luminal acidification. They also secrete HCO<sup>3</sup> <sup>−</sup> through the CFTR channel. Lastly, basal cells transmit physiological cues, in particular during sexual arousal, which regulate the activity of principal and CCs.

Zhou et al., 2018) most importantly, those studies uncovered the existence of small vesicles, called epididymosomes, which enable material transfer between epithelial and sperm cells and support their maturation, in the absence of intrinsic de novo transcription and translation events (**Figure 2A**; Sullivan et al., 2007; Frenette et al., 2010; for review see Sullivan et al., 2007; Zhou et al., 2018; Trigg et al., 2019).

#### Epididymis Luminal Milieu, Ionic Fluxes, and pH

One important feature of epididymal maturation is the establishment of an acidic luminal fluid, which is required for sperm quiescence during their maturation and storage (Shum et al., 2011). Such specific luminal environment starts to be established within the efferent ductules, which exert an intensive reabsorption of the fluid released with spermatozoa from the testis (Clulow et al., 1998). The acidic pH of the epididymal luminal fluid is related to particular ionic composition, with low level of sodium, Cl<sup>−</sup> and HCO<sup>3</sup> <sup>−</sup> ions, in comparison to that of other organ fluids or blood plasma (Wales et al., 1966; Levine and Marsh, 1971; Jenkins et al., 1980). Overall, it is conferred by specific secretive and absorptive properties of each epithelial cell type and complex intercellular cross-talks (**Figure 2**). First, are involved the CCs, which are categorized as mitochondriarich cells, and actively secrete protons via the V-ATPase proton pump, a multi-protein complex located at their apical side. In those cells, activation of the soluble adenylate cyclase (sAC) and a PKA-dependent pathway, trigger the accumulation of the V-ATPase pump at the plasma membrane from intracytoplasmic storage vesicles (Pastor-Soler et al., 2003; Belleannee et al., 2011; Battistone et al., 2018). The luminal ATP also stimulates membrane addressing of the V-ATPase pump in CCs, through pH-activated ATP purinergic membrane receptors such as P2 × 4 and elevation of the intracellular Ca2<sup>+</sup> (Belleannee et al., 2011; Battistone et al., 2018). In addition, CCs also express the cytosolic carbonic anhydrase type II, which catalyzes hydration of carbon dioxide to HCO<sup>3</sup> <sup>−</sup> and is therefore essential for acid/base transport (Breton, 2001). The PCs, which constitute the most abundant cell type of the epididymis are also very active in absorbing the HCO<sup>3</sup> <sup>−</sup> in the proximal region of the mouse epididymis (initial segment) and in secreting protons through the sodium/hydrogen exchanger NHE3, in the distal region (Park et al., 2017; **Figure 2**). Last, the basal cells are also critical as they transmit physiological cues which regulate the activity of both principal and CCs (Leung et al., 2004; Cheung et al., 2005; Shum et al., 2008). In particular, during sexual arousal, prior to ejaculation, basal cells activate the secretion of HCO<sup>3</sup> <sup>−</sup> by the PCs through the CFTR channel in a cAMP-PKA dependent manner (Park et al., 2017), an action which is hypothesized to prime the spermatozoa (Hagedorn et al., 2007; Pierucci-Alves et al., 2010; **Figure 2**). Interestingly, the luminal HCO<sup>3</sup> <sup>−</sup> may also be incorporated into the CCs via the sodium HCO<sup>3</sup> <sup>−</sup> co-transporter NBC (Jensen et al., 1999), and subsequently activate the sAC-PKA pathway triggering proton secretion. In this sense, CCs may behave as counteractors of luminal pH elevation, and be involved in the regulation of abnormal and/or sustained pH increase conditions. In addition, or alternatively, HCO<sup>3</sup> <sup>−</sup> secretion by the PCs may be part of an integrated paracrine mechanism involving a crosstalk between clear and PCs, and ultimately leading to proton secretion by CCs and lumen acidification.

Overall, within the epididymal milieu, the established acidic pH and the low HCO<sup>3</sup> <sup>−</sup> concentration need to be tightly regulated to insure proper sperm maturation and storage (for review see Bernardino et al., 2019). The importance of such epididymal intraluminal environment is well-demonstrated in various mouse models associated with abnormally elevated pH or ionic disequilibrium conditions, which all induce reduced sperm fertilization potential and male infertility (Yeung et al., 1998, 2004; Zhou et al., 2001; Bigalke et al., 2010; Weissgerber et al., 2012; for review see Zhou et al., 2018).

## SPERM CAPACITATION IN THE FEMALE GENITAL TRACT

#### Definition, Function, and Associated Sperm Changes

Following ejaculation, spermatozoa achieve their ultimate functional maturation, which is initiated by secretions from the male accessory glands (prostate and seminal vesicles) and fully completed in the female genital tract, a process called capacitation. The process of capacitation was discovered by Chang (1951) and Austin (1952), respectively, who experimentally demonstrated that spermatozoa from rabbits and rats, need to spend enough time within the fallopian tubes of the female genital tract, in order to acquire their fertilization. These pioneer studies were immediately followed by investigations aiming at defining the minimal conditions permitting to capacitate sperm cells in vitro and the first successful experiment of in vitro fertilization was performed with hamster eggs by Yanagimachi and Chang (1963). This led to major achievements in reproductive medicine through the development of assisted

reproduction technologies. Many additional investigations permitted to describe some of the molecular and biochemical events associated with capacitation, in humans and other species (see review Gervasi and Visconti, 2016), while sperm epididymal maturation was much less investigated and remains poorly understood.

It is now well-established that capacitation confers the sperm cells a hyperactivated motility characterized by increased flagellar amplitude and beating frequency, which enables sperm cells to penetrate through the cumulus cell layer surrounding the oocyte (Suarez, 2008). In addition, sperm acquire the ability to perform the acrosomal reaction and to specifically recognize and interact with oocyte (Tulsiani and Abou-Haila, 2011). This is associated with biochemical and electrophysiological changes occurring in the cytoplasm and the plasma membrane of the sperm cells, which together constitute some hallmarks of capacitation. Among these principal changes, were described an increase of the plasma membrane fluidity mainly due to cholesterol depletion, and complex ionic fluxes inducing alkalinization of the cytoplasm, plasma membrane hyperpolarization and flagellar protein phosphorylation (**Figure 3**; Visconti et al., 2011). HCO<sup>3</sup> <sup>−</sup>, Cl−, and Ca2<sup>+</sup> ions were described to be involved in those processes. In particular, Ca2<sup>+</sup> and HCO<sup>3</sup> <sup>−</sup> directly bind to the soluble adenylate cyclase (sAC) and stimulate cAMP production (Chen et al., 2000; Jaiswal and Conti, 2003). The resulting increase in intracellular cAMP concentration is responsible for the activation of the protein kinase A (PKA) and subsequent phosphorylation cascades of flagellar proteins that are indispensable for sperm fertilization (Visconti et al., 2011; **Figure 3**). Among the phosphorylated targets, both axonemal and peri-axonemal proteins of the sperm flagellum were identified. Hence signaling proteins such as the AKAP proteins (Carrera et al., 1996) together with enzymes involved in energetic metabolism (pyruvate dehydrogenase and aldolase) (Arcelay et al., 2008) that locate to the fibrous sheath of the sperm flagellum are phosphorylated upon capacitation; other peri-axonemal components of the flagellum, such as the ODF (Mariappa et al., 2010) were also identified. Importantly, structural protein of the axoneme such as Tubulin (Arcelay et al., 2008) and dynein chains (Baker et al., 2010), which orchestrate flagellar beating (see **Figure 1**), are phosphorylated upon capacitation. Ca2<sup>+</sup> also directly binds to CaM present in the sperm cells (head and flagellum) (Jones et al., 1980; Feinberg et al., 1981; Carrera et al., 1996) and regulates additional phosphorylation cascades initiated by the Calmodulin kinase (CaM kinase) (Gonzalez-Fernandez et al., 2012; Navarrete et al., 2015; see review Suarez, 2008). Overall, ion fluxes induced during capacitation tightly regulate both sperm flagellar beating and energy homeostasis.

#### Sperm Capacitation, Ionic Fluxes, and pH

Capacitation is induced by sperm exposition to high HCO<sup>3</sup> − concentration and basic pH in the female genital tract, as compared to the acidic epididymal milieu (Zhou et al., 2005). Hence in contrast to the recorded HCO<sup>3</sup> <sup>−</sup> concentrations of 2–7 nmol/L (pH 6.4; rats) in epididymal cauda (Levine and Marsh, 1971), sperm cells encounter HCO<sup>3</sup> <sup>−</sup> concentrations of 30 nmol/L in the vas deferens (pH 7.5; rats) (Levine and Marsh, 1971), and approximatively 25 to 90 nmol/L in the fallopian tubes (pH 7.61; rabbit) (Vishwakarma, 1962; see review Ng et al., 2018). A panoply of ion transporters located at the surface of murine and human sperm cells was identified and shown to mediate some of the complex ionic fluxes during capacitation (**Figure 4**; Puga Molina et al., 2017). Briefly, the combined proton extrusion and HCO<sup>3</sup> <sup>−</sup> influxes result in cytoplasm alkalinization. Proton extrusion from the sperm cells is mediated by the voltagegated H<sup>+</sup> channel (Hv1) in humans, and the Na+/H<sup>+</sup> exchangers (NHE), also called SLC9 proteins, in humans, mice and rats (Puga Molina et al., 2017). Bicarbonate transporters involved in capacitation include members of the SLC4 sodium-dependent transporter and SLC26 Cl−/HCO<sup>3</sup> <sup>−</sup> exchangers families (cf. infra), together with the CFTR Cl<sup>−</sup> channel, which functions by cooperating with SLC26 Cl−/HCO<sup>3</sup> <sup>−</sup> exchangers, in mouse and in humans (this later aspect regarding SLC26 proteins will be developed in the next section of the review). In addition, several isoforms of carbonic anhydrases (cytosolic and membrane) are present in the sperm and likely contribute in regulating sperm HCO<sup>3</sup> <sup>−</sup> concentration (Puga Molina et al., 2017). The resulting cytoplasm alkalinization regulates pHdependent channels; hence the Ksper and Slo outward potassium channels are activated (Navarro et al., 2007; Santi et al., 2010) while the inward sodium ENac channel is inhibited. This leads to hyperpolarization of the plasma membrane and in turn activates voltage- and pH-dependent channels. Among those, the CatSPER channel (cation channel sperm associated) is a multiprotein complex, which exclusively locates to the plasma membrane of the principal piece in human and mouse sperm flagellum, and mediates Ca2<sup>+</sup> influxes (**Figure 4**; see review Singh and Rajender, 2015). In consistence with their restricted expression and function during capacitation, mutations in some genes encoding for some of the above ionic transporters (CatSPER1, CatSPER2, SLC26A3, and SLC26A8) were associated with male infertility due to asthenozoospermia, a pathology defined by reduced or absence of sperm motility (Hildebrand et al., 2010; Ray et al., 2017; Wedenoja et al., 2017).

## SLC26 PROTEINS IN SPERM FUNCTION AND MALE FERTILITY

#### Overview of SLC26 Protein Family

The Solute carrier 26 (SLC26) members are evolutionary conserved transmembrane proteins that mediate the transport of various anions including Cl<sup>−</sup> (chloride), HCO<sup>3</sup> <sup>−</sup> (bicarbonate), SO<sup>4</sup> <sup>2</sup><sup>−</sup> (sulfate), iodide (I−), formate (HCOO−) and C2O<sup>4</sup> 2− (oxalate), and contribute to the composition and the pH of secreted fluids in the body (Alper and Sharma, 2013). SLC26 belong to the highly conserved superfamily of amino acid-polyamine-organocation (APC) transporters and SLC26 related proteins are present in various organisms including bacteria, yeast, algae, plants (SulP/Sultr proteins) and nonmammalian vertebrates. In mammals, 10 members (SLC26A1 to

SLC26A11; SLC26A10 being a pseudogene) have been identified, and are expressed throughout the body with organ-specific distribution (Alper and Sharma, 2013; **Table 1**). SLC26 proteins mainly function as secondary anion transporters (ion-coupled transporters), utilizing the electrochemical gradient of an ion to drive the transport of another solute against its gradient. Some of them also function as uncoupled electrogenic transporters similar to Cl<sup>−</sup> channels (SLC26A7, A9) (Ohana et al., 2011; Alper and Sharma, 2013; **Table 1**). A few exception are to be mentioned: first, in mammals, no proper anion transport activity was reported for SLC26A5 (Prestin) (in contrast to chicken, zebra fish and insects) (Schaechinger and Oliver, 2007; Hirata et al., 2012) and SLC26A5 is supposed to act as a motor protein and to control outer hair cells of the cochlea in an anion-dependent manner (Zheng et al., 2000; Rybalchenko and Santos-Sacchi, 2008); second, the activity of SLC26A8 and SLC26A11, has been poorly investigated, and it is therefore difficult to precisely state on their anion specificity and mode of transport.

SLC26 proteins share a common structure, including a highly conserved transmembrane region with 10 to 14 spans, supporting the anion transport activity, and a cytoplasmic region, which comprises the STAS domain (Sulfate Transporter and Anti-Sigma factor antagonist) involved in SLC26 trafficking (Sharma et al., 2011; Bai et al., 2016), protein-protein interaction and regulation (for reviews see Ohana et al., 2011; Alper and Sharma, 2013; **Figure 5**). Several SLC26 members also carry a PDZ binding domain at their carboxy-terminal extremity (for reviews see Ohana et al., 2011; Alper and Sharma, 2013). Interestingly, most SLC26 members were shown to physically interact with the Cystic Fibrosis Transmembrane conductance Regulator channel (CFTR; MIM 602421) via their STAS domain, and to stimulate the CFTR Cl<sup>−</sup> channel activity in various epithelia (Ko et al., 2004; Khouri and Toure, 2014; **Figure 5**). Such physical and functional cooperation highlights the crosstalks which are likely to exist between SLC26 and other ionic transporters. Interestingly, while SlC26 proteins were initially thought to function as monomers, some biochemical studies indicate that they can form homo- and hetero-dimers, providing an additional level of complexity in their mode of function and regulation (Chavez et al., 2012). Studies of the bacterial YeSLC26A2 protein indicated that homodimerization is supported by the transmembrane core and not by the cytoplasmic STAS domain (Compton et al., 2011) and recent work performed by Chang et al. (2019) achieved structural modeling of the membrane-embedded prokaryotic SLC26 dimer (SLC26Dg, Deinococcus Geothermalis. In mammals, such transmembrane

homodimerization property was also described for SLC26A5 (Prestin) (Liu et al., 2003; Compton et al., 2011).

SLC26 proteins constitute one of the main classes of transporters that are involved in HCO<sup>3</sup> <sup>−</sup> and pH homeostasis regulation; the other being SLC4, and HCO<sup>3</sup> <sup>−</sup> transporters (see review Bernardino et al., 2019). In humans, their importance in maintaining correct ionic equilibrium and pH in various tissues and differentiation processes is demonstrated by the identification of SLC26 "loss of function" mutations in several hereditary genetic diseases: nephrolithiasis (SLC26A1), diastrophic dysplasia (SLC26A2), chloride loosing diarrhea (SLC26A3), Pendred syndrome -deafness and goiter- (SLC26A4), non-syndromic deafness (SLC26A5) and in men with reduced fertility and asthenozoospermia (SLC26A3, SLC26A8) (Everett and Green, 1999; Dawson and Markovich, 2005; El Khouri and Toure, 2014; Seidler and Nikolovska, 2019; **Table 1**). All the above phenotypes are in line with the nearly restricted tissue expression profiles observed for most of SLC26 genes. Notably, mutant mouse models have been generated for all SLC26 members, and all reproduced the clinical features of the SLC26 human-related diseases when applicable (Liberman et al., 2002; Forlino et al., 2005; Schweinfest et al., 2006; Touré et al., 2007; Dallos et al., 2008; Dror et al., 2010; El Khouri et al., 2018). In addition, studies of members not so far associated with human diseases (SLC26A7 and A11) revealed their functions in various tissues such as kidney, gastro intestinal tract, enamel, vestibular membrane of the cochlea, and brain (Xu et al., 2009; Rahmati et al., 2013, 2016; Kim et al., 2014; Yin et al., 2017), predicting that further investigations might lead to the identification of novel SLC26 gene mutations associated with pathophysiological conditions in humans (**Table 1**). Lastly SLC26 function may be critical for cystic fibrosis condition (CF; MIM 219700), a disease which is due to mutations in CFTR, and characterized by general defective electrolyte transport, chronic lung infections and inflammation, respiratory failure, digestive symptoms and male infertility (i.e., congenital bilateral absence of the vas deferens). Hence, genetic variants in SLC26A9, impairing the established cross-talk and interaction with CFTR contribute to the severity of respiratory and gastrointestinal symptoms observed in cystic fibrosis (see review El Khouri and Toure, 2014).

## SLC26 and CFTR Protein Functions in Sperm Cells and Epididymis

Among SLC26 proteins, SLC26A3, A6, and A8 were reported to locate to the human and mouse sperm and their functions were investigated through a range of cellular, biochemical and electrophysiological approaches (Chan and Sun, 2014; El Khouri and Toure, 2014). In addition, the generation and availability of knock out mouse models for all three proteins permitted to investigate in vivo their function and confirm some of the findings. Herein we will describe SLC26 protein functions in the sperm and epididymal cells, following their chronological order

#### TABLE 1 |Principal features of SLC26 family members.


(Continued)

#### TABLE 1 | Continued


The table displays the tissues were SLC26 genes and proteins are mainly detected in humans and mice, together with their anion specificity and their mode of transport, when defined. The clinical signs of the human genetic diseases associated with SLC26 mutations and the phenotype of the corresponding mutant mouse models are also described. References for information summarized in the table are indicated. In blue are highlighted all the features in regard with male reproductive organs and this review.

of discovery; in addition, as the CFTR channel was identified as a main interactor of SLC26 proteins, we will also describe the current knowledge about CFTR function in the sperm cells and male reproductive tract (**Table 2**).

#### SLC26A8 Function in Sperm Cells

fcell-07-00230 October 16, 2019 Time: 17:33 # 11

SLC26A8 (also called Testis Anion transporter 1) was cloned almost concomitantly by Touré et al. in 2001 (Toure et al., 2001) and Lohi et al. (2002), and reported to be exclusively expressed in human testis and in the male germ cells (Toure et al., 2001; Lohi et al., 2002). It was shown to interact with MgcRacGAP, a regulator of small Rho GTPases, later identified to be required for cytokinesis, in somatic and germ cells (Maddox and Oegema, 2003; Lores et al., 2014). SLC26A8 was the first member to be investigated in male reproductive functions as the remarkable tissue-specificity suggested that it might fulfill critical function in the sperm cells. As mentioned above, the anion transport activity of SLC26A8 has been poorly investigated. First studies from Touré et al. (2001) indicated a Cl−-dependent SO<sup>4</sup> 2− transport when the protein was expressed in COS cells, suggesting that SLC26A8 might function as a coupled ion transporter. To date the activity of SLC26A8 toward the HCO<sup>3</sup> <sup>−</sup>, which physiological relevance in sperm cell function is established, has not been reported. Ubiquitous invalidation of Slc26a8 gene was performed in the mouse by homologous recombination, and resulted in male sterility due to total sperm immotility while viability was unaffected. Functional analysis of Slc26a8 null sperm indicated reduced ATP consumption and the absence of capacitation-associated protein phosphorylation (Touré et al., 2007). SLC26A8-null sperm also displayed structural defects of the annulus, which induced a hairpin bending of the flagellum (Touré et al., 2007). In line with this phenotype, the SLC26A8 protein was found to locate at the annulus and equatorial segment of mouse and human sperm (Touré et al., 2007; Lhuillier et al., 2008; Rode et al., 2012).

#### CFTR Function in Sperm Cells

In the same time, publications from different laboratories indicated the expression and functions of the CFTR channel during sperm capacitation. In addition to its expression in the respiratory, digestive and genital epithelia, the CFTR channel was shown to be expressed in mature sperm from mice, guinea pigs and humans where it contributes to Cl<sup>−</sup> and HCO<sup>3</sup> <sup>−</sup> fluxes during capacitation (see review Touré, 2017). Hence in 2007, Xu and coworkers identified the CFTR channel at the equatorial segment of the sperm head in both mouse and human sperm. They showed that sperm treatment with a selective inhibitor of CFTR (CFTRinh-172), prevents the increases in intracellular cAMP, and pH and the membrane hyperpolarization, which are required for proper capacitation and acrosome reaction. The study of a heterozygous mouse model for CF (CFTRtm1Unc) also indicated low fertilization capacity, with impaired sperm motility and capacitation (Xu et al., 2007). Hernández-González et al. (2007) concomitantly reported the expression of CFTR in mouse and human sperm but found it to be restricted to the midpiece of the flagella; they showed that CFTR is required for membrane potential hyperpolarization during capacitation by regulating the epithelial sodium channels (ENaC). Those findings were categorically confirmed by work from Figueiras-Fierro et al. (2013) who specified CFTR currents in mouse sperm cells by means of patch clamp measurements on wild-type vs. CFTR 1F508-null sperm and by using specific CFTR agonists and antagonists. More recently Puga Molina et al. (2017) also demonstrated that in human sperm, CFTR activity is required for capacitation-associated phosphorylation in a PKAdependent manner.

## SLC26A8 and CFTR Cooperation in Sperm Cells

Following the discovery of CFTR protein and activity in mature sperm, Rode et al. (2012) demonstrated that SLC26A8 interacts with the CFTR channel via the STAS domain; the molecular complex was also identified by immunoprecipitation on mouse testis protein extracts. By means of radioactive iodide efflux measurements in CHO-K1 cells and patch clamp experiments in Xenopus oocytes, they demonstrated that SLC26A8 stimulates the CFTR Cl<sup>−</sup> transport activity (Rode et al., 2012). They further studied the Slc26a8 knock out model and demonstrated that the absence of capacitation-associated protein phosphorylation and motility could be partially rescued when supplementing with cAMP permeant analogs. They also demonstrated that the soluble adenylate cyclase (sAC) relocates properly at the annulus of Slc26a8-null sperm, despite their structural defects. This indicated that overall, the motility and capacitation defects observed in SLC26A8-null sperm result from a functional deregulation; therefore in fine SLC26A8 localization at the annulus may be conciliated with the regulation of anion fluxes (Rode et al., 2012). Overall these data suggested that SLC26A8 actively and/or indirectly contribute to Cl<sup>−</sup> and HCO<sup>3</sup> <sup>−</sup> influxes required for the activation of c-AMP-dependent protein phosphorylation during capacitation.

#### SLC26A3 Function in Sperm Cells

SLC26A3, also called Down Regulated in Adenoma (DRA), was cloned in 1993 by Schweinfest et al. (1993) and later showed to encode for an intestinal anion transport molecule (Silberg et al., 1995) whose mutations lead to congenital Chloride Loosing Diarrhea (CLD; [MIM 214700]) (Hoglund et al., 1996), an autosomal recessive disorder due to defective intestinal electrolyte absorption (Kere et al., 1999; Wedenoja et al., 2011). SLC26A3 acts as a Cl−/HCO<sup>3</sup> <sup>−</sup> exchanger (PMID 10428871) with a 1:1 or 2:1 stoichiometry (see review Seidler and Nikolovska, 2019) and is principally expressed in the enterocytes of the gastrointestinal tract epithelium, in humans and rodents (Jacob et al., 2002), where it is responsible for high HCO<sup>3</sup> <sup>−</sup> output rates in the mid to distal region of the colon (Xiao et al., 2012, 2014). Chen et al. (2009) reported for the first time the expression of SLC26A3 in sperm cells from guinea pig. They showed that the protein locates to the equatorial segment and colocalizes with the CFTR channel. Importantly they demonstrated that Cl<sup>−</sup> was required for the HCO<sup>3</sup> <sup>−</sup>-dependent changes occurring during capacitation (pH and cAMP rise, protein phosphorylation); they proposed a cooperative model involving SLC26A3 for


TABLE 2 | Principal characteristicsand functions of SLC26A3, A6, A8, and CFTR in the sperm and epididymal cells.

The table summarizes the tissues within the male reproductive organs where SLC26A3, A6, A8, and CFTR genes are expressed and the detection of the corresponding proteins in sperm and epididymal cells. The sperm and epididymal phenotypes in the mouse knock out mouse models are reported together with the phenotype associated with mutations in those genes in humans. References for information summarized in the table are indicated.

HCO<sup>3</sup> <sup>−</sup> entry (in exchange of Cl<sup>−</sup> efflux) and CFTR for Cl<sup>−</sup> recycling pathway. In 2012, Chavez et al. (2012) reported the expression of SLC26A3 transcripts in mouse spermatogenic cells (spermatocytes, spermatids) and found SLC26A3 protein to be restricted to the sperm flagellum midpiece, similar to the CFTR channel. Chavez et al. (2012) also conducted a series of in vitro measurements on mouse epididymal sperm, using MQAE and DISC3, two fluorescent probes reflecting the intracellular Cl<sup>−</sup> content and membrane hyperpolarization, respectively. These assays were performed in presence or absence of a set of anion transport antagonists: the CFTRinh172, which specifically targets CFTR, or Tdap and 5099, which presumably target SLC26A3 (to date, their specificity and selectivity among SLC26 members were not proven). From those data, Chavez et al. (2012) concluded that both SLC26A3 and CFTR are involved in regulating the Cl<sup>−</sup> influx and HCO<sup>3</sup> <sup>−</sup>-induced hyperpolarization upon capacitation.

## SLC26A6 Function in Sperm Cells

In the course of their study, Chavez et al. (2012) similarly analyzed SLC26A6, also called PAT-1, CFEX. SLC26A6 is highly expressed in pancreas, kidney (Lohi et al., 2000; Waldegger et al., 2001) and in the intestine (Wang et al., 2002) where it functions as a Cl−/HCO<sup>3</sup> <sup>−</sup> exchanger, both in mouse and in humans; although the stoichiometry seems to differ between the two species as an electroneutral exchange was measured in human cells it is electrogenic in the mouse (Cl−/HCO<sup>3</sup> <sup>−</sup>, 1:2) (Seidler and Nikolovska, 2019). Chavez et al. (2012) demonstrated that in sperm cells, SLC26A6 co-localized with SLC26A3 and CFTR to the sperm flagellum midpiece. They also demonstrated by co-immunoprecipitation that, in vivo, SLC26A6 is part of the SLC26A3/CFTR protein complex. However, in vitro, the use of DOG and PMA, two compounds that are described to inhibit SLC26A6 through PKC activation, did not alter Cl<sup>−</sup> influx nor membrane hyperpolarization upon capacitation, suggesting that SLC26A6, in contrast to SLC26A3, is not critical for those processes (Chavez et al., 2012). Taken together the above studies indicate that in the sperm cells, the SLC26A3 and A8 proteins locate to the flagellum midpiece and annulus, respectively, and cooperate with the Cl<sup>−</sup> CFTR channel in regulating Cl<sup>−</sup> influx, membrane hyperpolarization and protein phosphorylation during capacitation. In addition, SLC26A3, A8 and CFTR are likely to colocalized at the equatorial segment of mouse and human sperm and to control acrosomal reaction. In mouse, although shown to locate to the sperm cells, the SLC26A6 protein seems dispensable for capacitation.

## SLC26A3 Function in Epididymal Cells

Last year, El Khouri et al. (2018) investigated in vivo, the function of SLC26A3 and SLC26A6 proteins, in sperm functionality by analyzing ubiquitous knock out mouse models previously generated to study their functions in the gastrointestinal system (Singh et al., 2010; Xiao et al., 2012). They showed that in addition to the previously reported phenotype of congenital diarrhea (Xiao et al., 2012; Singh et al., 2013), SLC26A3−null mice displayed severe lesions and abnormal cytoarchitecture of the cauda epididymis, which strongly impacted the reserve of sperm cells in epididymides (El Khouri et al., 2018). This phenotype is in line with the subfertility previously observed for a few CLD affected men (Hoglund et al., 2006). SLC26A3−null mice showed a drastic reduction of the cauda size with reduced tube sections observed within the epididymis while the caput regions appeared overall not affected; in addition, the number

of the presence of granuloma and fibrosis was observed, indicative of an inflammatory context and a disruption of the blood−epididymal barrier. The limited number of sperm cells, which was produced, failed to swim and was not responsive to the induction of capacitation as protein phosphorylation could not be obtained when supplementing with HCO<sup>3</sup> <sup>−</sup> and Ca2+. Similar to what was observed for SLC26A8-null sperm, a partial rescue of protein hyperphosphorylation was obtained when adding cAMP permeant analog, indicating a failure to activate the soluble adenylate cyclase. In addition, Slc26a3-null sperm exhibited abnormal morphology with increased proportion of bent and coiled flagellum, and the proportion of reacted acrosome was also significantly increased. Importantly, similar analyses were performed in parallel, on SLC26A6 knock out mice and no epididymal, nor spermatic defects were observed. This indicated that although co-expressed with SLC26A3 in sperm and epididymal cells, in vivo, SLC26A6 is not critical for sperm production and functionality. This could be due to subtle differences in anion transport activity as different stoichiometry and ion specificity were reported for these two members (i.e., A3 and A6); in addition, their activities could also vary between the different tissues where they are expressed, due to tissue-specific partners/regulators.

In humans, SLC26A3, SLC26A6, and CFTR proteins were detected on the luminal border of the apical mitochondria-rich cells (AMRC) of the ductus epididymis (Hihnala et al., 2006; Kujala et al., 2007) while SLC26A8 was absent (Kujala et al., 2007). In the mouse, El Khouri et al. (2018) detected SLC26A3 and SLC26A6 transcripts in epididymes while SLC26A8 was absent. Ruan et al. (2012) detected CFTR protein in the PCs of the mouse cauda epididymis while it was absent in other epithelial cell types. CFTR protein was aslo recently detected on the apical membrane of mouse caput epididymis and in the smooth muscle myoid cells (Sharma and Hanukoglu, 2019). Overall, such expression patterns are in line with the observed epididymal phenotype in SLC26A3 knock out mice and suggests an impairment of electrolyte homeostasis, which may impact the pH and ionic content of the epididymal milieu and prevent sperm maturation. Supporting this hypothesis, an increased amount of V ATPase protein was observed in SLC26A3−null epididymis caput compared to wild type tissues, a compensatory mechanism reflecting HCO3- and pH deregulation within the luminal fluid (Shum et al., 2009, 2011; Breton et al., 2016). As observed in the context of the sperm cells, although present within the epididymal ductus, SLC26A6 function seems to be dispensable for epididymal cytoarchitecture and functionality. Lastly, SLC26A8 protein was found absent from the epididymal tract (Kujala et al., 2007; Wedenoja et al., 2017; El Khouri et al., 2018) and rationally no epididymal phenotype was observed in the knock-out mice (Touré et al., 2007).

Taken together those studies indicate that among the three SLC26 proteins investigated in the male reproductive organs, only SLC26A3 appears critical for epididymal luminal milieu and sperm maturation. SLC26A3 is likely to contribute to the cellular cross-talks regulating HCO<sup>3</sup> <sup>−</sup> fluxes and leading to luminal acidification. SLC26A3 function in epididymal electrolyte regulation probably relies on interaction and/or cooperation with CFTR channel, and the Na+/H<sup>+</sup> antiporter 3, NHE3, which is expressed in both the intestine and in the male reproductive system (Melvin et al., 1999; Lamprecht et al., 2002; Hihnala et al., 2006). Hence Wang et al. (2017) recently reported that NHE3−deficient mice display ultrastructural defects of the epididymis and the vas deferens, as well as significant reduction of CFTR protein levels in these structures, ultimately leading to male infertility. Such phenotype is comparable to that of Slc26a3-null mice and support the hypothesis of a multi−channel protein complex, CFTR/SLC26A3/NHE3, involved in electrolyte regulation and luminal acidification of the epididymal ductus.

#### SLC26 Dysfunctions in Human Male Infertility

Following the identification of SLC26A8, genetic investigations in human infertility were promptly initiated owing to its exclusive expression in the testis and the male germ cells. Hence in 2005, Mäkelä et al. screened a cohort of 83 men with oligo- and azoospermia but did not identified variants in SLC26A8 genes associated to this phenotype (Makela et al., 2005). In 2013, based on the phenotype of Slc26a8-null mice, Dirami et al. (2013) screened a cohort of 146 men consulting for infertility and displaying moderate asthenozoospermia. Asthenozoospermia is defined by a reduction or an absence of sperm motility (less that 32% of progressive sperm, following the values established by the World Health Organization) (Cooper et al., 2010) and is found in nearly 80% of infertile men (Curi et al., 2003). Dirami and collaborators identified three heterozygous missense variants, c.260G > A (p.Arg87Gln), c.2434G > A (p.Glu812Lys), and c.2860C > T (p.Arg954Cys), which they showed to be absent from control individuals and to impact the functional cooperation between SLC26A8 and CFTR. They demonstrated, in vitro, that while physical interaction was not altered, all identified variants conducted to reduced protein amounts of SLC26A8 and that of the associated CFTR channel. They showed that in vitro, SLC26A8 protein amounts could be restored to control levels by proteasome inhibition, indicating that the variants impacted protein stability, likely by inducing deleterious protein conformational changes (Dirami et al., 2013). These three mutations were identified at the heterozygous state, in contrast to mutations identified in other SLC26-related diseases, which all segregate following an autosomal recessive mode (Dawson and Markovich, 2005); this could be attributed to the fact that only men displaying asthenozoospermia of moderate severity were screened in this study.

Following a similar strategy, Wedenoja et al. published in 2017, the screening of a cohort of 283 asthenozoospermic men and the identification of the c.2062 G > C (p.Asp688His) heterozygous variant in SLC26A3, in 3.2% of the patients (Wedenoja et al., 2017). Analysis of the variant's frequency in Exac database indicated that it is enriched in the Finnish population. Furthermore, functional studies showed that the p.Asp688His variant did not impact SLC26A3 intrinsic Cl−/HCO<sup>3</sup> <sup>−</sup> exchange activity, nor its protein amount. The p.Asp688His variant, which locates to the STAS domain, was able to interact with the CFTR channel but failed to stimulate CFTR Cl<sup>−</sup> transport activity, in vitro, in Xenopus oocytes. Here again, the variant was identified at the heterozygous level, but in this case, this might be consistent with the fact that SLC26A3 homozygous loss of function induce Chloride Loosing Diarrhea and subfertility, a much more severe phenotype (Wedenoja et al., 2011).

#### DISCUSSION

fcell-07-00230 October 16, 2019 Time: 17:33 # 15

SLC26 constitute one of the largest family of membrane proteins but their functions have only been recently investigated, mainly in the gastrointestinal and renal tissues where they appear critical for electrolyte transport and pH regulation. The male and female genital tracts also rely on pH homeostasis but few investigations were performed in the reproductive organs. As described in this review, recent work was performed by means of electrophysiological and in vitro studies on human and mouse sperm, which permitted to describe the critical role of SLC26A3 and A8 in regulating the electrophysiological and biochemical changes occurring in the sperm cells during capacitation. In addition, the phenotypical characterization of Slc26 knock-out mouse models together with translational studies of human infertility conditions, permitted to confirm their physiological relevance for sperm fertilization potential; importantly, the requirement of SLC26A3 for proper peididymal structure and functions was uncovered.

One of the main research gaps to be overtaken concerns the delineation of each SLC26 member contribution, in those processes. In this regard, a comprehensive and comparative analysis of SLC26 cellular and subcellular expression patterns, within the male reproductive tract, is cruelly missing. Hence, except SLC26A8, for which a clear-cut expression pattern is available (i.e., exclusively sperm specific), the expression pattern of SLC26 members is still unclear and the observed protein localizations, as in the case of SLC26A3, were found to diverge between different laboratories. The development of high-throughput sequencing technologies and single cell analyses has facilitated the access to large public expression datasets and therefore constitutes one asset to further progress in this field. In particular, few expression databases dedicated to the reproductive tissues have been established: The ReproGenomics viewer (Darde et al., 2015, 2019) and The Mammalian Reproductive Genetics database. When analyzing three distinct RNAseq datasets from mouse purified germ cells, conflicting results were obtained as compared to the assumed expression pattern of SLC26 members in the sperm cells. Hence while SLC26A6 and A8 transcripts were clearly detected in the mouse germ cells from spermatogonia to spermatid stage (Gan et al., 2013; Green et al., 2018; Lukassen et al., 2018), SLC26A3 transcripts were not detected at all. The analyses of the same datasets also exclude any expression of SLC26A1, A4, A5, and A9 in the mouse germ cells while SLC26A2, A7, and A11 were detected. Similar analyses of epididymal expression dataset, through the Mammalian Reproductive Genetics database, indicated that SLC26A3 transcripts are readily detected in the mouse epididymis, which is coherent with transcript and protein detection in mouse and in human epididymides that were reported by distinct laboratories, (Hihnala et al., 2006; El Khouri et al., 2018).

The discrepancy regarding SLC26A3 expression in the germline really question the presence and the function of SLC26A3, if any, in the mouse sperm cells. These conflicting data result from the limited biochemical tools available to analyze SLC26 protein expression, and potentially from antibody cross-reaction with different SLC26 members. Importantly, the situation is very different in humans as analysis of RNAseq data indicated that SLC26A3 transcripts are detected in human differentiating germ cells, as opposed to the mouse germ cells. This clearly alerts about the risk of generalizing data obtained from one specie to another. The differences in cytoarchitecture and compartmentations of the epididymis ductus between humans and mouse together with the difference in the sperm "status" used in studies (epididymal sperm in mouse vs. ejaculated sperm in human) constitute additional arguments, if required, to prohibit data transfer from one specie to another.

Considering the profound epididymal defects reported in the Slc26a3 knock out mouse model, an important connected point is to determine whether the dysfunctions observed in Slc26a3-null sperm when performing in vitro capacitation, could result from defects initially occurring during epididymal maturation. Hence it is highly probable that functional defects occurring within the epididymis could impair sperm "priming" and later prevent proper response to capacitation in the female genital tract. This point is particularly important to address, considering the uncertain expression of SLC26A3 in the sperm cells. The discrimination between those two intricated processes (epididymal priming and maturation vs. capacitation in the female genital tract) constitutes a requisite if one wants to better define sperm post-testicular maturation events; in the future this may be performed by generating conditional mutant mouse models with gene invalidation restricted to the epididymal epithelium or to the germ cell lineage.

Lastly, an important progress concerns the development of pharmacological compounds specifically targeting SLC26 proteins. To date some compounds such as Tenidap, 5099, DOG and PMA, were utilized, in vitro, to inhibit SLC26 protein functions in the sperm cells but their inhibitory mechanisms are sometimes indirect or unknown; in addition, no information is available regarding their selectivity among SLC26 members, which may limit the interpretation of the data and even lead to confusion in specifying the contribution of each member. A major work in this field was recently published by Haggie et al. (2018) who identified SLC26A3 specific antagonists with the aim of treating gastrointestinal defects. The authors performed a fluorescent high-throughput screening based on the Cl−/I<sup>−</sup> exchange activity mediated by SLC26A3 and a halide-sensitive yellow fluorescent probe (YFP). The strength of this study relies on the demonstrated selectivity of the identified compounds by means of various and complementary approaches: in vitro study on different representative members of the SLC26 family (human and mouse orthologs) together with in vivo studies using a mouse model with gastrointestinal defects resulting from SLC26A3 dysfunction. In the future, the use of such compounds will definitely help to better understand SLC26 functions and molecular mechanisms in the processes of sperm posttesticular maturation and fertilization potential. The study of other SLC26 members expressed in the epididymal and sperm cells, such as SLC26A2, A7 and A11, will also provide additional information regarding the multiple cross-talks and hierarchical regulatory mechanisms between SLC26 and other ion transporters.

#### AUTHOR CONTRIBUTIONS

fcell-07-00230 October 16, 2019 Time: 17:33 # 16

AT performed the request bibliography analyses and wrote the manuscript.

#### REFERENCES


Austin, C. R. (1952). The capacitation of the mammalian sperm. Nature 170:326.


#### FUNDING

This work was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), the Centre National de la Recherche Scientifique (CNRS), the Université de Paris, and the French National Research Agency (Grants: MUCOFERTIL ANR-12-BSV1-0011-01, MASFLAGELLA ANR-14-CE15-0002- 03, and DIVERCIL ANR-17-CE13-0023-02).

#### ACKNOWLEDGMENTS

The author is grateful to Elma El Khouri for help in illustrations together with Marjorie Whitfield and Pierre F. Ray for critical reading of the manuscript.


Breton, S. (2001). The cellular physiology of carbonic anhydrases. JOP 2, 159–164.


SLC9A3R1 in mouse sperm capacitation. Biol. Reprod. 86, 1–14. doi: 10.1095/ biolreprod.111.094037





55, 125–135. doi: 10.1002/(SICI)1098-2795(200002)55:2<125::AID-MRD1>3. 0.CO;2-Q




**Conflict of Interest:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Touré. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Transient Sperm Starvation Improves the Outcome of Assisted Reproductive Technologies

Felipe A. Navarrete<sup>1</sup> , Luis Aguila<sup>1</sup> , David Martin-Hidalgo1,2, Darya A. Tourzani<sup>1</sup> , Guillermina M. Luque<sup>3</sup> , Goli Ardestani<sup>1</sup> , Francisco A. Garcia-Vazquez4,5, Lonny R. Levin<sup>6</sup> , Jochen Buck<sup>6</sup> , Alberto Darszon<sup>7</sup> , Mariano G. Buffone<sup>3</sup> , Jesse Mager<sup>1</sup> , Rafael A. Fissore<sup>1</sup> , Ana M. Salicioni<sup>1</sup> , María G. Gervasi<sup>1</sup> \* and Pablo E. Visconti<sup>1</sup> \*

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Jean-Ju Lucia Chung, Yale School of Medicine, United States Polina V. Lishko, University of California, Berkeley, United States Deborah A. O'Brien, The University of North Carolina at Chapel Hill, United States

#### \*Correspondence:

María G. Gervasi mariagracia@vasci.umass.edu Pablo E. Visconti pvisconti@vasci.umass.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

> Received: 22 August 2019 Accepted: 17 October 2019 Published: 05 November 2019

#### Citation:

Navarrete FA, Aguila L, Martin-Hidalgo D, Tourzani DA, Luque GM, Ardestani G, Garcia-Vazquez FA, Levin LR, Buck J, Darszon A, Buffone MG, Mager J, Fissore RA, Salicioni AM, Gervasi MG and Visconti PE (2019) Transient Sperm Starvation Improves the Outcome of Assisted Reproductive Technologies. Front. Cell Dev. Biol. 7:262. doi: 10.3389/fcell.2019.00262 <sup>1</sup> Department of Veterinary and Animal Sciences, Integrated Sciences Building, University of Massachusetts Amherst, Amherst, MA, United States, <sup>2</sup> Research Group of Intracellular Signaling and Technology of Reproduction, Institute of Biotechnology in Agriculture and Livestock (INBIO G + C), University of Extremadura, Cáceres, Spain, <sup>3</sup> Instituto de Biología y Medicina Experimental, Consejo Nacional de Investigaciones Científicas y Técnicas, Buenos Aires, Argentina, <sup>4</sup> Department of Physiology, Veterinary School, International Excellence Campus for Higher Education and Research, University of Murcia, Murcia, Spain, <sup>5</sup> Institute for Biomedical Research of Murcia, IMIB-Arrixaca, Murcia, Spain, <sup>6</sup> Department of Pharmacology, Weill Cornell Medical College, Cornell University, New York, NY, United States, <sup>7</sup> Departamento de Genética del Desarrollo y Fisiología Molecular, Instituto de Biotecnología, Universidad Nacional Autónoma de México, Cuernavaca, Mexico

To become fertile, mammalian sperm must undergo a series of biochemical and physiological changes known as capacitation. These changes involve crosstalk between metabolic and signaling pathways and can be recapitulated in vitro. In this work, sperm were incubated in the absence of exogenous nutrients (starved) until they were no longer able to move. Once immotile, energy substrates were added back to the media and sperm motility was rescued. Following rescue, a significantly higher percentage of starved sperm attained hyperactivated motility and displayed increased ability to fertilize in vitro when compared with sperm persistently incubated in standard capacitation media. Remarkably, the effects of this treatment continue beyond fertilization as starved and rescued sperm promoted higher rates of embryo development, and once transferred to pseudo-pregnant females, blastocysts derived from treated sperm produced significantly more pups. In addition, the starvation and rescue protocol increased fertilization and embryo development rates in sperm from a severely subfertile mouse model, and when combined with temporal increase in Ca2<sup>+</sup> ion levels, this methodology significantly improved fertilization and embryo development rates in sperm of sterile CatSper1 KO mice model. Intracytoplasmic sperm injection (ICSI) does not work in the agriculturally relevant bovine system. Here, we show that transient nutrient starvation of bovine sperm significantly enhanced ICSI success in this species. These data reveal that the conditions under which sperm are treated impact postfertilization development and suggest that this "starvation and rescue method" can be used to improve assisted reproductive technologies (ARTs) in other mammalian species, including humans.

Keywords: sperm, blastocyst, embryo transfer, IVF, ICSI, capacitation

## INTRODUCTION

fcell-07-00262 November 1, 2019 Time: 17:32 # 2

Sperm capacitation is defined as all the biochemical and physiological events occurring after epididymal maturation that are necessary for the sperm to acquire fertilizing capacity (Yanagimachi, 1994; Gervasi and Visconti, 2016). Following this definition, capacitation ends once the sperm fuses with a metaphase II (MII)-arrested egg to become a zygote. The zygote follows a series of complex cell biological processes leading to the first cleavage, continuing with pre-implantation development and differentiation, implantation, and finally, birth of an offspring (White et al., 2018). Among these processes, capacitation, fertilization, and early embryo development can be mimicked in vitro in defined media (Ventura-Junca et al., 2015). Fertilization can also be achieved by intracytoplasmic sperm injection (ICSI) effectively bypassing many physiological events preceding gamete fusion (Palermo et al., 2017). These techniques are well established in many mammalian species including humans. However, the efficiency of these methods is species-specific, and success depends on other factors including the age of male and female gametes, environmental factors, and genetic background (Meldrum et al., 2016). Although many problems in capacitation can be overcome using ICSI, obtaining good quality preimplantation embryos at the blastocyst stage is a major limiting factor of successful pregnancies (Sadeghi, 2017).

Sperm capacitation depends upon crosstalk between metabolic and signaling pathways (Goodson et al., 2012). In mouse sperm, changes in motility pattern associated with capacitation (i.e., hyperactivation) are based on ATP produced by glycolysis (Miki et al., 2004). However, it is also clear that the Krebs cycle and oxidative phosphorylation are active in these cells (Goodson et al., 2012) and mammalian sperm capacitation media contain fuels for both glycolysis (i.e., glucose) and oxidative phosphorylation (i.e., pyruvate). To explore the specific role of these fuels for capacitation, initially, we incubated mouse sperm in media devoid of all nutrients (starvation step). Once sperm became motionless, defined energy molecules (e.g., glucose and pyruvate) were added back (rescue step). We hypothesized that after energy depletion, adding back nutrients to the sperm incubation media would induce recovery of sperm functionality. Indeed, the recovery was observed; however, the starvation and rescue protocol significantly increased the percentage of sperm achieving hyperactivated motility and also enhanced in vitro fertilization (IVF) rates when compared with sperm persistently incubated in standard capacitation media containing glucose and pyruvate. Also unexpectedly, the embryos fertilized with previously starved and recovered sperm were more efficient in reaching the blastocyst stage; and these blastocysts produced approximately three times more pups when transferred to pseudo-pregnant females.

The need for capacitation is often overcome by the use of ICSI. Although this procedure is highly successful in mice, ICSI with bovine sperm presents severe challenges and only succeeds when the egg is artificially activated (Catt and Rhodes, 1995; Malcuit et al., 2006). We now show that the starvation and rescue protocol improves ICSI success rates in the bovine system. Together, these results indicate that sperm incubation conditions can improve the outcome of embryo development and that capacitationassociated changes occurring in sperm before fertilization are still relevant for post-fertilization events. The molecular basis of these changes is still not known and warrants future investigation. Some of the hypotheses regarding these mechanisms will be briefly mentioned in the section "Discussion."

## RESULTS

Sperm from CD1 mice incubated in TYH media lacking glucose and pyruvate stopped moving after ∼40 min (**Supplementary Video S1**). We reasoned that these sperm had ceased swimming because they had consumed all available stored nutrients, and we refer to them as "starved." Once immotile, glucose and pyruvate were added back to reach standard concentrations in TYH media (5.56 mM glucose and 0.5 mM pyruvate, **Supplementary Video S1**). For the purpose of comparison, parallel control incubations were done resuspending sperm in standard TYH capacitating conditions persistently incubated in the presence of nutrients. Counter-intuitively, a higher percentage of starved and capacitated sperm were motile (**Figure 1A**) and became hyperactivated (**Figure 1B**) after addition of nutrients compared to control sperm persistently incubated in the presence of nutrients in standard TYH capacitating conditions. Hereafter, we refer to this procedure, in which sperm energy is recovered after starvation, using the acronym SER.

Considering that SER produced a higher percentage of hyperactivated sperm, we hypothesized that SER-treated sperm would also improve fertilization rates. IVF using sperm from CD1 mice is highly efficient (over 90%). Therefore, to test the effectiveness of SER in IVF and other assisted reproductive technologies (ARTs), we switched to using sperm from inbred C57BL/6J which have lower IVF and embryo development rates (Liu et al., 2009). C57BL/6J sperm were capacitated in standard TYH media with BSA and HCO<sup>3</sup> <sup>−</sup> (CAP) containing nutrients or incubated in nutrient-free CAP TYH media until becoming immotile and then rescued by addition of glucose and pyruvate (SER). Similar to sperm from CD1 mice, SER increased motility (**Figure 1C**) and hyperactivation rates (**Figure 1D**) of C57BL/6J sperm. In addition, SER-treated C57BL/6J sperm yielded more eggs reaching the two-cell stage when incubated with heterologous CD1 MII-arrested eggs (**Figure 2A**; n = 38) as well as when incubated with homologous MII-arrested eggs from C57BL/6J females (**Figure 2D**; n = 15). Representative images of two-cell stage embryos are displayed in **Figure 2G**.

More surprising than the improvement observed in fertilization rates, in both homologous and heterologous conditions, a significantly higher percentage of two-cell embryos reached blastocyst stage (**Figures 2B,E** blastocyst % from twocell). Representative images of blastocyst development are shown in **Figure 2H**. To compare the efficiency of the treatments, these results were also presented as the percentage of the initial number of eggs reaching blastocyst stage (**Figures 2C,F**, blastocyst % from initial number of eggs). For more information, these data are also presented in **Supplementary Figure S1** and tabulated

in **Supplementary Table S1** according to male mice age. To evaluate blastocyst developmental potential, in a fraction of the experiments described above (n = 19, n = 15 for heterologous CD1 female x C57BL/6J male IVF and n = 4 for homologous C57BL/6J female x C57BL/6J male), an equal number of blastocysts derived from CAP or SER-treated sperm were nonsurgically transferred to pseudo-pregnant females. Unexpectedly, the number of pups born from SER-treated sperm derived blastocysts was over three times higher than the number of pups born from control blastocysts (**Figure 3A** for heterologous IVF and **Supplementary Table S1** for heterologous and homologous IVF). The improvement in embryo development is reflected in the increased ratio of transferred blastocysts producing offspring (**Figure 3B** and **Supplementary Table S1**). In addition, blastocyst quality was also assessed by counting the total cell number present at the blastocyst stage (**Figure 3C**) as well as by blastocyst outgrowth (**Figure 3D**). These analyses revealed a higher number of cells and percentage of outgrowth of blastocysts derived from SER-treated sperm. Thus, these data indicate that transient nutrient withdrawal provides a qualitative improvement in the ability of sperm to produce embryos, and this effect is not simply due to increased ability to fertilize oocytes.

The aforementioned experiments indicate an overall better success in fertilization, embryo development, and pregnancy rates when SER-treated sperm are used. We next sought to determine if this methodology enhances IVF and embryo development rates in sub-fertile and sterile mouse models. Recently, we showed that sperm derived from FERDR/DR mice, in which the FER tyrosine kinase is replaced by a non-active mutant, are severely sub-fertile in vitro (Alvau et al., 2016). When compared to sperm incubated in standard capacitation media, SER-treated FERDR/DR sperm significantly increased fertilization rates (**Figure 4A**) and embryo development (**Figures 4B,C**). When these embryos were transferred to pseudo-pregnant females, live pups were born (**Figure 4D**). We next tested the SER treatment on sperm from CatSper1 KO mice, which are sterile (Ren et al., 2001). We recently found that a short incubation of CatSper1 KO sperm with Ca2<sup>+</sup> ionophore A<sup>23187</sup> induced fertilizing capacity in these normally infertile sperm (Navarrete et al., 2016). SER treatment alone did not rescue the fertilizing capacity of CatSper1 KO sperm. However, the two treatments proved to be synergistic; sequential treatment of CatSper1 KO sperm with SER and A<sup>23187</sup> significantly increased fertilization (**Figure 5A**) and embryo development (**Figure 5B**) rates when

compared with A<sup>23187</sup> alone. When blastocysts (**Figure 5C**) from these experiments were transferred to pseudo-pregnant females, live pups were born (Table in **Figure 5D**).

While ICSI in the mouse is quite successful, the efficiency of this technique in bovines is very poor when compared to other species (Salamone et al., 2017). Therefore, we used frozen bovine sperm and in vitro matured bovine oocytes to test if SER can also improve ICSI success rates. Contrary to mouse sperm, bovine sperm motility was reduced but did not completely stop during incubation in the absence of exogenous energy sources. Bovine capacitation media normally contain pyruvate and lactate but do not contain glucose; therefore, after 3 h in the absence (starvation step) of nutrients, pyruvate and lactate were added back, and normal motility was restored. Individual bovine sperm were injected into oocytes (ICSI) and the number of zygotes undergoing cleavage divisions was evaluated. The percentage of two-cell embryos was increased over threefold in SER-treated bovine sperm relative to controls persistently

embryo transfer. A fraction of the blastocysts derived from CAP or SER treatments as described in Figure 2 were used to evaluate the impact of these treatments in blastocyst potential. (A) Blastocysts derived from CAP- or SER-treated sperm were transferred to pseudo-pregnant recipient female mice. The total number of pups obtained from either CAP or SER treatment (n = 15) is indicated. Image shown is representative of a litter obtained after transfer of SER-derived embryos. (B) Percentage of pups born per number of embryos transferred is indicated. (C) Blastocysts derived from CAP- or SER-treated sperm were stained with Hoechst as indicated in the section "Materials and Methods." The number of stained nuclei that represents the number of cells in each blastocyst was counted (n = 3, 35 blastocysts/treatment counted). Representative images of stained embryos are shown. Scale bar = 25 µm. (D) Blastocyst derived from CAP or SER-treated sperm were assayed for outgrowth in vitro as explained in the section "Materials and Methods" (n = 10, 205 and 210 blastocysts counted for CAP and SER-derived blastocysts, respectively). In each panel, significant differences are indicated as: <sup>∗</sup>p < 0.05; ∗∗p < 0.01, and ∗∗∗p < 0.001. Experiments represented in this figure were done using blastocysts obtained by heterologous fertilization.

incubated in the presence of nutrients. Importantly, almost 17% of these two-cell embryos developed into blastocysts compared to zero blastocysts obtained from sperm incubated in standard capacitation conditions (**Table 1**).

#### DISCUSSION

Infertility and sub-fertility are critical health problems with social and economic consequences. About 9% of couples worldwide present signs of infertility. Since the first successful "Test-Tube" baby in 1978, over 5 million babies were born using ARTs. ARTs include IVF, artificial insemination (AI), ICSI, in vitro maturation of oocytes, superovulation, cryopreservation, and embryo transfer techniques. In addition to their use in human clinical application, ARTs are also used for agricultural domestic species and for production and maintenance of valuable laboratory animals which often present reproductive problems (Behringer et al., 2014). Despite their frequent use, ART methods are usually expensive, time consuming, and can be plagued by limited successful pregnancy rates (de Mouzon et al., 2010; Mantikou et al., 2013). At the basic research level, our group has recently shown that we can induce in vitro fertilizing capacity in sperm from sterile knock-out genetic models using a short incubation with the Ca2<sup>+</sup> ionophore A<sup>23187</sup> (Navarrete et al., 2016). Interestingly, short treatment with A<sup>23187</sup> also increased fertilization and embryo development rates of control inbred C57BL/6J mice sperm (Navarrete et al., 2016) suggesting that embryo development can be affected by differential sperm incubation conditions during capacitation.

Capacitation can be achieved in vitro in defined media containing ions, a cholesterol binding source, and energy substrates (reviewed in Gervasi and Visconti (2016)). In the mouse, capacitation and IVF can be achieved with TYH medium containing only glucose and pyruvate as energy compounds (Toyoda and Yokoyama, 2016). Here, we observed that mouse sperm incubated in the absence of these exogenous nutrients become immotile in less than 40 min and motility can be rescued with the addition of glucose and pyruvate. Once rescued, starved sperm achieved higher percentages of hyperactivation and IVF rates. We do not yet know the molecular basis for this improvement in sperm functional parameters. Two alternative possibilities to explain these results are that due to the need for energy sources molecules limiting capacitation competence are consumed during the starvation stage; or that during rescue, molecules supporting capacitation are upregulated. Another possibility is that the starvation followed by rescue protocol induces changes in metabolites, supporting an increase of certain metabolites that could be beneficial for the development of sperm hyperactivated motility and fertilization. Finally, we cannot discard changes in the production of radical oxygen species (ROS) which have been associated with deleterious effects on sperm motility as well as with beneficial effects on capacitation (de Lamirande and Gagnon, 1995; Aitken, 2017). Further investigations are being performed to test these hypotheses.

Unexpectedly, eggs fertilized with SER-treated sperm showed significantly greater percentage of two-cell embryos reaching the blastocyst stage. Moreover, at 3.5 days post-fertilization, blastocysts derived from SER-treated sperm had a higher percentage of outgrowth and higher number of total cells, parameters that are indicative of good implantation potential (Binder et al., 2015). Furthermore, when transferred to pseudopregnant females, blastocysts derived from SER-treated sperm gave rise to a higher number of pups than those obtained with sperm incubated in standard capacitation conditions. These experiments indicate that blastocysts derived from SERtreated sperm have advantages over those obtained from sperm incubated in standard capacitation media. In concordance with our findings, it has been proposed that sperm incubation conditions for IVF can affect embryonic development (Jaakma et al., 1997; Matas et al., 2003; Zheng et al., 2018). However, the molecular basis of sperm-derived post-fertilization effects remains unknown. We hypothesize that the post-fertilization effects observed after sperm SER treatment are related to epigenetic changes in the male gamete. Environmental factors can

influence epigenetic processes, change gene expression, and have a great impact on development. Epigenetic marks such as, DNA methylation, and histone tail modifications, show asymmetrical distribution between male and female gamete and play critical roles in embryo development (Marcho et al., 2015; Canovas et al., 2017). Recent evidence suggests that small non-coding RNAs in sperm have a role in non-genomic inheritance of paternal traits (Chen et al., 2016; Sharma et al., 2016; Conine et al., 2018). One intriguing possibility is that SER treatment can affect RNA levels including the pool of non-coding RNAs, which could impact post-fertilization development. On the other hand, it has been shown that the state of the sperm chromatin contains critical information for the regulation of gene expression after fertilization (Jung et al., 2017, 2019). One possibility is that SER treatment is inducing changes in the sperm chromatin prior fertilization, and that those changes have a direct impact in the development of the embryo. We are currently studying these possibilities.

Intracytoplasmic sperm injection is a technique by which sperm are directly introduced into the egg, bypassing the need for capacitation. In mice, ICSI is very successful, effectively rescuing several sperm sterile phenotypes (Kuretake et al., 1996; Yanagimachi et al., 2004; Ihara et al., 2005). In contrast, embryo development after ICSI is compromised in bovines, and successful development requires additional treatments with reagents that induce egg activation (Catt and Rhodes, 1995; Malcuit et al., 2006; Salamone et al., 2017). ICSI has also been achieved by pre-treating bovine sperm or eggs with diverse reagents (Rho et al., 1998; Aguila et al., 2017; Canel et al., 2017). At present, it is not clear why bovine sperm do not induce egg activation, but our data suggest that SER treatment of bovine sperm stimulates their ability to activate oocytes. One of the best characterized sperm contributions to early embryo development is the phospholipase C zeta (PLCζ) mediated induction of Ca2<sup>+</sup> oscillations (Saunders et al., 2002). We hypothesized that in bovine sperm, the PLCζ protein and its activity are closely guarded such that its release requires active nuclear remodeling. Consistently, our group has recently shown that injected bovine sperm are resistant to nuclear remodeling and decondensation by in vitro matured bovine oocytes and that treating these oocytes with ionomycin plus cycloheximide (CHX) to chemically induce acute activation and sperm head decondensation can partially overcome the poor success of this technique in this species (Aguila et al., 2017). In addition, it has been recently shown that the sperm-borne protein glutathione-S-transferase omega 2 facilitates nuclear decondensation after fertilization (Hamilton et al., 2019). These results suggest that during penetration through the cumulus and zona pellucida layers, bovine sperm undergo changes that induce or support nuclear remodeling, which facilitate the subsequent exposure and/or activation of PLCζ. Our data indicate that SER treatment of bovine sperm stimulates their ability to activate oocytes

following fertilization by ICSI. We hypothesize that in bovine sperm, SER conditions induce remodeling of the sperm head facilitating nuclear decondensation and PLCζ release, just as it occurs during the process of natural sperm entry and fertilization. It will be interesting to determine whether the mechanism increasing sperm functional parameters and improving embryo development success rates in mice are related to the molecular changes which increase egg activation rates in bovines.

In all ARTs, the limiting step for a successful pregnancy is the attainment of embryos competent for implantation that can develop into viable fetuses (Geary and Moon, 2006; Patrizio

TABLE 1 | Effect of SER treatment on the in vitro development of bovine embryos generated by ICSI, piezoelectric-aided.


Data followed by different superscripts (a,b) are significantly different (p < 0.05).

et al., 2007). Poor quality of embryos results in higher rates of implantation failure (Takeuchi et al., 2017). Not surprisingly, most clinical research in humans and species of economic relevance has focused on improving embryo culture techniques (Summers, 2013) and selecting the best embryos to be transferred to the female (Castello et al., 2016; Vaiarelli et al., 2016). In particular, embryos can be selected by monitoring embryo development in real time using imaging technology or by preimplantation genetic diagnostic (PGD; Castello et al., 2016; Vaiarelli et al., 2016). In comparison, less research has been conducted to investigate sperm contributions to embryo quality. For ICSI, recent efforts focused on selecting high quality sperm using high microscopy magnification (Goswami et al., 2018), a method, known as morphologically selected sperm injection (IMSI). IMSI has been used by different clinics with a variety of outcomes (De Vos et al., 2013; Luna et al., 2015). More recently, thermotaxis was used to select sperm for ICSI using temperature gradients (Perez-Cerezales et al., 2018). After ICSI, thermotaxis-selected sperm resulted in higher percentage of blastocyst development and better implantation rates than nonselected sperm (Perez-Cerezales et al., 2018). These investigations

suggest that the condition of the sperm prior to fertilization contributes to embryo development and that sperm selection prior to ICSI has the potential to be used in the clinic to improve pregnancy rates. Similarly, SER-treated sperm increased embryo development and pregnancy rates; however, instead of selecting the best sperm, SER appears to improve the overall sperm quality increasing their chances to produce high-quality embryos.

Translationally, SER treatment can be used for more efficient transgenic mouse production. SER treatment increased fertilization and embryo development rates of FerDR/DR mice which have a severely in vitro sub-fertile phenotype (Alvau et al., 2016). As another model, CatSper1 KO mice are infertile (Ren et al., 2001). In previous work, we showed that a short exposure of Ca2<sup>+</sup> ionophore A<sup>23187</sup> to sperm induced IVF rates of 30% in CatSper1 KO mice (Navarrete et al., 2016). SER treatment alone was not able to induce fertilizing capacity in sperm from sterile CatSper1 KO mice. One important difference between these two models is that while FERDR/DR mouse model is fertile in vivo, CatSper1 KO is completely sterile. We hypothesize that SER treatment is able to bypass lack of tyrosine phosphorylation in FERDR/DR because sperm from these mice are still able to fertilize. In this case, SER is only improving their ability to fertilize. On the other hand, CatSper1 KO mice are completely sterile in vivo and in vitro. When we combined both treatments, we found a synergistic increase of fertilization and embryo development rates in CatSper1 KO mice. These experiments suggest that an increase in sperm [Ca2+]<sup>i</sup> is still required for SER to work.

Overall, the sperm energy restriction treatment is a novel method that improves both sperm fertilization rates and embryo development after IVF. The molecular mechanisms triggered by the sperm SER treatment prior IVF or ICSI are still not understood. Our data raise a series of very important still unanswered questions that call for new investigations in the field of reproductive biology and early embryo development. Importantly, our results indicate that sperm capacitation conditions continue to be relevant after fertilization has occurred, opening new avenues to study the relationship between sperm capacitation and early embryo development. The overall SER improvements in sperm function before and after fertilization have the potential to be applied to other mammalian species including humans and to impact ART practices worldwide.

#### MATERIALS AND METHODS

#### Materials

Chemicals and other lab reagents were purchased as follows: 4- Bromo-Ca2<sup>+</sup> Ionophore A<sup>23187</sup> (B7272), bovine serum albumin (BSA, fatty acid-free, A0281) for media, pregnant mare serum gonadotropin (PMSG) (G4877), and human chorionic gonadotropin (hCG) (CG5) from Sigma (St. Louis, MO, United States). Non-surgical embryo transfer (NSET) device was acquired from Paratechs (Billerica, MA, United States). Light mineral oil (ES-005-C) and EmbryoMax <sup>R</sup> KSOM Medium (1X) with 1/2 Amino Acids (MR-106-D) were obtained from Millipore (Billerica, MA, United States). Paraformaldehyde was purchased from Electron Microscopy Sciences (Hatfield, PA, United States). Hoechst 33342 was obtained from ThermoFisher Scientific (Agawam, MA, United States).

#### Animals

All procedures involving experimental animals were performed in accordance with Protocol 2016-0026 approved by the University of Massachusetts Amherst Institutional Animal Care and Use Committee (IACUC). Male CD-1 (ICR) retired breeders, female CD1 (ICR) 8–12 weeks of age, and male CD-1 (ICR) vasectomized mice were obtained from Charles River Laboratories (Wilmington, MA, United States). Male C57BL/6J (different ages) and female C57BL/6J 8-weeks of age mice were obtained from The Jackson Laboratory (Farmington, CT, United States). Male Fer DR/DR mice (Craig et al., 2001) were obtained from Dr. Greer from Queen's University, Kingston (ON, Canada). Infertile CatSper1 KO (Ren et al., 2001) mice were obtained from Dr. Clapham and Dr. Chung from Harvard University, Boston (MA, United States).

### Computer-Assisted Sperm Analysis (CASA)

Sperm incubation for analysis of motility was performed in a HEPES modified Toyoda–Yokoyama–Hosi (HEPES-TYH) medium (Toyoda et al., 1971) containing 119.37 mM NaCl, 4.7 mM KCl, 1.71 mM CaCl2.2H2O, 1.2 mM KH2PO4, 1.2 mM MgSO4.7H2O, 0.51 mM Na-pyruvate, 5.56 mM glucose, and 20 mM HEPES (pH 7.4). This medium does not contain HCO<sup>3</sup> − or BSA and does not support capacitation; it was named (NON-CAP-HEPES-TYH) and abbreviated as NON-CAP in **Figure 1**. The same medium supplemented with 15 mM HCO<sup>3</sup> <sup>−</sup> and 5 mg/mL BSA support capacitation, it was named CAP-HEPES-TYH and abbreviated as CAP in **Figure 1**. Starvation was done in CAP medium in which glucose and pyruvate were omitted; this medium was named Free-CAP-HEPES-TYH and notated with the symbol (–) in **Figure 1**. Finally, rescue was performed by incubating sperm first in FREE-HEPES-TYH until motility stopped and then recovered in CAP-HEPES-TYH medium. Following the same abbreviation used throughout this manuscript, this procedure was abbreviated as SER in **Figure 1**.

To conduct computer-assisted sperm analyses (CASA), for each replicate, cauda epididymides from two mice were dissected, cut in half, and distributed in specific media as follows: (1) for non-capacitating conditions (NON-CAP), two half epididymis from each mouse were collected in NON-CAP-HEPES-TYH containing energy substrates but lacking NaHCO<sup>3</sup> and BSA; (2) for capacitating conditions (CAP), two half epididymis from each mouse were collected in CAP-HEPES-TYH containing energy substrates and supplemented with 15 mM NaHCO<sup>3</sup> and 5 mg/mL of BSA; (3 and 4) the remaining half epididymides were collected in FREE-CAP-HEPES-TYH medium devoid of glucose and pyruvate supplemented with 15 mM NaHCO<sup>3</sup> and 5 mg/mL of BSA. In all these cases, after 10 min of sperm swimout, epididymides were removed and samples were washed by two sequential centrifugations, using the respective incubation media. Sperm were then counted for each condition; equal number of sperm (∼2 million sperm) were then divided in

four aliquots (notice that starved sperm in FREE-CAP-HEPES-TYH are divided in two different aliquots) and incubated at 37◦C in 500 µL of their respective medium. After ∼40 min (when sperm in starvation conditions stop moving), to facilitate washing, additional 1.5 mL of the appropriate media (as defined in the next sentence) was added to each aliquot. Appropriate media mean: for condition 1 (NON-CAP), NON-CAP-HEPES-TYH; for condition 2 (CAP), CAP-HEPES-TYH, for condition 3 (–), FREE-CAP-HEPES-TYH; and for condition 4 (SER), same as for condition 2, CAP-HEPES-TYH. Sperm suspensions were then centrifuged for 5 min at 150 × g at room temperature, 1.7 mL of supernatants was removed leaving the bottom 300 µL in the tubes to avoid losing sperm cells. Additional 200 µL of the respective media was added and each aliquot incubated for 60 additional minutes. After this period, sperm suspensions (25 µL) were loaded into pre-warmed chamber slide (depth, 100 µm) (Leja slide, Spectrum Technologies) and placed on a microscope stage at 37◦C. Sperm motility was examined using the CEROS computer-assisted semen analysis (CASA) system (Hamilton Thorne Research, Beverly, MA, United States) as previously described (Sanchez-Cardenas et al., 2018). The default settings include the following: frames acquired: 90; frame rate: 60 Hz; minimum cell size: 4 pixels; static head size: 0.13–2.43; static head intensity: 0.10–1.52; and static head elongation: 5–100. Sperm with hyperactivated motility, defined as motility with high amplitude thrashing patterns and short distance of travel, were sorted and analyzed using the CASAnova software (Goodson et al., 2011). At least five microscopy fields corresponding to a minimum of 200 sperm were analyzed for each treatment in each experiment.

#### Sperm Preparation for in vitro Fertilization (IVF)

Media used for IVF assay did not contain HEPES; instead pH 7.4 was maintained using HCO<sup>3</sup> <sup>−</sup>/CO<sup>2</sup> buffer system. For these assays, we used Toyoda–Yokoyama–Hosi Complete TYH (C-TYH) medium (Toyoda et al., 1971) containing 119.37 mM NaCl, 4.7 mM KCl, 1.71 mM CaCl2.2H2O, 1.2 mM KH2PO4, 1.2 mM MgSO4. 7H2O, 25.1 mM NaHCO3, 0.51 mM Na-pyruvate, 5.56 mM glucose, and 4 mg/mL BSA, 10 µg/mL gentamicin and phenol red 0.0006% equilibrated at 5% CO<sup>2</sup> to reach pH 7.4. For media used for starvation, glucose and pyruvate were omitted from the medium described above and was called F-TYH.

For IVF, spermatozoa were collected from the cauda epididymis. In each case, the two epididymides from the same mouse were separated and incubated in different conditions. One of the cauda was placed in 2 mL of C-TYH, containing glucose and pyruvate and abbreviated as CAP in **Figures 2**–**5**. The other one was placed in 2 mL of F-TYH media, not containing nutrients; sperm in this medium will then be rescued and the treatment abbreviated SER in **Figures 2**–**5**. After 10 min of sperm swim-out, each cauda epididymis was removed from the tubes, and sperm were washed by two sequential centrifugations using the respective incubation medium as follows. First, sperm suspensions were centrifuged for 5 min at 300 × g at room temperature. After that, supernatants were removed, and sperm were re-suspended in 2 mL of C-TYH for the control, and 2 mL of F-TYH for the SER treatment. Samples were centrifuged for additional 5 min at 150 × g at room temperature, supernatants were removed, and sperm re-suspended in 500 µL of C-TYH for the CAP control and 500 µL of F-TYH for the SER treatment. Both sperm suspensions were incubated at 5% CO<sup>2</sup> at 37◦C until sperm from F-TYH stopped moving (∼40 min). Then, 1.5 mL of C-TYH was added to both CAP- and SER-treated sperm. Sperm suspensions were then centrifuged for 5 min at 150 × g at room temperature and 1.7 mL of supernatants was removed leaving the bottom 300 µL in the tubes. To maintain the initial sperm concentration, additional 200 µL of C-TYH was added to both samples.

## Ca2<sup>+</sup> Ionophore Treatment for IVF

Ca2<sup>+</sup> ionophore 4Br-A<sup>23187</sup> was used at a final concentration of 20 µM in C-TYH as previously described (Navarrete et al., 2016). Briefly, sperm were incubated with Ca2<sup>+</sup> ionophore 4Br-A<sup>23187</sup> for 10 min in 5% CO<sup>2</sup> at 37◦C, and subsequently washed twice by mild 5 min centrifugations (150 × g) in C-TYH media. For SER and Ca2<sup>+</sup> ionophore A<sup>23187</sup> combination, SER treatment was done as explained above with minor modifications. At 35 min of incubation in F-TYH, just before sperm stopped their movement, Ca2<sup>+</sup> ionophore 4Br-A<sup>23187</sup> was added for 5 min. Next, 1.5 mL of C-TYH was added and the sperm suspension was centrifuged for 5 min at 150 × g at room temperature. Supernatant was removed, and 2 mL of C-TYH was added. Then, the sperm sample was centrifuged once more for 5 min at 150 × g and the supernatant removed leaving the bottom 300 µL in the tube. Additional 200 µL of C-TYH was added to reach a final volume of 500 µL.

#### Sperm Motility Video Recordings

Sperm were incubated in F-TYH or C-TYH as described above. The respective sperm suspensions (25 µL) were loaded into glass-bottom culture dishes (MatTek Corp., Ashland, MA, United States). Videos were captured using NIS-Elements software on a Nikon TE300 inverted microscope (Chiyoda, Tokyo, Japan) fitted with 20X objective lenses (Plan Apo, NA 0.75) and equipped with a cMOS camera (Andor Zyla, Belfast, Northern Ireland). Temperature of the samples was maintained at 37◦C using a stage warmer (Frank E. Fryer scientific instruments, Carpentersville, IL, United States).

#### In vitro Fertilization

Metaphase II-arrested mouse oocytes were collected from 6–8 week-old super ovulated CD-1 (ICR) or C57BL/6J female mice (Charles River Laboratories, Wilmington, MA, United States) as previously described (Navarrete et al., 2015). Each female was injected with 7.5–10 IU PMSG and hCG 48 h apart. Superovulated females were then sacrificed 13 h post-hCG injection, oviducts were dissected, and cumulus-oocyte complexes (COCs) were collected in TL-HEPES medium [containing 114 mM NaCl, 3.22 mM KCl, 2.04 mM CaCl2.2H2O, 0.35 mM NaH2PO4.2H2O, 0.49 mM MgCl2.6H2O, 2.02 mM NaHCO3, 10 mM Lactic acid (sodium salt), and 10.1 mM Hepes]. Then, COCs were washed in C-TYH medium, and subsequently placed into a 90 µL drop of C-TYH covered with mineral oil previously equilibrated in an

incubator with 5% CO<sup>2</sup> at 37◦C. Fertilization drops containing two to three COCs (∼40–60 eggs) were inseminated with 100,000 sperm incubated in different conditions as described above. Fertilization dishes were kept in an incubator with 5% CO<sup>2</sup> at 37◦C. After 4 h of insemination, eggs were washed, put in fresh C-TYH media, and incubated overnight in 5% CO<sup>2</sup> at 37◦C. Fertilization rates (in percentage) were evaluated 24 h postinsemination and were calculated as the total number of two-cell embryos divided by the total number of inseminated oocytes multiplied by 100.

#### Embryo Culture and Embryo Transfer

Twenty-four hours post-fertilization, two-cell embryos were transferred to drops containing KSOM media and further incubated for 3.5 days in 5% CO<sup>2</sup> at 37◦C. At this stage, the percentage of blastocyst formation was evaluated. In some cases, 12–18 blastocysts were transferred to 2.5 days post-coitum (dpc) pseudo-pregnant CD-1 recipient females using the non-surgical uterine embryo transfer device NSET (Bin Ali et al., 2014). The number of blastocysts for transfer was selected considering the minimum number of blastocysts available which in all cases corresponded to those obtained from sperm incubated in standard capacitation media. The pseudo-pregnant CD-1 recipient females were obtained by mating with vasectomized males (obtained from Charles River) 2.5 days before the embryo transfers were performed. Only female mice with a visible plug were chosen as embryo recipients.

#### Blastocyst Nuclear Staining

A total cell count of 35 individual 3.5-day blastocysts for each condition (control or SER treatment) was performed. Embryos were removed from culture and washed two times in 100 µL drops of PBS containing 1 mg/mL BSA. Then, embryos were fixed in a 100-µL drop of paraformaldehyde solution [4% (w/v) in PBS, pH 7.4] for 20 min at room temperature and washed three times in 100 µL PBS-BSA. After that, embryos were transferred to a 100-µL drop of 1 µg/mL Hoechst 33342 (ThermoFisher Scientific, Agawam, MA, United States) in PBS-BSA and incubated for 10 min at room temperature. Then, embryos were washed two times in 100-µL drops of PBS-BSA and mounted on slides.

Images were taken using an A1 HD25 resonant scanning confocal microscope (Nikon) equipped with an sCMOS camera using a 40X (Plan Fluor, NA 1.3) objective. For each embryo, 10– 12 z-stacks of 0.9 µm were taken, and cell number was calculated as the total number of nuclei stained per embryo.

#### Blastocysts Outgrowth Assay

Blastocysts were collected and transferred gently into a culture plate coated with 0.1% gelatin (Sigma–Aldrich, St. Louis, MO, United States) and cultured in DMEM (Lonza, Allendale, NJ, United States) containing 10% fetal calf serum (Atlanta Biologicals, Flowery Branch, GA, United States) and 1X GlutaMAX (Thermo Fisher Scientific, Agawam, MA, United States). Outgrowth assays were conducted at 37◦C in a humidified atmosphere of 5% CO<sup>2</sup> for 3 days and were observed daily. Blastocysts outgrowth was considered positive when: (1) they were able to hatch out of the zona pellucida by 24 h, (2) they had attached to the culture plate by 48 h, and (3) they had formed inner cell mass (ICM) colonies with surrounding trophoblast cells at 72 h.

#### Bovine Sperm Preparation and SER Treatment

Bovine sperm were obtained from frozen semen samples kindly donated by American Breeder Services (DeForest, WI, United States). Sperm containing straws were thawed at 38◦C for 1 min. After that, sperm were separated using a Percoll gradient (45%/90%) by centrifugation 600 × g for 15 min.

After Percoll centrifugation, sperm (5 × 10<sup>6</sup> spermatozoa/mL) were washed in 3 mL of either Sp-TALP [containing 100 mM NaCl, 3.1 mM KCl, 2.04 mM CaCl2.2H2O, 0.35 mM NaH2PO4.2H2O, 0.4 mM MgCl2.6H2O, 25 mM NaHCO3, 21.6 mM lactic acid (sodium salt), and 1 mM pyruvate] (Parrish et al., 1988) or pyruvate-lactate-free Sp-TALP (free-Sp-TALP) medium, and centrifuged at 150 × g for 5 min at room temperature. After that, supernatants were removed, and sperm were re-suspended in 0.5 mL of Sp-TALP for the control, and 0.5 mL of free-Sp-TALP for the SER treatment. Both sperm suspensions were incubated at 5% CO<sup>2</sup> at 38◦C until sperm from free-Sp-TALP showed a considerable reduction in their motility (∼3 h). Then, 1.5 mL of Sp-TALP was added to both control and SER-treated sperm. The samples were then centrifuged for 5 min at 150 × g at room temperature, 1.7 mL of supernatants was removed leaving the bottom 300 µL. Sperm were used for ICSI as explained below.

#### Bovine Intracytoplasmic Sperm Injection

Intracytoplasmic sperm injection was carried out as previously described (Kurokawa and Fissore, 2003), according to standard protocols, using Narishige manipulators (Medical System Corp., Great Neck, NY, United States) mounted on a Nikon diaphot microscope (Nikon Inc., Garden City, NY, United States). Before ICSI, oocytes were denuded of granulosa cells by gently pipetting in the presence of 1 mg/mL of hyaluronidase and selected by the presence of the first polar body. All matured oocytes were randomly allocated to groups. One part of the sperm suspension was mixed with one part of PBS containing 10% polyvinylpyrrolidone (PVP, M.W. 360 kDa; Sigma). Bovine sperm were immobilized by applying a few piezo pulses to the sperm tail, and the whole sperm was injected. Sperm were delivered into the oocyte's cytosol using a piezo micropipette-driving unit (Piezodrill; Burleigh Instruments Inc., Rochester, NY, United States).

#### Statistical Analysis

Data from all experiments were analyzed using SIGMA plot software<sup>1</sup> . Data are expressed as the means ± SEM. The difference

<sup>1</sup>www.sigmaplot.com

between groups mean values was analyzed by one-way analysis of variance (ANOVA) followed by Tukey's test. P-values < 0.05 were considered significant, and statistical significances were indicated in the figure legends.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

#### ETHICS STATEMENT

fcell-07-00262 November 1, 2019 Time: 17:32 # 11

The animal study was reviewed and approved by the University of Massachusetts, Amherst Institutional Animal Care and Use Committee (IACUC), protocol 2016-0026.

#### AUTHOR CONTRIBUTIONS

FN, MG, and PV were responsible for the organization and design of the whole work, data analysis, and preparation of manuscript. FN, LA, DM-H, DT, GL, GA, FG-V, and MG performed the experiments. AS, LL, JB, MB, AD, JM, MG, RF, and PV contributed with experimental design, animal protocols, discussion of findings, and correction of the manuscript.

#### FUNDING

This study was supported by NIH grants HD-038082 (to PV), HD-088571 (to JB, LL, and PV), and HD-92499 (to RF). FG-V was supported by Jiménez de la Espada Mobility Program-Fundación Séneca 2015 (19935/EE/15). DM-H was a recipient of a postdoctoral fellowship from the Government of Extremadura (Spain) and the Fondo Social Europeo (PO14005).

### REFERENCES


#### ACKNOWLEDGMENTS

The authors would like to thank Dr. Peter Greer for the donation of FerDR/DR mice. Confocal images were taken at the Light Microscopy Facility and Nikon Center of Excellence at the Institute for Applied Life Sciences, University of Massachusetts Amherst, which is supported by the Massachusetts Life Sciences Center.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00262/ full#supplementary-material

FIGURE S1 | Sperm energy restriction and recovery (SER) treatment improves fertilization and embryo development rates in C57BL/6J male mice of different ages. Sperm obtained from C57BL/6J mice of different ages were incubated in C-TYH, containing glucose and pyruvate (CAP) or in F-TYH media, not containing nutrients; sperm in this medium will then be rescued by energy sources replenishment as explained in Methods, and the treatment abbreviated (SER). After treatment (CAP or SER) sperm from each condition were added to the insemination drops containing cumulus-enclosed female oocytes (COCs). In all cases, left panels represent percentages of fertilization calculated as the percentage of oocytes that reached 2-cell embryo stage. Middle panels represent percentages of blastocyst formation calculated as percentage of the 2-cell embryos that reached a 3.5-day blastocyst stage. Right panels represent percentages of total blastocysts calculated as the percentage of 3.5-day blastocyst stage embryos out of the total number of inseminated eggs. (A–C) Results from IVF and embryo cultures after heterologous fertilization between CD1 female oocytes and 2–6 months old C57BL/6J sperm (A, n = 18); 7–12 months old C57BL/6J sperm (B, n = 10); or 13–24 months old C57BL/6J sperm (C, n = 10). (D–F) Results from IVF and embryo culture after homologous fertilization between C57BL/6J female oocytes and 2–6 months old C57BL/6J sperm (D, n = 5); 7–12 months old C57BL/6J sperm (E, n = 5); or 13–24 months old C57BL/6J sperm (F, n = 5). In each panel, statistically significant differences are indicated as: <sup>∗</sup>p < 0.05; ∗∗p < 0.01, and ∗∗∗p < 0.001.

TABLE S1 | Compilation of IVF, embryo development, and pups born after embryo transfer.

VIDEO S1 | Sperm energy restriction and recovery (SER) treatment. Sperm were incubated in F-TYH at 5% CO<sup>2</sup> at 37◦C until stopped moving (∼40 min). Then, an aliquot was loaded into glass-bottom culture dishes and filmed in an inverted microscope as explained in Methods. C-TYH was added after 10 sec of recording and sperm regained motility.

M. Gertsenstein, K. Vintersten Nagy, and A. Nagy, (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press), 593–600.



by intracytoplasmic sperm injection. Mol. Reprod. Dev. 44, 230–233. doi: 10.1002/(sici)1098-2795(199606)44:2<230::aid-mrd12>3.3.co;2-j



implications for IVF in humans. Biol. Res. 48:68. doi: 10.1186/s40659-015- 0059-y


**Conflict of Interest:** LL and JB report owning equity interest in CEP Biotech which has licensed commercialization of a panel of monoclonal antibodies directed against sAC. The method described in this submission is the subject of intellectual property filed by the University of Massachusetts Amherst. It was fully funded by the National Institutes of Health and is subject to rights retained by the United States Government. AS is the founder of Sperm Capacitation Technologies Inc., which is seeking to commercialize the technology. PV is a scientific advisor to the company.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Navarrete, Aguila, Martin-Hidalgo, Tourzani, Luque, Ardestani, Garcia-Vazquez, Levin, Buck, Darszon, Buffone, Mager, Fissore, Salicioni, Gervasi and Visconti. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Proteomic Changes in Human Sperm During Sequential in vitro Capacitation and Acrosome Reaction

Judit Castillo<sup>1</sup> \*, Orleigh Adeleccia Bogle<sup>1</sup> , Meritxell Jodar<sup>1</sup> , Forough Torabi<sup>2</sup> , David Delgado-Dueñas<sup>1</sup> , Josep Maria Estanyol<sup>3</sup> , Josep Lluís Ballescà<sup>4</sup> , David Miller<sup>2</sup> and Rafael Oliva1,5 \*

<sup>1</sup> Molecular Biology of Reproduction and Development Research Group, Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Fundació Clínic per a la Recerca Biomèdica, Department of Biomedical Sciences, Faculty of Medicine and Health Sciences, Universitat de Barcelona, Barcelona, Spain, <sup>2</sup> LIGHT Laboratories, Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, United Kingdom, <sup>3</sup> Proteomics Unit, Scientific and Technical Services, Universitat de Barcelona, Barcelona, Spain, <sup>4</sup> Clinic Institute of Gynaecology, Obstetrics and Neonatology, Hospital Clínic, Barcelona, Spain, <sup>5</sup> Biochemistry and Molecular Genetics Service, Hospital Clínic, Barcelona, Spain

#### Edited by:

Tomer Avidor-Reiss, University of Toledo, United States

#### Reviewed by:

João Ramalho-Santos, University of Coimbra, Portugal Md Saidur Rahman, Chung-Ang University, South Korea Philip Chi Ngong Chiu, The University of Hong Kong, Hong Kong

> \*Correspondence: Judit Castillo juditcastillo@ub.edu Rafael Oliva roliva@ub.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 01 August 2019 Accepted: 06 November 2019

Published: 20 November 2019

#### Citation:

Castillo J, Bogle OA, Jodar M, Torabi F, Delgado-Dueñas D, Estanyol JM, Ballescà JL, Miller D and Oliva R (2019) Proteomic Changes in Human Sperm During Sequential in vitro Capacitation and Acrosome Reaction. Front. Cell Dev. Biol. 7:295. doi: 10.3389/fcell.2019.00295 The male gamete is not completely mature after ejaculation and requires further events in the female genital tract to acquire fertilizing ability, including the processes of capacitation and acrosome reaction. In order to shed light on protein changes experienced by the sperm cell in preparation for fertilization, a comprehensive quantitative proteomic profiling based on isotopic peptide labeling and liquid chromatography followed by tandem mass spectrometry was performed on spermatozoa from three donors of proven fertility under three sequential conditions: purification with density gradient centrifugation, incubation with capacitation medium, and induction of acrosome reaction by exposure to the calcium ionophore A23187. After applying strict selection criteria for peptide quantification and for statistical analyses, 36 proteins with significant changes in their relative abundance within sperm protein extracts were detected. Moreover, the presence of peptide residues potentially harboring sites for post-translational modification was revealed, suggesting that protein modification may be an important mechanism in sperm maturation. In this regard, increased levels of proteins mainly involved in motility and signaling, both regulated by protein modifiers, were detected in sperm lysates following incubation with capacitation medium. In contrast, less abundant proteins in acrosome-reacted cell lysates did not contain potentially modifiable residues, suggesting the possibility that all those proteins might be relocated or released during the process. Protein-protein interaction analysis revealed a subset of proteins potentially involved in sperm maturation, including the proteins Erlin-2 (ERLIN2), Gamma-glutamyl hydrolase (GGH) and Transmembrane emp24 domain-containing protein 10 (TMED10). These results contribute to the current knowledge of the molecular basis of human fertilization. It should now be possible to further validate the potential role of the detected altered proteins as modulators of male infertility.

Keywords: sperm, spermatozoa, capacitation, acrosome reaction, proteomics, mass spectrometry, TMT labeling

## INTRODUCTION

fcell-07-00295 November 18, 2019 Time: 15:9 # 2

Mammalian fertilization relies on the ability of spermatozoa to penetrate through the zona pellucida (ZP) and fuse with the oolemma. Testicular spermatozoa, however, although morphologically mature, lack this essential capacity and additional post-gonadal stages of maturation are required in both the male and the female reproductive tracts. After being released into the lumen of the seminiferous tubules, testicular spermatozoa are sequentially modified throughout their transit in the epididymis, acquiring forward progressive motility (Dacheux and Dacheux, 2014). However, maturation is not completed until spermatozoa undergo capacitation during their journey through the female genital tract (Visconti et al., 1998; Stival et al., 2016; Puga Molina et al., 2018a). Sperm are only capable of successfully fertilizing the oocyte after completing this essential post-ejaculatory step, which is a prerequisite to the acrosome reaction (Barros, 1967; Gaddum and Blandau, 1970; Visconti and Kopf, 1998; Visconti et al., 1998, 2011; De Jonge, 2005; Liu et al., 2007).

Sperm capacitation is a recognized phenomenon associated with acquisition of hyperactive motility comprising a complex set of highly regulated molecular and physiological events. As mature spermatozoa are no longer engaged in nuclear gene expression (Oliva, 2006; Jodar et al., 2013), understanding the role of protein modifiers, cell-cell communication and signaling pathways is essential (Baldi et al., 1996; Halet et al., 2003; De Jonge, 2005; Visconti et al., 2011; Buffone et al., 2012; Signorelli et al., 2012; Cuasnicú et al., 2016; Barrachina et al., 2018; Torabi et al., 2017b). For instance, many studies have correlated capacitation with an increase in cholesterol efflux from the sperm membrane (Visconti et al., 2002; Gadella et al., 2008) that modulates intracellular levels of calcium and bicarbonate ions (Visconti et al., 2011; Battistone et al., 2013; Li et al., 2014; Puga Molina et al., 2018a). As a consequence, other changes during capacitation include the hyperpolarization of the sperm plasma membrane (Puga Molina et al., 2017, 2018b), and the enhanced activity of protein kinases, phosphatases and acetylases (Ficarro et al., 2003; Ickowicz et al., 2012; Battistone et al., 2014; Sati et al., 2014; Wang J. et al., 2015; Puga Molina et al., 2017; Ritagliati et al., 2018). In addition, protein regulation through redox signaling has been proposed as one of the critical properties of sperm capacitation (O'Flaherty, 2015). All these changes occur prior to the acrosome reaction, a calcium-dependent process leading to a sequential release of the acrosomal contents (Buffone et al., 2008), which includes hydrolytic enzymes, such as acrosin and hyaluronidase, that allow sperm penetration through the ZP and subsequent fusion with the oocyte membrane (Brucker and Lipford, 1995; Patrat et al., 2000; Ickowicz et al., 2012; De Jonge and Barratt, 2013; Petit et al., 2013; Sánchez-Cárdenas et al., 2014; Singh and Rajender, 2015; Ito and Toshimori, 2016; Torabi et al., 2017a). The role of kinases and tyrosine phosphorylation has been also established as essential for sperm to correctly undergo acrosome reaction (Ickowicz et al., 2012; Breitbart and Finkelstein, 2015).

The analysis of protein profiles of lysates obtained from human sperm cell has revealed a complex set of proteins involved in processes culminating in oocyte fertilization, including capacitation, the acrosome reaction, oocyte penetration and sperm-oocyte fusion (Castillo et al., 2018). Indeed, capacitation is the event most studied by the application of quantitative proteomic strategies in human (Ficarro et al., 2003; Secciani et al., 2009; Wang J. et al., 2015; Yu et al., 2015; Hernández-Silva et al., 2019) as well as in model species, such as mice (Baker et al., 2010), boars (Bailey et al., 2005; Kwon et al., 2014), bulls (Park et al., 2012) and buffalo (Jagan Mohanarao and Atreja, 2011; Hou et al., 2019). Recently, Hernández-Silva et al. (2019) approached the proteomic study of human sperm capacitation from the alternative perspective of the seminal fluid components bathing sperm during and following ejaculation. Interestingly, they found a group of seminal fluid derived proteins attached to the sperm surface which inhibits the progress of capacitation by negatively affecting sperm hyperactivation and protein tyrosine phosphorylation (Hernández-Silva et al., 2019).

The study of the sperm protein composition is contributing not only to increase the knowledge of the male gamete structure and cargo, but also of past and future events related to sperm development and oocyte fertilization, respectively (Amaral et al., 2014a; Codina et al., 2015; Castillo et al., 2018). Mass spectrometry (MS) strategies are rapidly evolving, and novel approaches are currently available that could provide further insights into the maturation of the ejaculated sperm (Codina et al., 2015). Also, identifying protein changes in abundance and post-translational modifications during the different steps of sperm maturation is a good strategy to find candidates that could act as male infertility biomarkers with predictive, diagnostic and prognostic potentials (Rahman et al., 2013; Amaral et al., 2014a; Jodar et al., 2017). However, while the involvement of sperm proteins in ZP binding has been explored by proteomics (Redgrove et al., 2011; Petit et al., 2013), the sperm protein profile after the acrosomal exocytosis has not yet been addressed by MS. In the current study we employed a quantitative proteomics strategy based on isotopic peptide labeling using tandem mass tags (TMT), protein identification by liquid chromatography followed by tandem MS (LC-MS/MS), and strict and robust quantification and statistical analysis to evaluate changes in ejaculated spermatozoa after incubation in capacitation medium and calcium ionofore-induced acrosome reaction, aiming to improve the knowledge of the molecular mechanisms leading to post-ejaculation sperm maturation from a proteomic perspective.

## MATERIALS AND METHODS

## Biological Material and Sample Collection

Human semen samples were obtained from 3 donors of proven fertility attending the Assisted Reproduction Unit (FIVclinic) at the Clinic Institute of Gynaecology, Obstetrics and Neonatology, from the Hospital Clínic, Barcelona, Spain. The sample size was calculated by considering a balance between (1) a sufficient number of samples to statistically evaluate the significance of the resulting data, and (2) the maximum amount of analytical samples that can be analyzed by quantitative proteomics using

a unique set of Tandem Mass TagTM 6-plex (TMTsixplexTM) Reagents. The ejaculates were collected by masturbation into sterile containers following a minimum of 3 days abstinence. Semen was allowed to liquefy at room temperature and an initial evaluation of the seminal parameters was taken for concentration and motility assessment using the automatic semen analysis system CASA (Computer Assisted Semen Analysis; Proiser, Paterna, Spain). Additional parameters such as sperm variability and morphology were also assessed, using 0.5% (w/v) Eosyn Y and Diff-quickTM staining, respectively. All three ejaculates showed normal parameters according to the limits established by the World of Health Organization (WHO) (World Health Organization, 2010) and were classified as normozoospermic semen samples (**Supplementary Table S1**).

All samples were used in accordance with the appropriate ethical guidelines and Internal Review Board, and the biological material storing and processing was approved by the Clinical Research Ethics Committee of the Hospital Clínic of Barcelona. Written informed consents were obtained from all donors in accordance with the Declaration of Helsinki.

#### Sperm Preparation

Each ejaculate was processed under three sequential conditions (**Figure 1**): density-gradient centrifugation (DGC sperm), incubation in capacitation medium (CAP sperm) and incubation with the calcium ionophore A23187 to in vitro induce the acrosome reaction (AR sperm), following wellestablished protocols described elsewhere (World Health Organization, 2010; De Jonge and Barratt, 2013) with some minor modifications.

#### Density-Gradient Centrifugation

Spermatozoa were purified using a 60% Percoll <sup>R</sup> (Sigma-Aldrich, St Louis, MO, United States) gradient equilibrated with

Hepes-buffered Ham's F10 medium (Gibco, Life Technologies, United Kingdom) supplemented with sodium bicarbonate (0.2% NaHCO3; w/v; Merck, Darmstadt, Germany), sodium pyruvate (0.003% C3H3NaO3; w/v; Sigma-Aldrich) and sodium DL lactate solution (0.36% v/v, Sigma-Aldrich). In order to ensure the complete elimination of the gradient material, the recovered sperm cells were washed twice with supplemented Hepes medium and centrifuged at 400 g for 10 min. This procedure allowed the purification of the samples from cells other than sperm while maintaining proportions of sperm subpopulations similar to the native semen sample (Mengual et al., 2003; Zhao et al., 2007; Wang G. et al., 2013). The sperm purification efficiency was checked using phase-contrast microscopy. All purified sperm samples contained <1% of potential contaminating cells. For each sample, an aliquot of 20 million of DGC sperm was taken for subsequent procedures of protein solubilization, while the remaining material was subjected to incubation with capacitation medium.

#### Sperm Incubation With Capacitation Medium

Capacitation medium consisted of Hepes-buffered Ham's F10 medium supplemented with 3.5% of bovine serum albumin (BSA; Sigma-Aldrich). DGC sperm were incubated in 1 ml of capacitation medium for each 10 million of sperm, for 3 h at 37◦C with constant rotation. Aliquots of 20 million of CAP sperm from each biological replicates were taken for protein solubilization and the remaining material were used for the incubation with the calcium ionophore A23187.

#### Sperm Incubation With the Calcium Ionophore A23187 to Induce the Acrosome Reaction

To induce the acrosome reaction, 10 µL of the calcium ionophore A23187 (Sigma-Aldrich, stock solution 1.0 mM in DMSO) was added to approximately 10 million CAP sperm to get a final concentration of 10 µM, following WHO recommendations (World Health Organization, 2010). Spermatozoa were then incubated for 15 min at 37◦C. Negative control aliquots were incubated with DMSO (vehicle control) for an equal amount of time (De Jonge and Barratt, 2013). Subsequently, 70% ethanol was added to stop the reaction and the spermatozoa were recovered by centrifugation at 400 g for 20 min (World Health Organization, 2010). The viability of the cells was monitored by eosin staining (World Health Organization, 2010). For each replicate, an aliquot of 20 million of AR sperm was used for protein solubilization.

## Pisum sativum Agglutinin – Fluorescein Isothiocyanate (PSA-FITC) Labeling

The acrosome reaction was monitored by assessing the integrity of the sperm acrosome vesicle by the PSA-FITC labeling method (World Health Organization, 2010) (**Figure 1**). This fluorescent probe binds to the alpha-methyl mannose and labels the acrosomal content of sperm (Benoff et al., 1993). Briefly, 10 µl of each DGC, CAP, AR sperm and DMSO negative control (3 donors) were smeared on microscope slides in duplicates and were allowed to air dry. The slides were then fixed in 95% (v/v) ethanol for 30 min, and subsequently immersed in 25 µg/ml PSA-FITC staining solution, for 2 h at 4 ◦C. The slides were washed in distilled water, air-dried and covered with mounting medium containing DAPI (Vectashield <sup>R</sup> , Vector Laboratories, Burlingame, CA, United States). The slides were visualized using a BX50 microscope (Olympus, Hamburg, Germany) equipped with a triple band pass filter, and images were acquired with an Olympus DP71 camera. The sperm acrosome status was categorized as: (1) intact acrosome (more than half of the sperm head with bright and uniform fluorescence); (2) reacted acrosome (a band of fluorescence localized to the equatorial segment or no fluorescing stain at all in the acrosome region); and (3) partially intact acrosome (all other sperm cells) (World Health Organization, 2010; De Jonge and Barratt, 2013). A minimum of 200 sperm were counted per slide and the mean of the two slides per sample represented the final value. Differences between the sperm treatments were evaluated by repeated measures ANOVA with Holm–Sidak correction. P-values < 0.05 were considered significant. Determination of the efficiency of the induction of acrosome reaction was also used as an indirect measure of efficiency on sperm capacitation (Torabi et al., 2017a).

## Protein Solubilization and Quantification

From each sperm preparation (DGC, CAP and AR sperm; 3 donors), proteins were solubilized by incubating 20 million spermatozoa with 50 µL of lysis buffer containing 2% SDS and 1 mM Phenylmethylsulfonyl fluoride (PMSF), for 30 min at room temperature with constant gentle shaking. Lysates were centrifuged at 16,000 g for 10 min at 4◦C. Proteins in the soluble fraction were quantified using the BCA method (Thermo Fisher Scientific, Rockford, IL, United States), following manufacturer's recommendations.

## Sperm Peptide Isotopic Labeling (TMT 6-Plex)

A total of 9 protein extracts were used for the proteomic study, corresponding to the DGC, CAP and AR sperm preparations from three different semen donors (**Figure 1**). Differential peptide labeling was performed using Tandem Mass TagTM 6-plex (TMTsixplexTM) Reagents (Thermo Fisher Scientific). The labeling procedure was adapted from the manufacturer's instructions after several studies conducted in our laboratory (Amaral et al., 2014b; Azpiazu et al., 2014; Bogle et al., 2017; Barrachina et al., 2018). Briefly, proteins were reduced in 9.5 mM tris (2-carboxyethyl) phosphine (TCEP) for 1 h at 55◦C, and alkylated with 17 mM iodoacetamide (IAA) for 30 min in the dark. Six-volumes of cold acetone (−20◦C) were added and proteins were allowed to precipitate overnight at −20◦C. Samples were centrifuged at 17,500 g for 10 min, and resuspended in 100 mM triethylammonium bicarbonate (TEAB, Thermo Fisher Scientific) in order to reach a protein concentration of 1 µg/µl. Trypsin was then added at a 1:20 protease-toprotein ratio and the mixture was incubated overnight at 37◦C with constant and gentle shaking. Prior to peptide labeling, aliquots were taken from each of the 9 samples and combined

in equal amounts to represent the internal control sample. Subsequently, equal amounts of peptides from each sample and the internal control were labeled with TMT isobaric tags. After 1 h of incubation at room temperature, the reaction was quenched with 4 µl of 5% hydroxylamine for 15 min. Labeled peptides from each sample were then combined constituting two different multiplex pools, as indicated in **Figure 1**: Run 1 with 5 samples (TMT tags 127, 128, 129, 130, and 131) and 1 internal control (TMT tag 126), and Run 2 with 4 samples (TMT tags 127, 128, 129, and 130) and 1 internal control (TMT tag 126). The two multiplex pools containing labeled peptides were dried in a vacuum centrifuge to near dryness and resuspended in 20 µL of 0.5% trifluoracetic acid (TFA) in 5% acetonitrile. Subsequently, peptide purification was conducted by using Pierce C18 Spin Columns (Thermo Fisher Scientific) following manufacturer's indications.

### Peptide Analysis by LC-MS/MS

Labeled peptides were analyzed via LC-MS/MS with an LTQ-Orbitrap Velos (Thermo-Fisher Scientific) interfaced with an Eksigent nanoLC ultra 2D plus system (AB Sciex, Switzerland). Peptides were injected onto a Pepmap 100 trap column (300 µm × 5 mm, 5 µm, 100 Å) at a flow rate of 400 nL/min. For analytical separation, the trap was switched inline to an Acclaim Pepmap C18 column (75 µm × 15 cm, 3 µm, 100 Å) using a 240 min linear gradient from 5 to 30% acetonitrile in 0.1% formic acid at a flow rate of 400 nL/min. MS/MS analyses were performed using an LTQ Orbitrap Velos (Thermo Fisher Scientific) with a nanoelectrospray ion source. The LTQ-Orbitrap Velos settings included one 30,000 resolution scan for precursor ions followed by MS2 scans of the 20 most intense precursor ions in positive ion mode. MS/MS data acquisition was completed using Xcalibur 2.1 (Thermo Fisher Scientific). The fragmentation method used for identification of TMT labeled peptides was based on higher energy collisional dissociation (HCD) with 40% fixed collision energy (CE).

## LC-MS/MS Raw Data Analysis for Protein Identification and Quantification

LC-MS/MS data were analyzed by Proteome Discoverer 1.4.1.14 (Thermo Fisher Scientific). For database searching, raw MS files were submitted to an in-house Homo sapiens UniProtKB/Swiss-Prot 2018 database including Sus scrofa trypsin (HUMAN\_PIG\_Uniprot\_Release\_2018\_03.fasta). SEQUEST HT version 28.0 was used (Thermo Fisher Scientific). Searches were performed using the following settings: 2 maximum miss cleavage sites for trypsin, TMT as a N-Terminal modification, lysine-TMT (+229.163 Da) and methionine-oxidation (+15.995 Da) as dynamic modifications, cysteine-carbamidomethylation (+57.021 Da) as static modification, 20 ppm precursor mass tolerance, 0.6 Da fragment mass tolerance, and 5 ppm peak integration tolerance. The criteria used for protein identification was set as 1% FDR and a minimum of one peptide match per protein. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2019) partner repository with the dataset identifier PXD014871.

Quantitative analysis were performed simultaneously with protein identification using Proteome Discoverer software. For quantification, only unambiguous and unique peptides were considered. For that, the 'ungrouping' of proteins from their respective families was used during the quantification process. This avoids the possible ambiguity associated with different isoforms of the same protein (Amaral et al., 2014b; Bogle et al., 2017; Barrachina et al., 2018). The normalized TMT quantification values for each identified spectrum were extracted from the ratio between the reported ion intensities corresponding to each individual sample (TMT-127 to TMT-131) and the reporter ion intensity from the internal control (TMT-126). Quantification values were corrected according to the isotopic purities of the reporter ions provided by the manufacturer.

### Statistical Analysis of Proteomic Data Under Strict Quality Criteria

In order to avoid any chance of false positive findings, robust statistics were performed by applying strict selection criteria on the proteomic data. Specifically, only those proteins with at least 1 unique peptide quantified by ≥2 peptide-to-spectrum matches (PSMs) in all the samples, and with a coefficient of variation < 50% in at least 75% of the samples, were considered for further statistical analyses. This pipeline was set by our group in a previous publication (Barrachina et al., 2018). Statistical analysis were conducted in R (R Development Core Team, 2016) with the package "car" (Fox and Weisberg, 2011). Significant differences among the different sperm lysates in the three donors were evaluated using repeated-measures ANOVA test on logtransformed proteomic quantification values. P-values < 0.05 were considered significant if the assumption of sphericity assessed by Mauchly's test had not been violated. In contrast, Greenhouse–Geisser and Huynh–Feldt corrections were applied when data violated the sphericity assumption. Subsequently, the pairwise t-test combined with Holm–Sidak adjustment was applied to identify significant differences between pairs of sample preparations (DGC-CAP, CAP-AR, and DGC-AR). P-values < 0.05 after Holm–Sidak adjustment were considered significant. These results were further filtered by applying similar statistical analysis at peptide level, in order to discard those significant proteins whose corresponding individual peptides do not show statistically significant differences between sperm preparations. Protein abundance alterations were considered significant only when corrected p-values < 0.05 were found at both protein and peptide levels. GraphPad Prism 7 (GraphPad Software Inc., San Diego, CA, United States) was used for visualizing the results.

#### Functional Prediction of the Proteins Detected With Altered Abundance Using Public Databases

In order to predict the functional involvement of the proteins detected with abundance alterations after each sperm preparation

Castillo et al. Proteomics of Post-gonadal Sperm Maturation

(DGC, CAP, and AR), different data were retrieved from the UniProt Knowledgebase<sup>1</sup> and Pubmed<sup>2</sup> . Gene Ontology (GO) enrichment analyses on terms related to biological processes and cellular components were performed with the Gene Ontology Consortium database<sup>3</sup> supported by PANTHER v13.1. The significance of the enrichment analyses was calculated by a Fisher's exact test. P-values < 0.05 after FDR adjustment were considered statistically significant.

Gene Ontology Consortium database was also used to retrieve a list of proteins associated to GO terms related to capacitation and acrosome reaction processes. Specifically, the selected GO terms were "capacitation," "acrosome reaction," "regulation of acrosome reaction," "acrosome assembly," "acrosome matrix dispersal," "positive regulation of acrosome reaction" and "negative regulation of acrosome reaction," setting Homo sapiens as organism. This list of proteins was submitted, together with the list of altered proteins found in this study, to the String database<sup>4</sup> , in order to explore potential protein-protein interactions and functional association networks between altered proteins and proteins already known to have a role in ejaculated sperm maturation. For String analysis, the confidence threshold was set at 0.7 (high confidence).

Theoretical post-translational modifications (PTMs) in the differentially abundant peptides were predicted using data contained at the PhosphoSitePlus <sup>R</sup> database<sup>5</sup> .

#### RESULTS

#### Induction of the Acrosome Reaction With the Calcium Ionophore A23187

Before initiating the proteomic analysis, PSA-FITC labeling was performed to assess the integrity of the sperm acrosome in DGC, CAP and AR sperm, as well as to indirectly measure the level of capacitation of the sperm cells (Torabi et al., 2017a). As seen in **Figure 2**, the absence of signals in the majority of AR sperm (>74%) after exposure to the calcium ionophore A23187 demonstrates that they had mostly undergone acrosomal exocytosis (p < 0.01 one-way ANOVA, Holm–Sidak correction). The significant increase in the number of cells with reacted acrosome also indirectly confirmed the efficiency of sperm capacitation. The three samples showed >85% live cells after incubation with calcium ionophore. PSA-FITC staining revealed between 22 and 30% of the cells with a spontaneously reacted acrosome in the DMSO control, which could be considered higher than expected. However, no remarkable changes were found when DMSO cells were compared to DGC or CAP sperm. Taking into account that these levels were equivalent between the three semen samples, no bias on proteomic data was expected.

FIGURE 2 | PSA-FITC labeling for the assessment of acrosomal integrity. (A) Representative sperm PSA-FITC labeling of one donor sample. PSA-FITC labeling was conducted in purified sperm by density gradient centrifugation (DGC; upper left), sperm incubated with capacitation medium (CAP; upper right) and sperm in which the acrosome reaction was induced by incubation with a calcium ionophore A23187 (AR; lower left). Vehicle control is also shown (DMSO control; lower right). Acrosomal vesicles are shown in green and the sperm cells with reacted acrosomal vesicles are indicated by an arrowhead. Sperm nuclei are stained with DAPI (blue). Scale bar = 5 µM (B) Quantification of the number of sperm with an intact acrosome, partially intact acrosome and reacted acrosome in the three fertile donors used for this study. The number of spermatozoa was counted after PSA-FITC labeling on DGC, CAP, AR, and DMSO sperm. Significant differences in the percentage of sperm cells with reacted acrosomes are indicated with stars (p-value < 0.01; one-way ANOVA with Holm–Sidak correction).

#### Alterations in the Relative Abundance of Proteins After Differential Density Gradient Centrifugation, Incubation in Capacitation Medium and Induction of the Acrosome Reaction

A total of 3658 peptides corresponding to 781 proteins were identified by LC-MS/MS in the lysates from DGC, CAP and AR sperm from the three donor samples (**Supplementary Table S2**). TMT quantification values were determined for all samples in 1901 peptides from 484 proteins (**Supplementary Tables S2**, **S3**),

<sup>1</sup>https://www.uniprot.org/

<sup>2</sup>https://www.ncbi.nlm.nih.gov/pubmed

<sup>3</sup>http://geneontology.org/

<sup>4</sup>https://string-db.org/

<sup>5</sup>https://www.phosphosite.org

of which 860 peptides derived from 240 proteins met our strict quantification criteria (>1 PSM with <50% variability in >75% of the samples; **Supplementary Tables S2**, **S3**). Repeated-measures ANOVA test combined with post hoc pairwise t-test and Holm–Sidak correction revealed changes in the relative abundance of 48 sperm proteins. However, changes in the levels of albumin (ALB) and semenogelin 2 (SEMG2) were disregarded as they were most likely altered due to technical issues. Certainly, the marked increase of ALB seen in CAP sperm may have been due to the presence of bovine serum albumin in the capacitation medium. In fact, from the three peptides quantified for ALB, the only one not showing significant changes in abundance after incubation with capacitation medium was unique for Homo sapiens, while the other two were in common with the bovine sequence (data not shown). Similarly, the lower level of SEMG2 in CAP and AR sperm could be attributed to seminal plasma remnants in DGC sperm that were reduced or even eliminated after the consecutive sperm incubation conditions. Of note, no more proteins were found showing differences with a magnitude similar to SEMG2, which would discard additional artifactual extraneous contamination.

Relative inter-lysate changes in protein abundance were further validated by conducting quantitative analysis at the peptide level. This strategy revealed that peptides from some of the 48 proteins did not show significant differences following sequential incubation conditions. After p-value correction at both the peptide and protein levels, only 36 proteins showed significant inter-lysate differences and were considered for further analyses (**Table 1** and **Supplementary Table S4**). Of note, all these proteins have been identified in the human sperm cell by previous proteomic studies (Castillo et al., 2018). GO enrichment terms showed that this subset of 36 proteins was mainly involved in fertilizationrelated processes, including "fertilization," "sperm-egg recognition," and "acrosome reaction," and energy productionrelated processes, such as "ATP metabolic process" and "glycolytic process," among others (p < 0.05 after FDR correction; **Table 2**). Regarding localization in the sperm cell itself, the 5 most enriched cellular components were "sperm flagellum," "acrosomal vesicle," "mitochondrial proton-transporting ATP synthase," "acrosomal membrane," and "outer dense fiber" (p < 0.05 after FDR correction; **Table 2**).

Looking more closely at the 36 proteins with altered interlysate concentrations, 13 differed significantly between DGC and CAP sperm (9 with decreased and 4 with increased protein levels in CAP sperm lysates; **Figures 3A–D** and **Supplementary Figure S1**), and 13 differed significantly between DGC and AR sperm (8 with reduced abundance and 5 with increased abundance in AR sperm lysates; **Figures 3C–F** and **Supplementary Figure S2**). Interestingly, the relative abundance of 14 proteins was only altered after induction of the acrosome reaction (AR sperm; **Figures 3B–E**). Of those, 9 proteins showed lower and 5 higher levels of abundance in AR lysates (**Figure 3E** and **Supplementary Figure S3**).

#### Functional Roles of Sperm Proteins Differentially Solubilized After Consecutive Incubation in Capacitation Medium and Following Induction of the Acrosome Reaction

The small number of proteins with altered levels of abundance identified for each of the sperm preparation groups limited the usefulness of GO enrichment analysis for identifying potential functional roles of these proteins. Instead, this was evaluated by searching in public databases and close examination of the available literature. Eleven functional categories were identified in this way, including "sperm motility," "fertilization," "energy production," "signaling," "detoxification/antioxidant response," "protein degradation," "protein folding," "vesicular trafficking," "metabolism of folic acid," "RNA biogenesis" and "unknown." **Figure 4** shows the distribution of proteins in each of the functional groups according to the stage of post-ejaculatory sperm processing. The specific category for each protein is indicated in **Table 1**.

Following incubation of the DGC sperm in capacitation medium, differences in their relative abundance were detected in proteins involved in sperm motility, fertilization, energy production and signaling (**Figure 4**). Induction of the acrosome reaction resulted in the reduction of the abundance of proteins involved mainly in fertilization, which includes sperm-oocyte recognition, binding and fusion (**Figure 4**). In addition, reduced levels of proteins with known roles in sperm motility, energy production and metabolism of folic acid were also identified in AR lysates (**Figure 4** and **Table 1**). Interestingly, proteins with increased relative abundance following the acrosome reaction were related to energy production, signaling, protein degradation and vesicular trafficking (**Figure 4**). Regarding differences observed between DGC and AR groups, which tests for changes induced by the combination of both incubation in capacitation medium and the acrosome reaction on DGC sperm, proteins with altered abundance related to detoxification/antioxidant response, protein folding and RNA biogenesis were also revealed (**Figure 4**).

To further explore the potential functional roles of the lysate proteins with altered post-processing abundance or possible synergies between them, protein-protein interactions between these proteins and all proteins associated to GO terms related to capacitation or acrosome reaction were assessed using the STRING database. This analysis highlighted the interaction of altered proteins not previously related to these processes with other proteins known to play important roles in the maturation of ejaculated sperm (**Table 3** and **Supplementary Figure S4**). For example, ERLIN2, which differed in abundance between DGC and AR conditions, is known to interact with CFTR, a protein associated with the GO term for capacitation. Likewise, STRING analysis revealed an interaction between ATP5F1A (also known as ATP5A), the relative abundance of which was altered following incubation in CAP medium, with DLD, a protein related with the GO of "capacitation" (**Table 3** and **Supplementary Figure S4**).


TABLE 1 | Quantified proteins with an altered abundance after the consecutive incubation with capacitation medium and in vitro induction of the acrosome reaction.

DGC-CAP: proteins differentially detected after incubation with capacitation medium; CAP-AR: proteins differentially detected after induction of the acrosome reaction; DGC-AR: proteins differentially detected after incubation of capacitation medium followed by the induction of the acrosome reaction; ↑: increased abundance; ↓: decreased abundance; Related function: according to public databases.

#### Role of PTMs and Isoforms in Protein Abundance Alterations Detected After Incubation in Capacitation Medium and Following Induction of the Acrosome Reaction

To identify a possible mechanism behind the relative changes in protein abundance following the consecutive incubation of sperm with capacitation medium and the induction of the acrosome reaction, the presence of PTMs in residues from peptides with significant inter-lysate differences in relative abundance was explored by using data from the PhosphoSitePlus <sup>R</sup> database. Potentially phosphorylated, acetylated, ubiquitinated, succinylated and sumoylated residues were identified (**Figure 5** and **Supplementary Table S4**). Interestingly, no PTMs were described for the differential peptides identified for proteins with reduced abundance following induction of the acrosome reaction (**Figure 5B**). However, retrieved data revealed modified residues for all the significantly altered peptides corresponding to proteins with an increased relative abundance in AR lysates (**Figure 5B**). Described PTMs were also attributed to peptides corresponding to altered proteins after sperm incubation in capacitation medium, or after consecutive incubation with capacitation medium and following induction of the acrosome reaction (**Figures 5A–C**).

TABLE 2 | Gene Ontology (GO) terms enrichment analysis.

fcell-07-00295 November 18, 2019 Time: 15:9 # 9


GO terms corresponding to biological processes and cellular components that are significantly enriched in the set of sperm proteins with altered abundance after incubation with capacitation medium and induction of the acrosome reaction. P-value < 0.05 after FDR correction.

In addition to the above observations, assignment of the peptides with altered abundance to specific protein isoforms was also undertaken. Three peptides were quantified for protein ATP5IF1, which showed increased protein levels only in comparison between DGC and AR sperm lysates. Following statistical analysis, the only peptide shown to differ significantly was exclusive for the ATP5IF1 isoform 1 (**Table 1** and **Supplementary Table S4**).

#### DISCUSSION

Studying the molecular processes that govern how ejaculated spermatozoa mature in the female genital tract and become competent to fertilize the oocyte is now a key component of basic research in the field of Andrology. The fundamental cellular and molecular mechanisms and their components driving capacitation and acrosome reaction processes, however, are not yet completely understood. Identifying sperm components displaying alterations in abundance during in vitro postejaculated sperm maturation could aid to unravel hidden molecular aspects which can impact on male fertility, and the use of novel high-throughput molecular approaches is a good strategy to help achieve this goal.

There are published reports that made use of MS to help identify protein differences before and after the induction of sperm capacitation in human and model species (Ficarro et al., 2003; Bailey et al., 2005; Sleight et al., 2005; Secciani et al., 2009; Baker et al., 2010; Zigo et al., 2013; Kwon et al., 2014; Wang J. et al., 2015; Hernández-Silva et al., 2019; Hou et al., 2019). However, the number of studies focusing on human sperm and using the most advanced techniques is limited. In addition, to the best of our knowledge, acrosomal exocytosis has not yet been evaluated from a proteomic perspective. In the present study, we applied, for the first time, a high-throughput proteomic procedure combined with isotopic peptide labeling to help characterize not only changes in the proteomic profile of healthy spermatozoa after in vitro incubation with capacitation medium, but also after the release of the acrosomal contents. It is important, however, to be mindful of the fact that human semen samples are biologically very heterogeneous and that among the sperm population, only a few sperm will be capable of becoming competent to fertilize the oocyte. Both intra- and inter-individual sample variance might influence the results of 'omics- based research and, therefore, setting strict criteria for data analysis becomes essential to improve the impact of these data (Lualdi and Fasano, 2019). In the current study, we applied robust statistics at both the protein and peptide levels and considered potential technical bias. This workflow led us to identify 36 sperm proteins with a remarkably altered abundance in sperm lysates as a consequence of the incubation with capacitation medium, the calcium ionophore-induced acrosome reaction or both joint processes.

In order to understand whether the apparent differences in protein abundance are truly reflective of novel factors that may be critical during the acquisition of sperm fertilization capacity, it is necessary to consider their potential causes and effects. This is highly relevant since spermatozoa are inactive at both the transcriptional and the translational levels, excluding the possibility of increased gene expression accounting for rises in the relative abundance of some proteins in sperm lysates and placing more emphasis on dynamic molecular changes due, perhaps, to protein modification, degradation or translocation (Signorelli et al., 2012; Jodar et al., 2013; Amaral et al., 2014a; Castillo et al., 2018). We found that incubation with capacitation medium induced quantitative changes in proteins involved mainly in sperm motility and signaling, which are processes regulated by PTMs. Indeed, it is widely thought that important cellular changes facilitating sperm motility during capacitation is mediated by tyrosine phosphorylation through protein kinase A (PKA) activation and the down-regulation of Ser/Thr phosphatases by Src family kinases (Ficarro et al., 2003; Visconti et al., 2011; Battistone et al., 2013; Signorelli et al., 2013; Wang J. et al., 2015). The A kinase anchoring proteins (AKAPS) are actively involved in PKA-dependent protein tyrosine phosphorylation, and a decrease in the relative interlysate abundance of AKAP3 and AKAP4 was found in the current study after incubation with capacitation medium. A requirement for AKAP3 degradation in sperm capacitation has been reported (Hillman et al., 2013; Vizel et al., 2015), which supports our results. Likewise, we found relative alterations in other proteins known to be involved in PKA-dependent signaling processes, including ROPN1B (Carr et al., 2001; Fiedler et al., 2013) and the acquisition of sperm motility, including PGK2 (Danshina et al., 2010; Shen et al., 2013; Liu et al., 2016; Huang et al., 2017), ODF2 (Huang et al., 2015; Wang X. et al., 2015; Zhao et al., 2018), and RSPH1 (also known as TSGA2) (Hui et al., 2006; Shetty et al., 2007; Onoufriadis et al., 2014).

(repeated-measures ANOVA with Holm–Sidak correction) and fold changes (log ratio of TMT intensities of each sample with the internal control) for all quantified proteins (under strict criteria) after incubation with sperm capacitation medium (DGC vs. CAP), induction of the acrosome reaction (CAP vs. AR) and the combination of both treatments (DGC vs. AR). Fold change of –1 represents reduction to half, and +1 increasing to double (vertical dot lines). Horizontal dotted lines are set at p-value = 0.05. (D–F) heatmaps showing the ratio of TMT intensities from each sample with the internal control corresponding to proteins with statistically significant differences in their abundance after the different sperm treatments (DGC, CAP, and AR). Proteins with altered abundance were grouped depending on the treatment leading to the significant differences (DGC vs. CAP, CAP vs. AR, DGC vs. AR). Horizontal thick line separates proteins decreasing in abundance after the treatment (top) from those detected with higher protein levels (bottom).

While several proteomic reports can be found focused on sperm capacitation, this is the first study applying advanced proteomic strategies to analyze protein changes following the release of the acrosomal content. It is important to take into consideration that, although calcium influx is required to initiate the process, the use of a calcium ionophore does not mimic the natural trigger of the acrosome reaction, which requires binding to the zona pellucida (Liu and Baker, 1996a,b). However, tests simulating a zona-mediated physiological acrosome reaction are restricted by the limited availability of enough biological material for proteomic studies. Calcium ionophore A23187 is therefore an acceptable alternative when aiming to describe protein changes, since the results in terms of the quantitative loss of acrosomal contents should be equivalent. We found 9 proteins with a lower abundance after the acrosome reaction compared with lysates from CAP sperm, while 5 proteins seemed to increase their relative protein levels. Of note, none of the peptides from proteins with a lower abundance after the acrosome reaction contained residues with potential PTMs in their sequence. In contrast, all proteins with relatively higher abundance in AR lysates could be targets of protein modifiers. These results suggest that all proteins with apparently lower abundance in AR lysates may be released or relocated during the acrosome reaction, while those with relative increases in abundance may represent modified proteins

involved in signaling processes that have lost the corresponding PTM. In support of this hypothesis, many of the proteins with diminished protein levels after induction of the acrosome reaction are known to be components of the acrosomal vesicle, including ACR, ACRBP, ARV1, and SPACA7 and, therefore, those might be released together with the acrosomal content. However, potential changes in the distribution of sperm proteins should also be considered, as it has been observed in previous studies using a number of different methods (Torabi et al., 2017a). The lysis buffer used in this study to solubilize sperm proteins most likely gained access to compartments of the sperm during and following capacitation that were hidden beforehand. A similar effect would lead to the release of proteins following the acrosome reaction where the acrosomal vesicle is lost. This does not imply that our data is a technical artifact (notwithstanding the status of albumin and semenogelin, both disregarded); instead, we propose that it is mainly the dynamic changes in cellular and structural aspects of the sperm occurring during capacitation and the acrosome reaction that is the main driver of these results, although active enzymatic destruction of sperm proteins during these processes may also contribute. It should also be recognised that successive transitional stages take place before the complete release of the acrosome contents (Buffone et al., 2008). Therefore, further studies considering different time points during the acrosome reaction would shed light into the functional involvement of these proteins.

synergic effect of incubation with capacitation medium followed by induction of the acrosome reaction.

The finding of alterations in the abundance of sperm proteins already known to be involved in sperm capacitation and the acrosome reaction supports the reliability of the strategy followed by this study. Moreover, it also increases the interest in those (altered) proteins never associated previously with sperm functionality. For example, our results suggest that, in our in vitro conditions, sperm maturation induced changes in the ER lipid raft-associated 2 (ERLIN2) protein, which is known to mediate degradation of inositol 1,4,5-triphosphate receptors (Wang Y. et al., 2009; Wright et al., 2018). Interestingly, STRING analysis revealed the interaction of ERLIN2 with the cystic fibrosis transmembrane conductance regulator protein (CFTR), which is essential during sperm capacitation through PKAdependent phosphorylation, alkalization and hyperpolarization (Puga Molina et al., 2017, 2018b). The reduction in the abundance of gamma-glytamyl hydrolase (GGH) in AR sperm is also noteworthy. The interaction analysis revealed the association of GGH with the T-complex protein 1 subunit beta (CCT2), which is a chaperone involved in sperm-ZP binding (Redgrove et al., 2011). Only an involvement of GGH in the metabolism of folic acid is reported in the literature (DeVos et al., 2008). However, since both GGH and CCT2 showed significant alterations in protein abundance after the process of acrosome reaction, GGH is suggested as a potential candidate for further study. In addition, GGH might also interact with Glutathione S-transferase Mu 3 protein (GSTM3), which has been previously related to capacitation and ZP binding using MS (Ficarro et al., 2003; Petit et al., 2013). Another novel protein thought to be involved in the process of the acrosome reaction is the Transmembrane em24 domaincontaining protein 10 (TMED10), which, although not previously associated with sperm function, may be involved in vesicular trafficking in other cellular systems (Nakano et al., 2017; Luteijn et al., 2019).

TABLE 3 | Protein–protein interactions between altered proteins detected in this study and proteins associated to GO terms of capacitation and acrosome reaction.


Proteins found with altered abundance in this study are highlighted in gray. Confidence score > 0.7 is set as high by STRING database.

Sperm capacitation and the acrosome reaction are complex processes involving many proteins and signaling pathways. Indeed, the sheer complexity precludes our obtaining a complete proteomic picture of these processes in just one MS-based study. A combination of data from high-throughput studies applying different approaches, such as enrichment in peptides with specific PTMs, the conduction of cell fractionation or the use of isobaric tags for protein quantification, among others, is required (Amaral et al., 2014a; Codina et al., 2015). In this study, we have applied

corresponding to proteins identified with significantly different abundance after incubation with capacitation medium and in vitro induction of the acrosome reaction. (A) proteins found with altered abundance after incubation with (Continued)

#### FIGURE 5 | Continued

fcell-07-00295 November 18, 2019 Time: 15:9 # 13

capacitation medium. (B) proteins found with altered abundance after induction of the acrosome reaction. (C) proteins found with altered abundance after the synergic effect of incubation in capacitation medium followed by induction of the acrosome reaction. Arrows indicate whether the abundance of the proteins was increased or decreased after incubation with capacitation medium and in vitro induction of the acrosome reaction.

TMT labeling to quantify changes in sperm proteins in response to stimuli leading to an in vitro simulation of post-gonadal sperm maturation. We believe that our results increase the current knowledge about maturation of ejaculated sperm and suggest new players in the process. However, further experiments at peptide level are now required. By dissected the process of capacitation and the acrosome reaction at the peptide level, there is the possibility of understanding the roles of these proteins during the acquisition of competence to fertilize the oocyte and to confirm both the presence of modifications and alterations thereof in infertile patients. In addition, our results highlight the limitations of the currently available in vitro methods to mimic the in vivo situation. Capacitation and the acrosome reaction occur within the female reproductive system and thus, secretions from and components of the female tract may have an impact on sperm acquisition of fertilizing capacity. Therefore, the development of 3-D culture systems that mimic the oviduct and provide an in vivo-like environment to the sperm cell are desirable (Ferraz et al., 2017). A better understanding of the molecular mechanism mediating fertilization is essential in order to improve current infertility diagnosis and treatment.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study can be found in the ProteomeXchange Consortium via the PRIDE partner repository, dataset identifier PXD014871.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Clinical Research Ethics Committee of the Hospital Clínic of Barcelona. The donors provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

OB, FT, and RO designed the study. JB was involved in donor samples collection. JC, OB, and FT performed the research. JE performed the proteomic technology. JC, MJ, FT, and DD-D analyzed the data. JC, MJ, and RO interpreted the data. JC, MJ, FT, DM, and RO drafted the manuscript. All authors critically reviewed and approved the final version of the manuscript.

## FUNDING

This work was supported by grants to RO from the Spanish Ministry of Economy and Competitiveness (Ministerio de Economía y Competividad; fondos FEDER 'una manera de hacer Europa' PI13/00699, and PI16/00346), from Fundación Salud 2000 (SERONO 13-015), and from EU-FP7-PEOPLE-2011- ITN289880. JC is supported by the Sara Borrell Postdoctoral Fellowship from the Spanish Ministry of Economy and Competitiveness (Ministerio de Economía y Competitividad, Acción Estratégica en Salud, CD17/00109). MJ is granted by Government of Catalonia (Generalitat de Catalunya, pla estratègic de recerca i innovació en salut, PERIS 2016-2020, SLT002/1600337).

### ACKNOWLEDGMENTS

We thank Raquel Ferreti and Alicia Diez for their assistance in the collection and routine analysis of semen samples.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00295/ full#supplementary-material

FIGURE S1 | Proteins detected with altered abundance after incubation of the sperm cells with capacitation medium (DGC-CAP). The mean of the ratio between TMT intensities of the three donors and the internal control are shown for each protein at each sperm condition.

FIGURE S2 | Proteins detected with altered abundance after the combination of sperm incubation with capacitation medium and induction of the acrosome reaction (DGC-AR). The mean of the ratio between TMT intensities of the three donors and the internal control are shown for each protein at each sperm condition.

FIGURE S3 | Proteins detected with altered abundance after the induction of the acrosome reaction (CAP-AR). The mean of the ratio between TMT intensities of the three donors and the internal control are shown for each protein at each sperm condition.

FIGURE S4 | Protein–protein interaction networks between proteins found with altered abundance in this study and all those proteins associated with Gene Ontology terms related to capacitation and acrosome reaction. Proteins known to be functionally associated to the processes of capacitation and acrosome reaction were retrieved from the Gene Ontology Consortium database and submitted to STRING database together with the list of proteins found in this study with statistical significant differences in abundance. Only those protein-protein interactions with a high confidential score (>0.7) are shown.

TABLE S1 | Seminal parameters of the semen samples included in the study and obtained from 3 fertile donors.

TABLE S2 | Proteins identified in the three sperm donors and three sperm conditions by LC-MS/MS. The number of identified peptides, PSMs at both TMT runs, quantified peptides in all samples and quantified peptides that fit the strict selection criteria for statistics are indicated.

TABLE S3 | Peptides identified by LC-MS/MS containing quantification values.

TABLE S4 | Proteins identified with altered abundance. P-values after one way ANOVA for repeated measures and t-test with Holm–Sidak adjustment are provided for both proteins and peptides. P-values <0.05 are highlighted in green.

#### REFERENCES

fcell-07-00295 November 18, 2019 Time: 15:9 # 14




degradation of inositol 1,4,5-trisphosphate receptors in muscarinic receptorexpressing HeLa cells. Biochim. Biophys. Acta 1793, 1710–1718. doi: 10.1016/j. bbamcr.2009.09.004


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Castillo, Bogle, Jodar, Torabi, Delgado-Dueñas, Estanyol, Ballescà, Miller and Oliva. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Novel Techniques of Sperm Selection for Improving IVF and ICSI Outcomes

Iván Oseguera-López<sup>1</sup>† , Sara Ruiz-Díaz2,3† , Priscila Ramos-Ibeas<sup>3</sup> and Serafín Pérez-Cerezales<sup>3</sup> \*

<sup>1</sup> Unidad Iztapalapa, Universidad Autónoma Metropolitana, Mexico City, Mexico, <sup>2</sup> Mistral Fertility Clinics S.L., Clínica Tambre, Madrid, Spain, <sup>3</sup> Department of Animal Reproduction, Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria, Madrid, Spain

Almost 50% of the infertility cases are due to male factors. Assisted reproductive technologies (ARTs) allow to overcome the incapacity of these patients' spermatozoa to fertilize the oocyte and produce a viable and healthy offspring, but the efficiency of the different techniques has still the potential to improve. According to the latest reports of the European Society of Human Reproduction and Embryology (ESHRE) and the Centers for Disease Control and Prevention of the United States (CDC), the percentages of deliveries per ART cycle in 2014 and 2016 were 21 and 22%, respectively. Among the reasons for this relatively low efficiency, the quality of the spermatozoa has been pointed out as critical, and the presence of high percentages of DNA-damaged spermatozoa in patients' ejaculates is possibly one of the main factors reducing the ARTs outcomes. Thus, one of the main challenges in reproductive medicine is to ensure the highest quality of the spermatozoa used in ARTs, and specifically, in terms of genetic integrity. The latest techniques for the preparation and selection of human spermatozoa are herein discussed focusing on those proven to improve one or several of the following parameters: sperm genetic integrity, fertilization capacity, embryo production, and in vitro survival, as well as pregnancy and delivery rates following in vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI). In addition, we discuss the potential of techniques developed in non-human mammals that could be further transferred to the clinic.

Keywords: sperm selection, ICSI, IVF, ARTs, sperm quality

## INTRODUCTION

#### Current Efficiency of ARTs

According to the World Health Organization, infertility is defined as "a disease characterized by the failure to establish a clinical pregnancy after 12 months of regular, unprotected sexual intercourse or due to an impairment of a person's capacity to reproduce either as an individual or with his/her partner" (Zegers-Hochschild et al., 2017). Around 15–20% of couples within the reproductive age are infertile and in ∼50% of the cases, the male factor is present. Furthermore, in 30–40% of cases, the male infertility is idiopathic (Katz et al., 2017). In order to overcome infertility, a number of clinical treatments have been successfully developed and are englobed in what is known as assisted reproductive technologies (ARTs). These involve in vitro handling of oocytes, sperm, and embryos for their use in reproduction and comprise in vitro fertilization (IVF),

#### Edited by:

Zhibing Zhang, Virginia Commonwealth University, United States

#### Reviewed by:

Hongmin Qin, Texas A&M University, United States Fan Jin, Zhejiang University, China

> \*Correspondence: Serafín Pérez-Cerezales perez.serafin@inia.es; s.perez.cerezales@gmail.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 28 June 2019 Accepted: 07 November 2019 Published: 29 November 2019

#### Citation:

Oseguera-López I, Ruiz-Díaz S, Ramos-Ibeas P and Pérez-Cerezales S (2019) Novel Techniques of Sperm Selection for Improving IVF and ICSI Outcomes. Front. Cell Dev. Biol. 7:298. doi: 10.3389/fcell.2019.00298

embryo transfer, intracytoplasmic sperm injection (ICSI), embryo biopsy, preimplantation genetic testing, and gamete and embryo cryopreservation. The use of ARTs has been steadily rising in the developed countries in last decades (Ferraretti et al., 2017) and from 1997 to 2014, at least 1.5 million babies have been conceived by these techniques in Europe (De Geyter et al., 2018). ARTs are also used in the veterinarian clinical practice and livestock production. However, assisted insemination (AI) is still the preferential reproductive procedure in breeding programs because of its simplicity and effectiveness (Hansen, 2014; Knox, 2016). In cattle embryo production, because of its high efficiency, IVF is intensively used in routine breeding management and millions of calves have been born by this technique (Sirard, 2018). In contrast, ARTs are not applied in sheep and goat farming although the technology has already been developed (Paramio and Izquierdo, 2014). In porcine, the main disadvantage when producing embryos by IVF is the high level of polyspermy (Nagai et al., 2006). This problem can be bypassed by ICSI since normal blastocysts can be produced by this technique with the same efficiency than IVF (Wu et al., 2001). In livestock species, ICSI is not used to overcome male infertility unless the male has a high value, as for endangered species or stallion (Galli et al., 2014; Herrick, 2019), because of the low efficiencies of embryo production achieved in farm animals with this technique (Salamone et al., 2017). In stallion, only two foals have been reported to born following IVF using in vivo matured oocytes. As a result in this species, ICSI is the only alternative and has allowed the production of foals from in vitro matured oocytes mainly in the clinical management of infertile individuals (Galli et al., 2014).

Despite the wide use of ARTs in the last decades, their efficiency still has the potential to improve. Using both techniques (ICSI or IVF), the latest efficiencies reported by the American and European societies of reproduction and fertility are around 37% of pregnancies and 25% of deliveries per embryo transfer (CDC, 2018; De Geyter et al., 2018). These percentages were higher when fresh oocytes from donors were used, reporting 54 and 38% of pregnancies and deliveries per embryo transfer, respectively. In livestock species, ARTs and specially ICSI are also inefficient (García-Roselló et al., 2009; Salamone et al., 2017) or have not been successfully developed, as in the case of IVF in horse (Leemans et al., 2016) and ICSI in cattle (Li et al., 2004). These low efficiencies can be due to multiple factors such as suboptimal in vitro conditions, injuries associated to gametes and embryo manipulation, subjacent male and female factors, etc. We overview herein the different methods recently developed in order to improve ARTs, focusing on the strategy of spermatozoa selection.

#### Why It Is Important to Select the Spermatozoa? Semen Quality and ARTs Outcome

After the copula, from the millions of spermatozoa ejaculated, only a small number of about few hundreds make it to the region of the oviduct (ampulla), where they encounter the egg and fertilization occurs (Williams et al., 1993; Hino et al., 2016). Presumably, this subpopulation has been selected through the oviduct in a way that only those with the highest fertilization capability and the best features for supporting embryo development get the opportunity to fertilize the egg (Sakkas et al., 2015; Pérez-Cerezales et al., 2017). Thus, it has been proposed that one of the reasons for the relatively low efficiency of ARTs is that we currently lack an effective methodology to separate this specific sperm subpopulation for its use in ARTs (Sakkas et al., 2015; Pérez-Cerezales et al., 2017). This is especially relevant if we consider that both IVF and ICSI bypass the sperm selection operating in vivo, increasing the risk of fertilizing the oocyte with defective spermatozoa that could lead to developmental failure and even affect the offspring in the long run (Fernandez-Gonzalez et al., 2008). This risk could be higher in the clinical practice since the incidence of sperm abnormalities, including DNA fragmentation, is higher in infertile men (Saleh et al., 2003; Ozmen et al., 2007; Schulte et al., 2010).

The extent to which the initial quality of the sperm sample affects ARTs success is not clear. It has been generally accepted that depending on the nature of the male factor, IVF or ICSI should be followed as a preferable treatment. In contrast to IVF, ICSI bypasses the last barriers that the spermatozoa have to overcome during fertilization, increasing the risk of fertilizing the oocyte with defective spermatozoa. Therefore, the recommendation is to use always IVF as first choice and ICSI only in the following cases: testicular and epididymal spermatozoa, immotile but viable spermatozoa, asthenozoospermia, globozoospermia, teratozoospermia, and with frozen spermatozoa (Tournaye, 2012). However, because ICSI has been proven to report the same efficiencies as IVF (Schwarze et al., 2017), nowadays this is the most used technique in fertility treatments in developed countries, replacing IVF as the first choice (71.3 and 72% of total cycles were performed by ICSI in Europe in 2014 and in United States in 2016, respectively) (CDC, 2018; De Geyter et al., 2018). This goes against the effectiveness of the last spermatozoa selection barriers imposed by the oocyte during IVF. In a former study, Jones et al. (1998) reported a negative correlation between the presence of male infertility factor and the IVF outcome in terms of blastocyst production. However, in this study and another one conducted by Boulet et al. (2015), no differences were found in final ratios of live births among patients with or without male factor using IVF or ICSI (Jones et al., 1998; Boulet et al., 2015). Recently, Chapuis et al. (2017) reported that using ICSI, patients showing severe oligospermia showed reduced blastulation rate, whereas a reduced progressive motility affected the fertilization and cleavage rates when IVF was performed. Pregnancy rates also decreased in both IVF and ICSI when the father was older than 51, although female age could also have conditioned these results.

Chromatin integrity is the most studied parameter out of the classical and routinely evaluated features of spermatozoa. However, its power to predict ARTs outcome is also under discussion and there are a number of publications showing opposite results as reviewed by Schulte et al. (2010). Latter reviews including a meta-analysis from existing literature corroborate that abnormal chromatin is associated to male

infertility and suggest a negative correlation with the probability of ARTs to succeed (Simon et al., 2017b,c). Recent works proved that sperm DNA fragmentation delays embryo cleavage when donated oocytes are used for ICSI but does not affect the final quality of the embryo (Esbert et al., 2018; Casanovas et al., 2019). Also recently, Jerre et al. (2019) retrospectively analyzed the use of semen with different degrees of chromatin packaging for IVF and ICSI in 1602 pregnancies and found that, above certain threshold, spermatozoa porting immature chromatin slightly increased the risk of early miscarriage. The contradictory results reported in the literature could be related to the different methods used for evaluating genetic integrity, all susceptible of variability between different labs. For further insight into the principles of the methods employed for the determination of chromatin integrity, their advantages and disadvantages, and their correlation with male infertility, we address the reader to specific reviews available in the literature (Raheem and Walsh, 2016; Majzoub et al., 2019).

The difficulties to establish a clear correlation among the presence of male factor, sperm chromatin status, and ARTs success could be due to the multifactorial nature of fertility, where the oocyte, spermatozoid, endometrial environment, etc., are individually determinant on pregnancy failure. An underevaluation of the sperm quality could also be withholding this correlation, since different features are currently not taken into consideration in andrology laboratories as a routine, due to the complexity of the required analysis and to the lack of basic research on molecular markers of sperm quality. Recent studies have also revealed the complex composition of the ejaculate at different levels including epigenetics, indicating that a number of sperm subpopulations with a wide diversity of features coexist (see the next section). Thus, further research should be done to identify additional markers in the spermatozoa related to ARTs success that could serve to isolate the sperm subpopulation whose features support a better embryo development and pregnancy to term.

#### Sperm Heterogeneity

Any given sperm sample shows a complex heterogeneity revealed at different levels. The most obvious intra-sample diversity can be directly seen under the bright field microscope as different motions are shown by each spermatozoon. A closest examination also reveals a mosaic of subpopulations regarding various morphological features. This heterogeneity of the sperm motility and morphology is objectively confirmed by automatized techniques of sperm tracking and analysis (i.e., CASA and CAMA) that classify the spermatozoa into different subpopulations according to specific kinetical parameters (Martínez-Pastor et al., 2011) and morphologies (Yániz et al., 2015; Soler et al., 2016). Moreover, the composition of sperm subpopulations regarding motility is dynamic throughout time. For instance, in vitro conditions for capacitation cause time-dependent changes in the spermatozoa at the cellular level that can be recorded as changes in motility, such as the acquisition of the hyperactive motility. This motility type involves vigorous movements under low-viscous conditions produced by asymmetrical and high-amplitude waves in the flagella and resulting in erratic swimming trajectories (Suarez, 2008). However, not all the spermatozoa acquire the hyperactive motility under capacitation conditions, reaching a maximum of around 20% in humans (Burkman, 1984). Furthermore, human spermatozoa also show a heterogeneous response to chemical inductors of hyperactivation. Ooi et al. (2014), following a single cell analysis by high-speed video recording of adhered human spermatozoa, showed that in response to the potent hyperactivation inducer 4-aminopyridine (4AP), multiple patterns of flagellar beating are produced in each spermatozoa, configuring complex and heterogeneous responses to this type of stimulus. Moreover, using the chemoattractant progesterone, another inductor of hyperactivation, Armon and Eisenbach (2011) showed a variety of responses of individual spermatozoa within few seconds immediately after the sperm–progesterone interaction. These observations suggest the existence of different sperm subpopulations, each with a specific sensitivity to respond by chemotaxis, a mechanism leading the spermatozoa to orient their swimming within a chemical gradient (Eisenbach, 1999). In accordance, only 2–12% of the human spermatozoa are able to align the direction of their swimming toward follicular fluid in a chemotactical response (Cohen-Dayag et al., 1994) and only ∼5% of human spermatozoa migrate toward the warmer temperature in response to thermotaxis (Bahat et al., 2012), a mechanism leading the spermatozoa to orient their swimming within a temperature gradient (Bahat et al., 2003).

Sperm heterogeneity has been reported at every level and it can be noticed almost in any study and analysis. Accordingly, a number of research studies hypothesize that only a small subpopulation of spermatozoa within the ejaculate retain the ability to achieve fertilization (Holt and Fazeli, 2015; Sakkas et al., 2015; Pérez-Cerezales et al., 2017). This determines that a high number of spermatozoa are needed to be placed directly onto the oocyte for achieving fertilization in vitro. The standard ratio for IVF in humans is 50,000 or more motile sperm per oocyte (Stephens et al., 2013); however, the lower limits of this ratio for achieving fertilization depend on the sperm quality. For example, in a study with subfertile male patients showing diverse sperm qualities, the use of 5000 sperm/oocyte resulted in 37% of zygotes from total oocytes, a percentage that was significantly increased to 60% when using 20,000 sperm/oocyte (Tournaye et al., 2002). The meta-analysis conducted on available publications reporting randomized controlled trials for the treatment of male subfertility evidences a large variability of fertilization ratios, ranging from 50,000 to 10 × 10<sup>6</sup> sperm/oocyte for IVF (Tournaye et al., 2002). In mice, this limit was reduced to 5 motile sperm/oocyte achieving 60% of fertilization by prolonging the capacitated status of the spermatozoa via creatine supplementation of the medium (Umehara et al., 2018). Therefore, these results also indicate a transient availability of capacitated and "fertile spermatozoa" within the semen. Overall, these data indicate that from the whole spermatozoa population, only a small fraction is susceptible of acquiring the capacity to achieve fertilization at a given time point under in vitro conditions.

This observable intra-sample heterogeneity is surely determined by underlying differences among spermatozoa at the cellular and molecular levels. Using discontinuous Percoll gradient, Buffone et al. (2004) found at the different fractions (45, 65, and 90% of Percoll) that human spermatozoa possessed different characteristics regarding quality (motility and morphology) and capacity to show hyperactive motility and protein tyrosine phosphorylation in response to incubation under capacitating conditions. A recent study on bull sperm reported that within two different sperm populations separated by density gradient centrifugation (DGC), there were 31 proteins more abundant in low motile spermatozoa, while 80 proteins were more abundant in the high motile population (D'Amours et al., 2018). Following discontinuous gradient centrifugation of spermatozoa from normozoospermic men, Jenkins et al. (2014) proved the existence of differentially methylated regions in the DNA between high-quality and low-quality fractions. Immunocytological studies on the location of opsins in human and mouse spermatozoa revealed heterogeneous staining patterns for each of the studied proteins involved in the thermotaxis response (Pérez-Cerezales et al., 2015) and reported that a specific subpopulation (∼15% of the ejaculate) showing rhodopsin at a specific cellular location is the one with the capability of migrating toward the higher temperature within a gradient (Pérez-Cerezales et al., 2018).

The heterogeneity of the spermatozoa is also reflected at the genetic and epigenetic levels. Thus, single cell analysis of DNA fragmentation by the Comet assay reveals high intra-sample heterogeneity in human spermatozoa, as described by Simon et al. (2017a). We have observed as well that the level of DNA fragmentation of individual spermatozoa ranged from 0 to 70% in normozoospermic men and from 0 to 65% in epididymal mouse spermatozoa (Pérez-Cerezales et al., 2018). The sperm sample also shows various subpopulations at the chromatin packaging level, as evaluated by the sperm chromatin structure assay (SCSA). This assay discriminates subpopulations with high and low chromatin compaction, with different degrees among individual spermatozoids (Evenson, 2016). Moreover, about 9% of the human spermatozoa port some chromosomic alteration (Martin, 2008) and a high percentage of them carries punctual mutations (Wang et al., 2012). In addition, single cell analysis of the telomere length of human spermatozoa also revealed intra-sample heterogeneity with no differences between normal or abnormal spermatozoa (Antunes et al., 2015). These new findings included in the –omics fields reveal the high level of complexity of the ejaculate and the intimate association between these features and male infertility (for a comprehensive review on sperm –omics, we direct the reader to Sinha et al., 2017). The deep study of the genome, epigenome, transcriptome, proteome, and metabolome at the single cell level or in specific sperm subpopulations is expected to reveal higher levels of complexity than the currently known. This type of analyses could help to identify and correlate the presence/absence of specific sperm subpopulations with fertility, and these subpopulations could also be selected for improving ARTs outcome.

In conclusion, it is widely accepted that not all the spermatozoa from an ejaculate are equally good for achieving fertilization in vivo or in vitro (Holt and Fazeli, 2015). Due to the heterogeneity of the ejaculate, sperm selection prior to ARTs has been considered an important step for ensuring a successful outcome and a strategy to improve ARTs efficiency.

#### ROUTINE PREPARATION OF SEMEN FOR ARTs

Sperm selection/preparation techniques should be operatively simple and economic in order to fit in the routines of the human and veterinary clinics. Also, they always must ensure the enrichment of the sample in high-quality spermatozoa in the shortest time possible. Besides removing low quality spermatozoa, including those immotile, sperm preparation techniques should allow to eliminate other cells such as leukocytes and bacteria, as well as toxic or bioactive substances like reactive oxygen species (ROS) (Henkel and Schill, 2003).

Because they satisfy all these requirements, swim-up (SU) and DGC (**Figure 1**) are the most extended techniques for the preparation of spermatozoa (Henkel and Schill, 2003). SU was first described by Mahadevan and Baker (1984) and the general principle is the recovery of motile spermatozoa that migrate toward a cells-free medium usually placed above the sperm sample. Different variants of SU are available involving centrifugation (Mahadevan and Baker, 1984), straight migration from unprocessed semen (Aitken and Clarkson, 1988), recovery of spermatozoa from non-resuspended pellet (Carreras et al., 1990), or spermatozoa sedimentation by gravity prior to SU [migration sedimentation (MS) method] (Kiratli et al., 2018). On the other hand, DGC is based on the capacity of motile spermatozoa to progress through a gradient of density constituted by colloidal particles during centrifugation (Henkel and Schill, 2003). Variants of the method consist in different types of gradient: continuous or discontinuous (Henkel and Schill, 2003) and substances used to generate the gradient of density: ficoll (Harrison, 1976), PVP-coated silica particles or Percoll, or the current commercial variants such as IxaPrep <sup>R</sup> (MediCult, Copenhagen, Denmark), SilSelect <sup>R</sup> (FertiPro N.V., Beernem, Belgium), PureSperm <sup>R</sup> (NidaCon Laboratories AB, Gothenburg, Sweden), or ISolate <sup>R</sup> (Irvine Scientific, Santa Ana, CA, United States) (Henkel and Schill, 2003) used in human clinics to substitute Percoll due to its toxicity (Mousset-Siméon et al., 2004) and side effects on the sperm function (Strehler et al., 1998). In all the SU and DGC variants, it has been shown that the recovered sample is enriched in motile spermatozoa showing normal morphology. It has even been observed that both techniques select those spermatozoa with the longest telomeres (Zhao et al., 2016), this being an indicator of correct spermatogenesis (Rocca et al., 2016).

A number of studies have tried to clarify which of these two methods is more efficient; however, the results available in the literature are multiple and contradictory. In a first study approaching sperm selection for ICSI in patients with cryptozoospermia or severe oligoasthenoteratozoospermia (OAT), Sanchez et al. (1996) showed that the MS method (a type of SU) yielded a sperm fraction with better motility, vitality,

morphology, and chromatin condensation than when using minipercoll (a type of DGC), achieving 39.7% of pregnancies (n = 159 cycles). Also, Zini et al. (2000) showed that DGC is less efficient than SU for the separation out of spermatozoa with less DNA integrity. These results were contradictory to those reported by Sakkas et al. (2000) indicating the opposite. Amiri et al. (2012) reported that spermatozoa from normozoospermic donors selected by DGC showed less DNA fragmentation, higher normal morphology, and higher motility than when selected by SU (24 vs 32%, 28 vs 14%, and 76 vs 52%, respectively). Supporting these results, Wang et al. (2014) proved that DGC reduces DNA fragmentation in patients with oligozoospermia and astenozoospermia. Karamahmutoglu et al. (2014) reported that the use of spermatozoa of subfertile patients selected by DCG for intrauterine insemination significantly improved the pregnancy rate compared to the use of SU (18 vs 7%). In contrast, Oguz et al. (2018) reported that SU enriched the sample in spermatozoa with less DNA fragmentation in patients with mild or idiopathic male factor, while DGC did not allow such enrichment in those same patients. A variant of SU, the direct micro-SU, has shown comparable fertilization percentages by ICSI to those of DCG but higher blastocyst development in vitro and pregnancy rates (42 vs 26%, respectively), also reducing the abortion rate (13 vs 29%, respectively) (Palini et al., 2017).

In livestock species, SU and DGC are also the most used techniques for sperm selection (Arias et al., 2017) and although Percoll is currently avoided in human ART because of safety issues, as above indicated, it is still a method of choice for animal ART (Cesari et al., 2006). In rams, it has been reported that SU gives better results in terms of reducing the presence of apoptotic spermatozoa vs DGC with Percoll (67 vs 72%, respectively). However, DGC with Percoll delivered a higher percentage of capacitated spermatozoa than the SU method (Martí et al., 2006). In bovine, when comparing DGC (BoviPureTM) (Nidacon, Sweden) vs SU for selecting frozenthawed spermatozoa, BoviPureTM resulted in spermatozoa showing a greater progressive motility and viability, as well as a higher blastocyst yield (31.79 vs 21.91%, respectively) (Samardzija et al., 2006). Arias et al. (2017) reported that Percoll gradient recovered more spermatozoa with intact plasma and acrosomal membranes (89.8 and 87.5%) than BoviPure gradient (83.3 and 80.4%), but ROS levels were higher with Percoll separation. In addition, Percoll gradient resulted in lower blastocyst yield in comparison to BoviPureTM (∼20 vs 30%, respectively) using frozen-thawed semen (Lee et al., 2009). Thus, BoviPureTM should be considered the best frozen-thawed sperm selection technique for in vitro production of bovine embryos.

Both SU and DGC are regularly used in laboratories around the world prior to IVF and ICSI but, as discussed in the section "Introduction," the success rates of both techniques should be improved in both human and veterinary practices. Both methodologies for sperm selection are exclusively based on the motile capacity of sperm, which does not mean that all motile sperm are of the highest quality. For instance, it has been pointed that both SU and DGC are not efficient methods to select spermatozoa in terms of apoptosis, DNA integrity, membrane maturation, and sperm ultrastructure (Said and Land, 2011). Furthermore, centrifugation steps inherent to both techniques generate ROS, which have a detrimental effect on sperm quality (Aitken and Clarkson, 1988; Henkel and Schill, 2003). Therefore, new sperm selection methods should be based on sperm characteristics that are better connected to fertilization ability and quality of the spermatozoa, as well as on their contribution to support embryo development to term.

#### DIRECT SELECTION OF IMMOTILE SPERM

It is important to note that samples in which spermatozoa are immotile, or where the percentage of motile spermatozoa is low, are not suitable for SU and DGC methodologies. This is the case of sperm samples obtained by testicular sperm extraction (TESE) in azoospermic (AO) patients. Nowadays, in the routine practice of fertility clinics, only vague and subjective morphological

criteria are followed to select the immotile spermatozoa prior to ICSI, like the identification of spermatozoa with normal head and tail. However, in the last decades, a number of efforts have been made in order to find a selection method for this type of sample, aiming to discriminate viable spermatozoa directly under the micromanipulator irrespectively of their motility.

Marques de Oliveira et al. (2004) proposed the mechanical touch technique (MTT). According to these authors, MTT allows to identify immotile but viable spermatozoa because their tail is flexible when applying a lateral force to the flagella with the ICSI micropipette. In contrast, flagella of non-viable spermatozoa remain rigid to the same force. These authors conducted the only published clinical trial of MTT and found no differences in the ICSI outcome, including deliveries, between motile spermatozoa, and those immotile selected by MTT, in both fresh or frozen samples (Marques de Oliveira et al., 2004). Unfortunately, this study was never corroborated and completed by others. It would be desirable to report as well the comparison between the use of viable vs non-viable spermatozoa selected by MTT for ICSI. Without this information, it is difficult to envisage the actual utility of this technique.

Another sperm selection method that has been investigated is the hipo-osmotic swelling (HOS) test. This technique is based on the fundament that the tails of viable spermatozoa swell or curl under hypo-osmotic conditions due to normal membrane function, thus allowing their identification and recovery under the microscope, potentially for their use for ICSI (Verheyen et al., 1997). Sallam et al. (2005) conducted a randomized study with 79 couples following TESE-ICSI. In this study, where both frozen and fresh samples with total absence of sperm motility were used, pregnancy rates were 20.5% for the 44 couples following HOS and 2.9% for the 35 couples following morphological evaluation (normal head and tail) prior to ICSI. Fertilization and grade I and II embryos rates were also significantly higher in the HOS group. However, when data were analyzed separately, considering the type of sperm sample (fresh or frozen), pregnancy results were not significantly different.

Under polarized light microscopy, the head of viable spermatozoa is birefringent. On this basis, Baccetti (2004) proposed to use this property as a parameter for sperm selection. Indeed, the selection of immotile spermatozoa with birefringent head for ICSI has shown to increase clinical pregnancy and implantation rates compared to control immotile spermatozoa (58 vs 9% and 42 vs 12%, respectively) (Gianaroli et al., 2008) and to increase clinical pregnancy when compared to those selected by HOS (45 vs 11%, respectively) (Ghosh et al., 2012).

The use of chemical inducers of sperm motility has also been investigated. Thus, phosphodiesterase inhibitors of the xanthine family, such as pentoxifylline (PTF), demethylxanthine theophylline (TPF), and papverine have shown to activate the motility in a fraction of immotile testicular sperm (Ta¸sdemir et al., 1998; Ebner et al., 2011; Terriou et al., 2015). The use of this motile fraction, whether using PTF, TPF, or papverine, has allowed to achieve normal fertilization, pregnancy, and birth after ICSI (Terriou et al., 2000; Kovaci ˇ c et al., 2006 ˇ ; Amer et al., 2013; Sandi-Monroy et al., 2019). Furthermore, Mangoli et al. (2011) showed that the use of PTF compared with HOS selected spermatozoa, yielded higher fertilization rate, and doubled clinical pregnancies (32 and 16%, respectively, n = 25). Interestingly, the successful treatment of male infertility associated to the Kartagener's syndrome has been reported with both PTF and HOS, achieving the delivery of healthy babies in ICSI cycles (Hattori et al., 2011; Montjean et al., 2015). These findings show the therapeutic potential of both methodologies that could be useful for the treatment of other male infertility cases.

Motility of immotile but vital spermatozoa can also be induced with a single laser shot to the tip of the flagellum (Aktan et al., 2004). Thus, the laser-assisted immotile sperm selection (LAISS) has been proposed as an alternative to the use of chemicals like xanthines, avoiding in this way their potential toxic effects. This method has shown to significantly increase cleavage and birth rates after ICSI using both testicular and ejaculated sperm when compared to control groups (Aktan et al., 2004; Nordhoff et al., 2013). Moreover, LAISS has shown its utility to restore fertility following ICSI in specific cases of male patients showing primary cilia dyskinesia (Gerber et al., 2008) and the Kartagener's syndrome (Ozkavukcu et al., 2018). Despite the potential of this technique, its complexity and cost have possibly prevented its use in clinical routines and explain the scarce number of published studies.

In addition to these specific procedures, the methodologies described in the following sections, that do not require sperm motility, in principle are also suitable for sperm selection in immotile samples, although their efficiency for this type of samples should be further explored.

#### SPERM SELECTION BASED ON MEMBRANE CHARACTERISTICS

The outer membrane of the spermatozoa is crucial for their functionality as it takes part in a number of basic processes such as cell metabolism, capacitation, ova binding, acrosome reaction, etc. Its accessibility and relation with sperm vitality and quality make this organelle, its integrity, and variable characteristics, a logic target for the development of selection methods of highquality spermatozoa for their use in ARTs.

## Annexin V Magnetic Activated Cell Sorting (AV-MACS)

Magnetic activated cell sorting (MACS) is a method that allows the separation of cell populations based on their surface antigens (Plouffe et al., 2015). Coating magnetic nanoparticles with a molecule with affinity for these antigens allows the trapping of the desired cell subpopulation and its separation within a column subject to a strong magnetic field. Thus, cells expressing the antigen stay in the column while other cells flow through for downstream applications. The loss of membrane integrity is an early event of the apoptotic response

and leads to the externalization of phosphatidylserine (Elmore, 2007). On this basis, magnetic nanoparticles coated with Annexin V, a molecule showing high affinity for phosphatidylserine (Vermes et al., 1995), bind to apoptotic spermatozoa that are retained in the MACS column, allowing the recovery of non-apoptotic spermatozoa for their use in ARTs (Gil et al., 2013) (**Figure 2A**).

The use of Annexin V MACS (AV-MACS) for sperm selection in the human clinical practice was first reported by Grunewald et al. (2001), demonstrating the utility of this technique for the enrichment of the sample with non-apoptotic spermatozoa retaining membrane integrity. Since then, a number of studies have shown the enrichment of high-quality spermatozoa with reduced levels of DNA fragmentation using AV-MACS (**Table 1**). Said et al. (2006) employed AV-MACS for the selection of semen from 35 healthy donors, reporting an enrichment of spermatozoa with lower DNA fragmentation and higher oocyte penetration capacity. Using this method, Zahedi et al. (2013) also reported an enrichment of spermatozoa with lower DNA fragmentation in both fertile and infertile (terato- and asthenozoospermic) patients. Additionally, Lee et al. (2010) and Degheidy et al. (2015) proved that spermatozoa selected by AV-MACS from idiopathic infertile patients and patients diagnosed with varicocele showed lower DNA fragmentation than the original sample and kept intact sperm motility. However, a number of studies comparing AV-MACS with SU and DGC question the actual utility of this technique for the enrichment in spermatozoa with higher quality. Therefore, Tavalaee et al. (2012) found that combined AV-MACS followed by DGC on semen from infertile patients with different etiologies was more efficient at enriching the sample in spermatozoa with lower DNA fragmentation than both procedures autonomously or DGC followed by AV-MACS. Nadalini et al. (2014) obtained better results with DGC followed by SU than with DGC followed by AV-MACS in terms of sperm quality as determined by motility, morphology, and DNA fragmentation. Cakar et al. (2016) did not find significant differences in any of the analysis of sperm quality, including DNA fragmentation, by using AV-MACS, SU, DG, SU/AV-MACS, and DG/AV-MACS, perhaps due to the low sample size of the study. In contrast, Berteli et al. (2017) found a better quality in terms of motility and DNA fragmentation in spermatozoa selected by AV-MACS followed by DCG than with the other combinations tested: DCG and AV-MACS alone or DCG followed by AV-MACS. Similar to these results, Zhang et al. (2018) reported that the combination of DCG followed by AV-MACS in immotile sperm samples resulted in the recovery of a sperm population with lower DNA fragmentation compared to DGC alone. In agreement, Esbert et al. (2017) reported that DGC followed by AV-MACS reduced the number of spermatozoa porting chromosomal abnormalities.

Regarding the use of the spermatozoa selected by AV-MACS for ARTs, reported results are scarce and unclear (**Table 2**). An early study employing 196 oligoasthenozoospermic patients reported an increase of cleavage and pregnancy rates when AV-MACS was applied prior to ICSI in comparison to DGC (Dirican et al., 2008). However, another study registering 237 infertile couples following ICSI with donated oocytes did not find significant differences in terms of embryo quality and in the ratios of fertilization, implantation, pregnancy, and live birth when the semen was cryopreserved and then selected by SU or SU followed by AV-MACS (Romany et al., 2014). García-Ferreyra et al. (2014) reported higher, but not statically significant pregnancy and implantation rates when semen showing basal high DNA fragmentation was selected by AV-MACS followed by DGC, compared to DGC alone in ICSI treatments. These results are supported by those reported by Stimpfel et al. (2018) in couples of teratozoospermic patients and women over 30 years old undergoing ICSI, showing higher quality blastocysts when

#### TABLE 1 | DNA fragmentation of spermatozoa selected by AV-MACS.


DNA fragmentation was analyzed in all cases by TUNEL. Values of % DNA fragmentation are expressed as % ± SEM, SD or (interquartile range). <sup>∗</sup>P < 0.05, ∗∗∗P < 0.001 and different letters indicate differences between groups (P < 0.05).

spermatozoa were selected by DGC/SU followed by AV-MACS, compared to DGC/SU. Recently, Ziarati et al. (2019) reported a significant increase in the percentage of high-quality embryos and clinical pregnancies employing DGC followed by MACS in 80 infertile couples showing male factor and undergoing ICSI when compared to DGC itself.

Studies conducted in livestock species using MACS are missing. In rabbits, AV-MACS did not show a clear enrichment in non-apoptotic spermatozoa and did not affect the reproductive outcome when the selected semen was used for AI (Vasicek et al., 2014).

These studies collectively show that the combination of DGC and MACS might provide slight benefits in patients with male factor undergoing ICSI in terms of clinical pregnancy. However, it is not clear to which extent live birth rate is actually improved. Thus, the type of donor or patient and the combination of AV-MACS with other sperm preparation methods could be determinant for improving ARTs. More experiments, registering larger number of patients, and exploring the different variables involved, are needed in order to certificate the utility of AV-MACS for the human clinical practice.

#### Hyaluronic Acid (HA) Binding

The hyaluronic acid (HA) is one of the main components of the extracellular matrix surrounding the cumulus-oocyte complex (COC) (Dandekar et al., 1992), and those spermatozoa that follow an adequate spermatogenesis and maturation exhibit binding sites to it (Cayli et al., 2003; Huszar et al., 2006). Therefore, two methodologies of sperm selection have been developed based on the spermatozoa-HA interaction: (i) recovering those spermatozoa trapped on the surface of HA coated dishes (Huszar et al., 2006) (**Figure 2B**) and (ii) picking up those spermatozoa moving slow when swimming in a medium containing HA in solution (Barak et al., 2001).

According to published data, both methods have shown their capacity to select spermatozoa with lower DNA fragmentation (**Table 3**). In an early work, Nasr-Esfahani et al. (2008), employing HA-coated dishes, reported a significant inverse correlation among the percentage of HA-bounded spermatozoa and protamine deficiency, DNA fragmentation, and abnormal sperm morphology in the original sample. While Razavi et al. (2010) reported that spermatozoa recovered from HA-coated dishes show the same level of DNA fragmentation than the unselected ones, other laboratories reported higher DNA integrity in spermatozoa selected by HA-binding methods (Parmegiani et al., 2010a; Yagci et al., 2010; Mongkolchaipak and Vutyavanich, 2013) Furthermore, Parmegiani et al. (2010a) and Huang et al. (2015) showed lower DNA fragmentation in spermatozoa selected in HA solution compared to SU, and by HA coated dishes compared to DGC, respectively. However, both studies reported similar levels of DNA fragmentation when compared to spermatozoa selected under the microscope regarding motility and morphological features. Similarly, Mongkolchaipak and Vutyavanich (2013) did not find any difference between DGC/HA coated dishes and DGC/intracytoplasmic morphologically selected sperm injection


TABLE 2 | Reproductiveoutcomes of ARTs employing spermatozoa selected by AV-MACS.

All the reported cases followedICSI. Values are expressed as % ± w/o, SEM, SD, or (interquartile range). ∗P<0.05, ∗∗P<0.01.

(IMSI) respect to sperm DNA fragmentation. These results question the utility of sperm selection based on HA-binding.

Although the use of HA was first conceived for substituting the routinely used PVP as an agent to decrease sperm motion prior ICSI (Sbracia et al., 1997; Barak et al., 2001), the potential as a selective agent was soon explored (Cayli et al., 2003). Thus, Park et al. (2005) used HA solution to select boar spermatozoa and found a significant increase of embryos with normal chromosomal counts and less chromosome abnormalities when compared to conventional ICSI. Few months later, similar results were reported in humans showing that sperm selection by HA led to a reduction of chromosomal disomy frequencies, diploidy, and sex chromosome disomy; all abnormalities associated to embryos produced by ICSI (Jakab et al., 2005). All these publications did not report clear effects on ICSI outcomes; however, this was possibly not carefully studied until 2008. Since then, a number of publications have specifically checked the capability of sperm selection by HA-binding for improving ICSI outcome (**Table 4**). In a first attempt, Nasr-Esfahani et al. (2008) found an increase of the fertilization rate using HA-coated dishes but there was no detectable effect on pregnancy or implantation rates. Parmegiani et al. (2010b) studied the reproductive outcome in 331 patients undergoing ICSI using HA solution (293 patients) or conventional ICSI (86), finding no differences in clinical pregnancies or live birth ratios but an improvement in embryo quality and implantation rate. In another study, the same authors reported comparable results employing 206 oligozoospermic patients (Parmegiani et al., 2010a). Similar results were reported in subsequent studies using either HA-coated dishes or HA solution (Choe et al., 2012; Parmegiani et al., 2012; Majumdar and Majumdar, 2013; Mokánszki et al., 2014). In contrast, Worrilow et al. (2013) reported that the use of HA-coated dishes significantly increased pregnancy (Worrilow et al., 2013; Mokánszki et al., 2014) and live birth rate (Mokánszki et al., 2014) for those patients undergoing ICSI which semen showed a HA-binding capacity ≤ 65%, as measured by the Hyaluronan binding assay. This effect proves that the method could be useful only in specific cases and indicates the need of a preliminary semen analysis for determining the suitability of the technique. Therefore, Erberelli et al. (2017) reported a significant improvement of ICSI using HA-coated dishes in patients with various categories of male factor. Their results also suggested that the benefit of this method could be higher for teratozoospermic patients. In a randomized study employing a large number of patients (2772 couples), Miller et al. (2019) did not find any significant benefit of HA-coated dishes compared to conventional ICSI and either detected an association between results of Hyaluronan binding assay and ICSI outcomes, contradicting previous works above described. These results collectively show a poor capacity of HA binding methods for improving ICSI; however, more studies are needed to identify the types of patients that could benefit from this procedure.

#### Zeta Method

Sperm membrane is negatively charged (Engelmann et al., 1988). On this basis, methods for separating X- and Y-bearing


Analysis of DNA fragmentation was conducted by Sperm Chromatin Dispersion (SCD) test, Acridin Orange fluorescence (AOF), or TUNEL. Values of % of DNA fragmentation are expressed as % ± SEM. ∗∗P < 0.01 and different letters indicate differences between groups (P < 0.05).

#### TABLE 4 | Reproductiveoutcomes of ARTs spermatozoa selected by hyaluronic acid.


All the reported cases followedICSI. Values are expressed as %±w/o, SEM, SD, or (interquartile range). ∗P<0.05, ∗∗P<0.01, ∗∗∗P<0.001.

Novel Sperm Selection for ARTs

fcell-07-00298 November 29, 2019 Time: 17:0 # 11

spermatozoa (Engelmann et al., 1988) and for the separation of pure sperm heads from disintegrated mammalian spermatozoa (Chaudhuri and Datta, 1994) were developed decades ago. Later, the methods developed by Chan et al. (2006) and Ainsworth et al. (2005) (**Figure 2C**) allowed to collect the charged spermatozoa adhered to the wall of a centrifuge tube or migrating within an electric field, respectively. With both methodologies, downstream analysis of the selected spermatozoa revealed an increased percentage of spermatozoa with higher quality and porting highintegrity DNA (Kheirollahi-Kouhestani et al., 2009; Razavi et al., 2010; Zarei-Kheirabadi et al., 2012; Zahedi et al., 2013) (**Table 5**). This procedure, known as Zeta method, has allowed the selection of spermatozoa with lower DNA fragmentation compared to the HA-coated dish selection (Razavi et al., 2010). Also, sperm DNA fragmentation has been proven to be lower using DGC-Zeta than DGC by itself (Zarei-Kheirabadi et al., 2012) or MACS-DGZ (Zahedi et al., 2013). However, despite these promising results, only one randomized study has been published using spermatozoa selected by Zeta method in patients undergoing ICSI (Esfahani et al., 2016). In this research, significant increases in top quality embryos [45.83 ± 3.11% vs 35.38 ± 4.64% (P = 0.04)] and pregnancy rates [39.2 vs 21.8% (P = 0.009)]

were obtained with DGC/Zeta compared to DGC, respectively. However, further analysis is needed to know the potential of Zeta method for improving ARTs.

#### Future of Sperm Selection Based on Membrane Characteristics

As exposed above, sperm selection by AV-MACS and HA does not seem to contribute in general terms to a significant improvement of ARTs outcome, and Zeta method needs to proof its efficiency in larger studies. However, the use of the properties of the plasma membrane for sperm selection could really improve ARTs if markers with higher potential for discriminating highquality spermatozoa are discovered. Thus, new antigens could be used as targets for selection by MACS, by coated plates or even by fluorescence-activated cell sorting (Funaro et al., 2013). However, this research direction is currently poorly explored, possibly because of the lack of basic knowledge about markers discriminating the various sperm populations contained in the ejaculate. Heidari et al. (2018) explored three surface proteins, HSPA2, Dj-1, and serum amyloid P, as biomarkers of sperm DNA integrity and found that the abundance of these three


Analysis of DNA fragmentation was conducted by Sperm Chromatin Dispersion (SCD) test, Acridin Orange fluorescence (AOF), or TUNEL. Values of % of DNA fragmentation are expressed as % ± SEM or SD. <sup>∗</sup>P < 0.05 and different letters indicate differences between groups (P < 0.05).

proteins in the semen was directly correlated to sperm quality, DNA integrity, and embryo quality. Mostek et al. (2018) found that high-quality bull ejaculates showed higher abundance of extracellular sperm surface proteins and that sperm proteins in low-quality ejaculates are characterized by high carbonylation levels. All these membrane features and new ones yet to be discovered could be used with the aim of improving ARTs through sperm selection.

#### SPERM SELECTION BASED ON SPERM MORPHOLOGY – INTRACYTOPLASMIC MORPHOLOGICALLY SELECTED SPERM INJECTION (IMSI)

The morphometric evaluation of spermatozoa is widely applied to analyze sperm quality due to its correlation to fertility (Yániz et al., 2015; Soler et al., 2016). Furthermore, the introduction of computer-enhanced digital microscopy has enabled the analysis and quantification of detailed features of the cell that applied to motile spermatozoa configure the "Motile Sperm Organelle Morphology Examination" (MSOME) (Bartoov et al., 2002). The use of MSOME for selecting spermatozoa for ICSI is known as IMSI. This technique consists in the selection and direct capture of those spermatozoa with a low number of vacuoles and a nucleus of normal morphology under a microscope equipped with a micromanipulation system and a magnification system of 6300x (Bartoov et al., 2002) (**Figure 3**). The power of selecting spermatozoa with higher DNA integrity has been proved by different authors, although contradictory results are also present in the literature (**Table 6**). Using this methodology, Franco et al. (2008) revealed that the presence of large vacuoles in the sperm nucleus is associated with a higher DNA fragmentation compared to those spermatozoa with normal nucleus. The correlation between the presence of vacuoles, as detected by MSOME, and DNA fragmentation was later confirmed by Wilding et al. (2011). However, Leandri et al. (2013) did not find significant differences in this respect between the spermatozoa selected by IMSI and those selected as conventionally done during regular ICSI. Other authors have suggested that IMSI could be useful only in specific cases of male infertility. Accordingly, when spermatozoa from infertile donors showing more than 13% of DNA fragmentation were selected by IMSI, those showing normal morphology under high magnification delivered less DNA fragmentation than those classified as motile and normal using conventional magnification of 200x (Hammoud et al., 2013). Moreover, the authors reported in the same work that those selected under lower magnification showed the same DNA fragmentation than the unsorted spermatozoa. Results reported by Lavolpe et al. (2015) suggest that in patients whose semen scored ≤ 4% according to the strict morphology index, the presence of vacuoles in the nucleus of the spermatozoa was less related to DNA fragmentation and chromatin compaction than in those patients with strict morphology index ≥ 14%.

Recently, in the largest study reported to date where 873 sperm samples were analyzed by MSOSE, the presence of head vacuoles

was not associated to sperm DNA fragmentation and live birth rate (Fortunato et al., 2016). Authors claim that vacuoles are physiological features that do not alter sperm functionality. The controversy about the convenience of IMSI for improving ARTs exists from the first moment the method was described. We address here the reader to an excellent systematic review on the literature published between 2001 and 2013, where the authors conclude that IMSI was only proven to improve reproductive outcome in cases of recurrent implantation failure following ICSI (Boitrelle et al., 2014). The same year, other authors published a meta-analysis on 13 publications confronting IMSI vs conventional ICSI in cases with previous ICSI failures and when the male factor was the cause of infertility, and concluded that in both cases, IMSI improved the reproductive outcome (Setti et al., 2014). However, authors pointed to the lack of randomized studies to corroborate their conclusions. Since then, together with the study of Fortunato et al. (2016), others have shown the failure of IMSI for improving ARTs (**Table 7**). Thus, Leandri et al. (2013), having examined 458 couples, did not find significant differences between IMSI or ICSI in the reproductive outcome, regardless of the initial characteristics of semen regarding DNA fragmentation, chromatin compaction, morphology, or motility. Setti et al. (2015) showed no benefits of IMSI for couples with poor ovarian response. In contrast, Shalom-Paz et al. (2015) reported that IMSI significantly increased implantation and clinical pregnancy rates in patients with repeated IVF-ICSI failure. However, Gatimel et al. (2016) failed in confirming these results. Kim et al. (2014) and Goswami et al. (2018) showed that IMSI applied to OAT patients of various severities resulted in significantly higher implantation, pregnancy rates, and live birth rates when compared to a previous ICSI cycle performed to the same patients, suggesting an actual utility of IMSI for these specific cases. Also, in another study on 170 patients,


TABLE 6 | DNA fragmentation of spermatozoa selected by IMSI.

Analyses of DNA fragmentation were conducted by, Acridin Orange fluorescence (AOF) or TUNEL. Values of % of DNA fragmentation are expressed as % ± SEM or SD. Values separated by double slash indicate correspondence to the employed method of DNA fragmentation in that column. In Lavolpe et al. (2015) % of DNA fragmentation are intervals according to different patterns of vacuole number, morphology, and distribution. <sup>∗</sup>P < 0.05, ∗∗∗P < 0.001 and different letters indicate differences between groups (P < 0.05).

the results suggest that IMSI can improve pregnancy rates in patients with severe sperm pathologies affecting various sperm parameters (Schachter- Safrai et al., 2019). Overall, although the literature dissuades the routine use of IMSI, it seems that it could be highly indicated for the treatment of severe cases of male factor. Larger head-to-head randomized studies discriminating seminal parameters in infertile pathologies are still needed in order to elucidate under which circumstances IMSI could benefit reproductive outcome in the human clinical practice.

#### SPERM SELECTION BASED ON GUIDANCE MECHANISMS

As we discussed in the section "Introduction," guidance within the female genital tract has been suggested as a physiological mechanism for the selection of those spermatozoa able to fertilize the egg and ensuring the ulterior embryo development to term (Pérez-Cerezales et al., 2017). Hence, a strategy to develop new sperm selection methodologies is to employ in vitro the same principles that govern the sperm selection operating within the female genital tract. Consequently, there are three known mechanisms that have been proposed to guide the spermatozoa within the oviduct and they have been tested accordingly for their capacity to select spermatozoa with the aim of improving ARTs.

#### Rheotaxis

Miki and Clapham (2013) proved for the first time that both human and mice spermatozoa orientate their swimming against a fluid flow, process known as rheotaxis. Latter, El-Sherry et al. (2014) and Romero-Aguirregomezcorta et al. (2018) confirmed the occurrence of this phenomenon also in bovine, stallion, and ram spermatozoa, suggesting rheotaxis as a conserved feature of mammalian spermatozoa. Furthermore, the existence in mice of oviductal flow toward the uterus intensified after copula supports rheotaxis as a long-distance guidance mechanism within the oviduct (Miki and Clapham, 2013).

De Martin et al. (2017) conducted a first study employing rheotaxis to select spermatozoa from normozoospermic donors and reported an enrichment in spermatozoa with higher chromatin compaction (99%) compared to the sample before selection (71%) and to spermatozoa selected by DCG (83%). Zaferani et al. (2018) reported the design of a microfluidic device for selecting motile spermatozoa on the basis of rheotaxis; however, these authors did not

Novel Sperm Selection for ARTs

fcell-07-00298 November 29, 2019 Time: 17:0 # 15

#### TABLE 7 | Reproductiveoutcomes of ARTs spermatozoa selected by IMSI.


Values are expressed as%±w/o, SEM, or SD.∗P<0.05,∗∗P<0.01,∗∗∗P<0.001.

analyze the quality of the selected spermatozoa. Nagata et al. (2018) used another microfluidic device to select frozen bull spermatozoa by rheotaxis (**Figure 4A**). These authors reported a reduction of the level of DNA fragmentation (0.37%) in the spermatozoa selected by rheotaxis when compared to unselected semen (7%). The use of these selected bull spermatozoa for artificial insemination delivered pregnancy results similar to semen without selection, but using a 20 times lower dose (1 million sperm/insemination against 20 million sperm/insemination, respectively).

Thus, the results reported to date are promising, but the passive nature of rheotaxis, which is based on the hydrodynamics of the motile spermatozoa (Zhang et al., 2016), indicates that this mechanism has poor selective potential beyond the separation of those spermatozoa with a correct swimming behavior. As a matter of fact, half of the spermatozoa orient their swimming by rheotaxis, independently of being or not capacitated when directly observed under the microscope (Miki and Clapham, 2013). Although this is the in vitro situation, the in vivo picture could be drastically different. Moreover, it has been proposed that the planar move of non-capacitated spermatozoa would make them easier to stick to the epithelium preventing their migration by rheotaxis (Miki and Clapham, 2013) under the physiological conditions of viscosity. In contrast, the rotation along their longitudinal axis that determines the swimming of capacitated spermatozoa under a viscous medium could allow their detachment from this epithelium, allowing their free swimming against the fluid current (Miki and Clapham, 2013). In this way, rheotaxis would be part of a selection system of capacitated spermatozoa allowing only this subpopulation to migrate toward the in vivo fertilization site. In vitro developed procedures could test this hypothesis for sperm selection by rheotaxis considering these physiological conditions and then configure an effective way of selecting capacitated spermatozoa for improving ARTs.

#### Chemotaxis

Chemotaxis is the mechanism of navigation that the spermatozoa use in the proximity of the oocyte to orient their swimming in a gradient of chemical substances released by the COC where progesterone (P4) seems to be the main chemoattractant (Oren-Benaroya et al., 2008). It should be noted that only capacitated spermatozoa respond by chemotaxis, so this property could allow the selection of this specific subpopulation (Oren-Benaroya et al., 2008). Thus, Gatica et al. (2013) selected spermatozoa from both normozoospermic donors and subfertile patients by employing a simple device named "Sperm Selection Assay" (SSA) in which a P4 gradient was established (**Figure 4B**). In this work, it was found that in both cases, the selected spermatozoa were three times more capacitated, showed less DNA fragmentation and less oxidative stress than the semen before the selection. Li et al. (2018) also reported an improvement in the sperm quality from normozoospermic donors in terms of normal morphology, lower DNA fragmentation, and lower percentage of apoptotic spermatozoa when using another device composed of a system of microfluidic channels. Dominguez et al. (2018) proved that the use of the SSA device allows the selection of higher quality bull spermatozoa, improving cleavage rates using both sexed and unsexed semen for IVF. However, the effectiveness of sperm selection by chemotaxis in the improvement of ARTs in clinical practice has not been studied to date.

## Thermotaxis

Sperm thermotaxis consists in the orientation of the spermatozoa movement toward the highest temperature in a gradient (Bahat et al., 2003). Evidence indicates that this mechanism allows sperm to orient itself in the fallopian tubes to ascend to the ampulla (Pérez-Cerezales et al., 2017). As in chemotaxis, only capacitated spermatozoa respond to thermotactic stimuli, so based on it, this subpopulation could be selected (Bahat et al.,

2003). In a recent publication, employing a simple method (**Figure 4C**), we showed in mice and in normozoospermic patients that spermatozoa selected by SU/thermotaxis possess higher DNA integrity compared to the seminal sample prior to selection and after selection by SU (Pérez-Cerezales et al., 2018). In addition, the use of these selected spermatozoa for ICSI in mice increased cleavage rates and blastocyst production, as well as implantation and live birth rate as opposed to the use of spermatozoa selected by SU. Although this is the only work related to the use of spermatozoa selected by thermotaxis in ARTs, a recent study carried out in bulls has shown that seminal samples delivering high pregnancy rates after artificial insemination show a greater response to thermotaxis (Mondal et al., 2016). However, as in the case of rheotaxis and chemotaxis, there are still no available publications showing the capability of sperm selection by thermotaxis to improve the ARTs efficiency in the human clinic.

#### SPERM SELECTION BASED ON SPERM MOTILITY (MICROFLUIDICS)

As we have seen in some of the previous methodologies, microfluidic systems are becoming transversal devices for sperm selection based on several fundaments. Therefore, microfluidic systems have been tested for sperm selection by rheotaxis (Zaferani et al., 2018), chemotaxis (Ko et al., 2018; Nagata et al., 2018), and thermotaxis (Ko et al., 2018), as well as by other physical properties that operate at small scale, allowing the selection of motile spermatozoa (**Table 8**). Nosrati et al. (2014) reported the improvement of human spermatozoa selected by a microfluidic device composed of 500 parallel microchannels in terms of reducing more than 80% DNA fragmentation. Similar results were reported by Kishi et al. (2015), showing a significant reduction of DNA fragmentation in spermatozoa selected by a commercially available device based on microfluidics and flow dynamics (Sperm Sorter Qualis <sup>R</sup> , Menicon, Kasugai, Japan) when compared to SU or unselected spermatozoa. Using the same device, Shirota et al. (2016) also reported the selection of spermatozoa with lower DNA fragmentation compared to spermatozoa selected by DGC followed by SU in healthy donors. In addition, Nagata et al. (2018) reported a robust selection of human spermatozoa porting low DNA fragmentation by utilizing a diffuser-type microfluidic sperm sorter (DMSS). Quinn et al. (2018) employed another commercially available device, a singleuse chip with an inlet sample chamber connected to an outlet collection chamber by a microfluidic channel (FERTILE, Zymot, DxNow Inc., Gaithersburg, MD, United States), and showed a very effective selection of spermatozoa with low fragmented DNA compared to unselected spermatozoa or those selected by DGC-SU.

Due to the novelty of these methodologies, there is only one study that employs the selected spermatozoa in ARTs. In a recent publication in which spermatozoa were selected from 122 patients with infertility of unknown etiology using a commercial microfluidic device (61 patients) (Fertile Chip <sup>R</sup> , KOEK Biotechnology, Turkey) or by SU (61 patients), no differences were found in terms of fertilization, pregnancy, and live birth rates during ICSI between both sperm selection methods (Yetkinel et al., 2019). However, these authors observed


Analyses of DNA fragmentation were conducted by sperm chromatin structure assay (SCSA), sperm chromatin dispersion (SCD) test, or TUNEL. Values of % of DNA fragmentation are expressed as % ± SEM or SD. In Nosrati et al. (2014), DNA fragmentation separated by slash corresponds to the microchannel lengths 6/7.5/9 mm, respectively. Negative linear correlation between microchannel lengths was reported (R<sup>2</sup> = 0.99). <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Different letters indicate differences between groups (P < 0.005).

a higher quality in those embryos produced with spermatozoa selected by Fertile Chip <sup>R</sup> . Further studies are then necessary to determine under which conditions these devices are suitable for improving ARTs outcomes.

#### CONCLUSION

The efficiency of ARTs has a margin for improvement. Sperm selection may be an important factor for achieving higher live birth rates in ARTs, especially in infertility cases where the male factor is present. However, the methodologies developed to date have not proven to be useful for their routine application in the clinical practice and seem to be effective only in specific cases of male infertility. Some novel methods described here based on the physiological selection operating in vivo and on microfluidic environments have delivered promising results yet to be confirmed in large studies in the context of clinical practice. These studies should be randomized and strict in the confrontation of results with sperm samples of different qualities minimizing the female factor. It may even be necessary to combine several sperm selection methodologies to increase the efficiency of the ARTs.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

IO-L and SR-D conducted literature review and wrote the manuscript. PR-I wrote the manuscript. SP-C defined the topic, conducted literature review, and structured and wrote the manuscript.

#### FUNDING

This work was supported by Grant RTI2018-096736-A-I00 from the Spanish Ministry of Science, Innovation, and Universities. IO-L and SR-D were supported by CONACYT fellowship of the Mexican Government (283833) and "doctorados industriales 2018," fellowship of Comunidad of Madrid (IND2018/BIO-9610). PR-I was supported by a Talent Attraction Fellowship from Comunidad de Madrid. SP-C was supported by a Ramón y Cajal contract from the Spanish Ministry of Science, Innovation, and Universities (RYC-2016-20147).

#### ACKNOWLEDGMENTS

We thank Anna Riabzev for making the drawing.

over a wide range. PLoS One 7:e41915. doi: 10.1371/journal.pone.00 41915



functional human sperm even in subfertile samples. Mol. Hum. Reprod. 19, 559–569. doi: 10.1093/molehr/gat037


sperm injection. Curr. Opin. Obstet. Gynecol. 18, 260–267. doi: 10.1097/01.gco. 0000193018.98061.2f





**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Oseguera-López, Ruiz-Díaz, Ramos-Ibeas and Pérez-Cerezales. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Successful ICSI in Mice Using Caput Epididymal Spermatozoa

Raúl Fernández-González\*, Ricardo Laguna, Priscila Ramos-Ibeas, Eva Pericuesta, Víctor Alcalde-Lopez, Serafín Perez-Cerezales and Alfonso Gutierrez-Adan

Departamento de Reproducción Animal, INIA, Madrid, Spain

Spermatozoa undergo their last phase of spermiogenesis, known as maturation, as they pass through the epididymis. A recent report indicates that mouse immature spermatozoa retrieved from the caput epididymis for intracytoplasmic sperm injection (ICSI) give rise to embryos with multiple developmental defects. Further, these embryos were unable to develop to term after their transfer to surrogate mothers. Herein, we examined the potential of mouse caput spermatozoa to produce normal embryos by comparing the use of caput vs. cauda epididymal spermatozoa for in vitro fertilization (IVF) or ICSI. Two methods for the separation of sperm heads prior to ICSI were also compared: freezing/thawing or drawing through a syringe. We found that in contrast to caudal spermatozoa, caput spermatozoa failed to produce embryos via IVF, confirming their immature state. However, regardless of the method employed for the separation of sperm heads, similar efficiencies of blastocyst production in vitro and development to term after transfer to surrogate mothers were observed following ICSI using both caput and cauda epididymal spermatozoa. It therefore seems that mice spermatozoa recovered from the caput epididymis are as valid as caudal spermatozoa for the production of embryos and offspring by ICSI.

Keywords: ICSI, caput, epididymis, mice, sperm

## INTRODUCTION

Spermatozoa acquire their final mature state during passage through the epididymis. Mammalian immature sperm produced in the testis move from the ducts of the testis into the caput epididymis, where they are completely immobile and unable to fertilize an egg cell; then they are transported through the epididymis to the cauda region as a result of rhythmic contractions of this organ, where they undergo complex modifications, acquire motility and become fertilization-competent, denominating then mature sperm (Gervasi and Visconti, 2017). Thus, in theory, ejaculated sperm should have a better fertilizing capacity than sperm recovered from the testis as they have transited through the epididymis, where the final steps of maturation take place. These steps include compaction of sperm chromatin, formation of disulfide bridges between protamines and epigenetic remodeling (Ariel et al., 1994; Haidl et al., 1994; Perez-Cerezales et al., 2012). Chromatin maturation is required to fully stabilize and protect DNA from damage induced by endogenous nucleases or reactive oxygen species during sperm passage through the epididymis (Irvine et al., 2000; Perez-Cerezales et al., 2012). The importance of sperm maturation in the epididymis has also been confirmed by in vitro fertilization (IVF) assays performed with spermatozoa collected from different sites of the epididymis. In these assays, sperm retrieved from the distal part acquire a higher motility

#### Edited by:

Xin Zhiguo Li, University of Rochester, United States

#### Reviewed by:

Giuliano Callaini, University of Siena, Italy Duancheng Wen, Weill Cornell Medicine, Cornell University, United States

#### \*Correspondence:

Raúl Fernández-González raulfg@inia.es; agutierr@inia.es

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 29 July 2019 Accepted: 03 December 2019 Published: 17 December 2019

#### Citation:

Fernández-González R, Laguna R, Ramos-Ibeas P, Pericuesta E, Alcalde-Lopez V, Perez-Cerezales S and Gutierrez-Adan A (2019) Successful ICSI in Mice Using Caput Epididymal Spermatozoa. Front. Cell Dev. Biol. 7:346. doi: 10.3389/fcell.2019.00346

and ability to fertilize the oocyte (Sullivan and Mieusset, 2016). The epididymis is also able to eliminate defective or immature germ cells (Ramos-Ibeas et al., 2013).

Recently, it has been reported that small RNAs are trafficked to mammalian sperm during the process of post-testicular maturation in the epididymis (Sharma et al., 2018). It has been also described that 2-cell embryos generated via intracytoplasmic sperm injection (ICSI) using sperm obtained from the proximal (caput) versus distal (cauda) epididymis show similar RNA expression, but as development progresses, their gene expression profiles diverge (Conine et al., 2018). Further, embryos arising from caput spermatozoa (caput embryos) have been shown to have multiple developmental defects, and when transferred to surrogate mothers were unable to reach term (Conine et al., 2018). However, the injection of cauda- specific microRNAs (miRNAs) into caput zygotes was observed to suppress the specific gene expression profile of caput embryos and rescue their potential for development. These results indicate that dynamic small RNAs are essential for sperm maturation and also necessary for successful development and were taken as evidence that sperm-borne small RNAs derived from near the cauda region are critical for fetal development (Conine et al., 2018; Sharma et al., 2018). However, both testis spermatids and sperm support full-term development when injected into oocytes (Kimura and Yanagimachi, 1995; Moreira et al., 2016). In this context, the findings of Conine et al. (2018) suggest that the ability of sperm to support full-term embryo development after ICSI is acquired in the testis, then lost in the caput epididymis, and re-acquired during epididymal transit prior to arriving at the cauda. However, this hypothesis contradicts the results reported by others showing that ICSI produces normal offspring using caput sperm (Suganuma et al., 2005; Zhou et al., 2019).

Our hypothesis is that the developmental potential of caput and cauda fertilized embryo is similar and that the discrepancies of the results reported by Conine et al. (2018) respect to the literature could be due to differences in ICSI protocols, mainly in the technique used to separate the head from the tail of the sperm, since the different techniques can produce different degrees of alterations in the sperm (Tateno, 2008). Mouse spermatozoa are particularly sensitive to mechanical damage (Mazur et al., 2000), and these damages can be more decisive for the sperm of the caput, since it has been shown that there are more DNA fragmentation and less chromatin compaction in the caput than in the cauda epididymal spermatozoa (Perez-Cerezales et al., 2012), and that sperm obtained from the caput are more susceptible to nuclease-dependent DNA damage than sperm obtained from other epididymal regions (Yamauchi et al., 2007). Moreover, it has been reported that ICSI in mice using DNAfragmented sperm has been linked to altered gene expression at the preimplantation stage, reduction of fetal development, an increased risk of genetic and epigenetic abnormalities both in embryos and offspring (Ramos-Ibeas et al., 2014). Separation of the sperm heads from the tail prior to ICSI is commonly conducted by sonication, by fine needle shearing, by unprotected freezing, by piezo pulses, or by detergent treatment, and all these protocols may produce different level of chromosome aberrations in mouse (Martin et al., 1988; Tateno, 2008).

Our study addresses the reproductive potential of caput spermatozoa used for IVF or ICSI. As a positive control, we used caudal spermatozoa, regularly used for these procedures in mice. Our objective was to test the reported inability of caput spermatozoa to produce viable ICSI-derived mice by injecting into oocytes caput or cauda epididymal sperm recovered from standard B6D2F1 hybrid and inbred B6 mouse strains using similar protocols to previously described (Conine et al., 2018).

### MATERIALS AND METHODS

#### Handling of Mice and Sperm Samples

Experiments in mice were carried out in strict accordance with recommendations of the guidelines of the European Community Council Directive 86/609/EEC. The study protocol was approved by the Committee on the Ethics of Animal Experiments of the INIA (2016 permit number CEEA 2014/025). Spermatozoa were collected from the caput and caudal epididymis of 20-weekold B6D2F1 males (C57/BL/6JxDBA/2J) and B6 (C57BL/6NHsd) mice (**Figure 1**). Sperm were released by making an incision in the epididymal tissue and then squeezing to release its contents. The sperm were then suspended in M2 medium (Sigma–Aldrich, Madrid, Spain) for the ICSI experiments, or in human tubal fluid medium (HTF) supplemented with 1% BSA for IVF tests. To capacitate the sperm for the IVF experiments, samples were incubated for 30 min under an atmosphere of 5% CO<sup>2</sup> at 37◦C.

#### Intracytoplasmic Sperm Injection (ICSI) and in vitro Fertilization (IVF)

For our ICSI experiments, the caput and caudal epididymis of the same animal were dissected separately, and sperm collected in M2 medium (6 adult individuals) as described previously (Perez-Cerezales et al., 2012). The sperm was mixed with five volumes of a 10% solution of polyvinyl-pyrrolidone in M2. ICSI was performed in M2 medium at room temperature (Ramos-Ibeas et al., 2014). Sperm heads were decapitated by the freezingthawing method (Moreira et al., 2003) or by drawing the sperm repeatedly through a fine-gauge needle (25G) into a 1 mL syringe 20–30 times (Conine et al., 2018). In parallel, oocytes were collected at metaphase II from B6D2F1 8-week-old females, which were superovulated by standard intraperitoneal injection (Moreira et al., 2005), and incubated with hyaluronidase (300 µg/mL) to remove cumulus cells. After allowing their recovery for 15 min, sperm were injected into the oocytes in M2 at room temperature. Surviving zygotes were cultured until the blastocyst stage to check their development or until the 2-cell stage for the embryo transfer experiments.

For the IVF experiments, oocytes were obtained from superovulated female mice as described above. The method used has been described elsewhere (Hourcade et al., 2010). Briefly, 2.5– 10.0 µL of fresh caput or cauda epididymal sperm were added to each fertilization drop to achieve a final concentration of ∼1– 2 × 10<sup>6</sup> spermatozoa/mL. Four hours after oocyte and sperm co-incubation at 37◦C in a humidified atmosphere of 5% CO<sup>2</sup> in air, putative zygotes were washed and cultured in KSOM.

Statistical differences between caput- and cauda- derived embryos and offspring were assessed using an unpaired t-test and ANOVA (SigmaStat package), respectively. Significance was set at p < 0.05.

#### RESULTS

Sperm from the cauda and caput epididymis were collected from six 20-week-old sexually mature B6D2F1 and four inbred B6 male mice (Perez-Cerezales et al., 2012) and prepared according to requirements for IVF or ICSI (**Figure 1**). For IVF, the sperm were capacitated for 30 min and for ICSI the sperm were frozen without a cryoprotectant (Moreira et al., 2003) or drawn repeatedly through a fine-gauge needle into a 1 mL syringe 20–30 times (Conine et al., 2018) to obtain tail-free sperm heads. Our IVF experiments revealed that only caudal sperm had the capacity to fertilize oocytes. Thus, using caudal sperm, 96.29 ± 3.70% of fertilized oocytes (n = 48) underwent 2-cell cleavage and all of these embryos developed to blastocysts. In contrast, no fertilization was produced with sperm obtained from the caput epididymis: only 4 out of 128 (3.25 ± 3.25%) fertilized oocytes developed to the 2-cell stage, and none developed into blastocysts. This result could be explained by a lack of motility of caput sperm.

We next microinjected caput and caudal sperm (heads decapitated by freezing-thawing) of B6D2 mice into oocytes and found that embryos produced with both sperm types developed in vitro to the blastocyst stage at similar efficiencies (**Figure 2**). Given that caput-derived embryos were perfectly capable of developing to the blastocyst stage in vitro (57% of blastocysts produced by caput sperm vs. 63% by caudal sperm, four replicates and more than 100 embryos produced in each group), we then generated both caput- and caudaderived embryos by ICSI using paired sperm samples from the same animal, and surgically transferred 15–20 2-cell embryos into pseudopregnant surrogate mothers to analyze their in vivo developmental capacity. First, we used our classic method of fast freezing/thawing to separate sperm heads. As we show in **Table 1**, we found similar percentages of cleaved embryos and live offspring using caput or caudal sperm. We then used the drawing through a syringe method to decapitate sperm (Conine et al., 2018), and using both B6D2 and B6 mice, we also observed similar percentages of 2-cell embryos and live offspring derived from caput or caudal sperm. Interestingly, we noted a small but non-significant reduction in the number of live offspring when sperm were treated with the syringe, and also a non-significant reduction in the percentage of live offspring generated from caput sperm. In total, 20 and 30 pups were obtained by both procedures from caput and cauda sperm, respectively. Birth weights and blastocyst development were similar between the caput and cauda groups and both males and females were fertile (n = 8 per group).

## DISCUSSION

A recent report describes that embryos generated via ICSI using immature sperm derived from the caput epididymis show multiple defects during the peri-implantation period and quickly fail to undergo implantation (Conine et al., 2018). The authors


TABLE 1 | Development of oocytes injected with sperm from caput vs. caudal epididymis.

Four independent experiments were conducted. B6D2 = C57/BL/6 JxDBA/2 J; B6 = C57BL/6NHsd; DWG T syringe = Drawing through syringe. <sup>a</sup>All recipients were pregnant. P > 0.05 Chi-square analysis.

of this paper also observed that these defects could be rescued by the injection of small RNAs derived from extracellular vesicles (epididymosomes), which deliver their small RNA repertoire to mature sperm, suggesting that small RNAs in mature sperm play major roles during early mouse embryogenesis. Here, we report that similar embryo development and offspring production results for ICSI using immature sperm from the caput epididymis or mature sperm from the caudal epididymis.

In mice, ICSI is performed with isolated sperm heads. However, sperm heads can be obtained by different protocols and this could have an appreciable impact on the sperm. In our hands, both protocols used to generate individual sperm heads (freezing/thawing, the standard method used in our laboratory, and drawing through a syringe, the method employed by Conine et al. (2018) produced healthy offspring from immature caput sperm. However, the number of live offspring generated with caput sperm was slightly lower than that generated with caudal sperm. In effect, this could be due to differences in maturation and sperm chromatin status between caput and caudal sperm. We earlier reported that the immaturity of mouse spermatozoa in the caput epididymis was revealed by a higher permeability of membranes to propidium iodide (PI) (Perez-Cerezales et al., 2012). Sperm chromatin undergoes the formation of disulfide bridges between protamines during the transit through the epididymis, increasing compaction of the genetic material (Golan et al., 1996). Also, using the comet assay and sperm chromatin structure assay (SCSA), we previously described different maturation stages of spermatozoa along the epididymis consisting of more DNA fragmentation and less chromatin compaction in the caput- than cauda epididymal spermatozoa (Perez-Cerezales et al., 2012). These differences in chromatin structure could make caput sperm more sensitive to decapitation procedures. Perhaps, a more aggressive technique will lead to more severe chromatin alterations in spermatozoa from the caput epididymis, or leave DNA more exposed and thus susceptible to degradation by endogenous or exogenous nucleases. Oocytes can support the preimplantation development of sperm with damaged chromatin, but major developmental defects appear later, after implantation (Fernandez-Gonzalez et al., 2008). This is in line with clinical observations that ejaculated sperm with damaged DNA go through normal preimplantation embryonic stages but fail to develop beyond this stage (Tesarik et al., 2004). Further, another two procedures designed to separate the sperm head from tail have been used to produce normal offspring from caput sperm (Suganuma et al., 2005; Zhou et al., 2019). The sperm head can be separated from the midpiece and tail by applying one or a few piezo pulses (Suganuma et al., 2005) or through sonication or the use of detergents like Triton X-100 or 3-[(3-cholamidopropyl) dimethylammonio]-1 propanesulfonate (Zhou et al., 2019).

There are several methodological differences between Conine et al. (2018) and previous studies (Suganuma et al., 2005; Zhou et al., 2019) which could potentially account for the discrepancy about the ability of caput-derived embryos to implant. In our experiments we used similar protocols to Conine et al. (2018) with the exception of the mouse strain background (we use B6D2 hybrids and inbred B6, and Conine et al. (2018) use inbred FVB). Further, other inbred mice have been also used, like 129Sc (Suganuma et al., 2005), showing to produce normal caput-derived mice. Although we cannot eliminate the possibility that the ability of caput-derived embryos to implant could be affected by strain background our results rule out the general character of the findings reported by Conine et al. (2018) for the mouse species.

In summary, sperm from the caput region of the epididymis are incapable of IVF but, contrary to recent findings (Conine et al., 2018), we here confirm that healthy fertile mice can be produced by ICSI from caput sperm.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

#### ETHICS STATEMENT

The animal study was reviewed and approved by the Committee on the Ethics of Animal Experiments of the INIA (2016 permit number CEEA 2014/025).

#### AUTHOR CONTRIBUTIONS

RF-G and RL produced the ICSI mice and collaborated in the phenotyping. PR-I, EP, VA-L, and SP-C performed all the experiments and co-wrote the manuscript. AG-A conceived the experiments and co-wrote the manuscript.

## FUNDING

This work was supported by Grant RTI2018-093548-B-I00 from the Spanish Ministry of Science, Innovation and

#### REFERENCES


Universities. PR-I was funded by a Talent Attraction Fellowship from the Madrid Community (2017-T2/BIO-5182). SP-C was supported by the Ramón y Cajal Programme (RYC-2016-20147).


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Fernández-González, Laguna, Ramos-Ibeas, Pericuesta, Alcalde-Lopez, Perez-Cerezales and Gutierrez-Adan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Signaling Enzymes Required for Sperm Maturation and Fertilization in Mammals

Souvik Dey, Cameron Brothag and Srinivasan Vijayaraghavan\*

Department of Biological Sciences, Kent State University, Kent, OH, United States

#### Edited by:

Zhibing Zhang, Virginia Commonwealth University, United States

#### Reviewed by:

Andrew Burgess, Anzac Research Institute, Australia Haim Breitbart, Bar-Ilan University, Israel Ana Maria Salicioni, University of Massachusetts Amherst, United States

> \*Correspondence: Srinivasan Vijayaraghavan svijayar@kent.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 30 June 2019 Accepted: 03 December 2019 Published: 18 December 2019

#### Citation:

Dey S, Brothag C and Vijayaraghavan S (2019) Signaling Enzymes Required for Sperm Maturation and Fertilization in Mammals. Front. Cell Dev. Biol. 7:341. doi: 10.3389/fcell.2019.00341 In mammals, motility and fertilizing ability of spermatozoa develop during their passage through the epididymis. After ejaculation, sperm undergo capacitation and hyperactivation in the female reproductive tract – a motility transition that is required for sperm penetration of the egg. Both epididymal initiation of sperm motility and hyperactivation are essential for male fertility. Motility initiation in the epididymis and sperm hyperactivation involve changes in metabolism, cAMP (cyclic adenosine monophosphate), calcium and pH acting through protein kinases and phosphatases. Despite this knowledge, we still do not understand, in biochemical terms, how sperm acquire motility in the epididymis and how motility is altered in the female reproductive tract. Recent data show that the sperm specific protein phosphatase PP1γ2, glycogen synthase kinase 3 (GSK3), and the calcium regulated phosphatase calcineurin (PP2B), are involved in epididymal sperm maturation. The protein phosphatase PP1γ2 is present only in testis and sperm in mammals. PP1γ2 has a isoform-specific requirement for normal function of mammalian sperm. Sperm PP1γ2 is regulated by three proteins – inhibitor 2, inhibitor 3 and SDS22. Changes in phosphorylation of these three inhibitors and their binding to PP1γ2 are involved in initiation and activation of sperm motility. The inhibitors are phosphorylated by protein kinases, one of which is GSK3. The isoform GSK3α is essential for epididymal sperm maturation and fertility. Calcium levels dramatically decrease during sperm maturation and initiation of motility suggesting that the calcium activated sperm phosphatase (PP2B) activity also decreases. Loss of PP2B results in male infertility due to impaired sperm maturation in the epididymis. Thus the three signaling enzymes PP1γ2, GSK3, and PP2B along with the documented PKA (protein kinase A) have key roles in sperm maturation and hyperactivation. Significantly, all these four signaling enzymes are present as specific isoforms only in placental mammals, a testimony to their essential roles in the unique aspects of sperm function in mammals. These findings should lead to a better biochemical understanding of the basis of male infertility and should lead to novel approaches to a male contraception and managed reproduction.

Keywords: hyperactivation, fertility, PP1γ2, GSK3α, epididymal sperm maturation, PP2B, PKA, contraception

## INTRODUCTION

fcell-07-00341 December 16, 2019 Time: 15:39 # 2

Testicular spermatozoa in mammals are immotile and lack the ability to fertilize eggs. Motility initiation and fertilizing ability develop during their passage through the epididymis (Cooper, 1986; Bedford and Hoskins, 1990). Epididymal maturation is an absolute necessity to produce fertile spermatozoa. Sperm undergo hyperactivation in the female reproductive tract before penetration of the egg. Motile sperm failing to undergo hyperactivation cannot fertilize (Marquez and Suarez, 2008). Understanding the biochemical basis for how motility is attained and modified prior to fertilization is essential for understanding male fertility and infertility.

Sperm flagellar activity and also ciliary motility of somatic cells are known to be regulated by the intracellular mediators cAMP, calcium and pH. Capacitation is operationaly defined as functional modifications that enable sperm to fertilize eggs. Sperm undergo various changes during capacitation: (a) removal of membrane cholesterol followed leading to a decrease in cholesterol to phospholipid ratio (b) membrane hyperpolarization, rise in intra-sperm pH and cAMP levels, (c) increase in calcium uptake, and (d) an increase in protein tyrosine phosphorylation possibily due to increased protein kinase A (PKA) activity (Arcelay et al., 2008; Bailey, 2010; Alvau et al., 2016; Jin and Yang, 2017; Molina et al., 2018). It is generally accepted that these changes during capacitation lead to: (a) the ability of the sperm to bind the oocyte's extracellular matrix, the zona pellucida (ZP) (Si and Olds-Clarke, 1999; Topper et al., 1999) and subsequently undergo the acrosome reaction, (b) hyperactivation, a whiplash flagellar motion required to penetrate the egg (Ho and Suarez, 2001), and (c) the capacity to fuse with the oocyte (Evans and Florman, 2002). Hyperactivation, however, is regulated by similar yet distinct signaling events that distinguish it from capacitation (Marquez and Suarez, 2004).

Considerable progress has been made in understanding changes in sperm cAMP, calcium, and pH during sperm activation and hyperactivation. The proteins responsible for the generation of cAMP, calcium influx and changes in pHi (intracellular pH) have been identified and their functions elucidated by biochemical and genetic approaches. Despite this information, we still do not understand, in biochemical terms, how these second messengers enable sperm to acquire motility in the epididymis and how motility is altered in the female reproductive tract. Our understanding is incomplete because other signaling enzymes can profoundly alter the action and effects of second messengers in sperm. These signaling enzymes, protein phosphatase 1 (PP1), glycogen synthase kinase 3 (GSK3) and the calcium activated phosphatase (PP2B), affect sperm function in the epididymis and in the female reproductive tract. Several reviews detail the roles of sperm cAMP-PKA and calcium metabolism (Visconti et al., 2002; Burton and McKnight, 2007; Buffone et al., 2014; Stival et al., 2016; Freitas et al., 2017; Balbach et al., 2018; Leemans et al., 2019; Stewart and Davis, 2019). Following a brief summary of the actions of cAMP-PKA, calcium and pH, the remainder of the review will be focused largely devoted to examination of the key roles of PP1, GSK3, and calcium activated protein phosphatase, PP2B (PPP3R2/CC) in mediating the actions of the second messengers during sperm maturation and fertilization.

## ROLES OF cAMP AND PROTEIN KINASE A IN SPERM FUNCTION

Motility can be induced in demembranated testicular and caput epididymal sperm in the presence of ATP and cAMP, as well as appropriate calcium levels and pH (Lindemann, 1978; Mohri and Yanagimachi, 1980; Yeung, 1984; Lindemann et al., 1987; Bedford and Hoskins, 1990; Lesich et al., 2008). Early studies on the motility effects of phosphodiesterase inhibitors (such as caffeine, theophylline, and IBMX), on intact sperm led to the discovery of the role of cAMP in sperm motility regulation (Dougherty et al., 1976; Jiang et al., 1984). Phosphodiesterase inhibitors or cell permeable cAMP analogs can initiate motility in immotile spermatozoa and stimulate increased motility in motile sperm. Motility activation of demembranated sperm is abrogated in the presence of protein phosphatases in the reactivation medium, suggesting that cAMP mediates its motility effect through protein phosphorylation (Takahashi et al., 1985; Murofushi et al., 1986). An elevation of intra-sperm cAMP is also thought to be involved in motility initiation in the epididymis (Hoskins et al., 1974; Amann et al., 1982; Vijayaraghavan et al., 1985). A unique, hormone insensitive but bicarbonate sensitive, soluble adenylyl cyclase (sAC) is responsible for the synthesis of cAMP in sperm. Sperm cAMP levels should also be regulated by phosphodiesterases (PDEs) enzymes that degrade cAMP. Of the eleven PDEs, PDE1, PDE4, PDE8, PDE10, and PDE11 are present in testis and sperm (Omori and Kotera, 2007; Keravis and Lugnier, 2012).

It is well known that cAMP acts through a protein kinase (PKA). Knockout of the sperm sAC or PKA leads to infertility due to impaired sperm motility and the inability of sperm to undergo hyperactivation (Hess et al., 2005; Xie et al., 2006). Mature sperm contain PKA which is composed of regulatory subunit RIIα and a sperm specific catalytic subunit Cα2. Targeted knock-out of sperm-specific Cα2 results in male infertility. Spermatogenesis in these mutant mice is normal, while kinetic vigor and beat amplitude of epididymal sperm were markedly reduced. Mutant sperm are unable to undergo bicarbonate induced motility stimulation, and hyperactivation. It is known that PKA in cells owes it specificity and function to its localization through anchoring proteins known as AKAPs (Colledge and Scott, 1999). Sperm contain at least two PKA anchoring proteins AKAP 3 and 4 (also known as AKAP110 and 82, respectively) (Lin et al., 1995; Vijayaraghavan et al., 1999). Targeted deletion of AKAP 3 or 4 results in sperm dysfunction and male infertility (Jackson Labs Mouse Repository) (Miki et al., 2002). AKAP82 has been shown to play significant roles mediating the PKA action in murine and human sperm (Moss et al., 1999; Turner et al., 1999).

One of the downstream effects of PKA has been suggested to involve activation of protein tyrosine kinases (PTKs): proline−rich tyrosine kinase 2 (PYK2), ABL (Abelson murine leukemia viral oncogene homolog 1), SRC, and FER

(Baker et al., 2006, 2009; Alvau et al., 2016; Brukman et al., 2019). However, targeted disruption of these PTKs did not impair male fertility. Thus the exact role, of protein tyrosine phosphorylation in sperm function is still not well understood. PKA has also been suggested to regulate cAMP-phosphodiesterase activity. Phosphorylation by PKA has been shown to increase catalytic activities of PDE4 and PDE11 (Sette and Conti, 1996; Yuasa et al., 2000) suggesting a feedback regulation of sperm cAMP levels. Hyper-activated motility in sperm capacitation may also include phospholipase D-dependent actin polymerization (Itach et al., 2012). Thus while it is known that cAMP and PKA are essential in sperm the exact downstream biochemical actions of PKA are still unknown.

### THE ROLES OF pHi, CALCIUM AND CatSper CHANNELS IN SPERM FUNCTION

The role of pH in regulating flagellar motion was first recognized in the mechanism underlying activation of sea urchin sperm in sea water, which occurs due to increased pHi mediated by a sodium – proton exchange (Lee et al., 1983). Several studies have now shown that increasing intracellular pH activates motility of sperm in a number of species (Vijayaraghavan et al., 1985; Hamamah and Gatti, 1998; Carr and Acott, 2005; Nishigaki et al., 2014). A change in pHi is also thought to accompany sperm maturation in the epididymis (Vijayaraghavan et al., 1985). Acidic pH and the high concentration of lactate which acts as a membrane permeable proton carrier render sperm immotile in the luminal fluid of the cauda epididymis (Carr et al., 1985; Vijayaraghavan and Hoskins, 1988). Dilution of the quiescent caudal sperm, during ejaculation or in a buffer in vitro, results in initiation of vigorous motility. More recently, elevation of pHi has also been recognized to be a key event required for initiation of calcium influx during sperm hyperactivation. Knockout of channels, Slo3, NHE1 (Slca9c) or SLC26A3, responsible for changes in pH renders sperm infertile (Wang et al., 2007; Chen et al., 2009; Santi et al., 2010; Zeng et al., 2011; Chavez et al., 2014; Toure, 2019). A significant advance in understanding hyperactivation came from the discovery of calcium channels (CatSper) as essential mediators of sperm calcium influx. The calcium channel is composed of several essential subunits (Qi et al., 2007; Chang and Suarez, 2011; Singh and Rajender, 2015; Williams et al., 2015; Sun et al., 2017). Alkalinization of the sperm cytosol followed by activation of calcium channels (CatSper) triggers hyperactivation (Carlson et al., 2003; Qi et al., 2007; Lishko and Kirichok, 2010; Chung et al., 2011; Chavez et al., 2014; Lishko and Mannowetz, 2018; Orta et al., 2018). Prevention of the increase in pHi or loss of CatSper leads to infertility. A recent study shows that CatSper channels also mediate Zn2+-dependent stimulation of sperm hyperactivation (Allouche-Fitoussi et al., 2018). A model for sperm hyperactivation incorporating the functions of cAMP-PKA, calcium, and pHi is shown in **Figure 1**. This model is based on those presented in reviews noted earlier on the mechanisms regulating sperm function. Absent in theses models are the essential functions of the signaling enzymes PP1γ2, GSK3α, and PP2B required for sperm motility initiation and fertility.

## PROTEIN PHOSPHATASE ISOFORM, PP1γ2 IN SPERM FUNCTION

The presence of a protein kinase in a cell requires a corresponding protein phosphatase. The phosphorylation status of a protein is the result of the opposing activities of protein kinases and protein phosphatases. Based on the observation that protein phosphatases prevent motility initiation of demembranated spermatozoa and also enzyme activity measurements in sperm extracts (Swarup and Garbers, 1982; Takahashi et al., 1985; Murofushi et al., 1986), it was long suspected that a protein phosphatase regulates flagellar motility. However, the identity of the phosphatase and details of its regulation were not known. Research on protein phosphatases was boosted by the discovery of compounds, calyculin A, okadaic acid, and microcystin, isolated from marine organisms that are potent inhibitors of protein phosphatases (Cohen, 1990; Cohen et al., 1990; Fernandez et al., 2002). The inhibitors display distinct inhibition profiles against the serine/threonine phosphatases PP1, PP2A, and PP2B enabling their identification in cellular extracts (da Cruz e Silva et al., 1995). The protein phosphatase inhibitors, calyculin A and okadaic acid, were among the most potent in initiating and stimulating motility of sperm (Smith et al., 1996, 1999; Vijayaraghavan et al., 1996). The inhibitors initiated and stimulated epididymal sperm motility at nanomolar and micromolar concentrations, respectively. The inhibition profile of enzyme activity in sperm extracts suggested that the predominant phosphatase in sperm was protein phosphatase 1 (PP1) along with measurable levels of PP2A. There are four isoforms of PP1, PP1α, PP1β, PP1γ1, and PP1γ2, encoded by three genes (Okano et al., 1997; Lin et al., 1999). The amino acid sequences of all four proteins are essentially identical except at their C-termini. The two PP1 isoforms, PP1γ1 and PP1γ2, are alternate transcripts from a single gene, Ppp1c (**Figure 2A**). Based on enzyme activity profiles and western blot analysis, we found that the predominant serine/threonine protein phosphatase in spermatozoa is PP1γ2. High PP1γ2 activity is associated with low sperm motility, while low PP1γ2 activity is associated with vigorous motility (Smith et al., 1996, 1999; Vijayaraghavan et al., 1996). A decline in PP1γ2 activity occurs during epididymal sperm maturation, due to a decrease in its catalytic activity. Other laboratories have shown that the phosphatase inhibitors also promote hyperactivated sperm motility and acrosome reaction (Furuya et al., 1992a,b; Signorelli et al., 2013; Rotfeld et al., 2014; Matsuura and Yogo, 2015; Tsirulnikov et al., 2019). The enzyme PP1γ2 is present in spermatozoa of a wide range of mammalian species including human and non-human primates (Chakrabarti et al., 2007a; Vijayaraghavan et al., 2007).

In most tissues and cells, the loss of any one of the PP1 isoforms is compensated by one of the other isoforms. In yeast the loss of its endogenous protein phosphatase (GLC7) can be functionally replaced by one of the four mammalian PP1

isoforms, highlighting their functional equivalence (Gibbons et al., 2007). Because PP1γ2 was implicated in sperm motility it was of interest to see how its loss would affect sperm function. Loss of Ppp1cc leads to defects in spermiogenesis and lack of sperm in the epididymis (Varmuza et al., 1999; Chakrabarti et al., 2007b). The enzyme PP1γ2 has a dual role, one, during spermatogenesis and the other in sperm after their exit from the seminiferous tubules. This dual role of PP1γ2 is intriguing because loss of the enzymes of cAMP metabolism and action, sAC or PKA in testis, does not impair sperm morphogenesis or sperm formation. It is likely that other serine/threonine protein kinases along with PP1γ2 are responsible for regulation of protein phosphorylation during spermatogenesis (Kawa et al., 2006; Xu et al., 2007; MacLeod et al., 2014; Cruz et al., 2019). Due to its requirement in spermatogenesis, it is not possible to obtain sperm lacking PP1γ2, which is a limitation for the study of the enzyme in mature sperm.

The only phenotype resulting from the knockout of Ppp1cc is male infertility. Females lacking Ppp1cc are normal and fertile (**Table 1**). Conditional knockout in post-meiotic developing germ cells also has the same phenotype as the global loss of Ppp1cc showing the requirement of Ppp1cc only in differentiating germ cells in testis (Sinha et al., 2013). It should be noted that the Ppp1cc gene is responsible for expression of both the PP1 isoforms: PP1γ1 and PP1γ2. Since, PP1γ2 is the predominant isoform in testis, it strongly suggests, but does not prove that the reason for male infertility in mice lacking Ppp1cc is likely due to the absence of only PP1γ2 in differentiating spermatogenic cells. It was later confirmed that, despite the global absence of PP1γ1, transgenic expression of PP1γ2 driven by the PGK2 promoter in spermatocytes and spermatids of Ppp1cc null mice restored spermatogenesis, sperm function, and fertility (Sinha et al., 2012). These data provide compelling evidence that the PP1γ2 isoform expressed only in developing germ cells is sufficient for normal sperm function and fertility.

The phosphatase isoform, PP1γ2, is present only in eutherian mammals. Sperm from non-mammalian species and invertebrates contain one of the three PP1 isoforms – PP1α,

retained as an extended exon leading to the eight amino acid C-terminus of PP1γ1 (note that "exon 8" is part of its 30UTR). The PP1γ1 encodes a protein containing 323 amino acids derived from the seven exons along with the 8 amino acid C-terminus from the extended exon 7. In the post-meiotic germ cells in testis the intron 7 is spliced out, thus, producing a shorter Pp1γ2 transcript of approximately 1.7 kb. Exon 8 codes for the 22 amino acid C-terminus in PP1γ2. Thus, the amino acid sequences of PP1γ1 and PP1γ2 are identical in all respects except for their extreme C-termini. (B) Constructs for generating transgenic PP1γ1 mice. Rescues I–III constructs contain the entire or a portion of intron 7 which is part of the 30UTR of the messenger RNA for PP1γ1. There was little or no transgenic expression of PP1γ1 in testis of mice generated from these constructs. The last rescue construct (Rescue IV) lacks the 0.9 kb region of intron 7 following the stop codon in PP1γ1 mRNA. Transcript from this construct will resemble PP1γ2 mRNA except that PP1γ1 protein will be produced. The transgenic mice produced from this construct expressed high testis levels of transgenic PP1γ1 and rescued spermatogenesis but not sperm fertility.

PP1β, or PP1γ1 – which is able to support sperm motility and fertility in these species. The fact that PP1γ2 alone is sufficient for male fertility does not necessarily suggest that the other PP1 isoforms would be unable to functionally replace PP1γ2. Is the requirement for PP1γ2 an evolutionary accident or is there an isoform specific function for it in mammalian sperm? Can expressing the PP1γ1 isoform in testis, restore spermatogenesis and fertility of Ppp1cc null mice? Employing the same strategy used in the PP1γ2 transgenic rescue approach, the PGK2 promoter was used to drive transgenic expression of PP1γ1. This approach which was successful for PP1γ2 expression (Sinha et al., 2012), failed to transgenically express PP1γ1 (Dudiki et al., 2019) in testis. The transgene construct in all these failed attempts contained portions of the intron preceding exon 8, which is part of the 30UTR of the mRNA for PP1γ1 (**Figure 2B**). In the fourth attempt, removal of this entire portion of the 30UTR in the cDNA in the transgene construct (**Figure 2B**) led to robust expression of PP1γ1 in developing spermatocytes and spermatids (Dudiki et al., 2019). Transgenic expression of PP1γ1 in testis of Ppp1cc null mice was able to fully restore spermatogenesis. However, sperm function and fertility were severely compromised in these PP1γ1 rescue mice. Motility of PP1γ1-bearing sperm was diminished and their flagellar beat amplitude was severely dampened (Dudiki et al., 2019). Fertility defects in the rescue mice were most likely due to the inability of sperm bearing PP1γ1 to undergo hyperactivation. Thus PP1γ2 is essential in sperm for its normal function and fertility.



A specific isoform requirement of a protein is usually thought to arise due to isoform specific binding partners for that protein or due to its unique biochemical activity. An isoform specific function could also be due to the restricted spatiotemporal expression of the protein isoform during cell or tissue development. Despite the knowledge that PP1γ2 is the only PP1 isoform expressed in developing spermatocytes and spermatids (Chakrabarti et al., 2007b). it was anticipated that specific binding proteins for PP1γ2 exist in testis and sperm. However, binding partners of sperm PP1γ2 identified so far are ubiquitous in tissues and organisms and are known to bind to all PP1 isoforms (Goswami et al., 2019). That is, these proteins would bind to any PP1 isoform if present in sperm, just as they do in other cells and tissues. The three protein regulators of PP1γ2 identified are PPP1R2 (inhibitor I2), PPP1R7 (SDS22), and PPP1R11 (inhibitor I3). These three regulators/inhibitors are evolutionarily ancient and conserved across species (Heroes et al., 2013), play key roles in mitosis and other cellular functions (Peggie et al., 2002; Wang et al., 2008; Eiteneuer et al., 2014). Thus, loss of any one them is likely to cause embryonic lethality. It is intriguing that the sperm specific isoform, PP1γ2, is regulated by these ancient, ubiquitous, and essential PP1 binding proteins. The three regulators share localization with PP1γ2 in the head and the principal piece of sperm. The association of inhibitors to PP1γ2 changes during epididymal sperm maturation. In immotile caput epididymal sperm, PPP1R2 and PPP1R7 are not bound to PP1γ2, whereas in motile caudal sperm, all three inhibitors are bound as hetero-dimers or hetero-trimers (Goswami et al., 2019) (**Table 2** and **Figures 3A,B**). In caudal sperm from male mice lacking sAC and GSK3 (see below), where motility and TABLE 2 | The binding profile of the regulators with PP1γ2 in caput and caudal epididymal sperm is summarized in the table along with data with caudal sperm from Gsk3α knockout, PKA Cα2 knockout and sAC knockout mice where the binding status of sds22 resembles that in wild type caput sperm.


Thus, epididymal sperm maturation appears to be altered in these knockout mice.

fertility are impaired, the association of PP1γ2 to the inhibitors resembles immature caput sperm. In sperm containing PP1γ1 the association of these inhibitors are altered resembling that of PP1γ2 in immotile caput epididymal sperm (Goswami et al., 2019) (**Table 2**). It is known that binding of inhibitor 2 to PP1 is regulated by GSK3 (**Figure 3B**). It is likely that binding of the other two inhibitors, PPP1R7 and PPP1R2, to PP1γ2 are also regulated by phosphorylation. Changes in the associations of the regulators with PP1γ2, are likely part of biochemical mechanisms responsible for the development of motility and fertilizing ability of sperm.

A recently identified potential binding protein for PP1γ2 is CCDC181 (Schwarz et al., 2017). It is likely that CCDC181 binds to PP1γ1 and PP1γ2 with differing affinities leading to preferential localization of PP1γ2 to the flagellum. Determination of the protein targets of PP1γ2 in the flagellum and how CCDC181 regulates PP1γ2 is under active investigation.

#### GLYCOGEN SYNTHASE KINASE 3α, GSK3α, IN SPERM

The enzyme GSK3, a serine/threonine protein kinase, was named GSK-3 because it was discovered after PKA and phosphorylase kinase (GSK-1 and GSK-2). Two other GS kinases, GSK-4 and -5, named based on their relative elution profiles in phosphocellulose chromatography of muscle extracts, were later renamed as casein kinase 1 and 2, respectively (Parker et al., 1982). The enzyme GSK3, retained its original name even though it was later found to be a key signaling component of a large number of cellular processes (Kaidanovich-Beilin and Woodgett, 2011; Medina and Wandosell, 2011). An array of functions attributed to GSK3 include insulin action, regulation of cell survival, apoptosis, embryonic development, Wnt/β-catenin and hedgehog signaling, and growth factor action. It is also a target for drug development in several clinical disorders including cancer (Jope et al., 2007).

In mammals, GSK3 is ubiquitous and is expressed as two isoforms, GSK3α and GSK3β, encoded by different genes. The catalytic domains of the two isoforms are 98% identical while

GSK3α, PKA Cα2, PP1γ2 and PP2B (PPP3R2/CC) during epididymal sperm maturation and sperm hyperactivation in female reproductive tract. Curved orange

arrow(s) indicate relative degree of activities of the enzymes; straight black arrow(s) denote relative level of the ion/protein inside the cell.

their N- and C-termini are distinctive (Woodgett, 1990). While there are reports ascribing distinct roles for each of the isoforms (McNeill and Woodgett, 2010) under most circumstances the two isoforms are redundant and functionally interchangeable. Knockout of Gsk3β in mice causes late embryonic lethality (Hoeflich et al., 2000). The inability of GSK3α to substitute for GSK3β in the developing embryo may be due to the non-overlapping expression of the two isoforms. Conditional knockout of the floxed Gsk3β alleles on a Gsk3α null background show that complete loss of both isoforms impairs signaling and tissue function. However, one allele of Gsk3β or Gsk3α on a Gsk3α or Gsk3β null background, respectively, is sufficient to maintain normal Wnt signaling and tissue function (Doble et al., 2007; McNeill and Woodgett, 2010) highlighting the functional redundancy of the two isoforms in most tissues and cell types.

The protein, GSK3 was first discovered as an enzyme responsible for activation of PP1γ2 in bovine sperm (Vijayaraghavan et al., 1996). Both α and β isoforms of GSK3 are present in sperm. Immotile caput sperm contain four-fold higher GSK3 activity than motile caudal epididymal sperm. Both tyrosine phosphorylation (which stimulates catalytic activity) and serine phosphorylation of GSK3 (an inhibitory mechanism) increase significantly in sperm during their passage through the epididymis (Somanath et al., 2004). Incubation of motile or immotile sperm with compounds that activate PKA (e.g., dbcAMP) or inhibit protein phosphatase (e.g., calyculin A) is accompanied by increases in GSK3 serine phosphorylation and motility stimulation. GSK3 tyrosine phosphorylation which is believed to be autoregulatory, remains unchanged during capacitation, while only GSK3α ser21 phosphorylation is altered during this event (Dey et al., 2019a,b).

It was recently shown that Gsk3α null mice exhibit male infertility (Bhattacharjee et al., 2015, 2018). Knockout of GSK3α in post-meiotic testicular germ cells, using the Cre-Lox strategy, also results in male infertility. Mice with a testis knockout of GSK3β are normal and fertile. Thus, GSK3α has an isoform specific function in sperm. Analysis of sperm lacking GSK3α showed that adenine nucleotide levels, energy metabolism,

and protein phosphatase and kinase activities were affected suggesting impaired sperm maturation in the epididymis. A recent report also documents the role for GSK3 and a noncanonical Wnt signaling during epididymal sperm maturation: loss of Wnt signaling in sperm results in male infertility (Koch et al., 2015). The activity of GSK3α isoform has also been correlated with human sperm motility (Freitas et al., 2019). The inability of GSK3β to replace GSK3α, only in testis and sperm, is surprising given the fact that the two isoforms are functionally interchangeable in most cellular contexts and in tissues. Thus, despite the presence of both GSK3 isoforms, mammalian sperm are unique in their requirement for the GSK3α isoform. Taken together, these studies, support the notion that GSK3α is essential for epididymal sperm maturation, motility, and fertilization.

## CALCINEURIN IN SPERM FUNCTION

Calcineurin (also known as PP2B or PPP3C) is a serine/threonine phosphatase regulated by calcium. In response to an elevation of cellular calcium, calmodulin binds to a calmodulin binding region of the catalytic subunit PPP3C. This binding causes an auto-inhibitory arm of calcineurin to move away from the substrate binding site thus activating the enzyme by enabling its access to substrates (Rusnak and Mertz, 2000; Parra and Rothermel, 2017). The catalytic activity of the enzyme is also regulated by calcium binding to a regulatory subunit (PPP3R2). Regulation and function of calcineurin in several cell types has been extensively studied (Rusnak and Mertz, 2000; Parra and Rothermel, 2017). More than two decades ago a role for a calcium regulated protein phosphatase was proposed in the regulation of sperm motility (Tash et al., 1988; Tash and Bracho, 1994). In non-mammalian sperm, calcineurin has been shown to have role in activation of progressive motility and egg activation (Levasseur et al., 2013; Krapf et al., 2014). The catalytic and regulatory subunits of calcineurin are present as testis-specific isoforms, PPP3CC and PPP3R2. It was shown by super resolution microscopy that the catalytic subunit of calcineurin, PPP3CC, is localized near the quadrilateral structures of CatSper in the axoneme (Chung et al., 2014). In CatSper1-deficient spermatozoa, PPP3CC can be seen localized mostly to the axoneme but disappears from the quadrilateral structures. Another report showed that pharmacological inhibition of calmodulin affects protein tyrosine phosphorylation seen during sperm capacitation (Navarrete et al., 2015). In another study, micromolar amounts of FK506 has been demonstrated to prevent sperm acrosomal exocytosis (Castillo Bennett et al., 2010). A recent report now shows that knockout of either Ppp3CC or Ppp3R2 present only in testis and sperm resulted in male infertility (Miyata et al., 2015). Sperm numbers and testis weights in these knockout mice are normal; but sperm motility is impaired with a stiffened mid-piece. The Ppp3CC or Ppp3R2 knockout mice are infertile in vivo. Sperm from these knockout mice also cannot fertilize eggs in vitro. Surprisingly, wild type sperm treated with the calcineurin inhibitors, FK506 and cyclosporine, did not affect in vitro fertilization. Thus, infertility was thought to be due to impaired sperm function in the male reproductive tract. Calcineurin inhibitors injected into mice resulted in reversible male infertility. The investigators concluded that calcineurin was required for epididymal sperm maturation: genetic disruption or pharmacological inhibition in vivo affected sperm maturation causing infertility (Miyata et al., 2015).

The questions of how calcineurin may act and how it is activated during epididymal sperm maturation were not addressed. Recent data show that calcineurin and GSK3 are interrelated in their roles in epididymal sperm maturation and absence of calcineurin increases GSK3 phosphorylation resulting in its lower catalytic activity (Dey et al., 2019b). It is suspected that calcineurin regulates mitochondrial energization directly and glycolysis indirectly through its effect on GSK3. It is likely that high calcium levels in immature sperm (Vijayaraghavan and Hoskins, 1989, 1990) activates calcineurin.

#### INTERRELATIONSHIP BETWEEN PKA, PP1, GSK3α, AND PP2B

As described earlier the requirement and the roles of cAMP and the kinase activated by it, PKA, in sperm are well known. The relationship between PKA and GSK3 in sperm was indicated by the fact that cAMP analogs increased GSK3-α/β phosphorylation, without any isoform specificity. Phosphorylation of both GSK3 α and β isoforms were reduced and its catalytic activity increased in sperm with diminished cAMP (using KH7, a sAC inhibitor), or due to knock out of sAC. That is, GSK3 is a target of PKA phosphorylation. This relationship between GSK3 and PKA was further validated in GSK3α knockout mice. Loss of GSK3α or β by targeted disruption or pharmacological inhibition of the enzyme significantly reduced sperm cAMP levels (Dey et al., 2018). The decrease in cAMP levels was attributed to increased phosphodiesterase activity. Together these data support the possibility that GSK3 and cAMP form an interrelated regulatory loop (Dey et al., 2018).

Inhibition of the predominant sperm protein phosphatase PP1γ2 by calyculin A significantly increased phosphorylation of both GSK3 isoforms (Ser21/9) in caudal epididymal sperm and a concomitant decrease in its catalytic activity. Conversely, increased GSK3 activity is associated with increased PP1γ2 activity. One of the ways by which GSK3 regulates PP1 activity is by its phosphorylation of the inhibitor I2 as discussed earlier. Phosphorylated I2 dissociates from PP1 leading to its activation (**Figure 3B**). The roles of binding and dissociation of PP1γ2 binding proteins due to their reversible phosphorylation is a feature of the regulation of the phosphatase in sperm (Goswami et al., 2019). Sperm PP2A also targets GSK3, without any isoform specificity; however, its role in regulating GSK3 has not been investigated in detail (Dudiki et al., 2015).

New information shows that not only is GSK3 a target of PP1γ2, but it is also regulated by calcineurin. While PP1γ2 acts on both isoforms of GSK3 (Somanath et al., 2004),

calcineurin preferentially dephosphorylates only the GSK3α isoform (Dey et al., 2019a,b). Phosphorylation of the GSK3α (Ser21) isoform is elevated in sperm lacking calcineurin (Dey et al., 2019a,b). Following a decrease in their catalytic activities in caudal epididymal sperm during motility initiation, surprisingly the catalytic activities of both GSK3α and calcineurin increase during sperm capacitation and hyperactivation, recapitulating the situation in caput sperm (**Figure 3C**) (Dey et al., 2019a,b). Increased calcineurin activity, presumably following increased sperm Ca2<sup>+</sup> during hyperactivation is responsible for decreased phosphorylation and increased catalytic activity of GSK3α. Pharmacologic inhibition of calcineurin during capacitation abrogated this decrease in phosphorylation and the increase in GSK3 activity. Increased activities of GSK3α and PP2B appear to be characteristics of not only capacitated sperm, but also, paradoxically, of immature caput epididymal sperm. Thus, phosphatases (PP1γ2 and PP2B) and the kinases (PKA and GSK3) are mechanistically interrelated during epididymal initiation of motility and also during fertilization of the egg (summarized in **Figure 3C**).

#### THE PROTEINS PP1γ2, GSK3α, PKA Cα2, AND PPP3R2/CC PRESENT ONLY IN SPERM ARE CONSERVED IN MAMMALS

As noted earlier the PP1γ2 isoform in sperm, is present only in mammals. Examination of the available annotated genome databases shows that monotremes and other non-mammalian vertebrates contain the Ppp1cc gene but the PP1γ2 isoform cannot form due to the absence of splice sites at exon 7 in the gene. Also absent is a region corresponding to exon 8, which is highly conserved and present in the Ppp1cc in all mammals (**Figure 4A**). This PP1γ2 isoform derived from exon 8 has a

conserved in all mammals only a portion of the N-terminus sequence of PPP3R2 is shown.

unique 22 amino acid C-terminus present and conserved in all mammals (**Figure 4A**). The somatic isoform, PP1γ1 has a six amino acid C-terminus derived from the extended exon 7. Aside from these differences in the C-termini, the rest of 315 amino acid sequences of the both PP1γ1 and PP1γ2 are identical. Why mammalian sperm contain only PP1γ2 is unknown.

It is known that knockout of the enzyme synthesizing cAMP in sperm, sAC, and that of PKA result in male infertility. It is intriguing that the catalytic subunit of PKA is also expressed as a sperm specific isoform (PrkaCA also known as PKA Cα2) with a unique six amino acid N-terminus due to expression of an alternate exon, exon 1b (Agustin et al., 2000; Desseyn et al., 2000; San Agustin and Witman, 2001) (**Figure 4B**). The exon 1a codes for 12 amino acids of the N-terminus of the ubiquitous and the somatic cell version of the catalytic subunit. Aside from this difference in their N-termini the primary sequences of the rest of the catalytic subunits are identical. The reason this different N-terminus is required in testis and sperm remains puzzling because both sperm and somatic cell forms of the PKA catalytic subunits have identical biochemical properties (Vetter et al., 2011). Removal of exon 1b leading to the loss of PKA in sperm, renders males infertile (Nolan et al., 2004). Whether replacement

of the sperm form of the enzyme with the somatic form of the PKA catalytic subunit would sustain normal sperm function is not known. However, it is intriguing that exon 1b, the sperm specific isoform of PKA, is present only in mammals (Soberg et al., 2017). Non-mammalian species only contain the isoform derived from utilization of exon 1a (**Figure 4B**).

A germ cell–enriched protein, viz. sperm PKA interacting factor (SPIF), was found to be co-expressed and co-regulated with PKACα2 and with t-complex protein (TCP)-11 (Stanger et al., 2016). These three proteins constitute part of a novel trimeric complex in murine spermatozoa. During capacitation, the SPIF undergoes phosphorylation leading to a molecular rearrangement that brings PKACα2 and TCP11 into close proximity of each other. These results could explain how PKA Cα2 functions as a specific isoform complexed with SPIF and TCP11 during capacitation and fertilization.

The GSK3α isoform is essential in mammalian sperm despite the fact both GSK3α and β isoforms are interchangeable in other cells and tissues. GSK3α arose in vertebrates presumably by gene duplication of GSK3β which is the only isoform in invertebrates. Among vertebrates the GSK3α isoform is absent in birds (Alon et al., 2011). Sequence comparison of the extended glycine rich N-terminus present only in GSK3α shows that this sequence segment is highly conserved only in placental mammals. This extended N-terminus is present, but not conserved in non-mammalian vertebrates (**Figure 4C**). We predict that sperm from non-mammalian vertebrates contain only GSK3β and not GSK3α. In fact, we have shown that Xenopus sperm contains only GSK3β despite the fact both GSK3 isoforms are present in the genome of this species. These observations are compatible with the possibility that mammalian sperm contain a GSK3α-specific binding protein. Efforts to identify a isoform-specific binding protein using a two-hybrid approach with testis cDNA yielded a number of GSK3 binding proteins, but these GSK3 interactors were not isoform specific (Freitas et al., 2019). However, a yeast two-hybrid approach using human fetal cDNA yielded four GSK3α-specific binding proteins one of which appears to play a role in regulating circadian rhythm (Zeidner et al., 2011). One of these four GSK3α binding proteins is CENPV which is highly expressed in testis (NCBI, mouse ENCODE transcriptome data). CENPV also binds to tubulin (Honda et al., 2009) and is therefore, expected to be localized along the length of the flagellum. Thus, GSK3α in the flagellum is likely to orchestrate phosphorylation of proteins involved in regulating sperm motility and hyperactivation. The requirement of sperm GSK3α with its conserved of N-terminus suggests a role for it in maturation and fertilization events unique to mammals.

A relatively recent addition to the list of signaling enzymes regulating sperm function is the calcium regulated phosphatase, PP2B. Catalytic activity of PP2B is required in sperm during their passage through the epididymis as is the case with GSK3α. Sperm and testis express specific isoforms of the catalytic and regulatory calcineurin, PPP3CC and PPP3R2, respectively. Examination of the genomic sequences of several species shows that PPP3R2 is present only in eutherian mammalians and its amino acid sequence is remarkably conserved in the 121 mammals for which annotated genomic databases exist (**Figure 4D**).

Several predominantly testis-expressed proteins present in mammalian sperm have been identified as mammal-specific. These include proteins involved in DNA binding, sperm egg binding or those required for required for the unique structural features of mammalian sperm, such as protamine 3, SMCP, and ADAM proteins (Cho, 2012; Luis Villanueva-Canas et al., 2017). However, the observation that the four signaling enzymes suggested to be mechanistically interrelated are mammal specific isoforms is significant, suggesting a unique function in male gametes (**Figure 3C**). The specific isoforms PP1γ2, GSK3, PKA and PP2B play key roles in regulation sperm motility and hyperactivation, a phenomenon unique only to mammals (**Figure 5**). It would appear that their roles in these physiological functions in mammals arise due to their location in the flagellum. **Figure 5** shows the intrasperm localization of these enzymes. It would be interesting to determine if one or more of CatSper complexes located along the flagellum are mammal specific. The two possible binding proteins for PP1γ2 in the flagellum are CCDC181 (Schwarz et al., 2017) and PPP1R32 (Cifuentes et al., 2018), both of which are expressed predominantly or exclusively in the testis. As described above, the GSK3α binding protein CENPV should be present bound to the flagellum because CENPV also binds to tubulin, which in sperm is only present in tail (Honda et al., 2009). Sperm calcineurin has been shown by high resolution microscopy to be localized along the flagellum (Chung et al., 2014). The nature of the binding protein for calcineurin in sperm is not known. Finally, PKA is known to bind AKAP3, which is present along the sperm flagellum (Vijayaraghavan et al., 1999). How the unique N-terminus of the sperm specific PKA catalytic subunit affects this localization is not known. Thus, all the four signaling proteins localized in the flagellum (PP1γ2, GSK3α, PP2B, and PKA) are likely to determine the phosphorylation status of proteins orchestrating motility and metabolism required for normal sperm function (**Figure 3C**).

In summary, considerable data show that the enzymes PP1γ2, GSK3 and PP2B, along with PKA, are mechanistically interrelated in regulating the two physiological processes unique to mammals: epididymal sperm maturation and sperm hyperactivation preceding fertilization (**Figure 5**).

#### AUTHOR CONTRIBUTIONS

SV and SD contributed to the conception and outline of the review. SD and CB organized the database. SV wrote the first draft of the manuscript. SD wrote the sections of the manuscript. All authors contributed to the manuscript reading and revisions.

#### FUNDING

This work was funded by the National Institute of Health (SV): R03HD096176 and R21HD086839.

## REFERENCES

fcell-07-00341 December 16, 2019 Time: 15:39 # 12


acrosomal exocytosis. J. Biol. Chem. 285, 26269–26278. doi: 10.1074/jbc.M109. 095752


signaling shown by using an allelic series of embryonic stem cell lines. Dev. Cell 12, 957–971. doi: 10.1016/j.devcel.2007.04.001


fcell-07-00341 December 16, 2019 Time: 15:39 # 13


fertility and sperm cell hyperactivated motility. Proc. Natl. Acad. Sci. U.S.A. 104, 1219–1223. doi: 10.1073/pnas.0610286104


fcell-07-00341 December 16, 2019 Time: 15:39 # 14


a novel, sperm-specific protein kinase A-anchoring protein. Mol. Endocrinol. 13, 705–717. doi: 10.1210/mend.13.5.0278


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Dey, Brothag and Vijayaraghavan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Kinase Anchor Protein 4 Is Vulnerable to Oxidative Adduction in Male Germ Cells

Brett Nixon1,2 \*, Ilana R. Bernstein1,2, Shenae L. Cafe1,2, Maryse Delehedde<sup>3</sup> , Nicolas Sergeant3,4, Amanda L. Anderson1,2, Natalie A. Trigg1,2, Andrew L. Eamens1,2 , Tessa Lord1,2, Matthew D. Dun5,6, Geoffry N. De Iuliis1,2 and Elizabeth G. Bromfield1,2,7

<sup>1</sup> Priority Research Centre for Reproductive Science, School of Environmental and Life Sciences, Discipline of Biological Sciences, The University of Newcastle, Callaghan, NSW, Australia, <sup>2</sup> Pregnancy and Reproduction Program, Hunter Medical Research Institute, New Lambton Heights, NSW, Australia, <sup>3</sup> SPQI – 4BioDx-Breeding Section, Lille, France, <sup>4</sup> University of Lille, INSERM UMRS, Lille, France, <sup>5</sup> Cancer Signalling Research Group, School of Biomedical Sciences and Pharmacy, Faculty of Health and Medicine, University of Newcastle, Callaghan, NSW, Australia, <sup>6</sup> Priority Research Centre for Cancer Research Innovation and Translation, Hunter Medical Research Institute, New Lambton Heights, NSW, Australia, <sup>7</sup> Department of Biochemistry and Cell Biology, Faculty of Veterinary Medicine, Utrecht University, Utrecht, Netherlands

#### Edited by:

Xin Zhiguo Li, University of Rochester, United States

#### Reviewed by:

Woo-Sung Kwon, Kyungpook National University, South Korea Philip Chi Ngong Chiu, The University of Hong Kong, Hong Kong

\*Correspondence: Brett Nixon brett.nixon@newcastle.edu.au

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 11 July 2019 Accepted: 20 November 2019 Published: 20 December 2019

#### Citation:

Nixon B, Bernstein IR, Cafe SL, Delehedde M, Sergeant N, Anderson AL, Trigg NA, Eamens AL, Lord T, Dun MD, De Iuliis GN and Bromfield EG (2019) A Kinase Anchor Protein 4 Is Vulnerable to Oxidative Adduction in Male Germ Cells. Front. Cell Dev. Biol. 7:319. doi: 10.3389/fcell.2019.00319 Oxidative stress is a leading causative agent in the defective sperm function associated with male infertility. Such stress commonly manifests via the accumulation of pathological levels of the electrophilic aldehyde, 4-hydroxynonenal (4HNE), generated as a result of lipid peroxidation. This highly reactive lipid aldehyde elicits a spectrum of cytotoxic lesions owing to its propensity to form stable adducts with biomolecules. Notably however, not all elements of the sperm proteome appear to display an equivalent vulnerability to 4HNE modification, with only a small number of putative targets having been identified to date. Here, we validate one such target of 4HNE adduction, A-Kinase Anchor Protein 4 (AKAP4); a major component of the sperm fibrous sheath responsible for regulating the signal transduction and metabolic pathways that support sperm motility and capacitation. Our data confirm that both the precursor (proAKAP4), and mature form of AKAP4, are conserved targets of 4HNE adduction in primary cultures of post-meiotic male germ cells (round spermatids) and in mature mouse and human spermatozoa. We further demonstrate that 4HNE treatment of round spermatids and mature spermatozoa results in a substantial reduction in the levels of both proAKAP4 and AKAP4 proteins. This response proved refractory to pharmacological inhibition of proteolysis, but coincided with an apparent increase in the degree of protein aggregation. Further, we demonstrate that 4HNE-mediated protein degradation and/or aggregation culminates in reduced levels of capacitation-associated phosphorylation in mature human spermatozoa, possibly due to dysregulation of the signaling framework assembled around the AKAP4 scaffold. Together, these findings suggest that AKAP4 plays an important role in the pathophysiological responses to 4HNE, thus strengthening the importance of AKAP4 as a biomarker of sperm quality, and providing the impetus for the design of an efficacious antioxidant-based intervention strategy to alleviate sperm dysfunction.

Keywords: 4-hydroxynonenal, A-kinase anchor protein 4, male germ cells, oxidative stress, spermatozoa, sperm capacitation, sperm motility

## INTRODUCTION

fcell-07-00319 December 19, 2019 Time: 16:47 # 2

A male factor is implicated in approximately half of all cases of infertility (Dohle et al., 2005), and in the majority of these individuals, the underlying etiology is attributed to dysregulation of sperm function (Aitken, 2018). Among the myriad of causative agents, oxidative stress has repeatedly been implicated as a key contributor to the loss of sperm developmental and functional competence (Tremellen, 2008; Aitken et al., 2010). Indeed, notwithstanding the important physiological role that reactive oxygen species (ROS) serve to drive the latter phases of sperm maturation (Aitken and Nixon, 2013), exposure to even modest levels of ROS has the potential to induce a state of oxidative stress that deleteriously affects the unique physiology of the male gamete (Aitken et al., 2016). Notably, the legion of pathologic impacts resulting from elevated ROS vary considerably depending on the timing of exposure, with recent evidence suggesting that the round spermatid stage of development is particularly sensitive to oxidative attack. Specifically, post-ROS exposure, round spermatids rapidly succumb to a novel form of regulated cell death, termed ferroptosis (Bromfield et al., 2019). By contrast, equivalent levels of ROS elicit functional lesions in the mature spermatozoon, which compromise their fertilization potential, but does not negatively impact their viability (Bromfield et al., 2015). Such differential pathogenesis may be attributed to the highly specialized architecture of the male germ cell, which depending on their stage of differentiation, features an abundance of substrates for free radical attack, minimal antioxidant defense enzymes, and limited capacity for self-repair when oxidative damage is sustained (Walters et al., 2018b).

Irrespective of the timing of oxidative stress exposure, a virtually ubiquitous response to this insult is lipid peroxidation, a process whereby hydroxyl radicals attack the unsaturated fatty acyl chains of membrane phospholipids (Hauck and Bernlohr, 2016). The ensuing peroxidation reactions generate appreciable levels of highly reactive short chain carbonyl compounds. Among the most abundant and cytotoxic of these secondary oxidation products is 4-hydroxynonenal (4HNE) (Petersen and Doorn, 2004); an aldehyde that has become a major focus in the field due to the pathophysiological impact it exerts on both male (Aitken et al., 2012; Baker et al., 2015; Bromfield et al., 2017b) and female germ cells (Lord et al., 2015; Mihalas et al., 2017, 2018). Under normal physiological conditions, aldehyde-metabolizing enzymes function to detoxify 4HNE, and to abrogate its cellular accumulation, thus limiting its ability to propagate oxidative cellular damage (Hauck and Bernlohr, 2016). However, when challenged with either chronic or acute oxidative stress, 4HNE can accrue within the cellular environment, and, owing to its inherent stability and rapid diffusion, direct the formation of Michael adducts with nucleophilic sites in DNA, lipids and proteins (LoPachin et al., 2009). Curiously, despite 4HNE being capable of disseminating across large functional boundaries, not all elements of the cellular proteome display equivalent susceptibility to 4HNE carbonylation, with only a relatively small number of putative targets identified in spermatozoa to date (Aitken et al., 2012).

In seeking to reconcile the mechanisms underpinning the pathophysiology of 4HNE accumulation, recent studies have reported the application of affinity-based isolation techniques coupled with mass spectrometry to identify 4HNE adducted proteins harbored by oxidatively stressed human spermatozoa (Baker et al., 2015). Among the dominant 4HNE targets identified was A-kinase anchoring protein 4 (AKAP4). AKAP4 is encoded by a single gene located on the X chromosome and synthesized as a precursor pro-polypeptide before undergoing cleavage of the first 188 amino acids to release the mature AKAP4 protein (Carrera et al., 1996; Turner et al., 1998). AKAP4 is subsequently incorporated into the fibrous sheath (Brown et al., 2003); a unique cytoskeletal structure surrounding the axoneme and outer dense fibers that extends throughout the principal-piece of the sperm flagellum (Eddy et al., 2003). Aside from structural contributions to the fibrous sheath, AKAP4 has been implicated as a subcellular scaffold responsible for the tethering of cyclic AMP-dependent protein kinase (PKA), thereby compartmentalizing PKA within the immediate proximity of its enzymatic substrates (Eddy, 2007). Via the targeted positioning of PKA, AKAP4 exerts influence over the specificity of the signal transduction and metabolic processes that support sperm motility and capacitation (Luconi et al., 2011; Sergeant et al., 2019). The importance of AKAP4 in the organization and integrity of the fibrous sheath has been elegantly highlighted by gene ablation studies in which male mice lacking AKAP4 are rendered infertile due to lesions in the progressive motility profiles of their spermatozoa (Miki et al., 2002; Fang et al., 2019). Indeed, while the number of spermatozoa produced by Akap4 knockout animals remains unchanged, these cells display aberrant fibrous sheath development, a shortened flagella, and a substantially reduced abundance of signal transduction and glycolytic enzymes usually associated with the fibrous sheath (Miki et al., 2002).

These findings take on added significance in view of the dramatic under-representation of AKAP4 in the spermatozoa of infertile human patients (Moretti et al., 2007; Redgrove et al., 2012; Frapsauce et al., 2014). More recent work has also established positive correlations between the levels of AKAP4, and/or that of the proAKAP4 precursor molecule, with key sperm quality and fertility indicators in a number of livestock species (Peddinti et al., 2008; Blommaert et al., 2019; Sergeant et al., 2019). Taken together, these cross species analyses identify the potential use of proAKAP4 and AKAP4 as diagnostic biomarkers of overall semen quality (Sergeant et al., 2019). At present however, it remains uncertain what factor(s) contribute to the striking differences in proAKAP4 and AKAP4 levels documented in livestock (Blommaert et al., 2019) and human spermatozoa (Jumeau et al., 2018). Here, we sought to validate proAKAP4 and AKAP4 as targets for chemical alkylation by 4HNE, and to explore the consequences of 4HNE-mediated alkylation of proAKAP4 and AKAP4 during key phases of sperm development.

#### MATERIALS AND METHODS

fcell-07-00319 December 19, 2019 Time: 16:47 # 3

#### Ethics Statement

All experimental procedures involving animals were conducted with the approval of the University of Newcastle's Animal Care and Ethics Committee (ACEC) (approval numbers: A-2013– 322, A-2018-826). Experiments involving human spermatozoa were performed with semen samples obtained with informed written consent from a panel of healthy normozoospermic donors assembled for the Reproductive Science Group at the University of Newcastle. Volunteer involvement and all experimental procedures were performed in strict accordance with institutional ethics approvals granted by the University of Newcastle Human Research and Ethics Committee (approval number H-2013-0319).

#### Reagents

Unless specified, chemical reagents were obtained from Sigma-Aldrich (St. Louis, MO, United States) and were of research grade. Cell culture reagents were purchased from Sigma-Aldrich or Thermo Fisher Scientific (Waltham, MA, United States). The following primary antibodies were used to characterize proteins of interest: monoclonal anti-AKAP4 antibody clone 7E10 (4BDX-1602; 4BioDx, Lille, France), monoclonal anti-proAKAP4 antibody clone 6F12 (4BDX-1701; 4BioDx), rabbit polyclonal anti-4HNE (HNE11-S; Alpha Diagnostic International, San Antonio, TX, United States), rabbit polyclonal anti-androgen receptor (SAB4501575; Sigma-Aldrich), rabbit polyclonal anti-GAPDH antibodies (G9545; Sigma-Aldrich), monoclonal anti-phosphotyrosine (PT66) (P5872; Sigma-Aldrich), rabbit polyclonal anti-phospho (Ser/Thr) PKA substrate (9621; Cell Signaling, Danvers, MA, United States), and rabbit polyclonal anti-amyloid fibrils OC (ab2286; Merck Millipore, Kenilworth, NJ, United States). Appropriate horseradish peroxidase (HRP)-conjugated and Alexa Fluor-conjugated secondary antibodies were obtained from Sigma-Aldrich and Thermo Fisher Scientific, respectively. Bovine serum albumin (BSA) and 3-[(3-cholamidopropyl)dimethylammonio]-1 propanesulfonate (CHAPS) were obtained from Research Organics (Cleveland, OH, United States), Dulbecco's Modified Eagle Medium (DMEM) was purchased from Thermo Fisher Scientific, Tris was purchased from ICN Biochemicals (Castle Hill, NSW, Australia), nitrocellulose was purchased from GE Healthcare (Buckinghamshire, United Kingdom), Mowiol 4-88 was purchased from Calbiochem (La Jolla, CA, United States), and the paraformaldehyde used in this study was purchased from ProSciTech (Thuringowa, QLD, Australia).

#### Mouse Germ Cell Isolation

Swiss mice were obtained from a breeding colony held at the institutes' central animal house and maintained according to the recommendations prescribed by the ACEC. Mice were housed under a controlled lighting regimen (16L:8D) at 21–22◦C and were supplied with food and water ad libitum. Prior to dissection, animals were euthanized via CO<sup>2</sup> inhalation. Enriched populations of spermatocytes and spermatids were isolated from dissected adult mouse testes using density sedimentation at unit gravity as described previously (Nixon et al., 2014; Bromfield et al., 2017b). Briefly, testes were disassociated and tubules were sequentially digested with 0.5 mg/ml collagenase/DMEM and 0.5% (v/v) trypsin/EDTA to remove extra-tubular contents and interstitial cells. The remaining cells were loaded atop a 2–4% (w/v) BSA/DMEM gradient to separate male germ cell types according to their density. Consistent with previous data (Bromfield et al., 2019), this method resulted in the isolation of pachytene spermatocytes (>90% purity) and round spermatids (>85% purity) with minimal somatic cell contamination. Unless stated otherwise, spermatocytes and spermatids were pooled from 2 to 4 mice to achieve adequate cell numbers for subsequent analyses. Experiments were conducted on three independently pooled samples.

#### Preparation of Mouse and Human Spermatozoa

Mature spermatozoa were isolated from the cauda epididymides of adult mice by retrograde perfusion and subsequently allowed to disperse into Biggers, Whitten and Whittingham medium (BWW) as previously described (Walsh et al., 2008; Dun et al., 2012). By contrast, corpus spermatozoa were recovered by placing each sampled corpus segment into a 500 µl droplet of modified BWW. After making multiple incisions with a razor blade, the spermatozoa were gently washed into the medium via mild agitation. The resulting suspensions were next placed atop a 27% Percoll density gradient and subjected to centrifugation at 400 × g for 15 min at room temperature (RT). The pellet, consisting of an enriched population of >95% corpus spermatozoa was resuspended in fresh BWW and then re-centrifuged at 400 × g for 2 min at RT to again pellet the cells and to allow for the removal of excess Percoll (Zhou et al., 2018). These cell preparations were then pooled with those of the cauda epididymis and assessed for viability and motility as previously described (Zhou et al., 2018), prior to being allocated to appropriate treatment groups in preparation for further analysis. In all samples, >80% of the mouse spermatozoa were deemed viable and motile prior to treatment.

Human semen samples were collected following sexual abstinence of at least 2 days. Following collection, all samples were maintained at 37◦C and sample analysis was initiated after completion of liquefaction and within 1 h of ejaculation. Each sample was analyzed for total sperm count, motility, morphology and vitality as described previously (Walters et al., 2018a); with all samples used in this study exceeding WHO reference values of 58% live cells, 15 × 10<sup>6</sup> cells/ml, 1.5 ml ejaculate volume, 40% total motility and 4% normal morphology. After assessment, semen samples were fractionated over a discontinuous Percoll density gradient (comprising 40 and 80% Percoll suspensions) by centrifugation at 500 × g for 30 min at RT (Redgrove et al., 2011). This study was restricted to the use of good quality spermatozoa collected from the base of the 80% Percoll suspension as opposed to the poor quality cells partitioning at the 40/80% Percoll interface. After centrifugation, sperm pellets were resuspended in BWW and washed by centrifugation at

500 × g for 15 min at RT. Post-washing, the cells were again resuspended in fresh BWW prior to experimental use. A routine assessment of semen parameters was conducted for all donors in accordance with World Health Organization (WHO) criteria and corresponding to the checklist published by Bjorndahl et al. (2016). At least 100 cells were assessed for determination of cell motility, viability and morphology, with at least five microscope fields of view being examined for each count. Sperm morphology was assessed in accordance with WHO criteria using bright field microscopy (×400 magnification; Olympus CX40; Olympus Corporation, Tokyo, Japan). Sperm motility was assessed using phase contrast microscope optics (×400 magnification), with cells being classified as either motile (that is, sperm that displayed any form of motility, ranging from rapid progressive to nonprogressive) or immotile.

#### 4HNE Treatment

Oxidative stress was induced in germ cells and mature spermatozoa via direct 4HNE (Cayman chemicals, Ann Arbor, MI, United States) challenge, in accordance with our previously established protocols (Bromfield et al., 2015, 2017b; Walters et al., 2018a). For both germ cells and mature spermatozoa, 4HNE treatment regimens (consisting of 4HNE concentrations of either 50 or 100 µM and exposure periods of either 1 or 3 h at 37◦C) were selected based on our prior in vitro experimentation, in which these conditions were shown to robustly promote the loss and/or dysregulation of alternative 4HNE targeted proteins in male germ cells (Bromfield et al., 2015, 2017a). We also note that the concentrations of 4HNE used herein fall within the range normally associated with low levels of oxidative stress in vivo (Uchida, 2003; Chen and Niki, 2006), but are considered sub-lethal such that they are below those required to elicit significant elevation of apoptotic hallmarks (i.e., caspase activation and annexin V binding) or DNA fragmentation (i.e., TUNEL positivity) under comparable in vitro incubation periods to those assessed in our study (Aitken et al., 2012). Postincubation, residual 4HNE was removed via centrifugation at 500 × g for 3 min at RT, and resuspension in pre-warmed DMEM or BWW for germ cells and spermatozoa, respectively. Consistent with our previous studies (Bromfield et al., 2015, 2017a; Walters et al., 2018a), the use of these conditions resulted in the death of <20% of the cells under examination (**Supplementary Figure S1**). In accordance with previous studies (Aitken et al., 2012; Bromfield et al., 2017b, 2019), we also noted that these experimental treatments elicited robust levels of oxidative stress as measured by labeling of the target germ cell populations with MitoSOX Red (MSR) and dihydroethidium (DHE) probes (reference values: ∼40–45% human spermatozoa stained positive for MSR and DHE following 50 µM 4HNE treatment for 3 h). By contrast, equivalent analysis of untreated germ cells confirmed relatively low basal levels of oxidative stress (reference values: <10% human spermatozoa stained positive for MSR and DHE following 50 µM 4HNE treatment for 3 h). Where indicated, an additional control group was incorporated in which male germ cells were pretreated with reduced glutathione (GSH; 4 mM) for 10 min prior to the addition of 4HNE (100 µM). The cells were then co-incubated with both GSH and 4HNE throughout the exposure period (i.e., 3 h for round spermatids and 1 h for spermatozoa). Alternatively, following the completion of 4HNE exposure, the cells were washed twice with BWW (via centrifugation at 500 × g for 3 min at RT) to remove residual 4HNE before being resuspended in either DMEM (round spermatids) or BWW (spermatozoa) supplemented with reduced GSH (4 mM) and incubated for an additional 30 min. At the completion of either treatment regimen, gametes were washed twice in BWW via centrifugation at 500 × g for 3 min at RT before being processed in accordance with the relevant protocols described below.

#### Sodium Dodecyl Sulfate (SDS) Polyacrylamide Gel Electrophoresis and Immunoblotting

Following 4HNE treatment, cells were centrifuged at 500 × g for 3 min at RT, and the resulting pellets resuspended in SDS-based protein extraction buffer as previously described (Reid et al., 2012). Protein extracts were then boiled in the presence of NuPAGE LDS sample buffer (Thermo Fisher Scientific) containing 8% β-mercaptoethanol, and subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) using 4–12% Bis–Tris pre-cast gels (Thermo Fisher Scientific). Post-electrophoresis, separated proteins were transferred to nitrocellulose membranes using standard Western blotting techniques (Towbin et al., 1979). To detect proteins of interest, membranes were blocked in 3% BSA (w/v) in Tris-buffered saline supplemented with 0.1% Tween-20 (v/v; TBST, pH 7.4), and then probed with either anti-proAKAP4, anti-AKAP4, anti-4HNE, anti-phospho-PKA substrate, or anti-phosphotyrosine antibodies, as appropriate. All primary antibodies were diluted 1:1000 in TBST supplemented with 1% BSA and probing reactions were conducted overnight at 4◦C on a rotating platform. Membranes were next washed in three changes of TBST (10 min/wash) and thereafter, the appropriate secondary antibodies were applied to each membrane for 1 h at RT on a rotating platform. The membranes were again washed with three changes of TBST (10 min/wash), and labeled proteins were then visualized using an enhanced chemiluminescence detection kit according to the manufacturer's instructions (ECL plus, GE Healthcare). After visualization of the abundance of proteins of interest, all immunoblotted membranes were stripped prior to being re-incubated with anti-GAPDH antibodies (1:4000 dilution), and with the corresponding secondary antibody (washing was performed as outlined above), to demonstrate equivalent protein loading. Band density was quantified in each of three replicate blots using ImageJ software (version 1.48v; National Institute of Health, Bethesda, MD, United States) and the abundance of each protein of interest determined relative to GAPDH labeling intensity. Full length immunoblots of the antipro-AKAP4, AKAP4, and GAPDH antibodies used throughout this study are presented in **Supplementary Figure S3**.

#### Immunocytochemistry

Following 4HNE treatment, round spermatids, pachytene spermatocytes, and mature human and mouse spermatozoa, were

fixed in 4% paraformaldehyde, washed 3× with 0.05 M glycine in phosphate-buffered saline (PBS). Each cell preparation was then carefully pipetted onto a poly-L-lysine-coated glass coverslip and the cells allowed to settle via overnight incubation at 4◦C. Cells were permeabilized with 0.2% (v/v) Triton X-100, then placed in a humidified chamber and blocked with 3% (v/v) BSA/PBS for 1 h at RT. Each coverslip was then washed in PBS and incubated with either anti-proAKAP4, anti-AKAP4, anti-4HNE, anti-phospho-PKA substrate, or anti-phosphotyrosine primary antibodies as appropriate. All primary antibodies were diluted 1:100 in 1% (v/v) BSA/PBS before being applied to coverslips and incubated overnight at 4◦C. Coverslips were next washed with three changes of PBS (5 min/wash) before applying the appropriate Alex Fluor-conjugated secondary antibodies (diluted 1:100 in 1% (v/v) BSA/PBS and incubating for 1 h at RT. Coverslips were washed with changes of PBS (3 × 5 min/wash) before mounting in a solution consisting of 10% (v/v) Mowiol 4-88 (Calbiochem) supplemented with 30% (v/v) glycerol in 0.2 M Tris (pH 8.5) and 2.5% (v/v) 1,4-diazabicyclo-(2.2.2)-octane (DABCO). Labeled cells were examined using a Zeiss LSM510 laser scanning confocal microscope (Carl Zeiss).

#### Proximity Ligation Assay

Duolink in situ primary ligation assays (PLAs) were conducted in accordance with the manufacturers' instructions (Sigma-Aldrich), on fixed cells adhered to poly-L-lysine-coated coverslips. Briefly, samples were blocked in Duolink blocking solution and then incubated with appropriate primary antibody pairings (anti-proAKAP4, anti-AKAP4, anti-4HNE or anti-amyloid fibrils OC antibodies) overnight at 4◦C. Oligonucleotide-conjugated secondary antibodies (PLA probes; anti-rabbit plus, Duo92002; anti-mouse minus Duo82004; Duolink) were then applied for 1 h at 37◦C and ligation of the PLA probes was performed. The fluorescent signal generated when molecules are in close association (<40 nm) was visualized using fluorescence microscopy and the number of cells deemed positive for PLA staining was recorded for untreated and 4HNE-treated populations. The specificity of the PLA reaction was ensured by performing proximity ligation with antibodies to the target antigens combined with anti-androgen receptor antibodies with which they should not interact.

#### Pharmacological Inhibition of proAKAP4 and AKAP4 Proteolysis

To prevent 4HNE-mediated proteolysis of proAKAP4 and AKAP4 in male germ cells and mature spermatozoa, the cells were treated with several broad-spectrum pharmacological suppressors of proteolytic activity (Complete Mini Protease Inhibitor Cocktail; Roche), or with a selective inhibitor of proteasomal activity (MG132). Alternatively, cells were treated with a selective inhibitor of arachidonate 15-lipoxygenase (PD146176; Tocris Bioscience, Bristol, United Kingdom), an enzyme involved in the propagation of oxidative stress via the metabolism of polyunsaturated fatty acids to 4HNE (Walters et al., 2018b). The concentrations of each inhibitor (Complete Mini Protease Inhibitor Cocktail at 1× working concentration; MG132, 10–25 µM; PD146176, 1 µM), were selected based on published IC<sup>50</sup> values and from our previous use of these reagents to effectively block the loss of other 4HNE targeted proteins from the germ cell proteome (Bromfield et al., 2017b; Walters et al., 2018a). Germ cells and mature spermatozoa were pretreated with each inhibitor for 15 min prior to exposure to 4HNE (protease inhibitor cocktail and MG132) or H2O<sup>2</sup> (in the case of PD146176) and the inhibitor was retained for the duration of treatment thereafter to ensure adequate inhibition of each target. A DMSO vehicle control (1 µM) was also included in these experiments.

#### Functional Assessment of Human Spermatozoa

Computer Assisted Sperm Analysis (CASA): the movement characteristics of human spermatozoa were assessed using a Hamilton-Thorn motility analyzer (HTMA IVOS II; Hamilton-Thorn Research, Danvers, MA, United States). The settings for human spermatozoa were: 10× NH 160 mm objective, negative phase-contrast optics, recording rate 60 frames/s, minimum head brightness 171, minimum cell size 5 µm<sup>2</sup> , maximum head size 50 µm<sup>2</sup> and measured in a chamber of 20 µm depth. The criteria of sperm movement assessed were average path velocity (VAP), curvilinear velocity (VCL), straight line velocity (VSL), progressive motility, characterized by a VAP of >25 µm/s and a STR (straightness) of >80%.

Acrosome reaction: Following the induction of capacitation as previously described (Zhou et al., 2017), spermatozoa were induced to acrosome react by supplementation of media with 2.5 µM A23187 for 30 min. The spontaneous rates of acrosome loss were assessed via the inclusion of a capacitated sperm control group, which were prepared under identical incubation conditions with the exception that they did not receive an A23187 stimulus. At the completion of this induction period, the cells were then incubated in pre-warmed hypo-osmotic swelling media (HOS; 0.07% w/v sodium citrate; 1.3% w/v fructose) for another 30 min at 37◦C. After being fixed in 4% PFA, spermatozoa were aliquoted onto 12-well slides, air-dried and permeabilized with ice cold methanol for 10 min. Cells were then incubated with fluorescein isothiocyanate (FITC) conjugated PSA (Pisum sativum agglutinin) (1 µg/µl) at 37◦C for 15 min, and the acrosomal status of viable cells (possessing coiled tails as a result of incubation in HOS medium) were verified using fluorescence microscopy as previously described (Zhou et al., 2017).

#### Statistics

All experiments were replicated at least three times, with each biological replicate comprised of germ cells or mature spermatozoa from at least three mice, or in the case of human spermatozoa, three healthy normozoospermic individuals. Data are expressed as mean values ± SE. Experimental results were analyzed using two-tailed unpaired Student's t-tests (for two way comparison of 4HNE treatments to that of untreated controls) or by one-way analysis of variances (ANOVA) using Microsoft Excel (Version 14.0.0); post hoc comparison of group means was by

Fisher's PLSD (protected least significant difference). Differences were considered significant if a p < 0.05 was obtained.

#### RESULTS

#### ProAKAP4 and AKAP4 Are Targeted for 4HNE Adduction in Post-meiotic Male Germ Cells

Previous studies have established that several residues within the AKAP4 primary structure are vulnerable to 4HNE adduction in human spermatozoa. However, the implications of this form of chemical alkylation remain unknown. Here, we therefore, initially assessed the stability of both the AKAP4 and the precursor form, proAKAP4, in mouse germ cell populations treated with exogenous 4HNE (50 or 100 µM) for either 1 or 3 h at 37◦C (**Figure 1**). Consistent with our previous findings (Bromfield et al., 2017a, 2019), the modest level of oxidative stress generated under these exposure regimens led to <20% loss of germ cell viability (**Supplementary Figure S1**). However, the 1 and 3 h 4HNE treatments did significantly reduce the abundance of both proAKAP4 (**Figures 1A,B**) and AKAP4 (**Figures 1C,D**) in post-meiotic round spermatids (P < 0.05). Further, the degree of reduction was similar for both the 50 or 100 µM 4HNE treatments (**Figures 1E,F**). As expected, based on its expression profile (Johnson et al., 1997), the proAKAP4 precursor was not detected in equivalent lysates of untreated pachytene spermatocytes (**Figures 1A,B**), a finding that confirmed germ cell preparation purity. To account for the possibility of 4HNE-mediated epitope masking, recombinant AKAP4 protein was treated under identical conditions to those imposed on germ cells (i.e., 50 µM 4HNE for 1 h at 37◦C) prior to being prepared for immunoblotting. **Supplementary Figure S2** clearly demonstrates that 4HNE treated recombinant AKAP4 was detectable at an equivalent level of efficiency to the untreated control with both anti-proAKAP4 and anti-AKAP4 antibodies. Together, these data suggest that the demonstrable reduction in proAKAP4 and AKAP4 immunoreactivity in round spermatids challenged with 4HNE reflects a genuine reduction in the abundance of both precursor and mature forms of AKAP4. This result led us to next attempt to uncover the causative nature of the observed response.

Toward this goal, round spermatids were treated with 4HNE prior to assessing the cumulative levels of this aldehyde, and the extent of 4HNE co-localization with the proAKAP4 precursor; the isoform determined most affected by 4HNE treatment (**Figure 1E**). Consistent with the immunoblotting data, the accumulation of 4HNE throughout the cytosol of round spermatids was accompanied by a reciprocal reduction in the amount of immunoreactive proAKAP4 detected within these cells (**Figure 2A**). In addition, the residual proAKAP4 remaining in round spermatids appeared to co-localize with 4HNE labeling (**Figure 2A**). This co-localized accumulation profile for 4HNE and proAKAP4, was confirmed via use of the proximity ligation assay, which revealed intense labeling foci for the paired proAKAP4 and 4HNE antibodies throughout the round spermatid cytosol (**Figure 2B**). Notwithstanding the basal levels of 4HNE detected in untreated round spermatids via conventional immunolabeling (**Figure 2A**), this naïve cell population was characterized by very few PLA positive foci when assessed using the same antibody pairing (anti-proAKAP4 and anti-4HNE) (**Figure 2B**). The specificity of PLA labeling was confirmed via inclusion of two controls, namely; antibodies against the androgen receptor protein that is not known to interact with proAKAP4, as well as the use of a single antibody only (anti-proAKAP4) control. **Figure 2C** clearly shows that each control failed to generate a positive PLA signal (red fluorescence). This result confirms the close proximity of 4HNE to the proAKAP4 precursor in the cytosol of round spermatids.

The preceding data strongly suggests that elevated levels of intracellular 4HNE leads to proAKAP4 adduction, possibly destabilizing the structure of the precursor; a modification that would likely interfere with the formation of mature AKAP4. The ubiquitin-proteasome pathway has been previously implicated in the selective elimination of a variety of oxidatively damaged proteins (Davies, 2001; Shringarpure et al., 2001). We therefore, next assessed the contribution of the proteasomal pathway to the clearance of modified proAKAP4. For this purpose, round spermatids were co-incubated with 4HNE in tandem with MG132, a cell-permeable synthetic peptide that is widely used to repress proteasome activity. Unexpectedly however, MG132 coincubation failed to rescue 4HNE-mediated loss of proAKAP4 in round spermatids (**Figures 3A,B**). Similarly, co-incubation with a broad-spectrum protease inhibitor cocktail also had minimal effect on the abundance of proAKAP4 in 4HNE-treated round spermatids, when used either alone, or in tandem with MG132 (data not shown). This result led to the conclusion that 4HNEmediated loss of proAKAP4 in round spermatids is unlikely to involve conventional proteolytic degradation.

#### Acute 4HNE Exposure Reduces proAKAP4 and AKAP4 Abundance in Mouse and Human Spermatozoa

Given the pronounced 4HNE-mediated loss of proAKAP4 and AKAP4 in post-meiotic round spermatids, we next extended our analyses to assess the consequences of 4HNE treatment of mouse and human spermatozoa; terminally differentiated cells that have a reduced capacity to ameliorate the deleterious effects of oxidative insult compared to that of precursor germ cells. In mouse spermatozoa, 4HNE treatment reduced the abundance of both proAKAP4 and AKAP4 (**Figures 4A,D**). However, proAKAP4 appeared more sensitive in these cells, experiencing a highly significant ∼40% reduction (P < 0.01), compared to the untreated controls (**Figures 4A,D**). Consistent with our data on the round spermatid (**Figure 2**), the application of PLA revealed intense co-labeling of 4HNE with proAKAP4 and AKAP4 within the expected domains of the sperm flagellum (**Figure 4B**). More specifically, PLA fluorescence was restricted to the proximal portion of the sperm flagellum for the 4HNE and proAKAP4 antibody pairing, whilst the 4HNE and AKAP4 pairing allowed for the visualization of fluorescent foci throughout the entire principal piece (**Figure 4B**). Together, these data indicate that

residual proAKAP4 and AKAP4 proteins that remain within the flagellum of 4HNE treated mouse spermatozoa likely harbor substantial levels of aldehyde adducts. It is also important to note here that, as observed for 4HNE-treated round spermatids, co-incubation with the proteasome inhibitor, MG132, failed to rescue the loss of either proAKAP4 or AKAP4 in 4HNE-treated mature mouse spermatozoa (**Figures 4C,D**).

To determine if AKAP4 is similarly targeted for 4HNEmediated adduction in human spermatozoa, these cells were treated as per their mouse counterparts. Initially, proAKAP4 and AKAP4 abundance was assessed in the spermatozoa of three healthy normozoospermic donors, with each sample subjected to acute 4HNE treatment (50 or 100 µM) for 1 h at 37◦C. As shown in the representative immunoblots in **Figure 5A**, sperm from each donor responded to 4HNE challenge in an equivalent manner, that is; proAKAP4 and AKAP4 abundance was significantly reduced in a dose-dependent manner (p < 0.01). For the 100 µM 4HNE treatment, both the precursor and mature form of AKAP4 were reduced by as much as 70% compared to the untreated control samples (**Figure 5B**). Accordingly, immunocytochemistry revealed elevated levels of 4HNE and a concomitant reduction in proAKAP4 and AKAP4 abundance within the flagellum of each 4HNE-treated human sperm sample assessed (**Figures 6A,B**). Furthermore, PLA fluorescence was observed in the majority of these cells post-4HNE treatment when the 4HNE antibody was paired with either the proAKAP4 or AKAP4 antibody (**Figures 6C–F**). Although basal levels of PLA fluorescence were detected in untreated control spermatozoa, both the distribution and intensity of this labeling appeared equivalent to that detected in the irrelevant antibody control samples, suggesting that the detected signal was non-specific in nature (**Figures 6C,D**). Overall, these data are in general agreement with that generated for mouse spermatozoa (**Figure 4**), thus indicating cross-species conservation in terms of the vulnerability of both the precursor and mature form of AKAP4 to 4HNE insult as well as the fate of the alkylated proteins.

#### Consequences of 4HNE-Adduction of ProAKAP4 and AKAP4

To explore the causal nature of 4HNE-mediated attenuation of proAKAP4 and AKAP4 abundance, populations of round spermatids, mouse and human spermatozoa were treated with reduced glutathione (GSH) prior to, or after, implementing the 4HNE exposure. As anticipated based on the ability of GSH to directly alkylate and thereby reduce the bioavailability of 4HNE, pre-incubation of male gametes with this antioxidant conferred at least some degree of protection to proAKAP4 and AKAP4

proteins. Indeed, the abundance of both proteins detected in pre-GSH treated round spermatids was statistically indistinguishable from that of the untreated control samples. An equivalent trend was observed in mouse and human spermatozoa, although in both instances the abundance of proAKAP4 and AKAP4 proteins was still reduced in pre-GSH treated cells versus that of their untreated counterparts. In terms of the response to the opposing post-GSH treatment, this again varied depending on the cell type analyzed. Thus, in the case of round spermatids, which retain some capacity for protein synthesis, proAKAP4 and AKAP4 levels were recovered to levels that proved statistically similar to that of untreated controls (**Figures 7A,B**). However, the retrospective application of GSH failed to rescue either proAKAP4 or AKAP4 levels in 4HNE treated mouse or human spermatozoa (**Figures 7C–F**). These results prompted us to explore whether alternative strategies could be used to prevent 4HNE-mediated reduction of proAKAP4 and AKAP4 in mature spermatozoa. With this goal in mind, we next investigated whether the abundance of the two AKAP4 isoforms could be maintained via the administration of a pharmacological inhibitor that acts upstream of 4HNE production. Specifically, spermatozoa were subjected to oxidative stress in the form of an H2O<sup>2</sup> insult either alone, or in the presence of PD146176, a selective inhibitor of the lipoxygenase enzyme, arachidonate 15-lipoxygenase (ALOX15); an enzyme that drives a positive feedback loop of lipid peroxidation leading to amplification of 4HNE generation in male germ cells (Walters et al., 2018a). Accordingly, via the administration of H2O<sup>2</sup> alone, we were able to reproduce the previously observed reduction in the abundance of both proAKAP4 and AKAP4 from mouse and

human spermatozoa (**Figures 8A–D**). Further, in H2O<sup>2</sup> alone treated samples, proAKAP4 and AKAP4 were reduced to levels similar to those elicited by the direct application of 4HNE. Unexpectedly however, co-incubation of PD146176 together with H2O2, led to only a partial recovery of proAKAP4 and AKAP4 levels in mouse spermatozoa (**Figures 8A,C**), and proved completely ineffective in the rescue of proAKAP4 and AKAP4 in human spermatozoa (**Figures 8B,D**).

In view of these data, we next attempted to determine if 4HNE adduction can act as a catalyst to destabilize either the precursor or mature form of AKAP4, in order to drive the protein toward aggregate formation. Thus, PLA was again employed to assess the dual labeling of either the proAKAP4 or AKAP4 antibody together with an amyloid fibrils OC-specific antibody (**Figure 8**). The amyloid fibril OC antibody was selected for inclusion in this analysis based on its recognition of generic epitopes common to many amyloid fibrils and fibrillary oligomers, but not prefibrillar oligomers or natively folded precursors. As shown in **Figure 9A**, PLA fluorescence was observed for both antibody pairings throughout the cytosol of 4HNE-treated round spermatids, with the labeling appearing more intense at the higher 4HNE concentration of 100 µM (**Figure 9A**). PLA fluorescence, albeit less intense, was also observed for both antibody pairings within the flagellum of 4HNE-treated mouse (**Figure 9B**) and human spermatozoa (**Figure 9C**). In all instances, PLA labeling was attenuated via prior incubation of the germ cells with GSH before they received 4HNE-exposure (**Figures 9A–C**). Taken together, these data suggest that in male germ cells, 4HNE adduction can result in the aggregation of both the precursor and mature forms of AKAP4.

A potential consequence of 4HNE-mediated protein degradation and/or aggregation is dysregulation of the cAMP signaling framework assembled around the AKAP4 scaffold in mature spermatozoa. To address this possibility, human spermatozoa were subjected to 4HNE treatment before being driven to capacitate and assessed for the phosphorylation of PKA and tyrosine kinase substrates. Immunocytochemical analysis demonstrated that our 4HNE treatment regimen did not overtly impact the localization of proteins being phosphorylated during the capacitation of human spermatozoa, with prominent staining being evident throughout the sperm flagellum (**Figures 10A,B**). However, immunoblotting confirmed that 4HNE treatment did in fact lead to reduced levels of phosphorylation among a portion of the intermediary PKA targets, particularly the prominent bands resolving at ∼35 and 100 kDa (**Figure 10C**). In addition, 4HNE elicited a more pronounced, global reduction in the phosphorylation levels of downstream tyrosine kinase substrates (**Figure 10D**). In both instances, these effects were reduced by prior incubation of the

antibodies to confirm equivalent protein loading. (B) The adduction of proAKAP4 and AKAP4 with 4HNE was assessed through the application of a proximity ligation assay, whereby fixed spermatozoa were incubated with target primary antibodies (anti-proAKAP4 or anti-AKAP4 and anti-4HNE) or (C) negative controls (anti-proAKAP4 and anti-AKAP4 and anti-androgen receptor; anti-proAKAP4 alone) and oligonucleotide-conjugated secondary antibodies (PLA probes). PLA probes were then ligated and the signal was amplified. The red fluorescent signals generated when target antigens reside within <40 nm were visualized using fluorescence microscopy and representative images of each treatment group are presented. (D) Protein abundance was quantified by band densitometry and mean values (±SEM) are presented relative to the corresponding GAPDH control (n = 3). Differing lowercase letters denote statistical significance (p < 0.05) as determined by ANOVA. Scale bars = 5 µm. UT = untreated control.

human spermatozoa with reduced GSH before the addition of 4HNE (**Figures 10A–D**).

Notably, the deleterious impact of 4HNE extended to alterations in the motility profile of capacitated human spermatozoa (**Figures 11A–D**), as well as the ability of these cells to undergo acrosomal exocytosis (**Figure 11E**). Interestingly in this regard, in addition to a subtle, albeit significant reduction in the proportion of human spermatozoa exhibiting some form of motility (i.e., total motile sperm count; **Supplementary Figure S1E**), the assessment of key parameters of sperm movement, including average path velocity, curvilinear velocity, straight line velocity, and progressive motility, were all significantly attenuated after incubation with 4HNE

Differing lowercase letters denote statistical significance (p < 0.05) as determined by ANOVA. UT = untreated control.

(**Figures 11A–D**). Similarly, human spermatozoa experienced a significant, ∼25% reduction in their ability to complete a calcium ionophore (A23187) induced acrosome reaction (**Figure 11E**). Once again, these damaging effects of 4HNE were ameliorated via pre-treatment of spermatozoa with GSH.

## DISCUSSION

Driven by unprecedented rates of recourse to assisted reproductive technologies, there remains a pressing need to develop robust biomarkers capable of predicting fertilization success and discriminating the causative agents that can be targeted to prevent reproductive failure (Aitken et al., 2010). Consistent with these objectives, here, we examined the effects of sub-lethal doses of oxidative insult on the expression of proAKAP4 and of AKAP4; proteins that hold pivotal roles in the formation and structure of the sperm flagellar, in addition to coordination of capacitation-associated signaling (Luconi et al., 2011). We demonstrate that both proAKAP4 and AKAP4 are highly sensitive to oxidative challenge throughout the development of the male germ cell. Indeed, exposure to the lipid aldehyde, 4HNE, led to a substantive loss of both proAKAP4 and AKAP4 in round spermatids and mature spermatozoa. In addition, the residual proAKAP4 and AKAP4 that remained post-4HNE treatment were demonstrated to harbor the burden of 4HNE chemical alkylation; a modification that may perturb the stability of the protein native state, resulting in misfolding and nucleation of protein aggregates as has been shown in somatic cells (Maniti et al., 2015).

As the focus of this study, AKAP4 is a member of a large family of structurally diverse, but functionally conserved, anchoring proteins that coordinate the subcellular distribution of cAMP-mediated signaling networks (Colledge and Scott, 1999). Accordingly, a defining feature of the AKAP family is the presence of at least one PKA anchoring domain that serves to position the PKA holoenzyme at locations where it can rapidly respond to fluctuations in cAMP production. However, the multivalent nature of the AKAP scaffold has also implicated this protein in the physical tethering of several kinases, phosphatases, ion channels, and GTP binding proteins (Colledge and Scott, 1999). This property enables AKAPs to orchestrate the formation of macromolecular complexes compatible with the integration of cAMP and alternate intracellular signaling networks (Colledge and Scott, 1999). In the context of spermatozoa, the compartmentalization of cAMP signaling machinery coordinated by AKAP4 serves to regulate the specificity of signal transduction pathways responsible for the development, activation and maintenance of motility (Luconi et al., 2011; Sergeant et al., 2019). In accordance with this fundamental role, AKAP4 expression displays high evolutionary conservation, with Akap4 transcripts and/or immunoreactive orthologs of AKAP4 having been characterized in the testes and spermatozoa of species as phylogenetically diverse as eutherian mammals (humans, rodents, bovine, equine, porcine) (Johnson et al., 1997; Turner et al., 1998; Moss et al., 1999; Teijeiro and Marini, 2012; Blommaert et al., 2019; Sergeant et al., 2019), marsupials (tammar wallaby, opossum) (Hu et al., 2009), monotremes (platypus) (Hu et al., 2009), and reptiles (crocodile) (Nixon et al., 2019).

In the mouse testes, AKAP4 is synthesized as a precursor of ∼100 kDa (proAKAP4) during the post-meiotic phase of spermatogenesis, with transcripts first detected in early stage round spermatids (Johnson et al., 1997; Brown et al., 2003). Following synthesis, proAKAP4 is processed via the proteolytic cleavage of an N-terminal prodomain, yielding the mature AKAP4 protein (∼82 kDa) (Carrera et al., 1994). AKAP4 is subsequently transported to the principal piece of developing flagellum to become the most abundant structural element of the nascent fibrous sheath; accounting for as much 50% of the protein within this specialized domain. The fate of proAKAP4 is somewhat less certain, with emerging evidence supporting interspecies differences in both the extent of its processing and distribution within the mature sperm flagellum. Thus,

proAKAP4 has been reported to localize along the entire length of the principal piece of mouse testicular spermatozoa before becoming restricted to the proximal portion of the flagellum in mature cauda epididymal spermatozoa (Johnson et al., 1997); suggestive of an additional wave of proAKAP4 processing during epididymal sperm maturation. By contrast, and in agreement with our own data, proAKAP4 is retained within the flagellum of mature human spermatozoa, where it co-localizes with AKAP4 throughout the fibrous sheath (Turner et al., 1998; Jumeau et al., 2018). Notably however, striking differences in proAKAP4 abundance have been documented in the semen of normozoospermic individuals, wherein the quantity of the protein is positively correlated with sperm motility; albeit in the high-quality cells recovered by density gradient centrifugation (Jumeau et al., 2018). Whilst these data invite speculation that the extent of proAKAP4 processing may reflect the integrity of post-testicular sperm maturation, and hence contribute to

differences in the motility profile of human spermatozoa, there is currently no causal evidence to substantiate such an association. Rather, it has been postulated that proAKAP4 may serve as a "reservoir" that can be activated to generate mature AKAP4 and thus rescue sperm motility, if and when required (Jumeau et al., 2018; Sergeant et al., 2019).

Our collective findings confirm previous work in identifying AKAP4 as a primary target of 4HNE adduction in mature human spermatozoa. Indeed, studies by Baker et al. (2015) identified at least three AKAP4 peptides that harbor either one or two 4HNE modified residues after exogenous treatment of spermatozoa with the aldehyde. This study not only confirmed a significant 11-fold enrichment in modified AKAP4 peptides isolated after 4HNE treatment, but also revealed basal levels of endogenous 4HNE modification of AKAP4 in untreated spermatozoa (Baker et al., 2015); data that further reinforces the inherent sensitivity of AKAP4 to chemical alkylation

by ANOVA. Scale bars = 5 µm. UT = untreated control.

completion of 4HNE exposure, the cells were washed to remove residual 4HNE before resuspended in media supplemented with GSH (4 mM) and incubated for an additional 30 min (Post-GSH). At the completion of either treatment regimen, gametes were washed before being processed for immunoblotting with either anti-proAKAP4 or anti-AKAP4 antibodies. Each blot was stripped and re-probed with anti-GAPDH antibodies to confirm equivalent protein loading. This analysis was performed in triplicate and representative blots are depicted. (B,D,F) Protein abundance was quantified by band densitometry and mean values (±SEM) are presented relative to the corresponding GAPDH control (n = 3). Differing lowercase letters denote statistical significance (p < 0.05) as determined by ANOVA. UT = untreated control.

reactions. Though the full extent of AKAP4 damage elicited by insertion of bulky 4HNE (C-9) carbonyl adducts has yet to be investigated, our in silico modeling of 4HNE modified residues (K279, K331, C<sup>566</sup> and C570), indicates that these lie outside of the prodomain (M1-N188) and those regions implicated in the binding of PKA regulatory subunits (i.e., F219– A<sup>232</sup> and I336–K345) (Miki and Eddy, 1998, 1999). The 4HNE modified residues do however, reside close to several putative phosphorylation sites and include cysteine residues that may be involved in stabilization of AKAP4 tertiary structure. By analogy with other 4HNE targets, it is reasonable to suspect that these modifications could elicit protein mis-folding, poor substrate recognition, and/or degradation of the protein itself (Carbone et al., 2004a,b); lesions that likely contribute to the dysregulation of sperm motility witnessed in this study as well as that commonly reported in cells burdened by excessive ROS production (Aitken et al., 2010).

In extrapolating these data beyond motility, the importance of the signaling framework coordinated by AKAP4 (Luconi et al., 2011) raises the question of whether cAMP-responsive events associated with sperm capacitation are similarly affected by 4HNE adduction. In addition to the supporting data presented here on phosphorylation profiles and acrosome reaction rates, previous studies have also shown that exogenous 4HNE administration does compromise important correlates of the capacitation cascade including global increases in tyrosine phosphorylation (Baker et al., 2015) and downstream functional endpoints such as zona pellucida adhesion (Bromfield et al., 2015). However, as an important caveat in the interpretation of these data, we note that alternative elements of the

motility apparatus (dynein, outer dense fiber protein 1) (Baker et al., 2015), capacitation signaling (PKA) (Baker et al., 2015), and membrane remodeling proteins (heat shock protein A2) (Bromfield et al., 2015) have been validated as primary targets of 4HNE-mediated modification. Such pervasive broad-spectrum effects make it extremely challenging to disentangle whether individual sperm proteins such as AKAP4 play a dominant role in the pathophysiological responses to 4HNE.

Of concern, these pleiotropic effects are elicited at concentrations of 4HNE that are well within the range attained under conditions of oxidative stress, which can reach as high as 5 mM (Uchida, 2003). They also proved refractory to a lipoxygenase inhibition strategy designed to disrupt the positive feedback loop of lipid peroxidation and hence prevent the induction of redox cycling cascades. In somatic and male germ cells alike, 4HNE adducted proteins are commonly targeted for proteolysis in order to mitigate the risk they pose to cellular homeostasis (Grune et al., 1995; Shang et al., 2001; Carbone et al., 2004b; Bromfield et al., 2017a). Here, we confirm that, despite differences in their solubility and putative interaction networks (Nipper et al., 2006), the innate stability of both proAKAP4 and AKAP4 is also significantly impacted by 4HNE adduction. Curiously however, our data do not support the mobilization of conventional proteolytic degradation pathways in the clearance of proAKAP4 and AKAP4, neither of which were rescued by pharmacological interventions intended to suppress proteasomal activity. Similarly, incubation of mature spermatozoa with GSH post-4HNE treatment also failed to rescue proAKAP4 and AKAP4 levels. A possible explanation for this response lies in the chemistry of 4HNE, which has the potential to form either Michael addition to thiol or amino compounds (via the C3 of the C2=C3 double bond) or Schiff bases (between the C1 carbonyl group and primary amines) (Dalleau et al., 2013). Notably, the kinetics of the Schiff base formation are inherently slow and reversible, whereas those of the Michael-adducts are more rapid and stable; hence Michael-adducts predominate (Dalleau et al., 2013). Although retro-Michael cleavage can occur, leading to resolution of 4HNE adducts, this is by no means a universal phenomenon and is instead heavily influenced by the context of the cellular environment in which this response is measured (Schaur et al., 2015). Irrespective, at present we remain uncertain what mechanism(s) may account for the loss of AKAP4 expression, however, we do not consider this an artifact caused by 4HNE adducts masking antibody recognition

FIGURE 9 | Assessment of the aggregation potential of 4HNE modified proAKAP4 and AKAP4 in male germ cells. Populations of (A) round spermatids, (B) mouse spermatozoa, and (C) human spermatozoa were incubated with 4HNE (50 or 100 µM) for 1 h to induce oxidative stress before being fixed with paraformaldehyde. The propensity of 4HNE adducted proAKAP4 and AKAP4 to form aggregates was assessed through the application of PLA, whereby fixed spermatozoa were incubated with target primary antibodies (anti-proAKAP4 or anti-AKAP4 and anti-amyloid fibrils OC antibodies) and oligonucleotide-conjugated secondary antibodies (PLA probes). In addition to an untreated control (UT), gametes were also pre-treated with reduced glutathione (GSH; 4 mM) for 10 min prior to the addition of 4HNE (100 µM). The cells were then co-incubated with both GSH and 4HNE throughout the exposure period (Pre-GSH). PLA fluorescent signals were visualized using fluorescence microscopy and representative images of each treatment group are presented. This experiment was replicated three times and representative images are depicted. Scale bars = 10 µm.

FIGURE 10 | 4HNE adduction of proAKAP4 and AKAP4 attenuates capacitation-associated signaling in human spermatozoa. Human spermatozoa treated with 4HNE (100 µM for 1 h at 37◦C) were induced to capacitate (Capacitated) or held in a non-capacitated state (Non-cap) using standard protocols (Mitchell et al., 2007). Spermatozoa were subsequently prepared for either (A,B) immunocytochemistry or (C,D) immunoblotting with anti-phospho-PKA substrate or anti-phosphotyrosine antibodies. A densitometric trace of representative immunoblots is included to highlight changes in phospho-labeling in untreated (UT; black trace), GSH pre-treated (Pre-GSH; blue trace) and 4HNE treated (4HNE; red trace) spermatozoa. This experiment was replicated with samples from three separate donors and representative images are immunoblots are presented. Scale bars = 10 µm.

FIGURE 11 | 4HNE negatively impacts the motility profile and ability of human spermatozoa to complete an acrosome reaction. (A–D) Human spermatozoa treated with 4HNE (100 µM for 1 h at 37◦C) were induced to capacitate (Capacitated) or held in a non-capacitated state (Non-cap) using standard protocols (Mitchell et al., 2007). Computer Assisted Sperm Analysis (CASA) was subsequently used to objectively track the movement characteristics of human spermatozoa and the assessment criteria of (A) progressive motility, (B) curvilinear velocity (VCL), (C) straight line velocity (VSL), and (D) average path velocity (VAP) are presented. In addition to an untreated control (UT), spermatozoa were also pre-treated with reduced glutathione (GSH; 4 mM) for 10 min prior to the addition of 4HNE and the cells were then co-incubated with both GSH and 4HNE throughout the exposure period (Pre-GSH). (E) Alternatively, the spermatozoa were assessed for their competence to complete a calcium ionophore (A23187) induced acrosome reaction, via staining with fluorescently labeled PSA as previously described (Zhou et al., 2017). The spontaneous rates of acrosome loss were assessed via the inclusion of a vehicle control group (DMSO). These experiments were replicated with samples from three separate donors and representative images are and mean values (±SEM) are presented. Differing lowercase letters denote statistical significance (p < 0.05) as determined by ANOVA.

motifs. Indeed, we were able to demonstrate anti-proAKAP4 and anti-AKAP4 antibody recognition of 4HNE-treated recombinant AKAP4 protein. Moreover, these antibodies also readily labeled the residual proAKAP4 and AKAP4 in situ post-4HNE treatment, a setting in which the proteins harbored signatures of 4HNE adduction and aggregation. These latter findings are supported by independent evidence that heavily oxidized proteins resisting degradation have a propensity to crosslink and aggregate (Castro et al., 2017). Indeed, evidence from the somatic cell literature, has established that 4HNE oxidative modifications can inactivate target proteins by rendering them prone to formation of oligomers; both in solution and in the intracellular environment (Maniti et al., 2015). Accordingly, we have also shown that 4HNE treatment can result in significant accumulation of cytosolic protein deposits in the male germline (Cafe et al., 2019). We therefore, hypothesize that 4HNE induced aggregation of the proAKAP4 and AKAP4 proteins could reduce their activity and the efficacy with which they are recovered from cell lysates; although this prospect awaits further investigation.

#### CONCLUSION

Here, we have confirmed the vulnerability of both proAKAP4 and AKAP4 to 4HNE modification throughout male germ cell

development. Such modification leads to a range of adverse sequela that not only affect proAKAP4 and AKAP4 stability, but also compromises their functions in mature spermatozoa. These findings provide a physiological explanation for the loss of sperm motility and capacitation competence commonly encountered in response to oxidative stress and raise the prospect that such defects may help reduce the likelihood of sperm harboring oxidative DNA lesions from participating in fertilization. In this sense, oxidative damage of both proAKAP4 and AKAP4 may be considered a protective mechanism that limits the transmission of an altered male genome to the next generation. In any event, these data add to a growing body of literature emphasizing the need for novel therapeutic interventions to alleviate the burden of oxidative-stress mediated dysfunction in the male germline.

#### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by The University of Newcastle Human Ethics Committee. All donors provided their written informed consent to participate in this study. The animal study was reviewed and approved by The University of Newcastle Animal Care and Ethics Committee.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

BN conceived the study and wrote the first draft of the manuscript. IB, SC, AA, and NT were responsible for the study execution and data analysis. MD, NS, AE, TL, GD, MDD, and EB contributed to the conception and design of the study, and participated in the data analysis. All authors contributed to the manuscript revision, read, and approved the submitted version.

#### FUNDING

This project was supported by a National Health and Medical Research Council of Australia (NHMRC) Project Grant (APP1163319) awarded to BN and EB. SC and NT are the recipients of University of Newcastle Postgraduate Research Scholarships. EB is the recipient of a NHMRC CJ Martin Early Career Fellowship. BN is the recipient of a NHMRC Senior Research Fellowship. MDD is the recipient of a Cancer Institute NSW Early Career Fellowship.

#### ACKNOWLEDGMENTS

The authors gratefully acknowledge the support of the University of Newcastle, Priority Research Centre for Reproductive Science.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00319/ full#supplementary-material

anchor protein 4 (AKAP4) and its precursor (proAKAP4) in equine semen: promising marker correlated to the total and progressive motility in thawed spermatozoa. Theriogenology 131, 52–60. doi: 10.1016/j.theriogenology.2019. 03.011



**Conflict of Interest:** NS and MD are co-founders of SPQI – 4BioDx (Lille, France), a commercial company that markets the proAKAP4 and AKAP4 reagents used in this study.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Nixon, Bernstein, Cafe, Delehedde, Sergeant, Anderson, Trigg, Eamens, Lord, Dun, De Iuliis and Bromfield. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Amino Acids of Seminal Plasma Associated With Freezability of Bull Sperm

Muhammet Rasit Ugur<sup>1</sup> , Thu Dinh<sup>1</sup> , Mustafa Hitit1,2, Abdullah Kaya<sup>3</sup> , Einko Topper<sup>4</sup> , Bradley Didion<sup>4</sup> and Erdogan Memili<sup>1</sup> \*

<sup>1</sup> Department of Animal and Dairy Sciences, Mississippi State University, Starkville, MS, United States, <sup>2</sup> Department of Animal Genetics, Kastamonu University, Kastamonu, Turkey, <sup>3</sup> Department of Reproduction and Artificial Insemination, Selçuk University, Konya, Turkey, <sup>4</sup> Alta Genetics, Inc., Watertown, WI, United States

#### Edited by:

Xin Zhiguo Li, University of Rochester, United States

#### Reviewed by:

Marc Yeste, University of Girona, Spain Arumugam Kumaresan, National Dairy Research Institute (ICAR), India

\*Correspondence:

Erdogan Memili em149@ads.msstate.edu; em149@msstate.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 01 July 2019 Accepted: 03 December 2019 Published: 14 January 2020

#### Citation:

Ugur MR, Dinh T, Hitit M, Kaya A, Topper E, Didion B and Memili E (2020) Amino Acids of Seminal Plasma Associated With Freezability of Bull Sperm. Front. Cell Dev. Biol. 7:347. doi: 10.3389/fcell.2019.00347 Sperm cryopreservation is an important technique for fertility management, but postthaw viability of sperm differs among breeding bulls. With metabolites being the end products of various metabolic pathways, the contributions of seminal plasma metabolites to sperm cryopreservation are still unknown. These gaps in the knowledge base are concerning because they prevent advances in the fundamental science of cryobiology and improvement of bull fertility. The objective of this study was to test the hypothesis that seminal plasma amino acids are associated with freezability of bull sperm. To accomplish this objective, amino acid concentrations in seminal plasma from seven bulls of good freezability (GF) and six bulls of poor freezability (PF) were quantified using gas chromatography–mass spectrometry (GC–MS). Multivariate and univariate analyses were performed to identify potential freezability biomarkers. Pathways and networks analyses of identified amino acids were performed using bioinformatic tools. By analyzing and interpreting the results we demonstrated that glutamic acid was the most abundant amino acid in bull seminal plasma with average concentration of 3,366 ± 547.3 nM, which accounts for about 53% of total amino acids. The other most predominant amino acids were alanine, glycine, and aspartic acid with the mean concentrations of 1,053 ± 187.9, 429.8 ± 57.94, and 427 ± 101.3 nM. Pearson's correlation analysis suggested that phenylalanine concentration was significantly associated with post-thaw viability (r = 0.57, P-value = 0.043). Significant correlations were also found among other amino acids. In addition, partial least squares-discriminant analysis (PLS-DA) bi-plot indicated a distinct separation between GF and PF groups. Phenylalanine had the highest VIP score and was more abundant in the GF groups than in the PF groups. Moreover, pathway and network analysis indicated that phenylalanine contributes to oxidoreductase and antioxidant reactions. Although univariate analysis did not yield significant differences in amino acid concentration between the two groups, these findings are significant that they indicate the potentially important roles of amino acids in seminal plasma, thereby building a foundation for the fundamental science of cryobiology and reproductive biotechnology.

Keywords: amino acids, seminal plasma, freezability, bull sperm, metabolomics

## INTRODUCTION

fcell-07-00347 December 26, 2019 Time: 16:35 # 2

There is an urgent need for more efficient, sustainable, and profitable cattle farming to feed the ever-increasing world population. Artificial insemination (AI) using cryopreserved sperm is a significant tool for the agri-food industry to improve modern animal production. The first attempt on sperm cryopreservation was made in 1776 (Royere et al., 1996); since then significant progress has been made using various cryoprotective agents and protocols during the last two centuries. Such progress, however, has not yet achieved the desired level of success because post-thaw survivability of sperm cells is disappointingly low, <50%, despite the best effort put forward in developing preservation techniques (Nijs et al., 2009). During cryopreservation, sperm cells undergo cellular and molecular changes, among which are membrane damage, oxidative stress, DNA fragmentation, reduced mRNA–protein interactions, as well as epigenetic modifications (O'Connell et al., 2002; Flores et al., 2011; Valcarce et al., 2013). Such modifications have detrimental effects on sperm physiology and thus on fertility.

Bovine seminal plasma is composed of secretions from testis, epididymis, and accessory sex glands. Such mixture contains proteins, ions, and metabolites including amino acids, lipids, monosaccharides, nucleosides, minerals, electrolytes, and steroid hormones (Egea et al., 2014; Cheng et al., 2015). As metabolites are the end-products of metabolic pathways, they play significant roles in sperm physiology such as energy metabolism, motility, and regulation of metabolic activities (Bieniek et al., 2016). While some components of seminal plasma have positive influences on sperm cryotolerance, others have detrimental effects (Yeste, 2016; Recuero et al., 2019). Regardless, metabolites in seminal plasma can be used to estimate bull fertility and sperm freezability. Hamamah et al. (1993) analyzed seminal plasma from fertile and infertile men using <sup>1</sup>H nuclear magnetic resonance (NMR) spectra, and found significant differences in concentrations of glycerylphosphorylcholine citrate (GPC), and lactate between azoospermic and oligoasthenozoospermic patients. Lin et al. (2009) characterized the metabolite profiles of primate sperm to investigate the association between metabolism and energy supply. The association between glycolytic substrates and energy production, which is essential for motility, was determined using metabolomics approach in mouse spermatozoa (Goodson et al., 2012). More recently, using both NMR and gas chromatography– mass spectrometry (GC–MS), total of 96 metabolites and more than 10 biological pathways were identified in human sperm (Paiva et al., 2015).

Free amino acids of seminal plasma have various functions including reducing free radicals, protecting cells against denaturation, and providing oxidizable substrate to spermatozoa (Mann and Lutwak-Mann, 1981). However, identities and roles of seminal plasma amino acids during cryopreservation are not fully understood. Alanine, glycine, glutamine, histidine, and proline have been used as cryoprotectant agents for various species as they either inhibit lipid peroxidation or modulate osmotic mechanism (Heber et al., 1971; Renard et al., 1996; Trimeche et al., 1999; Jaiswal and Eisenbach, 2002; Dvoˇráková et al., 2005; Sangeeta et al., 2015). In addition to stabilizing proteins, amino acids also possess antioxidant properties to protect sperm cells from cold shock (Atessahin et al., 2008). For example, proline improves motility and protects sperm cells against damages caused by free radicals by stabilizing the membrane structure and function during the freezing (Rudolph et al., 1986; Smirnoff and Cumbes, 1989). Additionally, alanine and glutamine also affect the motility and viability of the sperm (Koskinen et al., 1989; Khlifaoui et al., 2005; Amirat-Briand et al., 2009) by to some extent improving the cryoprotective effects of glycerol.

Recently, we have identified 63 seminal plasma metabolites of which 21 were amino acids from bulls with different field fertility scores (Velho et al., 2018) demonstrating the importance of metabolite profiles between low and high fertility bulls. Seminal plasma addition before freezing also influenced on post-thaw bull sperm kinematics (Nongbua et al., 2018). To investigate further the impacts of seminal plasma composition on sperm cells, in this study we ascertained the relationship between freezability and amino acids in bull seminal plasma.

#### MATERIALS AND METHODS

### Semen Collection and Determination of Bull Semen Freezability

Seminal plasma samples from 13 bulls with various freezability and semen freezability data were provided by a commercial breeding company (Alta Genetics Inc., Watertown, WI, United States). The bulls were housed in the same nutrition and management environment to prevent sample variation. Semen was collected using artificial vagina and protease inhibitor was added immediately. Semen was then centrifuged at 800 × g for 15 min to separate the seminal plasma and sperm. This seminal plasma was transferred into sterile microcentrifuge tubes and centrifuged again at 800 × g for 15 min to completely eliminate sperm in the sample. Following this second centrifugation, seminal plasma was transferred into new tubes and shipped to Mississippi State University (MSU) in a liquid nitrogen tank.

Bull semen was extended with commercial egg-yolk–trisbased extender, and then frozen at Alta Genetics using standard protocols (Pace et al., 1981). Briefly, fresh semen was collected from bulls via artificial vagina, and semen was evaluated for concentration, volume, color, and motility. Then, semen was extended with one-step egg-yolk–tris–glycerol extender. The extender included 20% egg-yolk and 6% glycerol. This is called initial extension which includes fourfold dilution with extender at 32◦C. The extended semen was cooled down to 5 ◦C within 90 min. Then, the remaining extender was added at 5◦C to complete extension, and packaged into quarter cc straws (250 µl) and let semen equilibrate for 3–4 h. Following the equilibration process, straws were frozen using automated freezer machine. The freezing was completed within 14 min (temperature from 5◦C to −196◦C), and stored in a liquid nitrogen tank.

Post-thaw sperm viability was assessed using flow cytometry (CyFlow SL, Partec, Germany). Fluorescent stain combinations of SYBR-14 with propidium iodide (SYBR-14/PI, Live/Dead Sperm Viability Kit L-7011, Thermo Fisher Scientific) were used as described previously (Garner et al., 1994; Nagy et al., 2004). Membrane integrity of 10,000 sperm cells from each collection was measured with the highest accuracy and objectivity. We verified that biological sample preparations, instrument configurations, and data analysis were compliant with the recommendations set by the International Society for Advancement of Cytometry on the minimum information necessary. The CyFlow SL (Partec, Münster, Germany) instrument equipped with 488 nm blue state laser allowed excitation of SYBR14 and PI to measure sperm viability. It was also fully equipped with five parameters: FSC, SSC, red, green, and orange/yellow colors. With the Partec FloMax software, the instrument allowed a real-time data acquisition, data display, and data evaluation.

The quality control measures and repeatability of flow cytometric sperm viability analysis were routinely verified by control samples which consisted of positive (100% dead sperm) and negative control (100% live sperm) of standard biological samples and their mixture of different ratios (100/100, 75/25/50/50, 25/75, and 100/100% dead and live sperm combinations). Another quality measure we used was the control of reagents (SYBR-14 and PI). The reagents and biological standards were used to calibrate the instrument settings and data processing. In the calibration, non-sperm particles were gated out and not included in the calculations. Partially stained (green and red) moribund sperm were considered as dead in the analysis. The percentage of live (green) sperm is used as a measure of sperm freezability parameter, the percentage of sperm that maintained membrane integrity during freeze–thawing process. The following formula was used to count the percentage of viable sperm: The% Viable sperm = The number of viable sperm/Total sperm (viable + dead + moribund) × 100.

Collectively, a unique freezability phenotype was generated to characterize variation among bulls for their lifetime postthaw viability of sperm. For this particular research, we used post-thaw viability data generated over 8 years period (between 2008 and 2016). The database included 100,448 ejaculates from 860 Holstein bulls that were collected at least 20 different times in approximately 3 months period. The average and standard deviation of post-thaw viability for individual bulls were calculated, and then bulls were ranked based on average postthaw semen viability. The threshold was the population average which consisted of the 100,448 ejaculates from 860 Holstein bulls. The average post-thaw viabilities of all bulls ranged from 33.03 to 67.3% (population average 54.7 ± 5.4%). The bulls were then classified as GF and PF based on average postthaw viability score and the differences from the population average. The population average was our threshold to classify GF and PF groups. Bulls with greater sperm post-thaw viability than population average grouped as GF while those lower than average were considered as PF. Total of 13 bulls were selected with high confidence among 860 bulls for the presented study (**Table 1**).

TABLE 1 | Semen freezability phenotypes of the Holstein bulls used for GC–MS analysis: (A) Bulls 1–7 were defined as good freezability (GF) and Bulls 8–13 were grouped as poor freezability (PF) and (B) Percent differences of good and poor freezing phenotypes from the population average.


(B)

Bulls were classified as good freezability and poor freezability based on average post-thaw viability scores and the percent differences from the population average (P < 0.001).

#### Sample Preparation for Gas Chromatography–Mass Spectrometry Analysis

The amino acid analyses of seminal plasma from 13 bulls with various freezability were performed using EZfaast Amino Acid kit (©Phenomenex Inc., Torrance, CA, United States) as previously described by Kaspar et al. (2008). All samples were prepared and analyzed according to protocol that provided by Phenomenex Inc., Torrance, CA, United States. Briefly, 10 µl of seminal plasma and 25 µl of internal standard solution (norvaline 0.02 nM and N-propanol 10%) were pipetted into glass sample preparation vials. Solution in the sample preparation vial was passed through the sorbent tip using a syringe. Then, 200 µl of N-propanol

Ugur et al. Amino Acids and Sperm Freezability

was pipetted into the same vial and passed through the sorbent tip and into syringe barrel. Drained liquid from sorbent tip was discarded. One hundred and twenty microliters of sodium hydroxide and 80 µl of N-propanol were pipetted into same glass vial, and the particles inside the sorbent tip were ejected into solution. A volume of 50 µl of chloroform–propyl chloroformate, and 100 µl of iso-octane were transferred to the tube and the resulted mixture was vortexed for 1 min after each adding. Transparent part of the (upper) organic layer transferred into autosampler vial, and evaporation of the solvent was achieved using a TurboVap <sup>R</sup> LV evaporator (Biotage, Charlotte, NC, United States) with a gentle stream of nitrogen at 30◦C. The extract was then suspended in 50 µl of solution containing isooctane (80%) and chloroform (20%) and transferred to an amber glass vial having a fixed insert (Agilent Technologies, Santa Clara, CA, United States) for the analysis using GC–MS.

## Amino Acid Analysis Using Gas Chromatography–Mass Spectrometry

We used recommended GC-MS parameters to analyze the seminal plasma amino acids and the reference standards in an Agilent 7890A GC System that was coupled to an Agilent 5975C inert XL MSD with triple-axis mass detector, an Agilent 7693 Series Autosampler, and a capillary column (ZebronTM EZ-AAA 10 m × 0.25 mm; ©Phenomenex, Santa Clara, CA, United States). The derivatized mixture (1.5 µl) was injected into the inlet that was heated at 250◦C with 1:15 split ratio. Following the injection of the sample at 3 ml/min, standard septum purge was performed using helium carrier gas at 1 ml/min constant flow rate. Auxiliary, ion source, and quadrupole were heated at 310, 240, and 180◦C, respectively. The oven was programmed initially at 110◦C, and ramped up to 320◦C within 11 min. The solvent delay time was at 1.30 min. The MS was operated in selected ion monitoring (SIM) mode and appropriate ion sets were selected. All amino acids were identified based on retention times, target and qualifier ions in comparison with authentic standards supplied by ©Phenomenex Inc., Torrance, CA, United States (**Table 2**). For calibration, increasing volumes of the diluted standards (0, 5, 10, 40, 80, 160, and 200 nmol/ml) were as described above. Abundances of the target ions of amino acids were divided by abundance of target ion of the internal standard (norvaline) and the unitless ratios were used to calculate amino acid concentrations using internal standard calibration.

#### Statistical Analysis

The associations between freezability of sperm and concentration of seminal plasma amino acids were assessed using both univariate and multivariate approaches. For univariate approach, a generalized linear mixed model was used to determine the statistical significance between GF and PF groups. The variance was estimated by the GLIMMIX procedure of SAS 9.4 (SAS Institute Inc., Cary, NC, United States). The Kenward–Roger approximation was used to calculate the degree of freedom in case of heterogeneous variances. In addition, correlations among seminal plasma amino acids and correlations between seminal plasma amino acid concentrations and freezability TABLE 2 | Selected ions and retention times for the SIM analysis of 22 amino acids, dipeptides and internal standard (norvaline).


All amino acids were identified based on retention time, target and qualifier ions.

scores were determined using Pearson's correlation (Xia and Wishart, 2011). For multivariate analyses, MetaboAnalyst 3.0<sup>1</sup> (Xia et al., 2015) was used. For each variable, an observation was subtracted from the overall mean and the difference was divided by the standard deviation. This scaling or normalization of the data allowed us to bring the variances of all variables to the value of 1 while preserving the relative variability among observations within a variable. Following the normalization of data, partial least squares regressiondiscriminant analysis (PLS-DA) was performed and the bi-plot was constructed. The VIP scores in PLS-DA were calculated to identify significance of variables on phenotype. A VIP score >1.5 was considered as significant for group separation, and the significance level of 0.05 was used to determine statistical significance for other analyses. GraphPad Prism 8 was used (GraphPad Software, Inc., La Jolla, CA, United States) to generate some of the figures.

#### Pathway and Network Analysis

Pathways and network analyses of amino acids were performed using bioinformatic tools. The pathway-based compoundreaction-enzyme-gene networks were identified using MetScape 3.1 (Karnovsky et al., 2012) which was plug in Cytoscape 3.7.1<sup>2</sup> . Interactomes of gene products defined by MetScape 3.1

<sup>1</sup>http://www.metaboanalyst.ca

<sup>2</sup>https://cytoscape.org/

were identified using the biological networks gene ontology tool (BiNGO) within Cystoscope 3.7.1. A merged network was created in Cystoscope by entering subjected genes to analyze the interactome of genes for Bos Taurus, and significance level was set as 0.05.



#### RESULTS

#### Amino Acid Concentration in Bull Seminal Plasma

Twenty-one amino acids were detected in bull seminal plasma (**Figure 1**). Free amino acid concentrations of bull seminal

FIGURE 2 | Concentrations of the most and the least abundant amino acids in bull seminal plasma. (A) The most abundant amino acids in bull seminal plasma was glutamic acid. Alanine, glycine, aspartic acid, and serine were the other predominant amino acids in bull seminal plasma. (B) The least predominant amino acids were tyrosine, methionine, alpha aminobutyric acid, allo-isoleucine, and asparagine.

plasma are depicted in **Table 3**. Among these, glutamic acid was the most abundant amino acid in seminal plasma with an average concentration of 3366 ± 547.3 (mean ± SD) nM, which accounts for approximately 53% of all the amino acids. The other most predominant amino acids were alanine, glycine, aspartic acid, and serine with mean concentrations of 1053 ± 187.9, 429.8 ± 57.94, 427 ± 101.3, and 278.2 ± 40.14 nM, respectively (**Figure 2A**). The least abundant were tyrosine, methionine, alpha-aminobutyric acid, allo-isoleucine, and asparagine with mean concentrations of 12.87 ± 1.91, 8.97 ± 1.76, 7.87 ± 2.53, 5.91 ± 2.13, and 2.92 ± 1.25 nM, respectively (**Figure 2B**).

#### Identification of Potential Freezability Biomarkers

There was no significant difference in amino acid concentrations between GF and PF groups (P > 0.05). However, phenylalanine concentration was significantly correlated with average post-thaw viability (r = 0.57, P-value = 0.044). Additionally, there were significant correlations among individual amino acids (**Figure 3**), such as the concentration of proline was positively correlated with leucine (r = 0.90, P-value < 0.0001), iso-leucine positively correlated with valine (r = 0.92, P-value < 0001), and the concentration of threonine was positively correlated with alanine (r = 0.95, P-value < 0001).

The multivariate analysis, PLS-DA, of seminal plasma amino acids showed a distinct separation between GF and PF bulls (**Figure 4**). PLS-DA was used for the classification. A variable importance in projection (VIP) score, which is widely used in PLS-DA, rank the amino acids considering their significance in discrimination between the GF and PF bulls. VIP score is referred as a weighted sum of squares of the PLS loadings. The X-axis specifies the VIP scores to each variable on the Y-axis. Therefore, amino acids with VIP score >1.5 was identified as phenylalanine, and VIP score in the corresponding heat map demonstrated that phenylalanine is more abundant in seminal plasma of the GF bulls than in that of PF bulls (**Figure 5**).

most important amino acid features identified by PLS-DA. Colored boxes on right indicate concentration of corresponding amino acid from GF and PF samples. VIP score is a weighted based on PLS-DA model.

#### Pathways and Networks of Seminal Plasma Amino Acids

Pathway and network analyses of the amino acids with highest VIP scores (phenylalanine and threonine) and the most abundant amino acids (glutamic acid, alanine, and glycine) were performed using MetScape (3.1.3) (Karnovsky et al., 2012). By analyzing the results, we showed that phenylalanine was involved in tyrosine metabolism, and interacted with several compounds and genes (**Figure 6A**). The interactome of phenylalanine showed that this amino acid contributes to a number of cellular and biological processes, such as antioxidant detoxification, metabolic processes of reactive oxygen species, and oxidoreductase activity (**Table 4**). Threonine was involved in glycine, serine, alanine, and threonine metabolism and it shows significant gene ontology in terms of cellular amino acids and derivative metabolic processes (**Figure 6B**). Glutamic acid was correlated with many genes, enzymes, and other reactions (**Figure 6C**), most of which occur in mitochondria. It has a significant interactome regarding oxidoreductase activity, regulation of cell death, and the oxoacid metabolic process. It also contributes to histidine metabolism, the urea cycle, and the metabolism of arginine, proline, glutamate, aspartate, and asparagine, and Vitamin B9 (folate) metabolism. Alanine created a significant gene ontology in terms of ligase activity and forming carbon–oxygen bonds, and is also involved in pathways of glycine, serine, alanine, and threonine metabolism (**Figure 6D**). Finally, glycine is involved in seven different biological and cellular pathways, and has generated significant gene ontology in terms of oxidoreductase activity (acting on the CH-NH<sup>2</sup> group of donors), sarcosine oxidase activity, and D-amino-acid oxidase activity. All findings are summarized in **Table 4**.

#### DISCUSSION

Successful sperm cryopreservation is an imperative element of fertility management and assisted reproductive studies (ART). The contributions that seminal plasma metabolites have on sperm cryopreservation are still largely unknown. In this present study, we performed GC–MS analyses to investigate the amino acid profiles of bull seminal plasma and classify potential biomolecular markers of freezability. Consecutively, bioinformatic tools were used to identify network and biological pathways of seminal plasma amino acids. To the extent of our knowledge, our study is the first to conduct an extensive assessment of amino acids in bull seminal plasma considering association of specific seminal plasma amino acids with freezability phenotypes.

Seminal plasma is a complex fluid composed of a broad range of metabolites such as organic compounds and energy substrates. Biochemical compositions of seminal plasma differ among species and even among individual males (Killian et al., 1993). This may be due to different management and feeding variations as well as metabolic activity of sperm. These metabolites in seminal plasma have functional roles in sperm preservation, motility, and control of metabolic activity (Bieniek et al., 2016). Amino acids and peptides are the major biochemical compounds found in bovine sperm and its seminal plasma. There is a wide range of amino acids in seminal plasma of which concentrations of many rise after ejaculation due to the massive proteolytic activities occurring in semen (Mann, 1964). Amino acids function as oxidizable substrates for the energy supply, causing reactions in semen (Neumark and Schındler, 2007).

The most abundant amino acid present in seminal plasma is glutamic acid accompanied by a considerable level of glutamic oxaloacetic transaminase (GOT) activity (Flipse, 1960). As in earlier bull semen studies, the abundance of glycine, alanine, serine, aspartic acid, and glutamic acid is found to be high and high levels of amino acids in seminal plasma are higher than in sperm (Roussel and Stallcup, 1967). In a recent study aimed at analyzing metabolomes of seminal plasma from bulls with somewhat higher vs. lower fertility, researchers have identified 63 metabolites, in seminal plasma, of which 21 are amino acids that can be potential biomarker of fertility. Abundances of Lleucine and ornithine differed between the fertility groups, and the levels of fructose were correlated with those of glutamic acid and amino-butyrolactone (Velho et al., 2018). In other studies, researchers have determined different numbers of amino acids and peptides, in seminal plasma of bull, where glutamic and aspartic acid were the most abundant, and were associated with fertility and pregnancy rates (al-Hakim et al., 1970; Holden et al., 2017). Also, seminal plasma of human and other species were found to contain large numbers of amino acids (Engel et al., 2019; Santiago-Moreno et al., 2019). Fertility and sperm freezability are not always related. This current study was aimed at ascertaining

TABLE 4 | The interactome of amino acid shows that amino acid contributes to a great number of cellular and biological processes, such as antioxidant detoxification, reactive oxygen species metabolic processes, and oxidoreductase activity.


seminal plasma amino acids associated with sperm freezability. In the current study, we identified the glutamic acid as the most abundant amino acid. We also demonstrated that glutamic acid was correlated with a number of genes, enzymes, and other reactions, most of which occur in mitochondria. This provides an important evidence of interactome regarding oxidoreductase activity, regulation of cell death, the oxoacid metabolic process, and significant possibility of influence on cell energy production.

The other most predominant amino acids revealed in our study were alanine, glycine, aspartic acid, and serine. When these amino acids in seminal plasma were compared to those found in human seminal plasma (Li et al., 2019), profiles of some seminal plasma amino acids were similar to those profiles we found such as serine, glycine, and glutamic acids. The least abundant in our study, on the other hand, are tyrosine, methionine, alphaaminobutyric acid, allo-isoleucine, and asparagine, and are found to be similar with the low levels of amino acids of methionine and tyrosine in bull (Assumpção et al., 2005). Also, in domestic fowl, valine, serine, glycine, and alanine were the most abundant amino acids followed by glutamic acid (Santiago-Moreno et al., 2019). The alanine created a significant gene ontology in terms of its ligase activity and formed carbon–oxygen bonds, and is also involved in pathways of glycine, serine, alanine, and threonine metabolism.

During the cryopreservation process, sperm undergo critical cryo-injury based on membrane damage, oxidative stress, and DNA fragmentation which reduce post-thaw viability of sperm cells. Even though the exact cryo-protectant mechanisms of amino acids have not been clearly understood, it is presumed they may bind phospholipid membrane bilayers and stabilize the cell membranes (Bilodeau et al., 2001). In addition, osmoregulative and antioxidative features may provide resilience during freezing–thawing (Kruuv and Glofcheski, 1992; Farshad and Hosseini, 2013). However, there are not a great number of studies that have investigated the protective influence of amino acids against cryo-injury. Previous studies have claimed that seminal plasma supplementation of amino acids into semen extenders improved sperm viability, acrosome integrity and membrane integrity of sperm (Ali Al Ahmad et al., 2008), and post-thaw semen quality (Saravia et al., 2009). More specifically, in human research, it was found that addition of glutamine to semen as a cryoprotectant agent increased post-thaw motility in human sperm (Atessahin et al., 2008). In animal studies, supplementation of extender solutions with glutamine, glycine, and cysteine enhanced acrosome and membrane integrity of buffalo bull semen (El-Sheshtawy et al., 2008). Additionally, there was a positive correlation between membrane integrity and the concentration of valine, isoleucine and leucine, and lysine (Santiago-Moreno et al., 2019).

One of the most common negative consequences of cryopreservation of sperm cell is DNA damage, and majority of DNA lesions in sperm cells is caused by oxidative stress (Zribi et al., 2010). Seminal plasma content plays a significant role in protection against oxidative stress. Aitken and Baker (2004) clarified that taurine and hypotaurine are the amino acids that reduce oxidative stress through binding to the oxidizing agents. In addition, supplementation of donkey semen with glutamine reduced DNA fragmentation index (Bottrel et al., 2018). Sangeeta et al. (2015) reported that supplementation of ram sperm with L-glutamine and L-proline reduced lipid peroxidation and increased acrosomal integrity. Glutamic acid is the key component of glutathione which has been demonstrated to inhibit cellular damage resulting from lipid peroxidation and reactive oxygen species (Arai et al., 1999). In the present study, we showed that phenylalanine is more abundant in seminal plasma of the GF bulls than in that of PF bulls. It has significant gene ontology terms for antioxidant activity, response to oxidative stress, and oxidoreductase activity through

its actions on peroxide as acceptor, and metabolic processes of oxygen as well as reactive oxygen species. We postulate that phenylalanine could have an antioxidant effect, and increased concentrations of phenylalanine in seminal plasma may reduce DNA damage caused by oxidative stress. Moreover, PLS-DA results demonstrate a distinct separation between GF and PF groups. Thus, the abundance of glutamic acid may explain protective effects of seminal plasma during cryopreservation. Furthermore, glutamine may play an important role in gene expression redox-potential, and cell integrity (Curi et al., 2005). It has been assumed electrostatic interactions between plasma membrane phospholipids and amino acids help to generate a layer on the sperm surface, and which thus protects the sperm cell from cryo-injury (Anchordoguy et al., 1988; Kundu et al., 2001).

#### CONCLUSION

We have found that glutamic acid, alanine, and glycine are the predominant metabolites in bull seminal plasma. It is clear that there is a distinct separation of the amino acid profiles for the seminal plasmas of GF and PF bulls. According to our findings, phenylalanine could be considered as a freezability biomarker, and may be used as a cryoprotectant supplement. In addition, amino acid profiles of the seminal plasma could be used to determine the freezability phenotypes. These findings help us better understand the exact mechanisms of cryopreservation for sperm cells as well as other cell types.

## REFERENCES


### DATA AVAILABILITY STATEMENT

The raw data generated from this study will be available through the corresponding author.

#### AUTHOR CONTRIBUTIONS

MU, AK, TD, and EM conceptualized the study. MU, TD, and MH curated the data. MU, TD, MH, and EM carried out the investigations. AK, BD, and ET provided the essential samples and phenotypic data. MU, TD, MH, AK, and EM wrote the original draft, and reviewed and edited the manuscript.

#### FUNDING

This project was supported by the Agriculture and Food Research Initiative Competitive Grant No. 2017-67016- 26507 from the USDA National Institute of Food and Agriculture. Partial funding was provided by the Mississippi Agricultural Forestry Experiment Station, and by Alta Genetics Inc., Watertown, WI, United States. MH was funded through a competitive international postdoctoral fellowship by the Scientific and Technological Research Council of Turkey (TUBITAK). MU was funded through a competitive graduate fellowship from the Turkish Ministry of National Education.



early embryo development. Cryobiology 67, 84–90. doi: 10.1016/j.cryobiol.2013. 05.007


deoxyribonucleic acid integrity. Fertil. Steril. 93, 159–166. doi: 10.1016/j. fertnstert.2008.09.038

**Conflict of Interest:** ET and BD employed by the company Alta Genetics, Inc.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Ugur, Dinh, Hitit, Kaya, Topper, Didion and Memili. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Membrane Potential Assessment by Fluorimetry as a Predictor Tool of Human Sperm Fertilizing Capacity

Carolina Baro Graf1,2† , Carla Ritagliati<sup>1</sup>† , Valentina Torres-Monserrat<sup>3</sup> , Cintia Stival<sup>1</sup> , Carlos Carizza<sup>3</sup> , Mariano G. Buffone<sup>4</sup> \* and Dario Krapf1,2 \*

<sup>1</sup> Laboratory of Cell Signal Transduction Networks, Instituto de Biología Molecular y Celular de Rosario (IBR), CONICET-UNR, Rosario, Argentina, <sup>2</sup> Laboratorio de Medicina Reproductiva, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, Rosario, Argentina, <sup>3</sup> Fertya, Medicina Reproductiva, Rosario, Argentina, <sup>4</sup> Instituto de Biología y Medicina Experimental (IBYME), CONICET, Buenos Aires, Argentina

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Elisabetta Baldi, University of Florence, Italy Katerina Komrskova, Czech Academy of Sciences, Czechia

#### \*Correspondence:

Mariano G. Buffone mgbuffone@ibyme.conicet.gov.ar Dario Krapf krapf@ibr-conicet.gov.ar †These authors share first authorship

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

Received: 15 August 2019 Accepted: 18 December 2019 Published: 17 January 2020

#### Citation:

Baro Graf C, Ritagliati C, Torres-Monserrat V, Stival C, Carizza C, Buffone MG and Krapf D (2020) Membrane Potential Assessment by Fluorimetry as a Predictor Tool of Human Sperm Fertilizing Capacity. Front. Cell Dev. Biol. 7:383. doi: 10.3389/fcell.2019.00383 Mammalian sperm acquire the ability to fertilize eggs by undergoing a process known as capacitation. Capacitation is triggered as the sperm travels through the female reproductive tract. This process involves specific physiological changes such as rearrangement of the cell plasma membrane, post-translational modifications of certain proteins, and changes in the cellular permeability to ions – with the subsequent impact on the plasma membrane potential (Em). Capacitation-associated Em hyperpolarization has been well studied in mouse sperm, and shown to be both necessary and sufficient to promote the acrosome reaction (AR) and fertilize the egg. However, the relevance of the sperm Em upon capacitation on human fertility has not been thoroughly characterized. Here, we performed an extensive study of the Em change during capacitation in human sperm samples using a potentiometric dye in a fluorimetric assay. Normospermic donors showed significant Em hyperpolarization after capacitation. Em values from capacitated samples correlated significantly with the sperm ability to undergo induced AR, highlighting the role of hyperpolarization in acrosomal responsiveness, and with successful in vitro fertilization (IVF) rates. These results show that Em hyperpolarization could be an indicator of human sperm fertilizing capacity, setting the basis for the use of Em values as a robust predictor of the success rate of IVF.

Keywords: membrane potential, sperm capacitation, in vitro fertilization, acrosome reaction, human infertility

#### INTRODUCTION

Infertility is a worldwide public health problem affecting ∼1 in 7, or ∼80 million couples worldwide (Slama et al., 2012; Datta et al., 2016). Although the causes of infertility are heterogeneous, the male factor is now conceived as important as the female etiology, accounting for at least 50% of cases (HFEA, 2014). Infertile couples rely on assisted reproductive technology (ART) to conceive, which include: IUI (intra-uterine insemination), in vitro fertilization (IVF), and ICSI (intracytoplasmic sperm injection). These treatments are expensive, invasive and risky. Excluding the female contribution, the decision on which treatment to choose often relies on general parameters of sperm in their basal state (i.e., non-capacitated sperm). However, these parameters are not true indicators of treatment success. Thus, a detailed understanding of how both normal and dysfunctional spermatozoa behave is necessary to develop a platform for new diagnostic tools and to infer best treatment options (Barratt et al., 2017, 2018).

Immediately after ejaculation the mammalian sperm is unable to fertilize the egg until it goes through a maturation process known as capacitation. Two capacitation events are essential for successful sperm penetration into oocytes: a vigorous motility called hyperactivation (HA) and the ability to undergo the acrosome reaction (AR) in response to a physiological agonist (Yanagimachi, 1994a,b). In addition, a series of sequential and concomitant biochemical processes must occur during capacitation, including the hyperpolarization of the plasma membrane (Zeng et al., 1995; Arnoult et al., 1999; Muñoz-Garay et al., 2001; Demarco et al., 2003; Ritagliati et al., 2018). In sperm, as in most cells, the internal ion concentrations are markedly different from those in the extracellular medium. At rest, the balance of ion fluxes, gradients, and permeabilities results in an electric potential, known as the resting Em (Visconti et al., 2011; Stival et al., 2016). Mammalian sperm encounter environments with very different ionic composition on their journey to meet the egg. Sperm must regulate their Em and adapt to the changes of external ion concentration. In turn, Em modulates membrane ion channels and transporters, including the sperm-specific Ca2<sup>+</sup> channel CatSper and voltage-gated proton channel Hv1 (Darszon et al., 2011; Lishko et al., 2012). Ion channels and ionic gradients play key roles in orchestrating intracellular signaling pathways.

Hyperpolarization of the plasma membrane during capacitation has been thoroughly studied in mouse sperm. Initially, mouse sperm are relatively depolarized (Em between −35 and −45 mV) and become hyperpolarized (Em around −70 mV) upon capacitation (Zeng et al., 1995; Arnoult et al., 1999; Stival et al., 2015; Ritagliati et al., 2018). This hyperpolarization has been shown to be both necessary and sufficient for sperm to undergo [Ca2+]<sup>i</sup> increases and the AR in response to an agonist (De La Vega-Beltran et al., 2012). Despite the different roles that membrane hyperpolarization might play during capacitation, Em changes have not been extensively studied in human sperm. Linares-Hernández et al. (1998) reported that the Em of non-capacitated sperm is around −40 mV, whereas Patrat et al. (2005) reported that capacitated sperm exhibit an Em of about −58 mV. On the other hand, López-González et al. (2014) used a potentiometric dye in a flow cytometry assay to show that a subpopulation of human sperm undergoes Em hyperpolarization upon capacitation, which correlated with an increase in intracellular pH and Ca2<sup>+</sup> concentration. However, they determined relative membrane potentials through the median fluorescence observed in each condition and not absolute values. Furthermore, the physiological relevance of Em hyperpolarization has not been elucidated yet.

Mouse sperm lacking either the K<sup>+</sup> channel Slo3 or its auxiliary subunit Lrrc52 have markedly reduced fertility (Santi et al., 2010; Zeng et al., 2011, 2015). Thus, it could be hypothesized that malfunction of K<sup>+</sup> channels in human sperm might also contribute to the occurrence of subfertility in men. Brown et al. (2016) used whole-cell patch-clamp electrophysiology to assess the biophysical characteristics of sperm from men undergoing fertility treatments and compared to those from fertile, healthy donors. In approximately 10% of the samples from infertile patients there was either a negligible outward conductance or an enhanced inward current, both of which caused Em depolarization. Analysis of clinical data from IVF patients showed a significant association of depolarized Em assessed by patch-clamp with low fertilization rate (Brown et al., 2016). These results point toward the correlation of the Em value of a sperm sample with its fertilizing capacity. However, patchclamp is a complex, laborious and time-consuming technique, not compatible with prediction tools needed for the clinic. In addition, in view of the heterogeneity of sperm populations, it is highly possible that values acquired by patch-clamp techniques from a few cells would not be representative of those undergoing capacitation, which could lead to incorrect conclusions.

We aimed to analyze in depth human sperm Em during capacitation, its role in the acquisition of HA and AR, and its association with fertilization competence. Therefore, we characterized the Em upon capacitation in 60 human sperm samples using a potentiometric dye in a fluorimetric assay (Baro Graf et al., 2019). Sperm from normospermic donors showed a significant Em hyperpolarization during capacitation. In addition, Em values from capacitated sperm correlated significantly with the sperm capacity to undergo induced AR, underlying the essential role of hyperpolarization in acrosomal responsiveness. More importantly, this study shows that Em hyperpolarization is associated with successful IVF rates. Our results pave the way for the application of Em measurement as a useful tool to predict the success rate of IVF procedures.

## MATERIALS AND METHODS

#### Ethical Approval

Volunteer donors and patients were provided with written information about the study prior to giving informed consent. The study protocol was approved by the Bioethics Committee of the Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, protocol #564/2018. The studies are in compliance with the Declaration of Helsinki principles.

#### Human Sperm Preparation

Semen samples were obtained by masturbation from healthy donors after 2–5 days of abstinence and analyzed following WHO recommendations (World Health Organization, 2010). Samples that fulfilled semen parameters (total fluid volume, sperm concentration, motility, viability and morphology) according to WHO normality criteria, were considered as normospermic and those that did not fulfilled any of them as non-normospermic. Samples were allowed to liquefy for 1 h at room temperature, then, sperm ejaculates were allowed to swim-up in noncapacitating media at 37◦C for 1 h. The non-capacitating medium used was HEPES-buffered human tubal fluid (HTF) containing 90.7 mM NaCl, 4.7 mM KCl, 0.3 mM KH2PO4, 1.2 mM MgSO4, 2.8 mM glucose, 3.4 mM sodium pyruvate, 1.6 mM CaCl2, 60 mM sodium lactate and 23.8 mM HEPES (pH 7.4), which was supplemented with 20 mM bicarbonate and 5 mg/ml BSA to obtain the capacitating medium. Cells were left to capacitate at 37◦C as detailed for each experiment, in the same procedure as that used to prepare spermatozoa for IVF.

In Fertya (Assisted Reproduction Medical Clinic), commercially available HTF media was used for sperm preparation. The spermatozoa were separated from semen by two-layer density gradient centrifugation (Irvine Scientific Isolate) and then washed and concentrated with Quinn's Advantage Medium with HEPES (Sage) which was supplemented with 10% Serum Protein Substitute (Sage). Washed spermatozoa were incubated to capacitate at 37◦C pH 7.4 for 3–4 h prior to performing the IVF.

#### Membrane Potential Assay in Cell Populations

Cells were loaded with 1 µM of the membrane-potentialsensitive dye DISC3(5) (Molecular Probes) for at least 5 min. No mitochondrial un-couplers were used because their contribution to the resting potential has been determined to be insignificant (Guzmán-Grenfell et al., 2000). Sperm were then transferred to a gently stirred cuvette at 37◦C, and the fluorescence was monitored with a Varian Cary Eclipse fluorescence spectrophotometer at 620/670 nm excitation/emission wavelengths. Recordings were initiated when steady-state fluorescence was reached and calibration was performed at the end of each measure by adding 1 µM valinomycin and sequential additions of KCl for internal calibration curves, as previously described (Ritagliati et al., 2018; Baro Graf et al., 2019). Sperm Em was obtained from the initial fluorescence (measured as Arbitrary Fluorescence Units) by linearly interpolating it in the theoretical Em values from the calibration curve against arbitrary fluorescence units of each trace. This internal calibration for each determination compensates for variables that influence the absolute fluorescence values.

#### Acrosome Status Assay

After incubation in the respective conditions, progesterone (21 µM) was added and incubated for another 30 min. Alternatively, in the NC<sup>0</sup> condition, 1 µM valinomycin was added 5 min before the progesterone. Cells were seeded on eight-well glass slides. After air-drying, sperm were fixed with 3.7% formaldehyde in PBS for 15 min at room temperature, permeabilized with 0.5% Triton X-100 for 5 min, washed with PBS and incubated with PBS containing 1% BSA and FITCconjugated pisum sativum lectin (1/200) for 1 h at room temperature. Before mounting, samples were washed with PBS (four times for 5 min each time). Epifluorescence microscopy was used to assess acrosome status. At least 200 sperm were analyzed in each condition.

#### Sperm Motility Analysis

Sperm suspensions were loaded on a 30-µm deep slide and placed on a microscope stage at 37◦C. Sperm movements were examined using computer-assisted semen analysis (CASA) system (IVOS Sperm Analyzer, Hamilton Thorn). Thirty frames were acquired at a rate of 60 Hz. At least 200 sperm were analyzed in each condition. The following parameters were measured: mean path velocity (VAP, µm/sec), curvilinear velocity (VCL, µm/sec), straight-line velocity (VSL, µm/sec), linearity (LIN, %), amplitude of lateral head displacement (ALH, µm), and straightness (STR, %). Sperm were considered hyperactivated when presenting VCL ≥ 150 µm/sec, LIN < 50%, and ALH ≥ 5 µm. At the time of the analysis, 0.5 mg/ml BSA was added to non-capacitated conditions to avoid sperm adherence to the slide.

#### Fertilization Rate at IVF

In vitro fertilization was performed 3–4 h post sperm preparation in 4-well dishes (Oosafe) with 0.5 ml of Quinn's Advantage Fertilization medium supplemented with 10% Serum Protein Substitute (Sage) and covered with 0.5 ml of tissue culture oil (Sage). In order to assure good oocyte quality, this study only involved female patients with no detected female factors. All patients were below 38 years old. One thousand sperm were incubated per cumulus-oocyte complex. A maximum of five cumulus-oocyte complex were incubated per well in K-System G185 incubator (37◦C, 6.2% CO2, 5% O2) during 4 h. After this period, oocytes were denudated and incubated in individual drops in QA Protein plus Cleavage medium (SAGE <sup>R</sup> ) covered with 0.5 ml of tissue culture oil (Sage) in the same conditions as described above. At 18 h post insemination, fertilization was analyzed with an inverted microscope. Successful fertilization was considered when two pronuclei (2PN) and two distinct or fragmented polar bodies were observed. When at least 60% of eggs were fertilized, the IVF procedure was classified as successful, according to local regulations.

#### Statistical Analyses

Data are expressed as mean ± standard deviation (SD) or standard error of the mean (SEM), as indicated in each figure, of at least six independent experiments for all determinations. Statistical analyses were performed using the GraphPad Prism 6 software (La Jolla, CA, United States). Student's t test was used to compare mean values between control and tested groups, while the difference between mean values of multiple groups was analyzed by one-way analysis of variance (ANOVA) with multiple comparison tests. A probability (p) value p < 0.05 was considered statistically significant.

## RESULTS

#### Human Sperm Hyperpolarize During Capacitation

In this work, we performed a fluorimetric population assay to determine the absolute Em values of non-capacitated and capacitated sperm. When constant concentrations of sperm and probe are used, this method provides a highly reproducible value of plasma membrane potential after calibration using valinomycin (a K<sup>+</sup> ionophore) and sequential additions of KCl (Baro Graf et al., 2019) (**Figure 1A**). A total of 60 sperm samples were analyzed, among which 49 corresponded to normospermic and 11 to non-normospermic donors, as categorized on the basis of total sperm number, ejaculate volume and motility

according to WHO guidelines (see section "Materials and Methods" for details). Consistent with previous observations using flow cytometry (López-González et al., 2014), we found a large heterogeneity among Em values of samples from different individuals. However, normal sperm samples exhibited a significant hyperpolarization after capacitation, while subnormal samples did not (**Figure 1B**). The Em changes (EmCAP – EmNC) were also very variable between individuals. Both in normospermic and non-normospermic donors we could identify samples with depolarizing, hyperpolarizing or unchanged behaviors (**Figure 1C**). However, while in normospermic donors the majority of the samples (53.6%) hyperpolarized, among donors with sub-normal parameters, only 27.3% hyperpolarized and 63.6% depolarized (**Figure 1D**). When samples with different behaviors were pooled and further analyzed, Em changes became more evident. These results demonstrate that in most normospermic individuals there is a capacitation-induced hyperpolarization, with an initial non-capacitated Em value of −37.7 ± 9.9 mV, that shifts to –57.8 ± 12.9 mV upon capacitation (**Figure 1E**).

## Hyperpolarization Correlates With Acrosomal Responsiveness

The AR is a key step in fertilization. In mammalian sperm, the AR is mediated by intracellular calcium fluxes. Since Ca2<sup>+</sup> channels

in sperm cells are voltage-dependent, it has been hypothesized that gating might be controlled by Em.

We evaluated the role of the Em in the acquisition of acrosomal responsiveness to progesterone (Pg) in samples from normospermic donors. Both Em and AR were evaluated immediately after swim-up (NC0), as well as after 5 h in either non-capacitating (NC5) or capacitating (CAP) media. For AR analysis, cells were further incubated for 30 min in the absence or presence of Pg. Non-capacitated sperm obtained after swim-up (NC0) did not increase the percentage of acrosome reacted sperm in the presence of Pg. However, they acquired acrosomal responsiveness after pharmacological hyperpolarization with valinomycin (1 µM for 5 min) (**Figures 2B,C**). These results point toward the pharmacological hyperpolarization sufficiency for acrosomal responsiveness, in agreement to previous results in mouse sperm (De La Vega-Beltran et al., 2012). On the other hand, capacitated human sperm hyperpolarized (**Figure 2A**) and showed significantly increased induced AR (**Figures 2B,C**). Furthermore, a correlation analysis between Em change upon capacitation (EmCAP – EmNC0) and the induced AR (induced – spontaneous), although not statistically significant (p = 0.06), showed a clear negative correlation (r = −0.6337) (**Figure 2D**). Interestingly, as observed in the correlation analysis between absolute Em values and induced AR for each condition (**Figures 2E,F**), there is a strong negative correlation for sperm incubated in capacitating (r = −0.8234) media, but not in NC<sup>0</sup> (r = 0.0025).

#### Role of Em in Hyperactivated Motility

Hyperactivation is essential for mammalian sperm to fertilize the oocyte (White and Aitken, 1989; Suarez et al., 1993); we therefore aimed to determine if hyperpolarization of human sperm would affect their motility. Hyperactivated motility is defined by an increase in the VCL and the ALH and a decreased LIN: VCL ≥ 150 µm/s, LIN ≤ 50% and ALH ≥ 5 µm/s, being VCL the most characteristic parameter. **Figure 3** shows Em values (**Figure 3A**), percentage of hyperactivated sperm (**Figure 3B**) and VCL (**Figure 3C**) of sperm incubated in either non-capacitated or capacitated conditions at different time points. As can be observed, sperm progressively hyperpolarized in time with a maximum at 5 h (which was the longest time assayed), while the maximum HA rate and VCL were at 3 h (**Figures 3A–C**). In order to assess the role of hyperpolarization in sperm motility, we performed a correlation analysis of Em values with HA and

FIGURE 2 | Hyperpolarized membrane potentials correlate with acrosomal responsiveness. Sperm samples from normospermic donors obtained after swim-up (NC0) and upon 5 h incubation in non-capacitating (NC5) and capacitating (CAP) media were analyzed. (A) Em measurements. (B,C) Cells were further incubated for 30 min in the absence (-) or presence (+) of 21 µM progesterone (Pg). The percentage of acrosome reacted sperm was assessed by FITC-PSA staining as described in section "Materials and Methods." Data represent mean ± SEM from at least six independent experiments. Paired Student's t test was performed. Statistically significant differences between the indicated conditions or with the control NC<sup>0</sup> (asterisks on each column bar) are as follows: <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.005, ∗∗∗∗p < 0.0001. (D–F) Correlation analysis between the change on Em (D, 1 EmCAP – EmNC0) or the Em values in CAP (E) and NC<sup>0</sup> (F) and the induced AR. The correlation coefficients (r) and p values are indicated.

VCL after 3 and 5 h of capacitation (**Supplementary Figure S1**). Though not significant, there is a tendency in the correlation between Em at 3 h of capacitation and VCL (**Supplementary Figure S1A**). On the other hand, the correlation was weaker at 5 h of capacitation (**Supplementary Figure S1B**), which can be explained by the fact that at 5 h sperm further hyperpolarized, while HA did not continue increasing (**Figure 3**). Finally, the analysis between Em change upon capacitation (EmCAP3 – EmNC0) and the VCL or HA values, showed a weak correlation (**Supplementary Figure S1C**).

#### Em Analysis Predicts IVF Outcome

Aiming to further investigate the biological role of Em on human sperm fertilization competence, we determined the Em of non-capacitated and capacitated sperm from IVF patients. One important drawback in correlation analyses of human sperm is the biological variation between samples obtained from the same donors on different days. Thus, in order to avoid this issue, the fluorimetric Em measurements were performed on aliquots of the ejaculate used on the day of the IVF treatment, allowing direct comparison of Em values with fertilization success.

In vitro fertilization patients were classified according to their fertilization rate, satisfying local clinical parameters for successful (fertilization rate ≥ 60%) and unsuccessful IVF procedures (fertilization rate < 60%). These two groups showed significant differences between their mean IVF success rates (**Figure 4A**). In a retrospective analysis, sperm samples with successful IVF rates exhibited a significant Em hyperpolarization during capacitation (**Figure 4B**). On the other hand, patients with IVF rates under 60% did not hyperpolarize (**Figure 4B**). As can be seen in **Figure 4C**, 78.6% of successful IVF samples hyperpolarized during capacitation. However, the majority of samples from poor IVF rate patients (66.7%) showed a depolarizing behavior. These results indicate that the IVF outcome might be dependent on a capacitation-associated hyperpolarization. To test this hypothesis, we performed correlation analysis of absolute Em values obtained after capacitation and IVF rates (**Figure 4D**), and also of Em changes upon capacitation (EmCAP – EmNC) against IVF rates (**Figure 4E**). In agreement with our previous results (**Figure 2**) where we observed a correlation between capacitated sperm Em and induced AR, there was a significant correlation between capacitated sperm Em (EmCAP) and IVF rate. Interestingly, a predictive analysis showed that hyperpolarizing samples have significantly higher IVF rates in comparison with depolarizing samples, as depicted in **Figure 4F**. Finally, a ROC curve was constructed to assess the effectiveness of capacitated sperm Em in predicting IVF outcomes (**Figure 5**). The area under the curve was 0.8571 ± 0.098 (95% CI = 0.6647-1), indicating that capacitated sperm Em (EmCAP) is a good parameter to discriminate between successful IVF rate (>60%) and unsuccessful IVF rate (<60%). The cut off value for EmCAP with the highest sensitivity and specificity was −48.6 mV (100% sensitivity and 71.4% specificity). Considering our results, a depolarized Em could be related to IVF failure in idiopathic subfertile patients.

### DISCUSSION

A total of 60 sperm samples were analyzed, among which 49 corresponded to normospermic donors. These samples exhibited a significant Em hyperpolarization after incubation in capacitating media, while sub-normal samples did not hyperpolarize. This raises the question whether functional defects in the capacitation-associated hyperpolarization relate to subnormal parameters. Although further work is needed regarding more non-normospermic donors, our results may suggest that normal semen parameters can be associated to the relative permeability of the plasma membrane, ion channels regulation and metabolic state of the sperm resulting in their ability to hyperpolarize in capacitating conditions.

The AR, a key step in fertilization, is strictly dependent on an increase in intracellular Ca2<sup>+</sup> (Romarowski et al., 2016). In 1998, a voltage-dependent Ca2<sup>+</sup> influx caused by sperm depolarization was described in human sperm, which was enhanced when

depolarization was preceded by hyperpolarization (Linares-Hernández et al., 1998). This was consistent with a hypothesis stating that capacitation-associated hyperpolarization is required to remove voltage Ca2<sup>+</sup> channels inactivation. Furthermore, the steroid hormone progesterone, which is the only well-characterized biological agonist of the AR in human sperm, activates CatSper, induces Ca2<sup>+</sup> influx, membrane depolarization and AR (Aitken, 1997; Kirkman-Brown et al., 2000; Patrat et al., 2005; Lishko et al., 2011). In mouse sperm it has been shown that Em hyperpolarization is necessary and sufficient for cells to acquire acrosomal responsiveness (De La Vega-Beltran et al., 2012). However, in human sperm, the correlation between Em and acrosomal responsiveness had not been thoroughly demonstrated yet. The results presented herein indicate that Em hyperpolarization plays an important role in human sperm AR. We have shown that: (1) capacitated sperm exhibit significant Em hyperpolarization and induced AR; (2) non-capacitated depolarized sperm only gain acrosomal responsiveness after pharmacological hyperpolarization; and (3) there is a strong and significant negative correlation between Em and induced AR in capacitated sperm. Altogether, these data support an important role of Em hyperpolarization for the acquisition of acrosomal responsiveness.

It is generally accepted that good sperm motility is a central component of male fertility. Individuals with poorly motile or immotile sperm are considered infertile or subfertile, and in need of ART procedures. In fact, asthenozoospermia is the commonest problem underlying male subfertility (Van Der Steeg et al., 2011) and because the root cause of this condition is usually not known, treatments for this problem are non-specific. In this context, we aimed to understand whether human sperm Em hyperpolarization and HA were linked. A correlation tendency was observed at 3 h between both parameters. It is well established that intracellular calcium plays a pivotal role in sperm motility regulation. Alasmari et al. (2013) showed that defects in Ca2<sup>+</sup> signaling lead to poor HA and that the ability to undergo Ca2<sup>+</sup> -induced HA affects sperm fertilizing capacity. This is in agreement with our results and with the hypothesis that the Em might play a role in regulating Ca2<sup>+</sup> channels and, consequently, in intracellular calcium and indirectly in sperm motility (reviewed in Ritagliati et al., 2018).

Regardless the amount of work invested toward understanding the molecular basis of sperm capacitation,

studies of male factor issues attending reproductive clinics seem to be completely dissociated from basic science knowledge. Regretfully, semen diagnostic analysis hardly involves sperm function evaluation, i.e., capacitation parameters required for fertilization. With this in mind, we aimed to analyze whether Em hyperpolarization relates to human sperm fertilizing capacity. It has been recently proposed that a certain degree of Em shift is associated with normal sperm function, as assessed by electrophysiology and IVF outcome (Brown et al., 2016). However, this study involved laborious techniques that hampered the analysis of many cells per patient. Thus, we pursued the study of sperm Em from patients attending a fertility clinic using a robust and relatively simple technique. In the fluorimetric population assay, the sample's behavior is followed throughout the experiment and is performed in the physiological working conditions allowing more accurate correlations (Baro Graf et al., 2019). Em measurements were performed on aliquots of the same samples used in IVF procedures. Our data show that almost 80% of successful IVF samples hyperpolarized. On the other hand, there was no hyperpolarization in samples with unsuccessful IVF rates. In fact, the majority of these samples (66.7%) depolarized during incubation in capacitating conditions. Accordingly, there is a significant strong correlation between capacitated sperm Em and IVF rate. These data strongly suggest that human sperm Em changes have an implication in sperm fertilizing capacity. Interestingly, in a predictive study, samples that depolarized upon capacitation exhibited lower fertilization rates. After a ROC analysis we determined that the Em absolute value from capacitated sperm can be considered as a good parameter to predict IVF rates, with a cut-off value of −48.6 mV. Although further work is needed in order to increase the number of patients analyzed, and to evaluate whether this technical approach could also predict IUI success, this study has the potential to add diagnostic tools to help predict the success of reproductive techniques.

During preparation of this manuscript, we contacted the group of Dr. Celia Santi at Washington University School of Medicine in Saint Louis, who independently achieved the same results using a different methodology. Both this manuscript and Santi's work are intended to be published concurrently (Puga Molina et al., 2020).

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the Bioethics Committee of the Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario. The patients/participants provided their written informed consent to participate in this study.

#### AUTHOR CONTRIBUTIONS

CB, CR, and CS conducted the experiments. VT-M and CC designed and conducted the IVF procedures. CB, CR, MB, and DK conceived the study. CB, CR, and DK wrote the manuscript. All authors analyzed the data and revised the final version of the manuscript.

#### FUNDING

This work was supported by grants from the Agencia Nacional de Promoción Científica y Tecnológica de Argentina PICT 2015- 3164 and PICT 2017-3217 (awarded to DK). CB, CR, and CS are recipients of a fellowship from CONICET, Argentina.

#### ACKNOWLEDGMENTS

We thank the Male Contraceptive Initiative for support granted to CB and the Statistics and Data Processing Department from Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, Argentina.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00383/ full#supplementary-material

#### REFERENCES

fcell-07-00383 January 17, 2020 Time: 13:8 # 9



**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Baro Graf, Ritagliati, Torres-Monserrat, Stival, Carizza, Buffone and Krapf. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Quantitative Intracellular pH Determinations in Single Live Mammalian Spermatozoa Using the Ratiometric Dye SNARF-5F

#### Julio C. Chávez, Alberto Darszon, Claudia L. Treviño and Takuya Nishigaki\*

Departamento de Genética del Desarrollo y Fisiología Molecular, Instituto de Biotecnología, Universidad Nacional Autónoma de México, Cuernavaca, Mexico

Intracellular pH (pH<sup>i</sup> ) plays a crucial role in mammalian sperm physiology. However, it is a challenging task to acquire quantitative single sperm pH<sup>i</sup> images due to their small size and beating flagella. In this study, we established a robust pH<sup>i</sup> imaging system using the dual-emission ratiometric pH indicator, SNARF-5F. Simultaneous good signal/noise ratio fluorescence signals were obtained exciting with a green high-power LED (532 nm) and acquiring with an EM-CCD camera through an image splitter with two band-pass filters (550–600 nm, channel 1; 630–650 nm, channel 2). After in vivo calibration, we established an imaging system that allows determination of absolute pH<sup>i</sup> values in spermatozoa, minimizing cell movement artifacts. Using this system, we determined that bicarbonate increases non-capacitated human pH<sup>i</sup> with slower kinetics than in mouse spermatozoa. This difference suggests that distinct ionic transporters might be involved in the bicarbonate influx into human and mouse spermatozoa. Alternatively, pH<sup>i</sup> regulation downstream bicarbonate influx into spermatozoa could be different between the two species.

#### Keywords: intracellular pH, alkalization, spermatozoa, dual emission, image splitter, ratiometric

## INTRODUCTION

The pH is fundamental for most proteins to ensure their proper function, as it influences the electrostatic status of their side chains that, in turn, affect protein structure (folding and conformation) and their interaction with other molecules (Zhou and Pang, 2018). Therefore, intracellular pH (pHi) changes serve as crucial signals in many cell types.

In spermatozoa, pH<sup>i</sup> critically regulates motility (Ho et al., 2002; Nishigaki et al., 2014). In mammals, spermatozoa remain quiescent in the epididymis due to the acidic environment created by vacuolar-type H+-ATPase (V-ATPase) found in the apical plasma membrane of epithelial cells (Acott and Carr, 1984; Brown et al., 1997). Flagellar beating is suppressed in acidic environments as dynein ATPases, the motor molecules that propel the flagellum, are highly pH<sup>i</sup> dependent (Crhisten et al., 1983). Upon ejaculation and contact with the seminal fluid sperm pH<sup>i</sup> increases, and the flagellum starts beating. The initial flagellar beat is symmetric with low amplitude and high frequency. Subsequently in the oviduct, the flagellar beat pattern becomes vigorous (asymmetric with high amplitude and low frequency), a process called hyperactivation (Ho and Suarez, 2001). Hyperactivated motility is essential for mammalian spermatozoa since it is required to approach the oocyte and to penetrate its investments (Stauss et al., 1995; Suarez and Pacey, 2006). In order to induce and maintain hyperactivation, an increase in intracellular Ca2<sup>+</sup> concentration ([Ca2+]i)

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Shaomin Shuang, Shanxi University, China Lin Yuan, Hunan University, China

#### \*Correspondence: Takuya Nishigaki

takuya@ibt.unam.mx

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 13 August 2019 Accepted: 13 December 2019 Published: 17 January 2020

#### Citation:

Chávez JC, Darszon A, Treviño CL and Nishigaki T (2020) Quantitative Intracellular pH Determinations in Single Live Mammalian Spermatozoa Using the Ratiometric Dye SNARF-5F. Front. Cell Dev. Biol. 7:366. doi: 10.3389/fcell.2019.00366

is required (Ho et al., 2002), which is mediated through a sperm-specific Ca2<sup>+</sup> channel, named CatSper (Ren et al., 2001). Although there are species-specific activation mechanisms of CatSper (Lishko et al., 2011), this channel is moderately voltage dependent and highly up regulated by intracellular alkalization (Kirichok et al., 2006). In mouse, the sperm-specific Na+/H<sup>+</sup> exchanger (sNHE) is essential for the regulation of sperm motility and has been proposed as an activator of CatSper by elevating pH<sup>i</sup> (Wang et al., 2003; Navarro et al., 2008). On the other hand, in human spermatozoa, a voltage-gated H<sup>+</sup> channel (Hv1) has been documented to be the main H<sup>+</sup> transporter that activates CatSper rather than sNHE (Lishko et al., 2010). In sea urchin sperm, CatSper is a predominant player in chemotaxis toward spermattracting peptides (Seifert et al., 2015; Espinal-Enríquez et al., 2017) and sNHE has been shown to be critical for modulating CatSper activity (González-Cota et al., 2015; Windler et al., 2018).

External bicarbonate (HCO<sup>−</sup> 3 ) is fundamental for capacitation in mammalian spermatozoa (Lee and Storey, 1986; Visconti et al., 1995). Both the pH and the HCO<sup>−</sup> 3 concentration of the oviductal fluid are higher in uterine and tubal fluids compared to plasma (Vishwakarma, 1962). Moreover, pH in the rhesus monkey female tract elevates dramatically, concomitantly with ovulation (Maas et al., 1977), which might promote sperm capacitation in vivo. In mammalian spermatozoa, several HCO<sup>−</sup> 3 transporters were reported as candidates to mediate HCO<sup>−</sup> 3 influx across the plasma membrane such as Na+/HCO<sup>−</sup> 3 cotransporter (NBC) (Demarco et al., 2003), Cl−/HCO<sup>−</sup> 3 exchangers (Chavez et al., 2012), and CFTR (Hernández-González et al., 2007; Xu et al., 2007), as well as its indirect entrance via CO<sup>2</sup> diffusion with subsequent hydration by intracellular carbonic anhydrases (CA) (Wandernoth et al., 2010; José et al., 2015). Besides an increase in the pH<sup>i</sup> , a cytosolic HCO<sup>−</sup> 3 elevation is crucial for activation of the sperm soluble adenylyl cyclase (Okamura et al., 1985; Buck et al., 1999).

To understand how sperm pH<sup>i</sup> is regulated, it is indispensable to determine where and when it changes in individual cells. Although sperm pH<sup>i</sup> measurements in suspension have been performed using fluorescence indicators for more than three decades (Schackmann and Boon Chock, 1986; Darszon et al., 2004; Hamzeh et al., 2019), there are few reports of imaging single sperm pH<sup>i</sup> (Zeng et al., 1996; Chávez et al., 2014, 2018; González-Cota et al., 2015). All these experiments were performed with BCECF (Rink et al., 1982), the most popular fluorescent pH<sup>i</sup> indicator in cell physiology. This fluorescence probe is ratiometric but requires dual-excitation (Rink et al., 1982). Consequently, there is a time lag between two subsequent images excited by two different wavelengths and therefore, cell movement artifacts can be significant. Furthermore, BCECF is highly phototoxic to cells (Nishigaki et al., 2006), which was also confirmed in this study.

To overcome the BCECF disadvantages stated above we employed SNARF-5F acetoxymethyl ester (AM) (Liu et al., 2001) whose fluorescence spectra changes (shift of the peak wavelength) depending on pH (pKa: 7.2). This dye allowed us to perform dualemission ratiometric pH<sup>i</sup> imaging using an image splitter with a single EMCCD camera. In this report, we detail our pH<sup>i</sup> imaging setup and conditions. Furthermore, we found kinetic differences in the pH<sup>i</sup> changes induced by HCO<sup>−</sup> 3 in human and mouse spermatozoa which could suggest that HCO<sup>−</sup> 3 influx pathways are distinct in human and mouse spermatozoa.

#### MATERIALS AND METHODS

#### Materials

Dimethyl sulfoxide (DMSO, cat. D2650), ammonium chloride (NH4Cl, cat. A9434), nigericin (cat. N7143), progesterone (cat. P8783), and concanavalin A (cat. C2010) were purchased from Sigma–Aldrich. Pluronic F-127 (cat. P6867), SNARF-5F AM (5-(and-6)-carboxylic acid, acetoxymethyl ester) (cat. S23923), and 2<sup>0</sup> , 7<sup>0</sup> -Bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (BCECF AM) (cat. B1170) were obtained from ThermoFisher Scientific.

#### Biological Sample Collection Human Spermatozoa

Human spermatozoa samples were obtained from healthy donors under written informed consent and with the approval of the Bioethics Committee of the Instituto de Biotecnología, Universidad Nacional Autónoma de México (IBt-UNAM). Only ejaculates that fulfilled the World Health Organization guidelines were used in all the experiments (Cao et al., 2010). Motile cells were recovered using the swim-up technique in HTF medium (in mM: 90 NaCl, 4.7 KCl, 1.6 CaCl2, 0.3 KH2PO4, 1.2 MgSO4, 2.8 glucose, 0.3 pyruvic acid, 23.8 HEPES, and 21.4 lactic acid, 25 NaHCO3) pH 7.4 (Mata-Martínez et al., 2013). Briefly, 400 µl of liquefied semen was placed in glass test tubes and 1 ml HTF medium was carefully added on the top of the semen without mixing the phases. Samples were incubated for 1 h at 37◦C under 5% CO2. The upper layer (700 µl) with motile spermatozoa was then collected. Cell density was determined using a Makler chamber and adjusted to 10 × 10<sup>6</sup> spermatozoa/ml.

#### Mouse Spermatozoa

All experimental protocols were approved by the Bioethics Committee of the IBt-UNAM). Motile spermatozoa were obtained from epididymal cauda of 3-month-old CD-1 mouse by placing incised epididymis in an Eppendorf tube containing 1 ml of in TYH medium (in mM: 119 NaCl, 4.7 KCl, 1.7 CaCl2, 1.2 KH2PO4, 1.2 MgSO4, 5.6 dextrose, 0.5 pyruvic acid, and 20 HEPES) pH 7.4. Spermatozoa were allowed to swim-out during 15 min at 37◦C. The upper layer (800 µl) with motile spermatozoa was collected and the cell density was adjusted to 10 × 10<sup>6</sup> spermatozoa/ml using a Makler counting chamber (Irvine Scientific, Santa Ana, CA, United States).

#### In vitro Fluorescence Spectra of SNARF-5F

Fluorescence spectra of SNARF-5F were determined with a Perkin-Elmer LS 55 (Perkin-Elmer, Waltham, MA, United States) fluorescence spectrometer using the software FL WinLab version 4.00.03 for data acquisition (**Figure 1**). SNARF-5F non-permeable and AM versions were used at 20 µM final

concentration in 50 mM potassium phosphate buffer (see table in **Supplementary Figure S1**). Multiple spectra were acquired using various excitation wavelengths (405, 440, 465, 488, 510, 532, and 543 nm) and different pH<sup>e</sup> (5.5, 6.0, 6.4, 6.8, 7.0, 7.2, 7.4, 7.8, and 8.2) (**Supplementary Figure S1**).

#### SNARF-5F and BCECF Incorporation Into Spermatozoa

Motile mouse/human spermatozoa (10 × 106/ml) were incubated with 20 µM SNARF-5F AM in the presence of 0.1% pluronic F-127 during 90 min at 37◦C with 5% CO<sup>2</sup> in the dark. The cells were washed once by centrifugation at 200 × g for 5 min and resuspended with fresh medium. To obtain fluorescence spectra of SNARF-5F incorporated into spermatozoa, human spermatozoa loaded with SNARF-5F AM were diluted in the media of different pH<sup>e</sup> as described above and treated with 0.1% Triton-X 100 detergent. Fluorescence spectra of the lysed spermatozoa were acquired at excitation wavelengths 488 and 532 nm. For single cell recordings, spermatozoa were attached to Concanavalin A-treated coverslips for 2–3 min and mounted in recording chambers.

For BCECF experiments, motile mouse/human spermatozoa (10 × 106/ml) were incubated with 1 µM BCECF AM without pluronic acid during 15 min at 37◦C with 5% CO<sup>2</sup> in the dark and the unincorporated dye was removed by centrifugation as the case of SNARF-5F.

#### Imaging Setup

Single cell images were acquired using two different setups: (1) Olympus iX71 LED-light source epifluorescence microscope and (2) Olympus iX81 laser widefield/total internal reflection fluorescence (TIRF) microscope (Olympus, Japan). The LED setup was equipped with a PlanApo N 60X/1.42 oil objective and a 3A 532 nm LED coupled to Opto-LED light controller (Cairn Research, United Kingdom). The laser setup was equipped with an Apo N (TIRF) 60X/1.49 oil objective and a 488 nm laser with high speed imaging shutter. To acquire dual emission images of SNARF-5F, an image splitter OptoSplit II (Cairn Research, United Kingdom) was used for both setups, LED (with band-pass filter ET 530/30X) and laser. To acquire the images with SNARF-5F, a wide bandpass filter ET 575/50 M (channel 1) and a band-pass filter ET 640/20 M (channel 2) were employed as dual emission filters combined with a dichroic mirror DC 610lp (Chroma Technology, United States) (**Figure 2**). For BCECF experiments, a 3.15 A 465 nm LED (Luminus Devices, Woburn, MA, United States) with bandpass filter HQ 480/40X was used for excitation light, combined with a dichroic mirror (Q505lp) and an emission filter (HQ 535/50M) (Chroma Technology, United States). Each setup has a 512 × 512 Andor iXon 3 EMCCD camera (model X3 DU897E-CS0) (Oxford Instruments, United Kingdom).

Images were acquired with the software Andor iQ version 2.9.1 (LED set-up) (Oxford Instruments, United Kingdom) and Xcellence version 1.2 (laser set-up) (Olympus, Japan). Fluorescence images of both SNARF-5F and BCECF were taken with 1 × 1 binning, 5 images/s (5 ips), with an exposure time of 10 ms for the LED setup and 30 ms (laser potency 35%) for the laser setup. Images were analyzed with ImageJ version 1.52n (NIH, United States), obtaining mean fluorescence intensities, selecting heads and flagellum as regions of interest.

an emission cube (orange square) (ET575/50M, ET640/20M, and dichroic mirror DC610 LP), that divides the emission in reflected (channel 1, corresponding to emission wavelength 575 nm) or transmitted (channel 2, corresponding to emission wavelength 640 nm) images. The image splitter is coupled to the microscope on one side, and to the detector (CCD camera) on the other. Fluorescence images obtained from epifluorescence (laser set-up) microscope with 60× (Plan Apo N, 1.49 numerical aperture) objective, using 20 µM non-permeable SNARF-5F (B) or 20 µM SNARF-5F-AM loaded in human and mouse (not shown) spermatozoa in the presence of nigericin 10 µM (C) in 50 mM potassium phosphate buffers at indicated pH. (D) Ratio values were obtained (referred from panel C) at each pH<sup>e</sup> (6.0, 6.5, 7.0, 7.5, and 8.0) in human and mouse spermatozoa, using channel 1 (575 nm) and channel 2 (640 nm) fluorescence values. (E) Lineal correlation between pH<sup>e</sup> and fluorescence ratio in human and mouse spermatozoa, obtaining R <sup>2</sup> = 0.99 and 0.98, respectively. Numbers 1 (yellow) and 2 (red) to the left in panels B and C refer to channel 1 (emission 575 nm) and channel 2 (emission 640 nm). Scale bar in panels B and C is equal to 10 µm; n = 3.

## In vivo pH<sup>i</sup> Calibration

To convert fluorescence data to pH values in vitro, the following equation is commonly used:

$$\mathrm{pH} = \mathrm{p} \mathrm{K\_a} - \log\left[\frac{R - R\_{\mathrm{B}}}{R\_{\mathrm{A}} - R} \times \frac{F\_{\mathrm{B}(\lambda\_2)}}{F\_{\mathrm{A}(\lambda\_2)}}\right]$$

where R is the ratio Fλ1/Fλ<sup>2</sup> of fluorescence intensities (F) measured at two wavelengths λ<sup>1</sup> and λ<sup>2</sup> and the subscripts A and B represent the values at the acidic and basic conditions, respectively (Whitaker et al., 1991). However, it is difficult to maintain live spermatozoa in highly acidic and alkaline condition to obtain FB(λ<sup>2</sup> )/FA(λ2) values from the same cells. Therefore, we performed in vivo pH<sup>i</sup> calibration by fixing external pH (pHe) between 6.0 and 8.0 as reported previously (Grillo-Hill et al., 2014). Briefly, spermatozoa suspensions were incubated for 15 min with calibration medium (in mM: 120 KCl, 25 HEPES, 1 MgCl2, and 0.01 nigericin, at different pHe: 6.0, 6.5, 7.0, 7.5, and 8.0, adjusted with KOH). We measured the fluorescence intensity at two emissions, 575 (channel 1) and 640 nm (channel 2), always subtracting the fluorescence background value in each channel. As fluorescence ratio values (RF640/F575) have a lineal relation with pH<sup>i</sup> values (between 6.0 and 8.0) (**Figures 2D,E**), we used the following equations to estimate pH<sup>i</sup> from the fluorescence ratio values, for human: pH<sup>i</sup> = (RF640/F<sup>575</sup> + 6.96)/1.22 and for mouse: pH<sup>i</sup> = (RF640/F<sup>575</sup> + 7.52)/1.32.

#### Statistical Analysis

The results are expressed as the mean ± SEM of at least three independent experiments (three different donors or mice), with a minimum of 200 cells per condition. The data were analyzed by a comparison test between the groups, using the non-parametric Mann–Whitney U-test with 95% statistical significance. The paired tests were carried out comparing the head and the flagellum of the same cell. Additionally, the Bonferroni correction was used when multiple comparisons were made.

## RESULTS

#### Emission Spectra of SNARF-5F With Distinct Excitation Wavelengths

To perform ratiometric fluorescence measurements with a good signal to noise ratio (S/N ratio), it is important to acquire bright fluorescence images in both channels. In other words, if we detect dim fluorescence signals in one channel, the S/N ratio of the dual-emission ratio values become undesirably low even when we detect bright signals in the other channel. As spermatozoa possess a quite reduced cytoplasm, it is crucial to use an appropriate excitation wavelength and emission filters. Therefore, we first determined the fluorescence spectra of SNARF-5F at several pH<sup>e</sup> values (5.5–8.2), exciting with various wavelengths (405–543 nm). As shown in **Supplementary Figure S1**, the

longer excitation wavelength (longer than 465 nm) gives the higher fluorescence intensities in all emission wavelengths we explored (550–750 nm). Namely, 543 nm produced the highest fluorescence values.

**Figure 1** illustrates the fluorescence spectra of SNARF-5F at different pH<sup>e</sup> (5.5–8.2) excited at 488 and 532 nm. At both exciting wavelengths, the fluorescence intensities around 575 nm (the first peak) decrease when the pH increases, while those of around 640 nm (the second peak) increase at the same conditions. When exciting at 532 nm, the relative fluorescence intensities within the first peak at different pH<sup>e</sup> are much smaller than those within the second peak, as reported in the original article of SNARF-5F (Liu et al., 2001). Conversely, the relative fluorescence intensities within the two peaks became almost equal when 488 nm was used as excitation light (**Figure 1A**). In spite of this favorable feature, their absolute fluorescence intensities are small. Considering these characteristics, we selected 532 nm as the best compromise between brightness and peak balance for this study.

To evaluate the incorporation of the membrane permeant dye SNARF-5F AM into the spermatozoa, we incubated human spermatozoa with 20 µM of this dye for 90 min. After the excess dye was washed out by centrifugation, the cells were lysed with 0.1% Triton X-100 and the fluorescence spectra were acquired (**Figure 1B**). The spectra of the dye incorporated into human spermatozoa were not identical to SNARF-5F in vitro, suggesting that SNARF-5F AM was not completely hydrolyzed in the cell and/or some of the dye was bound to certain molecules of the cell. Nevertheless, SNARF-5F AM incorporated into the cell responded to pH<sup>e</sup> changes similarly to SNARF-5F AM.

#### Dual-Emission pH<sup>i</sup> Imaging System and in vivo Calibration of pH<sup>i</sup>

A conventional dual-emission fluorescence imaging setup usually is composed of an epi-fluorescence microscope, a CCD camera, and a filter wheel, which exchanges two emission filters alternatively. In this type of setup, there is always a time lag between the image in one channel and the image in the other channel. Since spermatozoa are small and motile cells, the presence of a time lag between two images of each channel is undesirable. Therefore, we used an image splitter (Kinosita et al., 1991) that allows the simultaneous capture of the images from the two channels with a single camera (**Figure 2A**). In a common configuration of dual-emission ratiometric imaging, emission lights are divided into two components (two channels) at the isosbestic point, around 595 nm in the case of SNARF-5F AM excited by 532 nm. However, because the fluorescence intensity of the first peak (575 nm) is lower than the second peak (640 nm) as described previously (**Figure 1**), we separated the emission light at 610 nm (about 15 nm longer than the isosbestic point) by a dichroic mirror. Consequently, we collected a wide range of wavelengths, 550–600 nm, as the fluorescence signals in the first channel (channel 1). Then, we collected 630–650 nm wavelengths as the longer wavelength fluorescence (channel 2). This configuration gives us comparable fluorescence intensities from the two channels without the insertion of a neutral density filter (**Figure 2A**).

**Figure 2B** shows fluorescence images (gray scale) of SNARF-5F in media at different pHs, clearly demonstrating the opposite changes of fluorescence intensity between Channel 1 and Channel 2. Since fluorescence spectra of SNARF-5F and SNARF-5F AM incorporated into human spermatozoa show a slight difference (**Figures 1A,B**), we performed in vivo calibration using human spermatozoa to convert the ratio fluorescence values into pH<sup>i</sup> values. To perform in vivo calibration, spermatozoa pH<sup>i</sup> was equilibrated to the pH<sup>e</sup> using high K<sup>+</sup> (120 mM) media in the presence of 10 µM nigericin (an ionophore that facilitates K+/H<sup>+</sup> exchange across the lipid bilayer). **Figure 2C** shows fluorescence images (pseudo color) of human spermatozoa, whose pH<sup>i</sup> was fixed at different pH<sup>e</sup> (6.0–8.0).

The mean ratio values of fluorescence intensities of the two channels (F640/F575) in distinct pH<sup>i</sup> are summarized in **Figure 3C** and these ratio values are plotted as a function of pH<sup>i</sup> (**Figure 2D**). The ratio values increase proportionally to pH<sup>i</sup> between 6.0 and 8.0 with excellent linearity (human spermatozoa: RF640/F<sup>575</sup> = 1.22 × pH<sup>i</sup> – 6.96, R <sup>2</sup> = 0.99; mouse spermatozoa: RF640/F<sup>575</sup> = 1.32 × pH<sup>i</sup> – 7.52, R <sup>2</sup> = 0.98) (**Figure 2E**).

### Phototoxicity of SNARF-5F to Spermatozoa

BCECF is known to be quite phototoxic to spermatozoa and this effect can be easily detected as flagellar beat attenuation and as a decrease in the fluorescence intensity (photo-bleaching) during the intense exposure of excitation light (Nishigaki et al., 2006; González-Cota et al., 2015). In this study, we confirmed the phototoxic effect of BCECF on sperm using the same setup utilized for SNARF-5F (**Supplementary Figure S2**). Particularly, the 488 nm laser excitation attenuated the flagellar beat of human and mouse spermatozoa after around 60 and 20 s illuminations, respectively. Subsequently, notable photobleaching of BCECF was observed in both human and mouse spermatozoa (**Supplementary Figures S2A,C**). On the other hand, LED illumination caused less photobleaching in human spermatozoa (**Supplementary Figure S2B**), but certain level of photobleaching was still observed in 40% of mouse spermatozoa (**Supplementary Figure S2D**). This result suggests that mouse sperm are more susceptible to oxidative stress than human spermatozoa.

In contrast, SNARF-5F incorporated into spermatozoa is much less toxic to the cells than BCECF (**Figure 3**). The fluorescence intensities excited by LED and 488 nm laser (epifluorescence mode) did not cause photo-bleaching of the dye during our experimental periods (5 ips for 250 s) (**Figures 3A,B**). However, we observed a slight photobleaching of SNARF-5F when excited by the 488 nm laser in the TIRF configuration (**Figure 3C**). This photobleaching was negligible when we reduced the frequency of image acquisition from 5 to 2.5 ips (data not shown).

#### Comparison of Epi-Fluorescence and TIRF Images

In our previous study of pH<sup>i</sup> imaging (epi-fluorescence mode) of sea urchin spermatozoa using BCECF, fluorescence intensities of

the flagellum were much lower than those of the heads and their fluorescence signals were noisy with a poor S/N ratio (González-Cota et al., 2015). We thought that the use of TIRF would improve this aspect, avoiding the saturation of fluorescence signal in the head. However, the difference of SNARF-5F images between the two configurations (epi-fluorescence and TIRF) was relatively small in both human (**Figure 4A**) and mouse spermatozoa (**Figure 4B**). Particularly, in mouse spermatozoa, the TIRF fluorescence signals in the head and the flagellum (primarily mid piece) are very similar to the epi-fluorescence images. This result is probably due to the thin hook-like shape of the mouse sperm head. As a consequence, an important difference between the two systems (epi-fluorescence and TIRF) is not significant for mouse spermatozoa.

## Spermatozoa Responses to pH<sup>i</sup> Manipulation

Using the epi-fluorescence configuration with the LED as a light source, we acquired fluorescence images upon pH<sup>i</sup> manipulations. During image acquisition, we added HTF or TYH medium as control in human and mouse spermatozoa, respectively. As additional control, 10 mM NH4Cl and 5 mM HCl were added to increase and reduce the pH<sup>i</sup> , respectively. The upper panels of **Figures 5A,B** show human sperm fluorescence signals from the two channels in the head and the flagellum, respectively. Changes of fluorescence intensities were observed during the additions even in control conditions (indicated with arrows). Also, fluorescence signals from some cells are noisy probably due to the continuous movement associated to the flagellar beat. Once the dual emission signals were converted into the ratio and pH<sup>i</sup> values (**Figures 5A,B**, lower panels), the problems of addition artifacts and movement were eliminated in the both regions, demonstrating the advantage of the dual-emission ratiometric imaging. Additionally, the effects of NH4Cl and HCl can be clearly observed as an increase and a decrease of the ratio and the pH<sup>i</sup> , respectively. **Figure 6** basically demonstrates the same results as **Figure 5** but using mouse spermatozoa. In this experiment, the fluorescence signal of the flagellum arises mainly from the mid piece since mouse spermatozoa flagellum is much longer than that of human spermatozoa and therefore it is difficult to capture the

image of the entire flagellum of mouse spermatozoa. In these experiments, the average pH<sup>i</sup> of non-capacitated human and mouse spermatozoa was 6.72 ± 0.19 (SEM) and 6.63 ± 0.23 (SEM), respectively; n = 3.

#### Response to HCO<sup>−</sup> 3

To obtain new insights of mammalian spermatozoa pH<sup>i</sup> regulation, we determined the effect of HCO<sup>−</sup> 3 (10 and 25 mM) on pH<sup>i</sup> in non-capacitated human and mouse spermatozoa using our dual-emission imaging system (**Figure 7**). In these experiments, we confirmed that HCO<sup>−</sup> 3 increases the pH<sup>i</sup> of human (**Figure 7A**) and mouse (**Figure 7B**) spermatozoa, in a concentration-dependent manner. We did not observe statistical differences in the pH<sup>i</sup> increase induced by HCO<sup>−</sup> 3 between human and mouse spermatozoa (**Figure 7C**). However, we found a significant difference in the kinetics of the pH<sup>i</sup> increase between the two species. Namely, HCO<sup>−</sup> 3 rapidly increases the pH<sup>i</sup> of mouse spermatozoa, and the time to reach 50% of the maximum pH<sup>i</sup> increase (t50) was around 10 s. In contrast, HCO<sup>−</sup> 3 increases human sperm pH<sup>i</sup> gradually with a longer t<sup>50</sup> (40 s) in our experimental conditions (**Figure 7D**). Moreover, the pH<sup>i</sup> increase in human spermatozoa was slightly, but significantly, slower in the flagellum compared to the head with 10 and 25 mM HCO<sup>−</sup> 3 additions.

## Response to Progesterone in Human Spermatozoa

In the literature, there is some controversy about the effect of progesterone on human sperm pH<sup>i</sup> . A decrease (Garcia and Meizel, 1996; Cross and Razy-Faulkner, 1997), no change (Fraire-Zamora and González-Martínez, 2004) or a slow increase (Hamamah et al., 1996) in pH<sup>i</sup> have been reported by different groups in response to this hormone. Therefore, we determined the effect of different progesterone concentrations on pH<sup>i</sup> in human spermatozoa. **Figure 8** shows that progesterone at 500 nM (I), 1 µM (II), and 10 µM (III) did not change pH<sup>i</sup> neither in the head (**Figure 8A**) nor in the flagellum (**Figure 8B**) of these cells. As a control we tested 1 µM monensin, a Na<sup>+</sup> ionophore that exchanges Na+/H<sup>+</sup> (Babcock, 1983). As anticipated, this ionophore alkalized pH<sup>i</sup> in these cells in both head (**Figure 8A**, IV) and flagellum (**Figure 8B**, IV).

## DISCUSSION

## Advantages of the New System to Determine Spermatozoa pH<sup>i</sup>

In this study we established a dual-emission ratiometric imaging system using SNARF-5F AM, which has negligible photo-toxicity compared to BCECF (**Figure 3** and **Supplementary Figure S2**). Our system allows determining mammalian spermatozoa (head and flagellum) pH<sup>i</sup> with minimum artifacts associated to cell movements and focus alteration upon addition or exchange of bath solutions (**Figures 5**–**7**). Commonly, the ratio of dual fluorescence signals utilizes the dye isosbestic point (595 nm in our condition). However, the first peak fluorescence intensity (575 nm) of SNARF-5F excited at 532 nm is much smaller than the second peak (640 nm). Therefore, we divided the fluorescence at 610 nm, 15 nm longer than the isosbestic point, and employed a wide band-pass filter (550–600 nm) for Channel 1. In this configuration, fluorescence signals of the two channels are comparable (**Figure 2**), which is a critical point to obtain the ratio values with a good S/N ratio. This type of optical filter configuration (division of fluorescence signals not at the isosbestic point) could be applied to other dual-emission indicators such as GEM-GECO (Zhao et al., 2011) and Asante

Calcium Red (Hyrc et al., 2013) because their dual-emission signals are quite asymmetric.

## pH<sup>i</sup> Calibration

We observed a slight difference between the fluorescence spectra of SNARF-5F in vitro and inside human spermatozoa. Therefore, in order to convert the fluorescence emission values into the pH<sup>i</sup> , we performed an in vivo pH<sup>i</sup> calibration with human and mouse spermatozoa using a high K<sup>+</sup> solution combined with nigericin in order to equal pH<sup>i</sup> to the pHe. This protocol is based on the assumption that the cytoplasmic K<sup>+</sup> concentration is 120 mM in human and mouse spermatozoa, as determined in bovine spermatozoa (Babcock, 1983). Therefore, depending on the real cytoplasmic K<sup>+</sup> concentration in human and mouse spermatozoa, the absolute pH<sup>i</sup> values could be different. In

our conditions, we determined that the pH<sup>i</sup> value in noncapacitated human and mouse spermatozoa is 6.72 ± 0.19 and 6.63 ± 0.23, respectively. These values were measured in the head, but no significant differences were observed in the flagellum (see below). There are several reports of pH<sup>i</sup> determinations (most of them in cell population experiments and a few using single cell determination) of non-capacitated spermatozoa from distinct mammalian species: 6.24 (Parrish et al., 1989a) and 6.7 (Vredenburgh-Wilberg and Parrish, 1995) in bovine spermatozoa, 6.55 (Balderas et al., 2013), 6.54 (Zeng et al., 1996), and 6.8 (Carlson et al., 2007) in mouse sperm, and 6.7 (Hamamah et al., 1996; Fraire-Zamora and González-Martínez, 2004) and 6.94 (Cross and Razy-Faulkner, 1997) in human spermatozoa. Our findings that pH<sup>i</sup> values of non-capacitated mammalian spermatozoa are >6.5 are consistent with the

FIGURE 7 | HCO<sup>−</sup> 3 increased pH<sup>i</sup> in a concentration-dependent manner in both, head and flagellum, regions using human and mouse spermatozoa. Representative recordings from human (A) and mouse (B) spermatozoa, measuring pH<sup>i</sup> using 20 µM SNARF-5F in head (Top) and flagellum (Bottom) regions. The perfused addition of medium (HTF and TYH for human and mouse, respectively) (left, control in gray rectangle), 10 (center in green rectangle) or 25 mM (right in green rectangle) HCO<sup>−</sup> 3 are showed. As positive controls, perfused addition of 10 mM NH4Cl (red rectangle) and 5 mM HCl (purple rectangle) are showed in each panel. Traces in each panel show representative single cell pH<sup>i</sup> . Same color in both, head and flagellum, indicate to the same cell. Maximum change in pH<sup>i</sup> (1pH<sup>i</sup> ) (C) and average of t<sup>50</sup> (D), time to reach 50% of the maximum fluorescent intensity, before and after 10 (blue bars), 25 (green bars) mM HCO<sup>−</sup> 3 addition, in head (shaded) or flagellum (diagonal lines) regions. The bars in C,D indicated means ± SEM. Different letters indicate significant differences at the p ≤ 0.05 level, according to Mann–Whitney U-test; n = 5.

single cell pH<sup>i</sup>

fcell-07-00366 January 8, 2020 Time: 18:33 # 10

. The same color at both emission wavelengths and in both regions (head and flagellum) corresponds to the same cell; n = 3.

report that detergent-demembranated bovine spermatozoa do not exhibit motility at pH 6.5, although they are highly motile at pH 7.0 (Ho et al., 2002).

## Regional pH<sup>i</sup> Difference in the Head and the Flagellum

We did not observe significant differences between the head and the flagellum in the basal pH<sup>i</sup> in non-capacitated spermatozoa, although the head pH<sup>i</sup> tends to be slightly higher than the flagellar pH<sup>i</sup> both in human (6.72 ± 0.19 and 6.69 ± 0.24, respectively) and mouse spermatozoa (6.63 ± 0.23 and 6.60 ± 0.26, respectively). Our results are similar to those reported in bovine spermatozoa (Vredenburgh-Wilberg and Parrish, 1995). In general, the epifluorescent signal from an indicator incorporated into the sperm head is generally much higher than in the flagellum, independently of the species. Therefore, we examined if TIRF microscopy would reduce the fluorescence difference between the head and the flagellum. However, we did not observe significant differences nor advantages of TIRF microscopy compared to epifluorescence microscopy (**Figure 4**) either in mouse or human spermatozoa for measuring pH<sup>i</sup> . With these data, we can conclude that epifluorescence microscopy with SNARF-5F AM allows performing reliable single spermatozoa pH<sup>i</sup> imaging with a good S/N ratio in both spermatozoa head and flagellum (mid piece of flagellum in the case of mouse).

#### Difference of pH<sup>i</sup> Responses to HCO<sup>−</sup> 3 Between Human and Mouse Spermatozoa

HCO<sup>−</sup> 3 is an essential ion for mammalian sperm to acquire the ability to fertilize the oocyte (Lee and Storey, 1986). In fact, the HCO<sup>−</sup> 3 concentrations in rabbit uterine and tubal fluids are approximately twice as high as in the blood plasma, which results in pH values of 7.4 and 8.1–8.3, respectively (Vishwakarma, 1962). In rhesus monkeys, the pH and HCO<sup>−</sup> 3 concentration in the oviduct lumen change during the menstrual cycle. Namely, these values are similar to those of the blood plasma during the follicular phase, but they suddenly increase concomitantly with ovulation (Maas et al., 1977). This observation supports the importance of HCO<sup>−</sup> 3 for fertilization in mammals. The principal role of cytoplasmic HCO<sup>−</sup> 3 in mammalian spermatozoa is considered to be the activation of the soluble adenylyl cyclase, which increases cAMP (Okamura et al., 1985; Buck et al., 1999; Chen et al., 2000), leading to protein kinase A (PKA) stimulation. The enhanced PKA activity increases flagellar beat frequency (Wennemuth, 2003) and elevates CatSper activity (Carlson et al., 2003; Orta et al., 2018), among many other things.

In this work, we observed that HCO<sup>−</sup> 3 elevates the pH<sup>i</sup> in both human and mouse spermatozoa (**Figure 7D**). In contrast, Carlson et al. (2007) reported that HCO<sup>−</sup> 3 did not induce a pH<sup>i</sup> increase in mouse spermatozoa. Interestingly, we found a difference in the kinetics of HCO<sup>−</sup> 3 -induced pH<sup>i</sup> increase between the two species (**Figure 7D**), namely a faster increase in mouse compared to human spermatozoa, but of similar magnitude (**Figure 7C**). So far, several mechanisms have been reported for HCO<sup>−</sup> 3 influx, such as the NBC (Demarco et al., 2003), Cl−/HCO<sup>−</sup> 3 exchangers (Chavez et al., 2012), and the CFTR channel (Hernández-González et al., 2007; Xu et al., 2007). However, the physiological relevance of each transporter is unclear as well as differences between the two species. In addition to HCO<sup>−</sup> 3 transporters, CO<sup>2</sup> diffusion with subsequent hydration by intracellular CA contributes to an increase in cytoplasmic HCO<sup>−</sup> 3 concentration (Carlson et al., 2007; Wandernoth et al., 2010; José et al., 2015). Curiously, a general CA inhibitor, ethoxyzolamide, potently affects human but not mouse sperm motility (José et al., 2015), suggesting a difference in the involvement of CAs in the motility of the two species. Another

explanation is that human spermatozoa may have higher pH buffering capacity than mouse spermatozoa. This might be correlated to the time required for capacitation (>6 h in human compared to 1–2 h in mouse spermatozoa). Indeed, the pH<sup>i</sup> of mammalian spermatozoa studied so far increases around 0.14–0.4 units during capacitation (Parrish et al., 1989b; Vredenburgh-Wilberg and Parrish, 1995; Hamamah et al., 1996; Zeng et al., 1996; Cross and Razy-Faulkner, 1997; Fraire-Zamora and González-Martínez, 2004; Balderas et al., 2013). A significant part of this pH<sup>i</sup> change can be attributed to the HCO<sup>−</sup> 3 influx into the cell. Therefore, further studies are required for a better understanding of the mechanism of HCO<sup>−</sup> 3 -induced pH<sup>i</sup> increase during capacitation. The pH<sup>i</sup> imaging system established in this study should contribute to this issue.

#### Effect of Progesterone in Human Spermatozoa

Progesterone increases [Ca2+]<sup>i</sup> in human spermatozoa at concentrations as low as 300 nM, through CatSper activation (Tesarik et al., 1992; Harper et al., 2003; Achikanu et al., 2018). Recently it was described that the progesterone receptor in these cells is a α/β hydrolase domain-containing protein (ABHD2), which depletes the endocannabinoid 2-arachinoylglycerol (2AG) from membrane and removes CatSper inactivation (Miller et al., 2016; Mannowetz et al., 2017). In contrast, there is inconsistency regarding how progesterone affects pH<sup>i</sup> . Some groups suggest that this hormone acidifies, others that it alkalizes or does not induce pH<sup>i</sup> changes (Garcia and Meizel, 1996; Hamamah et al., 1996; Cross and Razy-Faulkner, 1997; Fraire-Zamora and González-Martínez, 2004). In the present work, progesterone did change pH<sup>i</sup> in human spermatozoa even at concentrations as high as 10 µM (**Figure 8**). Our result supports that progesterone activates CatSper in a pH-independent manner, possibly exclusively via ABHD2-2AG.

Progesterone-induced Ca2<sup>+</sup> influx through CatSper may affect the activity of transporters and enzymes that can affect pH<sup>i</sup> such as PMCA (Wennemuth et al., 2003; Okunade et al., 2004) and NOX5 (Baker and Aitken, 2004; Musset et al., 2012). Both PMCA and NOX5 may acidify pH<sup>i</sup> when they are activated, namely when the [Ca2+]<sup>i</sup> is high. However, the pH<sup>i</sup> acidification together with the membrane potential depolarization caused by Ca2<sup>+</sup> influx through CatSper and electron efflux through NOX5 could activate Hv1 channel (Lishko et al., 2010; Berger et al., 2017) and may rapidly neutralize the acidification (alkalize the pHi). Depending on the experimental conditions, one activity (acidifying or alkalizing) may exceed the other when progesterone stimulates human CatSper. This may account for part of the discrepancies regarding human sperm pH<sup>i</sup> responses to progesterone. Further studies are required to confirm this hypothesis.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

## ETHICS STATEMENT

The studies involving human participants/donors were reviewed and approved by the Bioethics Committee of the Instituto de Biotecnología. The donors provided their written informed consent to participate in this study. The animal study was reviewed and approved by the Bioethics Committee of the Instituto de Biotecnología.

## AUTHOR CONTRIBUTIONS

TN conceived the project. JC performed most of the experiments and prepared the figures. All authors proposed the experiments, discussed the results, and wrote, revised, and approved the manuscript.

## FUNDING

This work was supported by the PAPIIT DGAPA (Grant Numbers IA200419 to JC, IN200919 to AD, IN202519 to CT, and IN205719 to TN) and CONACyT (Fronteras 71).

## ACKNOWLEDGMENTS

We thank Gastón Contreras from the National Laboratory in Advanced Microscopy Facilities, in addition to José L. De la Vega-Beltrán, Yoloxóchitl Sánchez-Guevara, and Paulina Torres for technical assistance. Also, we would like to thank Shirley Ainsworth for library services, and the people of the animal facilities Elizabeth Mata, Graciela Cabeza, and Sergio González. We acknowledge Juan Manuel Hurtado, Roberto Rodríguez, Omar Arriaga, and Arturo Ocádiz for computer services.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00366/ full#supplementary-material

FIGURE S1 | Spectral characterization of the pH sensitive dye SNARF-5F. Representative emission spectra in 405, 440, 465, 488, 510, 532, and 543 nm excitation wavelengths, in 50 mM potassium phosphate buffers at the proportions indicated in Table 1, obtaining pHe: 5.5, 6.0, 6.4, 6.8, 7.0, 7.2, 7.4, 7.8, and 8.2. The lines are representative fluorescence spectra at indicated excitation wavelength in each pHe; n = 3.

FIGURE S2 | Single cell pH<sup>i</sup> measurements with BCECF using laser of led excitation causes significant photobleaching and a reduced response to alkalization and acidification. Representative normalized recordings using BCECF-loaded spermatozoa in human (A,B) and mouse (C,D). Laser (A,C) and LED (B,D) were used as the excitation light source. Arrows indicate the manual addition of 10 mM NH4Cl and 5 mM HCl in each panel. Each trace represents the response of a single cell; n = 3.

## REFERENCES

fcell-07-00366 January 8, 2020 Time: 18:33 # 12



**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Chávez, Darszon, Treviño and Nishigaki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Membrane Potential Determined by Flow Cytometry Predicts Fertilizing Ability of Human Sperm

Lis C. Puga Molina<sup>1</sup> , Stephanie Gunderson<sup>1</sup> , Joan Riley<sup>1</sup> , Pascal Lybaert<sup>2</sup> , Aluet Borrego-Alvarez<sup>1</sup> , Emily S. Jungheim<sup>1</sup> and Celia M. Santi<sup>1</sup> \*

<sup>1</sup> Department of Obstetrics & Gynecology, Washington University School of Medicine, St Louis, MO, United States, <sup>2</sup> Laboratory of Experimental Hormonology, School of Medicine, Université Libre de Bruxelles, Brussels, Belgium

Infertility affects 10 to 15% of couples worldwide, with a male factor contributing up to 50% of these cases. The primary tool for diagnosing male infertility is traditional semen analysis, which reveals sperm concentration, morphology, and motility. However, 25% of infertile men are diagnosed as normozoospermic, meaning that, in many cases, normalappearing sperm fail to fertilize an egg. Thus, new information regarding the mechanisms by which sperm acquire fertilizing ability is needed to develop a clinically feasible test that can predict sperm function failure. An important feature of sperm fertilization capability in many species is plasma membrane hyperpolarization (membrane potential becoming more negative inside) in response to signals from the egg or female genital tract. In mice, this hyperpolarization is necessary for sperm to undergo the changes in motility (hyperactivation) and acrosomal exocytosis required to fertilize an egg. Human sperm also hyperpolarize during capacitation, but the physiological relevance of this event has not been determined. Here, we used flow cytometry combined with a voltage-sensitive fluorescent probe to measure absolute values of human sperm membrane potential. We found that hyperpolarization of human sperm plasma membrane correlated positively with fertilizing ability. Hyperpolarized human sperm had higher in vitro fertilization (IVF) ratios and higher percentages of acrosomal exocytosis and hyperactivated motility than depolarized sperm. We propose that measurements of human sperm membrane potential could be used to diagnose men with idiopathic infertility and predict IVF success in normozoospermic infertile patients. Patients with depolarized values could be guided toward intracytoplasmic sperm injection, preventing unnecessary cycles of intrauterine insemination or IVF. Conversely, patients with hyperpolarized values of sperm membrane potential could undergo only conventional IVF, avoiding the risks and costs associated with intracytoplasmic sperm injection.

Keywords: membrane potential, human, sperm, IVF, normozoospermic infertility, flow cytometry, capacitation

## INTRODUCTION

Infertility, defined as the inability of a couple to achieve pregnancy after 12 months of unprotected intercourse, affects between 10 and 15% of couples around the world (Sharma et al., 2013). Many such couples turn to in vitro fertilization (IVF), which has been used to conceive over 6.5 million babies. Approximately 50% of infertility cases are due to a male factor (Kumar and Singh, 2015).

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

Elisabetta Baldi, University of Florence, Italy Katerina Komrskova, Academy of Sciences of the Czech Republic (ASCR), Czechia

> \*Correspondence: Celia M. Santi santic@wustl.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 15 August 2019 Accepted: 20 December 2019 Published: 21 January 2020

#### Citation:

Puga Molina LC, Gunderson S, Riley J, Lybaert P, Borrego-Alvarez A, Jungheim ES and Santi CM (2020) Membrane Potential Determined by Flow Cytometry Predicts Fertilizing Ability of Human Sperm. Front. Cell Dev. Biol. 7:387. doi: 10.3389/fcell.2019.00387

To diagnose these men, assisted reproduction specialists rely on semen analysis, which provides information about sperm concentration, morphology, and motility. However, this method does not reveal abnormalities in sperm from some infertile men (Bracke et al., 2018), who are then described as having normozoospermic idiopathic infertility. One possibility is that sperm from these men are unable to fertilize an egg because they are unable to capacitate, a process in which sperm become hyperactive and prepared to undergo acrosomal exocytosis. Together, hyperactivation and acrosomal exocytosis allow sperm to bind to and fuse with an oocyte (Yanagimachi, 1994). In natural pregnancies, capacitation is triggered by factors in the female reproductive tract (Austin, 1951; Chang, 1951). In IVF, sperm is capacitated by incubation in a chemically defined media containing Ca++, HCO<sup>3</sup> <sup>−</sup>, energy sources, and a cholesterol acceptor. Given that capacitation is required for fertilization, a test that could assess the ability of sperm to undergo this process could be a valuable addition to IVF diagnostics.

In many species, sperm capacitation is accompanied by sperm plasma membrane hyperpolarization (an increase in intracellular net negative charge) (Zeng et al., 1995; Hernández-González et al., 2007; López-González et al., 2014; Escoffier et al., 2015). In mouse sperm, capacitation-associated hyperpolarization is largely driven by activation of the sperm-specific potassium (K+) channel SLO3 (Santi et al., 2010; Chávez et al., 2013). Slo3 knockout mice are infertile, and their sperm are unable to undergo hyperactivation or acrosomal exocytosis (Santi et al., 2010; Zeng et al., 2011). Human sperm also hyperpolarize during capacitation (López-González et al., 2014), and two studies reported that more depolarized human sperm membrane potentials values were associated with lower fertility (Calzada and Tellez, 1997; Brown et al., 2016). However, these studies did not address the effect of changes of sperm membrane potential on hyperactivation or acrosomal exocytosis. In addition, the methods used in both studies to assess membrane potential are technically difficult and not suitable to be implemented in a clinical setting. Other methods that evaluate the capacitating state of the sperm, such as the CapScore test, require staining, counting, and correctly identifying staining patterns in more than 150 individual sperm (Moody et al., 2017), which also is challenging to implement clinically.

Here, we used flow cytometry in combination with the voltage-sensitive fluorescent dye DiSC3(5) to create a calibration curve and measure absolute sperm membrane potential values. Using this method, we found that human sperm membrane hyperpolarization correlated with sperm fertility capacity in IVF patients. We propose that our method could be used clinically to diagnose male infertility and predict IVF success of normozoospermic patients.

#### MATERIALS AND METHODS

#### Reagents

All reagents to prepare non-commercial Human Tubal Fluid (HTF) and Toyoda–Yokoyama–Hosi (TYH) media, as well as valinomycin, FITC-labeled Pisum sativum agglutinin, Tween 20, 2<sup>0</sup> ,70 -Bis-(2-Carboxyethyl)-5-(and-6)- Carboxyfluorescein, Acetoxymethyl Ester (BCECF-AM), and calcium ionophore A23187, were from Sigma (St. Louis, MO); 3, 3<sup>0</sup> -dipropyl- thiadicarbocyanine iodide [DiSC3(5)] was from Invitrogen (Carlsbad, CA, United States); Hoechst 33342 was from Cayman Chemicals, (Ann Arbor, MI, United States); paraformaldehyde was from Electron Microscopy Sciences (Hatfield, PA, United States); Dulbecco's Phosphate Buffered Saline (DPBS) was from GIBCO Life Technologies (Gaithersburg, MD, United States); and Quinn's Advantage Human (HTF) fertilization media and human serum albumin were from CooperSurgical (Trumbull, CT, United States).

#### Mouse Samples and Ethics Statement

Washington University School of Medicine (WUSM) policy states that all research involving animals be conducted under humane conditions, with appropriate regard for animal welfare. WUSM is a registered research facility with the United States Department of Agriculture (USDA) and is committed to comply with the Guide for the Care and Use of Laboratory Animals (Department of Health and Human Services) and the provisions of the Animal Welfare Act (USDA and all applicable federal and state laws and regulations). The WUSM Animal Care and Use Committee ensures compliance with all applicable federal and state regulations for the purchase, transportation, housing, and research use of animals. WUSM has filed appropriate assurance of compliance with the Office for the Protection of Research Risks of the National Institute of Health. The WUS Animal Care and Use Committee approved the method of euthanasia for mice: CO<sup>2</sup> asphyxiation followed by rapid cervical dislocation. These methods are consistent with the recommendation of the panel on euthanasia of the American Veterinary Medical Association and YARC Standard Euthanasia guidelines for rodents. Caudal epididymal sperm were collected from 90-day-old Slo3 knock-out (generated as previously described [Santi et al., 2010]) or wildtype mice.

#### Human Samples and Ethics Statement

This study was approved by the Washington University Institutional Review Board and was performed in collaboration with the Washington University Fertility and Reproductive Medicine Center. Semen samples were obtained via masturbation after 3–5 days of abstinence in private collection rooms at the fertility clinic. Only donors and conventional IVF patients' samples that met the World Health Organization [WHO] (2010) (≥32% progressive motility; ≥40% total motility; ≥15 × 10<sup>6</sup> cells/ml). For studies involving semen donors, de-identified human sperm samples were obtained within 2 h of collection from men who were asked to have a semen analysis performed.

Couples undergoing IVF met inclusion criteria if their oocytes were subjected to conventional in vitro fertilization or split insemination [half intracytoplasmic sperm injection (ICSI), half conventional in vitro fertilization] or conventional in vitro fertilization, and if the female partner had a diagnosis of unexplained infertility, tubal infertility, or uterine factor infertility. Patients were excluded from this study if they had male factor infertility. Additionally, to assure good-quality

oocytes for this study, patients were excluded if females had polycystic ovarian syndrome, endometriosis, anovulation, or diminished ovarian reserve. The average age of the 49 healthy men included in this study was 34.63 (range 26–48) years. On the day of oocyte retrieval, male partners gave written informed consent in accordance with the Declaration of Helsinki. At the time of consent, male partners also completed a brief health questionnaire. For IVF, around 14 × 10<sup>6</sup> /ml sperm were incubated overnight with mature oocytes including the zona pellucida and cumulus cells. For ICSI, oocytes with zona pellucida and without cumulus cells were used. Sperm samples left over after IVF were obtained approximately 4 h after initial collection and incubated overnight at 37◦C with 5% CO2. IVF cycle data including fertilization rate were collected for analysis. The fertilization rate was calculated as the number of normally fertilized mature oocytes with two pronuclei divided by total number of inseminated mature oocytes. The Washington University Fertility and Reproductive Medicine Center defines successful IVF as a fertilization rate >70% (fertilization ratio >0.7); this cut-off is determined by calculating the mean fertilization ratio in conventional IVF cases quarterly and yearly.

For studies involving frozen human samples, five vitrified washed normozoospermic semen samples that were used to conceive a pregnancy were obtained from the Washington University Reproductive Medicine Center cryobank. The samples were produced by anonymous donors, aliquoted, frozen, and used for both intrauterine insemination and IVF. Remaining aliquots of samples that were used to conceive a pregnancy were donated for the purposes of research.

## Media

Non-commercial HEPES-buffered HTF (4.67 mM KCl, 0.31 mM KH2PO2, 90.69 mM NaCl, 1.20 mM MgSO4, 2.78 mM glucose, 1.6 mM CaCl2, 0.51 mM sodium pyruvate, 60 mM sodium lactate, 23.8 mM HEPES, and 10 µg/ml Gentamicin, (pH 7.4 with NaOH) was used to measure human sperm membrane potential. Non-commercial HEPES-buffered HTF was supplemented the same day with 25 mM NaHCO<sup>3</sup> and 0.5% w/v BSA A-7906 (Sigma) for capacitating conditions, and only with 0.5% w/v BSA for non-capacitating conditions. IVF samples were processed in commercial Quinn's Advantage Human (HTF) fertilization media [CooperSurgical (Trumbull, CT, United States)] supplemented with 3 mg/ml human serum albumin (CooperSurgical) and pre-incubated overnight at 37◦C with 5% CO<sup>2</sup> to equilibrate pH. The same media was used to capacitate IVF and donor sperm samples.

Mouse capacitating modified (TYH) medium [119.30 mM NaCl, 4.70 mM KCl, 1.71 mM CaCl2, 1.20 mM KH2PO4, 1.20 mM MgSO4, 0.51 mM sodium pyruvate, 5.56 mM glucose, 20 mM HEPES, 15 mM NaHCO<sup>3</sup> and 5 mg/ml BSA A-2153 (pH 7.4 with NaOH)] was used to capacitate mouse sperm.

#### Human and Mouse Sperm Capacitation

Ejaculated human sperm were allowed to swim-up in commercial or non-commercial HTF capacitating medium for 1 h at 37◦C with CO2. The highly motile sperm recovered after swim up were washed, centrifuged for 5 min at 400 × g, and incubated overnight at 37◦C with 5% CO2. For experiments involving comparisons between non-capacitating and capacitating conditions, human sperm were allowed to swim-up in noncapacitating HTF medium at 37◦C for 1 h. Then the sample was divided into two, and an equal volume of non-capacitating medium or 2-fold concentrated BSA and NaHCO<sup>3</sup> were added to prepare non-capacitating or capacitating media, respectively. Then, the samples were incubated overnight at 37◦C with 5% CO<sup>2</sup> for capacitating conditions, or at 37◦C for non-capacitating conditions.

Frozen human sperm samples were thawed for 20 min at room temperature, resuspended in 1.5 ml final volume of commercial Quinn's Advantage Human (HTF) fertilization media (CooperSurgical) medium and washed twice with centrifugation at 400 × g for 5 min. Sperm were allowed to swimup in 300 µl of media with 5% CO<sup>2</sup> at 37◦C for 1 h. Highly motile sperm were recovered from the supernatant. Half were used to determine membrane potential, and half were incubated overnight with 5% CO<sup>2</sup> at 37◦C for to determine membrane potential after capacitation.

Mice were euthanized, and cauda epididymal sperm were placed in 1 ml of modified TYH medium. After 15 min at 37◦C (swim-out), suspended sperm were collected. Sperm concentration was adjusted to a final maximum concentration of 10 × 10<sup>6</sup> cells/ml and incubated for 75 min at 37◦C for capacitation.

## Determination of Membrane Potential by Flow Cytometry

After overnight incubation, human sperm samples were centrifuged at 400 × g for 5 min and resuspended in noncapacitating HTF medium and 0.1 µM BCECF-AM. The samples were covered to be protected from light, incubated at 37◦C for 10 min, washed again to eliminate non-incorporated dye, and resuspended in non-capacitating HTF media with 25 mM NaHCO3. Before recording, 5 nM DiSC3(5) was added, and data were recorded as individual cellular events on a FACSCanto II TM cytometer (Becton Dickinson). Forward scatter area (FSC-A) and side-scatter area (SSC-A) data were collected from 20,000 events per recording. Threshold levels of two-dimensional dot plot SSC-A vs. side-scatter height (SSC-H) were set for gating singlets. BCECF fluorescence was detected with the filter for fluorescein isothiocyanate (FITC; 530/30) and used to exclude non-viable cells as previously described (Puga Molina et al., 2017). Positive cells for DiSC3(5) fluorescence was detected with the filter for allophycocyanin (APC; 585/40). Because DiSC3(5) is positively charged, cell hyperpolarization increases dye incorporation and intracellular fluorescence. Recordings were initiated after reaching steady-state fluorescence (1–3 min).

Mouse sperm samples were centrifuged at 200 × g for 10 min and resuspended in TYH media without BSA. Before recording, 5 nM DiSC3(5) and 900 nM Hoechst 33342 were added. Data were recorded as individual cellular events on a FACSCanto II TM cytometer (Becton Dickinson).

FSC-A and SSC-A data were collected from 20,000 events per recording. Threshold levels of two-dimensional dot plot SSC-A vs. SSC-H were set for gating singlets. Hoechst 33342 fluorescence was detected with the filter for pacific blue (Pacific Blue; 450/50) and used to exclude non-viable cells. DiSC3(5) fluorescence was detected with the filter for allophycocyanine (APC; 585/40).

In both mouse and human sperm experiments, membrane potential fluorescent signal was calibrated by adding 1 µM valinomycin and sequential adding KCl corresponding to 4.98, 7.48, 12.48, 22.48, and 42.48 mM K<sup>+</sup> for HTF and 5.90, 8.4, 13.4, 23.4, and 43.4 mM K<sup>+</sup> for TYH (Chávez et al., 2013). Theoretical values were obtained by using the Nernst equation, assuming an intracellular K<sup>+</sup> concentration of 120 mM (Linares-Hernández et al., 1998) and considering room temperature as 298.15◦K. The final sperm membrane potential was obtained by linearly interpolating the median fluorescence of the unknown sample to the calibration curve of each trace.

## Determination of Membrane Potential by Spectrophotometry

After swim-up incubation, human sperm samples were centrifuged at 400 × g for 5 min, resuspended in noncapacitating HTF media with 5 mg/ml of BSA A-2153 (Sigma), and 4 × 10<sup>6</sup> sperm were transferred to a gently stirred cuvette at 310.15◦K. DiSC3(5) was added to a final concentration of 1 µM. Fluorescence was monitored with a Varian Cary Eclipse fluorescence spectrophotometer at 620 nm excitation and 670 nm emission as described before (Chávez et al., 2013). Calibration was performed by adding 1 µM valinomycin and sequential additions of KCl as described for flow cytometry experiments. The final sperm membrane potential was obtained by linearly interpolating the theoretical membrane potential values versus arbitrary units of fluorescence of each trace.

## Acrosomal Status

To assess acrosomal status, 50 µl human sperm samples were treated with 10 µM of the calcium ionophore A23187 or DMSO (vehicle) for 30 min before the end of the capacitation incubation time. After incubation, sperm were fixed at 4◦C by adding 50 µl of 12.5% w/v paraformaldehyde in 2.28 M Tris. Sperm were then spotted onto slides, dried, washed 3 times for 5 min and one time for 15 min with DPBS containing 0.1% v/v Tween 20 (t-DPBS), and stained with 100 µg/ml of FITC-labeled Pisum sativum agglutinin dissolved in t-DPBS. After washing, slides were observed under an EVOS <sup>R</sup> FL Cell Imaging System epifluorescence microscope at 40×. Sperm were scored as acrosome intact if a bright staining was observed in the acrosome, or as acrosome reacted when either fluorescent staining was restricted to the equatorial segment or no labeling was observed. The percentage of sperm that underwent induced acrosomal exocytosis was obtained by subtracting the percentage of acrosome reacted sperm incubated with DMSO from the percentage of acrosome reacted sperm incubated with A23187.

## Computer-Assisted Sperm Analysis (CASA)

Aliquots (3 µl) of sperm suspension were placed into a 20 micron Leja standard count 4 chamber slide, pre-warmed at 37◦C. CASA analysis was performed with a Hamilton–Thorne digital image analyzer (HTR-CEROS II v.1.7; Hamilton–Thorne Research, Beverly, MA, United States). Phase alignment was checked, and the settings used for the analysis were selected as follows: objective 1: Zeiss 10XNH; min total count: 200; frames acquired, 30; frame rate, 60 Hz; camera exposure: 8 ms; camera gain: 300; integrated time: 500 ms; elongation max%: 100; elongation min%: 1; head brightness min 170; head size max: 50 µm<sup>2</sup> ; head size min: 5 µm<sup>2</sup> ; static tail filter: false; tail brightness min: 70; tail brightness auto offset: 8; tail brightness mode: manual; progressive STR (%): 80; progressive VAP (µm/s): 25. The criteria used to define hyperactivated sperm were: curvilinear velocity (VCL) >150 µm/s, lateral head displacement (ALH) >7.0 µm, and linearity coefficient (LIN) <50% (Mortimer et al., 1998).

## Calculations and Statistical Analyses

Data are expressed as mean ± standard error of the mean (SEM). Independent experiments were carried out with samples from different donors or patients. A probability (p) value p < 0.05 was considered statistically significant. A value of p < 0.05 was indicated with an <sup>∗</sup> , p < 0.01 with ∗∗, and p < 0.001 with ∗∗∗. Calculations were performed with Microsoft Office 365 ProPlus spreadsheet, and statistical analyses were performed with GraphPad Prism version 6.01, GraphPad Software (La Jolla, CA, United States). To test whether values were in a Gaussian distribution, D'Agostino-Pearson, Shapiro-Wilk, and Kolmogorov-Smirnov tests were run in parallel. Parametric or non-parametric comparisons were used as dictated by the data distribution. Spearman test was used for correlations. The differences between means from two groups were analyzed by paired or unpaired t-test, depending on the condition tested. The receiving operating characteristic (ROC) curve and the maximum Youden index were used to determine the optimal cut-off value for depolarized and hyperpolarized samples.

## RESULTS

#### Flow Cytometry Can Be Used to Measure Absolute Values of Sperm Membrane Potential

Our method of calculating absolute sperm membrane potential values is based on the properties of the cationic voltage-sensitive dye DiSC3(5), which distributes across cellular membranes in response to electrochemical gradients. DiSC3(5) enters the sperm upon membrane hyperpolarization, increasing intracellular fluorescence. DiSC3(5) has been used to measure mammalian sperm membrane potential by spectrophotometry (Espinosa and Darszon, 1995; Muñoz-Garay et al., 2001; Demarco et al., 2003; Santi et al., 2010). Since flow-cytometry in association with DiSC3-(5) was not previously used to determine absolute sperm membrane potential values, we first validated our method. To do

so, we conducted DiSC3(5) flow cytometry experiments with both wild type (WT) and Slo3−/<sup>−</sup> mouse sperm and compared our values to those previously measured by spectrophotometry (Santi et al., 2010). Forward scatter area (FSC-A) and side scatter area (SSC-A) light detection were used to identify sperm cells based on size and granularity (**Figure 1A**). To exclude non-single cells (doublets or debris), singlets were selected with SSC-A and sidescatter height (SSC-H) (**Figure 1B**; Escoffier et al., 2015). Hoechst 33342 dye was used to differentiate between live and dead sperm (**Figure 1C**), and DiSC3(5) fluorescence was analyzed only in live sperm (see section Materials and Methods).

To calibrate the fluorescent signal, we constructed a calibration curve by using the K<sup>+</sup> ionophore valinomycin and sequential additions of known concentrations of KCl. Addition of valinomycin causes membrane hyperpolarization up to the value of the K<sup>+</sup> equilibrium potential, causing DiSC3(5) to enter the

FIGURE 1 | Flow cytometry can be used to determine absolute values of membrane potential in wild type and Slo3−/<sup>−</sup> mouse sperm. (A) Forward-scatter (FSC) and side-scatter (SSC) light data were collected. Threshold levels for side-scatter area (SSC-A) and forward-scatter area (FSC-A) were set to exclude signals from cellular debris or cells with abnormal morphology. (B) Threshold levels of two-dimensional dot plot of SSC-A vs. side-scatter height (SSC-H) were set to include only single cells and exclude clumps. (C) Cells with bright Hoechst staining were excluded to measure DiSC3(5) fluorescence only in live sperm. (D,E) Upper panel: Representative dot plot of DiSC3(5) fluorescence from Slo3−/<sup>−</sup> (D) and WT (E) mouse sperm incubated in capacitated conditions. Calibration was performed by adding 1 µM valinomycin and sequentially adding KCl, corresponding to the theoretical plasma membrane potential values of –77.4, –68.3, –56.3, –42.0, and –26.2 mV ([K+]<sup>i</sup> of 120 mM). Black crosses indicate the median DiSC3(5) fluorescence of sperm within the gates (rectangles) defined for each addition. Lower panels: Histograms of DiSC3(5) fluorescence from selected gates as indicated in flow cytometry dot plots. (F) Linear regression of the calibration curves constructed from the median DiSC3(5) fluorescence from Slo3−/<sup>−</sup> and WT sperm obtained in D and E. Interpolated values for each curve are indicated with X and dotted lines. Slo3−/<sup>−</sup> r <sup>2</sup> = 0.9948; WT r <sup>2</sup> = 0.9311. (G) Membrane potential of Slo3−/<sup>−</sup> and WT capacitated sperm. <sup>∗</sup>P = 0.0265 by unpaired t-test.

sperm, leading to increased intracellular fluorescence. Sequential KCl additions cause depolarization of the plasma membrane to the corresponding K<sup>+</sup> equilibrium potential values and decreased intracellular fluorescence. The membrane potential values after addition of valinomycin and KCl were calculated according to the Nernst equation, assuming a [K+]<sup>i</sup> of 120 mM (**Figures 1D,E**). The absolute sperm membrane potential values were obtained by linearly interpolating the theoretical membrane potential values versus arbitrary units of fluorescence of each trace (**Figure 1F**). The membrane potential values obtained with this technique from WT (−59.4 ± 5.3 mV) and Slo3−/<sup>−</sup> (−37.5 ± 3.3 mV) sperm (**Figure 1G**) were similar to those reported by spectrophotometry by our group (−60 mV for WT and −40 mV for Slo3−/−) (Santi et al., 2010). Thus, we concluded that our flow cytometric method of measuring absolute membrane potential values was valid for mouse sperm.

#### Human Sperm Incubated in Capacitation Conditions Are More Hyperpolarized Than Non-capacitated Sperm

Previous flow cytometry studies of human sperm membrane potential showed that non-capacitated sperm were more depolarized than sperm incubated in capacitating media (López-González et al., 2014; Brukman et al., 2019), but absolute membrane potential values were not reported. Here, we used our flow cytometric technique to measure the absolute membrane potential values in both capacitated and non-capacitated sperm from normozoospermic donors. The fluorescence probe BCECF-AM was used as an indicator of live sperm, and only BCECFpositive cells were analyzed (Puga Molina et al., 2017, 2018b; **Supplementary Figures S1A–C**). We conducted a calibration curve as we had done for mouse sperm (**Figures 2A–C**). We found that human sperm incubated in capacitation conditions were, on average, significantly more hyperpolarized than noncapacitated sperm (−45.2 ± 3.2 vs. −35.7 ± 2.8 mV, n = 13, respectively) (**Figure 2D**). However, some donor samples did not undergo hyperpolarization in capacitating media.

The spectrophotometry assay is widely used to measure absolute membrane potential values in sperm (Linares-Hernández et al., 1998; Patrat et al., 2002; Santi et al., 2010; Chávez et al., 2013). Therefore, we compared the membrane potential value obtained by spectrophotometry and flow cytometry in a donor sample. We obtained similar membrane potential values with the two techniques (**Supplementary Figure S1**). This result further indicates that our flow cytometric method could be used to determine human sperm membrane potential.

### Human Sperm Membrane Potential Is a Good Indicator of the Ability of Sperm to Undergo Sapacitation

Mouse sperm that do not hyperpolarize, do not hyperactivate and do not undergo regulated acrosomal exocytosis, the two endpoints of sperm capacitation (Demott and Suarez, 1992; Yanagimachi, 1994; Buffone et al., 2009). To determine whether capacitated human sperm membrane potential correlated with the ability of sperm to undergo acrosomal exocytosis, we used our flow cytometry method to calculate membrane potential in 17 capacitated donor normozoospermic sperm samples. In parallel, we treated the sperm with the calcium ionophore A23187, which induces acrosomal exocytosis (Köhn et al., 1997; Moody et al., 2017; Puga Molina et al., 2017) and determined the percentage of sperm in each sample that underwent acrosomal exocytosis. A scatter plot comparing the two values revealed that the membrane potential values obtained by flow cytometry correlated significantly with the percentage of cells that underwent acrosomal exocytosis (**Figure 3A**).

To determine whether capacitated sperm membrane potential correlated with sperm hyperactivation (HA), we used computerassisted sperm analysis to measure three parameters associated with HA: curvilinear velocity (VCL), amplitude of lateral head displacement (AHL), and linearity (LIN). Sperm with values of VCL>150 µm/s, AHL>7.0 µm and LIN <50% were considered hyperactivated (Mortimer et al., 1998). The percentage of sperm in each of 27 normozoospermic donor samples, correlated significantly with membrane potential values (**Figure 3B**). Specifically, VCL and ALH values (**Figures 3C,D**), but not LIN values (**Figure 3E**), correlated with sperm membrane potential. Together, these data indicate that human sperm membrane potential is a good indicator of the ability of sperm to undergo capacitation.

#### Membrane Potential of Capacitated Human Sperm Correlates With Fertilization Ratio in IVF Patients

Next, we wanted to determine whether the membrane potential could indicate the ability of sperm to fertilize an egg. To address this question, we performed our flow cytometric assay on excess samples from 49 normozoospermic men whose sperm was used for conventional IVF. We found that the percentage of fertilized oocytes (reported as fertilization ratio) from these patients significantly correlated with the absolute value of membrane potential of capacitated sperm. Those with more negative membrane potential had higher fertilization ratios than those with more positive membrane potential (**Figure 4**).

If IVF failure were caused by impaired sperm capacitation, this defect would be bypassed by ICSI. Thus, we reviewed the ICSI outcome of 14 patients who underwent this procedure and whose sperm resulted in an IVF ratio of less than 0.7. In three cases, the fertilization ratio was the same in ICSI and conventional IVF. However, for the other 11, the ICSI fertilization ratio was higher than the IVF ratio (mean values n = 14 0.8648 vs. 0.4208; Wilcoxon test, P = 0.0005) (**Supplementary Figure 2A**). This result suggests that the low IVF ratios were related to impaired sperm capacitation.

## Fertile Sperm Are Hyperpolarized

Given the correlation we observed between sperm membrane potential values and fertilization ratios in IVF, we wondered whether we could detect a significant difference in sperm membrane potential values between patients with successful and non-successful IVF, which are defined as fertilization ratio

>0.7 or <0.7, respectively (Harris et al., 2019). This cutoff was determined by calculating the mean fertilization ratio in conventional IVF cases quarterly and yearly. Thus, we dichotomized the samples into those with fertilization ratio >0.7 (n = 27) and those with fertilization ratio <0.7 (n = 22) and compared their sperm membrane potential values. Whereas sperm samples with fertilization ratio >0.7 had an average membrane potential of −58.03 ± 3.00 mV, sperm samples with fertilization ratio <0.7 had an average membrane potential of −40.61 ± 2.60 mV (**Figure 5A**).

We next investigated whether sperm samples capable of fertilizing an egg, hyperpolarize during capacitation. These experiments require measuring the membrane potential before (time = 0 h) and after capacitation (time = 18 h). This kind of experiment was not possible to do using fresh sperm because these samples were obtained after finishing clinical procedures and were kept in capacitating media for 2–4 h. Therefore, we did not have a real 0 time point. For this reason, we decided to use and aliquot of frozen samples from anonymous donors that had resulted in pregnancies. We used flow cytometry in these frozen samples to measure membrane potential before (t = 0) and after capacitation (t = 18 h). We found that all these fertile sperm samples underwent hyperpolarization after capacitation, changing from −47.9 ± 4.4 to −62.5 ± 4.1 mV (**Figure 5B**). Although there could be many differences between frozen and fresh sperm samples, it is noteworthy that the mean membrane potential value of frozen sperm from fertile donors after capacitation was similar to the membrane potential value of fresh sperm samples from patients with successful IVF (fertilization ratio >0.7) (**Figure 5A**). In addition, the mean membrane potential values of the frozen sperm from fertile donors before capacitation, was similar to that of fresh sperm from patients with fertilization ratio <0.7. These findings suggest that infertile sperm fail to hyperpolarize in capacitating conditions: even though the samples with lower fertilization ratio were incubated in capacitated conditions, they apparently failed to undergo hyperpolarization and failed to capacitate and fertilize. The frozen samples hyperpolarized and all of them were capable of fertilizing an egg.

#### Membrane Potential May Be Used to Predict Fertility Success

Finally, we wanted to determine whether membrane potential values could be used to predict sperm capacitation state and fertilization success. To do this, we first determined a value of

membrane potential that allowed us to characterize a sample as either hyperpolarized or depolarized. We used the receiving operating characteristic curve and the maximum Youden index to determine the optimal cut-off value. We considered samples with fertilization ratio >0.7 as the reference for a positive outcome and <0.7 as a negative outcome. The optimal cut-off with 81.82% (59.72 to 84.81%) sensitivity and 88.89% (70.84 to 97.65%) specificity was −46 mV (**Figure 6A**). We then applied this cut-off value and found that those samples with a sperm membrane potential value more hyperpolarized than −46 mV had a significantly higher IVF fertilization ratio than those with a sperm membrane potential value

more depolarized than −46 mV (0.46 ± 0.06, n = 21, vs. 0.80 ± 0.03, n = 28) (**Figure 6B**). Furthermore, the percentages of sperm that underwent induced acrosomal exocytosis (**Figure 6C**) and hyperactivated motility (**Figure 6D**) were significantly higher in samples with sperm membrane potential values more hyperpolarized than −46 mV. However, ICSI fertilization ratios were not significantly different between sperm samples with membrane potential values more positive and more negative than −46 mV (**Supplementary Figure S2**). This result is not surprisingly because ICSI bypasses sperm capacitation.

#### DISCUSSION

Here, we report a new method that combines flow-cytometry with the voltage sensitive dye DiSC3(5) to quantify human sperm membrane potential. This technique could be developed as a new tool to predict sperm fertility capacity. This method has several advantages over other available methods to evaluate the ability of human sperm to capacitate. It is less subjective and time consuming than the CapScore test, which assays the localization pattern of a ganglioside (GM1) within the plasma membrane as a marker of capacitation. This method requires staining, counting, and correctly identifying staining patterns in more than 150 individual sperm (Moody et al., 2017). Patch clamp can be used to measure sperm membrane potential, but the flow cytometry method is simpler and can evaluate membrane potential in higher number of sperm than this technique (Brown et al., 2016). Finally, our flow cytometry method excludes nonviable sperm and requires a smaller sperm sample than does spectrophotometry (0.25 × 10<sup>6</sup> sperm vs. 4 × 10<sup>6</sup> sperm).

We argue that quantification of human sperm membrane potential by flow cytometry is a valid method to predict human sperm fertility success for several reasons. First,

FIGURE 5 | Fertile sperm are hyperpolarized. (A) Sperm membrane potential (Em) values from patients with in vitro fertilization ratios <0.7 (n = 22; –40.61 ± 2.60 mV) or >0.7 (n = 27, –58.03 ± 3.00 mV). ∗∗∗∗P < 0.0001 by unpaired t-test. (B) Membrane potential of fertile frozen sperm samples from patients that conceived a pregnancy, incubated in commercial capacitating media for 1 h (before capacitation, –47.92 ± 4.38 mV) or 18 h (after capacitation, –62.51 ± 4.05 mV). n = 6, <sup>∗</sup>P = 0.0146 by paired t-test.

hyperpolarization is a key event in sperm becoming competent to fertilize an egg in many mammalian species (Zeng et al., 1995; Hernández-González et al., 2007; López-González et al., 2014). In murine and bovine sperm, lack of hyperpolarization is associated with failure to undergo acrosomal exocytosis (Zeng et al., 1995; De La Vega-Beltran et al., 2012). An increase in membrane K<sup>+</sup> permeability and the subsequent sperm plasma membrane hyperpolarization are essential for sperm capacitation in mice (Chávez et al., 2013). Furthermore, human sperm also hyperpolarize under capacitating conditions (López-González et al., 2014). Brown et al. showed an association between depolarized membrane potential values and poor fertilization rate, as 2 of 19 IVF patients that they analyzed had highly depolarized membrane potential and a low fertilization ratio (Brown et al., 2016).

Second, we validated our method by showing that it produced similar values for wild-type and Slo3−/<sup>−</sup> mouse sperm as have been reported by other methods (Santi et al., 2010). Third, the mean membrane potential value of human sperm from donors capacitated in commercial IVF media (−57.7 ± 4.1 mV, N = 22, **Supplementary Figure S3**) is consistent with the value reported by Patrat et al. (2002) from spectrofluorimetric assays (−57.8 ± 2.20 mV, n = 12). In addition, we obtained similar values of human sperm membrane potential by spectrophotometry and flow cytometry. Fourth, the mean value of sperm membrane potential we calculated from patients with fertilization ratio <0.7 (−40.61 ± 2.60 mV) was similar to the value Calzada and Tellez (1997) obtained via a radioactivity assay of sperm from patients with idiopathic infertility (−35.00 ± 1.60 mV). Our value was also similar to the mean membrane potential value of non-capacitated sperm that Linares-Hernández et al. (1998) measured with spectrophotometry (−40.00 ± 6.00 mV). Conversely, we found that membrane potential values of capacitated sperm from patients with successful IVF (−58.03 ± 3.00 mV), from frozen fertile samples (−62.5 ± 4.1 mV), and from fertile donors included in this study (−64.1 ± 7.9 mV

(n = 32, <sup>∗</sup>P = 0.049); and (D) hyperactivated motility (n = 41, <sup>∗</sup>P = 0.031). All P-values were calculated by a Kolmogorov-Smirnov test.

n = 4, dark gray, **Supplementary Figure S3**) were similar to the reported values of sperm membrane potential from fertile donors, measured by a radioactivity assay (Calzada −75.0 ± 1.9 mV, n = 10).

Fifth, we argue that our method is valid because we found that human sperm, like mouse sperm, hyperpolarized when incubated in capacitating media. However, the extent of this hyperpolarization was smaller in human sperm (mean difference 9.24 mV) than in mouse sperm (mean difference ∼15–30 mV) (Espinosa and Darszon, 1995; Zeng et al., 1995; Muñoz-Garay et al., 2001; Demarco et al., 2003; Hernández-González et al., 2006; Santi et al., 2010; De La Vega-Beltran et al., 2012). This difference in membrane potential between capacitated and non-capacitated human sperm is similar to the 5 mV difference reported by using the patch clamp technique (Brown et al., 2016). It is also noteworthy that human capacitated sperm membrane potential values were more heterogeneous than were mouse capacitated sperm values (coefficients of variation 25.7% vs. 0.29%, respectively), even though all the samples analyzed were normozoospermic (≥32% progressive motility; ≥40% total motility; ≥15 × 10<sup>6</sup> cells/ml). This variability, along with the small difference in membrane potential values between capacitated and non-capacitated sperm, could explain why changes in membrane potential of human sperm during capacitation are difficult to detect by other methods (Puga Molina et al., 2018a).

Sixth, by using this methodology, we found that membrane hyperpolarization positively correlated with the two endpoints of sperm capacitation: hyperactivated motility and acrosomal exocytosis. Hyperpolarization of the plasma membrane is necessary and sufficient for acrosomal exocytosis of mouse sperm (De La Vega-Beltran et al., 2012), but this is the first report that correlates membrane potential with both hyperactivation and the acrosomal exocytosis and highlights the importance of membrane potential in human sperm capacitation.

Finally, we reported here that hyperpolarization measured by our method correlates with IVF success. Only two previous studies attempted to find an association between membrane potential values and human fertility, but the techniques used to measure sperm membrane potential (patch clamp or radioactivity assays) are extremely challenging and not suitable to clinical implementation (Calzada and Tellez, 1997; Brown et al., 2016). Brown et al. showed that sperm plasma membrane depolarization, measured with the patch clamp technique, was associated with low IVF rates. However, these authors did not report a correlation between the values of membrane potential and fertilization ratio. Additionally, they did not address whether the mean values of sperm membrane potential from successful and unsuccessful IVF significantly differed (Brown et al., 2016). Interestingly our results agree with the results also reported in this issue which were independently obtained by Dr. Krapf's group showing that IVF rates correlate to

human sperm membrane potential values measured by the spectrophotometry technique (Baro Graf et al., 2020).

Until now, the fluorescent probes DiSC3(5) or DiSBAC2(3) have only been used in combination with flow cytometry to report relative changes in sperm membrane potential (López-González et al., 2014; Escoffier et al., 2015; Puga Molina et al., 2017, 2018b; Brukman et al., 2019) and not absolute values. However, we were able to build upon spectrofluorimetric assays in which DiSC3(5) fluorescence was conveniently calibrated by using valinomycin and different KCl concentrations, permitting measurement of sperm membrane potential values (Zeng et al., 1995; Demarco et al., 2003; Chávez et al., 2013). Importantly, DiSC3(5) has no toxic effects on sperm function at the concentrations used (López-González et al., 2014). When constant concentrations of sperm and dye are used, DiSC<sup>3</sup> provides reproducible estimates of plasma membrane potential and good signal-to-noise ratio. One potential challenge with using DiSC3(5) is that it binds to mitochondria in their normal energized state, and the resulting mitochondrial fluorescence could contribute to the fluorescence signal (Rink et al., 1980). However, we previously measured DiSC3(5) fluorescence in the presence and absence of several mitochondrial uncouplers (CCCP, antimycin, and oligomycin) and showed that, in all conditions, the plasma membrane potential of valinomycintreated mouse sperm responds to the external K<sup>+</sup> concentration in accordance with the Nernst equation (Chávez et al., 2013). In addition, Baro Graf et al. (2019) showed that the mitochondrial electron transport inhibitor rotenone does not affect DiSC3(5) membrane potential measurements in human sperm.

In summary, our data indicate that the absolute value of sperm membrane potential is a good indicator of human sperm fertilizing ability. We found that a membrane potential value of −46 mV could be used to classify the sperm samples as depolarized or hyperpolarized and predict their fertilization success. However, a larger cohort of patients is needed to determine a more accurate cut-off value of membrane potential for use in a clinical setting. Our findings also suggest the possibility of using capacitated sperm membrane potential values as a novel tool to diagnose idiopathic normozoospermic male infertility. In addition, this method could be developed as a fast, simple, clinically feasible assay to personalize fertility treatment plans for normozoospermic infertile men (Oehninger et al., 2014; Palermo et al., 2015). For example, normozoospermic men whose sperm have hyperpolarized values of membrane potential could be guided by clinicians to approaches such as intrauterine insemination or conventional IVF. Conversely, men with sperm with depolarized values of membrane potential might be able to avoid the emotional, physical, and financial costs associated with repeated intrauterine or IVF cycles by being guided to ICSI. Given that few IVF clinics are likely to have a flow cytometer,

#### REFERENCES


future work will be aimed at developing a more automated method to measure sperm membrane potential.

## DATA AVAILABILITY STATEMENT

All data generated in this study are included in the article/**Supplementary Material**.

## ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Washington University Institutional Review Board. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Animal Care and Use Committee at Washington University.

#### AUTHOR CONTRIBUTIONS

LP, CS, SG, and PL conceived and planned the experiments. LP and SG performed the experiments, analyzed the data, and obtained consent from patients. LP and CS discussed the results and wrote the manuscript. AB-A performed spectrophotometry assays and helped with mouse experiments. All authors contributed to revising the manuscript and read and approved the submitted version.

## FUNDING

This work was supported by NIH grants R01HD069631 and R01HD095628 to CS.

## ACKNOWLEDGMENTS

We thank Yang-Yang Yuan for helping with mice and sperm samples, the Fertility and Reproductive Medicine Center for providing human sperm samples, Paul Schlesinger for the use of Spectrophotometer, Jeremias Incicco for their help with statistical analysis, and Deborah Frank for editorial help.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2019.00387/ full#supplementary-material

potential analysis. Front. Cell Dev. Biol. 7:101. doi: 10.3389/fcell.2019. 00101

Baro Graf, C., Ritagliati, C., Torres-Monserrat, V., Stival, C., Carizza, C., Buffone, M. G., et al. (2020). Membrane potential assessment by fluorimetry as a predictor tool of human sperm fertilizing capacity. Front. Cell Dev. Biol. 7:383. doi: 10.3389/fcell.2019. 00383


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Puga Molina, Gunderson, Riley, Lybaert, Borrego-Alvarez, Jungheim and Santi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Essential Role of Sperm-Specific PLC-Zeta in Egg Activation and Male Factor Infertility: An Update

Alaaeldin Saleh<sup>1</sup> , Junaid Kashir2,3,4, Angelos Thanassoulas<sup>1</sup> , Bared Safieh-Garabedian<sup>1</sup> , F. Anthony Lai1,5 and Michail Nomikos<sup>1</sup> \*

<sup>1</sup> Member of QU Health, College of Medicine, Qatar University, Doha, Qatar, <sup>2</sup> College of Medicine, Alfaisal University, Riyadh, Saudi Arabia, <sup>3</sup> Department of Comparative Medicine, King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia, <sup>4</sup> School of Biosciences, Cardiff University, Cardiff, United Kingdom, <sup>5</sup> Biomedical Research Center, Qatar University, Doha, Qatar

#### Edited by:

Tomer Avidor-Reiss, The University of Toledo, United States

#### Reviewed by:

John Parrington, University of Oxford, United Kingdom Karl Swann, Cardiff University, United Kingdom Rafael A. Fissore, University of Massachusetts Amherst, United States

#### \*Correspondence:

Michail Nomikos mnomikos@qu.edu.qa; mixosn@yahoo.com

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology Received: 09 September 2019 Accepted: 14 January 2020

Published: 29 January 2020

#### Citation:

Saleh A, Kashir J, Thanassoulas A, Safieh-Garabedian B, Lai FA and Nomikos M (2020) Essential Role of Sperm-Specific PLC-Zeta in Egg Activation and Male Factor Infertility: An Update. Front. Cell Dev. Biol. 8:28. doi: 10.3389/fcell.2020.00028 Sperm-specific phospholipase C zeta (PLCζ) is widely considered to be the physiological stimulus responsible for generating calcium (Ca2+) oscillations that induce egg activation and early embryonic development during mammalian fertilization. In the mammalian testis, PLCζ expression is detected at spermiogenesis following elongated spermatid differentiation. Sperm-delivered PLCζ induces Ca2<sup>+</sup> release via the inositol 1,4,5 trisphosphate (InsP3) signaling pathway. PLCζ is the smallest known mammalian PLC isoform identified to date, with the simplest domain organization. However, the distinctive biochemical properties of PLCζ compared with other PLC isoforms contribute to its unique potency in stimulating cytosolic Ca2<sup>+</sup> oscillations within mammalian eggs. Moreover, studies describing PLCζ "knockout" mouse phenotypes confirm the supreme importance of PLCζ at egg activation and monospermic fertilization in mice. Importantly, a number of clinical reports have highlighted the crucial importance of PLCζ in human fertilization by associating PLCζ deficiencies with certain forms of male factor infertility. Herein, we give an update on recent advances that have refined our understanding of how sperm PLCζ triggers Ca2<sup>+</sup> oscillations and egg activation in mammals, while also discussing the nature of a potential "alternative" sperm factor. We summarise PLCζ localization in mammalian sperm, and the direct links observed between defective PLCζ protein in sperm and documented cases of male infertility. Finally, we postulate how this sperm protein can be used as a potential diagnostic marker, and also as a powerful therapeutic agent for treatment of certain types of male infertility due to egg activation failure or even in more general cases of male subfertility.

Keywords: sperm, phospholipase C zeta, PLC zeta, egg activation, fertilization

## SPERM PLCζ IS THE PRIMARY STIMULUS FOR EGG ACTIVATION AND EARLY EMBRYONIC DEVELOPMENT

In mammalian fertilization, the fertilizing spermatozoon stimulates egg activation, a fundamental event that initiates embryonic development (Nomikos et al., 2017a). It is well established that the most crucial event of egg activation is an acute increase in cytosolic free Ca2<sup>+</sup> concentrations, which in mammals occurs in the form of long-lasting Ca2<sup>+</sup> oscillations that commence at or directly following gamete fusion, and persist for several hours beyond meiotic completion (Stricker, 1999; Malcuit et al., 2006; Kashir et al., 2013a). This Ca2<sup>+</sup> signaling paradigm is essential

for the completion of the multiple events of egg activation. However, previous archetypes of our understanding regarding the distinct events of egg activation and the events controlling them are continuously being unraveled and questioned, with specifics still being investigated. It is, however, clear that Ca2+ release is an integral component of egg activation in all species studied to date (Cran et al., 1988; Swann and Ozil, 1994; Jones, 1998; Nomikos et al., 2012; Limatola et al., 2019b). Over the last few decades, a number of sperm-derived molecules had been proposed as potential soluble sperm factors responsible for the generation of Ca2<sup>+</sup> oscillations during mammalian fertilization (for more information see Nomikos et al., 2012, 2013a, 2017a). The fact that sperm-induced Ca2<sup>+</sup> oscillations are caused by activation of the inositol 1,4,5-trisphosphate (InsP3) signaling pathway (Miyazaki et al., 1992) suggested that the sperm factor might itself be a phospholipase C (PLC) isoform (Jones et al., 1998).

In 2002, a novel testis-specific PLC, termed PLC zeta (PLCζ), was discovered (Saunders et al., 2002) and abundant experimental evidence has accumulated over the years suggesting that PLCζ fulfills all prerequisite criteria of the soluble sperm factor responsible for the generation of Ca2<sup>+</sup> oscillations at mammalian fertilization (Cox et al., 2002; Saunders et al., 2002, 2007; Knott et al., 2005; Kouchi et al., 2005; Nomikos et al., 2005, 2013b, 2017a; Swann et al., 2006; Yu et al., 2008; Kashir et al., 2012a). Upon sperm-egg fusion, PLCζ is proposed to be delivered by the fertilizing sperm into the ooplasm, triggering the Ca2<sup>+</sup> oscillations via the InsP<sup>3</sup> signaling pathway, through the hydrolysis of its membrane-bound phospholipid substrate, PIP<sup>2</sup> (Saunders et al., 2002; Nomikos, 2015). The importance of this sperm specific protein in mammalian fertilization has been further highlighted by numerous clinical studies directly linking defects or deficiencies in human PLCζ with documented cases of male factor infertility (Yoon et al., 2008; Heytens et al., 2009; Nomikos et al., 2011a, 2017b; Kashir et al., 2012b,c; Escoffier et al., 2016; Torra-Massana et al., 2019).

Intriguingly, two recent independent studies described the phenotype of a PLCζ "knockout" mouse (Hachem et al., 2017; Nozawa et al., 2018). By using multiple transgenic models of PLCζ "knockout" mice generated by CRISPR/Cas methodology, both studies reported that males can produce offspring, albeit with significantly reduced litter numbers (∼25%). Interestingly, both studies showed that sperm lacking functional PLCζ protein failed to induce Ca2<sup>+</sup> release when microinjected into mouse eggs by ICSI. However, in vitro fertilisation (IVF) with such sperm, produced atypical and delayed patterns of Ca2<sup>+</sup> oscillations (lower in number and frequency) with a high degree of polyspermy and activation failure, compared to the robust, physiological pattern triggered by physiological PLCζ-induced egg activation (Nozawa et al., 2018; Satouh and Ikawa, 2018).

Perhaps the atypical and delayed pattern of Ca2<sup>+</sup> release, observed alongside the low number of embryos and offspring, could be spontaneous activation, unrelated to Ca2<sup>+</sup> release, which is common in some strains of mice (Cheng et al., 2012), alongside with the introduction of PLCζ knockout sperm. Indeed, eggs that had been fertilized by knockout sperm also displayed multiple pronuclei, consistent with the inability of a sufficient polyspermy block (Nozawa et al., 2018). Critically, however, eggs fertilized with PLCζ knockout sperm exhibited a total of 3–4 oscillations in total, initiating following a 1-h delay. This was in contrast to normal fertilization where 3–4 oscillations were observed per hour over 3–4 h (Nozawa et al., 2018; Satouh and Ikawa, 2018). Such observations perhaps suggest that sperm containing a second molecule with Ca2<sup>+</sup> releasing activity, albeit weaker than PLCζ (Jones, 2018). From such results, one could potentially posit that perhaps PLCζ is not an absolute requirement for natural fertilization, and that perhaps an alternative "primitive" or "cryptic" sperm factor may also be involved in leading to egg activation (Nozawa et al., 2018; Satouh and Ikawa, 2018).

It is possible that such a factor could be one of the previously proposed unsuccessful candidates for the "sperm factor," including tr-kit (Sette et al., 2002), citrate synthase (Harada et al., 2007), or PAWP (Aarabi et al., 2010), which while not contributing to the majority of Ca2<sup>+</sup> release at oocyte activation, may have a contributory function, especially, in the absence of PLCζ. However, it is worth noting that none of the aforementioned proteins is able to elicit Ca2<sup>+</sup> release in the specific manner required for oocyte activation at physiological levels within sperm (Kashir et al., 2014; Nomikos et al., 2014; Satouh et al., 2015), while none of the alternatively proposed sperm factors (apart from PLCζ) has been shown to be directly involved in IP3-mediated Ca2<sup>+</sup> release (Kashir et al., 2014). Furthermore, we cannot exclude the possibility that another sperm-associated enzyme, which might be only able to achieve critical levels due to absence of PLCζ in the sperm of PLCζ knockout mice, might play the role of the "cryptic" factor triggering embryogenesis by a distinct mechanism.

Theories regarding RNA involvement are also questionable since the total amount of PLCζ RNA present within sperm may not be enough to elicit any Ca2<sup>+</sup> release. On the other hand, this may have been altered as part of genetic compensation.

Intriguingly, starfish eggs pre-injected with heparin (which also blocks InsP<sup>3</sup> receptor function) to disrupt cytoskeletal arrangement were unable to exhibit a rapid Ca2<sup>+</sup> wave response upon interaction with sperm, instead exhibiting a much more delayed pattern of release, and failed to prevent polyspermy. Furthermore, the amplitude of subsequent Ca2<sup>+</sup> peaks were reduced, exhibiting an effect similar to observations made with sperm from PLCζ-null mice. In starfish, it was suggested that heparin- or age-induced hyperpolymerization of the cortical actin disrupted actin cytoskeleton dynamics at fertilization influenced Ca2<sup>+</sup> release (Puppo et al., 2008; Santella et al., 2015; Limatola et al., 2019a), potentially also impacting upon subsequent events in egg activation such as cortical granule exocytosis. It is thus possible that due to the lack of a sufficient response at fertilization due to deficient/absent PLCζ, a similar effect was observed in the PLCζ-null mice, where similar symptoms of insufficient Ca2<sup>+</sup> release and increased polyspermy were also observed. Indeed, it may be the case that the low number and frequency

**Abbreviations:** ART, assisted reproductive technology; Ca2+, calcium; ICSI, intracytoplasmic sperm injection; InsP3, inositol 1,4,5-trisphosphate; PI(3)P, phosphatidylinositol 3-phosphate; PI(5P), phosphatidylinositol 5-phosphate; PIP2, phosphatidylinositol 4,5-bisphosphate; PLCζ, phospholipase C zeta.

of Ca2<sup>+</sup> observed could be due to events surrounding actin polymerization or associated InsP3-independent events of Ca2<sup>+</sup> release (such as influx). However, these mechanisms are poorly understood in mammals, and require further investigation to fully ascertain.

It is clear that both Hachem et al. (2017) and Nozawa et al. (2018) represent keystone studies that support the notion that PLCζ is the primary physiological stimulus that triggers the required specific pattern of Ca2<sup>+</sup> oscillations, ensuring monospermy and eventually successful egg activation and early embryonic development (Hachem et al., 2017; Nozawa et al., 2018; Swann, 2018). Moreover, the presence of an alternative factor in other species and especially in humans is still questionable, particularly taking into consideration all the documented cases of male factor infertility due to PLCζ deficiencies. However, this is an intriguing area of investigation that ongoing studies are now aiming to address. It would be interesting to examine how the increasing body of invertebrate animal work will direct the mammalian side of the coin in the future, particularly with relation to the early influence exerted by the egg actin cytoskeleton upon patterns of Ca2<sup>+</sup> release and fertilization as is being unraveled in starfish. Furthermore, integral studies are required in particularly livestock mammalian models to demonstrate whether PLCζ-loss resembles the mouse and/or human scenarios. Perhaps of particular interest should be attempts to generate transgenic knockout models of PLCζ in porcine or bovine systems. While this would of course be considerably harder to perform than in the mouse, such data would undoubtedly assist in ascertaining the validity of "cryptic factor" theories. Further experiments that would be prudent would be to examine the specific timing of the reduced profiles of Ca2<sup>+</sup> release in relation to PLCζ knockout sperm-egg fusion and fertilization. Are such reduced frequency and amplitude oscillations due to fusion of a single sperm, or the cumulative effect of multiple sperm-egg fusion events? It is necessary that such experiments are performed to ascertain fully the conflicting data generated from knockout studies thus far.

## SPERM PLCζ STRUCTURE AND DOMAIN ORGANIZATION

Phospholipase C zeta is currently the smallest known mammalian PLC isoform (∼70–75 kDa in size) with the most elementary domain organization (Cox et al., 2002; Saunders et al., 2002; Nomikos et al., 2013a, 2017a). Despite this, PLCζ exhibits uniquely supreme potency in triggering Ca2<sup>+</sup> oscillations within the fertilizing egg compared to other somatic PLCs. This is attributed to its novel biochemical characteristics, arising from the essential role of its domains that contribute to the unique biological function and mode of regulation of this distinctive PLC isozyme (Nomikos, 2015; Nomikos et al., 2017a). PLCζ domain structure consists of four tandem EF hand domains at the N-terminus, the catalytic X and Y domain in the center of the molecule, followed by a single C2 domain at the C-terminus (**Figure 1**; Nomikos et al., 2017a). All these domains are common to other PLC isoforms. The X and Y catalytic domains are

separated by a short segment, the XY-linker, which through its net positive charge plays an important role in targeting PLCζ to intracellular membranes by direct electrostatic interactions with its negatively charged substrate, PIP<sup>2</sup> (Nomikos et al., 2007, 2011b). The XY-linker region differs considerably between PLC isozymes (Nomikos et al., 2013a). By contrast, the XY catalytic domain between PLC isoforms is the most highly conserved region (Nomikos, 2015).

The XY domain of PLCζ shares ∼60% sequence similarity to that of all PLCs and is responsible for PIP<sup>2</sup> hydrolysis (Nomikos et al., 2005). The EF hands are Ca2+-binding motifs and in PLCζ these domains play a vital role in its high Ca2<sup>+</sup> sensitivity compared with the other somatic PLCs, allowing PLCζ to be active at resting Ca2<sup>+</sup> levels within the egg cytosol, when PLCζ enters after sperm-egg fusion (Nomikos et al., 2005). Additionally, we have demonstrated that the first EFhand domain of PLCζ, which contains a cluster of basic amino acid residues, plays an essential role together with the XY-linker region, in the interaction of PLCζ with the PIP2-containing membranes (Nomikos et al., 2015b). The C-terminal C2 domain of PLCζ, comprising ∼120 amino acid residues is essential for PLCζ function, as targeted deletion or replacement of this domain by the corresponding domain from PLCδ1 abolishes the Ca2+-oscillation-inducing activity of PLCζ in eggs, without altering its enzymatic activity or Ca2<sup>+</sup> sensitivity (Nomikos et al., 2005; Theodoridou et al., 2013). We have provided biochemical evidence that this domain directly interacts with the membrane phospholipids, PI(3)P, and PI(5)P and have suggested that C2 association with these phospholipids may facilitate in

the membrane targeting of PLCζ (Theodoridou et al., 2013; Nomikos, 2015).

#### PLCζ IN MAMMALIAN SPERM

Phospholipase C zeta mRNA has been identified during both early and late stages of spermatogenesis in mice and pigs (Yoneda et al., 2006; Young et al., 2009; Bedford-Guaus et al., 2011; Kaewmala et al., 2012). Specific localization patterns, however, remain elusive in the literature throughout the various spermatogenic cells within the testes (Kashir et al., 2018). Aarabi et al. (2012) indicated that PLCζ is integrated as part of the acrosome during the Golgi phase of human and mouse spermiogenesis (Aarabi et al., 2012), suggesting that observable PLCζ levels are diminished gradually throughout spermatid elongation (Kashir et al., 2018). PLCζ was originally identified in mouse sperm extract fractions that were able to induce Ca2<sup>+</sup> release. Subsequent immunofluorescence analysis indicated a post-acrosomal localization for PLCζ; a component of the postacrosomal sheath (Saunders et al., 2002; Fujimoto et al., 2004; Young et al., 2009). Nevertheless, PLCζ has been identified in the sperm of various mammalian species, and usually tends to be found within the sperm head in distinct subcellular regions, postulating differential functional roles for each population (Amdani et al., 2013; Kashir et al., 2014, 2018).

While in mouse and porcine sperm, PLCζ has been observed mainly at acrosomal and post-acrosomal regions (Fujimoto et al., 2004; Young et al., 2009; Nakai et al., 2011; Kaewmala et al., 2012), in equine sperm, PLCζ was recorded at the acrosome, equatorial segment, and head mid-piece, as well as the principle piece of the flagellum (Bedford-Guaus et al., 2011; Kashir et al., 2018). Several PLCζ populations were observed in humans in multiple studies including the acrosomal, equatorial and post-acrosomal regions of the sperm head, with a potential tail localization (Grasa et al., 2008; Yoon et al., 2008; Young et al., 2009; Kashir et al., 2013b, 2018; Escoffier et al., 2015; Yelumalai et al., 2015; Yeste et al., 2016). While there is consensus regarding PLCζ localization in mouse sperm, the veracity of the multiple populations identified in other mammalian sperm (particularly in humans) remains debated (Nomikos et al., 2017a; Kashir et al., 2018). In human sperm, this variation in PLCζ localization is not only limited to observations between different studies but substantial variability in the PLCζ localization pattern was found even within the same study (Kashir et al., 2013b).

Despite numerous efforts to examine PLCζ localization within mammalian sperm, significant concern surrounds the specificity of the majority of antibodies used to date. More specifically, most antibodies used in the literature are unable to demonstrate a consistent motif of recognizing a single band following immunoblotting of human sperm, often detecting multiple protein bands other than, or in addition to, that of the expected size for native PLCζ protein. Compounded by this non-specificity, multiple groups have identified varying populations between mouse and human sperm, even using the same antibodies, suggesting that varying protocols and the use of different antibodies are the main source of inconsistent results between studies (Grasa et al., 2008; Yoon et al., 2008; Heytens et al., 2009; Kashir et al., 2011a,b, 2013b; Aarabi et al., 2012).

Addressing such concerns, we recently generated highly epitope-specific PLCζ polyclonal antibodies against human, mouse, and porcine PLCζ, that exhibit high consistency throughout numerous studies for both recombinant and native PLCζ (Nomikos et al., 2013b, 2014, 2015a; Theodoridou et al., 2013). Furthermore, we have also developed specific antigen unmasking/retrieval protocols, which we previously demonstrated are essential to enhance the visualization efficacy of PLCζ in mammalian sperm (Kashir et al., 2017). Using these enhanced protocols and materials, we have identified PLCζ in the acrosomal and post-acrosomal, acrosomal and equatorial, and post-acrosomal and equatorial compartments of mouse, human, and porcine sperm, respectively. Furthermore, we have also consistently observed potential tail localization in all species (Kashir et al., 2017). **Figure 2** demonstrates the expression and distribution of PLCζ in mouse sperm using our specific polyclonal antibodies and our recently developed and reported protocols (Kashir et al., 2017). It is now imperative that these specific antibodies and protocols are applied in a systematic manner to examine whether particular localization patterns or profiles of PLCζ exhibit any relationships between male fertility parameters, or indeed between fertility treatment outcomes.

Another intriguing question is how PLCζ, despite its high Ca2+-sensitivity and its potent enzymatic activity, is kept in an inactive state within the sperm, especially when it is likely to be present in much higher concentrations in a single spermatozoon than within the fertilizing egg. Indeed, our previous work where it was shown that PLCζ is inactive in somatic cells even at levels over 1000 times that at which it is active in eggs (Phillips et al., 2011), suggests that either PLCζ has an essential binding-partner within the egg, or that other factors within sperm and somatic cells may inhibit its catalytic activity.

#### REDUCED EXPRESSION LEVELS AND ABNORMAL FORMS OF SPERM PLCζ LEAD TO MALE INFERTILITY

Infertility is estimated to affect ∼15% of couples, with male infertility affecting ∼7% of men worldwide (Kashir et al., 2010). While genetic causes of male infertility are estimated to underlie ∼30% of such cases (Harton and Tempest, 2012; Jungwirth et al., 2012; Hotaling, 2014), ∼50% of cases of male infertility remain unexplained (Kashir et al., 2018). While most forms of infertility can now be treated via a collection of laboratory techniques collectively termed ART, a number of conditions such as severe male infertility (19–57% of cases) cannot yet been treated (Botezatu et al., 2014). Despite the fact that powerful ART methods such as IVF or ICSI can successfully treat some infertility cases, it is concerning that this is achieved only after several fertility treatment cycles. A significant causative factor may be recurrent implantation failure, which even after fertility treatment leads to infertility (Polanski et al., 2014; Kashir et al., 2018).

Considering the indispensable contribution of PLCζ to fertilization, defects in either egg activation, or in PLCζ

protein itself, may underlie conditions of male infertility where fertilization failure occurs. The first evidence came from studies that reported sperm of infertile men, which consistently failed to fertilize eggs following routine IVF or ICSI, and were either unable to induce Ca2<sup>+</sup> release upon microinjection into mouse eggs, or produced highly abnormal Ca2<sup>+</sup> transients which were reduced in frequency and amplitude (Yoon et al., 2008; Heytens et al., 2009). Furthermore, such sperm also exhibited reduced or absent levels, as well as abnormal localization patterns, of PLCζ within the sperm head (Heytens et al., 2009; Kashir et al., 2011a,b, 2013b), suggesting that deficiencies in PLCζ protein may underlie currently unknown cases of male factor infertility. In a clinical scenario, in contrast to other causes, complete fertilization failure is attributed to egg activation failure in a species-specific manner (Kashir et al., 2010, 2018).

Moreover, PLCζ gene abrogation in patients diagnosed with egg activation deficiency is now increasingly being reported within the scientific literature. The first two PLCζ mutations were identified in the gene of an infertile male, whose sperm was unable to trigger the normal pattern of Ca2<sup>+</sup> oscillations, leading to egg activation failure and potentially to his infertility (Heytens et al., 2009; Kashir et al., 2011b, 2012b,c). Both mutations were reported within the active catalytic site domains of PLCζ (X and Y), disrupting local protein structural folding to cause reduction of enzymatic activity, subsequently leading to highly abnormal Ca2<sup>+</sup> transients unable to initiate egg activation (Kashir et al., 2011b). Both mutations were reported to be heterozygous, with one mutation being inherited from the patient's father and the other from the patient's mother, indicating for the first time that such maternally inherited lossof-activity mutations can lead to male infertility (Kashir et al., 2012b,c; Nomikos et al., 2017a). Subsequently, a further mutation in homozygosis was later reported by Escoffier et al. (2016) from two infertile brothers. This mutation is located within the C2 domain of PLCζ(Escoffier et al., 2016). Intriguingly, this PLCζ mutant displayed similar enzymatic activity to wild type PLCζ, but displayed a dramatically reduced relative binding-affinity to PI(3)P and PI(5)P-containing liposomes (Nomikos et al., 2017a). More importantly, this genetic report by Escoffier et al. (2016) and the identification of this novel missense homozygous PLCζ mutation in these infertile brothers after whole exomic sequencing, strongly indicates that absence or defects in PLCζ protein alone is sufficient to prevent human egg activation by the sperm, suggesting that PLCζ is essential for human egg activation and thus human fertilization.

Furthermore, single nucleotide polymorphisms (SNPs) have also been reported by Yoon et al. (2008) and more recently by Ferrer-Vaquer et al. (2016), either within the PLCζ coding sequence or its associated bi-directional promoter in human patients (Nomikos et al., 2017a).

Interestingly, Torra-Massana et al. (2019) very recently reported six new PLCζ mutations after screening an egg activation deficiency group, one of which was previously described (Kashir et al., 2012b,c), in addition to four novel single-nucleotide missense mutations, located in the EF-hands, the X catalytic and C2 domains; while the sixth mutation identified was a frameshift variant, which was predicted to generate a truncated protein at the X-Y linker region (Torra-Massana et al., 2019). While further analysis indicated a potential deleterious effect of some of these mutations upon PLCζ activity within eggs, further biochemical analysis is required to ascertain whether such variants of PLCζ are deleterious as claimed at physiological levels, needing their accurate quantification in sperm and eggs. However, it is now clear that deleterious mutations in PLCζ may be more widespread than previously thought, appearing not only in the catalytic active site, but also in other vital regulatory regions of this essential sperm protein, afflicting its membrane and/or substrate binding, its Ca2<sup>+</sup> sensitivity, as well as its enzymatic activity.

Despite the relatively high success rate of ICSI in overcoming cases of failed fertilization after IVF, ∼30% of such cases still repeatedly fail ICSI (Flaherty et al., 1998; van der Westerlaken et al., 2005). Low global pregnancy success rates following ART have been attributed to poor embryogenesis following fertility treatment (Fauque et al., 2007; Pelinck et al., 2010). Importantly, such poor embryonic competency can be directly linked to PLCζ and competency of egg activation (Nomikos et al., 2017a; Kashir et al., 2018). Injection of increasing levels of PLCζ in human eggs results in increasing frequencies and amplitudes of Ca2<sup>+</sup> oscillations (Yamaguchi et al., 2017), which in turn affects subsequent gene expression (Ducibella et al., 2002, 2006). Furthermore, the frequency and amplitude of Ca2<sup>+</sup> oscillations has been shown to play an important role in compaction, and blastocyst formation (Swann and Ozil, 1994; Miyazaki and Ito, 2006). Finally, taking into consideration that the rate of progression to the 2- and 4-cell stages of human eggs following fertilization has been suggested as an indicator of normal embryogenesis (Wong et al., 2010), PLCζ-driven Ca2<sup>+</sup> oscillations may not only be required for egg activation, but can also be equally important for subsequent embryogenesis. Thus, abnormalities in sperm PLCζ levels may underlie not only infertility through fertilization failure, but also cases of male subfertility, whereby enough PLCζ may be delivered into the eggs to trigger activation, but prove insufficient for embryonic competence.

## CLINICAL APPLICATIONS OF PLCζ AND FUTURE DIRECTIONS

Currently, cases of defective egg activation are clinically resolved using assisted egg activation (AOA), involving artificially mediated Ca2<sup>+</sup> release (Santella and Dale, 2015; Kashir et al., 2018). Ca2<sup>+</sup> ionophores like ionomycin, calcimycin are currently used to overcome unexplained fertilization failure in couples, who repeatedly fail ICSI cycles (Fawzy et al., 2018; Norozi-Hafshejani et al., 2018). Recently, a meta-analysis indicated that use of Ca2<sup>+</sup> ionophores significantly improved fertilization and implantation rates in ICSI (Murugesu et al., 2017). However, Ca2<sup>+</sup> ionophores induce a single Ca2<sup>+</sup> transient, unlike the endogenous specific physiological pattern of Ca2<sup>+</sup> oscillations observed during normal fertilization (Rinaudo et al., 1997). In fact, microinjection of human recombinant PLCζ yielded higher blastocyst development rates than Ca2<sup>+</sup> ionophore treatment (Sanusi et al., 2015). Thus, PLCζ has long represented a physiologically endogenous alternative method to clinically treat cases of egg activation failure/deficiency, which would involve the in vitro production of active, purified versions of human recombinant PLCζ protein. Indeed, following initial difficulties, multiple studies made advancements in the production of such a desired product (Kashir et al., 2011b; Yoon et al., 2012), culminating in efforts by Nomikos et al. (2013b) who made a significant breakthrough by generating purified, highly active recombinant PLCζ, capable of inducing physiological patterns of Ca2<sup>+</sup> oscillations following microinjection into mouse and human eggs (Nomikos et al., 2013b). More importantly, recombinant PLCζ was able to effectively rescue failed egg activation in a prototype of male infertility (Nomikos et al., 2013b). Further attempts to develop the use of recombinant PLCζ protein as a therapeutic agent in a clinical setting are currently in progress, in order to eliminate any potential cytotoxic effects during embryonic development and confirm the overall safety of exogenous PLCζ on the subsequent offspring.

Finally, PLCζ not only represents a promising clinical therapeutic agent, but also a potentially powerful diagnostic biomarker, which may help in determining the criteria and requirements of the fertility treatment of male patients, significantly decreasing the number of cycles needed for a successful pregnancy to occur. Taking into consideration that PLCζ analysis might be beneficial to identify not only cases of ICSI-failure but also cases of male subfertility, a simple immunocytological approach to routinely examine PLCζ protein in sperm is currently widely regarded as a cost-effective approach, which could be easily applied by the majority of IVF clinics worldwide. However, it is essential that studies focus on the reliable investigation of PLCζ parameters in relation to sperm health, using robust protocols and ultra-specific antibodies, before such clinical promise can be achieved. Indeed, significant issues remain regarding such analyses, particularly in humans, where PLCζ antibodies used recognize multiple protein bands in addition to the full-length PLCζ protein. The same antibodies have also been used to identify different immunoblotting profiles (for detailed review see Kashir et al., 2018). Predictably, such shortcomings have led to conflicting results between the association of specific PLCζ localization patterns and quantification levels.

## ETHICS STATEMENT

Use of mouse sperm cells was carried out in accordance with the principles of the Basel Declaration and recommendations of the Animal Care and Use Committee (ACUC) at the Office of Research Affairs (ORA) at the King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia. The protocols utilized for the relevant study (RAC-216004) were approved by the ACUC.

## AUTHOR CONTRIBUTIONS

AS, JK, AT, and MN prepared the first draft of the manuscript, which was revised and approved by all authors.

## FUNDING

This work was supported by the Qatar University student grant to AS and FL (QUST-1-CMED-2020-3). JK was supported by a Healthcare Research Fellowship Award (HF-14-16) made by Health and Care Research Wales (HCRW), alongside a National Science, Technology, and Innovation plan (NSTIP) project grant (15-MED4186-20) awarded by the King Abdulaziz City for Science and Technology (KACST).

## REFERENCES

fcell-08-00028 January 27, 2020 Time: 16:3 # 7


uncapacitated, capacitated, and ionophore-treated human spermatozoa. Hum. Reprod. 23, 2513–2522. doi: 10.1093/humrep/den280


phospholipase C zeta (PLCzeta) leads to male infertility. Hum. Reprod. 27, 222–231. doi: 10.1093/humrep/der384



**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Saleh, Kashir, Thanassoulas, Safieh-Garabedian, Lai and Nomikos. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Toward Development of the Male Pill: A Decade of Potential Non-hormonal Contraceptive Targets

Katarzyna Kent1,2,3, Madelaine Johnston1,3, Natasha Strump1,3 and Thomas X. Garcia1,2,3 \*

<sup>1</sup> Department of Pathology & Immunology, Baylor College of Medicine, Houston, TX, United States, <sup>2</sup> Department of Biology and Biotechnology, University of Houston–Clear Lake, Houston, TX, United States, <sup>3</sup> Center for Drug Discovery, Baylor College of Medicine, Houston, TX, United States

With the continued steep rise of the global human population, and the paucity of safe and practical contraceptive options available to men, the need for development of effective and reversible non-hormonal methods of male fertility control is widely recognized. Currently there are several contraceptive options available to men, however, none of the non-hormonal alternatives have been clinically approved. To advance progress in the development of a safe and reversible contraceptive for men, further identification of novel reproductive tract-specific druggable protein targets is required. Here we provide an overview of genes/proteins identified in the last decade as specific or highly expressed in the male reproductive tract, with deletion phenotypes leading to complete male infertility in mice. These phenotypes include arrest of spermatogenesis and/or spermiogenesis, abnormal spermiation, abnormal spermatid morphology, abnormal sperm motility, azoospermia, globozoospermia, asthenozoospermia, and/or teratozoospermia, which are all desirable outcomes for a novel male contraceptive. We also consider other associated deletion phenotypes that could impact the desirability of a potential contraceptive. We further discuss novel contraceptive targets underscoring promising leads with the objective of presenting data for potential druggability and whether collateral effects may exist from paralogs with close sequence similarity.

Keywords: contraception, drug target, knockout mouse, spermatozoa, druggability

## INTRODUCTION

Currently, fertility control approaches for men fall into one of two categories: hormonal and nonhormonal. Several approaches have been investigated involving injectable or transdermal regimes using testosterone alone or combined with other molecules (Bagatell et al., 1993; Ilani et al., 2012). Although claims of total reversibility and full recovery to fertility have been made with hormonal contraception in males (Pasztor et al., 2017), prolonged use of exogenous hormones is associated with off-target effects, such as decreased high density lipoprotein cholesterol levels and potential cardiovascular risk in otherwise healthy men (Meriggiola et al., 1995). Therefore, there has been considerable interest in alternative methods for safe and reversible fertility control. Several non-hormonal methods are currently under development including gel-based obstruction of vas deferens, contraceptive vaccines, sperm-specific calcium ion channel blockers, and antispermatogenic indenopyridines with varied effectiveness and risks involved (Aitken, 2002; Hild et al., 2007; Morakinyo et al., 2009; Baggelaar et al., 2019).

#### Edited by:

Zhibing Zhang, Virginia Commonwealth University, United States

#### Reviewed by:

C. Yan Cheng, Population Council, United States Sigrid Hoyer-Fender, University of Göttingen, Germany

> \*Correspondence: Thomas X. Garcia thomas.garcia@bcm.edu

#### Specialty section:

This article was submitted to Cell Growth and Division, a section of the journal Frontiers in Cell and Developmental Biology

> Received: 23 July 2019 Accepted: 22 January 2020 Published: 26 February 2020

#### Citation:

Kent K, Johnston M, Strump N and Garcia TX (2020) Toward Development of the Male Pill: A Decade of Potential Non-hormonal Contraceptive Targets. Front. Cell Dev. Biol. 8:61. doi: 10.3389/fcell.2020.00061

#### THE 'POPULAR' PROTEINS: WHAT HAS BEEN TARGETED SO FAR?

Past efforts on the development of a safe non-hormonal contraceptive for men has resided on a diverse array of targets identified through both forward and reverse approaches. Forward approaches include identification of a target protein prior to drug development, while reverse approaches identify a drug with contraceptive effect prior to identification of the target protein. Often reverse approaches fail to identify specific drug targets with minimal side effects as discussed below.

#### Targets Involved in Sertoli-Germ Cell Interactions

Adjudin, a derivative of 1H-indazole-3-carboxylic acid, was shown to have potent anti-spermatogenic activity in rats, rabbits, and dogs prior to identification of its targets (Mok et al., 2011). Through further studies, it was later found to disrupt the Sertoligerm cell junction proteins ACTB, ITGA6, ITGB1, MYH11, and OCLN (Mok et al., 2011, 2012; Mruk and Cheng, 2011). Unsurprisingly, administration of this drug resulted in significant toxicity, including effects on liver and skeletal muscle (Mruk et al., 2006) necessitating efforts to lower systemic toxicity by developing a conjugate capable of delivering Adjudin directly to the testis (Mruk et al., 2006), and by utilizing different formulations of the drug for its oral use to reduce the effective dose (Chen et al., 2016a). However, the efficacy of these approaches has yet to be determined.

H2-Gamendazole, an additional indazole carboxylic acid analog, blocks spermatogenesis by inhibiting production of inhibin B by primary Sertoli cells (Tash et al., 2008a), inhibiting HSP90AB1 and EEF1A1, and increasing Interleukin 1 Alpha expression in Sertoli cells (Tash et al., 2008b), which is a known disruptor of Sertoli cell-spermatid junction integrity (Nilsson et al., 1998). Although Gamendazole was shown to induce significant loss of spermatids (Tash et al., 2008b) and cause complete infertility after just a single dose, the damages to the seminiferous epithelium were not fully reversible (Tash et al., 2008a). The failure of some animals to regain fertility may have resulted from the combined effects of this drug on EEF1A1 and HSP90 in less differentiated spermatogenic cells. Therefore, further progress on this drug has been halted until these issues can be addressed. Generation of potentially safer analogs of gamendazole are currently under development (Kanishchev and Dolbier, 2018).

#### Targets That Reside on the Testis Side of the Blood–Testis Barrier

One major obstacle in the development of a safe, reversible, and non-hormonal male contraceptive is identifying small molecule inhibitors that can penetrate the highly selective blood–testis barrier (BTB), or Sertoli cell barrier (SCB). Established through a strong network of tight junction proteins between adjacent Sertoli cells, the BTB/SCB partitions the seminiferous epithelium into two compartments: basal and adluminal (lumen-facing) (Franca et al., 2016). The basal compartment contains the undifferentiated spermatogonia, differentiating spermatogonia, and early differentiating spermatocytes. Meanwhile, the lumenfacing compartment contains differentiated spermatocytes, and round and elongated spermatids that—due to homologous recombination during meiosis—are no longer recognized by the immune system and thus require an immune privileged location. Many molecules have been shown to not cross the BTB/SCB and are thus used as molecular tools for measuring an intact BTB/SCB in manipulated animal models (Chen et al., 2018).

To the best of our knowledge, the small-molecule inhibitor JQ1, which inhibits the bromodomain-containing, testis-specific protein, BRDT, is currently the only molecule that has been shown to be present beyond the BTB/SCB (Matzuk et al., 2012). By binding to the BRDT acetyl-lysine binding pocket, JQ1 disrupts spermatogenesis at the spermatocyte and round spermatid stages, producing a reversible contraceptive effect in mice (Matzuk et al., 2012) that phenocopies the male infertility phenotype of BRDT knockout mice (Shang et al., 2007). Development of JQ1 was done in a target-based drug discovery approach focused on finding inhibitors of BRD4, a close paralog of BRDT. JQ1 blocks the production of sperm in the testes by inhibiting BRDT, however, it also acts on other bromodomains (BRD) of the bromo-and-extra-terminal (BET) proteins known to regulate transcription and DNA repair, which include important BRDT paralogs with oncogenic effects (French et al., 2008). BRD4, has become known as an important therapeutic target in several types of cancers, including breast and prostate cancers, as well as glioblastoma multiforme (GMB) (Shi et al., 2014; Zhou et al., 2016; Wen et al., 2019). JQ1 possesses anti-tumor effects (Shao et al., 2014; Das et al., 2015) and has been identified as a promising inhibitor for treating GBM (Wen et al., 2019). Along with other related pan-BET BRD compounds JQ1 was also associated with improvement of associated memory and enhancement of special memory precision in mice, showing potential for treatment of dementia (Benito et al., 2017). Given the therapeutic potential of JQ1 significant efforts have been made to understand the mechanisms of its action on the BET BRD family, giving insight into the intricate protein– protein interactions (PPI) with transcriptional complexes, and most importantly, providing means for enhancing inhibition specificity for a single BET (Lambert et al., 2019).

## Sperm Maturation in the Epididymis

The epididymis is a prime target for the development of a male contraceptive. This is because sperm leaving the testis are neither motile nor able to recognize or fertilize an egg; they must transit through the epididymis to acquire these abilities (Sullivan and Mieusset, 2016). Knockout mice of several epididymis-specific proteins have confirmed that the epididymis is essential for sperm maturation in the mouse (Davies et al., 2004; Roberts et al., 2006; Sipila et al., 2009). To the best of our knowledge, inhibitors that target an epididymis-specific protein do not yet exist. However, one of the most promising reversible, non-hormonal male contraceptives developed thus far has been against EPPIN (epididymal protease inhibitor; SPINLW1), which was developed through a reverse approach. EPPIN is a protein secreted by Sertoli cells and epithelial cells of the epididymis that gets deposited on

the surface of maturing spermatozoa (Silva et al., 2012). The small molecule inhibitor EP055, which targets EPPIN, causes reversible infertility in primates (Silva et al., 2012) most likely by targeting the testes and epididymis where the drug can be found (O'Rand et al., 2018). In fact, deposition of EPPIN on the sperm surface appears to be greatest in the epididymis, indicating that the drug primarily acts in the epididymis (O'Rand et al., 2018). Current efforts are focused on increasing the half-life of the drug molecule (O'Rand et al., 2018).

Additionally, the inhibitors cyclosporine A and FK506, which are used as immunosuppressant drugs, target the sperm calcineurin subunits PPP3CC and PPP3R2 in the epididymis causing reversible effects on sperm morphology and motility (Miyata et al., 2015). Treatment of mice with cyclosporine A or FK506 creates phenocopies of the sperm motility and morphological defects apparent in knockout mice, which appear within 4 to 5 days of treatment and are reversed a week after discontinuation (Miyata et al., 2015). Unfortunately, cyclosporine A and FK506 are undesirable candidates due to their immunosuppressive effect (Stucker and Ackermann, 2011).

## FACTORS THAT MAKE A PROMISING CANDIDATE TARGET

When evaluating potential druggability in a target-based drug discovery process, one must consider the protein properties that are required for safe and effective inhibition. Among the most significant is tissue expression specificity to minimize potential adverse effects; protein function and whether protein activity or interaction with other proteins is potentially druggable; sequence similarity to closely related paralogs that may be ubiquitously expressed; whether genetically manipulated animal models demonstrate a functional requirement for the target of interest; and other considerations as discussed below.

## Male Reproductive Tract Specificity

By ensuring in the first steps of development that candidate drug targets are near exclusively expressed in the male reproductive tract, the potential for off-target effects in humans is minimized. Gamendazole, for example, targets HSP90AB1 and EEF1A1, which are highly expressed in non-reproductive tissues (Djureinovic et al., 2014; Uhlen et al., 2015). Therefore, significant toxicity as evidenced through previous studies (Tash et al., 2008a,b) could have been predicted based on target gene expression analysis. Likewise, the toxicity of Adjudin has resulted in attempts to target the drug specifically to the testes through conjugation with FSH (Chen et al., 2016a), a complicated approach which may minimize, but not completely remove, offtarget effects. Since the identified molecular targets of Adjudin are ACTB, ITGA6, ITGB1, MYH11, and OCLN (Mruk and Cheng, 2011; Mok et al., 2012), which are all widely expressed in nonreproductive organs (Djureinovic et al., 2014; Uhlen et al., 2015), likewise, the toxicity of Adjudin could have been predicted and avoided had the drug's development began with a target-based discovery approach.

## Protein Druggability

Protein druggability is often based on the protein family for which other members are known drug targets (Hopkins and Groom, 2002). For instance, an enzyme with a known binding site might be an easier target when compared to a novel protein that has not already been categorized. The development of an inhibiting drug for an uncategorized protein might seem challenging, however, disrupting PPI has recently been gaining attention as one of the possible methods. Generally, PPI are considered to be more challenging than traditional drug targets due to the smaller protein interfaces and difficulty with finding a sufficiently binding ligand capable of interrupting the interaction site at a suitable concentration (Whitty and Kumaravel, 2006). However, PPI targets are not deemed undruggable, based on the discovery of small molecules capable of deeper and higher affinity binding within the contact surfaces of the target protein (Wells and McClendon, 2007). Therefore, although not initially compelling, uncategorized genes have the likelihood of becoming potential drug targets by using druglike compounds that can modulate the PPI at multiple interface sites—increasing the ligand binding affinity—and consequently lowering the necessary drug dose administered (Fuller et al., 2009). **Table 1** presents the list of male reproductive tract-specific genes expressed at various stages of spermatogenesis with assigned categories of protein families. Besides a handful of enzymes (HFM1, MOV10L1, PGK2, PRSS37, and LRGUK), two transcription factors (SOX30 and TERB1), a few epigenetically active proteins (SCML2 and TDRD5) and a sperm specific ion channel KCNU1, most of the proteins discussed in this review belong to an unknown category. However, the contraceptive potential of these genes should not be overseen, but rather investigated for the identification of a high-affinity small molecule that can either (1) interfere with PPI, or (2) target the protein specifically for degradation as discussed below.

Many of the potential drug targets, such as non-enzymatic proteins, are uncategorized and identified as "undruggable" due to various challenges with existing targeting approaches. However, an emerging targeted protein degradation method called Proteolysis Targeting Chimeras (PROTACs) is a promising technique that can address these issues. The traditional approach to most enzymatic proteins serves to interfere with the functional aspect of the protein target, whereas PROTACs eliminate the protein through utilization of the ubiquitin-proteasome system to promote selective degradation (Bondeson et al., 2015; Lu et al., 2015; Olson et al., 2018). Thus, PROTACs do not require that an identified small molecule both bind with high affinity and reduce activity of the target protein (through interference of the binding pocket, etc.). PROTACs only requires that the identified small molecule bind with high affinity to the protein target. Additional design and chemistry are conducted to conjugate the small molecule to a high affinity E3 ubiquitin ligase ligand that results in target protein ubiquitination. There are currently various combinations of PROTACs developed to overcome the limitations of cell permeability, stability, solubility, selectivity, and tissue distribution (Brooks et al., 2005;

Kent et al.

fcell-08-00061 February 24, 2020 Time: 17:2 # 4


(Continued)

Potential Non-hormonal Male Contraceptive Targets

#### TABLE 1 | Continued


 TM, transmembrane. fcell-08-00061 February 24, 2020 Time: 17:2 # 5

Bechara and Sagan, 2013; Ottis et al., 2017; Bondeson et al., 2018). Therefore, PROTACs technology provides the potential to greatly promote the development of contraceptive drugs against the "undruggable" non-enzymatic protein targets.

### Sequence Similarity to Known Paralogs

The probability of two or more paralogs sharing a specific function increases with the percentage of sequence similarity (Zallot et al., 2016). If a reproductive tract-specific target is found to have high sequence similarity to a ubiquitously expressed protein or paralog, especially in the potentially druggable domain, this would make the target a poor choice for contraceptive development due to potential off-target effects. However, if a reproductive tract-specific protein has one or more ubiquitously expressed paralogs with low sequence similarity, this indicates that the function of the expressed proteins in the reproductive tract could be disrupted without significant offtarget risks. Considerably high contraceptive potential resides in genes without a confirmed paralog; however, one must remain cautious in assuming a complete lack of paralogs as the druggable domain may be present in completely unrelated proteins, and further investigation should be conducted before proceeding to contraceptive drug development.

#### Validation Through Ablation – Creating Functional Knockouts to Verify Contraceptive Potential

Mice serve as one of the most efficient and effective models to understanding human physiology for a variety of reasons. The mouse genome is very well-characterized with almost all genes sharing similar functions to human orthologs (Cheng et al., 2014; Lin et al., 2014). Mice are biologically very similar, yet because they are small and have very short lifespans compared to humans, developmental processes can be studied economically and at an accelerated rate. Genetically manipulated knockout mouse models have significantly advanced our understanding of male gamete differentiation and the molecular mechanisms underlying male fertility. Nevertheless, there is still much to be learned about these intricate processes considering only approximately half of the protein coding genes have been individually ablated in mice (Dickinson et al., 2016; Smith et al., 2018). By ensuring that a knockout animal model of an individual protein target of interest leads to a complete male infertility phenotype, not subfertility, confidence in potential drug efficacy is established.

The current popularity of CRISPR-generated knockout mouse models as a method to study the function of a gene in vivo has dominated over other approaches. While knockout methods lead to a complete depletion of gene expression, or complete elimination of gene function, methods involving gene knockdowns lead to reduction in protein expression, the results of which may yield valuable information. RNA interference (RNAi) was found to be useful for silencing gene expression in mammals and employed to study the functional relevance of genes in a relatively fast and easy way. Past reproductive studies have utilized short interfering RNA (siRNA) to determine functional significance of genes in male mouse fertility and has, in some cases, resulted in impaired or complete infertility phenotype in viable mice (Nagai et al., 2011; Welborn et al., 2015). Several interesting experiments that compared the phenotypes resulting from knocking-out and knocking-down mouse genes have revealed that the two methods are complimentary in conducting gene functional studies (De Souza et al., 2006). The focus of this review, however, lies in male reproductive genes whose functional relevance was determined through knockout mouse models, since incisive genetic approaches provide a more definitive foundation for establishing the contraceptive potential of a novel reproductive tract-specific gene. A variable and indeterminate level of off-target effects introduced from transfection or lentivirus-mediated infection cannot be excluded as confounding variables in knockdown experiments.

Many reproductive tract-specific genes have been identified as dispensable for male fertility when individually ablated in mice (Miyata et al., 2016; Holcomb et al., 2020; Lu et al., 2019). Functional compensation through upregulation of other functionally similar genes or molecular pathways may explain the lack of phenotype (Marschang et al., 2004; Hanada et al., 2009; Kashiwabara et al., 2016). Therefore, identifying a small molecule inhibitor against an individually dispensable target would yield no contraceptive effect.

As previously discussed, BRDT is a testis-specific gene that when individually ablated in mice results in arrest of male meiosis, azoospermia, and complete male infertility (Barda et al., 2016). Since BRDT has enough sequence dissimilarity— >55% sequence dissimilarity with its closest paralogs that are ubiquitously expressed, BRD4 and BRD2, development of BRDT inhibitors with enhanced specificity and drug selectivity is feasible and currently underway (Matzuk et al., 2012; Miller et al., 2016). On the other end of the spectrum, GJA1 is an example of a gene that when conditionally ablated in the Sertoli cells of male mice leads to male infertility (Brehm et al., 2007). However, because the gene is not exclusively expressed in the reproductive tract and (Djureinovic et al., 2014; Uhlen et al., 2015) other knockout animal models of this gene display a variety of nonreproductive phenotypes including severe heart abnormalities (Gutstein et al., 2001; Liao et al., 2001), thus demonstrating that this target is not suitable for the development of a safe male contraceptive.

## Human Mutations Implicated in Male Infertility

Mutations causing male infertility in humans might also be informative in the context of contraception and should be taken into consideration in the search for a potential nonhormonal target. There are 1,517 human genes associated with male infertility and/or either azoospermia, globozospermia, or oligospermia as reported by GeneCards and MalaCards (Stelzer et al., 2016; Rappaport et al., 2017). Of these 1,517 genes, 202 genes are reproductive tract-specific in humans as reported through the CITDBase Contraceptive Target Database (GTEx Consortium, 2013; Djureinovic et al., 2014; Uhlen et al., 2015; Schmidt et al., 2017; Lee, 2019; Samaras et al., 2019) and/or Djureinovic et al. (2014) and Uhlen et al. (2015) (**Figure 1**). Of these 202 genes, 21 genes do not have a corresponding

mouse ortholog and 3 of these genes (KLK2, CDY2A, and RHOXF2) may be of potential interest for further study as they encode an enzyme, an epigenetic protein, and a transcription factor, respectively; all proteins with potential druggable activity (**Supplementary Table S1**). The remaining 18 genes without a corresponding mouse ortholog encode proteins of unknown drug target type (**Supplementary Table S1**). Of the 202 reproductive tract-specific human genes associated with human infertility, 78 genes have a corresponding mouse ortholog displaying male infertility in the mouse (**Figure 1** and **Supplementary Table S1**). The remaining 124 genes either (1) have a single mouse ortholog each and none have been knocked-out in the mouse, (2) have two or more mouse ortholog genes, whereby none or an incomplete number of orthologs have been knocked-out, and of those that have been knocked-out, none display a male infertility phenotype, (3) have a single mouse ortholog or multiple mouse orthologs that when individually ablated all lack a male infertility phenotype, or (4) have multiple mouse orthologs that have all been knocked out and that individually do not display a male infertility phenotype (**Figure 1** and **Supplementary Table S1**). While the first two categories (categories 1 and 2) require further study for functional validation of the contraceptive potential of these genes, the latter categories (categories 3 and 4) do not require any further study. Genes listed in the first two categories (47 genes total) encode 2 enzymes (KLK3 and SPAM1), 1 kinase (TSSK2), 3 transcription factors (HSFY1, TGIF2LX, and TGIF2LY), and 41 proteins of unknown drug target type (**Supplementary Table S1**). Of the 202 reproductive tract-specific human genes associated with human infertility, 32 genes are mentioned in **Figure 2**, and four genes (BOLL, KCNU, SPATA22, and TEX101) are discussed in further detail within this review.

## Site of Target Expression

Potential contraceptives could hit targets expressed along various stages of sperm development in the testis and epididymis, including targets active during the maintenance of the progenitor spermatogonia pool, entry into and passage through the various stages of meiosis, spermatid development and release, and sperm maturation through the epididymis. The contraceptive potential at these various stages varies in advantages and disadvantages. Targeting genes involved in early sperm development, for instance, could potentially be more effective, as suggested by several previous reviews identifying groups of male reproductive tract-expressed genes as promising drug targets (Schultz et al., 2003; Archambeault and Matzuk, 2014; Payne and Goldberg, 2014; O'Rand et al., 2016). However, disrupting the early stages of spermatogenesis poses the risk of testicular atrophy, longer recovery, and an increased

phenotype as follows: red = reproductive tract-specific displaying male infertility phenotype; green = non-reproductive tract-expressed and male infertility phenotype; blue = reproductive tract-specific displaying fertile or subfertile phenotype; orange = reproductive tract-specific with unknown fertility phenotype; black = non-reproductive tract-expressed displaying fertile or subfertile phenotype; gray = non-reproductive tract-expressed with unknown fertility phenotype. Expression and phenotype data obtained from Contraceptive Infertility Target DataBase (CITDBase), Human Protein Atlas (HPA), Ensembl Biomart, Mouse Genome Informatics (MGI), and the International Mouse Phenotyping Consortium (IMPC). \*denotes genes implicated in human male infertility according to GeneCards and MalaCards.

possibility of irreversibility. Genes involved in later phases of spermatogenesis, namely spermiogenesis, acrosome and flagella formation, spermiation, and sperm maturation, would be more desirable contraceptive targets because testicular size would most likely remain unaffected, with a quicker, more reliable return to full fertility. Functional analysis of the expression patterns and specificity of male reproductive tract genes is imperative as it often provides additional insight into the molecular mechanisms of various stages of spermatogenesis, understanding of which is necessary in the process of developing a safe and effective non-hormonal male contraceptive.

#### Other Considerations

Identification of a drug-like small molecule that can effectively modulate the activity of a given target should also be assessed based on protein properties such as structure, size, and complexity (Kozakov et al., 2015). Targets that contain transmembrane helices could increase the difficulty of obtaining a properly folded and soluble protein for drug selection purposes. Likewise, due to glycosylation and processing through the secretory pathway, secreted proteins require protein production in eukaryotic expression systems, which could increase the difficulty of obtaining suitable quantities of purified protein. Therefore, whether drug development is feasible resides on careful consideration of the biophysical properties of the protein.

### TARGET GENES

#### Previously Reviewed 'Novel' Target Genes

Several notable reviews published in the last decade have mentioned promising, non-hormonal, contraceptive leads that include both meiotically and post-meiotically expressed genes that are testis-specific or epididymis-specific genes required for sperm maturation (O'Rand et al., 2011; Alves et al., 2014; Archambeault and Matzuk, 2014; Murdoch and Goldberg, 2014; Payne and Goldberg, 2014; Chen et al., 2016b; O'Rand et al., 2016; Drevet, 2018) (**Figure 2**). Some are, in fact, not reproductive tract-specific, but are still required for fertility, while others that are indeed reproductive tract-specific, lead to subfertility, not infertility, which is an ineffective and highly undesirable outcome for a contraceptive. Thus, the genes in **Figure 2** are color-coded according to reproductive tract-specificity and infertility phenotype in the mouse. It is worth noting that while the reviews were restricted to the past decade, the reproductive tract-specific genes mentioned in these reviews were first identified beyond the past decade including some that were reported in 2006 and earlier, such as TNP1 (Yu et al., 2000), CATSPER1 (Ren et al., 2001), and TEX14 (Greenbaum et al., 2006).

## In This Review

In this review, 45 genes are discussed in further detail as they were identified in the last decade as required for male fertility through knockout mouse studies, whereby many of these genes were not discussed or discussed minimally in any previous male contraceptive drug target review. Henceforth, these 45 genes will be listed according to their published expression patterns (**Figure 3**), which is an important criterium that should be considered in the selection of a safe and/or desirable contraceptive target. It is worth noting that nearly all of the genes mentioned below are male reproductive tractspecific or highly enriched as reported in the literature and with additional confirmation through the published expression data from CITDBase (GTEx Consortium, 2013; Uhlen et al., 2015; Schmidt et al., 2017; Lee, 2019; Samaras et al., 2019) and/or the Human Protein Atlas (Djureinovic et al., 2014; Uhlen et al., 2015). A graphical summary of the RNAseq-based expression data for these 45 genes is depicted in **Figure 4**. Those that are not reproductive tract-specific are identified as such through the level of non-reproductive tissue expression. The following discussion provides detailed information about individual genes with focus on determining the contraceptive potential of each. Additional relevant information for these genes is listed in **Table 1**.

## Spermatogonia

In the last decade, only four reproductive tract-specific genes with infertile mouse models have been identified that fit the criteria of being expressed as early as the spermatogonia stage: ASZ1 (Ma et al., 2009), MOV10L1 (Frost et al., 2010; Zheng et al., 2010), SCML2 (Hasegawa et al., 2015), and TEX101 (Fujihara et al., 2013; Li et al., 2013) (**Table 1**). With expression as early as the spermatogonia stage (Teng et al., 2006; Ma et al., 2009; Frost et al., 2010; Maezawa et al., 2018), a deficiency of these genes' functions will likely require at least 2 months prior to a contraceptive effect, at least 2 months for recovery, and in some cases, depending on the gene, the potential for irreversibility of the contraceptive effect. Nevertheless, since this form of contraceptive may be considered desirable due to the permanence of the effect, and the need for a compound that can traverse the BTB is not required, these genes are worthy of consideration. Since the conserved domains of human ASZ1, MOV10L1, and TEX101 show sequence similarity below 40%


FIGURE 4 | Digital PCR (heatmap) depicting the average transcripts per million (TPM) value per tissue per gene from human RNAseq data published by the Human Protein Atlas (Djureinovic et al., 2014; Uhlen et al., 2015). White = 0 TPM, Black ≥ 30 TPM. The genes are ordered from most reproductive tract-specific to least based on the level of non-reproductive tissue expression. "Fold" = testis/max normal. The data was obtained from a tab-separated file including Ensembl gene identifier, analyzed sample, and TPM value per gene that was downloaded from the Human Protein Atlas website. The expression profile of the housekeeping genes, GAPDH, is included.

to their respective, non-reproductive tract expressed paralogs, ANKRD34C, CT55, and CD177, then these protein targets have a reasonable potential for drug specificity with a low risk of collateral effects. However, as it has been previously reported that human SCML2 is expressed ubiquitously at low levels (Bonasio et al., 2014) (**Figure 4**), low but potentially physiologically relevant expression of SCML2 across many nonreproductive tissues may therefore be of concern in the process of drug development. Additionally, the closest paralog to SCML2, SCMH1, is ubiquitously expressed and shows 60% sequence similarity in its conserved domain. Based on this information, human SCML2 may not be an ideal contraceptive target.

#### Spermatocytes

The genes discussed here are expressed in spermatocytes during preleptotene, leptotene, zygotene, pachytene, and/or diplotene stages and typically serve an essential role in meiosis I or meiosis II. A deficiency of these early stage genes results in male infertility typically due to meiotic arrest, which indicates a strong contraceptive potential. Important to note: although these genes may not directly impact spermatogonial stem cell self-renewal, and a healthy pool of spermatogonial stem cells may still remain for the regeneration of spermatogenesis, recovery time following cessation of targeting these genes may still take considerably longer than drugs targeting later stages of spermatogenesis or sperm. Consistently, the onset of a reliable contraceptive effect could take weeks, which may be an undesirable timeframe for drug action.

In the last decade, at least twenty-five reproductive tractspecific genes with infertile mouse models have been identified that fit the criteria of being expressed in spermatocytes (**Table 1**). Of these, seven genes—MEIOB, MEIOC, PIH1D3, SPATA22, SYCE3, TERB1, and TOPAZ1—encode the most promising target candidates as they do not have an associated paralog, or conserved domain with sequence similarity to any known protein, and thus these proteins bear the lowest risks of generating off-target effects from inhibiting compounds. Additionally, HORMAD2 is also an attractive target considering it only has one paralog, HORMAD1, which is also highly enriched in the male reproductive tract and leads to male infertility phenotype when ablated in mice (Shin et al., 2010; Daniel et al., 2011; Kogo et al., 2012b). Although HORMAD1 shows low, but appreciable gene expression in two non-reproductive tissues (**Figure 4**) this fact may be offset by the sequence dissimilarity between HORMAD1 and HORMAD2 (50% whole protein and 40% conserved domain) which can aid in the identification of a selective inhibitor with careful drug selection and design. C14orf39 and CNBD2 would both be attractive candidates because they also lack paralogs, however both have tissue specificity issues of concern (**Figure 4**), and Cnbd2 null mice display incomplete penetrance (Krahling et al., 2013), which may translate to potential ineffectiveness in humans. Therefore, while MEIOB, MEIOC, PIH1D3, SPATA22, SYCE3, TERB1, and TOPAZ1 are ideal candidates, C14orf39 and CNBD2 are not.

The next most promising candidates (another six genes)— BTBD18, CCDC63, CCDC155, HFM1, MCMDC2, and TDRD5 encode proteins that show sequence similarity below 40% to their respective, non-reproductive tract expressed paralogs, and within the conserved domains of their paralogs, KLHL26, SSH2, CCDC114, SNRNP200, MCM3, and TDRD1; thus, these potential targets have a reasonable potential for drug specificity with a low risk of collateral effects. However, the following six candidates—BOLL, FBXO43, INSL6, NUP210L, SOX30, and SPDYA—may require careful drug selection and design since the conserved domains of these proteins are 48–60% similar to the conserved domains of their respective, non-reproductive tract expressed paralogs, DAZ4, FBXO5, RLN1, NUP210, SOX7, and SPDYC. With respect to NUP210L, spermatid-Sertoli cell interaction was severely impaired in the knockout mouse model, resulting in Sertoli cell degeneration (Walters et al., 2009). Since this may cause irreversible disruption of spermatogenesis, this candidate may pose potential irreversibility issues. LY6K would be an attractive candidate because its closest ubiquitously expressed paralog, GML, has only 28% sequence similarity to LY6K; however, low, but appreciable expression of LY6K in non-reproductive tissues is of concern (**Figure 4**).

The most difficult spermatocyte-expressed candidates to target specifically are RAD21L1 and PGK2, which have ubiquitously expressed paralogs, RAD21 and PGK1, that have 83 and 87% similarity in their respective conserved domains. Since human mutations in RAD21 and PGK1 are characterized by significant disorders affecting numerous organ systems— Cornelia de Lange syndrome 4 (Deardorff et al., 2012), Mungan syndrome (Bonora et al., 2015), and phosphoglycerate kinase 1 deficiency (Fermo et al., 2012)—extraordinary effort would need to be made to generate specific drug molecules against RAD21L1 and PGK2.

## Round and Elongated Spermatids

The proteins discussed here are expressed during the later stages of spermatogenesis and sperm maturation (**Table 1**). These proteins are found in round or elongating spermatids, either in the acrosome, acroplaxome, basal body, manchette, or flagellum during spermiogenesis, and these proteins function in either proper sperm head formation and attachment, midpiece formation, acrosome formation and attachment, generating sperm with normal motility, or generating sperm capable of normal sperm-zona binding. Targeting genes at these postmeiotic stages is more likely to act within a faster timeframe and lead to better and potentially faster recovery of fertility upon cessation of drug.

In the last decade, at least nine reproductive tract-specific genes with infertile mouse models have been identified that fit the criteria of being expressed in spermatids (**Table 1**). Again, the most attractive candidates are those that show excellent reproductive tract-specificity and that do not have any ubiquitously expressed paralogs, or that have ubiquitously expressed paralogs with low sequence similarity. CCDC62, SPACA1, and TCTE1 do not have any paralogs and TDRD12, CCDC42, PRSS37, and LRGUK have ubiquitously expressed paralogs, TDRD15, CFAP73, KLK15, and PPP1R42, that share less than 40% sequence similarity at the whole protein and conserved domain level. CALR3 has a ubiquitously expressed paralog, CALR, that shows 51% and 55% sequence

similarity at the whole protein and conserved domain level, respectively, which may necessitate careful drug selection and design to ensure off-target effects are minimized. RIMBP3, RIMBP3B, and RIMBP3C in humans are all testis-specific and RIMBP3B and RIMBP3C share conserved domains with identical protein sequences, indicating a strong evolutionary requirement for the function of these proteins and high likelihood of targeting more than one isoform with one small molecule inhibitor. However, the closest ubiquitously expressed paralog to these proteins is RIMBP2, which shares 76% sequence similarity in its conserved domain, which may be of concern in finding a specific drug molecule that targets this region.

#### Spermatozoa

Genes in this group are highly expressed in spermatozoa and disrupting their function would most likely affect spermatozoa maturation as they pass through the epididymis, while still maintaining the pool of testicular spermatogonia and spermatocytes, or inhibit proper spermatozoa function after sperm maturation. These targets are typically localized to the neck, principal piece, flagellum, or the central microtubule apparatus of mature spermatozoa. Since these targets act late, it is important to note that some—depending on the binding kinetics of the drugs and if the drugs are reversible or irreversible inhibitors—may only provide a momentary decrease in sperm function while in the male reproductive tract, and shortly thereafter, but not indefinitely in the female reproductive tract after ejaculation as the drug concentration invariably decreases over time. Thus, although this category of drug may be the most desirable due to having fastest onset of drug action, fastest recovery after cessation of drug, and no effect on testicular size, there may be additional challenges to address during drug development to ensure contraceptive efficacy.

In the last decade, at least seven reproductive tract-specific genes with infertile mouse models have been identified that fit the criteria of being present in mature sperm (**Table 1**). CATSPERD, KCNU1, PMIS2, and SUN5 show the lowest potentials for off-target effects as their closest ubiquitously expressed paralogs, CASC4, KCNMA1, SYNDIG1L, and SUN1, have less than 50% similarity at the whole protein and conserved domain level. After that, RSPH6A, has 52% (whole) and 63% (CD) similarity to its closest ubiquitously expressed paralog, RSPH4A, which may make it a promising lead with additional attention during drug selection and design to minimize offtarget effects. Although, CFAP54 would be an excellent candidate with less than 22% similarity (whole and CD) to its closest ubiquitously expressed paralog, NPBWR2, expression in cilia of the respiratory epithelium (Djureinovic et al., 2014; Uhlen et al., 2015) and additional low but potentially physiologically relevant expression across many non-reproductive tissues (**Figure 4**) indicate targeting this candidate may yield side effects. Likewise, ATP1A4 is not an ideal candidate due to having high sequence similarity—81% similarity (whole and CD)—to its closest ubiquitously expressed paralog, ATP1A2, and also having low but potentially physiologically relevant expression levels across many non-reproductive tissues (**Figure 4**).

## Additional Potential Targets

Not discussed in terms of potential drug target specificity, however of additional potential interest are the following genes that are also reproductive tract-specific in humans according to CITDBase (GTEx Consortium, 2013; Uhlen et al., 2015; Schmidt et al., 2017; Lee, 2019; Samaras et al., 2019) with mouse models displaying male infertility phenotype published in peer-reviewed journals in the last 10 years: 3 genes encoding enzymes [ENO4 (Nakamura et al., 2013), PNLDC1 (Nishimura et al., 2018), and SPINK2 (Lee et al., 2011)] and 8 genes encoding proteins of unknown drug target type [M1AP (Arango et al., 2013), MEIG1 (Zhang et al., 2009), MEIKIN (Kim et al., 2015), NXF2 (Pan et al., 2009), ODF1 (Yang et al., 2012), PPP3R2 (Miyata et al., 2015), SMC1B (Revenkova et al., 2010), and SPATA16 (Fujihara et al., 2017)]. Furthermore, the following genes are reproductive tractspecific in humans according to CITDBase (GTEx Consortium, 2013; Uhlen et al., 2015; Schmidt et al., 2017; Lee, 2019; Samaras et al., 2019) with mouse models displaying male infertility phenotypes as reported by the International Mouse Phenotyping Consortium (IMPC) (Munoz-Fuentes et al., 2018): 1 gene encoding an enzyme (ADAD2); 1 gene encoding an epigeneticrelated protein (PHF7); and 11 genes encoding proteins of unknown drug target type (ACTL7B, ADGB, ARRDC5, C11orf94, C16orf92, C3orf20, DNAH17, FBXO47, NUTM1, ODF4, TEX38). The expression pattern of these additional individually published and IMPC-reported genes are listed in **Figure 4**.

## CONCLUSION

Global demand for the development of a "male pill" is at an all-time high with the exponential growth of the human population. Efforts to reduce unplanned pregnancies are recognized as the search for a safe and effective method of male contraception has been a decades-long quest (Prasad and Rajalakshmi, 1976; Frick and Aulitzky, 1988; Herndon, 1992; Waites, 1993; Gottwald et al., 2006; Mruk, 2008). Identification of reproductive tract-specific targets through transcriptomic and proteomic approaches followed by validation of their functional requirement using knockout mice models has helped advance this quest. In this review, we discuss novel reproductive tractspecific protein targets that have been identified in the past 10 years, their potential druggability, the factors that contribute to their druggability and why they should be taken into consideration when selecting male contraceptive targets. We direct the reader to consider that optimal gene targets are those that contribute to the later phases of spermatogenesis as disrupting genes in the earlier phases could potentially lead to permanent infertility. Additionally, the reader is directed to consider the potential adverse effects that may exist when targets whose protein sequences bear high sequence similarity to other ubiquitously expressed proteins. This information can be valuable in future studies since off-target effects can considerably hamper development of a safe, non-hormonal male contraceptive. Continued persistence in the search for an optimal protein target should lead to a clinically approved, affordable product.

#### AUTHOR CONTRIBUTIONS

fcell-08-00061 February 24, 2020 Time: 17:2 # 13

KK and TG designed the research and wrote the manuscript. KK, MJ, NS, and TG performed the research and analyzed the data.

## FUNDING

This research was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD095341 to TG).

#### REFERENCES


## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2020.00061/ full#supplementary-material

TABLE S1 | Orthologous mouse genes and mouse model information of 202 reproductive tract-specific human genes associated with male infertility in humans. Mouse ortholog gene symbols and phenotype information collected from Ensembl BioMart and MGI. For publication references refer to MGI. "Incomplete mouse data available" means not all orthologs have been knocked-out. ‡Genes that have been knocked-out are indicated in bold. <sup>∗</sup> In instances where more than one mouse ortholog exists, the male infertile mouse ortholog is indicated in bold.



prophase via the recruitment of ATR activity. Genes Cells 17, 897–912. doi: 10.1111/gtc.12005




1-deficient mice. Proc. Natl. Acad. Sci. U.S.A. 97, 4683–4688. doi: 10.1073/pnas. 97.9.4683


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Kent, Johnston, Strump and Garcia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.