### Frontiers Copyright Statement

© Copyright 2007-2018 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-611-6 DOI 10.3389/978-2-88945-611-6

### 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

# SIGNALING PATHWAYS IN DEVELOPING AND PATHOLOGICAL TISSUES AND ORGANS OF THE CRANIOFACIAL COMPLEX

Topic Editors:

Thimios Mitsiadis, University of Zurich, Switzerland Claudio Cantù, Linköping University, Sweden Lucia Jimenez-Rojo, University of Zurich, Switzerland

The cover represents a cartoon of the main signaling pathways (BMP, TGF, Shh, Wnt, FGF, RA and Notch) operating during craniofacial development and pathology. Image: Thimios Mitsiadis.

Head formation requires the well-orchestrated and harmonized development of various tissues and organs within the craniofacial complex. A big variety of signaling pathways are involved in this process by controlling cell proliferation, migration, differentiation, tissue morphogenesis, homeostasis and regeneration. Deregulation and malfunction of these signaling molecules may lead to mild or severe craniofacial pathologies.

This eBook is a collection of articles dealing with a variety of important signals involved in the control of developmental and pathological events of craniofacial organs and tissues. These recent advances show the importance of signaling pathways in craniofacial physiology and pathology and generate important new knowledge aiming the development of new pharmaceutical products that mimic and/or block the actions of specific molecules.

Citation: Mitsiadis, T., Cantù, C., Jimenez-Rojo, L., eds (2018). Signaling Pathways in Developing and Pathological Tissues and Organs of the Craniofacial Complex. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-611-6

# Table of Contents

*07 Editorial: Signaling Pathways in Developing and Pathological Tissues and Organs of the Craniofacial Complex*

Thimios A. Mitsiadis, Claudio Cantù and Lucia Jimenez-Rojo

# CHAPTER ONE

# SIGNALING MOLECULES DURING TOOTH DEVELOPMENT


Thimios A. Mitsiadis, Pierfrancesco Pagella and Claudio Cantù

*29 Expression of Nerve Growth Factor (NGF), TrkA, and p75NTR in Developing Human Fetal Teeth*

Thimios A. Mitsiadis and Pierfrancesco Pagella

*39 Distribution of Syndecan-1 Protein in Developing Mouse Teeth* Anna Filatova, Pierfrancesco Pagella and Thimios A. Mitsiadis

# CHAPTER TWO

# SIGNALS INVOLVED IN ENAMEL AND DENTIN FORMATION AND PATHOLOGY


Sophia Houari, Sophia Loiodice, Katia Jedeon, Ariane Berdal and Sylvie Babajko

*69 The Role of Na:K:2Cl Cotransporter 1 (NKCC1/SLC12A2) in Dental Epithelium During Enamel Formation in Mice*

Rozita Jalali, Johannes C. Lodder, Behrouz Zandieh-Doulabi, Dimitra Micha, James E. Melvin, Marcelo A. Catalan, Huibert D. Mansvelder, Pamela DenBesten and Antonius Bronckers

*82 Detection of a Novel* DSPP *Mutation by NGS in a Population Isolate in Madagascar*

Agnès Bloch-Zupan, Mathilde Huckert, Corinne Stoetzel, Julia Meyer, Véronique Geoffroy, Rabisoa W. Razafindrakoto, Saholy N. Ralison, Jean-Claude Randrianaivo, Georgette Ralison, Rija O. Andriamasinoro, Rija H. Ramanampamaharana, Solofomanantsoa E. Randrianazary, Louise H. Ralimanana, Béatrice Richard, Philippe Gorry, Marie-Cécile Manière, Jeanne A. Rasoamananjara, Simone Rakoto Alson and Hélène Dollfus

# *90 Corrigendum: Detection of a Novel* DSPP *Mutation by NGS in a Population Isolate in Madagascar*

Agnès Bloch-Zupan, Mathilde Huckert, Corinne Stoetzel, Julia Meyer, Véronique Geoffroy, Rabisoa W. Razafindrakoto, Saholy N. Ralison, Jean-Claude Randrianaivo, Georgette Ralison, Rija O. Andriamasinoro, Rija H. Ramanampamaharana, Solofomanantsoa E. Randrianazary, Louise H. Ralimanana, Béatrice Richard, Philippe Gorry, Marie-Cécile Manière, Jeanne A. Rasoamananjara, Simone Rakoto Alson and Hélène Dollfus

# CHAPTER THREE

# SIGNALS INVOLVED IN DEVELOPMENT AND PATHOLOGY OF TISSUES AND ORGANS OF THE CRANIOFACIAL COMPLEX


Tathyane H. N. Teshima, Silvia V. Lourenco and Abigail S. Tucker

*129 Understanding Mechanisms of GLI-Mediated Transcription During Craniofacial Development and Disease Using the Ciliopathic Mutant,*  talpid2

Ya-Ting Chang, Praneet Chaturvedi, Elizabeth N. Schock and Samantha A. Brugmann

*142 Mesenchymal Remodeling During Palatal Shelf Elevation Revealed by Extracellular Matrix and F-Actin Expression Patterns*

Matthias Chiquet, Susan Blumer, Manuela Angelini, Thimios A. Mitsiadis and Christos Katsaros

*155 MORN5 Expression During Craniofacial Development and its Interaction With the BMP and TGF*β *Pathways*

Petra Cela, Marek Hampl, Katherine K. Fu, Michaela Kunova Bosakova, Pavel Krejci, Joy M. Richman and Marcela Buchtova

*167 MicroRNA Profiling During Craniofacial Development: Potential Roles for*  Mir23b *and* Mir133b

Hai-Lei Ding, Joan E. Hooper, Peter Batzel, B. Frank Eames, John H. Postlethwait, Kristin B. Artinger and David E. Clouthier

*183 Ephrin Ligands and Eph Receptors Show Regionally Restricted Expression in the Developing Palate and Tongue*

Guilherme M. Xavier, Isabelle Miletich and Martyn T. Cobourne

# CHAPTER FOUR

# SIGNALS INVOLVED IN HARD TISSUE FORMATION AND PATHOLOGY


Shuji Oishi, Yasuhiro Shimizu, Jun Hosomichi, Yoichiro Kuma, Hideyuki Maeda, Hisashi Nagai, Risa Usumi-Fujita, Sawa Kaneko, Naoki Shibutani, Jun-ichi Suzuki, Ken-ichi Yoshida and Takashi Ono

# CHAPTER FIVE

# STEM CELLS IN ORGANS AND TISSUES OF THE CRANIOFACIAL COMPLEX

*232 Calvarial Suture-Derived Stem Cells and Their Contribution to Cranial Bone Repair*

Daniel H. Doro, Agamemnon E. Grigoriadis and Karen J. Liu

*242 Monitoring Notch Signaling-Associated Activation of Stem Cell Niches Within Injured Dental Pulp*

Thimios A. Mitsiadis, Javier Catón, Pierfrancesco Pagella, Giovanna Orsini and Lucia Jimenez-Rojo

*251 The Effects of Photobiomodulation of 808 nm Diode Laser Therapy at Higher Fluence on the* in Vitro *Osteogenic Differentiation of Bone Marrow Stromal Cells*

Andrea Amaroli, Dimitrios Agas, Fulvio Laus, Vincenzo Cuteri, Reem Hanna, Maria Giovanna Sabbieti and Stefano Benedicenti


Emily J. Lodge, John P. Russell, Amanda L. Patist, Philippa Francis-West and Cynthia L. Andoniadou

# Editorial: Signaling Pathways in Developing and Pathological Tissues and Organs of the Craniofacial Complex

### Thimios A. Mitsiadis <sup>1</sup> \*, Claudio Cantù<sup>2</sup> and Lucia Jimenez-Rojo<sup>1</sup>

<sup>1</sup> Orofacial Development and Regeneration, Centre for Dental Medicine, Institute of Oral Biology, University of Zurich, Zurich, Switzerland, <sup>2</sup> Department of Clinical and Experimental Medicine, Faculty of Health Sciences, Wallenberg Centre for Molecular Medicine, Linköping University, Linköping, Sweden

Keywords: signaling pathways, tooth, enamel, palate, calvaria, craniofacial, stem cells, tissue regeneration

### **Editorial on the Research Topic**

### **Signaling Pathways in Developing and Pathological Tissues and Organs of the Craniofacial Complex**

### Edited and reviewed by:

Petros Papagerakis, University of Saskatchewan, Canada

> \*Correspondence: Thimios A. Mitsiadis thimios.mitsiadis@zzm.uzh.ch

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

> Received: 06 June 2018 Accepted: 09 July 2018 Published: 26 July 2018

### Citation:

Mitsiadis TA, Cantù C and Jimenez-Rojo L (2018) Editorial: Signaling Pathways in Developing and Pathological Tissues and Organs of the Craniofacial Complex. Front. Physiol. 9:1015. doi: 10.3389/fphys.2018.01015 The field of craniofacial biology has greatly benefited from the progress made the last decades in developmental biology, molecular biology, stem cell biology, and genomics. A detailed description of the action of signaling pathways that are involved in developmental processes has significantly increased our comprehension of the organization and function of molecular networks that are involved in the formation of the craniofacial complex (Kouskoura et al., 2011). Craniofacial development requires the patterning, outgrowth, fusion, and molding of various tissues with a heterogeneous developmental origin. The important role of signaling molecules driving epithelialmesenchymal interactions in craniofacial development has been revealed by sophisticated genetic manipulations in mice (Cobourne and Mitsiadis, 2006).

The importance of the various signaling pathways in embryonic development was revealed more than two decades ago. Since then, the concept of how signaling molecules bind to their receptors and are transported from the membrane to the cytoplasm and/or nucleus has been corroborated with new exciting findings. The multifunctional nature of the signaling molecules provides craniofacial tissues and organs with versatile means of driving their development and controlling cell behavior. These molecules have numerous effects on cell proliferation, migration, differentiation, tissue morphogenesis, homeostasis and regeneration, and their deregulation and malfunction lead to severe craniofacial pathologies. The effects of the signaling molecules can be different, even opposite, depending on the cell type on which they act, and the environmental conditions. Elucidating the functionality of the various signaling pathways raised the specter of complicated signal transduction processes (Kouskoura et al., 2011). The biochemical crosstalk between molecules previously considered as being dedicated only to specific signaling pathways revealed new principles of signal-driven transcriptional action. Target genes of the various signaling molecules may trigger stem cell differentiation, cell cycle arrest, apoptosis, and even immune responses. It is important to elucidate how cells transit from one state to another (e.g., from stem cells to differentiated cells) under the governance of precise signaling molecules (Mitsiadis et al., 2007).

Recent advances have showed the importance of signaling molecules in craniofacial physiology and disease and moved the field closer to a more comprehensive understanding of the context-dependent nature of their action. Mutations in regulators and selected components of the signaling pathways were a first indication of their medical relevance (Kouskoura et al., 2011). Research on the ligand-receptor and adaptor protein-transcription factors interactions will generate important knowledge aimed at the development of new pharmaceutical products that mimic and block these interactions (Yong-Ming et al., 2017; Zimmerli et al., 2017; Aung et al., 2018). Finally, understanding how and when the various signaling pathways may activate cell proliferation, migration and apoptosis could provide impetus to the development of new therapeutic approaches after injury or cancer within the craniofacial complex (Pagella et al., 2017).

In this research topic, prominent researchers within the craniofacial field have contributed with important discoveries and generated exciting results concerning the role of signaling pathways in the pathophysiology of the various craniofacial tissues. Numerous original articles have evidenced the big variety of signals that are necessary for controlling developmental and pathological events of the craniofacial organs.

The influence of systemic glucocorticoid administration in the composition of bone structures of the jaw of mini pigs is reported (Schulz et al.). The ion channel protein TRPM7 mediates the mineralization process of craniofacial hard tissues (Nakano et al.). The influence of photobiomodulation on the capacity of bone marrow cells to form alveolar and craniofacial bones is also discussed (Amaroli et al.). Calvaria formation is a complex process that necessitates the coordinated action of osteogenic stem cell populations (Doro et al.), controlled by specific signaling molecules such as Indian Hedgehog and the transcription factor Gli3 (Veistinen et al.). Palatogenesis is an equally complex process orchestrated by cytoskeleton and extracellular matrix rearrangement (Chiquet et al.) and expression of specific signaling molecules (Iyyanar and Nazarali; Xavier et al.).

The identification of stem cell compartments in conjunction with the discovery of the complex ensemble of signals that creates their microenvironment is essential for a successful regenerative approach. Therefore, the identification and characterization of stem cell niches within dental pulp and the analysis of Notch signals during tooth repair is of prime importance (Mitsiadis et al.).

Loss of the alveolar bone that surrounds teeth could be an actor of periodontal pathology, and this process could be

# REFERENCES


controlled via the Stemodia maritime L. antioxidant extracts (Teixeira et al.). Similarly, Wnt signaling (Mitsiadis et al.), VEGF and Hypoxia Inducible Factor (Oishi et al.) and Syndecan-1 (Filatova et al.) might affect alveolar bone remodeling and periodontal ligament homeostasis.

Defined molecular mechanisms such as GLI-mediated transcription (Chang et al.), Fibroblast Growth Factor 10 (Teshima et al.), MORN5 (Cela et al.) and the microRNA Mir23b and Mir133b (Ding et al.) are involved in the physiopathology of a variety of tissues and organs of the craniofacial complex. The Wnt pathway controls the growth of dental lamina (Putnová et al.) from where the future teeth will develop. Retinoic acid (Morkmued et al.), steroids (Houari et al.), and Na:K:2Cl Cotransporter-1 (Jalali et al.) control the formation of enamel. Generation of different dental epithelial spheres that contain stem cells might enhance our understanding of enamel formation (Natsiou et al.). Dentin sialophosphoprotein (DSPP) is also very important for proper dentin formation and mutations in the DSPP gene lead to dentinogenesis imperfecta (Bloch-Zupan et al.). The role of Nerve Growth Factor pathway is similarly important during human tooth development and repair (Mitsiadis and Pagella). Finally, the Hippo-YAP1/TAZ cascade is important for pituitary stem cell development (Lodge et al.).

The increased knowledge on the development, pathology and regeneration of tissues and organs of the craniofacial complex will most certainly orchestrate a significant shift toward novel diagnostic and therapeutic approaches. Although many questions concerning the mechanisms involved in craniofacial tissues development and regeneration have not yet been resolved, modern imaging tools, mathematics, bioinformatics, and genomics will help to elaborate new concepts and models that will change drastically this field. Further progress in treatments concerning craniofacial tissues and organs depends upon active and vigorous research programs.

# AUTHOR CONTRIBUTIONS

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

# ACKNOWLEDGMENTS

This work was supported by institutional funds from University of Zurich. The authors contributed to the writing, reading, and editing of the present editorial.


Oncotarget 8, 99213–99214. doi: 10.18632/oncotarget. 22281


**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 © 2018 Mitsiadis, Cantù and Jimenez-Rojo. 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.

# Angled Growth of the Dental Lamina Is Accompanied by Asymmetrical Expression of the WNT Pathway Receptor Frizzled 6

Iveta Putnová1, 2 , Hana Dosedelová ˇ 1, 2, Vitezslav Bryja<sup>3</sup> , Marie Landová<sup>1</sup> , Marcela Buchtová1, 3 \* and Jan Štembírek 1, 4

 Laboratory of Molecular Morphogenesis, Institute of Animal Physiology and Genetics, Academy of Sciences, Brno, Czechia, Department of Anatomy, Histology and Embryology, University of Veterinary and Pharmaceutical Sciences, Brno, Czechia, Department of Animal Physiology and Immunology, Institute of Experimental Biology, Masaryk University, Brno, Czechia, Department of Maxillofacial Surgery, University Hospital Ostrava, Ostrava, Czechia

### Edited by:

Claudio Cantù, University of Zurich, Switzerland

### Reviewed by:

Jean-Christophe Farges, Claude Bernard University Lyon 1, France Michel Goldberg, French Institute of Health and Medical Research (Inserm), France Catherine Chaussain, Paris Descartes University, France

> \*Correspondence: Marcela Buchtova buchtova@iach.cz

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 09 October 2016 Accepted: 11 January 2017 Published: 31 January 2017

### Citation:

Putnová I, Dosedelová H, Bryja V, ˇ Landová M, Buchtová M and Štembírek J (2017) Angled Growth of the Dental Lamina Is Accompanied by Asymmetrical Expression of the WNT Pathway Receptor Frizzled 6. Front. Physiol. 8:29. doi: 10.3389/fphys.2017.00029 Frizzled 6 (FZD6) belongs to a family of proteins that serve as receptors in the WNT signaling pathway. FZD6 plays an important role in the establishment of planar cell polarity in many embryonic processes such as convergent extension during gastrulation, neural tube closure, or hair patterning. Based on its role during hair development, we hypothesized that FZD6 may have similar expression pattern and function in the dental lamina, which is also a distinct epithelial protrusion growing characteristically angled into the mesenchyme. Diphyodont minipig was selected as a model species because its dentition closely resemble human ones with successional generation of teeth initiated from the dental lamina. We revealed asymmetrical expression of FZD6 in the dental lamina of early as well as late stages during its regression with stronger expression located on the labial side of the dental lamina. During lamina regression, FZD6-positive cells were found in its superficial part and the signal coincided with the upregulation of molecules involved in epithelial-mesenchymal transition and increased migratory potential of epithelial cells. FZD6-expression was also turned on during differentiation of cells producing hard tissues, in which mature odontoblasts, ameloblasts, or surrounding osteoblasts were FZD6-positive. On the other hand, the tip of successional lamina and its lingual part, in which progenitor cells are located, exhibited FZD6-negativity. In conclusion, asymmetrical expression of FZD6 correlates with the growth directionality and side-specific morphological differences in the dental lamina of diphyodont species. Based on observed expression pattern, we propose that the dental lamina is other epithelial tissue, where planar cell polarity signaling is involved during its asymmetrical growth.

Keywords: FZD6, successional dental lamina, WNT signaling, planar cell polarity (PCP), odontoblast, ameloblast, osteoblast, epithelial remnants

# INTRODUCTION

Tooth development is a complex process that is dependent on reciprocal and strictly regulated interactions between the ectoderm-derived epithelium and cranial neural crest-derived mesenchyme (Thesleff et al., 1995). Numerous regulatory genes associated with all stages of tooth formation (patterning, morphogenesis, cytodifferentiation, and mineralization) belong to evolutionarily conserved signaling pathways. They are necessary for individual steps of odontogenesis and are regulated by a precise timing mechanism (Thesleff, 2003; Mitsiadis and Luder, 2011). The WNT signaling pathway has previously been demonstrated to play an important role in mouse as well as human tooth development (Sarkar and Sharpe, 1999; Sarkar et al., 2000; Handrigan and Richman, 2010). FZD members belongs to a family of proteins that serve as receptors in WNT signaling pathways (Fischer et al., 2007; Dijksterhuis et al., 2016; Wang et al., 2016). Their functions include activation of the FZD/β-catenin, FZD/Ca+<sup>2</sup> and FZD/planar cell polarity signaling pathways (Schulte and Bryja, 2007). To date, 10 FZD proteins have been described in mammals (Schulte and Bryja, 2007; Dijksterhuis et al., 2016). While the expression of several ligands of WNT signaling during odontogenesis has been welldescribed (Cai et al., 2011; Lin et al., 2011; Wang et al., 2014), little attention has been paid to the expression of individual receptors. Here, we focus on FZD6, which is involved in PCP (planar cell polarity) signaling and is required for the transmission of polarity signals across the plasma membrane in epidermal cells (Wang et al., 2006).

Cellular communication mediated by WNT and FZD has been shown to be essential for proper embryonic development of invertebrates as well as vertebrates. FZD induction has been found in several developmental processes such as the polarized cell movements required for convergent extension during gastrulation in frog and fish (Borello et al., 1999), neural induction and patterning, cell proliferation, cell specification, stem cell differentiation, axonal outgrowth and guidance, and synaptogenesis (Logan and Nusse, 2004; Chien et al., 2009; Wang et al., 2010). Besides these early patterning processes, FZD3 and FZD6 have been reported redundantly to control neural tube closure and planar orientation of hair bundles on a subset of auditory and vestibular sensory cells. In the inner ear, these two proteins are located on the lateral faces of sensory and auxiliary cells in a pattern that correlates with the axis of planar polarity (Wang et al., 2006). The polarity of FZD6 localization with respect to the asymmetric position of the kinocilium is reversed between vestibular hair cells in the cristae of the semicircular canals and auditory hair cells in the organ of Corti (Wang et al., 2006). FZD6 controls macroscopic hair patterning in the mouse and is also expressed in the skin and hair follicles (Guo et al., 2004; Wang et al., 2010; Chang et al., 2016).

Mutations in PCP genes lead to a wide range of developmental defects, including a shortened body axis, a widened neural plate, and neural tube defects (NTDs; Simons and Mlodzik, 2008). In the case of targeted deletion of the Fzd6 gene, stereotyped whorls on the hind feet, variable whorls and tufts on the head and disorientation of hairs on the torso are evident (Guo et al., 2004). In the Fzd6−/<sup>−</sup> mouse, the orientations of the earliest born hair follicles are uncorrelated, but over time the follicles reorient to create patterns, which are typical for different body regions and are characterized by a high degree of local arrangement (Wang et al., 2010). Fifty percent of male newborns, but not female Fzd6−/<sup>−</sup> mice, displayed abnormal claw morphology. The claws are easily lost with age or under increased mechanical stress. The claw disappears or become rudimentary on the hind limbs at the age of 2–3 months. The reason for the significant misbalance between sexes is unknown but it could be due to the more aggressive behavior of males (Fröjmark et al., 2011). Similarly in humans, loss-of-function mutations caused recessive nail dysplasia (Fröjmark et al., 2011; Naz et al., 2012). While several ectodermal derivates have been found to be affected in transgenic mouse lines or humans with defective FZD6, to date, no tooth phenotype has been described.

Here, we analyzed the expression of FZD6, a transmembrane protein of the WNT family, that is known to regulate the number of epithelial differentiation-related genes. During human odontogenesis, Fzd6 was shown to exhibit weak mRNA expression in the dental epithelium of incisors and molars at 8 and 12 weeks of gestation (Wang et al., 2014). Later during 15 week, Fzd6 was observed in the inner and outer enamel epithelium and in the surrounding mesenchymal cells (Wang et al., 2014). However, the distribution of FZD6 on protein level has not been analyzed yet. We focused on premolar development in diphyodont dentition during early as well as late mineralization stages of odontogenesis to determine the distribution of its expression throughout development. Labiolingual differences during the initiation and regression of dental lamina were analyzed to uncover signaling involved in asymmetrical morphology and growth of the lamina. Therefore, the main aim of our study was to describe the expression pattern of FZD6 at the protein level during early odontogenesis in the minipig dentition with a special focus on the asymmetric distribution of FZD6 in the dental lamina during its angled growth and regression. Furthermore, changes in FZD6-positivity in odontoblasts and ameloblasts during their differentiation were determined.

# MATERIALS AND METHODS

# Embryonic Material

Selected developmental stages of the minipig (E29, E30, E36, E56, E67) were used to analyse the expression of FZD6 during odontogenesis. Minipig embryos and fetuses were obtained from Libechov animal facility (Lib ˇ echov, Czech Republic). The day ˇ after insemination was established as day 1 of gestation. Staged embryos and fetuses were obtained by hysterectomy. All samples were fixed in 4% neutral formaldehyde and decalcified in 12.5% EDTA in 4% PFA until the mandibular bones of embryos were soft enough for further processing. Sections were stained with Haematoxylin-Eosin and alternative slides were used for immunohistochemical labeling. All procedures were conducted following a protocol approved by the Laboratory Animal Science Committee of the Institute of Animal Physiology and Genetics, Academy of Sciences (approval no. 020/2010, Libechov, Czech ˇ Republic).

# Immunohistochemical Analysis

For detection of FZD6-positivity, we performed immunohistochemical labeling. After deparaffinization and rehydration, antigen retrieval was performed in a water bath (97◦C) in citrate buffer (pH = 6) for 20 min. Blocking serum was applied to the sections for 20 min and slides were incubated for 1 h at room temperature with primary FZD6 antibody (cat. no. G260, Antibodies online, 1:200 dilution). The secondary antibody was applied for 30 min. Streptavidin-FITC complex (1:250 dilution, cat. no. 554060, BD Pharmigen, Franklin Lakes, USA) was used for visualization of FZD6-positive cells (30 min). DAPI (cat. no. P36935, Invitrogen, Oregon, USA) or DRAQ5 (1:500 dilution, cat. no. 62254, Thermo Scientific, USA) were applied for the counterstaining. The photos taken under a fluorescence microscope Leica DM LB2 (Leica Microsystems, Germany) were merged together in Adobe Photoshop 7.0 (USA). High power images were taken on confocal microscope Leica SP5 using 40x (air) objectives (Leica Microsystems, Germany) with Leica Application Suite software.

# RESULTS AND DISCUSSION

# Asymmetrical Expression of FZD6 at Early Stages of Odontogenesis

PCP components are often localized asymmetrically in different types of cells. The Frizzled family was found to be required for producing the correct orientation of cuticular bristles and hairs (Guo et al., 2004). This process is referred to as tissue or planar polarity (Gubb and García-Bellido, 1982; Vinson et al., 1989). Such polarization is often precisely coordinated relative to the axes of a tissue or organ, but the mechanisms underlying this regulation are still poorly understood. As exact coordination of tissue polarity in the labio-lingual axis is necessary to direct the angled growth of dental lamina into the mesenchyme, we wondered whether protein expression FZD6 can be associated with morphological side-related differences in the dental lamina. Indeed, we observed significant differences in the level of FZD6-positivity in distinct areas of the dental lamina at all stages of minipig odontogenesis. Already at the epithelial thickening stage, a strong FZD6 signal was apparent on the labial side of the oral epithelium and in the dental epithelium, especially in its basal layer (**Figure 1A**). The signal became weaker toward its lingual side. Distinct expression was also obvious in the mesenchyme surrounding the epithelial thickening, where more FZD6-positivity was noticeable in the labial area (**Figure 1A**). Later, during dental lamina growth into the mesenchyme, stronger expression was observed also on the labial side of the oral dental interface in comparison with the lingual area (**Figures 1B,C,C**′ ). At the dental bud stage, FZD6 was only weakly expressed inside the tooth anlagen (**Figures 1D,D**′ ) while FZD6-immunopositivity was detected in the dental lamina connecting the tooth to the oral epithelium, where stronger expression was located on the labial side (**Figures 1D,D**′ ).

At the cap stage (**Figure 1E**), distinct expression was found throughout all thicknesses of the oral epithelium and in the dental lamina, but was not apparent inside the tooth anlagen. The signal spread into the cervical loop area during the early bell stage (**Figure 1F**). Almost no signal was visible in the stellate reticulum (**Figure 1F**). The dental papilla and surrounding mesenchyme were FZD6-negative. FZD6-signal was abundant in the basal layer of the oral epithelium at all analyzed stages and became localized to the superficial area of membrane in more differentiated superficial layers of the epithelium (**Figure 1**).

It is known that FZD6 is transducing signals in non-canonical WNT pathway and Wnt5a is one of its representative, which signals upstream of PCP pathway in mammals (Moon et al., 1993; Kilian et al., 2003). It was shown that WNT5a regulates tooth growth, cusp patterning and odontoblast differentiation in developing mouse molars and incisors (Lin et al., 2011). Previously, WNT5a expression was found not only in the dental mesenchyme but also in the dental epithelium in mouse during E14–E17 (Cai et al., 2011). However, we detected FZD6 even at very early stages of epithelial thickening and in the surrounding mesenchyme, and therefore another WNT ligand probably activates signaling at these early stages, which will be necessary to further analyse in future.

FZD can also mediate canonical signaling with activation of β-catenin. WNT3a is strongly expressed in the inner enamel epithelium of humans during the bell stage (Wang et al., 2014). In mice, Wnt3a is expressed in the enamel knot at the cap stage (Millar et al., 2003). These expression patterns do not seem to correlate with our observations of a strong signal located in the superficial part of the tooth anlagen whereas deeper parts including enamel knots and cervical loops were FZD6-negative (**Figure 1**).

FZD6 was also distributed in distinct areas of the mesenchyme surrounding epithelial thickenings during the early stages of odontogenesis; however, mesenchymal expression was downregulated and only small spots were visible on the mesenchymal cell surface at older stages. On the other hand, FZD6 was mostly expressed in the epithelium with uniform expression throughout the whole thickness of the oral epithelium. This finding is consistent with observations in FZD6 knockout mice, where numerous genes encoding keratins, keratinassociated proteins and transglutaminases and their substrates were significantly downregulated (Cui et al., 2013). Therefore, the mesenchymal expression pattern is dissimilar to the epithelial signal indicating distinct role of FZD signaling in the epithelium in contrast to the mesenchyme.

# FZD6 Expression in the Successional Dental Lamina

During later stages of odontogenesis, the dental lamina protruded deeply into the mesenchyme and morphological changes were obvious in its superficial part (**Figure 2**). In the most superficial area, cells connecting the lamina to the oral epithelium were FZD6-positive (**Figures 2A,B**, **3A,A**′ **,B,B**′ ). Expression in superficial cells even increased following disconnection of the lamina from the oral epithelium (**Figures 2C**, **3D,D**′ **,E,E**′ ). Cells on the labial side separating from the lamina were strongly FZD6-positive similar to the stalk of very flat cells connecting the dental lamina to the tooth (**Figures 2A–C**). On the other hand, the apical tip of the successional lamina was FZD6-negative (**Figures 2A–D**, **3C,C**′ **,F,F**′ ). Therefore, there are significant differences in FZD6 expression through the dental lamina with higher expression maintained in its superficial layers.

While, the most distal tip of the successional dental lamina was FZD6-negative, β-catenin was previously described in the tip and lingual side of the dental lamina in snake and alligator (Wu et al., 2013) and transcription factor Lef1 (Wnt/β-catenin pathway target gene) in corn snake and python dental lamina (Handrigan

FIGURE 1 | FZD6 expression is asymmetrical in the dental lamina. (A) At epithelial thickening stage, the oral epithelium and surrounding mesenchyme are FZD6-positive. (B,C,C′ in detail) Later, FZD6-positive signal is located on the labial side (arrow) of the protruding dental lamina. (D,D′ in detail) At dental bud stage, stronger FZD6-immunopositivity (arrow) was detected on the labial side of dental germ. (E,F) At bell stage, the labial cervical loop is more positive that the lingual one. Almost no signal was visible in the stellate reticulum. The dental papilla and surrounding mesenchyme were FZD6-negative. b, bone; de, dental epithelium; dl, dental lamina; dp, dental papilla; oe, oral epithelium; th, tooth. Scale bar = 100 µm.

and Richman, 2010; Gaete and Tucker, 2013). WNT ligands that signal through β-catenin are involved in stem/progenitor self-renewal and maintenance of cells in a proliferative and undifferentiated state while non-canonical signaling promotes their differentiation (Liu et al., 2009; Grumolato et al., 2010). As non-canonical signaling is known to inhibit WNT/β-catenin canonical signaling (Topol et al., 2003; Mikels and Nusse, 2006), this asymmetrical expression of non-canonical and canonical WNT molecules and the balance among them is critical for the regulation of dental progenitor cell lines similar to that shown in other systems (Grigoryan et al., 2008). Based on our evidence, FZD6 is not expressed in the areas of dental lamina with high cell proliferation or proposed localization of progenitor cells. Therefore, canonical and non-canonical WNT signaling exhibits distinct asymmetrical expression pattern through the lamina, which seems to be aligned with side-specific differences during successional dental lamina formation.

# FZD6 Was Strongly Expressed in the Epithelial Remnants and Pearls during Dental Lamina Regression

The minipig similar to human has a diphyodont type of dentition, where only two generation of teeth are initiated. During embryonic period, the dental lamina undergoes major morphological changes and becomes thinner and disconnected from the oral epithelium. Cells are elongated in the superficial area of the dental lamina (Buchtová et al., 2012). Furthermore, they are elongated in the area of the dental stalk connecting the tooth anlagen to the dental lamina. After the dental lamina had disconnected from the oral epithelium, stronger expression of FZD6 was evident in the superficial epithelial cells. In deeper parts of the lamina, the expression of FZD6 was apparent on the side facing the tooth, which undergoes regression (**Figure 2C**). There were also differences in the level of expression along the jaw axis. In the area between teeth, rudimental interdental lamina exhibited only weak FZD6 expression located in the most superficial area (**Figure 2D**).

Later during dental lamina regression, the dental lamina become fragmented, and only occasional epithelial islands remain in the superficial area (**Figure 4A**). Some of these fragments undergo further morphological changes and epithelial pearls become visible along the jaw (Buchtová et al., 2012). Expression was also found in the epithelial clusters during the process of pearl formation (**Figure 4C**) as well as in the basal layer of already formed pearls (**Figure 4B**), while the central area was negative (**Figure 4D**).

Interestingly, distinct expression was obvious especially in the basal layers of the lamina epithelium facing the tooth where clusters of elongated cells were moving out of the dental lamina. The expression of FZD6 during lamina regression coincides with the upregulation of molecules involved in epithelialmesenchymal transition of epithelial cells (Buchtová et al., 2012). Our observations that cells moving from the dental lamina are FZD6-positive is consistent with the previously known role of PCP proteins in notochord or somite cell elongation during their convergent extension (Keller et al., 2000; Keller, 2002; Seifert and Mlodzik, 2007) as well as regulation of cell polarity and directed motility during gastrulation in frogs and fish, and neural tube and eyelid closure in mammals (Wang et al., 2006; Seifert and Mlodzik, 2007).

# Frizzled 6 in Differentiation of Hard-Tissue Producing Cells

Both canonical and non-canonical WNT pathways were previously shown to be involved in the differentiation of hardtissue producing cells (Millar et al., 2003; Lin et al., 2011; Sakisaka et al., 2015). WNT signaling was shown to promote the differentiation of dental follicle cells into the cementoblast or osteoblast phenotype (Peng et al., 2010; Du et al., 2012; Sakisaka et al., 2015; Nemoto et al., 2016). Moreover, Wnt5a overexpression promotes the differentiation of dental papilla cells and increased the expression of mineralization-related genes (Peng et al., 2010).

In agreement with these findings, FZD6 expression was switched on during the differentiation of odontoblasts, ameloblasts, and osteoblasts in minipigs (**Figure 5**) at time when the production of hard tissues started. Differentiated and secretory ameloblasts were FZD6-positive while nondifferentiated cells of the inner enamel epithelium were negative (**Figures 5A,A**′ ). A similar pattern was observed in the dental papilla where differentiated odontoblasts exhibiting dentin production were FZD6-positive (**Figures 5A,A**′ ). Almost no signal was found in the stellate reticulum or stratum intermedium (**Figure 5A**). On the other hand, the outer enamel epithelium was positive, especially the superficial clusters of cells during disruption of the enamel organ by blood vessels. FZD6 signal was also located in the structures surrounding teeth in the jaw such as alveolar bones, in which osteoblasts were positive (**Figures 5B,B**′ ) along with the secretory area of salivary glands or their ducts (**Figures 5C,C**′ ).

While the tooth phenotype of FZD6 mutant mice has not yet been described, Wnt5a-deficient mice exhibit delayed odontoblast differentiation (Lin et al., 2011). Abnormal morphology of ameloblasts and defective odontoblast differentiation with absence of predentin formation were

also found in Ror2 mutant mice (Lin et al., 2011) in which ROR2 can serve as an alternative WNT receptor (Oishi et al.,

dental papilla; od, odontoblast; oee, outer enamel epithelium; sg, salivary gland. Scale bar = 100 µm.

2003). Odontoblasts in Wnt5a-deficient mice were shorter and thicker than in control animals. Similarly in Ror2-deficient mice, odontoblasts were polarized but appeared to be shorter in comparison to littermate control animals (Lin et al., 2011). While Wnt5a is expressed in the dental mesenchyme, its receptor is also expressed in the epithelium. It was proposed that another receptor must be involved in WNT5a-mediated signaling as

The tip of successional dental lamina was FZD6-negative while cells in the superficial part of the lamina were FZD6-positive. Nuclei are labeled by blue and FZD6 expression in green through the epithelial tissues. de, dental epithelium; dl, dental lamina; oe, oral epithelium; th, tooth.

defects in tooth development in Ror2 mutants occur much later than observed in WNT5a-deficient mice (Lin et al., 2011). Based on our observations, FZD6 could be one of these receptors.

Odontoblasts and ameloblasts are highly polarized cells with a characteristic morphology and arrangement of cellular compartments. In odontoblasts, FZD6 was expressed asymmetrically only on the side facing the dentin (**Figures 5A,A**′ ). In ameloblasts, FZD6 was observed on both sides of differentiated cells, and later, the signal showed a more uniform distribution through the cells. Differentiation of ameloblasts and odontoblasts is characterized by elongation of the cells and establishment of their polarity. Actin filament bundles exhibit a polarized distribution in rat ameloblasts and are abundant at the ameloblast junction area (Nishikawa and Kitamura, 1985, 1986). As actin cytoskeleton organization is downstream of the PCP signaling pathway, it is possible that FZD6 can be involved in the rearrangement of cellular polarity in the odontoblasts and ameloblasts during their differentiation.

# CONCLUSIONS

In summary, FZD6 was expressed asymmetrically in the dental lamina at early as well as late stages of diphyodont dentition (**Figure 6**). We observed significant differences in the level of FZD6 expression in distinct areas of the dental lamina with stronger expression in the labial and superficial parts (**Figure 6**). The apical tip of the successional lamina was FZD6-negative. During dental lamina regression, labial cells separating from the lamina were strongly FZD6-positive similar to the stalk of very flat cells connecting the dental lamina to the tooth germ. On the other hand, FZD6 was not expressed in the areas of dental lamina with high cell proliferation or proposed localization of progenitor cells.

FZD6 expression was also switched on during the differentiation of odontoblasts, ameloblasts, and osteoblasts in minipigs at time when the production of hard tissues started. In odontoblasts, FZD6 was expressed asymmetrically only on the side facing the dentin therefore it is possible that FZD6 can be involved in the establishment of cellular polarity in the odontoblasts during their differentiation. However, the exact role of FZD6 in the growth directionality of the lamina and differentiation of hard tissue producing cells has to be proven experimentally in future.

# AUTHOR CONTRIBUTIONS

MB and VB designed the project. IP, HD, ML, and JS performed and interpreted experiments. IP, MB, and JS wrote

# REFERENCES


manuscript with contribution of VB. All authors approved the manuscript.

# ACKNOWLEDGMENTS

This study was supported by the Czech Science Foundation (14-29273P to IP and JS) and Grant Agency of the University of Veterinary and Pharmaceutical Sciences Brno (108/2015/FVL to HD), Masaryk University (MUNI/A/0988/2016) and institutional support (RVO:67985904). This contribution is free of conflict of interest.


**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 © 2017 Putnová, Dosedelová, Bryja, Landová, Buchtová and Štembírek. ˇ 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) or licensor 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.

# Early Determination of the Periodontal Domain by the Wnt-Antagonist Frzb/Sfrp3

### Thimios A. Mitsiadis <sup>1</sup> \*, Pierfrancesco Pagella<sup>1</sup> and Claudio Cantù<sup>2</sup>

<sup>1</sup> Orofacial Development and Regeneration, Institute of Oral Biology, Centre for Dental Medicine, Medical Faculty, University of Zurich, Zurich, Switzerland, <sup>2</sup> Institute of Molecular Life Sciences, University of Zurich, Zurich, Switzerland

Odontogenesis results from the continuous and reciprocal interaction between cells of the oral epithelium and cranial neural crest-derived mesenchyme. The canonical Wnt signaling pathway plays a fundamental role in mediating these interactions from the earliest stages of tooth development. Here we analyze by in situ hybridization the expression patterns of the extracellular Wnt antagonist Frzb/Sfrp3. Although Frzb is expressed in dental mesenchymal cells from the earliest stages of odontogenesis, its expression is absent from a tiny population of mesenchymal cells immediately adjacent to the invaginating dental epithelium. Cell proliferation studies using BrdU showed that the Frzb expressing and Frzb non-expressing cell populations display different proliferative behavior during the initial stages of odontogenesis. DiI-mediated cell-fate tracing studies demonstrated that the Frzb expressing cells contribute to the formation of the dental follicle, the future periodontium. In contrast, the Frzb non-expressing cells give rise to the dental pulp. The present results indicate that Frzb is discriminating the presumptive periodontal territory from the rest of the dental mesenchyme from the very beginning of odontogenesis, where it might act as a barrier for the diffusion of Wnt molecules, thus regulating the activation of Wnt-dependent transcription within dental tissues.

### Edited by:

Agnes Bloch-Zupan, Université de Strasbourg, France

### Reviewed by:

Zhi Chen, Wuhan University, China Brad A. Amendt, University of Iowa, United States Ophir D Klein, University of California, San Francisco, United States

### \*Correspondence:

Thimios A. Mitsiadis thimios.mitsiadis@zzm.uzh.ch

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 02 August 2017 Accepted: 06 November 2017 Published: 21 November 2017

### Citation:

Mitsiadis TA, Pagella P and Cantù C (2017) Early Determination of the Periodontal Domain by the Wnt-Antagonist Frzb/Sfrp3. Front. Physiol. 8:936. doi: 10.3389/fphys.2017.00936 Keywords: Frzb/Srfp3, Wnt signaling, tooth development, cell fate, dental pulp, dental follicle, periodontium

# INTRODUCTION

Odontogenesis is characterized by the sequential and reciprocal interactions between cells of the oral epithelium and the cranial neural crest-derived mesenchyme and proceeds through a series of well-defined morphological steps, namely bud, cap, bell, and cytodifferentiation/mineralization stages (Mitsiadis and Graf, 2009). Ectomesenchymal cells form two distinct and specialized tooth components, the dental papilla that gives rise to the pulp and the dentin-producing odontoblasts, and the dental follicle, which surrounds the developing dental epithelium and forms the periodontium (Mitsiadis and Graf, 2009; Krivanek et al., 2017).

All stages of tooth development are mediated by the exchange of a big variety of signaling molecules between homotypic and heterotypic cell populations (Mitsiadis and Graf, 2009; Mitsiadis and Luder, 2011; Jussila and Thesleff, 2012; Balic and Thesleff, 2015). Among these molecules, the secreted lipid-modified Wnt glycoproteins trigger the evolutionarily conserved Wnt signaling pathway, a molecular cascade important for the development of virtually all organs (Clevers, 2006). One previous study has shown that several extracellular Wnt ligands and Wnt inhibitors are expressed in specific stages and compartments during odontogenesis (Sarkar and Sharpe, 1999). Although many reports addressed the role of the canonical Wnt/β-catenin-mediated signaling in the formation and regeneration of dental tissues (Aurrekoetxea et al., 2012, 2016; Liu et al., 2014; Zhang et al., 2016; Babb et al., 2017) a unifying picture of its activity during odontogenesis is still missing (Tamura and Nemoto, 2016). The critical requirement of Wnt signaling during tooth development has been already evidenced, since the genetic loss of β-catenin, or the specific abrogation of β-catenin-dependent transcription, leads to arrested tooth formation at early stages (Liu et al., 2008; Cantù et al., 2017).

The action of the Wnt ligands in the extracellular matrix is regulated by physiologically secreted Wnt antagonists (or negative regulators) such as the ones belonging to the family of Secreted Frizzled Related Proteins (SFRPs), which possess a cysteine-rich domain homologous to the Wnt-binding domain of Frizzled (Frzb) receptors (Cruciat and Niehrs, 2013) and are implicated as tumor suppressors in several forms of cancer (Zimmerli et al., 2017).

Frzb (known as Frzb1 or Sfrp3, Secreted Frizzled Related Protein 3) was initially identified as a chondrogenic factor during bone morphogenesis (Hoang et al., 1996). It was subsequently shown to modulate the activity of XWnt8 during Xenopus dorsoventral axis development (Leyns et al., 1997) and to repress canonical Wnt signaling in other contexts (Person et al., 2005). Here we identify Frzb as a novel marker of the neural crest-derived mesenchymal cells that contribute to dental follicle formation, the future periodontium. Frzb expression at the earliest stages of odontogenesis allows distinguishing two dental mesenchymal cell populations with clearly defined developmental fates.

# MATERIALS AND METHODS

# Cell Proliferation Analysis

All animals were maintained and handled according to the Swiss Animal Welfare Law and in compliance with the regulations of the Cantonal Veterinary Office, Zurich (License 11/2014). In vivo cell proliferation in dental tissues was analyzed by immunohistochemistry for phosphorylated Histone 3 (pH3; rabbit Ab, 1:200; Upstate, Charlottesville, VA) and bromodeoxyuridine (BrdU). For the latter, a BrdU cell proliferation kit (Boehringer Mannheim, Germany) was used. Foster mothers were injected intraperitoneally with 5 mg/ml of BrdU in PBS at a concentration of 50 mg/kg body-weight, 60 min before embryos were sacrificed. BrdUpositive cells in developing teeth of E13–E15 embryos were analyzed on 14µm cryosections after staining with an anti-BrdU antibody. Immunohistochemistry was performed as described earlier (Mitsiadis et al., 2008). Cells were counted with the CellCounter Plugin, ImageJ. Statistical Analysis was performed with GraphPad Prism 7 (t-test).

# Lineage Tracing Using DiI Labeling

DiI (1, 10, di-octadecyl-3, 3, 30,-tetramethylindo-carbocyanine perchlorate; Molecular probes cell tracker CM-DiI, C-7000) injection was performed to various locations of mesenchymal cells surrounding the dental epithelial ingrowths of cultured mandible slices to monitor cell kinetics. Briefly, E13 mouse mandibles were carefully dissected out from the head, placed upon a chopping plate of a McIlwain tissue chopper (Mickle Laboratory Engineering Co., Ltd., Guilford, UK), orientated to obtain frontal sections and finally cut into 250µm thick slices. Slices containing the molar tooth buds were selected and cultured. DiI, which is highly lipophilic dye intercalating into the cell membranes, was dissolved in ethanol (EtOH) at 2.5 µg/µl (stock solution) and then diluted 1–9 in 0.2 M sucrose. Thereafter, small amounts of DiI were injected using a mouthcontrolled micropipette made from a 50 mm borosilicate glass into different areas of the condensing dental mesenchyme, either in mesenchymal cells contacting the molar bud epithelium or in condensing cells located at a more distant area from the tooth epithelium (Mitsiadis et al., 2008; Gruenbaum-Cohen et al., 2009).

# Slice Cultures and Imaging

DiI labeled slices were placed upon Millipore filters coated with growth factor reduced Matrigel basement membrane matrix (BD Biosciences). The slices were completely encapsulated by an additional layer of Matrigel that served to structurally support the morphology of the explants during their development. The filters were supported above the culture medium by metal grids within Petri dishes. The culture medium was composed of Dulbecco's Minimum Essential Medium (DMEM) supplemented with 1% penicillin/streptomycin, 2 mM L-glutamine and 15% fetal calf serum (FCS). Slices were cultured up to 4 days in a 37◦C/5% CO2 air-jacketed incubator. After culturing, samples were fixed in 4% paraformaldehyde (PFA) for 30 min, washed with PBS and then embedded in wax and sectioned at 8µm.

The initial positions of the DiI injection and the subsequent location of the DiI-labeled cells were monitored throughout the culture period using a Leica dissecting microscope equipped with UV light (Leica Microsystems Ltd., Germany).

# In Situ Hybridization

Frzb in situ hybridization probe was kindly provided by Prof. De Robertis (Leyns et al., 1997). The labeled probe was ethanol-precipitated, resuspended in 100 mM DTT, diluted in hybridization solution (60% deionized formamide, 20 mM Tris-HCl, 5 mM EDTA, pH 8, 0.3 M NaCl, 0.5 mg/ml yeast RNA, 5% dextran sulfate). In situ hybridization was performed according to standard procedures (Mitsiadis et al., 1995). Briefly, slides were incubated with the probe at 60◦C. After intense washing, the slides were incubated in blocking solution (20% Normal Goat Serum) and anti-digoxigenin (DIG)-AP (alkaline phosphatase conjugate) Fab-fragment (Boehringer Mannheim, 1093 274) diluted 1:1,000 in blocking solution. The color reaction was developed using Nitro Blue Tetrazolium (NBT, Sigma N-6876) and 5-Bromo-4-Chloro-3-Indolyl Phosphate (BCIP, Sigma B-8503) in staining solution 2% NaCl, 5% MgCl2, 10% Tris-HCl pH 9.5, 1% Tween-20. In situ hybridization immediately followed by BrdU immunohistochemistry was performed in cryosectioned slides of E13–E15 mouse embryos to show the correlation between Frzb expression and cell proliferation (Mitsiadis et al., 2008). No hybridization signal was detected with the sense probe at these developmental stages.

# RESULTS

# Frzb Is Expressed in a Subpopulation of Dental Mesenchymal Cells

To determine the potential role of Frzb in odontogenesis, we analyzed its expression pattern during the early stages of mouse tooth development (**Figure 1A**). We monitored the expression of Frzb in the developing mouse tooth germs from embryonic day 11 (E11; initiation stage) to E15 (cap stage). Intense hybridization signal was observed in the mesenchyme of the mandible during the tooth initiation period (E11) (**Figure 1B**). During the dental epithelial invagination to the underlying mesenchyme (early bud stage, E12), Frzb mRNA was restricted in mesenchymal cells located at the areas of molar (**Figures 1C,D**) and incisor (**Figure 1E**) formation. At this stage, the hybridization signal was strikingly absent from a layer of mesenchymal cells nearby the epithelium (**Figures 1C–E**, red asterisk). However, Frzb was strongly expressed in mesenchymal cells that are not in close contact with the dental epithelium (**Figures 1C–E**). This observation was confirmed by transcript localization at E13 (late bud stage) (**Figures 1F,G**). At the cap stage (E14–E15), Frzb hybridization signal was absent from the cells composing the dental papilla, while Frzb expression was strong in the peripheral regions of the developing tooth germ (**Figures 1H,I**).

# Differential Proliferation of Presumptive Dental Papilla and Follicle Cells

We then wondered whether this distinct Frzb expression pattern in the developing dental mesenchyme could correlate with dissimilar proliferative behavior between these two cell populations (Frzb expressing and Frzb non-expressing cells). To test this hypothesis, pregnant females were injected with BrdU and tooth germs of E13–E15 embryos were analyzed for cell proliferation. At E13 (bud stage), two territories could be observed according to BrdU immunoreactivity. Cell proliferation was significantly lower in the mesenchyme nearby the epithelium (**Figure 2A**, red asterisk), which does not express Frzb (Frzb−, **Figure 2G**), when compared to the mesenchyme that is most distant from the epithelium (**Figure 2A**; green arrowheads) and expresses Frzb (Frzb+, **Figures 2G,J**). At the subsequent cap stage (E14–E15), the proliferation status of the two cell populations switched: abundant mitotic activity was observed in mesenchymal cells forming the dental papilla (**Figures 2B,C**; green arrowheads) outside the Frzb expression domain (Frzb−, **Figures 2H,I**). In contrast, proliferative activity was sporadically detected in the forming dental follicle (**Figures 2B,C**; green arrowheads) that expresses Frzb (Frzb+, **Figures 2H–J**). To confirm the differential proliferation status between cells of the dental follicle (Frzb<sup>+</sup> cells) and dental papilla (Frzb<sup>−</sup> cells), we stained E13–E15 tooth germs for phospho-Histone H3 (pH3), which marks cells in active mitosis (M phase) (**Figures 2D–F**). pH3 immunohistochemistry showed that Frzb<sup>+</sup> cells of the dental follicle proliferate significantly more than Frzb<sup>−</sup> cells of the presumptive dental papilla at E13 (**Figures 2D,K**). A proliferative switch occurred at E14–E15, when Frzb<sup>+</sup> cells of the follicle display a significantly lower mitotic activity than Frzb<sup>−</sup> cells of the papilla (**Figures 2F,I,K**).

# Frzb Expressing Cells Mark the Presumptive Dental Follicle

To test if the Frzb expressing cells selectively contribute to the formation of dental follicle, we labeled subsets of the Frzb expressing (**Figure 3A**) and non-expressing (**Figure 3E**) mesenchymal domains with DiI at the bud stage and followed the fate of DiI-positive cells (**Figures 3B–D,F–H**). DiI-positive cells that were located in the territory of Frzb expression participated in the formation of the dental follicle of the cap (2 days of culture) and early bell (4 days of culture) staged teeth (**Figures 3B–D**). As DiI does not allow simultaneous labeling of the entire Frzb expression domain, we marked different sub-regions in a significant number of experiments (n > 10). In all experiments realized, DiI staining was never observed neither in the dental papilla compartment of these developing teeth, nor in other tissues formed far away from the tooth germ (**Figures 3C,D,I,J**). Conversely, DiI labeled cells in direct contact with the dental epithelium (Frzb<sup>−</sup> cells) contributed to dental papilla formation at the cap (2 days of culture) and early bell (4 days of culture) stages (n > 10; **Figures 3F–H**). Sections of the tooth germs of the early bell stage indicated dental papilla cells stained with DiI (**Figures 3G,H,K,L**). Since DiI injection was performed in both tooth mesenchymal cell populations in this experiment, DiI labeled cells were present in both dental follicle and dental papilla cells. Due to the intrinsic limitations of this technique we rarely obtained DiI labeling in dental epithelial cells when injection was performed in mesenchymal cells immediately adjacent to dental epithelial bud (**Figures 3E,F,I**).

# DISCUSSION

Cranial neural crest-derived mesenchymal cells play a crucial role in tooth formation and are recognizable since the earliest stages of odontogenesis due to their positivity to classical neural crest or dental mesenchyme-specific markers such as Pax9 (Bonczek et al., 2017), Barx1 (Mitsiadis et al., 1998), and midkine (MK) (Mitsiadis et al., 1995). These cells have been historically considered as a homogeneous population of cells; nevertheless, the dental mesenchyme generates highly specialized adult soft tissues, such as the pulp and the periodontium (Jiménez-Rojo et al., 2014; Otsu et al., 2014; Mitsiadis et al., 2015). It is therefore of great interest the discovery of new markers that would allow distinguishing the subsets of mesenchymal cells possessing different differentiation commitments. Here we identify Frzb as a novel dental mesenchymal marker. Importantly, Frzb is specifically expressed from the earliest tooth developmental stages in mesenchymal cells that are not in direct contact with the dental epithelium. This specific expression pattern only partially overlaps with the expression domain of other well-established dental mesenchymal markers, such as Pax9 (**Figure 4A**). One previous report described the expression of

FIGURE 1 | Expression of Frzb transcript in developing tooth germs by digoxigenin in situ hybridization on cryosections (A) Schematic representation of the early developmental tooth stages (E11–E15). (B) At E11, Frzb is expressed in the mandibular and maxillary mesenchyme, while it is excluded from the tooth germ. (C–G) At the bud stage (E12–E13), Frzb is expressed in the dental mesenchyme, but excluded from the 1–3 cell layers immediately adjacent to the dental epithelium (red asterisks), both in molars (C,D,G) and incisors (E,F). (H,I) At the cap stage (E14–E15), Frzb expression consistently marks the dental follicle, while it is absent from the dental papilla. ab, alveolar bone; ae, aboral epithelium; de, dental epithelium; df, dental follicle; dp, dental papilla; eo, enamel organ; m, mesenchyme; md, mandible; mx, maxilla; oe, oral epithelium; n, nose. Scale bars: 100µm.

Frzb in the dental mesenchyme (Sarkar and Sharpe, 1999). However, the radioactive in situ hybridization technique that was used for the detection of the transcripts did not offer sufficient resolution to distinguish the absence of Frzb mRNA expression in the cell layers immediately adjacent to the dental epithelium. Based on Frzb expression we could identify two functionally distinct cell domains: one expressing Frzb that will give rise to the dental follicle, and another non-expressing Frzb that will form the dental papilla. This important finding shows for the first time the presence of two functionally distinct cell populations in the tooth mesenchyme and their commitment for generating specialized dental structures of the tooth at very early developmental stages (**Figure 4B**). During early odontogenesis (E11–E12), the thickened dental epithelium is surrounded by mesenchymal cells that condense and actively proliferate (Mitsiadis et al., 2003; Mitsiadis and Graf, 2009). These combined activities of cell migration, adhesion and proliferation within the mesenchyme are controlled by signaling and cell adhesion molecules such as fibroblast growth factors, midkine, syndecan. Dental mesenchyme is important for the tooth shape determination that involves alternated and well-orchestrated proliferations of distinct cell populations throughout odontogenesis. Initially, active proliferative events in the territory that will give rise to dental follicle/periodontium correlate with Frzb expression. This initial process of active cell proliferation within the Frzb expression domain determines the size of the periodontal domain that is important for proper tooth germ development and its integration in the growing

FIGURE 2 | Differential proliferative behavior of dental mesenchymal populations. (A) BrdU immunostaining showing intense proliferation in the presumptive dental follicle region (green arrowheads), and little proliferation in the presumptive dental papilla (red asterisk). (B,C) At E14–E15, cell proliferation is concentrated in the dental papilla, while it is very limited in the dental follicle (green arrowheads). (D) At E13, phosphorylated Histone H3 (pH3) immunostaining indicates cell mitosis in the presumptive dental follicle region (green arrowheads), and little mitotic activity in the presumptive dental papilla (red asterisk). (E,F) At E14 and E15, pH3 immunolabelling shows intense mitotic activity in dental papilla and moderate in dental follicle. (G–I) Combined BrdU immunostaining (red) and Frzb in situ hybridization (blue) shows the correlation between cell proliferation and Frzb expression within the dental follicle and papilla at E13 (G), E14 (H) and E15 (I). (J) Quantification of BrdU<sup>+</sup> cells in dental follicle and dental papilla at E13, E14, and E15. n = 5 vs. 5. (K) Quantification of pH3<sup>+</sup> cells in the dental follicle and dental papilla at E13, E14, and E15 (n = 5 vs. 5). de, dental epithelium; df, dental follicle; dm, dental mesenchyme; dp, dental papilla; eo, enamel organ; m, mesenchyme; md, mandible; oe, oral epithelium. Scale bars: 100µm. \*\*p < 0.001; \*\*\*p < 0.0001.

4. When DiI was injected at a distance from the dental epithelium, DiI+-cells (bright yellow color) were observed only in the dental follicle (I,J). Worth note that some epithelial cells next to the dental follicle were also injected with DiI during the procedure. In contrast, when both distant and adjacent mesenchymal cells were injected with DiI, fluorescence was observed in the dental pulp and dental follicle (K,L). de, dental epithelium; df, dental follicle; eo, enamel organ; m, mesenchyme; md, mandible; oe, oral epithelium; p, pulp. Scale bars: (B–D/F–H) 200µm; (I–L): 100µm.

surrounding environment (e.g., alveolar bone). Subsequently, once Frzb expressing cells have delimited the dental papilla domain, cells from this tiny Frzb-negative territory start to actively proliferate in order to increase the tissue size. This process is accompanied by morphological rearrangements within the growing epithelium that dictate the tooth shape.

The differential proliferative activities between dental pulp cells and dental follicle cells are not only limited to development. In our previous studies, we have shown that when pulp and follicle cells are co-cultured, they never intermingle and they assume an organization reminiscent of the in vivo situation, with follicle cells surrounding and engulfing pulp cells (Schiraldi et al., 2012). These cell populations appear to have a clear genetic memory and compete for their own territory even in vitro. Our results indicate that these two cell populations are functionally and molecularly discriminated from the earliest stages of odontogenesis, thus providing a developmental basis to their clear and persistent peculiarity. Future studies are required to understand the molecular mechanisms underlying the early specification and fate of these cell populations. In this regard, Frzb-based genetic lineage tracing fluorescent analysis on transgenic mouse models might be instrumental to understand the exact contribution of Frzb-expressing cells to tooth development as well as to the formation and the regeneration of the periodontium. Since Frzb is an extracellular Wnt antagonist, it is tempting to speculate that the restricted expression of Frzb to a subset of mesenchymal cells could generate an asymmetric barrier whose function is to limit the activation of the Wnt cascade only in certain cells. It is plausible, in fact, that distinct differentiation fates require cells to be Wnt-responsive, while others necessitate the inhibition of this pathway, as it occurs for the early development of the mouse heart (Marvin et al., 2001) and the eye lens (Cantù et al., 2014; Cvekl and Ashery-Padan, 2014). In this scenario, Frzb might restrict the activity of the secreted Wnt ligands to the oral epithelium and to the presumptive dental papilla. Consistently, many genes encoding for Wnt ligands are specifically expressed from the oral epithelium throughout tooth development (Sarkar and Sharpe, 1999) and genetic and molecular evidence indicates that the epithelium must remain responsive to canonical Wnt

signaling. When a mutated, transcriptionally silent, version of βcatenin is exclusively expressed in the epithelium (i.e., epithelial cells become unresponsive to canonical Wnt signaling), tooth development stops abruptly (Cantù et al., 2017). Intriguingly, Frzb expression coincides with the dental mesenchymal territory that receives the growing neurons (Pagella et al., 2014). Indeed, from the earliest stages of odontogenesis, nerve fibers grow toward the tooth germ and progressively innervate the dental follicle. On the contrary, these nerves do not penetrate the dental papilla mesenchyme until the late stage of tooth mineralization (Pagella et al., 2014). It is thus possible that Frzb expression could regulate tooth innervation and probably morphogenesis. Indeed, increasing evidence indicates that innervation plays an active role in organ morphogenesis and development (Kumar and Brockes, 2012; Pagella et al., 2014). Taken together the present results suggest that the developing dental follicle necessitates a specific inhibition of the activity of the Wnt ligands. Conditional knockout and overexpression studies will constitute fundamental approaches to determine the exact role of Frzb in odontogenesis.

Both the dental pulp (formed by the Frzb non-expressing dental papilla cells) and the periodontal ligament (formed by

# REFERENCES

Aurrekoetxea, M., Irastorza, I., García-Gallastegui, P., Jiménez-Rojo, L., Nakamura, T., Yamada, Y., et al. (2016). Wnt/β-catenin regulates the activity of epiprofin/Sp6, SHH, FGF, and BMP to coordinate the stages of odontogenesis. Front. Cell Dev. Biol. 4:25. doi: 10.3389/fcell.2016.00025

the Frzb expressing dental follicle) contain stem cells, the dental pulp and the periodontal ligament stem cell populations, which hold the promise of regenerative approaches (Huang et al., 2009; Mitsiadis et al., 2011; Jiménez-Rojo et al., 2014). The present findings increase the arsenal of markers that permits the specific isolation of mesenchymal cells with desired differentiation potential by the combined use of mesenchymal markers. An increased knowledge of the regulation of mesenchymal dental cells will allow the future development of novel treatments for dental tissue repair and regeneration.

# AUTHOR CONTRIBUTIONS

TM designed the project, performed the experiments, interpreted the data, and wrote the manuscript; PP and CC contributed to the interpretation of the data and wrote the manuscript.

# ACKNOWLEDGMENTS

We thank Prof. E. De Robertis (UCLA, USA) for kindly providing the plasmid used for the generation of the Frzb probe. This work was supported by funds of the University of Zurich.


autocrine Wnt/β-catenin signaling in response to tooth damage. Sci. Rep. 7:3102. doi: 10.1038/s41598-017-03145-6


**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 © 2017 Mitsiadis, Pagella and Cantù. 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) or licensor 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.

# Expression of Nerve Growth Factor (NGF), TrkA, and p75NTR in Developing Human Fetal Teeth

Thimios A. Mitsiadis \* and Pierfrancesco Pagella

Orofacial Development and Regeneration, Institute of Oral Biology, Center for Dentistry (ZZM), University of Zurich, Zurich, Switzerland

Nerve growth factor (NGF) is important for the development and the differentiation of neuronal and non-neuronal cells. NGF binds to specific low- and high-affinity cell surface receptors, respectively, p75NTR and TrkA. In the present study, we examined by immunohistochemistry the expression patterns of the NGF, p75NTR, and TrkA proteins during human fetal tooth development, in order to better understand the mode of NGF signaling action in dental tissues. The results obtained show that these molecules are expressed in a wide range of dental cells of both epithelial and mesenchymal origin during early stages of odontogenesis, as well as in nerve fibers that surround the developing tooth germs. At more advanced developmental stages, NGF and TrkA are localized in differentiated cells with secretory capacities such as preameloblasts/ameloblasts secreting enamel matrix and odontoblasts secreting dentine matrix. In contrast, p75NTR expression is absent from these secretory cells and restricted in proliferating cells of the dental epithelium. The temporospatial distribution of NGF and p75NTR in fetal human teeth is similar, but not identical, with that observed previously in the developing rodent teeth, thus indicating that the genetic information is well-conserved during evolution. The expression patterns of NGF, p75NTR, and TrkA during odontogenesis suggest regulatory roles for NGF signaling in proliferation and differentiation of epithelial and mesenchymal cells, as well as in attraction and sprouting of nerve fibers within dental tissues.

Edited by: Gianpaolo Papaccio,

Seconda Università degli Studi di Napoli, Italy

### Reviewed by:

Eumorphia Remboutsika, Biomedical Sciences Research Center (BSRC) "Alexander Fleming," Greece Giovanna Orsini, Marche Polytechnic University, Italy

> \*Correspondence: Thimios A. Mitsiadis thimios.mitsiadis@zzm.uzh.ch

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 13 July 2016 Accepted: 21 July 2016 Published: 03 August 2016

### Citation:

Mitsiadis TA and Pagella P (2016) Expression of Nerve Growth Factor (NGF), TrkA, and p75NTR in Developing Human Fetal Teeth. Front. Physiol. 7:338. doi: 10.3389/fphys.2016.00338 Keywords: human, tooth, development, nerve growth factor, odontoblast, ameloblast, TrkA, p75NTR

# INTRODUCTION

The roles of nerve growth factor (NGF) in the development, survival and maintenance of selected group of neurons of the peripheral and central nervous system have been thoroughly evaluated in the last few decades (Chao, 2003; Lu et al., 2005; Reichardt, 2006; Ichim et al., 2012; Lewis and Carter, 2014). Dependence of neurons on NGF varies as a function of the stage of development. Additional members of the NGF-related family of neurotrophic molecules include brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), neurotrophin-4 (NT-4; also known as NT-4/5 or NT-5), and neurotrophin-6 (NT-6; Chao, 2003; Lu et al., 2005; Reichardt, 2006; Lewis and Carter, 2014). NGF-related neurotrophins (NTFs) support the survival and outgrowth of various neuronal populations (Reichardt, 2006; Ichim et al., 2012). Target cells for NTFs bear specific cell-surface receptors and their presence is indicative of a potentially responsive cell (Chao, 2003; Reichardt, 2006). Two binding affinities of NTFs to their receptors, one high, the other low, have been described. A transmembrane glycoprotein called lowaffinity NGF receptor (p75NTR) binds all NTFs with low-affinity (Radeke et al., 1987; Chao, 1992; Reichardt, 2006). However, it is still unclear whether the low-affinity form is capable of mediating all biological responses of NTFs. The products of the tyrosine kinase trk family of proto-oncogenes bind also NTFs, and are components of the high-affinity receptor. The trk gene family is formed of three characterized genes, trkA, trkB, trkC (Chao, 1992; Barbacid, 1994; Reichardt, 2006; Lewis and Carter, 2014). The trkA gene encodes a 140 kDa glycoprotein with a tyrosine kinase activity, which functions as a NGF receptor (Klein et al., 1991). Functional high-affinity NGF binding requires either coexpression and binding to both p75NTR and TrkA (Kaplan et al., 1991) or binding to dimers of TrkA (Chao, 2003; Reichardt, 2006).

Novel roles for NTFs in embryonic development are proposed by the presence of p75NTR and Trk receptors during organ morphogenesis and differentiation of non-neuronal cells (Chesa et al., 1988; Yan and Johnson, 1988; Represa and Bernd, 1989; von Bartheld et al., 1991; Nakamura et al., 2007; Di Girolamo et al., 2008; Truzzi et al., 2011; Tomellini et al., 2015). Indeed, expression of both p75NTR and NGF in the developing rodent teeth (Yan and Johnson, 1988; Byers et al., 1990; Mitsiadis et al., 1992, 1993; Mitsiadis and Luukko, 1995) suggests that NTFs play multiple roles in odontogenesis, dental cell function, and tooth homeostasis. The tooth develops as a result of sequential and reciprocal interactions between the oral ectoderm and the cephalic neural crestderived mesenchyme (Mitsiadis and Graf, 2009). Differentiation of tooth-specific cells gives rise to the mesenchymal-derived odontoblasts that produce the organic matrix of dentine, and the epithelial-derived ameloblasts that elaborate the enamel matrix proteins. In rodents, concomitant expression of p75NTR and NGF in dental mesenchyme is correlated with odontoblast differentiation, whereas in dental epithelium their co-expression corresponds mostly to proliferative phenomena (Mitsiadis et al., 1992, 1993; Mitsiadis and Luukko, 1995). These findings indicate that NGF may be implicated in morphogenetic and mineralization events by affecting either proliferation or differentiation of dental cells (Mitsiadis et al., 1993).

Although numerous studies are undertaken in rodents to understand the role of NGF signaling in tooth development and regeneration, only limited studies exist in humans. Previous data have focused only on the localization of p75NTR in both embryonic and adult teeth. These reports have shown that in the developing fetal teeth p75NTR is expressed transiently in both dental papilla mesenchyme and inner dental epithelium (Christensen et al., 1993), whereas in adult teeth the receptor is present only in unmyelinated axons and Schwann cells of the pulp (Maeda et al., 1992). To date, there is no available data on the distribution of both NGF and TrkA proteins in the developing human teeth. The present study was conducted to localize areas and specific dental cells that express NGF, p75NTR, and TrkA in developing human teeth, in order to better understand the mode of NGF action in dental tissues.

# MATERIALS AND METHODS

# Embryonic Tissues

# Tissues

Human fetal tissues were obtained from legal abortions. The material comprised teeth from 19 fetuses (5–23 gestational weeks). The gestation age was estimated from the fetal foot length and from the last menstruation of the mother. Embryos were non-infected, and all tissues were both macroscopically and microscopically normal. The fetuses were immediately fixed in 10% buffered formalin for 48 h to 5 days according to the fetus size. Maxillary and mandibular jaws from 5 to 15 weeks old embryos and fetuses were embedded in Paraplast at 56◦C, while the samples ranged in age from 19 to 23 gestational weeks (g.w.) were decalcified for 3 weeks in formic acid/10% formalin prior to embedding in Paraplast. Four to six micrometer thick sections were used for immunohistochemistry. This study was carried out in compliance with the French legislation, after approval of the Regional Ethics Committee of Development and Reproduction of the U.F.R. of Medicine of Reims-France (INSERM 314 Reims).

# Materials

# Antibodies

Preparation, purification and characterization of polyclonal anti-NGF antibodies have been described (Mitsiadis et al., 1992, 1993). Affinity purified mouse anti-human p75NTR monoclonal antibody was the kind gift of Dr. E. M. Johnson Jr. and Dr. C. Osborn (St. Louis, USA). The purification and characterization of the 20.4 antibody, which recognizes the human p75NTR, has been already described (Ross et al., 1984; Grob et al., 1985; Chesa et al., 1988). Polyclonal TrkA antibody was purchased from Abcam (ab76291).

# Chemicals

Vector Vectastain ABC kit was purchased from Biosys (Compiègne, France). Other chemicals were obtained from Sigma (St. Louis, MO, USA).

# Immunohistochemistry

Immunoperoxidase staining was performed as previously described (Mitsiadis et al., 1992, 1993, 2003). Briefly, the sections were deparaffinized, treated with 0.4% pepsin, exposed to a 0.3% solution of hydrogen peroxide in methanol, and then incubated overnight at 4◦C in a humid atmosphere either with polyclonal anti-NGF and anti-TrkA antibodies diluted 1:300 in PBS containing 0.2% BSA, or with the monoclonal anti-p75NTR antibody 20.4 (1:1000 dilution). Positive peroxidase staining produces red/brown color on light microscopy. After staining the sections were mounted with Eukit. In control sections the antibodies were omitted.

# RESULTS

Tooth development proceeds in three well-characterized morphological stages: the bud, cap, and bell stages. At the 5–7th gestational week (g.w.), the epithelial dental buds represent the first epithelial ingrowth into the neural crest-derived mesenchyme at sites corresponding to the position of the future fetal teeth. At that stage, weak NGF immunostaining was evident only in the dental mesenchyme (**Figures 1A**, **2A**), a pattern similar to that observed for TrkA (**Figure 1C**). While p75NTR labeling was absent from both epithelial and mesenchymal components at the 5th g.w., intense immunoreactivity was seen only in nerve fibers next to the developing tooth germ (**Figures 1B, 3A**).

Three epithelial cell populations compose the dental epithelium at the cap stage of development (8–15th g.w.): the outer dental epithelium, the inner dental epithelium and the stellate reticulum. The mesenchyme surrounded by the inner dental epithelium gives rise to the dental papilla (future dental pulp), whereas the mesenchyme limiting the dental papilla and encapsulating the enamel organ forms the dental follicle, which gives rise to the supporting tissues of the tooth (future

mesenchyme (dm) at the early cap stage (B,C), while the staining is mostly seen in the dental follicle (df) at the late cap stage (D,E). Cells from the inner dental epithelium (ide) start to express p75NTR, while no staining is detected in the dental papilla (dp) at this late stage (E). Note that nerve fibers (nf) surrounding the dental follicle are strongly stained. (F) Higher magnification of the flank area of a tooth germ at the early bell stage of development (Figure 1H). Restricted p75NTR reactivity in undifferentiated cells of the inner dental epithelium. Note that the staining is absent in dental pulp (p). Dotted lines represent the border between the dental epithelial and mesenchymal components. Additional abbreviations: de, dental epithelium; m, mesenchyme; ode, outer dental epithelium; oe, oral epithelium; pa, preameloblasts; po, preodontoblasts; sr, stellate reticulum. Scale bars: 100 µm.

periodontium). During the cap stage, NGF expression was highly dynamic. At early cap stage NGF immunoreactivity was absent from the dental mesenchyme and the adjacent dental epithelium, while it was present in the dental epithelium located far from the mesenchyme (**Figure 2B**). At the late cap stage, however, NGF was clearly expressed in the inner and outer dental epithelia, in the dental papilla and dental follicle, while expression was very weak in the stellate reticulum (**Figure 1D**). p75NTR expression was also extremely dynamic at this stage. At the 8th g.w. (early cap stage), intense p75NTR staining was seen in the condensed dental mesenchyme as well as growing nerve axons, while p75NTR staining was absent from all cells of the dental epithelium (**Figures 3B,C**). At the late cap stage (12th g.w.), p75NTR expression decreased in the dental papilla, while the staining was concentrated in the dental follicle surrounding the dental epithelium (**Figure 3D**). At 15th g.w., p75NTR labeling was restricted in proliferating cells of the inner dental epithelium, in cell of the dental follicle, as well as in nerve fibers (**Figures 1E, 3E**). At the early cap stage (8th g.w.), TrkA immunoreactivity was absent from both the dental epithelium and dental mesenchyme, while staining was observed in the mesenchyme surrounding the developing tooth germ (**Figure 4A**). At the late cap stage (15th g.w.), TrkA staining was localized in the dental papilla, as well as in nerve fibers surrounding the tooth germ (**Figures 2F, 4B**).

Continuous growth of the tooth germ leads to the bell stage of development (18th–23rd g.w.). The cells of the inner dental epithelium and outer dental epithelium rapidly proliferate in an apical direction and the tooth shape (i.e., crown morphology) is now observable. Dentinogenesis is initiated at the tips of the cusps at the late bell stage (23rd g.w.). Changes occur in peripheral undifferentiated dental pulp cells, which are separated from the inner dental epithelium by a cell-free zone. Mesenchymal cells adjoining this zone polarize, differentiate into odontoblasts and start to secrete the organic matrix of predentinee/dentinee. Changes also occur in the adjacent inner

FIGURE 4 | Expression of TrkA during tooth development. (A) A tooth germ at the early cap stage of development. TrkA immunoreactivity in red color. (B) Higher magnification of a tooth germ at the late cap stage of development (Figure 1F). (C–E) Higher magnifications of a tooth germ at the early bell stage of development (Figure 1I), representing the tip of the cusp area (C), the cervical loop and dental follicle territories (D), and an area of the flanks of the forming tooth crown (E). (F–K) TrkA staining in tooth germs at the late bell stage of development. (G–K) Higher magnifications representing the tip of the cusp areas (G,I), areas of the flanks of the forming tooth crown (H,J), and the dental follicle territory (K). Dotted lines represent the border between the dental epithelial and mesenchymal components. Abbreviations: a, ameloblasts; cl, cervical loop; d, dentinee; de, dental epithelium; df, dental follicle; dm, dental mesenchyme; dp, dental papilla; ide, inner dental epithelium; m, mesenchyme; nf, nerve fibers; o, odontoblasts; ode, outer dental epithelium; p, dental pulp; pa, preameloblasts; si, stratum intermedium; sr, stellate reticulum. Scale bars: (A,B,F) 100 µm, (C–E, G–J) 25 µm.

dental epithelial cells that change their shape, stop dividing, polarize, and differentiate into preameloblasts/ameloblasts that assume their enamel-forming function. During the early bell stage (18th g.w.), NGF immunoreactivity was observed in all dental epithelial cells, whereas the labeling was absent from the dental pulp (**Figure 1G**). p75NTR staining was evident in proliferating cells of the inner dental epithelium but the staining was considerably decreased in preameloblasts at the tip of the cusp (**Figures 1H, 3F**). Strong TrkA labeling was detected in cells of the inner dental epithelium, preameloblasts, stratum intermedium, and cells of the outer dental epithelium (**Figures 1I**, **4C–E**), while a very weak staining was detected in pulp fibroblasts at the cusp area (**Figure 4C**). No TrkA immunoreactivity was detected in the stellate reticulum.

At the late bell stage (23rd g.w.) strong NGF staining was observed in undifferentiated (i.e., inner dental epithelium, outer dental epithelium, stellate reticulum, stratum intermedium) and differentiated dental epithelial (i.e., preameloblasts, ameloblasts) as well as in preodontoblasts, and odontoblasts (**Figures 2C–E**). Undifferentiated cells exhibited a strong NGF labeling on their surface (**Figures 2E–G**). Weak NGF immunoreactivity was also detected in the dental follicle (**Figures 2E–G**), while the staining was stronger in nerve fibers surrounding the tooth germ (**Figures 2F,G**). At this stage, p75NTR immunoreactivity was absent in ameloblasts, odontoblasts, and fibroblasts of the dental pulp, but persisted in undifferentiated inner dental epithelial cells near the cervical loop region (**Figure 3F** and data not shown). In contrast, TrkA labeling was detected in preameloblasts and differentiating ameloblasts, while the staining was considerably decreased in mature ameloblasts (**Figures 4F–J**). In the dental pulp, strong TrkA immunoreactivity was observed in differentiating odontoblasts, but this reactivity was significantly reduced in secreting odontoblasts (**Figures 4F–J**). TrkA immunostaining was also expressed in cells of the dental follicle (**Figure 4K**).

# DISCUSSION

Sequential and reciprocal interactions between oral epithelium and cranial neural crest-derived mesenchyme result in tooth formation and generation of specific hard tissues, the enamel and dentine (Mitsiadis and Graf, 2009; Jussila and Thesleff, 2012). The present study describes the distribution of NGF, p75NTR , and TrkA proteins in the developing human fetal teeth, and confirms that their temporospatial distribution in human dental tissues is very similar with that observed previously in rodents (Yan and Johnson, 1988; Byers et al., 1990; Mitsiadis et al., 1992, 1993; Mitsiadis and Luukko, 1995). NGF is expressed in the epithelium during the early stages of tooth development, while p75NTR and TrkA are first expressed in the mesenchyme. The expression of NGF, TrkA, and p75NTR in the tooth germ significantly precedes the onset of tooth innervation. This pattern of expression suggests a paracrine mode of action of NGF during this stage of odontogenesis and also indicates a role of this molecule in epithelial-mesenchymal interactions. Our present data in humans and previous findings in rodents show p75NTR expression in the condensed mesenchyme of the cap staged molars (Mitsiadis et al., 1992; Mitsiadis and Luukko, 1995). Adhesive functions have been proposed for p75NTR during neuronal development (Chao, 1992; Mirnics et al., 2005), suggesting that this low affinity NGF receptor may be implicated in the condensation of the neural crest-derived mesenchyme and its specification in dental mesenchyme. p75NTR expression in dental follicle coincides with the outgrowths of the maxillary and mandibular nerves around the tooth germs. One of the proposed functions of p75NTR is to increase the local concentrations of NGF (Chao, 1992), thus providing a tropic and trophic support for specific neurons that follow the concentration gradients of NGF.

At later stages of tooth formation, when cytodifferentiation starts, NGF and TrkA are distributed in proliferating cells of the inner dental epithelium, as well as in preameloblasts and the enamel secreting ameloblasts. However, p75NTR expression in dental epithelium is more restricted and confined to undifferentiated, still proliferating, inner dental epithelial cells. The same pattern of p75NTR distribution has been previously detected in human fetal teeth (Christensen et al., 1993) and in the developing rat molars (Byers et al., 1990; Mitsiadis et al., 1992, 1993; Mitsiadis and Luukko, 1995). Concomitant NGF, TrkA, and p75NTR expression in proliferating cells of the inner dental epithelium suggests that NGF may act as a mitogenic factor. In fact, this function of NGF has been already evidenced in the developing human skin (Yaar et al., 1991), human airway smooth muscles (Freund-Michel et al., 2006), and the whisker follicles of the mouse (Davies et al., 1987). During cytodifferentiation, NGF and TrkA are localized around the nucleus and in the apical part of differentiating and functional ameloblasts, whereas p75NTR is not expressed in these cells, thus indicating that NGF can accomplish its biological activities (i.e., differentiation, enamel matrix synthesis, and deposition) in these cells through the TrkA receptor.

In the dental pulp NGF and TrkA are co-expressed in polarizing and differentiated odontoblasts, also indicating that NGF, upon binding to the TrkA receptor, could regulate the function and differentiation of mesenchymal cells, as suggested by previous in vitro studies (Arany et al., 2009). The expression pattern of NGF in the pulp of developing human teeth is identical to that previously observed in the pulp of developing rodent teeth (Mitsiadis et al., 1992, 1993; Mitsiadis and Luukko, 1995), illustrating a close correlation between the appearance of NGF and the process of odontoblast differentiation. The present results also confirm prior studies on p75NTR expression in embryonic human teeth, where this low-affinity NGF receptor was absent from dental pulp cells at the bell stage (Christensen et al., 1993). However, in contrast to our previous findings in developing rodent teeth showing p75NTR expression in preodontoblasts and cells of the sub-odontoblastic layer (Mitsiadis et al., 1992, 1993), in human teeth these two cell populations lack p75NTR , thus indicating that the mode of NGF action could be variable according to the species.

In the developing human tooth germs, NGF, p75NTR, and TrkA are also expressed in the dental follicle (i.e., future

periodontium), where mark mainly the nerve fibers. It has been shown that NGF and TrkA play a fundamental role in the innervation and nociception of dental tissues, since mutations in the gene coding for TrkA (i.e., NTRK1) leads to pain insensitivity and in some cases to tooth agenesis (Bonkowsky et al., 2003; Gao et al., 2013). In rodents, tooth movement up-regulates NGF, p75NTR, and TrkA expression in the periodontal ligament, which is associated with increased sensory nerve fibers sprouting and pain perception (O'Hara et al., 2009). NGF and TrkA expression in the dental follicle indicates that these molecules might modulate nerve growth and sprouting in the periodontal ligament of human teeth in both physiological and pathological conditions. Similarly, NGF expression is up-regulated upon injury in the pulp of mouse molars, leading to massive sprouting of TrkA expressing nerve fibers toward the site of injury (Sarram et al., 1997). The above mentioned findings suggest that innovative strategies could be applied in dental clinics for patients pain relief, since TrkA-positive nerve fibers are important mediators of pain perception in dental tissues (Sullins et al., 2000; Gao et al., 2013; Kyrkanides et al., 2015). Inhibition of NGF/TrkA signaling could be an effective approach to eliminate or reduce pain associated with dental pathologies and clinical interventions. In this context, NGF/TrkA/p75NTR gained enormous attention as potential targets for treatments against cancer-associated pain, leading to the development and trial of several classes of smallmolecule-inhibitors and antibodies targeting this signaling axis (Demir et al., 2016).

Apart from its implication in pain perception, the neuroattractive effect of NGF signaling could be an important modulator of tooth development and regeneration. Innervation plays a key role in the development and the regeneration of orofacial organs (Pagella et al., 2014), and consists one of the fundamental component of stem cell niches (Katayama et al., 2006; Méndez-Ferrer et al., 2008; Knox et al., 2013; Pagella et al., 2015). Both the periodontium and the dental pulp host various stem cell populations that guarantee their repair and regeneration. It has been evidenced on mouse models that tooth injury leads to the activation of NGF signaling and the concomitant attraction of nerve fibers (Sarram et al., 1997; Yamashiro et al., 2000; Kaukua et al., 2014), which might provide necessary but so far unidentified cues for proper formation and regeneration of dental tissues.

In conclusion, the present findings suggest a regulatory role for NGF in both mesenchymal and epithelial components of the developing human teeth. NGF, p75NTR, and TrkA expression in

# REFERENCES


dental tissues (**Figure 5**) indicates a function for NGF signaling in cell proliferation, differentiation and mineralization events during odontogenesis, paralleling its role in the development of the nervous system. More precisely, expression of all these three molecules in inner dental epithelial cells correlates with their proliferation status, while co-expression of NGF and TrkA alone in dental epithelial and mesenchymal cells is associated with their differentiation into ameloblasts and odontoblasts. Moreover, expression of the NGF, p75NTR, and TrkA proteins in nerve fibers of the developing human teeth indicates that NGF signaling is also involved in their sprouting and attraction toward dental tissues.

# AUTHOR CONTRIBUTIONS

TM and PP equally contributed to the collection of data, analysis of results, writing, and editing of the manuscript.

# ACKNOWLEDGMENTS

We are grateful to Professor Dr. D. Laurent-Maquin and Dr. D. Gaillard (INSERM UMR-S 926 "Interfaces Biomatériaux/Tissus Hôtes," U.F.R. d'Odontologie de Reims, France) for kindly providing us with human tissues. This work was supported by funds of the University of Zurich. The authors confirm that there are no conflicts of interest associated with this work.


**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 © 2016 Mitsiadis and Pagella. 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) or licensor 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.

# Distribution of syndecan-1 protein in developing mouse teeth

# *Anna Filatova , Pierfrancesco Pagella and Thimios A. Mitsiadis\**

*Division of Orofacial Development and Regeneration, Faculty of Medicine, Institute of Oral Biology, ZZM, University of Zurich, Zurich, Switzerland*

### *Edited by:*

*Claudio Cantù, University of Zurich, Switzerland*

### *Reviewed by:*

*Jean-Christophe Farges, University Lyon 1, France Catherine Chaussain, Université Paris Descartes Paris Cité, France Giovanna Orsini, Polytechnic University of Marche, Italy*

### *\*Correspondence:*

*Thimios A. Mitsiadis, Division of Orofacial Development and Regeneration, Faculty of Medicine, Institute of Oral Biology, ZZM, University of Zurich, Plattenstrasse 11, 8032 Zurich, Switzerland e-mail: thimios.mitsiadis@ zzm.uzh.ch*

Syndecan-1 is a cell surface proteoglycan involved in the regulation of various biological processes such as proliferation, migration, condensation and differentiation of cells, intercellular communication, and morphogenesis. The extracellular domain of syndecan-1 can bind to extracellular matrix components and signaling molecules, while its intracellular domain interacts with cytoskeletal proteins, thus allowing the transfer of information about extracellular environment changes into the cell that consequently affect cellular behavior. Although previous studies have shown syndecan-1 expression during precise stages of tooth development, there is no equivalent study regrouping the expression patterns of syndecan-1 during all stages of odontogenesis. Here we examined the distribution of syndecan-1 protein in embryonic and post-natal developing mouse molars and incisors. Syndecan-1 distribution in mesenchymal tissues such as dental papilla and dental follicle was correlated with proliferating events and its expression was often linked to stem cell niche territories. Syndecan-1 was also expressed in mesenchymal cells that will differentiate into the dentin producing odontoblasts, but not in differentiated functional odontoblasts. In the epithelium, syndecan-1 was detected in all cell layers, by the exception of differentiated ameloblasts that form the enamel. Furthermore, syndecan-1 was expressed in osteoblast precursors and osteoclasts of the alveolar bone that surrounds the developing tooth germs. Taken together these results show the dynamic nature of syndecan-1 expression during odontogenesis and suggest its implication in various processes of tooth development and homeostasis.

**Keywords: syndecan-1, tooth, incisor, odontoblast, ameloblast, tissue interactions, stem cells, stem cell niches**

# **INTRODUCTION**

Syndecan-1 is a member of a family formed by four proteoglycans (PGs) containing a C-terminal cytoplasmic domain, a well-conserved single-pass transmembrane domain, and a large N-terminal extracellular domain (Jalkanen et al., 1987; Sanderson and Bernfield, 1988; Saunders et al., 1989; Bernfield et al., 1992; Eriksson and Spillmann, 2012; Pap and Bertrand, 2013). The extracellular domain contains motifs for glycosaminoglycan (GAG) attachment, proteolytic cleavage, and cellular interactions (Salmivirta et al., 1991; Bernfield et al., 1992; Morgan et al., 2007; Eriksson and Spillmann, 2012; Pap and Bertrand, 2013). The binding of the intracellular domain of syndecan-1 to cytoplasmic proteins influences the dynamics of the cytoskeleton and promotes intracellular and membrane trafficking (Morgan et al., 2007). On the cell surface syndecan-1 strongly interacts with heparanase, which increases syndecan-1 shedding at the extracellular domain by stimulating protease expression (Kokenyesi and Bernfield, 1994; Morgan et al., 2007; Pap and Bertrand, 2013). In various tumorigenic conditions, syndecan-1 has been detected in the cell nucleus (Szatmári and Dobra, 2013; Kovalszky et al., 2014; Stewart and Sanderson, 2014).

Syndecan-1 is involved in the epithelial–mesenchymal interactions that take place during organogenesis, principally through its ability to bind growth factors and modulate their downstream signaling (Vainio and Thesleff, 1992; Perrimon and Bernfield, 2000). Syndecan-1 binds to a wide range of heparin-binding proteins such as Fibroblast Growth Factors (FGFs), Midkine (MK), and Hepatocyte Growth Factor (HGF) (Mitsiadis et al., 1995; Perrimon and Bernfield, 2000; Häcker et al., 2005; Teng et al., 2012). Signaling of these growth factors seems to be precisely controlled by regulatory loops involving syndecan-1 expression levels.

Tooth represents a suitable model system for studying epithelial–mesenchymal interactions and morphogenetic events during embryonic development (Thesleff et al., 1995). Sequential and reciprocal interactions between the oral epithelium and the underlying cranial neural crest-derived mesenchyme progressively transform the tooth primordia into complex mineralized structures with distinct cell types (Mitsiadis and Graf, 2009). Signaling molecules such as FGFs, MK, Bone Morphogenetic Proteins (BMPs), Wnt, and Sonic hedgehog (Shh) are involved in these interactions from the earliest stages of tooth initiation until the mineralization events (Mitsiadis and Luder, 2011). The epithelial-derived ameloblasts and the mesenchyme-derived odontoblasts are the highest differentiated dental cells that synthesize and secrete the organic components of the enamel and dentin, respectively.

Filatova et al. Syndecan-1 in odontogenesis

A number of previous data have shown that syndecan-1 is expressed in the developing rodent molars and incisors (Vainio et al., 1989, 1991; Bai et al., 1994; Mitsiadis et al., 1995, 2008; Dias et al., 2005; Muto et al., 2007). Furthermore, a recent study has reported on the expression of syndecan-1 in developing human teeth (Kero et al., 2014). These findings have suggested that syndecan-1 is implicated in the subdivision of the mesenchyme into dental and non-dental territories, controls tooth morphogenesis and influences differentiation events. Although these earlier studies have shown syndecan-1 expression during precise stages of tooth development, there is no equivalent study regrouping the expression patterns of syndecan-1 during all stages of odontogenesis. For this reason, we have performed a systematic analysis of syndecan-1 protein distribution in the developing mouse molars and incisors, as well as in the alveolar bone.

# **MATERIALS AND METHODS**

### **ANIMALS AND TISSUE PREPARATION**

Swiss and C57Bl/6 mice were used at embryonic and postnatal stages. The Veterinary Office of the Canton of Zurich (Switzerland) has approved the protocol working on mice (license Nr. 151/2014). Embryonic age was determined according to the vaginal plug (day 0) and confirmed by morphological criteria. Animals were killed by cervical dislocation and the embryos were surgically removed into Dulbecco's phosphate-buffered saline (PBS), pH 7.4. Dissected heads from the embryonic day 13.5 (E13.5) to the E18.5 mouse embryos and post-natal day 1 (PN1) to PN8 mouse pups were fixed in 4% paraformaldehyde (PFA) in PBS for 24 h at 4◦C. After fixation the post-natal tissues were demineralized in acetic acid 0.1 N in 0.5% PFA in PBS for 5 days and then washed with PBS for 4 h. The heads were then dehydrated and embedded in paraffin wax. Seven micrometer serial sections were mounted on silanized slides and stored in air-tight boxes at 4◦C until immunohistochemistry.

### **IMMUNOHISTOCHEMISTRY ON TISSUE SECTIONS**

A rat anti-mouse syndecan-1 antibody (281-2) was used (50µg/ml). Preparation and characterization of this monoclonal antibody has been described earlier (Jalkanen, 1985). The specificity of this antibody has been verified previously by immunohistochemistry (Mitsiadis et al., 1995). A secondary goat antibody against rat IgG (1:500) was diluted in PBS and incubated at room temperature for 2 h.

Immunoperoxidase staining (ABC kit, Vector Laboratories, Burlingame, CA) was performed as previously described (Mitsiadis et al., 1995). Positive peroxidase staining produces red color on light microscopy. Replacement of primary antibody with normal rat serum served as a negative control.

### **RESULTS**

### **SYNDECAN-1 PROTEIN IN DEVELOPING MOLARS**

At E13, the tooth epithelium forms a bud, around which the mesenchyme condenses. Syndecan-1 immunoreactivity was mainly detected on the surfaces of condensing mesenchymal cells that are close to the dental epithelium, whereas the staining in more peripheral mesenchymal cells was absent (**Figures 1A,B**). A faint

magnification of the **(D)**. Abbreviations: ab, alveolar bone; cm, condensed mesenchyme; de, dental epithelium; df, dental follicle; dp, dental papilla; ED, embryonic day; iee, inner enamel epithelium; m, mesenchyme; oe, oral epithelium; oee, outer enamel epithelium; si, stratum intermedium; sr, stellate reticulum. Size bars, 200µm.

and sporadic staining was also detected on the surfaces of some dental epithelial cells.

During the cap and early bell stages (E15–E17), syndecan-1 staining was seen on the surfaces of both dental epithelium and dental mesenchymal cells (**Figures 1C–E**). In the epithelium, intense reactivity was detected on cells of the outer dental epithelium and stellate reticulum, where the staining was weaker in cells of the inner dental epithelium and stratum intermedium (**Figures 1C–E**). In the mesenchyme, strong staining was observed in dental follicle cells, as well as in cells of the dental papilla in contact with the basement membrane that will differentiate into odontoblasts (**Figures 1C–E**). A weaker immunostaining was seen in fibroblasts of the dental papilla (**Figure 1E**).

During cytodifferentiation (PN3), syndecan-1 labeling was absent from odontoblasts and ameloblasts located at the tip of the cusps (**Figure 2A**) and the more apical areas (**Figure 2C**). However, the staining became very strong in cells of the stratum intermedium (**Figures 2A,C**) and in a lesser degree in stellate

reactivity (red color) in tissues of developing tooth germs at post-natal day 3 (PN3) **(A,C)** and PN10 **(B,D)**. Abbreviations: a, ameloblasts; ab, alveolar bone; ca, cervical area; d, dentin; df, dental follicle; dp, dental pulp; e, enamel; m, mesenchyme; o, odontoblasts; oe, oral epithelium; oee, outer enamel epithelium; rs, root sheath; si, stratum intermedium; sr, stellate reticulum; v, vessels. Size bars, 100µm.

reticulum (**Figure 2A**). A faint reactivity was observed in the extracellular matrix of the dental pulp at the crown area (**Figure 2A**), while strong immunostaining was found in mesenchymal cells that are close to the forming epithelial root sheath at the cervical area (**Figure 2C**).

During mineral matrix deposition at P10, cells of the stratum intermedium, stellate reticulum and outer dental epithelium exhibited a very strong syndecan-1 reactivity (**Figures 2B**, **3A**), while a faint labeling was detected in mature ameloblasts (**Figure 2B**). In the mesenchymal components of the tooth germ, moderate immunoreactivity was found in dental follicle cells and cells at the central part of the pulp (**Figure 2D**), whereas the staining persisted in dental pulp cells that are located near to the growing epithelial root sheaths (**Figures 2D**, **3A**). A moderate syndecan-1 labeling was detected in epithelial cells forming the root sheath (**Figure 2D**) and in cells related to blood vessels (**Figure 2D**).

### **SYNDECAN-1 PROTEIN IN DEVELOPING INCISORS**

While at early stages of incisor development (E13.5) syndecan-1 presented similar distribution patterns to those observed in molar, at more advanced stages (E14.5–E16.5) an intensive labeling was detected in dental follicle mesenchymal cells (data not shown). A weak staining was observed in cells of the inner dental epithelium and stratum intermedium, whereas the immunoreactivity was absent from cells of the dental papilla mesenchyme (data not shown; Mitsiadis et al., 2008). From P1 to P8, syndecan-1 staining was detected on the surface of inner enamel epithelial cells and cells of the stratum intermedium, stellate reticulum and outer dental epithelium at the labial side of the incisor (enamel forming side), while in preameboblasts/ameloblasts the labeling was weak (**Figures 3B**, **4A**). Strong syndecan-1 reactivity was observed in mesenchymal cells located at the central part of the pulp and the cervical loop region (**Figures 3B,C, 4A,B**), whereas the staining was absent in functional odontoblasts (**Figures 3B**, **4A**). A similar pattern was seen in pulp cells located at the lingual side of the incisor (cementum forming side) (**Figure 4B**), where cells of the inner dental epithelium were negative for syndecan-1 (**Figures 4C,D**). Cells of the periodontal ligament exhibited very intense syndecan-1 immunoreactivity (**Figures 4B–D**).

### **SYNDECAN-1 PROTEIN IN THE ALVEOLAR BONE**

Alveolar bone formation and remodeling occurs during the mineralization stages of tooth development. At PN10, strong syndecan-1 immunoreactivity was detected in osteoblastic cells of the alveolar bone that surrounds the developing tooth germs (**Figures 5A–C**), as well as in osteoclasts (**Figures 5C–E**).

# **DISCUSSION**

Transcription factors, secreted growth factors and extracellular matrix molecules provide the molecular signals for the epithelial– mesenchymal crosstalk that controls tooth development (Thesleff et al., 1995; Mitsiadis and Graf, 2009; Mitsiadis and Luder, 2011). The present study shows that syndecan-1 is distributed in distinct areas of the epithelium and mesenchyme during the early and late stages of odontogenesis. Syndecan-1 is involved in the signaling between epithelium and mesenchyme that is necessary for normal progression of tooth morphogenesis (Vainio et al., 1989). Syndecan-1 expression in the dental mesenchyme by E13.5 is associated with the transfer of the odontogenic potential from the epithelium to the mesenchyme (Vainio et al., 1989; Mitsiadis and Graf, 2009). Previous studies have shown that expression of syndecan-1 at these early stages of odontogenesis correlates with proliferation of mesenchymal cells (Vainio et al., 1991; Vainio and Thesleff, 1992). More recent studies have revealed that syndecan-1 also stimulates cell proliferation and migration in various pathological conditions (Iozzo and Sanderson, 2011; Teng et al., 2012). During the early bell stage of odontogenesis, syndecan-1 is distributed in both epithelial and mesenchymal compartments of the tooth germs. Once more, expression of syndecan-1 in the dental follicle mesenchyme is associated with intensive proliferative activity. In contrast, increased expression in odontoblast precursors (preodontoblasts) indicate that syndecan-1 may play an additional role in differentiation events (Hall and Miyake, 1995).

**FIGURE 3 | Comparison of syndecan-1 protein distribution in first molars and incisors of PN10 mouse pups.** Immunohistochemistry on frontal sections. Sections through the molar **(A)** and incisor **(B,C)**. Syndecan-1 immunoreactivity in red color. **(C)** Higher magnification of **Figure 3B**.

Abbreviations: a, ameloblasts; ab, alveolar bone; ca, cervical area; cl, cervical loop; d, dentin; de, dental epithelium; df, dental follicle; dp, dental pulp; e, enamel; o, odontoblasts; oe, oral epithelium; oee, outer enamel epithelium; rs, root sheath; sr, stellate reticulum. Size bars, 200 µm.

During the mineralization stage, the tooth crown achieves its definitive shape and the root formation begins. Functional ameloblasts that synthesize and secrete enamel matrix proteins express very little, if not any, syndecan-1 protein. Nevertheless,

**FIGURE 5 | Syndecan-1 protein distribution in alveolar bone of PN3 and PN10 mouse pups.** Immunohistochemistry on frontal sections **(A–E)**. Syndecan-1 reactivity in red color. Abbreviations: ab, alveolar bone; ob, osteoblasts; oc, osteocytes; ocl, osteoclasts. Size bars, 200µm.

increased syndecan-1 expression is seen in other epithelial cell layers such as stellate reticulum and stratum intermedium. Decrease of syndecan-1 expression in dental pulp and dental follicle correlates with the progression of mesenchymal cell differentiation. However, syndecan-1 expression persists in mesenchymal cells located near to the growing epithelial root sheaths at the apical part of the tooth germ.

Rodent incisors are continuously erupting teeth, characterized by distinct zones of cell proliferation, differentiation, and maturation along their anterior–posterior axis (Mitsiadis et al., 2007; Mitsiadis and Graf, 2009). In incisors, syndecan-1 protein is synthesized in both epithelial and mesenchymal cells of its apical part (cervical loop region), preameboblasts, pulp fibroblasts, perivascular cells and cells of the periodontium. The presence of syndecan-1 in periodontal cells indicates its involvement in periodontium homeostasis and regeneration, as has been already suggested by previous studies (Worapamorn et al., 2000, 2001, 2002). Epithelial stem cells from the cervical loop generate transit amplifying progenitor cells that differentiate into all cell types of the incisor including the ameloblasts (Harada et al., 1999; Mitsiadis et al., 2007). Similarly, mesenchyme stem cells of the apical part of the incisor divide actively and generate the odontoblast progenitors (Mitsiadis et al., 2008). Syndecan-1 is found in the highly proliferative and specialized epithelial and mesenchymal compartments of the apical part of both molars and incisors as well as in perivascular pulp territories. These welldefined anatomical compartments of the teeth form distinct storage sites for stem cells that are defined as stem cell niches. Previous findings have indicated the existence of mesenchymal stem cell niches at the perivascular sites of the dental pulp (Shi and Gronthos, 2003; Lovschall et al., 2007; Mitsiadis et al., 2011). Similarly, many studies have shown the existence of epithelial and mesenchymal stem cell niches at the apical part of the teeth (Mitsiadis et al., 1999, 2011). Extracellular matrix molecules and adhesion molecules composing the dental stem cell niches may influence stem cell behavior (Mitsiadis et al., 2007). Syndecan-1 expression in these specialized dental territories suggests its implication in self-renewing, expansion, spread and progression of stem cells from the dividing and undifferentiated state to that of functional differentiated cells during tooth homeostasis and repair. Repair mechanisms that involve the activation and proliferation of stem cells in order to repopulate the dental injured tissues with the appropriate cell types (About and Mitsiadis, 2001; Mitsiadis and Rahiotis, 2004; Mitsiadis et al., 2011), may depend on syndecan-1 function. Indeed, it has been reported that heparan sulfate proteoglycans (HSPGs) affect proliferation, differentiation, and maintenance of the stem cells in developmental and repair processes (Häcker et al., 2005; Cool and Nurcombe, 2006; Bishop et al., 2007).

Signaling molecules such as Wnt, FGFs, Epidermal Growth Factor (EGF), and Notch are important regulators of stem cell fate and function (Häcker et al., 2005; Shaker and Rubin, 2010; Mitsiadis et al., 2011). It is likely that the diverse roles of these signaling molecules during odontogenesis are dependent on the simultaneous presence of syndecan-1, which is involved in the fine-tuning of their activity (Mali et al., 1993; Mitsiadis et al., 1995; Su et al., 2007). For example, the distribution of the syndecan-1 in the apical tooth areas correlates with the expression of FGF molecules (Harada et al., 1999, 2002; Mitsiadis et al., 2008), thus suggesting that syndecan-1 could modulate the

**(blue color for epithelium, red color for mesenchyme) in the developing mouse molars and incisors (from bud to late bell stages).** Abbreviations: a, ameloblasts; cla, cervical loop area; cm, condensed mesenchyme; de,

epithelium; o, odontoblasts; oe, oral epithelium; ode, outer dental epithelium; oee, outer enamel epithelium; rs, root sheath; si, stratum intermedium; sr, stellate reticulum.

effects of FGFs by either enhancing or reducing their activities (Mali et al., 1993; Bishop et al., 2007; Eriksson and Spillmann, 2012; Pap and Bertrand, 2013). Alternatively, syndecan-1 may act synergistically with these molecules to keep dental cells in an undifferentiated state (Mitsiadis et al., 2008).

Syndecan-1 can have distinct functions during proliferation and/or migration of cells in dental tissues. Several studies have shown that syndecans may be part of the MK and Heparin-Binding Growth Associated Molecule (HB-GAM or pleiotrophin) signaling during embryonic development (Mitsiadis et al., 1995; Nakanishi et al., 1997; Deepa et al., 2004). However, the patterns of syndecan-1, MK, and HB-GAM distribution differ as shown previously by comparative analysis of their expression in developing teeth (Mitsiadis et al., 2008), thus suggesting that syndecan-1 is only part of their signaling in precise developmental stages.

In post-natal life, syndecan-1 is found in both osteoblasts and osteoclasts of the alveolar bone. Syndecan-1 may enhance migratory or differentiation responses in osteoblasts and/or their progenitors. The present findings suggest that syndecan-1 may play a role in alveolar bone formation and remodeling. Indeed, syndecans are involved in skeletogenesis by regulating endochondral ossification and organizing the fine structure of the extracellular matrix (Pap and Bertrand, 2013).

The functional redundancy of syndecans in development has been long suspected due to their common ancestral origin and high sequence homology. Indeed, syndecan-1 knockout mice are healthy and do not exhibit any obvious organ abnormalities by the exception of mammary glands and corneal and skin re-epithelialization (Stepp, 2002; Liu et al., 2004). Furthermore, transgenic mice that overexpress syndecan-1 exhibit impaired wound healing due to the increased levels of shed syndecan-1 ectodomain at the injury site, where it acts as a dominant-negative regulator of syndecan-1 function (Elenius et al., 2004). Thus, syndecan-1 is involved in wound healing processes, and most notably in keratinocyte function and re-epithelialization.

In conclusion, the present results highlight the important and dynamic nature of syndecan-1 during tooth development and homeostasis (**Figure 6**). Syndecan-1 could be used as a marker of stem/progenitor cells that are activated during tooth repair processes. Syndecan-1 function on proliferative, migratory and differentiation events may greatly depend on the concomitant presence of signaling molecules.

### **ACKNOWLEDGMENTS**

This work was supported by the Swiss National Foundation (SNSF) Grant 31003A\_135633 (Thimios A. Mitsiadis, Anna Filatova) and by institutional funds from the University of Zurich (Thimios A. Mitsiadis, Pierfrancesco Pagella). The authors state that there not existing conflicts of interest. The authors wish to thank Professor Markku Jalkanen (Faron Pharmaceuticals, Turku, Finland) for the gift of the syndecan-1 antibody.

### **REFERENCES**

About, I., and Mitsiadis, T. A. (2001). Molecular aspects of tooth pathogenesis and repair: *in vivo* and *in vitro* models. *Adv. Dent. Res.* 15, 59–62. doi: 10.1177/08959374010150011501


muscle cell axis in response to pulp injury. *Int. J. Dev. Biol.* 51, 715–721. doi: 10.1387/ijdb.072393hl


**Conflict of Interest Statement:** The Guest Associate Editor Dr. Claudio Cantù declares that, despite being affiliated to the same institution as author Prof. Thimios Mitsiadis, the review process was handled objectively and no conflict of interest exists. 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.

*Received: 08 December 2014; paper pending published: 16 December 2014; accepted: 18 December 2014; published online: 15 January 2015.*

*Citation: Filatova A, Pagella P and Mitsiadis TA (2015) Distribution of syndecan-1 protein in developing mouse teeth. Front. Physiol. 5:518. doi: 10.3389/fphys. 2014.00518*

*This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology.*

*Copyright © 2015 Filatova, Pagella and Mitsiadis. 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) or licensor 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.*

# Retinoic Acid Excess Impairs Amelogenesis Inducing Enamel Defects

Supawich Morkmued1, 2, 3, 4, 5, Virginie Laugel-Haushalter 1, 2, 3, 4, Eric Mathieu<sup>6</sup> , Brigitte Schuhbaur 1, 2, 3, 4, Joseph Hemmerlé<sup>6</sup> , Pascal Dollé1, 2, 3, 4 , Agnès Bloch-Zupan1, 2, 3, 4, 7, 8, 9, 10 and Karen Niederreither 1, 2, 3, 4, 7 \*

<sup>1</sup> Developmental Biology and Stem Cells Department, Institute of Genetics and Molecular and Cellular Biology (IGBMC), Illkirch, France, <sup>2</sup> Centre National de la Recherche Scientifique, UMR 7104, Illkirch, France, <sup>3</sup> Institut National de la Santé et de la Recherche Médicale, U 964, Illkirch, France, <sup>4</sup> Université de Strasbourg, Illkirch, France, <sup>5</sup> Pediatrics Department, Faculty of Dentistry, Khon Kaen University, Khon Kaen, Thailand, <sup>6</sup> Université de Strasbourg, INSERM UMR\_1121, Biomaterials and Bioengineering, Strasbourg, France, <sup>7</sup> Faculté de Chirurgie Dentaire, Université de Strasbourg, Strasbourg, France, <sup>8</sup> Faculté de Médecine, Fédération de Médecine Translationnelle de Strasbourg, Université de Strasbourg, Strasbourg, France, <sup>9</sup> Hôpitaux Universitaires de Strasbourg, Pôle de Médecine et Chirurgie Bucco-Dentaires, Centre de Référence des Manifestations Odontologiques des Maladies Rares, CRMR, Strasbourg, France, <sup>10</sup> Eastman Dental Institute, University College London, London, UK

### Edited by:

Ophir D. Klein, University of California, San Francisco, USA

### Reviewed by:

Amel Gritli-Linde, University of Gothenburg, Sweden Lucia Jimenez-Rojo, University of Zurich, Switzerland

> \*Correspondence: Karen Niederreither niederre@igbmc.fr

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 08 November 2016 Accepted: 20 December 2016 Published: 06 January 2017

### Citation:

Morkmued S, Laugel-Haushalter V, Mathieu E, Schuhbaur B, Hemmerlé J, Dollé P, Bloch-Zupan A and Niederreither K (2017) Retinoic Acid Excess Impairs Amelogenesis Inducing Enamel Defects. Front. Physiol. 7:673. doi: 10.3389/fphys.2016.00673 Abnormalities of enamel matrix proteins deposition, mineralization, or degradation during tooth development are responsible for a spectrum of either genetic diseases termed Amelogenesis imperfecta or acquired enamel defects. To assess if environmental/nutritional factors can exacerbate enamel defects, we investigated the role of the active form of vitamin A, retinoic acid (RA). Robust expression of RAdegrading enzymes Cyp26b1 and Cyp26c1 in developing murine teeth suggested RA excess would reduce tooth hard tissue mineralization, adversely affecting enamel. We employed a protocol where RA was supplied to pregnant mice as a food supplement, at a concentration estimated to result in moderate elevations in serum RA levels. This supplementation led to severe enamel defects in adult mice born from pregnant dams, with most severe alterations observed for treatments from embryonic day (E)12.5 to E16.5. We identified the enamel matrix proteins enamelin (Enam), ameloblastin (Ambn), and odontogenic ameloblast-associated protein (Odam) as target genes affected by excess RA, exhibiting mRNA reductions of over 20-fold in lower incisors at E16.5. RA treatments also affected bone formation, reducing mineralization. Accordingly, craniofacial ossification was drastically reduced after 2 days of treatment (E14.5). Massive RNA-sequencing (RNA-seq) was performed on E14.5 and E16.5 lower incisors. Reductions in Runx2 (a key transcriptional regulator of bone and enamel differentiation) and its targets were observed at E14.5 in RA-exposed embryos. RNA-seq analysis further indicated that bone growth factors, extracellular matrix, and calcium homeostasis were perturbed. Genes mutated in human AI (ENAM, AMBN, AMELX, AMTN, KLK4) were reduced in expression at E16.5. Our observations support a model in which elevated RA signaling at fetal stages affects dental cell lineages. Thereafter enamel protein production is impaired, leading to permanent enamel alterations.

Keywords: retinoids, tooth, enamel, RNA-seq, mouse models, enamelin

# INTRODUCTION

Enamel formation is a unique biomineralization process involving a highly organized matrix protein scaffold deposition and degradation, leading to hydroxyapatite crystal nucleation generating a dense and tightly aligned network of hydroxyapatite crystals. Mature enamel is the body's hardest tissue. Ameloblasts are cells of ectodermal origin responsible for enamel development. These cells secrete enamel proteins, which are required for correct mineralization and structural maturation. Enamel proteins self-assemble and provide a matrix organization that aligns the thin ribbons of calcium phosphate deposited during enamel appositional growth. Enamel formation is characterized by inductive, secretory, and maturation stages. During the inductive stage, the inner enamel epithelium begins to differentiate. Then at the secretory stage, polarized differentiated ameloblasts release enamel proteins, contributing to the enamel matrix (Bei, 2009). Finally at the maturation stage, ameloblasts absorb water and organic matrix. This dehydration allows dense crystal deposition. Mature enamel is thus extremely strong, because of the density and fine organization of its crystal layers (Bei, 2009; Simmer et al., 2010).

Amelogenesis imperfecta (AI) refers to a group of rare genetic diseases presenting with defects in enamel formation either as isolated trait, or in association with other symptoms. Patient cases of AI are classified into hypoplastic, hypomineralized, or hypomaturation categories based on enamel quantitative or qualitative defects, i.e., thickness, hardness, and/or color. To date, mutations in over 30 genes are associated with non-syndromic or syndromic AI (Bloch-Zupan et al., 2012). These diseases can be recapitulated in several mouse models. For example, hypoplastic or aplastic enamel deficiencies are produced by amelogenin (Amel), ameloblastin (Ambn), and enamelin (Enam) mutations, recapitulating defects in patients (Gibson et al., 2001; Fukumoto et al., 2004; Masuya et al., 2005; Hu et al., 2008, 2014). Studies in mouse models have contributed to understanding how defective ameloblast protein secretion contributes to these diseases. For a given genetic defect, interfamilial, intrafamilial, and individual intraoral variations in phenotype severity are often seen, suggesting that environmental factors come into play. Indeed, ameloblasts are highly sensitive to their environment (Simmer et al., 2010). Body stressors including high fever (Ryynänen et al., 2014), excess fluoride (Yang et al., 2015), and endocrine disrupters such as bisphenol A disrupt ameloblast function (Jedeon et al., 2014). It is likely that a variety of factors contribute to clinical heterogeneity found in AI and to acquired enamel defects such as molar incisor hypomineralization (MIH) or hypomineralized second primary molars (HSPM; Alaluusua, 2010; Jeremias et al., 2013). Our objective is to discover novel factors initiating and influencing enamel regulatory networks, with the aim of designing new strategies to alleviate dental defects. To further characterize nutritional and environmental factors regulating enamel formation, we have focused on the role of retinoic acid (RA), the main active form of vitamin A that plays key roles during vertebrate development (Niederreither and Dollé, 2008). RA is the ligand for nuclear receptors (RARα, β, and γ), which bind as heterodimers with RXRs to DNA regulatory elements termed RA-response elements (RAREs, Rochette-Egly and Germain, 2009). Through this mechanism, RA regulates expression of various target genes (Balmer and Blomhoff, 2002). RA distribution within embryonic tissues is tightly controlled through an interplay between enzymes involved in its synthesis (mainly retinol dehydrogenase 10 and retinaldehyde dehydrogenases [Raldh]1, 2, and 3) and catabolism (Cyp26A1, B1, and C1). As both RA deficiency and excess results in diverse developmental defects, the distribution of active retinoid signaling requires tight regulation to limit potent adverse effects (Rhinn and Dollé, 2012).

Expression of RA receptors, synthesizing, and catabolizing enzymes has been detected in the developing teeth (Bloch-Zupan et al., 1994; Berkovitz et al., 2001; Cammas et al., 2007). Severe dietary vitamin A deficiency in rats leads to hypoplastic enamel, enamel organ metaplasia, dentine dysplasia, and delayed tooth eruption (Mellanby, 1941; Punyasingh et al., 1984; McDowell et al., 1987). Hypervitaminosis A during rodent pregnancy induces exencephaly and craniofacial malformations, along with tooth fusions and supernumerary incisors (Knudsen, 1967 and references therein). These effects may reveal an evolutionary role of RA signaling in the posterior pharyngeal region controlling tooth number (Seritrakul et al., 2012). In culture, RA excess can retard molar growth (Mark et al., 1992), reduce ameloblast differentiation (Kronmiller et al., 1992), and diminish tooth alkaline phosphatase production (Jones et al., 2008). Retinoids may also regulate incisor cervical loop maturation, increasing mitosis and laminin gene expression (Bloch-Zupan et al., 1994).

In vivo models to substantiate RA actions in tooth development are lacking. Interestingly, though, Cyp26b1 knockout mice show a defect in maxillary bone compaction around upper incisors (Maclean et al., 2009). Observing that the fetal tooth has robust expression of Cyp26b1 and Cyp26c1 RA-degrading enzymes, we hypothesized that in vivo, conditions of RA excess may have adverse effects on osteoblast and ameloblast growth regulatory networks. We report that mice born from dams exposed to RA during mid-late pregnancy using a food-based supplementation suffer adult-stage enamel hypoplasia. Effects were strongest when treatments began at the tooth dental lamina-placode stage (E12.5), and continued until early bell developmental stages. Reductions of Enam, Ambn, and Odam mRNA expression in E16.5 lower incisors were observed. High throughput RNA sequencing (RNA-seq) analysis of lower incisors revealed that RA excess perturbs neural crest lineage determinants and pro-ossification growth factor and transcription networks. Combinatorial changes in collagen, extracellular matrix, and calcium homeostasis genes occur at E14.5, followed by a decrease at E16.5 of transcripts encoding pre-ameloblast secretory-stage proteins. These alterations in gene expression are observed several days before the ameloblast lineage begins to differentiate. Retinoid excess targets fetal odontoblasts, along with a range of epithelial enamel protein targets. Our data provides potential avenues through which environmental and nutritional changes may alter the penetrance and expressivity of human enamel defects such as AI or MIH.

# MATERIALS AND METHODS

# Ethics Statement

All animals were maintained and manipulated under animal protocols in agreement with the French Ministry of Agriculture guidelines for use of laboratory animals (IGBMC protocol 2012– 097) and with NIH guidelines, provided in the Guide for the Care and Use of Laboratory Animals. CD1 mice were purchased from Charles River, France. All-trans-RA (Sigma) suspended in ethanol (5 mg/mL) was mixed into 50 mL water and 50 g powdered food to a final concentration of 0.4 mg/g food. The RA-containing food mixture (protected from light) was left in the cage for the mice to feed ad libitum and changed every day. Control groups consisted of CD1 mice given a matching food treatment, but with no added RA. Equal numbers of RA-treated samples vs. controls were randomly assigned to treatment or control groups.

# In situ Hybridization, Beta-Galactosidase (X-gal) Staining, and Skeletal Analysis

In situ hybridization was performed using digoxigeninlabeled RNA probes on 200µM vibratome sections of paraformaldehyde-fixed embryos, which were processed using an Intavis InSituPro robot, as described in detail on the website http://empress.har.mrc.ac.uk/browser/ (Gene Expression section). To analyze patterns of RA-response, we used the RAREhsp68-lacZ reporter transgenic line (Rossant et al., 1991). At least 20 randomized fetal samples from control and matching RA-treated samples were used to test each probe. All expression studies were confirmed in at least 3 independent experiments. X-gal assays were performed on 200µm vibratome sections. Whole-mount fetal alizarin red/alcian blue staining was carried out as described in http://empress.har.mrc.ac.uk/browser/ (Bone, Cartilage, Arthritis, Osteoporosis section).

# Real-Time Quantitative RT-PCR

RT-PCR assays were performed in duplicate on 3 RNA samples for control or RA-treated incisors dissected at E14.5 and E16.5. Total RNA (1µg) was subjected to real-time RT-PCR using SYBR Green Reagents (Qiagen). RNA was extracted using the RNeasy Micro-kit. Oligo-dT primed cDNAs were generated using the Superscript II kit (Invitrogen). The incorporation of SYBR Green into the PCR products was monitored in real-time with a Roche 480 LightCycler. Sequences of primers are given in supplemental Table S1. Target genes were normalized relative to the glyceraldehyde-3-phosphate dehydrogenase (Gapdh) housekeeping gene.

# X-Ray Microtomography

Seven week-old mice were analyzed by X-ray micro-computed tomography (micro-CT) using a Quantum FX micro-CT in vivo Imaging System (Caliper Life Sciences), which operates at 80 kV and 160µA, with high-resolution at 10–80µm pixel size, to assay skull and tooth morphology. Data reconstructions were performed with the Analyze software (v 11.0; Biomedical Imaging Resource, Mayo Clinic, Rochester, MN).

# Scanning Electron Microscopy

The lower incisors of 7 week-old control and RA-treated mice were dissected out of the alveolar bone, rinsed, dehydrated in a graded series of ethanol, and then transferred in a propylene oxide/epon resin (v/v) solution. After embedding the teeth in Epon 812 (Euromedex, Souffelweyersheim, France), they were sectioned sagittally and polished with diamond pastes (Escil, Chassieu, France). The embedded half incisors were etched with a 20% (m/v) citric acid solution for 2 min, rinsed with distilled water, dehydrated in a graded series of ethanol and dried at room temperature. The samples were then coated with a gold-palladium alloy using a HUMMER JR sputtering device (Technics, CA, USA) before performing scanning electron microscopy with a Quanta 250 ESEM (FEI Company, Eindhoven, The Netherlands) with an accelerating voltage of the electrons of 5 kV.

# RNA Sequencing

Total RNA was extracted in quadruplicate from lower incisors at E14.5 and 16.5 (2 days or 4 days after RA treatment) and respective controls. The mRNA-seq libraries were prepared as detailed in the Illumina protocol (supplemental Experimental Procedures). Sequence reads mapped to the mouse reference genome mm10/NCBI37 using Tophat. Only uniquely aligned reads were retained for further analysis. Quantification of gene expression was performed with HTSeq-0.6.1. (see http://wwwhuber.embl.de/users/anders/HTSeq/doc/overview.html). For each transcript the resulting reads per kilobase of exon model per million mapped reads (RPKM) were converted to raw read counts, which were then added for each gene locus. Data normalization was performed as described (Anders and Huber, 2010) and resolved using the DESeq Bioconductor package. Resulting p-values were adjusted for multiple testing, according to Benjamini and Hochberg (1995). Regulated transcripts with an RPKM of >2, an adjusted p < 0.050, and a log2 fold change of >0.3 and < −0.3 at E14.5 and >0.5 and < −0.5 at E16.5 were considered.

# RESULTS

To analyze RA-dependent tooth alterations, we employed a protocol where RA added to the food supply was administered to pregnant CD1 mice (Niederreither et al., 2002), at a concentration of 0.4 mg/g food beginning at E12.5 or later. The treatment period began after craniofacial neural crest migration into the head was complete (Minoux and Rijli, 2010), avoiding earlier stage lethality due to exencephaly. In another study, HPLC analysis carried out after similar RA treatment at 0.1 mg/g food showed that serum RA levels were moderately increased (∼20%) compared with untreated controls (Mic et al., 2003). Treated dams bore litters with 50% lethality, typically with 5– 7 pups. Incisors in both groups erupted at the same age. Once the pups reached adulthood (4–7 weeks-old), we compared 100 randomized control and RA-treated groups macroscopically for dental defects. The labial side of rodent incisor (analog of the crown and covered with enamel) normally has a yellow/orange pigmentation, due to iron present at a net weight of about

0.03% in enamel (Pindborg, 1953). It gives mouse teeth a characteristic color, which is a dark orange in the upper incisors (**Figures 1A,E**). When RA treatments were performed from E12.5 to 16.5, they were found to produce a chalky lightening and length reduction of incisors, changes more pronounced in lower incisors (**Figures 1B,F**). The whiter color and less shiny surface may reflect reduced enamel thickness typical of mouse enamel hypoplasia models (Gibson et al., 2001; Masuya et al., 2005; Hu et al., 2014). Treatments performed at later stages (E13.5–16.5: **Figures 1C,G**, or E14.5–17.5: **Figures 1D,H**) had less severe effects on enamel, suggesting RA has early roles in the oral epithelium starting at the placode stage of tooth initiation.

Samples shown in **Figures 1A,E** (WT) and B,F (RAtreated) were analyzed by X-ray micro-computed tomography (micro-CT). This analysis confirmed lower incisor shortening (**Figures 2A,B**; Figure S1). Optical sections revealed a reduction in enamel mineral density and thickness (molars in **Figures 2C,E**; incisors in **Figures 2D,F**). Alveolar bone at the level of the molars showed greater porosity in the RA-treated animals (boxes in **Figures 2C,E**; Figure S2). Although retinoid gradients shape pharyngeal tooth evolution (Seritrakul et al., 2012), our micro-CT analysis revealed normal molar eruption and cusp morphology (Figure S3), as predicted because RA treatments are initiated at E12.5, after neural crest has completed its migration into the jaw. This analysis also revealed no evident signs of dental attrition from both groups. **Figure 2G** shows scanning electron micrographs (SEM) of enamel prisms of control lower incisor, exhibiting a well-organized decussating pattern. This was disrupted following RA treatment. The most outer enamel was less mineralized when compared with the control tooth, as outer enamel appeared darker in RA-treated animals. Enamel rods of RA-supplemented animals were less densely packed, and as a consequence, areas normally filled with interprismatic enamel seemed empty and showed holes-like pattern (**Figure 2H**), similar to Enam haploinsufficent mice (Hu et al., 2014).

Histological analysis of 1-week old mice after hematoxylineosin staining confirmed that RA treatments produced a shorter, smaller, and disorganized layer of ameloblasts in both molars (**Figures 3A,B**) and incisors (**Figures 3C–H**). This was clearly seen in the secretory zone, where ameloblasts displayed disrupted epithelial sheet organization (**Figure 3H**). In all treated samples, ameloblast adhesion to enamel was impaired (**Figures 3B,D,G,H**). This was first seen histologically as a detachment of the basement membrane, likely causing preameloblast separation from forming odontoblasts (**Figure 3G**, red arrowhead). Non-cellular

FIGURE 1 | RA excess during pregnancy produces stage-specific whitening and size reductions of mouse incisors. Enamel of the upper (A) and lower (E) incisors of 7 weeks-old CD1 control (non-RA treated) mice has a characteristic yellow/orange color, which is consistently darker in the upper incisor. Retinoic acid treatment from E12.5 to 16.5 reduces incisor length (by ∼20%) and lightens enamel color (B,F), suggesting reduced iron accumulation typical of murine models of hypoplastic enamel formation. When RA treatment begins at E13.5 (C,G) or E14.5 (D,H), incisor length reductions and lightening of incisor color are progressively less severe.

organic material was present between layers (**Figure 3B**, black arrowhead). The lower incisor secretory zone enamel layer was slightly thinner (**Figure 3H**), suggesting RA treatment delayed enamel maturation and/or reduced overall mineralization.

Enamel matrix proteins are secreted by ameloblasts and form a matrix directing enamel mineral deposition. Among these proteins, amelogenin is the most abundant (Eastoe, 1979). Enamel fails to form or is hypoplastic in amelogenindeficient mice (Gibson et al., 2001), and in Enam- or Ambnnull mutant mice (Fukumoto et al., 2004; Masuya et al., 2005; Hu et al., 2014). Our real-time RT-PCR analysis of E16.5 lower incisors following E12.5–16.5 RA treatment revealed up to 20-fold reductions in Enam, Ambn, and Odam mRNAs (**Figure 4**). Notably, no RA-induced alterations in Enam, Ambn, or Odam were observed at E14.5 (data not shown).

To assess if reduced enamel protein expression was linked to ectopic activation of RA signaling, we used RARE-hsp68-lacZ transgenic mice as a reporter for RA activity (Rossant et al., 1991). No expression was seen in the tooth germ areas at E13.5 (data not shown), although activity was found in the mandible next to the tooth germs at E14.5 (**Figures 5A,B**). This retinoid reporter transgene exhibited low level of expression in both mandible and maxilla adjacent to the growing incisors and molars in E16.5 untreated fetuses (**Figure 5C** and data not shown). In E16.5

fetuses following RA treatment from E12.5 to 16.5, domains of RARE-lacZ activity broadly extended into alveolar bone and surrounding mesenchyme, but appeared absent from enamel organ and dental ectomesenchyme (**Figure 5D** and data not shown). Note that very low levels of RA signaling may not be detected by such a reporter transgene. Proper control of RA levels is necessary for early neural crest patterning, but RA deficiency does not alter first branchial arch formation (Niederreither et al., 2000). To explore why RA activity was absent from most tissues of the tooth buds, we examined expression of CYP26 family cytochrome P450 enzymes specifically involved in RA catabolism. At E13.5, Cyp26b1 expression (**Figure 5E**) was found to be prominent in mesenchyme surrounding the forming incisors. At E14.5, Cyp26c1 (**Figure 5F**) was prominently expressed in the enamel organ, whereas a low amount of Cyp26a1 transcripts was seen in the dental papilla (**Figure 5F**, insert).

Excesses of vitamin A or RA lead to skeletal fragility by reducing bone formation and by stimulating bone-resorbing osteoclasts (Henning et al., 2015). In the Cyp26b1 mutant, impaired RA catabolism causes long-bone fusions and induces premature osteoblast differentiation into mineralizing osteocytes, truncating bone development in the craniofacial region (Maclean et al., 2009). Alizarin red/alcian blue staining of E14.5 skulls 2 days into RA regime revealed shortened mandibles, and truncated regions of ossified maxilla and premaxilla in the

RA-treated animals (**Figures 6A,B**). The skull frontal plate was also smaller in treated animals, with less ossification. Malformed Meckel's cartilage and truncated mandibles could lead to incisor shortenings. Skeletal staining performed at E15.5 confirmed RA-reduced mineralization (**Figures 6C,D**). RT-PCR analysis showed reductions in Runx2 mRNA in both lower incisor and adjacent alveolar bone at E14.5 in RA-treated animals (**Figures 6E,F**). A 3-fold reduction in the expression of this key target might account for reduced bone mineral deposition (Ducy and Karsenty, 1998), and its mesenchymal localization (**Figure 6G**) implies indirect enamel effects. To exclude systemic non-specific RA effects, we cultured isolated lower incisors or lower incisors with adjacent alveolar bone from E13.5 embryos (Figure S4). When 10−<sup>8</sup> M RA was added to culture medium, Enam levels were dramatically reduced in isolated incisor cultures, indicating RA acted directly on tooth.

To compare global gene expression changes in control vs. RA-treated samples, E14.5 and E16.5 lower incisors were analyzed by high throughput RNA sequencing (RNA-seq). Principal component analysis (PCA plot, Figure S5) and scatter plots (Figure S6) revealed the overall changes observed at E14.5 were tightly clustered, statistically significant, yet often seen as net reductions of ∼20–50% (**Table 1**). RNA-seq analysis confirmed broad effects of RA treatment affecting mineralization-inducing pathways, extracellular collagens, and calcium networks. Genes reduced in expression at E14.5 included Runx, Dlx, Bmp, and Shh, all known inducers of inner enamel epithelium differentiation (Bei, 2009). Net reduction in Runx2 and Dlx5 (**Table 1**) may act combinatorially to reduce the bone biomarker BGLAP (osteocalcin, (Hassan et al., 2004)), a target reduced by ∼2-fold at E16.5 in RA-treated samples (**Table 2**). A novel RA-inhibited target is the matricellular protein Smoc2, whose mutation in human produces oligodontia, microdontia, abnormal root development, dentin dyplasia, and reduced alveolar bone growth (Bloch-Zupan et al., 2011). Retinoids are known to drive uncommitted progenitor cells into neuronal lineages (Maden, 2007). Consistently, DAVID (Database for Annotation, Visualization, and Integrated Discovery) analysis revealed a functional enrichment score of 1.4 E−<sup>6</sup> for neuronal differentiation pathways. Increased Neurogenin1 (Neurog1), Neurogenin2 (Neurog2), and the notch target Hes5, are typical of an RA-induced neuronal differentiation profile (Table S2). At E16.5, Enam was the most reduced target (**Table 2**), also markedly reduced by in situ hybridization analysis (Figure S7). Reductions in ameloblastin, X- linked amelogenin, amelotin, and kallikreinrelated peptidase 4 (Klk4), all encoding enamel proteins (Núñez et al., 2015), were observed (**Table 2**). Notably, these reductions at E16.5 occur much earlier than the characterized times of action of the corresponding proteins in inductive, secretory, or maturation stages of rodent enamel development. Odontoblast-originating signals control ameloblast induction (Bei, 2009). Reductions in mineralization targets (alkaline phosphatase), odontoblast structural proteins (dentin matrix protein 1, dentin sialophosphoprotein), ossification biomarkers (Bglap1/osteocalcin, Bglap2), and calcium homeostasis pathways (calcitonin, vitamin D receptor) were all observed in E16.5 RA-treated samples. Table S3 summarizes how RA excess at E16.5 increases the expression of genes involved in retinoid signaling (including Cyp26b1), Wnt signaling, and neuronal differentiation.

# DISCUSSION

# RA Excess Affects Enamel Matrix Protein Expression Prior to Ameloblast Differentiation

AI refers to rare, inherited diseases characterized by a defect in enamel formation with clinical heterogeneity even within the same family (Bloch-Zupan et al., 2012). These variations, also observed in acquired enamel defects, have been proposed to be due to environmental excess in fluoride (Yang et al., 2015), or to other nutritional, environmental, or behavioral changes (Li et al., 2013), along with genetic makeup of an individual. Retinoids are regulators of skeletal growth and patterning, known to lead to skeletal fragility when given in excess (Henning et al., 2015). Since reciprocal interactions between enamel organ and ectomesenchyme are necessary for alveolar bone, periodontal, and tooth differentiation, we examined if nutritional RA excess could have additional effects on developing teeth. Prior to this study, little was known about in vivo effects of RA on enamel cytodifferentiation. Here we find that RA treatment at a defined window of murine development resulted in permanent defects resembling human AI. Ameloblast differentiation begins at the advanced bell stage (∼E18.5 in mouse), when the inner enamel epithelial originating cells express enamel secretory proteins and follow

mesenchyme.

by processing enzymes (Bei, 2009; Bloch-Zupan et al., 2012). Assuming normal rodent nocturnal feed, in our experimental protocol RA exposure would begin at the bud stage of tooth formation (E13.0). We observed that at E14.5, RA excess impairs expression of Runx, Dlx, bone morphogenetic proteins, while levels of enamel secretory proteins are not altered in either our RNA-seq analysis or RT-PCR analysis at these same stages (data not shown). RA excess changes likely occur prior to ameloblast differentiation, with molecular alterations indicating effects on pro-mineralization signaling.

# Retinoids Have Early Targets in Mineralized Tissue, and Later Effects on Enamel Proteins

Enamel-reducing effects of RA supplementation at E13–14.5 coincide with the initial stages of intramembraneous ossification. An early target is Runx2, a master regulator of skeletal mineralization. Runx2−/<sup>−</sup> mouse mutants have a block in endochondral and intramembraneous osteoblast maturation, and site-specific reductions in collagen type I and alkaline phosphatase expression (Ducy and Karsenty, 1998). Runx2−/<sup>−</sup> mutants lack differentiated odontoblast and ameloblast matrices, indicative of late bell-stage defects (D'Souza et al., 1999; Bronckers et al., 2001). We observe RA-dependent Runx2 reductions at E14.5 bud stage incisors, but not at E16.5, suggesting earlier effects. Cleidocranial dysplasia, an autosomal dominant condition caused by mutations of RUNX2, likewise results in insufficient dentin and enamel mineralization (Xuan et al., 2010) and other dental anomalies (Camilleri and McDonald, 2006). Terminal ameloblast differentiation requires odontoblast-originating signals and matrix components (Balic and Thesleff, 2015). Reports of evolutionarily conserved Runx2 binding sites in Enam, Ambn, and Odam gene promoters Dhamija and Krebsbach, 2001; Lee et al., 2010, suggested a model of RA-inhibitory effects (**Figure 7**). Retinoid excess at E14.5 would induce relatively small, yet combined reductions in Runx2/3, Dlx3/5, and Bmp2/3, predominantly mesenchymal targets that surround the bud stage tooth (Figure S8). By E16.5 these factors collectively regulate early epithelial (enamel

organ) expression of Enam, Ambn, and Amelx. This is plausible because these enamel targets possess evolutionarily conserved binding site-motifs in their proximal promoters (Loots et al., 2002; Cartharius et al., 2005; **Figure 7C**; Table S4). While the bell stage tooth reacts to the high retinoid environment by up-regulating Cyp26b1 levels (Table S3), this change is insufficient to offset strong reductions in enamel matrix protein secretion, which then induce ameloblast differentiation defects.

# Many Enamel Targets Are in the Secretory Phosphoprotein-Encoding (SCPP) Gene Cluster

Our RNA-seq analysis on the whole lower incisors uncovered many genes significantly reduced at E16.5 that regulate ameloblast differentiation, enamel formation, and dentin/bone development (**Table 2**). Hence combinatorial deficits in enamel secretory protein expression included reductions in X-linked amelogenin (AMELX, OMIM: 300391) (Gibson et al., 2001; TABLE 1 | Genes encoding regulators of bone growth, collagens, and proteins involved in calcium signaling and homeostasis, retinoid, and Shh pathways, reduced in expression in E14.5 RA-treated lower incisors.


Data are presented as log2 fold changes in RA-treated vs. control samples: for instance, a FC log2 value of −1.00 will correspond to a 50% mRNA level in the RA-treated samples.

Barron et al., 2010), AMBN (OMIM: 601259) (Fukumoto et al., 2004), AMTN (OMIM: 610912) (Nakayama et al., 2015), along with Odam (OMIM: 614843). These genes belong to the evolutionarily-related SCPP gene cluster, a linked group of genes also containing members regulating skeletal mineralization (Kawasaki and Weiss, 2008). These SCPP enamel proteins contain structural domains promoting calcium sequestration and TABLE 2 | Summary of genes encoding enamel proteins, extracellular matrix components, proteins involved in bone growth pathways, or calcium and iron signaling/homeostasis, all reduced in expression according to RNA-seq analysis of E16.5 RA-treated lower incisors.


Data are presented as log2 fold changes (FC log2) in RA-treated vs control samples: for instance, a FC log2 value of −1.00 will correspond to a 50% mRNA level in the RAtreated samples. Adjusted p-value takes into account a false discovery rate wherein 5% of samples with a p-value of 0.050 will result in false positives.

regulating crystal adsorption (Addison and McKee, 2010), and their mutations produce hypomineralized enamel phenotypes including altered prism patterning and increased cellular apoptosis in both patients and mouse models (Fukumoto et al., 2004; Bei, 2009; Hu et al., 2014; Núñez et al., 2015). Reductions in the SCPP cluster appeared specific to enamel expression. Other enamel-regulating targets included the kallikrein-related

FIGURE 7 | Model of how RA excess impairs enamel formation. Dental mesenchyme (green) and oral epithelium (red) have reciprocal interactions during tooth development (Balic and Thesleff, 2015) (A,B). (A) Normally the entire developing tooth is shielded from RA by actions of Cyp26 RA-degrading enzymes. This allows proper expression of Runx2/3, Dlx3/5, and Bmp2/3 in the mesenchyme condensing next to the oral epithelium at the bud stage (B). These targets are reduced by excess RA. Enam, Ambn, and Amelx exhibit strong reductions at the late bell stage, impairing enamel crystallization (C). The binding site motifs of Runx2/3, Dlx3/5, and Smad2/3/4 were obtained from the JASPAR database (http://jaspar.genereg.net) and used to find potential binding sites in conserved regions using the rVISTA database (http://genome.lbl.gov/vista/index.shtml).

peptidase 4, essential for removing enamel proteins and biomineralization (OMIM: 603767), the peroxisome proliferatoractivated receptor (PPAR) alpha, required to achieve normal enamel mineralization (OMIM: 170998) (Sehic et al., 2009), and FAM20C, whose mutation produces severe enamel defects in human and mouse (Wang et al., 2012; Acevedo et al., 2015).

# RA Excess Alters the Osteoblast, Odontoblast, and Ameloblast Lineages

Over 80 years ago, severe nutritional vitamin A deficiency was reported to reduce enamel formation (Mellanby, 1941), but since this time no genetic models of RA deficiency have been reported with ameloblast or other primary tooth defects. No defects are observed in mouse mutants for the RAsynthesizing enzymes Raldh2 and Raldh3 (Dupé et al., 2003, our unpublished observations), implicating a more severe RA deficiency is required. More likely vitamin A intake levels are usually quite high, hence cellular RA levels need to be reduced during ossification. Both hypervitaminosis A and very low serum retinol levels produce skeletal fragility, poor bone health, and osteoporosis risk (Henning et al., 2015; Green et al., 2016). During mineralization, site-specific increases in CYP26 enzymes are required for bone formation (Minegishi et al., 2014). Cyp26b1−/<sup>−</sup> mutants have truncated, underossified mandibles, possibly due to RA excess perturbing neural crest migration, or alternatively to defects in osteoblast maturation. Incisor defects, while noted, were not characterized (Maclean et al., 2009). Human mutations (both null and hypomorphic) for this RA-catabolizing enzyme produce calvarial hypoplasia and craniosynostosis (Laue et al., 2011). In our experiments, Cyp26b1 is potently induced in RA-treated E16.5 incisors (Table S3). Even so, enamel defects are still observed.

The dentino-alveolar bone complex regulates tooth development. We observe rapid in vivo effects of RA reducing Runx2, and its collagenous and mineralization targets. This rapid rodent response to hypervitaminosis A (Lind et al., 2013) is similar to effects in humans (Henning et al., 2015). These RA excesses target enamel matrix protein production. Phenotypic differences in AI severity have been described between family members with identical mutations (see Prasad et al., 2016, for a recent list). Affected patients could be sensitized to otherwise benign alteration in vitamin consumption, RA catabolism pathways, or defects in the tooth and bone biosynthesis programs. Accumulating mutations might sensitize fetal development to environmental factors, including nutrition, explaining variability in AI morphogenetic phenotypes. Similar models have been proposed for RA interactions with the DiGeorge/chromosome 22q1 deletion syndrome (Maynard et al., 2013). Even physiological RA excesses, in the context of additional genetic alterations (which otherwise would produce benign changes) could have net consequences contributing to clinical variations in oral manifestations of rare diseases.

# AUTHOR CONTRIBUTIONS

Study design: SM, AB, and KN. Data collection: SM, VL, EM, BS, and KN. Data analysis: SM, VL, JH, AB, and KN. Drafting manuscript: SM, PD, AB, and KN. Revising manuscript content: SM, JH, PD, AB, and KN. Approving final version of manuscript: SM, VL, JH, EM, BS, PD, AB, and KN. KN takes responsibility for the integrity of the data analysis.

# FUNDING

This work was supported by a grant from the University Hospital of Strasbourg (API, 2009–2012, "Development of the oral cavity: from gene to clinical phenotype in Human"), the EUfunded projects (ERDF) A27 "Oro-dental manifestations of rare diseases" in the framework of the RMT-TMO Offensive Sciences initiative INTERREG IV and INTERREG V No. 1.7 RARENET, the Institut d'Etudes Avancées (Institute of Advanced Studies) de l'Université de Strasbourg (USIAS Fellows 2015), and the grant ANR-10-LABX-0030-INRT managed by the Agence Nationale de la Recherche under the frame program Investissements d'Avenir ANR-10-IDEX-0002-02. Sequencing was performed by the IGBMC Microarray and Sequencing platform, supported

# REFERENCES


by the France Genomics National Infrastructure, funded as part of the Investissements d'Avenir program (ANR-10-INB S-0009).

# ACKNOWLEDGMENTS

We wish to thank Amandine Velt, Céline Keime, Bernard Jost, Christelle Thibault-Carpentier, Doulaye Dembelé, Antonio Nanci, Greg Pratt, Rena D'Souza, Sylvain Provot, Dominique Hotton, Valérie Fraulob, Carole Haushalter, Muriel Rhinn, Anna Niewiadomska-Cimicka, and Claire Huber for gift of reagents, technical assistance, and/or knowledgeable insight.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00673/full#supplementary-material


early development. Hum. Mol. Genet. 22, 300–312. doi: 10.1093/hmg/d ds429


**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 the authors VL and AB and states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2017 Morkmued, Laugel-Haushalter, Mathieu, Schuhbaur, Hemmerlé, Dollé, Bloch-Zupan and Niederreither. 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) or licensor 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.

# Expression of Steroid Receptors in Ameloblasts during Amelogenesis in Rat Incisors

Sophia Houari 1, 2 † , Sophia Loiodice1, 2 †, Katia Jedeon1, 2, Ariane Berdal 1, 2, 3 and Sylvie Babajko1, 2 \*

<sup>1</sup> Paris Laboratory of Molecular Oral Pathophysiology, Centre de Recherche des Cordeliers, Institut National de la Santé et de la Recherche Médicale UMRS 1138, Université Paris-Descartes, Université Pierre et Marie Curie-Paris, Paris, France, <sup>2</sup> Université Paris-Diderot, Unité de Formation et de Recherche d'Odontologie, Paris, France, <sup>3</sup> Centre de Référence des maladies rares de la face et de la cavité buccale MAFACE hôpital Rothschild, AP-HP, Paris, France

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Javier Catón, CEU San Pablo University, Spain Victor E. Arana-Chavez, University of São Paulo, Brazil Supawadee Sukseree, Medical University of Vienna, Austria

\*Correspondence:

Sylvie Babajko sylvie.babajko@crc.jussieu.fr

† These authors have contributed equally to this work.

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 01 August 2016 Accepted: 13 October 2016 Published: 02 November 2016

### Citation:

Houari S, Loiodice S, Jedeon K, Berdal A and Babajko S (2016) Expression of Steroid Receptors in Ameloblasts during Amelogenesis in Rat Incisors. Front. Physiol. 7:503. doi: 10.3389/fphys.2016.00503 Endocrine disrupting chemicals (EDCs) play a part in the modern burst of diseases and interfere with the steroid hormone axis. Bisphenol A (BPA), one of the most active and widely used EDCs, affects ameloblast functions, leading to an enamel hypomineralization pattern similar to that of Molar Incisor Hypomineralization (MIH). In order to explore the molecular pathways stimulated by BPA during amelogenesis, we thoroughly investigated the receptors known to directly or indirectly mediate the effects of BPA. The expression patterns of high affinity BPA receptors (ERRγ, GPR30), of ketosteroid receptors (ERs, AR, PGR, GR, MR), of the retinoid receptor RXRα, and PPARγ were established using RT-qPCR analysis of RNAs extracted from microdissected enamel organ of adult rats. Their expression was dependent on the stage of ameloblast differentiation, except that of ERβ and PPARγ which remained undetectable. An additional large scale microarray analysis revealed three main groups of receptors according to their level of expression in maturation-stage ameloblasts. The expression level of RXRα was the highest, similar to the vitamin D receptor (VDR), whereas the others were 13 to 612-fold lower, with AR and GR being intermediate. Immunofluorescent analysis of VDR, ERα and AR confirmed their presence mainly in maturation- stage ameloblasts. These data provide further evidence that ameloblasts express a specific combination of hormonal receptors depending on their developmental stage. This study represents the first step toward understanding dental endocrinology as well as some of the effects of EDCs on the pathophysiology of amelogenesis.

Keywords: amelogenesis, steroid receptors, steroid hormones, endocrine disrupting chemicals, enamel mineralization

# INTRODUCTION

The environment has become increasingly contaminated by various pollutants which may have a role in the modern burst of diseases. Among environmental toxicants, endocrinedisrupting chemicals (EDCs) have been associated over these past 50 years with many existing or emerging diseases including hormone-dependent cancers, diabetes, obesity, and decreased fertility (De Coster and van Larebeke, 2012; Maqbool et al., 2016). This is supported by numerous epidemiological surveys (De Coster and van Larebeke, 2012; Grindler et al., 2015; Ehrlich et al., 2016) and experimental studies (Brieño-Enríquez et al., 2015; Chevalier et al., 2015; Robinson and Miller, 2015; Maqbool et al., 2016; Palanza et al., 2016; Ziv-Gal and Flaws, 2016 for recent reviews). Among the thousands of EDCs, bisphenol A (BPA) is one of the most active and ubiquitous due to its wide use by the plastic industry. The consequences of exposure to BPA have been studied in detail in the development and pathophysiology of multiple organs including gonads, brain, pancreas, liver, heart, and adipose tissue, acting on different effectors of the steroid axis (Chevalier et al., 2015; Robinson and Miller, 2015; Palanza et al., 2016; Seachrist et al., 2016; Ziv-Gal and Flaws, 2016 for recent reviews). BPA has also been shown to induce enamel hypomineralization in rats (Jedeon et al., 2013). Amelogenesis follows a well-known sequence of cell proliferation, differentiation, maturation, and death characterized by specific gene-expression patterns (Nanci, 2012). Ameloblasts sequentially secrete enamel matrix proteins (amelogenin, enamelin, ameloblastin) and proteases (KLK4 and MMP20). The proteases degrade the enamel matrix allowing subsequent mineral crystal growth under the correct pH and ionic conditions [aided by several solute carriers (SLCs) and ionhandling proteins]. BPA modulates the expression of at least one enamel key gene at each stage of amelogenesis, including enamelin, KLK4, and SLC26A4 (Jedeon et al., 2013, 2016a). The resulting rat enamel defects may be scored as those observed in human Molar Incisor Hypomineralization (MIH; Jedeon et al., 2013), a recently described enamel pathology (Weerheijm et al., 2001; Weerheijm and Merjare, 2003). The teeth of rats exposed to BPA and those of humans affected by MIH share similar structural and biochemical abnormalities. Thus, exposure of rats to BPA is a good experimental model of MIH (Jedeon et al., 2013). MIH mostly affects permanent first molars and incisors which are the first teeth to mineralize, from the third trimester of fetal life to four-5 years after birth (Weerheijm et al., 2001), corresponding to the window of the highest susceptibility to EDCs. This enamel disease presents a similar epidemiological evolution to EDCrelated diseases. It was almost non-existent before the 80s', but now affects ∼15–18% of 6 to 9-year-old children (Jälevik, 2010; Jedeon et al., 2015). It may therefore constitute a marker of exposure to pollutants that disrupt amelogenesis. The mechanism of action of BPA is still unclear but seem to modulate directly or indirectly the activity of multiple receptors (Acconcia et al., 2015). Among them, BPA has been shown to bind the estrogen receptors (ERα and ERβ) (Delfosse et al., 2012), GPR30 (or GPER) (Pupo et al., 2012) and ERRγ with a high affinity (Liu et al., 2012). It also directly or indirectly interferes with the activity of the androgen receptor (AR), the progesterone receptor (PGR), the glucocorticoid receptor (GR), and the PPARγ (Acconcia et al., 2015; Rehan et al., 2015). The mechanism of action of BPA in dental cells is even less evident as its putative receptors are poorly defined in dental tissues, except for ERα (Jedeon et al., 2014a).

The aim of this study was to systematically investigate the expression pattern of the putative BPA receptors and members of their family during amelogenesis in order to understand the effects of BPA on enamel as well as those of other EDCs acting through these receptors. These data may thus help to decipher the physiological endocrine-mediated regulations of amelogenesis and enamel pathologies resulting from endocrine disruption. To date, only the vitamin D pathway has been investigated in dental cells (Berdal et al., 1993; Descroix et al., 2010; Woo et al., 2015). Dental endocrinology needs to be explored in depth to understand the pathways of hormones effects on dental growth and enamel quality.

# MATERIALS AND METHODS

# Animals and Biological Samples

Two month-old Wistar rats were purchased from Janvier France Sarl (Le Genest Saint Isle, France) and bred in our animal house. All animals were fed ad libitum, and maintained in accordance with the guidelines for the care and use of laboratory animals from the French Ministry of Agriculture (A-75-06-12).

Three groups of three 30 day-old male and three other similar groups of female rats were constituted and used in this study. Rats were anesthetized by isoflurane inhalation, killed, and their mandibles immediately dissected. The incisors were extracted and soft dental tissues microdissected as previously described (Jedeon et al., 2013). Briefly, dental epithelial cells from the secretion stage and the maturation stage were separately dissected using the molar reference line for isolation, removing the underlying 2 mm-tissue corresponding to the transition stage (Smith and Nanci, 1989). The incisor wasn't opened during enamel organ dissection thus avoiding contamination by the mesenchyme. The anatomically distinguishable cervical loop was dissected from the apical end of the incisor. Microdissection quality was validated by RT-PCR using Enamelin primers for the secretion stage, and KLK4 or SLC26A4 for the maturation stage; Jedeon et al., 2016a). The absence of contamination by the mesenchyme and bone was verified using osteocalcin primers.

# RNA Extraction and Gene Expression Profiling

RNAs were extracted from microdissected cervical loop, and secretion- and maturation-stage cells of rat enamel organ using the RNeasy <sup>R</sup> Protect Mini Kit (Qiagen-France) according to the manufacturer's procedure. Spectrophotometry was used to assess the concentration and purity of RNA by measuring absorbance at 260 nm with a NanoDrop 1000 and RNA Integrity Number (RIN) (threshold > 9.5) with an Agilent Bioanalyzer, respectively. Reverse transcription was carried out with 1 µg total RNA for 50 min at 42◦C, using a random primer oligodT primer mix, according to the manufacturer's instructions (Superscript II <sup>R</sup> —Invitrogen). Real-time quantitative PCR was performed using the CFX96 device (Bio-Rad Laboratories, Hercules, CA, United States). SYBER green fluorescence corresponding to neosynthesized amplicons was quantified at the end of each of the 45 PCR cycles corresponding to a denaturation step of 2 s at 95◦C followed by a polymerization step of 30 s at 60◦C. Each PCR was independently repeated in triplicate and the results normalized against those for the three selected reference genes, RS15, GAPDH, and TBP1, for which the expression did not vary under our experimental conditions. Details of the primers

and the corresponding amplicon sizes are presented in **Table 1**. The standard curve method was used to calculate the values corresponding to the relative amounts of test and reference RNAs. Mean ratios of test RNA/standard RNA were calculated for each sample. Similar data were obtained using the 11Ct method.

RNAs extracted from microdissected maturation-stage cells of male rat enamel organ were used for microarray experiments performed with Affymetrix RatGene1.0 ST chip probes at the Genom'IC platform of Cochin Institute (Paris, France) to measure the relative level of each (steroid) receptor.

# Immunofluorescence Assays

Dental tissues were fixed by immersion in a 4% paraformaldehyde solution for 4 h. After washing in PBS, the samples were dehydrated in ethanol, rinsed in clearene (Leica-France) and paraffin-embedded (Paraplast plus, Sigma). Serial 8µm sections were cut using a microtome (RM 2145, Leica, France). Sections were deparaffinized and rehydrated in decreasing concentrations of ethanol. Slices were microwaved for 20 min, and the tissues permeabilized with 0.5% Triton



X-100 for 10 min. Sections were then washed in PBS and blocked with 10% normal goat serum in PBS for 1 h at room temperature. Slices were incubated overnight at 4◦C with primary rabbit polyclonal anti-AR (N-20:sc-816, Santa Cruz) (1:200), anti-VDR (ab3508, Abcam) (1:500), or anti-ERα (sc-542, Santa Cruz) (1:50) antibodies. Sections were incubated with secondary goat anti-IgG coupled to Alexa Fluor 594 antibody (A-11072, Life Technologies) (1:500) at room temperature for one h in the dark. After rinsing with PBS, sections were immersed in DAPI (010M4003-Sigma) (1:100000) for 5 min and finally mounted with Fluoromount (Southern Biotech, Clinisciences).

# Statistical Analysis

RT-qPCR data resulting from three independent analyses of three RNA samples of each tissue (loop, secretion, maturation, mesenchyme, and other tissues used as references) are presented as means ± SD. and were analyzed with GraphPad Prism Software Version 5.0 (GraphPad Software Inc., La Jolla, CA) using One way Analysis of Variance followed by Bonferroni's correction. Compared values were considered to be significantly different when <sup>∗</sup>p < 0.05, ∗∗p < 0.01, or ∗∗∗p < 0.001.

# RESULTS

# Expression Patterns of BPA Putative Receptors during Amelogenesis

We determined the specific pattern of expression for each highaffinity BPA receptor ERRγ, GPR30, ERα, and ERβ, and the other members of the ERR family, ERRα and ERRβ during amelogenesis by qPCR analysis of the enamel organ RNAs (**Figure 1A**).

Rat enamel organ cells expressed all the tested receptors except the ERβ, which was undetectable at all stages of amelogenesis (**Figure 1A**). The BPA receptors ERRγ, and to a lesser extent GPR30, were primarily expressed in early-stage ameloblasts (secretory and pre-ameloblasts). ERRγ expression was 5.0 to 6.7-fold higher in the cervical loop containing the precursors than in secretion and maturation stages containing differentiated ameloblasts. The other two members of the ERR family, the ERRα and ERRβ, were expressed throughout amelogenesis with a 3.6 and 1.3-fold accumulation in the maturation stage ameloblasts, respectively. The ERα presented a variable profile depending on the animal. Some animals expressed the ERα essentially in the cervical loop, whereas it was mostly in the maturation-stage ameloblasts in others.

Both males and females expressed similar levels of all receptors measured.

# Expression Pattern of Additional Steroid Receptors, GR, AR, MR, PGR, VDR, and Retinoid Receptors during Amelogenesis

We also measured the expression of all receptors known to be involved in the action of BPA, including the AR, PGR and GR/MR (**Figure 1B**). The AR exhibited the highest difference of expression which was 7.3-fold higher in maturation-stage

using the molar reference line for isolation (See Materials and Methods). The cervical loop (L) that contains dental precursor cells, was anatomically distinguishable. The highest expression level ratio calculated for each studied and reference gene, using the standard curve method was set to 100% to compare data from the three independent experiments. Males (black bars) and females (white bars) were treated separately. The compared values were considered to be significantly different when \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001. (A) BPA receptors, ERRγ, and to a lesser extent GPR30 and ERα, were mainly expressed in the cervical loop, whereas ERRα and ERRβ were mostly expressed in the maturation stage. ERα and ERRβ expression pattern varied considerably between samples. ERβ was undetectable. (B) The other receptors able to mediate the effects of BPA were also expressed in the rat enamel organ, especially during the maturation stage. VDR and RXRα, two key receptors in amelogenesis, were also mostly expressed during the maturation stage.

than in early-stage ameloblasts. AR mRNA was mostly detected in maturation-stage epithelium where its level of expression was 3.6- and 5.7-fold higher than in the mesenchyme and in testis, respectively (**Figure 2A**). Immunofluorescence assays also showed the presence of the AR protein in dental epithelium, exclusively in maturation-stage ameloblasts, but not in secretion-stage ameloblasts, nor in cells of the papillary layer (**Figure 2B**). Among the different receptors investigated, its localization was the most specific, restricted to maturation-stage ameloblasts.

FIGURE 2 | Specificity of steroid hormone and VD receptor expression in maturation-stage ameloblasts. (A) Expression levels calculated by the 11Ct method were compared between the cervical loop (L), secretion-stage cells (S), maturation-stage cells (M), mesenchymal cells (Mes) and other tissues used as references: testis for AR, kidney for MR, and ovary for PGR. The AR showed the most preferential expression in maturation-stage enamel tissue relative to all the other receptors tested with a level of expression even higher than that found in testis, used as the androgen responsive tissue. Results are from three independent analyses of three RNA samples of each tissue and are presented as the means ± SD. (B) Immunofluorescent assays for the AR, ERα, and VDR, three receptors involved in amelogenesis. The ER signal was very low in all cells of the enamel organ. The signals corresponding to the AR and VDR were clearly localized in maturation-stage ameloblasts (involved in enamel terminal mineralization). The AR and VDR were also slightly detected in the secretion-stage. A, ameloblasts; PL, papillary layer; SI, stratum intermedium. Scale bars, 10 µm.

The other receptors were mostly expressed in maturationstage ameloblasts with a 2.6-fold higher level of the GR than in the cervical loop (**Figure 1B**). The MR and PGR were also mostly expressed during the maturation stage, but with only small differences relative to other stages. The level of MR expression in the maturation-stage cells was 11.2-fold lower than in the kidney, and the level of PGR 8.1-fold lower than in the ovary used as positive controls (**Figure 2A**).

We also examined the expression patterns of the VDR and its partner the RXRα. Both VDR and RXRα mRNAs accumulated in the maturation-stage ameloblasts with a mean two-fold higher level than in the other compartments of enamel organ (**Figure 1B**). Immunohistological assays, showing the localization of the corresponding proteins, confirmed the RTqPCR data with a signal for the VDR throughout the enamel organ, but stronger in mature ameloblasts (**Figure 2B**).

We observed no major differences between males and females (**Figure 1B**).

# Comparison of Relative Expression Levels of Steroid, BPA, Retinoid, and Vitamin D Receptors In Maturation Stage Ameloblasts

We determined the relative expression levels of the studied receptors in maturation-stage ameloblasts by microarray analysis. The most highly expressed receptors were RXRα, RARα, and VDR (**Figures 3A,B**). Maturation-stage ameloblasts also expressed all members of the ketosteroid receptors, GR, MR, AR, and PGR. GR and AR levels of expression were similar whereas MR and PGR were 5.9- and 7.9-fold lower, respectively.

The other receptors (ERα, ERRβ, ERRγ) were weakly expressed in maturation-stage ameloblasts: ERRγ was one of the least expressed, with mRNA level that was 27.7 fold less than the VDR (**Figure 3**). The ERβ, GPR30, and PPARγ, three other putative BPA receptors were almost undetectable.

# DISCUSSION

The effects of vitamin D (VD) on bone and enamel mineralization are well-known, but little is known about all other endocrine regulations of dental growth and mineralization. Recent reports showing the effects of EDCs on enamel mineralization (Bloch-Zupan et al., 1994; Alaluusua et al., 2004; Jan et al., 2007; Jedeon et al., 2014b) suggest that amelogenesis may be regulated by endogenous steroid hormones. The present study shows that many steroid receptors are expressed by ameloblasts with a specific pattern depending on cell proliferation and differentiation, making ameloblasts responsive cells to steroid hormones. The VDR, which binds VD and forms active heterodimers with the RXRα, was the most highly expressed nuclear receptor along with the RXRα throughout amelogenesis. This is in accordance with previous data showing


FIGURE 3 | Relative level of expression of steroid hormone and vitamin receptors during the maturation stage. (A) The relative level of expression of each mRNA was determined by microarray analysis of RNAs extracted from maturation-stage enamel organ. Three main groups of receptors were distinguished: The RXRα and VDR were the most highly expressed (black bars); the GPR30, MR, PGR, ERα, ERRβ, and ERRγ weakly (white bars); and the RARα, GR, AR, and ERRα expression levels were intermediate (gray bars). PPARγ, GPR30 and ERβ were at the limit of the detection. Data resulted from microarray analyses of four RNA samples were presented as means ± SD and were compared using One way Analysis of Variance followed by Bonferroni's correction. The compared values were considered to be significantly different when \*p < 0.05, \*\*\*p < 0.001. ns, non significant. (B) Raw microarray data and statistical analysis for the calculated mean levels of expression of the studied receptors.

the presence of VDR (Berdal et al., 1993) and RARα/RXRα (Bloch-Zupan et al., 1994) in enamel organ cells, reflecting the importance of VD and vitamin A/retinol in tooth development reported many years ago. VDR/RXR heterodimers control ameloblast differentiation and the expression of key enamel genes such as amelogenin and calbindin D 28k (Berdal et al., 1993; Papagerakis et al., 1999). They were also the most highly expressed nuclear receptors in mesenchymal cells, including odontoblasts, in accordance with previously published data showing the effects of VD on dentin (Davideau et al., 1996). We also detected the GR, AR, and ERRα, among the most highly expressed steroid hormone receptors, throughout amelogenesis with the highest level of expression in maturationstage ameloblasts. The role of ERRα, and more generally of ERRs, in amelogenesis is unknown. Corticoids affect enamel hardness and mineralization (Pawlicki et al., 1992), and a responsive element for GR (GRE) has been found in the amelogenin promoter (Gibson et al., 1997). Concerning the AR, it has already been detected in dental pulp cells (Dale et al., 2002; Inaba et al., 2013). In addition, our past work, as well as the present study, show that AR expression in the maturation-stage ameloblasts which is higher than in the secretion-stage and mesenchymal cells, suggesting a selective role of androgens in enamel final mineralization (Jedeon et al., 2016b). Testosterone is able to modulate the expression of enamel key genes present in maturation-stage ameloblasts such as SLC26A4 (or pendrin) and KLK4 (Jedeon et al., 2016b). Moreover, the higher level of AR expression in dental epithelium than in testis suggests that ameloblasts are responsive to plasmatic testosterone and thus androgen regulation of final enamel mineralization. This is likely not the case for the PGR and MR which levels of expression in enamel organ were 10 to 20-fold lower than in ovary and kidney, respectively.

The generally higher expression of steroid hormone receptors in the maturation-stage ameloblasts suggests a hormonal control of final enamel mineralization, and thus of enamel quality rather than enamel quantity. This has been experimentally demonstrated in rodent models for the VD/VDR. The deletion of the VDR leads to enamel hypomineralization even in the presence of normal levels of calcium and phosphate (Descroix et al., 2010). Indeed, low serum levels of VD during infancy is associated to caries (Schroth et al., 2014). Dental decay is a complex process involving many factors such as saliva, oral microbiota, and lifestyle, but enamel quality is also an important parameter. Elevated VD serum levels are negatively correlated to MIH (Kühnisch et al., 2015) and to EDC contamination (Johns et al., 2016), suggesting that MIH may be due, at least in part, to endocrine disruption. Epidemiological data have shown that contamination by PCBs and dioxin, two different classes of EDCs, may be associated with enamel hypomineralization (Alaluusua et al., 2004; Jan et al., 2007). Our previous experimental data showed that rats exposed to low-dose genistein and vinclozolin, as well as BPA, present enamel hypomineralization similar to human MIH (Jedeon et al., 2013, 2014b), which is both a hypomineralizing and hypoplasic enamel pathology (Jedeon et al., 2013). The selective affection of MIH suggests disruption during a narrow time window compatible with the steroid hormone secretion pattern during enamel mineralization. The clinical characteristics of enamel defects in MIH also suggest that BPA disrupts amelogenesis throughout the process. It may directly or indirectly modulate receptor activities, not only in maturation-stage ameloblasts, but also in pre-secretory and proliferating cells of the cervical loop. The ERα has already been shown to mediate, at least in part, the short-term mitogenic effects of BPA in pre-ameloblastic cells, but not genomic effects (Jedeon et al., 2014a). Similar non-genomic effects of BPA involving GPR30 activation has been shown in prostate cancer cells (Prins et al., 2014). The three high affinity BPA receptors, ERRγ, GPR30, and ERs, were very weakly expressed in the maturation-stage ameloblasts. They are mainly detected in proliferating epithelial and mesenchymal precursor cells of the loop, especially the ERRγ, which is the highest affinity receptor for BPA (Okada et al., 2008; Acconcia et al., 2015). The ERRγ is the in vivo receptor of BPA involved in the mineralization process of otoliths in zebrafish (Tohmé et al., 2014).

Despite the preferential impact of BPA in males, we detected no major differences between males and females in the hormone receptor expression patterns, or their expression levels. One possible explanation is that this sexual discrepancy may be due to disrupted levels of estrogens or androgens (Scinicariello and Buser, 2016). BPA exerts its anti-androgenic effects by preventing AR activation and lowering the levels of endogenous testosterone. BPA and other anti-androgenic EDCs may exert their anti-androgenic effects on final enamel mineralization through the AR expressed in maturationstage ameloblasts (Jedeon et al., 2016b). Thus, the high testosterone levels in males following the birth, concomitant with amelogenesis, may cause a sexual dimorphism in enamel quality.

In conclusion, our data show that dental cells express many steroid receptors, of which the expression pattern depends on their stage of differentiation. This study provides clues for further studies of dental endocrinology which needs to be developed in depth to understand the effects of steroid hormone receptors and EDCs acting through such receptors on dental growth and enamel quality.

# AUTHOR CONTRIBUTIONS

SH, SL, and KJ: Contribution to the acquisition, analysis and interpretation of data, interpretation of data for the work, drafting the work, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. AB and SB: Substantial contributions to the conception and design of the work, interpretation of data for the work, drafting the work, writing the paper, final approval of the version to be published, and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

# FUNDING

This work was funded by the University Paris-Diderot, the French National Institute of Health and Medical Research (INSERM), the Institut Benjamin Delessert, and the Institut Français pour la Recherche Odontologique (IFRO).

# ACKNOWLEDGMENTS

We are grateful to Georges Zadigue for animal breeding. We thank Manon Le Normand and Khaled Salhi for their experimental contribution.

# REFERENCES


**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 © 2016 Houari, Loiodice, Jedeon, Berdal and Babajko. 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) or licensor 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 Na:K:2Cl Cotransporter 1 (NKCC1/SLC12A2) in Dental Epithelium during Enamel Formation in Mice

Rozita Jalali 1, 2 \*, Johannes C. Lodder <sup>3</sup> , Behrouz Zandieh-Doulabi <sup>1</sup> , Dimitra Micha<sup>4</sup> , James E. Melvin<sup>5</sup> , Marcelo A. Catalan5, 6, Huibert D. Mansvelder <sup>3</sup> , Pamela DenBesten<sup>7</sup> and Antonius Bronckers 1†

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Claudio Cantù, University of Zurich, Switzerland Supawadee Sukseree, Medical University of Vienna, Austria Mina Mina, University of Connecticut Health Center, United States

\*Correspondence:

Rozita Jalali r.jalali@acta.nl † Retired.

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 14 July 2017 Accepted: 31 October 2017 Published: 21 November 2017

### Citation:

Jalali R, Lodder JC, Zandieh-Doulabi B, Micha D, Melvin JE, Catalan MA, Mansvelder HD, DenBesten P and Bronckers A (2017) The Role of Na:K:2Cl Cotransporter 1 (NKCC1/SLC12A2) in Dental Epithelium during Enamel Formation in Mice. Front. Physiol. 8:924. doi: 10.3389/fphys.2017.00924

<sup>1</sup> Department of Oral Cell Biology, Academic Centre for Dentistry Amsterdam (ACTA), Amsterdam Movement Sciences, University of Amsterdam, VU University Amsterdam, Amsterdam, Netherlands, <sup>2</sup> Department of Functional Anatomy, Academic Centre for Dentistry Amsterdam (ACTA), MOVE Research Institute Amsterdam, University of Amsterdam, VU University Amsterdam, Amsterdam, Netherlands, <sup>3</sup> Department Integrative Neurophysiology, Center for Neurogenomics and Cognitive Research, VU University, Amsterdam, Netherlands, <sup>4</sup> Department of Clinical Genetics, VU University Medical Center, Amsterdam Movement Sciences, Netherlands, <sup>5</sup> Secretory Mechanisms and Dysfunction Section, NIDCR/NIH, Bethesda, MD, United States, <sup>6</sup> Departamento de Ciencias Químicas y Farmaceúticas, Facultad de Ciencias de la Salud, Universidad Arturo Prat, Iquique, Chile, <sup>7</sup> Department of Orofacial Sciences, School of Dentistry, University of California, San Francisco, San Francisco, CA, United States

Na+:K+:2Cl<sup>−</sup> cotransporters (NKCCs) belong to the SLC12A family of cation-coupled Cl<sup>−</sup> transporters. We investigated whether enamel-producing mouse ameloblasts express NKCCs. Transcripts for Nkcc1 were identified in the mouse dental epithelium by RT-qPCR and NKCC1 protein was immunolocalized in outer enamel epithelium and in the papillary layer but not the ameloblast layer. In incisors of Nkcc1-null mice late maturation ameloblasts were disorganized, shorter and the mineral density of the enamel was reduced by 10% compared to wild-type controls. Protein levels of gap junction protein connexin 43, Na+-dependent bicarbonate cotransporter e1 (NBCe1), and the Cl−-dependent bicarbonate exchangers SLC26A3 and SLC26A6 were upregulated in Nkcc1-null enamel organs while the level of NCKX4/SLC24A4, the major K+, Na<sup>+</sup> dependent Ca2<sup>+</sup> transporter in maturation ameloblasts, was slightly downregulated. Whole-cell voltage clamp studies on rat ameloblast-like HAT-7 cells indicated that bumetanide increased ion-channel activity conducting outward currents. Bumetanide also reduced cell volume of HAT-7 cells. We concluded that non-ameloblast dental epithelium expresses NKCC1 to regulate cell volume in enamel organ and provide ameloblasts with Na+, K<sup>+</sup> and Cl<sup>−</sup> ions required for the transport of mineral- and bicarbonate-ions into enamel. Absence of functional Nkcc1 likely is compensated by other types of ion channels and ion transporters. The increased amount of Cx43 in enamel organ cells in Nkcc1-null mice suggests that these cells display a higher number of gap junctions to increase intercellular communication.

Keywords: mineralization, ion transport, pH regulation, SLC26A, gap junctions

# INTRODUCTION

Ion transport by ameloblasts is critical for the formation of fully mineralized dental enamel. Disturbance in transport of mineral and/or bicarbonate ions (either local and/or systemic) during enamel development can lead to permanent enamel abnormalities (Lacruz et al., 2010a, 2012).

To form and mature apatite crystals, maturation ameloblasts transport mineral ions as calcium and phosphate into the forming enamel using apical plasma membrane transporter(s) such as Na+-Ca2+exchangers (NXC1,3) (Okumura et al., 2010) and Na <sup>+</sup>-K+-Ca2<sup>+</sup> exchangers (NCKX4/SLC24A family) (Hu et al., 2012; Parry et al., 2013; Bronckers et al., 2015a). However, formation of each unit cell of the hydroxyapatite crystal also releases ∼8 protons (Ryu et al., 1998) that need to be buffered to prevent acidification and arrest of mineral accretion. In order to buffer protons, ameloblasts have been proposed to secrete bicarbonate ions into the enamel space that neutralize H<sup>+</sup> (Smith, 1998). Bicarbonate is taken up by ameloblasts basolaterally from the extracellular fluid by bicarbonate-transporting proteins such as electrogenic (NBCe) and electroneutral (NBCn) Na+-dependent bicarbonate cotransporters expressed in the ameloblasts during secretionand maturation phase of amelogenesis (Josephsen et al., 2010; Lacruz et al., 2013; Jalali et al., 2014; Bori et al., 2016). During maturation phase ameloblasts also generate bicarbonates by producing carbonic acid (H2CO3) from CO<sup>2</sup> and H2O via cytosolic carbonic anhydrase 2 (Car2) (Bori et al., 2016). During the maturation phase the bicarbonate ions are secreted by ameloblasts into the enamel space by coordinated activity of basolateral AE2, and apically by CFTR and members of the SLC26A family: SLC26A1, SLC26A3/Dra, SLC26A4/Pendrin, SLC26A6/Pat1 and SLC26A7 (Lyaruu et al., 2008; Lacruz et al., 2010b, 2012, 2013; Bronckers et al., 2011, 2015a; Jalali et al., 2015; Yin et al., 2015).

Recent studies suggested that non-ameloblast epithelial cells connected to the basal side of the ameloblasts are also involved in transepithelial transport, based on the expression of NBCe1 (Josephsen et al., 2010; Lacruz et al., 2010b; Jalali et al., 2014) and Na+-K+-ATPase and ATPase activity in papillary layer (Garant and Sasaki, 1986; Glynn, 2002; Josephsen et al., 2010). Accumulation of Na<sup>+</sup> and K<sup>+</sup> and reduction of Cl<sup>−</sup> in hypomineralizing enamel of Cftr- null and Ae2- null mice suggested that in wild type mice Na<sup>+</sup> and K<sup>+</sup> are removed but that Cl<sup>−</sup> is taken up from the forming maturation enamel (Bronckers et al., 2015a). It was speculated that maturation ameloblasts remove Na<sup>+</sup> and K<sup>+</sup> from the enamel space by a yet unknown mechanism which is impaired in Cftr-null and Ae2-null mice. Potential transporters for potassium and sodium reabsorption in ameloblasts may correspond to Na+:K+:2Cl<sup>−</sup> (NKCC) cotransporters, which are sensitive to loop diuretics such as bumetanide and furosemide (Gagnon et al., 1998).

Na+-K+-Cl<sup>−</sup> cotransporters (NKCCs) aid in the active transport of sodium, potassium and chloride in or out of the cells (Haas, 1994). The SLC12A family contains two Na+:K+:2Cl<sup>−</sup> cotransporters, NKCC1 and NKCC2, encoded by two different genes (SLC12A2 and SLC12A1, respectively). NKCC1 is widely distributed throughout the body. In salivary glands, basolateral NKCC1 mediates the transport of sodium, potassium and chloride from blood into the acinar cells. Lack of functional NKCC1 results in dramatic reduction of the volume of secreted saliva (Evans et al., 2000). NKCC1 is also necessary for establishing the potassium-rich endolymph that bathes part of the cochlea, the organ necessary for hearing. Inhibition of NKCC1, as with furosemide or bumetamide, can result in deafness (Delpire et al., 1999). NKCC2 has a more restricted distribution and is specifically found in the apical membrane of cells in the thick ascending limb of the loop of Henle and the macula densa in nephrons where it serves both in sodium absorption and tubuloglomerular feedback (Lytle et al., 1995). Expression of Slc12a2/Nkcc1 messenger RNA was reported in mouse enamel epithelium in bud to bell stage teeth (embryonic stage E15–postnatal day 3), suggesting a possible involvement of NKCC1 in enamel organ development (Hübner et al., 2001).

In this study, we tested the hypothesis that NKCC1 plays a role in the ion transport by dental epithelium during enamel formation. The enamel organs of mice and HAT-7 cells, a rat ameloblast-like cell line derived from the cervical loop of a rat incisor (Kawano et al., 2002), were analyzed for expression of NKCC1 at the protein level. The effect of the null mutation of Nkcc1 on enamel development, cellsize and enamel mineralization was studied by histology, immunohistochemistry, micro-computed tomography and Western blotting. To understand the role of NKCC1 cell volume regulation, we exposed in vitro HAT-7 cells to bumetanide and measured cell volume using the calcein-quenching method (Ye et al., 2015). The effect of bumetamide was also tested on electrophysiology of HAT-7 membranes by patch clamp.

# MATERIALS AND METHODS

# Tissues

Nkcc1-null mutant mice used for this study were generated and genotyped as previously described (Flagella et al., 1999). One hemi-maxillary incisor of each mouse was used for immunohistological studies and the other one was freeze-dried for micro-CT analysis and western blotting. For each genotypic mouse strain, at least three wild-type mice and three null mutant mice were analyzed.

All experiments were approved by the Committee for Animal Care (Vrije Universiteit Amsterdam; ACTA-12-01) and by the Animal Care and Use Committee of the National Institute of Dental and Craniofacial Research, National Institutes of Health (ASP 13–686). The methods were carried out in accordance with the approved guidelines.

# Cell Culture

HAT-7 cells were grown in DMEM/F12 (Sigma-Aldrich, St. Louis, MO, USA) with 10% HyClone fetal bovine serum, 100 U/mL of penicillin, 10µg/mL of streptomycin (Sigma-Aldrich) and 10−<sup>5</sup> mM dexamethasone in humidified atmosphere containing 5% CO<sup>2</sup> at 37◦C (Bori et al., 2016).

ameloblasts/enamel organ; pulpa, pulp; tong: tongue; stom, stomac; m3calv, MC3T3 mouse calvarial cell line; intes, intestine; calv, calvaria.

# Histology

Mouse jaws were fixed by immersion in 5% paraformaldehyde in 0.1 M phosphate buffer pH 7.3 and embedded in paraffin. Calcified tissues from mice older than 2 weeks were first decalcified in 4% EDTA, pH 7.3 for 2–3 weeks at 4◦C. Salivary glands were fixed in Bouin's fixative (75 ml of saturated picric acid, 25 ml of 40% formaldehyde, and 5 ml of glacial acetic acid) at room temperature for 48 h and processed into 5–7µm thick paraffin sections. Dewaxed sections were stained with 1% hematoxylin (1 min) and eosin (5 min) (HE) or used for immunohistochemicals staining.

# Quantifications in Cells of the Enamel Organ

At 40x objective pictures of secretory and maturation stage of enamel organ were made and images selected based on the anatomical position in the tooth. The length of the long axis of secretory and maturation ameloblasts was measured in sagittal sections using imaging software (image J).

# Real Time Quantitative Polymerase Chain Reaction (RT-qPCR)

Total RNA was extracted from HAT-7 cells and various fresh mouse tissues using the NucleoSpin RNA/protein kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. First strand cDNA synthesis was performed in a 20µl reverse transcription reaction containing 200 ng of total RNA using VILO kit (Invitrogen) according to the manufacturer's instructions. Real-time PCR analysis was performed to analyze expression of Slc12a2 (Nkcc1) with the primer sequences (FW:5' GAAGAAAGTACTCCAACCAGAGATG 3'; REV: 5' CTGAAGTAGACAATCCTGTGATA 3'; size: 232 bps) and the housekeeping protein tyrosine 3-monooxygenase (Ywhaz) with sequences (FW:5'GATGAAGCCATTGCTGAACTTG3'; REV:5'CTATTTGTGGGACAGCATGGA3'; size:229 bps) shown by using the LightCycler 480 system based on SYBR Green I dye (Roche Applied Science, Indianapolis, IN, USA). The LightCycler reactions were prepared in 20µl total volume with 7µl PCR-H2O, 0.5µl forward primer (0.2µM), 0.5µl reverse primer (0.2µM), 10µl LightCycler Mastermix (LightCycler 480 SYBR Green I Master; Roche Applied Science, IN, USA), to which 2µl of 5 times diluted cDNA was added as PCR template. Controls in the real-time RT-PCR reaction included RT reactions without the reverse transcriptase (control for DNA carry over) and RT reactions without template (control for reagent contamination). With the Light Cycler software, the crossing points were assessed based on a standard curve of five serial dilutions ranging from 10 ng to 1.6 pg of cDNA. PCR efficiency (E) was automatically calculated using the fit point method (E = 10−<sup>1</sup> /slope). Gene expression data were used only if the PCR efficiency was within a 1.85–2.0 range. For each gene the amount of measured DNA was normalized to that of YWHAZ housekeeping gene to calculate relative gene expression. The relative gene expression in different tissues was normalized to kidney levels for each gene in the graphs.

# Immunohistochemistry

Dewaxed paraffin sections were rinsed in phosphate buffered saline (PBS) and subjected to antigen retrieval in 10 mM citrate buffer (pH 6.0) either at 60◦C overnight or for 20 min in microwave at 95◦C. Endogenous peroxidase was blocked with a peroxidase block solution (Envision kit, Dakocytomation) for 5 min. Sections were washed 3x in tris-buffered saline (TBS). Non-specific staining was blocked for 30 min with 2% BSA after which sections were incubated overnight at 4◦C with primary antibodies. These were (1) goat anti-NKCC1 (Santa Cruz, affinity purified, catalog number SC-21545), raised against the N-terminal end of human NKCC1. (2) Mouse anti-NCKX4 monoclonal antibodies (IgG2b isotype) from NeuroMab

FIGURE 2 | Immunostaining of NKCC1 protein in HAT-7 cells, developing mouse dental epithelium and mouse salivary glands. NKCC1 expression as green punctuate grains near the plasma membranes of cultured HAT-7 cells stained with anti-NKCC1 (A) or with non-immune IgG (B, control) and visualized using Alexa488-coupled secondary antibody. Panels (C,D) show salivary glands of a wild type (C) and Nkcc1 null mouse (G) stained with anti-NKCC1. NKCC1 staining in mouse dental epithelium at secretory stage (F) and in maturation stage (E,G) in papillary layer. Absence of staining with anti-NKCC1 in salivary gland (D) and incisor (H) of a Nkcc1-null mouse confirms the specificity of primary antibody. Note: all the stainings have been tested in triplicate in three mice. E, enamel; ES, enamel space; SA, ameloblasts secretory stage; MA, ameloblasts maturation stage; P, pulp; PL, papillary layer; SI, stratum intermedium.

(UC Davis/NIH NeuroMab Facility, catalog # N414/25). (3) Matched non-immune IgG (1:200–1:300) or normal serum (same concentration as primary antibodies) served as controls. After overnight incubation at 4◦C with primary antibodies, sections were washed three times in TBS and incubated with rabbit anti-goat secondary antibody conjugated to peroxidase (Thermo Scientific) for 1 h at room temperature. After washing staining was visualized using DAB (EnVision kit), counterstained with hematoxylin. For immunofluorescent staining, goat anti mouse–IgG conjugated to Alexa Fluor 488 (5µg/mL; Invitrogen) was used and counterstained with propidium iodine (Vector Laboratories, Burlingame, CA, USA). Immunohistochemistry images were acquired with a Leica EL6000 or Axio Zoom V16 microscope.

# Microcomputed Tomography (microCT)

To determine the degree of mineral content, hemi-maxillae were scanned at a resolution of 8µm voxels in a microCT-40 high-resolution scanner (Scanco Medical, AG, Bassersdorf, Switzerland) to measure mineral density in enamel. An internal standard made of solid-sintered apatite (5-mm diameter, 1.5– 2.0 mm thick, solid sintered) with density of 2.9 ± 0.2 g/mL (a gift from Himed; http://www.himed.com) was used as high-density standard. Beginning at the apical part of the incisor and moving toward the tip, cross-sectioned images through the incisors were collected at sequential intervals of 300µm in maturation-stage and 60µm in secretory-stage enamel. In each slice, the mineral density of enamel was measured halfway through the enamel layer at three sites within a circular area, with a diameter of 7µm

(Continued)

### FIGURE 3 | Continued

for enamel mineralization. Nckx4 expression (arrows) in WT (A) and Nkcc1-null maturation ameloblast (B). Total protein was extracted from WT and Nkcc1-null enamel organs (n = 3 mice) and NCKX4 expression analyzed by western blot (C). Graph bar shows semi-quantitative NCKX4 band density normalized to that of ß-actin (D). Mineral density measured by micro-CT plotted against slice numbers (H). Blue color represents WT and red Nkcc1-null upper incisor; maturation stage starts at the dotted line. In (I) the bar graphs with the same color (blue and red) represent measurements of mineral density in different stages of amelogenesis (sec, secretory; EM, early maturation; LM, late maturation). Panels (E–G) show 3D reconstruction, virtual cross section of WT and Nkcc1-null upper incisor respectively. Circles in (F) indicate sites of measurement per slice.

at the mesial, lateral, and central sides. Mean values and standard error of mean (SEM) of the mineral density were calculated and presented as mean ±SEM. Independent Student's t-test was used to compare the groups. Statistical significance was set at p < 0.05 level.

# Western Blotting

From freeze-dried upper incisors obtained from wild-type and Nkcc1−/<sup>−</sup> mice early maturation stage enamel organs were micro dissected. The apical half of the enamel organ was dissected, dissolved in non-reducing condition in SDS loading buffer (from Nucleospin Triprep kit, Macherey-Nagel, supplied by Bioke, Leiden, NL) and protein was measured using the BCA protein assay (Bio-Rad, Hercules, CA). Twenty micro gram of enamel organ denatured protein and 10µg of molecular weight markers [Novex <sup>R</sup> Sharp Pre-stained Protein Standard (# LC5800) or SeeBlue <sup>R</sup> Plus2 Pre-stained Protein Standard (#LC5925)] were subjected to electrophoresis in a 3–8% Tris acetate Nupage gel with Tris acetate running buffer for 60 min at 150 V or 4–12% Bis-Tris Protein Gels with MOPS buffer for 35 min at 200 V. and subsequently electroblotted by an iBlot device (Invitrogen) on nitrocellulose membrane according to the manufacturer's instructions. Membranes were blocked with BSA 2% for 1 h at room temperature and incubated overnight (4◦C) with the primary antibodies. Blots were washed three times in PBS and incubated with IRDye secondary antibodies (LI-COR). Visualization and quantification was carried out with the LI-COR Odyssey scanner and software (LI-COR Biosciences). Actin was detected at 680 nm wavelength (shown as red) and other primary antibodies and tubulin were detected at 800 nm wavelength (shown as green). Quantification was performed using Odyssey software. Intensity values of the bands were normalized for actin or tubulin and expressed as percentage of wild type (100%). For western blots the following primary antibodies were used: rabbit anti-SLC26A3/Dra (Research Genetics, Huntsville, AL, USA) (Jalali et al., 2015), rabbit anti-SLC26A6/Pat1 (donated by Dr. P. Aronson, Yale University, New Haven, CT, USA) (Jalali et al., 2015), rabbit anti-NBCe1 (Jalali et al., 2014) (donated by Dr. W.F. Boron, Case Western Reserve University, Cleveland, Ohio, USA), rabbit anti-NCKX4 (NeuroMab, UC Davis/NIH, # N414/25), rabbit anti-connexin (Abcam, #ab11370), mouse anti-β-actin antibody (Sigma, A2228) and rabbit anti-tubulin antibody (Abcam, ab59680). Secondary antibodies: IRDye 800CW conjugated goat anti-rabbit IgG (H+L) highly crossadsorbed (LI-COR; Product number: 926–32,211) and IRDye 680CW conjugated goat anti-mouse IgG (H+L) highly-cross adsorbed (LI-COR; Product number: 926–32,220). Dilutions: anti-β-actin and anti-tubulin (1:1,000); other primary antibodies (1:250); secondary antibodies (1:10,000).

# Imaging of Volume Decrease after Exposure to Bumetanide

HAT-7 cells were loaded with 5µM Calcein-AM (Molecular Probes) for 20 min at 37◦C. Changes in cell volume of single HAT-7 cells, plated on poly-lysine coated glass cover slips, were assessed by measuring calcein fluorescence using the calceinquenching method (Ye et al., 2015). Cells were bathed in isoosmotic solution (20–22◦C) and transferred to a continuously perfused (5 ml/min) recording chamber, equipped with a microscope with 10X objective. An image was taken every 30 s. At the start images were obtained for 5 min in the isoosmotic solution to establish the baseline. Cells were exposed to media supplemented with 10µM bumetanide for 30 min, after which the iso-osmotic solution was re-introduced and images were taken for another 10 min. The cell surface and average fluorescence of each cell in the acquired images was calculated using Image-J software. Changes in cell surface and average fluorescence were expressed as St/S<sup>0</sup> and Ft/F<sup>0</sup> ratios, respectively, where S<sup>0</sup> and F<sup>0</sup> are the average cell surface area and fluorescence under iso-osmotic treatment at the beginning of the experiment.

# Electrophysiological Recordings

Cover slips bearing HAT-7 cells were transferred to the recording chamber, containing 1.5 ml external solution. The external solution was changed at a rate of 1.5 ml/min using a gravity driven constant perfusion system. During the recordings, HAT-7 cells were perfused with standard artificial Cerebral Spinal Fluid (aCSF) containing (in mM): 126 NaCl, 3KCl, 10 Dglucose, 26 NaHCO3, 1.2 NaH2PO4, 2 CaCl<sup>2</sup> and 1 MgSO4, carboxygenated with 95% O<sup>2</sup> and 5% CO<sup>2</sup> to obtain pH 7.4 and an osmolality of 300 mOsm. For electrophysiological recordings, we used a EPC-8 amplifier and a Instrutech ICT-18 (all from Heka, D-67466 Lambrecht, Germany). Cells were identified and patched using an Olympus IX51 inverted microscope equipped with a LCAcn 40x 0.55nA ph2 objective (all Olympus corporation, Tokyo, Japan). Glass pipettes for whole-cell and cell attached recordings were made from borosilicate capillaries (OD 1.5 mm, ID 0.86 mm; Harvard Apparatus, Holliston, MA, USA) using a Sutter P-87 micro-electrode puller (Sutter instruments, USA) and displayed a resistance of 2.5–5 M. Glass microelectrodes were filled with intracellular solution containing (in mM): 110 K-Gluconate, 10 KCl, 10 HEPES, 0.4 NaGTP, 4 Mg2ATP and 10 K-Phosphocreatine (pH 7.3 adjusted with KOH, 290 mOsm). HAT-7 cells were gently

FIGURE 4 | SLC26a6, DRA, NBCe1 and CX43 proteins possibly compensate during lost Nkcc1 gene expression. Expression of SLC26a6 (A), DRA (B), NBCe1 (C), and Cx43 (D) protein extracts of enamel organs of WT and Nkcc1 null mice by western blotting. Graph bars (E–G) show semi-quantitative SLC26a6, DRA, NBCe1 bands normalized to ß-actin. The CX43 bands were normalized to tubulin (H). Table (I) presents a summary of the quantification.

lifted from the cover slip and placed in front of a piezodriven theta-barrel electrode (TGC 200; Harvard Apparatus, Holliston, MA, USA), filled with standard aCSF on one side and standard aCSF supplemented with 10µM bumetanide (B1158000 Sigma-Aldrich) on the other side. By changing the position of the barrel bumetanide was applied during Jalali et al. Nkcc1 in Mouse Enamel Organ

the whole-cell recording. Voltage ramps from −70 to 80 mV (500 ms) were applied under control and in the presence of 10µM bumetanide.

The cell attached recordings were made by filling the recording pipet with aCSF for the control recordings and with aCSF supplemented 10µM bumetanide for experimental recordings. All data was acquired using an internal 7-pole Bessel filter (5 kHz) and a sample frequency of 20 kHz. Recordings with an access resistance above 12 m were excluded form analysis.

# RESULTS

# Slc12a2 mRNA Expression in Mouse Tissues and Rat HAT-7 Cells

Transcripts for Nkcc1/Slc12a2 normalized for Ywhaz housekeeping gene were detectable in enamel organ and intestine (high), pulp and kidney moderate-(low); in the remaining tissues tested expression was very low or below detection limit (**Figure 1A**; Supplementary Figure 1). HAT-7 cells also expressed Nkcc1 transcripts (**Figure 1B**).

Bumetanide blocks activity of the NKCC's. To test whether this blocking agent also could affect Nkcc1 expression level in enamel epithelium, HAT-7 cells were exposed to various concentrations of this inhibitor for 45 min, washed, and incubated for 9 h in medium without inhibitor. Then total RNA was isolated and the amount of Nkcc1 transcripts measured by RT-qPCR (**Figure 1B**). Bumetanide did not change the number of transcripts of Nkcc1.

# NKCC1 Expression in HAT-7 Cells, Mouse Enamel Organ, and Salivary Glands

Anti-NKCC1 antibodies stained plasma membranes in HAT-7 cells as fine granular material (**Figure 2A**). Replacing primary (mouse) antibodies for normal non-immune mouse IgG failed to stain these membranes in HAT-7 cells (**Figure 2B**).

Strong immunostaining was detected in the basolateral plasma membranes of the acinar cells of salivary glands, a well-established site of NKCC1 expression (Evans et al., 2000) (**Figure 2C**).

In upper incisors, in presecretory stage and during the secretory and maturation stage, the plasma membrane of outer enamel epithelium cells was immunopositive for NKCC1 (**Figures 2E,F**; Supplementary Figure 2). No staining was seen in ameloblasts. Weaker staining was seen in dental epithelium between ameloblasts and outer enamel epithelium. Strong staining was apparent in the papillary layer (intracellular and membranes) during maturation (**Figures 2E,G**).

Sections from salivary glands (**Figure 2D**) and enamel organ from Nkcc1-null mice (**Figure 2H**) incubated with anti-NKCC1 failed to immunostain the tissues, validating the specificity of the antibodies.. In developing molar tooth germs similar stainings were obtained: a strong staining for NKCC1 in outer enamel but no staining in ameloblasts (Supplementary Figure 3).

FIGURE 6 | (A) Bumetanide 10µM increased whole-cell membrane currents across the voltage range from −70 to +80 mV. The black tracing curve c is the control and the red aCSF tracing curve b is the current in the presence of 10µM bumetanide. The green tracing curve b-c is the net current induced by 10µM bumetanide. (B) Outward currents measured at +80 mV (upper ) and inward current at −70 mV (lower). Bars show the average amplitude of control and 10µM bumetanide (n = 8).

# NKCC1 Expression is Important for Enamel Organ Function and Enamel Mineralization

Slc12a2 (or Nkcc1) knockout mice exhibit growth retardation with a 30% incidence of death by the time of weaning. They are deaf, have less body fat, reduced mean arterial blood pressure, exhibit reduced cAMP-induced short circuit currents in jejunum, cecum, and trachea, unusual head postures, and engage in circling behaviors and rapid spinning, but have difficulty maintaining their balance (Flagella et al., 1999). Incisors of Nkcc1-null mice showed no obvious gross changes. Analysis of upper incisor enamel of Nkcc1-null mice showed no significant change in mineral density in enamel during secretory stage but

a reduction at maturation stage, reaching a final density that was reduced about 10% compared to controls (P < 0.0001) (**Figures 3E–I**).

Next we examined expression of NCKX4 (sodium/calciumpotassium exchanger-4) a major transcellular calcium transporter in maturation ameloblasts, to determine whether calcium transport is affected in Nkcc1-null incisors. Nckx4 protein was strongly expressed in the apical membrane of maturation ameloblasts of wild type mice (**Figure 3A**), but its antibodyassociated staining was weaker in Nkcc1-null mice (**Figure 3B**). Western blot analysis (**Figure 3C**) of protein extracts from wild type and Nkcc1-null maturation enamel organ, showed a small (but not significant) reduction of NCKX4 protein in Nkcc1-null enamel organ (**Figure 3D**).

# Upregulation of SLC26A6, DRA, NBCe1 and Gap Junction (Cx43) Protein Expression in Nkcc1-Null Enamel Organ by Western Blotting

To examine if the expression of other Na<sup>+</sup> and Cl<sup>−</sup> ion transporters or/ and gap junction channels could compensate for the absence of NKCC1, we tested maturation stage enamel organs of Nkcc1-null mice for changes in protein levels of SLC26A6, DRA, NBCe1 and Cx43 by Western blotting (**Figures 4A–H**). Total protein extracted from Nkcc1 -null enamel organs showed a 303% increase of SLC26A6/PAT1 (P = 0.003) (**Figures 4A,E**) compared with wild type a, 143% increase for SLC26A3/DRA (P = 0.04) (**Figures 4B,F**), 47% increase for NBCe1 (P = 0.04) (**Figures 4C,G**) and 78% increase for Cx43 (P = 0.01) (**Figures 4D,H**). The results are summarized in **Figure 4I**.

# Regulation of Cell Volume Is Impaired in Nkcc1-Null Enamel Organ and HAT-7 Cells Exposed to Bumetanide

NKCC1 is one of the proteins involved in cell volume recovery after cell shrinkage in kidney (Walker et al., 2002). Analysis of histological sections of Nkcc1-null upper mouse incisors showed that late maturation ameloblasts lost their polarization and that the cells of the papillary layer were shorter compared to that seen in wild type controls. However, secretory stage ameloblasts and stratum intermedium seemed hardly or not affected in the null mutants in comparison to wild type mice (**Figures 5F,G,I,J**). To examine a possible role of NKCC1 in regulating cell volume in ameloblasts, we measured cell volume changes of single HAT-7 cells in vitro by measuring cell size (surface area measurements) and by using the calcein-quenching method before and after bumetanide (10 uM) treatment, (**Figures 5A–E**). Bumetanide decreased cell volume as measured directly (**Figure 5D**) and by loss of fluorescence (**Figure 5E**). In incisors of Nkcc1-null mice the values of the long axes of secretory ameloblasts were not different from those in wild types (**Figures 5F–H**) but were significantly shorter (∼25%; P < 0.0001; **Figures 5I–K**) in maturation ameloblasts.

# Bumetanide Increases Membrane Conductance in HAT-7 Cells

Bumetanide has been previously shown to induce potassium currents in canine kidney cells. To test whether bumetanide affects the membrane ionic conductance in HAT-7 cells, electrophysiological whole-cell voltage-clamp recordings were made from these cells. The membrane potential was ramped from −70 to +80 mV. In control conditions, the voltage ramp induced a small membrane current that reversed at −30 mV. Application of bumetanide (10µM) increased membrane currents across the entire membrane potential range (**Figure 6A**). The outward current measured at +80 mV increased about 3-fold, whereas the inward current measured at −70 mV was 20-fold enlarged (**Figures 6A,B**). The increase of membrane currents induced by bumetanide hardly reversed upon washout (not shown). To our surprise, the bumetanide-induced currents reversed at −30 mV. Given that for the recording solutions used in these experiments the equilibrium potential for potassium was −95 mV, for chloride −65 mV and for sodium and calcium above +60 mV, this suggests that most likely a mixed ionic current was activated by bumetanide-block of NKCC transporters.

# Bumetanide Increases Ion Channel Activity in HAT-7 Cells

It is not known whether the effect of bumetanide on the membrane conductance required block of NKCC transporters across the entire cell membrane, or whether this effect could also be induced locally. To determine this, we made cell-attached recordings from HAT-7 cells and applied bumetanide only to the membrane patch inside the patch pipette. The pipette potential was kept at 0 mV. In control recordings, ion channel activity and occasional discrete single channel openings were observed that were alternated by silent periods (n = 10, **Figure 7**). At this pipette potential, only outward currents were observed. During recordings in which bumetanide (10 uM) was included in the recording pipette, ion channel activity conducting outward currents increased (n = 10, **Figure 7**). No discrete amplitude levels could be distinguished, which suggests that blocking NKCC transporters with bumetanide augments the activity of multiple ion channels in the membrane patch. These results suggest that the effects of bumetanide on membrane ionic conductance could be induced locally in a patch of the cell membrane.

# DISCUSSION

The present study demonstrates that in mice enamel organ epithelium in secretory and maturation stage expresses NKCC1 at the mRNA and protein level. Without NKCC1 enamel was slightly less mineralized, at visual inspection without gross change in enamel structure. The upregulation of NBCe1, DRA and PAT1 in enamel organs of Nkcc1 null mice suggest that NKCC1 can be at least in part compensated by typical pH regulators as reported for salivary glands and duodenum epithelium in Nkcc1 null mice (Evans et al., 2000; Walker et al., 2002). This suggests that NKCC1 is involved in enamel mineralization possibly by providing amloblasts Cl<sup>−</sup> crucial for bicarbonate secretion but its role is limited. However, it cannot be excluded that some of the defects in enamel mineralization are secondary due to systemic changes brought about by changes in other tissues in Nckk1- null mice.

The reduction in mineral density of enamel in Nkcc1 null mice was relatively small (10%) in comparison with the reduction in Cftr-null mice (47%, Bronckers et al., 2015b) and in Ae2- null mice (57% Lyaruu et al., 2014), two transporters critical for pH regulation by maturation ameloblasts. The reduction in mineral density in Nkcc1 null mice could be due to: (1) a slightly increased acidification associated with reduced intracellular Cl<sup>−</sup> levels in ameloblasts resulting from less NKCC1-mediated Cl<sup>−</sup> transport in the papillary layer. The rather normal gross appearance of incisor enamel and the fact the mineral density changes appear quite late in comparison to the severe mineral reduction and porotic changes in enamel of Cftr-null and Ae2- null mice seen early in maturation make this option less likely. (2) Reduction of the cytosolic levels of K<sup>+</sup> or a reduced number of apical NCKX4 in maturation ameloblasts needed for NCKX4-mediated Ca2<sup>+</sup> transport by maturation ameloblasts. (3). Reduction in the levels of Na<sup>+</sup> in ameloblasts required for Na-Pi2b mediated phosphate transport (Bronckers et al., 2015a).

Ameloblast did not express NKCC1 but papillary layer cells highly expressed the cotransporter. To reach forming enamel ions transported by NKCC1 into the papillary layer need to pass from papillary layer into ameloblasts. The enhanced expression of connex43 points to a mechanism for such transport. Ameloblasts and papillary layer have been suggested to act as a functional unit in passing ions from outer enamel epithelium to ameloblasts through gap junctions (Josephsen et al., 2010). Gap junctions are structures passing through the plasma membranes of two adjacent cells forming small pores that enable direct intracellular transport of ions and small molecules/peptides (Toth et al., 2010). The significance of communication between enamel organ cells is exemplified in Gja1Jrt−/<sup>+</sup> mice in which enamel formation was severely affected by a decrease in the number of gap junctions in the enamel organ (Toth et al., 2010). Upregulation of connexin43 in enamel organ of Nkcc1-null mice may occur to increase the number of gap junctions to compensate for changes in ion transport due to absence of NKCC1.

Our histological and electrophysiological results suggest that NKCC1 is involved in amelogenesis. The distribution of NKCC1 in enamel organs, and the reduction in cell height of ameloblast layer in Nkcc1-null mice suggests that during amelogenesis the symporter NKCC1 serves to import Na+, K+, and Cl<sup>−</sup> from blood vessels, regulates cell volume of dental epithelium and likely water transport across the ameloblast layer. Transport of large amounts of Cl<sup>−</sup> and Na<sup>+</sup> across epithelial cells, for example by Cl<sup>−</sup> channels and cotransporters such as CFTR, NKCC2 and NKCC1, is required for fluid transport (e.g., in kidney) and maintenance or changes in cell volume. Volume changes in ameloblasts during normal amelogenesis are substantial indicated by the changes in the long axes of ameloblasts when moving from differentiation into secretion stage, reaching maximal values at full secretion stage, followed by a 50% reduction during transitional and early maturation stage. In Nkcc1-null mice the maturation ameloblasts were shorter than in wild type controls. Exposing HAT-7 cells to 10µM bumetanide caused cell volume decrease without altering the number of Nkcc1-mRNA transcripts suggesting that isoosmotically shrinkage accrued after adding bumetanide. These data are in line with a function of NKCC1 in cell volume regulation and ion transport.

In short, NKCC1 is highly expressed in non-ameloblast enamel organ epithelium. A reduction in mineral density of enamel, shortening of ameloblast cell body and upregulation of other ion transporters when NKCC1 is absent suggests that NKCC1 regulates cell volume and ion transport but can be partly compensated by enhanced activity of other ion transporters.

# AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: RJ, AB, PD, and JL; Performed the experiments: RJ, JL, BZ-D, DM, JM, and MC; Wrote the manuscript: RJ, AB; Edited the manuscript: JL, DM, JM, MC, HM, PD, and AB.

# REFERENCES


# ACKNOWLEDGMENTS

The authors thank Dr. P. Aronson (Yale University New Haven CT, USA) for providing anti-SLC26A6, Dr. W. F. Boron (Case Western, Cleveland, OH, USA) for anti-NBCE1. Ben K. A. Nelemans and Manuel Schmitz Orthopedic Surgery (VU Medical Center, MOVE Research Institute Amsterdam, the Netherlands) for their help in microscopic examination of the samples. The authors thank Dr. H. Harada (Department of Anatomy, Developmental Biology Division, Iwata Medical University Iwate, Japan) for donating HAT07 cells. This study was supported in part by the Intramural Research Program of the National Institute of Dental and Craniofacial Research (NIDCR), NIH (JEM, DE000738) and the NIDCR Veterinary Research Core (DE000740).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.00924/full#supplementary-material


**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 CC and handling Editor declared their shared affiliation.

Copyright © 2017 Jalali, Lodder, Zandieh-Doulabi, Micha, Melvin, Catalan, Mansvelder, DenBesten and Bronckers. 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) or licensor 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.

,

# Detection of a Novel DSPP Mutation by NGS in a Population Isolate in Madagascar

Agnès Bloch-Zupan1, 2, <sup>3</sup> \*, Mathilde Huckert 1, 4, Corinne Stoetzel <sup>4</sup> , Julia Meyer <sup>1</sup> , Véronique Geoffroy <sup>4</sup> , Rabisoa W. Razafindrakoto<sup>5</sup> , Saholy N. Ralison<sup>5</sup> , Jean-Claude Randrianaivo<sup>5</sup> , Georgette Ralison<sup>5</sup> , Rija O. Andriamasinoro<sup>5</sup> , Rija H. Ramanampamaharana<sup>5</sup> , Solofomanantsoa E. Randrianazary <sup>5</sup> , Louise H. Ralimanana<sup>5</sup> , Béatrice Richard<sup>6</sup> , Philippe Gorry <sup>7</sup> , Marie-Cécile Manière1, 2 Jeanne A. Rasoamananjara<sup>5</sup> , Simone Rakoto Alson<sup>5</sup> and Hélène Dollfus <sup>4</sup>

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Eumorphia Remboutsika, BSRC "Alexander Fleming", Greece Amel Gritli-Linde, University of Gothenburg, Sweden Giovanna Orsini, Polytechnic University of Marche, Italy

### \*Correspondence:

Agnès Bloch-Zupan agnes.bloch-zupan@unistra.fr

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 17 December 2015 Accepted: 12 February 2016 Published: 02 March 2016

### Citation:

Bloch-Zupan A, Huckert M, Stoetzel C, Meyer J, Geoffroy V, Razafindrakoto RW, Ralison SN, Randrianaivo J-C, Ralison G, Andriamasinoro RO, Ramanampamaharana RH, Randrianazary SE, Ralimanana LH, Richard B, Gorry P, Manière M-C, Rasoamananjara JA, Rakoto Alson S and Dollfus H (2016) Detection of a Novel DSPP Mutation by NGS in a Population Isolate in Madagascar. Front. Physiol. 7:70. doi: 10.3389/fphys.2016.00070 <sup>1</sup> Faculté de Chirurgie Dentaire, Université de Strasbourg, Strasbourg, France, <sup>2</sup> Centre de Référence des Manifestations Odontologiques des Maladies Rares, Hôpitaux Universitaires de Strasbourg, Pôle de Médecine et Chirurgie Bucco-dentaires Hôpital Civil, Strasbourg, France, <sup>3</sup> Centre National de la Recherche Scientifique-UMR7104, Institut de Génétique et de Biologie Moléculaire et Cellulaire, Institut National de la Santé et de la Recherche Médicale U 964, Université de Strasbourg, Illkirch, France, <sup>4</sup> Laboratoire de Génétique Médicale, Faculté de Médecine, Institut National de la Santé et de la Recherche Médicale U 1112, Université de Strasbourg, Strasbourg, France, <sup>5</sup> Institut d'Odonto-Stomatologie Tropicale de Madagascar, Université de Mahajanga, Mahajanga, Madagascar, <sup>6</sup> UFR Odontologie de Lyon, Université Claude Bernard, Lyon, France, <sup>7</sup> Research Unit of Theoretical & Applied Economics, GREThA (UMR Centre National de la Recherche Scientifique 5113), Université de Bordeaux, Pessac, France

A large family from a small village in Madagascar, Antanetilava, is known to present with colored teeth. Through previous collaboration and 4 successive visits in 1994, 2004, 2005, and 2012, we provided dental care to the inhabitants and diagnosed dentinogenesis imperfecta. Recently, using whole exome sequencing we confirmed the clinical diagnosis by identifying a novel single nucleotide deletion in exon 5 of DSPP. This paper underlines the necessity of long run research, the importance of international and interpersonal collaborations as well as the major contribution of next generation sequencing tools in the genetic diagnosis of rare oro-dental anomalies. This study is registered in ClinicalTrials (https://clinicaltrials.gov) under the number NCT02397824.

### Keywords: rare disease, dentinogenesis imperfecta, dental anomalies, dentin, mutations, NGS, human

# INTRODUCTION

Dentinogenesis imperfecta (DI) belongs to a group of rare genetic diseases affecting the formation/mineralization of tooth dentin and is transmitted, as recorded so far, in an autosomal dominant manner (Barron et al., 2008). A dominant negative effect of a modified dentin sialophosphoprotein (DSPP) has been suggested as the pathogenic mechanism underlying DI (Wang et al., 2011).

These disorders exist either in isolation, with clinical manifestations limited to the oral cavity and are named dentinogenesis imperfecta type II (DGI-II) or hereditary opalescent dentin (OMIM #125490) also called dentinogenesis imperfecta 1 (DGI-1) or Capdepont teeth, and dentinogenesis imperfecta type III (DGI-III; OMIM # 125500) (Shields et al., 1973; Kim and Simmer, 2007; Barron et al., 2008; de la Dure-Molla et al., 2015). They can be associated with other symptoms like progressive sensorineural hearing loss (OMIM # 605594) (Xiao et al., 2001) or encountered in syndromes like osteogenesis imperfecta and Goldblatt syndrome for example in which bone defects (a tissue similar to dentin) are key features of the clinical synopsis (Bloch-Zupan et al., 2012; Bloch-Zupan, 2014). Mutations described so far occur in one single gene DSPP (4q21.3), belonging to the SIBLINGs family and encoding three major noncollagenous dentin matrix proteins- dentin sialoprotein (DSP), dentin glycoprotein (DGP) and dentin phosphoprotein (DPP) (Zhang et al., 2001; MacDougall, 2003; Kim et al., 2005; Lee et al., 2008, 2009, 2011a, 2013).

In this paper, we focus on a large family from a small village in Madagascar, Antanetilava, known to present with colored teeth. The aim of the study is, through phenotyping and genotyping, to unravel the diagnosis and genetic origin of this rare familial condition. Through previous collaboration (1996) and successive visits in 2004, 2005 and 2012, we provided dental care to the inhabitants and linked the discoloration diagnosis to dentinogenesis imperfecta. Recently, using whole exome sequencing we confirmed this diagnosis by identifying a novel DSPP mutation segregating with the disease in this family.

# MATERIALS AND METHODS

# Patients

In 1996 (Razafindrakoto et al., 1996), the Strasbourg Faculty of Dentistry and the INSERM\_U424 (JV Ruch) were contacted by colleagues from l'IOSTM (Institut d'Odonto-Stomatologie Tropicale de Madagascar) of Mahajanga University to help in disseminating scientific data related to a specific family originating from the small isolated village of Antanetilava (18◦ 58′ 42.2′′S 47◦ 14′ 01.9′′E), in the middle of the luxuriant tropical forest, in the Toamasina province, located 40 km North-West from the city of Toamasina. This region on the East coast of Madagascar is known for its hot and humid climate. It is a rural area devoted essentially to agriculture where rice, manioc, potatoes, banana and yam are cultivated. Colleagues visited Antanetilava in 1994, when a total of 110 inhabitants lived there. Problems related to inherited tooth anomalies in the population of Antanetilava had been previously noticed and the aim of this first visit was to determine what the tooth defects were, to follow the disorder in the families and to estimate its frequency.

At that time 50 people (28 females and 22 males) belonging to 22 of 26 households in the village were examined. Eleven individuals (22%, 6 females and 5 males) presented with this colored teeth anomaly affecting both the primary and permanent dentition. Clinical examination revealed brown to blue-gray discoloration of the crowns. Severe attrition due to early enamel chipping was visible. A clinical diagnosis of dentinogenesis imperfecta type II was proposed. Radiographic examination was possible for one 23 year-old patient in the nearest local hospital of Toamasina. Progressive pulp chamber obliterations as well as absent root canals were noticed, confirming the diagnosis. A first pedigree was drawn and demonstrated that the genetic disease affected 5 generations and 46.7% of the family members.

After a 2004 preparatory mission, JM returned to Mahajanga and visited Antanetilava in 2005 after a difficult journey consisting of 6 h of bus, 8 h of bush taxi, 1 dugout canoe river crossing and a further hour of walking.

Twenty seven participants (25 affected) and 2 nonaffected family members were examined during this 2005 visit. Affected and unaffected family members gave written informed consent and, the study was approved by the village council. The orodental phenotypes were documented using the D[4]/phenodent registry: a Diagnosing Dental Defects Database (see www.phenodent.org, to access assessment form). This registry allows the standardization of data collection and assists in orodental phenotyping. It also facilitates providing clinical care to patients, a basis for genotype/orodental phenotype correlations, and sharing of data and clinical material between clinicians. D[4]/phenodent registry is approved by CNIL (French National commission for informatics and liberty) under the following number 908416.

This clinical study has been registered in Clinical trials (https://clinicaltrials.gov) under the number NCT02397824 and is registered by the French Ministère de l'enseignement supérieur et de la recherche, DGRI/Cellule de bioéthique (bioethics committee) under DC-2012-1677. It was acknowledged by the CPP (person protection committee) Est IV on the 11/12/2012.

# Mutation Analysis

JM collected DNA samples using Whatman FTA cards and Oragene <sup>R</sup> DNA kits.

Genomic DNA was isolated from the saliva of 14 family members (9 affected and 5 unaffected), during the 2012 mission, using the prepIT-L2P OG-250 Oragene <sup>R</sup> DNA kit (DNA Genotek Inc., Ontario, Canada) according to standard protocols.

A few attempts of direct DSPP Sanger sequencing were unsuccessful.

We performed whole-exome sequencing (IntegraGen, Evry, France) for five affected patients (III.15, III.32, IV.26, IV.57, and IV.65) and one healthy individual (IV.22). Exons of DNA samples were captured using in-solution enrichment methodology (SureSelect Human All Exon Kits Version 3, Agilent, Massy, France) with the company's biotinylated oligonucleotide probe library (Human All Exon v5+UTR— 75 Mb, Agilent). The genomic DNA was then sequenced on a sequencer as paired-end 2X75 base pair reads (Illumina HISEQ2000, Illumina, San Diego, USA) resulting in an average coverage of 200X. Image analysis and base calling was performed using Real Time Analysis (RTA) Pipeline version 1.9 with default parameters (Illumina). The bioinfomatic analysis of sequencing raw data was based on the pipeline provided by the company (CASAVA 1.8, Illumina and finally detects from 80965 to 82263 variants (SNPs and Indels) per proband (**Table 1**). Annotation, ranking, and filtering of genetic variants were performed with the VaRank program (Geoffroy et al., 2015). Very stringent criteria were used for excluding non-pathogenic variants, in particular: (1) variants represented with an allele frequency of more than 1% in dbSNP 138, the EXAC database or the NHLBI Exome Sequencing Project Exome Variant Server (EVS), (2) variants found at the homozygous state or more than once at the heterozygous state in 48 control exomes, (3) variants in the 5′ or 3′ UTR, (4) variants with intronic locations and no prediction of local splice effect, and (5) synonymous variants without prediction of local splice effect.

### TABLE 1 | Summary of the exome sequencing results.


Sanger sequencing (GATC Biotech, Applied Biosystems ABI 3730xlTM, Konstanz, Germany) was used to validate the mutations and verify segregation using the following primers.

Specific forward (F) and reverse (R) primers were designed to amplify the DSPP exon 5 region containing the mutation: DSPP-F (GTGACAGCAGCAATAGCAGTGATA) and DSPP-R (TCACTGGTTGAGTGGTTACTGTC) (expected product size of 376 bp (base pair). PCR amplifications were performed in a final volume of 50µl containing 0.2µM forward and reverse primers, 0.2 mM dNTPs, 1X GoTaq reaction buffer containing 1.5 mM MgCl2, 1.25 unit of GoTaq DNA polymerase (Promega), 50 ng of template DNA and 3% DMSO. Amplifications were performed for 40 cycles, each consisting of 30s denaturation at 94◦C, 30s annealing at 64.9◦C and 17s elongation at 72◦C.

# RESULTS

# Clinical Phenotype

The family history and pedigree, as established in 1996, was updated. It then included 137 individuals spanning 5 generations (**Figure 1**) and 4 related families. 55 individuals, 31 males and 24 females, were reported as affected and presenting with DI corresponding to a 40.1% prevalence of the disease within this population (1/2 in generation I, 2/3 in generation II, 4/13 or 30.8% in generation III, 16/25 (64%) in IV, 30/64 (46.9%) in V and 2/7 (28.6%) in VI).

Medical history was collected from the 25 affected of the 27 examined persons. Patients reported only infectious episodes like malaria, measles, and fever. Some affected individuals (3) presented a triangular face shape or a facial asymmetry. Most of affected persons (23) showed blue sclera. Disturbance of hearing was recorded for 5 affected individuals. 9 patients presented articular distortions or pain and 3 had nail dysplasia.

Dental history mentioned infections, early tooth mobility and loss and tooth extractions. Both the primary and permanent dentitions were affected. Teeth presented with the amber-gray color pathognomonic of heritable dentin defects (**Figure 2**). Some tooth shape/size anomalies were observed as scoop shaped incisors, absence of convex vestibular crown surface, flat aspect of crown occlusal surfaces and supernumerary cusps. Enamel, when visible, presented an irregular appearance. Tooth wear was considerable and was visible via the colored abnormal dentin after enamel shedding. Fifteen individuals (11 adults, 4 children) suffered from tooth mobility. Nine individuals experienced dental infections. A probable diagnosis of DI was made. Three affected patients benefitted from X-ray investigations through intraoral radiographs taken in a private practice in the Antananarivo town. These pictures showed complete pulp space obliteration and globular crowns with cervical constrictions (**Figure 2**) confirming the diagnosis.

FIGURE 1 | Family pedigree. A large pedigree, spanning 5-generations with 137 individuals, of which 55 are affected, is showing a dominant inheritance pattern. Arrows point to individuals whose exomes were sequenced (affected: III.15, III.32, IV.26, IV.57, IV.65; non-affected: IV.22). Sanger sequencing was performed for the following subjects: affected (III.15, III.32, III.33, IV.25, IV.26, IV.57, IV.59, IV.64, IV.65), non-affected (III.16, III.37, IV.21, IV.22, IV.23).

FIGURE 2 | Clinical description of the disease. (A) The remote village of Antananarivo. (B–J) Inhabitants'dentition showing the typical features of dentinogenesis imperfecta with the gray-brown discolouration of the dentin clearly visible after enamel cleavage and progressive tooth wear. (C) On the retro-alveolar radiography of the lower right premolar/molar sector of individual (B), cervical constriction, short roots and the disappearance of pulp spaces due to erratic dentin formation represent the characteristic hallmarks of dentinogenesis imperfecta. (H–J) In addition to dentin anomalies, hypoplastic enamel defects exist, with the presence of pits, striae and flattened buccal surfaces.

# Genotype

Using VaRank, to annotate rank, and filter the genetic variants, we identified, amongst 80965–82263 variants (SNPs and Indels) per proband, four candidate variants in five genes (DSPP [dentin sialophosphoprotein, OMIM: 125485], ABHD14A-ACY1 [ABHD14A-ACY1 readthrough (NMD candidate)], ABHD14A [abhydrolase domain containing 14A], HERC6 [HECT and RLD domain containing E3 ubiquitin protein ligase family member 6, OMIM: 609249] and THAP9 [THAP domain containing 9, OMIM: 612537]) with heterozygous variants present only in the affected patients (**Table 1**). We then focused the subsequent study on the heterozygous variant in DSPP, a known causal gene for DI: a heterozygous deletion (c.3676del [p.Ser1226Alafs\*88] [RefSeq NM\_014208.3]). Mutation is absent from the 1000 genome, NHLBI EVS and ExAC databases.

We identified through exome sequencing and confirmed by bidirectional Sanger sequencing analysis of DSPP, a heterozygous deletion of 1 base pair (bp) in exon 5 (p.[Ser1226Alafs<sup>∗</sup> 88];[=] or c.[3676delA];[=]) in 9 affected patients (**Figure 3**). Segregation analysis validated the absence of the mutation in unaffected individuals. The deletion was absent from dbSNP and the Exome Variant Server. These mutations are predicted to cause a frameshift from codon Ser1226 producing an early stop codon 87 amino acids after the deletion and deleting the protein of an important functional domain.

# DISCUSSION

This work is an extraordinary travel in time, human beliefs and mutual assistance, genetics, science, and new technologies allowing the understanding of the exceptional prevalence of DI in this remote village from Madagascar. The family believed that the tooth coloration and the disease was of nutritional origin.

deletion of 1T is indicated with an arrow, this deletion creates a shift in the reading frame in position 3676 of the cDNA reference sequence, resulting in 2 superposed sequences. On the scheme the numbering corresponds to the following references: 1. Rajpar et al. (2002); 2. Malmgren et al. (2004); 3. Xiao et al. (2001); 4. Zhang et al. (2007) and Qu et al. (2009); 5. Hart and Hart (2007); 6. Mcknight et al. (2008a); 7. Li et al. (2012) and Lee et al. (2013); 8. Wang et al. (2009); 9. Lee et al. (2008); 10. Holappa et al. (2006); 11. Kim et al. (2004); 12. Kim et al. (2005); 13. Song et al. (2006); 14. Lee et al. (2011b); 15. Kida et al. (2009); 16. Lee et al. (2009); 17. Zhang et al. (2001); 18. Wang et al. (2011); 19. Mcknight et al. (2008b); 20. Zhang et al. (2011); 21. Bai et al. (2010); 22. Nieminen et al. (2011); 23. Song et al. (2008); 24. Lee et al. (2011a); 25. Dong et al. (2005).

Some hypotheses were proposed like the large consumption of red rice, or drinking habits (acidic water source). Oral tradition of the family history described healthy ancestors. Tradition forbids women, after giving birth, to eat white rice and transgression of this law after a famine period was believed to be associated with the appearance of the dental defect among the population.

DI incidence is believed to reach 1 in 8000 individuals according to Barron et al. (2008). Due to the founder effect, well observed in generation I of the pedigree, and the geographic isolation of the studied population, this prevalence approximates 40% in this population.

DI is transmitted as an autosomal dominant trait and this is clearly visible with the parent to child transmission seen in the pedigree and the presence of affected members in each generation.

The medical history revealed hearing loss problems, which indeed have been reported as associated with DI and DSPP mutations (Xiao et al., 2001). Blue sclerae are a classical hallmark of osteogenesis imperfecta clinical synopsis and the association of DI with even a milder form of osteogenesis imperfecta was still a possible diagnosis (Wang et al., 2012).

The phenotype demonstrated both enamel and dentin defects as was previously also reported (Wang et al., 2011). Dspp is expressed by odontoblasts and transiently by preameloblasts (Bronckers et al., 1993; Ritchie et al., 1997; Begue-Kirn et al., 1998).

Difficulties throughout the years to sequence the DSPP gene, especially the DPP region, are due to the high GC rich contents and the number of repeats. As no mutation could be initially detected in this candidate gene and because of disease high frequency within this population we hypothesized that another unidentified gene might be involved. Thus, we used exome sequencing to look for the causative gene. But in fact, we identified a novel single base pair deletion within the end of the fifth DSPP exon leading to a premature stop codon. It has never been described in the literature.

Thirty nine mutations in the human DSPP gene causing dentin defects have been previously reported (**Figure 3**). Mutations (mostly substitutions) leading to a DI (DGI) phenotype are located mostly at the 5′ end of DSPP and also seem to cluster in exon 2 and around the splice boundaries of exon 3. In exon 5 at the 3′ end of DSPP, deletions causing frame shift mutations were responsible for DGI and dentin dysplasia (DD) (Wang et al., 2011). The mutation described herein is also localized at the end of exon 5. This exon codes for DPP (dentin phosphoprotein), which is one of the most abundant extracellular matrix components in dentin (after collagen type I COL1A1, COL1A2). DPP has a role in biomineralization by binding to collagen and calcium and promoting the nucleation and growth of hydroxyapatite crystals (Prasad et al., 2010). The discovered mutation is predicted to cause a frameshift from codon Ser1226 producing an early stop codon 87 amino acids after the deletion, depleting the protein of an important functional domain. This domain is called "Asp/Ser-rich" by UniProt (position 439-1301).

To date, only one other mutation has been identified in the 3 ′ end of exon 5 (Dong et al., 2005) and consisted of a 36 bp deletion and an 18 bp insertion with a phenotype of DGI type III. Authors reported affected family members with amber tooth discoloration, opalescent appearance, severe attrition of teeth, visible pulp chambers and shell teeth on radiographs differing from the DI phenotype reported in this paper.

Targeted next-generation sequencing technics for orodental disorders (Prasad et al., 2016) prove to be efficient methods to sequence DSPP gene allowing further mutations detection and helping providing accurate molecular and clinical diagnosis to rare disease patients. Differential clinical and molecular diagnosis between DI and mild forms of osteogenesis imperfecta presenting with opalescent teeth is important and will orientate patients toward appropriate integrated dental and medical care. These methods, as associated costs decrease, will be transposed from research results to diagnostic molecular findings.

# AUTHOR CONTRIBUTIONS

JM, RWR, SNR, JCR, GR, ROA, RHR, SER, LHR, JAR collected the salivary samples and detailed the patients' phenotype. JM travelled back and forth between France and Madagascar to develop the project and gathered funding. BR, PG tried to sequence DSPP gene using conventional techniques. MH, CS, VG identified the molecular basis of the disease through NGS assays. MH, CS, VG, MCM, SRA, JAR, HD, ABZ analyzed the data and wrote the manuscript. ABZ designed the study and was involved from conception, funding seeking to drafting and critical review of the manuscript. All authors therefore contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.

# FUNDING

This work was financed by grants from: the University of Strasbourg, the Hôpitaux Universitaires de Strasbourg (API, 2009-2012, "Development of the oral cavity: from gene to clinical phenotype in Human"), the EU-funded project (ERDF) A27 "Oro-dental manifestations of rare diseases," supported by the RMT-TMO Offensive Sciences initiative, INTERREG IV Upper Rhine program, a contribution from EAPD and the INTERREG V RARENET program. This study was also supported by the grant ANR-10-LABX-0030-INRT, a French State fund managed by the Agence Nationale de la Recherche under the frame programme Investissements d'Avenir labeled ANR-10-IDEX-0002-02. ABZ is a fellow of University of Strasbourg Institute for Advanced Study (USIAS) and received USIAS Fellowship. Funding, support as well as dental materials were gathered by JM for her travels from the Conseil Départemental de l'Ordre des Chirurgiens Dentistes du Bas-Rhin, Ville de Strasbourg, Conseil Général du Bas-Rhin, Bureau de la Vie Etudiante de l'Université de Strasbourg, Association Amicale des Etudiants en Chirurgie Dentaire de Strasbourg, Henry Schein, GC Europe, Megadental, Laboratoire Unodis Haguenau, Colgate, Pierre Fabre, Alpha Omega Alsace, Laboratoire Flecher Strasbourg, Crédit Mutuel Profession de Santé. A grant was received in Madagascar from Général Randrianazary "Secrétaire d'état à la gendarmerie" in 2012, to cover travel expenses.

# WEB RESOURCES

The URLs for data presented herein are as follows: dbSNP, http://www.ncbi.nlm.nih.gov/projects/SNP/ NHLBI Exome Sequencing Project (ESP) Exome Variant Server, http://evs.gs.washington.edu/EVS/ OMIM, http://www.omim.org/ PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/ RefSeq, http://www.ncbi.nlm.nih.gov/refseq/ SIFT, http://sift.bii.a-star.edu.sg/ VaRank, http://www.lbgi.fr/VaRank UCSC Genome Browser, http://genome.ucsc.edu

# ACKNOWLEDGMENTS

We are grateful to the families and village council for their participation and invaluable contribution. We are grateful to Emilson, Nirina, Liva, Judicaël, Prudence, Nelly-Olivia for their help. We would like to thank Prof. Y. Rumpler, University of Strasbourg who inspired this project and long lasting collaboration. Several persons contributed to this project, and we would like to acknowledge their contribution here: Mr. J. Zafy Jean, for guiding one of the visits, Miss Rakotonindrina Miary for transferring samples from Madagascar, Dr. C. Kaempf, president of the Conseil de l'Ordre des Chirurgiens Dentistes du Bas-Rhin, Dr. H. Randrianary and Dr. Y. Randriamahefa, presidents of the Conseil de l'Ordre des Chirurgiens-Dentistes

# REFERENCES


de Madagascar and from Toamasina province in 2004, Dr. A. E. Rakotoarivony, Director of IOSTM, Dr. S. Ralison, assistant in IOSTM in 2004. Prof. C. Holmgren gave generous advises about the ART (Atraumatic Restorative treatment) technique. Prof. Y. Haikel supported this project. We would like to thank also Dr. M. K. Prasad for critical reading of the manuscript. The authors declare no potential conflict of interest with respect to the authorship and/or publication of this article.


**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 © 2016 Bloch-Zupan, Huckert, Stoetzel, Meyer, Geoffroy, Razafindrakoto, Ralison, Randrianaivo, Ralison, Andriamasinoro, Ramanampamaharana, Randrianazary, Ralimanana, Richard, Gorry, Manière, Rasoamananjara, Rakoto Alson and Dollfus. 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) or licensor 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.

# Corrigendum: Detection of a Novel DSPP Mutation by NGS in a Population Isolate in Madagascar

Agnès Bloch-Zupan1, 2, 3 \*, Mathilde Huckert 1, 4, Corinne Stoetzel <sup>4</sup> , Julia Meyer <sup>1</sup> , Véronique Geoffroy <sup>4</sup> , Rabisoa W. Razafindrakoto<sup>5</sup> , Saholy N. Ralison<sup>5</sup> , Jean-Claude Randrianaivo<sup>5</sup> , Georgette Ralison<sup>5</sup> , Rija O. Andriamasinoro<sup>5</sup> , Rija H. Ramanampamaharana<sup>5</sup> , Solofomanantsoa E. Randrianazary <sup>5</sup> , Louise H. Ralimanana<sup>5</sup> , Béatrice Richard<sup>6</sup> , Philippe Gorry <sup>7</sup> , Marie-Cécile Manière1, 2 , Jeanne A. Rasoamananjara<sup>5</sup> , Simone Rakoto Alson<sup>5</sup> and Hélène Dollfus <sup>4</sup>

<sup>1</sup> Faculté de Chirurgie Dentaire, Université de Strasbourg, Strasbourg, France, <sup>2</sup> Centre de Référence des Manifestations Odontologiques des Maladies Rares, Hôpitaux Universitaires de Strasbourg, Pôle de Médecine et Chirurgie Bucco-dentaires Hôpital Civil, Strasbourg, France, <sup>3</sup> Centre National de la Recherche Scientifique-UMR7104, Institut de Génétique et de Biologie Moléculaire et Cellulaire, Institut National de la Santé et de la Recherche Médicale U 964, Université de Strasbourg, Illkirch, France, <sup>4</sup> Laboratoire de Génétique Médicale, Faculté de Médecine, Institut National de la Santé et de la Recherche Médicale U 1112, Université de Strasbourg, Strasbourg, France, <sup>5</sup> Institut d'Odonto-Stomatologie Tropicale de Madagascar, Université de Mahajanga, Mahajanga, Madagascar, <sup>6</sup> UFR Odontologie de Lyon, Université Claude Bernard, Lyon, France, <sup>7</sup> Research Unit of Theoretical & Applied Economics, GREThA (UMR Centre National de la Recherche Scientifique 5113), Université de Bordeaux Pessac, Pessac, France

Keywords: rare disease, dentinogenesis imperfecta, dental anomalies, dentin, mutation, NGS, humans

### Edited and reviewed by:

Thimios Mitsiadis, University of Zurich, Switzerland

### \*Correspondence:

Agnès Bloch-Zupan agnes.bloch-zupan@unistra.fr

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

> Received: 10 May 2016 Accepted: 06 July 2016 Published: 26 July 2016

### Citation:

Bloch-Zupan A, Huckert M, Stoetzel C, Meyer J, Geoffroy V, Razafindrakoto RW, Ralison SN, Randrianaivo J-C, Ralison G, Andriamasinoro RO, Ramanampamaharana RH, Randrianazary SE, Ralimanana LH, Richard B, Gorry P, Manière M-C, Rasoamananjara JA, Rakoto Alson S and Dollfus H (2016) Corrigendum: Detection of a Novel DSPP Mutation by NGS in a Population Isolate in Madagascar. Front. Physiol. 7:304. doi: 10.3389/fphys.2016.00304 **Detection of a Novel DSPP Mutation by NGS in a Population Isolate in Madagascar**

by Bloch-Zupan, A., Huckert, M., Stoetzel, C., Meyer, J., Geoffroy, V., Razafindrakoto, R. W., et al. (2016). Front. Physiol. 7:70. doi: 10.3389/fphys.2016.00070

Reason for Corrigendum:

**A corrigendum on**

The authors, Jeanne A. Rasoamananjara and Louise H. Ralimanana were inadvertently missed in the original manuscript and we wish to add their names for contribution to this work. All authors have agreed to this modification.

The authors apologize for this oversight. This error does not change the scientific conclusions of the article in any way.

The original article has been updated.

# AUTHOR CONTRIBUTIONS

ABZ, MH, CS, JM, VG, RWR, SNR, JCR, GR, ROA, RHR, SER, LHR, BR, PG, MCM, JAR, SRA and HD approved the content of Corrigendum.

**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 © 2016 Bloch-Zupan, Huckert, Stoetzel, Meyer, Geoffroy, Razafindrakoto, Ralison, Randrianaivo, Ralison, Andriamasinoro, Ramanampamaharana, Randrianazary, Ralimanana, Richard, Gorry, Manière, Rasoamananjara, Rakoto Alson and Dollfus. 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) or licensor 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.

# Regulation of Calvarial Osteogenesis by Concomitant De-repression of GLI3 and Activation of IHH Targets

Lotta K. Veistinen<sup>1</sup> , Tuija Mustonen1, 2, Md. Rakibul Hasan<sup>1</sup> , Maarit Takatalo<sup>1</sup> , Yukiho Kobayashi 1, 3, Dörthe A. Kesper <sup>4</sup> , Andrea Vortkamp<sup>4</sup> and David P. Rice1, 5 \*

<sup>1</sup> Orthodontics, Oral and Maxillofacial Diseases, University of Helsinki, Helsinki, Finland, <sup>2</sup> Minerva Research Institute, Helsinki, Finland, <sup>3</sup> Orthodontics, Tokyo Medical and Dental University, Tokyo, Japan, <sup>4</sup> Center of Medical Biotechnology, University of Duisburg-Essen, Essen, Germany, <sup>5</sup> Orthodontics, Oral and Maxillofacial Diseases, Helsinki University Hospital, Helsinki, Finland

### Edited by:

Giovanna Orsini, Università Politecnica delle Marche, Italy

### Reviewed by:

Pierfrancesco Pagella, University of Zurich, Switzerland Javier Catón, CEU San Pablo University, Spain

> \*Correspondence: David P. Rice david.rice@helsinki.fi

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 02 August 2017 Accepted: 29 November 2017 Published: 19 December 2017

### Citation:

Veistinen LK, Mustonen T, Hasan MR, Takatalo M, Kobayashi Y, Kesper DA, Vortkamp A and Rice DP (2017) Regulation of Calvarial Osteogenesis by Concomitant De-repression of GLI3 and Activation of IHH Targets. Front. Physiol. 8:1036. doi: 10.3389/fphys.2017.01036 Loss-of-function mutations in GLI3 and IHH cause craniosynostosis and reduced osteogenesis, respectively. In this study, we show that Ihh ligand, the receptor Ptch1 and Gli transcription factors are differentially expressed in embryonic mouse calvaria osteogenic condensations. We show that in both Ihh−/<sup>−</sup> and Gli3Xt−J/Xt−<sup>J</sup> embryonic mice, the normal gene expression architecture is lost and this results in disorganized calvarial bone development. RUNX2 is a master regulatory transcription factor controlling osteogenesis. In the absence of Gli3, RUNX2 isoform II and IHH are upregulated, and RUNX2 isoform I downregulated. This is consistent with the expanded and aberrant osteogenesis observed in Gli3Xt−J/Xt−<sup>J</sup> mice, and consistent with Runx2-I expression by relatively immature osteoprogenitors. Ihh−/<sup>−</sup> mice exhibited small calvarial bones and HH target genes, Ptch1 and Gli1, were absent. This indicates that IHH is the functional HH ligand, and that it is not compensated by another HH ligand. To decipher the roles and potential interaction of Gli3 and Ihh, we generated Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> compound mutant mice. Even in the absence of Ihh, Gli3 deletion was sufficient to induce aberrant precocious ossification across the developing suture, indicating that the craniosynostosis phenotype of Gli3Xt−J/Xt−<sup>J</sup> mice is not dependent on IHH ligand. Also, we found that Ihh was not required for Runx2 expression as the expression of RUNX2 target genes was unaffected by deletion of Ihh. To test whether RUNX2 has a role upstream of IHH, we performed RUNX2 siRNA knock down experiments in WT calvarial osteoblasts and explants and found that Ihh expression is suppressed. Our results show that IHH is the functional HH ligand in the embryonic mouse calvaria osteogenic condensations, where it regulates the progression of osteoblastic differentiation. As GLI3 represses the expression of Runx2-II and Ihh, and also elevates the Runx2-I expression, and as IHH may be regulated by RUNX2 these results raise the possibility of a regulatory feedback circuit to control calvarial osteogenesis and suture patency. Taken together, RUNX2-controlled osteoblastic cell fate is regulated by IHH through concomitant inhibition of GLI3-repressor formation and activation of downstream targets.

Keywords: calvarial development, hedgehog signaling pathway, osteoblast, cell differentiation, craniosynostosis

# INTRODUCTION

The majority of the bones of the face and calvaria are formed by intramembranous ossification. Growth occurs primarily at the bone edges in the sutural joints (Rice and Rice, 2008). We and others have shown that across the suture there is the complete spectrum of osteogenic differentiation from quiescent stem cells through all osteoprogenitor stages to functioning mature osteoblasts. The expansion of each cell population is tightly regulated and dependent on the functional demands of the growing skull (Lana-Elola et al., 2007; Zhao et al., 2015; Maruyama et al., 2016). Osteoprogenitors condense at the osteogenic fronts and in order for osteogenesis to proceed a constant supply of progenitor cells must be available (Rice et al., 2003; Lana-Elola et al., 2007). Mis-regulation of the osteogenic condensations will result in alterations in skull bone shape and size, as well as in the patency of the sutures with consequent effects on skull growth. Within this niche, GLI transcription factors regulate stem cell maintenance, osteoprogenitor proliferation and differentiation (Rice et al., 2010; Veistinen et al., 2012). GLI1-positive cells in the suture mesenchyme are a source of mesenchymal stem cells, and ablation of Gli1 in postnatal mice results in the premature fusion, or craniosynostosis, of all the calvarial sutures and the consequent secession of growth (Zhao et al., 2015). GLI3 loss of function mutations cause craniosynostosis resulting in fusion across the interfrontal suture in patients and across the lambdoid and interfrontal sutures in mice (McDonald-McGinn et al., 2010; Rice et al., 2010; Hurst et al., 2011). These differences in phenotypes can, in part, be explained by location-specific differences in mRNA expression. Gli1 is expressed evenly in all calvarial sutures while Gli3 transcripts are more highly expressed in the interfrontal and lambdoid sutures compared to other calvarial sutures (Rice et al., 2010; Zhao et al., 2015).

GLI proteins are regulators of Hedgehog (HH) signaling, with GLI3 mainly functioning as a repressor of HH target genes. In the absence of the HH ligand GLI3 is proteolytically cleaved into the truncated repressor isoform, GLI3R (Wang et al., 2000). GLI3 acts as a strong repressor during limb patterning and during endochondral ossification, with SHH regulating limb patterning and IHH regulating cartilage development (Litingtung et al., 2002; Hilton et al., 2005; Koziel et al., 2005).

Indian hedgehog (Ihh) is expressed in the embryonic calvaria, and in osteoblastic cells has been shown to upregulate an osteogenic master regulator Runx2 through GLI2 activator function (GLI2A) (Shimoyama et al., 2007; Rice et al., 2010). It has been proposed that IHH regulates several stages of osteoblast differentiation, either through Gli3R or Gli2A functions (Hilton et al., 2005; Joeng and Long, 2009). Although IHH is essential for osteoblast differentiation during endochondral ossification, it is not essential for intramembranous bone development. Osteoblasts appear to develop normally in calvarial bones of Ihh−/<sup>−</sup> mice although the bones are reduced in size (St-Jacques et al., 1999; Lenton et al., 2011). By mapping the gene expression of Ihh in the developing chick dentary bone and analyzing the retroviral missexpression (gain of function) of Ihh in the developing chick frontal bone, it has been proposed that IHH regulates the transition of osteoprogenitors to osteoblasts. It is suggested that IHH does this by restricting the differentiation of preosteoblasts so that they can proliferate for longer which increases the pool of progenitors (Abzhanov et al., 2007). There is also evidence that IHH positively regulates osteoprogenitor recruitment to the osteogenic front possibly by controlling Bone morphogenetic protein (BMP) signaling (Lenton et al., 2011).

We have shown earlier how Gli3Xt−J/Xt−<sup>J</sup> mice, which produce no functional GLI3 protein, can be used as a model to study craniosynostosis and suture biogenesis, and by genetically reducing the dosage of Runx2 the calvarial phenotype can be rescued (Rice et al., 2010; Tanimoto et al., 2012; Veistinen et al., 2012). The transcription factor TWIST1 negatively regulates osteogenesis by inhibiting Runx2 through direct DNA binding (Rice et al., 2000, 2010; Bialek et al., 2004), and in Gli3Xt−J/Xt−<sup>J</sup> mice the expression of Twist1 is reduced across the sutures. This aberrant signaling results in an increased pool of osteoprogenitors and enhanced osteogenic differentiation, with consequent ectopic bone formation and suture fusion. Support for this mechanism comes from premature activation of Runx2 in embryonic day (E) 9.5 mice, as well as from the deletion of one allele of Twist1 both of which result in enhanced osteogenesis and craniosynostosis (Bourgeois et al., 1998; Maeno et al., 2011). GLI3 repressor reduces RUNX2 activity by competing for the same DNA target sequences, thus negatively regulating osteoblast differentiation (Ohba et al., 2008; Lopez-Rios et al., 2012).

In this study we aimed to elucidate whether GLI3 controls IHH regulated intramembranous osteogenesis, and to test whether IHH, GLI3 and RUNX2 regulate and maintain the different stages of osteogenesis within the patent suture.

We show that IHH is the functional HH ligand in the embryonic mouse calvaria osteogenic condensations, where it regulates the progression of osteoblastic differentiation. We also demonstrate a location specific regulatory role for GLI3 repressor within the suture which is independent of IHH expression, as Ihh deletion does not rescue craniosynostosis exhibited by Gli3 Xt−J/Xt−<sup>J</sup> mice. In addition, we show that GLI3 represses the expression of Runx2-II and Ihh, and elevates Runx2-I. And that IHH is regulated by RUNX2 in calvarial osteoblasts. These data raise the possibility of a regulatory feedback circuit to control calvarial osteogenesis and suture patency.

# MATERIAL AND METHODS

# Mice

The Ihh+/<sup>−</sup> and Gli3+/Xt−<sup>J</sup> mice maintenance, breeding and PCR genotyping has been described previously (Koziel et al., 2005). Wild-type (WT) littermates NMRI mice were used as controls. All animal experiments were approved by the University of Helsinki, Helsinki University Hospital and the Southern Finland Council Animal Welfare and Ethics committees.

# In Situ Hybridization

E13.5 and E15.5 tissues were fixed in a copious volume of freshly prepared 4% paraformaldehyde overnight at 4◦C. They were then washed in PBS before dehydrating in an increasing strength ethanol or methanol series before in situ hybridization on either tissue sections or whole mounts, respectively.

Ihh, Ptch1, Gli1-3, Runx2, Osx, Bglap, and Ibsp riboprobes were prepared, and in situ hybridization was performed as described previously and briefly described below (Rice et al., 2006; Tanimoto et al., 2012).

# In Situ Hybridization on Tissue Sections

Sections (7µm) were deparaffinised, rehydrated and permeabilized with proteinase K. Tissues were hybridized overnight at 52◦C with <sup>35</sup>S-UTP labeled riboprobes. Hybridization was followed by high stringency washes at 50◦C and at 65◦C. Slides were then washed in NTE at 37◦ and treated with ribonuclease A to remove non-specifically bound and excess probe. The slides were coated with autoradiography liquid emulsion and exposed in a dark box for 10–18 days at 4◦C. The slides were developed, fixed and then counterstained with hematoxylin.

# In Situ Hybridization on Whole Mounts

Samples were rehydrated, bleached with H2O<sup>2</sup> and permeabilized with proteinase K. Tissues were prehybridized in PBST. Tissues were hybridized overnight at 64◦C with denatured digoxigeninlabeled probes followed by high stringency washes. Next tissues were washed MABT then pre-blocked at room temperature for 3 h prior to overnight incubation with anti-dig-antibody coupled to alkaline phosphatase 4◦C. Following extensive MABT and NTMT washes the color reaction was performed with NBT/BCIP.

# Immunohistochemistry

Immunohistochemical staining was performed as described previously (Rice et al., 2016). Briefly tissue sections were permeabilized and blocked, then incubated overnight at 4◦C with anti-IHH (AF1705, R&D systems), anti-GLI1 (L42B10, Cell Signaling), anti-GLI3 (AF3690, R&D systems) or anti-PTCH1(G-19, Santa Cruz Biotechnology). Signal visualization was performed using Enzmet HRP detection (Nanoprobes). Sections were counterstained with nuclear fast red.

# Skeletal Staining

Heads of mutant mice and WT littermates aged E16.5 and E18.5 were fixed in 95% ethanol overnight and stained with Alcian blue and Alizarin red and then cleared in 1% KOH and transferred to glycerol. For calculations of calvarial bone and suture size, images were captured using AnalySIS software (Soft Imaging System) and Olympus BX41 microscope and analyzed in Adobe <sup>R</sup> Photoshop CS4.

# Western Blot Analysis

Calvaria were dissected from E15.5 WT embryos and tissue samples taken from osteogenic front of the frontal bone and from interfrontal midsutural mesenchyme. Samples were pooled from three calvaria of the same litter and this was repeated 4 times. Age-matched brain tissue was used as control. Western blotting was carried out as described previously (Tanimoto et al., 2012). Briefly, 10 µg of each sample was probed for GLI3 antibody (AF3690, R&D Systems), GLI1 antibody (2643, Cell Signaling Technology), α-tubulin (T6199, Sigma-Aldrich). 20 µg of total protein from siRNA treated calvarial cells was probed for RUNX2 antibody (8486, Cell Signaling Technology), IHH antibody (AF1705, R&D Systems), and normalized against signal by Actin antibody (A2066 Sigma), detected by secondary antibodies goat anti-mouse IRDye 800CW (926-32210, LI-COR) and goat anti-rabbit 680RD (926-68071, LI-COR). Blots were analyzed with an Odyssey CLx infrared imager (model 9120) (Li-Cor) and Image-J (NIH) software. Statistical values were calculated using Student's t-test, with P-value below 0.05 indicated as significant.

# The Effect of Exogenous IHH on Primary Calvaria Derived Cells

Mouse E15.5 wild type (WT) primary calvaria derived cells (CDC) were isolated by trysinization and cultured for several days in DMEM containing high glucose and supplemented with 10% FBS. Cells were pooled from at least 2 calvaria. After 1st passage cells were seeded onto 6-well plates for experiments (100,000 cells per well). After 48 h of culture, cells were treated with recombinant human/mouse IHH protein (1705-HH-025 R&D systems): 50, 100, 250 ng/ml, and control cells with BSA (bovine serum albumin) for 1, 2, 6, and 24 h. RNA isolated from control and IHH treated CDC cells was quantified. 300 ng of RNA was used for cDNA synthesis using reverse transcriptase enzyme and random hexamer primers. Thereafter, RT-qPCR was performed using the primers Gli1 (FP:CAGCATGGGAACAGAAGGACT, RP:CTCTGGCTGCTC CATAACCC), Gli2 (FP:AACTTTTGTCTCCTCGGGTCC,RP: CTGCTGTCCTCCAAGAGACC) and Gli3 (FP:AAGCCCATG ACATCTCAGCC,RP:CTCGAGCCCACTGTTGGAAT). SYBR fast (Kapa Biosystems) qPCR master mix was used for real time quantitative PCR using Quant studio 3 from Thermo scientific. Mouse 18S rRNA gene was used as the reference gene and negative control with no cDNA was used. The experiment conducted 3 times, each time with different biological samples.

# Calvarial Osteoblasts and Explant Cultures

E15.5 mouse calvarial cells were isolated by four sequential trypsin-treatments of whole calvaria separated from the skin and meninges. The first set of cells after 15 min 0.25% trypsin incubation were discarded, and the cells from the following trypsin treatments were pooled and cultured in growth medium (DMEM containing 4,5 g/L glucose supplemented with 1 mM Na-pyruvate, 4 mM L-glutamine, 1% penicillin-streptomycine, and 10% FBS, (Lonza and Gibco).

Analysis of Wt and Gli3Xt−J/Xt−<sup>J</sup> calvarial cells: At passage 2 the cells were transferred into new growth medium until confluent from which cell lysates were derived. Western blot analyses were carried out with 20 µg of total protein.

siRNA experiments: At passage 2 the cells were transfected either with negative control siRNA (Ambion Silencer Select control #1 siRNA, 4390843, Thermo Fisher) or Runx2 siRNAs (Ambion 4390771 Runx2 Silencer Select Pre-designed siRNA, Thermo Fisher) using Lipofectamine RNAiMAX (Thermo Fisher). Cells were grown on six well plates with the siRNA transfection complex for 3 days, after which the medium was replaced by osteogenic medium (DMEM, 10 mM βglycerophosphate, (Sigma), 50µg/ml ascorbic acid, Sigma, and 100 ng/ml BMP2 (R&D Systems) for 24 h from which cell lysates were derived. Western blot analyses were carried out with 20 µg of total protein. siRNA experiments were carried out 3 times.

E15 mouse calvarias were dissected for Trowell type of organ culture (Rice et al., 2000). Affi-gel agarose beads (Bio-Rad) were washed with PBS, and circa 200 beads were soaked with premixed siRNA—RNAiMAX reagent (Thermo Fisher) at 37◦C for 30 min. Several beads were placed on the calvarias that were further cultured for 3 days in osteogenic medium. Efforts were made to place beads onto comparable areas of the calvaria that had endogenous expression of either Runx2 full length or Ihh (Rice et al., 2010). The areas of Ihh expression in WT E15.5 calvaria were further identified (**Figure 1B**) and beads impregnated with either negative control or anti-RUNX2 SiRNAs were placed on the parietal bone osteogenic fronts for IHH experiments, which can be identified in bright field microscopy (Kim et al., 1998). Beads placed on the parietal bone or osteogenic fronts for RUNX2 experiments.

# RESULTS

# Location Specific Regulatory Role of GLI3 Repressor

To determine the specific transcript (**Figures 1A–C**) and protein (**Figure 1D**) expression domains and protein activity of HH pathway members in the developing calvaria in situ hybridization and immunohistochemistry were performed. Detailed in situ hybridization was performed at different developmental stages of calvarial and suture maturation. In an attempt to get comparable data, at both E13.5 and E15.5 analysis was done on consecutive E15.5 calvarial tissue sections. Also, at E15.5 whole calvarial tissue was analyzed to determine whether all sutures have similar expression patterns.

Ihh mRNA and protein were detected in the osteogenic fronts of all the calvarial bones, in a very restricted group of osteoprogenitors (**Figures 1C,D**). In addition there was weak protein expression across the suture. The cell surface HH receptor Ptch1 and Gli1 and Gli2 expression domains were overlapping with that of Ihh. Ptch1 and Gli1 are induced by HH ligand and therefore indicate the limit of Hedgehog activation (Lee et al., 2016). In contrast, Gli3 was diffusely expressed across the sutural mesenchyme.

In general, the calvarial osteogenic fronts (E15.5) have similar expression patterns to the osteogenic condensations (E13.5). Indeed, the osteogenic fronts can be considered analogous to osteogenic condensations with similar steps and regulation of osteoblastic differentiation and development. We also observe a differential expression of Gli3 in different sutures which may, in part, explain the suture specific phenotype seen in mice and patients with GLI3 mutations.

To establish if GLI3R is the predominant protein isoform in the calvarial suture, we performed western blotting on tissue samples dissected from either the midsuture or the osteogenic front from WT calvaria (**Figure 1E**). The total amount of GLI3 protein was higher in the midsutural samples. Both isoforms were observed in both samples, but the level of GLI3R was higher in the suture compared to the osteogenic front, and the opposite was true for the GLI3FL (Full length) (**Figures 1F,G**). The ratio of GLI3R/GLI3FL was significantly higher in the suture mesenchyme indicating that GLI3R is predominate in the midsutural tissue.

To verify whether a sample was midsutural or from the osteogenic front, we blotted the samples against GLI1. GLI1 also acts as a read-out for HH signaling, and based on our in situ hybridization results, HH signaling is active only in the osteogenic front. The level of GLI1 protein was significantly higher in the osteogenic front samples compared to the samples isolated from the suture (**Figures 1F,G**). Taken together, HH signaling is activated in the osteogenic fronts and GLI3R is predominantly localized in the midsutural mesenchyme.

# GLI3 Represses the Expression of Runx2-II and Ihh, and Elevates Runx2-I

RUNX2-I and RUNX2-II isoforms differ in their N-termini due to alternative promoter usage which results in dissimilar functions during bone development (Fujiwara et al., 1999; Zhang et al., 2009). Across developing sutures, Runx2 isoforms are differentially expressed with Runx2-I primarily expressed in relatively undifferentiated osteoprogenitors and Runx2-II in cells further along in osteoblastic differentiation (Park et al., 2001). We have previously shown that Runx2 is ectopically expressed across the Gli3Xt−J/Xt−<sup>J</sup> suture (Rice et al., 2010). Here we analyzed their protein levels in osteogenic cells isolated from Gli3Xt−J/Xt−<sup>J</sup> calvaria. We found that RUNX2-II was upregulated and RUNX2- I downregulated, compared to wild type controls indicating that in Gli3Xt−J/Xt−<sup>J</sup> mice the osteoprogenitors within the suture were more differentiated than their WT littermates (**Figures 2B,C**). These results indicate that GLI3 maintains osteoprogenitors in an undifferentiated state by participating in the regulation of RUNX2. These data help explain the location specific differences seen in mRNA expression in Gli3Xt−J/Xt−<sup>J</sup> calvaria and the craniosynostosis observed.

As Runx2 expression is altered in Gli3Xt−J/Xt−<sup>J</sup> calvaria (**Figure 2**) (Rice et al., 2010), and RUNX2 is known to regulate limb growth through the induction of Ihh (Yoshida et al., 2004), we analyzed IHH protein levels in Gli3Xt−J/Xt−<sup>J</sup> cells. We found IHH to be upregulated (**Figure 2D**). This indicates that the lack of GLI3 has allowed either increased IHH expression or a broader IHH protein expression domain.

# Ihh Does Not Alter the Expression of Gli3 mRNA

In mesenchymal cells of the developing limb bud, Sonic hedgehog is known to up-regulate Gli transcription, while down-regulating Gli3 expression (Marigo et al., 1996). To test whether IHH has a direct effect on the expression of Gli3 in WT primary calvaria derived cells (CDC) we added recombinant mouse IHH, of varying concentration, to mouse E15.5 calvarial cells and

(Continued)

FIGURE 1 | osteogenic fronts calvarial bone primordia. Ptch1, Gli1,-2 and−3 are also expressed in the osteogenic fronts but in a wider domain than Ihh. Gli3 is also expressed more distant from the osteogenic fronts in the midsutural mesenchyme (white arrows). (E) Schematic to show the areas of tissue sampling from E15.5 WT calvaria interfrontal midsutural mesenchyme (green) and osteogenic fronts of frontal bones (red). (F) Western blot analysis (triplicate), GLI3R (83 kDa) is the upper band indicated by the arrowhead. Integrated density values obtained for GLI3; both GLI3 full-length (GLI3FL) (190 kDa) and GLI3R, and GLI1 levels were normalized against those of α-tubulin and compared against the samples (G). GLI3R/GLI3FL ratio is higher in the midsutural mesenchyme compared to the osteogenic front. Conversely, normalized GLI1 intensity is greater in the osteogenic fronts compared to the midsuture. Error bars standard deviation. Statistical values calculated using Student's t-test, with p < 0.05 considered significant (\*p < 0.05). b, brain; e, eye; f, frontal bone; if, interfrontal suture, ip, interparietal bone; l, lambdoid suture; p, parietal bone; s, sagittal suture; sut mes, sutural mesenchyme. Scale bars: (A–C) 250µm, (D) 100µm.

investigated Gli1, −2 and −3 expression levels RT-qPCR at different time points.

Exogenous IHH did not alter the mRNA levels of Gli3 or Gli2 (**Supplementary Figure 1**). At 24 h, IHH exposure resulted in a gradual up-regulation of Gli1 expression in a dose-dependent manner (not statistically significant) (**Supplementary Figure 1D**). As Gli1 is a known target of hedgehog signaling activation this response was used as a positive control. This is consistent with previous work in a chondrogenesis model using human bone marrow stromal/stem cells (Handorf et al., 2015).

# Ihh Deletion Does Not Rescue the Gli3**Xt**−**J**/**Xt**−**<sup>J</sup>** Craniosynostosis

GLI transcription factors are effectors of HH signaling, with GLI3 acting as a strong repressor of IHH signals in chondrocyte differentiation with the developmental defects in endochondral ossification observed in Ihh−/<sup>−</sup> mice being partially normalized by the loss of Gli3 (Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice) (Koziel et al., 2005). We considered an analogous role during osteoblast development in the calvaria, and compared the calvarial phenotypes of Ihh−/−, Gli3Xt−J/Xt−<sup>J</sup> and Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mutant mice, E16.5–18.5, to those of WT littermates (**Figure 3**).

Calvarial bones of Ihh−/<sup>−</sup> mice were smaller than WT. The size of the frontal and the interparietal bones was most affected, the difference being the greatest at E16.5 [frontal bone 43% smaller (p < 0.05), interparietal bone 33% smaller compared to WT]. Concomitant with the reduced bone size was a widening of all of the calvarial sutures [E16.5: posterior interfrontal suture 32% wider (p < 0.05), sagittal suture 43% wider, lambdoid suture 45% wider. E18.5: lambdoid suture comparable to WT, posterior interfrontal suture 47% wider (p < 0.05), sagittal suture 71% wider (p < 0.05)] (**Figures 3E–H,Q–T**). The posterior part of the frontal bone was reduced in height (in apical direction) and the interfrontal suture was the widest posteriorly. At E18.5, the overall length of Ihh−/<sup>−</sup> skull (nasal bones to the occiput) was slightly reduced (7%, not significant). This was in part due to the short cranial base (12% shorter, p < 0.05) (**Figures 3E,G**).

Gli3Xt−J/Xt−<sup>J</sup> mutant mice showed very specific stage and location differences in bone size and suture width (**Figures 3I–L,Q–T**). Already at E16.5 25% of Gli3Xt−J/Xt−<sup>J</sup> mice showed synostosis of the lambdoid suture. This was primarily due to an increase in the size of the interparietal bone which was enlarged from E16.5 onwards (**Figure 3T**) as opposed to a generalized increase in the size of the parietal bone. At E16.5 the width of the posterior interfrontal suture in Gli3Xt−J/Xt−<sup>J</sup> calvaria was significantly wider (26%, p < 0.05). However, the surface area of the frontal bones was only 15% smaller. This mismatch of the widened interfrontal suture in between the near normally sized frontal bones, indicated that frontal bone development was affected from an early stage and that extrinsic factors, for example brain shape and size and may affect frontal bone and interfrontal suture morphology. Later, the size of the frontal bones relative to WT increased. This was in part due to strong heterotopic ossification in the interfrontal suture, which culminated in partial fusion of the suture at E18.5. (**Figure 3K**). At every stage examined the width of sagittal suture was significantly wider (p < 0.05) (51% wider compared to WT at E16.5, 87% wider at E18.5), but the size of parietal bones was not significantly smaller compared to WT (15% smaller at E16.5, 12% smaller at E18.5). This sagittal suture widening may be caused also by extrinsic factors. At E16.5 and E18.5 overall length of Gli3Xt−J/Xt−<sup>J</sup> mutant skull was comparable to WT samples (**Figures 3I–L**).

In conclusion, Gli3Xt−J/Xt−<sup>J</sup> calvaria exhibited enlargement of the interparietal bone and heterotopic interfrontal bones with consequent interfrontal and lambdoid suture fusion. In contrast, the same sutures were wider and the bones smaller in Ihh−/<sup>−</sup> mice. These opposite differences led us to test whether Gli3Xt−J/Xt−<sup>J</sup> phenotype could be rescued by deleting Ihh. The craniosynostosis phenotype across the interfrontal and lambdoid sutures in Gli3Xt−J/Xt−<sup>J</sup> mice was unaffected by the loss of Ihh (**Figures 3M–P**), despite there being a small normalization of interparietal size at E16.5 (not significant) (**Figure 3T**). Thus, even in the absence of IHH ligand, Gli3 deletion is sufficient to drive precocious ossification of the suture mesenchyme.

# Unlike Ihh−/<sup>−</sup> Endochondral Bones, Osteoblastogenesis In Ihh−/<sup>−</sup> Calvaria Progresses Normally

Ihh−/<sup>−</sup> mice exhibit a severe disruption of endochondral osteogenesis with a complete lack osteoblasts. In contrast Ihh−/<sup>−</sup> mice do form calvarial bones (St-Jacques et al., 1999). Using Runx2, Osterix (Osx, Sp7), Integrin binding sialoprotein (Ibsp), and Bglap (Osteocalcin) as markers of increasing osteoblastic maturity, we analyzed osteoblastogenesis in Ihh−/<sup>−</sup> embryonic calvaria (**Figure 4**). We found that osteoblastogenesis progressed normally even though the calvarial bones were small and thin, and the sutures wide with marker expression reflecting this phenotype (**Figures 4E–H**).

FIGURE 2 | Gli3Xt−J/Xt−<sup>J</sup> calvarial osteoblasts in growth medium without osteogenic supplements show upregulated levels of RUNX2-II and IHH, and downregulated RUNX2-I, for 3 biological samples in each test group by six separate western blot analysis, representative results are shown. (B,C) Results are normalized against Actin antibody, and normalized values are shown with 1 representing the WT osteoblasts. (\*p < 0.05) Error bars standard deviation.

# De-repression of GLI3 in Ihh−/<sup>−</sup> Mice Results in a Similar Calvarial Osteoblastic Phenotype to Gli3**Xt**−**J**/**Xt**−**<sup>J</sup>** Mice

In Ihh−/<sup>−</sup> endochondral bones, we have previously shown that, de-repression of GLI3R (Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice) is sufficient to restore the RUNX2-positive osteoprogenitor population (Koziel et al., 2005) but the additional action of GLI2A is required to restore the progression of osteoblastogenesis to the RUNX2 positive/Osterix-positive cell stage (Hilton et al., 2005). In Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> calvaria the osteoblastic expression pattern was similar to that in Gli3Xt−J/Xt−<sup>J</sup> mice although the heterotopic ossifications were less extensive (**Figures 4M–P**). Interestingly Gli3Xt−J/Xt−<sup>J</sup> mice, which exhibit heterotopic ossifications in the interfrontal suture, showed expression of immature osteoblasts (Runx2 and Osx) extending across the full width of the suture and more mature osteoblastic expression markers (Ibsp and Bglap) were limited to isolated ossification islands (**Figures 4I-L**).

# Ihh Is the Functional HH Ligand in Late Embryonic Mouse Calvaria

Ptch1 and Gli1 are direct transcriptional targets of HH signaling and both are upregulated by IHH (Lee et al., 2016). In Gli3Xt−J/Xt−<sup>J</sup> mice, Ptch1 and Gli1 are expressed in the developing calvarial bones and in the abnormal craniosynostotic sutures. In Ihh−/<sup>−</sup> calvaria however, Ptch1 and Gli1 were absent, indicating that no other Hh ligands are functional to compensate for the lack of Ihh (**Figures 5A–F**).

Endochondral ossification defects observed in Ihh−/<sup>−</sup> mice can be partially normalized by deletion of Gli3 (Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice). This rescue is accompanied by a concomitant normalization of Ptch1 expression (Koziel et al., 2005). We therefore decided to test whether Ptch1 and Gli1 are normalized in the developing calvaria of Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice. However, similar to Ihh−/<sup>−</sup> calvaria, Ptch1 and Gli1 expression were absent in Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice (**Figures 5G,H**). This suggests that the mechanism of Ptch1 and HH signaling activation in the calvarial osteoprogenitors differs compared to that in chondrocytes in the developing growth plate.

# Ihh Is Regulated by Runx2 in Calvarial Osteoblasts and Calvarial Explants

In the developing growth plate, it has been shown that Runx2 directly regulates Ihh expression in chondrocytes (Yoshida et al., 2004). To test an analogous interaction during intramembranous osteogenesis, we silenced Runx2 in E15.5 primary calvarial osteoblasts by selective Runx2 siRNA. Silencing was verified by western blot analysis showing

FIGURE 3 | (to the WT n = 20). Gli3Xt−J/Xt−<sup>J</sup> mice (n = 24) have heterotopic ossifications in the interfrontal suture. The interparietal bone is larger and the lambdoid sutures are narrower with some samples already fused at E16.5. In contrast the sagittal suture is wider compared WT. Although the calvarial bone areas are reduced at E16.5, deletion of Ihh in Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> (DKO) (n = 6) mice does not prevent the lambdoid sutures fusing (arrows) and heterotopic bones forming in the interfrontal suture (arrowheads) by E18.5. f, frontal bone; ip, interparietal bone; is, interfrontal suture; ls, lambdoid suture; p, parietal bone; ss, sagittal suture; DKO, double knockout (Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> ). (\*, p < 0.1, \*\*, p < 0.01 compared to WT). Error bars standard deviation. Scale bar 2 mm.

sialoprotein (Ibsp) are all expressed in the WT frontal bones, shown by in situ hybridization (A–D arrows). All markers are also expressed in the Ihh−/<sup>−</sup> frontal bones, but the interfrontal suture is wider (E–H). In Gli3Xt−J/Xt−<sup>J</sup> (I–L) and Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> (M–P) calvaria Runx2, Osx, Bglap and Ibsp are expressed in the frontal bones, but expression also extends to the interfrontal suture (arrowheads). b, brain; e, eye; f, frontal bone; is, interfrontal suture; s, skin. Scale bar 500µm.

that Runx2 protein was downregulated (**Figures 6A,B**). Ihh protein levels were significantly (p < 0.001) downregulated with Runx2 siRNA, when compared to negative controls (**Figures 6A,B**). The regulation of Ihh by Runx2 was then tested in embryonic calvarial organ culture with reduced expression of Ihh observed adjacent to beads coated in Runx2 siRNA, assayed by whole mount in situ hybridization (**Figure 6C**).

# DISCUSSION

Suture biogenesis involves the balance between the promotion and suppression of osteogenesis. The correct location specificity of the positive and negative osteo-inductive signals is important so that aberrant bone formation does not take place. It is not only important that the size and shape of the bones are controlled and that ectopic bones do not form but also that sutures stay patent to allow for continued brain and craniofacial growth.

Here, we show how skull bone size and shape are tightly controlled by hedgehog signaling in maintaining the required balance of progenitor cells and functioning osteoblasts, with Gli3 having a pivotal role. Gli3R acts as an inhibitor of osteogenesis in a location specific manner by repressing the osteo-inductive key players, RUNX2-II and IHH. GLI3R can act to negatively regulate RUNX2 or the function of RUNX2 via a least three possible mechanisms: GLI3 inhibits RUNX2 by competitively

Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> calvaria expression of both Ptch1 and Gli1 is absent (G,H black arrows). In Ihh−/<sup>−</sup> and Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice Ptch1 and Gli1 are expressed in the skin where they are activated by Shh (C,D,G,H arrowheads). In Gli3Xt−J/Xt−<sup>J</sup> and Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice Gli1 is ectopically expressed in the apical midline of the malformed cerebrum, intracranial to the interfrontal suture where it is activated by Shh (F,H blue arrow). b, brain; e, eye; f, frontal bone; is, interfrontal suture; s, skin. Scale bar 1mm.

binding to the BGLAP (osteocalcin) promoter (Ohba et al., 2008; Lopez-Rios et al., 2012); GLI3 inhibits BMP2-DLX5 activation of RUNX2; and GLI3 upregulates Twist1 which directly represses Runx2 (Bialek et al., 2004; Tanimoto et al., 2012). RUNX2- I is down regulated in Gli3Xt−J/Xt−<sup>J</sup> sutures and RUNX2-II upregulated. This is consistent with the expression of Runx2- I by a more immature osteoprogenitor population than those cells that express Runx2-II (Park et al., 2001; Tanimoto et al., 2012).

The effects of Ihh deletion on endochondral ossification are dramatic with long bones lacking osteoblasts (St-Jacques et al., 1999). IHH promotes osteogenesis in the orthotopic bone collar of endochondral bones through simultaneous GLI3 suppression and GLI2 activation (Hilton et al., 2005; Joeng

and Long, 2009). Genetic expression of a constitutive Gli2 activator is insufficient to rescue the defective osteoblastic differentiation in Ihh−/<sup>−</sup> mice. However, the additional removal of Gli3 restores Runx2 and Osx expression (Joeng and Long, 2009). The regulation of osteogenesis in the calvaria differs from that in long bones as Ihh−/<sup>−</sup> calvaria express Runx2 and Osx (**Figures 4E,F**). That said, IHH positively regulates intramembranous osteogenesis with IHH duplications resulting in an overexpression of IHH being associated with craniosynostosis, and Ihh−/<sup>−</sup> mice exhibiting small calvarial bones with expanded sutures (**Figures 3E–H**) (Klopocki et al., 2011; Lenton et al., 2011). Despite us demonstrating that IHH is the functional ligand for HH regulated osteogenesis in the calvaria (**Figures 5C,D**), deletion of Ihh in Gli3Xt−J/Xt−<sup>J</sup> calvaria was not sufficient to rescue the craniosynostosis (**Figures 3M–P**). Despite the lack of IHH targets (Ptch1 and Gli1) in Ihh−/−;Gli3Xt−J/Xt−<sup>J</sup> mice (**Figures 5G,H**). This highlights the repressive role of GLI3 and also that

RUNX2; and GLI3R upregulates Twist1 which directly represses Runx2 (Bialek et al., 2004; Tanimoto et al., 2012). IHH promotes GLI2A which permits the progression of osteoblast differentiation from RUNX2-positive to RUNX2-positive/OSX-positive progenitors (Shimoyama et al., 2007). Our data suggests that RUNX2 may also participate in the regulation IHH. Figure is adapted from (Joeng and Long, 2009), and (Long and Ornitz, 2013) (Joeng and Long, 2009; Long and Ornitz, 2013). (B1) Osteogenesis in the normal suture. GLI3 primarily acts as an inhibitor of intramembranous ossification at the periphery of the bones calvarial bones. The truncated repressor form of GLI3 (GLI3R) is the predominant isoform in the undifferentiated sutural mesenchyme (yellow) and at the peripheral edge of the osteogenic condensation (blue). GLI3R restricts ossification and limits the proliferation rate of the mesenchymal osteoprogenitor cells by upregulating Twist1 (1). In midsutural mesenchyme Runx2-I expression by relatively undifferentiated progenitors is high, while in the osteogenic condensation the Runx2 promoter is activated and both Runx2 isoforms are upregulated. Ihh expression is highly localized within the osteogenic condensation possibly resulting in a short range activation of target genes, including GLI2A. GLI3 repressor limits bone formation at the edge of the osteogenic condensation. In the osteogenic condensation RUNX2-I and RUNX2-II act in a feedback loop by upregulating IHH (dashed line). Within the Ihh signaling domain (red) the truncation of GLI3FL to GLI3R does not occur and osteogenesis is (Continued)

FIGURE 7 | activated. Outside the IHH signaling domain, GLI3R is generated which inhibits osteogenesis and maintains suture patency. (B2) In Gli3Xt−J/Xt−<sup>J</sup> calvaria, as GLI3R does not function, the tight regulation of osteogenesis is lost. This permits the expression domains of IHH and RUNX2 to expand and bone to form across the suture. Key: GLI3 Full length (GLI3FL), GLI3 repressor (GLI3R), undifferentiated sutural mesenchyme (yellow), osteogenic condensation periphery (blue), osteogenic condensation center (red), bone matrix (gray).

craniosynostosis is not just excessive osteogenesis but a patterning defect.

Analysis of Ihh zebrafish mutants suggests that Ihh is involved in osteoprogenitor recruitment and proliferation and ultimately the regulation of dermal bone outgrowth and shape (Huycke et al., 2012). Also, following local retroviral mis-expression of Ihh in the developing chick frontal bone a developmentally later role for Ihh has been suggested where it regulates the transition from preosteoblastic progenitors to osteoblasts (Abzhanov et al., 2007).

Our results do not exclude the possibility that GLI3 has other roles during calvarial development even prior to the establishment of sutures. Inactivation of the BMP responsive transcription factors Msx1 and Msx2 from neural crest cells from E9.5–E10.5 causes heterotopic ossification in the interfrontal suture due to an alteration in the fate of the early migrating neural crest cells (Roybal et al., 2010). A population of calvarial mesenchymal cells arise from HH-responsive Gli1 positive cells in the cephalic paraxial mesoderm and although this transcriptional activation of Gli1 is transient, GLI3 may have a role in controlling migration and patterning of these cells (Deckelbaum et al., 2012).

During calvarial osteoblast differentiation Gli3 is expressed prior to Ihh expression and then it functions as a repressor of HH target genes. During limb patterning the situation is similar, as GLI3 and dHand prepattern the limb bud before dHand activates SHH signaling (te Welscher et al., 2002). Evidence also shows that mouse limbs expressing only GLI3R have only one digit and thus resemble Shh null allele mice. The primary mechanism by which SHH patterns the anteriorposterior limb is by inhibiting the GLI3R formation (Cao et al., 2013).

Two mouse models of craniosynostosis have been reported with abnormal processing of GLI3 as an etiological factor. An ENU-induced mouse model representing a hypomorphic allele of Ptch1 has bilateral craniosynostosis of the lambdoid suture, similar to Gli3Xt−J/Xt−<sup>J</sup> mice (Feng et al., 2013). This phenotype is caused by a ligand-independent activation of HH signaling and a decrease in GLI3R. Fuzzy is a regulator of cilia trafficking and consequently HH signaling. Despite, Fuzzy null allele mice have increased repression of GLI3R they also exhibit craniosynostosis, all be it a different suture from that fused in Gli3Xt−J/Xt−<sup>J</sup> and Ptch1 mutant mice. The coronal suture fusion seen in Fuz−/<sup>−</sup> mice is associated with elevated fibroblast growth factor signaling which is known to play a role in regulating suture patency (Tabler et al., 2013).

We found that in Gli3Xt−J/Xt−<sup>J</sup> cells, RUNX2 and IHH are elevated and silencing Runx2 decreases Ihh expression (**Figures 6A,B**). This suggests that RUNX2 may control Ihh expression during calvarial osteoblast differentiation, which would be analogous to its role during chondrogenesis (Yoshida et al., 2004). Ihh is expressed by progenitors in the osteogenic fronts (**Figures 1A,B**) and it signals to the cells within the osteogenic condensation to differentiate via a GLI2 driven mechanism. Proteolytic processing of GLI3 into the repressor isoform is inhibited in these cells allowing osteoblast differentiation to proceed. Runx2 is upregulated and in turn activates Ihh expression (**Figures 2A–C**).

# CONCLUSIONS

The stepwise differentiation of osteoblasts from mesenchymal progenitors through developmental stages characterized by the expression of SOX9, RUNX2, and OSX is well described (Long and Ornitz, 2013) (**Figure 7A**). IHH acts at 2 stages prior to and post Runx2 positive cells (Joeng and Long, 2009). IHH inhibits the post translational cleavage of GLI transcription factors into truncated repressor forms (Wang et al., 2000). Thus, when IHH is expressed GLI1 and −2 function as activators of downstream targets to regulate osteogenesis. GLI1 regulates calvarial stem cells while GLI2 acts an activator which, together with the simultaneous removal of GLI3R, allows the progression of uncommitted RUNX2-positive osteoprogenitors into committed RUNX2-positive/OSX-positive osteoprogenitors (Shimoyama et al., 2007; Joeng and Long, 2009; Zhao et al., 2015).

We and others have previously shown that GLI3R regulates calvarial osteogenesis by negatively regulating the action of RUNX2. With RUNX2, GLI3R competitively binds to the BGLAP (osteocalcin) promoter (Ohba et al., 2008; Lopez-Rios et al., 2012). GLI3R inhibits the Bmp2-Dlx5 activation of Runx2 and upregulates TWIST1 which directly represses RUNX2 (Bialek et al., 2004; Tanimoto et al., 2012).

We show that IHH is the functional HH ligand in the late embryonic mouse calvaria. But unlike Ihh−/<sup>−</sup> endochondral bones, osteoblastogenesis in Ihh−/<sup>−</sup> calvaria progresses normally. So IHH is not essential for calvarial osteoblastic differentiation (St-Jacques et al., 1999). As IHH regulates the conversion of full-length GLI3 into GLI3R, within the IHH signaling domain the truncation of GLI3FL to GLI3R does not occur and osteogenesis is activated. Outside the IHH signaling domain, in immature mesenchymal progenitor cells, GLI3R is generated which inhibits osteogenesis (**Figure 7B**). Here we demonstrate a location specific regulatory role for GLI3R within the suture which is independent of Ihh expression, as IHH deletion does not rescue craniosynostosis exhibited by Gli3Xt−J/Xt−<sup>J</sup> mice. We show that GLI3 represses the expression of Runx2-II and Ihh, and elevates Runx2-I. And that IHH is regulated by RUNX2 in calvarial osteoblasts and calvarial explants. This raises the possibility of a regulatory feedback circuit to control calvarial osteogenesis and suture patency.

While we have found some interesting results with regard to intra-suture regulation of, particularly, the interfrontal suture, we have not made extensive analyses of different sutures and at multiple developmental stages. With these limitations in mind, caution is emphasized when extrapolating conclusions we make from data mainly from the interfrontal suture to other sutures.

Taken together, we have presented evidence how GLI3R, RUNX2 and IHH regulate the stage and location of osteoblast differentiation in the calvarial sutures, which ultimately controls skull bone patterning, shape and size.

# AUTHOR CONTRIBUTIONS

Study design: LV, TM, MH, MT, AV, and DR, Data collection, analysis, and interpretation: LV, TM, MH, MT, YK, DK, AV, and DR, Drafting manuscript and final version approval: LV, TM, MH, MT, YK, DK, AV, and DR, Responsibility for the integrity of the data analysis: LV, TM, MH, MT, YK, DK, AV, and DR.

# FUNDING

This work was supported by Biocentrum Helsinki, The Research Foundation of Helsinki University Hospital (TYH2015319),

# REFERENCES


Jusélius Foundation (4702957), Academy of Finland (257472) and FINDOS.

# ACKNOWLEDGMENTS

We are grateful to Airi Sinkko and Marjatta Kivekäs for excellent technical assistance.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.01036/full#supplementary-material

Supplementary Figure 1 | IHH does not affect the expression of Gli3 mRNA. (A–D) Gli1,−2 and −3 expression levels in WT primary calvaria derived cells analyzed by RT-qPCR at 1, 2, 6, and 24 h in response to application of 50, 100 and 250 ng/ml of recombinant mouse IHH. Exogenous IHH did not alter the mRNA levels of Gli2 or Gli3. At 24 h, IHH exposure resulted in a gradual up-regulation of Gli1 expression in a dose-dependent manner (not statistically significant). As Gli1 is a known target of hedgehog signaling activation this response was seen as a positive control.


**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 © 2017 Veistinen, Mustonen, Hasan, Takatalo, Kobayashi, Kesper, Vortkamp and Rice. 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) or licensor 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.

# Hoxa2 Inhibits Bone Morphogenetic Protein Signaling during Osteogenic Differentiation of the Palatal Mesenchyme

### Paul P. R. Iyyanar\* and Adil J. Nazarali †

*Laboratory of Molecular Cell Biology, College of Pharmacy and Nutrition and Neuroscience Research Cluster, University of Saskatchewan, Saskatoon, SK, Canada*

Cleft palate is one of the most common congenital birth defects worldwide. The homeobox (*Hox*) family of genes are key regulators of embryogenesis, with *Hoxa2* having a direct role in secondary palate development. *Hoxa2*−/<sup>−</sup> mice exhibit cleft palate; however, the cellular and molecular mechanisms leading to cleft palate in *Hoxa2*−/<sup>−</sup> mice is largely unknown. Addressing this issue, we found that *Hoxa2* regulates spatial and temporal programs of osteogenic differentiation in the developing palate by inhibiting bone morphogenetic protein (BMP) signaling dependent osteoblast markers. Expression of osteoblast markers, including *Runx2*, *Sp7,* and *AlpI* were increased in *Hoxa2*−/<sup>−</sup> palatal shelves at embryonic day (E) 13.5 and E15.5. *Hoxa2*−/<sup>−</sup> mouse embryonic palatal mesenchyme (MEPM) cells exhibited increased bone matrix deposition and mineralization *in vitro*. Moreover, loss of *Hoxa2* resulted in increased osteoprogenitor cell proliferation and osteogenic commitment during early stages of palate development at E13.5. Consistent with upregulation of osteoblast markers, *Hoxa2*−/<sup>−</sup> palatal shelves displayed higher expression of canonical BMP signaling *in vivo*. Blocking BMP signaling in *Hoxa2*−/<sup>−</sup> primary MEPM cells using dorsomorphin restored cell proliferation and osteogenic differentiation to wild-type levels. Collectively, these data demonstrate for the first time that *Hoxa2* may regulate palate development by inhibiting osteogenic differentiation of palatal mesenchyme via modulating BMP signaling.

Keywords: Hoxa2, cleft palate, bone morphogenetic protein (BMP), osteoblast, osteoprogenitor, proliferation, RUNX2

# INTRODUCTION

Cleft palate is one of the most common structural birth defects in humans with an incidence of 1 in 700–1,000 live births (Dixon et al., 2011). Studies using mouse model which has a high similarity to human palate development helped to identify several key stages and cellular processes during palate formation (Yu et al., 2017). In mice, secondary palate development begins at embryonic day (E)11.5 and is completed with palatal fusion by E15.5 (Ferguson, 1988). During palate development, the

### Edited by:

*Thimios Mitsiadis, University of Zurich, Switzerland*

### Reviewed by:

*Alexandre Rezende Vieira, University of Pittsburgh, United States David Clouthier, University of Colorado Anschutz Medical Campus, United States*

> \*Correspondence: *Paul P. R. Iyyanar paul.iyyanar@usask.ca*

*†Deceased.*

### Specialty section:

*This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology*

Received: *01 September 2017* Accepted: *02 November 2017* Published: *14 November 2017*

### Citation:

*Iyyanar PPR and Nazarali AJ (2017) Hoxa2 Inhibits Bone Morphogenetic Protein Signaling during Osteogenic Differentiation of the Palatal Mesenchyme. Front. Physiol. 8:929. doi: 10.3389/fphys.2017.00929*

**106**

**Abbreviations:** AlpI, alkaline phosphatase I; ARS, Alizarin red S; BMP, bone morphogenetic protein; CNCC, cranial neural crest cells; d, day; E, embryonic day; MEPM, mouse embryonic palatal mesenchyme; pSMAD 1/5/8, phosphorylated SMAD 1/5/8; qRT-PCR, quantitative real-time PCR; Runx2, runt-related transcription factor 2.

vertical palatal shelves grow downward along the sides of the tongue until E13.5 and then elevate above the tongue at E14. The palatal shelves on either side contact each other forming midline epithelial seam at E14.5, which eventually disintegrates leading to palatal fusion by E15.5 (Kaufman, 1992). Impairment in any of these distinct stages during palatogenesis may result in cleft palate. The palate is comprised of the palatal process of the maxilla and the palatal process of the palatine bone derived from the cranial neural crest cells (CNCC) (Iwata et al., 2010), constituting the anterior and posterior part of the hard palate, respectively (Baek et al., 2011). While the structural changes during palate development are well defined, there is a scarcity of knowledge on the molecular mechanisms governing the patterning of the palate.

In murine models, deletion of about 280 genes are known to cause cleft palate (Funato et al., 2015). Among these genes, mutations of 55 genes are associated with cleft palate in humans (Funato et al., 2015). Mutation in the Hoxa2 gene is associated with cleft palate in humans (Alasti et al., 2008) and mouse models (Gendron-Maguire et al., 1993; Rijli et al., 1993). In Hoxa2−/<sup>−</sup> mice, the cleft palate phenotype was initially attributed to the physical obstruction of the tongue preventing the palatal shelves to elevate and fuse (Barrow and Capecchi, 1999). However, our group has previously demonstrated that Hoxa2 is expressed in the palatal shelves (Nazarali et al., 2000) and plays an intrinsic role in palatogenesis (Smith et al., 2009). The palatal shelves from Hoxa2−/<sup>−</sup> mouse embryonic maxilla devoid of tongue grown in rolling bottle cultures failed to fuse (Smith et al., 2009). Hence, tongue musculature may not be the principal reason for the cleft palate phenotype in Hoxa2−/<sup>−</sup> mice. Hoxa2 appears to be a key regulator of palatogenesis, yet the molecular signaling pathways downstream of Hoxa2 remain largely unknown.

Bone morphogenetic protein (BMP) signaling plays a critical role in palate development regulating cell proliferation (Zhang et al., 2002). Bmp4 is upstream of Bmp2 to induce proliferation in the palatal mesenchyme and is able to reverse the reduced cell proliferation and cleft palate phenotype in the Msx1−/<sup>−</sup> mice (Zhang et al., 2002). Defective cell proliferation observed in Pax9−/<sup>−</sup> embryos is consistent with the reduced Bmp4 expression in the palatal mesenchyme at E13.5 (Zhou et al., 2013). Similarly, reduced expression of Bmp2 is associated with reduced cell proliferation in the palatal shelves of Hand2 hypomorphic mice (Hand2LoxP/−) (Xiong et al., 2009). In addition, growing evidence highlight the importance of osteogenic differentiation in the elevation of palatal shelves and abnormal osteogenic differentiation could lead to cleft palate manifestations (Wu et al., 2008; Fu et al., 2017; Jia et al., 2017a,b). BMP signaling is critical for osteogenic differentiation in the palatal mesenchyme (Wu et al., 2008; Baek et al., 2011; Hill et al., 2014), where it is required for the expression of osteoblast markers such as Runx2, Sp7, and AlpI (Baek et al., 2011). During craniofacial development, Hoxa2 restricts the bone mineralization in the calvaria (Dobreva et al., 2006). Moreover, Hoxa2−/<sup>−</sup> mice exhibit ectopic Runx2-positive osteogenic center in the second pharyngeal arch that results in duplication of tympanic ring (Kanzler et al., 1998).

In this study, we tested the hypothesis that Hoxa2 inhibits osteogenic differentiation of the palatal mesenchyme in vivo and in vitro using Hoxa2−/<sup>−</sup> mice. Our findings reveal that Hoxa2 plays a critical role in the spatial and temporal regulation of osteogenic differentiation via modulating BMP signaling pathway in the developing palate.

# MATERIALS AND METHODS

# Animals

Wild-type and Hoxa2−/<sup>−</sup> embryos were obtained from timed pregnant Hoxa2+/<sup>−</sup> (heterozygous) mice. Genotype was confirmed using PCR as previously described (Gendron-Maguire et al., 1993). This research was approved by the University of Saskatchewan's Animal Research Ethics Board and adhered to the Canadian Council on Animal Care guidelines for humane animal use.

# Primary Mouse Embryonic Palatal Mesenchyme (MEPM) Cell Culture and Osteogenic Induction

Primary MEPM cells were isolated from micro-dissected palatal shelves of wild-type and Hoxa2−/<sup>−</sup> mouse embryos at E13.5. The palatal shelves were treated with 0.25% trypsin-EDTA for 15 min, passed through a 70µm cell strainer and cultured as monolayer cells (Iwata et al., 2012) in DMEM: Ham's F12 (1:1) media containing 10% FBS, 1% antibiotic-antimycotic solution (Sigma). Osteogenic differentiation was carried out as described previously (Kwong et al., 2008) with minor modifications. Briefly, MEPM cells were seeded on 0.1% gelatin or poly-D-lysine coated plates at a cell density of 5 × 10<sup>4</sup> cells per well in 24 well plates and cultured until they reached confluence. Osteogenic differentiation was induced with differentiation media (DMEM, 10% FBS, 2 mM L-glutamine and 1% antibiotic-antimycotic solution) supplemented with osteogenic inducing agents, including 50µg/ml L-ascorbic acid 2-phosphate sesquimagnesium salt (Sigma), 10 mM βglycerophosphate (Sigma), and 100 nM dexamethasone (Sigma). Cells were differentiated for up to 21 days (d) and samples were collected at d8, d15, or d21. To assess the impact of BMP signaling, MEPM cells were treated with dorsomorphin (5µM) or DMSO and were harvested at d8 for further experiments.

# Alkaline Phosphatase I (ALPI) Staining

ALPI staining in the palatal shelves in vivo was carried out as previously described (Baek et al., 2011). Embryonic mouse heads were fixed overnight in freshly prepared 4% paraformaldehyde at 4◦C and rehydrated in 30% sucrose at 4◦C. Frozen coronal sections (10µm) were prepared on slides coated with 0.5% gelatin. The sections were air dried for at least 2 h and then rehydrated with TBS with 0.08% tween-20 two times for 10 min each. Subsequently, the sections were treated with alkaline phosphatase buffer (100 mM NaCl, 100 mM Tris-HCl pH 9.5, 50 mM MgCl2, 0.1% Tween-20) for 20 min and stained with alkaline phosphatase buffer containing 4.5 µl/ml of 5-bromo-4-chloro-3-indolyl phosphate (Roche) and 3.5 µl/ml of nitro blue tetrazolium (Roche) for 10 min. The reaction was stopped with PBS containing 20 mM EDTA buffer and counter stained with nuclear fast red. The stained sections were dehydrated in a series of PBS/ethanol, ethanol/xylene and finally mounted in DPX mounting media (Sigma). For osteoblast differentiation in primary MEPM cells, ALPI staining was carried out following the aforementioned protocol after fixing the cells with 4% paraformaldehyde for 15 min.

# Alizarin Red S (ARS) Staining and Quantification

ARS staining and quantification was carried out as previously described (Gregory et al., 2004) with minor modifications. Briefly, monolayer MEPM cells were fixed with 4% paraformaldehyde for 15 min and stained with 250 µl of 40 mM ARS (Sigma) solution (pH 4.1) at room temperature for 20 min with gentle shaking. Excess dye was aspirated and washed with deionized water before imaging. ARS quantification was carried using an acid extraction method (Gregory et al., 2004). Standard plot of ARS concentration was constructed by serially diluting the 40 mM ARS in the buffer containing 10% (v/v) acetic acid and 10% (v/v) ammonium hydroxide. The absorbance values of the standard concentrations were used to interpolate the concentrations of the test samples.

# Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)

Total RNA was isolated from the micro-dissected palatal shelves using RNA mini spin column (Bio-Rad) as per the manufacturer's protocol. First strand complementary DNA synthesis (reverse transcription) was performed in 20 µl reactions with 500 ng of total RNA using High-Capacity complementary DNA Reverse Transcription Kit (Invitrogen). qRT-PCR was carried out as described in our previous study (Thangaraj et al., 2017) using SYBR green assay (Applied Biosystems) in 7300-real time PCR system (Applied Biosystems) with primers listed in **Table 1**.

# Western Blotting

Western bot analyses were carried out as previously described (Brown and Nazarali, 2010). Briefly, palatal tissues were homogenized in RIPA buffer. Total protein content was quantified using the Bradford assay and proteins were separated in 10% SDS-PAGE. Primary antibodies used were: RUNX2 (1:500; Abcam ab102711), SP7 (1:1500; Abcam ab22552), phosphorylated SMAD 1/5/8 (pSMAD 1/5/8) (1:500; Cell signaling 9511S), SMAD 1/5/8 (1:500; Santa Cruz Biotechnology sc-6031-R), and β-ACTIN (1:2,000; Developmental Studies Hybridoma JLA20). Densitometric analyses were carried out using AlphaView software.

# Immunohistochemistry

Embryonic mouse heads were fixed overnight with freshlyprepared 4% paraformaldehyde and rehydrated in 30% sucrose at 4◦C. Frozen coronal sections (10µm) were rehydrated with PBS for 45 min, permeabilized with 0.1% Triton X-100 and blocked with 3% skim milk containing 0.1% Triton X-100 in 1X PBS for 1 h at room temperature. Sections were then incubated overnight with the following primary antibodies: RUNX2 (1:200; Abcam ab23981) or SP7 (1:800; Abcam ab22552) in 1X PBS with 0.1% Triton X-100 at 4◦C. Double labeling was carried TABLE 1 | Primer sequences used for the relative quantification of the transcripts by qRT-PCR.


out by co-incubating: Ki67 (1:100; eBioscience 14569882) and RUNX2 (1:200; Abcam ab23981) overnight at 4◦C. Sections were then washed three times and treated with secondary antibodies conjugated with Alexa Fluor <sup>R</sup> 488 (1:200) or Alexa Fluor <sup>R</sup> 594 (1:400) in 1X PBS with 0.1% Triton X-100 at room temperature for 1.5 h. Cell counting analyses were carried out manually using ImageJ software platform (NIH).

# Cell Proliferation Assay

Cell proliferation assay was carried out in MEPM cells using cellcounting kit-8 (Dojindo) as previously described (Iwata et al., 2010). MEPM cells were incubated with CCK-8 reagent for 1 h and the absorbance measured at 450 nm was plotted to calculate the relative cell proliferation rate.

# Statistical Analyses

Statistical analyses were carried out using unpaired t-test in the case of two groups. One-way ANOVA or two-way ANOVA with Bonferroni multiple comparison test was used for one or two variate analyses, respectively. A p-value of <0.05 was considered significant.

# RESULTS

# Hoxa2−/<sup>−</sup> Mice Exhibit Increased Expression of Osteoblast Markers during Palate Development in Vivo

To investigate the role of Hoxa2 in osteogenesis of the palatal mesenchyme, we first examined changes in osteogenic differentiation in the embryonic palatal shelves of Hoxa2−/<sup>−</sup> mice at E16.5, a stage when both the prospective palatal process of the maxilla and the palatal process of the palatine bone evidently ossify (Baek et al., 2011). Staining for ALPI, a marker of osteoblast differentiation showed an expansion in ALPI expression domain in the Hoxa2−/<sup>−</sup> palatal mesenchyme compared to wild-type (**Figures 1A–H**). At the anterior region of the hard palate, ALPI staining was restricted to the nasal half in two condensations of the prospective palatal process of the maxilla on either side of the degraded midline epithelial seam in wild-type embryos (**Figures 1A,E**). In contrast, the domain of ALPI positive preosteoblast area was increased and expanded toward the oral side covering oral-nasal axis in Hoxa2−/<sup>−</sup> palatal shelves (**Figures 1B,F**). In the posterior region of the hard palate, ALPI staining was present in the ossifying centers of the palatal process of the palatine bone in wild-type embryos (**Figures 1C,G**), whereas there was an expansion in the expression domain of ALPI positive preosteoblasts toward the oral side in Hoxa2−/<sup>−</sup> embryos (**Figures 1D,H**).

Two well-known regulators of osteogenic differentiation RUNX2 (Komori et al., 1997) and SP7 (previously known as Osterix; Nakashima et al., 2002) have been implicated in the patterning of the palatal bones (Baek et al., 2011).

FIGURE 1 | Loss of *Hoxa2* leads to increased osteogenic differentiation of the palatal mesenchyme at E16.5. Position matched coronal sections of wild-type and *Hoxa2*−/<sup>−</sup> embryos at E16.5 were stained for ALPI (A–H), RUNX2 (I–L), and SP7 (M–P). Sections in the anterior region (A,B,E,F,I,J,M,N) were through the middle of the first molar tooth bud to detect osteogenic condensation of the palatal process of the maxilla. Sections in the posterior region (C,D,G,H,K,L,O,P) were through the osteogenic centers of the developing palatal process of the palatine bone. (A–D) ALPI staining (blue) counterstained with nuclear fast red. Scale bar, 100µm. Boxed regions in (A–D) highlighting the palate are enlarged (E–H). Scale bar, 50µm. (E,F) In the anterior hard palate, ALPI staining in the two condensations of the palatal process of the maxilla (marked in black dotted lines) was evidently increased in the *Hoxa2*−/<sup>−</sup> embryos (F) compared to wild-type (E). (G,H) In the posterior hard palate, ALPI stained developing palatal process of the palatine bone (marked in black dotted lines) in the *Hoxa2*−/<sup>−</sup> embryos (H) was increased compared to the wild-type (G), *<sup>n</sup>* <sup>=</sup> 5 biological replicates. (I–P) Immunohistochemical analyses of RUNX2 (green; I–L) and SP7 (red; M–P) in wild-type and *Hoxa2*−/<sup>−</sup> palate at E16.5. RUNX2 was increased in both anterior (J) and posterior regions (L) of the *Hoxa2* null hard palate, whereas SP7 was increased only in the anterior hard palate (N), *n* = 4 biological replicates. Scale bar, 50µm. M1, first molar; Mb, mandible; Mx, maxilla; NS, nasal septum; pppb, the palatal process of the palatine bone; ppmx, the palatal process of the maxilla; T, tongue.

To elucidate the spatial mis-regulation of palatal bone formation in Hoxa2−/<sup>−</sup> mice, expression pattern of these two osteoblast-specific transcription factors were assessed at E16.5. Immunohistochemical analyses revealed that RUNX2 (**Figure 1I**) and SP7 (**Figure 1M**) expressions were confined to the condensations of the palatal process of the maxilla at the anterior hard palate in wild-type embryos at E16.5, whereas RUNX2 (**Figure 1J**) and SP7 (**Figure 1N**) expression domains were increased in Hoxa2−/<sup>−</sup> embryos. In the posterior region, along the developing palatal process of the palatine bone, the expression of RUNX2 (**Figure 1L**) was increased in Hoxa2−/<sup>−</sup> embryos compared to wild-type (**Figure 1K**). In this region, the expression of SP7, a downstream target of RUNX2 and a marker of mature osteoblasts, was not evidently increased in the Hoxa2−/<sup>−</sup> palate (**Figure 1P**) compared to wild-type (**Figure 1O**). This suggests that cells toward the oral side of the palatal process of the palatine bone are at immature osteoblast stage and may not have developed bone matrix by E16.5. Collectively, these data indicate that Hoxa2 could be a potential inhibitor of osteogenic differentiation in the palatal mesenchyme, which may serve to spatially restrict the expression of osteoblast-specific proteins during palate development in vivo.

Furthermore, gene expression profiles of osteoblast markers were evaluated during the initiation of ossification of the palatal process of the palatal bone and the palatal process of the maxilla at E13.5 and E15.5, respectively. The loss of Hoxa2 in the developing palate resulted in an increase in mRNA expression of osteoblast markers such as Runx2, AlpI and Sp7 at both E13.5 (**Figures 2A–D**) and E15.5 (**Figures 2E–H**). At E13.5, mRNA expression of Runx2, AlpI and Sp7 were increased to ∼6.36-, ∼9.65-, and ∼2.62-fold, respectively, in Hoxa2−/<sup>−</sup> palate compared to wild-type (**Figures 2A–C**). At E13.5, mRNA expression of Bglap (previously known as Ocn) was not significantly altered (**Figure 2D**). At E15.5, mRNA expression of Runx2, AlpI, Sp7, and Bglap were upregulated ∼1.86-, ∼2.29-, ∼1.42-, ∼3.27-folds, respectively, in Hoxa2−/<sup>−</sup> palate compared to wild-type (**Figures 2E–H**). Consistent with this, protein expression of RUNX2 was upregulated at E13.5 and E15.5 to ∼1.4-fold (**Figures 2I,J**). SP7 protein expression, both long (**Figures 2I,K**) and short isoforms (**Figures 2I,L**) were upregulated at E15.5 to ∼1.4-fold. These data reveal that along with regulating the spatial patterning of osteogenic differentiation, Hoxa2 also regulates the expression of osteogenic markers at the molecular level in the developing palate.

# Hoxa2 Inhibits Osteoblast Differentiation of Mouse Embryonic Palatal Mesenchymal (MEPM) Cells in Vitro

To evaluate the potential of Hoxa2 in regulating the temporal pattern of osteogenesis, the primary mesenchyme cells from the wild-type and Hoxa2−/<sup>−</sup> palatal shelves were differentiated in vitro for up to 21 days (d). Osteogenesis of mesenchymal cells involves sequential stages of proliferation, osteogenic commitment around day8 (∼d8) followed by matrix deposition (∼d15) and mineralization (∼d21) (Gordon et al., 2010). Differentiating cells were stained for ALPI at d8 and Alizarin Red S (ARS) at d15 and d21. ALPI staining showed an increased osteoblast differentiation at d8 in Hoxa2−/<sup>−</sup> MEPM cells compared to the wild-type MEPM cells (**Figures 3A,B**). In addition, ARS staining followed by quantification of ARS extracted matrix showed that Hoxa2−/<sup>−</sup> MEPM cells exhibited increased extracellular matrix deposition ∼2-fold at d15 (**Figures 3C,D,G**) and increased mineralization ∼1.5-fold at d21 (**Figures 3E,F,H**) compared to the wild-type MEPM cells.

Next, the gene expression profiles of osteogenic markers were examined in wild-type and Hoxa2−/<sup>−</sup> MEPM cells during osteogenic differentiation in vitro. Runx2 mRNA expression was increased to ∼1.9-fold in the Hoxa2−/<sup>−</sup> MEPM cells compared to the wild-type during osteoblast commitment stage at d8 (**Figure 3I**). AlpI and Sp7 mRNA expression were increased ∼2.85 and ∼3.37-fold, respectively, during matrix deposition stage at d15 in the Hoxa2−/<sup>−</sup> MEPM cells (**Figures 3J,K**). Bglap mRNA expression was increased ∼6.37-fold during matrix deposition at d15 and ∼8.09-fold during matrix mineralization at d21 in the Hoxa2−/<sup>−</sup> MEPM cells (**Figure 3L**). Thus, loss of Hoxa2 results in upregulation of osteogenic marker expression in a stage-specific manner as early as d8 (osteogenic commitment stage). These data indicate that Hoxa2 may play a role in early osteoblast differentiation by inhibiting the transcription factors regulating osteogenic fate specification.

# Increased Osteoprogenitor Proliferation and Commitment in the Hoxa2−/<sup>−</sup> Palatal Mesenchyme during Early Palate Development

Hoxa2 peaks in its expression in the developing palate at E13.5 (Smith et al., 2009), a stage when the mesenchymal cells simultaneously proliferate and commit to form preosteoblasts of the prospective palatal process of the palatine bone. This suggests that the cleft palate phenotype in Hoxa2−/<sup>−</sup> mice, due to the failure of palatal shelves to elevate and reorient horizontally above the tongue after E13.5 (Barrow and Capecchi, 1999), may be a consequence of abnormal cell proliferation (Smith et al., 2013) and osteogenic differentiation (Wu et al., 2008; Fu et al., 2017; Jia et al., 2017a,b). To gain further insight into the role of Hoxa2 during this early stage of palate development, the rate of mesenchymal cell proliferation and the commitment of mesenchymal cells to osteoprogenitor fate was investigated in vivo at E13.5. Immunohistochemical staining of RUNX2 (**Figures 4A,B**) was used to evaluate osteoprogenitor commitment in the wild-type and Hoxa2−/<sup>−</sup> palatal shelves at E13.5. RUNX2 (**Figure 4A**) expression in the wild-type was restricted to the bend region in the nasal side of the palatal shelves, whereas the expression domain of RUNX2 (**Figure 4B**) was increased spatially toward the medial edge of the palate as well as to the oral side of the palate in the Hoxa2−/<sup>−</sup> mutants. This is similar to the aberrant expression patterns of RUNX2 observed at E16.5 in the Hoxa2−/<sup>−</sup> palatal shelves. In addition, the number of RUNX2-positive osteoprogenitor cells on the nasal side of the palatal shelves were significantly higher in

for up to 21 days (d). The differentiated cells were stained for ALPI at d8 (A,B), ARS at d15 (C,D), and d21 (E,F). ARS stained osteocyte matrices from wild-type and *Hoxa2*−/<sup>−</sup> MEPM cells were extracted and quantified at d15 (G) and d21 (H). Experiment was carried out three times and the data shown here are from a representative experiment with *<sup>n</sup>* <sup>=</sup> 3 biological replicates; mean <sup>±</sup> S.E.M; unpaired *<sup>t</sup>*-test, \**<sup>p</sup>* <sup>&</sup>lt; 0.05; \*\*\**<sup>p</sup>* <sup>&</sup>lt; 0.001). *Hoxa2*−/<sup>−</sup> MEPM cells displayed increased matrix deposition and mineralization at d15 (G) and d21 (H), respectively. (I–L) qRT-PCR analyses revealed that gene expression profile of osteogenic markers such as *Runx2* (I), *AlpI* (J), *Sp7* (K), and *Bglap* (L) were upregulated in the *Hoxa2*−/<sup>−</sup> MEPM cells in a stage-specific manner during osteoblast differentiation. Data was normalized to β*-actin* and represented relative to wild-type at d0 (*n* = 3 biological replicates; mean ± S.E.M; two-way ANOVA with Bonferroni post-hoc test, \**p* < 0.05; \*\**p* < 0.01; \*\*\**p* < 0.001).

the Hoxa2−/<sup>−</sup> mutants (∼64%) compared to wild-type (∼33%; **Figure 4C**).

Next, the rate of cell proliferation was assessed at E13.5 using Ki67 immunostaining. The percentage of Ki67-positive cells was significantly increased in the Hoxa2−/<sup>−</sup> palatal mesenchyme (∼50%) compared to wild-type (∼26%; **Figures 4D–F**). In the nasal side of the palatal shelves, the percentage of Ki67-positive cells was ∼53% in Hoxa2−/<sup>−</sup> embryos compared to ∼26% in the wild-type (**Figure 4F**). Interestingly, the nasal side mesenchyme displayed a higher proliferation rate of ∼53% compared to the oral side of ∼23% in Hoxa2−/<sup>−</sup> palatal shelves (**Figure 4F**). In addition, the percentage of proliferating osteoprogenitor cells (RUNX2-positive/Ki67-positive) in the nasal side of the Hoxa2−/<sup>−</sup> palatal shelves was higher (∼20%) compared to ∼11% in wild-type (**Figures 4G–I**). Furthermore, mRNA expression of cyclin D1 (Ccnd1), a critical G1 phase cell cycle regulator was also upregulated in the Hoxa2−/<sup>−</sup> palatal shelves from E12.5 to E14.5 (**Figure 4J**). These results indicate that Hoxa2 plays a critical role by inhibiting osteoprogenitor commitment and osteoprogenitor proliferation prior to the elevation and fusion of the palatal shelves.

# Increased Canonical BMP Signaling Pathway in the Hoxa2−/<sup>−</sup> Palatal Shelves

To understand the molecular signaling pathways underlying the aberrant cell proliferation and osteogenic differentiation in the Hoxa2−/<sup>−</sup> palatal shelves, BMP signaling was investigated as it is critical for cell proliferation (Zhang et al., 2002) and expression of osteogenic markers in the developing palate (Baek et al., 2011). First, the mRNA expression of BMP ligands critical for osteoblast differentiation such as Bmp2 and Bmp4 in the developing palatal shelves was examined. Bmp2 expression was upregulated to ∼3.57-fold at E13.5 and ∼1.96-fold at E15.5 in Hoxa2−/<sup>−</sup> palatal shelves compared to wild-type (**Figure 5A**). Similarly, Bmp4 expression was upregulated to ∼3.42-fold at E13.5 and to ∼1.81-fold at E15.5

(**Figure 5B**). Immunoblotting analyses revealed that canonical BMP signaling mediated by pSMAD 1/5/8 was also upregulated to ∼1.5-fold in the Hoxa2−/<sup>−</sup> palate at E15.5 (**Figures 5C,D**). These results indicate that canonical BMP signaling pathway may be downstream of the Hoxa2 gene network in palate development.

# Blocking Canonical BMP Signaling Rescues the Aberrant Cell Proliferation and Osteogenic Differentiation in Hoxa2−/<sup>−</sup> MEPM Cells

To determine if the upregulated canonical BMP signaling is functionally responsible for the increased mesenchymal cell proliferation and osteogenic differentiation observed in the Hoxa2−/<sup>−</sup> palate, dorsomorphin was used to inhibit BMP signaling during osteogenic differentiation of MEPM cells in vitro. Although at higher doses dorsomorphin (10–20µM) inhibits AMPK signaling (Zhou et al., 2001) and mTOR signaling (Vucicevic et al., 2011), it selectively inhibits BMP signaling at lower doses (Yu et al., 2008). Upon 5µM dorsomorphin treatment, upregulated mRNA expressions of Bmp2 (**Figure 6A**), Bmp4 (**Figure 6B**) and Runx2 (**Figure 6C**) in the Hoxa2−/<sup>−</sup> MEPM cells were restored to the wild-type levels. Moreover, increased protein expression of RUNX2 and pSMAD 1/5/8 (**Figure 6D**) in the Hoxa2−/<sup>−</sup> MEPM cells were reduced after dorsomorphin treatment. The increased cell proliferation (**Figure 6E**) and osteogenic differentiation (**Figure 6F**) in the Hoxa2−/<sup>−</sup> MEPM cells were also reduced after dorsomorphin treatment. These results indicate that the upregulated canonical BMP signaling is functionally responsible for the increased cell proliferation and osteogenic differentiation during palate development in Hoxa2−/<sup>−</sup> embryos. Altogether, the findings reveal that Hoxa2 inhibits osteoprogenitor proliferation and commitment, via BMP signaling, to control the spatial and temporal expression of osteoblast markers for proper palatogenesis.

# DISCUSSION

Mice lacking Hoxa2 exhibit cleft palate (Gendron-Maguire et al., 1993; Rijli et al., 1993; Barrow and Capecchi, 1999) and microtia (Minoux et al., 2013), which are consistent with Hoxa2 mutations in humans (Alasti et al., 2008). We have previously shown that Hoxa2 is expressed in the palatal shelves during development (Nazarali et al., 2000) reaching a maximal expression at E13.5 and regulates cell proliferation in the

developing palate (Smith et al., 2009). There are several lines of evidence that Hoxa2 regulates palate development intrinsically (Smith et al., 2009), yet the mechanism is largely unknown. In this study, we have found that Hoxa2 inhibits BMP signaling dependent osteogenic differentiation spatially and temporally to regulate palate formation. The present study deepens the current understanding of the role of Hoxa2 in palate formation and the mechanisms underlying the cleft palate phenotype in Hoxa2−/<sup>−</sup> mice linking Hoxa2, BMP signaling and osteogenesis.

Our findings here reveal that Hoxa2 controls the temporal and spatial expression pattern of osteoblast markers in the developing palatal mesenchyme. Ossifying domains characterized by RUNX2 and ALPI were increased in the palatal process of the maxilla and in the palatal process of the palatine bone in Hoxa2−/<sup>−</sup> mice. In contrast, SP7 a marker of mature osteoblasts was expanded only in the palatal process of the maxilla and not in the palatal process of the palatine bone at E16.5. This suggests that cells toward the oral side of the palatal process of the palatine bone are at immature osteoblast stage and may not have developed bone matrix by E16.5. Patterning of the palatal process of the palatine bone and of the maxilla are through independent skeletogenic processes (Baek et al., 2011). The palatal process of the palatine bone ossifies at E13.5, whereas the ossification of the palatal process of the maxilla begins only at E15.5. Consistent with this, qRT-PCR and immunoblot analyses revealed a corresponding upregulation of osteogenic markers in the Hoxa2−/<sup>−</sup> palate at these two critical stages E13.5 and E15.5. In addition, primary Hoxa2−/<sup>−</sup> MEPM cells displayed an increase in osteogenic differentiation and a stage-specific increase in the expressions of the osteoblast-specific transcripts indicating that Hoxa2 regulates temporal differentiation of mesenchyme cells to osteoblasts in the palate. Together, our results reveal that Hoxa2 functions as an inhibitor of osteogenic differentiation in the palatal mesenchyme during development. Our findings are in agreement with previous studies showing the role of Hoxa2 as an inhibitor of bone formation in other craniofacial regions (Kanzler et al., 1998; Dobreva et al., 2006).

Very little is known about the signaling network downstream of Hoxa2 during palatogenesis. Here, we have demonstrated that Hoxa2−/<sup>−</sup> palatal shelves exhibit upregulated canonical

BMP signaling in the developing palate, which in turn restricts the expression domain of osteogenic markers such as *Runx2*, *AlpI,* and *Sp7*. (B) In wild-type, *Hoxa2* expression peaks during early palatogenesis to control cell proliferation and to maintain mesenchymal cells in an undifferentiated stage by regulating BMP signaling pathway. (C) Loss of *Hoxa2* leads to upregulation of BMP signaling resulting in increased osteoprogenitor cell proliferation and osteogenic differentiation, possibly accounting for the failure in the elevation of palatal shelves resulting in manifestation of cleft palate.

BMP signaling mediated by pSMAD 1/5/8. In addition, the expression of BMP ligands such as Bmp2 and Bmp4 are upregulated in Hoxa2−/<sup>−</sup> palatal shelves in vivo and in Hoxa2−/<sup>−</sup> MEPM cells in vitro. BMP signaling plays a critical role in proliferation (Zhang et al., 2002; Baek et al., 2011) and osteogenic differentiation of the palatal mesenchyme (Wu et al., 2008; Han et al., 2009; Baek et al., 2011). Importantly, abnormal BMP signaling in the palatal mesenchyme leads to cleft palate manifestation (Zhang et al., 2002; He et al., 2008). Inactivation of Bmpr1a in the palatal mesenchyme (Osr2-IresCre; Bmpr1a<sup>f</sup> /<sup>f</sup> ) results in submucous cleft palate, absence in the patterning of the palatal process of the maxilla and defective palatal process of the palatine bone (Baek et al., 2011). Genome-wide mapping revealed that Bmp2, Bmp4 and Bmpr1a are possible targets of Hoxa2 (Donaldson et al., 2012) and HOXA2 protein binds to the intronic region of Bmp4 (Minoux et al., 2013) in the developing pharyngeal arch2. In this study, dorsomorphin was used to inhibit BMP signaling in the wild-type and Hoxa2−/<sup>−</sup> primary palatal mesenchymal cells during osteogenic differentiation. Dorsomorphin selectively inhibits BMP signaling at lower doses (Yu et al., 2008) and at higher doses dorsomorphin (10–20µM) also inhibits AMPK signaling (Zhou et al., 2001) and mTOR signaling (Vucicevic et al., 2011). In our study, dorsomorphin treatment not only rescued the upregulated gene expression of osteogenic factors such as Bmp2, Bmp4, and Runx2 but also the aberrant cell proliferation and osteogenic differentiation in the Hoxa2−/<sup>−</sup> MEPM cells. These experiments highlight the involvement of BMP signaling in the abnormal osteoprogenitor cell proliferation and osteogenic differentiation in the Hoxa2−/<sup>−</sup> palate, which could attribute to the cleft palate phenotype in these mutants.

To our knowledge, there is no report available on the characterization of osteoprogenitor cell proliferation and commitment in the palatal mesenchyme during development. In this study, we have unraveled the role of Hoxa2 in maintaining the palatal mesenchymal cells in an undifferentiated stage by inhibiting osteoprogenitor proliferation and commitment preventing abnormal ossification in the developing palate. Palatal mesenchymal cells derived from CNCC undergo osteogenic proliferation and commit to form osteoblasts (Iwata et al., 2010). Double immunolabeling analyses of RUNX2 and Ki67 at E13.5 revealed that among the total population of mesenchyme cells, there was a significantly higher number of (i) proliferating cells (Ki67-positive cells), (ii) osteoprogenitor cells (RUNX2 positive cells), and (iii) proliferating osteoprogenitor cells (RUNX2-positive /Ki67-positive cells) in the nasal side of the Hoxa2−/<sup>−</sup> palatal shelves compared to the wild-type. In the palatal mesenchyme, increased or decreased cell proliferation could result in failure of the palatal shelves to elevate and reorient above the tongue leading to cleft palate (Bush and Jiang, 2012; Smith et al., 2013). Recent studies show evidence for abnormal osteogenic signaling prior to the elevation of palatal shelves in several well-studied cleft palate mutant mice models including Pax9−/<sup>−</sup> mice (Jia et al., 2017a,b) and Osr2−/<sup>−</sup> mice (Fu et al., 2017). Consistent with our findings here in the Hoxa2−/<sup>−</sup> mice, Osr2−/<sup>−</sup> exhibit increased osteogenic centers of the palatal process of the palatine bone prior to the elevation of the palatal shelves at E13.5 and in addition to defective cell proliferation, enhanced osteogenesis could contribute to cleft palate phenotype in Osr2−/<sup>−</sup> mice (Fu et al., 2017). In addition, RNA-Seq data from Osr2−/<sup>−</sup> palatal shelves revealed upregulation of several positive regulators of osteogenesis including Runx2, Runx3, Sp7, and Bmp ligands- Bmp3, Bmp5, and Bmp7. Furthermore, Pax9−/<sup>−</sup> mice exhibit reduced cell proliferation and osteogenesis in the developing palate (Jia et al., 2017a). Restoration of reduced cell proliferation and osteogenesis by Wnt agonists (Dkk inhibitors) rescued the cleft palate phenotype in Pax9−/<sup>−</sup> mice (Jia et al., 2017a). The increase in cell proliferation in the nasal side of the Hoxa2−/−palate indicates a strong role for Hoxa2 in the spatial maintenance of mesenchymal cells in an undifferentiated state for temporal coordination of osteoblast differentiation (**Figure 7**). Our findings here exemplify the regional heterogeneity in proliferation and osteogenic differentiation by Hoxa2 along the oral-nasal axis in the palatal mesenchyme prior to the elevation of palatal shelves. Our data argue that improper BMP signaling leading to the increased osteoprogenitor cell proliferation and commitment could be a reason for the cleft palate pathogenesis in the Hoxa2−/<sup>−</sup> mice. Further studies are needed to address if the cleft palate phenotype in the Hoxa2−/<sup>−</sup> mice could be rescued using other mutant mice with impaired osteogenesis.

Our data demonstrate that Hoxa2 inhibits osteoprogenitor cell proliferation and osteogenic commitment via modulating BMP signaling in the mouse embryonic palatal mesenchyme. Hoxa2 regulates spatial and temporal programs of osteogenesis by maintaining mesenchymal cells in an undifferentiated stage until osteogenic clues arrive. In conclusion, our findings provide new insights into the signaling mechanism underlying the role of Hoxa2 during embryonic palate development.

# AUTHOR CONTRIBUTIONS

AN conceived and coordinated the study. PI designed the study, performed experiments, analyzed data and wrote the manuscript. AN and PI proofed, revised the manuscript for critical content and interpretation of data.

# FUNDING

This work was supported by discovery grant 171317–2012 from the Natural Sciences and Engineering Research Council of Canada (NSERC) to AN and in part by Subtelny Orthodontic Clinical Research Grant from American Cleft Palate-Craniofacial Association (ACPA-CPF), USA to PI.

# ACKNOWLEDGMENTS

PI acknowledges the Apotex Graduate Scholarship Award from the College of Pharmacy and Nutrition, University of Saskatchewan; Graduate Research fellowship and Saskatchewan Innovation and Opportunity Scholarships from the College of Graduate and Postdoctoral studies, University of Saskatchewan. Authors thank Drs. Brian F. Eames and Kendra L. Furber for critical suggestions on the manuscript and Larhonda Sobchishin for providing technical expertise. This manuscript is dedicated to the memory of our colleague and mentor, Dr. Adil J. Nazarali, who passed away before the submission of the manuscript.

# REFERENCES


**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 © 2017 Iyyanar and Nazarali. 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) or licensor 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.

# Multiple Cranial Organ Defects aft Conditionally Knocking Out Fgf10 i the Neural Crest er n

### Tathyane H. N. Teshima<sup>1</sup> , Silvia V. Lourenco<sup>1</sup> and Abigail S. Tucker <sup>2</sup> \*

<sup>1</sup> Department of Stomatology, School of Dentistry, University of Sao Paulo, São Paulo, Brazil, <sup>2</sup> Department of Craniofacial Development and Stem Cell Biology, King's College London, London, UK

Fgf10 is necessary for the development of a number of organs that fail to develop or are reduced in size in the null mutant. Here we have knocked out Fgf10 specifically in the neural crest driven by Wnt1cre. The Wnt1creFgf10fl/fl mouse phenocopies many of the null mutant defects, including cleft palate, loss of salivary glands, and ocular glands, highlighting the neural crest origin of the Fgf10 expressing mesenchyme surrounding these organs. In contrast tissues such as the limbs and lungs, where Fgf10 is expressed by the surrounding mesoderm, were unaffected, as was the pituitary gland where Fgf10 is expressed by the neuroepithelium. The circumvallate papilla of the tongue formed but was hypoplastic in the conditional and Fgf10 null embryos, suggesting that other sources of FGF can compensate in development of this structure. The tracheal cartilage rings showed normal patterning in the conditional knockout, indicating that the source of Fgf10 for this tissue is mesodermal, which was confirmed using Wnt1cre-dtTom to lineage trace the boundary of the neural crest in this region. The thyroid, thymus, and parathyroid glands surrounding the trachea were present but hypoplastic in the conditional mutant, indicating that a neighboring source of mesodermal Fgf10 might be able to partially compensate for loss of neural crest derived Fgf10.

### Keywords: Fgf10, ocular glands, thyroid, palate, cranial glands, CVP

# INTRODUCTION

Fgf10 is an essential signaling molecule from the fibroblast growth factor family and is involved in the development of many organs, signaling through Fgfr2b in the epithelium (Ohuchi et al., 2000). Patients with mutations in one copy of the Fgf10 ligand or its receptor have Lacrimo Acoustic Dental Digital (LADD) syndrome (OMIM 149730) or the related Aplasia of Salivary and Lacrimal Gland (ASLG) syndrome (OMIM 180920), characterized by defects in a variety of cranial glands (Rohmann et al., 2006). Mice with a complete knockout of Fgf10 die at birth due to a complete lack of lungs and limbs and formation of a cleft palate (Min et al., 1998; Sekine et al., 1999; Rice et al., 2004). As in patients, loss of Fgf10 also impacts on the development of a number of cranial glands, with null mutants showing complete loss of the salivary glands, thyroid gland, pituitary gland (Ohuchi et al., 2000), ocular glands (Govindarajan et al., 2000; Makarenkova et al., 2000) and the circumvallate papilla (CVP) housing the Von Ebner's glands in the tongue (Petersen et al., 2011). The salivary glands have been shown to arrest at the prebud stage, with heterozygous mice showing a delay in development that leads to later gland hypofunction (Jaskoll et al., 2005; May et al., 2015). The pituitary gland starts to initiate in the Fgf10 null with an infolding of the oral

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Amy Elizabeth Merrill, University of Southern California, USA Claudio Cantù, University of Zurich, Switzerland Vesa Kaartinen, University of Michigan, USA

> \*Correspondence: Abigail S. Tucker abigail.tucker@kcl.ac.uk

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 27 July 2016 Accepted: 10 October 2016 Published: 25 October 2016

### Citation:

Teshima THN, Lourenco SV and Tucker AS (2016) Multiple Cranial Organ Defects after Conditionally Knocking Out Fgf10 in the Neural Crest. Front. Physiol. 7:488. doi: 10.3389/fphys.2016.00488

**119**

epithelium to form Rathke's pouch at the back of the mouth, but the ectoderm is associated with high levels of apoptosis and the pouch disappears by E13.5 (Ohuchi et al., 2000). Other glands form but are reduced in size, such as the thymus glands (Ohuchi et al., 2000; Revest et al., 2001), and more subtle defects are also observed in the inner ear, in the patterning of the trachea cartilage rings, in the teeth and in the skin and hair follicles (Ohuchi et al., 2000; Pauley et al., 2003; Sala et al., 2011). In these organs loss of Fgf10 may be compensated for by the presence of Fgf7 or Fgf3, which can both bind to the same receptor (Zhang et al., 2006). In keeping with this loss of both Fgf10 and Fgf3 leads to a more severe defect in the inner ear (Wright and Mansour, 2003), and knockout of the receptor Fgfr2b leads to additional defects not observed in Fgf10 knockouts, such as arrest of tooth development at the bud stage (De Moerlooze et al., 2000).

During development Fgf10 is expressed in the mesenchyme that surrounds many developing organs (lungs, limbs, ocular glands, palatal shelves, salivary glands (Bellusci et al., 1997; Moustakas et al., 2011; Wells et al., 2013). In contrast its receptor, Fgfr2b, is expressed in epithelial structures overlying these regions, emphasizing the importance of epithelial-mesenchymal interactions (Peters et al., 1992; Orr-Urtreger et al., 1993; Rice et al., 2004). In the developing brain Fgf10 is expressed in the infundibulum, which signals to the developing oral epithelium during pituitary gland development (Takuma et al., 1998). Early on during facial development Fgf10 is expressed in the oral epithelium of the first pharyngeal arch (Kettunen et al., 2000; Wells et al., 2013), with expression also observed in the tooth germ epithelium in some species (Moustakas et al., 2011). In the otic region Fgf10 is first expressed in the mesoderm derived mesenchyme around the otic epithelium at E8.75 and then in the otic cup and otic vesicle at E9 and E9.5 (Wright and Mansour, 2003). Fgf10 is therefore expressed in a range of tissues during development.

In this paper we have conditionally knocked out Fgf10 specifically in neural crest derived tissue using the Wnt1cre transgenic line (Chai et al., 2000). Previously a conditional knockout of Fgf10 has been carried out using Dermo1 cre, which led to specific loss of Fgf10 in the mesoderm around the developing lungs, resulting in lung branching defects (Abler et al., 2009). By knocking out Fgf10 in neural crest derived tissues only we aim to investigate which phenotypes in the null mutant are a consequence specifically of Fgf10 expression in the neural crest. A number of tissues in the head are known to be derived from the neural crest. These include the mesenchyme around the developing salivary glands, thyroid and thymus glands, teeth, and the palatine bone (Chai et al., 2000; Jiang et al., 2000; Jaskoll et al., 2002; Müller et al., 2008; Johansson et al., 2015). The Fgf10 expressing mesenchyme that underlies the forming CVP in the tongue is also neural crest derived (Hosokawa et al., 2010). The origin of the tissue around the developing ocular glands has not been confirmed as the developing eye is surrounded by neural crest derived mesenchyme and lateral plate mesoderm, which together forms the periocular mesenchyme (Langenberg et al., 2008). In contrast the limbs and lungs are surrounded by mesoderm and so would be predicted to develop normally in the conditional Wnt1cre Fgf10 mice. The pituitary would also be expected to be normal in these conditional mutants as the source of the Fgf10 is the neuroectodermal infundibulum (Takuma et al., 1998). In addition we compare the conditional knockout to the phenotype in the null Fgf10 mouse to clarify the role of neural crest derived Fgf10 in a variety of craniofacial tissues, and identify a few discrepancies with the published literature.

# MATERIALS AND METHODS

# Transgenic Mice

Fgf10 floxed (Fgf10A02 tmc1c) mice on a C57Bl6 background were produced by MRC-Harwell as part of the International Mouse Phenotyping Consortium (IMPC; Pettitt et al., 2009; Skarnes et al., 2011; Bradley et al., 2012). Fgf10fl/fl females were crossed to Wnt1cre/Fgf10 fl/+ males to generate Wnt1creFgf10fl/fl embryos (3 litters), collected at E14.5, E15.5, and E19.5 (E14.5 n = 3; E15.5 n = 3; E19.5 n = 2). These conditional mutants were compared to Fgf10fl/fl littermates that did not carry the cre and were therefore phenotypically wildtype. A total of 6 Fgf10 null embryos generated on a mixed C57Bl6/CD1 background (E14.5, E15.5, E18.5) were used to compare the conditional phenotype with that of the complete null.

Wnt1cre males were mated to tdTomato reporter females (Gt(ROSA)26 Sor tm14(CAG-tdTomato)Hze JAX labs) to lineage trace the neural crest and were viewed with a Nikon SMZ25 fluorescence microscope.

The Wnt1cre mouse is widely used for neural crest specific knockout studies, however, it has been linked to elevated levels of Wnt signaling in the midbrain, particularly in Wnt1cre Tg/Tg embryos (Lewis et al., 2013). We used Wnt1cre Tg/+ mice for our crosses to reduce this effect. In addition, no facial phenotype was observed in Wnt1cre embryos compared to WT littermate controls (data not shown), agreeing with results that show that any midbrain dysmorphologies caused by the Wnt1cre line do not cause cranial shape changes (Heuze et al., 2014).

Pregnant mice were culled using schedule 1 culling methods at E14.5 to E19.5, just prior to giving birth. All procedures were carried out as agreed by the UK Home Office and King's College London. Animals were housed in approved nonspecific-pathogen-free conditions. Animal experiments conform to ARRIVE (animal research: reporting of in vivo experiments) guidelines.

Embryos were photographed using a Leica dissecting microscope.

# Skeletal Preps

E19.5 embryos were skinned and eviscerated before fixing in 95% Ethanol. Samples were then stained in alcian blue and alizarin red to stain cartilages and bones, respectively. Embryos were cleared in 0.5% KOH and stored in glycerol and photographed using a Leica dissecting microscope.

# Histology

Embryos were fixed in 4% paraformaldehyde and dehydrated through an ethanol or methanol series before embedding in wax. Sections were cut on a microtome at 8µm and slides were stained with a trichrome stain (Haematoxylin, alcian blue and sirrus red). Sections were photographed using a Nikon microscope.

# Thymus Analysis

To compare the size of the thymus glands in the conditional mutants the thymus glands from 3 Fgf10fl/fl mice and 4 Wnt1cre Fgf10fl/fl mice were assessed using histology sections at E14.5. The number of sections with a thymus was multiplied by the thickness of the sections (8µm) to give the total extent of the gland. This was then compared using a student t-test where significance was P < 0.05.

# Radioactive In situ Hybridisation

CD1 mice were used for expression of Fgf10. Fgf10 probe was a gift from Ivor Mason. In situ hybridisation on wax sections was carried out according to previously published protocols (Kettunen et al., 2000). Fgf10 antisense probe was synthesized using 35S labeled UTP and signal was identified using sliver emulsion, which when developed showed positive signal as white grains under darkfield. The magic wand tool in photoshop was used to pseudocolour the white grains red and this layer was overlain on top of the light field image to produce a final compound image.

# RESULTS

# Normal Lung, Limb and Pituitary Development but Defective Palate Formation in Wnt1cre Fgf10 Conditional Mutants

In agreement with previous studies we observed expression of Fgf10 in the developing lung, limb, ocular glands, palate, salivary glands, and epithelium of the developing maxilla and mandible (**Figure 1**). As expected conditional mutants (Wnt1creFgf10fl/fl) had normally developing lungs at E19.5 (**Figures 2A,B**; N = 2/2). Histology of the lungs matched that of littermate controls (Fgf10fl/fl; **Figures 2C,D**), as although there are Wnt1cre positive cells located within the developing lungs these are associated with the intrinsic nervous system, which does not express Fgf10 (Freem et al., 2010). The limbs also formed normally (N = 8/8), in contrast to complete loss of these structure in Fgf10 null mutants (**Figures 2E,F**; N = 6/6).

In the palate Fgf10 is expressed in the mesenchyme adjacent to the oral epithelium (**Figure 1D**). Palate development was disrupted in the conditional mutants with a failure in development of the palatal shelves at E15.5 (**Figures 3A,B**; N = 3/3), suggesting problems in shelf development similar to those observed in the null (Rice et al., 2004). Just before birth (E19.5) defects in formation of the palatine processes of the maxilla and palatine bone were clear, leaving the vomer visible when viewed from the oral side (**Figures 3C,D**; N = 2/2). Due to the cleft palate the conditional mutant would not be predicted to survive past birth, and, in agreement with this, no mutants were discovered at P1 (postnatal day 1) in one litter where the mother was left to litter down.

In the Fgf10 null the ectoderm part of the pituitary (Rathke's pouch), which forms the anterior lobe, is completely lost due to high apoptosis at early stages of development (Ohuchi et al., 2000). In contrast the posterior lobe, which is derived from the neuroectoderm, is apparent at E13.5 but regresses in the absence of the anterior lobe and has been reported to be lost by E15.5 (Ohuchi et al., 2000). In agreement with the published data, we observed a complete loss of the ectodermally derived portion of the pituitary at E15.5 in the Fgf10 null mice, but the posterior lobe was still present at this stage (**Figure 3G**). It was also still evident at E18.5 (**Figure 3H**), suggesting that this part of the pituitary is not dependent on the presence of the anterior lobe as previously proposed. The conditional mutant showed normal development of both the anterior and posterior lobe of the pituitary at E15.5 (**Figures 3E,F**), indicating no requirement for neural crest derived Fgf10 in its formation.

# Development of a Hypoplastic CVP and Loss of Salivary and Ocular Glands in Wnt1cre Fgf10 Conditionals

Salivary glands are absent in Fgf10 nulls. In keeping with this result the salivary glands were completely absent in the conditional mutant (**Figures 4A–C,E–G**; N = 8/8), although a mesenchymal capsule still formed despite the lack of any branching epithelium (**Figure 4G**), phenocopying the null phenotype (Wells et al., 2013). These results are in agreement with the neural crest origin of the salivary gland mesenchyme (Jaskoll et al., 2002).

Slightly unexpectedly, the conditional mutants also formed a circumvallate papilla (CVP) at the back of the tongue (**Figures 4D,H**; N = 3/3). The CVP was smaller in size compared to littermate controls and the two fingers of invaginating epithelium were reduced, similar to the phenotype observed in Eda pathway mutants (Wells et al., 2011). We checked the development of the CVP in Fgf10 null embryos, where the CVP has been recorded as missing, and found that the CVP was present but reduced in size in the Fgf10 null embryos at E15.5 (N = 3/3), similar to the phenotype in the conditional knockout, indicating that the CVP can initiate in the absence of Fgf10 (**Figure 4K**).

Fgf10 is expressed at high levels in the mesenchyme around the developing eye during the stages of ocular gland development (**Figure 1C**). At E15.5 the Harderian gland had initiated in littermate controls at the back of the eye, while this gland was missing in the conditional mutant, despite the presence of a mesenchymal capsule (**Figures 4I,J**; N = 3/3). This is in agreement with previous research that Fgf10 expression is essential for the formation of ocular glands, and confirms that the source of Fgf10 is the neural crest around the eye.

# Hypoplasia of Neck Glands but Normal Tracheal Cartilage Patterning in Conditional Mutants

The thyroid gland was present, but reduced in size in the conditional mutants (**Figures 5A,B**; N = 3/3). This suggests either that not all Fgf10 signaling required for formation of this gland is neural crest derived, or that in fact this gland can develop in the absence of Fgf10. To confirm this we looked at development of the thyroid in Fgf10 null mutants. A small

underlying rathke's pouch and in the oral epithelium (arrowhead). Scale bars in (B) = 200µm, same scale (A,C–F).

thyroid was observed in 2/3 cases, and in both cases was unilateral, indicating that the thyroid is able to initiate in the absence of Fgf10 (**Figure 5C**). The gland tissue was located in the correct place, under the cricoid cartilage, indicating that migration cues were unaffected in the mutant, however the gland did not extend as far anteriorly toward the thyroid cartilage. The null gland when present was smaller than that observed in the conditional mutant suggesting another source of Fgf10 might be available for development of this gland in the conditional mutant. Alternatively development of this gland might depend on interaction with other tissues, not affected in the conditional. As the parathyroids migrate to the thyroid we checked for the presence of these glands in our samples. The parathyroids were normally positioned next to the thyroid in the conditional mutant, but as with the thyroid were slightly hypoplastic (**Figures 5D,E**). No evidence of parathyroids were observed in the Fgf10 null mutant mice (N = 3; **Figures 5F**), suggesting again an alternative non-neural crest source for parathyroid gland development in our conditional mutants. The thymus glands in the Fgf10 null mutants have been shown to be hypoplastic (Ohuchi et al., 2000; Revest et al., 2001). In the conditional mutants the thymus glands were present (**Figures 5G,H**) but appeared slightly reduced in size in the conditional mutant at E15.5, although analysis at E14.5 showed

no statistically significant difference (P = 0.684). This is in contrast to the null where the thymus glands are much smaller by this stage (Revest et al., 2001). In each case, therefore, the neck glands were less severely affected in the conditional compared to the null mutant.

We therefore decided to confirm the position of the boundary between the neural crest and mesoderm in this region of the neck. Tracheas from Wnt1cre-tdTom reporter mice were dissected out with the glands removed at P0 to identify the limit of the neural crest, which was found to lie between the thyroid (Wnt1 positive) and cricoid (Wnt1 negative) cartilages, with the tracheal rings being mesodermal (**Figure 5K**). This therefore places the thymus, thyroid and parathyroids within the mesoderm, despite the glands themselves having a neural crest origin. In the Fgf10 null the trachea cartilages are severely mispatterned (Sala et al., 2011). We therefore investigated tracheal cartilage formation at

missing in the conditional mutant and the underlying vomer (arrowhead) is visible. (E–H) Developing pituitary gland. (E,F) The anterior and posterior lobes form as normal in the conditional mutant at E15.5. (G,H) The anterior lobe is missing but the posterior lobe is still evident in the Fgf10 null at E15.5 (G) and E18.5 (H). A = anterior lobe derived from oral ectoderm. P = Posterior lobe derived from neuroepithelium. BS = Basisphenoid. Scale bars in (A,B) = 500µm. Scale bars (C,D) = 200µm. Scale bars (E–H) = 200µm.

E19.5 by skeletal prep in the conditional mutants. As expected, given the limit of the neural crest in this region, the cartilage rings were unaffected in the conditional mutants (N = 2/2), matching the pattern in littermate controls (**Figures 5I,J**).

# DISCUSSION

The development of the ocular and submandibular and sublingual salivary glands was completely dependent on Fgf10 signaling from the neural crest derived mesenchyme, with development arresting at early initiation stages as in the null. This paper therefore confirms that the Fgf10 expressing ocular and salivary gland mesenchyme is derived from the neural crest. As expected, palate development was also disrupted after loss of Fgf10 in the neural crest derived mesenchyme of the developing palatal shelves, and the conditional mutation is likely to cause lethality.

In contrast to these cranial glands other more posterior glands, such as the thyroid, thymus, and parathyroids did not phenocopy the complete loss of Fgf10. The thymus, parathyroids and thyroid initiate within neural crest derived mesenchyme (Müller et al., 2008; Johansson et al., 2015) and then migrate more posteriorly to sit within mesodermally derived mesenchyme, as supported by our neural crest lineage analysis of the trachea, and previous lineage tracing that mentions the tracheal rings are not neural crest derived (Matsuoka et al., 2005). All of these glands are severely affected in the Fgf10 null but the conditional mutants had a milder phenotype. In all three glands it is possible that other Fgfs and alternative signaling pathways are able to compensate for the initial loss of Fgf10 in this tissue, allowing their initiation, while mesodermal Fgf10 may be able to act once the glands have reached their final positions in the neck. In agreement with this Fgf10 is expressed in the mesenchyme around the thymus at E13.5, a stage after the glands have reached their final position (Revest et al., 2001), and is strongly expressed in the ventral mesenchyme of the developing trachea from E14.5 (Sala et al., 2011). Mesodermal Fgf10 is therefore in the right place to be able to signal to the more posterior glands. It is also possible that signals between these tissues and other neighboring structures are important for their development, and that their presence is interdependent.

Early on in development Fgf10 is strongly expressed in the oral epithelium. It was therefore possible that some of the orally derived structures would have a reduced phenotype when compared to the Fgf10 null. The glands of the oral cavity, however, appeared to mimic the null phenotype indicating that only the loss of Fgf10 in the neural crest was critical. Moreover the epithelial expression appeared to have no influence on the developing teeth, the molars having only a very minor defect in relative size similar to the null (Ohuchi et al., 2000). It would therefore be interesting to see whether knocking out Fgf10 in the early epithelium has any effect on development of these key structures. As expected development of the pituitary, limbs and lungs was normal in the conditional knockout, in which Fgf10 was provided by the neuroepithelium and mesoderm rather than neural crest. The tracheal rings were also normal highlighting the fact that the Fgf10 expressing mesenchyme that forms the cartilage rings is not neural crest derived.

Our comparison of the conditional mutants with the Fgf10 null mutants revealed a few differences between the phenotype observed in our null mice and in previously published data. For example, although it has been reported that the thyroid fails

(arrow in D). (H,K) A CVP formed in the null and conditional but was smaller than in the WT. (I,J) Frontal sections of the eye at E15.5. The harderian gland at the back of the eye failed to form in the conditional mutant (asterix in J). Scale bars (A,B) = 500µm, same scale in (E,F). Scale bars (C,D,G,I) = 200µm, same scale in (H,J,K).

to form in Fgf10 null mice (Ohuchi et al., 2000) in our null mice a small amount of glandular tissue was present around the trachea in the region of the thyroid but this was only observed unilaterally. Interestingly, in the Ohuchi paper although the text states no gland forms the figures highlight a rudimentary thyroid. The thyroid therefore does appear to be able to form in the complete absence of Fgf10 but is severely hypoplastic, while we saw no evidence of a parathyroid.

We also observed development of a hypoplastic CVP in the tongue, which had previously been reported as missing in the Fgf10 mutant (Petersen et al., 2011). Fgf7 is also expressed in the mesenchyme of the developing tongue and may compensate for the loss of Fgf10 in this structure (Sohn et al., 2011). These differences with the published data may indicate variation due to genetic background. For our studies we investigated Fgf10 nulls on a mixed C57Bl6/CD1 background, while other papers have used a mixed C57bl6/CBA or C57Bl6/129SVJ or not reported the background used (Min et al., 1998; Sekine et al., 1999; Ohuchi et al., 2000; Rice et al., 2004; Petersen et al., 2011). In fact the arrest in limb development was reported to occur at slightly different time-points when the Fgf10 knockout was originally reported by two groups, with the difference being suggested to be due to genetic background (Min et al., 1998; Sekine et al., 1999). Our findings have therefore shed light on the structures affected by neural crest expressing Fgf10 but have also revealed some differences in the published literature which merit further investigation.

# ETHICS STATEMENT

All experiments were approved by the Home Office and conducted with the correct project and personal licenses. Experiments using GMOs were approved by the Kings Biological Safety Committee.

(A–C,G,H) = 200µm. Scale bars (D–F) = 100µm. Scale bars (I–K): 500 µm.

# AUTHOR CONTRIBUTIONS

AT and SL conceived the experiments, AT and TT conducted the experiments and undertook data acquision, AT, SL, and TT wrote the manuscript.

# FUNDING

TT was funded by São Paulo Research Foundation (FAPESP), grant 2015/02824-6. AT is funded by the Wellcome Trust (102889/Z/13/Z).

# REFERENCES


# ACKNOWLEDGMENTS

The Fgf10A02 tmc1c mice were obtained from the MRC-Harwell, which distributes these mice on behalf of the European Mouse Mutant Archive (http://www.emmanet. org). The MRC-Harwell is also a member of the International Mouse Phenotyping Consortium (IMPC), which funded the generation of the Fgf10A02 tmc1c mice. Associated primary phenotypic information may be found at http://www. mousephenotype.org.


**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 CC and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Teshima, Lourenco and Tucker. 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) or licensor 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.

# Understanding Mechanisms of GLI-Mediated Transcription during Craniofacial Development and Disease Using the Ciliopathic Mutant, talpid<sup>2</sup>

Ya-Ting Chang1, 2, Praneet Chaturvedi <sup>2</sup> , Elizabeth N. Schock 1, 2 and Samantha A. Brugmann1, 2 \*

*<sup>1</sup> Division of Plastic Surgery, Department of Surgery, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA, <sup>2</sup> Division of Developmental Biology, Department of Pediatrics, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA*

### Edited by:

*Thimios Mitsiadis, University of Zurich, Switzerland*

### Reviewed by:

*Vicki Rosen, Harvard University, USA Ophir D. Klein, University of California, San Francisco, USA Paul Trainor, Stowers Institute for Medical Research, USA*

> \*Correspondence: *Samantha A. Brugmann samantha.brugmann@cchmc.org*

### Specialty section:

*This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology*

Received: *30 July 2016* Accepted: *29 September 2016* Published: *17 October 2016*

### Citation:

*Chang Y-T, Chaturvedi P, Schock EN and Brugmann SA (2016) Understanding Mechanisms of GLI-Mediated Transcription during Craniofacial Development and Disease Using the Ciliopathic Mutant, talpid<sup>2</sup> . Front. Physiol. 7:468. doi: 10.3389/fphys.2016.00468* The primary cilium is a ubiquitous, microtubule-based organelle that cells utilize to transduce molecular signals. Ciliopathies are a group of diseases that are caused by a disruption in the structure or function of the primary cilium. Over 30% of all ciliopathies are primarily defined by their craniofacial phenotypes, which typically include midfacial defects, cleft lip/palate, micrognathia, aglossia, and craniosynostosis. The frequency and severity of craniofacial phenotypes in ciliopathies emphasizes the importance of the cilium during development of the craniofacial complex. Molecularly, many ciliopathic mutants, including the avian *talpid<sup>2</sup>* (*ta<sup>2</sup>* ), report pathologically high levels of full-length GLI3 (GLI3FL), which can go on to function as an activator (GLIA), and reduced production of truncated GLI3 (GLI3T), which can go on to function as a repressor (GLIR). These observations suggest that the craniofacial phenotypes of ciliary mutants like *ta<sup>2</sup>* are caused either by excessive activity of the GLI<sup>A</sup> or reduced activity of GLIR. To decipher between these two scenarios, we examined GLI3 occupation at the regulatory regions of target genes and subsequent target gene expression. Using *in silico* strategies we identified consensus GLI binding regions (GBRs) in the avian genome and confirmed GLI3 binding to the regulatory regions of its targets by chromatin immunoprecipitation (ChIP). In *ta<sup>2</sup>* mutants, there was a strikingly low number of GLI3 target genes that had significantly increased expression in facial prominences compared to the control embryo and GLI3 occupancy at GBRs associated with target genes was largely reduced. *In vitro* DNA binding assays, further supported ChIP results, indicated that the excessive GLI3FL generated in *ta<sup>2</sup>* mutants did not bind to GBRs. In light of these results, we explored the possibility of GLI co-regulator proteins playing a role in regulatory mechanism of GLI-mediated transcription. Taken together our studies suggest that craniofacial ciliopathic phenotypes are produced via reduced GLI<sup>T</sup> production, allowing for target gene transcription to be mediated by the combinatorial code of GLI co-regulators.

Keywords: primary cilia, craniofacial, talpid<sup>2</sup> , c2cd3, ciliopathies, GLI

# INTRODUCTION

Primary cilia are ubiquitous organelles that serve as cellular hubs for transduction of numerous signaling pathways. Most notably, cilia have been identified as transducers of the Hedgehog (Hh) pathway. Identification of the primary cilium as a signaling hub for the Hh pathway came from seminal experiments reporting that anterograde and retrograde intraflagellar transport (IFT) proteins in the cilium are required for Sonic Hedgehog (SHH) signal propagation (Huangfu et al., 2003; Huangfu and Anderson, 2005). When SHH ligand is present, it binds to its receptor Patched (PTCH), thus allowing Smoothened (SMO) to localize and accumulate in the primary cilium (Corbit et al., 2005; Rohatgi et al., 2007). Activated, ciliary SMO, in concert with Kif7, then promotes the dissociation of GLI from Suppressor of Fused (SUFU) (Humke et al., 2010; Tukachinsky et al., 2010; Li et al., 2012) and the subsequent post-translational processing of GLI proteins necessary for their function as activators and repressors (Goetz and Anderson, 2010).

In vertebrates, there are three members of the GLI transcription factor family: GLI1, GLI2, and GLI3. GLI1 and GLI2 are considered transcriptional activators, whereas GLI3 mostly behaves as a repressor. However, there have been examples of GLI2 functioning as a repressor in the absence of GLI3, and GLI3 functioning as an activator in the absence of GLI2 (Mo et al., 1997; Theil et al., 1999; Tole et al., 2000; Bai and Joyner, 2001; Persson et al., 2002; Rallu et al., 2002; Buttitta et al., 2003; Motoyama et al., 2003; Bai et al., 2004; Lei et al., 2004; McDermott et al., 2005; Pan et al., 2009). Full-length GLI2 and GLI3 can be processed via phosphorylation and other post-translational modifications into the activator isoform (GLIA) or truncated into the repressor isoform (GLIR) (Wang et al., 2000; Pan et al., 2006). Inhibition of GLI processing prevents production of GLI<sup>A</sup> and GLI<sup>R</sup> isoforms. Thus, an essential role of primary cilia is to establish the ratio of GLI<sup>A</sup> to GLI<sup>R</sup> proteins (Haycraft et al., 2005; Liu et al., 2005), which in turn controls transcription of SHH target genes.

Three basic models have been proposed to depict the potential mechanism of how SHH target genes are activated by a gradient of GLI isoforms: (1) the ratio sensing model, (2) the threshold repression model and (3) the threshold activation model (Falkenstein and Vokes, 2014). Ratio sensing, as the name implies, is based on the ratio of GLI<sup>A</sup> and GLI<sup>R</sup> rather than concentration of either. The net balance of GLI<sup>A</sup> to GLI<sup>R</sup> then determines if, and the extent to which, a target is activated or repressed. The other two models suggest threshold-specific mechanisms. The threshold activation model suggests that SHH targets are activated when GLI<sup>A</sup> reaches a threshold-specific concentration. On the other hand, in the threshold repression model (de-repression), the activation of SHH target genes is dictated by the alleviation of repression via loss of GLIR.

The talpid<sup>2</sup> (ta<sup>2</sup> ) is a naturally occurring avian mutant that is best characterized by severe polydactyly and its oral-facial phenotype (Abbott et al., 1959, 1960; Dvorak and Fallon, 1991; Schneider et al., 1999). The face of affected ta<sup>2</sup> embryos is characterized by a dysmorphic frontonasal prominence, facial clefting, hypoplastic maxillary prominences, incomplete fusion of the primary palate and hypoglossia (Chang et al., 2014). Our recent work determined that the ta<sup>2</sup> mutation affected ciliogenesis via a 19 bp deletion in C2CD3 (Brugmann et al., 2010; Chang et al., 2014), a centriolar protein required for ciliogenesis (Hoover et al., 2008). Our studies also determined that the ta<sup>2</sup> mutant genetically, biochemically and phenotypically phenocopied the human craniofacial ciliopathy, Oral-facialdigital syndrome 14 (OFD14) (Schock et al., 2015). ta<sup>2</sup> embryos, similar to many other ciliopathies, have a significant increase in GLI3FL and a reduction in the amount GLI<sup>T</sup> (Chang et al., 2014). However, the mechanism by which this disruption in GLI isoform production affects expression of GLI targets in the developing craniofacial complex remains unknown.

Herein, we utilize a combination of several biochemical techniques to determine the impact loss of cilia has on GLI function. Specifically, we examine the expression of GLI target genes and occupation of GLI binding regions (GBRs) associated with those targets in the developing frontonasal, maxillary and mandibular prominences (FNP, MXP, and MNP, respectively) in order to uncover the mechanism by which GLI mediated transcription is being impacted in ta<sup>2</sup> mutants. Understanding the full extent of molecular disruptions in ta<sup>2</sup> mutants will hopefully guide future therapeutic strategies for craniofacial ciliopathies, a rapidly growing group of disorders that currently have little to no therapeutic treatment.

# MATERIALS AND METHODS

# Embryo Preparation

talpid<sup>2</sup> (ta<sup>2</sup> ) heterozygous carriers were mated, eggs were collected and shipped from the UC Davis Avian Sciences Department. Embryos were incubated at 37◦ C for approximately 5–7 days when embryos reached Hamburger Hamilton stage 25–31 (HH25-31).

# Quantitative RT-PCR of GLI Targets

FNPs, MXPs, and MNPs were harvested from day 5 chick embryos. mRNA was prepared with TRIzol reagent (Thermo Fisher Scientific), and then converted to complementary DNA through reverse transcription reaction (High-Capacity cDNA Reverse Transcription Kit, Applied BiosystemsTM ). Different amounts of cDNA (40, 20, 10 and 5 ng) was used for quantitative PCR to test PCR efficiency and a linear range of duplication (SsoAdvancedTM Universal SYBR <sup>R</sup> Green Supermix, BioRad). Expression of genes known to play a role in craniofacial development were examined with the following primer sets: ALX4 (105 bp) F: GTTACGGTAAGGAGAGCAGTTT, R:CTTTCACTCCAGCCTCCTTC, BMPR1A (100 bp) F: GTGCTGTCGGACTGATTTCT, R:TGCCATCCAACGAAT GCT, WIF1 (100 bp) F: CAACCTGTTTCAATGGAGGAAC, R: GGCTGATGGCATTTACTGATTT, OSR2 (140 pb) F: CCACTTCACCAAGTCCTACAA, R: TCTCTTTGGAATGGAT GTACCG. The statistical significance of the data was evaluated through two-tailed Student's t-test. p-values less than or equal to 0.05 (95% confidence level) were considered as statistically significant differences.

# Western Blotting of GLI2/3 Proteins

FNPs, MXPs, or MNPs were pooled and lysed in RIPA buffer containing protease inhibitors and phosphatase inhibitors (1 mM Phenylmethylsulfonyl fluoride, 1 mM NaVO4, 1X complete protease inhibitor cocktail, EDTA-free), and slightly sonicated with a microprobe to recover chromosome bound GLI2 and GLI3 proteins. BCA assay (Pierce) was used to measure protein concentration of cell lysates. Proteins were boiled with 1X Laemmli sample buffer and run on 6% SDS-PAGE for GLI2 and GLI3, or 12% SDS-PAGE for GAPDH, which later were wet-transferred to Polyvinylidene difluoride (PVDF) membrane. Anti-GLI2, anti-GLI3 (Polyclonal goat IgG 1:500, R&D systems) and anti-GAPDH (FL-335, polyclonal rabbit IgG 1:4000, Santa Cruz Biotechnology) were prepared in 1X TBST (0.1% Tween-20)/6% nonfat milk, as well as secondary antibodies (antigoat and anti-rabbit, 1:10,000). Proteins were detected by Electrochemiluminescence assay (Amersham ECL Prime, GE Healthcare Life Science).

# DNA Binding Affinity Assay

PATCHED 1 promoter oligonucleotides were designed according to the location of a Gli binding region (GBR) at −2549 from the TSS site (GGAAGAAGTGTCAGTGTAAG AGTCTCCACGTGGGTGGTCAAGGCCATGGCTGCCTCAC GG). 100 pmole of biotin-conjugated positive oligonucleotides and complementary oligonucleotides (Integrated DNA Technologies) were annealed in 1X TE/50 mM NaCl buffer in a PCR cycler and incubated with Dynabeads Streptavidin (M280, Invitrogen). FNPs, MXPs, or MNPs from day 5 control or ta2 embryos were pooled and processed as described for Western blotting. Cell lysates were incubated with oligonucleotidesbound Dynabeads at 4◦C for 2 h. Beads were washed with 1 ml RIPA buffer three times and processed for Western Blotting analysis.

# Chromatin Immunoprecipitation

FNPs, MXPs, or MNPs from day 5 control or ta<sup>2</sup> embryos were harvested, pooled and crosslinked in 1% formaldehyde. Tissue was homogenized in RIPA buffer and sonicated (Bioruptor <sup>R</sup> Pico, Diagenode) at 5 cycles of 30 s on/45 s off. Sheared DNA was distributed around 0.3 kb to 1 kb on 1% agarose gel. Cell lysates were pre-cleaned with Dynabeads protein G (ThermoFisher Scientific) and quantified by BCA assay. Dynabeads for immunoprecipitation were blocked with 20µg/ml Glycogen, 20µg/ml BSA, 20µg/ml yeast RNA in RIPA buffer at 4◦C for an hour. GLI3 antibody (AF3690, R&D systems) and pre-blocked Dynabeads Protein G were incubated with 90% of cell lysates at 4◦C overnight. 10% of cell lysates were kept as Input. Beads were washed with IP wash buffer I (low salt; 50 mM HEPES-KOH pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% sodium deoxycholate, 1% Triton X-100), IP wash buffer II (high salt; 50 mM HEPES-KOH pH 7.5, 500 mM NaCl, 1 mM EDTA, 0.1% sodium deoxycholate, 1% Triton X-100), IP Wash Buffer III (LiCl containing buffer; 10 mM Tris-Cl pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5% Sodium deoxycholate, 0.5% NP-40) and TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA). ChIP samples were reverse-crosslinked by boiling with 10% Chelex-100 (BioRad), and treated with 0.2 mg/ml Proteinase K at 55◦C for 30 min. Immunoprecipitated DNA samples were analyzed with quantitative real-time PCR (BioRad) with primers to GBR of target genes. Error bars in all figures represent standard error of the mean (S.E.M.) from five to seven independent experiments. The statistical significance of the data was evaluated through two-tailed Student's t-test. p-values less than or equal to 0.05 (95% confidence level) were considered as statistically significant differences.

# GBR Analysis

Sequences for GBRs from previous publications (Vokes et al., 2007, 2008) were used with a custom perl script to search for all the exact matches of the various possible sequences of the consensus GBRs in the chicken genome. Acquired positions of the motifs in the genome were run through a second custom perl script to search for genes that encompass these motif sites at a distance of 20 kb from either end. Potential GLI targets were confirmed using chromatin-immunoprecipitation (ChIP) assays.

# RESULTS

### The Avian Ciliopathic Mutant, talpid<sup>2</sup> , has Craniofacial Anomalies Characteristic of a Ciliopathy

To understand the transcriptional networks affected in craniofacial ciliopathies we analyzed the talpid<sup>2</sup> (ta<sup>2</sup> ) mutant, a naturally occurring avian ciliopathic mutant that has been established as a model for the human craniofacial ciliopathy, Oral-facial-digital syndrome 14 (Chang et al., 2014; Schock et al., 2015). The ta<sup>2</sup> craniofacial phenotype characterized by facial and palatal clefting, micrognathia, and hypoglossia, is fully evident at day 7 (**Figure 1**). Although our previous work has characterized this phenotype (Chang et al., 2014; Schock et al., 2015), to determine the transcriptional networks that contribute, we first needed to identify when phenotypic onset occurred. At day 7 the frontonasal prominence (FNP) is shorter and wider and frequently does not fuse to adjacent prominences (**Figures 1A–B'**). Two days earlier at day 5, the MXPs were medially rotated, the nasal pits were larger, thus preventing the proper juxtapositioning of the FNP with adjacent prominences (**Figures 1C–D'**). Palatal views showed increased patency of the naturally cleft avian secondary palate in ta<sup>2</sup> embryos relative to controls at day 7 (**Figures 1E–F'**, dotted white lines). Two days earlier, at day 5, the dysmorphology and malposition of the MXPs is just becoming apparent (**Figures 1G–H'**). Dorsal views of the developing lower beak showed the MNP of ta<sup>2</sup> embryos failed to fuse completely and had a hypoplastic tongue (hypoglossia, dotted black line) (**Figures 1I–J'**, white arrow). At day 5 there was little difference in mandibular growth between control and ta<sup>2</sup> MNPs (**Figures 1K–L'**). From phenotypic evaluations, taken together with the fact that these anomalies were not readily identifiable at day 4, we determined that craniofacial anomalies were initiated at approximately day 5 of development. Thus, to determine the molecular networks responsible for these


phenotypes we carried out our analyses in the facial prominences of day 5 embryos.

# Loss of Cilia Results in Aberrant GLI Isoform Production

Our previous work determined that aberrant ciliogenesis in ta<sup>2</sup> embryos disrupted the production of GLI proteins in such a manner that there was increased GLIFL production and decreased GLI<sup>T</sup> production (Chang et al., 2014). The GLIFL isoform typically goes on to function as an activator, whereas the GLI<sup>T</sup> isoform goes on the function as a repressor. The excessive production of GLIFL is an extraordinarily common molecular phenotype in ciliopathies, including those with craniofacial phenotypes (Huangfu and Anderson, 2005; Davey et al., 2006; Tran et al., 2008; Tabler et al., 2013). To carefully exam the differences of GLIFL and GLI<sup>T</sup> protein levels in control and ta<sup>2</sup> mutants, we performed Western blot analysis of GLI2 and GLI3 in the three facial prominences affected in the ta<sup>2</sup> at day 5. We detected a very low level of GLI2FL and a substantial amount of GLI2<sup>T</sup> isoforms in the control facial prominences (**Figure 2A**). The loss of cilia in ta<sup>2</sup> embryos disrupted GLI processing and altered the production of GLI2 protein isoforms. In ta<sup>2</sup> facial prominences, we detected dramatically increased levels of GLI2FL, and low levels of GLI2T, relative to control prominences. We next examined the production of GLI3 isoforms. Western blot analyses showed that, similar to GLI2FL, GLI3FL was increased in ta<sup>2</sup> prominences relative to controls. Contrary to what was observed with the GLI2T, GLI3<sup>T</sup> was readily detectable in both control and ta<sup>2</sup> facial prominences. Specifically, ta<sup>2</sup> MNP had less GLI3<sup>T</sup> than control MNP. These data suggested that disrupted ciliogenesis affected the processing of GLI2 and GLI3 in distinct manners, yet the net result was an increase in GLIFL production. We did not observe a change in GLI1 protein levels between control and mutant embryos (Data not shown).

Since the GLIFL isoform typically goes on to act as an activator, these results suggest that craniofacial phenotypes in the ta<sup>2</sup> mutant could be caused by increased GLI activator function leading to increased expression of GLI target genes. To test this hypothesis, we utilized an in silico approach to look for potential GLI target genes within the avian genome. Using previously published sequences of GBRs (Vokes et al., 2007, 2008) we scanned the chicken genome for possible GLI targets. GBR positions were run through a custom perl script to search for genes that encompass these motif sites at a distance of 100 kb from the 5′ or 3′ end of the gene. From this list we identified several genes known to play a role in craniofacial development (**Supplemental Table 1**). Confirmation of our in silico approach was carried out on selected genes using ChIP assays (**Supplemental Figures 1 A–D**). Through these analyses, we selected four GLI3 target genes known to be expressed during, and have a role in, craniofacial development: ALX4 (Beverdam et al., 2001; Mavrogiannis et al., 2001), BMPR1A (Li et al., 2011, 2013; Saito et al., 2012), OSR2 (Lan et al., 2001), and WIF1 (Hsieh et al., 1999; Darnell et al., 2007). PTCH1, a known GLI target, was used as a positive control (**Supplemental Figure 1E**) and IgG IP as an antibody background control. The data were all normalized to IgG IP percentage, and the genes with relative enrichment >1 were considered positive for GLI3 binding. To determine if increased production of GLI3FL in ta<sup>2</sup> facial prominences correlated with increased expression of these target genes, we performed quantitative RT-PCR (qPCR) with mRNA from facial prominences of day 5 control and ta<sup>2</sup> embryos (**Figures 2B–E**). There was not an increase in ALX4 expression in any of the facial prominences (**Figure 2B**); however, we detected a significant decrease in ALX4 expression in the MXP. BMPR1A expression was also not significantly increased in any of the developing prominences (**Figure 2C**). No significant changes in OSR2 expression were detected in the FNP or MNP; however, a significant increase in expression was observed in the MXP (**Figure 2D**). Finally, WIF1 expression was not changed in the FNP, yet was significantly increased in the MXP and significantly decreased in the MNP (**Figure 2E**). Taken together, these data do not support the idea that increased production of GLI3FL directly and uniformly results in increased expression of GLI targets throughout the facial prominences. Additionally, these data suggest that each facial prominence interprets aberrant GLI production in a unique manner.

# Excess Production of GLIFL Does Not Correlate with an Increase of GLIFL Occupancy at GBRs

For GLIFL to function as an activator, it has to occupy the regulatory regions of GLI targets. We wondered if lack of uniform increases of target gene expression in ta<sup>2</sup> embryos was due to failure of GLIFL to recognize and occupy GBRs of target genes. To test this hypothesis, we performed an in vitro DNA binding assay using the PATCHED 1 (PTCH1) promoter. We synthesized a 60 base pair biotin-labeled oligonucleotide of the PTCH1 promoter containing an endogenous GLI binding motif found at position −2549 proximally upstream of the transcription start site (TSS). Through high affinity of Biotin-Streptavidin interaction, we were able to evaluate the DNA binding ability of GLI3 isoforms by Western blot. Pre-incubation of nonlabeled PTCH1 oligonucleotides depleted GLI3 protein signals, which confirmed the specificity of the GLI3-GBR interaction. Under the same exposure, the affinity based pull-down assay showed that GLI3<sup>T</sup> predominantly bound to the biotin-labeled oligonucleotides in both control and ta<sup>2</sup> mutants. Interestingly, despite the high level of GLI3FL production in ta<sup>2</sup> embryos, GLI3FL failed to bind to the biotin-labeled oligonucleotides (**Figures 3A,B**). On the contrary, comparison between control and ta<sup>2</sup> embryos indicated that the amount of GLI3<sup>T</sup> binding to the oligonucleotides correlates with the protein concentration (**Figures 3A,B**). Taken together these data suggest that despite increased GLI3FL production, target gene expression is not increased via increased GLI activator function because GLI3FL does not occupy the GBRs of target genes. In addition, the predominant binding of the GLI3<sup>T</sup> repressor at GBRs supports the idea of the threshold repression model, in which GLI targets underwent de-repression due to the removal of GLI<sup>T</sup> repressor.

FIGURE 2 | Excessive GLIFL production in ta<sup>2</sup> embryo does not correlate with increased gene activation of GLI targets. (A) Western blot of GLI2 and GLI3 proteins from the frontonasal prominence (FNP), maxillary prominence (MXP) and mandibular prominence (MNP) of day 5 control and *ta<sup>2</sup>* embryos. GAPDH was used as a loading control. (B–E) mRNA-qPCR analyses of *ALX4*, *BMPR1A*, *OSR2*, *WIF1* from FNP, MXP, and MNP of day 5 control and *ta<sup>2</sup>* embryos. The data was normalized by the individual facial prominences of control and *ta<sup>2</sup>* embryos. The asterisks indicate statistically significant differences and are assigned as followed: \**P* < 0.05, \*\**P* < 0.01. Error bars are based on the standard error of the means (S.E.M.). *n* = 4.

mandibular prominence (MNP) in day 5 control and *ta<sup>2</sup>* embryos. (B) Western blot of GLI3 proteins from the same lysates as (A). GAPDH as loading control of cell lysates.

# GLI3 Occupancy at GBRs within the Regulatory Regions of Target Genes Is Altered in ta<sup>2</sup> Mutants

In vitro DNA binding affinity assays suggested that there was not an increase of ectopic GLI3FL at GBRs of target genes in ta<sup>2</sup> mutants. To determine if GLI3 binding to GBRs was altered in vivo in ta<sup>2</sup> mutants compared to control embryos, we next performed ChIP-qPCR (**Figure 4**). In ta<sup>2</sup> mutants, GLI3 enrichment at GBRs associated with the craniofacial genes ALX4 and OSR2 was reduced in all facial prominences, yet only significantly in the MXP and MNP (**Figures 4A,B**). GLI3 enrichment at the GBRs associated with BMPR1A and WIF1, was also overwhelmingly reduced in facial prominences, yet only significantly in the ta<sup>2</sup> FNP and MXP (**Figures 4C,D**). GLI3 binding in the MNP of WIF1 was below detectable levels in controls, and thus could not be evaluated. These data suggested that GLI3 occupancy at the GBR of target genes is decreased. Our in vitro data suggested that GLI3FL did not bind at target gene GBRs (**Figure 3**). Thus, taken together these data indicate that the observed reduction of GLI3 binding is indicative of reduced GLI3<sup>T</sup> binding at the regulatory regions of target genes.

Reduced GLI3 binding at GBRs of GLI target genes appeared to be a general trend in ta<sup>2</sup> mutants, as several other target genes expressed in the craniofacial complex also exhibited reduced enrichment of GLI3 at associated GBRs (**Supplemental Figure 2**). Thus, these data suggested that altered expression of GLI targets could be caused by aberrant GLI3<sup>T</sup> binding in ta<sup>2</sup> mutants. Interestingly, there was not always a direct correlation with loss of GLI3<sup>T</sup> repressor binding and increased gene expression. Furthermore, different results among the three facial prominences examined pointed to context—and tissue-specific regulation of GLI targets (**Supplemental Figures 1A–D**). Several studies have proposed the possibility that GLI proteins work together with co-regulators to influence expression of targets genes (Brewster et al., 1998; Koyabu et al., 2001; Mizugishi et al., 2001; Lee et al., 2010; Peterson et al., 2012). To determine if target gene expression in the craniofacial complex was influenced by the action of GLI coregulators, we next examined the proximity of GBRs to motifs for potential GLI co-regulators.

# Motif Analyses Identified Sequences for Known GLI Co-Regulators Frequently Co-Localize with GBRs

GBRs have previously been identified in close proximity to binding motifs for other transcription factors, including members of the bHLH, SP, and Sox families (Vokes et al., 2008; Peterson et al., 2012; Aberger and Ruiz I Altaba, 2014). Furthermore, the co-occupancy and cooperativity of GLI with SOX transcription factors was previously shown to be essential for activating neural gene expression signatures (Peterson et al., 2012). To determine if expression of our identified GLI targets could be influenced by the presence or absence of GLI coregulators, we first examined the genomic sequence around the GBRs of our target genes for E-box, SP, and SOX binding motifs. (**Figure 5**, **Supplemental Figure 3**). We defined the area <1 kb upstream of the transcription start site (TSS) as the promoter region, <20 kb away from the TSS as proximal upstream or downstream, and <100 kb away from the TSS as distal upstream or downstream. We analyzed the sequence surrounding GBRs in our four selected GLI target genes for sequences predictive of E-box, SP, and SOX binding. All four of our identified GLI target genes contained at least one motif cluster containing a GBR, E-box, SP, and SOX binding site within 1 kb of each other (**Figures 5A–D**). Several other clusters containing three of the four motifs were also identified (**Figures 5A–C**; red box). The close proximity of these binding motifs suggested that expression of GLI targets in the developing craniofacial complex could be influenced by the cooperative function of GLI isoforms and co-regulator proteins.

# GLI Co-Regulators Have a Prominence Specific Expression Pattern That Changes When Primary Cilia Are Lost

Several transcription factors synergistically cooperate with GLI proteins to influence GLI target gene expression (Aberger and Ruiz I Altaba, 2014). Their co-occupancy at the promoter of GLI targets is required for the optimal activation/repressor. ChIPbased, high-throughput analyses uncovered several transcription factor motifs located close to GBRs in the cis-regulatory modules of GLI targets. Specifically, binding sequences for Sox (Peterson et al., 2012), bHLH (Lee et al., 2010), and SP proteins (Vokes et al., 2008) have been shown to exist in very close proximity

to GBRs. Our in silico analyses confirmed these sequences exist in near GBRs in four previously identified GLI-targets. We hypothesized that differential expression of these co-regulators could contribute to the differential gene expression of GLI targets in both control and ta<sup>2</sup> mutant embryos. We first investigated the expression of genes that could bind to motifs found in close proximity to GBRs, specifically SOX8, SP3, and HAND2 (**Figures 6A–C**). SOX8 expression was significantly reduced in the FNP and MXP of ta<sup>2</sup> mutants, yet was not significantly altered in the MNP (**Figure 6A**). SP3 expression was significantly increased in ta<sup>2</sup> MNP, yet not changed in the FNP and MXP (**Figure 6B**). HAND2 expression was reduced in the MXP, but not significantly changed between control and ta<sup>2</sup> FNPs and MNPs (**Figure 6C**). Thus, differential expression of GLI co-regulators could possibly contribute to altered target gene expression in ta<sup>2</sup> mutants and explain why changes in GLIFL and GLI<sup>T</sup> isoforms do not uniformly or directly correlate with changes in target gene expression.

Our previous analyses indicate that GLI target gene expression changed in a prominence specific manner (Chang et al., 2014). We wondered if differential expression of potential GLI co-regulators could contribute to this prominence-specific expression changes. SOX8 was robustly expressed within the FNP, with levels significantly higher than those in the MXP or MNP (**Figure 6A'**). SP3 was more robustly expressed in the MXP and MNP, relative to the FNP (**Figure 6B'**). HAND2 was robustly and exclusively expressed in the MNP (**Figure 6C'**). Thus, each of these potential co-regulators has a prominences specific expression pattern that could differentially influence GLI target gene expression. Collectively, the proximity of binding motifs, coupled with the differential expression of these coregulators could possibly contribute to altered expression of GLI targets in ta<sup>2</sup> embryos.

# DISCUSSION

Craniofacial ciliopathies are a rapidly growing group of disorders that severely impact craniofacial growth and development. Currently, there are little to no therapeutic options for these conditions. Although the molecular mechanism behind these disorders remains nebulous, many ciliopathies have aberrant production of GLIFL and GLI<sup>T</sup> isoforms. (Huangfu and Anderson, 2005; Davey et al., 2006; Tran et al., 2008; Tabler et al., 2013; Chang et al., 2014). Herein, we attempted to identify the mechanism by which aberrant GLI protein production impacts craniofacial development. To do so we used the avian ta<sup>2</sup> model, which has recently been characterized as a bona fide model for the human craniofacial ciliopathy Orofacial-digital syndrome 14 (Schock et al., 2015). In silico and ChIP assays identified GLI target genes in the avian genome that play a role in craniofacial development. RT-qPCR of mRNA levels verified some significant

changes in the expression of these GLI targets in the developing facial prominences, yet there was not a clear, linear relationship between changes in GLI isoform production and target gene expression. Motif cluster analysis supported the hypothesis that GLI proteins work in concert with co-regulators. GBRs associated with GLI target genes were found to be situated within 1000 bp of binding motifs for several, previously identified GLI co-regulators. Differential expression within developing facial prominences of predicted co-regulators supported a hypothesis in which expression of GLI target genes is dependent upon the cooperative function of GLI isoforms and co-regulator proteins.

Together, these data provide a better understanding of the complex nature of GLI-mediated transcription that occurs in normal and ciliopathic craniofacial development.

# GLI Binding Regions Are Present throughout the Avian Genome in Genes That Affect Craniofacial Development

Work in other species has identified GBRs and examined the role of GLI proteins as transcription factors that affect gene expression in numerous signaling pathways (Vokes et al., 2007, 2008).

Three models currently exist to explain GLIFL/GLI<sup>T</sup> function (Falkenstein and Vokes, 2014). First the GLIFL::GLI<sup>T</sup> ratio sensing model suggest that the relative levels of GLIFL to GLI<sup>T</sup> is integrated and results in graded levels of transcription of targets. This model suggests that response to changes in ratio, rather than concentration, affect target gene expression. The threshold activation model suggests that, rather than ratio of GLIFL::GLIT, a threshold-specific concentration of GLIFL activates target gene expression (Oosterveen et al., 2012). The threshold repression model depends on the removal of the GLI<sup>T</sup> repressor from target genes, allowing for transcriptional activation by other transcription factors to initiate gene expression. From our results, mRNA expression of most target genes was not dramatically altered despite excessive GLIFL production in ta<sup>2</sup> mutant, indicating that the ratio sensing model cannot explain our observation in ta<sup>2</sup> mutant. Secondly, the accumulated GLI3FL in the nucleus was expected to bind and activate target gene in the threshold activation model; however, instead we found reduced occupancy of GLI3 at target genes, along with lack of detectable GLIFL binding to GBRs (**Figures 3**, **4**), suggesting that a dysfunctional GLIFL that fails to promote GLIA-mediated activation is produced under ciliopathic conditions. Thus, these data do not support the threshold activation model for GLIFL function. Conversely, our results showed that reduced GLI<sup>T</sup> production could lead to a de-repression of target genes and potentially a gene activation, when the required transcription factors are available. The threshold repression model seems to be more applicable to the observation owing to less GLI3 occupancy and the failure of GLIFL to GLI<sup>T</sup> conversion. A more in depth understanding of GLI mechanisms of action will require tools with the ability to definitively decipher between GLIFL and GLI<sup>T</sup> in vivo.

# Excessive GLIFL Production Does Not Equate to Increased GLI<sup>A</sup> Activity in Ciliary Mutants

A number of disorders identified as ciliopathies have craniofacial abnormalities including Oral-facial-digital syndrome, Joubert syndrome, Bardet-Biedl syndrome, Meckel-Gruber syndrome, Ellis-van Creveld syndrome (Zaghloul and Brugmann, 2011). The phenotypes for syndromes such as these, while not identical, do have several phenotypes indicative of aberrant SHH signaling including widening of the midface, cleft lip/palate, micrognathia, craniosynostosis and oral/dental anomalies (Zaghloul and Brugmann, 2011). Increased production of GLIFL isoforms has been observed in several ciliary mutants (Huangfu and Anderson, 2005; Liu et al., 2005; Humke et al., 2010), thus a common interpretation of these data is that ciliopathic phenotypes were due to either increased activator function or skewed GLIFL::GLI<sup>T</sup> ratio in favor of GLIFL. Despite these common interpretation regarding the molecular mechanism causing ciliopathic phenotypes, examination of subsequent levels of GLI processing, binding, and transcriptional function in ciliary mutants has not been performed.

Our studies are among the first to examine how and if GLI proteins function in the developing craniofacial complex of ciliary mutants; however, some questions remain regarding the mechanistic reasons as to why GLIFL appears not to function in ciliary mutants. Prior to processing, GLI proteins associate with Suppressor of Fused (SUFU), a conserved protein known to regulate the activity of GLI proteins via modulating GLI processing, stabilization and subcellular localization (Barnfield et al., 2005; Humke et al., 2010; Tukachinsky et al., 2010; Wang et al., 2010). In the presence of a SHH signal, the SUFU-GLIFL complex traffics through the cilium (Eggenschwiler and Anderson, 2007). Activated ciliary SMO then works through KIF7 to promote the dissociation of the inhibitory SUFU-GLIFL complex (Humke et al., 2010; Tukachinsky et al., 2010; Li et al., 2012). Free GLIFL is then processed into an activator and moves to the nucleus to activate downstream targets. In the absence of the SHH ligand, SMO is not translocated into the cilium and thus cannot antagonize SUFU. SUFU remains in complex with GLIFL, GLIFL is proteolytically processed into GLI<sup>T</sup> and the SUFU-GLI<sup>T</sup> complex moves to the nucleus where it recruits the Sap18-Sin3 co-repressor complex to repress GLI target genes (Ding et al., 1999; Kogerman et al., 1999; Cheng and Bishop, 2002; Paces-Fessy et al., 2004). Furthermore, our recent work with the murine ciliary mutant, Kif3fl/fl ;Wnt1-Cre shows increased SUFU production and nuclear localization, as well as enhanced association of SUFU with GLI3 (Chang et al., in press). We hypothesize a similar mechanism is at play in ta<sup>2</sup> mutants. Specifically, excessive GLIFL produced in ta<sup>2</sup> mutants cannot dissociate from SUFU because the complex cannot undergo ciliary trafficking. We further hypothesize that lack of dissociation prevents GLIFL from occupying GBRs and directly activating target gene transcription. Our future studies will address the association of GLIFL and SUFU in ta<sup>2</sup> mutants and determine if their maintained association contributes to ciliopathic phenotypes.

# GLI-Binding to Target Genes Occurs in a Prominence Specific Manner

Frequently the craniofacial complex is thought of as a singular organ system; however, the prominences that make up the face develop independently prior to their fusion. There is evidence to support that these prominences have distinct molecular profiles and develop as separate developmental fields (Brugmann et al., 2007). Our ChIP-qPCR results detected differential GLI binding to target genes in facial prominences of control embryos (**Figures 2B–D**). Further, the expression of co-regulators also followed a prominence specific pattern in both control and ta<sup>2</sup> embryos (**Figures 6A'–C'**). These data supported the hypothesis that during normal development, as well as when cilia are lost, each facial prominence uses a unique mechanism to transduce a SHH-dependent GLI signal. Based on our examination of coregulators, we hypothesize that there is a combinatorial code of GLI isoforms and co-regulators that work together to precisely regulate target gene expression. Thus, when cilia are lost and GLI production is altered, target gene expression is dictated by how the combinatorial code of remaining GLI isoforms and co-regulators function together.

In sum, our data suggest that the increased or decreased production of GLI isoforms alone is not sufficient to explain how target gene expression will be altered. To understand the molecular mechanisms responsible for ciliopathic, GLImediated phenotypes, future studies will have to account for tissue specificity, the presence or absence of co-regulators and the mode of GLI function (activation or de-repression) to begin to address this process.

# AUTHOR CONTRIBUTIONS

YC designed and tested all primers, and performed and interpreted all ChIP and qPCR experiments, PC performed in silico GBR analysis, ES harvested and photographed embryos, SB conceived the project, analyzed data and wrote the manuscript with input from all authors.

# FUNDING

This work was funded by the National Institutes of Health, National Institute of Dental and Craniofacial Research (NIDCR) [R00-DE01985 and [R01-DE023804] to SB] and by C.T.O. and Research in Progress (RIP) funds from the Cincinnati Children's Research Foundation to SB.

# ACKNOWLEDGMENTS

We thank the UC Davis Avian Facility and Jackie Pisenti for husbandry of the talpid<sup>2</sup> line and providing talpid<sup>2</sup> eggs for this study.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00468

Supplementary Figure 1 | GLI3 binding to GLI targets is confirmed by ChIP. ChIP-qPCR analyses of GLI3 precipitated with GLI binding regions of *ALX4* (A), *BMPR1A* (B), *OSR2* (C), *WIF1* (D), and *PTCH1* (E) in facial prominences of control embryo. Error bars are based on the standard error of the means (S.E.M.). *n* = 3.

# REFERENCES


Supplementary Figure 2 | ChIP-qPCR analyses of GLI3 precipitated with GLI binding regions of PTCH1 (A), TFAP2A (B), CTNNB1 (C), WNT2B (D), DKK1 (E), FGFR (F), BMP2 (G), BMP4 (H), BMP7 (I), and EPHA4 (J) in control and ta<sup>2</sup> facial prominences. Error bars are based on the standard error of the means (S.E.M.). *n* = 3.

Supplementary Figure 3 | The schematic of the clusters of transcription factor motifs in GBRs of GLI targets. The *in silico* analyses of GLI (blue circle), E-box (magenta triangle), SP site (red rectangle), and SOX site (yellow circle) in the GBRs of GLI targets. We defined 5′ -untranslated region (UTR), gene and 3′ -UTR as intragenic region, <1 kb upstream of transcription start site (TSS) is promoter region, 1–20 kb away from TSS as proximal regulatory region, and <100 kb away from TSS as distal regulatory region. The numbers labeled below the symbols indicate the positions of the motifs according to the distance away from TSS sites. The position at the upstream of TSS site is assigned a negative symbol. The primer sets used for ChIP-quantitative PCRs are labeled above the specific GLI binding motif (red line). Clusters of four motifs are highlighted in a black box, clusters of three motifs are highlighted in a red box.

### Supplementary Table 1 | Identified Gli target genes.


**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 © 2016 Chang, Chaturvedi, Schock and Brugmann. 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) or licensor 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.

# Mesenchymal Remodeling during Palatal Shelf Elevation Revealed by Extracellular Matrix and F-Actin Expression Patterns

Matthias Chiquet <sup>1</sup> \*, Susan Blumer <sup>1</sup> , Manuela Angelini <sup>1</sup> , Thimios A. Mitsiadis <sup>2</sup> and Christos Katsaros <sup>1</sup>

<sup>1</sup> Department of Orthodontics and Dentofacial Orthopedics, Medical Faculty, School of Dental Medicine, University of Bern, Bern, Switzerland, <sup>2</sup> Orofacial Development and Regeneration, Center for Dental Medicine, Institute for Oral Biology, University of Zurich, Zurich, Switzerland

During formation of the secondary palate in mammalian embryos, two vertically oriented palatal shelves rapidly elevate into a horizontal position above the tongue, meet at the midline, and fuse to form a single entity. Previous observations suggested that elevation occurs by a simple 90◦ rotation of the palatal shelves. More recent findings showed that the presumptive midline epithelial cells are not located at the tips of palatal shelves before elevation, but mostly toward their medial/lingual part. This implied extensive tissue remodeling during shelf elevation. Nevertheless, it is still not known how the shelf mesenchyme reorganizes during this process, and what mechanism drives it. To address this question, we mapped the distinct and restricted expression domains of certain extracellular matrix components within the developing palatal shelves. This procedure allowed to monitor movements of entire mesenchymal regions relative to each other during shelf elevation. Consistent with previous notions, our results confirm a flipping movement of the palatal shelves anteriorly, whereas extensive mesenchymal reorganization is observed more posteriorly. There, the entire lingual portion of the vertical shelves moves close to the midline after elevation, whereas the mesenchyme at the original tip of the shelves ends up ventrolaterally. Moreover, we observed that the mesenchymal cells of elevating palatal shelves substantially align their actin cytoskeleton, their extracellular matrix, and their nuclei in a ventral to medial direction. This indicates that, like in other morphogenetic processes, actin-dependent cell contractility is a major driving force for mesenchymal tissue remodeling during palatogenesis.

Keywords: palate morphogenesis, palatal shelf elevation, extracellular matrix, actin, tissue remodeling, mouse embryo

# INTRODUCTION

The secondary palate, which separates the oral cavity from the nasal cavity, is a specific feature of mammals (Ferguson, 1988). Cleft lip/palate is the most frequent craniofacial malformation in humans (Jugessur et al., 2009a; Mossey et al., 2009; Dixon et al., 2011). Formation of the secondary palate during embryogenesis is a complex process involving extensive tissue growth and remodeling (Gritli-Linde, 2008; Jugessur et al., 2009b; Bush and Jiang, 2012). Palate development

### Edited by:

Giovanna Orsini, Marche Polytechnic University, Italy

### Reviewed by:

Gianpaolo Papaccio, Seconda Università degli Studi di Napoli, Italy Catherine Chaussain, Paris Descartes University, France

\*Correspondence: Matthias Chiquet matthias.chiquet@zmk.unibe.ch

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 11 July 2016 Accepted: 23 August 2016 Published: 07 September 2016

### Citation:

Chiquet M, Blumer S, Angelini M, Mitsiadis TA and Katsaros C (2016) Mesenchymal Remodeling during Palatal Shelf Elevation Revealed by Extracellular Matrix and F-Actin Expression Patterns. Front. Physiol. 7:392. doi: 10.3389/fphys.2016.00392

**142**

starts from the palatal shelves, extensions of the maxillary processes that are located left and right from the tongue. The palatal shelves initially have a vertical orientation, but around the 6th week of gestation in humans, or at embryonic day 14.5 (E14.5) in mice, they elevate rapidly into a horizontal position above the tongue. The shelves meet medially, and fuse by disappearance of the midline epithelium and merging of the mesenchyme (Gritli-Linde, 2007; Bush and Jiang, 2012). While much information exists concerning the mechanism of palatal shelf fusion (Kaartinen et al., 1995; Proetzel et al., 1995) (Jin et al., 2014), the process of shelf elevation still remains elusive, and knowledge about its cellular and molecular basis is scarce (Bush and Jiang, 2012).

Palatal shelves are known to be divided into distinct regions along their anterior-posterior axis. Their frontal and medial parts will give rise to the hard palate, their posterior part to the soft palate (Yu and Ornitz, 2011; Smith et al., 2012). It was debated whether palatal shelves elevate and fuse first from the front or from the back (reviewed in Brinkley and Vickerman, 1979). In vitro, their medial parts elevate first (Brinkley and Vickerman, 1979), whereas in the embryo, tongue movements strongly affect shelf elevation (Iseki et al., 2007; Kouskoura et al., 2016). This causes a remarkable variability in the time course of elevation in vivo between siblings, and even between the two shelves of one embryo (Luke, 1984; Iseki et al., 2007; Yu and Ornitz, 2011). A recent histomorphometric analysis concludes that in the embryo, elevation usually starts at the medial to posterior part of the palatal shelves, which then zipper in both directions (Yu and Ornitz, 2011).

The above findings imply that all parts of the palatal shelves have an intrinsic capability to elevate. However, it has been debated how this reshaping occurs, and several reports indicated regional differences in tissue reorganization during shelf elevation. Histological observations suggested that anteriorly, the palatal shelves perform a simple flipping movement (Coleman, 1965). However, other reports indicated that the medial and posterior parts of the shelves remodel extensively during elevation, such that their lingual side protrudes whereas the ventral tip regresses (Brinkley and Vickerman, 1979; Chou et al., 2004). A recent study used the expression of the Mmp13 gene to determine the origin of midline epithelial cells during palatal shelf elevation. Its results supported a flipping movement of the shelves in the front and extensive remodeling posteriorly (Jin et al., 2010), which was confirmed by careful histomorphometric analysis (Yu and Ornitz, 2011).

Although there is general agreement that extensive remodeling of the palatal shelves occurs during elevation, it is not known how different regions of the palatal mesenchyme move relative to each other. Moreover, the forces that drive tissue remodeling in the elevating shelves are still unknown. Distinct patterns of cell proliferation (Sasaki et al., 2004) and density (Brinkley and Bookstein, 1986) are unlikely to contribute substantially to the remodeling process, since it occurs in less than 1 day in mouse embryos (Yu and Ornitz, 2011). The finding of hyaluronidase-sensitive material in the "hinge" region of palatal shelves led to the hypothesis that rapid accumulation of hyaluronic acid or other glycosaminoglycans might provide the driving force for shelf elevation, by generating osmotic swelling pressure (Brinkley and Morris-Wiman, 1987). Alternatively, this process might be powered by coordinated cytoskeletal contractions of both epithelium and mesenchyme, similar to what is observed in many other morphogenetic tissue rearrangements (Wozniak and Chen, 2009). Although an early report suggested a role for actin-based contractility in palatal shelf elevation (Lessard et al., 1974), to our knowledge this idea has never been followed up.

Therefore, in the present study we address two important questions concerning the process of palatal shelf elevation. Firstly, is it possible to detect rearrangements of different parts of the palatal mesenchyme relative to each other during the elevation process? Secondly, can we find any evidence for actinbased contractility in the remodeling of palatal mesenchyme during elevation? To address these points, we took advantage of the fact that certain extracellular matrix (ECM) components have a restricted expression in defined regions of the palatal mesenchyme, which can be followed during shelf elevation. Furthermore, we stained sections through elevating palatal shelves for F-actin, and detected elongated cell nuclei that aligned with the actin network. Our results confirm and refine previous findings of extensive remodeling of the palatal mesenchyme at medial and posterior levels, and provide the first evidence for tensile stress acting within this tissue during the process of shelf elevation.

# MATERIALS AND METHODS

# Animals, Embryonic Tissues, and Cryosectioning

C57BL/6 wild-type mouse embryos were obtained from J.-F. Spetz at the Friedrich-Miescher Institute for Biomedical Research in Basel, Switzerland, or from in house breeding at the central animal facilities, Department of Clinical Research, University of Bern. After mating, appearance of a vaginal plug was considered embryonic day 0.5 (E0.5). Pregnant females were sacrificed at the desired stage (E13.5–E14.5), embryos were removed from the uterus and decapitated. Animal experiments were approved by the Cantonal Veterinary Offices of Basel, Zurich and Bern, Switzerland. The embryo heads were washed in ice-cold PBS, fixed in 4% paraformaldehyde in phosphate buffered saline (PBS; 150 mM NaCl, 20 mM Na-phosphate, pH 7.4) overnight, soaked for 24 h in 30% sucrose in PBS, embedded in Tissue Tek (O.C.T. compound; Sakura Finetek Europe B.V., Zoeterwoude, Netherlands), and frozen on an aluminum block cooled to −80◦C. Frozen heads were stored at −80◦C before sectioning. Serial frontal sections (12 µm thick) were prepared on a Cryocut E cryomicrotome (Reichert-Jung, Leica Microsystems, Heerbrugg, Switzerland). Sections were dried at 37◦C for 1–5 min, and stored at −80◦C for further use.

# Gene-Specific RNA Probes and In situ Hybridization

Total RNA was isolated from E14.5 C57BL/6 wildtype mouse embryos using an RNAeasy Mini Kit (Qiagen, Hombrechtikon, Switzerland), and reverse transcribed to cDNA using Moloney murine leukemia virus reverse transcriptase (Promega, Dübendorf, Switzerland). Gene-specific primers (Microsynth, Balgrach, Switzerland) were designed using free NCBI software (http://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi?

LINK\_LOC=BlastHome). Primers were fitted with BamH1 (forward primers) or HindIII (reverse primers) restriction sites at their 5′ ends, respectively. With these primers (**Table 1**), specific products were amplified from E14.5 mouse cDNA by PCR using Go Taq polymerase (Promega), cut with respective restriction enzymes, and cloned into pBluescript SK+ plasmid (Stratagene/Agilent, Santa Clara, USA). In the case of mouse tenascin-C (Tnc) and tenascin-W (gene name Tnn), full-length cDNA clones were obtained from R. Chiquet-Ehrismann (Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland), and restriction fragments of suitable size (**Table 1**) were cloned into pBluescript SK+. Plasmids were linearized by cutting upstream or downstream of the insert, respectively, and digoxygenin-labeled anti-sense or sense RNA probes were synthesized (Koch et al., 1995) with a labeling kit from Roche Diagnostics using T3 or T7 RNA polymerase (Promega). In situ hybridization was done with the labeled probes as published in detail before (Flück et al., 2000). After incubation with alkaline phosphatase conjugated anti-digoxygenin antibody (Roche Diagnostics), a color substrate solution was used to which 10% polyvinyl alcohol (MW 31′ 000-50′ 000; Sigma-Aldrich, Buchs, Switzerland) was added (Shen, 2001). After development, the sections were counterstained with Nuclear Fast Red (Sigma).

# Fluorescence Labeling of Tissue Sections

Monoclonal antibody mTn12 against mouse tenascin-C (Aufderheide and Ekblom, 1988) was obtained from

### TABLE 1 | Gene-specific RNA probes for in situ hybridization.

### MOUSE PERIOSTIN (Postn: NM\_015784.3)

Forward primer: 5′CCGGATCCGCAGCCGCCATCACCTCTGAC 3′ (nucleotide: 196-216)

Reverse primer: 5′ CCAAGCTTGCTTGTCTTGCCGCAGCTTGT 3′ (nucleotide: 964-944)

PCR product/probe length (without restriction sites): 769 bp

### MOUSE TENASCIN-C (Tnc: D\_90343.1)

Restriction fragment: Sac1 (nucleotide 5862)–EcoR1 (nucleotide 6813) Probe length: 950 bp

### MOUSE TENASCIN-W (Tnn: AJ\_580920.1)

Restriction fragment: HindIII (nucleotide 1190)–BamH1 (nucleotide 1440) Probe length: 250 bp

### MOUSE SMOC-2 (Smoc2: NM\_022315.2)

Forward: 5 ′ CCGGATCCCAGGTGCCCCGGTTCAA 3′ (nucleotide: 661-679) Reverse: 5 ′ CCAAGCTTCCAGGGTGTGGCTGGGGT 3′ (nucleotide: 1253-1235) PCR product/probe length (without restriction sites): 593 bp

### MOUSE TRANSFORMING GROWTH FACTOR-β3 (Tgfb3: NM 009368.3)

Forward primer: 5′ CCGGATCCCAACCCCAGCTCCAAGCG 3′ (nucleotide: 1578-1597)

Reverse primer: 5′ CCAAGCTTCCAGGTTGCGGAAGCAGT 3′ (nucleotide: 2046-2026)

PCR product/probe length (without restriction sites): 469 bp

R. Chiquet-Ehrismann (Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland). Cryosections from paraformaldehyde-fixed mouse embryo heads were blocked with 3% bovine serum albumin in phosphate-buffered saline (BSA/PBS) and incubated with anti-tenascin antibody (1:100 in BSA/PBS) for 1 h at room temperature. After washing with PBS containing 0.2% Tween-20, sections were incubated with Alexa488-labeled secondary antibody (Invitrogen; 1:1000 in BSA/PBS). For labeling the F-actin network on the same section, 1 µg/ml rhodamine-phalloidin (Sigma) was added to the secondary antibody solution. After washing again with PBS/Tween, sections were embedded in phosphate-buffered glycerol (pH 7.4) containing 1 µg/ml DAPI stain (Sigma) to label nuclei.

# Microscopy

Slides processed for in situ hybridization were viewed under bright field optics with 4x or 10x objectives on an Olympus BX-51 microscope. Slides triple-labeled with fluorescent dyes were observed on the same microscope by epifluorescence optics, using 10x and 20x fluorescence objectives and the following Olympus filter/mirror series: U-MWIBA3 for Alexa488; U-MWIGA3 for rhodamine; U-MNUA2 for DAPI. Digital images were recorded using a ProgRes CT3 CMOS camera and ProgRes Capture Pro software (Jenoptik, Jena, Germany). Slides from one experiment were photographed at identical camera settings, and resulting images were processed identically.

# Image Analysis

DAPI-labeled nuclei on tissue sections were color-coded depending on their aspect ratio (ratio of x to y axis) in the following way. Images were imported into ImageJ, binarized, and the Watershed tool was used to separate touching/overlapping nuclei. Particles (i.e., DAPI-stained nuclei) were fitted with ellipses, and their aspect ratio was analyzed. A ROI (region of interest) color coder plugin with the "Phase" LUT (look up table) was then used to color-code all nuclei with an aspect ratio of >2.5 and above in red, 2.0–2.5 in pink, and <2.0 in blue.

# RESULTS

# Periostin as a Mesenchymal Marker for Palatal Shelves during Elevation

To generate a fate map of palatal mesenchyme and epithelium during the process of shelf elevation, we screened the expression patterns of more than 20 ECM components by in situ hybridization on frontal sections of wild-type C56/BalbC mouse embryo heads during the relevant stages, i.e., E13.5–E14.5. Of these genes, the expression of eight was highly enriched in either all or just a specific region of the palatal shelves compared to the surrounding maxillary tissue: Periostin, tenascin-C, tenascin-W, collagen types II, IX, XI, and XIV, and SMOC-2.

Periostin (gene name Postn) is an ECM component belonging to the "matricellular" proteins (Murphy-Ullrich and Sage, 2014) and known to be involved in morphogenesis and wound healing (Walker et al., 2016). Periostin protein was demonstrated in mouse palatal shelves by immunofluorescence (Oka et al., Chiquet et al. Remodeling of Elevating Palatal Shelves

2012). Here, we used in situ hybridization to reveal the expression pattern of Postn mRNA in palatal shelves of wild-type mouse embryos immediately before (E13.5) and after (E14.5) elevation. Serial frontal cryosections through palatal shelves were hybridized with Postn-specific RNA probe at three levels: anterior (future anterior hard palate), medial (future posterior hard palate), and posterior (future soft palate). Anteriorly, Postn mRNA was expressed in a broad layer of mesenchyme on both the buccal (ventrolateral) and lingual (medial) aspect of E13.5 palatal shelves (**Figure 1A**). After shelf elevation (E14.5), the layer of Postn expression was located ventrolaterally in the shelves, and filled their medial tips (**Figure 1B**). At the medial level, a strong Postn signal covered the entire mesenchymal area of vertical palatal shelves at E13.5 (**Figure 1C**). After elevation at this level, Postn expression was observed in a broad area around the midline, from where it extended into a ventral stripe facing the oral cavity (**Figure 1D**). In contrast and remarkably, the dorsolateral regions of the elevated shelves were free of Postn signal (**Figure 1D**). A similar distribution was seen at the posterior level both before and after elevation, except that the signal was weaker in the central mesenchyme region of the shelves (**Figures 1E,F**).

# Tenascin-C Expression in the Medial Palatal Mesenchyme Both before an after Shelf Elevation

Tenascin-C (gene name Tnc) is another "matricellular" ECM protein with highly regulated patterns of expression in vertebrate morphogenesis and pathologies (Chiquet-Ehrismann and Chiquet, 2003). Like periostin, tenascin-C protein has been detected in the developing secondary palate before by immunofluorescence (Ferguson, 1988). At the anterior level at E13.5, the Tnc mRNA expression pattern resembled that of periostin: In vertical shelves, the signal was found in the sub-epithelial mesenchyme both on the lingual (medial) and the buccal (ventrolateral) aspect (**Figure 2A**). After elevation, Tnc mRNA was concentrated at the tips of the shelves and extended laterally into the subepithelial mesenchyme both dorsally and ventrally (**Figure 2B**). A different distribution was observed at the medial level, i.e., in the future posterior hard palate. There, Tnc expression covered the entire lingual half of the vertical shelves at E13.5, whereas the buccal half was devoid of signal except for a thin stripe underneath the epithelium (**Figure 2C**). In elevated shelves (E14.5) at this level, Tnc mRNA was observed in a broad mesenchymal region that extended from dorsal to ventral around the midline, and in a thin stripe underneath the ventral epithelium (**Figure 2D**). A similar distribution was seen for Tnc at the posterior level (future soft palate), although the signal was weaker (**Figures 2E,F**).

# Dorsomedial Tenascin-W Expression in Palatal Shelves Both before and after Shelf Elevation

Tenascin-W (gene name Tnn) is the latest member of this family of ECM proteins (Scherberich et al., 2004). During embryogenesis, its expression partially overlaps with that of tenascin-C but is even more restricted. Notably, tenascin-W was shown to be involved in osteogenesis (Martina et al., 2010) and is a marker for osteogenic areas in embryos (Scherberich et al., 2004). In anterior palatal shelves at E13.5, we found Tnn to be expressed exclusively in the dorsomedial quadrant of the mesenchyme, whereas the tips of the vertical shelves were devoid of signal (**Figure 3A**). At E14.5, the Tnn signal was restricted to the dorsal mesenchyme around the midline of the elevated shelves (**Figure 3B**). A similar distribution was found at the medial level both at E13.5 and E14.5, except that in the elevated shelves the area of Tnn expression was spread out laterally and filled the entire dorsal half of the shelf mesenchyme (**Figures 3C,D**). More posteriorly, the Tnn signal was present in addition around the developing os palatinum (**Figures 3E,F**).

# SMOC-2 Expression at Palatal Shelf Tips before, but Not at the Midline after Elevation

SMOC-2 (SPARC related modular calcium binding 2; gene name Smoc2) belongs to the SPARC/Osteonectin family of matricellular proteins (Vannahme et al., 2003) and is involved in keratinocyte migration (Maier et al., 2008) and angiogenesis (Rocnik et al., 2006). In mouse embryos, Smoc2 is prominently expressed in perichondrial mesenchyme (**Figure 4A**). Surprisingly, Smoc2 mRNA is also detected in defined areas of the oral epithelium. At a medial to posterior level before elevation (E13.5), the Smoc2 signal outlines the buccal surface of the palatal shelves and extends into the epithelium at their tips, but finishes abruptly at the lower end of their lingual aspect (**Figure 4A**). At E14.5, Smoc2 mRNA is present in the entire ventral epithelium of the elevated shelves, but stops shortly before the midline (**Figure 4B**). A similar expression pattern in palatal epithelium before and after elevation was found for type XIV collagen, also named undulin, a minor fibril-associated collagen (Ricard-Blum, 2011) (not shown). For comparison, sections from a similar level were hybridized with a Tgfβ3 probe. This growth factor is secreted by midline epithelial cells and required for palatal shelf fusion (Kaartinen et al., 1995). At E13.5, Tgfβ3 mRNA extended from the shelf tip about halfway up the lingual epithelium (**Figure 4C**), and at E14.5 was found at the midline (**Figure 4D**). Thus, Smoc2 and Tgfb3 exhibit a reciprocal expression pattern in the palatal epithelium, except for a short overlap at the shelf tip before, or close to the ventral midline after elevation, respectively.

# Mesenchymal and Epithelial Rearrangements during Palatal Shelf Elevation at the Medial Level

Since the mRNAs of the chosen ECM genes were enriched in distinct areas of the palatal shelves before and after elevation, this allowed the mapping of their combined expression patterns (**Figure 5**; compare with **Figures 1**–**4**). At the medial level, the combined results implied that elevation is not due to a simple rotation of the palatal shelves from the vertical into the horizontal plane. If this were the case, the Tnc-expressing lingual half of the vertical shelves should end up dorsally in the elevated secondary palate. Instead, the Tnc-expressing

FIGURE 1 | Periostin expression as a marker for palatal shelves during elevation. In situ hybridization for Postn mRNA on frontal sections of palatal shelves from wild-type mouse embryos at the anterior (A,B), medial (C,D), and posterior (E,F) level. Serial sections were obtained from an E13.5 (A,C,E) and an E14.5 (B,D,F) embryo, respectively. For description, see text. Abbreviations: palatal shelf (p), tongue (t). Bar, 400 µm.

area had moved to a broad zone at the midline, whereas the lateral parts of the elevated shelves were devoid of Tnc signal. The Postn signal, which filled the entire shelves at E13.5, was drawn out after elevation into a broad tringle that filled the entire mesenchyme around the midline, and tapered on the ventrolateral aspect of the shelves (cf. **Figure 1D**). Mesenchyme around the nasopharyngeal fold that forms at E.14.5 lacked expression of Postn (**Figure 5**), suggesting that maxillary tissue originally outside the palatal shelf proper is drawn into the elevated shelves at E14.5. Thus, at this level the palatal shelf bulges out medially whereas the original tip retracts during elevation. This involves massive reorganization of the palatal mesenchyme: The lingual mesenchyme of the E13.5 shelf, rather than that at the distal tip, forms the tissue around the midline at E14.5. The mesenchyme at the original tip ends up more laterally after elevation, whereas the originally buccal mesenchyme distorts and

FIGURE 2 | Tenascin-C expression primarily in the lingual/medial half of palatal shelves both before and after shelf elevation. In situ hybridization for Tnc mRNA on frontal sections of palatal shelves from wild-type mouse embryos at the anterior (A,B), medial (C,D), and posterior (E,F) level. Serial sections were obtained from an E13.5 (A,C,E) and an E14.5 (B,D,F) embryo, respectively. For description, see text. Abbreviations: palatal shelf (p), tongue (t). Bar, 400 µm.

spreads out along the ventrolateral aspect of the elevated shelves. Remodeling of the mesenchyme is paralleled by reorganization of the palatal epithelium. Smoc2 mRNA was detected in shelf tip epithelium before, but not in midline epithelium after elevation. Conversely, Tgfβ3 was expressed primarily in lingual epithelium of shelves before, and in midline epithelium after elevation (**Figures 4**, **5**).

# Anterior to Posterior Differences in ECM Rearrangement during Shelf Elevation

In wild-type Balb/C litters collected at E14.5, it is quite frequent to find embryos in which either one or both of the palatal shelves have not elevated yet (Luke, 1984; Yu and Ornitz, 2011). These mice can be compared to their siblings with already elevated shelves, and one representative example is shown in

**Figure 6**. The images show frontal sections at three levels through the palatal shelves of two different E14.5 embryos from the same litter, stained by fluorescence for actin (red) and tenascin-C (green). **Figure 6A** shows a still vertical palatal shelf of one E14.5 embryo at the anterior level, and it is noteworthy that a prominent epithelial invagination has formed on its buccal side. Staining for tenascin-C protein largely fills the shelf mesenchyme, except for its central area. At the same level in the sibling (**Figure 6B**), its elevated shelf occupies the space between nasal process and tongue, and the epithelial invagination has disappeared. Tenascin-C is present in the medial and ventrolateral parts of the shelf. Since tenascin-C positive regions largely correspond to each other before and after elevation, these observations agree with a flipping movement of the anterior palatal shelves. On the other hand, comparison of shelves from the same two embryos also confirms that

extensive mesenchymal reorganization takes place at the medial and posterior level during elevation. At the medial level, a small protrusion has formed on the dorsolingual aspect of the still vertical shelf (**Figure 6C**). Tenascin-C is found primarily in the lingual half of the shelf mesenchyme, with only a narrow ribbon underneath the buccal epithelium. In case of the elevated shelf at the same level (**Figure 6D**), the entire tenascin-C containing mesenchymal shelf ECM between the original tip and the dorsolingual protrusion appears compressed between nasal process and tongue. Remarkably, the palatal artery appears circular in cross-section by actin staining before elevation, but is flattened in the vertical direction after elevation (c.f. **Figures 6C,D**). At the posterior level, the dorsolingual protrusion is very prominent on the not yet elevated shelf, forming a second tip (**Figure 6E**). Tenascin-C protein is enriched in the mesenchyme between the two shelf tips, rather than at the tips themselves. At this level, it is most obvious that the entire tenascin-C containing mesenchyme has moved to the midline after elevation (**Figure 6F**). Rather than a flipping movement as seen at the front, this again implies a slug-like reshaping of the medial/posterior palatal shelf, with a deformation and medial movement of the lingual mesenchyme (including the palatine artery) and a retraction of the original shelf tip.

# Orientation of Actin Fibers and Nuclei in Palatal Mesenchyme during Elevation

Fluorescently labeled phalloidin is a specific marker for actin stress fibers of cells in culture (Small et al., 1999), and can also be used to label the cytoskeletal F-actin network of embryonic tissues (**Figures 7A,B**). As a major component of the contractile apparatus, actin stress fibers are known to align with the major direction of strain within a cell (Burridge and Wittchen, 2013). In **Figure 7A**, a higher magnification of a E14.5 palatal shelf at the medial level just before elevation (c.f. **Figure 6C**) is shown after phalloidin labeling. Note that the F-actin network of mesenchymal cells is aligned within this shelf, pointing toward the original ventral tip and the newly forming lingual protrusion, and forming an arc between them. In already elevated shelves of the same stage, the cellular actin network appears less organized, and no preferred direction is apparent in the shelf proper (**Figure 7B**). We also determined, on the same sections, the deformation of DAPI-labeled cell nuclei in the palatal mesenchyme. Using ImageJ, the ellipticity

level before (E13.5; above) and after (E14.5; below) elevation. The approximate areas of mesenchymal expression of periostin, tenascin-C, tenascin-W, and epithelial expression of Smoc2 and Tgfβ3 are indicated. Note that tenascin-C, tenascin-W, and Tgfβ3 expressing areas all end up at the midline after elevation, although their center of expression does not localize to the shelf tip before elevation. Conversely, Smoc2 mRNA is epithelially expressed at the shelf tip before, but excluded from midline epithelium after elevation.

of stained nuclei was measured, and nuclei with an aspect ratio of >2.5 were labeled in red. As can be seen in **Figure 7C**, elongated (red) nuclei are enriched in, and align with, the arclike F-actin network that is visible between the ventral and the lingual tips of this shelf before elevation. In contrast, elongated nuclei appear sparse and have no preferred direction in the elevated shelf (**Figure 7D**); they are more frequent in lateral maxillary mesenchyme. Note further that antibody to tenascin-C labels an ECM meshwork that lies parallel to the actin arc and the elongated nuclei in elevating shelves (**Figure 7E**). After elevation, the tenascin-C positive ECM has moved toward the midline as seen before, and the ECM fibrils appear randomly arranged (**Figure 7F**). These observations suggest that tensile stress builds up in the palatal shelves before elevation, which is released after the shelves have reached the horizontal position.

# DISCUSSION

# Remodeling of Palatal Shelf Mesenchyme during Elevation Revealed by ECM Expression Patterns

In the present study, we took advantage of the differential expression patterns of certain ECM genes to monitor tissue rearrangements during palatal shelf elevation. After screening over 20 ECM genes by in situ hybridization before and after elevation, we identified a few that exhibited expression patterns restricted to specific areas of the shelf mesenchyme or epithelium. Genes for ECM components are strongly expressed, and the turnover of their mRNAs is relatively slow (Eckes et al., 1996). We therefore argue that the observed changes in ECM expression patterns reflect morphogenetic rearrangements of palatal shelf tissue, rather than alterations in the rate of gene transcription. In the case of tenascin-C, in situ hybridization results were further confirmed by antibody stainings. This makes it even more unlikely that the observed dynamic changes in tenascin-C patterns during shelf elevation are due to rapid alterations in its synthesis.

Our results confirm the findings of Jin et al. (2010) concerning palatal shelf remodeling during elevation (see below). In addition, they extend earlier studies by providing more detailed information on the reshaping and medial translocation of the lingual shelf mesenchyme during this morphogenetic process. A further observation is noteworthy: During elevation, epithelial folds form dorsolaterally from the palatal shelves; these folds mark the lateral edges of the nasal cavity. Interestingly, the mesenchyme surrounding these folds express little or no periostin, which before elevation is a marker for the entire shelf proper. This suggests that at the epithelial folds, some originally nasal mesenchyme might be pulled into the elevating shelves. We have shown before that the epithelial folds are characterized by high expression of Mmp2 mRNA and gelatinase activity, which presumably is needed there to avoid closure of the nasal cavity (Gkantidis et al., 2012). From our results, these specialized structures might originate from tissue outside the original shelf proper.

The mechanism of shelf elevation during secondary palate formation in mammals has been disputed (Brinkley and Vickerman, 1979; Jin et al., 2010; Bush and Jiang, 2012). Early reports favored a simple flipping movement of the vertical palatal shelves into a horizontal position (Coleman, 1965), but it was soon postulated that posteriorly the palatal shelves reorganize differently. From histological observations, it appeared that their lingual/medial aspect bulged out whereas their original ventral tip retracted (Brinkley and Vickerman, 1979). This process requires extensive remodeling of the entire shelf, but the underlying mechanism remained obscure. Recently, Jin et al. (2010) found that matrix metalloproteinase Mmp13, a specific marker for midline epithelial cells, was expressed in still vertical E14.5 shelves posteriorly not at their tip but on their lingual/medial aspect, whereas anteriorly the Mmp13 signal extended into the shelf tip. The authors concluded that only anteriorly, the original tip epithelium will form the midline of elevated palatal shelves, whereas in middle and posterior regions, the midline epithelial cells are derived from the lingual epithelium of vertical shelves (Jin et al., 2010). These findings were in agreement with the earlier hypothesis that upon elevation, shelves undergo a flipping motion anteriorly, and a reshaping in middle and posterior regions.

(Jin et al., 2010) also showed that the mesenchymal transcription factor Goosecoid was expressed around the midline in E14.5 embryos, but in the lingual rather than the tip mesenchyme of non-elevated shelves of the same stage. This provided the first indication for large scale reorganization of

palatal shelves at the anterior (A,B), medial (C,D), and posterior (E,F) level were double-stained with rhodamine-phalloidin for actin (red) and antibody to tenascin-C (green), and viewed in a fluorescence microscope. Sections were obtained from two embryos of the same E14.5 litter; one in which the shelves had just started to elevate (A,C,E), and one in which they had already risen above the tongue (B,D,F). Abbreviations: palatal shelf (p), tongue (t), nasal septum (n). The palatine artery is marked with an arrow in (C,D), and shown enlarged in the inserts. Bar, 400 µm.

pre-existing mesenchyme during shelf elevation, rather than for differential cell growth as it has been suggested (Brinkley and Bookstein, 1986; Sasaki et al., 2004). In any case, the rapid and often asynchronous time course of shelf elevation (Yu and Ornitz, 2011), as well as our present results, strongly argue against differential cell growth as a mechanism for palatal shelf remodeling.

# Actin-Based Cellular Contractility Combined with Differential ECM Stiffness as a Possible Mechanism for Shelf Remodeling during Elevation

Cells attach to the ECM via integrins, which in turn are linked to the intracellular actin cytoskeleton. Embryonic cells and their

shelves were triple-stained with phalloidin for actin (A,B), DAPI for nuclei (C,D), and antibody to tenascin-C (E,F), respectively, and viewed in a fluorescence microscope. On the left (A,C,E), a shelf just before elevation is shown, and on the right (B,D,F) an already elevated shelf from an embryo of the same litter. The images (C,D) were processed by ImageJ, such that all elongated nuclei with an aspect ratio above 2.5 are false-colored in red, those with 2.0–2.5 in pink. Roundish nuclei (aspect ratio <2.0) are blue. Note that before elevation (C), many elongated (red/pink) nuclei are present between the original tip and the medial protrusion of the elevating shelf (arrowheads in A,C,E), and that their direction parallels that of the actin (A) and ECM (E) meshwork. The palatine artery is marked with an arrow in (A,B). Abbreviations: palatal shelf (p), tongue (t), nasal septum (n). Bar, 200 µm.

ECM thereby form a mechanical continuum, in which actingenerated tensile stresses are transmitted from one cell to the next (Wozniak and Chen, 2009). Stresses align the intracellular F-actin, and also reorient and organize extracellular ECM fibers (Burridge and Wittchen, 2013). Moreover, cell nuclei are mechanically connected to the actin cytoskeleton and deformed by tensile stresses that pull on the cell (Brosig et al., 2010). Thus, preferred orientations of the interconnected actin-integrin-ECM meshwork, together with elongated cell nuclei, reflect vectors of tensile stress that act within embryonic mesenchyme. The fungal toxin phalloidin specifically binds to cellular F-actin (Small et al., 1999). Tenascin-C is known to decorate ECM fibrils in embryonic tissues (Chiquet-Ehrismann and Chiquet, 2003). We used these two tools, together with measurements of the ellipticity of cell nuclei, to determine whether tensile stresses might occur in palatal shelves during their elevation. Our results clearly indicate that in the medial and posterior parts of the elevating shelf, the mesenchyme is tensed between the original tip and the newly forming lingual/medial protrusion. It is more than likely that such tensile stress is generated by F-actin dependent contraction of the shelf mesenchymal cells themselves, and that it is transmitted onto the ECM in which they are embedded.

Despite of several hypotheses that have been put forward over the years, the forces driving palatal shelf elevation were essentially unknown. Walker and Fraser (1956) first postulated an "internal shelf force." A popular hypothesis, still cited in recent reviews (Gritli-Linde, 2007), postulated that a cushion of hyaluronic acid in the "hinge" region of the palatal shelves provides a swelling force by generating osmotic pressure, which pushes up the shelf (Brinkley and Morris-Wiman, 1987). This idea was based on the presence of hyaluronidase-sensitive material in the respective region of the palatal shelf (Brinkley and Morris-Wiman, 1987), and on the fact that a glycosaminoglycan synthesis inhibitor caused cleft palate in mice (Brinkley and Vickerman, 1982). However, the main effect of the inhibitor was to cause a substantial shrinkage in shelf size compared to controls (Brinkley and Vickerman, 1982). Therefore, the inhibitor was likely to cause its damage long before shelf elevation occurred. In essence, there is no direct evidence that hyaluronic acid or other glycosaminoglycans are involved in generating the forces for shelf elevation.

Actin-based cellular contractility is well recognized for driving diverse morphogenetic movements during embryogenesis of both vertebrate and non-vertebrate species (Wozniak and Chen, 2009). Compared to alternatives such as differential cell growth or accumulation of ECM, this mechanism has a large advantage for the involved cells. Namely, they can control it rapidly and precisely in time and space, by activating the intracellular RhoA/ROCK signaling pathway that triggers actomyosin contraction (Burridge and Wittchen, 2013). Surprisingly, although actin contractility has been suggested as a mechanism for palatal shelf elevation early on (Lessard et al., 1974), this idea has apparently not been followed up in more recent years, but is now strongly supported by our current findings.

An actin-based mechanism for palatal shelf reorganization by no means implies that ECM itself has no function in the

# REFERENCES


process. ECM sustains and counteracts forces generated by cellular contractility, and its local stiffness determines to which extent embedded cells are able to deform the tissue (Shawky and Davidson, 2015). Thus, if palatal shelf elevation is driven by cellular actin contraction, differences in ECM composition and stiffness within the shelf might have substantial effects on reshaping of the tissue. Moreover, specific ECM components can modulate cell adhesion, and thereby affect cell contractility. For example, tenascin-C inhibits RhoA/ROCK signaling (and thus actomyosin contraction) by interfering with cell adhesion to fibronectin (Midwood and Schwarzbauer, 2002). It can therefore be speculated that cell contraction is diminished in tenascin-C rich regions within the palatal mesenchyme. Future studies involving a careful mapping of F-actin dynamics, as well as of ECM composition and stiffness, within elevating palatal shelves will be required to fully understand this complex morphogenetic process.

# AUTHOR CONTRIBUTIONS

MC, SB, TM, and CK contributed to the conception and design of the work; MC, SB, and MA were responsible for data acquisition, analysis, and interpretation; MC drafted the manuscript; SB, MA, TM, and CK critically revised the manuscript; MC, SB, MA, TM, and CK approved the manuscript; MC, SB, MA, TM, and CK agreed to be accountable for all aspects of the work.

# FUNDING

MC, TM, and CK were supported by grant 31003A\_146825 from the Swiss National Science foundation.

# ACKNOWLEDGMENTS

We are grateful to François Spetz and Dr. Ruth Chiquet-Ehrismann (Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland), Dr. Daniel Graf (University of Alberta, Edmonton, Canada), and Younes El Fersioui (Department of Orthodontics and Dentofacial Orthopedics, University of Bern, Switzerland) for providing us with tissue and reagents for this study. We thank Sabrina Ruggiero for technical help. MC would like to express his profound gratitude to Ruth Chiquet-Ehrismann, close collaborator, friend and partner for more than 30 years, for her unfailing support throughout his career. Ruth sadly and unexpectedly died on September 4, 2015.

Brinkley, L. L., and Vickerman, M. M. (1979). Elevation of lesioned palatal shelves in vitro. J. Embryol. Exp. Morphol. 54, 229–240.


and migration. Exp. Cell Res. 314, 2477–2487. doi: 10.1016/j.yexcr.2008. 05.020


**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 © 2016 Chiquet, Blumer, Angelini, Mitsiadis and Katsaros. 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) or licensor 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.

# MORN5 Expression during Craniofacial Development and Its Interaction with the BMP and TGFβ Pathways

Petra Cela1, 2 , Marek Hampl 1, 2, Katherine K. Fu<sup>3</sup> , Michaela Kunova Bosakova<sup>4</sup> , Pavel Krejci 4, 5, Joy M. Richman<sup>3</sup> and Marcela Buchtova1, 2 \*

1 Institute of Animal Physiology and Genetics, v.v.i., Academy of Sciences of the Czech Republic, Brno, Czech Republic, <sup>2</sup> Department of Animal Physiology and Immunology, Institute of Experimental Biology, Masaryk University, Brno, Czech Republic, <sup>3</sup> Life Sciences Institute, University of British Columbia, Vancouver, BC, Canada, <sup>4</sup> Department of Biology, Faculty of Medicine, Masaryk University, Brno, Czech Republic, <sup>5</sup> International Clinical Research Center, St. Anne's University Hospital, Brno, Czech Republic

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Amel Gritli-Linde, University of Gothenburg, Sweden Javier Catón, CEU San Pablo University, Spain David Clouthier, University of Colorado Anschutz Medical Campus, USA

> \*Correspondence: Marcela Buchtova buchtova@iach.cz

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 05 June 2016 Accepted: 17 August 2016 Published: 31 August 2016

### Citation:

Cela P, Hampl M, Fu KK, Kunova Bosakova M, Krejci P, Richman JM and Buchtova M (2016) MORN5 Expression during Craniofacial Development and Its Interaction with the BMP and TGFβ Pathways. Front. Physiol. 7:378. doi: 10.3389/fphys.2016.00378 MORN5 (MORN repeat containing 5) is encoded by a locus positioned on chromosome 17 in the chicken genome. The MORN motif is found in multiple copies in several proteins including junctophilins or phosphatidylinositol phosphate kinase family and the MORN proteins themselves are found across the animal and plant kingdoms. MORN5 protein has a characteristic punctate pattern in the cytoplasm in immunofluorescence imaging. Previously, MORN5 was found among differentially expressed genes in a microarray profiling experiment of the chicken embryo head. Here, we provided in situ hybridization to analyse, in detail, the MORN5 expression in chick craniofacial structures. The expression of MORN5 was first observed at stage HH17-18 (E2.5). MORN5 expression gradually appeared on either side of the primitive oral cavity, within the maxillary region. At stage HH20 (E3), prominent expression was localized in the mandibular prominences lateral to the midline. From stage HH20 up to HH29 (E6), there was strong expression in restricted regions of the maxillary and mandibular prominences. The frontonasal mass (in the midline of the face) expressed MORN5, starting at HH27 (E5). The expression was concentrated in the corners or globular processes, which will ultimately fuse with the cranial edges of the maxillary prominences. MORN5 expression was maintained in the fusion zone up to stage HH29. In sections MORN5 expression was localized preferentially in the mesenchyme. Previously, we examined signals that regulate MORN5 expression in the face based on a previous microarray study. Here, we validated the array results with in situ hybridization and QPCR. MORN5 was downregulated 24 h after Noggin and/or RA treatment. We also determined that BMP pathway genes are downstream of MORN5 following siRNA knockdown. Based on these results, we conclude that MORN5 is both regulated by and required for BMP signaling. The restricted expression of MORN5 in the lip fusion zone shown here supports the human genetic data in which MORN5 variants were associated with increased risk of non-syndromic cleft lip with or without cleft palate.

Keywords: cleft lip, maxillary prominence, mandibular prominence, frontonasal mass, BMP

# INTRODUCTION

The vertebrate face is formed very early in development from the paired maxillary and mandibular prominences and the single frontonasal mass surrounding the oral cavity. These facial prominences arise during early embryogenesis from interactions between neural crest derived mesenchyme and head ectoderm. The frontonasal mass grows out, contacts and fuses together with the maxillary prominences to form the upper jaw. The midline facial skeleton consisting of the nasal septum, prenasal cartilage and premaxilla are all derived from the frontonasal mass (Richman and Lee, 2003). Craniofacial development is complex process coordinated by a network of transcription factors and signaling molecules (Murray and Schutte, 2004; Chai and Maxson, 2006; Jiang et al., 2006; Brunskill et al., 2014; Kurosaka, 2015; Marcucio et al., 2015; Nimmagadda et al., 2015). Disruption of this tightly controlled cascade can result in clefts where the facial prominences fail to meet and fuse (Leslie and Marazita, 2013).

Cleft lip and/or cleft palate are the most common craniofacial birth defects in humans (Setó-Salvia and Stanier, 2014; Watkins et al., 2014). The majority of clefts appear as isolated or non-syndromic clefts, because they occur in isolation from other developmental abnormalities. The causes of clefting are thought to be multifactorial, including an interaction between genes and the environmental factors (Schutte and Murray, 1999; Dixon et al., 2011; Leslie and Marazita, 2013; Setó-Salvia and Stanier, 2014; Watkins et al., 2014). Identification of genes contributing to clefts formation is important not only for our understanding of facial development, but also for improved prevention and treatment of affected individuals. The chicken embryo is a valuable experimental model to study the signals that control lip fusion. The avian primary palate closely resembles the primary palate in mammals (Abramyan et al., 2015). Moreover, the face can be accessed directly in the living embryo through a window in the shell. The disruption of FGF (Szabo-Rogers et al., 2008), BMP (Ashique et al., 2002), SHH (Hu et al., 2015), and WNT signaling (Geetha-Loganathan et al., 2014) causes a cleft lip in chickens that resembles that of humans.

Previously, a microarray study was performed to profile gene expression in individual chicken facial prominences in stage 18 embryos (Buchtová et al., 2010). From the list of genes that were significantly more highly expressed in the maxillary prominence, we selected MORN5 (also known as C9orf113, C9orf18 or FLJ46909) for further studies because it was described as a cleft susceptibility gene (Letra et al., 2010). Microarray analysis revealed 24 times higher expression of MORN5 in the maxillary prominence compared to expression in the frontonasal mass at stage 18, while mandibular prominence showed 10 times higher expression than the frontonasal mass (Buchtová et al., 2010).

Members of the MORN family were named for the presence of multiple MORN motifs (Membrane Occupation and Recognition Nexus). There are five paralogous genes in the MORN family (MORN1-5). Limited functional information is available for a subset of MORN genes. MORN1 has been identified in the parasite Toxoplasma gondii and other Apicomplexan protists where it plays role during cell division (Ferguson et al., 2008; Lorestani et al., 2010). Human MORN2 was found to facilitate phagocytosis-mediated restriction of some bacteria in macrophages (Abnave et al., 2014). Expression of MORN3 was detected in mouse testis, where it regulates spermatogenesis (Zhang et al., 2015). Finally, MORN4 promotes axonal degeneration in mouse sensory axons (Bhattacharya et al., 2012).

In chicken, the MORN5 gene is located on the forward strand of chromosome 17. On the reverse strand, NDUFA8 and LHX6 genes are nearby to the MORN5 gene. The size of the MORN5 gene is 13.5 kb and there are 6 exons (only 5 exons are coding) with four splice variants. The MORN5 gene encodes a protein of 172 amino acids, which contains a histone H3 K4-specific methyltransferase SET7/9 N-terminal domain (SSF82185) and three MORN motifs (**Figure 1**).

As the gene expression pattern or possible function of MORN5 during development had not been investigated in any animal model, we aim to analyzed chicken MORN5 expression in embryos and its integration into signaling pathways.

# MATERIALS AND METHODS

# Embryonic Material

Fertilized chicken eggs (ISA brown) were obtained from the farm Integra (Žabcice, Czech Republic). Eggs were incubated in a ˇ humidified forced air incubator at 37.8◦C. Embryos were staged and morphological characteristic were described according to Hamburger and Hamilton (1951). All procedures were conducted following a protocol approved by the Laboratory Animal Science Committee of the Institute of Animal Physiology and Genetics (Libechov, Czech Republic). ˇ

# Section In situ Hybridization (ISH)

Chicken MORN5 was obtained as chicken EST clone CHEST ID 543 F09 (Biovalley, France), where the probe sequence was cloned into pBluescript II KS+ vector. The entire region containing the probe sequence flanked by T3 and T7 RNA polymerase sites was amplified using M13 primers (forward primer: 5′ - GTA AAA CGA CGG CCA G-3′ , reverse primer: 5′ -CAG GAA ACA GCT ATG AC-3′ ). Then, the amplicon was isolated via gel purification (QIAquick Gell Extraction Kit, Qiagen, Germany) and this linearized DNA fragment was used in RNA polymerase reactions. DIG labeled antisense riboprobe was synthesized with T3 RNA polymerase (antisense) or with T7 polymerase (sense controls).

Embryos were fixed in 4% paraformaldehyde (PFA), processed through ethanol and xylene into paraffin, and sectioned for ISH. Hybridization was performed with RNA probe at 60◦C overnight as described previously (Holland et al., 1996). Anti-digoxigenin sheep antibody conjugated with alkaline phosphatase (1:2000, Roche, USA) was applied overnight at 4◦C. Visualization was achieved by incubation with substrates for alkaline phosphatase (BM Purple AP, Roche, Germany) for several days. Slides were then counterstained with eosin. ISH was carried out on at least three embryos for each stage.

# Embryo Manipulations

Embryos were treated with beads soaked in All-trans retinoic acid (RA), Noggin protein, Tris or Dimethyl Sulfoxide (DMSO) as described (Lee et al., 2001). Since DMSO was the solvent for RA, we used DMSO bead as a control for RA treatment and Tris as a control for Noggin treatment. AG1-X2 beads (Bio-Rad Laboratories, Hercules, Canada) of 100µm in diameter were soaked in RA (cat. No. R2625 Sigma) at a concentration of 1 mg/ml for 1 h as previously described (Lee et al., 2001; Nimmagadda et al., 2015). Noggin proteins were soaked into Affigel blue beads (Bio-Rad Laboratories, Hercules, Canada) of 200µm diameter for a minimum of 1 h at a concentration of 1 mg/ml (cat. No. 1967-NG, R&D Systems, Minneapolis, USA). Control beads were soaked with DMSO or Tris. Two beads were implanted into the maxillary region on the right side of chicken embryo at stage HH15. For ISH and QPCR, samples were collected 24 h post-bead implantation, embedded into paraffin and processed for ISH.

# Immunofluorescence on Slides

Embryos were collected at stage HH24 for MORN5 protein detection. Chicken duodenum was used as a control according to the manufacturer's instruction. Samples were fixed in 4% PFA and processed into paraffin. Following deparaffinization and rehydration, antigen retrieval was carried out using citrate buffer for 1 min at 97◦C. Polyclonal antibody to MORN5 (1:50, cat. No. NBP1-91230, Novus Biologicals, USA) was applied overnight at 4 ◦C. The secondary anti-rabbit antibody (1:200, Alexa Fluor 594, cat. No. A-21207) was applied for 30 min at RT. Sections were washed in PBS and coverslipped with Prolong Gold anti-fade reagent containing DAPI (cat. No. P36935, Invitrogen, USA).

# Quantitative RT-PCR

Gene expression of MORN5 was analyzed on tissues isolated from normal chicken prominences at stage 15, 18, 20, and 26. Moreover, Noggin or RA treated maxillary prominences were dissected 24 h following bead implantation at stage 15. Prominences were pooled from at least 15 embryos to produce one sample and 4 biological replicates were analyzed. Total RNA was extracted using the Mini RNeasy Kit (Qiagen, Germany) according to the manufacturer's instructions. The total RNA concentration and purity of each sample were assessed by spectrophotometry using a NanoDrop1000 (Thermo Scientific, Waltham, USA). First-strand cDNA was synthesized using the SuperScript Vilo cDNA synthesis Kit (cat. No. 11754050, Thermo Fisher, USA). The qPCR was performed in 10µl final reaction volumes containing the one-step master mix (no AmpErase UNG, cat. No. 4324018, Applied Biosystems, Carlsbad, USA) mixed with MORN5 (TaqMan Assays, Assay ID: AJKAKYV, context sequence: TTCCTGAGAAATGCAGAC GATGAGG, FAM-MGB, Applied Biosystems, Austin, USA) on LightCycler <sup>R</sup> 96 (Roche, Manheim, Germany) with preheating at 95◦C/10 min, followed by 40 cycles of 95◦C/15 s and 60◦C/1 min. Gene expression levels were calculated using 11CT method with normalization against the HPRT1 level (TaqMan Assays, Assay ID: Gg033338900\_m1, context sequence: TTGAATCATATC TGTGTGATCAGTG, FAM-MGB, Applied Biosystems, Austin, USA), which was used as the housekeeping gene. Means of 3 technical replicates were generated for each of 3 biological replicates and these values were used for statistical analysis. All procedures were repeated in at least three independent experiments.

# Transfection with MORN5 Plasmids in Cell Cultures

The expression vector containing C-terminally FLAG-tagged human MORN5 was obtained from OriGene (Rockville, MD). HEK293T cells were obtained from ATCC (Manassas, VA) and propagated in DMEM media (Sigma-Aldrich, St. Louis, MO) with 10% fetal bovine serum, 1% Pen/Strep and 1% lglutamin (Invitrogen, Carlsbad, CA). Cells were transfected using FUGENE6 reagent according to manufacturer's protocol (Promega).

HEK293T cells grown on glass coverslips were fixed with 4% PFA (RT/15 min), permeabilized with 0.1%Triton-X100 in PBS (RT/5 min), and incubated with the following antibodies at 4◦C overnight: MORN5 (1:100, cat. No. NBP1-91230, Novus Biologicals), FLAG (1:200, cat. No. F1804, Sigma-Aldrich). The secondary antibody AlexaFluor 488 (1:500; cat. No. A21206, Life Technologies) or AlexaFluor 594 (1:500; cat. No. A21203, Life Technologies) were used. Coverslips were mounted into DAPIcontaining Mowiol. Images were taken on an LSM700 laser scanning microscope with acquisition done using ZEN Black 2012 software (Zeiss, Jenna, Germany).

# siRNA Targeting gMORN5 in Chicken Embryos

Silencer Select custom designed siRNA (gMORN5, cat. No. 4399666, Ambion, Austin, USA) was mixed with FUGENE 6 (Roche, Mannheim, Germany), and then was injected into the maxillary prominence of chicken embryos. Negative siRNA (Silencer select negative control No.1 siRNA, cat. No. 4390843, Ambion, Austin, USA) was used as a control. The first injection of siRNA was performed at stage HH20 and the second one after 24 h about stage HH24. One day later after the second injection, embryos had reached stage HH28 and maxillary prominences were dissected for RNA isolation. Tissues were dissected from 5 embryos to form one sample and three biological samples were used for treated embryos (MORN5 siRNA) as well as for control (Silencer select negative control No.1 siRNA) embryos.

# PCR Arrays

Total RNA was extracted from siRNA treated maxillary prominences using the Mini RNeasy Kit (Qiagen, Germany) according to the manufacturer's instructions. First-strand cDNA was synthesized using the SuperScript Vilo cDNA synthesis Kit (cat. No. 11754050, Thermo Fisher, USA). Downregulation of MORN5 expression after injection was first confirmed using qPCR before further processing for PCR Array analysis.

Custom made Chick-bone plates (KRD, Czech Republic) were used for analysis of BMP pathway genes. The PCR arrays were performed in 12µl final reaction volumes containing SYBR Premix Ex Taq II (cat. No RR0821A, Takara, Japan) on LightCycler <sup>R</sup> 96 (Roche, Manheim, Germany) with preheating at 95◦C/30 min, followed by 45 cycles of 95◦C/5 s, 60◦C/20 s and 72◦C/15 s. Data were statistically evaluated by 11CT method with normalization against HPRT1 levels. In each PCR array plate, there were three technical replicates for 24 genes, and 2 technical replicates for an additional 13 genes.

# Statistical Analysis

All results were expressed as means ± standard deviations (SD) of three samples for each treatment and were compared by unpaired two-tailed Student's t-test for qPCR and PCR Array. Differences were considered to be significant at p < 0.05.

# RESULTS

# Spatiotemporal Gene Expression Pattern of MORN5 in Facial Prominences

First, we analyzed spatiotemporal expression pattern of MORN5 in individual prominences of chicken face. Facial prominences begin to form during early embryonic development. In situ hybridization showed no expression in chicken face at Hamburger-Hamilton (HH) stage 15 (50–55 h of incubation, **Figures 2A–C**) which is shortly after neural crest cells have entered the face. Later at stage HH17 (52-64 h of incubation, **Figures 2D–F**), MORN5 expression appeared in the caudal part

FIGURE 2 | Gene expression of MORN5 in early chicken face. (A–C) Frontal sections of chicken face at stage HH15. (D–F) Frontal sections of chicken face at stage 17. (D,E) In the ventral part of maxillary and mandibular prominences, there was no expression. (F) MORN5 expression gradually appeared dorsally in caudal part of maxillary region. (G–I) Frontal sections of chicken face at stage HH18. MORN5 expression was strong in maxilla and also in central part of each mandibular prominence (G). MdP, mandibular prominence; MxP, maxillary prominence. Scale bars = 100µm.

of the presumptive maxillary mesenchyme close to the maxillomandibular cleft. At stage HH18, the bulge of the maxillary prominence contained high levels of MORN5 transcripts (65– 69 h of incubation, **Figures 2G–I**). Expression was also detected in the dorsal (oral side) part of the mandibular prominences close to the maxillo-mandibular cleft (**Figures 2G,H**). At stage HH20 (70–72 h of incubation), there continued to be restricted expression in caudal and medial domains within the maxillary prominences (**Figures 3A–C**). In the mandibular prominences, there was expression in the cranial mesenchyme on either side of the midline groove (**Figures 3A–O**) with the exceptions of mesenchymal condensations of Meckel's cartilage (**Figures 3B,C**). At stage HH24 (4 days of incubation), maxillary prominence enlarged and strong MORN5 expression was present throughout the mesenchyme (**Figures 3E–G**). Mandibular expression was similar to stage HH20 (**Figures 2E–G**). Thus,

FIGURE 3 | Gene expression of MORN5 in later stages of chicken embryo. (A–C) ISH analysis in frontal sections of chicken head at stage HH20. There was strong expression in the maxillary prominence. The expression appeared in the cranial part in each mandibular prominence and it continued in dorsal direction. (B) No expression was observed in mesenchymal condensations and close to the fusion region of mandibular prominences. (E–G) MORN5 expression in frontal section of chicken head at stage HH24. MORN5 expression was strong in the mesenchyme of maxillary prominences. MORN5 expression was localized in the dorsal part of mandibular prominence, but not in mesenchymal condensation and close to the midline. (I–K) Frontal sections of chicken head at stage HH27 showed prominent expression in the maxillary prominence. There was weak expression in the globular process. Prominent expression was observed in rostral part of maxillary prominence and also in the mandibular prominence with the exception of midline. (M–O) Frontal sections of chicken head at stage 29 where beak is evident. MORN5 expression was localized in rostral part of maxilla and in fusion region. In the mandibular prominence, there was strong expression but not in the midline. (D,H,L,P) ISH analysis using sense probe. FNM, frontonasal mass; GP, globular process; mc, mesenchymal condensation; MdP, mandibular prominence; MxP, maxillary prominence. Scale bars = 100µm.

MORN5 is expressed in a restricted pattern in neural crestderived mesenchyme but not in epithelium. Sense probe did not show signal in the maxillary prominence (**Figures 3D,H,L,P**).

# MORN5 Expression in the Lip Fusion Zone at Later Stages

The next critical phase of facial morphogenesis is the fusion of the lip. Between stage HH27–29, the cranial-medial edges of the maxillary prominences meet the lateral corners of medial nasal prominences (globular processes) and fuse (Abramyan et al., 2015). At stage HH27 (5 days of incubation), MORN5 expression was observed for the first time in the corners of the frontonasal mass (globular processes, **Figures 3I,J**). Expression in the maxillary prominences was high in the rostral-medial corner just where fusion with the globular processes will take place. There continued to be expression in the mandibular prominences similar to stage 24 (**Figures 3I–K**). At stage HH29 (6 days of incubation), MORN5 expression was located in the region of lip fusion (**Figures 3M,N**) as well as in the mandible. This is the first stage where expression of MORN5 in Meckel's cartilage was detected (**Figure 3M**). Further confirmation of the restricted expression in the lip fusion zone is shown in other embryos cut in the frontal (**Figures 4A–C**) or transverse plane at stage HH29 (**Figures 4A–F**). Note that mesenchymal bridging has taken place by stage HH29, unifying the domains of expression of MORN5 in the globular processes and maxillary prominences (**Figures 4B,C**).

To quantify the relative levels of expression between the stages of development, we performed QPCR for evaluation of MORN5 expression level in each prominence at four different stages (HH15, 18, 20, 26). Since stage HH15, we did not observe any expression of MORN5 by ISH, this level of expression was chosen as the reference value for 11Ct analysis for individual prominences. In the maxillary prominence, MORN5 expression gradually increased during development with the peak level seen at stage HH20 (**Figure 5A**). In the mandibular prominence, we observed significantly increased expression at stage HH20 and 26 compared to stage HH15 embryos (**Figure 5B**). In the frontonasal mass, MORN5 expression is very low except of the globular processes we were surprised to see a statistically significant increase in expression of stage HH20 embryos (**Figure 5C**). In the section of in situ experiments, we could not detect MORN5 at stage HH20 (data not shown) therefore sensitivity of QPCR is greater than in situ hybridization. By stage HH27, there is expression of MORN5 in the in situ experiments; however, QPCR data did not pick up a significant expression level in stage HH26 embryos (**Figure 5C**). Some of the variability may be due to the dissection process and whether the globular process was included in all the samples. We did not compare expression levels between the facial prominences due to the experimental design.

# MORN5 Protein Expression in the Face

To correlate MORN5 protein distribution with MORN5 gene expression, we performed immunofluorescence staining. MORN5 protein was localized in developing chicken face at stage

in the area where edges of the maxillary prominences grow together with medial nasal prominence. (D–F) Horizontal sections of chicken head. (E) Region of fusion had prominent MORN5 expression. (F) More caudal section (other sample). Scale bars = 100µm.

HH24, with the most prominent expression in individual cells in the maxillary and mandibular prominences (**Figures 6A–H**). Thus, only a subset of cells expressing MORN5 RNA expresses the protein. In positive control (adult chicken intestine), there was expected signal in Goblet cells, in the apical parts of enterocytes and in fibroblasts of the lamina propria (**Figures 6I–L**).

The specificity of the MORN5 antibody was also confirmed in HEK293T cells transfected with a MORN5-FLAG plasmid (**Figures 7A–C**). The staining of MORN5 and FLAG antibodies overlapped (**Figures 7A–C**). Similar to tissue section data, exogenous MORN5 protein was found in the cytoplasm in a punctate pattern (**Figures 7D,E**).

# Downregulation of MORN5 after Noggin a Retinoic Acid Treatment

Our study uncovered high levels of MORN5 expression in normal chicken embryos, however a previous study from

overexpression in comparison to stage HH15. (B) In the mandibular prominence, we observed significant expression at stage HH20 and 26. (C) Very low MORN5 expression was detected in the frontonasal mass at stage HH15 and 18, but at stage HH20 was MORN5 expression significantly increased. FNM, frontonasal mass; MdP, mandibular prominence; MxP, maxillary prominence. t-test; \*\*\*p < 0.001, \*\*0.001 < p < 0.01, \*p < 0.05.

our group discovered that MORN5 was downregulated in an experimental paradigm involving beads implanted into the chicken face (Nimmagadda et al., 2015). Beads soaked in the bone morphogenetic protein antagonist, Noggin and retinoic acid (RA) synergistically induced transformation of the maxillary prominence into the frontonasal mass (Lee et al., 2001). The tissues from embryos induced to form this duplicated beak were profiled using microarrays. A significant downregulation of MORN5 expression was observed in all the treatment groups compared to controls treated with DMSO-Tris beads (-3.77 fold Noggin-DMSO treatment, -3.68-fold Noggin-RA, -2-fold after RA-Tris treatment) (Nimmagadda et al., 2015). We wanted to follow up this findings since it appeared that the RA and BMP pathways were upstream regulators of MORN5 and that possibly MORN5 was one of a set of genes mediating the beak duplication phenotype. First, we validated the array results using QPCR on maxillary tissues collected from treated and control embryos. We found a significant downregulation of MORN5 after Nogin-RA and Noggin-DMSO treatment compared to Tris-DMSO controls (**Figure 8A**). Next, we asked whether there were any spatial differences in MORN5 expression induced by the bead implants using in situ hybridization. Control embryos implanted with beads soaked in DMSO-Tris showed strong expression in the maxillary region and maxillo-mandibular cleft (**Figure 8B**). In contrast, no expression was observed in the maxillary prominence of Noggin-RA or Noggin-DMSO treated embryos. Interestingly, there was residual expression of MORN5 observed in embryos treated with RA-Tris located just under the epithelium of maxilla-mandibular cleft (**Figure 8B**).

# Downregulation of MORN5 by siRNA Altered Gene Expression of BMP and TGFβ Pathways Members

We had discovered that BMP activity was required for MORN5 expression but next wanted to investigate the genes that might be downstream of MORN5. As the first group of potential targets, we studied genes that are known to be in the BMP pathway. MORN5 expression in the maxillary prominence was downregulated to 75% of control levels following transfection with siRNA (2 rounds of transfection: at stage 20 and 24; **Figure 9A**). We used a PCR array that included 34 genes specific for the BMP pathway with HPRT1 acting as the reference control gene (Table S1).

Eight genes showed a statistically significant increase in their expression caused by partial MORN5 silencing (**Figure 9B**). These included ENG (Endoglin), Gdf2 (Growth differentiation factor 2, also BMP9), PLAU (plasminogen activator, urokinase), FST (Follistatin), Runx1 (Runt-related transcription factor 1), ID1 (Inhibitor of DNA binding 1), TGFβR2 (Transforming growth factor beta receptor 2) and TGFβ3 (**Figure 9B**). The most striking increase was seen with GDF2 (increased 3.5-fold). Statistically significant downregulation was observed only in the case of BMP5 (**Figure 9B**). It is interesting that MORN5 normally repressesID1, a transcription factor that positively regulates BMP signaling. Although levels of ID1 were increased, which should imply higher BMP signaling, there is also decreased expression of the BMP5 ligand. It is likely that cytoplasmic MORN5 indirectly regulates the expression of these genes and that further work is needed to determine the intermediate mediators of BMP and TGFβ signaling affected by MORN5.

# DISCUSSION

Here, we found spatially and temporally restricted expression of MORN5 in the face area during embryonic development suggesting its role in patterning of the maxillary prominences. Moreover, there was expression in the globular processes of frontonasal mass just before their fusion with the maxillary

the mandibular prominence, there is very low signal in the mesenchyme. MORN5 expression showed dispersed pattern of distribution. (A,C,E,G) DAPI staining. (I–L) Chicken intestine was used as a positive control with strong positivity in Goblet cells (gc), in the apical parts of enterocytes and also spotted expression in fibroblasts (fb). MdP, mandibular prominence; MxP, maxillary prominence.

prominences. Previously, a human genetics study found that MORN5 was associated with non-syndromic cleft lip with or without cleft palate (NSCLP) (Letra et al., 2010). We are the first to document expression of MORN5 in the relevant parts of the face undergoing lip fusion. In addition, there is strong expression on the medial sides of the maxillary prominences, the sites where palatal shelves will arise. While the chicken has a naturally cleft palate, it is interesting that MORN5 is expressed in the intermediate stages of palatal shelf formation. Based on previous microarray studies carried out on the chicken face, MORN5 came up twice, once as a maxillary enriched gene (Buchtová et al., 2010) and second as a differentially expressed gene following Noggin and RA bead implants (Nimmagadda et al., 2015). Taken together, the human genetic and chicken data suggest that MORN5 is an important maxillary patterning and possibly lip fusion gene. Thus, it would be worthwhile targeting MORN5 using mouse models and to include this gene in human NSLCP studies.

Complex signaling interactions coordinate the outgrowth of facial prominences to form the adult face. Some of factors have been previously identified by whole genome expression screens or by candidate gene mapping. The BMP signaling pathway regulates many cellular processes of craniofacial development and it is necessary for mesenchymal outgrowth of facial prominences. The expression of BMPs in chicken face was found at the time prior and during lip fusion (Ashique et al., 2002). BMP4 transcripts were previously detected in the epithelium of the globular processes of frontonasal mass and MORN5 expression underlays the same area however entirely in the mesenchyme. Also in the maxillary and mandibular prominences, epithelial BMP4 expression was described in parallel areas to mesenchymal MORN5. Furthermore, BMP2 and BMP7 were previously detected in the mesenchyme of both facial prominences. Maxillary and mandibular prominences express also several downstream target of BMP signaling. MSX1 is strongly expressed in the maxillary prominence but in slightly different pattern than MORN5 in the mandible (Shigetani et al., 2000; Fuchs et al., 2010). There are additional transcription factors that appear to overlap with MORN5 specifically in the frontonasal mass globular processes and maxillary prominences, TBX22, DLX5, and MSX2 (Higashihori et al., 2010). These transcription factors may regulate expression

**162**

antibodies. Cells expressing transgenic MORN5 were also MORN5-positive pointing to the specificity of the antibody. (D,E) In addition, a cytosolic spotted pattern of MORN5 expression was present in non-transfected cells. Scale bars = 100µm.

of MORN5. Interestingly, in the microarray study on beak duplicated embryos, TBX22 was upregulated following Noggin-RA treatment and TBX22 acts as a transcriptional repressor. It will be necessary to analyze whether MORN5 is a target of TBX22 which is a known clefting gene in humans (Kantaputra et al., 2011). It is interesting to note that the gene LHX6 which is located 3′ to MORN5 on the opposite strand was reported to be highly expressed in the chicken face (Washbourne and Cox, 2006). Expression of LHX6 begins in the maxillo-mandibular cleft at stage 18 similar to MORN5 (Washbourne and Cox, 2006). There is also striking similarity of expression of LHX6 in the globular processes and medial maxillary prominences at stage 27. This suggests that the two genes may share some common enhancers that drive expression in particular regions of the face.

MORN5 knockdown revealed indirect roles for this gene in controlling the expression of BMP and TGFB signaling pathways. Since we have shown MORN5 is a cytoplasmic protein it is unlikely that it is directly involved in regulating gene transcription. Further biochemical studies are needed to determine the exact function of MORN5 in the cell. Nevertheless, the RNA changes we observed suggest a subset of genes are dependent on MORN5 for their expression. We did not study the TGFB pathway in our bead implantation studies; therefore, the PCR array data extended our original findings on MORN5

FIGURE 8 | QPCR analysis and ISH of MORN5 expression after bead implantation. (A) QPCR analysis showed 1.68 times downregulation after RA-Tris, 3.03 after Noggin-DMSO and 2.59 after Noggin-RA treatment in comparison to control DMSO-Tris. (B) Control embryos implanted with beads soaked in DMSO-Tris had strong expression in maxillary region and maxilla-mandibular cleft. In RA-Tris treated embryos was very weak MORN5 expression. After Noggin-DMSO treatment, expression was rapidly decreased. No expression of MORN5 was observed after Noggin-RA treatment. Nog, Noggin; RA, Retinoic acid; TD, Tris-DMSO. Scale bars = 100µm. t-test; \*p < 0.05.

function. The most highly upregulated gene, GDF2, is known to associate with Endoglin (Castonguay et al., 2011) a glycoprotein located on cell surfaces that serves as a co-receptor for members of the Transforming growth factor-β superfamily (Cheifetz et al., 1992). This suggests that activity of TGFβ family is normally repressed by MORN5. Other changes such as the increase in the antagonist FST (Follistatin) may indicate that MORN5 operates in another way to regulate TGF signaling. MORN5 may normally repress this antagonist, which binds members of the TGFβ superfamily with a particular focus on activin (Lambert-Messerlian et al., 2007). We have shown that FST does not induce skeletal changes in the palate as compared to Noggin (Celá et al., 2016) therefore FST regulation by MORN5 may serve different functions outside of facial morphogenesis. Several other genes known to be in the TGFB pathway and essential for mouse palate development were upregulated including RUNX1 (Yamashiro et al., 2002), TGFβ3 (Cui et al., 1998) and ID1 (Rice et al., 2005). In summary, we discovered that MORN5 is involved in TGFB signaling at all levels. In conclusion, BMP signaling is required for MORN5 expression and reduction of MORN5 derepresses several genes in the BMP and TGFβ signaling pathways. Furthermore, MORN5 has two potential roles in facial patterning, to specify

# REFERENCES


maxillary identity and to regulate lip fusion that warrant further study in animal models.

# AUTHOR CONTRIBUTIONS

MB, JR, and PC conceived the study. PC, MH, KF, MK conducted the experiments. PK provided intellectual contribution. PC, JR, and MB wrote the manuscript. All authors reviewed and approved the final manuscript.

# ACKNOWLEDGMENTS

This study was supported by the Grant Agency of the Czech Republic (14-37368G to MB lab), Agency for Healthcare Research of the Czech Republic (15-33232A to PK lab) and institutional support (RVO:67985904).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00378


fetal palate tissues. Pediatr. Dev. Pathol. 10, 436–445. doi: 10.2350/06-05- 0087.1


**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 © 2016 Cela, Hampl, Fu, Kunova Bosakova, Krejci, Richman and Buchtova. 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) or licensor 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.

# MicroRNA Profiling during Craniofacial Development: Potential Roles for Mir23b and Mir133b

Hai-Lei Ding1 † , Joan E. Hooper <sup>2</sup> , Peter Batzel <sup>3</sup> , B. Frank Eames 3, 4, John H. Postlethwait <sup>3</sup> , Kristin B. Artinger <sup>1</sup> \* and David E. Clouthier <sup>1</sup> \*

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Ralf Kist, Newcastle University, UK Brad A. Amendt, The University of Iowa, USA Claudio Cantù, University of Zurich, Switzerland

### \*Correspondence:

Kristin B. Artinger kristin.artinger@ucdenver.edu; David E. Clouthier david.clouthier@ucdenver.edu

### †Present Address:

Hai-Lei Ding, Laboratory of Anesthesiology, Xuzhou Medical College, Xuzhou, Jiangsu, China

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 01 December 2015 Accepted: 21 June 2016 Published: 14 July 2016

### Citation:

Ding H-L, Hooper JE, Batzel P, Eames BF, Postlethwait JH, Artinger KB and Clouthier DE (2016) MicroRNA Profiling during Craniofacial Development: Potential Roles for Mir23b and Mir133b. Front. Physiol. 7:281. doi: 10.3389/fphys.2016.00281 <sup>1</sup> Department of Craniofacial Biology, School of Dental Medicine, University of Colorado Anschutz Medical Campus, Aurora, CO, USA, <sup>2</sup> Department of Cell and Developmental Biology, School of Medicine, University of Colorado Anschutz Medical Campus, Aurora, CO, USA, <sup>3</sup> Department of Neuroscience, University of Oregon, Eugene, OR, USA, <sup>4</sup> Department of Anatomy and Cell Biology, University of Saskatchewan, Saskatoon, SK, Canada

Defects in mid-facial development, including cleft lip/palate, account for a large number of human birth defects annually. In many cases, aberrant gene expression results in either a reduction in the number of neural crest cells (NCCs) that reach the frontonasal region and form much of the facial skeleton or subsequent failure of NCC patterning and differentiation into bone and cartilage. While loss of gene expression is often associated with developmental defects, aberrant upregulation of expression can also be detrimental. microRNAs (miRNAs) are a class of non-coding RNAs that normally repress gene expression by binding to recognition sequences located in the 3′ UTR of target mRNAs. miRNAs play important roles in many developmental systems, including midfacial development. Here, we take advantage of high throughput RNA sequencing (RNA-seq) from different tissues of the developing mouse midface to interrogate the miRs that are expressed in the midface and select a subset for further expression analysis. Among those examined, we focused on four that showed the highest expression level in in situ hybridization analysis. Mir23b and Mir24.1 are specifically expressed in the developing mouse frontonasal region, in addition to areas in the perichondrium, tongue musculature and cranial ganglia. Mir23b is also expressed in the palatal shelves and in anterior epithelium of the palate. In contrast, Mir133b and Mir128.2 are mainly expressed in head and trunk musculature. Expression analysis of mir23b and mir133b in zebrafish suggests that mir23b is expressed in the pharyngeal arch, otic vesicle, and trunk muscle while mir133b is similarly expressed in head and trunk muscle. Functional analysis by overexpression of mir23b in zebrafish leads to broadening of the ethmoid plate and aberrant cartilage structures in the viscerocranium, while overexpression of mir133b causes a reduction in ethmoid plate size and a significant midfacial cleft. These data illustrate that miRs are expressed in the developing midface and that Mir23b and Mir133b may have roles in this developmental process.

Keywords: craniofacial development, neural crest cell, mouse, zebrafish, RNA duplex, facebase

# INTRODUCTION

Morphogenesis of the vertebrate face is a complex event requiring coordination among a variety of signaling cascades. Human craniofacial birth defects occur at a world-wide rate of 1:800 births (Schutte and Murray, 1999; Spritz, 2001), with structural defects often resulting from failure of spatio-temporal integration of these signaling cascades. While the vertebrate mid-face has a complex embryonic origin, most facial birth defects result from disruption of cranial neural crest cell (NCC) patterning and differentiation (Chai and Maxson, 2006; Knight and Schilling, 2006; Walker and Trainor, 2006; Dixon et al., 2011; Clouthier et al., 2013). Cranial NCCs arise at the border between the neural and non-neural ectoderm and subsequently migrate to the frontonasal region (Le Douarin, 1982; Bronner-Fraser, 1995). These cells eventually give rise to most of the bone, cartilage, and connective tissue of the mid-face and neck (Couly et al., 1996; Köntges and Lumsden, 1996). NCC development relies on intricate regulation of patterning cues within specific boundaries of the head. While boundaries are established through continuous refinement of gene expression, the mechanisms required for this refinement are less clear.

MicroRNAs (miRNAs) are a class of small, noncoding RNAs that inhibit translation of target mRNAs by binding to a recognition sequence almost always in the 3′ untranslated region (UTR) of the target mRNA (Lee et al., 1993; Wightman et al., 1993; Lai, 2002; Lewis et al., 2003). This binding results in decreased mRNA translation through a number of mechanisms that can include cleavage and degradation of the target mRNA, translational repression, deadenylation of the 3′ -poly(A) tail and thus mRNA decay and miRNA sponging (Hutvágner et al., 2001; Zheng and Cullen, 2003; Wu et al., 2006; Hausser and Zavolan, 2014; Jens and Rajewsky, 2015). miRNAs have been implicated in many clinically relevant processes, including development, cancer, and stem cell maintenance and differentiation (Chen et al., 2012; Fernández-Hernando et al., 2013; Kuppusamy et al., 2013; Morceau et al., 2013; Parpart and Wang, 2013). miRNAs are also involved in numerous aspects of craniofacial development, including palatogenesis, odontogenesis, and upper and lower jaw development (Tavares et al., 2015). One of the first examples of miRNA action in facial development is mir140, whose regulation of pdgfra is required for NCCs to migrate past the optic stalk on their way from the hindbrain to the future palate (Eberhart et al., 2008). A SNP in the human MIR140 gene leads to reduced Mir140 expression in murine palatal mesenchymal cells (Li et al., 2011). In addition, this SNP is associated with increased risk of nonsyndromic cleft palate in mothers who smoke during pregnancy (Li et al., 2010). These findings suggest that miRNA function is evolutionarily conserved and illustrates a role for miRNAs in human palate development. Another miRNA family involved in craniofacial development is the MIR17 and MIR92 family, which has been linked to Feingold syndrome in human patients (Kannu et al., 2013; Tassano et al., 2013). Targeted knockouts of Mir17 and Mir92 in mice results in hypoplasia of most skull bones, including reduced ossification and cleft palate, similar to human patients (Ventura et al., 2008; de Pontual et al., 2011; Li et al., 2012; Wang et al., 2013). In addition, others, such as Mir196, Mir199, and Mir200 have likely roles in determining craniofacial size, bone and cartilage formation, and cartilage size, respectively (Watanabe et al., 2008; Desvignes et al., 2014). However, targeted deletion of the miRNA processing enzyme DICER in NCCs results in hypoplasia of most craniofacial structures, suggesting that numerous other miRNAs are also likely involved in this process (Huang et al., 2010; Zehir et al., 2010; Nie et al., 2011; Oommen et al., 2012).

Here, we use data from a high-throughput miRNA sequencing project of developing mouse facial structures to identify many miRNAs that are potentially involved in craniofacial development. We have examined the expression of a number of these miRNAs in both mouse and zebrafish. Further, we have performed functional analysis of four of these miRNAs in zebrafish. Our in situ hybridization and overexpression analyses provide evidence that Mir23b and Mir133b are important regulators of craniofacial development.

# MATERIALS AND METHODS

# miRNA-Seq Differential Expression Analysis

RNA-Seq count data with gene assignments have been deposited in the FaceBase repository for craniofacial research (www.facebase.org). Specific FaceBase accession numbers are: E10.5 FB00000662.01, E11.5 FB00000663.01, E12.5 FB00000664.01, FB00000664.01, E13.5 FB00000665.01, E14.5 FB00000666.01. Counts of overlapping and nested sequences assigned to each mature (1035 sequences) or hairpin (306 sequences) mouse miRNA were summed to obtain consolidated counts per mature or hairpin miRNA (724 and 255, respectively). A threshold of 1 was used for the RNA-seq analysis. After low filtering (mean counts per sample >5; 629 mature miRs and 163 hairpin miRNAs), we used DESeq to identify miRNAs differentially expressed by anatomy or age. For principal components analysis and hierarchical clustering, count data was normalized using the regularized log transformation in the DESeq R package. A p < 0.01 after multiple testing correction (Benjamini and Hochberg method) was used, as well as no fold-change threshold (Benjamini and Hochberg, 1995).

# Mouse and Zebrafish Maintenance

All mouse embryo collection was performed using 129S6 mice (Taconic), with the day of the copulation plug denoted as 0.5 days. Embryo collection and fixation were performed as previously described (Clouthier et al., 1998). Zebrafish were maintained according to common lab practice (Westerfield, 2000), with embryo staging according to established methods (Kimmel et al., 1995). The wild type lines used are AB and an in-cross line maintained between the AB and TL lines (TAB). This study was carried out in accordance with the recommendations of the Institutional Animal Care and Use Committee (IACUC) of the University of Colorado Anschutz Medical Campus as laid out in protocols approved by IACUC. Nomenclature for miRNAs is consistent with that described in Desvignes et al. (2015). For simplicity, pri-MiR(mouse) and pri-miR (zebrafish) were annotated without the "pri" in the text below.

# Cloning and Probe Synthesis

To detect expression of mouse MiRNAs, we used both Locked Nucleic Acid (LNA; Obernosterer et al., 2007) and conventional riboprobes against the primary pri-MiR transcripts, both labeled with digoxigenin (He et al., 2011). For LNA-based experiments, Exiqon LNA Mir133b (#616614-360) and Mir23b (#615366-360) probes were used. To generate pri-MiR probes, PCR primers were designed to amplify about 200–600 nt of the primary MiR transcript centered on the mature miRNA sequence for mouse and zebrafish. All primer sequences are given in Supplemental Table 1. To generate mouse pri-MiRs, PCR primers were used with genomic tail DNA obtained from tails of 2 month old mice. To generate zebrafish pri-miRs, PCR primers were used with 5 day post fertilization (dpf) zebrafish embryos. PCR products were cloned into the pCR <sup>R</sup> II-TOPO <sup>R</sup> Vector (Invitrogen). Inserts were validated by sequencing. Plasmids were linearized and transcribed using the DIG Labeling Kit (Roche). Probes were purified as previously described (Clouthier et al., 1998). The opposite stand sense probes were used as controls for specificity for the antisense expression (data not shown).

# Mouse and Zebrafish Whole Mount In situ Hybridization (ISH)

Whole mount ISH in mouse was performed as previously described (Clouthier et al., 1998). For sectional miRNA ISH, serial 10µm frontal sections through the head of E12.5 and/or E13.5 mouse embryos were collected on Superfrost Plus slides (Fisher). Subsequent ISH was performed as previously described (Hendershot et al., 2007). Hybridizations were performed at 70◦C overnight. Colorimetric detection was performed with BM purple (Roche). Signal detection required an average of 2–3 days in staining solution. After developing, slides were rinsed in TBST, coverslipped and photographed using Nomarski optics on an Olympus BX51 compound microscope fitted with a DP71 digital camera. Whole-mount ISH in zebrafish was performed as previously described (Johnson et al., 2011). BM Purple (Roche) was used as a substrate for the alkaline phosphatase reaction. After staining, embryos were mounted in 80% glycerol or 3% methylcellulose and imaged under Nomarski optics on the Olympus BX51 compound microscope as described for mice.

# Overexpression Analysis and Bone and Cartilage Staining

MiRNA duplexes were created by annealing 5p and 3p MiRs together at 95◦C and slowly cooling to room temperature (MiR133b-3p: UUUGGUCCCCUUCAACCA; MiR133b-5p: GCUGGUCAAAUGGAACCA; MiR23b-3p: AUCACAUUG CCAGGGAUU; MiR23b-5p: GGUUCUUGGCAUGCUGA). Thirty-three micrometers of MiR23b duplex (5′—3′ ), 6.25µM of MiR133b (5′—3′ ), and 33µM of standard control miRNA (5′ - CTTACCTCAGTTACAATTTATA -3 duplexed with 5′ - TAAATTGTAACTGAGGTAAGAG-3′ ) were injected into single cell zebrafish embryos and allowed to grow for 6 dpf. Zebrafish embryos were fixed and stained with alizarin red and alcian blue (Birkholz et al., 2009) with minor modifications. Larvae were fixed in 2% PFA/PBS for 1 h at room temperature, washed in 100 mM Tris (pH 7.5)/10 mM MgCl2, and stained overnight in 0.04% alcian blue (Anatech Ltd) in 80% EtOH/100 mM Tris (pH 7.5)/10 mM MgCl2. Larvae were then rehydrated in 80% EtOH/100 mM Tris (pH 7.5)/10 mM MgCl2, followed by 50 and 25% EtOH/100 mM Tris (pH 7.5), cleared in 25% glycerol/0.1% KOH, then stained 30 min with 0.01% alizarin red (Sigma) in 25% glycerol/100 mM Tris (pH 7.5) and stored in 50% glycerol/0.1% KOH. Stained skeletons were dissected, flat-mounted and imaged as previously described (Johnson et al., 2011).

# RESULTS

# Identification of miRNAs Involved in Midfacial Development from miRNA-Seq Data

We have previously conducted massively parallel miRNA sequencing (miRNA-Seq) on miRNAs extracted from embryonic age (E) 10.5, E11.5, E12.5, E13.5, and E14.5 mouse maxillary prominences, frontonasal prominences, and palatal shelves, with the data deposited in FaceBase (www.facebase.org). Between E10.5 and E11.5, NCC-derived mesenchyme that will give rise to craniofacial structures undergoes significant patterning events that establishes the positional and functional identity of the mesenchyme. Between E12.5 and E14.5, gene expression in the facial region is driving differentiation into the bone and cartilage of the upper and lower face. Because we are interested in the mechanisms that govern this second event, we focused our analysis of miRNA expression between E12.5 and E14.5. Using a combination of biostatical approaches and packages (Section Materials and Methods), we identified 262 differentially expressed mature miRNAs and 71 hairpin miRNAs at a false discovery rate-adjusted p > 0.01 (data not shown). Of these, 49 mature miRNAs and 11 hairpin miRNAs were differentially expressed between maxillary prominences, frontonasal prominences, and palatal shelves (**Figure 1A**); some also differed by age (**Figure 2**). **Table 1** shows the top differentially expressed miRNAs as total counts, while **Figure 2** shows the log2fold change in expression of specific miRNAs between facial prominences at E13.5. Total counts of each miRNA are also shown in **Table 1** so that the relative expression of miRNAs can be compared.

# Principal Components Analysis of miRNAs Expressed in the Face

Principal Components Analysis **(**PCA) is a primary quality control assay, with the expectation that replicates will cluster and that principal components might reflect known qualities of the sample conditions (e.g., age, anatomy) or of the experiment (e.g., batch or read depth). After sequence consolidation and low-expression filtering, normalized counts for 979 miRNAs (724 mature miRNAs and 255 hairpin miRNAs) were used to analyze the principal components of the variability among the 24 samples (**Figure 1B**). The first principal component, which

ward. D2 method) shows groups of miRNAs enriched in the frontonasal prominence (FnP), palate (Pal), the maxillary prominence (Max), or in combinations thereof. The dendrogram at the top shows the clustering of the samples; the three major divisions correspond to FnP, Pal, and Max, and duplicate samples cluster together. (B) Principal Component Analysis (PCA) analysis of data. Data is color coded by age (inset), with tissue source denoted by symbol (inset). Principal Component 1 (PC1) distributes samples by age. Principal Component 2 (PC2) is orthogonal to PC1 but does not correspond to any obvious property of the samples.

accounts for 26% of the variability, distributed the samples by age (color-coded). Neither prominence (symbols) nor read depth (not shown) correlated with any of the top five principal components. Biological duplicates from E10.5-E12.5 tended to cluster, indicating biological reproducibility at those ages. Together these demonstrate the biological reproducibility for key miRNAs between replicates and within prominences.

# Expression Analysis of Selected miRNAs in the Mouse at E12.5

For our in situ hybridization (ISH) analysis of miRNA expression in the developing midface, we focused on E12.5 mouse embryos, as at this stage, the lateral and medial nasal prominences, palatal shelves, and maxilla are well-defined and are undergoing extensive growth. We initially examined miRNA expression in E12.5 mouse embryo using whole mount ISH and LNA probes against Mir23b, Mir24.1, and Mir666 (Supplemental Figure 1). LNA probes specifically detect the mature Mir sequence. However, the overall level of expression of these Mir's as detected in whole mount analysis was low (likely due in part to poor probe penetration; Supplemental Figure 1), leading us to examine Mir expression using frozen sectional ISH analysis. When using LNA probes against Mir23b and Mir133b, robust expression was present in a variety of facial structures, though overall background staining on the sections was high (Supplemental Figure 2). To next assess whether we could improve the signal to noise ratio of the staining, we PCR-generated probes against the pri-miRNA transcript, an approach we have previously used to examine Mir expression in embryos (He et al., 2011). Probes encompassed the pri-miRNA sequence (see (Section) Materials and Methods). These probes detect both pri-miRNAs and mature miRNAs, so expression is expected in both the nucleus and cytoplasm. However, the relative abundance of mature miRNAs for any single species

### TABLE 1 | microRNAs that are differentially expressed between prominences at E13.5.


(Continued)


### TABLE 1 | Continued

Colored and bolded areas indicate groups of miRs that are highly expressed within a specific prominence. Green, Frontal nasal prominence (Fn); Brown, Maxilla (Mx); Pink, Palate (Pal).

### TABLE 2 | Expression profiles of miRNAs in the developing mouse midface.


compared to that of pri-miRNA species makes it far more likely that these probes detect the mature miRNA. Further, we have shown that the use of pri-miRNA probes for miRNA ISH provides comparable results to those using LNA probes (He et al., 2011). Based on our bioinformatics analysis (**Figures 1**, **2**), we examined the expression of 13 pri-miRNAs in E12.5 mouse embryos and found that many were expressed in specific embryonic tissues, including facial structures (**Table 2** and Supplemental Figures 3–9). Based on this initial analysis of ISH data, we focused our subsequent analysis on MiR23b and MiR133b.

# Expression of MiR23b and MiR133b in Mouse Facial Structures at E12.5

In E12.5 wild type mouse embryos, MiR23b expression appeared throughout the facial region including the nasal epithelium, mystacial vibrissae, and vomeronasal nasal organs (**Figures 3A–C**), upper incisors (**Figure 3D**), tongue connective tissue and epithelium (**Figures 3E,F**), and maxillary epithelium (**Figure 3F**). Expression was also observed in the anterior palatal shelves, with expression more prominent on the nasal side of the shelves (**Figures 3E,F**) and in the trigeminal ganglion (**Figure 3G**). In addition, MiR23b was expressed along the

palatal shelf epithelium, again starting at the midline of the shelf and continuing on the oral side (**Figures 3H,I**). Overall, the expression pattern of MiR23b supported our analysis (**Figure 1**), though it was difficult to assess the qualitative differences in expression between prominences compared to the quantitative differences of miRNA-seq. In the genome, MiR23b is part of the 850 bp Mirc23 cluster, which contains Mir23b, Mir27b, and Mir3074.1. In addition, Mir24.1 is 5 kb away from Mir23b. As previously described for many clustered miRNAs (Lagos-Quintana et al., 2003; Lim et al., 2003), Mir24.1 had an expression pattern similar to that observed for Mir23b, including expression in the nasal epithelium (Supplemental Figures 3A–C), tongue (Supplemental Figures 3E,F,H,I) and maxillary process epithelium (Supplemental Figure 3D), though expression in the palatal shelf mesenchyme and overlying epithelium (Supplemental Figures 3D,F,H,I) and trigeminal ganglia (Supplemental Figure 3G) was weak.

Like MiR23b, MiR133b was also strongly expressed in the craniofacial region at E12.5. Weak expression was observed in the nasal epithelium, mystacial vibrissae and maxilla (**Figures 4A–C**), but strong expression was observed in facial musculature, including the intrinsic and extrinsic musculature of the tongue and eye (**Figures 4D–I**) and facial muscles (including the masseter muscle (arrow; **Figure 4F**). Expression of MiR133b was not observed in the maxilla or palatal shelves, suggesting that the expression observed by miRNA-seq might reflect presence of other tissue in dissected samples. Like Mir23b, Mir133b exists in a cluster with Mir206, which had a similar pattern of expression to that of Mir133b. Message was primarily detected in facial muscles (Supplemental Figures 4E,F), with message also detected in the cornea and lens of the eye (Supplemental Figure 4A). Further, like the comparison between MiR23b and MiR24.1, expression of MiR206 was much weaker than the expression observed for MiR133b.

FIGURE 4 | Expression of Mir133b in mouse facial muscles at E12.5. ISH analysis in frozen frontal sections through the head of E12.5 mouse embryos. (A–C) Mir133b is weakly expressed in the nasal epithelium (ne; A), mystacial vibrissae (mv; B), and maxilla (mx); (D–F). (mx); (D–F). Mir133b is strongly expressed in facial muscles, including the masseter (ma), and in the muscles of the eye (em; D) and tongue muscle (t; E–I) Expression is not observed in the palatal shelves (ps; E,F). (G–I) Posterior expression continues continues in the tongue muscles. e, eye; oc, oral cavity.

# Expression of mir23b and mir133b is Conserved in Zebrafish Facial Structures

Based on analysis of mir140 action, miRNA function during facial development is also present in zebrafish embryos. Similarly, we found that a selected subset of the miRNAs from **Table 1** were also expressed in the head region of 30-hours post fertilization (hpf) zebrafish embryos (**Table 3** and Supplemental Figure 10). To examine whether the pattern of expression of mir23b and mir133b was also conserved between mouse and zebrafish embryos, we examined expression of both miRNAs in 30–72 hpf embryos. Expression of mir23b was detected in the head and pharyngeal arch mesenchyme (**Figure 5A**) and in the somitic mesoderm (**Figure 5D**) at 30 hpf. Diffuse expression was also present in the eye (**Figure 5A**). At 48 hpf, mir23b expression was still present in the head and pharyngeal arch mesenchyme while also appearing in the otic vesicle (**Figure 5B**). Expression was also observed in the trunk muscle and notochord (**Figure 5E**). By 72 hpf, expression remained in cranial muscle (**Figures 5C,G**) while also being present in the somite-derived trunk muscle and notochord (**Figure 5F**) and in ceratobranchial structures (arrows, **Figure 5G**).

At 30 hpf, mir133b expression was also observed in the head region (around the eye and portions of the brain; **Figure 6A**), though expression was weaker than that of mir23b. Expression was quite strong in the developing somites (**Figure 6D**). By 48 hpf, expression around the otic vesicle, eye, and posterior brain was increased (**Figures 6B,G**). As in 30 hpf embryos, mir133b was strongly expressed in the somites (**Figure 6E**). By 72 hpf, mir133b expression was observed in trunk muscles (**Figure 6F**), otic vesicle and heart (**Figure 6C**). Expression was also present in the facial muscles, including the anterior mandibularis (arrow, **Figure 6H**) and muscles developing near the ceratobranchials, including the stemohyoideus (arrowheads, **Figure 6H**).

TABLE 3 | Expression profiles of miRNAs in the developing zebrafish.


# mir23b and mir133b Overexpression Results in Viscerocranial and Neurocranial Defects in Zebrafish

Two potential methods for assessing function of genes in zebrafish are over-expression and gene inactivation. While Crispr-Cas9-mediated gene inactivation is underway, we began our analysis of potential function by injecting 1–2 cell zebrafish embryos with duplex RNA for MiR23b and MiR133b examining cartilage development at 6 dpf. Overexpression of miRNA duplex is more likely to generate a phenotype, as increasing the repressor function will result in reduced gene expression of the miRNA gene targets. In the viscerocranium, MiR23b duplex injection resulted in aberrant development of Meckel's cartilage and the ceratohyal (**Figure 7B**). In addition, ectopic cartilage extended from the basihyal to the medial aspect of Meckel's cartilage (23/128; 18%; **Figure 7B**). Two bilateral ectopic cartilages also extended anteriorly from Meckel's cartilage. Morphological changes were also present in the neurocranium, including a slight increase in the width of the ethmoid plate likely resulting from a shortening of the trabeculae (74/128; 58%; **Figures 7F,I**) as compared to standard control MiR injected embryos (n = 10; **Figures 7D,H,I**) and control non-injected embryos (**Figures 7A,E**).

Injection of duplex RNA for MiR133b resulted in the formation of a cleft in the ethmoid plate (40/89; 45%; **Figure 7G**) and a general reduction in the overall size of the neurocranium, with a significant reduction in both the ethmoid plate width and length (46/89; 52%; **Figures 7G,I**). Injected animals also showed a mild change in the viscerocranium, including a reduction in the size of the ceratohyal (**Figure 7C**).

# DISCUSSION

miRNAs have well defined roles in numerous developmental process, including craniofacial development (Tavares et al., 2015). We have shown here that numerous miRNAs are expressed within the developing midface, with expression patterns for some of these suggesting distinct functions in a variety of tissues. In addition, we have shown that over-expression of mir23b and mir133b results in changes in craniofacial cartilage morphogenesis. Together, results from our study illustrate the utility of an approach to quickly assess potential miRNA function during vertebrate morphogenesis.

# Identification of miRNAs in Craniofacial Development

In addition to the identification of novel miRNAs expressed in the craniofacial complex, several miRNAs identified in this study as differentially expressed between facial prominences and over time have been identified to play a role in development in other systems. Mir199a, which we found to be enriched in the palate as compared to the FNP at E13.5, is involved in aspects of chondrogenesis and osteogenesis (Suomi et al., 2008; Lin et al., 2009). Similarly, Mir92a which is in the Mirc1 cluster on mouse chromosome 14 containing Mir17, Mir18, Mir19a, Mir20a, Mir19b-1, and Mir92a-1, is required to promote proliferation of orofacial development (Ning et al., 2013). Thus, these data confirm and validate the deep sequencing data. Other groups have performed miRNA expression profiling of the developing mouse orofacial region using microarrays (Mukhopadhyay et al., 2010) and have found similar differential expression across time for miRNAs that included Mir133a and Mir133b. Interestingly, Mir23b was not identified by microarray analysis. This may be due to the retrospective nature of microarray studies, as they are limited by the total number of miRNAs that are present on the microarray chip. This approach may exclude all of the transcribed miRNAs in the genome. By definition, the inclusion of miRNAs on microarray chips means that the miRNA has already been identified and annotated. Furthermore, because microarrays are generally based on genome sequence, they may not assay miRNAs that originate by post-transcriptional editing (de Hoon et al., 2010). In contrast, miRNA-seq is prospective: it identifies all miRNAs present in a sample, regardless of whether they have been annotated or not or whether they have experienced editing. Thus, miRNA-seq has the power to identify novel and edited miRNAs in addition to known miRNAs.

miRNA-seq was recently used on avian embryos, in which dissection of the chick, quail and duck FNPs followed by miRNAseq identified 170 miRNAs that were differentially expressed between the three species (Powder et al., 2012). Interestingly, several miRNAs were found to be avian-specific, suggesting that miRNAs may have promoted the evolutionary diversification of facial shape and beak formation.

# Potential Mir23b and Mir133b Functions and Targets

Here we have shown that Mir23b is expressed in the developing face of mouse embryos and in the head of zebrafish embryos and that its overexpression in zebrafish embryos results in ectopic cartilage structures in the viscerocranium. This may indicate roles for Mir23b in regulating either patterning of the NCC-derived mesenchyme or later chondrogenesis. One of the interesting aspects our data analysis (**Figure 1**) is that Mir27b expression is also present in the developing midface, with its expression mirroring that of Mir23b. In mouse, Mir23b is part

of a miRNA cluster that includes Mir23b, Mir27b, Mir3074.1, and Mir24.1. In zebrafish, this corresponds to mir23b, mir27d, and mir24.1. mir23b and mir27b are separated by less than 200 bp, though it is not clear that their expression is co-regulated. In the fetal mouse liver, the Mir23b cluster regulates cell fate switch between hepatocytes and bile duct cells by regulating expression of Smad3, 4, and 5 and thereby repressing TGFβ signaling (Rogler et al., 2009). This negative regulation of TGF-β signaling is interesting, as proper regulation of TGF-β signaling is crucial for proper craniofacial development (Behnan et al., 2005). Marfan syndrome results from aberrant TGFβ signaling due to the failure of fibrullin-1 to bind to and sequester the latent form of TGF-β (Neptune et al., 2003; Chaudhry et al., 2007). Conversely, loss of TGF-β signaling is also detrimental to midfacial development, with loss of numerous TGF family members leading to facial birth defects (Iwata et al., 2011), illustrating the importance of miRNA regulation. While our miRNA-seq data suggests that the two miRNAs are co-expressed, further analysis of embryonic expression and loss-of-function phenotypes for individual miRNAs and the clusters will be required to fully understand their function. This is especially true for Mir23b and Mir27b, as while both work concurrently to drive cardiomyocyte development from ES cells in vitro, Mir23b subsequently controls the later beating phenotype of differentiated cells while Mir27b functions to inhibit this event (Chinchilla et al., 2011; Wang et al., 2012). Our analysis of expression and function in zebrafish suggest that mir23b is expressed in the pharyngeal arch mesenchyme and potentially functions to promote proliferation of chondrocytes, such that when overexpressed, ectopic cartilage arises.

Mir133b resides in a cluster that includes Mir206 and Mir133b. This family is referred to as myomiRs because they regulate genes involved in adult muscle formation. In addition,

Mir133b is down-regulated in several cancers, including muscle rhabdomyosarcoma, osteosarcoma, and prostate, colorectal and gastric cancers (Namløs et al., 2012; Qin et al., 2012; Mo et al., 2013). In these abnormal contexts, MIR133b in human cervical carcinoma targets EGFR and FGFR1, similarly acting as a tumor suppressor (Namløs et al., 2012). Interestingly, both EGF and FGF signaling have been implicated in craniofacial development. Mutations in EGF signaling have been implicated in human clefting (Falagan-Lotsch et al., 2015) and cranial suture formation in mice (Rawlins et al., 2008). More roles have been ascribed for FGF signaling in craniofacial development, with FGFR1 specifically linked to skeletal dysplasias and craniosynostosis in both humans and mice, suggesting a mechanism by which Mir133b may regulate craniofacial development (Moosa and Wollnik, 2016). Changes in facial muscle can affect development of facial bone and cartilages (Reider et al., 2012). From our expression analysis, it is plausible that overexpression of Mir133b may lead to changes in facial muscle which affects subsequent viscerocranium development.

In addition, Mir133b is involved in the differentiation of dopaminergic neurons. Zebrafish mir133b is expressed in the midbrain at low levels and regulates pitx3 to control dopaminergic neuron differentiation (Sanchez-Simon et al., 2010). In contrast, mice in which Mir133b has been inactivated have normal dopaminergic neuron numbers and normal PITX3 protein levels (Heyer et al., 2012) even though Mir133b can target the Pitx3 message. This suggests the possibility that other Mir133 family members with the same seed sequence (GGACCAAA; i.e., Mir133a, Mir133c) can compensate for loss of Mir133b. Mir133b is clustered in the genome with Mir 206.1, which does not have the same seed sequence but is predicted to bind some of the same targets in mouse including

Histone Deacetylase 4, DNA Polymerase α, and Connexin43 (Anderson et al., 2006; Chen et al., 2006; Kim et al., 2006; Goljanek-Whysall et al., 2012). mir133b has the potential to target the same set of myogenic targets in zebrafish, but only Histone Deacetylase 4 contains a seed sequence for mir133b. More work is needed to distinguish between these possibilities. While we have not determined the numbers of dopaminergic neurons in zebrafish in which mir133b is over-expressed, we do see specific defects in cartilage differentiation, including hypoplasia of the ethmoid plate and specific gaps or missing cartilage in the viscerocranium, suggesting that mir133b may act non-cell autonomously to regulate cartilage differentiation. While there is little known about a direct role of mir133b on the development of the craniofacial cartilage, these results suggest that it is involved in both muscle and neuronal development, both of which can influence, either directly or indirectly, formation of the craniofacial complex. In summary, our results illustrate multiple MiRNAs are expressed in the developing face. This analysis will allow for more focused analysis of miR function in both zebrafish and mouse craniofacial development.

# AUTHOR CONTRIBUTIONS

Experiments were designed by DC and KA in consultation with JP. All mouse and zebrafish experiments were performed by HD with help from BE. The RNA-seq was performed by BE and JP with bioinformatics performed by PB and JH. HD, KA, and DC analyzed and interpreted the data, and wrote the manuscript. All authors commented on the manuscript.

# FUNDING

This work was supported by a grant from NIH/NIDCR (DE020076) to DC, KA, and JP.

# REFERENCES


# ACKNOWLEDGMENTS

We appreciate the excellent fish care provided by Morgan Singleton.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00281


encompassing miR-17-92 and GPC5. Eur. J. Med. Genet. 56, 452–457. doi: 10.1016/j.ejmg.2013.06.001


contributes to pathogenesis in Marfan syndrome. Nat. Genet. 33, 407–411. doi: 10.1038/ng1116


Zheng, Y., and Cullen, B. R. (2003). Sequence requirements for microRNA processing and function in human cells. RNA 9, 112–123. doi: 10.1261/rna.2780503

**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 © 2016 Ding, Hooper, Batzel, Eames, Postlethwait, Artinger and Clouthier. 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) or licensor 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.

# Ephrin Ligands and Eph Receptors Show Regionally Restricted Expression in the Developing Palate and Tongue

### Guilherme M. Xavier 1, <sup>2</sup> , Isabelle Miletich<sup>1</sup> and Martyn T. Cobourne1, 2 \*

<sup>1</sup> Department of Craniofacial Development and Stem Cell Biology, King's College London Dental Institute, Guy's Hospital, London, UK, <sup>2</sup> Department of Orthodontics, King's College London Dental Institute, Guy's Hospital, London, UK

The Eph family receptor-interacting (ephrin) ligands and erythropoietin-producing hepatocellular carcinoma (Eph) receptors constitute the largest known family of receptor tyrosine kinases. Ephrin ligands and their receptors form an important cell communication system with widespread roles in normal physiology and disease pathogenesis. In order to investigate potential roles of the ephrin-Eph system during palatogenesis and tongue development, we have characterized the cellular mRNA expression of family members EphrinA1-A3, EphA1–A8, and EphrinB2, EphB1, EphB4 during murine embryogenesis between embryonic day 13.5–16.5 using radioactive in situ hybridization. With the exception of EphA6 and ephrinA3, all genes were regionally expressed during the process of palatogenesis, with restricted and often overlapping domains. Transcripts were identified in the palate epithelium, localized at the tip of the palatal shelves, in the mesenchyme and also confined to the medial epithelium seam. Numerous Eph transcripts were also identified during tongue development. In particular, EphA1 and EphA2 demonstrated a highly restricted and specific expression in the tongue epithelium at all stages examined, whereas EphA3 was strongly expressed in the lateral tongue mesenchyme. These results suggest regulatory roles for ephrin-EphA signaling in development of the murine palate and tongue.

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Catherine Chaussain, Université Paris Descartes, France David Clouthier, University of Colorado Anschutz Medical Campus, USA Claudio Cantù, University of Zurich, Switzerland

> \*Correspondence: Martyn T. Cobourne martyn.cobourne@kcl.ac.uk

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 14 December 2015 Accepted: 08 February 2016 Published: 23 February 2016

### Citation:

Xavier GM, Miletich I and Cobourne MT (2016) Ephrin Ligands and Eph Receptors Show Regionally Restricted Expression in the Developing Palate and Tongue. Front. Physiol. 7:60. doi: 10.3389/fphys.2016.00060 Keywords: palatogenesis, tongue development, ephrin, Eph, gene expression, in situ hybridization

# INTRODUCTION

The formation of a palate separating the oral and nasal cavities is a developmental process characteristic of higher vertebrates and requires complex and highly coordinated molecular interactions (reviewed in Ferguson, 1988; Cobourne, 2004; Dudas et al., 2007; Gritli-Linde, 2007). In the embryo, the primary palate is a derivative of the frontonasal process, whilst the secondary palate forms from the paired palatal shelves of the maxillary process, themselves a derivative of the first pharyngeal arch. The palate is formed by elevation and fusion of the maxillary palatal shelves, with each other posteriorly, with the primary palate anteriorly and the nasal septum superiorly (reviewed in Dudas et al., 2007). The palatal structures are built from cranial neural crest (CNC)-derived ectomesenchyme, mesoderm and the oro-pharyngeal ectoderm (reviewed in Ferguson, 1988). In mice, the palate is formed relatively late in organogenesis, with the palatal shelves initially appearing at embryonic day (E) 11.5 and growing vertically adjacent to the developing tongue from E12.5 to E14.0. However, by E14.5 the shelves have elevated above the tongue and grown to meet their counterpart at the midline, where the layers of epithelium adhere and then fuse with each other to achieve continuity in the roof of the oral cavity (**Figures 1A–L**).

Palatal shelf elevation is a rapid process, accompanied and facilitated by changes within the extracellular matrix of the palatal shelf mesenchyme and the coordinated movement of other craniofacial structures. It is generally accepted that elevation of the palatal shelves above the tongue and their associated change in orientation from a vertical to horizontal position, arises from a combination of intrinsic and extrinsic forces, including descent of the tongue (Ferguson, 1988). The complexity of palatogenesis means that in humans it is frequently disturbed, resulting in the birth defect of cleft palate (reviewed in Ferguson, 1988; Cobourne, 2004). The causes of cleft palate as a malformation can be broadly categorized on an embryological basis as a lack of adequate growth in the palatal shelves, failure to elevate above the tongue or a breakdown in the mechanism of fusion between the shelves. In addition, cleft palate can also arise secondary to other craniofacial malformations, such as micrognathia and basoccipital or basisphenoid fusion, craniosynostosis and both muscle and tongue abnormalities

FIGURE 1 | Frontal sections through the developing craniofacial region of the early mouse embryo between E13.5 and E16.5. At E13.5, the palatal shelves are positioned vertically adjacent to the developing tongue. At E14.5 the shelves have elevated above the tongue and grown to meet their counterpart at the midline, where the layers of epithelium adhere and begin to fuse with each other. At E15.5, continuity has been achieved and the palate separates the oral and nasal cavities. At E16.5, palatogenesis is essentially complete. g, genioglossus; hf, hair follicle; itb, incisor tooth bud; Mc, Meckel's cartilage; mes, medial epithelium seam; mtb molar tooth bud; nc, nasal cavity; ns, nasal septum; pb, presphenoid bone; ps, palatal shelves; sm, submandibular gland; t, tongue. Scale bar in L = 500 µm for (A–L).

(reviewed in Ferguson, 1988; Rice et al., 2004; Casey et al., 2006; Chai and Maxson, 2006; Gritli-Linde, 2007; Xiong et al., 2009).

Development of the vertebrate tongue involves contributions from CNC cells derived from pharyngeal arches 1–3 and the somitic myoblasts (Parada and Chai, 2015). The oral portion or anterior two thirds of the murine tongue emerges from the floor of the early oral cavity as a set of mesenchymal swellings derived from the first branchial arch. A medial lingual swelling initially forms, but this is rapidly engulfed by two lateral lingual swellings that will form the anterior two thirds proper. The posterior third or pharyngeal component is derived from two further swellings within the third branchial arch, the copula and hypopharyngeal eminence (Noden and Francis-West, 2006; Hosokawa et al., 2010). In the mouse embryo, the process of tongue development begins around E10.5, with a noticeable tongue bud evident by E12.5, which undergoes rapid enlargement and differentiation to form a large muscular organ by E16.5 (Parada et al., 2012; **Figures 1A–L**).

The Eph family receptor-interacting (ephrin) ligands and erythropoietin-producing hepatocellular carcinoma (Eph) receptors have been extensively studied since their discovery (Hirai et al., 1987). Ephs constitute the largest known family of receptor tyrosine kinases, comprising at least 16 distinct receptors that are highly conserved (Hirai et al., 1987; Jones et al., 1995; Scales et al., 1995; Lackmann and Boyd, 2008; Islam et al., 2010). Based on structural features in their ligandbinding domains and their ephrin-binding preferences, Ephs are classified into 10 EphA and 6 EphB receptors. The EphA group preferentially bind glycosylphosphatidylinositol (GPI) linked ligands of the ephrin-A subclass; whilst the EphB group preferentially interact with transmembrane ligands of the ephrin-B subclass (reviewed in Lackmann and Boyd, 2008). However, EphA4 binds both classes of ephrin and EphB2 can bind ephrinA5 (Himanen et al., 2004; Dravis and Henkemeyer, 2011).

Together, Eph receptors and their ligands, form an important cell communication system with widespread roles in normal physiology and disease pathogenesis (Pasquale, 2005, 2010). Eph–ephrin complexes emanate bidirectional signals, forward signals that depend on Eph kinase activity propagated in the receptor-expressing cell and reverse signals, that depend on Src family kinases propagated in the ephrin-expressing cell. Ephrindependent but kinase-independent Eph signals can also occur (Gu and Park, 2001; Matsuoka et al., 2005; Miao et al., 2005). Eph signaling is known to control cell morphology, adhesion, migration, and invasion by modifying organization of the actin cytoskeleton and influencing the activities of integrins and intercellular adhesion molecules (Pasquale, 2005, 2010; Klein, 2012).

There is evidence from both humans and mice for the potential involvement of specific ephrin and Eph family members during palate development. In the human craniofrontonasal syndrome, mutations in EPHRINB1 give rise to a range of cranial defects, including cleft lip and palate (Twigg et al., 2004; Wieland et al., 2004; Torii et al., 2007); whilst targeted disruption of EphrinB1 in mice results in craniofacial and other skeletal defects, including cleft palate (Orioli et al., 1996; Compagni et al., 2003; Davy et al., 2004). Additionally, engineering of compound transgenic mice for EphB2 and EphB3 leads to cleft palate; suggesting that a combination of EphB3 protein and EphB2 forward signaling is important for palate development (Risley et al., 2009).

EphA-family receptor expression patterns have previously been described in the developing palate (Agrawal et al., 2014); however, only limited tongue expression data was shown. On the basis of this previous data and a rudimentary PCR-based screen of EphA transcriptional activity in the developing palate (data not shown) we have investigated expression of EphA-family members and their ephrin-A ligands during murine palate and tongue development. We also mapped ephrinb2 expression in these regions, given that this ligand interacts with EphA4, and ephrinB2 reverse signaling is known to be important for normal closure of the secondary palate (Dravis and Henkemeyer, 2011). In addition, EphB1 and EphB4 expression was analyzed. EphB1 has also been associated with cleft lip and palate in human populations (Watanabe et al., 2006) and previously identified as the preferred receptor of ephrinB2 in the mechanism of axonal pathfinding (Chenaux and Henkemeyer, 2011); whilst EphB4 only binds ephrinB2 amongst all the ephrin-B family ligands (Sakano et al., 1996).

We find widespread expression of these family members during murine palatogenesis. In addition, regionally-restricted expression of many members in the developing tongue, suggests some commonalities during the coordinated development of the palate and tongue.

# MATERIALS AND METHODS

Mouse plasmids containing cDNA were linearized with the appropriate restriction enzymes and antisense <sup>35</sup>S-UTP radiolabeled riboprobes generated using specific RNA polymerases (**Table 1**).

CD-1 mice were time-mated and pregnant females sacrificed with cervical dislocation. Matings were set up such that noon of the day on which vaginal plugs were detected was considered as E0.5. Embryos were collected between E13.5 and E16.5, fixed in 4% (w/v) paraformaldehyde at 4◦C overnight, washed in PBS, dehydrated through a graded series of ethanols, embedded in paraffin wax and sectioned at 7µm, prior to section in situ hybridisation.

Radioactive section in situ hybridisation was carried out as previously described (Xavier et al., 2009). Light and dark-field images of sections were photographed using a Zeiss Axioscop microscope and merged in Adobe Photoshop CS2.

# RESULTS AND DISCUSSION

EphrinA1 transcripts were identified in the palate epithelium from E13.5 to E16.5 (**Figures 2A–D**), particularly at the tip of the palatal shelf at E13.5 (**Figure 2A**, highlighted) with strong expression throughout the oral surface of the palatal shelf epithelium at E14.5 (**Figure 2B**, highlighted). In contrast,

### TABLE 1 | Plasmids used for the generation of riboprobes.


EphrinA1 was only expressed at background levels between E13.5 and E16.5 (**Figures 2A–D**). EphrinA2 showed no specific epithelial expression in the palate at E13.5, although transcripts were present in the mesenchyme (**Figure 2E**, highlighted); however, by E14.5 distinct transcriptional activity was observed in the MES (**Figure 2F**, highlighted). During subsequent development at E15.5–E16.5 EphrinA2 upregulated in the palatal shelf epithelium (**Figures 2G,H**). EphrinA3 was not detected above low-level background signal in the developing palate between E13.5 and E16.5 (data not shown). However, at E13.5 transcripts were identified in epithelium of the developing vomeronasal organ and nasal cavity (**Figure 2I**), expression domains that were maintained between E14.5 and E16.5 (**Figures 2J–L**).

EphA1 was generally expressed in the palatal shelf mesenchyme at E13.5, and in a complementary manner to its ligand EphrinA1 (see **Figure 2A**), was upregulated in mesenchyme at the tip of the shelves (**Figure 3A**, highlighted). Lower-level expression was maintained in the palatal mesenchyme at later stages (**Figures 3B–D**), but at E15.5–E16.5

EphA1 was clearly upregulated in the oral epithelium after palatal shelf fusion (**Figures 3C,D**). In contrast, no expression was detected in epithelium of the MES (**Figure 3B**), which is in agreement with previous findings (Agrawal et al., 2014). EphA2 was detected in the palatal shelf epithelium from E13.5– E16.5 (**Figures 3E–H**); although no transcriptional activity was observed in the MES (**Figure 3F**). EphA2 has been shown to function as a positive regulator of mammary epithelial proliferation and branching (Vaught et al., 2009; Park J. E. et al., 2013) and it is known that growth of the palatal shelves is controlled by reciprocal epithelial-mesenchymal interactions along the antero-posterior axis (Bush and Jiang, 2012; Economou et al., 2013). Based on the distinctive expression pattern within the epithelium, EphA2 may be important for normal growth of the early palatal shelves.

EphA3 was intensely expressed throughout the palatal shelves at E13.5 (**Figure 3I**), although this expression became localized to regions of epithelium at E14.5 (**Figure 3J**, highlighted) in contrast to previous observations, transcripts were also detected in the midline during the process of fusion, including the MES and regions of adjacent mesenchyme (**Figure 3J**, arrowhead). EphA3 remained enriched in these regions of the palate epithelium and mesenchyme during subsequent stages of palatogenesis between E15.5 and E16.5 (**Figures 3K,L**; Agrawal et al., 2014). EphA4 was also strongly expressed throughout the palatal shelves prior to elevation at E13.5 (**Figure 3M**), progressively localizing to the oral epithelium and MES during later development (**Figures 3N–P**). Despite this dynamic expression pattern, an absence of both EphA3 and EphA4 function does not result in any overt developmental phenotype in the mouse, including the palate. Redundant roles played by other family members may explain the lack of palate phenotype in compound EphA3−/−; EphA4−/<sup>−</sup> mutant embryos (Agrawal et al., 2014). EphA5 hybridization signals were present in a patchy distribution within the mesenchymal component of the palatal shelves at E13.5, (**Figure 4A**, highlight); whilst during later stages, expression was detected throughout the epithelium and very strongly in mesenchyme at the lateral edges of the palate, with this strong expression also observed in the nasal cavity epithelium (**Figure 4B**). Following fusion at E15.5, EphA5 was localized to the palatal epithelium (**Figure 4C**, arrowed); however, at E16.5, marked up-regulation was observed in the mesenchyme, but restricted to medial regions of the fused shelves (**Figure 4D**). EphA6 transcripts were not detected at any significant level in the palatal shelves at E13.5 (data not shown) although some upregulation was seen in mesenchyme of the nasal cavity

(**Figure 4E**, arrowheads). Transcripts were detected in palatal epithelium of the oral cavity during fusion at E14.5, but they were absent from the MES (**Figure 4F**) and no expression was observed following fusion of the palatal shelves at E15.5 (**Figure 4G**) and E16.5 (data not shown). Interestingly, there was strong localized expression of EphA6 in epithelium of the oral commissure at E15.5 (**Figure 4G**, arrowed) and intense expression also identified in the lens and neural layer of the retina at E16.5 (**Figure 4H**). EphA7 was consistently detected in the palatal shelf epithelium throughout palatogenesis, but only weakly in the mesenchyme (**Figures 4I–L**, highlight in **Figure 4K**). However, at E15.5 strong midline expression was detected (**Figure 4K**, arrowhead). This expression pattern was different from that described in previously published data, where EphA7 was mainly observed in the mesenchyme (Agrawal et al., 2014). EphA8 showed intense expression in both the epithelium and mesenchyme of the palatal shelves at E13.5 (**Figure 4M**), with lower-level expression at later stages; again, with the exception at E15.5, where strong expression was detected in the midline mesenchymal region (**Figures 4N–P**, arrowhead in **Figure 4O**), which also differs from that previously reported (Agrawal et al., 2014). Recently, in vivo expression of EphA8-Fc was reported to result in neuroepithelial cell apoptosis and a subsequent decrease in brain size (Kim et al., 2013). These findings are in agreement with previous studies that demonstrated that Ephrin-Eph signaling plays a critical role in determining the size of the neuroepithelial cell population during early embryonic brain development (Holmberg et al., 2000; Park E. et al., 2013). EphA8 may therefore have a role in mediating epithelial apoptosis during the process of palatal shelf fusion.

EphrinB2 was expressed in the epithelium and (more weakly) in the mesenchyme during palatogenesis, particularly at the tip of the palatal shelves at E13.5 and in the MES at E14.5 (**Figures 5A–D**, highlighted in **Figure 5A**). EphB1 transcriptional activity was weak but widespread in the palatal shelf mesenchyme at E13.5–E14.5; however, by E15.5 expression was up-regulated in the midline of the embryonic palate, returning to previous levels by E16.5 (**Figures 5E–H**). EphB4 was also weakly expressed in the palatal shelf mesenchyme throughout palatogenesis, but with strong midline expression at E14.5 in the MES during shelf fusion (**Figures 5I–L**, arrow in **Figure 5J**). EphB1 expression has been previously reported in the venous vasculature throughout embryonic development to adulthood (Li and Mukouyama, 2013). Additionally, EphB1

has been observed in the mouse retina (Birgbauer et al., 2000), during the early stages of embryonic rat spinal cord development (Jevince et al., 2006) and in the basal ganglia nuclei (Richards et al., 2007). Interestingly, behavioral evaluation of EphB1 null mice in an open-field environment has revealed the presence of spontaneous locomotor hyperactivity (Richards et al., 2007). During palatogenesis streams of directional cell migration (both in the anterior and posterior aspect) have been demonstrated to

occur and are thought to be of importance for palate patterning (shaping) and elevation (He et al., 2008). Interestingly, a cellular migration system solely dependent on EphrinB2–EphB4 signal transduction has demonstrated that EphB4 is capable of triggering the regulation of cell migration (Sturz et al., 2004). Taken together, these results suggest that these genes could also be involved in cell migration events that take place during palate development.

Diagrams not drawn to scale.

FIGURE 6 | The main domains of expression associated with the EphA family of receptors in the developing tongue at (A) E13.5 and (B) E14.5.

We also identified the expression of numerous Ephs during murine tongue development between E13.5 and E16.5. The main domains of expression associated with the EphA group in the tongue at E13.5 and E14.5 are summarized in **Figures 6A,B**. EphA5 and EphA7 presented with ubiquitous expression at E14.5 (see **Figures 4B,J**, respectively). During all stages examined, EphA1 and EphA2 demonstrated distinctive expression in the tongue epithelium (**Figures 3A–H**), whereas EphA3 was strongly expressed in the lateral tongue mesenchyme between E13.5 and E16.5 (**Figures 3I–L**). Although EphA4 was also detected in the lateral tongue mesenchyme at earlier stages (**Figures 3M,N**); by E15.5, transcriptional activity was down-regulated (**Figure 3O**) and restricted to patchy regions of the epithelium (**Figure 3O**), although at E16.5, expression was increased in the mesenchyme (**Figure 3P**). EphA6 presented weak and widespread expression in the mesenchyme during tongue development (**Figures 3E–G**). However, at E14.5, a marked upregulation was observed in the inter-molar eminence of the tongue (**Figure 4F**). Similarly to EphA3, EphA7, and EphA8 transcriptional activity were also markedly increased in the lateral mesenchyme of the tongue during development (**Figures 4I–O**). However, by E16.5 EphA8 expression was down-regulated and more restricted to the epithelial compartment (**Figure 4P**). Rapid depression of the tongue in embryogenesis is critical for proper palatogenesis. Any delay in this process can disturb palatal shelf elevation and hence, lead to cleft palate (Nie, 2005). For these events to take place a coordinated balance between apoptosis and proliferation is essential (Parada et al., 2012; Parada and Chai, 2015).

The Eph and ephrin family-member gene expression in the developing palate and tongue described here is summarized in **Figures 6**, **7**. These dynamic domains suggest important potential roles for these molecules in both epithelium and mesenchyme during development of these regions. Further analysis using animal models will be required to delineate the precise requirements during these developmental processes. However, the co-expression of EphA3, A4, and A8 in the palatal shelves makes it difficult to test the hypothesis that these genes are involved in palatogenesis. Considering the known promiscuous interactions between Ephs and ephrins, it is likely EphA3, A4, and A8 may also play redundant roles during palate development. Analysis of a triple loss-of-function mouse model may be required to definitively address this question.

# CONCLUSIONS

Eph receptors A3, A4, and A8 are very strongly expressed within palatal shelf mesenchyme during early palatogenesis and both EphA1 and A5 are up-regulated at the shelf tip during this stage. Eph receptors A3, A4, and A8 are also strongly expressed in lateral regions of the tongue at these stages, suggesting some co-ordination in the regulation of palatogenesis and tongue development. EphA and ephrinA-family members are also expressed in palatal shelf epithelium (EphA2, EphA7, ephrinA1) and mesenchyme (EphA1, A3, A4, A5, A6, A8, and ephrinA2, A3) suggesting the possibility of epithelial-mesenchymal interactions being mediated by these proteins during development of the palate.

# AUTHOR CONTRIBUTIONS

MC and IM conceived the experiments, GX and IM conducted the experiments and undertook data acquisition, GX, IM, and MC wrote the manuscript.

# ACKNOWLEDGMENTS

GMX is the recipient of a National Institute of Health Research, UK Clinical Lectureship. GMX is supported by the Academy of Medical Sciences (Wellcome Trust, British Heart Foundation, Arthritis Research, UK). We thank David Anderson, Andrea Ballabio, Tyler Cutforth, David Feldheim, Mark Henkemeyer, and David Wilkinson for kindly providing plasmid cDNA.

# REFERENCES


mammalian small intestine. Dig. Dis. Sci. 55, 2478–2488. doi: 10.1007/s10620- 009-1102-z


**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 © 2016 Xavier, Miletich and Cobourne. 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) or licensor 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 Critical Role of TRPM7 As an Ion Channel Protein in Mediating the Mineralization of the Craniofacial Hard Tissues

Yukiko Nakano1, <sup>2</sup> , Michael H. Le<sup>1</sup> , Dawud Abduweli <sup>1</sup> , Sunita P. Ho<sup>3</sup> , Lillia V. Ryazanova<sup>4</sup> , Zhixian Hu<sup>4</sup> , Alexey G. Ryazanov <sup>4</sup> , Pamela K. Den Besten1, 2 and Yan Zhang1, 2 \*

<sup>1</sup> Department of Orofacial Sciences, University of California, San Francisco, San Francisco, CA, USA, <sup>2</sup> Center for Children's Oral Health Research, University of California, San Francisco, San Francisco, CA, USA, <sup>3</sup> Preventive and Restorative Dental Sciences, University of California, San Francisco, San Francisco, CA, USA, <sup>4</sup> Department of Pharmacology, Robert Wood Johnson Medical School, Piscataway, NJ, USA

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Thomas G. H. Diekwisch, Texas A&M University Baylor College of Dentistry, USA Claudio Cantù, University of Zurich, Switzerland

> \*Correspondence: Yan Zhang yan.zhang2@ucsf.edu

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 13 April 2016 Accepted: 13 June 2016 Published: 06 July 2016

### Citation:

Nakano Y, Le MH, Abduweli D, Ho SP, Ryazanova LV, Hu Z, Ryazanov AG, Den Besten PK and Zhang Y (2016) A Critical Role of TRPM7 As an Ion Channel Protein in Mediating the Mineralization of the Craniofacial Hard Tissues. Front. Physiol. 7:258. doi: 10.3389/fphys.2016.00258 Magnesium ion (Mg2+) is the fourth most common cation in the human body, and has a crucial role in many physiological functions. Mg2<sup>+</sup> homeostasis is an important contributor to bone development, however, its roles in the development of dental mineralized tissues have not yet been well known. We identified that transient receptor potential cation channel, subfamily M, member 7 (TRPM7), was significantly upregulated in the mature ameloblasts as compared to other ameloblasts through our whole transcript microarray analyses of the ameloblasts. TRPM7, an ion channel for divalent metal cations with an intrinsic serine/threonine protein kinase activity, has been characterized as a key regulator of whole body Mg2<sup>+</sup> homeostasis. Semi-quantitative PCR and immunostaining for TRMP7 confirmed its upregulation during the maturation stage of enamel formation, at which ameloblasts direct rapid mineralization of the enamel matrix. The significantly hypomineralized craniofacial structures, including incisors, molars, and cranial bones were demonstrated by microCT analysis, von Kossa and trichrome staining in Trpm71kinase/<sup>+</sup> mice. A previously generated heterozygous mouse model with the deletion of the TRPM7 kinase domain. Interestingly, the skeletal phenotype of Trpm71kinase/<sup>+</sup> mice resembled those found in the tissue-nonspecific alkaline phosphatase (Alpl) KO mice, thus we further examined whether ALPL protein content and alkaline phosphatase (ALPase) activity in ameloblasts, odontoblasts and osteoblasts were affected in those mice. While ALPL protein in Trpm71kinase/<sup>+</sup> mice remained at the similar level as that in wt mice, ALPase activities in the Trpm71kinase/<sup>+</sup> mice were almost nonexistent. Supplemented magnesium successfully rescued the activities of ALPase in ameloblasts, odontoblasts and osteoblasts of Trpm71kinase/<sup>+</sup> mice. These results suggested that TRPM7 is essential for mineralization of enamel as well as dentin and bone by providing sufficient Mg2<sup>+</sup> for the ALPL activity, underlining the key importance of ALPL for biomineralization.

Keywords: TRPM7, enamel, dentin, bone, ion transport, biomineralization, alkaline phosphatase, magnesium homeostasis

# INTRODUCTION

Magnesium is the fourth most common cation in the human body, and is the second most abundant cellular cation (Romani, 2011). Intracellularly, by binding to the enzymes, magnesium functions as an essential activator of enzymes (Cowan, 2002; Maguire and Cowan, 2002; Sreedhara and Cowan, 2002), and by binding to nucleic acids, it contributes to the second messenger systems and modification of nucleic acid structure (Neitzel et al., 1991; Barciszewska et al., 2001). Moreover, it binds to cellular membrane components, including ion channels, and affects fluidity and permeability of molecules (Wolf and Cittadini, 2003; Wolf et al., 2003). Magnesium deficiency in humans is known to be associated with skeletal diseases, including hypocalcemia and osteoporosis, due to impaired parathyroid hormone (PTH) secretion, renal and skeletal resistance to PTH and vitamin D, and increasing inflammatory cytokines, like interleukin (IL) -1 and tumor necrosis factor (TNF) -α (Weglicki et al., 1996; Rude and Gruber, 2004; Rude et al., 2004; Rude and Shils, 2006). In animal models treated with low Mg2<sup>+</sup> diet, dentin and enamel mineralization defects are reported (Irving, 1940; Bernick and Hungerford, 1965; Trowbridge et al., 1971).

Transient receptor potential melastatin-subfamily member 7 (TRPM7) is a permeable ion channel for divalent metal cations, preferentially permitting the flow of Mg2<sup>+</sup> and Ca2<sup>+</sup> (Nadler et al., 2001; Monteilh-Zoller et al., 2003; Penner and Fleig, 2007). TRPM7 has an essential role in the regulation of both cellular and whole body Mg2<sup>+</sup> homeostasis, modulating fundamental cellular processes including cell division, growth, survival, differentiation, and migration (Ryazanova et al., 2010; Yee et al., 2014). The c-terminus of TRPM7 is a serine/threonineprotein kinase domain which functions as an intracellular sensor of magnesium status, and thus, provides coordination of cellular and systemic responses to magnesium deprivation (Ryazanova et al., 2014). In the whole body, TRPM7 is ubiquitously expressed and homozygous deletion of TRPM7 kinase domain is embryonic lethal, indicating that this molecule has a fundamental and non-redundant role in cellular physiology (Nadler et al., 2001; Ryazanova et al., 2010). Heterozygous KO mice for TRPM7 kinase domain (Trpm71kinase/<sup>+</sup> mice) are viable, but there is a change in magnesium homeostasis or hypomagnesemia-like phenotype. Sensitivity to intracellular Mg2<sup>+</sup> levels is a critical mechanism to regulate the Mg2<sup>+</sup> influx through TRPM7 channel into the cells, and the Trpm71kinase/<sup>+</sup> mice shows increased sensitivity to the inhibition by Mg2<sup>+</sup> (Ryazanova et al., 2010, 2014).

Although, the magnesium deficiency is known to cause skeletal and tooth defects, the role of TRPM7 in hard tissue formation including the tooth mineralization has not been determined. Through, two whole transcript microarray analyses of varying stages of differentiating ameloblasts, we found that Trpm7 was upregulated in secretory ameloblasts as compared to presecretory ameloblasts (GEO accession number GSE59214; Liu et al., 2015), and in maturation as compared to secretory ameloblasts (GEO accession number GSE57224; Zhang et al., 2014). Taking into consideration the significance of the maturation stage in enamel formation, where the majority of enamel mineralization occurs (Robinson et al., 1995; Smith, 1998), together with the importance of intracellular Mg2<sup>+</sup> homeostasis in the skeletogenesis, we hypothesize that TRPM7 potentially contributes to the enamel matrix mineralization. In this study, we therefore confirmed the expression and synthesis of TRPM7 in differentiating ameloblasts, and further investigated the function of TRPM7 associated with the mineralization of craniofacial hard tissues using Trpm71kinase/<sup>+</sup> mice model. We determined a relationship of TRPM7 and tissuenonspecific alkaline phosphatase activity to critically regulate the mineralization of craniofacial hard tissue.

# METHODS

# Animals

All animal procedures were performed upon the approval by the Institutional Animal Care and Use Committee (IACUC) of the University of California, San Francisco and Rutgers Robert Wood Johnson Medical School, and adhered to the principles outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Trpm71kinase/<sup>+</sup> and wt mice were provided by Dr. Alexey G. Ryazanov (Rutgers Robert Wood Johnson Medical School). As previously described, Trpm71kinase/<sup>+</sup> mice were genetically modified by replacing exons 32–36 of Trpm7 gene, the kinase domain, with the Neo gene cassette (Ryazanova et al., 2010). At postnatal day 14 days, mice were euthanized, and whole heads were dissected out and fixed with 4% paraformaldehyde (PFA) overnight.

C57BL/6J female mice were maintained at the UCSF animal facility. At postnatal day 0 (P0), 5 (P5), and 10 (P10), mice were euthanized, developing first molars were harvested and processed for total RNA extraction. For immunohistochemical staining, 7-week old female C57BL/6J mice were anesthetized with 240 mg/kg tribromoethanol (Sigma-Aldrich, St. Louis, MO), fixed with 4% PFA for overnight.

# Total RNA Extraction and Semi-Quantitative PCR (qPCR)

Total RNA was purified from developing molar tooth organ using the RNeasy Mini kit (Qiagen, Germantown, MD). The tooth organs were not homogenized therefore RNA would be primarily extracted from the exposed enamel epithelium overlying the tooth bud. cDNA was obtained by reverse transcription of the mRNA using Superscript III First-Strand Synthesis Supermix for qRT-PCR (Life Technologies, Grand Island, NY).

Expression of Trpm7 was examined by semi-quantitative PCR with FastStart Universal SYBR Green Master Kit (Roche Diagnostics, Indianapolis, IN) using the ABI 7500 system (Applied Biosystems, Foster City, CA). The primer sequences for Trpm7 are: sense 5′ -ATGGCACTGTTGGAAAGTATGG-3′ , antisense 5′ -CGCCTTCAAATATCAAAGCCAC-3′ ; Eef1a1 was used as a reference gene, the primer sequences are: sense 5′ -CAA CAT CGT CGT AAT CGG ACA-3′ , antisense 5′ -GTC TAA GAC CCA GGC GTA CTT-3′ . The expression levels of target gene was analyzed using the 11Ct method (Livak and Schmittgen, 2001). The relative expression levels of Trpm7 of P5 and P10 enamel organs were calculated based on the expression levels of P0 enamel organs. Significance of differences was determined using 1CT values by the multiple t-test with Bonferroni correction following ANOVA (Baker et al., 2006; Yuan et al., 2006).

# Micro-Computed Tomography (MICROCT)

Whole heads from wt and Trpm71kinase/<sup>+</sup> mice were fixed in 4% PFA overnight, and then imaged using a Micro XCT-200 system (Xradia, Pleasanton, CA). All scans were done at an operating voltage of 90 KVp and 66 µA of current, at an optical magnification 2x. A binning of 2 was used for 3D image reconstruction. All scans were done using the same experimental settings, including the distances between specimen, detector, and source. Virtual sections were converted to bmp images using the Xradia TXM3DViewer 1.1.6. software. Appropriate imaging planes were selected from three orthogonal sections centered at a level containing an entire sagittal slice of the incisor or entire frontal slice of the first molar containing the mesial and buccal cusps and mesial root inside the reconstructed space using Xradia TXM3DViewer 1.1.6. software.

# Immunohistochemistry

The mandibles and maxillae were decalcified in 8% EDTA (pH 7.3) at 4◦C for 2 wks (7-week-old C57BL/6J mice) or 1 wk (wt and Trpm71kinase/<sup>+</sup> mice). The jaws were further dehydrated through a graded series of ethanol, followed by a routine paraffin embedding. The paraffin blocks were sectioned at the thickness of 5 µm. After dewaxing, the sections used for TRPM7 immunostaining were subjected to the antigen retrieval in 1% SDS in 0.1 M Tris-HCL buffer (pH 9.0) for 5 mins at room temperature (Brown et al., 1996; Emoto et al., 2005), and then treated with 1% H2O<sup>2</sup> for 5 min at room temperature. Afterward, all sections were incubated with the blocking reagent containing 10% swine and 5% goat sera for 2 h at room temperature followed by an incubation with either rabbit anti-TRPM7 antibody (Abcam, Cambridge, United Kingdom) overnight at 4◦C or rabbit anti-human ALPL (Abcam) antibody overnight at 4◦C. Sections were further incubated with biotin conjugated swine anti-rabbit F(ab')<sup>2</sup> secondary antibody (Dako Denmark A/S, Glostrup, Denmark) for 1 h at room temperature. Next, the sections incubated with anti-TRPM7 antibody were incubated with intestinal alkaline phosphatase (ALPase) conjugated streptavidin (Vector Laboratories Inc., Burlingame, CA) for 30 mins, and immunoreactivity was visualized using a Vector <sup>R</sup> Red kit (Vector Laboratories Inc.) resulting in pink/red color for positive staining. To block the endogenous tissue-nonspecific ALPase activity, 1 mM levamisole was added to the visualization reagent. The sections incubated with anti-ALPL antibody were incubated with horseradish peroxidase (HRP) conjugated streptavidin (Vector Laboratories Inc.) for 30 mins, and immunoreactivity was visualized using an ImmPACT tM DAB peroxidase substrate kit

(Vector Laboratories Inc.), resulting in dark brown color for positive staining. Nuclear counterstaining was performed with methyl green (Dako Denmark A/S). Normal rabbit IgG was used as the negative control.

# Histological Analysis

The fixed mandibles and maxillae were dehydrated in acetone, and embedded in Technovit 8100 resin (Heraeus Kulzer GmbH, Hanau, Germany). Undecalcified sections were obtained at the thickness of 3 µm. Mineralization of the tissue was visualized by 2.5% silver nitrate staining (von Kossa stain). Bone and dentin mineralization was further assessed by following Goldner's Trichrome stain protocol (Goldner, 1938). For general morphological analysis, sections were stained with 1% toluidine blue.

# Enzyme Histochemistry for ALPase Activity

To detect ALPase activity, the undecalcified Technovit sections were incubated with the modified Burstone's reagent comprising 1.5 mM Naphthol AS-MX phosphate, 0.5 mM Fast Red Violet LB salt and 3 mM MgSO<sup>4</sup> in 0.1 M Tris–HCl buffer (pH 9.2) at 37◦C for 60 mins (Nakano et al., 2004). Some of the sections were preincubated with 0.1 M Tris–HCl buffer (pH 7.3) supplemented with 50 mM MgSO<sup>4</sup> for 1 day at 4◦C (Yoshiki et al., 1972; Nakano et al., 2004).

# RESULTS

# Expression and Synthesis of Trpm7 Progressively Increased with Ameloblast Differentiation

Analysis of our previous microarray data (GSE59214 and GSE57224) revealed that expression of Trpm7 was progressively upregulated from pre-secretory ameloblasts, to secretory ameloblasts, then to maturation ameloblasts. To confirm this, we used qPCR to compare relative Trpm7 expression in the ameloblasts obtained from developing mouse enamel organs at three differentiation stages; pre-secretory/P0, secretory/P5 and early maturation/P10 stages. The qPCR analysis showed that the relative expression level of Trpm7 compared with pre-secretory ameloblasts was 5.4-fold in secretory ameloblasts and 16.08-fold in maturation stage ameloblasts (**Figure 1A**). TRPM7 protein was detected in ameloblasts of all stages, and the intensity of the immunostaining increased as differentiation of ameloblasts advanced, with the highest immunostaining signal in maturation ameloblasts (**Figures 1Ba–e**). Moreover, we found that the odontoblasts (Od) and osteoblasts (Os) were also immunostained for TRPM7 (**Figures 1Bb,f,g**). No immunoreaction was detected on the negative control sections (**Figure 1Bh**).

# Significantly Hypomineralized Enamel, Dentin and Cranial Bones Are Found in Trpm71kinase/<sup>+</sup> Mice

To further understand the TRPM7 functions in enamel and dentin formation, and craniofacial skeletogenesis, we compared the craniofacial structure of Trpm71kinase/<sup>+</sup> mice to wt controls. Incisors of P14 wt mice were hard and translucent (**Figure 2A**), whereas Trpm71kinase/<sup>+</sup> mice were soft, and the red vasculature enriched dental pulp was easily seen through the enamel and dentin layers (red box in **Figure 2B**). Moreover, craniofacial bones were also soft and easily bent like a piece of paper.

Consistent with the gross morphology, the 3D microCT of craniofacial structure showed that in wt mice, the well mineralized incisors, molars, and craniofacial bones were

matrix of Trpm71kinase/<sup>+</sup> mice is negative for Von Kossa staining (F), and Trichrome staining on alveolar bone showed bone in Trpm71kinase/<sup>+</sup> mice is light pink (H), similar to osteoid. Unlike Von Kossa positive stained wt molar dentin (I,M), the entire dentin of the Trpm71kinase/<sup>+</sup> first molar is negative for Von Kossa staining (J) and shows pink/pre-dentin status by Trichrome staining (L). Interestingly, in the second molar of Trpm71kinase/<sup>+</sup> mice, both Von Kossa (N) and Trichrome (P) staining shows the coronal dentin as partially mineralized, but the root dentin is not mineralized. Enamel matrix of molars likely chipped off during the sectioning (seen as an enamel space/EnS), and the remaining matrix is positive for Von Kossa staining (N). Scale bars: 100 µm

distinguishable at the same intensity (**Figure 2C**), while in Trpm71kinase/<sup>+</sup> mice, mineralization was only detected in the crowns of molars, and in smaller area of the incisors specifically at incisal end (**Figure 2D**). Virtual sagittal sections from the Trpm71kinase/<sup>+</sup> hemimandibles further demonstrated limited mineralization in the crown of the molar and the incisor (**Figures 2F,I,J**) as compared to wt controls (**Figures 2E,G,H**).

Histological assessment on the undecalcified mandibular sections showed von Kossa positive stained (in dark) enamel (En), dentin (Dn), and alveolar bone (**Figures 3A,E,I,M**), and a lack of von Kossa positive staining on the Trpm71kinase/<sup>+</sup> incisor enamel and dentin, molar root and alveolar bone (**Figures 3B,F,J,N**). Trichrome staining showed that dentin and bone in the wt mice were stained in blue (**Figures 3C,G,K,O**), while incisor dentin, molar root dentin and bones in Trpm71kinase/<sup>+</sup> mice were stained in pink, although the width of dentin and bone layer was comparable (**Figures 3D,H,L,P**), indicating that the dentin and bone matrix in the Trpm71kinase/<sup>+</sup> mice was deposited at the same levels as wt mice but the mineralization did not proceed normally. The coronal dentin of the first molars of Trpm71kinase/<sup>+</sup> mice was negative for von Kossa staining (**Figure 3J**), while the coronal dentin of the second molar was positive for the same staining (**Figure 3N**). Consistently, by staining with Trichrome, the mineralization status of the coronal dentin was also confirmed as pink/unmineralized, in the first molars (**Figure 3L**), and blue gray/mineralized, in the second molar (**Figure 3P**).

FIGURE 4 | There are no obvious morphological change in ameloblasts of Trpm71kinase/+ mice. Morphodifferentiation of ameloblasts (Am) from pre-secretory stage to secretory stage in Trpm71kinase/<sup>+</sup> mice (F–J) is similar to that of wt mice (A–E). Am, ameloblast; SI, stratum intermedium; PL, papillary layer; scale bars: 25 µm

# Ameloblast Morphology Is Not Significantly Altered in Trpm71kinase/<sup>+</sup> Mice

The morphology of ameloblasts is associated with the functional property of each differentiation stage. As mentioned above, TRPM7 was immunolocalized in ameloblasts in a stage-specific manner. Therefore, we examined the morphology of ameloblasts by Toluidine blue staining in order to determine whether TRPM7 is critical for the morphological differentiation of ameloblasts. The morphological property of the ameloblasts in the Trpm71kinase/<sup>+</sup> mice at all stages remained similar to that of the wt mice throughout the differentiation. Only the height of ameloblast layer appeared to be slightly shorter (**Figure 4**).

# Acellular Cementum Is Absent in Trpm71kinase/<sup>+</sup> Mice

Through further morphological studies, we found that acellular cementum in molars, normally stained as a deep blue line by toluidine blue in undecalcified sections (McKee et al., 2011; **Figures 5A,B**), was absent in the Trpm71kinase/<sup>+</sup> mice (**Figures 5C,D**). While the most of the root analog surface was covered by the acellular cementum in wt incisors (**Figures 5E–H**), acellular cementum was only detected near the incisal end on the incisor of Trpm71kinase/<sup>+</sup> mice (**Figures 5I–L**).

# Ex vivo AlPase Activity Is Abolished in the Trpm71kinase/<sup>+</sup> Mice, But Can Be Rescued in Sections Pre-incubated with Mg2<sup>+</sup> Solution

We found that the mineralization phenotypes in the Trpm71kinase/<sup>+</sup> mice, including hypomineralization in bone, incisor and molar root and absence of acellular cementum, resembled those of the tissue-nonspecific alkaline phosphatase (Alpl) KO mice (Beertsen et al., 1999; McKee et al., 2011). These phenotypes led us to examine the abundance of ALPL and ex vivo ALPase enzyme activity in osteoblasts (Os), odontoblasts (Od), and ameloblasts (Am). The intensity and localization of ALPL immunostaining in Os, Od, and enamel organ cells, including Am, stratum intermedium (SI) and papillary layer cells (PL), was similar in wt and Trpm71kinase/<sup>+</sup> mice (**Figure 6**). However, ALPase activity present in wt mice (**Figures 7A,D,G,J**) was absent in the Trpm71kinase/<sup>+</sup> mouse, except the stratum intermedium cells (**Figures 7B,E,H,K**). Nevertheless, the ex vivo ALPase activity was largely rescued in sections of Trpm71kinase/<sup>+</sup> mineralizing tissue by the pre-incubation of the sections with MgSO<sup>4</sup> solution prior to staining for ALPase activity (**Figures 7C,F,I,L**).

# DISCUSSION

Tooth enamel is the hardest mineralized tissue in the vertebrate body comprised of 92 volume% inorganic mineral, which is higher than those in dentin and bones (61 and 64.5 volume% respectively; Kumar, 2014). Such a high mineralization rate of enamel is primarily achieved by maturation ameloblasts, which are able to efficiently remove the hydrolyzed enamel matrix proteins and deposit minerals into enamel matrix during the final phase of mineralization. We found that transient receptor potential cation channel, subfamily M, member 7 (TRPM7) was significantly upregulated in maturation ameloblasts. Characterization of craniofacial structures of mice Trpm71kinase/<sup>+</sup> mice, showed a severe hypomineralized phenotype in all mineralized tissues, indicating a universal role of TRPM7 in tissue mineralization.

Although, some unique molecular mechanisms are employed during enamel mineralization, there are many common factors that regulate the mineralization of enamel, dentin and bone. For instance, tissue-nonspecific alkaline phosphatase (ALPL) is known to be a key regulator of bone and tooth mineralization. Mice lacking Alpl show defects in bone, dentin and enamel mineralization and absence of acellular cementum (Waymire et al., 1995; Narisawa et al., 2001; McKee et al., 2011). In humans, mutations in ALPL genes cause hypophosphatasia, a rare inherited disorder characterized by deficiency of serum and bone alkaline phosphatase activity, resulting in the defective bone and tooth mineralization (Mornet, 2007). Physiological functions of ALPL are not fully understood yet, nevertheless, in bone and probably dentin, where the mineralization occurs in the type 1 collagen dominant matrix, ALPL is shown to initiate and direct mineralization by removing pyrophosphate (PPi), a mineralization inhibitor, (Addison et al., 2007) and antagonizing generation of PPi (Johnson et al., 2000; Hessle et al., 2002). Though ALPL has been shown to be present in secretory and maturation stage of ameloblasts (Bevelander

incisor root analog dentin near the incisal end in Trpm71kinase/<sup>+</sup> mice (I–L). Scale bars: 25 <sup>µ</sup><sup>m</sup>

and Johnson, 1949; Gomez and Boyde, 1994), the significance of ALPL in enamel formation has not been understood yet.

Divalent metal ions, including Mg2+, Zn2+, and Ca2+, are essential for ALPase activity (Stec et al., 2000; Hoylaerts et al., 2015). Magnesium deficient rats showed significantly reduced plasma ALPase activity, which was partially restored by an in vitro magnesium supplementation (Heaton, 1965). TRPM family, which consists of eight members, plays an essential role for magnesium entering the cells (Ryazanova et al., 2014). In analysis of our previously published microarray data (Zhang et al., 2014; Liu et al., 2015), we identified that TRPM7 was the only gene, among the well-known magnesium transporters (such as TRPM6, MAGT1, MRS2, PPM1G, MMGT1, and NIPAL1), which was significantly upregulated in ameloblasts as compared to other epithelial cells.

Trpm71kinase/<sup>+</sup> mice are known to have significantly lower plasma, bone and urine magnesium levels as compared with wt mice, and also show behavioral defects (clasping, tremors, and seizures) and other hypomagnesemia-like phenotype (Ryazanova

FIGURE 6 | Trpm7 gene deficiency does not affect the abundance of ALPL protein. In both wt and Trpm71kinase/<sup>+</sup> mice, ALPL was immunolocalized on the basal and lateral plasma membrane of osteoblasts (Os) (A,B) and odontoblasts (Od) (C,D) at the similar levels. In enamel organs of both wt and Trpm71kinase/<sup>+</sup> mice, ALPL was immunolocalized on plasma membrane of stratum intermedium cells (SI) and the cytoplasm of ameloblasts (Am) at secretory stage (E,F), and on the plasma membrane and cytoplasm of papillary layer cells (PL) and maturation ameloblasts (G,H). There was no notable difference in immunoreactive activities between wt and Trpm71kinase/<sup>+</sup> mice. Scale bars: 25 <sup>µ</sup><sup>m</sup>

et al., 2010). Interestingly, a seizure is also one of the behavioral phenotypes displayed by Alpl KO mice, and has been linked to a lack of vitamin B6 metabolized by ALPL (Waymire et al., 1995; Whyte et al., 1995; Narisawa et al., 1997; Mackey et al., 2006). Magnesium deficiency is shown to impair vitamin B6 status by inhibiting plasma ALPase activity in rat (Planells et al., 1997), suggesting that the activity of tissue non-specific ALPase (ALPL) is repressed under hypomagnesaemia condition.

Our microCT and morphological analyses on the hemimandibles of Trpm71kinase/<sup>+</sup> mice revealed similarity in skeletal and dental phenotypes, including partially mineralized molar crowns and absence of acellular cementum, between Alpl KO mice (Beertsen et al., 1999; McKee et al., 2011) and Trpm71kinase/<sup>+</sup> mice. These observations suggested a possibility that ALPL in the Trpm71kinase/<sup>+</sup> mice might be affected. In support of this possibility, we found that although Alpl protein content was similar in wt and Trpm71kinase/<sup>+</sup> mouse ameloblasts, odontoblasts and osteoblasts, ex vivo ALPase activity was dramatically reduced in Trpm71kinase/<sup>+</sup> mice. Restoration of ex vivo ALPase activity by Mg2<sup>+</sup> preincubation of sections prior to staining for ALP activity, indicates that magnesium deficiency is a major cause for the deficient ALPase activity of Trpm71kinase/<sup>+</sup> mice. It is worth noting that there are also numerous enzymes that require Mg2<sup>+</sup> for their activities. Therefore, hypomineralized tooth and bone in the Trpm71kinase/<sup>+</sup> mice could be also due to the cumulative deficiency of some other enzymes as well.

Nevertheless, in this study, the ex vivo supplementation of magnesium could not completely restore the ALPase activity to the levels displayed in the wt cells. This phenomenon is similar to what is found in ex vivo ALPase activity in tissue sections from Mg2<sup>+</sup> deficient rats (Heaton, 1965). The TRPM7 channel is also permeable to Ca2+, and therefore the lack of TRPM7 could also reduce the amount of available intracellular calcium. Optimized Ca2<sup>+</sup> is critical for ALPase activity (Hoylaerts et al., 2015), and therefore, although Mg2<sup>+</sup> concentrations are restored, reduced calcium concentrations could impact ALPase activity. In addition, magnesium deficiency is known to have a secondary effect on the metabolism of Ca2+, K<sup>+</sup> and inorganic phosphate (Konrad et al., 2004; Rude and Shils, 2006), therefore, the hypomineralization of Trpm71kinase/<sup>+</sup> mice was also attributed to an altered intra and/or extra cellular Ca2<sup>+</sup> and inorganic phosphate.

In this study, we found that TRPM7 was highly expressed by ameloblasts, odontoblasts and osteoblasts, which are cells responsible for the formation of enamel, dentin and bone respectively. All the mineralized tissues in the TRPM7 kinase domain deficient mice are hypomineralized, with reduced ALPase activity. Unlike bone and dentin, where ALPL functions in the initial phase of mineralization, in enamel formation, it is thought that ALPL functions at the maturation stage, where the final but majority of mineralization takes place (Takano and Ozawa, 1980). Our finding that ubiquitous inhibition of mineralization in all stage of enamel formation suggested that the initiation of enamel matrix mineralization at the secretory stage is also regulated by ALPL. In support of this, our immunohistochemistry showed that ALPL was apparently present in the cytoplasm of secretory ameloblasts. Therefore, similar to bone and dentin, ALPL also possibly contributes to the initiation of enamel matrix mineralization at the secretory stage by fine-tuning the local concentration of PPi.

Of course, there are also numerous enzymes that require Mg2<sup>+</sup> for their activities. Therefore, hypomineralized tooth and bone in the Trpm71kinase/<sup>+</sup> mice could be also due to the cumulative deficiency of some other enzymes as well.

In conclusion, our findings suggest that TRPM7 plays a critical role in the mineralization of enamel as well as dentin and

(Continued)

### FIGURE 7 | Continued

activity on osteoblasts. (D) ALPase activity was detected at the basolateral surface of odontoblasts (Od) in wt mice. (E) While in Trpm71kinase/<sup>+</sup> mice, no ALPase activity was seen in the odontoblasts. (F) Mg2<sup>+</sup> pre-treatment retrieved ALPase activity on odontoblasts. (J) ALPase activity was detected in stratum intermedium (SI), secretory ameloblasts (Am), and (G) papillary layer cells (P) at maturation stage in wt mice. In Trpm71kinase/<sup>+</sup> mice, ALPase activities were restricted only on the stratum intermedium at secretory stage, not in ameloblasts (H,K). Mg2<sup>+</sup> pre-treatment retrieved the ALPase activity on ameloblasts and papillary layer cells (I,L). Dn, dentin; pre-Dn, pre-dentin; En, enamel; Scale bars: 25 µm.

craniofacial bone via regulating activity of ALPL by transporting Mg2+, which is necessary for ALPL activity, into the cells.

# AUTHOR CONTRIBUTIONS

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

# REFERENCES


# ACKNOWLEDGMENTS

This research was funded by NIDCR grant R03 DE019507-02 to YZ and funds from the Department of Orofacial Sciences at UCSF to YZ, YN, NIDCR grant R56 DE13508 to PD. The authors thank the Biomaterials and Bioengineering Micro-CT Imaging Facility (http://bbct.ucsf.edu) at UCSF for the use of MicroXCT-200.


**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 © 2016 Nakano, Le, Abduweli, Ho, Ryazanova, Hu, Ryazanov, Den Besten and Zhang. 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) or licensor 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.

# Site-Specific Variations in Bone Mineral Density under Systemic Conditions Inducing Osteoporosis in Minipigs

Matthias C. Schulz <sup>1</sup> \*, Jan Kowald2, 3, Sven Estenfelder 2, 4, Roland Jung<sup>5</sup> , Eberhard Kuhlisch<sup>6</sup> , Uwe Eckelt <sup>1</sup> , Ronald Mai <sup>1</sup> , Lorenz C. Hofbauer <sup>7</sup> , Christian Stroszczynski <sup>8</sup> and Bernd Stadlinger <sup>9</sup>

### Edited by:

Giovanna Orsini, Università Politecnica delle Marche, Italy

### Reviewed by:

Nenad Filipovic, University of Kragujevac, Serbia Ariane Berdal, UMRS 1138 Institut National de la Santé et de la Recherche Médicale (INSERM), Université Paris Diderot, France

### \*Correspondence:

Matthias C. Schulz Matthias.Schulz@ uniklinikum-dresden.de

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

> Received: 18 March 2017 Accepted: 01 June 2017 Published: 20 June 2017

### Citation:

Schulz MC, Kowald J, Estenfelder S, Jung R, Kuhlisch E, Eckelt U, Mai R, Hofbauer LC, Stroszczynski C and Stadlinger B (2017) Site-Specific Variations in Bone Mineral Density under Systemic Conditions Inducing Osteoporosis in Minipigs. Front. Physiol. 8:426. doi: 10.3389/fphys.2017.00426 <sup>1</sup> Department of Oral and Maxillofacial Surgery, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>2</sup> Department of Radiology, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>3</sup> Division of Nephrology, Department of Internal Medicine III, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>4</sup> Department of Internal Medicine III, University of Ulm, Ulm, Germany, <sup>5</sup> Experimental Center, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>6</sup> Institute for Medical Informatics and Biometry, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>7</sup> Division of Endocrinology, Diabetes and Bone Diseases, Department of Medicine III, Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany, <sup>8</sup> Department of Radiology, University Hospital Regensburg, Regensburg, Germany, <sup>9</sup> Clinic of Cranio-Maxillofacial and Oral Surgery, University of Zurich, University Hospital Zurich, Zurich, Switzerland

Osteoporosis is a systemic bone disease with an increasing prevalence in the elderly population. There is conflicting opinion about whether osteoporosis affects the alveolar bone of the jaws and whether it poses a risk to the osseointegration of dental implants. The aim of the present study was to evaluate the effects of systemic glucocorticoid administration on the jaw bone density of minipigs. Thirty-seven adult female minipigs were randomly divided into two groups. Quantitative computed tomography (QCT) was used to assess bone mineral density BMD of the lumbar spine as well as the mandible and maxilla, and blood was drawn. One group of minipigs initially received 1.0 mg prednisolone per kg body weight daily for 2 months. The dose was tapered to 0.5 mg per kg body weight per day thereafter. The animals in the other group served as controls and received placebo. QCT and blood analysis were repeated after 6 and 9 months. BMD was compared between the two groups by measuring Hounsfield units, and serum levels of several bone metabolic markers were also assessed. A decrease in BMD was observed in the jaws from baseline to 9 months. This was more pronounced in the prednisolone group. Statistically significant differences were reached for the mandible (p < 0.001) and the maxilla (p < 0.001). The administration of glucocorticoids reduced the BMD in the jaws of minipigs. The described model shows promise in the evaluation of osseointegration of dental implants in bone that is compromised by osteoporosis.

Keywords: animal models, bone, bone mineralization, oral cavity, osteoporosis

# INTRODUCTION

Osteoporosis is a systemic disease that affects bone mass and architecture, increasing the risk of bone fractures (Rachner et al., 2011). Osteoporosis is prevalent in the United States, Europe, and Japan, causing over 4.5 million fractures annually (WHO Scientific Group on the Assessment of Osteoporosis at Primary Health Care Level, 2004). As consequence of an aging population, an increasing number of people suffer from osteoporosis. The most common form of osteoporosis is post-menopausal osteoporosis in females, due to reduced estrogen levels. Other risk factors for the development of osteoporosis include patient age and long-term systemic exposure to exogenous glucocorticoids. The clinical diagnosis of osteoporosis is usually confirmed radiographically, via dual-energy X-ray absorptiometry (DXA) based bone densitometry (Felsenberg and Gowin, 1999) and, more recently, with high resolution quantitative computed tomography (hrQCT; Snedeker et al., 2006; Graeff et al., 2013).

The influence of both local and systemic diseases on the osseointegration of dental implants has been a repeated point of discussion. For example, it has been shown that periodontitis in conjunction with smoking (Heitz-Mayfield and Huynh-Ba, 2009), and radiotherapy prior to implant placement (Chambrone et al., 2013), negatively affect implant integration. With regard to osteoporosis, there are contradictory reports on the success rates of dental implants placed in osteoporotic bone. Whereas, some studies describe no influence of osteoporosis on the implant osseointegration (Dvorak et al., 2011a), other authors report higher loss rates of implants in osteoporotic bone (Moy et al., 2005; Alsaadi et al., 2008). It is uncertain whether osteoporosis commonly induces areas of rarefaction in the alveolar processes of the jaws that are of clinical relevance to implant success. While some authors report that osteoporotic skeletal changes are reflected in alveolar bone changes (Jonasson et al., 2006), others have found no correlation between the condition of the general skeletal and alveolar bone structures (Kingsmill and Boyde, 1999).

Such possible correlations can be analyzed in animal models. In osteoporosis research, rodent models have been used for analysis of the integration of biomaterials (Stadlinger et al., 2013). In rodent studies, ovariectomy serves to induce an osteoporotic state. Large animal models more closely resemble human anatomy and human patterns of wound healing, allowing the application of regular-sized surgical instruments and dental implants. However, due to practicalities, it has been uncommon to employ systemically impaired large animal models for the analysis of dental implant integration in bone or soft tissue (Thoma et al., 2012). Sheep, minipigs, and dogs may serve for such studies. The establishment of an osteoporotic state in large animals is possible by various means. Usually osteoporosis is induced by ovariectomy, however this has been reported to have a limited influence on nulliparous pigs (Scholz-Ahrens et al., 1996). Other approaches involve a calcium-deficient diet, the systemic application of glucocorticoids, or a combination of those approaches. Recently, an animal model has been reported, showing mandibular bone loss following a hypothalamicpituitary disconnection (Oheim et al., 2014).

The application of glucocorticoids in minipigs has been shown to induce reduction of BMD in the lumbar spine as assessed by quantitative computed tomography (QCT; Scholz-Ahrens et al., 2007). QCT enables repeated measurements of the entire skeleton over time and is therefore suitable for large animal studies on bone structure.

In most pre-clinical studies, implant integration is analyzed in pristine bone under unimpaired osseous conditions. The aim of this minipig study was to establish a large animal model with glucocorticoid-induced osteoporosis, for the analysis of implant osseointegration under impaired systemic osseous conditions. To achieve this effect, the experimental animals received systemic glucocorticoid therapy for 9 months. During this period, the animals underwent three QCT assessments. The hypothesis was that glucocorticoid application for 9 months would induce a significant decrease of BMD in both the lumbar spine and the alveolar bone of the jaws.

# MATERIALS AND METHODS

# Experimental Design

The study protocol was approved by the Commission for Animal Studies at the District Government, Dresden, Germany (file reference: 24-9168.11-1/2008-43). The study was conducted in accordance to the recommendations and guidelines of the Commission for Animal Studies at the District Government, Dresden, Germany.

For the study, 37 adult female miniature pigs (nulliparous, aged 18–20 months, average weight 72.5 ± 16.7 kg) of the breed Mini-Lewe were used. The animals were randomly divided into two groups: Group I (29 animals) received daily oral prednisolone, 1 mg per kg body weight in the initial phase. After 2 months, the dosage was reduced to 0.5 mg prednisolone per kg body weight per day. Prednisolone administration was performed according to the protocol described by Scholz-Ahrens (Scholz-Ahrens et al., 2007). Group II (eight animals), which received neither prednisolone nor a carrier, served as control. Before the commencement of prednisolone, and at the 6- and 9-month time-points of prednisolone administration, QCT was performed for each animal, and blood samples for laboratory assessment and measurement of biochemical markers of bone turnover were also taken.

# Nutrition

Prior to the start of the study, the animals were fed a diet consisting of pollard, wheat bran, and corn silage obtained from a local food producer. The pollard contained 0.57% of calcium and 0.55% of phosphorus. The content of vitamin D3 was 328 international units (IU) per kg. After initiation of the glucocorticoid therapy, the experimental animals were fed a calcium-deficient diet omitting the pollard. The diet contained 58.3% wheat bran and 41.7% corn-based silage. The wheat bran contained 0.07% calcium and 0.05% phosphorus. The content of vitamin D3 was <300 IU per kg. The diet of the control group remained unchanged. Water was available ad libitum. The components of the animal nutrition were analyzed prior to the study (LUFA-ITL GmbH, Kiel, Germany).

# Sedation

Computed tomography investigations and the collection of the blood samples were performed under sedation, under the surveillance of a veterinarian. The medication for sedation involved a mixture of 1 mg/kg body weight midazolam (Ratiopharm GmbH, Ulm, Germany), 10 mg/kg body weight ketamine (Riemser Arzneimittel AG, Greifswald, Germany), and 0.05 mg/kg body weight atropine (Eifelfango Chem.-Pharm. Werke, Bad Neuenahr-Ahrweiler, Germany), administered intramuscularly.

# Quantitative Computed Tomography

The computed tomography scanner was a multislice-CT (SOMATOM 16, Siemens Healthcare, Erlangen, Germany). The minipigs were positioned ventrally due to veterinary aspects of respiration, with mandible and maxilla parallel to the horizontal plane. The images were acquired by applying the spiral technique with a collimation of 16 × 0.75 mm. The data acquisition was performed axial using a slice-thickness of 0.75 mm. The field of view was chosen 265 × 265 mm for the jaw scan and 200 × 200 mm for the scan of the lumbar vertebrae with a matrix of 512 × 512. Thus, a pixel spacing of 0.426 × 0.426 mm for the jaw scan and of 0.445 × 0.445 mm resulted. The optimal set-up conditions for the jaw scan were a tube voltage of 120 kV and 36 mAs. For the vertebra region the same tube voltage was used with 274 mAs.

The data was acquired by a filtered back projection with convolution kernel H60s for the jaw scan and B70s for the vertebra scan. Saving was performed as digital imaging and communication in medicine (DICOM) standard using the Leonardo workstation (LEONARDO VD10B; Syngo VX49B; Siemens Healthcare Diagnostics GmbH, Eschborn, Germany). The analyses and measurements of the CT scans were performed with interactive data language software (IDL 7.0.0, ITT-Visual Information Solutions, Boulder/CO, USA) by two experienced examiners (JK, SE). To calculate the inter- and intra-observer reproducibility, measurements of eight randomly selected animals were repeated twice by each examiner with a time interval of 2 weeks. Calibration was performed using a reference phantom containing a bone-like and a water-like phase (Osteo Phantom, Siemens Healthcare, Erlangen, Germany) placed parallelly to the horizontal plane. A gray scale of Hounsfield units (HU) served for the analysis of the bone mineral density. A threshold was defined in order to ensure that no soft tissue (e.g., fat) or dental structures (e.g., roots) were included in the measurements. In preliminary tests, a mean density of the mandibular and maxillary cortical bone of 1386 HU was assessed (data not shown). Furthermore, the gray values of the soft tissues in the bone marrow were measured to have a peak value below 0 HU. According to the resulting histograms, the thresholds at 0 HU for the minimum and at 1,350 HU for the maximum were defined. As mineralized dental structures show comparable or higher density values than cortical bone they could be precluded from the measurements by the threshold, likewise.

For the radiographic analysis of the jaw region, the radiographs were positioned parallelly to the transversal plane. The vertical position of the transversal cutting plane was determined by the root tips of the three premolar teeth in each jaw. The regions of interest in the mandible and maxilla were defined as follows: The oral and buccal borders were defined by the inner layer of the cortical bone. The distal edge of the canine root served as the anterior border and the mesial root of the third molar as the distal border. The matching of the radiographs of the baseline QCT and of the QCTs after 6 and 9 months was achieved by using constant anatomical structures e.g., pterygoid process, the bony nasal septum and the tuber maxilla in the upper jaw. In the mandible, the profile of the oral and vestibular cortical bone was used for the matching of the images. For the measurement of the BMD in lumbar vertebrae I–III, the radiographs were positioned parallel to the axial plane. The region of interest was defined through the ventral cortical bone and the cortical bone of the spinal canal. The lateral border was defined by a dense trabecula connecting the ventral cortical bone and the spinal canal in dorsomedial direction. The regions of interest are depicted in **Figures 1**, **2** for the jaw region and in **Figure 3** for the vertebral region. For the measurement of the bone density, the gray values of the pixels in the regions of interest were assessed and displayed in Hounsfield Units using the IDLsoftware. The mean value and standard deviation of each region of interest was calculated. In the upper and lower jaw, the left and right side were measured separately, and were later pooled for the statistical evaluation.

FIGURE 1 | The region of interest for the analysis of the computed tomography scans is depicted in the maxilla. The area is extended between the distal edge of the canine root and the mesial edge of the third molar. Laterally, the cortical bone served as border.

FIGURE 2 | The region of interest for the analysis of the computed tomography scans is depicted in the mandible. The area is extended between the distal edge of the canine root and the mesial edge of the third molar. Laterally, the cortical bone served as border.

FIGURE 3 | In the vertebrae, the region of interest was defined as a polygon being limited by the ventral and dorsal cortical bone and dorso-laterally by a dense trabecula between the ventral cortical bone and the spinal canal.

# Serum Parameters

Blood samples were taken before the commencement of glucocorticoid administration and after 6 and 9 months, respectively. The samples were always obtained from the ear vein at the same time of the day using serum gel monovettes (Sarstedt AG & Co., Nürmbrecht, Germany) and stored at 4◦C until centrifugation. Subsequently, the following parameters were analyzed using an automatic test system (cobas c, Roche Diagnostics, Mannheim, Germany). The serum level of ionic calcium and ionic phosphorus were analyzed by photometric test kits (CA2 and PHOS2). The level of total alkaline phosphatase was measured by an extinction test kit (ALP2). For the analysis of the levels of beta crosslaps an immunoassay test kit (ECLIA) was used. All test kits were obtained from Roche Diagnostics, Mannheim, Germany. Bone specific alkaline phosphatase was measured using an immunoassay (MicroVueTM BAP, Quidel Corporation, San Diego, CA, USA). The serum level of 25(OH) Vitamin D was analyzed by a chemiluminescence assay (LIAISON <sup>R</sup> 25 OH Vitamin D TOTAL Assay). The analysis of the osteocalcin level was performed using a twosite immunoradiometric assay (CA 72-4 IRMA). Both assays were obtained from DiaSorin Inc., Stillwater, MN, USA. Osteoprotegrin (OPG) levels were analyzed by an enzymelinked immune sorbent assay (Osteoprotegrin ELISA kit, Immundiagnostik AG, Bensheim, Germany). The serum level of 1,25(OH)2-Vitamin D was measured using a radioimmunoassay (1,25-Dihydroxy Vitamin D RIA, Immunodiagnostic Systems GmbH, Frankfurt am Main, Germany). All analyses were performed by the Institute for Clinical Chemistry and Laboratory Medicine (Medical Faculty "Carl Gustav Carus," Technische Universität Dresden, Dresden, Germany) using standardized laboratory methods.

# Statistical Analysis

The degree of agreement between the two observers and the degree of agreement between the two measurements were assessed by a reliability analysis. The inter- and intra-observer reliability is described by the intraclass correlation coefficient (ICC) taking a two-way random model with absolute agreement definition. The animal experiment was planned and carried out as longitudinal study. Consequentially the analyses were done by ANOVA using linear mixed models. The models included fixed effects associated with treatment, time, and the interaction between treatment and time. The residual covariance matrix was specified as a heterogeneous compound symmetry (CSH), so that the variances differ across the levels of time. Estimated means and their 95%-confidence limits are given. Post-hoc tests and confidence limits are alphaadjusted using the Tukey–Kramer method. The statistical analysis was performed using SPSS 19 (SPSS Inc., Chicago, Illinois, USA) and SAS 9.3 (SAS Institute Inc., Cary, NC, USA).

# RESULTS

From the initially 37 minipigs, two animals of the prednisolone group died due to unknown cause and were excluded from the analysis. Thus, 35 animals completed the study period and could be included in the evaluation.

# Quantitative Computed Tomography Intra- and Inter-Observer Control

The intra-observer analysis showed an ICC of 0.992 (95% confidence interval: 0.987–0.995) for observer I and an ICC of 0.991 (95%-confidence interval: 0.987–0.994) for observer II. The inter-observer analysis showed an ICC of 0.992 (95%-confidence interval: 0.988–0.995) for the first time point and an ICC of 0.990 (95%-confidence interval: 0.984–0.993) for the second time point. A statistically significant difference between both observers could not be found, neither for the first (p = 0.592) nor for the second time point (p = 0.224).

### Control Group—Development of Bone Mass with Aging

With regard to BMD-values, the control group showed a general decrease from baseline to the 6-month examination. This decrease was statistically significant for all regions of interest. From 6 to 9 months, a statistically significant increase of BMD was obvious for all examined regions. When comparing the baseline values to those obtained after 9 months, an overall decrease of BMD was found. However, this decrease was only statistically significant for the maxilla. For vertebrae and mandible no significant changes could be found (vertebrae: p = 0.988; mandible: p = 0.995).

### Effects of Glucocorticoid Exposure on Bone Mass

At the baseline point of the study, the values for BMD in the lumbar vertebrae and the maxilla differed statistically significantly between control and treatment group. For the mandible, no statistically significant difference could be observed (p = 0.973). In the computed tomography results after 6 months, the BMD showed a decrease in all evaluated regions. This decrease was statistically significant for both the control and the glucocorticoid group. Between the groups, no statistically significant differences could be found for any region. After 9 months, an increase of BMD in the control group was obvious for all evaluated regions compared to the 6-month values. This increase was statistically significant for all regions. However, the values after 9 months did not reach the baseline level again. Compared to baseline values, the decrease was statistically significant for the maxilla (p < 0.001). For the lumbar vertebrae and the mandible, the differences were not statistically significant (lumbar vertebrae: p = 0.981; mandible: p = 0.559). An example for the rarefied trabecular structure in the jaw bone is given in **Figures 4A,B** for the maxilla and in **Figures 5A,B** for the mandible.

After 9 months, there was an increase in the bone density values in the lumbar vertebrae in the glucocorticoid group compared to the 6-month value. This was statistically significant. In the jaw region, a slight decrease of BMD was observed. However, this decrease did not reach statistical significance when compared to the values after 6 months (maxilla: p = 0.551; mandible: p = 0.559). Compared to the baseline, the values after 9 months of glucocorticoid administration were lower. For the lumbar vertebrae region, this difference was not statistically significant (p = 0.981). In the jaw region, the difference reached statistical significance. When compared to the control group the values for BMD differed statistically significant for all examined regions in the glucocorticoid group. The detailed chronological data and statistical significant differences are shown in **Table 1**.

# Serum Parameters

### Control Group

In the control group, the level of ionic phosphorus decreased slightly during the period of study while values of ionic calcium, alkaline phosphatase, osteocalcin, and 1,25(OH)2-vitamin D varied. The values of bone specific alkaline phosphatase and beta crosslaps increased. During the study period, an increase could be observed for OPG serum levels. The means and their standard deviations are shown in **Table 2**. Statistical significant differences are depicted in **Figure 6**.

### Effects of Glucocorticoid Exposure on Serum Parameters

The serum values for ionic phosphorus showed a decrease over the study period. Ionic calcium values varied over the study period as in the control group. From the baseline to

FIGURE 4 | Trabecular rarefication over time shown in the maxilla. Compared to the baseline (A) the spongious area in the region of interest appears in darker gray values after 9 months (B) indicating a loss of mineralization.

FIGURE 5 | Trabecular rarefication over time shown in the mandible. Compared to the baseline (A) the spongious area in the region of interest appears in darker gray values after 9 months (B) indicating a loss of mineralization.

### TABLE 1 | Values of bone mineral density in Hounsfield Unit (HU).


Means and standard deviations (SD) are depicted. The statistical significant differences of the pairwise comparisons are marked (\*, \*\*, +, \$, 2, 3, 6, 8, 9, 11, 12: p < 0.001;", #, <sup>∧</sup>, 1, 4, 5, 7, 10: p < 0.05).

### TABLE 2 | The development of serum parameters over the study period.


Means, their 95% confidence intervals (CI) and standard deviations (SD) are depicted.

the measurement after 6 months, alkaline phosphatase and bone specific alkaline phosphatase decreased. After initiating the administration of glucocorticoids, the values for osteocalcin decreased after 6 months compared to the baseline. The serum levels for beta-crosslaps, OPG, and 1,25(OH)2-vitamin D varied. The detailed data is depicted in **Table 2** and statistical significant differences are depicted in **Figure 6**.

### Inter-Group Comparison

At baseline, the values for all serum parameters differed between the groups. A decrease could be observed for ionic phosphorus in both groups, but was more pronounced in the glucocorticoid group. The decrease differed statistically significant for both time points (p = 0.008; p = 0.009). When comparing bonespecific alkaline phosphatase, osteocalcin, and 25(OH)-vitamin D, the development in both groups differed significantly (alkaline phosphatase: p = 0.013; p = 0.017; osteocalcin: p < 0.001; 25(OH)-vitamin D: p < 0.001). No statistically significant differences were found for the trends of the levels of alkaline phosphatase (p = 0.076; p = 0.111), OPG (p = 0.266; p = 0.129). For 1,25(OH)2-vitamin D, the development initially differed between the groups (p = 0.006) but not for the entire study period (p = 0.443).

# DISCUSSION

The osseointegration of dental implants is commonly analyzed in animal models prior to clinical application in humans. In treatment clinics however, the situation is often different to ideal pre-clinical laboratory conditions. Many patients suffer

from local bone defects due to bone resorption following tooth extraction or as a consequence of local trauma and require bone augmentation prior to implant therapy. Further, an increasing number of patients suffer from systemic diseases such as diabetes mellitus or osteoporosis, and are receiving medical therapy. These factors might interfere with bone healing, osseointegration of implants, and the health of oral soft tissues (Sachelarie et al., 2016).

# Bone Mineral Density

In the present study, BMD was measured using QCT, which is frequently used in animal studies (Scholz-Ahrens et al., 2007; Veigel et al., 2011). The measurements are focused on the trabecular bone mineral density, as glucocorticoid administration seems to have an important effect (Wetzsteon et al., 2009; Paggiosi et al., 2015). Furthermore, the gray values of the cortical bone and the dental structures are in the same range. Thus, cortical bone has usually been excluded from the measurements to avoid a biased analysis through inclusion of dental structures (roots). Scholz-Ahrens et al. found a statistically significant decrease in BMD in lumbar vertebrae after 8 months of glucocorticoid treatment (Scholz-Ahrens et al., 2007). This partially agrees with the results of this current study, which showed a statistically significant decrease in lumbar vertebrae after 6 months with a slight increase of BMD after 9 months. This might be related to an increase of body weight which is described for minipigs after completed adolescence (Richel and Waldmann, 2015). The growing weight might be a stimulus for bone formation in the vertebral region due to an increase of stress. An unexpected finding in the current study were the statistically significant differences considering the baseline values of bone mineral density. One possible reason for the different baseline data might be inter-animal variance. All came from the same herd, however, showed a variance in age and weight. Another reason might be the different group size (8 vs. 27 animals completing the study). If the control group would have had an equal group size to the glucocorticoid group, the differences might have been lower due to the compensation of extremal values.

The results for maxillofacial BMD in animals undergoing the induction of osteoporosis are different to those of lumbar bone. In a study using ovariectomized sheep, a pronounced horizontal and vertical bone loss in the mandible was described after 3 and 12 months (Johnson et al., 1997). However, a metric measurement of bone volume was not performed. Using the same animal model, another study measured BMD 12 months after ovariectomy, and showed no changes in the region of the diastema or the mandibular ramus (Johnson et al., 2002). Bone density in the alveolar regions however was significantly decreased, by 27.8%. This is in accordance with our results, which showed a progressive rarefication of the trabecular structure of the maxilla and the mandible. This rarefication of trabecular bone was found to be more pronounced in mandibular bone. In the experimental group, both a progressive decrease over time and a decrease in comparison to the control group were recorded.

# Animal Models

The consideration of bone quality raises the question of appropriate animal models to simulate human osteoporosis. Osteoporotic large animal models have been established in sheep, dog, and pigs (Volozhin et al., 1990; Scholz-Ahrens et al., 2007; Dvorak et al., 2011b; Zhang et al., 2014; Kielbowicz et al., 2015). The most frequently described large animal model seems to be the sheep (Dvorak et al., 2011b; Veigel et al., 2011; Kielbowicz et al., 2015). With regard to the oral and maxillofacial situation, the minipig seems to be the most suitable model. First, it is an omnivore, so the diet has more similarities to a human diet than do sheep and dog models. Furthermore, the bone and oral tissues structure as well as bone turnover is quite similar to humans (Hönig and Merten, 1993; Hönig et al., 1997; Wang et al., 2007; Swindle et al., 2012). In a review, Pearce et al. concluded that both bone composition and bone remodeling in pigs are largely similar to that of humans, while micro- and macrostructure can be considered as moderately similar (Pearce et al., 2007). Regarding the age, the minipigs included in the current study can be considered as young adults. The life span of the breed Mini-Lewe was described with 10–15 years (Leucht et al., 1982). Richel observed in a longitudinal study of 144 Mini-Lewe pigs no further increase in the withers height and crown rump length after an average age of 20 months. Day 600 post-partum was defined as the time point of completed adolescence (Richel and Waldmann, 2015). This is in accordance with the findings of the current study, as no bone growth could be determined when comparing the matched images of the CT scans during the study period. Unlike in humans, in non-primate mammals there is no description of a physiological menopause state in the literature (Finn, 2001). Therefore, the use of young adult minipigs seems to be appropriate as there is no change in estrous cycle expected with increasing age. Additionally, when implant osseointegration is evaluated, the anatomy of the jaw of minipigs allows the application of the same instruments as used for human patients (Stadlinger et al., 2012).

# Serum Parameters

Any systemic animal model needs to be validated by systemic parameters. Along with the measurement of bone density, the analysis of serum parameters is of particular interest, providing better understanding of bone metabolism.

### Inorganic Phosphorus

The values of the inorganic phosphorus decreased in both groups which was more pronounced in the glucocorticoid group. When comparing the current values to the literature the values of the control group were in the same range (2.38 ± 0.86 to 3.10 ± 0.65 mmol<sup>∗</sup> l −1 ; Mieth, 1978; Leucht et al., 1982; Gusewski, 1983; Richel and Waldmann, 2015). Increasing age was discussed as a possible reason for this decrease by Gregor et al. but could not be confirmed in other evaluations (Gregor, 1979; Richel and Waldmann, 2015). In a recent study, inorganic phosphorus values between 1.55 and 2.24 mmol<sup>∗</sup> l <sup>−</sup><sup>1</sup> were measured in female Aachen minipigs, a breed descending from the Mini-Lewe strain (Pawlowsky et al., 2017). However, these differences might be due to a different state of growths as the used animals were aged 6 months. When comparing both groups during the study, lower phosphorus levels were obvious in the glucocorticoid group. This state might be considered as hypophosphatemia which possibly is caused by the reduced intake of vitamin D3 of the glucocorticoid group. Furthermore, hypophosphatemia could be corresponding with lower levels of 1,25(OH)2-vitamin D3 in the glucocorticoid group what could be observed in the present study. One possible reason for the lower phosphorus values might be an interaction of glucocorticoids and the phosphate balance (Levi et al., 1995).

### Calcium

The values of serum calcium varied over the study period in both groups ranging from 1.368 to 2.396 mmol<sup>∗</sup> l −1 . Values <2.1 mmol<sup>∗</sup> l −1 in humans are considered as hypocalcemia (Fong and Khan, 2012). In the literature, values between 3.1 and 2.5 mmol<sup>∗</sup> l −1 are stated (Mieth, 1978; Leucht et al., 1982; Gusewski, 1983; Richel and Waldmann, 2015). Symptoms of acute decreased serum calcium levels include muscle spasms, heart failure, and neuromuscular irritability whereas a slow decrease of calcium levels might remain without symptoms (Fong and Khan, 2012). As no of the animals in the present study developed comprehensibly any of these symptoms it is unlikely that there was an apparent hypocalcemia.

### Alkaline Phosphatase and Bone-Specific Alkaline Phosphatase

The values of total alkaline phosphatase and bone-specific alkaline phosphatase showed a significant decrease after glucocorticoid administration for 6 months. These findings were consistent with those of Ikeda et al., who observed a reduction of bone-specific alkaline phosphatase and osteocalcin values in serum of adolescent miniature pigs under glucocorticoid treatment (Ikeda et al., 2003). This might be explained by decreased bone formation due to supressed osteoblast activity (Lukert et al., 1986; Prummel et al., 1991). Clinically, a statistically significant correlation between bone alkaline phosphatase as well as osteocalcin levels and the age was found in postmenopausal women (Lumachi et al., 2009). However, a correlation between these bone markers and the BMD was not observed.

### β-Crosslaps

The C-terminal telopeptide of the type I-collagen is termed as β-Crosslaps. It is considered as bone resorption marker (Okabe et al., 2004). In a clinical study, the changes of β-Crosslaps in serum correlated with the annual changes in BMD in postmenopausal women (Christgau et al., 1998). They are of crucial value to monitor the effect of antiresorptive therapies (Christgau et al., 1998). Furthermore, an increased level of β-Crosslaps and OPG was observed in patients suffering from Wilsons disease and osteoporosis (Hegedus et al., 2002). No such results could be observed in the present study. A reason might be a wide interindividual variety of the serum levels. However, there is little literature describing the levels of β-Crosslaps in pigs.

### Osteocalcin

The osteocalcin serum levels in the glucocorticoid group decreased significantly after 6 months of prednisolone administrations. This was not observed in the control group. In human studies, osteocalcin was described as a sensitive marker for osteoblastic depression under glucocorticoid therapy (Lukert et al., 1986; Ekenstam et al., 1988). The effects were pronounced when doses of glucocorticoids between 5 and 30 mg were administered daily over a period of at least 3 months (Lukert et al., 1986). In our study, a daily average dosage correlating to 35–45 mg prednisolone in human subjects (based on body weight) over a period of more than 4 months was assumed. This implies a comparable effect on the osteocalcin level (Scholz-Ahrens et al., 2007). However, the baseline values of both groups differed significantly, limiting this comparison. One possible reason for low values of osteocalcin might have been the degradation of osteocalcin by proteases of erythrocytes and leukocytes. In cases of difficult blood drawing degradation might have started already before further processing of the blood samples.

# OPG

OPG is a glycoprotein that regulates bone density (Yano et al., 1999). A significant correlation was found between OPG serum level and age (Yano et al., 1999; Fahrleitner-Pammer et al., 2003). There was no such finding in the present study. Possibly, the age range of the minipigs was too small for the observation of such an effect. In contrast to our study, where young adult animals were examined the mentioned study evaluated patients with an age ranging from young adults to elderly patients. Furthermore, there is an increased level of OPG in patients suffering from osteoporosis (Yano et al., 1999). In the present study, slightly increased levels of OPG were observed in the glucocorticoid group during the study period. This, however, showed no significant statistical difference when compared to the control group. In a study of ovariectomized rats, a transient increase of OPG was found in the initial stage of osteoporosis, which might correlate with the increase of OPG observed in our study (Miyazaki et al., 2004).

# 1,25-Dihydroxyvitamin D and 25(OH)-Hydroxyvitamin D

The levels measured for 1,25-dihydroxyvitamin D were variable over the study period in both groups. This is in accordance to the literature, where increases, decreases, and an absence of alterations of the 1,25-dihydroxyvitamin D levels in individuals undergoing glucocorticoid therapy are described (Hodsman et al., 1991; Prummel et al., 1991; van der Veen and Bijlsma, 1992). However, in contrast to our study, higher glucocorticoid doses given intravenously cause a pronounced decrease (van der Veen and Bijlsma, 1992). To minimize extraneous influences, animals were housed in a stable and both groups received an equally vitamin D3 deprived (<300 IU/kg) daily diet. Comparable to other studies, no strong correlation could be found between the changes in bone metabolism and serum levels of 1,25-dihydroxyvitamin D (Seeman et al., 1980).

In summary, the serum parameters in the current study showed a wide range of values. In comparison to the literature, there are concordances but also some discrepancies. One potential reason is the fact, that different breeds of minipigs are being compared. Furthermore, in contrast to humans, there are no standard values for most parameters in minipigs. For some parameters, there were no reference values of pigs available at all. Thus, only a comparison to values obtained from other species was possible. Here, further studies are needed to elucidate potential correlations. Another possible reason might be the use of different analytical methods (Richel and Waldmann, 2015).

# Methods of Osteoporosis Induction

The induction of an osteoporotic state in large animal models has been described by means of ovariectomy, calcium-deficient diet, glucocorticoid administration, and combinations of these approaches (Scholz-Ahrens et al., 1996, 2007; Kielbowicz et al., 2015). In comparison to deficient nutrition and glucocorticoid application, ovariectomy carries the disadvantage of a surgical intervention, which is a cause of stress for the animals. Furthermore, the effect of ovariectomy on nulliparous animals has been reported to be limited (Scholz-Ahrens et al., 1996). This would apply to the animals in the current study. Additionally, ovariectomy creates a sudden change in the hormonal balance, which is not entirely comparable to the postmenopausal situation, where the hormone levels change more gradually. In distinction, the administration of glucocorticoids over a period of time might lead to a gradual induction of osteoporosis, more closely mimicking the natural disease condition. However, a longer period of glucocorticoid administration may be required to show measurable effects on bone density. One limitation of this study might be the unequal size of the animal groups. This might cause some bias in the statistical evaluation e.g., statistically significant differences in BMD at the baseline. The reason was to minimize the total number of animals used. This was similar to other studies, where unequal group sizes for treatment and control groups have also been used (Scholz-Ahrens et al., 2007; Kielbowicz et al., 2015).

Compared to intramuscular application of glucocorticoids one limitation of oral administration is the fact that the individual animal dose is not fully calculable, because of the uncertainty about complete ingestion and absorption. Furthermore, glucocorticoids have various metabolic effects not limited to the stimulation of bone resorption and induction of osteoblast apoptosis (Dalle et al., 2001; Spreafico et al., 2008). Regarding the suitability as a model for dental implant research, glucocorticoid administration might lead to delayed wound healing. Morin and Fardet described altered wound healing in a retrospective study in humans, when glucocorticoids were taken for more than 6 months (Morin and Fardet, 2015). Dental implant clinics are seeing an increasing number of patients with progressive osteoporosis secondary to the use of glucocorticoid medications for the treatment of rheumatoid arthritis or following organ transplantation. Thus, despite possible criticisms of the animal model, glucocorticoid-induced osteoporosis might be a fair representation of a common clinical situation.

# CONCLUSION

There are advantages and disadvantages of every animal model which involves artificial induction of a compromised metabolic state. However, the induction of an osteopenic state in the minipig by oral administration of glucocorticoids is predictable, and is minimally invasive when compared to surgical ovariectomy. In summary, the induction of osteoporosis in the minipig appears to be a promising model for the evaluation of the osseointegration of dental implants under systemically impaired osseous conditions.

# AUTHOR CONTRIBUTIONS

MS was involved in the data acquisition, analysis and interpretation, drafted the manuscript, approved the final version of the manuscript, and agreed to be accountable for integrity and accuracy of the work. JK, SE, and RJ were involved in the data acquisition, drafted the manuscript, approved the final version of the manuscript, and agreed to be accountable for integrity and accuracy of the work. EK and LH were involved in data analysis and interpretation, drafted the manuscript, approved the

# REFERENCES


final version of the manuscript, and agreed to be accountable for integrity and accuracy of the work. RM, UE, and CS were involved in study conception, drafted the manuscript, approved the final version of the manuscript, and agreed to be accountable for integrity and accuracy of the work. BS was involved in the conception of the study, data interpretation, drafted the manuscript, approved the final version of the manuscript, and agreed to be accountable for integrity and accuracy of the work.

# ACKNOWLEDGMENTS

The authors are grateful to Mrs. Sylvia Albrecht for the support in acquiring the CT data. Furthermore, the authors thank PD Dr. Volker Hietschold for supporting them with the IDL software. Mrs. Andrea Groß is acknowledged for her technical support. The authors thank Dr. Ian Chambers for the English editing.


phosphate transporter mRNA and protein abundance and glycosphingolipid composition. J. Clin. Invest. 96, 207–216. doi: 10.1172/JCI118022


1,25-dihydroxyvitamin D in health and in chronic glucocorticoid excess. J. Clin. Invest. 66, 664–669.


**Conflict of Interest Statement:** The study was supported by Dentsply Sirona Implants, Mannheim, Germany. LH was supported by the Transregio 67 subproject B2 of the German Research Foundation DFG (Deutsche Forschungsgemeinschaft).

The other 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 © 2017 Schulz, Kowald, Estenfelder, Jung, Kuhlisch, Eckelt, Mai, Hofbauer, Stroszczynski and Stadlinger. 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) or licensor 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.

# Stemodia maritima L. Extract Decreases Inflammation, Oxidative Stress, and Alveolar Bone Loss in an Experimental Periodontitis Rat Model

Alrieta H. Teixeira1, 2, Jordânia M. de Oliveira Freire<sup>1</sup> , Luzia H. T. de Sousa<sup>1</sup> , Antônia T. Parente<sup>3</sup> , Nayara A. de Sousa<sup>3</sup> , Angela M. C. Arriaga<sup>4</sup> , Francisca R. Lopes da Silva<sup>4</sup> , Iracema M. Melo<sup>2</sup> , Igor I. Castro da Silva<sup>2</sup> , Karuza M. A. Pereira<sup>5</sup> , Paula Goes <sup>6</sup> , José J. do Nascimento Costa<sup>7</sup> , Gerardo Cristino-Filho<sup>3</sup> , Vicente de Paulo T. Pinto<sup>3</sup> , Hellíada V. Chaves <sup>2</sup> and Mirna M. Bezerra1, 3 \*

### Edited by:

Claudio Cantù, University of Zurich, Switzerland

### Reviewed by:

Timothy C. Cox, University of Washington, United States Harald Osmundsen, University of Oslo, Norway

\*Correspondence:

Mirna M. Bezerra mirnabrayner@gmail.com

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 12 September 2017 Accepted: 17 November 2017 Published: 01 December 2017

### Citation:

Teixeira AH, Freire JMd, de Sousa LHT, Parente AT, de Sousa NA, Arriaga AMC, Lopes da Silva FR, Melo IM, Castro da Silva II, Pereira KMA, Goes P, Costa JJdN, Cristino-Filho G, Pinto VdPT, Chaves HV and Bezerra MM (2017) Stemodia maritima L. Extract Decreases Inflammation, Oxidative Stress, and Alveolar Bone Loss in an Experimental Periodontitis Rat Model. Front. Physiol. 8:988. doi: 10.3389/fphys.2017.00988 <sup>1</sup> RENORBIO, Federal University of Ceará, Fortaleza, Brazil, <sup>2</sup> Dentistry School, Federal University of Ceará, Sobral, Brazil, <sup>3</sup> Medical School, Federal University of Ceará, Sobral, Brazil, <sup>4</sup> Department of Organic and Inorganic Chemistry, Federal University of Ceará, Fortaleza, Brazil, <sup>5</sup> Department of Morphology, Medical School, Federal University of Ceará, Fortaleza, Brazil, <sup>6</sup> Department of Pathology and Legal Medicine, Medical School, Federal University of Ceará, Fortaleza, Brazil, <sup>7</sup> UNINTA University Center, Sobral, Brazil

Periodontitis is very prevalent worldwide and is one of the major causes of tooth loss in adults. About 80% of the worldwide population use medicinal plants for their health care. Stemodia maritima L. (S. maritima) antioxidant and antimicrobial effects in vitro as well as anti-inflammatory properties. Herein, the potential therapeutic effect of S. maritima was assessed in rats subjected to experimental periodontitis (EP). EP was induced in female Wistar rats by nylon thread ligature around 2nd upper left molars for 11 days. Animals received (per os) S. maritima (0.2; 1 or 5 mg/kg) or vehicle (saline + DMSO) 1 h before ligature and then once daily for 11 days. The naive group had no manipulation. After this time-point, the animals were terminally anesthetized, and the maxillae were removed for morphometric and histological analyzes (HE). Gingival tissues were dissected to cytokine levels detection (TNF-α, IL1-β, CINC-1, and IL-10), enzymes superoxide dismutase (SOD), and catalase (CAT) analysis, as well as gene expression (TNF-α, IL-1β, RANK, and iNOS) by qRT-PCR. Systemic parameters (weight variation, plasma levels of hepatic enzymes aspartate aminotransferase (AST) and alanine aminotransferase (ALT), creatinine, total alkaline phosphatase (TALP), and bone alkaline phosphatase (BALP) were performed. Histological analysis of the stomach, liver, kidney, and heart was also performed. S. maritima (5 mg/kg) decreased alveolar bone loss, TNF-α and CINC-1 gingival levels, oxidative stress, and transcription of TNF-α, IL1-β, RANK, and iNOS genes. It elevated both BALP activity and IL-10 gingival levels. The animals showed no any signs of toxicity. In conclusion, S. maritima reduced pro-inflammatory cytokine production, oxidative stress, and alveolar bone loss in a pre-clinical trial of periodontitis. S. maritima is a potential tool for controlling the development of periodontitis.

Keywords: Stemodia maritima L., periodontitis, inflammation, oxidative stress, bone loss

# INTRODUCTION

Current concepts define periodontitis as a chronic inflammatory disease that compromises the integrity of the tooth-supporting tissues. It is an imbalance between the polymicrobial biofilm and the immune-inflammatory response and includes genetic and environmental risk factors (Genco and Borgnakke, 2013; Hajishengallis, 2014). The periodontal tissue breakdown is mainly mediated by the release of pro inflammatory cytokines (TNF-α, IL-1, and IL-8/CXCL8 (in humans) or CINC-1/CXCL1 (in rats) as well as the production of reactive oxygen species (ROS) (Liao et al., 2014; Hienz et al., 2015; Silva et al., 2015; Gomes et al., 2016).

Mechanical therapy and surgical procedures have been used to treat periodontitis (Wang et al., 2014). Nevertheless, these procedures are not always satisfactory. Thus, adjunctive therapies may be necessary including antibiotics and non-steroidal antiinflammatory drugs. The major disadvantage of these agents is the development of bacterial resistance and gastric/renal toxicity. Thus, the search for newer and safer therapeutic agents continues to overcome these limitations. Phytochemicals isolated from plants are considered good alternatives to synthetic chemicals. In recent years, the use of plant extracts has gained popularity (Alviano and Alviano, 2009; Chandra Shekar et al., 2015; Kala et al., 2015; Ramesh et al., 2016) and some of these extracts have been used to treat periodontitis or repair bone defects (Guimarães et al., 2016; Lima et al., 2017; Oliveira et al., 2017).

In this regard, earlier chemical studies on Stemodia maritima L. (Plantaginaceae family, formely Scrophulariaceae), which is used in traditional medicine to treat inflammatory disorders, reported the isolation of flavonoids, diterpenes, and other compounds associated with larvicidal and antiviral activities (Rodrigues et al., 2010; Russell et al., 2011). Our group studied the phytochemicals in S. maritima L., and described the isolation of the natural compound, stemodinol along with seven known compounds (da Silva et al., 2014). Antioxidant and antimicrobial activities were observed on crenatoside, stemodinoside B, and stemodin. These are natural compounds also derived from S. maritima L. (da Silva et al., 2014).

Despite these data, there is no study in the literature focusing on inflammatory bone resorption using S. maritima L. extract or its bioactive phytochemicals. Thus, considering the involvement of both polymicrobial biofilm (Genco and Borgnakke, 2013; Hajishengallis, 2014) and oxidative stress (Tóthová et al., 2015; Lima et al., 2017) in periodontitis, and the potential therapeutic effect of S. maritima L., the goal of this study was to investigate the efficacy of S. maritima L. extract in a rat periodontitis model.

# MATERIALS AND METHODS

# Plant Material

S. maritima L. (S. maritima or Sm) is a common shrub that grows widely in the northeast region of Brazil, near the seacoasts. It belongs to Plantaginaceae family, tribe Gratioleae Benth (Albach et al., 2005). Dr. F. S. Cavalcanti and Prof. E. P. Nunes identified the plant and a voucher specimen (# 38483) was registered at Prisco Bezerra Herbarium, Federal University of Ceará, Fortaleza, Brazil. For this study, the extract was obtained from fresh leaves of Sm collected during the flowering stage along Fleixeiras Beach, Ceará, Brazil (da Silva et al., 2014).

# Animals

The experimental procedures and treatments performed on animals were approved by the Animal Research Ethics Committee of Federal University of Ceará (Permit number: 08/2013) in accordance with the guidelines from the Brazilian Society of Laboratory Animal Science - SBCAL/COBEA). Female Wistar rats, 10-weeks-old and weighting 200 ± 20 g were housed in appropriated cages of polypropylene and maintained on a 12 h-12 h light-dark cycles with a constant room temperature of 25◦C and received water and food ad libitum.

# Experimental Design

For the initial experiment, 30 animals were randomly divided into five specific groups (n = 6 in each group): unchallenged group (naive), three experimental periodontitis-challenged groups (EP) receiving oral gavage (per os) in different concentrations of S. maritima (0.2, 1, or 5 mg/kg) or 0.9% saline solution + DMSO (vehicle). Experimental Periodonttis (EP) was performed as previously described (Bezerra et al., 2002). Briefly, the animals were anesthetized with ketamine and xylazine (90:10 mg/kg, i.p.) and a sterilized nylon suture thread (3.0 Nylpoint, Ceará, Brazil) was placed around the second left maxillary molar. To facilitate suture placement, a guide through the medial and distal interproximal spaces was made using a 5.0 suture needle (Point Suture, Ceará, Brazil). Groups received vehicle and S. maritima (per os) 1 h before periodontitis induction and once daily, during 11 days. After this time, blood samples were obtained and an overdose of ketamine and xylazine (300:30 mg/kg; i.p.) were used to euthanize the animals. The maxillae were removed to analyze the total bone resorption area. Hystopathological analyses was performed with another animals following the same above designed groups (n = 30).

Sm at a dose of 5 mg/kg was found to be the most effective dose against alveolar bone loss (ABL), and therefore this dose was selected for the experiments. Subsequent series of experiments were conducted with the following groups: non-ligated (naive animals not subjected to EP); ligature only (vehicle), and ligature plus treatment with S. maritima 5 mg/kg diluted in 0.9% saline solution + DMSO (S. maritima 5) to quantification of gingival levels of cytokines TNF-α, IL-1β, CINC-1 and IL-10 (n = 18), antioxidant enzymes superoxide dismutase (SOD) and catalase (CAT) (n = 18) and qRT-PCR analysis of TNF-α, IL-1β, RANK, and iNOS (n = 18). All these assays were performed using the

**Abbreviations:** cDNA, Complementary Deoxyribonucleic Acid; CINC-1, Rat cytokine-induced neutrophil chemoattractant – 1; DMSO, Dimethyl sulfoxide; dNTP, Deoxynucleotide; DTT, Dithiothreitol; EDTA, Ethylenediaminetetraacetic acid; HE, Hematoxylin and eosin; IL-10, Interleukin 10; IL1-β, Interleukin 1 beta; NBT, Nitro-blue tetrazolium; qRT-PCR, Reverse Transcription Polymerase Chain Reaction; RANK, Receptor activator of nuclear factor-kappaB; RNA, Ribonucleic Acid; TNF-α, Tumor Necrosis Factor-alpha.

gingival tissues from surrounding maxillary left molars (the half of the maxillae with ligature).

# Measurement of Alveolar Bone Loss

After the period of EP the maxillae were removed, divided in half and fixed in buffered formalin (10%) for 24 h. Following, the maxillae were defleshed and kept in 8% sodium hypochlorite for 4 h (Pimentel et al., 2012). After that, the specimens were washed in running water and stained with methylene blue (1%) to differentiate bone from teeth. Then, hemi-maxillae were fixed in a piece of wax with their occlusal planes parallel to the ground and long axes perpendicular to the camera and photographed with a 6.1-megapixel digital camera (Canon <sup>R</sup> 60D). ImageJ <sup>R</sup> software (National Institute of Health, Bethesda, MD, USA) were used for measurement of ABL, as described previously (Kuhr et al., 2004). The buccal area (mm<sup>2</sup> ) corresponding to the exposed roots and coronary surface of the molars was calculated and obtained and subtracted from the correspondent area (mm<sup>2</sup> ) of the normal right hemi-maxillae.

After definition of the most effective dose, one sample of each group was scanned using a high resolution microcomputed tomography (micro-CT) system (SkyScan 1174; BrukermicroCT, Kontich, Belgic, 50 kV and 800µA) with a 0.5 mm aluminium filter and 15% beam hardening correction and ring artifacts reduction. We used 180 degrees of rotation and exposure range of 1 degree, time of scanning 38 min. Each specimen was scanned with acquisition of images every 0.7◦ , filed in TIFF format, with resolution of 19.7µm and saved on a hard disk. After reconstruction of the images (NRecon v1.6.9; Bruker-microCT) 3D models were created with help of CTAn software v.1.12 program (Bruker-microCT) in accordance with the recommended guidelines.

# Histopathological Analysis

The maxillae with ligature were fixed in 10% neutral-buffer formalin (10%) and demineralized in formic acid (10%). The specimens went through the process of dehydration, paraffin embedment, and after were sectioned in a mesio-distal plane for hematoxylin and eosin (HE) staining. The area between the first and second molars where ligatures had been placed was sectioned in 4µm thickness sections and evaluated under light microscopy considering the inflammatoryparameters such as cell influx and alveolar bone and cementum integrity. Previously standardized scores ranging from 0 to 3 were used (Lima et al., 2000; Leitão et al., 2005). Score 0 indicates absence of or only discrete presence of inflammatory cell infiltration cellular infiltration (restricted to the marginal gingiva, preserved alveolar process and cementum; score 1 represents moderate inflammatory cellular infiltration all over the insert gingiva, some minor alveolar process resorption and intact cementum; score 2 represents accentuated cellular infiltration in both gingiva and periodontal ligament, accentuated degradation of the alveolar process, and partial destruction of cementum; score 3 indicates accentuated cellular infiltration with complete resorption of the alveolar process and intense destruction of cementum. Two examiners who were masked to the identity of samples performed histologic evaluation.

Plasma Bone Alkaline Phosphatase (BALP)

The plasma concentration of bone alkaline phosphatase was obtained from blood samples collected from the retro-orbital plexus on the 11th day using the thermo-activation method, as previously described (Whitby and Moss, 1975). The samples were heated up to 56◦C for 10 min. Serum levels of BALP were calculated by subtracting the concentration of the heated alkaline phosphatase in serum from the concentration of the total alkaline phosphatase (TALP) in serum. The analysis was performed according to the manufacturer's instructions (Labest, Lagoa Santa, MG, Brazil).

# Quantification of Cytokine Levels in Gingival Tissue

The stored gingival tissues were used to determine the concentrations of TNF-α, IL-1β, CINC-1 and IL-10 using specific commercially available kits (DuoSet Elisa kit, R&D Systems Inc., MN, USA). The inflammatory mediators' levels were determined by enzyme-linked immunosorbent assay (ELISA) using respective standard curves. Results were shown as picogram/ml (pg/ml). All kits were used according manufacturer's instructions.

# Superoxide Dismutase (SOD) and Catalase (CAT) Levels in Gingival Tissue

The gingivae removed on 11th day were used to evaluate the effect of S. maritima 5 mg/kg on the oxidative stress. The SOD activity was assayed according the protocol previously described (Beauchamp and Fridovich, 1971). In a dark room, the gingival samples were homogenized in 20µl of ice-cold phosphate buffer at 15,000 G for 20 min. The supernatants were mixed with a solution comprised of phosphate buffer (50 nM), EDTA (100 nM) and L-methionine (19.5 mM) in a pH of 7.8. Then, 150 ml of a solution of riboflavin (10 nM) and nitro NBT (750 nM) was added and the mixture was exposed to light (20 W) for 15 min. The absorbance of the samples was measured at 560 nm. The results are expressed as grams of SOD per ml.

The measurement of O<sup>2</sup> production rate and H2O in proportion of H2O<sup>2</sup> was calculated to obtain CAT activity. Briefly, 20µl of gingival homogenate was mixed with a solution comprised of 3% H2O<sup>2</sup> and Tris-HCl EDTA buffer (5 nM, pH 8.0). The absorbance was measure immediately and 6 min after preparing the samples at a 230 nm wavelength (Maehly and Chance, 1954).

# qRT-PCR Analysis of TNF-α, IL-1β, RANK, and iNOS Levels

Total RNA was extracted from gingivae tissues using the TRIzol reagent (Invitrogen, São Paulo, Brazil). The RNA concentration was estimated by reading the absorbance at 260 nm and was checked for purity at 280 nm in a spectrophotometer (Amersham Biosciences, Cambridge, England). For each sample, the RNA concentrations used to synthesize cDNA were adjusted to 1,000 ng/mL. Before the reverse transcription reaction, samples of RNA were incubated for 5 min at 70◦C and then cooled in ice. The reverse transcription was performed in a total volume of 20µL composed of 10µL of sample RNA, 4µL reverse transcriptase buffer (Invitrogen), 8 units RNAsin, 150 units of reverse transcriptase Superscript III, 0.036U random primers, 10 mM DTT and 0.5 mM of each dNTP (Invitrogen). The mixture was incubated at 42◦C for 1 h, subsequently at 80◦C for 5 min and finally stored at −20◦C. The negative control was prepared under the same conditions but without addition of reverse transcriptase.

Quantitative real-time polymerase chain reaction (qRT-PCR) was performed in triplicate to determine the gingival levels of mRNA for TNF-α, IL1-β, RANK and iNOS. Each reaction contained 10µL of SYBR <sup>R</sup> Green Master Mix (Applied Biosystems, Warrington, UK), 7,3µL of ultra-pure water, 1µL of cDNA and 0.5µM of each primer and was performed in StepOne Real-Time PCR (Applied Biosystems, Warrington, UK) thermocycler. The thermal cycling profile for the first round of qRT-PCR was initial denaturation and activation of the polymerase for 10 min at 95◦C, followed by 40 cycles of 15 s at 95◦C, 30 s at 58◦C, and 30 s at 72◦C. The final extension was for 10 min at 72◦C. The primers were designed by using the PrimerQuest <sup>R</sup> Tool https://www.idtdna.com/Primerquest/ Home/Index. The primers used in this study are shown in **Table 1**. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as endogenous control for normalization of messenger RNA expression. The specificity of each primer pair was confirmed by melting curve analysis of qRT-PCR products. Relative quantifications of mRNA were carried out using the comparative threshold (CT) cycle method. The delta-delta-Ct method was used to transform the Ct values into normalized relative expression levels (Livak and Schmittgen, 2001).

# Subchronic Toxicity Evaluation

Variation of body mass, organ weight alteration and biochemical and histopathological parameters were evaluated in animals treated daily during 11 consecutive days with a single dose of S. maritima 5 mg/kg and saline + DMSO. Peripheral blood



samples were collected in order to obtain biochemical analysis of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels and creatinine (Labtest, Lagoa Santa, MG, Brazil).

After sacrificing the animals, the liver, kidney and heart were removed and weighed. The stomach was also removed for histological analysis. The organs were fixed with formalin and then dehydrated with increasing concentrations of ethanol and embedded in paraffin. The blocs were sliced in 5µm thick sections, stained with hematoxylin-eosin (HE) and observed at light microscope (Leica DM 2000, Wetzlar, Germany).

# Statistical Analysis

Shapiro-Wilk normality test was performed to analyze the data. Results are presented as means ± standard error (SEM) or as medians when appropriate. ANOVA followed by Tukey test or Games-Howell test were used to compare means and Kruskal– Wallis and Dunn tests were used to compare medians. P < 0.05 was considered significant. Analyses were performed using IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY or GraphPad Prism 6 software, San Diego, CA, USA.

# RESULTS

# Alveolar Bone Loss and Histopathological Analysis

No significant ABL was observed in the naive group (**Figures 1a,b**). Data indicated that S. maritima 5 mg/kg was the most effective dose protecting against ABL. Gavage (per os) administration of S. maritima 5 mg/kg 1 h before the placement of the ligature and once daily for 11 days resulted in a significant (P < 0.05) inhibition of ABL (2.37 ± 0.53) (**Figures 1a,bG,H**), compared to the group that received the vehicle only (4.47 ± 0.22) (**Figures 1a,bD,E**).

Histopathologic analysis of animals subjected to EP (vehicle group) demonstrated accentuated inflammatory cell infiltration, breakdown of alveolar bone, collagen fiber derangement within the periodontal ligament, and resorption of cementum, receiving a median score of 3 (range 2 to 3) (**Table 2**). The periodontium of animals treated with S. maritima 5 mg/kg showed preservation of the alveolar process and cementum, reduction of the inflammatory cell infiltration, and partial preservation of collagen fibers of the periodontal ligament (**Figures 1bC,F,I**), receiving a median score of 1 (range 1 to 2) (**Table 2**). These values were statistically different (P < 0.05), compared with the vehicle group.

# Plasma Bone-Specific Alkaline Phosphatase (BALP)

Experimental periodontitis decreased BALP serum levels and the treatment with S. maritima 5 mg/kg resulted in a significant increase in the BALP serum levels (P < 0.05), when compared with the vehicle group (**Figure 2**).

FIGURE 1 | (a) Effect of oral gavage of vehicle (saline + DMSO) and Sm extracts concentrations on alveolar bone loss (0.2, 1, and 5 mg/kg) in experimental periodontitis. Data represent the mean ± SEM of six animals/group. \*P < 0.05 was considered significantly different compared to the naïve control group; \*\*P < 0.05 was considered significantly different compared to vehicle group (saline + DMSO). (ANOVA and Games-Howell post-hoc test). (b) Macroscopic view (first column), microCT images (second column) and histological aspects (third column) of Naive, Vehicle and Sm 5 mg/kg. Data represent the mean ± SEM of six animals/group. (A–C) Indicates normal maxilla (naive), showing integrity of its components (c, Cementum; d, Dentine and ab, Alveolar bone). (D–F) Shows maxilla subjected to experimental periodontitis that received only the vehicle (saline + DMSO), showing severe bone resorption, inflammatory infiltrate in gingiva and periodontal ligament, extensive cementum destruction and total resorption of the alveolar process. (G–I) Indicate maxilla after 11 days of experimental periodontitis treated with Sm 5 mg/kg showing discrete cell influx and preservation of the alveolar process and cementum. Magnification x100.

TABLE 2 | Effect of oral gavage of Sm extracts and vehicle (saline + DMSO) on histopathologic score of rat maxillae.


\*P < 0.05 vs. naïve group (CONTROL); \*\*P < 0.05 vs. vehicle group (animals submitted to experimental periodontitis and treated with saline + DMSO) (Kruskal-Wallis followed by Dunn's test).

# Effects of S. maritima on Cytokine Levels in Gingival Tissue

Periodontitis challenge was associated with significant (P < 0.05) increase of pro-inflammatory cytokines (TNF-α, IL-1β, and CINC-1) in gingival tissue; at the same time it was observed a significant (P < 0.05) decrease in gingival levels of IL-10, an anti-inflammatory cytokine, when compared to naive group. Treatment with Sm 5 mg/kg significantly (P < 0.05) decreased TNF-α and CINC-1 but no significant difference was observed in IL-1β gingival levels; also, S. maritima 5 mg/kg increased IL-10 levels, when compared to vehicle treated group (**Figures 3A–D**).

# Superoxide Dismutase and Catalase Levels in Gingival Tissue

In the present study periodontitis induction resulted in a significant (P < 0.05) reduction of SOD and CAT levels in gingival tissue, when compared with naive group. S. maritima 5 mg/kg significantly (P < 0.05) increased both markers for

plasma bone alkaline phosphatase (BALP) in experimental periodontitis in rats. Data represent the mean ± SEM of six animals/group. \*P < 0.05 compared with naïve group (control); \*\*P < 0.05 compared with vehicle group (animals submitted to experimental periodontitis and treated with saline + DMSO) (ANOVA and Games-Howell post-hoc test).

oxidative stress in gingival tissue, when compared with vehicle group (**Figures 4A,B**).

# qRT-PCR Analysis of TNF-α, IL-1β, RANK and iNOS Levels

The vehicle groups showed significant increase in TNFα (**Figure 5A**), IL-1β (**Figure 5B**), RANK (**Figure 5C**) and iNOS (**Figure 5D**) mRNA expression, when compared with

± SEM of six animals/group. \*P < 0.05 compared with naïve group (control); \*\*P < 0.05 compared with vehicle group (animals submitted to experimental periodontitis and treated with saline + DMSO) (ANOVA and Tukey's post-hoc test).

the naive control group. Administration of S. maritima 5 mg/kg significantly reduced mRNA expression in all parameters evaluated compared to vehicle group (**Figure 5**).

# Analysis of Toxicity

No signs of systemic illness, adverse pharmacological events or changes in behavior were observed throughout the experimental period. S. maritima 5 mg/kg or vehicle did not affect the animals' body mass or the wet weight of the livers, kidneys or hearts compared to the naive control throughout the study period. Gross necropsy findings did not show any abnormalities. A single-dose of S. maritima (5 mg/kg) or vehicle (saline + DMSO) over 11 consecutive days had no significant histological alterations in the hepatic, renal parenchyma, and cardiac tissue. The values obtained for ALT/AST and creatinine did not differ from the control group.

# DISCUSSION

Ligature-induced periodontitis is a well-established animal model. While it has some limitations, the structure and organization of the periodontal tissues are similar to those of

humans. Further, rat models have fewer limitations than in vitro models that cannot reproduce the complexity of interactions among the oral microbiome, environmental factors, and the immune/inflammatory host response (Struillou et al., 2010; Oz and Puleo, 2011; Hajishengallis et al., 2015). In this model, ligature causes mechanical trauma, and affects tissue integrity to induce an inflammatory response, increase oxidative stress, and destroy the periodontal ligament leading to significant bone loss (Bezerra et al., 2000; Guimarães et al., 2016; Lima et al., 2017; Oliveira et al., 2017). Alternative and preventive treatment options are essential to overcome the adverse effects of both antimicrobial and anti-inflammatory agents as an adjunct to conventional mechanical and surgical treatments (Chandra Shekar et al., 2015; Ramesh et al., 2016).

Chemical studies on S. maritima reported the isolation of diterpenes and flavonoids showing antiviral, cytotoxic, and larvicidal activities (Hufford et al., 1991; Rodrigues et al., 2010). Further, it was showed that S. maritima compounds have in vitro inhibitory activity of both lipids peroxidation and cyclooxygenase 1 and 2 (Hufford et al., 1992; Russell et al., 2011). Besides, our research group demonstrated that S. maritima has antioxidant activity in vitro and it also has activity against some bacterial strains (da Silva et al., 2014). However, it is important to mention that there is no study in the literature using S. maritima extract or its identified compounds on bone resorption models. These data encouraged us to investigate whether S. maritima extract could be useful to ameliorate the bone loss during periodontitis.

The protective effect of S. maritima on alveolar bone loss was associated with an increase in plasma bone-specific alkaline phosphatase (BALP) suggesting that 5 mg/kg S. maritima prevents bone resorption and stimulates bone formation. Many studies have verified the effects of medicinal plants on ABL using a similar periodontitis model (Sezer et al., 2013; Hatipoglu et al., 2015; Saglam et al., 2015; Guimarães et al., 2016). Sezer et al. (2013) showed that the systemic use of Ginko biloba extract on reducing ABL. Hatipoglu et al. (2015) assessed ABL from microcomputed tomography (micro CT) images, and verified less bone resorption in animals treated with Crataegus orientalis extracts. Recently, Guimarães et al. (2016) observed the ability of the Matricaria recutita extract to inhibit TNF-α and IL-1β cytokines. These treatments prevented the osteoclast activation via RANKL-OPG. Lima et al. (2017) demonstrated anti-inflammatory and anti-oxidant activities with Calendula officinalis.

Cytokines play a significant role in periodontitis. Here, we showed that S. maritima (5 mg/kg) significantly increased IL-10 gingival levels, an anti-inflammatory cytokine, while decreasing the pro-inflammatory cytokines TNF-α and CINC-1. Alternative therapeutic approaches based that inhibit TNF-α production have been successfully used for the pre-clinical and clinical treatment of chronic inflammatory diseases, particularly rheumatoid arthritis and temporomandibular joint disorders (Feldmann, 2002; Araújo et al., 2013; Freitas et al., 2016; Alves et al., 2017). Further, when mRNA expression for TNF-α, IL-1β, iNOS, and RANK were evaluated, S. maritima 5 mg/kg significantly reduced mRNA expression for all these genes compared to the vehicle group. Cytokines amplify the inflammatory response in periodontitis (Duarte et al., 2015). It seems contradictory that S. maritima 5 mg/kg reduced mRNA expression for IL-1β without affecting IL-1β gingival levels. Here, we hypothesized that the IL1-β levels detected by ELISA in gingival samples might derive from a pre-formed pool because the increase in IL-1β levels after periodontitis challenge precedes the increase in mRNA. Studies analyzing the effects of resveratrol, a naturally occurring product found in numerous plants showed similar results (Casati et al., 2013; Tamaki et al., 2014).

The release of large amounts of NO by iNOS plays a major role in immune-inflammatory events including periodontitis (Leitão et al., 2005; Tamaki et al., 2014; Martins et al., 2016). TNF-α and IL-1 trigger the transcription of the iNOS resulting in the increase production of NO. On the other hand, the significant increase of IL-10 levels in gingival tissues could also decrease iNOS gene expression (Gadek-Michalska et al., 2013). Some components of medicinal plants might inhibit nuclear transcription factor-kB binding activity and downregulate the expression of iNOS (Kim et al., 2005; Cai et al., 2008). These results are in accordance with our present data since the extract of S. maritima decreased transcription of iNOS genes.

Oxidative stress plays a central role in periodontitis (Tóthová et al., 2015; Lima et al., 2017). This study demonstrated that experimental periodontitis in rats leads to oxidative stress as indicated by a significant reduction in SOD and CAT levels in gingival tissue. This effect was reduced in animals treated with S. maritima (5 mg/kg), which suggests that the beneficial effects of S. maritima are at least partially related to its antioxidative properties. Calendula offinalis extract also presents ability to reduce oxidative stress by increasing superoxide dismutase (SOD), catalase (CAT) and reduced gluthatione (GSH) enzymes and decreasing malonaldeyde (MDA) levels in gingival tissue (Lima et al., 2017).

The regulation of the critical cytokine macrophage colonystimulating factor, RANK ligand is essential to the differentiation of osteoclasts. Our results indicated that ligature-induced periodontitis in rats is associated with an increase in the RANK mRNA levels in periodontal tissues, and the treatment with S. maritima (5 m/kg) could reduce RANK expression.

Medicinal plants have been historically used for treatment of numerous human diseases (Varela-López et al., 2015; de Oliveira et al., 2016; Ramesh et al., 2016). The leaves and stem of S. maritima can treat stomach pain and fluid retention, although the literature still lacks studies that confirm the safety of this plant. Thus, in this study we used biochemical and histopathological analysis to show that the administration of S. maritima (5 mg/kg) did not promote any signs of toxicity when administered for 11 consecutive days.

This study observed that 5 mg/kg S. maritima extract reduced alveolar bone loss, inflammation, and oxidative stress in a ligature-induced model of periodontitis in rats without causing any systemic changes. These data are in accordance with previous results from our group, which demonstrated via in vitro assays that S. maritima has anti-inflammatory, antioxidant and antibacterial properties (da Silva et al., 2014). We also found evidence that, at least in part, the effectiveness of S. maritima depends upon a positive balance between pro and anti-inflammatory cytokines that decrease TNF-α, IL-1β, and CINC-1 gingival levels while increasing IL-10. This effect might also modulate both iNOS activity and RANK levels, to improve antioxidative events.

To the best of our knowledge, there are no existing reports evaluating the efficacy of S. maritima extract in a preclinical trial of rat periodontitis. Although additional studies are needed, these data suggest that S. maritima is a potential tool for controlling the development of periodontitis.

# AUTHOR CONTRIBUTIONS

AT and JF treated the animals, performed all assays, analysis and interpretation of data. LdS induced periodontitis. AA and FL performed the collection and extraction of plant material. NdS performed cytokines analysis. IM, KP, and IC performed histolopathogical analysis. AP assisted in laboratory experiments. JC performed quantitative qRT-PCR analysis. GC-F and VP performed and analyzed the S. maritima Linn toxicity profile. MB, HC, and PG designed and supervised the study. AT and MB wrote the paper. All authors critically reviewed and approved the manuscript.

# ACKNOWLEDGMENTS

This work was supported by Brazilian grants from Conselho Nacional de Pesquisa (CNPq) (grant #471974/2013-7), and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (Capes) / Fundação Cearense de Apoio ao Desenvolvimento Científico e Tecnológico (Funcap) (grant # AE1-0052- 000180100/2011). The authors are grateful to Bruno C. Vasconcelos from the School of Dentistry, Federal University of Ceará, Sobral, Ceará, Brazil for the microCT images**.**

# REFERENCES

Albach, D. C., Meudt, H. M., and Oxelman, B. (2005). Piecing together the "new" Plantaginaceae. Am. J. Bot. 92, 297–315. doi: 10.3732/ajb.92.2.297

Alves, S. M., Abreu, S. C., Lemos, J. C., Gomes, F. I. F., Alves, S. M., do Val, D. R., et al. (2017). Anti-inflammatory and anti-nociceptive effects of strontium ranelate on the zymosan-induced temporomandibular joint inflammatory hypernociception in rats depend on TNF-α inhibition. Pharmacol. Rep. 69, 764–772. doi: 10.1016/j.pharep.2017.03.007

Alviano, D. S., and Alviano, C. S. (2009). Plant extracts: search for new alternatives to treat microbial diseases. Curr. Pharm. Biotechnol. 10, 106–121. doi: 10.2174/138920109787048607


administration in preventing alveolar bone loss in rats with experimental periodontitis. PLoS ONE 10:e0128134. doi: 10.1371/journal.pone.0128134


bone loss in experimental periodontitis in rats. J. Appl. Oral Sci. 23, 33–41. doi: 10.1590/1678-775720140288


prevention and treatment of periodontal diseases. Antioxidants 4, 447–481. doi: 10.3390/antiox4030447


**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 © 2017 Teixeira, Freire, de Sousa, Parente, de Sousa, Arriaga, Lopes da Silva, Melo, Castro da Silva, Pereira, Goes, Costa, Cristino-Filho, Pinto, Chaves and Bezerra. 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) or licensor 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.

# Intermittent Hypoxia Influences Alveolar Bone Proper Microstructure via Hypoxia-Inducible Factor and VEGF Expression in Periodontal Ligaments of Growing Rats

Shuji Oishi <sup>1</sup> , Yasuhiro Shimizu<sup>1</sup> , Jun Hosomichi <sup>1</sup> \*, Yoichiro Kuma<sup>1</sup> , Hideyuki Maeda<sup>2</sup> , Hisashi Nagai <sup>3</sup> , Risa Usumi-Fujita<sup>1</sup> , Sawa Kaneko<sup>1</sup> , Naoki Shibutani <sup>1</sup> , Jun-ichi Suzuki <sup>4</sup> , Ken-ichi Yoshida<sup>2</sup> and Takashi Ono<sup>1</sup>

### Edited by:

*Thimios Mitsiadis, University of Zurich, Switzerland*

### Reviewed by:

*Harald Osmundsen, University of Oslo, Norway Supawadee Sukseree, Medical University of Vienna, Austria*

> \*Correspondence: *Jun Hosomichi hosomichi.orts@tmd.ac.jp*

Specialty section: *This article was submitted to*

*Craniofacial Biology, a section of the journal Frontiers in Physiology*

Received: *19 July 2016* Accepted: *05 September 2016* Published: *16 September 2016*

### Citation:

*Oishi S, Shimizu Y, Hosomichi J, Kuma Y, Maeda H, Nagai H, Usumi-Fujita R, Kaneko S, Shibutani N, Suzuki J-i, Yoshida K-i and Ono T (2016) Intermittent Hypoxia Influences Alveolar Bone Proper Microstructure via Hypoxia-Inducible Factor and VEGF Expression in Periodontal Ligaments of Growing Rats. Front. Physiol. 7:416. doi: 10.3389/fphys.2016.00416* *<sup>1</sup> Department of Orthodontic Science, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, Tokyo, Japan, <sup>2</sup> Department of Forensic Medicine, Graduate School of Medicine, Tokyo Medical University, Tokyo, Japan, <sup>3</sup> Department of Legal Medicine (Forensic Medicine), Keio University School of Medicine, Tokyo, Japan, <sup>4</sup> Department of Advanced Clinical Science and Therapeutics, The University of Tokyo, Tokyo, Japan*

Intermittent hypoxia (IH) recapitulates morphological changes in the maxillofacial bones in children with obstructive sleep apnea (OSA). Recently, we found that IH increased bone mineral density (BMD) in the inter-radicular alveolar bone (reflecting enhanced osteogenesis) in the mandibular first molar (M1) region in the growing rats, but the underlying mechanism remains unknown. In this study, we focused on the hypoxia-inducible factor (HIF) pathway to assess the effect of IH by testing the null hypothesis of no significant differences in the mRNA-expression levels of relevant factors associated with the HIF pathway, between control rats and growing rats with IH. To test the null hypothesis, we investigated how IH enhances mandibular osteogenesis in the alveolar bone proper with respect to HIF-1α and vascular endothelial growth factor (VEGF) in periodontal ligament (PDL) tissues. Seven-week-old male Sprague–Dawley rats were exposed to IH for 3 weeks. The microstructure and BMD in the alveolar bone proper of the distal root of the mandibular M1 were evaluated using micro-computed tomography (micro-CT). Expression of HIF-1α and VEGF mRNA in PDL tissues were measured, whereas osteogenesis was evaluated by measuring mRNA levels for alkaline phosphatase (ALP) and bone morphogenetic protein-2 (BMP-2). The null hypothesis was rejected: we found an increase in the expression of all of these markers after IH exposure. The results provided the first indication that IH enhanced osteogenesis of the mandibular M1 region in association with PDL angiogenesis during growth via HIF-1α in an animal model.

Keywords: bone mineral density, hypoxia inducible factor, intermittent hypoxia, periodontal ligament, vascular endothelial growth factor

# INTRODUCTION

Intermittent hypoxia (IH) during sleep has been implicated in the pathogenesis of obstructive sleep apnea (OSA; Noda et al., 1998; Lal et al., 2012), although the role of IH in the growth of children with OSA has not been clarified. In particular, it has been reported that pediatric OSA is frequently associated with impairment in the growth and development of craniofacial and otolaryngological tissues, as well as with neuromuscular diseases (Balbani et al., 2005; Huang and Guilleminault, 2012). Hypertension, cardiac remodeling, and other complications of OSA have been studied using rodent models of IH induced by short cycles of hypoxia–normoxia (Skelly et al., 2012; Maeda et al., 2013; Nagai et al., 2015). Previously, we demonstrated that IH exposure induces a decrease in the volume of the nasal cavity, reflecting impaired development of maxillofacial bones (Kuma et al., 2014). Other researchers showed that IH increased the bone mineral density (BMD) in the inter-radicular alveolar bone in the mandibular first molar (M1) region during growth (Oishi et al., 2015), but the underlying mechanism remains unknown.

Hypoxia is critical to the remodeling and repair of damaged bones via hypoxia-inducible factor (HIF; Maes et al., 2012), the key stimulator of vessel formation and angiogenesis. The gene encoding the prominent angiogenic factor vascular endothelial growth factor (VEGF) is the primary target for HIF-1α (Riddle et al., 2009). A recent study provided evidence supporting the view that HIFs and VEGF play essential roles in coupling between angiogenesis and osteogenesis during bone formation and repair (Schipani et al., 2009).

Alveolar bone proper is covered with collagen fibers in the periodontal ligament (PDL; Shimizu et al., 2014). The PDL is a specialized soft connective tissue that connects the tooth with the alveolar bone socket, thereby promoting the development and maintenance of periodontium (Kaku and Yamauchi, 2014). PDL tissues are comprised of PDL cells (Shimizu et al., 2014), collagen fibers (Kaku and Yamauchi, 2014), blood vessels (Muramoto et al., 2000), nerve elements (Muramoto et al., 2000), extracellular substances (Kaneko et al., 2001), osteoclasts (Kaneko et al., 2001), and osteoblasts (Mayahara et al., 2012), and provides progenitor cells for bone formation and remodeling. Alkaline phosphatase (ALP) and bone morphogenetic protein-2 (BMP-2) are known to induce osteogenesis and the osteogenic transformation of PDL cells (Kuru et al., 1999; Selvig et al., 2002). ALP activity reflects early osteogenic differentiation in the presence of osteoblasts (Kuru et al., 1999). Previous data showed that ALP activation and BMP-2 upregulation in PDL cells induce periodontium osteogenesis in response to growth hormones (Li et al., 2001) or matrix Gla proteins (Li et al., 2012). In addition, mechanical forces such as moderate occlusal stimuli and dissipation of masticatory force are transmitted from the teeth through the PDL to the progenitor cells, thereby promoting bone remodeling in the periodontal tissue (Chen et al., 2005). As such, PDL tissues maintain proper alveolar bone homeostasis via ALP and BMP-2.

Bone remodeling, a complex process by which old bone is continuously replaced by new tissue, is affected by a variety of biochemical and mechanical factors (Hadjidakis and Androulakis, 2006). With regard to signaling pathways, the Wnt (Wang et al., 2014), OPG/RANKL/RANK (Hsu et al., 2006), and HIF pathways (Mamalis and Cochran, 2011) are well-known in the control of bone remodeling. In this study, we specifically focused on the HIF pathway to assess the effect of IH by testing the null hypothesis that no significant differences in the mRNAexpression levels of relevant factors associated with the HIF pathway in PDL occur between control rats and growing rats with IH. To this end, we analyzed the microarchitecture and mineral density of the alveolar bone proper around the mandibular M1 after 3 weeks of IH in growing rats, with reference to HIF1-α, VEGF, ALP, and BMP-2 mRNA expression.

# MATERIALS AND METHODS

# Experimental IH Model

Experiments were conducted on twelve 7-week-old male Sprague–Dawley rats randomly divided into two groups. Experimental rats were exposed to IH at a rate of 20 cycles per h (nadir, 4% oxygen; peak, 21% oxygen; 0% carbon dioxide) for 3 weeks (IH group), and control rats breathed room air (C group). The control cage was placed next to the cage equipped with the IH apparatus, and all rats underwent their respective treatments for 8 h per day during the 12-h "lights on" period (Maeda et al., 2013; Nagai et al., 2015). The experiments were conducted while the rats were 7 to 10 weeks of age, when craniofacial bones actively develop (puberty), as documented by studies of craniofacial growth (Spence, 1940) and puberty onset (Cheung et al., 2001) in male rats. All rats were allowed free access to food and water throughout the experimental period, as previously described (Skelly et al., 2012; Maeda et al., 2013; Nagai et al., 2015). Immediately after the IH-exposure period, all rats were anesthetized by a sodium pentobarbital injection and sacrificed. All experimental procedures were performed according to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication 85-23, revised 1996). The animal protocol was approved by the Animal Experimental Committee of Tokyo Medical University (approval number S-26063).

# Three-Dimensional Microcomputed Tomography (Micro-CT) Analysis

Changes in the bony microstructure of the alveolar bone proper around roots of the mandibular M1 and in the PDL space were investigated using micro-CT with a desktop X-ray micro-CT system (SMX-100CT; Shimadzu, Kyoto, Japan) with a scanning resolution of 20-µm intervals on individual images. The region of interest (ROI) for structural morphometry was enlarged to a 40-µm radius around the distal root (**Figure 1A**; Shimizu et al., 2014). Each ROI was analyzed with respect to BMD, bone volume/tissue volume (BV/TV), trabecular thickness (Tb.Th), and trabecular number (Tb.N) using three-dimensional image-analysis software (TRI/3D-BON; Ratoc System Engineering, Tokyo, Japan). Additionally, to evaluate changes from mechanical stimuli in the PDL space, tissue volume (TV) around the distal root of the mandibular M1 was computed (**Figure 1B**; Shimizu et al., 2014).

# Reverse Transcription Quantitative Real-Time PCR (RT-qPCR) Analysis

PDL tissues were removed from extracted roots of the mandibular M1. Total RNA was isolated from PDL tissues using the PureLinkTM FFPE Total RNA Isolation Kit (Invitrogen, CA, USA) according to instructions provided by the manufacturer (Afanador et al., 2005; Luan et al., 2007). cDNA was synthesized from total RNA with reverse transcription random primers using High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, Foster City, CA, USA). Quantitative PCR assays were carried out in triplicate for each sample using a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). PCR analyses were conducted with gene-specific primers and fluorescently labeled TaqMan probes (Takara Bio, Shiga, Japan). Appropriate primers were chosen for real-time PCR amplification of VEGF (forward primer: 5′ -ccttgctgctctacctccac-3 ′ , reverse primer: 5′ -ccacttcgtgatgattctgc-3′ ), HIF-1α (forward primer: 5′ -ctaccagaagggcaggatacag-3′ , reverse primer: 5 ′ -gcaggcagatgaaataccagtc-3′ ), ALP (forward primer: 5′ acgtggctaagaatgtcatc-3′ , reverse primer: 5′ -ctggtaggcgatgtcctta-3 ′ ), BMP-2 (forward primer: 5′ -tcaagccaaacacaaacagc-3′ , reverse primer: 5′ -acgtctgaacaatggcatga-3′ ), and Hprt-1 (forward primer: 5 ′ -cagactttgctttccttgg-3′ , reverse primer: 5′ -tccactttcgctgatgacac-3 ′ ). The thermocycling conditions used were 95◦C for 30 s, followed by 40 cycles of 95◦C for 5 s and 60◦C for 34 s. Gene expression levels were calculated according to the 11CT method of relative quantification. The threshold cycle (Ct) value of the target mRNAs (VEGF, HIF1α, ALP, or BMP-2) was normalized to the Ct values of the internal control (Hprt-1) in the same sample (1Ct = Cttarget – CtHprt−1), followed by normalization to the control (11Ct = 1CtIHgroup – 1CtCgroup). The fold change in expression was calculated as the relative quantification value (RQ; 2−11Ct; Livak and Schmittgen, 2001).

# Statistical Analysis

Statistical calculations were performed using statistical analysis software (IBM SPSS Statistics Version 20.0 Chicago, IL, USA). We first examined the normality and variance of the data using

the F-test. The control and experimental groups were compared using the Mann–Whitney U-test for nonparametric data, and statistical significance was established at a p level of less than 0.05.

# RESULTS

# Body Weight Changes in Rats after IH

The median body weight (mean ± standard error) of rats exposed to IH for 3 weeks (277.5 ± 3.8 g) was significantly lower than that of control (C) rats (356.0 ± 9.8 g). Previous reports using this animal model showed that the correlation between the body weight and whole-body growth in this IH model was low (Maeda et al., 2013; Kuma et al., 2014; Oishi et al., 2015).

# Three-Dimensional Micro-CT Analysis

We investigated changes in the bony microstructure of the alveolar bone proper around the distal roots of mandibular M1 using micro-CT. Micro-CT images of alveolar bone proper around the distal root of mandibular M1 showed higher bone volume (BV) density in the IH rats than in the C rats (**Figure 2**). In addition, micro-CT analyses demonstrated significant increases in the BMD, BV/TV, Tb.Th, and Tb.N in the distal root alveolar bone proper of IH rats, compared with those of C rats (**Figure 3**).

# Quantitative Real-Time PCR

We evaluated the relative expression levels of osteogenesis– angiogenesis coupling markers (**Figure 4**) and osteogenic markers (**Figure 5**) in PDL tissues by RT-qPCR analysis. IH

FIGURE 3 | Comparisons of bone morphology between the C and IH groups by micro-CT analysis. Cancellous bone of the alveolar bone proper of the distal roots in the mandibular first molar region was compared between the C and IH groups. Box edges represent the upper and lower quantiles, the middle lines in the boxes represent the medians, and the whiskers represent the maxima and minima. BMD, bone mineral density; BV/TV, bone volume/tissue volume; Tb.Th, trabecular thickness; Tb.N, trabecular number. \**p* < 0.05 by the Mann–Whitney *U*-test.

increased the mRNA levels of HIF-1α and VEGF by 1.71- and 1.97-fold, respectively, in PDL tissues compared with those observed in C rats (**Figure 4**). Similarly, ALP and BMP-2 mRNA levels were increased 2.75- and 2.57-fold in PDL tissues compared to the corresponding levels in C rats (**Figure 5**). These changes are not likely to have been mediated by mechanical forces, as IH had no effect on the PDL space surrounding the distal root of the mandibular M1 (**Figure 6**).

# DISCUSSION

The null hypothesis that no significant differences would occur in the mRNA-expression levels of some factors associated with

the HIF pathway in PDL between control rats and IH rats was rejected: IH enhanced osteogenesis in the mandibular M1 region in association with VEGF, ALP, and BMP-2 gene up-regulation in PDL tissues. These molecules are abundantly expressed in osteoblasts and promote angiogenesis in association with osteogenesis via the HIF-1α pathway (Mamalis and Cochran, 2011).

Previously, we reported that morphological changes occur in craniofacial bone after IH exposure in growing rats. IH suppressed development of the nasal cavity and decreased the size of the mandibular and viscerocranial bones, which could have disturbed nasal breathing (Kuma et al., 2014; Oishi et al., 2015). Oxygen is indispensable for enzymatic reactions to promote tissue development, whereas HIF and VEGF play pivotal survival roles under hypoxic conditions (Maes et al., 2012). In the elderly,

it was reported that OSA is associated with an increase in BMD, which suggests that IH can stimulate bone remodeling (Sforza et al., 2013). Moreover, activation of the HIF pathway by hypoxiamimicking agents prevents bone loss in estrogen-deficient mice, but increases BMD and trabecular microarchitecture (Peng et al., 2014). Data from the present study revealed enhanced BMD and bone development in the alveolar bone proper in the mandibular M1 region after IH exposure (**Figures 2**, **3**). The 3 weeks of IH exposure starting at the age of 7 weeks in rats reflects pre-puberal development (Sengupta, 2013). Given that remodeling of interradicular alveolar bone, such as alveolar bone proper, in the mandibular M1 region is promoted by signals from PDL tissues (Kaku and Yamauchi, 2014), we considered that IH induced the peri-M1 changes via enhanced coupling in the signaling pathways for osteogenesis and angiogenesis in the PDL.

HIF-1α, a member of the HIF subfamily, is a ubiquitously expressed transcription factor that regulates cellular adaptation under hypoxia (Liu and Simon, 2004). VEGF is transcriptionally activated by HIF and positively regulates angiogenesis (Miyagawa et al., 2009; Wan et al., 2010). Previous findings have indicated that HIF-1α promotes both angiogenesis and osteogenesis via VEGF upregulation in osteoblasts (Wang et al., 2007). In addition, it was reported that VEGF stimulates the differentiation and chemotactic migration of osteoblastic cells (Hankenson et al., 2015), whereas HIFs and VEGF are involved in skeletal development and bone homeostasis (Wan et al., 2010; Maes et al., 2012). IH stimulates vessel network formation and VEGF production in a highly correlated fashion (Ehsan and George, 2015). The vascular system not only supplies nutrients and oxygen to developing bones, but also delivers critical signals that stimulate mesenchymal cell differentiation toward an osteogenic phenotype, whereas HIF triggers the initiation and promotion of angiogenic-osteogenic cascade events (Mamalis and Cochran, 2011). It was also shown that hypoxia promotes VEGF production in PDL cells (Motohira et al., 2007) and that the PDL plays an important role in tooth eruption (Kaku and Yamauchi, 2014). Consistent with these findings, we found increased HIF-1α and VEGF mRNA expression in peri-M1 PDL tissues after IH exposure (**Figure 4**), and the lack of change observed in the PDL space (**Figure 6**) suggests that, not mechanical force, but chemical stimulation related to HIF-1α from PDL tissues induced peri-M1 osteogenesis. Collectively, our findings indicate that HIF-1α mediated VEGF induction in PDL tissues and that VEGF induced peri-M1 osteogenesis.

We evaluated osteogenesis by studying ALP and BMP-2 expression. Given that ALP is a well-known indicator of osteoblastic differentiation, ALP expression indicates the presence of osteoblasts and osteogenic activity (Kuru et al., 1999; Nettelhoff et al., 2016). BMP-2, a member of the transforming growth factor beta superfamily, induces osteogenesis from PDL cells and the regeneration of alveolar bone by promoting osteoblastic differentiation (Selvig et al., 2002). We found significantly higher ALP and BMP-2 expression levels in PDL tissues in rats after IH exposure (**Figure 5**), which are thought to induce osteogenesis in the alveolar bone proper.

Cephalometric analysis indicated small mandibular sizes and posterior displacements in patients with OSA (Rivlin et al., 1984). Children with OSA present with increased over jet, reduced over bite, narrowed upper dental arches, and shorter lower dental arches (Pirilä-Parkkinen et al., 2009). Oral appliances and mandibular-advancement devices have been used to treat abnormal craniofacial development, such as morphological changes and intermaxillary relations in patients with OSA (Almeida et al., 2006; Hou et al., 2006). It has been suggested that a priori changes in craniofacial bones induce pathogenesis in patients with OSA.

Data from previous reports have indicated that the morphologies of mandibular and dental changes play primary roles in the development of OSA pathophysiology, whereas the present data indicated that the changes of molecular mechanism about PDL tissues in IH rat model. The osteogenesis– angiogenesis coupling phenomenon with HIF-1α and VEGF was involved in the increased BMD observed in developing alveolar bone under IH. Collectively, our data demonstrated for the first time that short periods of IH exposure can enhance peri-M1 abnormally osteogenesis via the HIF-1α-VEGF pathway in growing rats. Similar mechanisms may occur in the craniofacial bones in young children under IH exposure. Our results suggested the necessity of early treatment in children with OSA to maintain normal bone growth. Furthermore, we also demonstrated the usefulness of this IH model in studying the molecular mechanism underlying the morphological changes occurring in craniofacial bones in children with OSA. Although IH has been strongly implicated in OSA pathogenesis, OSA is associated with multifactorial pathogenesis, such as hypercapnia, intrathoracic negative pressure, and sympathetic overactivation (Fletcher, 2001). IH exposure as a single pathogenesis of OSA, a limitation of this study, is that the morphological changes observed in rats may differ somewhat from those that occur in children with OSA; thus, further morphological studies on pediatric OSA are required. Further study is also expected under different experimental settings, involving differences in IH exposure and age.

In conclusion, we demonstrated that IH increases BMD in alveolar bone proper around the roots of the first mandibular molar. Our data suggest the involvement of an osteogenesis– angiogenesis coupling phenomenon with HIF-1α and VEGF in PDL tissues. IH was found to significantly increase BMD and alter bone microstructure, potential risk factors for homeostasis

# REFERENCES


disturbance in alveolar bone proper in growing IH rats. The expression levels of HIF-1α, VEGF, ALP, and BMP-2 transcripts were up-regulated in PDL tissues subjected to IH exposure for 3 weeks. Although the signaling pathway underlying IHinduced changes in the bony microstructure is not yet fully elucidated, these findings improve our current understanding of the molecular mechanisms underlying the impacts of IH on bone homeostasis.

# AUTHOR CONTRIBUTIONS

SO, YS, JH, YK, HM, HN, RU, SK, NS, JS, KY, and TO conceived of and designed the study; SO performed the experiments; SO, YS, JH, and TO analyzed the data; SO, YS, JH, YK, HM, HN, RU, SK, NS, JS, KY, and TO interpreted the results of the experiments; SO and YS prepared the figures; SO drafted the manuscript; YS, JH, JS, and TO edited the manuscript; All authors approved the final version of the manuscript.

# FUNDING

This study was financially supported in part by Grants-in-Aid for Scientific Research (25463170, 26463089) from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (KAKENHI).

associated with bone mineral density at different skeletal sites in men. Hum. Genet. 118, 568–577. doi: 10.1007/s00439-005-0062-4


movement of teeth. J. Histochem. Cytochem. 55, 127–140. doi: 10.1369/jhc.6A70 18.2006


**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 © 2016 Oishi, Shimizu, Hosomichi, Kuma, Maeda, Nagai, Usumi-Fujita, Kaneko, Shibutani, Suzuki, Yoshida and Ono. 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) or licensor 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.

# Calvarial Suture-Derived Stem Cells and Their Contribution to Cranial Bone Repair

### Daniel H. Doro, Agamemnon E. Grigoriadis and Karen J. Liu\*

*Centre for Craniofacial and Regenerative Biology, King's College London, Guy's Hospital, London, United Kingdom*

In addition to the natural turnover during life, the bones in the skeleton possess the ability to self-repair in response to injury or disease-related bone loss. Based on studies of bone defect models, both processes are largely supported by resident stem cells. In the long bones, the source of skeletal stem cells has been widely investigated over the years, where the major stem cell population is thought to reside in the perivascular niche of the bone marrow. In contrast, we have very limited knowledge about the stem cells contributing to the repair of calvarial bones. In fact, until recently, the presence of specific stem cells in adult craniofacial bones was uncertain. These flat bones are mainly formed via intramembranous rather than endochondral ossification and thus contain minimal bone marrow space. It has been previously proposed that the overlying periosteum and underlying dura mater provide osteoprogenitors for calvarial bone repair. Nonetheless, recent studies have identified a major stem cell population within the suture mesenchyme with multiple differentiation abilities and intrinsic reparative potential. Here we provide an updated review of calvarial stem cells and potential mechanisms of regulation in the context of skull injury repair.

### Edited by:

*Thimios Mitsiadis, University of Zurich, Switzerland*

### Reviewed by:

*Natalina Quarto, University of Naples Federico II, Italy Franz E. Weber, University of Zurich, Switzerland Martin James Stoddart, AO Research Institute Davos, Switzerland*

> \*Correspondence: *Karen J. Liu karen.liu@kcl.ac.uk*

### Specialty section:

*This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology*

Received: *29 September 2017* Accepted: *10 November 2017* Published: *27 November 2017*

### Citation:

*Doro DH, Grigoriadis AE and Liu KJ (2017) Calvarial Suture-Derived Stem Cells and Their Contribution to Cranial Bone Repair. Front. Physiol. 8:956. doi: 10.3389/fphys.2017.00956*

Keywords: skull, calvaria, bone, stem cells, repair, cranial

# INTRODUCTION: SKELETAL STEM CELLS (SSCs)

Skeletal stem cells (SSCs) provide the bones with a supply of osteochondroprogenitors during development, modeling and life-long homeostasis (Park et al., 2012). More importantly, these cells are crucial for appropriate self-healing in response to injury or general bone loss. Bone marrow (BM) residing stem cells have been suggested to assist the repair by establishing a pro-regenerative hematopoietic microenvironment, but also by supplying progenitors for all the newly formed skeletal components (Grcevic et al., 2012; Park et al., 2012).

Defining specific stem cell subsets and their ontogeny within the stromal system remains a challenge, as the discovery of numerous markers and isolation methods reveals a large phenotypic heterogeneity of the SSC population (Bianco and Robey, 2015). Fundamental studies leading to the initial characterization of SSCs were based on heterotopic transplantation of bone marrow cell suspensions or bone-devoid BM explants into extramedullary sites. Both populations were able to generate an ectopic ossicle, similar in tissue composition and architecture to trabecular bone and its stroma, fulfilling the first of two essential criteria that define a stem cell population—the in vivo multipotency (Friedenstein et al., 1968; Tavassoli and Crosby, 1968). The second criterion, self-renewal capacity, was met later as isolated clonogenic cells were found able to reconstitute progenitors with identical phenotype, potency and equivalent anatomical location within the ectopic bony structure formed (Friedenstein et al., 1974; Kuznetsov et al., 1997). Further knowledge about phenotypic identity and anatomical location of SSCs has been accumulated upon discovery of cell-surface markers like human-CD146 (Sacchetti et al., 2007) and mouse-CD45/CD200 (Chan et al., 2015), or even proteins expressed by specific SSC subpopulations like NESTIN (Méndez-Ferrer et al., 2010) and GREM1 (Worthley et al., 2015). Taken together, these studies agree on the widely reported perivascular niche of SSCs in the long bone marrow (**Figure 1**). Cells labeled with the CD146 marker are found in the adventitial layer of sinusoidal walls with typical reticular morphology, reminiscent of previously described ARCs (adventitial reticular cells) (Sacchetti et al., 2007).

Across studies, however, the distinctively identified populations vary in their multipotency, as some of them can generate in vivo osteoblasts, chondrocytes and stromal cells, but not adipocytes (Worthley et al., 2015). Rather than conflicting findings, these results point toward phenotypic heterogeneity of the SSCs, revealing that what were once generically called bone marrow mesenchymal stem cells actually comprise a number of subsets with subtly different properties. Many aspects regarding ontogeny and regulatory networks in long bone stromal SSCs have yet to be elucidated. Nevertheless, the knowledge accumulated over the years far exceeds that of stem cell populations serving the calvarial bones, which only recently started to be characterized in detail. Our aim here is to provide a comprehensive summary of recently reported stem cell populations within the cranial suture mesenchyme, their roles during postnatal turnover and wound repair and proposed mechanisms of regulation.

# LONG BONE SSCs: WHERE DO THEY COME FROM?

Embryonic development of the long bones occurs via endochondral ossification and begins with the condensation of mesenchymal cells derived from lateral plate or paraxial mesoderm (reviewed in Olsen et al., 2000). These cells differentiate into chondrocytes and lay down a cartilaginous anlagen, which is later replaced by the osteoblasts and bone formation (see **Figure 1**). Chondrocytes in the center enlarge to a hypertrophic state allowing bone growth and inducing ossification in the peripheral layer of undifferentiated cells around the cartilage, the perichondrium (Kronenberg, 2003). Secretion of vascular endothelial growth factor (VEGF) attracts blood vessels and chondroclasts, cells specialized in digesting the cartilage matrix, which will then invade the ossified bone collar (Arai et al., 2002). Upon apoptosis of the hypertrophic chondrocytes, the hollow cavities left in the matrix are invaded by blood vessels and osteoblasts, which will lay down bone matrix on top of a calcified cartilage scaffold to form primary spongy or trabecular bone. Once the marrow space is established, the bone is finally invaded by stromal osteoprogenitors, some of which are recruited to the perivascular niche of large caliber sinusoids where they will reside as bone marrow skeletal stem cells. Prior to invasion and formation of bone marrow stroma, these cells are reportedly derived from proliferating perichondrial precursors. Phenotypic similarity of cells in the perichondrium with primitive stromal cells was observed in early studies (Bianco et al., 1993). Their origin was later confirmed by immunostaining of the perichondrial marker ALCAM (activated leukocyte cell adhesion molecule) (ALCAM). FAC-sorted ALCAM positive cells showed characteristics of mesenchymal stem cells and ability to differentiate into all skeletal lineages (Arai et al., 2002). Moreover, osterix expressing perichondrial/periosteal cells, labeled prior to vascular invasion, give rise to trabecular osteoblasts, osteocytes and stromal cells inside the developing bone, confirming the unequivocal origin of skeletal stem cells in the undifferentiated perichondrium (Maes et al., 2010; Staines et al., 2013).

# ANATOMICAL AND DEVELOPMENTAL DIFFERENCES POINT TO DISTINCT STEM CELL NICHE IN CALVARIAL BONES

In contrast with the extensively studied stem cell niche in the endochondral bones, the source of SSCs in cranial bones was, until recent years, neglected or generally assumed to be identical to the long bones. Given the limited bone marrow space in these bones and very distinct developmental history, it is unsurprising that the local stem cell population would reside in a different compartment (**Figure 1**). The majority of craniofacial bones derive from the neural crest, with a few exceptions like mesodermally-derived parietal bones (Jiang et al., 2002). The latter will compose, together with crest-derived frontal bones, the top part of the skull called the cranial vault, or calvaria, which is fully made via intramembranous ossification. Although early mesenchymal condensations are also required in membranous bone formation (Hall and Miyake, 2000), these cells do not assume a chondrogenic fate or lay down cartilaginous scaffolds as in endochondral bones. Rather, upon expansion of the condensed mesenchyme, the cells in the center differentiate directly into osteoblasts, which will then secrete unossified matrix (osteoid) that will be later mineralized into compact bone. Growth of the immature bone then occurs at the leading edges of separate bones, dependent on proliferation and subsequent differentiation of osteoprogenitors at the commonly called osteogenic front (Rice et al., 2003). As the osteogenic fronts of two bones approach each other, they are interposed by undifferentiated mesenchyme, forming fibrous joints called sutures. Throughout embryonic and postnatal development, the sutures remain as an active site of bone formation in the expanding skull. The studies of Lana-Elola et al. (2007) helped to elucidate the mechanisms of bone growth and the contribution of different lineages within the suture environment. Using vital dye labeling, it was found that the primary mechanism for parietal bone growth is via proliferation and differentiation of leading edge osteoprogenitors with a surprisingly minor contribution of sutural mesenchymal cells, provided they are adjacent to the osteogenic fronts (Lana-Elola et al., 2007). Growth at the suture is also finely controlled by signaling from the underlying dura mater (Kim et al., 1998). Absence of dura mater leads to osseous obliteration of the coronal

suture, arresting the active growth at that site (Opperman et al., 1995). Moreover, the orientation of the dura mater under the cranial vault is critical for temporal control over suture patency or fusion (Levine et al., 1998). Growth and remodeling of the skull is also influenced by the overlying periosteum, a stratified membrane of mesenchymal cells including fibroblasts and osteoprogenitors (Brey et al., 2007). Mechanisms of periosteal activity during appositional growth and bone resorption of the cranial vault remain unclear. However, the membrane has been suggested to regulate osteogenesis via paracrine signaling (Cadet et al., 2003) and to provide osteoprogenitors that support craniofacial repair (Ochareon and Herring, 2011). Unlike the thoroughly described bone marrow-residing SSCs as the main source of osteoprogenitors during long bone repair (Zhou et al., 2014), the contribution of calvarial specific stem cells for cranial regeneration was unknown until very recently. Although a clonal population with multipotent MSC differentiation properties has long been isolated from rat calvaria (Grigoriadis et al., 1988), the visual evidence of the niche where the calvarial SSCs reside was only provided in the last few years.

# SSCs RESIDING IN THE CRANIAL SUTURE ARE THE MAJOR CONTRIBUTORS TO INJURY REPAIR

In the same way as BM-SSCs, calvarial stem cell research was largely propelled by the identification of specific markers expressed by resident populations. Only in the last 2 years, three populations were identified within the sutural mesenchyme and proposed as major calvarial skeletal stem cells, or subsets of it. Namely, Gli1 positive (Gli1+), Axin2-expressing (Axin2+) and postnatal Prx1-expressing (Prx1+) cells, as per definition in the original references (Zhao et al., 2015; Maruyama et al., 2016; Wilk et al., 2017). The authors used similar approaches to characterize the suspected SSC population with regards to location, stem celllike properties and other features shared by the cells described which we attempt to summarize in this section.

# Finding a Good Calvarial Stem Cell Marker

Zhao et al. (2015) hypothesized that Gli1 expressing cells were the MSCs for craniofacial bones, as they were for the incisor mesenchyme they reported earlier (Zhao et al., 2014). The zinc finger protein GLI1 is a well-known transcriptional effector of Hedgehog signaling (Hooper and Scott, 2005). Seidel et al. (2010) had previously proposed it as a dental stem cell marker upon discovery of Hh responsiveness (Gli1lacZ) from slow cycling cells of the dental epithelium (Seidel et al., 2010). The stem cell identity of these cells was later confirmed in the studies of Zhao et al. (2014). The AXIN2 protein, a negative regulator of Wnt signaling, was previously implicated in calvarial morphogenesis with Axin2 knockout mice showing craniosynostosis (Yu et al., 2005). Maruyama et al. (2011) described the essential role of AXIN2 in orchestrating the signaling network (Wnt, BMP, FGF) that regulates mesenchymal cell fate determination (Maruyama et al., 2011). The protein was recently proposed as a calvarial stem cell marker as Axin2-expressing cells in the midline of the suture mesenchyme were found slow-cycling in nature (Maruyama et al., 2016). Finally, the transcription factor Prx1 (also referred to as Prrx1) was previously shown to be highly expressed during limb bud formation and craniofacial development (Martin and Olson, 2000). Interestingly, PRX1 expression seemed to avert differentiation of early progenitors into committed osteoblasts, suggesting that the transcription factor could potentially mark SSCs in the calvarial bones (Lu et al., 2011). This hypothesis was successfully tested in the latest of the above-mentioned studies by Wilk et al. (2017).

# Distribution within the Cranial Mesenchyme

Gli1+ cells of the cranium are found in the entire periosteum, dura and suture mesenchyme at birth. However, from postnatal day 21 onwards these cells are gradually restricted to the cranial sutures where they remain throughout adulthood (Zhao et al., 2015). Axin2 was previously found to be expressed in the midline of sagittal suture skeletogenic mesenchyme. In a more recent study, Axin2 expression was reported in all calvarial sutures, colocalizing with slow cycling cells that retained EdU (Maruyama et al., 2016). Unlike the Gli1+ cells, no obvious mention or visual evidence of early expression in the dura or periosteum is seen. In fact, Axin2 expression is reportedly restricted to the midline of the sutural mesenchyme from postnatal day 10, whereas Gli1+ tracing seems to target a larger area within the suture. Wilk et al. (2017) investigated postnatal Prx1+ cells and found them exclusively in the calvarial sutures from 2 weeks of age until late adulthood. Like Axin2-expressing cells, Prx1+ were also not found in the periosteum or dura mater, although no mention or visual evidence of early postnatal expression is given (as in Zhao et al., 2015). Overall, the markers described seem to identify cells in a common niche. Whether the populations overlap in identity and function, or represent distinct functional subsets of a main stem cell source is still under debate as further discussed below. No expression of the above mentioned markers was detected in the posterior frontal suture, which likely correlates with the fact that this is the only calvarial suture fused at that stage in mouse development.

# Self-Renewal and Ability to Generate All Calvarial Tissues

As previously mentioned in this review, self-renewal ability and in vivo multipotency are the cardinal requirements of a bone fide stem cell. While the in vivo differentiation potential can be assessed in many ways, some authors argue that selfrenewal should only be assessed via heterotopic transplantation of non-induced cultures (Bianco and Robey, 2015). A typical SSC after being cultured and clonally isolated should still be able to generate all the skeletal compartments (multipotency) and "a cell compartment anatomically, phenotypically and functionally equivalent to the one originally explanted" (selfrenewal) (Bianco and Robey, 2015). This remains the mainstay of MSC investigation criteria, as it honors the concept of stem cell autonomy in making, without the need for any artificial cues, organized skeletal tissue, including a reservoir of stem cells. In the case of calvarial stem cells, this would mean the ability to make in vivo, after culture and clonal isolation, all the components of a membranous bone and a mesenchymal equivalent to the cranial sutures.

Although heterotopic transplantation of putative calvarial SSCs was not the approach taken, the evidence presented in the studies reviewed herein strongly supports the ability of the putative stem cells to generate all calvarial compartments in vivo. Firstly, single pulse labeling (induced by tamoxifen in Gli1-CreERT2 and Prx1-creER mice, or doxycycline in Axin2 cre-Dox) provides that only the cells initially labeled within the sutural mesenchyme and their progeny are being tracked (Zhao et al., 2015; Maruyama et al., 2016; Wilk et al., 2017). In all studies, the co-expression of osteogenic markers like runx2, osterix, collagen type 1, osteocalcin and sclerostin was not observed in the labeled suture mesenchyme, confirming that the proposed stem cell populations were undoubtedly distinct from any postnatal preosteoblasts and mature osteoblasts. Gli1+ cells were permanently labeled at 1 month of age and found increasingly abundant in the suture mesenchyme, periosteum, dura mater and parts of the calvarial bones up to 8 months of age (Zhao et al., 2015). In a similar manner, derivatives of the Axin2+ cells permanently labeled at 1 month, accumulated continuously in the sagittal suture until after 1 year of development, with a small number of cells found embedded in the bone matrix near the osteogenic fronts. However, this study did not report any contributions of the cells in question to the mature bone and surrounding tissues (dura mater and periosteum). Conversely, a large contribution of Prx1+ cells to all calvarial tissues, except bone marrow osteoblasts, was observed in 6–8 week old mice (Wilk et al., 2017). Prx1 expression is detected as early as embryonic day 15, if not earlier, in the developing skull (Ouyang et al., 2013). Unfortunately, the conclusions about postnatal Prx1+ contribution to adult tissues were drawn from constitutive Prx1-cre lineage tracing analysis. This could include, although unlikely, cells that were specified immediately before the formation, or turn-over of the calvarial components analyzed, without definitive proof that they came from the sutural niche. Moreover, although Zhao and Wilk make no specific claims about self-renewal properties of the stem cell populations reported, Maruyama goes so far as to say that the accumulation of Axin2+ cells within the 1 year period reflects a self-renewal capability of those cells (Maruyama et al., 2016); however, this could also be due to a high proliferation capacity (Bianco and Robey, 2015). Nonetheless, Maruyama provides rigorous and more compelling evidence of in vivo stem cell behavior by transplanting isolated multi-colored (R26RConfetti) Axin2<sup>+</sup> cells to an extra-skeletal site (kidney capsule), which the other authors did not attempt. Ectopic bones labeled by single fluorescent colors were observed 3 weeks after transplantation, indicating that these cells can clonally expand in vivo and differentiate autonomously into the osteogenic lineage (Maruyama et al., 2016). However, contrary to bone marrow residing skeletal stem cells, Axin2<sup>+</sup> SCs from the suture were not able to generate cartilage, unless artificially stimulated by BMP2, which drifts away from the principle of non-cued multipotency toward

skeletal lineages expected from bona fide SSCs. While this could be a specific feature of membranous bone residing stem cells, ectopic chondrogenesis has been previously reported in the sutural mesenchyme, suggesting that they possess the intrinsic ability to differentiate into the chondrogenic lineage (Grigoriadis et al., 1988; Maruyama et al., 2011).

# MSC In vitro/Ex vivo Behavior

Aside from rigorously established criteria, MSCs are popularly defined by their in vitro culture properties, including the ability to differentiate into various cell types (osteoblasts, chondrocytes, adipocytes and, arguably, neurons) (Keating, 2012). Likewise, although with considerable disagreement within the stem cell field, MSCs are loosely defined based on the expression profile of numerous surface markers (Dominici et al., 2006). Typical MSC markers CD90, CD73, CD44, Sca1, and CD146 were not detected in the majority of Gli1<sup>+</sup> cells in vivo. However, after 1 week of culture, fluorescent-activated Gli1<sup>+</sup> sorted cells express high levels of such MSC markers. Gli1<sup>+</sup> cells are also clonogenic in vitro and subcultured clones were successfully differentiated into osteo-, chondro-, and adipogenic lineages under appropriate conditions (Zhao et al., 2015). With the exception of the recently reported Leptin receptor (Zhou et al., 2014), none of the bone marrow SSC markers investigated by Maruyama was distinctively expressed in the Axin2<sup>+</sup> cells of the suture (Mcam/CD146, Nestin and Gremlin1). Moreover, the in vitro differentiation ability of these cells was not assessed in this particular study. Axin2<sup>+</sup> isolated cells were, nevertheless, able to form colonies in vitro and to differentiate, after heterotopic transplantation, into the osteogenic lineage and into chondrogenic, although only upon external stimulation (BMP2), as previously mentioned (Maruyama et al., 2016). Postnatal Prx1<sup>+</sup> cells are also clonogenic and exhibit a SSC-like profile ex vivo, including: upregulation of SSC markers Pdgfrα and Mcam/CD146; downregulation of cell cycle and DNA replication markers Ccne2, Mcm4, and Pcna typical of quiescent cells; and high levels of transcripts associated with stem cell homing (Itga2, Itga3, and Itga6). Upon in vitro and in vivo stimulation with recombinant WNT3A, these cells are pushed toward osteogenic differentiation, judging by upregulated Osx and Col1 and reduced expression of chondrogenic markers Sox9 and Col2 (Wilk et al., 2017). Altogether, the variable approaches, contrasting profiles and requirement of cueing factors for multi-differentiation in these studies points to the unreliability of loosely defined criteria to identify SSCs, based on which virtually every tissue can provide "stem cell" equivalents.

# Contribution to Repair

The notion that skeletal stem cells work as postnatal contributors for tissue regeneration represents not only a general understanding of the organ physiology, but the very motivation for cautious and rigorous investigation of the stem cell populations, aiming at their translational potential. Although classically established requirements for SSCs do not include their regenerative properties, more than it does their multi-lineage differentiation ability in the context of organogenesis and postnatal turn-over, investigating the role of these cells during injury repair is essential, as the pursuit for stem cell-based therapies demand such knowledge. Therefore, the mechanisms of SSC recruitment and contribution to calvarial wound repair should accompany the research of any reported populations, as the three current studies highlighted here have done well to address. Gli1<sup>+</sup> cells respond immediately to injury in the suture by rapidly activating proliferation within 24 h post incision at the sagittal midline. Contribution of Gli1<sup>+</sup> cells to calvarial repair is observed in the vast infiltration of conditionally labeled cells into a 1 mm wide parietal wound situated 2 mm away from the sagittal suture. Two weeks after the injury is made, the majority of cells in the area are strongly labeled and, at 1 month, the repaired bone is mostly composed of Gli1<sup>+</sup> osteocytes as well as positively labeled periosteum and dura mater. Orthotopic transplantation of a lineage traced calvarial bone piece containing sagittal suture into a 4mm defect, revealed a significant number of labeled cells within 1 month-regenerated periosteum, dura mater and bone, whereas an explant lacking the suture failed to generate any of these components. Even with intact dura mater and periosteum, calvarial explants devoid of suture were incapable of regenerating the wounded bone in the recipient mice. The results indicate that, contrary to the previous idea that progenitors involved in cranial repair mainly resided in the periosteum, the sutural mesenchyme and not the surrounding membranes possesses the regenerative capacity (Zhao et al., 2015). Similarly to the Gli1+ population, Axin2+ lineage traced cells here drastically expanded in the sutural mesenchyme in response to a 1.4 mm parietal bone defect. Four weeks after the surgical procedure, large infiltration of the injury site by Axin2<sup>+</sup> cells is observed and co-localization with OSX (osteoprogenitors) and SOST (osteocytes) expression strongly indicates direct contribution to bone repair. Axin2<sup>+</sup> cells were also shown to improve injury repair when directly transplanted into the wound site. While Axin2 negative cells from the suture mesenchyme provide no better healing than that seen in non-transplanted injuries, Axin2<sup>+</sup> cell transplantation significantly increased the repaired area 2 and 4 weeks after surgery. As in the previous experiment, the lineage labeling overlapped with markers of osteoprogenitors and osteocytes, confirming direct contribution of the Axin2+ population to repair of the wounded area (Maruyama et al., 2016).

Finally, Prx1<sup>+</sup> cells actively contributed to regeneration of subcritical size defects in both frontal and parietal bones, although in this study a smaller 100µm wide injury was made. Nevertheless, lineage traced Prx1<sup>+</sup> cells were increasingly present in the wound area 5, 10, and 30 days post-surgery, when they co-localized to osteoblasts and osteocytes embedded in the newly formed bone. Moreover, the regeneration of, questionably, critical-sized (2 mm) defects was greatly improved by heterotopic transplantation of minced sutures into the wound, where Prx1 labeling was seen in the majority of cells integrating the repaired bone 4 weeks after surgery. Interestingly, the repair of parietal subcritical defect did not occur in suturectomized mice after 4 weeks, whereas the 0.5 mm wound is fully repaired in mice with intact sagittal and coronal sutures (Wilk et al., 2017). Taken together, these studies confirm the unequivocal and potentially exclusive contribution of the sutural mesenchyme to calvarial injury repair.

# Requirement for Development, Homeostasis and Injury Repair

Overall, the studies described provide compelling evidence for direct (by providing osteoprogenitors) as well as indirect (via paracrine signaling) contribution of the putative stem cell populations to calvarial repair. Predominant infiltration of Gli1+/Axin2+/Prx1<sup>+</sup> cells is seen in all the subcritical size defect assays, whereby the healing process should be accomplished by natural mechanisms occurring during normal bone physiology. Moreover, transplantation of these cells to the wound site showed significant repair improvement in all the attempts described in the previous section, supporting the translational potential of calvarial SSCs in a number of clinical scenarios. In two of the studies, notwithstanding, the authors found it appropriate to investigate whether the stem cell population described was indispensable for skull development, homeostasis and injury repair. While this is not a particular requirement for determining a resident stem cell, the findings provide valuable information on the ontogeny of the sutural niche and other calvarial compartments, as well as potential mechanisms of congenital disorders like craniosynostosis. Conditional ablation of Gli1<sup>+</sup> cells, for instance, led to severe craniofacial phenotypes in Gli1-CreERT<sup>2</sup> ; DTAflox/flox mice (diphtheria toxin A). When ablation was induced at 1 month of age, fusion of ordinarily patent coronal and frontal-premaxilla sutures was observed after 1 month, leading to general reduction of DTA-mice body size at 2 months post-induction. At this stage, all the craniofacial sutures were fused and the bones of Gli1<sup>+</sup> ablated mice presented severe osteoporosis. Repair of 1mm parietal defects was also compromised in DTA mice when compared to fully regenerated controls. Altogether, the results indicate that Gli1<sup>+</sup> cells are largely required for craniofacial bone turnover and injury healing. Moreover, pathological fusion of the cranial sutures in Gli1-ablated mice suggests that a deficient supply of undifferentiated mesenchymal cells hinders the patency expected in calvarial sutures of adult mice (Zhao et al., 2015).

Two different strategies confirmed Prx1<sup>+</sup> SSCs requirement for injury repair. Firstly, 8 week-old Prx1-creER; Rosa26DTA/<sup>+</sup> mice were used to perform global ablation of Prx1<sup>+</sup> cells during 5 days before and 5 days after 0.5 mm parietal wound surgery. Twenty 8 days later, micro-computed tomography and histological analysis showed drastic healing impairment in Prx1<sup>+</sup> ablated mice against complete regeneration in the controls. Even in mice injured 2 months after ablation, parietal defects failed to regenerate as non-ablated mice. Secondly, excision of sagittal and right coronal sutures yielded non-regenerated wound in the right parietal bone, whereas removal of the opposing sutures (frontal and left coronal) did not affect the healing of parietal subcritical defect. This strongly suggests that calvarial bones depend on the SSCs of adjacent sutures to perform injury repair. Unlike Gli1<sup>+</sup> SSCs, however, global ablation of postnatal Prx1<sup>+</sup> cells did not result in craniosynostosis or any other major craniofacial phenotype. Indeed, Gli1 expression is thought to identify a broadly distributed population of stem cells in the suture mesenchyme, of which Prx1+ cells constitute a subset seemingly dispensable for postnatal development and turnover. Ablation of embryonic Prx1+ cells leads to incomplete calvarial bone formation, suggesting that this specific subpopulation might only be required in earlier stages of development (Wilk et al., 2017). Whereas Maruyama did not perform any Axin2<sup>+</sup> SSC ablation studies, Axin2-null mice are known to exhibit premature fusion of the cranial sutures (Yu et al., 2005), although this does not necessarily correlate with any particular requirement of the population concerned. Reduction of Axin2-expressing cells is seen upon ossification of the frontal suture (Maruyama et al., 2016). However, this is likely an effect on the entire undifferentiated mesenchyme at the interfrontal suture as it commits to the osteogenic lineage and promotes the fusion of the osteogenic fronts.

# Aspects of Regulatory Signaling

Throughout the skeleton, as a local niche is established, mesenchymal stem cells are subject to recruitment, proliferation, induction of mitotic quiescence and lineage commitment. Over the years, extensive research of bone marrow SSCs has shown that these steps are regulated by largely convoluted signaling pathways including: platelet-derived growth factor (PDGF) during perivascular recruitment (Sacchetti et al., 2007); parathyroid hormone (PTH) during expansion of the bone marrow stroma (Kuznetsov et al., 2004); TGF-β1 as the cells are maintained in a quiescent state (reviewed in Pepper, 1997; Jain, 2003) and extensive crosstalk between TGFβ/BMP, Wnt/βcatenin, FGF and Hedgehog pathways, which generally control MSCs lineage fate decision (reviewed in Cook and Genever, 2013). Given the complexity of the network necessary for developing and maintaining a stem cell niche, it is no surprise that regulatory signaling receives little attention in recent papers that attempt to describe calvarial SSCs. In fact, most of the preliminary evidence regarding regulation of calvarial SSCs arose from in vitro or unconvincing in vivo assays that were based on mechanisms previously established for stem cells in the long bones. Zhao et al. set out to investigate whether Gli1<sup>+</sup> cells were, like BM-SSCs, regulated by hedgehog signaling in the context of lineage commitment. No Sonic hedgehog (Shh) signal was detected in the suture region using a R26Tdtomato reporter. However, Indian hedgehog (Ihh), a regulator of skeletal development, was found in cells of the osteogenic fronts flanking the cranial sutures. Deletion of the hedgehog receptor Smoothened (Smo) in Gli1<sup>+</sup> cells did not affect suture patency, proliferation of Gli1<sup>+</sup> cells, or their ability to generate periosteum, dura and osteocytes; nor did it induce apoptosis in the sagittal suture. Nonetheless, all the craniofacial bones in SmoKOGli1−cre mice exhibited severe osteoporosis and reduced bone volume after 8 months. In vitro analysis of sutural mesenchyme cells revealed up-regulated Gli1 expression, increased osteogenic activity and upregulation of osteodifferentiation markers upon IHH treatment, whereas the hedgehog inhibitor GDC0449 induced the opposite effects. Neither treatment affected proliferation or apoptosis of cultured mesenchymal cells, suggesting that Hedgehog signaling might be important for regulation of Gli1<sup>+</sup> SSC differentiation, but not their maintenance (Zhao et al., 2015). Maruyama and Wilk's insights on regulation of Axin2<sup>+</sup> and Prx1+ cells, Doro et al. Calvarial Stem Cells and Repair

respectively, are merely suggestive, as these cells present in vivo expression of canonical Wnt inhibitors in their quiescent state and lineage fate decision changes upon exogenous activation of Wnt or BMP signaling. Transplanted Axin2<sup>+</sup> cells were able to generate ectopic mineralized bone, but no cartilage after 7 days, contrasting the multi-lineage potential of long bone marrow SSCs. However, exogenous stimulation with BMP2 shifted Axin2<sup>+</sup> cells from an osteogenic to a chondrogenic fate, shown by detection of Alcian blue (cartilage), but not von Kossa (bone) staining in the structure generated (Maruyama et al., 2016). Expression of Wnt inhibitors Dkk1 and Sost by Prx1<sup>+</sup> sorted cells supports the well-established idea that skeletal progenitors are maintained in an undifferentiated status, or yield chondrogenic differentiation in the absence of canonical Wnt activation (Day et al., 2005; Hill et al., 2005). Indeed, in vivo expression of chondrogenic markers Sox9 and Col2 was diminished in pnPRX1<sup>+</sup> cells upon subperiosteal injection of Wnt agonist WNT3A to the cranial sutures, whereas osteogenic markers Osx and Col1 were significantly upregulated. Commitment to the osteogenic fate has also observed when these cells were treated in culture with recombinant WNT3A, endorsing the previously described role of Wnt/β-catenin pathway in differentiation of mesenchymal progenitors (Wilk et al., 2017). Altogether, recently published studies provide informative, yet inconclusive findings to characterize molecular regulation of calvarial SSCs, leaving exciting open prospects for future stem cell research.

# DISCUSSION

As once wisely stated, "All definitions of stem cells are functional in nature, and all types of stem cells are defined by functional assays" (Bianco et al., 2013). That is, one must see that hasty and careless attempts to identify stem cells in the adult body will not prompt us to replace rigorously established criteria with simple surface marker profiling, or loosely defined in vitro properties. From the very concept of stemness, proposing a stem cell candidate requires that a single cell within the niche described can generate multiple and fully differentiated tissues in vivo and produce a reservoir of cells with identical phenotype and multipotency. At the risk of raising unrealistic expectations for the clinical use of stem cells, many authors over the years claimed to have isolated post-natal fibroblastic populations from virtually every tissue with differentiation abilities that extrapolate the tissue-specific potential of the identified progenitor, much like an embryonic stem cell (da Silva Meirelles et al., 2006). This apparent "pluripotency" attributed to putative post-natal stem cells is often based on chemically induced in vitro differentiation studies (Robey et al., 2014), which more accurately describe the artificial reprogramming of a cell than its in vivo potency. The very use of the term mesenchymal stem cell (MSC) to describe bone marrow stromal progenitors is misleading in that it is based on the erroneous and unproven hypothesis that stem cells in the bone marrow could generate various differentiated tissues beyond the skeletal lineages. It also assumes that MSCs throughout the body have a common developmental origin (Caplan, 1991) and can be found in a number of different organs sharing the same differentiation potential as BMSCs and, as previously hypothesized, equivalent perivascular niche (Chen et al., 2012). In reality, evidence originating from more stringent in vivo assays indicates that the bone marrow harbors stem cells exclusively for skeletal-tissues, namely, bone, cartilage, adipose tissue and stroma (Owen and Friedenstein, 1988; Sacchetti et al., 2007). Therefore, as previously proposed (Bianco et al., 2008), "skeletalstem cell" is a more adequate designation, for it describes the real potential of the so called bone marrow stromal cells and implies that stem cell populations identified in other tissues are not expected to behave exactly like SSCs in the long bones. In fact, the concept of tissue-specific stem cells as opposed to a homogeneous MSC population with multiple locations, accords with a more reasonable prediction that most adult tissues would maintain a stock of undifferentiated yet committed progenitors to support physiological turn-over and regeneration in response to injury.

The search for a calvarial stem cell then must indeed be grounded on very stringent criteria. However, the assessment of in vivo multipotency and self-renewal in this population need not be guided inflexibly by the sequential assays classically employed for BM-SSCs investigation (Bianco et al., 2013; Bianco and Robey, 2015), lest one must conclude that: either the membranous bones in the skull lack a resident stem cell population, or the well supported contribution of sutural mesenchymal progenitors to all calvarial components reflects a distinct role/mechanism of skeletal turn-over and injury repair, even though the same is performed by bone marrow SSCs in the long bones. For instance, it is fair to require that in vivo multipotency is assessed exclusively by heterotopic transplantation of clonally isolated calvarial cells rather than alternate in vitro assays. But, to expect the formation of an ectopic ossicle with all the compartments of an endochondral bone, contradicts the idea of organ-specific stem cells and resonates with the old farfetched concept of "pluripotent" MSCs. Moreover, the ability to self-renew is easily observed in ectopically transplanted BM-SSCs, as these cells generate organized bone marrow tissue and reconstitute a pool of identical undifferentiated progenitors (Sacchetti et al., 2007). However, since the stem cells committed to an intramembranous fate will not generate the compartments produced during an endochondral ossification, a different readout must be required as proof of self-renewal ability. Maruyama has shown that ectopically transplanted Axin2<sup>+</sup> sutural SCs can clonally expand in vivo, but also generate bony structures, judging by monochromatic expansion of R26-confetti traced cells (Maruyama et al., 2016). Whether this could be taken as convincing evidence of stemness is still debatable, although it is surely more rigorous than otherwise defined in vitro MSC criteria (Dominici et al., 2006). Also, lineage-tracing approaches like the ones employed for the calvarial stem cell niche investigation might not be the mainstay for assessing bona fide stem cell criteria. Nonetheless, the conclusions drawn with regards to multi-lineage differentiation ability and injury repair contribution in vivo are well supported with single-pulse conditional labeling methods.

In summary, the stringent criteria defended by "rigorous in vivo assay" apologists stems from a genuine concern about negligent clinical application of poorly characterized stem cell candidates. Experienced stem cell scholars report the uprising of unrealistic expectations of systemic regenerative therapy using bone marrow stromal cells, when in fact, effective translational success, to date, was always obtained from local transplantation of organ-specific progenitors [e.g., bone marrow orthotopic transplantation (Horwitz et al., 2002); skin regeneration (Green, 1989; De Luca et al., 2006); cornea regeneration (Pellegrini et al., 1997; Rama et al., 2010)]. More importantly, stem cell investigation must be guided by translational demands, in which the primary ambition is to find: cells that can be easily isolated, whether prospectively (based on surface markers), or by plastic adherence of clonal populations; cells that can be clonally expanded in vitro; cells that can be locally and not systemically delivered and that will directly engraft or at least support regeneration by establishing a healing microenvironment; or ideally cells that can be regulated in vivo by pharmacological stimulation. With that in mind, the recent discovery of the calvarial stem cell niche is well evidenced and presents putative populations that fulfill the basic stem cell criteria, as dictated by translational demands. Whereas Gli1<sup>+</sup> cells are distributed across a large portion of the sutural mesenchyme, expression of Axin2 and Prx1 seems to identify more discrete subsets that may or may not be part of a major population (Wilk et al., 2017). Partial overlapping of the proposed markers and slight differences in functional assays should not imply contradictions in calvarial stem cell properties. Rather, the suture mesenchyme most likely comprises a heterogeneous niche in which distinct stem cell subsets can be identified, isolated and characterized with clinical application purposes. Since all recent studies reported in vitro clonogenicity of the candidate populations described, a step forward in understanding their biology and potency would be to transplant the expanded colonies to an ectopic site, as the approach addresses the autonomy of the cells in a nonskeletogenic environment.

In humans, while the existence of a calvarial niche residing in the sutural mesenchyme has not been examined, expression of these putative stem cell markers (Gli1, Axin2, and Prx1) is

# REFERENCES


well predicted based on fetal studies (Homayounfar et al., 2015), analyses of pathogenic skull bones (Coussens et al., 2007) and on our knowledge of signaling pathways crucial in calvarial development. Clearly, the calvarial stem cell niche needs to be further defined in humans if we are to harness these cells for improved skull repair. Isolation and thorough analysis of these cells will then allow us to design new therapies, including introduction of synthetic scaffolds designed to enrich for these cells, or carrying chemotactic cues for to attract these cells to the wound site. Moreover, from the perspective of future clinical translation, two potential autologous stem cell sources are the bone marrow and adipose-derived stem cells. Both of these are heterogenous populations which could be further examined for expression of Gli1, Axin2, and Prx1 positive sub-populations. These markers have not been well characterized in human stem cell populations, although the relevant signaling pathways are implicated in cancer stem cells.

Finally, although the mesenchyme interposing all calvarial bones is predominantly neural crest derived, very little attention was given in the studies reviewed to the distinct embryonic origins of frontal and parietal bones and how this might determine how surrounding stem cells are regulated. Even the mesoderm derived bone is hypothesized to require neural crest (Sox10+) lineages residing in the periosteum (Isern et al., 2014). Thus, given the inherent multi-potency of the neural crest cells, osteoblasts derived from that lineage may provide a more plastic milieu than the mesoderm.

# AUTHOR CONTRIBUTIONS

DD, AG, and KL sketched out the initial draft of the review. DD wrote the review and designed the figure. AG and KL revised and edited the review.

# FUNDING

This work was funded by a Brazil CAPES studentship (DD) and BBSRC - BB/I021922/1 (KL).


**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 FW and handling Editor declared their shared affiliation.

Copyright © 2017 Doro, Grigoriadis and Liu. 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) or licensor 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.

# Monitoring Notch Signaling-Associated Activation of Stem Cell Niches within Injured Dental Pulp

Thimios A. Mitsiadis <sup>1</sup> \*, Javier Catón<sup>2</sup> , Pierfrancesco Pagella<sup>1</sup> , Giovanna Orsini <sup>3</sup> and Lucia Jimenez-Rojo<sup>1</sup>

<sup>1</sup> Orofacial Development and Regeneration, Faculty of Medicine, Institute of Oral Biology, ZZM, University of Zurich, Zurich, Switzerland, <sup>2</sup> Department of Medical Basic Sciences, Faculty of Medicine, University CEU-San Pablo, Madrid, Spain, <sup>3</sup> Department of Clinical Sciences and Stomatology, Polytechnic University of Marche, Ancona, Italy

Dental pulp stem/progenitor cells guarantee tooth homeostasis, repair and regeneration throughout life. The decision between renewal and differentiation of these cells is influenced by physical and molecular interactions with stromal cells and extracellular matrix molecules forming the specialized microenvironment of dental pulp stem cell niches. Here we study the activation of putative pulp niches after tooth injury through the upregulation of Notch signaling pathway. Notch1, Notch2, and Notch3 molecules were used as markers of dental pulp stem/progenitor cells. Upon dental injury, Notch1 and Notch3 are detected in cells related to vascular structures suggesting a role of these proteins in the activation of specific pulpal perivascular niches. In contrast, a population of Notch2-positive cells that are actively proliferative is observed in the apical part of the pulp. Kinetics of these cells is followed up with a lipophilic DiI labeling, showing that apical pulp cells migrate toward the injury site where dynamic regenerative/repair events occur. The knowledge of the activation and regulation of dental pulp stem/progenitor cells within their niches in pathologic conditions may be helpful for the realization of innovative dental treatments in the near future.

Edited by:

Gianpaolo Papaccio, Università degli Studi della Campania "L. Vanvitelli" Naples, Italy

### Reviewed by:

Jean-Christophe Farges, Claude Bernard University Lyon 1, France Zhi Chen, Wuhan University, China

> \*Correspondence: Thimios A. Mitsiadis thimios.mitsiadis@zzm.uzh.ch

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

> Received: 03 April 2017 Accepted: 18 May 2017 Published: 30 May 2017

### Citation:

Mitsiadis TA, Catón J, Pagella P, Orsini G and Jimenez-Rojo L (2017) Monitoring Notch Signaling-Associated Activation of Stem Cell Niches within Injured Dental Pulp. Front. Physiol. 8:372. doi: 10.3389/fphys.2017.00372 Keywords: stem cells, dental pulp, niches, tooth injury, regeneration, cell proliferation, Notch signaling, Notch2

# INTRODUCTION

Biological repair and regeneration is an attractive alternative and/or complement to prosthetic replacement of tissues and organs. Cell-based therapeutic approaches consist of in vitro manipulation of stem cells and their consequent administration to patients (Passier et al., 2008; Segers and Lee, 2008; Djouad et al., 2009; Robinton and Daley, 2012; Shevde, 2012; Bender, 2016). Stem cells are defined by their dual capacity of self-renewal and multipotency (referred to as stemness) (Thomson et al., 1998; Shevde, 2012). These properties make stem cells extremely interesting for clinical tissue engineering applications (Bianco and Robey, 2001; Mitsiadis et al., 2012; Aurrekoetxea et al., 2015; Mele et al., 2016). Stem cells have been identified in the pulp of deciduous and adult permanent teeth (Mitsiadis et al., 2015; Miran et al., 2016). These cells are able to differentiate both in vivo and in vitro into many cell types such as odontoblasts, osteoblasts, chondrocytes, adipocytes, and neuronal cells (Bluteau et al., 2008; Mitsiadis et al., 2015). There is increasing evidence for the existence of more than one stem/progenitor cell populations within the dental pulp (Mitsiadis et al., 2011; Ducret et al., 2016).

Dental injuries are often lethal for the odontoblasts at the proximity of the lesion site, an event that triggers activation of dental pulp stem/progenitor cells. These cells proliferate, migrate, and finally differentiate into odontoblast-like cells that elaborate the reparative dentin (Mitsiadis and Rahiotis, 2004). However, the nature and exact location of these mesenchymal cell populations are not yet known. Niches consist of specific and protected anatomic locations housing stem/progenitor cells and enabling them to self-renew. Stromal cells belonging to a niche control stem cell behavior through cell-cell interactions, soluble factors, and specialized extracellular matrices (Scadden, 2006; Djouad et al., 2009; Shaker and Rubin, 2010; Oh and Nör, 2015; Pagella et al., 2015). This particular microenvironment permits stem/progenitor cells to survive, to change their number and fate, regulating thus their participation in tissue maintenance, repair and/or regeneration. Therefore, it is essential to identify stem cell niches within the dental pulp in order to understand the mechanisms and the microenvironment that support the survival of stem/progenitor cells in teeth.

Notch molecules are important regulators of the stem cell fate, with capacity to induce cell proliferation and/or differentiation (Hori et al., 2013). The close association of dental pulp mesenchymal cells and neo-vessels in dental diseases (e.g., carious lesions, injuries) and their relation to Notch signaling pathway may be critical in the regulation of stem cells to differentiate into odontoblast-like cells (Lovschall et al., 2007; Mitsiadis et al., 2011; Oh and Nör, 2015). Notch proteins form a family of evolutionary conserved transmembrane receptors that determine cell fate (Artavanis-Tsakonas and Muskavitch, 2010). In mammals, the four Notch receptors (i.e., Notch1, Notch2, Notch3, Notch4) are activated following direct contact with their five ligands: Jagged1 (Jag1), Jag2, Delta-like1 (Dll1), Dll3, Dll4. Upon ligand-receptor binding, the Notch protein is cleaved and its intracellular domain (NICD) translocates to the nucleus, where it associates with the DNA binding protein RBP-Jk to activate transcription (Artavanis-Tsakonas and Muskavitch, 2010; Hori et al., 2013). It has been reported that Notch activation promotes stem cell maintenance (Androutsellis-Theotokis et al., 2006; Artavanis-Tsakonas and Muskavitch, 2010).

Although Notch signaling has been exhaustively studied during tooth development, pathology and repair, its role in regulating the behavior of dental pulp stem/progenitor cells after injury remains elusive (Mitsiadis et al., 1999, 2003; Sun et al., 2014). To address such questions, experiments using loss- and/or gain-of-function transgenic animal models are necessary. In the present manuscript, as a first attempt to investigate this issue, we studied the correlation between the expression of Notch receptors and migration of apical pulp cells upon injury at the tooth crown.

# MATERIALS AND METHODS

# Tissue Preparation and Dental Explant Cultures

All mice (C57Bl/6; postnatal day 6–8) were maintained and handled according to the Swiss Animal Welfare Law and in compliance with the regulations of the Cantonal Veterinary office, Zurich (License 11/2014). The licensing committee of the Gesundheltsdirektion Kanton Zürich approved all experimental protocols (Versuch Nr. 11/2014 "Study of the function and potency of human and mouse dental stem cells after in vitro culture"). Mice were sacrificed by cervical dislocation and their mandibles were dissected out and cultured in vitro. Cavity preparations were performed in 20 first mandibular mouse molars using a 20-gauge drill, while 4 intact molars were used as a control. Four hours after the operation, the first molars were carefully extracted from the mandibles. Mandibles and dental explants were cultured in a medium consisted of Dulbecco's Modified Eagle Medium (DMEM; GibcoBRL), 20% fetal calf serum (FCS; GibcoBRL), 20 units/ml penicillin/streptomycin (GibcoBRL), and glutamine at 37◦C/5% CO2. Most of the molars were used for DiI labeling experiments, cell proliferation essays, and whole mount immunohistochemistry, while other molars were sectioned with a cryostat (12–14µm sections) and used for immunohistochemistry and cell proliferation essays. Molars from more aged mice could not be used, as it was impossible to remove the hard mineralized tissues (i.e., dentin, enamel) without causing sparse damage to the pulp tissue. Cultures were maintained for at most 4 days, since pulps cultured for longer periods showed clear signs of pulp necrosis.

# Immunohistochemistry and Immunofluorescence

Polyclonal antibodies against the mouse Notch1, Notch2, and Notch3 proteins were used (Mitsiadis et al., 1995, 1999, 2003). Immunohistochemistry on sections were performed as described previously (Mitsiadis et al., 1995, 1999, 2003). Briefly, the antibodies were applied for 2 h at 37◦C, thereafter the sections were incubated with the biotinylated secondary antibody for 1 h and finally with peroxidase-conjugated streptavidine for 10 min. Peroxidase was revealed with 3-amino-9-ethylcarbazole (AEC) reaction solution. In control sections the primary antibodies were omitted.

For immunofluorescence on the entire dental pulp, tissues were first fixed in 1% paraformaldehyde (PFA) at 4◦C overnight and then washed with phosphate-buffered saline (PBS)/0.2% bovine serum albumin (BSA)/0.3% Triton X-100 overnight. Samples were then incubated with Notch2 primary antibody during 5 days at 4◦C. Thereafter, tissues were washed with PBS and post-fixed 30 min at RT in 4%PFA and incubated in HCl (2 M) for 30 min at 37◦C prior to the incubation with the anti-BrdU antibody for 7 days at 4◦C. Secondary antibodies were added separately and each of them incubated during 3 days at 4 ◦C. Samples were then mounted with 2.5% 1,4-diazobicyclo- [2.2.2]-octane (DABCO, Sigma, D2522) mounting medium and analyzed using a Leica SP5 confocal microscope.

# Cell Proliferation Analysis

Cell proliferation in cultured dental explants was analyzed by using a bromodeoxyuridine (BrdU) cell proliferation kit (Boehringer Mannheim). The explants were cultured for an additional time (i.e., 2–4 h) with BrdU, according to the manufacturer's instructions. BrdU-positive cells were monitored after staining with an anti-BrdU antibody and detected by means of a Vectastain ABC kit (Vector). Whole mount immunohistochemistry was performed as earlier described (Mitsiadis et al., 1995).

# DiI Labeling and Fate Mapping of Dental Cells

Immediately after extraction, cells located at the apical part of dental pulp of both intact and injured first molars were labeled with DiI (Molecular probes cell tracker CM-DiI, C-7000). DiI is a lipophilic dye that intercalates in the cell membrane marking small groups of cells. DiI was prepared in ethanol (EtOH) at 2.5µg/µl. This stock solution was then diluted 1–9 in 0.2 M sucrose and warmed. DiI was injected by a mouth-controlled borosilicate glass micropipette. After DiI labeling, dental explants were cultured for 4 days. Thereafter, explants were fixed with 4%PFA in Dulbecco's PBS at 4◦C for 2 h. After fixation, the mineralized part of the majority of the teeth was carefully removed in order to obtain dental pulp tissues free of dentin and enamel. The fate of the labeled cells was assessed in transmitted light and fluorescence images, which were captured with a Zeiss Axioscope equipped with a CCD camera. The transmitted light and fluorescence images were finally merged.

# RESULTS

# Distribution of Notch Proteins in Injured Teeth

Upon injury of developing first molars (**Figure 1A**), immunohistochemistry on sections showed that Notch1 and Notch3 expression in the central part of the dental pulp is confined to cells forming vascular structures (**Figures 1B,C**). In contrast, Notch2 staining is detected in cells of the apical pulp (**Figure 1D**). This was confirmed by whole mount immunohistochemistry, clearly showing Notch2-positive cells at the apical part of the pulp (**Figures 1E,F**). Immunostaining against the three Notch proteins was not detected in the pulp of intact teeth (data not shown) that is in accordance with our previous results (Mitsiadis et al., 1999, 2003; Sun et al., 2014).

# Correlation of Notch2 Protein Distribution and Cell Proliferation at the Apical Part of the Injured Pulp

In an attempt to ascertain whether Notch2 protein expression in the apical pulp was correlated with cell proliferation, healthy and injured dental pulp explants were cultured in presence of BrdU for up to 4 h before fixation. BrdU whole mount immunostaining revealed that while in the intact teeth only few cells proliferate at the apical part of the pulp (**Figure 2A**), cell divisions significantly enhance in this area upon tooth crown injury (**Figure 2B**). Double immunofluorescence on sections of injured pulp explants, analyzed in confocal microscopy, revealed that proliferating cells at the apical pulp part also express the Notch2 protein (**Figures 2C–E**).

# Lineage Tracing of Apical Dental Pulp Cells in Intact and Injured Teeth

We then monitored the movement of dental pulp cells located at the apical part of the forming root in cultured first

is detected in perivascular cells of the central part of the pulp. (D) Notch2 (N2) staining is mainly found in apical pulp cells at the tooth apex. (E,F) Whole mount immunostaining showing Notch2 protein distribution at the apical part of an injured pulp. Scale bars: 50µm (B–D), 200µm (E,F).

molars. For this purpose, DiI was injected in apical pulp cells (**Figures 3A–C,G–I**, **4A–C**) of injured and intact molars (time of injection indicated as T0) and then the molars were cultured for up to 4 days (T4). The development of the dental explants proceeded normally during the culture period. After fixation of the dental explants, their hard tissues (i.e., dentin, enamel) were carefully removed in order to keep intact the morphology of the pulp tissue. The experiments were performed in molars from postnatal day 6–8 mouse pups since the integrity of pulp tissues from older mice was severely impaired upon removal of the dental hard tissues. Dental pulps were then photographed using a fluorescence microscope. After merging fluorescence and bright-field images, DiI-positive cells in injured teeth were observed not only in apical pulp cells but also in migrating cells that form a line joining the apical part with the injury site of the tooth (**Figures 3D–F**). In contrast, in pulps of intact teeth, labeled cells did not migrate and remained as cohesive patches in the apical pulp, where DiI was injected (**Figures 3J–L**). Similar results showing migration of DiI-positive cells from the apical part of the pulp toward the injury site were also obtained in tooth explants after 1 day of culture (**Figure 4**).

# DISCUSSION

Severe dental injuries (i.e., deep cavity preparations) induce apoptosis of odontoblasts and activation of complex regenerative mechanisms within the dental pulp (Tziafas et al., 2000; Mitsiadis and Rahiotis, 2004; Mitsiadis et al., 2008, 2011). Cells of the subodontoblastic layer, which express the Notch2 receptor (Mitsiadis et al., 2003; Mitsiadis and Rahiotis, 2004), could replace the apoptotic odontoblasts and differentiate into odontoblast-like cells. However, sub-odontoblastic cells are equally eliminated by apoptosis after a severe stress. Therefore, the number of neighboring cells that are able to differentiate into odontoblastlike cells decreases considerably and alternative sources of pulp stem/progenitor cells should be activated to ensure dental tissue repair. Although dental pulp contains a significant amount of stem cells (Bluteau et al., 2008; Catón et al., 2011), severe injury might induce stemness in cells that in physiological conditions do not behave as stem cells (Potten and Loeffler, 1990). Human dental pulp stem cells (hDPSCs) isolated from permanent teeth (Gronthos et al., 2000) and the apical pulp of shed primary teeth (called SCAP) (Shi and Gronthos, 2003) are characterized by expression of several mesenchymal stem cell markers (e.g., CD29, CD73, CD105, CD44) and extracellular matrix molecules such as collagen, vimentin, laminin, and fibronectin (Bluteau et al., 2008; Pagella et al., 2015). hDPSCs are multipotent and have shown a great potential for repair and regeneration since they can differentiate into various cell types such as adipocytes (Waddington et al., 2009), chondroblasts/chondrocytes (Waddington et al., 2009), osteoblasts/osteocytes (de Mendonça Costa et al., 2008; Graziano et al., 2008), myocytes (Kerkis et al., 2008), cardiomyocytes

(Gandia et al., 2008), and odontoblasts (Cordeiro et al., 2008; Nedel et al., 2009). This potential has made hDPSCs an attractive choice for tissue engineering, especially when they can be used as autologous transplants (Mitsiadis et al., 2015). The best example showing the success of such strategies is the first clinical trial using hDPSCs in patients for alveolar bone reconstruction (d'Aquino et al., 2009).

hDPSCs are thought to reside in one or more distinct storage sites, also called stem cell niches (Mitsiadis et al., 2015; Pagella et al., 2015). The concept of stem cell niches refers to defined anatomical compartments that include cellular and acellular (e.g., extracellular matrix) components. The niches provide highly specialized microenvironments, enabling stem cells to survive, self-renew, change their number and fate, and finally to participate in tissue repair (Scadden, 2006; Pagella et al., 2015). Similarly, niche-derived signals influence the function of dental pulp stem/progenitor cells. In addition, intercellular signaling pathways such as bone morphogenetic proteins (BMPs), Wnt, epidermal growth factor (EGF) and Notch are important regulators of stem cell function (Artavanis-Tsakonas and Muskavitch, 2010). Under the influence of these signaling pathways, dental stem cell might migrate from nearby or distant niches to the area of injury, where they will engraft and participate in the regeneration process. These cells will give rise to odontoblast-like cells that form the reparative dentin (Mitsiadis et al., 2015; Miran et al., 2016).

Significant molecular changes accompany dental pulp regeneration. For example, BMPs and transforming growth factor beta (TGFβ) are released from dentin after injury and contribute to reparative dentin formation (Tziafas et al., 2000; About and Mitsiadis, 2001; Mitsiadis and Rahiotis, 2004; Mitsiadis and Graf, 2009). Similarly, although Notch

pictures. DiI labels cells in red color while DAPI stains in blue color the nuclei of pulp cells in cryosections. (D–L) Migration of DiI-labeled cells from the root apex toward the injury site - whole pulp view. Bright-field pictures (D,G,J), dark-field pictures (E,H,K), merged pictures (F,I,L). (D–I) Dental pulps immediately after injury and DiI injection (T0). (J–L) Dental pulps 1 day after cavity preparation and DiI injection (T1). Black and white arrows indicate the area of the tooth injury. DiI-positive pulp cells are seen in red color. Light blue arrowheads indicate the final location of migrating. (M–R) Migration of DiI-labeled cells from the apical part of the pulp toward the injury site—cryostat sections. DiI-labeled cells in red color. Bright-field pictures (M,P), dark-field pictures (N,Q), merged pictures (O,R). Black and white arrows indicate injured teeth. (M–N) Intact dental pulps 1 day after DiI injection (T1). Note that in DiI-positive cells remain in the apical part of the pulp. (P–R) Dental pulps 1day after cavity preparation and DiI injection (T1). Light blue arrowheads indicate the final location of migrating DiI-positive cells. Abbreviations: p, pulp; T0, time zero days; T1, time 1 day. Scale bars: 100µm (A,B,D–R), 20µm (C).

receptors are absent in adult dental pulps, their expression is re-activated after injury (Mitsiadis et al., 1999, 2003; Lovschall et al., 2007). In injured rodent teeth, the Notch1 and Notch3 receptors, as well as the Delta-1 ligand are expressed in cells related to vascular structures, either near or far away of the injury site, while Notch2 is strongly expressed in dental pulp mesenchymal cells at the root apex (Mitsiadis et al., 1999). A similar pattern of Notch2 expression was observed in injured human permanent teeth (Mitsiadis et al., 2003), suggesting the existence of a pool of putative stem/progenitor cells at the apical part of the pulp. Previous findings in injured teeth showed reactivation of Notch3 in pericytes that might be another

source for dental pulp stem cells (Lovschall et al., 2007). This is in line with previous findings indicating the existence of putative dental pulp perivascular niches (Shi and Gronthos, 2003; Oh and Nör, 2015). However, it has not yet proven that these Notch expressing cell populations participate to the process of dentin/pulp regeneration and can differentiate into odontoblast-like cells after injury. The activation of Notch signaling in human dental pulp stem/progenitor cells in vitro by either Jagged1 or the intracellular domain of Notch1 leads to inhibition of odontoblast differentiation without affecting cell proliferation (Zhang et al., 2008). In contrast, the activation of Notch signaling by Delta1 stimulates both cell differentiation and proliferation (He et al., 2009). Together these results suggest that Notch receptors may act as both positive and negative regulators of DPSCs depending on the ligand that they bind. Proliferative events in the apical pulp correlate to Notch2 expression after tooth crown injury. Our results are in line with recent findings showing that Notch2 signaling controls proliferation of various stem cell populations (e.g., bone marrow stem cells) and cancer cells (Huang et al., 2015; Sato et al., 2016; Wu et al., 2016).

In intact teeth, no obvious movement of the DiI-labeled apical pulp cells was observed. In contrast, in injured teeth, these cells migrate from the apex toward to the wounded area, a site where the reparative dentin will form. The migrating cells were seen in the line unifying the apex with the injury site. These results indicate that the apical part of the pulp contains a pool of putative stem/progenitor cells capable of migrating toward the injured site of the tooth. Our findings show a correlation between DiIlabeled cells, proliferating cells and Notch2-positive cells at the apical part of the injured dental pulp, thus suggesting a regulatory role for Notch2 signaling in proliferation and migration of pulp stem/progenitor cells. Loss- and gain-of-function experiments will be needed to investigate the functional role of Notch2 signaling in this process.

Although cells located at the apical pulp can end up far away (i.e., tooth crown) and participate in healing, other pulp stem/progenitor cell populations may also participate in tooth repair (e.g., cells derived from the perivascular niches), either concomitantly with apical pulp cells or at different time points (Mitsiadis et al., 2011, 2015; Oh and Nör, 2015; Ducret et al., 2016). A hypothetical schematic model of the pulp stem cell niches, their activation upon tooth injury, and stem cell kinetics within the pulp during repair is shown in **Figure 5**. To determine whether Notch2-positive cells directly contribute to these cellular events upon tooth injury, genetic lineage tracing experiments using a Notch2-creERT driver (Fre et al., 2011), combined with immunostaining for odontoblast differentiation markers, have to be undertaken. The present findings highlight the importance and dynamic nature of the apical pulp cells for tooth repair. These cells can reach the injury site and may participate in the formation of reparative dentin. Here we establish a conceptual framework where important cellular and molecular mechanisms operating during dental regenerative processes start to be elucidated. Such knowledge might be helpful for the realization of innovative dental treatments.

# AUTHOR CONTRIBUTIONS

TM, Contributed to the conception of the hypothesis of the study and in the development of the model, involved in the evaluation of the results and preparation of the manuscript. He also provided approval for the publication of this version. JC, Contributed to the conception of the hypothesis of the study and to the development of the model, the acquisition and the analysis of data for the work. He was also involved in the preparation of the manuscript and provided approval for the publication of this version. PP, Contributed to the analysis and the interpretation of data for the work. He was also involved in the preparation of the manuscript and sanctioned the publication of this version. GO, Contributed to the development of the model and the interpretation of data for the work. She was also involved in the preparation of the manuscript and provided approval for the publication of this version. LJ, Contributed to the hypothesis of the study, collaborated in the development of the model and was

# REFERENCES


involved in the evaluation of the results and preparation of the manuscript. She also provided approval for the publication of this version.

# ACKNOWLEDGMENTS

This work was supported by funds from the University of Zürich. The authors did not receive any grant from commercial or profit sectors.


**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 © 2017 Mitsiadis, Catón, Pagella, Orsini and Jimenez-Rojo. 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) or licensor 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 Effects of Photobiomodulation of 808 nm Diode Laser Therapy at Higher Fluence on the in Vitro Osteogenic Differentiation of Bone Marrow Stromal Cells

Andrea Amaroli <sup>1</sup> , Dimitrios Agas <sup>2</sup> , Fulvio Laus <sup>2</sup> , Vincenzo Cuteri <sup>2</sup> , Reem Hanna<sup>1</sup> , Maria Giovanna Sabbieti <sup>2</sup> and Stefano Benedicenti <sup>1</sup> \*

<sup>1</sup> Department of Surgical and Diagnostic Sciences, Laser Therapy Center, University of Genoa, Genoa, Italy, <sup>2</sup> School of Biosciences and Veterinary Medicine, University of Camerino, Macerata, Italy

### Edited by:

Claudio Cantù, University of Zurich, Switzerland

### Reviewed by:

Patrizia Ferretti, University College London, United Kingdom Giovanna Orsini, Università Politecnica delle Marche, Italy

\*Correspondence:

Stefano Benedicenti stefano.benedicenti@unige.it

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 22 August 2017 Accepted: 07 February 2018 Published: 23 February 2018

### Citation:

Amaroli A, Agas D, Laus F, Cuteri V, Hanna R, Sabbieti MG and Benedicenti S (2018) The Effects of Photobiomodulation of 808 nm Diode Laser Therapy at Higher Fluence on the in Vitro Osteogenic Differentiation of Bone Marrow Stromal Cells. Front. Physiol. 9:123. doi: 10.3389/fphys.2018.00123 The literature has supported the concept of mesenchymal stromal cells (MSCs) in bone regeneration as one of the most important applications in oro-maxillofacial reconstructions. However, the fate of the transplanted cells and their effects on the clinical outcome is still uncertain. Photobiomodulation (PBM) plays an important role in the acceleration of tissue regeneration and potential repair. The aim of this in vitro study is to evaluate the effectiveness of PBM with 808 nm diode laser therapy, using a flat-top hand-piece delivery system at a higher-fluence (64 J/cm<sup>2</sup> ) irradiation (1 W, continuous-wave) on bone marrow stromal cells (BMSCs). The BMSCs of 3 old female Balb-c mice were analyzed. The cells were divided into two groups: irradiated group and control group. In the former the cells were irradiated every 24 h during 0 day (T0), 5 (T1), 10 (T2), and 15 (T3) days, whereas the control group was non-irradiated. The results have shown that the 64 J/cm<sup>2</sup> laser irradiation has increased the Runt-related transcription factor 2 (Runx2). Runx2 is the most important early marker of osteoblast differentiation. The higher-fluence suppressed the synthesis of adipogenic transcription factor (PPARγ), the pivotal transcription factor in adipogenic differentiation. Also, the osteogenic markers such as Osterix (Osx) and alkaline phosphatase (ALP) were upregulated with an increase in the matrix mineralization. Furthermore, western blotting data demonstrated that the laser therapy has induced a statistically valid increase in the synthesis of transforming growth factor β1 (TGF-β1) but had no effects on the tumor necrosis factor α (TNFα) production. The data has statistically validated the down-regulation of the important pro-inflammatory cytokines such as interleukin IL-6, and IL-17 after 808 nm PBM exposition. An increase in anti-inflammatory cytokines such as IL-1rα and IL-10 was observed. These in vitro studies provide for first time the initial proof that the PBM of the 808 nm diode laser therapy with flat-top hand-piece delivery system at a higher-fluence irradiation of 64 J/cm<sup>2</sup> (1 W/cm<sup>2</sup> ) can modulate BMSCs differentiation in enhancing osteogenesis.

Keywords: laser therapy, low-level-laser therapy, phototherapy, stem cells, bone, anti-inflammatory effect, differentiation, cytokines

**251**

# INTRODUCTION

Within dentistry, tissue regenerative concepts have been developed to be applied in the clinical areas of periodontology and implantology (Izumi et al., 2003; Egusa et al., 2012). Oro-maxillofacial reconstruction procedures can assist in facial tissue repair and regeneration, such as may be the result of trauma or neoplasm. In Europe, about 1.5 million patients undergo craniofacial reconstruction procedures annually. However, ∼20% of them continue to experience functional deficiencies despite the surgical interventions. Furthermore, about 30,000 patients per year would develop donor-site morbidity following oral and maxillofacial reconstruction (Czerwinski et al., 2010; Rodriguez y Baena et al., 2016).

The research area of utilizing stem cells toward restoring the structure and function of damaged tissues or organs in medicine and dentistry (Huang et al., 2009; Meirelles Lda and Nardi, 2009) is moving fast and attracting the interest of researchers. Many studies have shown that the use of mesenchymal stromal cells (MSCs) for bone regeneration is one of the most important applications in oro-maxillofacial reconstruction (Egusa et al., 2012; Zigdon-Giladi et al., 2014), periodontology and implantology (Izumi et al., 2003; Egusa et al., 2012). Despite the positive results that were achieved in bone regeneration through utilizing the MSCs, the destiny of the transplanted cells and their effect on the clinical outcome remains uncertain. This could be related to the fact that the transplanted cells die very quickly or migrate out of the transplantation site according to several animal studies (Meijer et al., 2008; Zimmermann et al., 2011; Egusa et al., 2012). Therefore, the challenge of stem cell research is related to the identification of supporting therapies, which can improve and enhance the effectiveness of various tissues repair and regeneration.

Utilization of a low-energy light intensity within the visible red and near infrared portion of the electromagnetic spectrum has been shown to stimulate irradiated cellular activity (Avci et al., 2013). This phenomenon has been referred to as photobiomodulation (PBM) (Avci et al., 2013; Amaroli et al., 2015a). Many cellular responses have been observed in in vitro models after irradiation with light sources of different wavelengths at specific energy density, which has induced PBM effects. These cellular responses resulted in: an increase of mitochondrial respiration and ATP production (Karu, 2010; Amaroli et al., 2015b,c, 2016c), the synthesis of proteins (Avci et al., 2013; Lipovsky et al., 2013) and in cells migration and proliferation (Avci et al., 2013; Amaroli et al., 2015a). Furthermore, reduction in the inflammation, acceleration of wound healing (Avci et al., 2013; Amaroli et al., 2018) and bone formation (Santinoni et al., 2017) were observed. Few studies have investigated the use of PBM on MSCs' stimulation (Lipovsky et al., 2013; Kushibiki et al., 2015). These studies have pointed out that the effectiveness of PBM on MSCs was related to enhanced angiogenic effects of adipose-derived stromal cells (Park et al., 2014) and to the modulation of gene expression, proliferation, intracellular cAMP levels, and osteogenic differentiation (Kushibiki et al., 2015). However, despite in vivo studies (animal models and randomized controlled clinical trials) with a positive PBM outcome, it remains a controversial subject as a consequence of conflicting effects produced by various operating parameters such as wavelength, fluence, laser power output, exposure time, number of applications (Jenkins and Carroll, 2011), and the beam profile (Amaroli et al., 2016b).

Recently, the AB2799 flat-top hand-piece has been developed and marketed. It can generate a more homogenous irradiation at ∼1 cm<sup>2</sup> (Spot size) from contact to 105 cm distance from the target tissue (Amaroli et al., 2016b; Selting, 2016). This probe allows the use of relatively higher power densities and fluences with lower risk of causing collateral thermal damage to adjacent tissues, in comparison to standard (Gaussian profile) handpieces (Amaroli et al., 2015a,b; Selting, 2016). Previous studies by Amaroli and co-workers have shown that, the use of the 808 nm diode laser with the flat-top hand-piece at a higher-fluence of 64 J/cm<sup>2</sup> and power output of 1 Watt (W) in continuous-wave (CW) emission, resulted in increased mitochondrial activity (Amaroli et al., 2016b), in stimulation of oxygen consumption (Amaroli et al., 2015b) and in ATP production (Amaroli et al., 2015c, 2016c) in the unicellular organism Paramecium primaurelia. In the same unicellular model, modulation of calcium fluxes and intracellular calcium concentration (Amaroli et al., 2015c, 2016a) was also observed, as well as an increment of fission rate rhythm (Amaroli et al., 2015a). In addition, it was evidenced that this higher-fluence did not generate genotoxic damage or adverse effects (Amaroli et al., 2017) while it was able to induce anti-inflammatory effects that promoted the wound healing in an animal model (Cuteri et al., 2016; Amaroli et al., 2018).

The purpose of this study was to assess the in vitro ability of 808 nm diode laser therapy with flat-top hand-piece delivery system at a higher-fluence and power (64 J/cm<sup>2</sup> ; 1 W, 1 W/cm<sup>2</sup> ) on inducing osteoblast maturation and on modulating the bone marrow stromal cells (BMSCs) secretion of important inflammatory mediators.

Latterly, increasing findings point to the properties of BMSCs; a cell population isolated from the bone marrow stroma and constituted by multiple cell types including multipotent progenitors for skeletal lineages (Bianco and Robey, 2015). In addition to their ability to differentiate into skeletal tissue cells, the BMSC possess immunomodulatory characteristics (Kuroda et al., 2017). The current research is increasingly demonstrating the importance of the interaction of the bone marrow stem/progenitor and the mature cell in their role in bone metabolism, turnover, and regeneration (Agas et al., 2015; Kassem and Bianco, 2015). Within this investigation, the BMSCs of 3 month old female Balb-c mice were analyzed through histochemical staining (alkaline phosphatase and alizarin red S assays), immunoassay (cytokines and chemokines assays) and western blotting (expression of runtrelated transcription factor 2, osterix, peroxisome proliferatoractivated receptor gamma, transforming growth factor β1, tumor necrosis factor α).

# MATERIALS AND METHODS

# Laser Unit

In this study, a diode laser (λ 808 nm) (Doctor Smile–LAMBDA Spa–Vicenza, Italy) with the AB2799 hand-piece was utilized to irradiate the BMSCs. According with the technical data (http://www.doctor-smile.com/assets/prodotti/accessori/pdf/

ST\_FLATTOP\_EN.pdf), the AB2799 hand-piece with a flat-top profile delivers more homogenous irradiation over ∼1 cm<sup>2</sup> surface area and has the same irradiation spot area and energy from contact to 105 cm of distance from the target tissue in comparison to the standard hand-piece, which delivers Gaussian profile irradiation (Amaroli et al., 2016b; Selting, 2016).

# Bone Marrow Stromal Cells (BMSCs) Derivation

All the described animal-related procedures were conducted according to Directive 2010/63/EU of the European Parliament and of the Council of 22nd September 2010 on the protection of animals used for scientific purposes (Article 3, Paragraph 1), the Italian Legislation (D. Lgs. n. 26/2014, Article 3, Paragraph 1, Letter a), which does not require any approval by the competent authorities. Three month old female Balb-c mice (Harlan Italy SrL, Correzzana, Milano, Italy) were used. The mice were kept in a laminar-flow cage in a standardized environmental condition. Food (Harlan, Italy) and water were supplied ad libitum. The mice were sacrificed by cervical dislocation.

Long bones (femurs, tibiae, and humeri) were dissected to be free from any adhering tissues. Bone ends were removed and the marrow cavity flushed out. Cells were pooled and plated on 100 mm culture dishes. Cells were grown in Roswel Park Memorial Institute (RPMI) culture medium added with 10% heat-inactivated-fetal calf serum (HIFCS), penicillin (100 U/ml), and streptomycin (50µg/ml) (all from Invitrogen Life Technologies, Milan, Italy) for 10 days at 37◦C in a humidified atmosphere of 5% CO<sup>2</sup> in order to generate monolayers of nonhematopoietic adherent cells (Bianco et al., 2013) (referred as "BMSCs"). The culture medium was replaced every 3 days.

# BMSCs Cultures for the Experimental Studies

Cells were detached using 0.25% trypsin for 2 min at room temperature and plated as follow: for the western blotting and the cytokines and chemokines assays (sections Western Blotting and Cytokines and Chemokines Assay, respectively), the BMSCs were plated on a 24-well culture plate at the density of 10 × 10<sup>5</sup> cells/well; for Alkaline Phosphatase assay and Alizarin Red S histochemical staining (section Alkaline Phosphatase Assay and Alizarin Red S Histochemical Staining) the BMSCs were plated on six well culture plates at the density of 10 × 10<sup>5</sup> cells/well; for the immunolabeling of Run-related transcription factor 2 (Runx2) or osterix (Osx)(section Single Immunolabeling) the BMSCs were plated at the density of 10 × 10<sup>4</sup> cells/well on round coverslips, which previously sterilized and inserted in the six well culture plates.

Cells were plated in RPMI culture medium with added HIFCS, except for the cultures used for Alkaline Phosphatase (ALP) assay and Alizarin Red S staining (section Alkaline Phosphatase Assay and Alizarin Red S Histochemical Staining) that were plated in osteogenic medium (RPMI, 10% HIFCS, 8 mM β-glycerophosphate, and 50µg/ml ascorbic acid). The same culture conditions were provided during the conduction of the experiments.

One day after cell seeding (passage 1), all the cultures were irradiated with 808 nm wavelength laser or untreated as below described. Culture medium was changed every 3 days.

# Irradiation Parameters

The culture dishes (**Table 1A**) were uncovered and then wrapped up in the aluminum foils (**Table 1B**) with a hole (**Tables 1C,D**) of a diameter corresponding to the diameter of the laser spot area of the hand-piece to avoid energy dispersion (Johnstone et al., 2014). The cultures were then irradiated using the 808 nm diode laser with the AB2799 flat-top hand-piece delivery system (**Table 1E**). The laser irradiation was performed with the handpiece in contact with the aluminum foil hole at 1 cm distance from the cells.

The cells were irradiated with 1 W in CW (power density 1 W/cm<sup>2</sup> ) for 60 s of irradiation exposure time to generate a final fluence of 64 J/cm<sup>2</sup> . This application of laser irradiation was repeated every 24 h over different periods of time (Day 0: T0; 5 days: T1; 10 days: T2; 15 days: T3). The related control cultures (Day 0: C0; 5 days: C1; 10 days: C2; 15 days: C3) were maintained

TABLE 1 | Description of the irradiation parameters and the experiments' design.


# EXPERIMENTAL DESIGN

and 15 days

in identical conditions except that the laser device was switched off.

# Western Blotting

The proteins from irradiated BMSCs or untreated (control) cells were extracted in cell lysis buffer (Cell Signaling, EuroClone) at the end of each set of irradiation (Day 0, 5 days, 10 days, 15 days) and the concentration was determined by the bicinchoninic acid protein assay reagent (Pierce, EuroClone).

The Western blotting was performed as previously described (Agas et al., 2016). Membranes were immunoblotted in blocking buffer with specific antibodies: rabbit anti-runt-related transcription factor 2 (Runx-2) antibody (1:800 dilution, Cell Signaling, Euroclone, Milano, Italy); rabbit anti-Osterix (Osx), rabbit anti-peroxisome proliferator-activated receptor gamma (PPARγ) antibody (1:600 dilution, Santa Cruz Biotechnology— DBA, Milano, Italy); rabbit anti- transforming growth factor β 1 (TGF-β1) antibody (1:600 dilution, Abcam, Prodotti Gianni, Milano, Italy); rabbit anti-tumor necrosis factor α (TNFα) (1:500 dilution, BioLegend, Microtech SrL, Napoli, Italy). After washing blots, they were incubated with horseradish peroxidase (HRP)-conjugated donkey anti-rabbit immunoglobulin G (IgG) or with HRP-conjugated rabbit anti-mouse IgG (Cell Signaling, Euroclone Milano, Italy). The immunoreactive bands were visualized using luminol reagents/ECL film according to the manufacturer's instructions. In order to normalize the bands, filters were stripped and re-probed with a monoclonal anti-α-tubulin (Sigma-Aldrich, Milano, Italy). The bands' density was densitometrically quantified by NIH Image.

# Single Immunolabeling

At the end of each set of irradiations, the untreated and irradiated BMSCs were fixed in 4% PFA and permeabilized with 0.3% Triton X-100 as previously described (Sabbieti et al., 2010). The cultures were then incubated for 2 h at room temperature with the following primary antibodies: rabbit anti-runt-related transcription factor 2 (Runx-2) antibody (1:100 dilution, Cell Signaling, Euroclone, Milano, Italy) or rabbit anti-Osterix (Osx) antibody (1:50 dilution, Santa Cruz Biotechnology—DBA, Milano, Italy). After washing the cells with PBS, they were incubated with Alexa Fluor-488 chicken anti-rabbit IgG or with Alexa Fluor 594 goat anti-rabbit IgG (both 1:100 dilution, Life Technologies, Monza, Italy) for 1 h at room temperature. The reaction controls were performed by complexing the primary antibody with a relative blocking peptide or by omitting the primary antibody. Coverslips were mounted on slides with PBS/glycerol (1:1). The slides were imaged using fluorescent microscopy on Zeiss Axioplan microscopy. Fluorescence analysis was performed by a fluorimeter Tecan Infinite with excitor filter 590 nm and emission 635 nm for Alexa Fluor 594, or 485, and of 535 nm for Alexa Fluor 488. A Tecan Infinite fluorescence reader quantified the amount of Alexa Fluor 594-labeled anti-Osx and Alexa Fluor 488-labeled anti-Runx2.

# Alkaline Phosphatase Assay and Alizarin Red S Histochemical Staining

The cells maintained in osteogenic medium (Naganawa et al., 2006, 2008) were irradiated or untreated as described above in section BMSCs Cultures for the Experimental Studies. At the end of each set of irradiations (Day 0, 5 days, 10 days, 15 days), the cells were fixed in 4% paraformaldehyde (PFA) for 20 min at room temperature. The alkaline phosphatase (ALP) staining was performed with a commercial kit (Sigma-Aldrich) according to the manufacturer's instructions. NIH Image has measured the ALP colonies. For the Alizarin Red S staining, the untreated and irradiated cells were stained to assess the mineralized matrix as previously described (Sabbieti et al., 2013). The cells' layers were briefly rinsed with PBS and fixed in 4% PFA. Then, the cultures were stained with 2% Alizarin Red S (pH 7.2, Sigma-Aldrich) for 20 min at 37◦C. The cultures were examined under light microscopy (Zeiss Axioplan; Zeiss S.p.A., Milano, Italy). The stain was desorbed and the collected solutions were distributed as 100 µL/well on 96-well plates for absorbance reading at 590 nm by spectrophotometry (Tecan Infinite reader; Tecan, Milano, Italy).

# Cytokines and Chemokines Assay

The cytokine/chemokine profiles in supernatants of the cultured BMSCs population in the laser irradiated group or untreated group (control) were assessed at the end of each set of irradiation (Day 0, 5 days, 10 days, 15 days) by using Mouse Cytokine Array Panel A kit (R&D Systems, Milano, Italy) according to the manufacturer's instructions as previously described (Sabbieti et al., 2015).

# Statistical Analysis

Data were analyzed by using one-way ANOVA followed by Tukey's pairwise comparisons. The letters a, b, c, or d above a particular column indicate statistically significant difference for that group in comparison to another group within a particular graph. Model p-value and sample number labeled, with standard error depicted for each treatment site. Values of <sup>∗</sup>p < 0.05 were considered significant.

The significance of difference between two groups was evaluated with an unpaired two-tailed Student's t-test and differences were considered significant at <sup>∗</sup>p < 0.05.

The results are a representative of those obtained by independent experiments repeated at least three times (samples per group n = 3).

# RESULTS

# Effects of Photobiomodulation on Osteoblast Differentiation

In order to assess the effects of laser therapy on osteoblast maturation, BMSCs were treated with PBM therapy at different time points, which corresponded at 0 (T0), 5 (T1), 10 (T2), and 15 (T3) days. The Western blotting analyses of laser treated cells indicated a significant increase of both Runx2 and Osx transcription factors that started at the time point T1 and reached maximal level at the time point T2 (**Figure 1A**). At the time point T3 the increase of Osx was maintained, while Runx2 synthesis was decreased. Notably, a decrease of the adipogenic transcription factor PPARγ was evident (**Figure 1A**). In addition, the findings from in situ investigations to examine the subcellular distribution of Runx2 and Osx showed an increase of the Runx2 and the Osx labeling at cytoplasm and perinucleare level in irradiated BMSCs at time point T1. Interestingly, the fluorescence analysis at the time point T2 revealed a strong accumulation of the mean fluorescence intensity for Runx2 (247.17% vs. untreated control) and osterix (380.43% vs. untreated control) particularly at perinuclear and nuclear level (**Figure 1B)**. This evidence indicates the correct trafficking of Runx2 and Osx to the nucleus. This process is required for their function as regulators of bone formation gene expression. No differences were found at T0 time point between laser-treated and untreated cells (data not shown). Since the alkaline phosphatase (ALP) is an important marker of osteogenic differentiation,

FIGURE 1 | (A) Time-course effects of laser irradiation on Runx2, osterix, and PPARγ synthesis in BMSCs by western blotting analysis. Cells were laser irradiated over different period of time (Day 0: T0; 5 days: T1; 10 days: T2; 15 days: T3). Control cultures (Day 0: C0; 5 days: C1; 10 days: C2; 15 days: C3) were maintained in identical conditions except that the laser device was switched off. At the end of each treatment, proteins from BMSCs irradiated and untreated were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and probed with rabbit anti-Runx2, rabbit anti-osterix, or rabbit anti-PPARγ antibodies; filters were stripped and re-probed mouse anti-α-tubulin antibody to show equal amount of loading. Graphic represents results of three independent experiments. Data were analyzed by using one-way ANOVA. Lowercase letters denote homogeneous subsets. Error bars represent ± SE (\*p < 0.05) (A). The labeling pattern of Runx2 or osterix in BMSCs laser irradiated and untreated. Representative images of Runx2 or osterix localization using a rabbit anti-Runx2 antibody (green: Alexa Fluor 488 staining) and the rabbit anti-osterix antibody (red: Alexa Fluor 594 staining). Bar, 50µm. Fluorescence analysis from a pool of three different experiments was quantified by a Tecan Infinite fluorescence reader. The values that analyzed by Magellan v4.0 software are reported as means ± SD and statistically analyzed; \*p < 0.05 (B).

we examined the effects of laser on ALP positive colonies formation. At 10 days of the treatment, an increase of ALP positive colonies was observed and its maximum number was reached after 15 days (**Figures 2A,B**). Additionally, the Alizarin Red S staining was used to measure the calcium deposition that is considered as a late differentiated marker; a notable increase of calcium deposition was found after 15 days of laser treatment (**Figures 3A,B**).

# Effects of Low-Level-Laser Therapy on the Cytokines Synthesis and Release by BMSCs

Predominantly, stromal stem/progenitor and mature cells release the signaling molecules which are released within bone marrow reservoir. They exert pivotal roles in regulating bone remodeling, in particular the transforming growth factor β1 (TGF-β1), which is involved in bone formation.

It is worth noting that TNFα is involved in inflammatory bone erosion.

The Western Blotting data have demonstrated that the laser therapy had induced a statistically valid increase in the synthesis

FIGURE 3 | Time-course effects of laser irradiation on calcium deposition. BMSCs grown in osteogenic medium, as detailed in the Materials and Methods section, were laser irradiated over different period of time (Day 0: T0; 5 days: T1; 10 days: T2; 15 days: T3). Control cultures (Day 0: C0; 5 days: C1; 10 days: C2; 15 days: C3) were maintained in identical conditions except that the laser device was kept off. The results show the Alizarin red S staining observed with a light microscope (Bar, 100µm) (A) and are in accordance with quantitative analysis of Alizarin red staining in which the values are reported as means ± SD and statistically analyzed; \*p < 0.05. vs. the corresponding untreated BMSCs (B).

of TGF-β1 but had no effects on the TNFα production (**Figure 4**). However, the analysis of the BMSCs supernatants provided slight evidence but statistically validated the downregulation of the important pro-inflammatory cytokines such as interleukin (IL)-6, and IL-17 after laser irradiation. In addition, an increase in anti-inflammatory cytokines such as IL-1ra and IL-10 was observed (**Figure 5**).

# DISCUSSION

Stem cell research is related to identifying optimal therapies that can be utilized in promoting tissue healing. Photobiomodulation could be a useful adjunctive, as it plays a role in acceleration of tissue regeneration and repair (Avci et al., 2013). Such application on stem cells may be therefore beneficial in many areas of biomedical application and ultimately, this would augment the success of regenerative medicine (Abrahamse, 2012). Our data appears to support this line of thought. In fact, it was found that the 808 nm laser irradiation at a fluence of 64 J/cm<sup>2</sup> increased the BMSCs' Runx2, which is one of the early cell markers that promotes MSCs differentiation into immature osteoblasts and inhibits lineage development into adipocytes

FIGURE 4 | Time-course effects of laser irradiation on TGF-β1 and TNFα synthesis in BMSCs by western blotting analysis. Cells were laser irradiated over different periods of time (Day 0: T0; 5 days: T1; 10 days: T2; 15 days: T3). Control cultures (Day 0: C0; 5 days: C1; 10 days: C2; 15 days: C3) were maintained in identical conditions except that the laser device was switched off. At the end of each treatment, proteins from BMSCs irradiated and untreated were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and probed with rabbit anti-TGF-β1 rabbit anti-TNFα; then, filters were stripped and re-probed mouse anti-α-tubulin antibody to show equal amount of loading. Graphic represents results of three independent experiments. Data were analyzed by using one-way ANOVA. Lowercase letters denote homogeneous subsets. Error bars represent ± SE (\*p < 0.05).

(Komori, 2010). Wu et al. (2012) study has shown that irradiation with gallium-aluminum-arsenide (GaAlAs) red laser (wavelength 660 nm) at different fluences of 1 J/cm<sup>2</sup> , 2 J/cm<sup>2</sup> , or 4 J/cm<sup>2</sup> with power density 10 mW/cm<sup>2</sup> can induce the generation of insulin-like growth factors 1 (IGF-1), to promote both the proliferation and the osteogenic differentiation of mouse bone marrow mesenchymal stem cells (D1 cells), whereas it may induce bone morphogenetic protein 2 (BMP2) expression primarily to enhance osteogenic differentiation. It is known that IGF-1 can regulate Runx2 DNA binding (Qiao et al., 2004; Guntur and Rosen, 2013) and BMP2 is able to induce osteoblast differentiation through Runx2-dependent activating transcription factor 6 (ATF6) expression (Jang et al., 2012). Therefore, our protocol of utilizing the 808 nm laser therapy at a higher-fluence can photobiomodulate the BMSCs in a way which is similar to what has been observed in Wu et al. (2013) results, when 660 nm was utilized to irradiate the cells at low-fluence. However, previous results by Bouvet-Gerbettaz et al. (2009) have failed to induce osteoblast differentiation with 808 nm wavelength at low-fluence (4 J/cm<sup>2</sup> ). In contrast, Soleimani et al. (2012) has observed that 810 nm GaAlAs with range of fluences of 3–6 J/cm<sup>2</sup> had a positive effect

on the BMSCs differentiation. However, the author did not provide an explanation of the mechanism of action of the laser therapy.

In our study an induction of MSC to osteoblast precursors and an inhibition of their commitment to adipocytes lineage was demonstrated by the increase of Runx2 and the decrease of PPARγ (Komori, 2010) in culture exposed to irradiation for 5 (T1) and 10 days (T2). In addition, the differentiation of osteoblasts precursors into mature osteoblasts was demonstrated in laser-irradiated cells by the increase of Osx (Zou et al., 2006) overall, evident after 15 days (T3) of irradiation. At the same time point (T3) an enhanced ALP positivity and matrix mineralization (revealed by Alizarin red staining analysis) was found in irradiated cultures, compared to the untreated control cells. As anticipated, for the evaluation of matrix deposition both untreated (control) and treated cultures were grown on osteogenic medium to accelerate and increase the mineralization process, since cells growth in RPMI culture medium showed slow matrix deposition after laser treatment. In addition, the untreated cultures did not show any mineralization, in line with the fact that our experiments were conducted on a cell population involving overall undifferentiated cells.

Taken together, these data point to the fact that laser irradiation not only induced the MSCs development toward immature osteoblasts but also promoted osteoblast maturation.

It has been demonstrated that within the bone marrow, the process of MSCs maturation along osteoblastic lineage is finely regulated by signals, which are released in the marrow microenvironment. For instance, TGF-β1, besides its role in promoting osteoblastic precursors or matrix-producing osteoblasts through chemotactic attraction (Park et al., 2014), also blocks the apoptosis of osteoblasts (Huang et al., 2009) and enhances the osteoblast proliferation (Horwitz et al., 2002). If it is considered that our laser irradiation of BMSCs has no effect on the anti-proliferative and pro-apoptotic TNF-α protein, we would speculate that our therapy might maintain the cell proliferation process. This assumption has been supported by our previous results which have shown that laser therapy has positive effects on the protozoa and mammalian mitochondria activities (Amaroli et al., 2015b,c, 2016b) and stimulates the cell proliferation (Amaroli et al., 2015a) without inducing cellular damage (Amaroli et al., 2017). Furthermore, the literature has supported the fact that MSCs proliferation can be induced after laser irradiation (Abramovitch et al., 2005; Hou et al., 2008; Soleimani et al., 2012; Wu et al., 2012; Giannelli et al., 2013). In addition, it has been demonstrated that elevated levels of TGF-β1 play an important role in downregulating the release of the proinflammatory cytokine such as IL-6 and IL-17, thus promoting favorable conditions for bone regeneration (Zhou et al., 2008).

In laser treated BMSCs, we have consistently observed an increase in the synthesis of TGF-β1 associated with a decrease in the secretion of the pro-inflammatory molecules Il-1β, IL-6, and IL-17. Furthermore, the production of the anti-inflammatory interleukins IL-1ra, IL-10 was increased.

Although the changes in the cytokine levels were slight, it should be considered that the BMSCs used in these experiments were derived from healthy mice with a bone marrow environment under steady-state condition. Therefore, such small fluctuations in cytokines release may be interpreted as specific responses to the treatment.

Moreover, our data has showen that utilization of a higherfluence at a higher-power setting has confirmed the antiflogistic propriety which has coincided with the findings of the wound healing model study in muscle and epithelial tissues (Amaroli et al., 2018).

The literature has clearly demonstrated that PBM has no deleterious effects on MSC (Kushibiki et al., 2015; Marques et al., 2016). However, the current research in literature has not provided a conclusive outcome (Marques et al., 2016) but it

### REFERENCES


has only pointed out the effectiveness of the red wavelengths (Kushibiki et al., 2015; Marques et al., 2016). This evidence stresses the value of the results of our current study and the possible applications of our in vivo phototherapy investigations. In fact, the results of the most in vitro studies on PBM were obtained using wavelengths in the range 600–700 nm. However, through in vivo studies, the light-energy in this range of wavelengths can quickly disperse and does not penetrate to deeper target tissue layers in order to provide therapeutic effects. On the other hand, the wavelengths ranging from 800 up to 1100 nm which have a longer optical penetration depth can target deeper tissues (Avci et al., 2013; Pandeshwar et al., 2016). Therefore, the wavelengths in the range of 390–600 nm are used to treat in vitro culture or superficial tissues while the longer wavelengths in the range of 600–1100 nm are used to treat deeper-seated tissues. Moreover, the wavelengths in the range of 700–750 nm have been found to have a limited biochemical activity and therefore, they are not often used (Avci et al., 2013; Pandeshwar et al., 2016).

# CONCLUSION

In conclusion, our data prove for the first time that 808 nm diode laser irradiation, delivered by the flat-top hand-piece at the higher-fluence and -power of 64 J/cm2 and 1 W (CW) respectively, promotes BMSCs differentiation toward osteogenesis. Within the limits of our evaluation, our results suggest an additional possible laser effect based on its ability to increase TGFβ synthesis and to facilitate osteoblast differentiation by creating an anti-inflammatory effect on bone marrow stroma cells.

# AUTHOR CONTRIBUTIONS

AA, DA, VC, MS, FL, and SB: Conceived and designed the experiments; AA, DA, VC, MS, and FL: Performed the experiments, collected data and performed the data analyses; AA, DA, VC, MS, FL, and SB: Provided analysis tools; AA, DA, VC, MS, FL, RH, and SB: Wrote the manuscript.

# FUNDING

Funds from any funding agency were not procured.

# ACKNOWLEDGMENTS

Thank you to Prof. Steven Parker, for the support, the critical reading and the very careful review of our paper.

mesenchymal stem cells seeded on a three-dimensional biomatrix. Lasers Med. Sci. 20, 138–146. doi: 10.1007/s10103-005-0355-9

Agas, D., Concetti, F., Capitani, M., Lacava, G., Concetti, A., Marchetti, L., et al. (2016). Administration of DNA plasmid coding protein aggregating domain induces inflammatory bone loss. Curr. Gene Ther. 16, 144–152. doi: 10.2174/15665232166661603311 25355


cells engraft and stimulate growth in children with osteogenesis imperfecta: implications for cell therapy of bone. Proc. Natl. Acad. Sci. U.S.A. 99, 8932–8937. doi: 10.1073/pnas.132252399


phosphatidylinositol 3-kinase/ERK-dependent and Akt-independent signaling pathway. J. Biol. Chem. 279, 42709–42718. doi: 10.1074/jbc.M404480200


**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 © 2018 Amaroli, Agas, Laus, Cuteri, Hanna, Sabbieti and Benedicenti. 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 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.

# Generation of Spheres from Dental Epithelial Stem Cells

Despoina Natsiou<sup>1</sup> , Zoraide Granchi <sup>2</sup> , Thimios A. Mitsiadis <sup>1</sup> and Lucia Jimenez-Rojo<sup>1</sup> \*

<sup>1</sup> Orofacial Development and Regeneration, Centre for Dental Medicine, Institute of Oral Biology, University of Zurich, Zurich, Switzerland, <sup>2</sup> Genomescan B.V., Leiden, Netherlands

The in vitro three-dimensional sphere model has already been established as an important tool in fundamental sciences. This model facilitates the study of a variety of biological processes including stem cell/niche functions and tissue responses to injury and drugs. Here we describe the complete protocol for the in vitro formation of spheres originated from the epithelium of rodent incisors. In addition, we show that in these spheres cell proliferation is maintained, as well as the expression of several key molecules characterizing stem cells such as Sox2 and p63. These epithelial dentospheres could be used as an in vitro model system for stem cell research purposes.

### Edited by:

Gianpaolo Papaccio, Seconda Università Degli Studi di Napoli, Italy

### Reviewed by:

Eumorphia Remboutsika, BSRC "Alexander Fleming", Greece Michel Goldberg, Institut National de la Santé et de la Recherche Médicale (INSERM), France

### \*Correspondence:

Lucia Jimenez-Rojo lucia.jimenezrojo@zzm.uzh.ch

### Specialty section:

This article was submitted to Craniofacial Biology and Dental Research, a section of the journal Frontiers in Physiology

Received: 15 December 2016 Accepted: 05 January 2017 Published: 19 January 2017

### Citation:

Natsiou D, Granchi Z, Mitsiadis TA and Jimenez-Rojo L (2017) Generation of Spheres from Dental Epithelial Stem Cells. Front. Physiol. 8:7. doi: 10.3389/fphys.2017.00007 Keywords: incisor, tooth, regeneration, epithelial stem cells, mouse, spheres, stem cell niche, 3D culture

# INTRODUCTION

Adult stem cells reside in specific and well-defined areas of most organs and tissues called stem cell niches. Niches provide to stem cells proper signals in order to regulate their function and maintenance according to the requirements of each specific tissue (Li and Xie, 2005; Pagella et al., 2015; Kirkeby et al., 2016). Therefore, the niches contain sources of undifferentiated cells that are involved in both tissue homeostasis and reparative processes after injury (Jiménez-Rojo et al., 2012; Van Keymeulen and Blanpain, 2012). The presence of stem cells is more important in organs with rapid renewal such as hairs, skin and intestine (Blanpain and Fuchs, 2014; Tetteh et al., 2015). The identification of different epithelial stem cell populations with various functions and plasticity during homeostasis and regeneration suggests the complexity of adult epithelial stem cell niches (Blanpain and Fuchs, 2014; Tetteh et al., 2015).

Although monolayer stem cell cultures were established more than three decades ago, three dimensional (3D) in vitro systems have recently emerged in order to preserve the physiological microenvironment of the cultured stem cells, thus providing an important tool for both basic and clinical research (Edmondson et al., 2014; Fatehullah et al., 2016; Kretzschmar and Clevers, 2016). Sphere-forming assays have been used to define stemness of adult epithelial cells within organs and tissues (e.g., intestine, mammary glands, and lungs; Dontu et al., 2003). In contrast to the two-dimensional (2D) monolayer culture, 3D culture systems allow cells to grow by forming aggregates/spheroids or organoids (Sasai et al., 2012; Edmondson et al., 2014).

In rodent incisors, dental epithelial stem cells reside in their most posterior part, the cervical loop area (Mitsiadis et al., 2007). The niche formed at the cervical loop is composed by a variety of epithelial cell populations and is separated from the surrounding mesenchyme by a basement membrane (Kieffer-Combeau et al., 2001; Mitsiadis et al., 2007). Hereby, we describe a technique to obtain epithelial dentospheres from the cervical loop of the continuously growing mouse incisor. This method allows evaluating the stemness and plasticity of dental epithelial cells, thus providing with essential information before proceeding with cell-based regenerative approaches in clinics.

**262**

# MATERIALS AND METHODS

All mice (C57Bl/6) were maintained and handled according to the Swiss Animal Welfare Law and the study was approved by the Cantonal Veterinary office, Zurich (License 11/2014).

# Dissection of Cervical Loops from Mouse Incisors (Stereomicroscope)


# Isolation of Single Dental Epithelial Cells (under the Laminar Flow)


# Embedding of Single Dental Epithelial Stem Cells in Matrigel (under the Laminar Flow)


# Harvesting the Resulting Spheres and Processing Them for Histology

	- xl. Analyse the slides with Leica DM6000 FS microscope and take pictures with the Leica DFC420C camera.

# RESULTS

We have described the method for the dissection of the cervical loop from mouse mandibular incisors and its further dissociation into a single cell suspension (**Figure 1**). We have cultured the isolated epithelial stem cells in a three-dimensional (3D) culture system. Epithelial dentospheres have been obtained by incubating dental epithelial stem cells in a 3D culture system using two different culture media (**Figure 2A**). Thereafter, we have analyzed histologically the dentospheres on 5 µm sections (**Figure 2B**). Interestingly, although the number of

the obtained spheres was not altered when using the two different media (**Figure 2C**), the morphology and size of the spheres (**Figures 2B,D**) was affected and was distinct according to the medium used. More precisely, the spheres formed in presence of Dulbecco's Modified Eagle's Medium/Nutrient F-12 Ham (DMEM/F12) medium containing Epidermal Growth Factor (EGF) and basic Fibroblast Growth Factor (bFGF) differentiated into a stratified squamous-like epithelium with a keratinized center (**Figure 2A**). In contrast, dental epithelial cells cultured in Keratinocyte serum free medium (KSFM) containing EGF and Bovine Pituitary Extract (BPE) formed smaller spheres that lacked any defined organization and not showing a squamous differentiation (**Figures 2A,B**). Histological sections were used to further analyse the protein expression profile of the spheres by both immunohistochemistry and immunofluorescence (**Figure 3**). Staining against the dental epithelial stem cell markers Sox2 and p63 indicate the presence of stem cells inside the epithelial dentospheres obtained after culture in presence of both media. In addition, cells within the spheres proliferate, as it is visualized by nuclear Bromodeoxyuridine (BrdU) incorporation, keep their epithelial identity (indicated by Keratin14 expression), and lack differentiated dental cells (indicated by the absence of Amelogenin). Interestingly, the presence of a set of cells expressing the epidermal terminal differentiation marker Keratin10 was observed within spheres cultured in DMEM/F12 implemented with a cocktail of B27/EGF/bFGF molecules (**Figure 3**). These differentiated cells were less than 15% of the total cell number within the sphere and were situated at the central part of the sphere (**Supplementary Figure 1**).

# DISCUSSION

Here we describe in detail the various steps leading to the generation of epithelial dentospheres, which were then analyzed histologically and molecularly. Sphere-forming assays were firstly used in the neural stem cell field, where neurospheres were generated from cells of the adult central nervous system (Reynolds and Weiss, 1992). Afterwards, these assays have been largely used to study the behavior and stemness of putative stem cells (Dontu et al., 2003; Yoshida et al., 2005; Edmondson et al., 2014). In teeth, spheres (or dentospheres) have been generated mostly from their mesenchymal component such as the dental pulp and the dental follicle (Miura et al., 2003; Sasaki et al., 2008; Stevens et al., 2008; Abe et al., 2011; Keeve et al., 2013). The formation of spheres originated from dental epithelial tissues retained less attention and therefore remains

FIGURE 3 | Molecular profile of epithelial dentospheres. Immunohistochemistry and immunofluorescence on epithelial dentospheres showing BrdU incorporation in proliferating cells, and expression of specific markers showing the epithelial identity (Krt10, Krt14) and stemness (p63, Sox2) of cells. Absence of Amelogenin indicates that cells within dentospheres are not differentiated into dental specific epithelial cells (ameloblasts). Abbreviations: BrdU, Bromodeoxyuridine; Krt10, Keratin10; Krt14, Keratin14. Scale bars: 30 µm.

poorly unexplored. The few realized studies have shown that dental epithelial cells isolated from the cervical loop area of postnatal mouse incisors are able to form spheres in serum-free 3D culture systems (Chang et al., 2013a,b; Chavez et al., 2013, 2014). Interestingly, the resulting spheres presented varying sizes and morphogies depending on the composition of the culture media used in the assays. For instance, when using a medium similar to that used previously for generating neurospheres, which consists of Dulbecco's Modified Eagle's Medium/Nutrient F-12 Ham (DMEM/F12) medium containing EGF and bFGF, dental epithelial stem cells formed well-delimited round-shaped spheres (Chavez et al., 2013, 2014). However, dental epithelial stem cells cultured in an epithelial-specific medium formed spheres that are smaller in size when compared to those obtained with the above mentioned medium (Chang et al., 2013a,b).

Here we show that appropriate selection of signaling molecules is important in order to allow maintenance and proliferation of stem cells within the spheres. Indeed, cells cultured in presence of DMEM/F12 with B27/EGF/bFGF generated spheres with multiple layers, which closely resemble to keratinized stratified squamous epithelial structures. In contrast, in presence of a medium composed by KSFM with EGF/BPE, the spheres adopted a more irregular morphology, where layers were not clearly distiguisable in histological sections. Immunohistochemistry and immunofluorecence revealed that spheres generated in both culture conditions contain cells that proliferate, as shown by BrdU labeling and, furthermore, express the stem cell markers Sox2 and p63. Spheres formed in presence of DMEM/F12 supplemented with B27/EGF/bFGF molecules exhibited a regionalization with clear undifferentiated and differentiated territories, both histologically and molecularly. Indeed, proliferative events and expression of specific stem cell markers were evident in the most external layers of the sphere, thus resembling to the basal layer of keratinized epithelial structures where stem cells reside. In the same spheres, differentiated epithelial cells located in the center could be considered as equivalent to cells of the suprabasal layers of a keratinized epithelium. In contrast, in spheres formed in presence of KSFM with EGF/BPE all cells exhibited a uniform morphology and expressed the Sox2 stem cell marker, but not a terminal differentiation marker, thus suggesting that these spheres contain only cells with stemness.

These results strongly suggest that the epithelial stemness has been retained in the spheres generated by both protocols used. This is reinforced by the capacity of cells originated by the spheres to de novo recreate spheres (secondary spheres) when placed in identical culture conditions.

# REFERENCES


In summary, the methods described here are valuable to assess the stemness of dental epithelial cells, making epithelial dentospheres one promising model for studying tooth pathology and regeneration.

# AUTHOR CONTRIBUTIONS

DN, ZG, and LJ performed the experiments. All authors contributed equally to the experimentation plan, writing and analyzing results.

# ACKNOWLEDGMENTS

This work was supported by the Swiss National Foundation (SNSF) grants 31003A\_135633 (TM) and 3100A0-118332 (ZG and TM), as well as by institutional funds from University of Zurich (DN, TM, and LJ).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00007/full#supplementary-material

Supplementary Figure 1 | Percentage of undifferentiated vs. differentiated cells within the spheres. Keratin10 positive terminally differentiated cells constitute less than 15% of the total cell number within the epithelial dentospheres.


**Conflict of Interest Statement:** The handling Editor declared a past coauthorship with one of the authors TM and states that the process nevertheless met the standards of a fair and objective review.

The other 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 © 2017 Natsiou, Granchi, Mitsiadis and Jimenez-Rojo. 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) or licensor 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.

# Expression Analysis of the Hippo Cascade Indicates a Role in Pituitary Stem Cell Development

Emily J. Lodge, John P. Russell, Amanda L. Patist, Philippa Francis-West and Cynthia L. Andoniadou\*

Craniofacial Development and Stem Cell Biology, Dental Institute, King's College London, London, UK

The pituitary gland is a primary endocrine organ that controls major physiological processes. Abnormal development or homeostatic disruptions can lead to human disorders such as hypopituitarism or tumors. Multiple signaling pathways, including WNT, BMP, FGF, and SHH regulate pituitary development but the role of the Hippo-YAP1/TAZ cascade is currently unknown. In multiple tissues, the Hippo kinase cascade underlies neoplasias; it influences organ size through the regulation of proliferation and apoptosis, and has roles in determining stem cell potential. We have used a sensitive mRNA in situ hybridization method (RNAscope) to determine the expression patterns of the Hippo pathway components during mouse pituitary development. We have also carried out immunolocalisation studies to determine when YAP1 and TAZ, the transcriptional effectors of the Hippo pathway, are active. We find that YAP1/TAZ are active in the stem/progenitor cell population throughout development and at postnatal stages, consistent with their role in promoting the stem cell state. Our results demonstrate for the first time the collective expression of major components of the Hippo pathway during normal embryonic and postnatal development of the pituitary gland.

### Edited by:

Thimios Mitsiadis, University of Zurich, Switzerland

### Reviewed by:

Eumorphia Remboutsika, BSRC Alexander Fleming, Greece Claudio Cantù, University of Zurich, Switzerland

### \*Correspondence:

Cynthia L. Andoniadou cynthia.andoniadou@kcl.ac.uk

### Specialty section:

This article was submitted to Craniofacial Biology, a section of the journal Frontiers in Physiology

Received: 10 February 2016 Accepted: 14 March 2016 Published: 31 March 2016

### Citation:

Lodge EJ, Russell JP, Patist AL, Francis-West P and Andoniadou CL (2016) Expression Analysis of the Hippo Cascade Indicates a Role in Pituitary Stem Cell Development. Front. Physiol. 7:114. doi: 10.3389/fphys.2016.00114 Keywords: pituitary, Hippo, YAP1, TAZ, pituitary stem cells, Rathke's pouch

# INTRODUCTION

The pituitary gland is a critical endocrine organ that controls multiple essential physiological processes such as metabolism, stress response, growth and reproduction. It is not surprising therefore, that abnormal pituitary function leads to human disease, including hypopituitarism and pituitary tumors, which can be associated with high morbidity and mortality. Hypopituitarism has an estimated prevalence of 45.5 per 100,000 (Schneider et al., 2007) and clinically relevant pituitary adenomas are reported to have a mean prevalence of 94 per 100,000 although up to one in six individuals are found to carry pituitary microadenomas (Ezzat et al., 2004; Daly et al., 2006). Understanding the genes and pathways that control normal pituitary development and function, and their likely involvement in disease, is required to speed up the discovery of new tools to improve patient management.

The pituitary develops from two discrete embryonic tissues; oral ectoderm, which gives rise to the endocrine anterior pituitary comprised of the intermediate and anterior lobes, and neural ectoderm (ventral diencephalon), which gives rise to the posterior pituitary, which is connected with the hypothalamus. In mice, the hypophyseal placode, the primordium of the anterior pituitary, is first identifiable as a thickening in the oral ectoderm at 8.0 days post coitum (dpc). From

**269**

9.0dpc the placode invaginates, forming the anterior pituitary primordium termed Rathke's pouch (RP). The overlying ventral diencephalon then evaginates toward and contacts RP by 10.5dpc to form the infundibulum (de Moraes et al., 2012; Rizzoti, 2015). Subsequently, RP detaches from the oral epithelium, to form the definitive pouch by 12.5dpc. The definitive pouch retains a central lumen that is lined by SOX2<sup>+</sup> uncommitted progenitor cells (de Moraes et al., 2012; Rizzoti, 2015). Descendants of these SOX2<sup>+</sup> cells restrict their fate to three lineages (Fauquier et al., 2008; Andoniadou et al., 2013; Rizzoti et al., 2013), which are characterized by expression of transcription factors PIT1, TPIT, and SF1. PIT1<sup>+</sup> progenitor cells differentiate into prolactin-secreting lactotrophs, growth hormone-secreting somatotrophs and thyroid-stimulating hormone-secreting thyrotrophs; TPIT<sup>+</sup> progenitors give rise to adrenocorticotrophic hormone-secreting corticotrophs in the AP and melanocyte-stimulating hormone-secreting melanotrophs in the IL. Lastly, SF1<sup>+</sup> progenitors produce luteinizing hormoneand follicle-stimulating hormone-secreting gonadotrophs. A proportion of SOX2<sup>+</sup> cells (3–5% of total pituitary cells) persist into adult life (Fauquier et al., 2008; Jayakody et al., 2012; Andoniadou et al., 2013). Postnatally, these SOX2<sup>+</sup> cells are predominantly found in a thin epithelial layer between the anterior and intermediate lobes of the pituitary (marginal zone), and groups of SOX2<sup>+</sup> cells are also dispersed within the parenchyma.

Multiple signals are required for correct pituitary development, however the activity or role of the Hippo pathway has not been previously studied. The infundibulum expresses FGF8, FGF10, and BMP2 from 9.0dpc (Treier et al., 1998, 2001), which diffuse to form a dorsal-ventral gradient. Changes in the extent of these expression domains within the infundibulum directly influence anterior pituitary size. In the absence of FGF signaling, RP is initially specified but cells fail to proliferate and undergo apoptosis (De Moerlooze et al., 2000; Ohuchi et al., 2000). FGF activity is mediated through the transcription factor LIM Homeobox 3 (LHX3) (Ericson et al., 1998), required for progenitor specification and proliferation (Sheng et al., 1997). SHH is expressed in the non-hypophyseal oral ectoderm and ventral diencephalon surrounding the infundibulum and signals to the developing RP (Treier et al., 2001; Khonsari et al., 2013). Loss of SHH signaling leads to a reduction in pituitary tissue, a phenotype attributed both to defective patterning and proliferation. The mesenchyme around the developing pituitary, derived from the neural crest rostrally and from the paraxial mesoderm caudally (Jiang et al., 2002; McBratney-Owen et al., 2008), expresses WNT and BMP signals that also influence morphogenesis, proliferation and cell-fate specification (Treier et al., 1998; Davis and Camper, 2007). WNT ligands have a role in promoting pituitary progenitor proliferation and PIT1-lineage specification as well as for correct expression of FGF and BMP factors (Cha et al., 2004; Potok et al., 2008; Gaston-Massuet et al., 2011; Andoniadou et al., 2013). Similarly, BMP2 and BMP4 are required for pituitary growth and lineage specification (Takuma et al., 1998; Treier et al., 1998).

The Hippo pathway regulates organ size through the control of stem cell activity, proliferation and apoptosis. The pathway is an inhibitory phosphorylation cascade first identified in Drosophila, where mutations in the Hippo (Hpo) kinase led to over-proliferation in imaginal discs and tissue overgrowth in the adult fly (Wu et al., 2003). The mammalian core Hippo pathway consists of MST1/MST2 kinases (Hpo homologs, a.k.a. STK4/STK3), which activate LATS1/LATS2 kinases, leading to phosphorylation of effectors YAP1 and TAZ (WWTR1) at multiple sites. These are then retained in the cytoplasm via 14- 3-3 (Hao et al., 2008) or ubiquitinated and degraded (Zhao et al., 2011). When the Hippo pathway is not active, YAP1 and TAZ can enter the nucleus and bind to transcription factors TEAD1-4 (Ota and Sasaki, 2008; Zhao et al., 2008), to activate transcription of stemness, proliferation and anti-apoptotic genes. Loss of function of core kinases leads to increased proliferation in several tissues, most obviously in the liver (Lu et al., 2010), heart (Heallen et al., 2011, 2013), and intestine (Cai et al., 2010; Imajo et al., 2015).

YAP1/TAZ regulate proliferation, survival and differentiation and are active in many stem cell populations including embryonic stem cells (Lian et al., 2010). Many factors have been implicated in the regulation of YAP1/TAZ via modulation of the core Hippo pathway. These include the proto-cadherins FAT4 and DCHS1 (Cappello et al., 2013; Bagherie-Lachidan et al., 2015) where loss of FAT4/DCHS1 has been reported to result in an increase in YAP1 and TAZ activity in neuronal cells and nephron progenitors. SOX2 has been shown to antagonize the Hippo pathway, leading to the nuclear accumulation of YAP1 and TAZ (Basu-Roy et al., 2015). Additionally, SOX2 has been shown to induce transcription of Yap1 in mesenchymal stem cells and osteoprogenitors (Seo et al., 2013), whilst in developing lungs, YAP1 can induce Sox2 expression (Mahoney et al., 2014). Therefore YAP1/TAZ is strongly associated with the stem cell state. In this manuscript we have analyzed in detail the expression of components of the Hippo pathway in the developing pituitary gland and demonstrate its activity in SOX2<sup>+</sup> cells during embryonic and postnatal development.

# MATERIALS AND METHODS

# Animals and Tissue Processing

Procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986, subject to KCL local Ethical Review. Wild type CD1 females were mated with wild type CD1 males for the generation of embryos. Sox2-Egfp animals have been previously described (Ellis et al., 2004). These were maintained as heterozygotes on a CD1 background. Midday of the day of vaginal plug was considered as 0.5 days post coitum (dpc). Dissected embryos and postnatal tissues were fixed in 10% neutral buffered formalin (NBF) at room temperature for 36 h, then dehydrated through a graded ethanol series and processed for paraffin embedding as previously described (Gaston-Massuet et al., 2008; Sajedi et al., 2008). Samples were sectioned along the sagittal plane for embryos between 9.5dpc and 13.5dpc and frontal plane for older embryos and postnatal pituitaries, at a thickness of 7 µm for immunofluoresence and 4 µm for RNAscope mRNA in situ hybridization.

# Immunofluorescence

Samples were dewaxed in histoclear twice for 10 min, followed by rehydration through a descending ethanol series then washed in water. Antigen retrieval was carried out in citrate-based Declere Unmasking Solution (Cell Marque) in a steamer twice for 30 min, followed by washing in PBT. For tyramide specific amplification [TSA, for antibodies against YAP1 (Cell Signaling 4912, 1:1000), pYAP (S127) (Cell Signaling 4911S, 1:1000) and TAZ (Sigma HPA007415, 1:1000)], slides were washed in TNT buffer (0.1M Tris-HCl, pH7.5, 0.15M NaCl, 0.05% Tween-20). Slides were blocked for 1 h in TNB [0.1M Tris-HCl pH7.5, 0.15M NaCl, 0.5% Blocking Reagent (FP1020, Perkin Elmer)], and incubated with primary antibody overnight at 4◦C in TNB. Following washes, species-specific biotinylated antibody was applied for 1 h at room temperature in TNB. Following washes in TNT, slides were incubated in ABC solution (Vector Laboratories, pk-6100) for 30 min in the dark then for 10 min at room temperature in TSA-Cy3 diluted in Stock Solution (Perkin Elmer, NEL760001). Slides were washed and mounted with soft-set mounting medium with DAPI (Vector Laboratories, Z1007) ready for imaging. For double immunofluorescence with GFP, the above conditions were used, with the inclusion of chicken anti-GFP primary antibody (Abcam ab13970, 1:350) and secondary goat anti-chicken Alexa Fluor 488 (Invitrogen A11039, 1:500). For antibodies against SOX2 (Abcam ab97959, 1:300) and Endomucin (Abcam ab106100, 1:500), blocking was carried out in Blocking Buffer (0.15% glycine, 2 mg/ml BSA, 0.1% Triton-X in PBS) with 10% sheep serum for minimum 1 h at room temperature. Primary antibody solution was applied overnight at 4◦C, diluted in Blocking Buffer with 1% sheep serum. Slides were washed and then incubated in goat anti-rabbit biotinylated (Abcam ab6720, for α-SOX2) or goat anti-rat Alexa Fluor 633 (Life Technologies A-21094, for α-Endomucin) for 1 h at room temperature, diluted to 1:350 in Blocking Buffer with 1% sheep serum. After washing, slides were incubated in Streptavidin-488 (Life Technologies S11223) at 1:500 dilution in Blocking Buffer with 1% sheep serum for 1 h a room temperature. Slides were washed and mounted as above.

# mRNA In situ hybridization

Tissue sections cut at 4 µm thickness were processed for mRNA in situ detection using the RNAscope 2.0 Fast Red Detection Kit (Advanced Cell Diagnostics), according to manufacturer's recommendations. For 10.5dpc and 12.5dpc embryos, pretreatment was carried out at the recommended "mild" timings and for older embryos or postnatal tissues, "standard" timings were used. RNAscope probes used: Hesx1, Sox2, Mst1 (Stk4), Mst2 (Stk3), Lats1, Lats2, Yap1, Tead1, Tead2, Tead3, Tead4, Fat3, Fat4, and Dchs1 (Advanced Cell Diagnostics). To control for background, we used a negative control probe against the Bacillus subtilis dihydrodipicolinate reductase, dapB (Advanced Cell Diagnostics) (Figure S1). Sections were weakly counterstained with hematoxylin.

# Microscopy

For fluorescent images, slides were visualized on a TCS SP5 confocal system (Leica Microsystems (UK) Ltd). Images were captured using a HCX Plan-Apochromat CS 20x/0.7 dry objective and HCX Plan-Apochromat CS 63x/1.3 Glycine objective (both Leica Microsystems (UK) Ltd). The DAPI, AlexaFluor 488, 594, and 633 conjugate dyes were excited with 405, 488, 561, and 633 nm lasers respectively. Z-stack images were acquired at a total thickness of 3 µm. Images were processed for maximum intensity z-projections using Fiji (Schindelin et al., 2012). For brightfield images, slides were scanned using a NanoZoomer-XR Digital slide scanner (Hamamatsu). Panels were compiled using Adobe Photoshop to create the figures.

# RESULTS

To determine reliability of the RNAscope in situ hybridization in the pituitary we first validated this method by analysis of Hesx1, which has a known pattern, only expressed during development. It is expressed strongly in oral ectoderm subsequently fated to become Rathke's pouch at 8.5dpc (Cajal et al., 2012), and persists in RP epithelium until 11.5dpc with levels of expression decreasing thereafter (Thomas and Beddington, 1996). Using the RNAscope in situ hybridization method for sensitive mRNA detection, we confirmed strong expression of Hesx1 in RP at 10.5dpc (**Figure 1A**, arrowhead), which extended rostrally in the oral epithelium. No expression was detected in the pharyngeal endoderm as previously reported (posterior limit of expression noted by arrow). Expression of Hesx1 was barely detectable at 12.5dpc and 13.5dpc (**Figures 1B,C**) with presence of only sporadic transcripts (**Figure 1C**′ magnified boxed region in C). Additionally, we investigated expression of Sox2, which marks progenitors/stem cells in the pituitary, using this method. Robust expression of Sox2 transcripts was detected in RP and the developing ventral diencephalon, with complete absence of expression in mesenchyme surrounding the pouch at all stages (**Figure 1**). Expression of Sox2 is known to be downregulated in committed lineages of the pituitary gland, which we confirmed at 13.5dpc; there is an absence of transcripts in the ventral anterior pituitary where cells are undergoing commitment (**Figure 1F**, asterisk). Expression persisted dorsally, specifically in the marginal epithelium surrounding the pouch where the uncommitted cells reside (arrowheads in **Figure 1F**).

# Expression of Upstream Hippo Regulators

We next sought to characterize the expression of proposed upstream regulators of the Hippo cascade, homologs of Drosophila Ds and Ft, whose protein products act as ligandreceptor pair. We analyzed expression of Dchs1 and Fat4 that have closest homology to Ft and Ds, as well as Fat3, which is detectable in developing pituitary tissue (Karine Rizzoti, personal communication). Previous studies have reported absence of expression of the remaining homologs Fat1, Fat2, and Dchs2 in the pituitary gland (Diez-Roux et al., 2011). At 10.5dpc we did not observe Dchs1 expression in RP epithelium but transcripts were detected in the caudal mesenchyme as well as in the ventral diencephalon (**Figure 2A**). Expression in RP was observed at 12.5dpc at low levels and persisted until at least 17.5dpc where it was detected both in the anterior and posterior pituitary and in the hypothalamus, with lowest expression in cells lining the third ventricle (**Figures 2B–E**). At 10.5dpc we

observed robust expression of Fat3 in the developing pouch, in ventral diencephalon and caudal mesenchyme (**Figure 2F**). Expression in RP persisted at lower levels until at least 17.5dpc (**Figures 2G–J**, arrowheads in **Figures 2G,H**), where strongest expression was detected in cells lining the third ventricle (arrowheads in **Figures 2I,J**) as well as in posterior pituitary tissue (arrow in **Figure 2J**). Fat4 transcripts were present from 10.5dpc throughout RP (arrowheads in **Figure 2K**) and oral epithelium but excluded from the pharyngeal endoderm (arrow). Strong expression was detected in surrounding mesenchyme. At 12.5dpc very strong expression was detected at the rostral tip (arrowheads in **Figure 2L**) with low levels of transcripts in RP epithelium. There was also expression in the infundibulum (arrow in **Figures 2L,M**) and surrounding mesenchyme. Fat4 was still expressed at 17.5dpc in sporadic cells of the anterior and posterior pituitary (arrows in **Figures 2N,O**) and surrounding mesenchyme (arrowheads in **Figure 2N**).

# Expression of Pathway Kinases

In multiple tissues, MST1 and MST2 have redundant functions. At all stages analyzed, expression of both Mst1 and Mst2 was observed throughout the developing pituitary between 10.5dpc and 17.5dpc (**Figures 3A–J**). We detected salt and pepper expression in Rathke's pouch, the infundibulum and their subsequent derivatives. Both genes were also expressed in neural structures from 10.5dpc with few sporadic cells displaying transcripts in surrounding mesenchyme at 10.5dpc and 12.5dpc.

We observed strong Lats1 expression at all stages throughout the oral epithelial-derived and neural tissues (**Figures 3K–O**). From 13.5dpc we observed a ventral bias in the developing pituitary, with the dorsal epithelium surrounding the lumen (future intermediate lobe) displaying lower expression (arrowhead in **Figure** 3**M**). As seen for Mst1/Mst2, expression in the mesenchyme was only detected in occasional single cells at 10.5dpc but was abundant from 12.5dpc. Expression of Lats2 was very low at all stages analyzed, across the developing pituitary and surrounding tissues (**Figures 3P–T**), but detectable (**Figure 3T**′ , magnified boxed area in **Figure 3T**).

# Expression of Hippo Pathway Effectors

When YAP1 and TAZ are not phosphorylated by LATS kinases they can associate with TEAD transcription factors in the nucleus to promote expression of target genes. We sought to determine the expression patterns of Yap1 and Tead1-Tead4 in the developing pituitary. We were not able to determine the expression of Taz mRNA using this method. Yap1 showed robust expression in Rathke's pouch and surrounding tissues at 10.5dpc (**Figure 4A**). By 12.5dpc there was strong expression in the dorsal aspect of the pouch in the epithelium (**Figure 4B** arrowhead), reduced expression in the expanding ventral portion and no expression in the rostral tip (arrow). This pattern was maintained at 13.5dpc and new tissue in the ventral region that is undergoing commitment showed low expression (arrowhead in **Figure 4C**). At 15.5 and 17.5dpc, Yap1 transcripts remained strong in the intermediate lobe, marginal zone of the anterior lobe and in scattered groups of cells throughout the anterior lobe (**Figures 4D,E**, black arrowheads). Yap1 was expressed in neural tissue and surrounding mesenchyme at all stages, maintained until 17.5dpc when it was expressed in the posterior lobe, the cell layer surrounding the third ventricle (white arrowheads in **Figures 4D,E**) and in mesenchyme-derived connective tissue surrounding the gland. From the four Tead

genes, Tead2 expression was the strongest (**Figures 4K–O**). This was reminiscent of the expression pattern of Yap1: strong expression in RP and all surrounding tissues at 10.5dpc (**Figure 4K**), strong expression in RP epithelium at 12.5dpc and 13.5dpc (**Figures 4L,M**, arrowheads) but no expression in the rostral tip (**Figures 4L,M**, arrows) and reduced expression in the ventral pituitary primordium at 13.5dpc (**Figure 4M**, asterisk). Tead2 expression remained very high at 15.5dpc in the marginal zone (**Figure 4N**, black arrowhead) and intermediate lobe and in scattered groups of cells throughout the anterior lobe. Levels of expression were reduced but present at 17.5dpc (**Figure 4O**, black arrowhead). The posterior lobe also expressed high levels of Tead2, as did cells surrounding the third ventricle (**Figures 4N,O**, white arrowheads). Tead1 and Tead3 were expressed at low levels at all stages examined (**Figures 4F-J, P-T**) and Tead4 was not expressed (**Figures 4V-Y**) except for low levels of transcript in RP at 10.5dpc only (**Figures 4U,U**′ arrowheads).

# Localization of YAP1 and TAZ Proteins

In order to infer activity of the Hippo kinase cascade, we investigated the localization of effector proteins TAZ, total YAP1, as well as the inactive phosphorylated form of YAP1 (S127). TAZ and YAP1 had similar localization at 10.5dpc; they appeared nucleo-cytoplasmic with a bias for the apical cytoplasm of RP epithelium (**Figures 5A,F**, arrowheads). Inactive YAP1, marked by pYAP1 was strongly cytoplasmic and also displayed an apical bias (**Figure 5K**, arrowheads). All three antibodies marked cells in the mesenchyme and neural tissue. At 12.5dpc and 13.5dpc YAP1 and TAZ both localized mostly in nuclei of cells in RP epithelium, with stronger expression in the dorsal RP epithelium at 12.5dpc, which persisted at 13.5dpc for YAP1 (yellow arrowheads in **Figures 5B,G,H**). In the ventral portion of the epithelium there was nuclear localization in a thin cell layer surrounding the cleft. Little expression was observed in more ventral regions (asterisk in **Figures 5B,C,G,H**), and no expression in the rostral tip (arrowheads in **Figures 5B,C,G,H**). Phosphorylated YAP1 was cytoplasmic in cells both in the dorsal and ventral regions at both stages but completely absent from the rostral tip (arrowheads in **Figures 5L,M**). Expression was stronger in the ventral epithelium than the dorsal (yellow arrowheads in **Figures 5L,M**), the reverse of the observed pattern for total YAP1 and TAZ. At 15.5dpc and 17.5dpc YAP1 and

detected in surrounding mesenchyme. (P–T) Very low levels of Lats2 transcripts are detected in all tissues at all stages analyzed (note positive expression in T', magnification of boxed region in T at 17.5dpc). Abbreviations: rp, Rathke's pouch; vd, ventral diencephalon; m, mesenchyme; inf, infundibulum; sph, sphenoid; rt, rostral tip; pl, posterior lobe; al, anterior lobe; il, intermediate lobe; hy, hypothalamus; 3v, third ventricle. Sagittal sections between 10.5dpc-13.5dpc and frontal between 15.5dpc and 17.5dpc. Axes in (A) applicable to (A–C,F–H,K–M,P–R: d, dorsal; v, ventral; r, rostral; c, caudal). Axes in (D) applicable to (D,E,I,J,N,O,S,T: d, dorsal; v, ventral; ri, right; le, left). Scale bars 200 µm.

TAZ were nucleo-cytoplasmic in the marginal zone epithelium surrounding the cleft on both sides, in the intermediate and anterior lobes (arrowheads in **Figures 5D,E,I,J**). TAZ protein was detected in a broader domain surrounding the epithelium than YAP1 (**Figure 5E**). They were both present in cells scattered around the anterior pituitary and in structures resembling blood vessels (arrows in **Figures 5D,E,J**). Inactive phosho-YAP1 was present in the cytoplasm of cells in the marginal zone epithelium (arrowheads in **Figures 5N,O**) and in many cells throughout the anterior and intermediate lobes at both stages. The posterior lobe stained with all three antibodies, nucleo-cytoplasmic for TAZ and cytoplasmic for YAP1 and phospho-YAP1.

We next investigated the localization of TAZ, YAP1, and phospho-YAP1 in sections of postnatal pituitary glands at P21, a time following the peak postnatal proliferative stage, when the gland is still expanding and undergoing major hormonal profile changes at weaning. We found that YAP1 and TAZ were primarily nuclear in epithelial cells lining the pituitary cleft (**Figures 6B,C**), which express the stem cell marker SOX2 (**Figure 6A**, arrowheads). Nucleo-cytoplasmic staining was also seen in cells associated with blood vessels (arrows in inserts in **Figures 6B,C**). The pattern of blood vessels was revealed by staining using antibodies against endomucin (**Figure 6A**, arrows). Nucleo-cytoplasmic staining for YAP1 and TAZ was also seen in scattered cells throughout the anterior lobe, where higher proportions were positive for TAZ (yellow arrowheads in **Figure 6B**, inserts). Inactive phosphorylated YAP1 was localized in the cytoplasm of cells in the marginal zone epithelium, stronger in some regions (arrowheads in **Figure 6D** indicating stronger staining). There was diffuse expression in the anterior pituitary and cytoplasmic staining in cells lining blood vessels (arrows in **Figure 6D**). In order to determine localization of TAZ, YAP1 and pYAP1 specifically in SOX2<sup>+</sup> cells we carried out double immunofluorescence staining on sections at P21 from Sox2Egfp/<sup>+</sup> animals, using antibodies against GFP to mark SOX2<sup>+</sup> cells (**Figures 6E–G**). We find nuclear localisation of both TAZ and YAP1 in cells positive for GFP (arrows in **Figures 6E,F**). Levels of pYAP1 in the epithelium do not correlate with GFP

FIGURE 4 | Expression of the Hippo pathway effectors during embryonic development of the pituitary gland. RNAscope mRNA in situ hybridization using probes against Yap1, Tead1, Tead2, Tead3, and Tead4 on sections of wild type CD1 embryos between 10.5dpc and 17.5dpc. (A–E) Yap1 transcripts are detected in neural tissue, mesenchyme and Rathke's pouch epithelium between 10.5dpc and 13.5dpc (A–C, arrowheads indicating RP expression). Note the dorsal expression bias at 12.5dpc and 13.5dpc and absence of transcripts in the rostral tip (arrows in B,C). Transcripts persist in all tissues at 15.5dpc and 17.5dpc especially in the periluminal region (black arrowhead in D) and epithelium surrounding the third ventricle (white arrowheads in D,E). (F–Y) Expression of Tead1, Tead2, Tead3, and Tead4 encoding TEAD transcription factors. Tead1 and Tead3 transcripts are detectable in all tissues at low levels (F–J,P–T), higher in RP (arrowheads in F,H,P–R). Tead2 is highly expressed in all tissues at 10.5dpc (K), and from 12.5dpc becomes restricted to the ventral diencephalon in neural tissue (L,M) and to the epithelium surrounding the third ventricle (white arrowheads in N,O). Tead2 is strongly expressed in the periluminal epithelium of the anterior pituitary primordium (black arrowheads L–O) but excluded from the rostral tip (arrows L,M). Tead4 transcripts are barely detectable (U–Y). Abbreviations: rp, Rathke's pouch; vd, ventral diencephalon; m, mesenchyme; inf, infundibulum; sph, sphenoid; rt, rostral tip; pl, posterior lobe; al, anterior lobe; il, intermediate lobe; hy, hypothalamus; 3v, third ventricle. Sagittal sections between 10.5dpc and 13.5dpc and frontal between 15.5dpc and 17.5dpc. Axes in (A) applicable to (A–C,F–H,K–M,P–R,U–W: d, dorsal; v, ventral; r, rostral; c, caudal). Axes in (D) applicable to (D,E,I,J,N,O,S,T,X,Y: d, dorsal; v, ventral; ri, right; le, left). Scale bars 200 µm.

positivity, with some GFP<sup>+</sup> cells showing stronger staining for pYAP1 protein (yellow arrowheads in **Figure 6G**) and lower levels in others (green arrowheads in **Figure 6G**).

# DISCUSSION

Coordinating proliferation, differentiation and cell death is critical for normal development of tissues and for maintaining the balance of cells during long-term homeostasis. The Hippo kinase cascade has been shown to mediate these processes through the inhibition of proliferation and promotion of differentiation and cell death. In this manuscript we reveal that the Hippo signaling cascade is active during all stages of embryonic pituitary development assessed and continues to act in the postnatal organ.

The genes Mst1 and Mst2 encoding the core Hippo kinases, are both expressed throughout the developing pituitary and Lats1 is expressed at high levels during development. Since we barely detected expression of Lats2, we hypothesize the main kinase upstream of YAP1/TAZ in the gland is likely to be Lats1. From the four Tead genes that encode the pathway transcription factors, Tead2 is the highest expressed making it likely to act as the main regulator of downstream target gene transcription. Interestingly, expression of Yap1 is very similar to Tead2, which are both

FIGURE 5 | Expression of YAP1 and TAZ proteins during embryonic development. Immunofluorescence using specific antibodies against total TAZ protein, total YAP1 protein and phosphorylated YAP1 (S127) in red. Nuclei are counterstained with DAPI (blue). (A–J) Localization of effectors TAZ (A–E) and YAP1 (F–J) using antibodies recognizing total protein. Note the nuclear localization in periluminal cells of Rathke's pouch (yellow arrowheads in B,C,G,H, white arrowheads in D,E,I,J). No YAP1/TAZ proteins are detected in the rostral tip (arrowheads in B,C,G,H) and there is a reduction in expression in ventral regions (asterisks in B,C,G,H). Note expression in structures resembling capillaries from 15.5dpc (arrows in D,E,J). (K–O) Immunofluorescence to detect the phosphorylated form of YAP1 at S127. Protein detection indicates Hippo kinase cascade activity at all stages, primarily in periluminal RP epithelium (white arrowheads in K,N,O). Note the ventral bias of protein localization at 12.5dpc and 13.5dpc (yellow arrows in L,M) and complete absence of protein from the rostral tip (white arrowheads in L,M). Boxed inserts show higher magnifications of the epithelium. Examples of cytoplasmic localization are noted by white arrows and examples of nuclear localization by yellow arrows. Abbreviations: rp, Rathke's pouch; vd, ventral diencephalon; m, mesenchyme; oe, oral ectoderm; inf, infundibulum; sph, sphenoid; rt, rostral tip; pl, posterior lobe; al, anterior lobe; il, intermediate lobe. Sagittal sections between 10.5dpc and 13.5dpc and frontal between 15.5dpc and 17.5dpc. Axes in (A) applicable to (A–C,F–H,K–M: d, dorsal; v, ventral; r, rostral; c, caudal). Axes in (D) applicable to (D,E,I,J,N,O: d, dorsal; v, ventral; ri, right; le, left). Scale bars 100 µm and 20 µm in boxed inserts.

strongest expressed in the regions rich in stem/progenitor cells. Expression of both is completely absent from the rostral tip of the pituitary at 12.5dpc and 13.5dpc, as is expression of YAP1 and TAZ proteins and of phosphorylated YAP1, despite positive expression of the kinases Mst1, Mst2, and Lats1 in this tissue. This suggests that the Hippo cascade is not regulating the rostral tip, but this region highly expresses Fat4, which can act as a receptor upstream of the Hippo cascade. Taken together, we hypothesize that Fat4 is not acting upstream of Hippo in the rostral tip during development. Expression of Dchs1 that complements the receptor-ligand interaction is strong in mesenchyme surrounding Rathke's pouch, likely acting in concert with FAT4 at the rostral tip. Both genes are expressed at low levels in the anterior pituitary and a possible role upstream of the Hippo cascade cannot be excluded. From this gene family, Fat3 is also a candidate to encode a protocadherin upstream of the pathway; its expression in RP resembles that of Yap1 and Tead2, as well as Sox2. These demonstrate a dorsal bias in expression, at the region of the future intermediate lobe and stem cell-containing region of the anterior lobe. Expression in the stem cell-rich periluminal zone persists in the anterior pituitary at later stages.

We observe strong nuclear localization of YAP1 and TAZ throughout the stem cell-rich regions of developing Rathke's pouch and the postnatal anterior pituitary and reveal that phosphorylation of YAP1 at S127 occurs, suggesting kinase activity within the SOX2<sup>+</sup> stem cell pool. The S127 residue is in one of the five LATS phosphorylation consensus motifs and results in YAP1 regulation by the Hippo pathway through 14-3-3 binding and cytoplasmic retention (Zhao et al., 2007). Activity of the kinase cascade on the stem/progenitor pool in the pituitary gland may function to regulate stem cell numbers or behavior. Additionally, we observe localization of all three proteins in blood vessels in the gland, which will need to be taken into account during any interpretation of future functional data for this pathway in the pituitary. At 12.5dpc the ventral epithelium of Rathke's pouch has stronger pYAP1 expression, suggesting that the Hippo cascade may have higher activity in the ventral, more committed aspect, corroborated by stronger nuclear expression of total YAP1 and TAZ in the dorsal epithelium. Several inputs can influence Hippo cascade activity, such as mechanotransduction, polarity and G-proteincoupled receptor signaling. More recently, negative regulation of the cascade by the transcription factor SOX2 was shown, which is expressed by many stem cell types, including pituitary stem cells throughout development and postnatal stages. SOX2 has been shown to antagonize the function of the Hippo cassette in two ways: by directly regulating Yap1 transcription (Seo et al., 2013) as well as by antagonizing NF2 and WWC1, homologs of Merlin and Kibra respectively, that positively regulate MST1/2 thus resulting in reduced phosphorylation of YAP1/TAZ (Basu-Roy et al., 2015). This places SOX2 as a likely candidate upstream

of the Hippo cascade in the pituitary gland, and its potential role to maintain the stem/progenitor cell state through this pathway can be addressed in future.

Our results suggest there is appropriate expression of Hippo pathway components to support a possible functional role in the pituitary gland. A previous study reporting over-proliferation of uncommitted pituitary tissue in the absence of LATS1 kinase (St John et al., 1999) supports that this pathway may be acting to regulate homeostasis in the pituitary. It remains to be determined if the function of LATS1 in the gland is mediated through YAP1/TAZ. In summary, our results highlight that Hippo signaling is active in the pituitary gland, both during development and at postnatal stages and reveal the expression patterns of its major components. Our studies suggest a possible role for this pathway in the regulation of the uncommitted stem/progenitor cell pool, with its function remaining to be elucidated.

# AUTHOR CONTRIBUTIONS

CA, EL conceived the study; EL, JR conducted the experiments with support from AP; EL, JR, and CA analyzed the results. PF provided intellectual contribution. CA, EL wrote the manuscript. All authors reviewed and approved the final manuscript.

# ACKNOWLEDGMENTS

This work was supported by the Medical Research Council (MR/L016729/1), The Royal Society (RG130699) and Society for Endocrinology Early Career Grant to CA. EL is supported by the King's Bioscience Institute and the Guy's and St Thomas' Charity Prize PhD Programme in Biomedical and Translational Science. JR is supported by a Diana Trebble PhD Scholarship. The authors state that they have no existing conflicts of interest. The authors thank Prof. JP Martinez-Barbera and Prof. Abigail Tucker for critical reading of the manuscript.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2016.00114

Figure S1 | Negative controls for RNAscope method in the pituitary gland. RNAscope mRNA in situ hybridization using probes against dapB, encoding bacterial dihydrodipicolinate reductase, on wild type CD1 embryos at stages between 10.5dpc and 13.5dpc. (A–E) Represenative examples of negative controls used in rounds of RNAscope to determine background levels of expression. For each stage analyzed, rare to no red dots were observed. Inserts are magnifications of boxed regions. Axes in (A) applicable to (A–C); axes in (D) applicable to (D–E). Scale bars 250 µm in (A–D) and 100 µm in (E).

# REFERENCES


**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 © 2016 Lodge, Russell, Patist, Francis-West and Andoniadou. 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) or licensor 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.

digital media

of impactful research

article's readership