EXTRACELLULAR ENZYMES IN AQUATIC ENVIRONMENTS: EXPLORING THE LINK BETWEEN GENOMIC POTENTIAL AND BIOGEOCHEMICAL CONSEQUENCES

EDITED BY : Maria Montserrat Sala, Judith Piontek, Sonja Endres, Anna Maria Romani, Sonya Dyhrman and Andrew Decker Steen PUBLISHED IN : Frontiers in Microbiology and Frontiers in Marine Science

#### Frontiers Copyright Statement

© Copyright 2007-2019 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 978-2-88963-004-2 DOI 10.3389/978-2-88963-004-2

#### 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

## EXTRACELLULAR ENZYMES IN AQUATIC ENVIRONMENTS: EXPLORING THE LINK BETWEEN GENOMIC POTENTIAL AND BIOGEOCHEMICAL CONSEQUENCES

Topic Editors:

Maria Montserrat Sala, Institut de Ciències del Mar (ICM-CSIC), Spain Judith Piontek, Leibniz Institute for Baltic Sea Research Warnemünde, Germany Sonja Endres, Business Development and Technology Transfer Corporation of Schleswig-Holstein (WTSH), Germany Anna Maria Romani, University of Girona, Spain Sonya Dyhrman, Columbia University, United States; Lamont-Doherty Earth Observatory, United States

Andrew Decker Steen, University of Tennessee, United States

Image: Maria Montserrat Sala

Microbial extracellular enzymes are fundamental to the cycling of elements in aquatic systems. The regulation of these enzymatic reactions in oceans, lakes and streams is under complex multiple control by environmental factors and the metabolic capacities of different taxa and communities. While the environmental control of enzyme-mediated processes has been investigated for over 100 years, in recent years tremendous progress in techniques to characterize the metabolic potential of microbial communities ("omics" techniques) has been made, such as high-throughput sequencing and new analytical algorithms.

This book explores the controls, activities, and biogeochemical consequences of enzymes in aquatic environments. It brings together experimental studies and fieldwork conducted with natural microbial communities in marine and freshwater ecosystems as well as physiological, biochemical and molecular studies on microbial

communities in these environments, or species isolated from them. Additionally, the book contributes to the ongoing debate on the impact of anthropogenic climate change and pollution on microbes, extracellular enzymes and substrate turnover.

Citation: Sala, M. M., Piontek, J., Endres, S., Romani, A. M., Dyhrman, S., Steen, A. D., eds. (2019). Extracellular Enzymes in Aquatic Environments: Exploring the Link Between Genomic Potential and Biogeochemical Consequences. Lausanne: Frontiers Media. doi: 10.3389/978-2-88963-004-2

# Table of Contents


Avery Bullock, Kai Ziervogel, Sherif Ghobrial, Shannon Smith, Brent McKee and Carol Arnosti


Jaroslav Vrba, Markéta Macholdová, Linda Nedbalová, Jiří Nedoma and Michal Šorf

*137 Identification and Characterization of a Novel Salt-Tolerant Esterase From the Deep-Sea Sediment of the South China Sea*

Yi Zhang, Jie Hao, Yan-Qi Zhang, Xiu-Lan Chen, Bin-Bin Xie, Mei Shi, Bai-Cheng Zhou, Yu-Zhong Zhang and Ping-Yi Li

*147 Characterization of a New S8 Serine Protease From Marine Sedimentary*  Photobacterium *sp. A5–7 and the Function of its Protease-Associated Domain*

Hui-Juan Li, Bai-Lu Tang, Xuan Shao, Bai-Xue Liu, Xiao-Yu Zheng, Xiao-Xu Han, Ping-Yi Li, Xi-Ying Zhang, Xiao-Yan Song and Xiu-Lan Chen

# Editorial: Extracellular Enzymes in Aquatic Environments: Exploring the Link Between Genomic Potential and Biogeochemical Consequences

Maria M. Sala<sup>1</sup> \*, Judith Piontek <sup>2</sup> , Sonja Endres <sup>3</sup> , Anna M. Romani <sup>4</sup> , Sonya Dyhrman5,6 and Andrew D. Steen<sup>7</sup>

<sup>1</sup> Department of Marine Biology and Oceanography, Institut de Ciències del Mar (ICM-CSIC), Barcelona, Spain, <sup>2</sup> Biological Oceanography, Leibniz Institute for Baltic Sea Research Warnemünde, Rostock, Germany, <sup>3</sup> Business Development and Technology Transfer Corporation of Schleswig-Holstein (WTSH), Kiel, Germany, <sup>4</sup> Department of Environmental Sciences, Institute of Aquatic Ecology, University of Girona, Girona, Spain, <sup>5</sup> Department of Earth and Environmental Sciences, Columbia University, New York, NY, United States, <sup>6</sup> Lamont-Doherty Earth Observatory, Palisades, NY, United States, <sup>7</sup> Departments of Microbiology and Earth and Planetary Sciences, University of Tennessee, Knoxville, TN, United States

#### Edited by:

Hongyue Dang, Xiamen University, China

#### Reviewed by:

George S. Bullerjahn, Bowling Green State University, United States

> \*Correspondence: Maria M. Sala msala@icm.csic.es

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 15 April 2019 Accepted: 11 June 2019 Published: 26 June 2019

#### Citation:

Sala MM, Piontek J, Endres S, Romani AM, Dyhrman S and Steen AD (2019) Editorial: Extracellular Enzymes in Aquatic Environments: Exploring the Link Between Genomic Potential and Biogeochemical Consequences. Front. Microbiol. 10:1463. doi: 10.3389/fmicb.2019.01463 Keywords: enzymes, microorganisms, aquatic environment, degradation, bacterial activity

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

#### **Extracellular Enzymes in Aquatic Environments: Exploring the Link Between Genomic Potential and Biogeochemical Consequences**

Microbes drive the Earth's biogeochemical cycles, exerting profound control over the global cycling of carbon and other elements (Falkowski et al., 2008). In aquatic systems, the importance of microbial extracellular enzymes to the mobilization, transformation, and turnover of organic and inorganic compounds in aquatic environments has been proved since the 80's (Hoppe, 1983; Chróst, 1989) and was summarized in the book "Microbial enzymes in aquatic environments" (Chróst, 1991). Since then, the field has advanced considerably, with new observations, assay methods, and molecular-level studies (Arnosti et al., 2014) and the measurement of extracellular enzyme activities has become standard in many labs. We now have rates of enzymatic activities in a wide variety of freshwater and marine environments, from polar to tropical and from surface to deep ocean, and from isolates obtained even from extreme environments. Additionally, in recent years measurement of enzyme activities has become an important tool to assess the impact of anthropogenic changes on microbial communities and biogeochemical cycles, such as in the events of oil spills, acidification, or global warming (e.g., Piontek et al., 2010; Sala et al., 2016; Ziervogel et al., 2016; Freixa et al., 2017).

In this special issue, twelve articles highlight new findings on extracellular enzyme activities in aquatic environments, bringing together experimental and field studies conducted in marine and freshwater ecosystems as well as physiological, biochemical, and molecular studies on microbial communities or species isolated from those environments.

The first part of the volume is devoted to field studies, both in marine and freshwater ecosystems. To begin, the perspective article by Baltar advocates the need to go "beyond the living things" and study cell-free enzymatic activities to fully constrain the future and evolution of marine biogeochemical cycles. In marine environments, Hoarfrost and Arnosti show that the spectrum of substrates hydrolyzed in mesopelagic and deep waters of the Atlantic Ocean is positively related to the strength of stratification depth patterns, which may influence the efficiency of the biological carbon pump. Apart from their enzyme activities, knowing the types of bacteria that metabolize polymers can help make the critical connection between the taxonomic composition of microbial communities and their biogeochemical function. Liu et al. show, by using stable isotope probing, that a more diverse group of bacteria is involved in metabolizing peptides in normoxic surface water than in hypoxic bottom seawater from the Gulf of Mexico. In freshwaters, kinetic measurements of 5 substrates for exo- and endo-acting extracellular peptidases in 28 freshwater bodies in the Pocono Mountains (Mullen et al.) show variable ratios between aminopeptidases (APs) and trypsin, highlighting that measuring only Leu- AP activity may underestimate the total peptidolytic capacity in an environment. Spatial, but not seasonal, variability is also observed in a multiseason investigation in two North Carolina rivers examining the activities of extracellular enzymes used to hydrolyze polysaccharides and peptides (Bullock et al). Collectively, these studies expand our understanding of the role of microbial enzymes in the biogeochemistry of aquatic ecosystems.

Experimental approaches are increasingly used to tease apart the complex role aquatic microbial enzymes have on the biogeochemistry and functioning of ecosystems. Traving et al. in a mesocosm experiment observed the effect of increased loads of dissolved organic matter (DOM) in bacterioplankton community composition and a stimulation of protease activity. This suggests that parts of future elevated riverine DOM supply to the Baltic Sea will be efficiently mineralized by microbes and will have consequences in bacterioplankton and phytoplankton community composition and function. Indeed, organic matter released by phytoplankton fuels bacterial growth and the transformation of this DOM plays a role in the formation of chromophoric DOM which is ubiquitous in the ocean. Kinsey et al. investigated CDOM formation mediated by microbial processing of phytoplankton-derived aggregates. Measurements of hydrolytic enzyme rates along with the fluorescent properties of organic matter suggest that

#### REFERENCES


bacterial degradation activity changes the composition of chromophoric dissolved organic matter to more humic-like compounds. Another experimental study by Kamalanathan et al. reports on the role of bacterial extracellular enzymes during exposure to hydrocarbons and dispersant in mesocosm tanks and observed enhanced EPS (extracellular polymeric substances) production and extracellular enzyme activities in the oil amended treatment. These studies, which span both the field and laboratory, highlight the role of microbial enzymes in processing organic matter and the ramifications for diversity, resource remineralization, and community responses to anthropogenic perturbations.

Two of the papers in this issue address methodological concerns or improvements on enzyme activity measurements. Obayashi et al. use kinetic experiments to suggest improvements in material and methods for measurements of extracellular protease activities, and specifically highlight the relevance of using low protein binding materials. Also, Vrba et al. propose the fluorescence-labeled enzyme activity (FLEA) assay based in a novel substrate, ELF97 phosphate, that allows tagging extracellular phosphatase activity on single cells in an epifluorescence microscope and has shown to be useful in green algae cultures. This assay is shown being a strong tool for exploring plankton P metabolism.

The last part of the issue is focused on the characterization of new enzymes: a novel salt-tolerant esterase from deep-sea sediment of the South China Sea (Zhang et al.) and a new S8 serine protease from marine sedimentary Photobacterium sp. A5-7 (Li et al.).

The broad-range of articles presented in this topic deepens our understanding on the controls, activities, and biogeochemical consequences of microbial enzymes in aquatic environments. This body of work has, not only increased our knowledge, but also identified challenges and questions that remain open and should be addressed in the future.

#### AUTHOR CONTRIBUTIONS

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

### FUNDING

This study was supported by grant ANIMA (CTM2015-65720-R) funded by the Spanish Government to MS and by the NSF grant OCE-1464392 to AS.


following the Deepwater Horizon oil spill. Deep Sea Res. II 129, 241–248. doi: 10.1016/j.dsr2.2014.04.003

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

Copyright © 2019 Sala, Piontek, Endres, Romani, Dyhrman and Steen. 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.

fmicb-08-02438 December 28, 2017 Time: 15:56 # 1

# Watch Out for the "Living Dead": Cell-Free Enzymes and Their Fate

#### Federico Baltar1,2 \*

<sup>1</sup> Department of Marine Science, University of Otago, Dunedin, New Zealand, <sup>2</sup> NIWA/University of Otago Research Centre for Oceanography, Dunedin, New Zealand

Microbes are the engines driving biogeochemical cycles. Microbial extracellular enzymatic activities (EEAs) are the "gatekeepers" of the carbon cycle. The total EEA is the sum of cell-bound (i.e., cell-attached), and dissolved (i.e., cell-free) enzyme activities. Cell-free enzymes make up a substantial proportion (up to 100%) of the total marine EEA. Although we are learning more about how microbial diversity and function (including total EEA) will be affected by environmental changes, little is known about what factors control the importance of the abundant cell-free enzymes. Since cell-attached EEAs are linked to the cell, their fate will likely be linked to the factors controlling the cell's fate. In contrast, cell-free enzymes belong to a kind of "living dead" realm because they are not attached to a living cell but still are able to perform their function away from the cell; and as such, the factors controlling their activity and fate might differ from those affecting cell-attached enzymes. This article aims to place cell-free EEA into the wider context of hydrolysis of organic matter, deal with recent studies assessing what controls the production, activity and lifetime of cell-free EEA, and what their fate might be in response to environmental stressors. This perspective article advocates the need to go "beyond the living things," studying the response of cells/organisms to different stressors, but also to study cell-free enzymes, in order to fully constrain the future and evolution of marine biogeochemical cycles.

Keywords: marine biogeochemical cycling, carbon cycle, organic matter hydrolysis, extracellular enzymatic activity, cell-free enzymes, climate change, warming

### IMPORTANCE OF MICROBES AND THEIR EXTRACELLULAR ENZYMATIC ACTIVITIES (EEA)

The marine environment plays a critical role in global biogeochemical cycles (Daily, 2003; Harley et al., 2006; Hutchins and Fu, 2017). Microbes are the engines driving Earth's biogeochemical cycles (Falkowski et al., 2008). These tiny organisms have the set of core genes coding for the enzymes of the major reactions responsible for transforming energy and matter into (usable) substrates essential for life (Falkowski et al., 2008). We live in a time of change, and anthropogenic impacts can alter the structure and functioning of marine microbial communities, and consequentially, the role of the ocean in the global biogeochemical cycles. Thus, if we aim to understand what the future of marine biogeochemical cycling is going to be, we need to understand what the fate of microbes, and their enzymes, will be.

When it comes to the consumption of organic matter for transformation and recycling, microbes seem to have a preference for specific types of organic matter. According to the

#### Edited by:

Maria Montserrat Sala, Institute of Marine Sciences (CSIC), Spain

#### Reviewed by:

Sachia Jo Traving, University of Copenhagen, Denmark Carol Arnosti, University of North Carolina at Chapel Hill, United States

#### \*Correspondence:

Federico Baltar federico.baltar@otago.ac.nz

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 26 September 2017 Accepted: 24 November 2017 Published: 04 January 2018

#### Citation:

Baltar F (2018) Watch Out for the "Living Dead": Cell-Free Enzymes and Their Fate. Front. Microbiol. 8:2438. doi: 10.3389/fmicb.2017.02438

**9**

fmicb-08-02438 December 28, 2017 Time: 15:56 # 2

"size-reactivity" model, heterotrophic microbes preferentially degrade high molecular weight dissolved organic matter (DOM) because it tends to be more bioavailable than the low molecular weight DOM (Benner and Amon, 2015). But, that food selectivity comes at a price: heterotrophic prokaryotes will need to hydrolyze most of those molecules into subunits small enough to be incorporated, because most molecules need to be smaller than 600 Da to pass through the prokaryotic cell wall (Weiss et al., 1991). For that purpose they use extracellular enzymes; so due to the central role of those enzymes they are referred to as the 'gatekeepers' of the C cycle (Arnosti, 2011). However, an alternative polysaccharide uptake mechanism of bacteria was recently revealed, which allows them to directly incorporate large molecular weight DOM compounds (Cuskin et al., 2015; Reintjes et al., 2017). Yet, not all high molecular weight DOM can be transported by this mechanism (i.e., biochemical and microbiological studies suggest that the binding/hydrolysis/transport is very selective for specific polysaccharides), and that mechanism still involve extracellular hydrolysis – but the binding proteins hang onto the pieces, such that they are transported into the cell with no loss to the external environment (Cuskin et al., 2015). Nevertheless, extracellular enzymatic activities (EEAs) are found from epipelagic to bathypelagic waters, commonly observing an increasing ratio of EEA to cell abundance with depth (Hoppe and Ullrich, 1999; Hoppe et al., 2002; Baltar et al., 2009).

#### THE LIVING AND THE 'LIVING DEAD': CELL-ATTACHED VERSUS CELL-FREE EEA

Extracellular enzymes exist in two forms; cell-bound (i.e., cell-attached), and dissolved (i.e., cell-free, operationally defined as passing through a 0.22 µm filter). The total EEA is the result of the combination of cell-associated and cell-free enzymes (**Figure 1**). Using the Internet as an analogy, we could say that cell-bound EEA would be the equivalent of a "wired" Internet connection, whereas cell-free EEA would be a "wireless" connection (i.e., would still provide the end product – data transfer/hydrolysate – even if not physically connected to the hardware/cell). Different sources of cell-free enzymes have been suggested, including the active release by bacteria in response to an appropriate substrate (Alderkamp et al., 2007) and bacterial starvation (Albertson et al., 1990), and changes in cell permeability (Chrost, 1990). These studies indicate that marine bacteria can release enzymes into the environment not only to facilitate the hydrolysis of specific substrates during exponential growth on specific substrates, but also that starved cells during stationary phase showed a greater release of extracellular enzymes than at the onset of starvation. In fact, even during this starvation period, de novo protein synthesis occurs for the production and/or release of the cell-free enzymes into the surrounding environment, highlighting the importance of this process for some marine bacteria (Albertson et al., 1990). Besides direct release, there are other indirect sources of cell-free enzymes, including grazing on bacterial communities (Bochdansky et al., 1995), and viral lysis (Karner and Rassoulzadegan, 1995).

The question then is whether the strategy of releasing enzymes into the aquatic environment is energetically and ecologically efficient. Extracellular enzymes released by bacteria have been shown to produce sufficient products from particulate organic carbon to support the growth of the bacteria in the absence of any other significant source of organic carbon and without direct contact between the cell and particulate substrate (Vetter and Deming, 1999). These empirical results obtained by Vetter and Deming (1999) support the predictions they formulated in a model (Vetter et al., 1998), suggesting that dissolved extracellular enzymes are advantageous when bacteria are attached to particles and when the substrate is within a well-defined distance from the enzyme source. Furthermore, other model simulations later revealed that when enzymeproducing microbes competed with non-producers: (1) nonproducers were favored by higher enzyme costs, (2) producers were favored by lower rates of enzyme diffusion, and (3) non-producers and producers coexisted in highly organized spatial patterns at intermediate enzyme costs and diffusion rates (Allison, 2005). These studies are also in agreement with the recently proposed solutions to the 'public good dilemma' in bacterial biofilms (Drescher et al., 2014). In that study, the secretion of public goods (i.e., chitinases) that potentially could be exploited by non-producing cells was explained by two mechanisms: cells can produce thick biofilms that confine the goods to producers, or fluid flow can remove soluble products of chitin digestion, denying access to non-producers. More recently, another model suggested that not only might cell-free enzymes be a profitable strategy to microbes live near or attached to particles, but went further to indicate that cell-free enzymes might be a viable strategy even for free-living cells, if substrate utilization is viewed as a cooperative effort (Traving et al., 2015). These authors also suggested that even small amounts of long-lived cell-free enzymes released by cell-free microbes could contribute significantly to the dissolved enzyme activity pool.

Once these enzymes are free, they can retain activity (i.e., are able to carry out their function) until they reach a specific substrate and then hydrolyze it. There are not many studies looking at the lifetime of these cell-free enzymes in seawater, but the ones available suggest that they can have half-life times of up to 20 days (Ziervogel et al., 2010; Steen and Arnosti, 2011; Baltar et al., 2013), and that deep water enzymes seem to have longer lifetimes than surface ones (Baltar et al., 2013). These observations are especially interesting because although it was originally thought that only cell-bound activity would be relevant in the marine environment (Rego et al., 1985; Chrost and Rai, 1993), evidence have been accumulating clearly indicating that the fraction of dissolved EEA is comparable to the bound fraction (Karner and Rassoulzadegan, 1995; Li et al., 1998; Baltar et al., 2010, 2013, 2016; Duhamel et al., 2010; Allison et al., 2012). In fact, in many cases, it has been observed that the proportion of dissolved EEA can reach a value of 100%, meaning that all the enzymatic activity, in those waters at the time of sampling, was being performed

by cell-free enzymes (e.g., Karner and Rassoulzadegan, 1995; Baltar et al., 2010, 2016). Thus, a disconnect between rates of EEA and community composition is not surprising (D'ambrosio et al., 2014). When cell-free enzymes are responsible for a high proportion of the total EEA, a disconnection is produced between the microbes and the enzymatic activities; that is, a decoupling of in situ hydrolysis rates from actual microbial dynamics. Thus, a high proportion of dissolved EEA could indicate a greater importance of the history of the water mass than of the actual processes occurring at the time of sampling (Karner and Rassoulzadegan, 1995; Baltar et al., 2010, 2016; Arnosti, 2011).

would be more closely related to an "alive" activity, whereas cell-free EEA would be in the "living dead" realm.

Potential differences in temporal scales of activity for cell-attached and cell-free enzymes raises the question about controls on these pools of enzymes. The cell-bound EEA will be closely linked to the living cell's behavior (i.e., be affected by the same factors controlling the growth, activity and diversity of living cells). In contrast, the cell-free enzymes will not necessarily be affected by the same factors influencing the cells, but will remain active for a prolonged period of time, and probably be affected by different factors (or in a different way in response to the same factors) than the cell-bound EEA. Bearing this in mind, using another analogy, we could say that the cell-bound EEA is an activity of the "living," since that activity is performed attached to a cell which is alive, and thereby that activity will change in response to the cell's needs and in response to the cells/community dynamics (**Figure 1**). Whereas the cell-free EEA is closer to an activity of the "living dead," in the sense that it is not "alive" anymore because it is separated from the living cell, but still remains in some way "alive" in the sense that it can still perform its function when it encounters the right substrate.

However, it is important to realize that particle encounter is not necessarily the end of the line for an enzyme, and a large fraction of the cell-free enzymes may be trapped by particles including colloids or liposomes. In fact, the lifetime of cell-free enzymes can be extended if they are associated with particles (Gianfreda and Scarfi, 1991; Ziervogel et al., 2007), since surface associations can offer an improved resistance to physicochemical degradation (Lähdesmäki and Piispanen, 1992), and protection from remineralization (Lozzi et al., 2001). Evidence of bacterial cell-free enzymes embedded in an exopolymeric matrix have been reported (Decho, 1990), where cell-free EE attached to this matrix can form a complex similar to the enzyme–humic complexes in soils (Chrost, 1990). Also, some enzymes might be associated with particles due to trapping of digestive enzymes within partially degraded bacterial membranes which act as micelles (liposomes) (Nagata and Kirchman, 1992).

#### FATE OF THE 'LIVING DEAD' CELL-FREE EEA

The activities of extracellular enzymes control the rate at which organic matter is processed in the ocean. Given the evidence of

fmicb-08-02438 December 28, 2017 Time: 15:56 # 3

fmicb-08-02438 December 28, 2017 Time: 15:56 # 4

high activities of cell-free enzymes, understanding the controls on the lifetimes and activities of these cell-free enzymes is essential.

There are very few studies on this topic but they are already starting to reveal some of the key factors controlling the relative importance of cell-free EEA, and temperature seems to be a major one. In a Baltic Sea seasonal (1.5 y-long) study, a significant inverse relation was found between the proportion of dissolved relative to total EEA (Baltar et al., 2016). In a lab experiment, incubating microbial communities from the Great Barrier Reef waters (Australia) at three different temperatures (i.e., in situ, +3 and −3 ◦C), a significant inverse relation was found again between temperature and the relative proportion of dissolved to total EEA (Baltar et al., 2017). These results are consistent with the increased observed in the proportion of dissolved relative to total EEA with depth as the temperature drops along the whole water column in a (sub) tropical Atlantic Ocean transect (Baltar et al., 2010). These studies signpost, from different angles (i.e., seasonal, climate change lab, and transect cruise studies), temperature as the main factor affecting the relative importance and activity of cell-free EEA. All those studies seem to indicate that the warmer the temperature the lower the proportion of dissolved EEA. This is also consistent with the longer lifetime of cell-free enzymes found for deep relative to the surface waters (Baltar et al., 2013).

Another factor that seems to be significantly contributing to controlling the lifetimes of cell-free enzymes is ultraviolet radiation (UVR). Very few studies are available on this topic too. The effect of UVR on cell-free enzymes directly was tested in Arctic seawater, finding that although natural illumination did not produce significant effects of photodegradation, a reduction in cell-free enzyme activity was found at artificially high UVR doses (i.e., UV-B intensity 5–10 times higher than in situ) (Steen and Arnosti, 2011). Interestingly, these authors found a significant effect of UVR on leucine aminopeptidase and alkaline phosphatase but not on beta-glucosidase at any treatment level. A recent study with cell-free enzymes from New Zealand waters revealed that environmentally relevant UVR irradiances reduced cell-free enzyme activities up to 87% in 36 h when compared to dark controls, likely a consequence of photodegradation (Thomson et al., 2017). This study also revealed that the magnitude of the effect of UVR on cell-free enzymes varied depending on the UVR fraction. Interestingly, consistent with the findings from Steen and Arnosti (2011), the effect of UVR differed depending on the enzyme; significantly decreasing the activity of cell-free leucine aminopeptidase and alkaline phosphatase, but not affecting β-glucosidase. This indicates that UVR (at ambient levels of radiation) can be a key factor reducing the activity (and lifetime) of cell-free enzymes (Thomson et al., 2017). Also, the fact that UVR effects vary among different enzymes indicates that UVR might change the spectrum of the EEA and thereby the composition of the resulting organic matter pool. Moreover, this effect of UVR on the cell-free EEA might help explain why the proportion of dissolved EEA tends to be lower in surface waters, and why the proportion of cell-free EEA is higher in winter and lower in summer.

More research is needed to fully constrain the factors and the mechanisms controlling the activity and lifetime of cell-free EEA as other factors, like pH, might be relevant. Moreover, the activities measured using externally-added substrates such as Leucine-MCA (7-amido-4-methylcoumarin) or MUF (4-Methylumbelliferyl)-β-glucose as substrates, reflect the activity of an unknown number/type of different enzymes that cleave the same substrate (Steen et al., 2015). Furthermore, enzymes of different primary and tertiary structure may hydrolyze the same substrate, but these enzymes would likely be susceptible to different degrees to UV radiation, heat inactivation, etc. Nevertheless, based on the evidence available thus far, and bearing in mind the projected warming ocean environment and the variable UVR light regime, it seems like there could be major changes in the activity of cell-free EEA and their contribution to organic matter remineralization in the near future.

#### CONCLUSION AND FUTURE OUTLOOK

It is clear now that, in any given marine location at any given time, cell-free EEA can be at least as important as cell-attached EEA. This has some strong implications on how we look and interpret many of the microbial parameters we measure since it can imply a decoupling between the community composition/function and the actual hydrolysis rates we measure. For instance, this complicates the study of functional redundancy in the marine environment if based on these extracellular enzymes, since changes in the microbial community composition might happen at a different temporal scale to the changes in EEA. Thus, it is important to consider the need to determine the cell-free fraction of EEAs (and not only the total fraction), if our aim is to link EEA to other microbial parameters.

In the near future, the advance in technology (e.g., gene- and protein-based techniques as well as organic matter characterization tools, etc.) will allow for a deeper understanding of the function and fate of cell-free EEA. For example, abundant periplasmic proteins were recently discovered using metaproteomics on seawater concentrates of the cell-free fraction (Xie et al., 2017). Based on that, these authors suggested that free proteins released from microbes could be important to ecosystem function (Xie et al., 2017). They admitted not having a clear explanation for the high presence of periplasmic proteins in the "non-bacterial" world. But, having in mind what has been discussed in this article, it is possible that many of those so-called "non-bacterial world" proteins by Xie et al. (2017) could indeed be cell-free (here described as "living dead") extracellular enzymes. This is an example of how the use of novel technologies can help to make new discoveries in this field. However, we should not only focus on the development of new technologies but also complement those with the use of classical rate measurements. A more refined characterization of the DOM pool (e.g., LC–MS/MS) coupled with a combined determination of hydrolysis rates and gene/protein expression might open new avenues and discoveries in this field of research. Furthermore, these kinds of experiments, performed under different anthropogenic stressors, might also help elucidate how the role of these enzymes might change in response to different climatic scenarios. These studies will probably confirm that the 'engines' of the marine biogeochemical cycles are not only the microbial cells but that there are other processes taking place away from cells which are also an important part of that engine.

To conclude, the findings discussed in this article advocate for the need to go beyond the "living things," – and study not only how the living cells/organisms will respond to anthropogenic perturbations, but also how the "living dead" cell-free active molecules will, if we really aim to fully constrain the future of marine biogeochemical cycles.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

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

#### ACKNOWLEDGMENTS

FB was supported by a Rutherford Discovery Fellowship (Royal Society of New Zealand). The author would like to acknowledge the support and insightful comments of the reviewers, which clearly helped improve the overall merit of the manuscript.


Frontiers in Microbiology | www.frontiersin.org

fmicb-08-02438 December 28, 2017 Time: 15:56 # 5


extracellular enzymes. Microb. Ecol. 36, 75–92. doi: 10.1007/s00248990 0095


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

Copyright © 2018 Baltar. 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.

fmicb-08-02438 December 28, 2017 Time: 15:56 # 6

# Heterotrophic Extracellular Enzymatic Activities in the Atlantic Ocean Follow Patterns Across Spatial and Depth Regimes

#### Adrienne Hoarfrost\* and Carol Arnosti

Department of Marine Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States

Heterotrophic microbial communities use extracellular enzymes to initialize degradation of high molecular weight organic matter in the ocean. The potential of microbial communities to access organic matter, and the resultant rates of hydrolysis, affect the efficiency of the biological pump as well as the rate and location of organic carbon cycling in surface and deep waters. In order to investigate spatial- and depth-related patterns in microbial enzymatic capacities in the ocean, we measured hydrolysis rates of six high-molecular-weight polysaccharides and two low-molecular-weight substrate proxies at sites spanning 38◦S to 10◦N in the Atlantic Ocean, and at six depths ranging from surface to bottom water. In surface to upper mesopelagic waters, the spectrum of substrates hydrolyzed followed distinct patterns, with hydrolytic assemblages more similar vertically within a single station than at similar depths across multiple stations. Additionally, the proportion of total hydrolysis occurring above the pycnocline, and the spectrum of substrates hydrolyzed in mesopelagic and deep waters, was positively related to the strength of stratification at a site, while other physichochemical parameters were generally poor predictors of the measured hydrolysis rates. Spatial as well as depth-driven constraints on heterotrophic hydrolytic capacities result in broad variations in potential carbon-degrading activity in the ocean. The spectrum of enzymatic capabilities and rates of hydrolysis in the ocean, and the proportion of organic carbon hydrolyzed above the permanent thermocline, may influence the efficiency of the biological pump and net carbon export across distinct latitudinal and depth regions.

Keywords: carbon cycling, extracellular enzymes, heterotrophy, functional biogeography, deep ocean, microbial activity, biogeochemistry

#### INTRODUCTION

Microbial communities are major drivers of organic carbon cycling in the ocean. The carbon cycling capacities of these communities ultimately affect the inventories of oxygen and CO<sup>2</sup> in the atmosphere, the magnitude and composition of organic carbon export from the surface to the deep ocean, and resource availability to higher trophic levels (Azam and Malfatti, 2007; Jiao et al., 2010). Although 99.9% of autochthonous organic carbon is remineralized before it reaches sediments, a large standing pool of dissolved organic carbon (DOC) persists in the water column (Hedges, 1992),

#### Edited by:

Maria Montserrat Sala, Consejo Superior de Investigaciones Científicas(CSIC),Spain

#### Reviewed by:

Hila Elifantz, Bar-Ilan University, Israel Zhanfei Liu, University of Texas at Austin, United States

\*Correspondence: Adrienne Hoarfrost adrienne.l.hoarfrost@unc.edu

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Marine Science

Received: 12 April 2017 Accepted: 12 June 2017 Published: 23 June 2017

#### Citation:

Hoarfrost A and Arnosti C (2017) Heterotrophic Extracellular Enzymatic Activities in the Atlantic Ocean Follow Patterns Across Spatial and Depth Regimes. Front. Mar. Sci. 4:200. doi: 10.3389/fmars.2017.00200 demonstrating that some fraction of marine organic carbon is not readily amenable to microbial degradation. The microbial enzymatic capacities to access organic carbon is a potentially important driver shaping the ocean carbon reservoir, but the factors that determine whether and how much organic matter is remineralized are poorly understood (Arnosti, 2011). Biogeographical patterns in microbial communities and their enzymatic capacities (Arnosti et al., 2012; Gomez-Pereira et al., 2012; Sunagawa et al., 2015), and their net effect on organic carbon transformations in the ocean, may in turn be shaped by a poorly constrained set of factors (Hanson et al., 2012).

The enzymatic capacity of a microbial community is a critical determinant of the breadth of organic compounds which may be recycled. Most organic carbon is biosynthesized as high molecular weight compounds, which are typically hydrolyzed by both endo-acting (mid-chain cleaving) and exo-acting enzymes (Warren, 1996). In order for heterotrophs to access natural organic matter, they must produce the appropriate enzymes to hydrolyze a particular substrate into sizes small enough to be transported into the cell. These enzymes have highly targeted structural specificities and are very diverse (Aspeborg et al., 2012; Teeling et al., 2012), reflecting both the complexity of natural organic matter and of the microbial communities that access it.

Differences in microbial enzymatic capacities thus may result in functional biogeographical patterns in carbon export and recycling. Field studies measuring activities of extracellular enzymes that degrade organic carbon have demonstrated that the types of substrates hydrolyzed and their rates of hydrolysis vary along latitudinal gradients (Arnosti et al., 2011, 2012), a pattern that parallels biogeographical patterns in microbial community composition (Fuhrman et al., 2008). Beyond community composition, the genetic capacity to hydrolyze individual substrates may also follow biogeographical patterns. Although, most of the enzymes involved in the extracellular breakdown of organic carbon have not yet been annotated, a targeted study of the biogeography of enzymes in glycosyl hydrolase family 5 revealed wide diversity and variation in relative abundance across the North Atlantic (Elifantz et al., 2008). Genetic distributions of polysaccharide-degrading enzymes in common heterotrophic marine microbial clades also vary considerably across oceanic provinces (Gomez-Pereira et al., 2012), as do the activities of polysaccharide hydrolyzing enzymes (Arnosti et al., 2012).

Organic carbon cycling capabilities are thus heterogeneously distributed across both microbial phylogenies (Zimmerman et al., 2013) and the natural environment. In order to examine latitudinal and depth-related patterns in hydrolytic capabilities of heterotrophic microbial communities, we measured extracellular enzyme activities across a broad range of latitude and depth, identified geospatial patterns in those activities, and investigated potential environmental factors affecting heterotrophic enzymatic activity. The hydrolysis of six high-molecularweight and two low-molecular-weight organic substrates was measured across 48 degrees of latitude and from surface to bottom waters in the South and Equatorial Atlantic Ocean. These data enable us to explore the connectivity of hydrolytic capacities between stations and depths, and the relationship of hydrolysis rates to stratification and physicochemical parameters. These factors may shape biogeographical patterns in heterotrophy, and ultimately affect the location and magnitude of organic carbon remineralization, and thus carbon sequestration, by the biological pump.

### METHODS

#### Seawater Sampling

Seawater was collected via Niskin rosette equipped with a conductivity-temperature-depth sensor (CTD) at nine stations spanning 38◦ S to 10◦N in the subtropical to equatorial Atlantic Ocean (**Figure 1**). Hydrolysis rates of polysaccharides and of monomeric substrates were measured at 6 stations, ranging from 38◦ S to 3.5◦N; only activities of monomeric substrates could be measured at the northernmost three stations due to lack of time for extended substrate incubations.

At each station (stations 2, 4, 7, 10, 15, 18, 21, 22, and 23; part of the DeepDOM cruise, Kujawinski, 2013), seawater was collected from six depths: surface (SuW, 5 m), deep chlorophyll maximum (DCM, ∼50–100 m), mesopelagic (meso, 250 m), Antarctic Intermediate Water (AAIW, ∼750–850 m), North Atlantic Deep Water (NADW, 2,500 m), and bottom water (bot, ∼3,700–4,600 m). Specific depths of DCM, AAIW, and bottom water were chosen according to a maximum in fluorescence (DCM), a minimum in salinity and peak in oxygen (AAIW), and a few meters above bottom (bot), respectively (Supplementary Figure 1), and thus varied by station.

At each depth, 1 L glass Duran bottles were rinsed three times with seawater from the corresponding depth, then filled without using tubing from a single Niskin bottle. 200 mL of seawater from each depth was autoclaved in a separate glass Duran bottle for use in killed control incubations.

#### Physical and Chemical Parameters

Temperature, salinity, oxygen, and fluorescence were measured via CTD (Supplementary Figure 1), and used to calculate potential temperature (θ), potential density (σθ), and buoyancy frequency (N<sup>2</sup> ) (Supplementary Figure 2). CTD data for every cast throughout the cruise, and nutrient data collected from discrete depths and analyzed by K. Longnecker, are available through the BCO-DMO database (Supplementary Figure 1, Kujawinski, 2013). CTD and nutrient data used for analysis in this study are provided through the associated BCO-DMO repository (Hoarfrost and Arnosti, 2016), and can be reproduced using scripts provided at the associated Github repository (Hoarfrost, 2016).

### α-Glucosidase and Leucine Aminopeptidase Activities

Two substrate proxies, α- glucose linked to 4 methylumbelliferone (α-Glu; Chem Impex 21676) and leucine linked to 4-methylcoumarinyl-7-amide (L-MCA; Sigma 62480- 44-8), were used to measure the activities of α-glucosidase and leucine aminopeptidase, respectively, after the method of Hoppe (1983). The enzymes hydrolyzing these substrates act on the α-1→4-linked terminal glucose of oligo- and polysaccharides, and N-terminal leucine residues of peptides or proteins, respectively. Recent work has demonstrated that L-MCA can also be hydrolyzed by enzymes other than leucine aminopeptidase (Steen et al., 2015), but this widely-used method still provides a measure of peptidase activity in the environment. For each substrate, triplicate aliquots of 4mL of live seawater and one autoclaved seawater killed control were incubated in plastic cuvettes at as close to in situ temperature as possible. Available incubation temperatures were 3, 12, 15, 18, 25, and 28◦C. Two cuvettes with 4 mL of live or autoclaved seawater and no added substrate served as live and killed blank incubations, respectively.

Saturating concentrations were determined at each station via a saturation curve conducted over 24 h using surface water, testing increasing concentrations of substrate. The saturating concentration was identified as the concentration of substrate at which addition of higher concentrations of substrate does not induce higher rates of activity. Since enzymatic activity is typically highest in surface or near-surface waters (e.g., Baltar et al., 2009; Steen et al., 2012), and leucine aminopeptidase activity is typically higher than α-glucosidase activity (Baltar et al., 2010, 2013), saturation concentrations determined for leucine-MCA in surface waters were used for all depths and substrates at each station. Substrates were added at saturating concentrations 100 µM at stations 2, 4, and 7; 75 µM at stations 10, 15, 18, 22, and 23; and 50 µM at station 21.

Incubations were sampled at four timepoints, and fluorescence was measured in a Turner Biosystems spectrophotometer (TBS-380). Later timepoints were chosen based on the rate of activity at earlier timepoints. A typical timecourse for a rapidly-hydrolyzed substrate was 6, 12, and 24 h; for a low- to no-hydrolysis substrate, 24, 48, and 72 h. No incubation was sampled later than 72 h. All incubations were sampled at 24 h to provide a common timepoint reference.

Rates reported here are maximum hydrolysis rates, typically at T3 for α-glucosidase and at T1 for leucine aminopeptidase. T3 for α-glucosidase was typically sampled at 36–48 h in shallow, more active waters, or 60–72 h in deeper, less active waters. T1 for leucine aminopeptidase was typically sampled at 4–6 h in shallow waters, or 24 h in deeper waters. α-glucosidase activities sampled at later timepoints may include a growth response, whereas the shorter timecourse of leucine aminopeptidase incubations likely does not include a growth response. In all cases, rates represent potential hydrolysis rates, since added substrate competes with naturally-occurring substrate for enzyme active sites.

#### Polysaccharide Hydrolysis Measurements

Activities of extracellular enzymes that hydrolyze six different fluorescently labeled polysaccharides were measured at all six depths between 38◦ S and 3.5◦N. These substrates arabinogalactan, chondroitin sulfate, fucoidan, laminarin, pullulan, and xylan—were chosen for their diverse monosaccharide compositions and macromolecular structures. All of these polysaccharides are found in marine environments, and/or enzymes and genes corresponding to their hydrolysis have been identified in marine prokaryotes (e.g., Alderkamp et al., 2007; Wegner et al., 2013; Xing et al., 2015). Furthermore, the activities of enzymes hydrolyzing these substrates have been detected in a wide variety of marine environments (e.g., Arnosti, 2008; Arnosti et al., 2009, 2011).

Fluorescently-labeled polysaccharides were prepared after the method of Arnosti (1996, 2003). Each polysaccharide was incubated in triplicate live incubations in 17 mL sterilized glass vials, and one killed control incubation using autoclaved seawater. In addition, incubations without substrate with live seawater and autoclaved seawater were used as live and killed blank controls, respectively. Seawater was sterilized for 20 min in an autoclave upon recovery and incubations initiated after autoclaved water had cooled in an ice bath. Substrate was added at concentrations sufficient to detect fluorescence of the substrate, at 3.5 µM monomer-equivalent concentrations in all cases, except for fucoidan, which was added at a concentration of 5 µM due to its lower fluorescence intensity. Substrate addition is kept to the lowest concentration that is technically feasible in order to minimize growth responses due to added substrate. All samples were incubated at as close to in situ temperature as possible.

Each incubation was sampled at four timepoints: 0, 5, 12, and 21 days. Kill and live blank controls were sampled at T<sup>0</sup> and Tfinal only. Due to time constraints, at Station 15 the final timepoint was taken at day 20 instead of 21, and at Station 18 only three timepoints were collected, at 0, 5, and 12 days. At each timepoint, ∼1.8 mL was withdrawn from each incubation, filtered through a 0.2 µm pore-size syringe filter and stored at −20◦C until analysis.

Enzyme activity was measured by tracking hydrolysis of the high-molecular-weight substrate into lower-molecularweight hydrolysis products over time, as determined using gel permeation chromatography with fluorescence detection (Arnosti, 1996, 2003). Hydrolysis rates were calculated from these shifts in molecular size distribution over time from the size-separated chromatograms using the scripts hosted at the associated Github repository (Hoarfrost, 2016). Chromatograms were manually curated after processing to verify chromatographic changes and to identify incubations with zero activity or non-hydrolytic fluorescence of the free fluorophore label, which can produce artificially high hydrolysis rates. Those incubation sets were tagged and their calculated rates adjusted by setting the activity to zero, or recalculating the rate while neglecting the free fluorophore portion of the chromatogram, respectively.

#### Statistical Analyses

#### Polysaccharide Hydrolytic Diversity Using Shannon Diversity Indices

Shannon indices, which reflect both the number of substrates hydrolyzed as well as the evenness of hydrolysis rates, were used to calculate hydrolytic diversity at all sites (Steen et al., 2010). The Shannon index is expressed as <sup>H</sup> = −P<sup>n</sup> i=1 pi ln(pi), where n

is the total number of substrates and p<sup>i</sup> is the hydrolysis rate of the ith substrate normalized to the summed hydrolysis rate of all substrates at that site. H is equal to zero when only one substrate is hydrolyzed, and is maximal at 1.79 when all six substrates are hydrolyzed at equal rates.

#### Hydrolytic Compositional Dissimilarity among Sampling Sites Using Bray-Curtis Dissimilarity

The Bray-Curtis Dissimilarity, BC, is used to describe the compositional dissimilarity between two sites (Bray and Curtis, 1957). As applied here, "composition" is defined as the hydrolytic composition, or the assemblage of substrates hydrolyzed and their relative rates of hydrolysis. BC is a unitless index between 0 and 1, with a minimum of 0 when the two sites have exactly the same composition (e.g., all the same substrates are hydrolyzed at the same rate), and a maximum of 1 when none of the same substrates are hydrolyzed at the two sites. The pairwise BC dissimilarity matrix was calculated between every site with every other site.

The Bray-Curtis Dissimilarity between two sites i and j is calculated as BCij = 1 − 2Cij S<sup>i</sup> + S<sup>j</sup> , where Cij is the sum of the lesser hydrolysis rates for only those substrates that were hydrolyzed at both sites i and j, and S<sup>i</sup> and S<sup>j</sup> are the total hydrolysis rates at site i and site j respectively.

#### Multiple Regression Analysis of Environmental Parameters vs. Hydrolytic Activity

Multivariate linear regression models between polysaccharide hydrolysis rates and up to ten environmental parameters in situ potential temperature, incubation temperature, salinity, oxygen, chlorophyll a, buoyancy frequency, phosphate, total nitrogen, DOC, and silicate—were generated. By testing several permutations of models considering different combinations of environmental parameters, the best fit multiple regression model was selected by manually maximizing correlation coefficient values (Supplementary Table 1).

#### Reproducibility

The scripts to process the GPC chromatograms and calculate rates, manipulate physicochemical data, perform statistical analyses, and generate the figures in this paper were all written in the R programming language (R Core Team, 2014), and can be reproduced using the scripts hosted at the associated Github repository (Hoarfrost, 2016). The raw data is hosted on BCO-DMO (Hoarfrost and Arnosti, 2016), and instructions to download raw data and run scripts can also be found in the Readme for the Github repository.

### RESULTS

#### Physical Context and Water Masses

The transect covered a broad range of latitude and physicochemical conditions, as well as several distinct water masses (Supplementary Figures 1, 3). Antarctic Intermediate Water (AAIW) flowing south to north was detectable as a minimum in salinity at ca. 750–850 m depth throughout the transect. North Atlantic Deep Water (NADW) flowing north to south was identified as a maximum in oxygen between ca. 1500–4,000 m water depth. A large circulation-driven oxygen minimum zone encompassed stations 10–23. At the southernmost station sampled, station 2, the influence of Antarctic circulation was still apparent, with circumpolar deep waters bounding NADW above and below, and Antarctic Bottom Water (AABW, detectable as a temperature minimum) in the bottom water sample (Supplementary Figures 1, 3). At the northernmost station (station 23), the Amazon River plume was sampled in surface waters, detectable by a sharp halocline. The strength of the pycnocline generally increased from south to north, such that the southernmost stations were less stratified than the northernmost stations (Supplementary Figures 2, 4). A south to north gradient in chlorophyll a concentrations was also evident at the deep chlorophyll maximum, which increased from ca. 0.242 mg m−<sup>3</sup> at station 2 to over 1 mg m−<sup>3</sup> at station 15 (Kujawinski, 2013). DOC concentrations ranged from ca. 70–82 µM in surface, ca. 62–77 µM in DCM, and ca. 47–58 µM in mesopelagic depths, and did not directly track chlorophyll a concentrations (Kujawinski, 2013).

### Polysaccharide Hydrolysis Rates and Patterns

Polysaccharide hydrolysis rates and patterns varied across depths as well as stations, as evident for individual substrates (**Figure 2**), by the summed hydrolysis rates (Supplementary Figure 5), and by the diversity of substrates hydrolyzed at a given depth (**Figure 3**). Some polysaccharides—such as laminarin—were hydrolyzed at nearly every station and depth, whereas fucoidan was not measurably hydrolyzed at any site, and arabinogalactan was hydrolyzed only in surface waters of station 15. Chondroitin, pullulan, and xylan were hydrolyzed only at particular stations and depths: chondroitin was the only substrate other than laminarin hydrolyzed below 250 m, but at some stations it was not hydrolyzed at any depths. Pullulan was hydrolyzed only

above the pycnocline, while xylan was also hydrolyzed in shallow waters but only at some stations (**Figure 2**).

Hydrolysis rates (**Figure 2**) and summed hydrolysis rates (Supplementary Figure 5) decreased with depth. This decrease was more abrupt and occurred at shallower depths at more stratified stations, with most hydrolytic activity occurring in the surface and DCM depths. At less stratified stations (where maximum buoyancy frequency in the water column was lower), hydrolytic activities decreased more gradually with depth (**Figure 2**), and a greater proportion of summed activity occurred below the pycnocline (**Figure 4**, R <sup>2</sup> = 0.66, P = 0.048).

The transect covered a gradient in water column productivity (as represented by chlorophyll a fluorescence) and in water column stratification. At the more northerly stations where stratification was stronger and chlorophyll a concentrations were higher, the highest hydrolytic diversity and rates of enzymatic activity were measured. Additionally, the depth at which the highest hydrolysis rate was observed at a particular station was at shallower depths at northerly, more stratified stations than at southerly, less stratified stations (**Figure 4**, **Figure 2**, Supplementary Figure 5).

Hydrolytic diversity also decreased with depth from shallow to deeper waters (**Figure 3**), and sites with higher overall rates of activity also had higher hydrolytic diversity (Supplementary Figure 6). Maximum hydrolytic diversity was typically measured at the surface or DCM, although station 2 exhibited highest hydrolytic diversity at mesopelagic depths, probably because the same assemblage of substrates was hydrolyzed at surface, DCM, and mesopelagic depths at station 2, but with different degrees of evenness.

station is positively correlated with the maximum buoyancy frequency at that station. R <sup>2</sup> = 0.66, P = 0.048.

In the upper 250 m of the water column, the assemblage of polysaccharide substrates hydrolyzed at a given station followed distinct patterns. Comparing Bray-Curtis dissimilarities among surface, DCM, and mesopelagic depths for each station, hydrolytic assemblages clustered strongly when grouped by station (**Figure 5A**, PERMANOVA P = 0.007). This result contrasted with grouping by depth sampled, which did not produce any distinguishable effect on Bray-Curtis distances between assemblages (**Figure 5B**, PERMANOVA P = 0.399). A similar analysis could not be done for the full depth range due to the lack of any detectable hydrolytic activity at many of the deeper depths.

#### Monomeric Substrate Hydrolysis Rates

Hydrolysis rates of monomeric substrates also varied by station, with maximal activity at the surface or DCM, and decreasing activity with depth (**Figure 6**), with the exception of a single replicate for α-glucose at station 7 where high activity was observed at mesopelagic depths. Depth-related decreases in αglucosidase and leucine aminopeptidase activities, unlike the polysaccharide hydrolase activities, did not correspond to the degree of water column stratification. Below 250 m, α-glucosidase activity was undetectable at all sites even after 72 h of incubation, whereas leucine activities were very low in deep water, but nonzero.

#### Relationship between Polysaccharide Hydrolysis and Environmental Parameters

The strength of the relationship between polysaccharide hydrolysis rates and up to 10 environmental parameters was investigated by fitting the multiple regression model that maximized R 2 -values (Supplementary Table 1). Overall, environmental parameters poorly explained the observed variation in hydrolysis rates (R <sup>2</sup> = 0.22). Since many of these parameters co-correlate with each other, one or two of these parameters generally explained as much or nearly as much of the variation in hydrolysis rates as all ten environmental variables. Temperature and chlorophyll a accounted for most of the relationship in the overall model (R <sup>2</sup> = 0.19), while the inclusion of the additional eight environmental variables only slightly improved the model (R <sup>2</sup> = 0.22). This result is mainly due to the difference in temperature and chlorophyll a in shallow vs. deep waters corresponding with higher rates of hydrolysis in shallower waters, since models using samples from just shallow or just deep water yielded very poor fits. The high frequency of zero hydrolysis rates did not appear to bias the model, however, since models using only non-zero rates yielded similar fits as the overall model (Supplementary Table 1).

When broken down by individual substrate, models were generally better fitted than the model of aggregated hydrolysis rates (Supplementary Table 1). However, the combination of environmental variables that best fit the data differed by substrate: for chondroitin, temperature only; for laminarin, temperature and chlorophyll a; for pullulan, temperature and buoyancy frequency; for xylan, chlorophyll a and salinity. Arabinogalactan and fucoidan were not modeled individually due to the lack of non-zero hydrolysis rates across all sites.

#### DISCUSSION

Microbial communities rely on extracellular enzymes to hydrolyze high molecular weight organic matter prior to uptake. The structural specificities of the enzymes active at a given site and depth determine which substrates are available for further metabolism, while relative rates of hydrolysis reflect the potential speed of substrate processing. Site- and depth-related differences in hydrolysis rates and capacities imply differential remineralization of organic matter across latitude and depth in the ocean. The overall patterns of enzyme activities observed along this transect—spatial differences in hydrolytic diversity in surface waters and a decrease in the spectrum of polysaccharides hydrolyzed with depth—are consistent with studies of surface waters from other parts of the world's oceans (e.g., Arnosti et al., 2011), and add considerably to the very few other depth profiles of polysaccharide hydrolase activities in the ocean (Steen et al., 2012; D'Ambrosio et al., 2014).

Distinct functional assemblages characterized individual stations along the latitudinal gradient, such that the diversity of substrates hydrolyzed was more similar from surface to mesopelagic depths at a single station than at similar depth levels across different stations (**Figure 5**). These spatial and depth-related patterns in hydrolytic diversity, hydrolysis rate, and functional similarity together suggest that the vertical transfer of enzymatic capabilities through the upper depths of the water column—whether through cells, cellular material, or active enzymes—influences the hydrolytic signature of a station, but that this vertical transfer may be more limited at more stratified stations.

Patterns in hydrolytic assemblages among deeper water masses remain to be investigated, since activities were low or not measurable over the timescale of incubation at many of the deeper depths. A lack of measurable polysaccharide hydrolysis at deep sites may indicate that the heterotrophic community had no capacity to detect or to hydrolyze the substrates tested, or that the 21-day incubation timescale was insufficient to measure hydrolysis. In particular, low hydrolytic activities, or activities that require the growth of potentially slow-growing and/or rare members of the microbial community might not be detectable over a 21-day time course (Arnosti, 2008), since a sufficient fraction of the total added polysaccharide pool must be hydrolyzed to detect activity. The observation that leucine aminopeptidase was hydrolyzed—albeit at low rates—in bottom waters at almost all stations, however, demonstrates that an active heterotrophic community was present at these depths.

Measurable hydrolysis of leucine-MCA and MUF-α-glucose in deep waters has also been reported at other sites in the South and Equatorial Atlantic Ocean (Baltar et al., 2009, 2010, 2013). The leucine-aminopeptidase activities of 1–4 nM h−<sup>1</sup> reported by Baltar and colleagues in deep water are considerably higher than the 0–0.35 nM h−<sup>1</sup> in the present study, although the range of leucine-aminopeptidase activities measured in surface waters are similar between this study and previous studies (Baltar et al., 2009, 2010, 2013). The range of α-glucosidase activities measured in the present study in surface water (0–20 nM h−<sup>1</sup> ) are much greater than reported in previous studies (∼0–0.25 nM h−<sup>1</sup> ), perhaps because of a growth response during the extended timecourse of our incubations (maximum of 72 h, vs. maximum of 48 h in Baltar et al., 2009, 2010, 2013). No α-glucosidase activity was detected at any depths below 250 m in this study, even after 72 h incubation, whereas previous investigations measured low but non-zero α-glucose hydrolysis rates in deep water (∼0–0.8 nM/h, Baltar et al., 2009, 2010, 2013). For both leucine aminopeptidase and α-glucosidase activities, the subtraction of killed control fluorescence from live incubation fluorescence may have contributed to the lower rates measured in deep water in our experiments.

Multiple factors may contribute to the patterns of enzyme activities we measured. Environmental parameters alone are not likely to be the principal drivers for these patterns: the environmental variables measured at these stations poorly predicted observed rates, in univariate as well as multivariate models (Supplementary Table 1). While the specific environmental variable(s) that best fit each model varied by individual substrate (Supplementary Table 1), a causal explanation for the correlation strengths between hydrolysis rates of individual substrates and specific environmental factors is not obvious in most cases.

Instead, the patterns of enzymatic activities observed along this transect may be tied to the biogeography of the underlying microbial communities (Rusch et al., 2007; Fuhrman et al., 2008; Zinger et al., 2011; Sunagawa et al., 2015) and their functional capacities (DeLong et al., 2006; Shi et al., 2011). For example, the capacity to produce three extracellular enzymes (alkaline phosphatase, chitinase, and β-N-acetyl-glucosaminidase) commonly measured in field studies varies on very fine phylogenetic scales across all annotated prokaryotic genomes (Zimmerman et al., 2013). The heterogeneous distribution of heterotrophic genetic capacities among microbial phylogenies, and a varying distribution of these capabilities among surface water and subsurface environments (DeLong et al., 2006; Elifantz et al., 2008; Shi et al., 2011; Gomez-Pereira et al., 2012) results in functional stratification and resource partitioning along depthand horizontal gradients. Differences in community composition and function are driven by a complex combination of factors that may include organic carbon composition and concentration (McCarren et al., 2010), distribution limitation (Follows et al., 2007; Hellweger et al., 2014), environmental selection (Ladau et al., 2013), or a confluence of interacting factors that defy simple categorization (Hanson et al., 2012).

Irrespective of the underlying factors, the relationship between water column stratification and the fraction of hydrolysis occurring in the shallow surface or DCM relative to deeper mesopelagic waters potentially has implications for the location of nutrient regeneration and for carbon export in the ocean. This point can be illustrated with a simple conceptual box model showing hypothetical DOC generation from particles sinking through the water column at the different stations along the transect (**Figure 7**). While there are many interacting factors that affect carbon export (De La Rocha and Passow, 2007), for the purposes of this discussion we consider only the degree of stratification and the spatial patterns of hydrolysis measured in this study. Assuming that particles are being hydrolyzed at the summed hydrolysis rates measured at our sites, DOC will be generated from particles during their passage through the water column at a rate related to the depth-integrated hydrolytic capacities at that location and the sinking rate of particles. For the purpose of this conceptual calculation, we assume a constant particle sinking rate of 100 m day−<sup>1</sup> and divide the upper water column into three boxes: DCM—centered at the DCM sampling depth for that station and arbitrarily set at

a thickness of 50 m; surface–all depths above the DCM; and upper mesopelagic—from below the DCM to 300 m. To estimate carbon remineralization, carbohydrates hydrolyzed from the particulate to the dissolved phase are then converted to DOC generated, assuming 6 C per monosaccharide produced. In this scenario, the total quantity of DOC generated in the upper 300 m of the water column, as well as the depth at which this DOC would be generated, varies greatly along the transect (**Figure 7**). At the productive and more strongly stratified station 15 (2.7◦ S), for example, most of the DOC would be generated in the surface and DCM, and labile DOC would likely quickly be respired to CO<sup>2</sup> which would remain in the surface ocean. The highest overall quantity of DOC, however, would be produced at station 7 (22.5◦ S), where more than half of the total generated DOC would be in the mesopelagic zone, and thus below the permanent thermocline. Labile DOC that is respired to CO<sup>2</sup> would likely remain below the thermocline, and would not exchange with surface waters or with the atmosphere on short timescales (Kheshgi, 2004).

The efficiency with which the biological pump removes surface-derived carbon from the upper ocean thus depends in part on the quantity of carbon remineralized from a sinking particle while it is still above the permanent thermocline (De La Rocha and Passow, 2007). This calculation in turn depends on the relative lability of the organic carbon in the sinking particles (Engel et al., 2009), the hydrolytic capacities of the microbial communities acting on them as they sink, the extent of water column stratification, and the residence time of the particle at different depths (Prairie et al., 2015). Holding all other factors constant, the biological pump would be more efficient at less stratified stations—such as stations 2, 4, and 7—where a larger proportion of the hydrolytic capacity occurs in the mesopelagic zone, below the permanent thermocline (**Figure 7**).

Biogeographical patterns in carbon cycling activities, and their relationship to oceanographic features, are of crucial significance to our ability to predict future conditions. For example, if increasing global temperatures result in a more stratified ocean (Capotondi et al., 2012), the quantity of organic matter sequestered by the biological pump below the thermocline may decrease. Such a decrease in turn would place greater influence on the relative hydrolytic capacities of microbial communities in the surface ocean, rather than in deeper waters, in determining the overall efficiency of the biological pump. The effects of an increase in stratification on carbon export will also depend on its impact on the biogeography of microbial communities and hydrolytic activities themselves. Increased stratification may in turn have complex downstream consequences for higher trophic levels that function in both shallow and deep waters, and depend on the availability of particulate organic carbon in both depth regions. Disentangling the roles of environmental characteristics, microbial community composition, functional capacities, and activities in regulating the marine carbon cycle is a prerequisite for a better understanding of the modern ocean, and of its sensitivity to perturbations in the future.

#### AUTHOR CONTRIBUTIONS

AH and CA designed experiments. AH conducted field experiments, collected and processed samples, and processed data. AH and CA analyzed results and wrote the manuscript.

#### FUNDING

The authors would like to thank Liz Kujawinski for the invitation to join the DeepDOM cruise, and Krista Longnecker for providing nutrient measurements. Funding for this

#### REFERENCES


work was provided by the National Science Foundation (OCE-1332881 to CA).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmars. 2017.00200/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.

Copyright © 2017 Hoarfrost and Arnosti. 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.

# Different Bacterial Communities Involved in Peptide Decomposition between Normoxic and Hypoxic Coastal Waters

#### Shuting Liu<sup>1</sup> , Boris Wawrik<sup>2</sup> and Zhanfei Liu<sup>1</sup> \*

<sup>1</sup> Marine Science Institute, The University of Texas at Austin, Port Aransas, TX, USA, <sup>2</sup> Department of Microbiology and Plant Biology, The University of Oklahoma, Norman, OK, USA

Proteins and peptides are key components of the labile dissolved organic matter pool in marine environments. Knowing which types of bacteria metabolize peptides can inform the factors that govern peptide decomposition and further carbon and nitrogen remineralization in marine environments. A <sup>13</sup>C-labeled tetrapeptide, alaninevaline-phenylalanine-alanine (AVFA), was added to both surface (normoxic) and bottom (hypoxic) seawater from a coastal station in the northern Gulf of Mexico for a 2-day incubation experiment, and bacteria that incorporated the peptide were identified using DNA stable isotope probing (SIP). The decomposition rate of AVFA in the bottom hypoxic seawater (0.018–0.035 µM h−<sup>1</sup> ) was twice as fast as that in the surface normoxic seawater (0.011–0.017 µM h−<sup>1</sup> ). SIP experiments indicated that incorporation of <sup>13</sup>C was highest among the Flavobacteria, Sphingobacteria, Alphaproteobacteria, Acidimicrobiia, Verrucomicrobiae, Cyanobacteria, and Actinobacteria in surface waters. In contrast, highest <sup>13</sup>C-enrichment was mainly observed in several Alphaproteobacteria (Thalassococcus, Rhodobacteraceae, Ruegeria) and Gammaproteobacteria genera (Colwellia, Balneatrix, Thalassomonas) in the bottom water. These data suggest that a more diverse group of both oligotrophic and copiotrophic bacteria may be involved in metabolizing labile organic matter such as peptides in normoxic coastal waters, and several copiotrophic genera belonging to Alphaproteobacteria and Gammaproteobacteria and known to be widely distributed may contribute to faster peptide decomposition in the hypoxic waters.

#### Keywords: peptide, bacteria, DNA stable isotope probing, nitrogen, hypoxia, Gulf of Mexico

#### INTRODUCTION

Proteins and peptides are key components of labile dissolved organic matter (DOM) that supports bacterial growth (Azam, 1998). Small peptides (ca. <600 Da) are key immediate products of microbial protein decomposition owing to the size constraints of bacterial cell membrane transport systems, i.e., porins (Weiss et al., 1991). After proteins are degraded to small peptides, these small peptides can be either taken up directly by bacteria, or hydrolyzed further to individual amino acids via extracellular enzymes with subsequent uptake. The interaction between peptide decomposition and bacteria plays an important role in the cycling of carbon and nitrogen, regeneration of nutrients, and preservation of refractory dissolved organic nitrogen (DON) in marine environments (Aluwihare et al., 2005; Nagata, 2008).

#### Edited by:

Andrew Decker Steen, University of Tennessee, USA

#### Reviewed by:

Xavier Mayali, Lawrence Livermore National Laboratory, USA Craig E. Nelson, University of Hawaii at Manoa, USA Nagissa Mahmoudi, Harvard University, USA

> \*Correspondence: Zhanfei Liu zhanfei.liu@utexas.edu

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 16 November 2016 Accepted: 20 February 2017 Published: 07 March 2017

#### Citation:

Liu S, Wawrik B and Liu Z (2017) Different Bacterial Communities Involved in Peptide Decomposition between Normoxic and Hypoxic Coastal Waters. Front. Microbiol. 8:353. doi: 10.3389/fmicb.2017.00353

Our previous studies have demonstrated that small peptides decompose more quickly in bottom hypoxic than surface normoxic (normal oxygen-saturated) waters in the northern Gulf of Mexico (nGOM), and that the growth of certain bacterial genera such as Vibrio, Marinobacterium, Neptuniibacter, Pseudoalteromonas, Thalassomonas, Amphritea, Roseobacter and Ruegeria, appears to respond to peptide addition (Liu et al., 2013; Liu and Liu, 2016). These results suggest that some bacterial groups may be more effective at metabolizing peptide-derived organic matter in hypoxic seawater, but direct evidence linking specific bacterial lineages to peptide decomposition has not been reported. Knowing which types of bacteria metabolize peptides may be useful in the assessment of factors that control hypoxia formation, as decomposition of labile organic matter leads to consumption of dissolved oxygen (DO) (Wright et al., 2012; Liu et al., 2013). As succession of microbial communities often occurs along with development of oxygen minimum zones (Crump et al., 2007; Zaikova et al., 2010; Parsons et al., 2015), studying the response of microbial communities to labile organic matter at different DO levels can also provide clues about linkages between microbial niche specialization and their resource utilization (Nelson and Wear, 2014).

Previous studies have demonstrated that some bacterial groups can outcompete others during the utilization of labile DOM (Eilers et al., 2000; Teske et al., 2011; Liu et al., 2015). For instance, the bacterial community shifted to Alphaproteobacteria and Betaproteobacteria dominated phylotypes in mesocosm tanks with diatom blooms that produced labile proteins, peptides and polysaccharides exudates (Murray et al., 2007). After bovine serum albumin (BSA) amendment, Gammaproteobacteria became the dominant bacterial class in the Chesapeake Bay water, while Bacteroidetes became dominant in the lower Delaware Bay water (Harvey et al., 2006). However, phylogenybased incubation studies provide only indirect evidence of the role that different bacterial groups play in labile DOM mineralization, and only a few studies to date have linked specific bacteria groups with labile DOM decomposition directly using radioisotope-labeled substrate and microautoradiography combined with fluorescent in situ hybridization (MAR-FISH) technique (Tabor and Neihof, 1982; Ouverney and Fuhrman, 1999). For example, Cottrell and Kirchman (2000) identified that Bacteroidetes and Gammaproteobacteria actively utilized <sup>3</sup>H-labeled protein in two estuarine waters. While powerful, hybridization techniques such as MAR-FISH are a targeted approach and require the design of unique probes to detect individual phylogenetic groups. These techniques often do not allow the identification of active bacteria beyond limited taxonomic depth due to probe hybridization constraints. In contrast, DNA-stable isotope probing (SIP) techniques provide an opportunity to interrogate activity in situ and to identify bacteria at several phylogenetic levels without a priori selection of specific phylotypes. SIP techniques were initially applied to identify bacteria that can degrade one-carbon (C1) compounds or specific pollutants in many environmental studies, such as discovering novel bacteria that degrade methanol, toluene, or alkanes in soils, sediments or marine seeps (Radajewski et al., 2000; Neufeld et al., 2007b; Luo et al., 2009; Redmond et al., 2010; Kleindienst et al., 2014). More recently, the application of DNA-SIP has been extended to marine environments. Examples include studies of urea uptake by marine pelagic bacteria and archaea in Arctic water, comparing bacteria incorporating glucose and cyanobacteria exudates in the Sargasso Sea, and exploring acetate-utilizing bacteria at the oxic–anoxic interface in the Baltic Sea (Gihring et al., 2009; Wawrik et al., 2009; Nelson and Carlson, 2012; Wawrik et al., 2012a; Berg et al., 2013; Connelly et al., 2014), demonstrating the utility of SIP to understand the marine C and N cycles.

The objective of this study was to gain insight into the identities of bacteria that utilize peptides in seawater. In particular, nGOM normoxic and hypoxic waters were targeted because they are characterized by contrasting biogeochemical processes and are known to harbor different bacterial communities that differentially respond to peptide addition. The13C-labeled tetrapeptide alanine-valine-phenylalaninealanine (AVFA) was used as a model compound and incubated in both surface normoxic and bottom hypoxic seawater in the nGOM. The AVFA sequence is derived from the protein sequence of the large subunit of ribulose-1,5-biphosphate carboxylase/oxygenase (RuBisCO) that is ubiquitous in photosynthesis and has been used to investigate peptide hydrolysis (Liu et al., 2010, 2015; Orellana and Hansell, 2012; Liu and Liu, 2014, 2015). Although individual peptides are often undetectable in natural seawater due to their rapid turnover, they play an important role in supporting bacterial growth as intermediates released from sloppy-feeding or lysis of cells (Nagata, 2008; Sipler and Bronk, 2015).

### MATERIALS AND METHODS

#### Seawater Sampling

Surface (2 m) and bottom (16 m) seawater were collected at Sta. C6 (28◦ 52<sup>0</sup> N, 90◦ 30<sup>0</sup> W) in the nGOM during a May 2013 cruise on the R/V Pelican. This station, with a depth of 18 m and ca. 20 km offshore, is heavily influenced by Mississippi River discharge and often subjected to hypoxia during summer (Rabalais et al., 2001). Seawater was sampled using 10 L Niskin bottles mounted on a conductivity-temperature-depth (CTD) rosette (Seabird 911). Temperature, salinity, DO and chlorophyll a of seawater were monitored through the CTD device (**Table 1**). Seawater was filtered immediately onboard through a 0.2 µm Nylon filter (diameter 47 mm, Whatman) and preserved under −20◦C for the analysis of dissolved organic carbon (DOC), total dissolved nitrogen (TDN), total dissolved amino acids (TDAAs), dissolved combined amino acids (DCAAs), dissolved free amino acids (DFAAs) and nutrients.

#### Peptide Incubation

Peptides <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA were custom-synthesized (C.S Bio, Menlo Park, CA, USA), and had a >95% compound purity (Liu et al., 2013). In <sup>13</sup>C-AVFA, 17 (all three carbons in A, all five carbons in V and six carbons of the aromatic ring in F) out of total 20 carbon atoms were labeled isotopically. As DNA of specific bacteria groups might only be partially


TABLE 1 | Chemical parameters of initial surface (2 m) and bottom (16 m) seawater at Sta. C6.

ud, under detection limit (ca. 0.03 µM).

fmicb-08-00353 March 3, 2017 Time: 17:6 # 3

labeled (<100%) with <sup>13</sup>C, bacteria groups incorporating <sup>13</sup>C can be spread through all SIP fractions with different density (see related method below). The unlabeled <sup>12</sup>C-AVFA incubation was thus included as a reference for comparison with <sup>13</sup>C-AVFA incubation to confirm isotopic enrichment in DNA. The unlabeled SIP gradient should contain less DNA in the more dense fractions than that in the samples incubated with <sup>13</sup>C substrates (Neufeld et al., 2007a). AVFA was incubated onboard in the surface normoxic and bottom hypoxic seawater. Briefly, either <sup>12</sup>C-AVFA or <sup>13</sup>C-AVFA was respectively amended in a series of 125 mL amber round bottles filled with 120 mL seawater at final concentrations of 0.25–0.47 µM. Duplicate incubations were conducted in the dark for 48 h at 24◦C, close to the ambient seawater temperature (**Table 1**). At different time points (0, 8, 13, 24, and 48 h), 1 mL aliquots of unfiltered water were collected and fixed with formaldehyde at a final concentration of 3% and stored at 4◦C for bacterial abundance analysis. The remaining 119 mL were filtered through the 0.2 µm Nylon filter and preserved at −20◦C for the analysis of peptides, amino acids, ammonium, and orthophosphate (Pi). The filters were preserved in 750 µL 1× STE (10 mM Tris-HCl [pH 8.0], 0.1 M NaCl, 1 mM EDTA [pH 8.0]) buffer at −20◦C for DNA extraction and sequencing. DO was not monitored throughout the incubation, but the parallel incubation experiment showed that it remained relatively constant throughout the 72 h (Liu and Liu, 2016). Two kinds of controls were included for the incubation experiment: a seawater control without peptide amendment and a killed control with 0.48–0.58 µM <sup>12</sup>C-AVFA and 180 µM HgCl<sup>2</sup> to inhibit bacterial activity (Lee et al., 1992). The incubation and aliquot sampling for the controls followed the same procedures as described above, but only AVFA was analyzed in the killed control.

#### Chemical Analyses

DOC and TDN of the filtered initial seawater were analyzed using a Shimadzu total organic carbon (TOC-V) analyzer coupled with a TNM-1 TDN analyzer with <6% error between duplicates (**Table 1**). DFAA were calculated as sum of individual amino acids analyzed in high performance liquid chromatography (HPLC, Shimadzu Prominence) equipped with a fluorescence detector after pre-column o-phthaldialdehyde (OPA) derivatization (Lindroth and Mopper, 1979; Lee et al., 2000). TDAA were analyzed in the same way as DFAA but after hydrolysis in 6 N HCl under nitrogen at 110◦C for 20 h (Kuznetsova and Lee, 2002). DCAA were calculated as TDAA subtracting DFAA. Nitrate, nitrite and orthophosphate (Pi) were measured following established protocols (Strickland and Parsons, 1968; Jones, 1984).

Alanine-valine-phenylalanine-alanine was analyzed in an HPLC-mass spectrometry (HPLC-MS) system (Shimadzu Prominence) following the method in Liu and Liu (2014). In brief, the mobile phase A was 10 mM ammonium acetate and mobile phase B was methanol. Samples were eluted through a C<sup>18</sup> column (Alltima 5 µm, 150 mm × 4.6 mm) and a six-way valve was programmed to direct the sea salt peak to waste before introducing the AVFA peak to the MS detector that is equipped with an electrospray ionization (ESI) source and a quadrupole mass analyzer. <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA were quantified in positive ion mode under selective ion monitoring (SIM) at m/z = 407 and 424, respectively.

Alanine-valine-phenylalanine-alanine hydrolysis products including peptide fragments (AV, VF, FA, VFA) and amino acids (A, V, F) were analyzed by HPLC after pre-column OPA derivatization (Liu et al., 2013). Standard deviations of amino acid analysis among replicates were 10–20%. Ammonium, a main metabolite of AVFA, was analyzed using HPLC with post-column OPA derivatization (Gardner and St. John, 1991).

### Bacterial Abundance Analysis

Bacterial cells in the formaldehyde-preserved samples were stained with SYBR Green II (Molecular Probes, 1:100 v/v) and enumerated in a flow cytometer (BD Accuri C6) under blue laser excitation at 488 nm (Marie et al., 1997; Liu et al., 2013). Bacterial cells were counted in a fixed volume mode with a flow rate below 300 events per second and cell counts were determined in a dot plot of side scatter (SSC-H) vs. green fluorescence signal (FL1-H) on a logarithmic scale.

#### DNA Extraction and Ultracentrifugation in CsCl Gradients

DNA was extracted from filtered cells using MoBio PowerSoil <sup>R</sup> DNA isolation kits (MoBio Laboratories, Carlsbad, CA, USA). Leftover STE buffer was spun in a centrifuge and the supernatant was discarded. The pellet was re-dissolved in 100 µL lysis solution from MoBio Powerbead tube and then combined with filter in the tube for DNA extraction. A subsample (ca. 10 µL) at each incubation time point was saved for microbial community structure analysis (unamended control 0 h, one duplicate of <sup>12</sup>C-AVFA 2 m 0 h, and one duplicate of <sup>12</sup>C-AVFA 16 m 0 h samples were lost during DNA extraction), and the remainder (ca. 80 µL) was for the ultracentrifugation in CsCl gradients. Duplicate DNA samples (each account for ca. 89% (80 µL out of 90 µL) of extracted DNA) from all three time points (13, 24, and 48 h)

for surface and bottom seawater incubations respectively were pooled (i.e., six samples were pooled) to obtain sufficient DNA for ultracentrifugation and fractionation. The pooled DNA was precipitated using isopropanol, and the DNA pellet was then re-suspended in 50 µL TE buffer (50 mM Tris-HCl, 15 mM EDTA [pH 8.0]) as previously described (Wawrik et al., 2009). Four pooled DNA samples (12C-AVFA surface, <sup>13</sup>C-AVFA surface, <sup>12</sup>C-AVFA bottom, <sup>13</sup>C-AVFA bottom) were prepared for ultracentrifugation. Note that ultracentrifugation and following fractionation were not performed on killed control samples and unamended samples. CsCl gradient ultracentrifugation and fractionation followed protocols as described previously (Buckley et al., 2007; Luo et al., 2009; Wawrik et al., 2012a). In brief, 61–170 ng DNA, quantified through a Qubit <sup>R</sup> 2.0 Fluorometer (Life Technologies), were mixed with 0.26 mL TE buffer and 4.45 mL of 1.295 g mL−<sup>1</sup> CsCl in gradient buffer A (15 mM Tris-HCl [pH 8.0], 15 mM KCl, 15 mM EDTA [pH 8.0], 2 mg mL−<sup>1</sup> ethidium bromide) in 4.7 mL polyallomer Optiseal tubes (Beckman). The tubes were centrifuged in a Beckman rotor VTi 65.2 at ca. 140,000 × g for 48 h. After ultracentrifugation, 30 150 µL fractions were collected from each tube in a Beckman fraction recovery system by replacing samples with mineral oil on top of the tubes at a constant rate using a peristaltic pump. The density of each fraction was calculated based on the refractive index that was measured in a Reichert AR200 refractometer (Wawrik et al., 2009). DNA was purified from each CsCl fraction by isopropanol precipitation and dissolved in 50 µL sterile nuclease-free water.

### Quantitative PCR (qPCR) of 16S rRNA Gene

The purified DNA from each SIP fraction was used to determine 16S rRNA gene copy numbers of bacteria via quantitative polymerase chain reaction (qPCR). Primers were 27F (5<sup>0</sup> -AGA GTT TGA TCM TGG CTC AG-3<sup>0</sup> ) and 519R (5<sup>0</sup> -GWA TTA CCG CGG CKG CTG-3<sup>0</sup> ) (Nakatsu and Marsh, 2007). Every 30 µL reaction mix for qPCR included 13.9 µL 2X Power SYBR Green PCR master mix (Applied Biosystems), 13.9 µL nuclease-free water, 200 nM (final concentration) of each primer, and 2 µL DNA template. qPCR was conducted in a realtime PCR system (Applied Biosystems, ABI 7300) followed the program: 2 min at 50◦C, 8 min at 95◦C, 40 cycles of 30 s at 95◦C, 1 min at 55◦C and 1 min at 72◦C. Genomic DNA of Roseobacter denitrificans Och 114 (DSMZ 7001) was used as the standard DNA for bacteria (standard concentrations ranging from 10−<sup>4</sup> to 10 ng µL −1 ). The qPCR detection limit for bacteria was ca. 0.002 ng (corresponding gene copy number of 442) and cycle threshold was ca. 31. The qPCR data was then normalized to the highest quantities of 16S rRNA gene copy numbers observed among all fractions in that gradient, in order to account for the differential abundances and distributions of their DNA in gradients (Wawrik et al., 2009; Connelly et al., 2014). These normalized numbers were named as ratios of quantities in which the highest normalized frequency measured equaled 1.

### 16S rRNA Gene PCR, Barcoding, and Illumina Sequencing

DNA from each incubation time point and each fraction collected from SIP gradients was amplified by PCR using Phusion high-fidelity DNA polymerase (Thermo Scientific) and barcoded for Illumina sequencing. PCR reactions utilized the universal forward primer 519F containing a 5<sup>0</sup> M13 tag (50 -GTA AAA CGA CGG CCA GCA CMG CCG C-3<sup>0</sup> ) and the reverse primer Bac-785R (5<sup>0</sup> -TAC NVG GGT ATC TAA TCC-3<sup>0</sup> ) as previously described (Wawrik et al., 2012b; Klindworth et al., 2013). PCR started with 94◦C for 2 min, followed by 28–32 cycles of 95◦C for 30 s, 52.8◦C for 30 s, and 72◦C for 30 s, then 72◦C for 5 min. The number of PCR cycles was optimized based on qPCR results and agarose gel check of PCR products to make sure enough PCR products were obtained but not reaching PCR plateau. PCR products were purified by QIAquick PCR purification kit (Qiagen) and barcoded using a unique M13-contianing primer for each sample that contained a 12 bp barcode for bioinformatical parsing of data (Wawrik et al., 2012b). Barcode tagging was checked by gel electrophoresis, amplifications were mixed at equimolar ratios, and sent to the Oklahoma Medical Research Foundation for MiSeq PE250 Illumina sequencing. Sequence reads were trimmed to a quality score of Q30 and adapter and primer sequences were trimmed from raw Illumina sequences. Overlapping forward and reverse reads were stitched and all nonoverlapping sequence reads were discarded. Processed sequences were clustered into OTUs using UCLUST, checked for chimeras using USEARCH and classified into taxonomy through the QIIME pipeline (Caporaso et al., 2010b). A randomly chosen set of representative sequences from each OTU was aligned to the SILVA small-subunit rRNA reference alignment<sup>1</sup> using the PyNAST algorithm (Caporaso et al., 2010a). Sequences were assigned to the genus level at the 95% identity as a compromise between resolution and conservative interpretation due to the short reads (250 bp) used here (Connelly et al., 2014). Sequences were deposited in National Center for Biotechnology Information (NCBI) GenBank under BioProject accession number PRJNA297372.

Bacterial community structures (relative abundance of genera presented in percentage) of the initial samples were assumed to be the same among unamended control, <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA samples. As unamended control and one duplicate of the <sup>12</sup>C-AVFA 0 h samples were lost, initial bacterial community data of <sup>13</sup>C-AVFA samples was more reliable to represent the initial bacterial community structure of all. To be consistent with the pooled SIP samples, results of the bacterial community structures from the three time points (13, 24, and 48 h) were also pooled and compared with others by non-metric multidimensional scaling (NMDS) using Matlab <sup>R</sup> . Analysis of similarity (ANOSIM) was applied to compare the bacterial community structures between surface and bottom seawater and between unamended control and AVFA treatment time-point samples using vegan package in R (Oksanen et al., 2016).

<sup>1</sup>http://www.arb-silva.de

Calculating percentage enrichment of each bacterial taxa in the <sup>13</sup>C-AVFA samples relative to the <sup>12</sup>C-AVFA SIP samples followed the protocol of Bell et al. (2011). In brief, 16S rRNA gene copy numbers for each SIP fraction were quantified through qPCR. Then the proportion of each bacterial taxonomic group sequences in a given density range was multiplied by the 16S rRNA gene copy number in that same density range, and the derived copy number of each bacterial taxonomic group was normalized to total gene copy number within that density range to correct for the slight difference in total DNA between the <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA samples and this normalized gene copy number was referred as relative gene copy number. Percentage enrichment of a certain bacterial taxonomic group was defined as dividing the difference of the relative copy number summed in the heavy density fractions between the <sup>13</sup>C-AVFA samples and the <sup>12</sup>C-AVFA samples by the relative copy number in the <sup>12</sup>C-AVFA samples within the same density range. The percentage enrichment was used as an indicator of the relative amount of <sup>13</sup>C enrichment among bacterial groups. A higher percentage enrichment in one bacterial taxa relative to other

taxa indicates increased cell replication responding to amended <sup>13</sup>C-AVFA. For example, 100% enrichment of a given taxonomic

unamended (no-AVFA) controls. Data points were presented as average ± absolute error of duplicate samples except control samples.

group in the same density range indicates that its 16S rRNA gene copy number in the <sup>13</sup>C treatment is twice as abundant as that in the <sup>12</sup>C treatment. To estimate the error/noise level of the percentage enrichment, 95% confidence interval was calculated from all positive percentage enrichment values among bacterial taxa for both surface and bottom samples.

### RESULTS

### Peptide Decomposition

The <sup>12</sup>C- and <sup>13</sup>C-AVFA decomposition patterns were nearly identical during the 48 h incubation, as expected (**Figures 1A,B**). The AVFA concentrations decreased linearly with time in both the surface (2 m) and bottom (16 m) seawater, but the decomposition rate in the bottom seawater (0.018– 0.035 µM h−<sup>1</sup> ) was twice as high as in the surface seawater (0.011–0.017 µM h−<sup>1</sup> ). The decomposition rate of AVFA in the bottom seawater was significantly higher than that in the surface seawater (t-test, p < 0.03). AVFA was completely degraded within 24–48 h in the surface water and within 13–24 h in the bottom water. In contrast, AVFA concentration in the killed

control remained nearly unchanged during the 48 h incubation (**Figure 1C**), indicating that the peptide disappearance in the seawater was due to microbial activity.

were presented as average ± absolute error of duplicate samples except control samples.

AV, FA, VF, and VFA produced during hydrolysis of AVFA (Liu et al., 2013) remained at low levels <0.012 µM throughout the incubation, but more amino acids and peptide fragments were produced in the surface than in the bottom incubation (**Figure 2**). Concentrations of amino acid F were significantly higher in the surface than in the bottom seawater (t-test, p = 0.04). Concentrations of free amino acids (A, V, and F) were 2–30 times greater than those of peptide fragments. As individual amino acids have been shown in similar incubations to be taken up at different rates (Liu et al., 2013), particularly amino acid A was taken up much faster than V and F, production of amino acids might not follow the stoichiometry of added peptides. F was the dominant amino acid, reaching up to 0.17 µM in the surface at 24 h and 0.082–0.11 µM in the bottom at 8 h, and then decreased to the background level at the end of the incubation. V and A followed a similar pattern to F, but with smaller changes, indicating they were taken up faster than F, consistent with a previous study at this station (Liu et al., 2013). Compared to the AVFA treatment, concentrations of amino acids in the control without peptide amendment remained relatively low (<0.011 µM) and constant throughout the incubation.

Ammonium is a key metabolite of peptides (Liu et al., 2013). During the first 24 h in the surface seawater incubation, ammonium concentrations increased by 0.66–1.45 µM in the <sup>12</sup>C- and <sup>13</sup>C-AVFA samples (Supplementary Figures S1A–C). In contrast, ammonium concentrations in the bottom seawater changed little before AVFA was completely degraded (0–13 h). However, the difference of produced ammonium between surface and bottom seawater was not significant (t-test, p = 0.38). After 13 h, ammonium concentrations kept increasing to 2.6– 2.9 µM in the surface seawater and remained constant at

about 2.8 µM or increased by 1.5 µM to reach 4.1 µM in the bottom seawater. In the control without AVFA, ammonium concentrations increased by 0.97 µM in the surface seawater and decreased by 0.63 µM in the bottom seawater during the 48 h incubation.

Pi is an essential element for bacterial growth (Elser et al., 2000; Karl, 2014; Liu and Liu, 2016). However, P<sup>i</sup> concentrations remained relatively constant throughout the 48-h incubation in both the peptide and control treatments (Supplementary Figures S1D–F). P<sup>i</sup> concentrations in the bottom water (1.1–1.5 µM) were more than one order of magnitude higher than those in the surface water (0.02–0.09 µM).

### Bacterial Abundance and Community Structure

In the surface <sup>12</sup>C- and <sup>13</sup>C-AVFA incubations, bacterial abundance increased by 31–57% within the initial 8–13 h, and then decreased afterward, while in the bottom, bacterial abundance increased by 44–45% during the initial 24 h and then decreased afterward (**Figures 1D,E**). Bacterial abundances in the control either decreased over time in the surface seawater or remained nearly constant in the bottom seawater (**Figure 1F**). Ambient surface water bacterial communities were dominated by Synechococcus (15–49%), whereas bottom samples were more evenly populated by Rhodobacteraceae (11–13%), Acidimicrobiaceae OCS155 marine group (3–8%), Saprospiraceae (5–7%), Planctomycetaceae (2–7%), SAR11 clade Surface 1 (3–6%), and Acidimicrobiales TM214 (3–5%) (Supplementary Figures S2A,B). Different initial surface community structures between <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA samples might be related to the loss of one duplicate <sup>12</sup>C-AVFA DNA sample or contamination of Synechococcus in one sample. Thus, initial surface community structure of the <sup>13</sup>C-AVFA duplicate samples might be more reliable than that of the <sup>12</sup>C-AVFA samples. Throughout the incubation, the relative abundance of Rhodobacteraceae, Thalassococcus, and Ruegeria increased by 6– 14, 7–13, and 1–3%, respectively, in both surface and bottom seawater, while other bacterial genera developed differently in the surface and bottom seawater incubations. For instance, OTUs classified within the Roseovarius clade increased by 7% only in the surface seawater, whereas Colwellia increased by 3% only in the bottom seawater. The surface and bottom bacterial community structures were well-separated in the NMDS plot (Supplementary Figure S2C); ANOSIM showed significant difference between the surface and bottom bacterial community structures (p = 0.002), further suggesting that bacterial community structures developed differently between the two water layers. To further evaluate the bacterial community development in unamended controls vs. AVFA treatments, NMDS was also plotted separately on surface and bottom bacterial communities (**Figure 3**). At the later incubation time points, unamended treatments and AVFA treatments formed separate clusters especially at NMDS1 axis, but this separation was not significant as shown in ANOSIM analysis (p > 0.05), indicating peptide amendement did not significantly change bacterial community structures during the 48-h incubation as compared to controls.

## Identifying Bacteria that Incorporated Peptides through DNA-SIP

Quantitative polymerase chain reaction of SIP fractions indicates that the density distribution of the 16S genes in bulk DNA of the <sup>13</sup>C-AVFA samples shifted to heavier densities as compared to the <sup>12</sup>C-AVFA samples in both the surface and bottom incubations (**Figures 4A,B**). The 16S PCR products from respective fractions were bar-coded and sequenced using Illumina Miseq to generate sequence libraries. A positive percentage enrichment is an indicator of the bacterial potential in incorporating <sup>13</sup>C (Bell et al., 2011). Estimated error/noise of percentage enrichment was 46 and 65% at the class and genus level, respectively, as calculated from the 95% confidence interval. From this error/noise estimate, percentage enrichment above 84% at the class level and above 168% at the genus level was above the 95% confidence interval. <sup>13</sup>C uptake, as suggested by SIP, was more evenly distributed among the bacterial classes in the surface seawater than in the bottom seawater (**Figures 4C,D**). Flavobacteria, Sphingobacteria, Alphaproteobacteria, Acidimicrobiia, Verrucomicrobiae, Cyanobacteria, and Actinobacteria showed highest enrichment ranging from 96 to 275% in the surface water, whereas Alphaproteobacteria and Gammaproteobacteria showed highest enrichment ranging from 175 to 279% in the bottom water.

Communities taking up <sup>13</sup>C in the surface and bottom seawater also differed at the level of dominant genera (>0.1% of the total bacterial community) (**Figures 4E,F**). In the surface seawater, Saprospiraceae, Tropicibacter, Roseovarius, Owenweeksia, Formosa, Flavobacteria NS4 marine group, and Microbacteriaceae SV1-8 dominated the <sup>13</sup>C uptake. In the bottom samples, major <sup>13</sup>C enriched groups included Thalassococcus, Rhodobacteraceae, Ruegeria, Colwellia, Balneatrix, and Thalassomonas. The extent of taxonomic enrichment in the heavier fractions varied widely among different bacterial genera, ranging from 86 to 498% in the surface incubation and from 4 to 646% in the bottom incubation. Within the same class, the enrichment of Roseovarius and Thalassococcus was almost twice as high as that of other genera in Alphaproteobacteria, and the enrichment of Colwellia was more than three times higher than that of other genera in Gammaproteobacteria.

### DISCUSSION

#### Faster AVFA Decomposition in the Hypoxic than in the Normoxic Seawater

Peptide decomposition in the bottom incubation was twice as fast as that in the surface incubation (**Figures 1A,B**). Normalized to initial bacterial abundance, cell-specific rate of peptide decomposition in the bottom incubation (9.0 × 10−9–1.5 × 10−<sup>8</sup> nM h−<sup>1</sup> ) was 1.3–1.4 times as high as that in the surface incubation (6.4 × 10−9–1.2 × 10−<sup>8</sup> nM h−<sup>1</sup> ). Also, AVFA decomposition produced less hydrolyzed

control samples. Stress was 9.2e-07 at 2 m and 3.3e-06 at 16 m, indicating excellent fitting of solution to recreate the dissimilarity (stress < 0.02). Bacterial composition formed separate clusters between unamended control and AVFA incubations at the later time points at both 2 and 16 m as shown in red circles, but ANOSIM showed this separation was not significant (p > 0.05).

fragments, including amino acids and peptides, in the bottom than surface incubations (**Figure 2**), indicating direct uptake of AVFA or tightly coupled hydrolysisuptake in the bottom water but extracellular hydrolysis in the surface water. In contrast to previous studies, which

used relatively high concentrations of added peptides (5–10 µM) (Liu et al., 2013; Liu and Liu, 2016), the much lower concentrations of AVFA (0.25–0.47 µM) added here accounted for only 14–84% of ambient DCAA that consist of all hydrolyzable proteins and peptides in seawater (**Table 1**).

normalized to the highest quantities of 16S rRNA gene copy numbers observed among all fractions in that sample, and 1 equals the highest value observed. Data (Continued)

#### FIGURE 4 | Continued

fmicb-08-00353 March 3, 2017 Time: 17:6 # 10

points were presented as average ± standard deviation of three replicate qPCR measurements. Gray bars indicate heavy density ranges used for percentage enrichment calculations in (C–F). (C,D) Percentage enrichment of major bacterial classes in the heavy density range in the <sup>13</sup>C-AVFA SIP fractions compared to the <sup>12</sup>C-AVFA SIP fractions. The 13, 24, and 48 h DNA samples were pooled together for SIP results. Bacterial class chosen were at least 0.1% abundance of the community. Flavo, Flavobacteria; Sphingo, Sphingobacteria; Alpha, Alphaproteobacteria; Acidi, Acidimicrobiia; Verruco, Verrucomicrobiae; Cyano, Cyanobacteria subsection I; Actino, Actinobacteria; Beta, Betaproteobacteria; Plancto, Planctomycetacia; Gamma, Gammaproteobacteria. Class with percentage enrichment >84% at 95% confidence interval was in bold. (E,F) Percentage enrichment of major bacterial genera within each class (listed above the bars) in the heavy density range of the <sup>13</sup>C AVFA sample SIP fractions compared to the <sup>12</sup>C AVFA sample SIP fractions in the surface 2 m and bottom 16 m seawater. Bacterial genera chosen were at least 0.1% abundance of the community. Bacterial class abbreviation was same as before. Genus with percentage enrichment >168% at 95% confidence interval was in bold.

As individual peptides may exist only at trace levels in ambient seawater, adding low concentration of peptides may simulate natural processes better. However, the low concentration amendments conducted here resulted in uptake patterns generally consistent with previous studies (Liu et al., 2013; Liu and Liu, 2016). Peptide decomposition all followed zero-order reaction with a linear decrease of concentration with time, indicating that the peptide decomposition in this study and previous studies was limited by the availability of enzymes produced by bacteria and the added peptide concentrations were probably all above a threshold of enzyme capacity in the ambient seawater. Faster peptide decomposition and less fragments produced in the bottom than in the surface incubations at both high and low added concentrations suggest that added peptides within this concentration range (0.25–10 µM) may trigger similar peptide decomposition mechanisms and bacterial response.

The peptide decomposition mechanism can be interrogated through a mass balance of the fate of added nitrogen, which may include: (1) extracellular hydrolysis to produce peptide fragments and amino acids, (2) remineralization to ammonium, and (3) incorporation into bacterial biomass. The percentage of extracellular hydrolysis can be estimated using the amino acid F and peptide fragments containing F, as bacterial uptake of F is limited within 24 h (Liu et al., 2013). The degree of remineralization can be estimated via changes in ammonium concentrations in peptide treatments compared to controls assuming nitrification is negligible during the 24 h (Liu et al., 2013). To calculate the incorporation percentage to microbial biomass, we assume a carbon conversion value of 20 fg C per bacterial cell and a C/N ratio of 4 for bacteria (Lee and Fuhrman, 1987). Based on these parameters, extracellular hydrolysis (40–56%) dominated the decomposition of AVFA in the surface water, whereas biomass production (4–20%) dominated in the bottom water throughout the incubation, leaving a major fraction (29–81%) of the AVFA nitrogen uncounted for in both layers, possibly in other forms of DON (**Figure 5**), which might be semilabile and refractory DON formed via microbial production or transformation processes (Jiao et al., 2010; Benner and Amon, 2015; Walker et al., 2016). For example, at 24 h, ca. 40% of decreased AVFA in the surface seawater was hydrolyzed to peptide fragments and amino acids, ca. 6% was converted to ammonium, 2–11% to bacterial biomass (8.1 × 104–3.3 × 10<sup>5</sup> cells per mL), and about 50% to other DON. In contrast, in the bottom seawater, less than 5% was hydrolyzed to peptide fragments and amino acids, hardly any ammonium was produced, and 18–28% was incorporated into bacterial biomass (7.7 × 105–9.0 × 10<sup>5</sup> cells per mL) at 13 h when AVFA disappeared, resulting in about 70–80% of AVFA nitrogen as other DON. This contrasting pattern suggests that the fast disappearance of AVFA in the bottom water incubation may relate to the higher percentage of peptide incorporation into bacterial biomass, i.e., bacterial growth. The efficiency of AVFA decomposition may depend on the fraction of nitrogen allocated to those fast-growing bacteria.

### Uptake of Peptide in the Normoxic vs. Hypoxic Seawater

In the surface seawater incubation, the incorporation of <sup>13</sup>C was greatest for Flavobacteria, Sphingobacteria, Alphaproteobacteria, Cyanobacteria, Acidimicrobiia, Verrucomicrobiae, and Actinobacteria (**Figure 4C**), indicating that both oligotrophic (such as Cyanobacteria) and copiotrophic bacteria were involved in peptide decomposition in the surface seawater. Copiotrophic, perhaps r-selected, bacteria use labile organic matter in nutrient-enriched environments. This is in contrast to more K-selected oligotrophic species that maintain efficient metabolism by growing more slowly on complex refractory substrates (Fierer et al., 2007; Wang et al., 2015). At the genus level, Saprospiraceae (Sphingobacteria), Tropicibacter (Alphaproteobacteria), Roseovarius (Alphaproteobacteria), Owenweeksia (Flavobacteria), Formosa (Flavobacteria), Flavobacteria NS4 marine group (Flavobacteria), and Microbacteriaceae SV1-8 (Actinobacteria) took up the most <sup>13</sup>C in the surface seawater (**Figure 4E**). Sphingobacteria showed a responsive role during peptone incubation in the seawater (Simon et al., 2012). The Roseobacter clade is often associated with plankton aggregates (Moran et al., 2007; Teeling et al., 2012; Yau et al., 2013), so Tropicibacter and Roseovarius belonging to the Roseobacter clade may be opportunistic in nutrient exploitation. Therefore, the observation that these populations can utilize the added peptide is expected. Flavobacteria are often effective in degrading high-molecularweight DOM including proteins (Cottrell and Kirchman, 2000). Some Actinobacteria can produce a wide range of bioactive metabolites including extracellular peptidases that are sometimes involved in pathogenic processes (Ventura et al., 2007; Chen et al., 2011), suggesting their potential in peptide utilization. Consistent with our results, Orsi et al. (2016) also found that diverse bacterial taxa, such as Flavobacteria, Verrucomicrobia, Gammaproteobacteria, Alphaproteobacteria, Actinobacteria, and Planctomycetes, utilized added dissolved proteins in coastal California waters. The phylogenetic widespread of bacterial classes incorporating peptides in this study agrees with ecological theory and previous studies indicating that heterogeneity of the coastal oceans favors generalist bacteria in DOC utilization (Mou et al., 2008). Alternatively, this diverse bacterial pattern may result from the significant production of individual amino acids from extracellular hydrolysis (**Figures 2A,C**). Since uptake of amino acids is generally constitutive among marine bacterial taxa (Payne and Gilvarg, 1971; Poretsky et al., 2010) and uptake of amino acids is also part of peptide metabolizing process, bacterial groups possessing the ability to take up amino acids A, V, and F should be widespread, thus increasing the range of bacteria taxa showing positive percentage enrichment in the surface seawater.

In contrast to the surface seawater incubation, bacteria incorporating <sup>13</sup>C in bottom waters were associated with fewer taxonomic groups, primarily belonging to the Alphaproteobacteria and Gammaproteobacteria (**Figure 4D**). The bacteria that metabolized the peptide differed between the surface normoxic and bottom hypoxic seawater. This depthdifferential response was strikingly consistent with the study of Nelson and Carlson (2012), showing that while both oligotrophic and copiotrophic bacteria incorporated amended DOC sources such as Synechococcus exudate, Synechococcus lysate, and gluconic acid in the euphotic seawater, no oligotrophic bacteria showed evidence of incorporation of amended DOC sources in the mesopelagic seawater. This concomitant pattern in different seawater systems implies that different water parameters between depths are likely to be the driving force in bacterial response to DOM sources. The dominant percentage enrichment of Alphaproteobacteria and Gammaproteobacteria in the bottom seawater suggested they can outcompete other bacteria in incorporating AVFA, thus leading to the faster decomposition of the peptide in the bottom seawater. This result is consistent with previous studies in DOM utilization, which quantified this process either directly through tracing radioisotope incorporation by bacteria or indirectly via analyzing changes of bacterial community structure (Gihring et al., 2009; McCarren et al., 2010; Carney et al., 2015). For example, the percentage of Gammaproteobacteria consuming proteins was higher than their abundance percentage among all bacterial phylogenetic groups in estuarine and coastal environments, indicating they are efficient at metabolizing proteins (Cottrell and Kirchman, 2000). Alphaproteobacteria or Gammaproteobacteria can dominate the bacterial community during DOM incubation in certain marine environments, indicating they can outcompete other bacteria in using DOM substrates (Harvey et al., 2006).

At the genus level, the highest percentage enrichment in the bottom seawater occurred to the Thalassococcus (Alphaproteobacteria), Rhodobacteraceae (Alphaproteobacteria), Ruegeria (Alphaproteobacteria), Colwellia (Gammaproteobacteria), Balneatrix (Gammaproteobacteria), and Thalassomonas (Gammaproteobacteria) (**Figure 4F**). Thalassococcus has been shown to be capable of utilizing phthalate (Iwaki et al., 2012), but its ability to metabolize peptides, as suggested here, has not yet been explored. Previous studies have shown that Rhodobacterales are often one of the dominant groups in coastal seawaters, accounting for as high as 75% of the Alphaproteobacteria (Dong et al., 2014). Their abundance is thought to be related to DOC concentrations in nutrient-enriched habitats and they are frequently involved in taking up labile organic molecules, such as peptides and amino acids, as detected by metaproteomics (Dong et al., 2014; Fodelianakis et al., 2014). Our previous study also showed that populations of Ruegeria, Thalassomonas, Pseudoalteromonas, and Neptuniibacter grew rapidly when AVFA was amended to the same Sta. C6 bottom water (Liu et al., 2013). These genera contain many copiotrophs. Copiotrophs, such as Ruegeria, Vibrio, Alteromonas, and Colwellia, grow rapidly when substrates are available, but also maintain growth potential under starvation conditions. This is akin to a "feast or famine" strategy that allows adaptation to rapidly changing environments (Eilers et al., 2000; Christie-Oleza et al., 2012). Their high capability to assimilate peptide is thus consistent with their ecology strategy. The growth of Pseudoalteromonadaceae and Colwellia increased when peptone was incubated in the Southern Ocean seawater (Simon et al., 2012). Consistently, copiotrophic bacteria, such as Vibrio, Roseobacter, Pseudoalteromonas, Photobacterium, Marinomonas, Marinobacter, and Alteromonas, dominated the incorporation of

DOC sources from Synechococcus exudate or lysate in seawater culture incubations (Nelson and Carlson, 2012). Particleattached Colwellia and Pseudoalteromonas also showed high incorporation of proteins in marine microcosms (Mayali et al., 2015). DOC-related transporter genes, such as amino acids, oligopeptides, carbohydrates, carboxylic acids, polyamines, and lipids transporters, in coastal seawater were associated with Rhodobacterales (primarily Roseobacter), Rickettsiales, Flavobacteriales, and five orders of Gammaproteobacteria, including Alteromonadales, Oceanospirallales, Pseudomonadales, Vibrionales, and an uncharacterized taxon related to sulfuroxidizing symbionts (Poretsky et al., 2010). Most of these bacteria also assimilated the peptide used in our study.

Changes in bacterial community structure that developed through incubations were not significantly different among the peptide treatment and control samples (**Figure 3**, ANOSIM p > 0.05, Supplementary Figure S2). These data suggest that peptide addition at relatively low concentrations had a minimal effect on the overall community structure. This also highlights that bacterial community structure cannot necessarily be used to infer roles of individual bacterial populations in incubation experiments. In contrast, the SIP technique directly links bacterial taxa with a metabolic function such as peptide decomposition. A direct comparison between the relative change in bacterial community structure and AVFA utilizing taxa, as determined via SIP, points to the interpretation that abundant bacterial taxa are not necessarily the most active ones (**Table 2**). For instance, Saprospricaea, Escherichia–Shigella, Balneatrix, and Thalassomonas accounted for <2% of communities and changed only <1% throughout the incubation while their abundance enrichment in heavy SIP fractions was 90–215%. Microbacteriaceae remained below 5% and their abundance did not increase with time during incubations, while frequencies increase by nearly 500% in heavy SIP fractions from surface seawater incubations. Similar observations have been reported elsewhere (Zemb et al., 2012), and indicate that some bacteria can be highly enriched in <sup>13</sup>C, but they may represent only a small proportion of the overall community. These rare bacteria may have long generation time with 10s of hour or more (Brock, 1971). During our short 48 h incubation, certain bacteria might be at lag phase of growth, which changed little in the community structure. For example, if these rare bacteria only doubled once during 48 h, their increase from ca. <1% to ca. <2% would not contribute much the overall community structure. Alternatively, these rare bacteria might have utilized the assimilated peptides mostly for respiration instead of for biomass building, leading to the mismatch between abundance and SIP incorporation. These data showed the potential role of some rare and uncultivable bacteria in peptide utilization, which is often overlooked based on bacterial community structure analysis. This uncoupling between microbial abundance and activity is also consistent with other studies showing uncoupled pattern between rDNA and rRNA for some bacterial populations (Campbell and Kirchman, 2013; Caruso et al., 2013; Hunt et al., 2013), reflecting bacterial activity

TABLE 2 | Comparison between the relative percentage change of bacteria genera (average percentage at 13+24+48 h relative to percentage at 0 h in quasi-replicates (n = 4) of <sup>12</sup>C-AVFA and <sup>13</sup>C-AVFA treatments) in the bacterial community structure and the percentage enrichment of bacteria genera showing positive enrichment in the SIP heavy fractions.


Percentage enrichment above 168 at 95% confidence interval was in bold.

rates are not necessarily correlated to their abundance and these two parameters may be controlled by different factors.

### Factors Leading to the Development of Different Bacterial Communities

It is intriguing that bacterial communities that incorporated the added peptide differed in surface and bottom incubations. The two layers differed in chemical and biological parameters (**Table 1**, **Figures 1D–F**, **3**, and Supplementary Figure S2), such as DO, DOC, and initial bacterial community structure, which probably contributed to the development of different bacterial communities, but the role of these factors seems to be limited (Liu et al., 2013; Liu and Liu, 2016). Other than these parameters, high levels of P<sup>i</sup> (>0.4 µM) in the bottom seawater may stimulate the growth of fast-growing bacteria with high RNA content (Liu and Liu, 2016), such as Alphaproteobacteria and Gammaproteobacteria, consistent with the Growth Rate Hypothesis (Elser et al., 2000; Makino et al., 2003). The fastgrowing bacteria may lead to faster peptide decomposition observed in the bottom than in the surface seawater. Assuming 0.2 pg dry mass bacterial cells and a P content of 1.3% (Sterner and Elser, 2002), the bacterial abundance increase observed here would have required 0.01–0.08 µM P<sup>i</sup> . These small values are close to the standard deviation (ca. 0.02 µM) of P<sup>i</sup> measurement, which may explain why no obvious decrease of P<sup>i</sup> was observed during our incubations (Supplementary Figures S1D,E). On the other hand, these results further suggest that the level of P<sup>i</sup> , rather than its absence, is the key factor limiting the development of fast-growing bacteria, supporting our previous hypothesis (Liu and Liu, 2016). The unique development of certain alphaproteobacterial and gammaproteobacterial genera may also explain the much lower production of AVFA fragments during the bottom water incubation compared to the surface water incubation (**Figures 2B,D**). Either these bacteria directly took up the peptide, or the hydrolysis and subsequent uptake of the fragments were tightly coupled (Fuhrman, 1987; Kuznetsova and Lee, 2002; Liu et al., 2013). These two processes cannot be differentiated with these data, but regardless, both pathways differ from that of the surface incubation, where hydrolysis and uptake seem uncoupled.

## Factors to be Considered for the DNA-SIP Approach

A successful DNA-SIP experiment depends on the amount of isotopically labeled substrate being assimilated and the length of the incubation time (Radajewski et al., 2003; Neufeld et al., 2007a). The substrate concentration must be high enough to ensure sufficient isotopic labeling of nucleic acids relative to unlabeled background substrates that are relatively abundant. However, if the substrate concentrations are too high, the incubation may deviate from the in situ situation. In our incubations, we added relatively low concentrations (0.25–0.47 µM) of AVFA to minimize disturbance to the natural substrates. qPCR results support the notion that sufficient isotope was incorporated into bacterial DNA. Successful uptake of peptide by bacteria is also indicated through the increased bacterial abundance in peptide treatments compared to control. Longer incubation time often results in greater isotope incorporation, but may also lead to crossfeeding, such as bacterial assimilation of labeled byproducts, intermediates or dead cells, produced from substrate metabolism (Neufeld et al., 2007a,c; Wang et al., 2015). To reduce crossfeeding, we applied relatively short incubation time (48 h) that was nonetheless sufficient to allow complete peptide loss.

A potential limitation of the DNA-SIP approach is that the buoyant density of DNA varies with G+C content. As G+C content may vary among different bacteria, this may result in a loss of power to identify bacteria that have incorporated the labeled substrate based on density shift (Buckley et al., 2007). However, it is more problematic for <sup>15</sup>N than for <sup>13</sup>C substrates given the greater buoyant density differential for nucleic acids labeled with <sup>13</sup>C. The density shift in our results was >0.01 g mL−<sup>1</sup> , equating to ca. 28% of <sup>13</sup>C incorporation, which is more than the minimum percentage (20%) that is typically required for separating <sup>13</sup>C and unlabeled organisms (Uhlik et al., 2009). Note that the overall buoyant density differed somewhat between the surface and bottom DNA fractions (**Figures 4A,B**). It is unclear why this difference was observed, but may be related to the different bacterial community composition in the surface and bottom incubations, as %G+C contents of DNA vary among different bacterial taxa and higher %G+C leads to heavier density (Buckley et al., 2007; Holben, 2011). However, it is presumed that this density difference will not affect our ability to identify bacteria incorporating <sup>13</sup>C, because the taxonomic percentage enrichment was derived relative to the corresponding <sup>12</sup>C-AVFA incubations within the surface or bottom samples. We note that DNA from three incubation time points (13, 24, 48 h) was pooled. This approach results in 'smearing' of the signal by spreading the DNA of active bacterial taxa across the gradient density range to a greater degree than for a single time point. However, this smearing should not be problematic with respect to the objectives of this study, because major bacterial taxa in the bacterial community structure were similar at all these three time points and the chemistry data showed a continuous pattern among these three time points. AVFA were completely degraded during 24–48 h in the surface seawater and during 13–24 h in the bottom seawater. Bacterial cell replication and DNA synthesis may have time lag after peptide incorporation due to their 9–12 h generation time (Eilers et al., 2000). Bacterial abundance was still increasing after 13 h in the bottom seawater (**Figures 1D,E**), indicating bacteria were still utilizing peptides for their growth shortly after peptide was completely degraded. To be consistent between two depths and make sure enough <sup>13</sup>C signal is obtained, pooling the last three time points seems an appropriate choice. As surface and bottom incubations were treated in the same way for SIP samples, the comparison between two waters still holds with pooled samples. While the exact degree of isotopic labeling or peptide incorporation may therefore not be attainable from our experiments, the high degree of enrichment observed for some bacteria (**Figures 4C–F**) supports the notion of active <sup>13</sup>C incorporation.

#### CONCLUSION

fmicb-08-00353 March 3, 2017 Time: 17:6 # 14

Work presented here builds on prior observations with respect to the inferred role of bacteria in peptide decomposition (Liu et al., 2013; Liu and Liu, 2016). Here, we directly linked specific bacterial taxa with peptide decomposition in surface and bottom waters in the hypoxic region of northern Gulf of Mexico. Major conclusions and implications from this study are as follows:


#### REFERENCES


Brock, T. D. (1971). Microbial growth rate in nature. Bacteriol. Rev. 35, 39–58.

Buckley, D. H., Huangyutitham, V., Hsu, S. F., and Nelson, T. A. (2007). Stable isotope probing with 15N achieved by disentangling the effects of genome G+C hypotheses about diverse microbial groups and their functions in marine environments. As hypoxia may be intensified in the future scenario, investigating bacterial decomposition of labile DOM under different nutrient conditions is necessary to pinpoint the factors controlling hypoxia formation.

#### AUTHOR CONTRIBUTIONS

All authors listed have made substantial, direct and intellectual contributions to the work and approved its final version for publication.

#### FUNDING

This work is funded by the Chemical and Biological Oceanography Programs of the National Science Foundation (OCE-1129659 & OCE-1634630).

#### ACKNOWLEDGMENTS

We thank the help from the crew of R/V Pelican. We appreciate J. Liu for his help with the incubation experiment, Dr. C. Shank for analyzing DOC samples and Dr. T. Villareal for his help with bacterial abundance analysis. We thank comments from Dr. W. Gardner, Dr. D. Erdner, Dr. J. McClelland and Dr. D. Kirchman. We are grateful for the DNA sequencing by the Oklahoma Medical Research Foundation.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.00353/full#supplementary-material

content and isotope enrichment on DNA density. Appl. Environ. Microbiol. 73, 3189–3195. doi: 10.1128/AEM.02609-06


the sea. Proc. Natl. Acad. Sci. U.S.A. 107, 16420–16427. doi: 10.1073/pnas. 1010732107


fmicb-08-00353 March 3, 2017 Time: 17:6 # 16

unusual sulfur biogeochemistry in a hypersaline Antarctic lake. ISME J. 7, 1944–1961. doi: 10.1038/ismej.2013.69


**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 Liu, Wawrik 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.

# Potential Activities of Freshwater Exo- and Endo-Acting Extracellular Peptidases in East Tennessee and the Pocono Mountains

Lauren Mullen<sup>1</sup> , Malcolm X Shabazz High School Aquatic Biogeochemistry Team<sup>2</sup> , Kim Boerrigter2,3, Nicholas Ferriero<sup>2</sup> , Jeff Rosalsky<sup>4</sup> , Abigail van Buren Barrett<sup>1</sup> , Patrick J. Murray<sup>2</sup> and Andrew D. Steen<sup>1</sup> \*

<sup>1</sup> Department of Earth and Planetary Sciences, University of Tennessee, Knoxville, Knoxville, TN, United States, <sup>2</sup> Malcolm X Shabazz High School, Newark, NJ, United States, <sup>3</sup> Harvard College, Cambridge, MA, United States, <sup>4</sup> Pocono Environmental Education Center, Dingmans Ferry, PA, United States

#### Edited by:

George S. Bullerjahn, Bowling Green State University, United States

#### Reviewed by:

Kaarina Sivonen, University of Helsinki, Finland Izabela Marques Dourado Bastos, University of Brasília, Brazil

> \*Correspondence: Andrew D. Steen asteen1@utk.edu

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 29 September 2017 Accepted: 16 February 2018 Published: 06 March 2018

#### Citation:

Mullen L, Malcolm X Shabazz High School Aquatic Biogeochemistry Team, Boerrigter K, Ferriero N, Rosalsky J, Barrett AB, Murray PJ and Steen AD (2018) Potential Activities of Freshwater Exoand Endo-Acting Extracellular Peptidases in East Tennessee and the Pocono Mountains. Front. Microbiol. 9:368. doi: 10.3389/fmicb.2018.00368 Proteins constitute a particularly bioavailable subset of organic carbon and nitrogen in aquatic environments but must be hydrolyzed by extracellular enzymes prior to being metabolized by microorganisms. Activities of extracellular peptidases (proteindegrading enzymes) have frequently been assayed in freshwater systems, but such studies have been limited to substrates for a single enzyme [leucyl aminopeptidase (Leu-AP)] out of more than 300 biochemically recognized peptidases. Here, we report kinetic measurements of extracellular hydrolysis of five substrates in 28 freshwater bodies in the Delaware Water Gap National Recreation Area in the Pocono Mountains (PA, United States) and near Knoxville (TN, United States), between 2013 and 2016. The assays putatively test for four aminopeptidases (arginyl aminopeptidase, glyclyl aminopeptidase, Leu-AP, and pyroglutamyl aminopeptidase), which cleave N-terminal amino acids from proteins, and trypsin, an endopeptidase, which cleaves proteins mid-chain. Aminopeptidase and the trypsin-like activity were observed in all water bodies, indicating that a diverse set of peptidases is typical in freshwater. However, ratios of peptidase activities were variable among sites: aminopeptidases dominated at some sites and trypsin-like activity at others. At a given site, the ratios remained fairly consistent over time, indicating that they are driven by ecological factors. Studies in which only Leu-AP activity is measured may underestimate the total peptidolytic capacity of an environment, due to the variable contribution of endopeptidases.

Keywords: extracellular enzymes, aminopeptidase, endopeptidase, freshwater, trypsin, protein

## INTRODUCTION

Kinetics of extracellular enzymes can give insight into the rates and pathways of organic matter processing in the environment (Schimel and Weintraub, 2003; Arnosti et al., 2014; Sinsabaugh et al., 2014). Diverse classes of extracellular enzymes have been observed in freshwaters, including peptidases, polysaccharide hydrolases, phosphatases, lipases, peroxidases, and laccases (Findlay and Sinsabaugh, 1999). Peptidases can be particularly valuable to microbial communities, because proteins provide organic N as well as C, and because protein-like organic matter is on

average more bioavailable than bulk natural organic matter (Fellman et al., 2010). The peptidases are a highly diverse class of enzymes: more than 300 distinct peptidases have been identified by function (McDonald et al., 2009) and many more have been identified by structure (Rawlings et al., 2016). Nevertheless, environmental studies have focused almost exclusively on the activity of a single extracellular peptidase, leucyl aminopeptidase (Leu-AP), which preferentially cleaves leucine from the N-terminus of proteins or peptides (Arnosti et al., 2014). In seawater, a diverse suite of aminopeptidases and endopeptidases (which cleave peptide bonds within proteins; Hooper, 2002), from bacteria as well as protists, are required for complete breakdown of proteins (Thao et al., 2014). Ratios of LeuAP and other aminopeptidases to endopeptidases can vary widely (Obayashi and Suzuki, 2005, 2008; Steen and Arnosti, 2013). The extent of this variation in freshwaters, and therefore the extent to which potential LeuAP activities represent total peptidolytic potential of freshwater ecosystems, remains unknown.

In order to constrain the degree to which LeuAP activities represent the total range of extracellular peptidases active in freshwaters, we assayed the potential activities of five different classes of extracellular peptidases in 28 freshwater bodies in southwest Pennsylvania (PA) and east Tennessee (TN) between 2013 and 2016. In addition to Leu-AP, we used substrates that putitively assay arginyl aminopeptidase (ArgAP), prolyl aminopeptidase, glycyl aminopeptidase (GlyAP), pyroglutamyl aminopeptidase, and trypsin. The first four of these are aminopeptidases while trypsin is an endopeptidase. Uniquely among amino acids in the substrates assayed here, pyroglutamic acid is not directly encoded by DNA and is not typically abundant in biomass. However, it does exist in low quantities in some environmentally relevant biomolecules, for instance, bacteriorhodopsin (Gerber et al., 1979). Freshwater organic matter contains a complex mixture of proteins and proteinlike molecules that require a diverse suite of extracellular enzymes to efficiently remineralize (Arnosti et al., 2014). A better understanding of the nature of extracellular peptidases in aquatic environments could therefore shed light on the mechanisms by which organic matter is oxidized in such systems.

#### MATERIALS AND METHODS

#### Sites and Sample Collection

Samples were collected from 28 locations in and around Knoxville, TN, United States, and in the Pocono Mountains of eastern PA near the Pocono Environmental Education Center in 2013, 2015, and 2016 (PEEC; **Figure 1** and **Table 1**). Water samples were collected by hand in acid-rinsed, 1-L polyethylene bottles. In situ water temperature was measured at the time of sampling. For the Knoxville samples, pH was measured by electronic pH meter (Accumet AB150) upon return to the University of TN. For the PEEC samples, pH was measured using pH strips (2013–2015, 2016 YSI Pro DSS Sonde). A portion of the collected samples have missing pH values; these samples have been recorded as n.m in **Table 1**. Methods were developed through progressive years of the study and evolved to be more efficient; pH data were collected for most samples but were neglected in the early studies. Samples were kept at in situ temperature in the dark and returned to the lab within 1 h for enzyme assays. To assess temporal variability in peptidase activities, a short time series of six samples each was collected from the Third Creek (3rd) and Volunteer Landing (VL) sites during the period from June 8 to July 6, 2015.

#### Enzyme Assays

Enzyme assays were performed using fluorogenic substrates (Hoppe, 1983) according to a modified version of the protocol described by Steen and Arnosti (2011). The following


n.m. indicates 'not measured'.

substrates were used: Arg-7-aminomethylcoumarin (AMC), Gly-AMC, Leu-AMC, Pyr-AMC, and Z-GlyGlyArg-AMC. Details of substrates are given in **Table 2**.

The four aminopeptidase substrates were chosen to represent a broad range of amino R group chemistries, including nonpolar (Leu), polar (Arg), small (Gly), and pyroglutamic acid, which is non-proteinogenic and which has an unusual cyclic R group. Z-GlyGlyArg-AMC, the only endopeptidase substrate used due to cost constraints, was chosen because hydrolysis of it was consistently observed (Obayashi and Suzuki, 2005, 2008). We note that the Z-(carboxybenzyl-) group on this substrate is a bulky protecting group that prevents sequential hydrolysis of the substrate by aminopeptidases. Throughout this manuscript, we use the Enzyme Commission (EC) system to refer to the enzymes that hydrolyze these substrates, in which enzymes are classified according to their function without regard to structure (Webb, 1992) because we lack any data (e.g., nucleic acid sequences) on enzyme structure. This is a shortcut: multiple peptidases are capable of catalyzing the hydrolysis of each of these substrates, as discussed below. For instance, both trypsin (EC 3.4.21.4) and oligopeptidase B (EC 3.4.21.83) are capable of catalyzing the hydrolysis of peptide bonds with N-adjacent Arg, despite major structural differences. The enzyme names used here are therefore consistent with specific enzyme classes, but not necessarily diagnostic of them.

In 2013, saturation curves (measurements of substrate hydrolysis rate as a function of substrate concentration at 0, 50, 100, 150, 200, 250, 300, and 400 µM) were measured at each site, with a single live replicate and matching killed control (boiled for ca. 5 min) at each concentration, plus triplicate live measurements at 400 µM. In 2014–2016, triplicate, saturating concentrations of 400 µM substrate were used in each incubation as well as a single killed control; 40 µL of substrate (10 mM stock concentration, dissolved in 90% MilliQ-H2O/10% DMSO) was added to 100 µL of phosphate buffer (100 mM, pH 7.5) and 860 µL unfiltered sample, in a 1-mL methacrylate cuvette.

#### TABLE 2 | Substrates used.

fmicb-09-00368 March 6, 2018 Time: 14:19 # 4


Suppliers and product numbers are examples. Different suppliers for identical molecules were used over the course of the study.

The cuvette was capped and mixed by hand. Measurements were taken approximately every 20 min for 2 h using a Promega Glomax Jr. (Ex 365 nm, Em 410–460 nm), Promega Quantifluor ST (Ex 365–395 nm, Em 440–470 nm), or Turner Biosystems TBS-380 fluorescence detector (Ex 365–395 nm, Em 440– 470 nm), each set to UV mode. Samples were incubated at in situ temperature (PA samples; in situ temperatures ranged from 21.4 to 25.6◦C or at room temperature (20–21◦C; TN samples). For every sample, a calibration curve was made using AMC standard dissolved in MilliQ-H2O mixed with 860 µL sample, 100 µL phosphate buffer, and an addition of MilliQ-H2O to bring the total volume to 1000 µL.

pH dependence for Gly-AMC and Leu-AMC hydrolysis was measured at Ardena Brook and Belmar Inlet in 2016. For pH optimum measurements, the procedure was the same, but the buffer was phosphate-citrate universal buffer, and the pH was manipulated from 5.0 to 9.0. For these measurements, a standard curve was created at each pH, and each sample was calibrated with the corresponding calibration curve.

#### Data Analysis and Quality Control

Enzyme activities were calculated using R. All data and scripts are included as supplemental data, and deposited at http://github.com/adsteen/PEEC\_MXSHS. Data were manually checked for linearity, and obvious outlier fluorescence data points were removed from the dataset based on the observation that our fluorescence detectors sometimes exhibit shot noise. Samples with outlier v<sup>o</sup> values were not removed. Vmax and K<sup>m</sup> were calculated using the non-linear least-squares fitting algorithm implemented by the nls() function in base R. Fits for which estimated Vmax and K<sup>m</sup> were both greater than 0, and for which the standard error of estimated K<sup>m</sup> was less than the estimated value of Km, were considered valid. As a second quality control step, fits meeting those criteria but which qualitatively did not appear to fit the data well were omitted from analysis. Note that measured Kms are effective Kms, since multiple enzymes almost certainly hydrolyzed each substrate.

#### RESULTS

#### Potential Kinetics of Extracellular Peptidases

Collectively, potential enzyme activities were distributed approximately log-normally, with a geometric mean Vmax of 91 nM h−<sup>1</sup> , a median of 73 nM h−<sup>1</sup> , and an interquartile range from 21 to 520 nM h−<sup>1</sup> (**Supplementary Figure S1**). In Knoxville, activities were highest in Douglas Lake (DL), the TN River at VL, and a small outdoor, constructed goldfish pond (EBF). DL showed high activity both times that it was sampled. The highest activities of LeuAP were observed in the EBF and at VL. At PEEC, the highest activity was measured at sites BB1, BB2, and BS (two approximately 30-m diameter, shallow catchment ponds, and a highly turbid wetland, respectively) which were characterized by high potential ArgAP and LeuAP activities. Trypsin-like activities were consistently high in DL and at the Hesler Biology Plant Pond. One-way repeatedmeasures ANOVA of log-transformed Vmax (n = 188) revealed significant differences in Vmax among substrates (p < 0.001). Pairwise paired t-tests of difference in Vmax means among samples revealed statistically significant differences in Vmax among each pair of substrates except ArgAP and trypsin-like enzymes, which were indistinguishable (p > 0.05; p-values corrected for multiple comparisons by the Bonferroni–Holm algorithm). Vmax of LeuAP was greatest, followed by ArgAP and trypsin-like enzymes, and then by GlyAP, and finally PyrAP (**Figure 2**).

Of the 50 sample/substrate combinations for which saturation curves were created in 2013, 20 were able to be fit to the

Michaelis–Menten function, vo= Vmax[S] Km+[S] , where v<sup>o</sup> is the observed rate of reaction, [S] is the substrate concentration, Vmax is the theoretical maximum rate of reaction at infinite substrate concentration, and K<sup>m</sup> is the effective half-saturation constant. In general, samples for which v<sup>0</sup> in live samples was considerably greater than boiled controls yielded valid Michaelis–Menten fits, whereas those in which v<sup>0</sup> was low did not. Thus, effective K<sup>m</sup> values could be estimated for each peptidase except Pyr-AP, which exhibited consistently low v0. Effective K<sup>m</sup> values ranged from a minimum of 15.6 µM to a maximum of 869 µM with a median of 101 µM and interquartile range from 66.3 to 273 µM (**Figure 3**).

All potential peptidase activities were significantly intercorrelated after log transformation (**Figure 4**; ArgAP-LeuAP: slope = 0.93 ± 0.05, n = 39, r <sup>2</sup> = 0.90, p < < 0.01; GlyAP-LeuAP: slope = 0.69 ± 0.04, n = 40, r <sup>2</sup> = 0.87, p < < 0.01; trypsin-like enzyme-LeuAP: 0.94 ± 0.10, n = 40, r <sup>2</sup> = 0.68, p < < 0.01; PyrAP-LeuAP: slope = 0.63 ± 0.11, r <sup>2</sup> = 0.54, n = 30). At individual sites, ratios of potential trypsin-like enzymes:LeuAP ranged from 0.037 to 9.3. These ratios were roughly log-normally distributed with a geometric mean of 0.53, a median of 0.46, and an interquartile range from 0.18 to 1.4 (**Supplementary Figure S2**). GlyAP and LeuAP pH dependences were indistinguishable at each site, although they were different among sites (**Supplementary Figure S3**). At Ardena Brook, both aminopeptidases were most active at pH 7.5, while at Belmar Inlet, both aminopeptidases were most active at or above pH 8.5. Potential peptidase activities were not significantly correlated to in situ temperature, probably because cell density or other ecological factors exerted stronger control over enzyme activity than temperature in the relatively narrow temperature range (18–32◦C) sampled here.

#### Intertemporal Stability of Peptidase Activity Ratios

Time-series measurements from Third Creek (an urban, anthropogenically impacted creek in Knoxville, TN, United States; Im et al., 2014) and in the TN River at VL (upstream of most of Knoxville's drainage basin) indicated that patterns of enzyme activity were relatively stable on a timescale of weeks (**Figure 5**). Third Creek consistently displayed higher activity of ArgAP and trypsin-like enzymes than the TN River, which displayed higher activities of LeuAP and GlyAP. PyrAP was always negligible (but sometimes detectable) at both sites. Sites BB1 and BB2 were also sampled over multiple years and consistently showed higher LeuAP than trypsin-like potential activity.

#### DISCUSSION

The shape of the saturation curves and the fact that substrate hydrolysis rates in untreated samples were generally substantially higher than those in boiled samples indicate that the substrate hydrolysis observed here reflects activities of enzymes rather than abiotic processes. The median K<sup>m</sup> value here, 101 µM, is somewhat higher than the median hydrolase K<sup>m</sup> reported in a meta-analysis of extracellular enzyme kinetics, suggesting a moderately high concentration of enzyme-labile proteinaceous organic matter in the systems assayed here (Sinsabaugh et al., 2014). Potential peptidase activities (Vmax) in this study varied over four orders of magnitude among environments and were all mutually inter-correlated. Vmax values were not significantly correlated to in situ temperature, likely because ecological factors (e.g., cell density and organic matter concentration) were more important than the kinetic effect of temperature in driving enzyme activity, and because the range of temperatures sampled (18–32◦C) was relatively narrow. Those correlations could indicate that the assays used here report activities of two distinct enzymes, expression of which is correlated at the community level. Alternately, correlations between two substrate hydrolysis rates could indicate that the same enzyme or set of enzymes hydrolyzes multiple fluorogenic substrates. Both factors likely led to the observed data. Extracellular aminopeptidases in freshwater are relatively promiscuous, and multiple classes of aminopeptidase can hydrolyze the same substrates (Steen et al., 2015). In that study, ArgAPs were responsible for more hydrolysis of LeuAMC than were Leu-APs. The tight inter-correlation between hydrolysis rates of LeuAMC, ArgAMC, and GlyAMC, combined

with the evidence for promiscuity among aminopeptidases, suggests that those substrates may have been hydrolyzed by the same enzyme or set of enzymes. This is further supported by the fact that pH dependence of GlyAP and LeuAP, which were indistinguishable from each other at two different sites, despite varying among sites (**Supplementary Figure S3**). The fact that Leu-AMC was consistently the fastest hydrolyzed substrate suggests that LeuAP, rather than GlyAP or ArgAP, was responsible for most of that hydrolysis.

The correlations between Leu-AMC and Z-GlyGlyArg-AMC hydrolysis rates and between Leu-AMC and Pyr-AMC hydrolysis rates (r <sup>2</sup> = 0.68 and 0.63, respectively) were considerably looser than the correlations between Leu-AMC and Arg-AMC or Gly-AMC hydrolysis rates (r <sup>2</sup> = 0.90 and 0.87, respectively). Correspondingly, the ratios of Z-GlyGlyArg-AMC and Pyr-AMC to Leu-AMC hydrolysis rates at individual sites were significantly more variable than ratios of Arg-AMC and Gly-AMC to Leu-AMC hydrolysis rates (**Figure 4** and **Supplementary Figure S2**). These facts suggest that, while Leu-AMC, Gly-AMC, and Arg-AMC were likely hydrolyzed by the same set of enzymes, different sets of enzymes hydrolyzed Z-GlyGlyArg-AMC and Pyr-AMC. This makes sense from a biochemical perspective: the unusual cyclic lactam structure of pyroglutamc acid is a poor fit for the active site of a typical aminopeptidase, and indeed N-terminal pyroglutamic acid acts to protect peptides from intracellular hydrolysis by aminopeptidases (Kumar and Bachhawat, 2012). Aminopeptidases specific for pyroglutamic acid have been identified (EC 3.4.19.3, Awadé et al., 1994), and pyroglutamic acid is a minor component of some proteins relevant to aquatic systems, such as bacteriorhodopsin (Blanck et al., 1989). Thus, it is plausible that the hydrolysis of Pyr-AMC observed in these samples was due to pyroglutamic aminopeptidase, but given the low activities observed, we cannot exclude the possibility that Pyr-AMC hydrolysis was primarily due to some other set of peptidases, possibly including peptidases that were not directly assayed here.

Z-GlyGlyArgAMC is a nominally a substrate for trypsin, a broad-spectrum endopeptidase (i.e., peptidase that hydrolyzes proteins from the middle rather than the ends) (Obayashi and Suzuki, 2005). The bulky Z- group effectively prevents hydrolysis by aminopeptidases (Nakadai and Nasuno, 1977), and given the broad range of observed ratios of Z-GlyGlyArg-AMC:Leu-AMC hydrolysis rates, it is likely that that Z-GlyGlyArg-AMC and the single-amino acid substrates were hydrolyzed by distinct enzymes. Thus, the broad correlation in hydrolysis rates between those two substrates is probably due to community-level coexpression of trypsin-like enzymes and the common set of aminopeptidases that hydrolyzed Leu-AMC, Arg-AMC, and Gly-AMC.

It has long been recognized that a variety of peptidases are potentially present in aquatic environments (Christian and Karl, 1998). Early evidence suggested that assays of a single peptidase substrate provide a reasonable approximation of the total peptidolytic potential of a microbial community, because extracellular peptidases are frequently capable of acting on a wide range of peptides (Nomoto et al., 1960). The promiscuity of aquatic peptidases was used as justification for fluorogenic substrate-based enzyme assays when that technique was first adopted for aquatic samples (Hoppe, 1983), and indeed it appears that Leu-AMC hydrolysis is caused by a range of aminopeptidases in aquatic environments (Steen et al., 2015).

The results presented here place further constraint on the degree to which measurement of the hydrolysis rate of a single substrate is a useful measure of the total peptidolytic capacity of an ecosystem. LeuAP potential activity does correlate well with the activity of other aminopeptidases across a broad range of systems. For studies that examine systems in which activity varies by several orders of magnitude (for instance, studies that use LeuAP as a proxy for N demand across diverse environments, e.g., Sinsabaugh et al., 2009), LeuAP activity correlates well enough with endopeptidase activity that the additional information, time, and expense required to assay multiple peptidases are not justified given the novel information those measurements yield. In studies that have a narrower domain, for instance, time-series analyses in which LeuAP activity might vary within an order of magnitude (Allison et al., 2012; Mahmoudi et al., 2017) – this assumption is more dangerous, as changes in the ratio of endopeptidases : aminopeptidases could obscure patterns observed in just one peptidase. In this study, the ratio of trypsin-like potential activity to LeuAP potential activities ranged from 0.037 to 9.3. If the sum of trypsin-like activity and LeuAP activity places a lower bound on the total peptidolytic capacity of a system, then LeuAP could represent anywhere from 9.7 to 96% of total peptidolytic capacity, representing about an order of magnitude of potential error. Furthermore, since the endopeptidase:aminopeptidase activity appears to be a non-stochastic feature of ecosystems, this error would be systematic rather than random. Studies in which the range of LeuAP activities is narrower than an order of magnitude or so, assaying a broader set of peptidases, including endopeptidases and aminopeptidases, may yield a more complete picture of the potential for protein degradation.

Heterotrophic processes in aquatic systems are often described in chemically non-specific terms, such as "N acquisition" or "protein degradation." This is a useful way to distill important ecological patterns from the tremendously complex set of biochemical pathways that may be active in a system. It also flows from the limitations of organic geochemistry analytical technology: at present, it is relatively straightforward to measure the concentration of "hydrolysable amino acids" (i.e., proteinlike material) in aquatic systems, but very challenging to measure concentrations of specific proteins (Moore et al., 2012). Microorganisms sense and interact with the world at much finer chemical resolution. Those fine-scale interactions with the environment are reflected in the expression of specific extracellular enzymes. Measuring a broader set of extracellular enzymes can therefore yield insight into how microorganisms interact with their chemical environment. These results indicate that Leu-AMC hydrolysis is an acceptable proxy for total peptidolytic capacity of an environment only when the potential LeuAP activities vary over several orders of magnitude. When potential LeuAP activities span about an order of magnitude or less, variability in the aminopeptidase:endopeptidase ratio may cause total peptidolytic capacity to become decoupled from potential LeuAP activity. In such a data set, assays for multiple peptidases should be included to capture variability in total community peptidolytic potential.

### MEMBERS OF THE MALCOLM X SHABAZZ HIGH SCHOOL AQUATIC BIOGEOCHEMISTRY TEAM

Ivy Adu-Poku, Saskieya Anderson, Kiara Amore, Tequan Anderson, Elmy Antonio, Delvon Artis, Monai Barnes, Manusha Bearfield, Kim Boerrigter, Antwon Bowman, Zion Brummel, Abdul Bryant, Tyre Bush, Clervens Clerjuste, Latonya Coates, Ameena Corney, Jamie Crosby, Abdul Crowley, Michelle Culver, Leon Cummings, Zyion Eastmond, Genaro Falcon, Ciara Gilette, Lexus Gonzalez, Brianna Grant, Dequan Graves, Christian Hardy, Jamar Harris, Kevin Harrison, Deshawn Hart, Ismail Hinds, Latifa Hinds, Samantha Hunter, Imade Igiebor, Teniyjaa Jacobs, Malachi Jefferson, Shani Ketema, Zack Ketema, Ahmad Lancaster, Camille Laurel, Ferard Majette, Imani McCray, Francis Mensa, Elizabeth Mensah, Justin Mestre, Tai-Xian Mitchell, Davon Moody, Ivana Negron, Zaire O'Neill, Okoye Onyebuchi, Nasir Parham-Sanders, Madelyn Perez, Chris Pitt, Briana Racine, Acestra Robinson-Williams, Elvis Sanchez, Jabryant Vines, Jacelyn Valenciano, Nadira Wilkins, Anzunai Williams, Ma-Lisa Winborne, Henasia Wilson, and Yasmiah Wilson.

### AUTHOR CONTRIBUTIONS

LM and ABB helped design the sampling plan, performed experiments, analyzed the data, and helped write the manuscript. MXSHS-ABT (a group author) and NF performed experiments and analyzed the data. KB designed and performed experiments and analyzed the data. JR helped design the sampling plan. PM initiated the project, helped design the sampling plan, and performed experiments. AS initiated the project, helped design the sampling plan, performed experiments, analyzed the data, and helped write the manuscript.

#### FUNDING

Funding was provided by the National Science Foundation (OCE-1357242), the Geraldine R. Dodge Foundation, the Foundation for Newark's Future, and Newark Public Schools. Additionally, the Malcolm X Shabazz Aquatic Geochemistry

#### REFERENCES


Team was funded by Wilma Baldridge Schwades, Christopher Cerf, Jacob Etter, Dianna Thompson, Bob and Ruth Steen, Erika Amir-Lin, Suzy Schmidt, Ben Cady, Ben Bond-Lamberty, Franklin Parker, gofundme.com, Elnardo Webster, Olivier Humblet, Robert Tortoriello, Rosa Leon, Nancy Ferguson, Jenna Schmidt, Cameron Huges, Rachel Yoho, Rebecca Ghent, Adam Sobel, Vivian McNeill, Liz Kujawinski, Carol Arnosti, Bill Armbruster, Elliot and Natalia Ikheloa, Yvette Jordan, Morgan Wurch, Keith Britt, Nagissa Mahmoudi, Monika Kopacz, Danielle Lee, Jordan Bird, Dug Steen, Agnes Taylor, Steven Hauck, Clark Short, Susan Cheng, Mel Goldsipe, Katrina Carter, Laura Newsome, Jill Marshall, Helena Pound, Naupaka Zimmerman, Amparo Jiminez, Moses Olivia, Dorothea Boerrigter, Patricia MacQueen, Sarah Sheffield, Mohammad Moniruzzaman, Liz Agee, Lori Fenton, Max Czapanskiy, Orlando Campos, Zena Cardman, Paula Gagliardi, Camille Thomas, Richard Lewis, James Hodges, Cecilia Kruczek, and 28 anonymous donors.

### ACKNOWLEDGMENTS

We gratefully acknowledge logistical support from the staff of the Pocono Environmental Education Center, the faculty and administration of Malcolm X Shabazz High School. Karen Lloyd cofounded the Malcolm X Shabazz Aquatic Geochemistry Team and provided key operational support. Mr. Jamal Hall created amazing graphics.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00368/full#supplementary-material

FIGURE S1 | Density plot of all Vmax values measured in this study, showing the approximate log-normal distribution of activities.

FIGURE S2 | Distribution of ratios of trypsin-like potential activity to LeuAP potential activity, showing the approximate log-normal distribution of activities.

FIGURE S3 | pH dependence of LeuAP (marked LeuAMC) and GlyAP (marked GlyAMC) at Ardena Brook (ARD) and Belmar Inlet (BLMR).


sequencing methods for membrane proteins. Proc. Natl. Acad. Sci. U.S.A. 76, 227–231. doi: 10.1073/pnas.76.1.227


**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 Mullen, Malcolm X Shabazz High School Aquatic Biogeochemistry Team, Boerrigter, Ferriero, Rosalsky, Barrett, Murray and Steen. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner 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 Multi-season Investigation of Microbial Extracellular Enzyme Activities in Two Temperate Coastal North Carolina Rivers: Evidence of Spatial but Not Seasonal Patterns

Avery Bullock, Kai Ziervogel † , Sherif Ghobrial, Shannon Smith, Brent McKee and Carol Arnosti\*

Department of Marine Sciences, University of North Carolina, Chapel Hill, NC, United States

#### Edited by: Maria Montserrat Sala,

Consejo Superior de Investigaciones Científicas (CSIC), Spain

#### Reviewed by:

Michael J. Wilkins, The Ohio State University, United States Nina Welti, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia

> \*Correspondence: Carol Arnosti arnosti@email.unc.edu

#### † Present Address:

Kai Ziervogel, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, United States

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 12 August 2017 Accepted: 12 December 2017 Published: 22 December 2017

#### Citation:

Bullock A, Ziervogel K, Ghobrial S, Smith S, McKee B and Arnosti C (2017) A Multi-season Investigation of Microbial Extracellular Enzyme Activities in Two Temperate Coastal North Carolina Rivers: Evidence of Spatial but Not Seasonal Patterns. Front. Microbiol. 8:2589. doi: 10.3389/fmicb.2017.02589 Riverine systems are important sites for the production, transport, and transformation of organic matter. Much of the organic matter processing is carried out by heterotrophic microbial communities, whose activities may be spatially and temporally variable. In an effort to capture and evaluate some of this variability, we sampled four sites—two upstream and two downstream—at each of two North Carolina rivers (the Neuse River and the Tar-Pamlico River) ca. twelve times over a time period of 20 months from 2010 to 2012. At all of the sites and dates, we measured the activities of extracellular enzymes used to hydrolyze polysaccharides and peptides, and thus to initiate heterotrophic carbon processing. We additionally measured bacterial abundance, bacterial production, phosphatase activities, and dissolved organic carbon (DOC) concentrations. Concurrent collection of physical data (stream flow, temperature, salinity, dissolved oxygen) enabled us to explore possible connections between physiochemical parameters and microbial activities throughout this time period. The two rivers, both of which drain into Pamlico Sound, differed somewhat in microbial activities and characteristics: the Tar-Pamlico River showed higher β-glucosidase and phosphatase activities, and frequently had higher peptidase activities at the lower reaches, than the Neuse River. The lower reaches of the Neuse River, however, had much higher DOC concentrations than any site in the Tar River. Both rivers showed activities of a broad range of polysaccharide hydrolases through all stations and seasons, suggesting that the microbial communities are wellequipped to access enzymatically a broad range of substrates. Considerable temporal and spatial variability in microbial activities was evident, variability that was not closely related to factors such as temperature and season. However, Hurricane Irene's passage through North Carolina coincided with higher concentrations of DOC at the downstream sampling sites of both rivers. This DOC maximum persisted into the month following the hurricane, when it continued to stimulate bacterial protein production and phosphatase activity in the Neuse River, but not in the Tar-Pamlico River. Microbial community activities are related to a complex array of factors, whose interactions vary considerably with time and space.

Keywords: enzyme activities, bacterial production, DOC, Hurricane Irene, peptidase, glucosidase, Neuse River, Tar River

## INTRODUCTION

Riverine systems are important sources of organic carbon and nutrients for coastal and estuarine systems (Paerl et al., 1998; Stow et al., 2001; Lin et al., 2007). The availability of organic matter that can be processed within rivers is dependent on multiple physical, biological, and chemical factors, including the nature and extent of allochthonous input via runoff and groundwater, as well as autochthonous production within the system (Spencer et al., 2012). The quantity and quality of organic carbon and nutrients ultimately delivered to estuaries and coasts is partially the outcome of organic matter processing by heterotrophic microbial communities within the rivers. These communities facilitate the transformation and respiration of organic matter, and regeneration of nutrients (Blackburn et al., 1996). The extent to which organic matter is processed and transformed within a riverine system is thus dependent in part on the capabilities of heterotrophic microbial communities. The initial step of organic matter transformation is typically hydrolysis via extracellular enzymes, since high molecular weight organic matter is too large to be transported directly into microbial cells. The heterotrophic microbial community therefore must utilize extracellular enzymes to hydrolyze high molecular weight organic matter to sizes sufficiently small for uptake (see Arnosti et al., 2014 for a review). Though not the sole sources of extracellular enzymes, bacterioplankton are assumed to be the major producers of extracellular enzymes in aquatic systems (Hoppe et al., 2002; Vrba et al., 2004). Only a sub-fraction of a microbial community may produce specific extracellular enzymes, but the products of hydrolysis potentially might be accessed by a wider range of organisms. The activities of extracellular enzymes may therefore benefit a wider community, and measurement of extracellular enzyme activities can represent the potential to initiate organic matter remineralization at the community-level.

A number of biological, chemical, and physical factors can influence the production of extracellular enzymes (Allison and Vitousek, 2005; Artigas et al., 2009), while the degradation of organic matter can be dependent upon such factors as substrate type (McCallister et al., 2006), availability (Sinsabaugh and Moorhead, 1994), and community nutrient demands (Rier et al., 2011). In addition, studies have shown that organic matter concentration and type change seasonally in freshwater watersheds (Singh et al., 2013). These seasonal changes are due largely to sorption of DOM on mineral soil surfaces and/or microbial breakdown of leaf litter. Singh et al. (2013) found that stormflow in summer contained DOM that was more humic in character than in spring and winter, as a result of more influence from the watershed during higher discharge periods. Therefore, changes in organic matter supply, environmental conditions, or microbial community composition across spatiotemporal scales may be reflected in the enzymatic profiles and activities of a microbial community.

Since organic matter type, concentration, and microbial community composition likely influence enzymatic activities, spatial and temporal variability of enzymatic activities could vary widely. Freshwater systems such as creeks and streams have been shown to be a medley of different microbial community activities, responding to temporally-changing environmental gradients (Frossard et al., 2012). Capturing these varying dynamics is challenging; prior studies typically have focused on sampling a range of stations over a limited timescale or on sampling a few sites over a longer period of time (e.g., Artigas et al., 2009; Millar et al., 2015).

We investigated spatial and as well as seasonal variations in microbial activities and organic matter remineralization in two distinct river systems in central and eastern North Carolina: the Neuse River and the Tar-Pamlico River. These rivers were each sampled at four different sites (two upstream sites, two downstream sites) ∼12 times over a 20-month period. We carried out this extended sampling program in an effort to capture temporal variability over a range of sites. Part of this temporal variability was caused by Hurricane Irene's passage over eastern North Carolina (August 2011), an event that provided the opportunity to measure changes in microbial activities in the rivers in response to a large-scale influx of precipitation and laterally-flowing water. In order to investigate a greater range of heterotrophic capabilities, we measured the activities of enzymes capable of hydrolyzing small substrate proxies typically used to assess glucosidase and leucine aminopeptidase activities, and also used a suite of polysaccharide substrates that can measure the endo-acting activities of enzymes that cleave specific polysaccharides mid-chain. We sought to investigate the manner in which changing biological, physical, and chemical parameters may affect organic carbon cycling, as measured via activities of extracellular enzymes.

#### MATERIALS AND METHODS

#### Study Sites

The Neuse and Tar-Pamlico Rivers, extending from central to eastern North Carolina, feed into Pamlico Sound (Paerl et al., 2010; **Figure 1**). The Albemarle-Pamlico Sound estuary system is the second largest estuary system in the United States (Paerl et al., 2010), and provides significant nursery area for commerciallyimportant fisheries on the U.S. Atlantic coast (Burkholder et al., 2006). Although the Albemarle-Pamlico Sound, as well as other estuary systems, serves as an important link between terrestrial/riverine systems and the marine environment (Paerl et al., 1998), the dynamics of organic matter processing occurring in the Neuse and Tar-Pamlico Rivers are not well-studied.

The two river systems have contrasting watersheds (Burgess, 2013a,b). The Neuse River is heavily urbanized upstream, with a population of over 1.5 million residing within its watershed (Burgess, 2013a), and is also subject to heavy industrialized agricultural use. The Tar-Pamlico River is a smaller, less developed river both in terms of agricultural and urban development, but it is the largest tributary of the Pamlico River Estuary (Overton et al., 2012; **Table 1**). The mean discharge to the Albemarle-Pamlico estuarine system is 190 m<sup>3</sup> s −1 for the Neuse River and 148 m<sup>3</sup> s −1 for the Tar-Pamlico River (Lin et al., 2007). Downstream salinities also differ between rivers: the Tar-Pamlico endmember salinities are close to zero (compared to the

TABLE 1 | Characteristics of the Neuse and Tar-Pamlico Rivers (O'Driscoll et al., 2010).


Neuse) because the Tar-Pamlico estuary is relatively narrow and the influence of ocean waters due to wind mixing is much more limited relative to that in the Neuse estuary, which has a more open morphology.

Surface water samples were collected from four stations in each river: N1, N2, N6, and N7 in the Neuse River, and T1, T2, T5, and T6 in the Tar River (**Figure 1**). These stations were chosen to include the different land-use impacts of the rivers (urbanization of upstream stations; agricultural effects on downstream stations), as well as the transition from a freshwater to an estuarine ecosystem. Collection dates (**Supplementary Table 1**) varied slightly among sites, due to the complex logistics of sampling eight different locations spread across a considerable distance. Due to the timing of our study, we also partially captured the influence of a major storm event (Hurricane Irene) on the lower stations of the Tar-Pamlico and Neuse Rivers in August 2011.

### Sample Collection

Surface water samples were collected over a 20-month period (November 2010 to June 2012) from each of the four stations in the Neuse and Tar-Pamlico Rivers (**Figure 1**). Samples were collected in 33 L Nalgene carboys and stored at in-situ temperatures during transportation back to UNC-Chapel Hill. Measurements of enzyme activities and bacterial productivity were initiated upon return to the lab, and samples for cell counts were preserved (see below). Dissolved oxygen, temperature, salinity, and pH data were collected on site using a YSI (YSI Inc. 556MPS) (**Supplementary Table 1**). River discharge and gage height for most stations was obtained from the USGS's monitoring website (http://waterdata.usgs.gov/nc/nwis/rt). Note that the USGS does not have a monitoring site at T6, and USGS data collection was ended at N7 in 2009, so for the downstream stations we used gage height (**Supplementary Figure 1**) and discharge data from T4 and T5 for the Tar River, and N5 and N6 for the Neuse River (**Figure 1**; **Supplementary Table 4**). Hurricane Irene struck eastern North Carolina on 27–28 August 2011. Post-hurricane sampling of the Tar-Pamlico River occurred at Stns. 5 and 6 on August 29, 2011. Somewhat later posthurricane sampling was carried out at Stns. T5 and T6 and Stn. N7 on Sept. 14th (Stns. T1 and T2 and Stns. N1 and N2 were sampled on Sept. 12th); all sampling dates are shown in **Supplementary Table 1**.

#### Extracellular Enzyme Activities

Activities of exo-acting (terminal-unit cleaving) as well as endoacting (mid-chain cleaving) enzymes were measured using two different methods. Low molecular weight substrate proxies [4-methylumbelliferone- (MUF-) and 4-methylcoumarinyl-7 amide- (MCA-) labeled substrates] were used to measure αand β-glucosidase, leucine aminopeptidase, and phosphatase activities, after the method of Hoppe (1983; Hoppe et al., 1988). Triplicate water samples from each station were amended with substrate proxies to a final concentration of 400µM (this concentration was chosen at the start of the project, from a saturation curve made to determine the appropriate saturation concentration of each substrate in the river water). Killed controls consisted of autoclaved water to which substrate was added. Samples were incubated for a period of 3–5 h at in situ or near in-situ temperature; an initial time-zero measurement was taken at the start of this period, and two to three subsequent time points were measured during this period. For each measurement, a 1-ml aliquot was taken from the incubating sample and combined with 1 ml of 20 mM borate buffer, and fluorescence was measured using single-cell fluorometers (Turner Biosystem TBS-380 or a Promega Quantifluor-ST). A dilution curve was made with each fluorophore in autoclaved river water to determine a fluorescence-hydrolysis rate conversion factor for each river. Hydrolysis rates were then calculated using the conversion factors and fluorescence measurements.

The activities of extracellular enzymes responsible for endo-acting (mid-chain cleaving) hydrolysis of a specific set of polysaccharides were measured using six distinct fluorescently labeled (FLA) polysaccharides (Arnosti, 1996, 2003). Arabinogalactan, chondroitin, fucoidan, laminarin, pullulan, and xylan (all obtained from Sigma-Aldrich USA) were labeled with fluoresceinamine as described in Arnosti (2003). These polysaccharides were selected because they are derived from a range of terrestrial (xylan, arabinogalactan) and marine (laminarin, xylan, fucoidan, pullulan, chondroitin) sources, and/or enzymes hydrolyzing these polysaccharides have been identified in marine bacteria and in marine bacterial genomes (for details, see e.g., Bold, 1985; Arnosti, 2000; Alderkamp et al., 2007; Wegner et al., 2013). In addition, the activities of enzymes hydrolyzing all of these polysaccharides have been measured in marine (Arnosti et al., 2011) as well as freshwater systems (Ziervogel et al., 2014). Because of the time and resources required for measurements with FLA-polysaccharides, polysaccharide hydrolysis rates were measured in duplicate, only at the upriver-most and downriver-most station in each river. At these stations (Stns. T1, T6; N1, and N7), duplicate live water samples, as well as an autoclaved control water sample for each station, were separately amended with one of each of the six substrates to a final concentration of 3.5 nM monosaccharide equivalent. A time-zero measurement was immediately taken, and the samples were then incubated in the dark at near in-situ temperature, with subsamples withdrawn periodically. After processing the samples, we found that, with very few exceptions, all polysaccharides were hydrolyzed at 3 days, the first time-point after the zero-time sample, so the data reported are all from this time point. Inconsistencies in sampling timepoints after 3 days in any case preclude use of later timepoints across the dataset.

Samples for measurement of polysaccharide hydrolase activities were collected by filtering 1–3 ml of sample water through a 0.2µm cellulose acetate-membrane + GF-prefilter syringe filter (Sartorius Stedim Biotech, Germany), and freezing samples at −20◦C until analysis. Hydrolysis was measured via changes in the molecular weight distribution of the FLAlabeled polysaccharide using gel permeation chromatography, as described in detail in Arnosti (2003). Several samples were lost prior to analysis from the Stn. T1 sample set: (date/substrate): 01/11 (fucoidan), 04/11 (pullulan), 06/11 (fucoidan), and 06/12 (xylan). At Stn. T6, missing samples were as follows: 09/11 (fucoidan), 11/11 (fucoidan), and 06/12 (laminarin).

#### Bacterial Cell Counts and Production

Aliquots of water were fixed for bacterial cell counts, following Porter and Feig (1980). Staining was carried out with 4′ , 6 diamidino-2-phenylindole (Sigma-Aldrich USA), and slides were counted under an epifluorescence microscope (Olympus U-RFL, Olympus USA) using MetaMorph Microscopy software (Molecular Devices USA). 10 fields of view were counted per slide, with duplicate slides made for each river station.

Bacterial production was measured using <sup>3</sup>H-leucine incorporation (Kirchman, 2001). These measurements were only initiated in January 2011, so no data are available for samples collected during November/December 2010. Water from the upstream- and downstream-most stations (the same stations used to measure polysaccharide hydrolysis) in each river, plus autoclaved control water, was amended with <sup>3</sup>H-leucine to a final concentration of 20 nM. Samples were incubated for 1–2 h; following this incubation period, reactions were terminated using 100% trichloroacetic acid (TCA). Samples were then concentrated and washed with 80% ethanol before drying over night. Samples were then amended with scintillation liquid and allowed to sit for a 2-day period before analysis in a scintillation counter (Perkin Elmer TriCarb 3110 TR).

#### Dissolved Organic Carbon

Water samples from each station were filtered through 0.2µm cellulose acetate-membrane + GF-prefilter syringe filter (Sartorius Stedim Biotech, Germany) into pre-combusted glass scintillation vials and frozen at −20◦C until further analysis. Dissolved organic carbon concentrations from these samples was measured via high temperature catalytic oxidation and non-dispersive infrared detection on a Shimazdu TOC-L series instrument (Shimadzu Corp. Kyoto). Samples were acidified to a pH < 2 and sparged with commercially obtained CO<sup>2</sup> free, zero-grade air for 10 min for inorganic carbon removal. Standards were generated from dilution of commercially prepared potassium hydrogen phthalate [KHP] (La-Mar-Ka Inc., Baton Rouge, LA) with 18.2 M ultrapure water.

#### Statistical Analyses

Environmental data and microbial activity measurements described above were analyzed to look for correlations using the corrplot package in R (R Core Team, 2014), run in R version 3.3.3. The same package was used to look for correlations among activities of individual polysaccharide hydrolases. T-test analyses of potential effects of season, station, and the effects of Hurricane Irene on microbial activity measurements were also analyzed using R.

### RESULTS

### Environmental and Hydrological Characteristics

Environmental data including dissolved oxygen (DO), pH, and salinity (**Supplementary Table 1**), as well as river discharge (**Figure 2**), were used establish a picture of the seasonal environmental characteristics and dynamics of both rivers, and to investigate potential connections between community activity and river characteristics. River discharge indicated that the upstream and downstream stations were hydrologically decoupled (**Figure 2**). Discharge volume for Stns. T1 and T2 tracked together, and were distinctly separated from discharge volume at Stns. T4 and T5. Discharge data from the Neuse River likewise showed decoupling of upstream (Stns. N1 and N2) stations from the downstream-most station for which discharge data are available (Stn. N6). These patterns were supported by other chemical and physical data (**Supplementary Table 1**): in the Neuse River, salinity remained near zero upstream (Stns. N1, N2); the downstream stations exhibited greater fluctuations in salinity, varying between freshwater and estuarine conditions. Salinity of the Tar-Pamlico River, however, remained close to zero, even at the station farthest downriver (Stn. T6; **Supplementary Table 1**). Temperature ranges changed seasonally in both rivers, with the annual variations in the Neuse River between 4 and 30◦C, and near 0 to 29◦C in the Tar-Pamlico River. At each sampling point, temperatures were broadly comparable among stations, although downriver stations were frequently slightly warmer than upriver stations, and the temperature difference between Stns. T1 and T6 was typically a few degrees greater than between Stns. N1 and N7 (**Supplementary Table 1**). Dissolved oxygen (DO) followed an inverse relationship with temperature in both rivers. The range of DO for both downstream stations was greater than the DO ranges upstream.

DOC concentrations showed broad patterns across spatial and temporal scales. In the Neuse River, DOC was

consistently highest (near or above 1,200 µmol C L −1 ) at the downstream-most station, Stn. N7 (**Figure 3**; **Table 2**), while DOC concentrations at Stns. N1, N2, and N6 ranged from ∼400 to 800 µmol C L−<sup>1</sup> . Stn. N2 exhibited the least temporal variation in concentration. No seasonal trends were evident, but the highest DOC measured at Stn. N7 (1,861 µmol C L−<sup>1</sup> ) was in September 2011, following Hurricane Irene in August 2011. In the Tar-Pamlico River, DOC concentrations at most stations and seasons exhibited wider temporal variability than in the Neuse River, ranging from ca. 300 to 1,000 µmol C L−<sup>1</sup> . There was no distinct seasonal trend (**Figure 3**; **Table 2**), but DOC concentrations were higher downstream than upstream (**Table 2**), and the highest DOC concentrations (exceeding 1,700 µmol C L−<sup>1</sup> ) were also recorded at the downriver stations, Stns. T5 and T6, in September 2011, following Hurricane Irene (**Figure 3**).

### Microbial Cell Counts and Leucine Incorporation

Seasonally, bacterial numbers were highest in the late winter and early spring (Feb.–April) of 2011 (t-test; p < 0.01). In both rivers, months sampled in the winter and late spring of 2012 had

lower bacterial abundance than their 2011 counterparts. Bacterial abundance (**Supplementary Table 2**) varied by a factor of 10 over the time course of the study. Bacterial protein production (**Figure 4**) showed slight increases during the spring through fall months, with minima occurring during the winter months (Jan/Feb) for both rivers (**Table 2**). When normalized on a percell basis (**Supplementary Table 2**), the summer and late fall months showed highest bacterial production. For most months, bacterial protein production rates normalized to cell abundance were higher downstream (Stns. T6, N7) compared to upstream (Stns. T1, N1) in both rivers (**Supplementary Table 2**). The highest bacterial protein production measured during the study (per cell, as well as on a volume basis) was measured in the Neuse River at Stn. N7 in Sept. 2011, after the passage of Hurricane Irene (**Figure 4**).

### Activities of Glucosidase, Peptidase, and Phosphatase Enzymes

Leucine amino peptidase (Leu-MCA), glucosidase (α- and β-glu), and phosphatase activities were measured immediately upon return of the samples to the lab to assess microbial heterotrophic activities across locations and seasons. Leu-MCA hydrolysis rates were highest at the downstream-most station for the Tar-Pamlico River (**Figure 5**; **Table 2**). Averaged across all timepoints, Leu-MCA hydrolysis rates were 182, 147, 156 nmol L−<sup>1</sup> h −1 for Stns. T1, T2, and T5, respectively, and approximately double-−324 nmol L−<sup>1</sup> h <sup>−</sup>1–at Stn. T6. For the Neuse River, averaged across all timepoints, Leu-MCA hydrolysis was 137, 144, and 93 nmol L <sup>−</sup><sup>1</sup> h −1 at Stns. N1, N2, and N6, respectively, and considerably higher (239 nmol L−<sup>1</sup> h −1 ) for Stn. N7. Although in both cases minimum rates occurred during winter months (**Table 2**), there were no overall seasonal trends for either river.

Glucosidase hydrolysis rates were generally higher in the Tar-Pamlico than the Neuse River, although this difference was statistically significant only for β-glu activities (**Table 2**). Glucosidase hydrolysis rates averaged close to 30 nmol L−<sup>1</sup> h −1 in the Tar-Pamlico River and ∼10 nmol L−<sup>1</sup> h −1 in the Neuse River. In both rivers, β-glu activities were generally a factor of 2–3 higher than α-glu activities (**Figures 6**, **7**; **Supplementary Table 5**); β-glu activities also showed a greater dynamic range (difference between lowest and highest rates). In both rivers, glucosidase activities were lowest in January and TABLE 2 | P-values (T-tests) to determine statistical significance (as shown by P < 0.05; bold font) of microbial activities, cell counts, and DOC concentration by location and time.


β-glu, β-glucosidase activities; α- glu, α-glucosidase activities; Leu, leucine aminopeptidase activities; phosph, phosphatase activities; sum FLA, summed polysaccharide hydrolase activities; bact prod, bacterial protein production; DOC, dissolved organic carbon concentration. Hurricane Irene-affected times and locations were defined as Stations T5/T6 and N6/N7 in Sept 2011 (and for T5/T6, also in Aug 2011) compared to all other months at the same stations.

February (**Table 2**), but otherwise varied considerably by month and station.

Phosphatase activities typically were higher in the Tar-Pamlico than in the Neuse River (**Figure 8**; **Table 2**). In both rivers, phosphatase activities were typically low in January and February, and were considerably higher during other months of the year (**Table 2**). Activities did not differ systematically between upstream and downstream locations (**Table 2**). At Stn. N7, however, a notably high maximum (ca. 500 nmol L−<sup>1</sup> h −1 ) was measured in Sept. 2011, the month after Hurricane Irene, when bacterial productivity and the DOC concentration at this station also reached maxima (**Figures 3**, **4**).

symbols (Stns. T5 and T6 only) show samples collected in August 2011, shortly after Hurricane Irene crossed through eastern North Carolina.

## Activities of Polysaccharide-Hydrolyzing Enzymes

In both rivers, a broad spectrum of polysaccharide hydrolase activities was measured, with all six polysaccharides hydrolyzed at many sampling dates and stations (**Supplementary Table 3**). Summed polysaccharide hydrolysis rates were frequently lowest in the winter months (**Table 2**), but otherwise varied considerably (**Figure 9**). The relative contribution of each polysaccharide hydrolase activity to summed activities was quite dissimilar, however. Chondroitin and xylan hydrolysis together averaged 58–65% of the total contributions to the summed polysaccharide hydrolysis rates across all seasons and stations in both rivers, irrespective of whether summed activities were high or low (**Supplementary Table 3**). For most stations, hydrolysis rates generally decreased in the order xylan, chondroitin >> laminarin > arabinogalactan with smaller contributions from fucoidan and pullulan (**Supplementary Table 3**). The annual range of summed hydrolysis rates was similar among all stations, from 3 to 65 nmol monomer L−<sup>1</sup> h −1 at Stn. N1, 6–87 nmol monomer L−<sup>1</sup> h −1 at Stn. N7, 3–81 nmol monomer L−<sup>1</sup> h −1 at Stn. T1, and 4–71 nmol monomer L−<sup>1</sup> h −1 at Stn. T6 (**Figure 9**).

### Correlations among Microbial Activities and Environmental Parameters

Correlation analysis of environmental parameters and microbial activities (**Figure 10**; p-values in **Supplementary Table 6**) showed some expected as well as unexpected correlations. Unsurprisingly, dissolved oxygen (DO) was strongly inversely correlated with temperature, and discharge, conductivity, and gage height also showed moderate correlation. Summed polysaccharide hydrolase activities (FLA; **Figure 10**) were positively correlated with temperature, and thus also inversely correlated with DO, although other enzyme activities did not show a notable correlation with temperature (or DO). Correlations among the other enzyme activities varied: βand α-glu were strongly correlated with each other, and were more weakly correlated with Leu-MCA and phosphatase.

symbols (Stns. T5 and T6 only) show samples collected in August 2011, shortly after Hurricane Irene crossed through eastern North Carolina. y-axis is a logarithmic

Cell counts were inversely correlated with DOC, but leucine incorporation (bacterial protein production) showed only a strong inverse correlation with DO. Correlation analysis among the individual polysaccharide hydrolase activities (**Figure 11**; p-values in **Supplementary Table 7**) showed that summed activities were most strongly correlated with xylan and chondroitin hydrolysis, followed by arabinogalactan and laminarin hydrolysis.

#### Effects of Hurricane Irene

scale.

Post-Hurricane Irene sampling (September 2011, plus August 2011 for Stns. T5 and T6) showed that phosphatase activities and DOC concentrations were considerably elevated compared to other sampling dates at the downriver stations (T5, T6, N6, N7; **Table 2**; **Figures 3**, **8**). Although bacterial protein productivity was greatly elevated at Stn. N7 in September 2011 (**Figure 4**), the statistical significance of this measurement could not be calculated due to the small number of observations.

## DISCUSSION

Microbial processing of organic matter in riverine systems can be influenced by a variety of physical, chemical, and biological factors that vary across a range of spatial and temporal scales (Singh et al., 2013). Most previous studies of microbially-driven carbon cycling in rivers have focused on either sampling a range of sites across a limited time period, or have investigated fewer sites over an annual cycle (e.g., Artigas et al., 2009; Tiquia, 2011; Frossard et al., 2012; Millar et al., 2015). In an effort to investigate some of the complexities of these interactions in rivers, we sampled both the Neuse and Tar-Pamlico Rivers across multiple seasons and sites. The range of Leu-MCA, β-glu, and phosphatase activities measured across sites and seasons in the Tar-Pamlico and Neuse Rivers proved to be similar to or slightly higher than rates measured across a broad range of riverine sites sampled at one season (e.g., Williams et al., 2012; Millar et al., 2015), as well as seasonal studies at fewer locations (Wilczek et al., 2005), suggesting that the Tar-Pamlico and Neuse Rivers

scale.

are not atypical in their enzymatic activities. Cell counts were also generally similar to those reported from other riverine locations (Wilczek et al., 2005; Williams et al., 2012; Millar et al., 2015).

Two broad-scale spatial patterns in microbial enzyme activities emerged over the course of the study: higher peptidase activities at T6 compared to the other stations in the Tar-Pamlico River, and higher β-glucosidase and phosphatase activities in the Tar-Pamlico River compared to the Neuse River (**Table 2**; **Figures 5**, **6**, **8**). These spatial patterns in microbial enzyme activities contrast in particular with a lack of spatial patterns for microbial cell counts, and bacterial productivity that showed a different spatial pattern: upriver/downriver contrasts, and higher values at Stn. N7 than N1 (**Table 2**). Moreover, bacterial protein production correlated (inversely) only with dissolved oxygen, but not with any of the other activity measurements (**Figure 10**).

Microbial sources likely account for most of the enzymes active in the water column, but individual microbes can differ considerably in terms of activity, as exemplified by the differences in cell-count normalized bacterial production (**Supplementary Table 2**), which also demonstrated no distinct spatial patterns. Moreover, the capabilities of distinct members of microbial communities to carry out specific enzymatic function differs substantially, even among closely-related microbes (Xing et al., 2014). Since the measured enzyme activities are also an outcome of the kinetic characteristics of enzymes and their active lifetimes in the water column, as well as the quantity of enzymes produced, a lack of correlation between microbial cell numbers or bacterial productivity and enzyme activities is not entirely surprising. A lack of correlation between either cell counts or bacterial productivity and glucosidase and Leu-MCA activities has also been observed in other freshwater environments (e.g., Sieczko et al., 2015). The observation that summed polysaccharide hydrolase activities do not demonstrate the same spatial patterns seen for β-glucosidase activities—there is no difference between the activities measured in the Tar-Pamlico and in the Neuse River (**Table 2**)—is likely due to the fact that the overall ability to produce specific extracellular enzymes is non-uniformly distributed among members of microbial

communities (Zimmerman et al., 2013). Furthermore, the longer incubation times for the polysaccharide hydrolase measurements (3 days, compared to hours for the β-glucosidase activities) allows time for growth and induction responses to polysaccharide addition, which may have masked any initial differences among sites.

Given prior observations of a limited spectrum of polysaccharide-hydrolyzing enzyme activities in aquatic systems (e.g., Ziervogel and Arnosti, 2009; Arnosti et al., 2011; Ziervogel et al., 2014), the hydrolysis of all six polysaccharide substrates at every station in the Neuse and Tar-Pamlico Rivers at timepoints throughout the year was remarkable (**Supplementary Table 3**). This breadth of hydrolytic capabilities has seldom been observed in other locations; even nutrient addition has not led to hydrolysis of some polysaccharides in some locations (Steen and Arnosti, 2014). Typically, over time-courses of incubations lasting well over 3 days, only a subset of polysaccharides was hydrolyzed (e.g., Steen et al., 2008; Arnosti et al., 2011; Ziervogel et al., 2014). Moreover, this broad range of hydrolytic capabilities in the Neuse and Tar-Pamlico Rivers was observed at timepoints when summed hydrolysis rates were low, as well as at times when summed hydrolysis rates were high, as at Stn. N1 in February and June 2012, when summed hydrolysis rates were 6 and 65 nmol monomer L−<sup>1</sup> h −1 , respectively (**Figure 9**; **Supplementary Table 3**).

The broad hydrolytic capabilities of microbial communities in the Neuse and Tar-Pamlico Rivers may be due to the extensive and diverse watersheds of both rivers, as well as the occurrence of seasonal flooding events, leading to substantial terrestrial input into the systems, which provides the microbial community with a greater quantity and diversity of organic matter sources. In particular, the lower reaches of both rivers are subject to frequent overbank flooding because there are no large dams to control flooding and because these low-elevation coastal plain rivers have low banks that help facilitate frequent flooding (Peng et al., 2004; Reed et al., 2008). Although previous studies in other riverine systems have suggested that freshwater and estuarine organic matter of autochthonous, as opposed to allochthonous, origin is of greater importance to the microbial community for uptake

hydrolase data are shown in Supplementary Table 3.

(McCallister et al., 2006), significant input of terrestrial organic matter may well influence, and possibly increase, hydrolytic capabilities. Such a broad range of DOM sources may also account for the observation that a site within Pamlico Sound also showed hydrolysis of all six polysaccharide substrates, in contrast to a nearby site on the continental shelf, where only four of the substrates were hydrolyzed (D'Ambrosio et al., 2014). Throughout all seasons, the comparatively high contributions of xylan hydrolysis to summed hydrolysis rates may be an additional indication of the importance of terrestrial sources of organic matter to both rivers, since xylan is a major constituent of land plants (Ebringerova and Heinze, 2000), as well as some algae. High xylan hydrolysis rates have also been measured in the Chesapeake Bay (Steen et al., 2008) and Delaware River (Ziervogel and Arnosti, 2009). Comparably high contributions of both chondroitin and xylan to summed hydrolysis rates have to date been observed in the Delaware River (Ziervogel and Arnosti, 2009), and in the Gulf of Mexico, at sites that presumably also have the potential to be influenced by terrigenous input from the Mississippi River (Arnosti et al., 2011; Steen et al., 2012).

Spatial differences in DOC concentrations were evident most notably elevated downstream DOC concentrations, particularly at Stn. N7—but this pattern is not reflected in most of the enzyme activities measured in the two rivers (**Table 2**). The elevated DOC concentration particularly at Stn. N7 may in part reflect lateral input at the downstream locations, as well as the hydrologic disconnect shown by discharge and gage height (**Figure 2**; **Supplementary Figure 1**) of upstream and downstream stations. DOC contributed via different hydrologic flow paths can differ considerably in source as well as composition (Singh et al., 2013). Moreover, lateral input of DOC may include comparatively more microbially recalcitrant organic matter that survives photochemical and microbiological processing, and does not enhance bacterial activity. A lack of clear relationships between DOC and enzyme activities was also reported for the large tributaries of the lower Mississippi


FIGURE 10 | Correlation plot for physical/chemical and microbial activity data. Gage, gage height; disch, discharge; temp, temperature; cond, conductivity; DO, dissolved oxygen; pH, pH; DOC, dissolved organic carbon; cells, cell counts; phos, phosphatase activities; incorp, bacterial protein production (leucine incorporation); leu, leucine-aminopeptidase (Leu-MCA); aglu, a-glucosidase; bglu, b-glucosidase; FLA, summed polysaccharide hydrolase activities. Colors show intensity of positive (blue) and negative (red) correlations; numbers show corresponding correlation coefficients.

River, where site-to-site differences for a single river were as large as between-river variations in enzyme activities (Millar et al., 2015). A similar lack of relationship between the origin and optical properties of DOM and glucosidase or Leu-MCA activities was reported for the Danube floodplain (Sieczko et al., 2015). The Danube floodplain sites also showed little systematic variation in bacterial abundance or productivity, despite differences in glucosidase and peptidase activities (Sieczko et al., 2015).

The lack of broad seasonal trends in microbial activities and abundance was something of a surprise. Although glucosidase, peptidase, and phosphatase activities were lowest in Jan/Feb (**Table 2**), when temperatures were also lowest (**Supplementary Table 1**), low activities were at times also measured in other months—for example, in June—when seasonal temperatures were near or at their maximum. Across the entire study period, temperature correlated with summed polysaccharide hydrolase activities (**Table 2**), but summed activities were sometimes low at times when water temperatures were high (e.g., June 2011 at Stn. N7; **Figure 9**) Other studies have shown seasonality to be an important factor associated with changes in microbial community activities (Artigas et al., 2009), with temperature as an important controlling variable (Wilczek et al., 2005). A high-resolution investigation of enzyme activities at a single site in the coastal Pacific, however, demonstrated that substantial variations in enzyme activities occurred on timescales shorter than 1 month, and that seasonal patterns were not clearly evident (Allison et al., 2012).

An investigation of microbial community composition along the salinity gradient of the Columbia River and its estuary, extending into the coastal ocean, also found that strong spatial patterns overwhelmed seasonal patterns, which were more

evident within individual groups of bacteria (Fortunato et al., 2012). Metabolic potential, as represented by metagenomes, varied comparatively little along this same gradient, but metatranscriptomic data showed considerable variability that was unrelated to season or salinity (Fortunato and Crump, 2015), suggesting that the communities were reacting to localized environmental conditions. In the case of the Tar-Pamlico and Neuse Rivers, factors that are unrelated to season—such as land use, and abundance of natural land cover (**Table 1**) may correlate more strongly with enzyme activities (Williams et al., 2012). The interplay between factors that do and do not have consistent seasonal trends may thus help drive enzymatic activities, and obscure seasonal correlations. However, evidence suggests that microbial communities in freshwater, estuarine, and marine systems are able to respond rapidly to increased organic matter inputs (Williams and Jochem, 2006; Allison et al., 2012). Such rapid responses may be the reason that consistent, longterm seasonality is not evident in the Neuse and Tar-Pamlico Rivers, as microbial communities may respond to factors that change on timescales different than the length of time between sampling dates.

The passage of Hurricane Irene across the eastern half of North Carolina in August 2011, however, is an example of an event that left a signature discernable in the lower reaches of the Tar-Pamlico and Neuse Rivers even two-plus weeks post-event. Phosphatase activities, DOC concentrations, and cell counts showed significant differences in September 2011 (and August, for Stns. T5 and T6) compared to other sampling months at these sites (**Table 2**). (Note that the August sampling was a day after Hurricane Irene passed through North Carolina; the September samplings were ca. 2 weeks posthurricane). DOC concentrations at Stns. T5, T6, and N7 all were far higher in September 2011 than in any other month at the same location (**Figure 3**), despite the probability that the concentrations measured 2 weeks post-event were lower than the maximum concentrations. Maximal DOC input into the Neuse estuary lagged maximum discharge by ∼1 week (Brown et al., 2014), presumably due to lateral inputs into the upper reaches of the Neuse River in response to extensive flooding. Flooding and elevated discharge were likely responsible for changes in microbial community composition in the Tar-Pamlico River, where microbial community composition downstream (Stn. T6) shifted considerably post-hurricane (Balmonte et al., 2016). Prior to the hurricane, the community composition of Stns. T1 and T6 were notably dissimilar. Immediately post-hurricane (August 2011) as well as 2 weeks later (Sept. 2011), there was evidence of coupling between upstream and downstream stations, as well as post-hurricane microbial input from terrestrial sources. These distinct microbial signatures were less evident by November 2011 (Balmonte et al., 2016).

Although DOC concentrations measured in Sept. 2011 were similar in the Neuse and Tar-Pamlico Rivers (**Figure 3**), the composition of this DOC was likely different, given the difference in watersheds (**Figure 1**; **Table 1**) and the notable differences in responses of the microbial communities in the two rivers to this DOC. Bacterial production on a cell-specific basis reached a maximum in Sept. 2011 at Stn. N7 more than an order of magnitude higher than otherwise measured at this station, and more than four times greater than at Stn. T6 at the same time (**Supplementary Table 2**); bacterial production at Stn. N7 was also maximal at this station on volume-specific basis (**Figure 4**). Phosphatase activities were also greatly elevated at Stn. N7, but not at Stn. T6 (**Figure 8**). In the Tar-Pamlico River, by contrast, bacterial production, glucosidase, and Leu-MCA activities at Stn. T6 were not notably elevated even during the August 2011 sampling (**Figures 5**–**7**), the day after the passage of Hurricane Irene. Although no data on the chemical characteristics of Hurricane-Irene associated DOC are available from the Tar-Pamlico and Neuse Rivers, Hurricane Irene-associated water collected within a Maryland watershed showed distinct spectroscopic characteristics compared to water collected at other times, likely due to differences in sources and flow-paths (Singh et al., 2013). The differences in land use, drainage, and flow paths thus may have led to considerable compositional differences in the DOC added to the Neuse and Tar-Pamlico Rivers as a consequence of Hurricane Irene. Together, these data suggest that there was a microbial response to the DOC added to the Neuse River, but not the Tar-Pamlico River, post Hurricane Irene, but this response did not involve the glucosidase, Leu-MCA, or polysaccharide hydrolase enzymes whose activities we measured, or (alternatively), any enzymatic response in the Neuse River was shorter-lived than the elevation of DOC concentration, two-plus weeks postevent.

Complex trends of organic carbon remineralization characterize microbial activities in the Tar-Pamlico and Neuse River systems. Broad-scale spatial patterns—in particular, higher β-glucosidase and phosphatase activities in the Tar-Pamlico compared to the Neuse River, as well as higher downstream Leu-MCA activities in the Tar-Pamlico River—are evident in this study, but no single factor can be pinpoint as the most influential in shaping community activities; even large-scale events such as a hurricane's passage elicited different responses in the two rivers. Future studies of similar spatiotemporal scales, ideally including focused investigation of DOC characteristics and flow paths, will be necessary for clearer understanding of the factors that drive microbial community activities and organic matter remineralization across aquatic gradients.

#### REFERENCES

Alderkamp, A.-C., van Rijssel, M., and Bolhuis, H. (2007). Characterization of marine bacteria and the activity of their enzyme systems involved in degradation of the algal storage glucan laminarin. FEMS Microbiol. Ecol. 59, 108–117. doi: 10.1111/j.1574-6941.2006.00219.x

#### AUTHOR CONTRIBUTIONS

CA and BM designed the study. AB, KZ, SG, and SS collected the samples, carried out the incubations, and collected the data. AB, KZ, SG, SS, and CA analyzed the data. AB and CA wrote the manuscript, with input from all co-authors. We are very grateful to the two reviewers, whose thoughtful comments considerably improved the manuscript.

#### FUNDING

This project was initiated and carried out with funding from the Eddie and Jo Allison Smith Family Foundation, with matching funding from UNC's Institute for the Environment and the Wallace Genetic Foundation. Additional support was provided by NSF (OCE-1332881 and OCE-1736772) to CA.

#### ACKNOWLEDGMENTS

We thank JP Balmonte, Kim DeLong, Sarah Underwood, and Benjamin Rhodes for assistance with lab and fieldwork. We also thank Anna Jalowska and Alexander Stephan for creating the map used as **Figure 1**.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2017.02589/full#supplementary-material

Supplementary Figure 1 | Gage height (ft) for the Tar-Pamlico and Neuse rivers. Solid circle show time points at which were collected; lines connect consecutive sampling dates.

Supplementary Table 1 | Temperature, salinity, dissolved oxygen, and pH at the date and time of sampling for all stations and sites.

Supplementary Table 2 | Leucine incorporation, cell counts, leucine incorporation per cell, and DOC concentration for all stations and sites.

Supplementary Table 3 | Polysaccharide hydrolase activities (nmol monomer L <sup>−</sup><sup>1</sup> h −1 ) for all stations and sites. Ara, arabinogalactan; cho, chondroitin sulfate; fu, fucoidan; lam, laminarin; pul, pullulan; xyl, xylan.

Supplementary Table 4 | Gage height and discharge on the date of sampling for each station and date.

Supplementary Table 5 | α-glucosidase, β-glucosidase, and leu-aminopeptidase activities (nmol L−<sup>1</sup> h −1 ) for each station and date.

Supplementary Table 6 | Data shown in Figure 10; P-values as calculated using the corrplot package in R. Values where P < 0.05 are shown in bold font.

Supplementary Table 7 | Data shown in Figure 11. P-values as calculated using the corrplot package in R. Values where P < 0.05 are shown in bold font.


Bold, H. C. (1985). Algae. Englewood Cliffs, NJ: Prentice Hall.


Methods Microbiol. 30, 227–238. doi: 10.1016/S0580-9517(01) 30047-8


status. Hydrobiology 569, 113–127. doi: 10.1007/s10750-006- 0126-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 © 2017 Bullock, Ziervogel, Ghobrial, Smith, McKee and Arnosti. 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 Effect of Increased Loads of Dissolved Organic Matter on Estuarine Microbial Community Composition and Function

Sachia J. Traving<sup>1</sup> , Owen Rowe2,3† , Nina M. Jakobsen<sup>4</sup> , Helle Sørensen<sup>4</sup> , Julie Dinasquet<sup>5</sup>† , Colin A. Stedmon<sup>6</sup> , Agneta Andersson2,3 and Lasse Riemann1,5 \*

#### Edited by:

Anna M. Romaní, University of Girona, Spain

#### Reviewed by:

Hans-Peter Grossart, Leibniz Institute of Freshwater Ecology and Inland Fisheries, Germany Guang Gao, Nanjing Institute of Geography and Limnology (CAS), China

> \*Correspondence: Lasse Riemann lriemann@bio.ku.dk

#### †Present address:

Owen Rowe, Department of Food and Environmental Sciences, University of Helsinki, Helsinki, Finland; Julie Dinasquet, Laboratoire d'Océanographie Microbienne, Observatoire de Banyuls sur mer, Banyuls sur mer, France

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 24 November 2016 Accepted: 20 February 2017 Published: 09 March 2017

#### Citation:

Traving SJ, Rowe O, Jakobsen NM, Sørensen H, Dinasquet J, Stedmon CA, Andersson A and Riemann L (2017) The Effect of Increased Loads of Dissolved Organic Matter on Estuarine Microbial Community Composition and Function. Front. Microbiol. 8:351. doi: 10.3389/fmicb.2017.00351 <sup>1</sup> Centre for Ocean Life, Marine Biological Section, University of Copenhagen, Helsingør, Denmark, <sup>2</sup> Umeå Marine Sciences Centre, Umeå University, Hörnefors, Sweden, <sup>3</sup> Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden, <sup>4</sup> Laboratory for Applied Statistics, Department of Mathematical Sciences, University of Copenhagen, Copenhagen, Denmark, <sup>5</sup> Marine Biological Section, University of Copenhagen, Helsingør, Denmark, <sup>6</sup> Centre for Ocean Life, National Institute of Aquatic Resources, Technical University of Denmark, Charlottenlund, Denmark

Increased river loads are projected as one of the major consequences of climate change in the northern hemisphere, leading to elevated inputs of riverine dissolved organic matter (DOM) and inorganic nutrients to coastal ecosystems. The objective of this study was to investigate the effects of elevated DOM on a coastal pelagic food web from the coastal northern Baltic Sea, in a 32-day mesocosm experiment. In particular, the study addresses the response of bacterioplankton to differences in character and composition of supplied DOM. The supplied DOM differed in stoichiometry and quality and had pronounced effects on the recipient bacterioplankton, driving compositional changes in response to DOM type. The shifts in bacterioplankton community composition were especially driven by the proliferation of Bacteroidetes, Gemmatimonadetes, Planctomycetes, and Alpha- and Betaproteobacteria populations. The DOM additions stimulated protease activity and a release of inorganic nutrients, suggesting that DOM was actively processed. However, no difference between DOM types was detected in these functions despite different community compositions. Extensive release of re-mineralized carbon, nitrogen and phosphorus was associated with the bacterial processing, corresponding to 25–85% of the supplied DOM. The DOM additions had a negative effect on phytoplankton with decreased Chl a and biomass, particularly during the first half of the experiment. However, the accumulating nutrients likely stimulated phytoplankton biomass which was observed to increase towards the end of the experiment. This suggests that the nutrient access partially outweighed the negative effect of increased light attenuation by accumulating DOM. Taken together, our experimental data suggest that parts of the future elevated riverine DOM supply to the Baltic Sea will be efficiently mineralized by microbes. This will have consequences for bacterioplankton and phytoplankton community composition and function, and significantly affect nutrient biogeochemistry.

Keywords: bacterioplankton community composition, community functions, extracellular enzymes, 16S rRNA, climate change, dissolved organic matter, generalized linear models, Baltic Sea

## INTRODUCTION

fmicb-08-00351 March 7, 2017 Time: 18:38 # 2

Climate change is projected to increase precipitation in the northern hemisphere, by as much as 30% in the Baltic region (Andersson et al., 2015). This will lead to a 15–20% increase in freshwater runoff (Meier et al., 2012) and parallel increases in loadings of riverine dissolved organic matter (DOM) and nutrients to coastal systems. The input of DOM and nutrients will conceivably affect the composition, function and activity of the recipient microbial communities and thereby impact the entire ecosystem (Sandberg et al., 2004; Wikner and Andersson, 2012). The elevated riverine DOM supply will likely increase bacterial respiration, as humic material is typically carbon rich and fuels bacterial respiration (Benner, 2003; Fasching et al., 2014). In coastal systems riverine DOM may support a significant fraction of bacterioplankton activity, thereby decoupling the activity of heterotrophic bacteria from phytoplankton production (Kemp et al., 1997; Sandberg et al., 2004; Wikner and Andersson, 2012; Figueroa et al., 2016). Moreover, an increased allochthonous carbon load supporting bacterial activity may lead to increased competition for inorganic nutrients between bacteria and phytoplankton (Thingstad et al., 2008; Wikner and Andersson, 2012), which may further weaken the link between phytoplankton production and bacterial activity and expand zones of net heterotrophy.

Bacterial communities are tightly coupled to the concentration and composition of DOM, with shifts in DOM composition inducing changes in bacterial community composition and functionality (e.g., Gasol et al., 2002; Kritzberg et al., 2004; Judd et al., 2006; Alonso-Sáez and Gasol, 2007). Therefore, the concentration and characteristics of riverine DOM may be important for the ultimate fate of this additional organic matter, such as the partitioning between local remineralization, passage through the food web or export to adjacent shelf seas. In addition, the increases in DOM will likely lead to "brownification" of the recipient waters and increased light attenuation due to higher concentrations of humic substances in DOM (Roulet and Moore, 2006), which is commonly observed in coastal regions (Sanden and Håkansson, 1996; Aksnes and Ohman, 2009; Frigstad et al., 2013). This may affect phytoplankton biomass and productivity ultimately leading to carbon limitation of bacterioplankton growth (Thingstad et al., 2008).

The Baltic Sea consists of basins characterized by distinct hydrology and differing land usage. The consequences of climate change are therefore expected to vary regionally (Rönnberg and Bonsdorff, 2004). The north receives large inputs of riverine DOM rich in humic substances (Andersson et al., 2015), while the south receives a comparatively nutrient rich riverine DOM inflow (Stepanauskas et al., 2002). The riverine loadings differ significantly in C:N:P stoichiometry depending on region and can potentially have very different effects on the recipient ecosystem. However, it remains unclear how the character and concentration of DOM supplied by rivers will influence the bacterioplankton and thereby the fate of carbon and nutrients in the system.

The objective of this study was to experimentally investigate the effects of elevated DOM on a recipient coastal microbial community and examine the impact on functional activity and population structures. Specifically, we focused on the response of the microbial community to differences in DOM characteristics.

## MATERIALS AND METHODS

### Collection Site and Mesocosm Setup

The experiment was performed from May 21–June 25 2012 at Umeå Marine Sciences Centre (UMF), Sweden, in 12 indoor mesocosm tanks of 4.87 m height (water column) and a diameter of 0.76 m. The water was collected on May 21 and 22 from a regularly sampled station in the Bothnian Sea (63◦ 320 05.2<sup>00</sup> N 19◦ 560 09.6<sup>00</sup> E) at 4 m depth, using a rotating pump (Flygt DS 3057.181 MT-230), with a 1.5 mm mesh pre-filter. The salinity of the water was 3.8 (Seaguard CTD, Aanderaa) and in situ temperature was 6◦C. The water was transported in 1 m<sup>3</sup> polythene containers to the field station and thereafter carefully pumped into the mesocosms, ensuring an equal distribution of the water between tanks. All mesocosms received inorganic nutrients on May 21, at concentrations of 0.7 µmol l−<sup>1</sup> nitrogen and 0.09 µmol l−<sup>1</sup> phosphorus, to prevent nutrient exhausting during a 4 days acclimation period. During this period the temperature was incrementally increased to 15◦C. To prevent stratification, a constant and gentle bubbling was applied at 0.6 m depth and convective stirring was generated by maintaining the upper section of each mesocosm tank at 14.8◦C, the middle section at 15.0◦C, and the lower section at 15.2◦C, the latter supported by a 250 W electrical heater. Light was provided on a 12 h light:dark cycle, using 150 W halogen lamps (MASTERColour CDM-T 150W/942 G12 1CT. Phillips©), yielding a photosynthetically active radiation (PAR) light level of ∼400 µmol photons m−<sup>2</sup> s −1 immediately below the surface.

### Treatments

The 12 mesocosm tanks received one of four treatments, each replicated in triplicate. Two treatments (North and South) received additions of soil extracted DOM (**Table 1**), prepared from soil samples collected along the riverbanks of two contrasting rivers that discharge into the Baltic Sea. The North treatment was collected from the Öre river in northern Sweden (63◦ 320 40.7<sup>00</sup> N 19◦ 420 35.6<sup>00</sup> E), the catchment area of this river being predominantly characterized by coniferous and deciduous forest. The South treatment was collected from the riverbanks



In parentheses, the increase after the initial ∼22% addition; Section "Materials and Methods." Ctrl<sup>N</sup> and Ctrl<sup>S</sup> received inorganic N and P in the stoichiometry and concentrations supplied with the DOM in the North and South treatments.

of the Reda river, Poland (54◦ 380 35.80<sup>00</sup> N, 18◦ 270 41.28<sup>00</sup> E), with a catchment area characterized by agricultural activities and some broad-leaf forest. Soil extracts were prepared and stored according to the procedure described in Lefebure et al. (2013). In brief, soil extracts were mixed with Milli-Q water and ion exchange resin (Amberlite IRC 7481) for 48 h at 4◦C, and then filtered through a 90 µm mesh. The carbon (C), nitrogen (N), and phosphorus (P) content of the extracts were determined using a Shimadzu TOC-5000 carbon analyzer and a Braan and Luebbe TRAACS 800 autoanalyzer. The experiment was designed to simulate future predictions of increased riverine DOM load to the Baltic (Eriksson Hägg et al., 2010). The DOM treatments aimed to increase dissolved organic carbon (DOC) in the water with roughly 50% (North) and 100% (South) relative to the DOC concentration (∼333 µmol l−<sup>1</sup> ) in the Bothnian Sea. Two additional treatments were set up as controls for the DOM addition, receiving only inorganic N and P (**Table 1**), corresponding to the inorganic N and P concentrations in the North and South, from here on referred to as "CtrlN" and "CtrlS," respectively.

Three days prior to first sampling DOM treatments received an initial large DOM addition, corresponding to ∼22% of the total amount of DOC added during the experiment, and this was then followed by smaller daily doses (**Table 1**). Control treatments also received corresponding N:P additions. The experiment had a duration of 32 days and sampling was initiated on May the 28th by the addition of seven equally sized young-ofthe-year perch (Perca fluviatilis) to each mesocosm. The fish were added in order to complete a pelagic food web, encompassing bacteria to zooplanktivorous fish. Every second day 40 l of water from 2 m depth was removed, and replaced with 40 l of 0.2 µm filtered Bothnian Sea water pumped into the marine research station from a point ∼2 km offshore (63◦ 330 15.6<sup>00</sup> N 19◦ 500 08.4<sup>00</sup> E). Samples were collected on days: 0, 3, 7, 10, 14, 17, 21, 24, and 28. The majority of parameters were sampled/processed on all days and extracellular enzymes were measured on two additional occasions, days 5 and 12.

### Water Chemistry, Phytoplankton Biomass, and Community Composition

Colored dissolved organic matter (CDOM) samples were filtered through pre-combusted Whatman <sup>R</sup> GF/F filters, and the filtrate was immediately frozen at −20◦C until further processing. Samples were measured within 2 months of the experiment on a Horiba Scientific Aqualog according to Murphy et al. (2011). CDOM absorption properties were characterized by calculating the spectral slope across 300–650 nm (Stedmon et al., 2000). Samples for inorganic nutrients and DOC were filtered through 0.22 µm cellulose-acetate filters (Gelman Supor <sup>R</sup> ) and immediately measured. Inorganic nutrients (NH<sup>4</sup> <sup>+</sup>, NO<sup>3</sup> −, NO<sup>2</sup> <sup>−</sup>, and PO<sup>4</sup> <sup>3</sup>−) were measured in whole water samples using continuous flow analysis on a Quaatro system (Seal Analytical), following methods outlined in Grasshoff et al. (1983) and Helcom guidelines (HELCOM, 2014). Samples for measurement of total dissolved nitrogen (TDN), total dissolved phosphorus (TDP), and DOC were filtered through 0.2 µm Acrodisk (Supor <sup>R</sup> ) filters (Pall Corporation). TDN and TDP samples were subjected to oxidative digestion using the peroxodisulfate/boric acid system (Koroleff, 1983), and then analyzed using the same method as for nitrate and phosphate. DOC was approximated as Non Purgeable Organic Carbon and analyzed using high temperature catalytic oxidation followed by non-dispersive infrared sensor detection of the gaseous CO2. The instrument used was a Shimadzu TOC-L Total Organic Carbon Analyzer (Shimadzu Corporation). Dissolved organic nitrogen (DON) was calculated by subtracting NH<sup>4</sup> <sup>+</sup>, NO<sup>3</sup> <sup>−</sup>, NO<sup>2</sup> <sup>−</sup> from TDN, and total dissolved phosphorus (DOP) was calculated by subtracting PO<sup>4</sup> <sup>3</sup><sup>−</sup> from TDP.

Chlorophyll a (Chl a) and phytoplankton community composition were measured as detailed in Lefebure et al. (2013). In brief, Chl a samples were filtered onto Whatman <sup>R</sup> GF/F filters (100 ml in duplicates), extracted in 95% ethanol for 24 h in the dark, and measured on a Perkin Elmer LS 30 fluorometer. Samples for phytoplankton community composition were preserved in Lugol's solution and later counted using microscopy and converted to biomass (µg C l−<sup>1</sup> ) using carbon conversion (Menden-Deuer and Lessard, 2000).

#### Bacterial Functions

Bacterial production was assayed using <sup>3</sup>H-thymidine incorporation (Fuhrman and Azam, 1982). Samples were incubated with thymidine (final concentration of 24 nM, and an activity of 84.4 Ci mmol−<sup>1</sup> ) at 15◦C for ∼1 h, and analyzed using a Beckman 6500 scintillation counter. Bacterial production was estimated using a conversion factor of 1.4 × 10<sup>18</sup> cells µmol−<sup>1</sup> thymidine (Wikner and Hagström, 1999) and 20.4 fg C cell−<sup>1</sup> (Lee and Fuhrman, 1987). Samples for bacterial abundance were preserved in a 4% final concentration of formaldehyde, stained with acridine orange, and enumerated using an epifluorescence microscope (Zeiss Axiovert 100, Thornwood, NY, USA) and image analysis (Blackburn et al., 1998).

Extracellular enzyme activities were assayed using fluorogenic 4-methylumbelliferone (MUF) and 7-amino-4-methylcoumarin (MCA) substrates (Sigma-Aldrich, St. Louis, MO, USA). The enzyme assays were prepared according to Hoppe (1983), modified to a reaction volume of 200 µl with 400 µM substrate (final concentration). The enzymes assayed and substrates used were: protease (L-Leucine-MCA), alkaline phosphatase (MUF-phosphate disodium salt), β-NAGase (MUF-N-acetyl-β-D-glucosaminide), lipase (MUF-oleate) and α- and β-glucosidases (MUF-α-D-glycopyranoside and MUFβ-D-glycopyranoside. respectively). Assays were measured in triplicates at 355 nm excitation and 460 nm emission on a FLUOstar OPTIMA plate reader (BMG, Labtech GmbH, Ortenberg, Germany). Assays were incubated at 15◦C in the dark and followed for 5 h.

### Bacterial Community – Extraction of Nucleic Acids and 16S rRNA Amplicon Sequencing

Samples (500 ml) for DNA and RNA extraction were filtered onto separate 0.22 µm cellulose-acetate filters (Gelman Supor <sup>R</sup> ), except for the South treatment which clogged at 300 ml. DNA

filters were stored in 1 ml Tris-EDTA buffer (VWR, Radnor, PA, USA) at −20◦C, while RNA filters were covered with 1 ml RNAlater <sup>R</sup> (Ambion <sup>R</sup> , Life Technologies, Carlsbad, CA, USA), flash frozen in liquid nitrogen, and stored at −80◦C until further processing. Extractions of DNA and RNA were done using the E.Z.N.A <sup>R</sup> Tissue DNA kit (OMEGA biotek, USA) and PowerWater <sup>R</sup> RNA isolation kit (MO BIO Laboratories, Inc., USA), respectively. Synthesis of cDNA was done according to Bentzon-Tilia et al. (2015) using TaqMan reverse-transcription reagents (Applied Biosystems, Foster City, CA, USA) and the 806r primer listed below. 16S ribosomal RNA (cDNA) and DNA amplicons of the V4 region of bacterial and archaeal communities were obtained using the primers 515f (5<sup>0</sup> -126 GTGCCAGCMGCCGCGGTAA) and 806r (50 -GGACTACHVGGGTWTCTAAT) (Caporaso et al., 2012). Polymerase chain reactions (PCRs) were performed in 25 µl reaction volume containing 1 (DNA) and 10 (cDNA) ng template, primers, MyTaq DNA polymerase reagents (Saveen & Verner AB, Limhamn, Sweden), 0.07 nmol BSA (Sigma-Aldrich, St. Louis, MO, USA) and 2 nmol MgCl<sup>2</sup> (DNA Diagnostic, Risskov, Denmark). The PCR conditions included an initial denaturing step at 94◦C for 3 min followed by 29 cycles of 94◦C for 45 s, 50◦C for 1 min, 72◦C for 1 min 30 s, and a final step of elongation at 72◦C for 10 min. Triplicate PCR reactions were pooled for each sample, purified using the Agencourt AMPure XP purification kit (Beckman Coulter, Inc., Brea, CA, USA), and quantified using the Quant-iTTM PicoGreen <sup>R</sup> quantification kit (Invitrogen, Waltham, MA, USA) and a FLUOstar OPTIMA plate reader (BMG Labtech GmbH, Ortenberg, Germany). PCR amplicons were pooled at equimolar concentrations and submitted for commercial paired-end sequencing on an Illumina MiSeq.

Sequence reads were assembled, trimmed to a mean length of 252 nucleotides, and de-multiplexed using QIIME v1.9 (Caporaso et al., 2010). Removal of singletons and clustering of operational taxonomic units (OTUs) at 97% similarity was done in USEARCH v1.8 (Edgar, 2010) using the UPARSE-OTU algorithm (Edgar, 2013) with implicit chimera check. Taxonomy was assigned in QIIME using uclust (Edgar, 2010) and the Greengenes v.13.8 reference database (McDonald et al., 2012). Chloroplasts and mitochondrial reads were removed before downstream analysis. OTUs only occurring once in the dataset and/or including < 10 reads in total were excluded. Henceforth, 16S rRNA and rRNA gene amplicons are referred to as rRNA and rDNA, respectively. Sequences were deposited in GenBank NCBI (accession numbers KX178204-KX179464). In the rRNA dataset three samples were lost during processing – (day, tank): (21, 4), (17, 5), (28, 10).

#### Data Analysis

All data and statistical analyses were carried out in R (R Core Team, 2016) using the R packages mvabund v. 3.11.7 (Wang et al., 2012), pgirmess v. 1.6.4 (Giraudoux, 2016), FactoMineR v. 1.31.4 (Husson et al., 2015), and vegan v. 2.3-2 (Oksanen et al., 2015). For testing differences between treatments, linear mixed models were used; with day, treatment, and their interaction as fixed effects and mesocosm tank as a random effect. Data were log-transformed if appropriate. The average outcome over time was compared for North and CtrlN, South and CtrlS, and for North and South. A total of 45 tests were carried out, and the p-values were Bonferroni–Holm adjusted in order to ensure a family wise error rate of at most 5%. Patterns of bacterial community compositions were analyzed using Bray– Curtis distances and visualized by non-metric multidimensional scaling (NMDS). Specifically for the NMDS and descriptions of relative composition of communities, samples were subsampled to a depth of 2280 reads per sample to accommodate for the smallest sample in the set.

Generalized linear models (GLM) were applied to identify the OTUs contributing most to the differences between the four treatments at the beginning (day 0) and end (day 28) of the experiment. The aim of the GLM approach described here is similar to that of the SIMilarity PERcentages analysis (SIMPER, Clarke, 1993), but also takes the mean-variance relationship in the rDNA and rRNA data, into account (Warton et al., 2012). For this analysis, datasets were filtered to include only OTUs with a relative abundance > 1% in at least one observation, leaving 87 and 95 OTUs in the rDNA and rRNA datasets, respectively. For each of these OTUs, and separately for days 0 and 28, univariate negative binomial regression models were fitted with abundance as the response variable. In each model, treatment was included as an explanatory variable while the logarithm of the total abundance of OTUs in each observation was included as an offset variable, in order to take into account the varying sample sizes. The models were fitted using a log link function and assuming an unknown over-dispersion parameter in each model, which was estimated from the data. Likelihood ratio (LR) test statistics for the hypothesis of no treatment effect were computed for each of the univariate models. Corresponding p-values adjusted for the multiple tests performed within the rDNA and rRNA datasets, respectively, for each of the days 0 and 28, were computed by means of a step-down resampling procedure using residual permutation (as implemented in Wang et al., 2012). These p-values were further adjusted by multiplication with a factor four (a Bonferroni-type correction) to ensure a family-wise error rate of <5% among all tests pertaining to the GLM models. OTUs with the largest LR test statistics are interpreted as the OTUs that exhibit the most notable difference in relative abundance between treatments. For an adjusted p-value of less than 0.05, the difference between treatments is considered statistically significant.

## RESULTS

### DOM and Nutrients

Dissolved organic carbon increased significantly during the experiment (**Figure 1A**) in both North and South compared to their controls (adjusted p << 0.0001) and also differed between North and South (adjusted p < 0.002). However, the increase was less than expected based on the added DOC (**Figure 1A**). The total amount of DOC removed corresponded to 25 and 37% of the added DOC in the North and South treatments, respectively (**Figure 1B**), calculated as the difference between added and measured DOC. In comparison, 12 and 13% of the

FIGURE 1 | (A) Concentrations of dissolved organic carbon (DOC) over time in the four treatments. The dotted lines ("North pre" and "South pre") represent predictions of accumulated DOC calculated from the respective DOM additions assuming no biological consumption, in North and South, respectively. Error bars represent standard deviations (n = 3). (B) The stoichiometry of the consumed DOM in the treatments. In the DOM treatments, consumption of DOC is calculated as % difference between the predicted DOC accumulation and measured DOC concentrations, and in the controls the consumption is calculated as the difference between measured start and end concentrations. Consumption of dissolved organic nitrogen (DON) and phosphorus (DOP) was calculated as for DOC. Values above each bar are the consumed concentrations (µmol l−<sup>1</sup> ).

DOC was removed in the Ctrl<sup>N</sup> and CtrlS, respectively (calculated as the difference between measured DOC at start and end). For DON and DOP, the communities in both treatments appeared to effectively mineralize a large fraction. In the North treatment 63 and 81% of the added DON and DOP, respectively, was removed and similar high removal was observed in the South treatment (57 and 85% DON and DOP, respectively), despite the higher DOM additions in South. Small amounts of DON accumulated in Ctrl<sup>N</sup> and CtrlS, corresponding to 7 and 3%, respectively, while 30% of the DOP was removed. The C:N:P stoichiometry of the removed DOM was 239:36:1 and 114:15:1 for North and South, respectively. This corresponded roughly to the stoichiometry of the added DOM, 387:41:1 for North and 156:18:1 in South.

Inorganic nutrients were also affected by the DOM additions, which caused significantly higher concentration of NH<sup>4</sup> + (adjusted p < 0.02), NO<sup>3</sup> <sup>−</sup> (adjusted p < 0.001), total N (adjusted p << 0.0001), and total P (adjusted p << 0.0001) in North and South, when compared to their respective controls (**Figure 2**). In addition, NO<sup>2</sup> <sup>−</sup> (adjusted p < 0.001) and PO<sup>4</sup> <sup>3</sup><sup>−</sup> (adjusted p < 0.0001) were also significantly higher in South compared to CtrlS, an effect not detected in the North treatment. Between North and South PO<sup>4</sup> <sup>3</sup>−, total N and P (adjusted p < 0.0001) were significantly different.

The CDOM slope (S) for the UVA-visible wavelength range (300–650 nm) was used as an indicator of differences in DOM quality between treatments. The UVA-visible range was used due to the direct link between organic matter and light attenuation. Other CDOM parameters measured behaved as expected based on the S values, e.g., short wavelength slopes correlated positively with S, and SUVA was inversely correlated to S. Hence, these were not included in further analyses. For both DOM treatments the slopes decreased with time while the controls showed little systematic change and no noteworthy differences. The DOM additions resulted in lower S values with the lowest values in South (**Figure 3A**). The increases in CDOM absorption over time in the DOM treatments resulted in decreasing PAR levels at 1 m depth during the experiment (**Figure 3B**).

#### Phyto-and protozooplankton Biomass and Community Composition

Concentrations of Chl a and phytoplankton biomass decreased over time in all treatments accompanied by shifts in phytoplankton community composition (**Figures 4**, **5**). Initially diatoms, chlorophytes and ciliates made up the majority of the community in all treatments. On day 17, Ctrl<sup>N</sup> and Ctrl<sup>S</sup> were dominated by cyanobacteria whereas North and South were dominated by cryptophytes, prasinophytes and a group of unidentified flagellates. On day 28, total biomass had increased again being highest in North and South treatments with diatoms, chlorophytes and prasinophytes being the dominant groups, whereas chrysophytes, prymnesiophytes, and cyanobacteria dominated in the control treatments.

#### Bacterial Response to DOM Additions

Patterns of bacterial abundance and production followed similar trends in the four treatments (**Figure 6**), with no significant differences in the bacterial abundance or production between DOM treatments, and compared to their respective controls.

Extracellular enzyme activities responded to DOM additions, but only protease had significantly higher activity (adjusted p < 0.0002), in North and South, compared to their controls (**Figure 7**). Furthermore, protease activity was not significantly different between North and South.

#### Shifts in Bacterial Community Composition

A total of 1,866,318 high quality reads remained after quality and chimera check, with 1,637 reads assigned to Archaea.

Clustering at 97% similarity resulted in a total of 1,261 OTUs. The composition in the total community (rDNA) changed in response to treatment and time, becoming more dissimilar over time and resulting in North and South communities shifting apart from each other, and from the two controls (**Figure 8A**). A similar shift in composition was observed in the active community (**Figure 8B**).

To further investigate the observed relative changes in community composition, a GLM analysis was applied to identify the OTUs for which relative abundance differed between the four treatments. This was done for the beginning (day 0) and the end (day 28) of the experiment. The OTUs with the most notable and significant differences between treatments, as indicated by the size of their LR test statistics and adjusted p-values, are reported together with their relative abundances (**Table 2** and **Figure 9**). At day 0 in the total communities the relative abundance of six OTUs were significantly different between treatments, five Bacteroidetes OTUs (OTU\_19, 36, 51, 55, and 761) and one Alphaproteobacteria (OTU\_7). In the active communities at day 0, two OTUs (OTU\_280 and 19, from Betaproteobacteria and Bacteroidetes, respectively) exhibited significant differences between treatments.

By the end of the experiment the community composition had changed, and the OTUs showing significant differences between treatments in the GLM models had also changed. On day 28, four OTUs were significant in the total communities, two OTUs from Gemmatimonadetes (OTU\_46 and 61) and two OTUs from Bacteroidetes and Planctomycetes (OTU\_ 29 and 154, respectively). In the active communities on day 28 only a single OTU from Bacteroidetes (OTU\_761) showed a significant difference between treatments.

## DISCUSSION

Elevated inputs of riverine DOM have been predicted for the future Baltic Sea (Andersson et al., 2015). The present experiment suggests that a fraction of the terrestrial DOM supplied was available to the microbial community over timescales similar to that of mixing in coastal waters. The organic matter addition was processed by the microbial community and resulted in an accumulation of inorganic nutrients, and this bacterioplankton mediated change in resource utilization stimulated phytoplankton biomass despite the simultaneous decrease in light penetration due to light absorption by CDOM.

### Fate of DOM

Dissolved organic matter flocculation can be an important removal process in estuarine systems (Sholkovitz, 1976; Sholkovitz et al., 1978). However, the majority of studies suggest that this is highly salinity-dependent. DOM flocculation is low

at salinities < 5 and a minor factor in systems with relatively stable salinities (Sholkovitz et al., 1978; Søndergaard et al., 2003). As salinities were constant at 3.8 in this experiment it is unlikely that abiotic flocculation contributed significantly to DOM removal. Microbial activity has also been suggested to contribute to DOC flocculation in freshwater with estimated removal rates between 0.3 and 1.2 µmol C l−<sup>1</sup> d −1 (Wachenfeldt et al., 2009). That would correspond to flocculation causing a C removal of 2 to 18% in our experiment, a relatively small fraction of the total C removed from the North and South treatments. We therefore interpret the removed DOM in the mesocosms to be primarily caused by microbial activity. The microbial communities in the DOM treatments consumed organic resources in ratios reflecting the C, N, P – stoichiometry of the input DOM, suggesting a relatively flexible consumption, as observed in bacterial isolates and natural aquatic assemblages (Makino et al., 2003; Godwin and Cotner, 2015). The observed differences in bacterial processing of DOM did not translate into differences in bacterial production or abundances between treatments. This suggests that the additional DOM consumption in North and South was not channeled into the bacterial biomass. However, grazing by hetero- and mixotrophic protists and viral lysis are the major processes controlling bacterial removal, unfortunately neither group was quantified so the fate of the DOM that was channeled into the bacterial commmunity remains speculative.

As a consequence of bacterial degradation the stoichiometry of the input DOM will affect the resource landscape of the recipient water due to the flexible consumption of the bacterioplankton. This seems to be supported by the significantly higher inorganic nutrient concentrations in the DOM treatments compared to the controls, likely causing the increased phytoplankton biomass towards the end of the experiment. The accumulated DOM, i.e., the fraction not utilized by bacteria, was rich in organic C, causing an increased coloration of the water and decreasing levels of PAR, and the CDOM characteristics (S values) in the DOM

FIGURE 5 | Phyto- and protozooplankton community composition shown as stacked bars, indicating the total biomass. Chl a is indicated by gray dots. All values are averages from the triplicate mesocosms.

treatments were lower, typical of terrestrial DOM (Stedmon et al., 2000). In our experiment, the increased light attenuation likely opposed more extensive phytoplankton growth induced by the available re-mineralized inorganic nutrients. Hence, a future elevated outflow of DOM may affect the relative importance of inorganic nutrients and light in the Baltic Sea – the two main factors controlling local phytoplankton growth and structure (Andersson et al., 1996).

The DOM additions stimulated protease activity with significantly higher activities measured in the presence of the terrestrial DOM. This is similar to findings in previous studies demonstrating strong protease stimulation by humic-rich DOM (Stepanauskas et al., 1999). The increased protease activity was consistent with the efficient consumption of DON, and lead to a release and accumulation of inorganic nitrogen. Alkaline phosphatase activity was not significantly different between treatments suggesting that the microbial communities in all four treatments were deficient in P. Indeed total P concentrations in all four treatments were relatively low (<0.9 µmol l−<sup>1</sup> ) with more than 80% bound in the organic fraction. Extracellular phosphatases are expressed by both prokaryotes and eukaryotes, and the activity is controlled by external dissolved inorganic P concentrations and intracellular P demand (Hoppe, 2003). Natural P concentrations in the Northern parts of the Baltic are low due to relatively little anthropogenic activity and removal of inorganic P through flocculation and sedimentation (Andersson et al., 1996; Forsgren et al., 1996). The activity of the proteases and alkaline phosphatases was likely the primary drivers in generating inorganic nutrients. No difference was detected in the activities of the glucosidases or β-NAGase, which suggests that

the C sources being hydrolyzed were similar in the treatments and that the added organic C did not stimulate the activities of these enzyme groups. Similarly, no difference was detected in lipase activity. This interpretation disregards the effect of time, and the variability in activities did indicate distinct enzyme responses. Extracellular enzymes vary in how their synthesis is controlled, as evident from the protease and alkaline phosphatase activities measured here, and to which environmental cues they respond (Arnosti, 2011). Moreover, they have a variable distribution and phylogenetic placement in natural bacterial communities (Elifantz et al., 2008; Zimmerman et al., 2013). It may therefore be hard to discern clear-cut responses in extracellular enzyme activities to environmental perturbations when working with natural bacterial communities, like in the present experiment.

The bacterial community composition changed in response to treatment, causing the North and South communities to separate from the two controls, which were indistinguishable. There was also a large effect of incubation on the composition patterns, evident as a relatively uniform temporal trend in all four treatments, a common observation in incubation experiments (Piquet et al., 2010). The active communities had a similar response pattern, suggesting that the DOM additions exerted selective pressure on the present (DNA) as well as the active (RNA) bacterial communities in the DOM treatments. The divergence between the North and South communities was likely driven by the different DOM stoichiometry and concentration. Changes in community composition may influence bacterial C:N:P stoichiometry due to differences in growth and cellular content, and likewise the input stoichiometry of resources

influence community composition (Makino et al., 2003). Furthermore, community composition and succession may be linked to functionality (Kirchman et al., 2004; Teira et al., 2008). However, we did not observe any significant differences between North and South in the protease activities or bacterial production despite large differences in community composition. Hence, with respect to these functionalities the communities of the two treatments were functionally redundant.


TABLE 2 | A list of the operational taxonomic units (OTUs) that contributed significantly to the community composition differences between the four treatments, ordered according to the size of their likelihood-ratio (LR) test statistics and adjusted p-value from the generalized linear model (GLM) analyses.

Only OTUs which were significant are reported here. Taxonomy was assigned using Greengenes v. 13.8.

To examine the communities and identify putative bacterial populations responding to DOM load, individual populations (i.e., OTUs), which contributed significantly to the differences in community composition between the treatments, were identified for the start and end of the experiment. In the day 0 communities, five Bacteroidetes populations and one Alphaproteobacteria population (Rhodobacter) were identified, and all except OTU\_761 had the highest relative abundance in North or South. Bacteroidetes are common and widespread in marine and coastal bacterioplankton communities (DeLong et al., 1993; Cottrell and Kirchman, 2000; Kirchman, 2002). They are often linked to the degradation of high molecular weight organic matter (Covert and Moran, 2001; Gómez-Pereira et al., 2012) and enriched in carbohydrate-active hydrolases and other functions compliant with a lifestyle of utilizing high molecular weight DOM (Davey et al., 2001; Cottrell et al., 2005; Teeling et al., 2012). Most of the Bacteroidetes populations were assigned to Flavobacteriaceae, a widespread group commonly found in soil, fresh and marine waters, and many isolates are known enzyme producers (Bernardet and Nakagawa, 2006). Another Bacteroidetes was a Fluviicola population which dominated in the control treatments. Known Fluviicola strains are from freshwater environments and they utilize carbohydrates for growth (Muramatsu et al., 2012). Alphaproteobacteria also contributed to the difference between communities with a Rhodobacter population. These are also common members of coastal bacterioplankton and often observed as particle-colonizers (González and Moran, 1997; Crump et al., 1999). In the active communities two populations contributed to the differences between treatments. One of these populations also appeared in the total community (OTU\_19) while the other population was related to Delftia (Betaproteobacteria). Both populations were highest in the South community. Previous studies have linked members of Betaproteobacteria, including a relative to Comamonadaceae which Delftia is grouped in, to the utilization of riverine DOM (Kisand and Wikner, 2003). Consequently, the DOM had dramatically altered the community structure, just 3 days after the initial DOM additions.

At the end of the experiment community compositions had changed, and the populations contributing to the differences between treatments also differed. Two populations from Gemmatimonadetes were present in the DOM treatments, with the highest abundances in North. A single population in the active communities contributed significantly to the treatment differences at the end, the Bacteroidetes population OTU\_761 (Fluviicola). This population was the only one occurring both at the start and end, where it had increased its relative abundance in North and decreased in the two controls, suggesting it responded to the DOM. Both Gemmatimonadetes and Planctomycetes inhabit aquatic and terrestrial ecosystems (Ward et al., 2006). Gemmatimonadetes are one of the major phyla in soil communities (DeBruyn et al., 2011), and the populations occurring at the end of the experiment may well originate from the added DOM as the soil extracts were not sterile. However, estuarine communities constantly receive an inflow of terrestrial and freshwater bacteria from surrounding sources, and the brackish conditions and long water retention times (>5 years) in the Northern Baltic Sea has previously been hypothesized to promote relatively stable bacterioplankton communities of uniquely adapted bacteria (Riemann et al., 2008). Furthermore, all the populations responding to the treatments during the experiment, represent phylogenetic groups which have previously been linked to the

degradation of riverine DOM (Kisand et al., 2002; Kisand and Wikner, 2003), and which are commonly identified as responsive in experiments amended with DOM (Pinhassi et al., 2004; Teeling et al., 2012; Lindh et al., 2015). Moreover, the dominance of Bacteroidetes in the GLM analysis suggests that this phylum is particularly responsive to elevated levels of DOM. Moreover, the different response patterns in OTUs contributing to the day 0 and day 28 community differences, suggest that bacterial populations may respond to the particular DOM composition, and that the continued treatment differences over time, were maintained by underlying shifts within each community.

## CONCLUSION

fmicb-08-00351 March 7, 2017 Time: 18:38 # 13

The results presented here indicate that increased DOM inflow will have a major effect on the recipient ecosystem. The DOM caused shifts in bacterioplankton community composition and stimulated protease activity and bacterial DOM consumption. Despite that the different types of DOM selected for distinct communities, only some functions responded, and these did not differ between the DOM treatments. The bacterioplankton utilization of DOM corresponded to approximately 25–85% of the supplied DOM depending on type (DON, DOP, or DOC), and the microbial activity caused a release of inorganic nutrients. This conceivably stimulated a succession in phytoplankton community composition as well as increasing biomass towards the end of the experiment. The increased phytoplankton biomass indicates that access to mineralized inorganic nutrients partially outweighs the detrimental effect of extensive light attenuation associated with the high DOM levels. Our experimental data suggest that a large fraction of future elevated DOM inflow to the Baltic Sea will be mineralized by bacterioplankton with consequences for nutrient biogeochemistry and primary production in coastal regions.

## AUTHOR CONTRIBUTIONS

ST, OR, AA, and LR conceived the study. ST, OR, and AA carried out the experimental work. ST, NJ, HS, CS, JD, and

### REFERENCES


LR performed the analyses and interpretations. ST and LR wrote the paper including the comments and revisions from all co-authors. All authors have read and approved the submitted version.

### FUNDING

This project was funded by the EU project MESOAQUA (No. 228224), the Marine Strategic Environment initiative EcoChange – Ecosystems dynamics in the Baltic Sea in a climate change perspective (FORMAS), the VKR Centre of Excellence in Ocean Life funded by the Villum foundation, the BONUS BLUEPRINT and COCOA projects that have received funding from BONUS, the joint Baltic Sea research and development program (Art 185), funded jointly from the European Union's Seventh Program for research, technological development and demonstration, and The Danish Council for Strategic Research.

## ACKNOWLEDGMENTS

The authors would like to thank H. Larsson, A. Brutemark, R. Lefébure, F. Miranda, J. Paczkowska, F. Chiriboga, P. Byström, U. Båmstedt, E. Lindehoff, B. Deutsch, and the staff at Umeå Marine Sciences Centre (UMF), Umeå University, for providing their help and expertise to accomplish this mesocosm experiment.

community sequencing data. Nat. Methods 7, 335–336. doi: 10.1038/nmeth. f.303



coastal site. FEMS Microbiol. Ecol. 73, 68–82. doi: 10.1111/j.1574-6941.2010. 00882.x


**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 Traving, Rowe, Jakobsen, Sørensen, Dinasquet, Stedmon, Andersson and Riemann. 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.

# Formation of Chromophoric Dissolved Organic Matter by Bacterial Degradation of Phytoplankton-Derived Aggregates

Joanna D. Kinsey <sup>1</sup> \*, Gabrielle Corradino<sup>1</sup> , Kai Ziervogel <sup>2</sup> , Astrid Schnetzer <sup>1</sup> and Christopher L. Osburn<sup>1</sup>

<sup>1</sup> Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC, United States, <sup>2</sup> Ocean Process Analysis Laboratory, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH, United States

Organic matter produced and released by phytoplankton during growth is processed by heterotrophic bacterial communities that transform dissolved organic matter into biomass and recycle inorganic nutrients, fueling microbial food web interactions. Bacterial transformation of phytoplankton-derived organic matter also plays a poorly known role in the formation of chromophoric dissolved organic matter (CDOM) which is ubiquitous in the ocean. Despite the importance of organic matter cycling, growth of phytoplankton and activities of heterotrophic bacterial communities are rarely measured in concert. To investigate CDOM formation mediated by microbial processing of phytoplankton-derived aggregates, we conducted growth experiments with non-axenic monocultures of three diatoms (Skeletonema grethae, Leptocylindrus hargravesii, Coscinodiscus sp.) and one haptophyte (Phaeocystis globosa). Phytoplankton biomass, carbon concentrations, CDOM and base-extracted particulate organic matter (BEPOM) fluorescence, along with bacterial abundance and hydrolytic enzyme activities (α-glucosidase, β-glucosidase, leucine-aminopeptidase) were measured during exponential growth and stationary phase (∼3–6 weeks) and following 6 weeks of degradation. Incubations were performed in rotating glass bottles to keep cells suspended, promoting cell coagulation and, thus, formation of macroscopic aggregates (marine snow), more similar to surface ocean processes. Maximum carbon concentrations, enzyme activities, and BEPOM fluorescence occurred during stationary phase. Net DOC concentrations (0.19–0.46 mg C L−<sup>1</sup> ) increased on the same order as open ocean concentrations. CDOM fluorescence was dominated by protein-like signals that increased throughout growth and degradation becoming increasingly humic-like, implying the production of more complex molecules from planktonic-precursors mediated by microbial processing. Our experimental results suggest that at least a portion of open-ocean CDOM is produced by autochthonous processes and aggregation likely facilitates microbial reprocessing of organic matter into refractory DOM.

Keywords: excitation-emission matrix (EEM), base-extracted particulate organic matter (BEPOM), marine snow, aggregates, hydrolytic enzyme activities, phytoplankton, fluorescence

#### Edited by:

Judith Piontek, GEOMAR Helmholtz Centre for Ocean Research Kiel (HZ), Germany

#### Reviewed by:

Teresa S. Catalá, University of Oldenburg, Germany Yumiko Obayashi, Ehime University, Japan

> \*Correspondence: Joanna D. Kinsey jdkinsey@ncsu.edu

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Marine Science

Received: 07 July 2017 Accepted: 14 December 2017 Published: 04 January 2018

#### Citation:

Kinsey JD, Corradino G, Ziervogel K, Schnetzer A and Osburn CL (2018) Formation of Chromophoric Dissolved Organic Matter by Bacterial Degradation of Phytoplankton-Derived Aggregates. Front. Mar. Sci. 4:430. doi: 10.3389/fmars.2017.00430

## INTRODUCTION

Oceanic primary production (PP) contributes between 35 and 65 petagrams (Pg) of carbon annually to global net PP (Field et al., 1998; del Giorgio and Duarte, 2002; Carr et al., 2006; Chavez et al., 2011), mainly through autochthonous production by photosynthetic phytoplankton. Phytoplankton release 2–50% of photosynthetically fixed carbon as dissolved organic matter (DOM) through active exudation and passive leakage (Thornton, 2014), contributing to the largest reservoir of reduced carbon on earth (662 Pg C) (Hansell et al., 2009). Additional phytoplanktonic DOM is released through viral infection, sloppy feeding by predators such as zooplankton, and cell death, with zooplankton contributing their own DOM through excretion of organic matter and fecal pellets (Urban-Rich et al., 2006; Steinberg et al., 2008; Saba et al., 2011). A significant, yet poorly characterized, portion of primary production is exported out of the euphotic zone and into the twilight zone (Siegel et al., 2016). The vertical export has been estimated to range from 5 to more than 12 Pg C y−<sup>1</sup> (e.g., Boyd and Trull, 2007; Henson et al., 2011) spanning annual anthropogenic CO<sup>2</sup> emissions (∼7.0 Pg C y−<sup>1</sup> ) (Siegenthaler and Sarmiento, 1993). Sinking aggregates (or marine snow) are an amalgam of intact phytoplankton cells, detritus, fecal pellets, silica and/or carbonate frustules (that provide ballast) and transparent exopolymeric material (TEP) (Alldredge and Silver, 1988). These aggregates exhibit a wide variability with respect to chemical composition (Wakeham and Lee, 1989, 1993; Minor et al., 2003) and may result in significant export of organic matter to depth (e.g., Hansell et al., 2009; Carlson et al., 2010) and contribution to refractory DOM (RDOM) formation (Lechtenfeld et al., 2015).

DOM is a complex mixture of organic compounds that play an important role in marine biogeochemical cycling. The optically active fraction, chromophoric DOM (CDOM), is essential in both physical and biological processes, controlling light attenuation and photochemical reactions in the surface ocean, impacting the depth of primary production and screening out harmful ultraviolet light (e.g., Arrigo and Brown, 1996; Blough and Del Vecchio, 2002; Mopper et al., 2015). A small fraction of DOM fluoresces, allowing for the identification of DOM sources using excitation-emission matrices (EEMs). Two types of fluorescence are typically described, amino acid-like compounds and humiclike compounds (Coble, 2007), and have been attributed to autochthonous (e.g., primary production) and allochthonous (e.g., terrestrial) production, respectively (Coble, 1996; Yamashita and Tanoue, 2003). Fluorescence of base-extracted particulate organic matter (BEPOM), has been established as a relatively new technique that allows for the optical properties and sources of intracellular fluorophores in POM to be determined (Brym et al., 2014). This work has suggested strong discrete fluorophores including amino acids in estuarine and coastal ocean POM, while studies investigating CDOM production from phytoplankton have confirmed strong amino acid-like fluorescence in axenic cultures with additional fluorescence in the region of humiclike fluorescence for some species (Romera-Castillo et al., 2010; Fukuzaki et al., 2014). Additionally, humic-like fluorescence has been found to increase in degradation studies (Rochelle-Newall and Fisher, 2002; Stedmon and Markager, 2005; Biers et al., 2007; Shimotori et al., 2009; Romera-Castillo et al., 2011) and with apparent oxygen utilization (AOU), especially in the Pacific and Southern oceans (Chen and Bada, 1992; Yamashita et al., 2007; Yamashita and Tanoue, 2008; Jørgensen et al., 2011; Catalá et al., 2015). This suggests a linkage to microbial oxidation and organic matter degradation (Hayase and Shinozuka, 1995; Yamashita et al., 2007; Jørgensen et al., 2011) and may be especially important as an indicator for biogeochemical cycling since DOM serves as a substrate for heterotrophic microbial remineralization of carbon and other trace elements. Microbes utilize extracellular enzymes to catalyze high-molecular weight DOM into smaller compounds that can be transported across cell membranes of bacteria (Arnosti, 2011) or through osmotic uptake by fungi and labyrinthulomycetes (Bochdansky et al., 2017). The carbon is then incorporated into biomass, respired to CO2, or excreted as dissolved organic carbon (DOC) in the form of metabolically transformed products (Arnosti, 2011).

Given the potential importance of heterotrophic bacterial processing of organic matter for both biogeochemical cycling and contribution to deep-sea fluorescence, we aimed to investigate the role of microbes in the formation of openocean CDOM from phytoplankton-derived POM to better understand sources of open-ocean fluorescence. Our study builds upon previous culture studies that investigated CDOM production by incorporating fluorescence of base-extracted POM and hydrolytic enzyme activities in addition to traditional CDOM fluorescence measurements. Roller bottle incubations were used to form planktonic aggregates of four monocultures, sampling at initial, exponential, and stationary growth phases, and after 6 weeks of degradation in the dark. Bacterial abundance and two classes of hydrolytic extracellular enzymes indicative of carbohydrate and peptide hydrolysis (glucosidase and leucine-aminopeptidase, respectively) were measured in conjunction with carbon concentrations and CDOM and BEPOM fluorescence. We demonstrate that the production of CDOM fluorescence is directly related to microbial processing of phytoplankton-derived material.

### MATERIALS AND METHODS

### Phytoplankton Growth Experiments and Sample Collection

Growth experiments were conducted with four phytoplankton taxa commonly found in coastal waters. Three strains were obtained from the NCMA Bigelow collection (https:// ncma.bigelow.org/) including diatoms Skeletonema grethae (CCMP2801, isolated May 2004 off North Carolina) and Leptocylindrus hargravesii (CCMP1856, isolated Feb 1997 in the Gulf of Mexico), and one haptophyte Phaeocystis globosa (CCMP2754, isolated Aug 2003 in the Atlantic Bight). The fourth was a coastal non-axenic isolate of Coscinodiscus sp. (COS1, isolated April 2015 off North Carolina). Two growth experiments were conducted, one with Skeletonema, Leptocylindrus, and Phaeocystis spp. grown in parallel and the other with only Coscinodiscus sp. All cultures were grown in optically clear 0.2 µm-filtered artificial seawater (ASW) prepared with baked (550◦C, 6 h) or ultrapure salts to minimize background CDOM fluorescence. A total of twelve 1.9 L borosilicate bottles per organism were inoculated with an initial concentration of ∼21,000–66,000 cells mL−<sup>1</sup> with the exception of the much slower growing Coscinodiscus sp. with ∼5 cells mL−<sup>1</sup> . All inoculated bottles, along with control bottles of ASW containing no phytoplankton, were amended with f/20 growth media (one-tenth of f/2 media, Guillard and Ryther, 1962). Inoculated bottles and controls were grown on a Wheaton roller culture apparatus rotating at ∼1 rpm (Shanks and Edmondson, 1989). Cultures were maintained at 18◦C and a 12:12 L:D cycle at ∼90 µEinstein m−<sup>2</sup> s <sup>−</sup><sup>1</sup> under cool white fluorescent light in a culture incubator (Percival Scientific, Iowa).

Growth was monitored every other day by in vivo fluorescence and cell counts until particles formed. This information guided more comprehensive sampling for additional chemical and biological parameters (see further detail below) during each growth phase. Skeletonema sp., Leptocylindrus sp., and Phaeocystis sp. were grown for 3–4 weeks under the above conditions, sampling at the beginning of the experiment (T0), during exponential growth (T1), and early stationary phase (particle formation; T2). Immediately after sampling for T2, the remaining culture and control bottles were placed in the dark for 6 weeks to promote bacterial degradation of particles (T3). The Coscinodiscus sp. experiment lasted 6 weeks with no degradation phase and was sampled at the beginning of the experiment (T0), twice during exponential growth [early-exponential and lateexponential (T1)], and early stationary phase (particle formation; T2). Changes in cell densities were used to calculate specific growth rates of phytoplankton during exponential phases (Brand et al., 1981). Cell counts (mL−<sup>1</sup> ) were conducted using an Olympus BX53 compound microscope after preservation with acid Lugol's solution (5%) with settling volumes depending on growth phase (Utermöhl, 1958).

At each time point, three culture bottles of each organism and one control bottle (0.2µm filtered ASW with f/20 nutrients) were sacrificed. Sample aliquots were gently vacuum filtered through pre-combusted (450◦C, minimum 6 h) 0.7µm (nominal mesh size) glass fiber filters. Filters were stored at −20◦C until analyzed for particulate organic carbon (POC), POC/PON, and base-extracted POM (BEPOM) fluorescence followed by base-extracted POC (BEPOC). Filtrate was collected in polycarbonate bottles for absorbance and fluorescence and analyzed within 24 h or in pre-combusted borosilicate vials, acidified with 85% phosphoric acid (H3PO4) to pH 2, and stored at 4◦C for dissolved organic carbon (DOC) analysis.

#### Optical Measurements

POM filters were base-extracted (BEPOM) in 10 mL of 0.1 M sodium hydroxide (NaOH) for 24 h at 4◦C in the dark. After 24 h, extracts were neutralized with concentrated hydrochloric acid, then filtered using a 0.2µm porosity Sterivex-GP polyethersulfone filter (EMD) (Brym et al., 2014). The filtrate was analyzed for absorbance and fluorescence in the same manner as CDOM.

Absorbance of CDOM and BEPOM was measured from 200 to 800 nm in a 1 or 10 cm quartz cell (Starna Cells, Inc.) on a Varian 300 UV spectrophotometer with ultrapure water as the reference blank for DOM samples and neutralized 0.1 M NaOH solution as the reference blank for BEPOM samples. Blank-corrected absorbance values were converted to Napierian absorption coefficients (αλ) (Osburn et al., 2012).

CDOM and BEPOM fluorescence were measured using a Varian Eclipse spectrophotometer with excitation (Ex) from 240 to 450 nm at 5 nm intervals and emission (Em) from 300 to 600 nm every 2 nm. Samples were run with slit widths set at 5 nm for both Ex and Em modes, an integration time of 0.0125 s, scan rate of 2,400 nm m−<sup>1</sup> , and at either 800 or 950 V, depending on sample response. Fluorescence measurements were corrected for excitation energy and emission detector response using correction factors supplied by the manufacturer and any innerfilter effects were corrected following Tucker et al. (1992). Final

TABLE 1 | Central regions of EEM fluorescence attributed to different sources of organic matter, modified from Coble (2007) by Stedmon and Nelson (2015).


Ex, peak excitation wavelength(s); Em, peak emission wavelength(s).

BEPOM fluorescence values were corrected for dilution and filtration volumes and both BEPOM and CDOM values were calibrated in quinine sulfate equivalents (QSE, where 1 QSE = 1 ppb quinine sulfate) (Lawaetz and Stedmon, 2009). Fluorescence results were concatenated into excitation-emission matrices (EEMs) for visualization as contour plots. The humification index (HIX) was determined by dividing the emission intensity in the 435–480 nm region by the intensity in the 300–345 nm region (Ohno, 2002).

#### Carbon Concentrations and POC/PON

DOC and BEPOC (from BEPOM samples) concentrations were measured on an OI Analytical 1030D TOC analyzer in combustion mode (Lalonde et al., 2014). Prior to analysis all acidified samples were sparged with ultrahigh purity argon. DOC concentrations were blank corrected using 18.2 M ultrapure Milli-Q (Millipore) water and calibrated with caffeine standards (0–5 mg C L−<sup>1</sup> ) and Hansell deep sea reference (DSR) water. BEPOC concentrations were blank-corrected against ultrapure water and caffeine standards ranging from 0 to 20 mg C L−<sup>1</sup> . Final BEPOC concentrations were corrected for extraction volume and initial filtration volume. Filters for POC concentrations and POC/PON ratios were dried overnight at 60◦C followed by flash combustion to CO<sup>2</sup> and N<sup>2</sup> using a Thermo Flash 1112 elemental analyzer with acetanilide as a standard and corrected for volume filtered.

#### Bacteria Abundance

Bacterial cells were enumerated by flow cytometry (Gasol and Del Giorgio, 2000). At each sampling point 1 mL of experimental or control water was fixed with 0.1% glutaraldehyde (final concentration) for 10 min at room temperature in the dark, and stored at −80◦C. Prior to analysis, thawed samples were pipetted through a cell strainer (Flowmi, 70µm porosity) and stained with SYBR Green I for 15 min on ice in the dark. Counts were performed with a FACSCalibur flow cytometer (Becton-Dickinson) using fluorescent microspheres (Molecular Probes) of 1µm in diameter as internal size standard. Cells were enumerated according to their right angle scatter and green fluorescence using the FloJo 7.6.1 software. This method quantifies free bacteria and does not account for bacteria attached to the aggregate particles.

## Hydrolytic Enzyme Activity

Hydrolytic enzyme activities were determined using L-leucine-4-methylcoumarinyl-7-amide (MCA) hydrochloride, 4-methylumbelliferyl α-D-glucopyranoside, and 4-methylumbelliferone (MUF) β-D-glucopyranoside (Sigma-Aldrich) as substrate proxies for leucine-aminopeptidase, α-glucosidase, and β-glucosidase activities, respectively (Hoppe, 1983). For each bottle and substrate proxy, 196 µL of unfiltered experimental or control water was added in duplicate to a pure-grade black 96-well plate (Brand Life Sciences) containing a single substrate proxy at saturation levels (final concentration 200µM). Fluorescence (excitation 370 nm, emission 440 nm) was measured in a Tecan Infinite 200 Pro microplate reader immediately following the addition of the substrate and several more times over 7–20 h. The well plates were incubated in the dark at ∼20◦C. MCA and MUF standard solutions prepared in ASW were used to determine hydrolysis rates. Killed controls (boiled sample water) and ultrapure water samples showed little change over the incubations.

#### Statistical Analysis

All statistical analyses were performed using SigmaPlot Version 12.5 (Systat Software Inc.). Results were not normally distributed thus non-parametric Spearman correlation coefficients (r) were computed to examine relationships between all variables. An α-value of ≤0.05 was used for all statistical analyses.

### RESULTS

#### Phytoplankton Growth

Exponential growth (T1) was reached within 15–23 days in all experiments. Specific growth rates (µ) for each of the strains were 0.25 d−<sup>1</sup> for Skeletonema sp., 0.21 d−<sup>1</sup> for Leptocylindrus sp., 0.17 d−<sup>1</sup> for Phaeocystis sp., and 0.11 d−<sup>1</sup> for Coscinodiscus sp. Macroscopic particles formed within 3–8 days of the onset of exponential growth. Particle characteristics varied for each organism, from lightly aggregated particles for Skeletonema sp., to small well suspended particles for Leptocylindrus and Phaeocystis spp. Coscinodiscus sp. aggregates started out stringy and suspended becoming rounded, dense aggregates over time.

### Fluorescence Characteristics

BEPOM samples for genera Skeletonema, Leptocylindrus, and Phaeocystis during exponential growth (T1) were dominated by low-intensity fluorescence (<0.15 QSE) in the region of 275/340 (excitation/emission, peak T) with additional diffuse secondary emission occurring in the regions of 260/400–460 (peak A) and 320–360/420–460 (peak C) (**Figures 1**, **2**, **Table 1**). BEPOM peaks A and C became more distinct and shifted to longer wavelengths over the duration of the experiment (**Figure 3**). Peak A red-shifted in emission wavelengths to 260/440–500 (peak A<sup>∗</sup> ) and peak C red-shifted in both excitation and emission to 350–400/450–500 (peak C<sup>∗</sup> ), resulting in a

TABLE 2 | Spearman correlation coefficient (r) between select CDOM fluorescence peaks and select BEPOM fluorescence peaks (T, M, A, and C) for all data (n = 12) and excluding degradation phase data in parenthesis (n = 9).


Values of r are significant at p-values ≤ 0.05 (values in bold). Coscinodiscus sp. data only for initial through stationary phase, no degradation phase data available.

three-peak pattern with peak T over growth and degradation. For example, BEPOM peak A emission fluorescence in Skeletonema sp. occurred at 400–410 nm for exponential and stationary growth phases but shifted to longer wavelengths (∼490 nm) after 6 weeks of degradation (**Figure 3**). Coscinodiscus sp. showed only faint secondary emission during both exponential growth samplings, but still followed the three-peak pattern observed in the other cultures. By stationary phase (T2), Coscinodiscus sp. had minimal secondary emission in the region of peaks C and C<sup>∗</sup> . Maximum fluorescence for all fluorescent peaks occurred during stationary phase (T2) with a sharp decrease (ca. 70%) over the 6 weeks of degradation (T3, **Figures 1**, **2**).

In contrast to the distinct fluorophores in the BEPOM samples, CDOM samples from exponential growth (T1) had low (<0.3 QSU), unstructured fluorescence with slightly enhanced fluorescence in the region of peak T and B (275/305), most commonly associated with protein-like fluorescence (**Figures 4**, **5**, **Table 1**). Higher overall fluorescence intensities occurred during stationary (T2) and degradation (T3) phases and demonstrated a three-peak pattern similar to the BEPOM samples, though CDOM maximum fluorescence intensities were about half of BEPOM maximum fluorescence intensities. While CDOM shared the three-peak features observed in the BEPOM, the three peaks were much broader and less distinct in CDOM than in BEPOM and increased in fluorescence throughout growth and degradation for all areas of fluorescence (**Figures 4**, **5**). Overall, CDOM fluorescence patterns were similar between all the phytoplankton species, with peaks A and T being the most prominent. CDOM peaks A and C were significantly correlated with BEPOM peaks A and C for many of the cultures, and all were significantly correlated when degradation (T3) time points were excluded (**Table 2**).

The humification index (HIX) is a florescence index of the degree of organic matter degradation, with higher values characteristic of higher molecular weight, aromatic compounds (Huguet et al., 2009). CDOM HIX values increased over the duration of the experiment for all of the cultures, whereas BEPOM HIX values were more variable with no clear trend among the cultures (**Figure 6**).

#### Carbon Concentrations and POC/PON

Carbon concentrations were similar between all three carbon pools (particulate, base-extracted, and dissolved) for all four cultures (**Figure 7**, **Table 3**). In general all carbon concentrations increased through stationary phase and decreased over the 6 weeks of degradation. The exception to this was Phaeocystis sp. DOC concentrations that leveled off between stationary and degradation phases, and for Coscinodiscus sp. which did not include a degradation stage. The largest increase occurred in the particulate pool with POC increasing between initial (T0) and stationary phase (T2) by 1.56 mg C L−<sup>1</sup> for genera Skeletonema and 5.08 mg C L −1 for Leptocylindrus. For base-extracted POC, Phaeocystis sp. had the smallest increase (0.33 mg C L−<sup>1</sup> ) compared to Leptocylindrus sp. (1.33 mg C L−<sup>1</sup> ) over the experiment. DOC concentrations had a similar range, with Coscinodiscus sp. increasing by 0.29 mg C L−<sup>1</sup> and Leptocylindrus sp. increasing by 1.14 mg C L−<sup>1</sup> between initial (T0) and stationary phase (T2).

Base-extraction efficiencies (BEPOC concentration divided by POC concentration times 100) varied between 4 and 88%, with Leptocylindrus sp. having a slightly narrower range (4–36%) (**Table 3**). These extraction efficiencies were consistent with those reported in Brym et al. (2014). Greater extraction efficiencies occurred earlier in growth (e.g., exponential phase 21–78%) and decreased with the duration of the experiment with the lowest extraction efficiencies occurring after 6 weeks of degradation (4– 24%, T3). Overall BEPOC concentration was correlated to POC concentration (r = 0.75, p < 0.001).

POC/PON ratios ranged from 5.2 to 10.9 for time points between exponential (T1) and degradation growth phases (T3) (**Table 3**). Higher POC/PON ratios occurred for Leptocylindrus sp. (6.9–9.6) and Coscinodiscus sp. (6.9–10.9), while Skeletonema sp. (6.7–7.2) and Phaeocystis sp. (5.2–6.8) had slightly lower POC/PON ratios; however, these differences were not significant.

#### Bacterial Abundance and Activities Bacterial Cell Counts

The greatest increase in bacterial cell numbers occurred for genera Phaeocystis and Leptocylindrus, which had maximum bacterial cell numbers during stationary phase of growth (T2), reaching 5.5 ± 2.4 × 10<sup>6</sup> cells mL−<sup>1</sup> and 4.3 ± 1.0 × 10<sup>6</sup> cells mL−<sup>1</sup> , respectively (**Figure 8**, **Table 3**). By the end of the 6 weeks of degradation (T3), bacterial cell numbers decreased by nearly half to 3.6 ± 0.7 × 10<sup>6</sup> cells mL−<sup>1</sup> and 2.4 ± 0.5 × 10<sup>6</sup> cells mL−<sup>1</sup> , respectively. Bacterial cell numbers in the Skeletonema culture increased throughout the growth experiment, but only increased from 0.02 × 10<sup>6</sup> at the

initial sampling (T0) to 1.0 × 10<sup>6</sup> during degradation (T3). The Coscinodiscus culture had the lowest overall bacterial cell numbers, only reaching 0.2 × 10<sup>6</sup> cells mL−<sup>1</sup> during stationary growth (T2).

#### Hydrolytic Enzyme Activities

Aminopeptidase enzyme activities were typically three to five orders of magnitude higher than either glucosidase activities, with the Phaeocystis culture having the highest activities during stationary phase (T2, **Figure 8**, **Table 3**). Activities of β-glucosidase were typically a factor of two greater than α-glucosidase activities, except for Phaeocystis sp. which had β-glucosidase activities a factor of six to seven higher than α-glucosidase. Due to sampling constraints, no data was available for late-exponential (second T1) or stationary (T2) phases for Coscinodiscus spp.

Enzyme activities were highly correlated to fluorescent components, especially for glucosidase activities with genera Leptocylindrus and Phaeocystis and for aminopeptidase with Skeletonema sp. (**Table 4**). α- and β-glucosidase activities were also correlated to all carbon concentrations for genera Leptocylindrus and Phaeocystis.

### DISCUSSION

#### Formation of CDOM by Marine Plankton in Culture

Previous laboratory experiments have relied on high nutrients (e.g., Rochelle-Newall and Fisher, 2002; Romera-Castillo et al., 2010) or additions of carbon substrates (e.g., Gruber et al., 2006; Goto et al., 2017) to observe CDOM formation by phytoplankton and/or bacteria. In our experiments, we produced CDOM fluorescence and phytoplankton densities on the same magnitude as open ocean waters using moderate nutrient additions and rotating bottles to keep cells in suspension. Using this approach, our maximum CDOM fluorescence for all cultures was low and ranged between 0.5 and 2 QSE (**Figure 4**), comparable to fluorescence intensities in several oceanic regions (0.5–6 QSE) (Nelson and Gauglitz, 2016; Netburn et al., in press). Additionally, trans-oceanic Pacific and Atlantic CDOM fluorescence ranged between 0.1 and 1 QSE for humic-like components and 0.1 to 11 QSE for protein-like components when data was modeled by parallel factor analysis (PARAFAC) (Murphy et al., 2008). Together, our results and previously published data, demonstrate that fluorescence intensities produced in culture experiments are similar to oceanic fluorescence intensities, even allowing for instrument variability. The low fluorescence intensities produced further suggest future culture experiments should use media with minimal background fluorescence to observe the low CDOM production by plankton.

Base-extraction of chromophoric and fluorophoric material from marine particles is a relatively new way to analyze POM and allowed for direct comparison between the two organic matter pools. The available field studies have been conducted in estuaries and coastal waters, but show remarkable consistency in BEPOM fluorescence. Overall fluorescence was low, typically

Frontiers in Marine Science | www.frontiersin.org


less than 0.25 QSE, with maximum fluorescence corresponding to chlorophyll maxima and decreasing with depth (Brym et al., 2014; Ziervogel et al., 2016; Netburn et al., in press). For our cultures, maximum BEPOM fluorescence ranged between 0.4 and 3 QSE during stationary phase and decreased to 0.1– 0.9 QSE during degradation (**Figure 1**). The elevated BEPOM fluorescence in cultures relative to natural samples was likely the result of measuring the cultures at the peak of phytoplankton biomass in a controlled environment compared to natural phytoplankton blooms that have more complex growth dynamics and influenced by other processes such as aggregation, grazing, and hydrodynamics.

### Spectral Properties of Phytoplankton-Derived CDOM and BEPOM

All four monocultures produced a three-peak pattern that was dominated by discrete BEPOM fluorophores, suggestive of distinct compounds such as aromatic amino acids (**Figure 1**). This three-peak pattern has been described previously in natural BEPOM samples of estuarine (Brym et al., 2014) and oceanic (Ziervogel et al., 2016; Netburn et al., in press) origin. The prominence of the protein-like peak strongly resembles tryptophan, which has been shown to be dominant during phytoplankton growth (Stedmon and Markager, 2005; Murphy et al., 2008; Jørgensen et al., 2011). Distinct fluorescence in the region of peaks A and C have been attributed to fluorophores present in derivatives of benzoic acids (including phenols) and flavins, respectively (Wolfbeis, 1985; Wünsch et al., 2015). Other possibilities include siderophores (Fukuzaki et al., 2014). More generally, peaks A and C have previously been attributed to, and are abundant in, terrestrially-derived humic substances. Given our algal growth medium lacked material of terrestrial origin, as it was prepared with artificial seawater with f/20 nutrients, the observed formation and subsequent red-shift of peaks A and C were likely the result of microbially-transformed planktonic material rather than terrestrially-derived fluorescence.

CDOM fluorescence showed similar peak regions to BEPOM, but were generally broader and unstructured in emission and increased through growth and degradation phases (**Figure 4**). We interpret these findings to suggest that degradation processes occurred on the algal-derived material once in the dissolved phase. In previous studies, CDOM generated in phytoplankton and bacterial cultures showed similar fluorescence patterns to those presented here. Culture work on diatoms, dinoflagellates, and prasinophytes illustrated CDOM production by phytoplankton at intensities observed in the ocean (Romera-Castillo et al., 2010). Fukuzaki et al. (2014) presented EEM spectra measured on DOM from axenic diatoms, dinoflagellates, chlorophytes, cryptophytes, haptophytes, and raphidophytes monocultures. Though the dominant fluorophores differed between individual species, the ubiquitous fluorescence patterns produced overall by a range of genera further supports planktonic-sources of open


TABLE 4 | Spearman correlation coefficient (r) between bacterial abundance or enzyme activity and DOC, BEPOC, POC and select BEPOM and CDOM fluorescence peaks (T, M, A,

January 2018 | Volume 4 | Article 430

and

C).

Values of r are significant at p-values ≤ 0.05 (values in bold). For all relationships

 n = 12, except for Coscinodiscus

 enzyme activities, where n = 6 since no data were available for the last two time points of that experiment.

ocean CDOM. However, these studies were only able to suggest that CDOM was formed contemporaneously with plankton growth as no fluorescence measurements on POM material were performed (Romera-Castillo et al., 2010; Fukuzaki et al., 2014). Additionally, these previous studies as well as our own still show elevated fluorescence in the regions of peaks A, C, and T, unlike open-ocean fluorescence which is low overall with maximum fluorescence in the area of peak A (Jørgensen et al., 2011). This suggests longer degradation experiments than performed in the present study and others (e.g., Gruber et al., 2006; Fukuzaki et al., 2014) are required to fully transform the protein and humic-like fluorescence into the low CDOM fluorescence observed in the ocean.

Results from our phytoplankton growth experiments, viewed through BEPOM to represent the particle phase, in addition to CDOM, strongly indicate that the unstructured CDOM fluorescence signal was a direct product of the microbial loop operating on planktonic biomass. Correlations between CDOM and BEPOM fluorescence peaks A and C (**Table 2**) were strongest when samples from degradation time points were excluded, due to the continual increase in CDOM fluorescence and concurrent decrease in BEPOM fluorescence (**Figures 4**, **5**, **Tables 1**, **2**). The overall increase in CDOM fluorescence and humification index (**Figure 6**), along with general decreases in all three carbon pools (**Figure 7**) and high microbial activities (**Figure 8**) between stationary and degradation time points, suggest microbial alteration of the DOM pool resulted in its compositional alteration to more fluorescent components. CDOM fluorescence increased in all peak regions throughout growth and degradation, including peaks A and C that have previously been described as humic-like substances arising from terrestrial material (Coble, 2007; Stedmon and Nelson, 2015). However, with no terrestrial material in our culture experiments, the humic-like fluorescence has to be a direct product of autochthonous production and microbial transformation. Similar processes occurring in the open-ocean water column would explain the global observations of elevated fluorescence and correlations with AOU in the deep ocean (Chen and Bada, 1992; Yamashita et al., 2007; Yamashita and Tanoue, 2008; Jørgensen et al., 2011; Catalá et al., 2015) that has been linked to the remineralization of organic matter and formation of RDOM (Yamashita et al., 2007; Jiao et al., 2010; Hansell, 2013; Lechtenfeld et al., 2015).

One advantage of incorporating BEPOM measurements into field sampling would be to observe subtle fluorescence shifts that are more distinct than in the dissolve phase and indicate microbial transformation of autochthonous material (**Figure 3**). A few previous studies have observed red-shifts in peaks A and C in CDOM, but these signals could easily be misinterpreted as terrestrial humic-like material or overshadowed by strong terrestrial influence, especially in coastal and freshwater systems. Murphy et al. (2008) hypothesized degradation of their coastal and open-ocean CDOM samples resulted in a PARAFAC component (C3) with excitation maxima at 260 and 370 nm and an emission maxima at 490 nm, similar to the locations of our red-shifted peaks A and C. Additionally, a similar PARAFAC component (C2, Ex 240 and 370 nm, Em 480) was determined from a meridional transect in the Atlantic Ocean by Kowalczuk et al. (2013), with its intensities doubling below the mixed layer. Burdige et al. (2004) also observed red-shifts in humic-like fluorescence of peaks A (239/429) and M (328/422) to longer wavelengths, resulting in peaks A' (248/461) and C (360/460) in sediment pore waters, and attributed the red-shift to diagenetic transformations of the original fluorophores. The red-shift in natural samples were all explained as a result of microbial degradation processes and most were at locations in the open ocean, far removed from terrigenous inputs of DOM. Thus, these studies support our interpretation that the appearance of redshifted peaks in our plankton experiments were due to microbial degradation and that structurally complex substances were formed from planktonic precursors. The structural complexity may arise as a direct result of marine snow formation and microbial processing. Aggregation during particle formation could bring aromatic aldehydes and ketones in close proximity with hydroxylated benzoic acid derivatives through the microbial reprocessing into more complex molecules such as RDOM (Lechtenfeld et al., 2015). The presence of these molecules in the same moiety could facilitate charge transfer interactions between electron acceptors and donors, which has been suggested for lignin derivatives (Del Vecchio and Blough, 2004; Baluha et al., 2013; Sharpless and Blough, 2014). While the biochemical pathways and ecological ramifications of the planktonic CDOM phenomenon remain elusive, we can be certain that not all oceanic CDOM is terrestrially-derived and plankton are important sources of RDOM to the deep ocean.

#### Organic Matter Remineralization

Positive correlations between bacterial cell numbers and the different carbon pools (**Table 4**) suggest a close link between bacterial growth and activities in our experiments, with the microbes actively transforming organic matter compounds, such as carbohydrates, to shape the CDOM and BEPOM pools. Furthermore, decreases in BEPOM fluorescence (ca. 70%), BEPOM extraction efficiencies, and BEPOC concentrations during the 6 weeks of degradation suggest that substantial amounts of base-extractable POM components were degraded or solubilized to the dissolved phase by bacteria. This solubilized material was then rapidly removed from the dissolved phase, likely though organic matter remineralization to CO2, resulting in net increases in DOC concentrations ranging between 0.19 and 0.46 mg C L−<sup>1</sup> , regardless of the maximum carbon concentrations produced in each culture (**Figure 7**). This net DOC production was on the same order of magnitude as RDOC concentrations observed in the deep ocean (0.41–0.96 mg C L−<sup>1</sup> ) (Hansell et al., 2009).

Strong correlations between bulk glucosidase activities and organic matter fluorescence indicate that either carbohydrate hydrolysis products were a major source of CDOM likely resulting from POM sources, or that bacterial formation of CDOM is an energetically demanding process. Goto et al. (2017) demonstrated that a strain of Alteromonas grown solely on glucose produced humic-like CDOM fluorescence with emission properties similar to what we measured in our cultures. In contrast, amino acids and other nitrogen compounds apparently played a minor role in the formation of the pool of fluorescing organic matter as indicated by fewer correlations between aminopeptidase activities and organic matter parameters (**Table 4**). Rather, intermediate reactions, such as the reaction of aldehydes with organic amines may incorporate N into structures that can produce charge transfer, thus creating "humic-like" CDOM fluorescence on these purely planktonic precursors (Kieber et al., 1997; Brandes et al., 2004; Del Vecchio and Blough, 2004). Furthermore, the overall high rates of aminopeptidase (**Figure 8**, **Table 3**) indicate rapid turnover of nitrogen compounds and recycling of nutrients throughout the phytoplankton growth experiments. The coupling of BEPOM analysis of fluorescence along with measurements of enzyme activities has clearly opened a new dimension of observation of processes by which CDOM in the ocean may be formed by plankton.

#### CONCLUSIONS

Overall, our study demonstrates that phytoplankton-derived organic matter directly results in the formation of unstructured "humic-like" CDOM fluorescence, facilitated by aggregation and microbial processing of precursor POM biomolecules containing discrete, yet unidentified, fluorophores. We have demonstrated the ability to measure fluorescence from the cellular biomass of POM, extracted into dilute base, but without the need to use resin extraction techniques to concentrate the signal, which may alter the composition (and optical properties) of CDOM. The overall trends in growth and production of fluorescent CDOM

#### REFERENCES


follows global ocean observations in fluorescence patterns and intensity, as well as, net production of DOC. Furthermore, the significant correlations between BEPOM and CDOM fluorescence, especially peaks A and C, challenges the paradigm that humic-like fluorescence in the open ocean is the result of terrestrially-derived material (Andrew et al., 2013; Jørgensen et al., 2014). Finally, hydrolytic enzymatic rates suggest that the bacterial community actively transforms organic matter with significant turnover of carbon and nitrogen, making bacteriallymediated production of RDOM from planktonic biomass a key pathway in global element cycles (e.g., Jiao et al., 2010; Lechtenfeld et al., 2015; Moran et al., 2016).

### AUTHOR CONTRIBUTIONS

All authors contributed to the design of the study and data interpretation. JK and GC performed the incubation experiment. JK, GC, and KZ ran sample analyses. JK wrote the initial draft of the manuscript and all authors contributed to its revision.

### FUNDING

Financial support for this work was provided by the National Science Foundation Chemical Oceanography award 1459406.

### ACKNOWLEDGMENTS

The authors thank Daniel Wiltsie, Mackenzie Fiss, and Alexandra Gizzi for their assistance in sample collection and analyses.


from fluorescence spectroscopy and parallel factor analysis. Mar. Chem. 157, 170–184. doi: 10.1016/j.marchem.2013.10.004


Geochemistry, eds M. H. Engel and S. A. Macko (Boston, MA: Springer), 145–169. doi: 10.1007/978-1-4615-2890-6\_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 © 2018 Kinsey, Corradino, Ziervogel, Schnetzer and Osburn. 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.

# Extracellular Enzyme Activity Profile in a Chemically Enhanced Water Accommodated Fraction of Surrogate Oil: Toward Understanding Microbial Activities After the Deepwater Horizon Oil Spill

#### Edited by:

Andrew Decker Steen, University of Tennessee, Knoxville, United States

#### Reviewed by:

Zhanfei Liu, University of Texas at Austin, United States John Paul Balmonte, University of North Carolina at Chapel Hill, United States

\*Correspondence:

Manoj Kamalanathan manojka@tamug.edu; manojkamalanathan711@gmail.com

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 21 November 2017 Accepted: 10 April 2018 Published: 24 April 2018

#### Citation:

Kamalanathan M, Xu C, Schwehr K, Bretherton L, Beaver M, Doyle SM, Genzer J, Hillhouse J, Sylvan JB, Santschi P and Quigg A (2018) Extracellular Enzyme Activity Profile in a Chemically Enhanced Water Accommodated Fraction of Surrogate Oil: Toward Understanding Microbial Activities After the Deepwater Horizon Oil Spill. Front. Microbiol. 9:798. doi: 10.3389/fmicb.2018.00798 Manoj Kamalanathan<sup>1</sup> \*, Chen Xu<sup>2</sup> , Kathy Schwehr<sup>2</sup> , Laura Bretherton<sup>1</sup> , Morgan Beaver<sup>2</sup> , Shawn M. Doyle<sup>3</sup> , Jennifer Genzer<sup>1</sup> , Jessica Hillhouse<sup>1</sup> , Jason B. Sylvan<sup>3</sup> , Peter Santschi2,3 and Antonietta Quigg1,3

<sup>1</sup> Department of Marine Biology, Texas A&M University at Galveston, Galveston, TX, United States, <sup>2</sup> Department of Marine Science, Texas A&M University at Galveston, Galveston, TX, United States, <sup>3</sup> Department of Oceanography, Texas A&M University, College Station, TX, United States

Extracellular enzymes and extracellular polymeric substances (EPS) play a key role in overall microbial activity, growth and survival in the ocean. EPS, being amphiphilic in nature, can act as biological surfactant in an oil spill situation. Extracellular enzymes help microbes to digest and utilize fractions of organic matter, including EPS, which can stimulate growth and enhance microbial activity. These natural processes might have been altered during the 2010 Deepwater Horizon oil spill due to the presence of hydrocarbon and dispersant. This study aims to investigate the role of bacterial extracellular enzymes during exposure to hydrocarbons and dispersant. Mesocosm studies were conducted using a water accommodated fraction of oil mixed with the chemical dispersant, Corexit (CEWAF) in seawater collected from two different locations in the Gulf of Mexico and corresponding controls (no additions). Activities of five extracellular enzymes typically found in the EPS secreted by the microbial community – α- and β-glucosidase, lipase, alkaline phosphatase, leucine amino-peptidase – were measured using fluorogenic substrates in three different layers of the mesocosm tanks (surface, water column and bottom). Enhanced EPS production and extracellular enzyme activities were observed in the CEWAF treatment compared to the Control. Higher bacterial and micro-aggregate counts were also observed in the CEWAF treatment compared to Controls. Bacterial genera in the order Alteromonadaceae were the most abundant bacterial 16S rRNA amplicons recovered. Genomes of Alteromonadaceae commonly have alkaline phosphatase and leucine aminopeptidase, therefore they may contribute significantly to the measured enzyme activities. Only Alteromonadaceae and Pseudomonadaceae among bacteria detected here have higher percentage of genes

for lipase. Piscirickettsiaceae was abundant; genomes from this order commonly have genes for leucine aminopeptidase. Overall, this study provides insights into the alteration to the microbial processes such as EPS and extracellular enzyme production, and to the microbial community, when exposed to the mixture of oil and dispersant.

Keywords: enzymes, aggregates, EPS, oil, corexit, bacteria

#### INTRODUCTION

The Deepwater Horizon (DwH) oil spill in the Gulf of Mexico was a catastrophic event that released 4.9 million barrels of oil (Turner et al., 2014). This was followed by the use of ∼2.9 million liters of the dispersing agent Corexit in an attempt to clear the oil from the surface of the ocean and protect coastal ecosystems (Kujawinski et al., 2011). Months after this event, the oil disappeared from the surface ocean, with factors such as natural light induced photo-oxidation, volatilization, sedimentation, and microbial degradation playing an important role (Mendelssohn et al., 2012). The ability of some bacterial taxa to degrade oil has been well established (Bailey et al., 1973), and oil biodegradation during the DwH spill has been extensively reported (Hazen et al., 2010; Valentine et al., 2010; Kostka et al., 2011). Following the oil spill, extracellular polymeric substances (EPS) combined with particulates and other materials, operationally defined as marine snow (Quigg et al., 2016), were observed in large amounts (Passow et al., 2012). Much of this marine snow was also found to have oil included in the aggregated material, so that it was referred to as 'marine oil snow' (MOS) by Passow et al. (2012). Enhanced EPS production in the form of transparent exopolymer particles (TEP) was observed in the presence of oil; this material might have acted as a bio surfactant (Kleindienst et al., 2015b).

Given that EPS production by phytoplankton can vary from 3 to 40% of the total primary productivity (Engel, 2002), and considering primary productivity accounts for 45–50 Pg C yr−<sup>1</sup> in the ocean (Longhurst et al., 1995; Field et al., 1998), the amount of carbon released as EPS could range between 1.5 and 20 Pg C yr−<sup>1</sup> . Furthermore, bacteria are also known to produce EPS (Manivasagan and Kim, 2014), and these EPS have been implicated to play a protective role against adverse environmental condition (Poli et al., 2010). EPS is heterogeneous in composition, consisting mainly of carbohydrates, proteins, monomers of sugars, amino acids, and uronic acids (Underwood et al., 2004; Xu et al., 2011a,b; Quigg et al., 2016). EPS can act as a carbon, nitrogen and phosphorus substrate that assists the growth of bacteria and mixotrophic phytoplankton, thereby boosting microbial activity (Decho, 1990). Several studies have shown that extracellular enzymes facilitate breakdown of complex polymers in EPS to less complex molecules (Jones and Lock, 1993; Romaní and Sabater, 2000; Espeland et al., 2001), which heterotrophic microbes would otherwise not be able to use. Heterotrophic microbes secrete several types of extracellular enzymes (exoenzymes) that assist them in degrading the various components of EPS (Decho, 1990) such as α- and β-glucosidase, alkaline phosphatase, leucine amino-peptidase, and lipase (Yamada and Suzumura, 2010). As enzymes are specific to the substrates they act upon and microbes vary in their ability to produce different kinds of enzymes, the differences in activities between enzymes can be used as an indirect indicator of microbial functional diversity in the system and/or the nutrient composition of the system (Caldwell, 2005).

In addition to the EPS, the oil and Corexit can act as a source of carbon to the microbes during the DwH oil spill, which may have affected EPS production and enzyme activities. Oil and Corexit has also known to cause a significant change in the microbial community, favoring hydrocarbon degraders (Bacosa et al., 2015b; Kleindienst et al., 2015a; Doyle et al., 2018). Therefore, how the addition of oil and Corexit can affect EPS production, extracellular enzyme production, microbial community and aggregate formation needs to be examined. In this study, we conducted mesocosm scale experiments with seawater from two sites in the Gulf of Mexico. A chemically enhanced water accommodated fraction of oil (CEWAF) prepared from a mixture of oil and Corexit (ratio of 20:1) in seawater and Controls (no additions) were used to test the effects of oil/dispersant mixture on the microbial community, their enzymatic activity and EPS production in three different layers within the mesocosm tanks (surface, water column, and bottom).

### MATERIALS AND METHODS

#### Mesocosm Study

Two mesocosm studies were carried out using seawater collected from the Gulf of Mexico during July 2016. The first site located at 27◦ 53N, 94◦ 2W (salinity: 30.77 ppt, pH: 8.38, temperature: 30.8◦C) was chosen as an offshore, open ocean site (∼174 km from shore) while the second site at 29◦ 22N, 93◦ 23W (salinity: 31.13 ppt, pH: 8.02, temperature: 30.5◦C) was chosen as a coastal site (∼20 km from shore). These will be referred to as offshore and coastal respectively. The seawater was supplemented with nutrients at f/20 concentrations (Guillard and Ryther, 1962) before starting the six mesocosms (3 controls, 3 CEWAF). The seawater was used directly as a Control treatment (87 L each mesocosm). Macondo surrogate oil (25 ml) and dispersant in the ratio of 20:1 were combined to produce a chemically enhanced water accommodated fraction (CEWAF) of oil according to Wade et al. (2017). Briefly, the oil and Corexit were added together before being transferred to the corresponding seawater and mixed in 130 L circulating baffled tanks for 24 h under low light and ambient temperature (∼21◦C). At the end of this period, the 87 L of CEWAF was transferred to the mesocosm tanks by pumping from the bottom of the baffled tank in order to avoid the surface slick. The initial nitrate and phosphate concentration were 97 (±4.7) µMol.L−<sup>1</sup> and 11.2 (±0.3) µMol.L−<sup>1</sup> in the Control and 119 (±1.9) µMol.L−<sup>1</sup> and

5.5 (±0.4) µMol.L−<sup>1</sup> in the CEWAF of the offshore mesocosm. Whereas, the initial nitrate and phosphate concentration were 133.3 (±33.8) µMol.L−<sup>1</sup> and 10.2 (±0.2) µMol.L−<sup>1</sup> in the Control and 130.0 (±7.4) µMol.L−<sup>1</sup> and 11.0 (±0.3) µMol.L−<sup>1</sup> in the CEWAF of the coastal mesocosm respectively. The initial estimated oil equivalent (EOE) concentration in the offshore and coastal CEWAF treatment were 39.06 (±0.77) mg.L−<sup>1</sup> and 81.06 (±20.50) mg.L−<sup>1</sup> , respectively. The incubation time for the offshore and coastal mesocosms were 96 hrs and 72 h respectively. The incubation time for both the mesocosm were decided based on the percentage of oil consumed/remaining in the CEWAF tanks. Due to technical issues associated with replicating CEWAF with the same initial oil concentration, the percentage oil concentration was chosen as the deciding parameter over actual oil concentration in terminating the study. The offshore mesocosms took 96 h to reach ∼20% of the initial oil concentration, whereas it only took 72 h in the coastal mesocosm.

#### Sample Collection

Samples for enzyme activity, EPS composition, and total organic carbon (TOC), dissolved organic carbon (DOC), and particulate organic carbon (POC) analysis were collected from three different layers of the mesocosm tanks on the last day. The mesocosm tanks were 74.5 cm long and 43 cm wide. Samples collected in the top 2–5 cm of the tanks were designated as the surface. Samples collected through a spigot mounted on the side of the tanks were designated as water column. Finally, samples collected from the floor of the tanks were designated as the bottom. A syringe was used to collect the samples from the surface and bottom respectively. The samples were collected from these three layers in order to account for the differences in aggregation (higher in the bottom layer for Control and in the surface for CEWAF treatment) observed during the experiment. Samples for bacterial community composition and micro-aggregate counts/sizes were collected concurrently from the water column layer.

#### Enzyme Assays

Enzyme activities for α- and β-glucosidase, alkaline phosphatase, leucine amino-peptidase, and lipase were measured on the last day according to Yamada and Suzumura (2010). The enzymes and the substrates used in this study are listed in **Table 1**. The substrates were dissolved in milli-Q water so that the final stock solutions were 1 mM. The substrate for 4-Methylumbelliferyl oleate was dissolved in minute volume (250 µl) of DMSO and the concentration was adjusted with milli-Q water to a final concentration of 1 mM, Substrates were then added to Control and CEWAF samples in triplicate to a final concentration of 0.2 mM. The samples were then incubated at room temperature in the dark for 3 h. After incubation, the reactions were stopped by the addition of 1 mL of borate buffer solution (0.4 M) adjusted to pH 8.0 for 7-amido-4-methylcoumarin (AMC) tagged substrates or pH 10.0 for 4-methylumbelliferyl (MUF) tagged substrates. The fluorescence intensity was then measured at excitation/emission wavelengths (nm) of 380/440 (AMC) or 365/448 (MUF) using a spectrofluorometer (Shimadzu RF-5300). The measurements were then corrected with the blank values obtained using heated seawater samples (80◦C for 15 min) in duplicate at the beginning of the incubation. Respective substrates corresponding to the different enzymes were added to the blank samples prior to incubation and measurement.

### EPS Analysis

Extracellular polymeric substances composition was measured in terms of carbohydrate, protein, and uronic acid content and total EPS was calculated by summing these parameters. Particles were collected with a polycarbonate filter (0.4 µm, Millipore, United States), and the attached EPS from the particles was then extracted with 0.35 M EDTA followed by an ultrafiltration step to remove the salts and excessive EDTA (Xu et al., 2009, 2011a,b). EPS from the dissolved phase was directly obtained by concentrating and desalting using an Amicon Ultra-15 centrifugal filter unit with ultracel-3 membrane (Millipore, 3 kDa). The carbohydrate concentration in the EPS was determined by anthrone method with glucose as the standard (Yemm and Willis, 1954). The protein content of EPS was determined with the help of a Pierce BCA protein assay kit based on a modified bicinchoninic acid method with bovine serum albumin as the standard (Smith et al., 1985). Uronic acids in the EPS were estimated by the addition of sodium borate (75 mM) in concentrated sulfuric acid and m-hydroxydiphenyl according to Blumenkrantz and Asboe-Hansen (1973) with glucuronic acid as the standard for this assay.

## TOC, DOC, and POC Analysis

TOC and DOC were determined using a Shimadzu TOC-L analyzer (Xu et al., 2011b). For POC analysis, water sample was filtered through a pre-combusted GF/F membrane (0.7 µm, Whatman, United States), and then quantified using a Perkin Elmer Series II CHNS 2400 analyzer, after HCl-fuming to remove the carbonates. Acetanilide (71.09%) was used as the analytical standard (Xu et al., 2011a). Samples from the offshore mesocosm were limited, therefore only samples from the coastal mesocosm were analyzed.

### Dissolved Oxygen (DO) Concentration

A calibrated 556 MPS YSI meter (Yellow Springs, OH, United States) fitted with a DO/Temperature sensor (5563-10) was used to measure the DO (mg L−<sup>1</sup> ) directly in the surface and


bottom of each mesocosm tank on the last day of each mesocosm experiment.

### Microbial and Micro-Aggregate Counts/Sizes

Direct cell counts were performed on the samples collected from the water column on the last day in three replicate tanks per treatment. Samples were visualized with an epifluorescence microscope (Zeiss Axio Imager.M2) after staining the fixed samples with DAPI (45 µM final concentration) for 5 min in the dark and filtering them onto 25 mm, 0.2 µm black polycarbonate filters, according to Doyle et al. (2018). Microbial cell counts were performed at 1000× magnification and, due to their much larger size, micro-aggregates were quantified at 400× magnification. For micro-aggregate abundance, the presence of a micro-aggregate was counted, not the number of cells present per micro-aggregate. Micro-aggregates were defined as groups of cells in clumps 10–200 µm in diameter, often found gathered around drops of oil.

Size fraction analysis of aggregates was also performed using Z1 dual-threshold Coulter counter (Beckman Coulter). It should be noted that these aggregates do not exactly correspond to the microbial micro-aggregates measured above, and likely include those as well as other particles in that size fraction. Samples (15 mL) from the water column were taken on the last day and analyzed immediately. Particles of four different size ranges (5–10, 10–20, 20–50, and >50 µm) were counted with a 100 µm aperture. A sample of filtered seawater was used as blank (typically less than 10 particles were counted). Samples were diluted with filtered seawater (0.2 µm) if the particle coincidence at the aperture exceeded 5%, where particle coincidence is the chance of more than one particle passing through the aperture at once.

### Bacterial Community Composition

Prokaryotic community composition (Bacteria and Archaea) was analyzed as described in detail in Doyle et al. (2018). Briefly, samples (150 ml) collected from the water column in three replicate tanks per treatment concurrently with other samples were pre-filtered through 10 µm filters to remove most eukaryotic cells followed by filtration onto 47 mm 0.22 µm Supor PES filter membranes (Pall). Total DNA was extracted from filters using FastDNA Spin kits (MP Biomedical). 16S rRNA gene (hyper-variable V4 region) was PCR amplified with GoTaq Flexi DNA Polymerase (Promega) according to Caporaso et al. (2012), with specifics in Doyle et al. (2018). Amplifications were performed using the 515F-806R universal primer pair according to recent revisions (Apprill et al., 2015; Parada et al., 2016), which included Golay barcodes and adapters for Illumina MiSeq sequencing. The products were combined and quantified with the QuantiFluor dsDNA System (Promega), pooled and purified with an UltraClean PCR Clean-Up Kit (MoBio Laboratories). The library, along with the three sequencing primers, were sent to the Georgia Genomics Facility (Athens, GA, United States) for MiSeq sequencing (v2 chemistry, 2 × 250 bp). Sequence processing was carried out using mothur v.1.36.1 (Schloss et al., 2009) following a modified version of the protocol described in Kozich et al. (2013). Analysis of rarefaction curves was conducted using all available reads. Generation of operational taxonomic units (OTUs) and analysis of alpha and beta diversity was conducted using a dataset subsampled to 39,054 samples per read (Supplementary Table S1). Goods coverage was >0.99 for all samples.

Raw DNA sequence data used in this project can be found in the NCBI Genbank database under accession numbers SAMN07795505-SAMN07795510 (Offshore) and SRR6176504, SRR6176505, SRR6176497, SRR6176488, SRR6176473 and SRR6176442 (Coastal).

#### Screening of Microbial Genomes

To determine which of the most relatively abundant bacterial families detected are potentially capable of extracellular enzyme production, we searched the Integrated Microbial Genomes (IMG) database (Markowitz et al., 2006) for genomes within these orders using the "Find Functions" search and the "Enzymes (list)" filters for each bacterial order searched, similar to previous work (Jacobson Meyers et al., 2014). EC 3.4.11.1 was used to search for the gene(s) encoding leucine aminopeptidase, EC 3.1.3.1 was used to search for the gene(s) encoding alkaline phosphatase, EC 3.1.1.3 was used to search for genes encoding for lipase, EC 3.2.1.20 was used to search for genes encoding α-glucosidase, and EC 3.2.1.21 was used to search for genes encoding β-glucosidase. Genomes were scored as positive if they contained a gene that encoded for an exoenzyme, and then the number of genes within the family was tallied to calculate the percentage of genomes within that family capable of producing each enzyme. This analysis depends on a few assumptions: (a) if an organism is a member of a clade in which a specific enzyme is more abundant, then that organism is more likely to have the enzyme, (b) the annotations in IMG are reliable, and (c) that the exoenzymes assayed are expressed extracellularly, rather than intracellularly.

#### Statistical Analysis

The results were statistically analyzed by using one-way ANOVA with multiple comparisons of the mean of each group with the mean of every other group using a Tukey test. These statistical analyses were performed using GraphPad Prism software (version 7.0f).

#### RESULTS

Measurement of α-glucosidase activities in different layers of the CEWAF tanks revealed highest activity at the surface in both the offshore and the coastal mesocosms (One way ANOVA: p < 0.02; **Figure 1A**). Similar patterns were seen for both β-glucosidase and alkaline phosphatase, although the differences in alkaline phosphatase were not statistically significant (**Figures 1B,C**). In the Control tanks, the activities of α-glucosidase, β-glucosidase and alkaline phosphatase were significantly higher at the bottom than in other layers in the offshore mesocosm (One way ANOVA: p < 0.0005; **Figure 1**). While the activities were similarly higher at the bottom layer in the Control tanks of the coastal mesocosm, the differences

were statistically significant only for alkaline phosphatase (One way ANOVA: p < 0.0015). Comparison of Control vs. CEWAF tanks in the coastal mesocosm revealed significant differences for α-glucosidase, β-glucosidase and alkaline phosphatase only at the surface (Two way ANOVA: p < 0.004). However, for the offshore mesocosm, α-glucosidase and β-glucosidase activities were significantly higher in CEWAF compared to Control in all three layers (Two way ANOVA: p < 0.0001), whereas alkaline phosphatase was significantly higher only at the surface (Two way ANOVA: p = 0.0011).

Measurement of lipase activities revealed slightly different profile compared to α-glucosidase, β-glucosidase and alkaline phosphatase (**Figure 1D**). For CEWAF tanks, lipase activities were highest in the surface compared to other layers in the offshore mesocosm (Two way ANOVA: p < 0.002), similar to that observed for α-glucosidase, β-glucosidase and alkaline phosphatase. However, in the coastal mesocosm, lipase activities were the lowest in the water column (Two way ANOVA: p < 0.002), and similar between surface and bottom layer in CEWAF tanks. Comparison of Control vs. CEWAF tanks in the offshore mesocosm revealed significant differences only at the surface for the offshore mesocosm (Two way ANOVA: p < 0.0001). However, in the coastal mesocosm, lipase activities were significantly higher in CEWAF than Control in both surface and bottom layers (Two way ANOVA: p < 0.004).

Measurement of leucine aminopeptidase activities in different layers of the CEWAF tanks revealed highest activity at the surface in the offshore mesocosm (One way ANOVA: p < 0.0001; **Figure 1E**). A similar pattern was observed for leucine aminopeptidase in the coastal mesocosm, however the differences were not statistically significant (**Figure 1**). In Control tanks, the activities were higher at the bottom for leucine aminopeptidase in the offshore mesocosm (One way ANOVA: p < 0.0001), whereas the activities were similar in the coastal mesocosm. CEWAF had significantly higher leucine aminopeptidase activities than the Control in all the layers for the offshore mesocosm (Two way ANOVA: p < 0.02), however, significant differences were only seen at the surface for the coastal mesocosm (Two way ANOVA: p = 0.02). Overall, most of the enzymes showed higher activities at the surface for CEWAF treatments and at the bottom for Control treatment in the offshore and/ or coastal mesocosms (**Figure 1**). In addition, overall enzyme activities were significantly higher in CEWAF treatment than in the Control for both offshore (Unpaired t-test: p = 0.0057) and coastal (Unpaired t-test: p = 0.0312) experiment (Supplementary Figure S1).

#### EPS Composition Across the Layers

The polysaccharide, protein, and uronic acid content was measured to determine the overall EPS composition (**Figure 2**). Total EPS concentration was highest in the water column in both treatments for both mesocosms (Two-way ANOVA: p < 0.0001). In Control tanks, significantly higher concentration of EPS was observed at the bottom layer compared to the surface layer in the offshore mesocosm (Unpaired t-test: p = 0.009). Similar trends

FIGURE 2 | (A–D) Average polysaccharide, protein and uronic acid content (±SD) of EPS at measured in the surface, water column and bottom layers of offshore and coastal mesocosms.

were observed in the coastal mesocosm samples, however, the differences were not statistically significant (**Figure 2D**). In the CEWAF tanks, there was more EPS in the surface than the bottom layer in both the mesocosms (Unpaired t-test: p < 0.003). Exposure to CEWAF treatment significantly increased the amount of polysaccharide, proteins and uronic acids produced in the water column and in the surface (Two-way ANOVA: p < 0.0001; **Figure 2**). Comparison of EPS composition in the water column of CEWAF relative to the Control, showed higher production of proteins in both the coastal (5.3 fold) and offshore (8.1 fold) mesocosms followed by uronic acids (4.4 and 2.6 fold) and carbohydrates (2.9 and 2.4 fold). Comparison of EPS composition in the surface of CEWAF relative to the Control showed similar patterns to that observed in the water column.

#### TOC, DOC, and POC

In the coastal mesocosm, TOC and DOC concentrations were highest in the surface layers in both treatments (Two-way ANOVA: p < 0.0025) (**Figure 3A**). Overall, there was significantly

Kamalanathan et al. Exoenzyme Activities in an Oil-Spill

TABLE 2 | Average dissolved oxygen concentration (±SD) at the surface and bottom in control and CEWAF treatments of the offshore and coastal mesocosm tanks.


more TOC and DOC in the CEWAF tanks than in the Control tanks at all layers (**Figure 3B**) (Two-way ANOVA: p = 0.0007). POC concentrations were significantly higher in the CEWAF treatment in the surface layer than in the Control (Unpaired t-test: p < 0.00001) (**Figure 3C**). The bottom POC concentration was higher than both the water column (Two-way ANOVA: p = 0.0007) and surface layers (Two-way ANOVA: p = 0.0056) in the Control treatment. In the CEWAF treatment, the POC was higher at the surface than in the water column and bottom (**Figure 3C**), although these differences were not statistically significant (One-way ANOVA: p > 0.2571).

#### Dissolved Oxygen Concentration

The DO concentration in the CEWAF mesocosms was very low on the last day, decreasing to almost zero at the bottom of the offshore and coastal tanks (**Table 2**). A decrease in DO in the CEWAF treatments from 0.3 (±0.12) mg L−<sup>1</sup> at the surface to 0.1 (±0.04) mg L−<sup>1</sup> at the bottom of the tanks was observed in offshore and 0.4 (±0.01) mg L−<sup>1</sup> to 0.22 (±0.04) mg L−<sup>1</sup> in coastal tanks (**Table 2**). These concentrations are considered to be hypoxic to anoxic. By comparison, the Control treatments had a significantly higher dissolved oxygen concentration throughout the tank with an average value of 6 (±0.37) mg L−<sup>1</sup> in offshore and 7.5 (±1.11) mg L−<sup>1</sup> in the coastal mesocosms from surface to bottom.

#### Microbial and Aggregate Counts

Compared to the Control, bacterial counts in the CEWAF treatments were approximately 3.3 fold higher in offshore and nearly 2 fold higher in coastal seawater (**Figure 4**). Similarly, the micro-aggregate counts in CEWAF were nearly 14 fold higher in offshore and 12 fold higher in coastal as compared to Control treatments (Two-way ANOVA: p < 0.0001). Further analysis of the aggregates based on size showed significantly higher particle concentrations in 5–10, 10–20, and 20–50 µm size range in the CEWAF treatment compared to the Control (Two-way ANOVA: p < 0.0001) (**Figure 5**). Control tanks had significantly higher number of particles than CEWAF treatments in the size range of >50 µm in the coastal mesocosm (Two-way ANOVA: p < 0.0001); however, no differences were observed for the same size range in offshore tanks. Particles abundance was more similar between coastal and offshore tanks across all the size ranges except for 10–20 µm, where it was higher in the coastal mesocosm (Two-way ANOVA: p < 0.0001) (**Figure 5**).

FIGURE 4 | (A,B) Average bacterial and micro-aggregate numbers (±SD) in Control and CEWAF treatments from the offshore and coastal mesocosms.

#### Microbial Community Composition

Prokaryotic communities in Control treatments were more diverse in both mesocosms than in the CEWAF treatments (Supplementary Figure S2 and Supplementary Table S1). Based on 16S rRNA amplicon sequence data, Bacteria were detected at much higher relative proportion than Archaea in both mesocosm experiments (**Figure 6**). Microbial community

composition was dramatically different between the Control and CEWAF samples for both the coastal and offshore mesocosms. The mean relative abundance of the family Alteromonadaceae was highest, followed by Piscirickettsiaceae, Rhodobacteraceae, Flavobacteriaceae, Pseudomonadaceae, and Rhodospirillaceae. Among these abundant families, the genera with the highest relative abundances was Marinobacter, followed by Alteromonas (both belonging to the family Alteromonadaceae), Methylophaga (Piscirickettsiaceae), unclassified Oceanospirillales, Cycloclasticus (Piscirickettsiaceae), Aestuariibacter (Alteromonadaceae) and Tenacibaculum (Flavobacteriaceae). Three different OTUs of Marinobacter (family Alteromonadaceae), OTU3, OTU8 and OTU11, represented 21% of the CEWAF community in the offshore and 45% of the CEWAF community in the coastal experiments, respectively (Supplementary Table S2). OTU level responses were not always uniform within a genus. For example, the relative abundance of the two most abundant OTUs of Methylophaga (family Piscirickettsiaceae) was different between treatments, with OTU2 having highest relative abundance in CEWAF in the offshore mesocosm and OTU5 exhibiting higher relative abundance in the Control for both mesocosm experiments.

Compared to all the extracellular enzymes, α-glucosidase was present in the lowest percentage (<75%) of genomes amongst all the abundant bacterial orders (**Figure 7**). β-glucosidase was 100% positive in Cellvibrionaceae, members of which were abundant in both the CEWAF treatments. In addition, β-glucosidase was highly positive (76-99%) in Pseudomonadaceae, which was abundant in both CEWAF treatments and the offshore Control treatments (**Figure 7**). Lastly, Erythrobacteraceae, which had higher relative abundance in the offshore mesocosm than the coastal mesocosm were highly positive for β-glucosidase as well (**Figure 7**). Alteromonadaceae and Pseudomonadaceae had the highest percentage of genomes positive for lipase (51–75%) (**Figure 7**). Both these orders, especially Alteromonadaceae, were significantly abundant in both the CEWAF treatments. Relative to other extracellular enzymes, leucine amino-peptidase was highly positive (76–100%) amongst all the abundant bacterial orders with the exception of Flavobacteriaceae and Saprospiraceae (**Figure 7**). Alkaline phosphatase was highly positive (76–99%) amongst Alteromonadaceae, Pseudomonadaceae and Cellvibrionaceae, which were abundant in both the CEWAF treatments (**Figure 7**). Amongst bacterial orders abundant in the Control treatments, alkaline phosphatase were 100% positive in

Saprospiraceae and highly positive in Erythrobacteraceae (**Figure 7**).

### DISCUSSION

The goal of this study is to better understand the role of extracellular enzymes, EPS production, and microbial community composition in the presence of a mixture of oil and dispersant. Five different enzymes: α- and β-glucosidase, lipase, alkaline phosphatase and leucine amino-peptidase were studied in the surface, water column and bottom of the mesocosm tanks filled with either coastal or offshore water from the Gulf of Mexico with and without CEWAF. Elevated extracellular enzyme activities have been previously reported in oil containing aggregates (Ziervogel et al., 2012). Also dense microbial colonization and active degradation of oil and components of EPS have been reported (Ziervogel et al., 2012, 2016; Arnosti et al., 2016), suggesting a link between extracellular enzymes and microbial aggregates.

α and β-glucosidase help in the digestion of complex polysaccharides (Reese et al., 1968). These enzymes help in cleavage of glycosidic bonds with α and β-glucosidase acting at α and β -linkages allowing the release of glucose molecule from polysaccharides (de Melo et al., 2006). In the Control treatments, highest α and β-glucosidase activity was seen in layers whereby the polysaccharide concentration in the EPS was lower and vice-versa. Similar patterns but relatively higher levels of activity, were observed in CEWAF treatments as well. This inverse relationship between glucosidase activity and EPS polysaccharide concentration across the three layers is indicative of active breakdown of polysaccharides by these enzymes.

α and β-glucosidase have been shown to help the turnover of polysaccharides secreted by phytoplankton and bacteria, as EPS fueling heterotrophic metabolism in the ocean (Piontek et al., 2010). We therefore hypothesize that the increased activity of these enzymes at the bottom for the Control and surface for CEWAF might help breakdown of the EPS, thereby enhancing the availability of simple carbon in the form of glucose. In addition, compared to the Control, a relatively higher glucosidase activity was observed in CEWAF in both the offshore and coastal mesocosms. Enhanced polysaccharide degradation by extracellular enzymes, especially laminarin, which are targets of glucosidase was observed by Arnosti et al. (2016) in oil associated aggregates. In addition, the polysaccharide concentrations decreased in the Control treatments from ∼0.15 mg.L−<sup>1</sup> at 0 hrs (data not shown) to ∼0.0003 mg.L−<sup>1</sup> at 96 h in the bottom of offshore and ∼0.18 mg.L−<sup>1</sup> (data not shown) to 0.0002 mg.L−<sup>1</sup> in the coastal mesocosm. Similarly in the CEWAF tanks, the polysaccharide concentrations decreased from ∼0.19 mg.L−<sup>1</sup> (data not shown) at 0 h to ∼0.0009 mg.L−<sup>1</sup> at 72 h in the surface of offshore and ∼0.36 mg.L−<sup>1</sup> (data not shown) to 0.002 mg.L−<sup>1</sup> for the coastal mesocosm. We therefore hypothesize that CEWAF induced enhanced EPS secretion in the form of polysaccharide at an earlier time point of this experiment that may have been rapidly degraded by these glucosidases.

Ziervogel et al. (2012) previously showed a correlation between lipase activity and oil degradation. Therefore, the higher lipase enzyme activity at the surface and at the bottom of the mesocosm tanks for the CEWAF treatments, along with other hydrocarbon degrading enzymes (not measured in this study), might have played a role in breakdown of the oil components. Analysis of relative prokaryotic abundance showed the presence of lipase producing families, such as Alteromonadaceae, Oceanospirillaceae, and Pseudomonadaceae, in relatively higher numbers. Several other studies focusing on DwH oil spill have reported the presence of these families in samples containing high concentrations of oil

(Dubinsky et al., 2013; Lamendella et al., 2014; King et al., 2015; Campeão et al., 2017). Additionally, the lipase producing genomic capabilities detected among members of the families Altermonadaceae, and Pseudomonadaceae suggest that these taxa might have played a significant role in the degradation of oil in the CEWAF treatment. The products of these extracellular hydrocarbon degrading enzymes might have led to increased carbon availability in the CEWAF treatment.

Leucine aminopeptidase (LAP) helps microbes acquire nitrogen by breaking down proteins and peptide molecules (Fruton and Mycek, 1956). Likewise, alkaline phosphatase is responsible for degradation of organic phosphates, providing a source of phosphorus to microbes (Martínez, 1968). Both leucine amino-peptidase and alkaline phosphatase activity patterns had higher activities at the surface than in the water column and the bottom of the mesocosm tanks for CEWAF treatments. EPS protein content was higher in the water column of the mesocosms than at the surface or bottom, similar to that observed for polysaccharides. Moreover, the higher leucine amino-peptidase and alkaline phosphatase activities at the surface for the CEWAF treatment pattern matches well with α- and β-glucosidase activities. This suggests active assimilation of nitrogen and phosphorus through leucine amino peptidase and alkaline phosphatase along with carbon through α and β-glucosidase from the EPS, which could have contributed in the generation of microbial biomass. There are other reports where enzyme activities have been shown to correlate with the C:N:P requirements of microbes (Sinsabaugh et al., 2008, 2009). Therefore, we hypothesize that the products of these extracellular enzyme activities observed in our mesocosm tanks may have supported microbial requirements that lead to increased microbial biomass observed in the CEWAF tanks.

Analysis of alkaline phosphatase, leucine amino-peptidase and β-glucosidase in whole water and <10 µm size fractions were statistically indistinguishable (Whitaker, personal communication), indicating that at least large eukaryotes > 10 µm are not responsible for the vast majority of activity measured. These assumptions aside, similar to other studies (Gutierrez et al., 2013; Kleindienst et al., 2015b; Yang et al., 2016a,b), our observations showed that Alteromonadaceae, represented largely by Marinobacter, Alteromonas, and Aestuariibacter, are the most abundant bacterial order detected and also very commonly have alkaline phosphatase and leucine aminopeptidase. This indicates a potentially large contribution to overall enzyme activity from these genera to alkaline phosphatase and leucine aminopeptidase enzyme activity rates. In particular, for AP and Lipase, the percentages of enzyme positive genomes are highly uneven, indicating that a consortia of microbes is necessary to efficiently hydrolyze EPS in situ.

It is also interesting that the family Piscirickettsiaceae is abundant but seems to only use leucine amino-peptidase out of the five enzymes in any significant proportion. All other abundant prokaryotes potentially contribute to 3–4 of the measured enzymes in a similar proportion within the family, but Piscirickettsiaceae operate outside this trend. This may indicate that this abundant family gets its phosphorus and carbon through other means. The two genera common in our mesocosm experiments, Methylophaga and Cycloclasticus, are putative hydrocarbon degraders (Gutierrez and Aitken, 2014). Therefore, they may have used hydrocarbons for carbon and enzymes other than lipase or glucosidases to metabolize that carbon-source.

Some members of the most abundant families in our mesocosms are known as "cheaters"; abundant community members that have minimal to no contribution in terms of enzyme activity (Allison, 2005). For example, members of the family Planctomycetaceae and SAR11 clades (Unclassified, Surface 1, and Surface 2) were abundant in the Control treatments but genomes from these lineages do not harbor abundant exoenzymes: < 10% of Pelagibactereaceae genomes have any of the enzymes assayed here, with the exception of leucine aminopeptidase, which is present in 47% of the genomes queried. Interestingly, there were almost no families lacking extracellular enzyme producing ability (cheaters) in the CEWAF treatments. This could be due to the potential toxicity of oil and Corexit, or the hypoxic environment in the CEWAF treatments restraining the microbial community exclusively to a few dominant families that are being selected due to their ability to tolerate and degrade hydrocarbons (Hamdan and Fulmer, 2011; Bacosa et al., 2012, 2015b; Liu et al., 2017). Such reduced microbial diversity limited to fewer families such as Alteromonadaceae, Pseudomonadaceae, Oceanospirillaceae, Piscirickettsiaceae, and Idiomarinaceae has been reported in other studies focusing of DwH oil spill (Kleindienst et al., 2015b; Yang et al., 2016a,b).

The overall abundance of potential oil degraders such as members of Alteromonadaceae, Pseudomonadaceae, Cellvibrionaceae, and Piscirickettsiaceae in the CEWAF treatments suggests oil may have acted as the primary carbon source. However, since the glucosidase were higher in CEWAF treatments, we hypothesize the products of glucosidases may have provided additional carbon source as well. Moreover, the heavy supply of carbon also increases the demand for other elements such as nitrogen and phosphorus, which are not present in oil. We assume, some portions (in addition to the supplemented nutrients in the mesocosm tanks) of the required high levels of nitrogen and phosphorus to support the growth may have been provided by means of extracellular enzyme reactions such as alkaline phosphatase and leucine aminopeptidase. The overall activities of enzymes were relatively higher in CEWAF treatments than Control, and similar observations were made by Kleindienst et al. (2015b). Therefore, the products of these enzymes may have played a role in the high cell numbers seen in this treatment relative to Control. Several reports indeed have shown that the products of these extracellular enzyme can support the growth of bacteria (Frankenberger and Dick, 1983; Vetter and Deming, 1999). Apart from higher cell counts, statistically higher micro-aggregate counts was also observed in the CEWAF treatments compared to Control. This can be explained by relatively higher protein content of the EPS produced in response to CEWAF. We hypothesize that increase in proteins content may have enhanced the amphiphilic nature of the EPS, which may have facilitated better interaction of the EPS with the oil. This in-turn may lead to formation of

more micro-aggregates in CEWAF treatment. Such role of EPS interaction with oil in the formation of aggregates has been previously suggested (Passow et al., 2012; Kleindienst et al., 2016; Quigg et al., 2016). Although, the enzyme activities and EPS production were higher in the CEWAF treatments than in Control, any interpretation on the effect of dispersant addition alone has to be taken with caution as the study did not compare CEWAF to WAF or a Corexit only treatment. Kleindienst et al. (2015b) suggested Corexit could suppress microbial activity, on the other hand, Bacosa et al. (2015a,b) showed positive effect on growth and alkane degradation in the presence of the dispersant Corexit. There were many differences between the experimental conditions used in Bacosa et al. (2015a,b) and Kleindienst et al. (2015b) such as temperature (8◦C vs. ambient temperature similar to our study) and site of sampling depth (1500 m vs. surface similar to our study). Kleindienst et al. (2015b) worked with a closed bottle system in the dark for 6 weeks while our study lasted a few days in an open system. These differences and others in experimental conditions may have influenced the outcome of dispersant addition on microbial activity. Therefore, despite the observation of higher microbial activity in our CEWAF treatments, further studies are essential to gain a more detailed understanding the effects of dispersant addition alone. Different levels of enzyme activities, EPS production, cell and aggregate counts were observed in response CEWAF of the offshore relative the coastal mesocosm. We hypothesize that these differences could be primarily due to the different initial conditions in the offshore and coastal waters.

#### CONCLUSION

Our study shows that addition of oil and dispersant Corexit enhances extracellular enzyme activity and EPS production in relative to Control, and a similar comparison between oil and dispersant mixture with oil only treatment is needed. Microbial community in CEWAF treatment was mostly dominated by hydrocarbon degraders such as members of Alteromonadaceae, Pseudomonadaceae and Cellvibrionaceae. The higher protein content of EPS in response to CEWAF may have facilitated increased aggregation. However, further studies comparing CEWAF treatment with WAF with time course measurements are needed to discern the effect of dispersant addition on extracellular enzyme and EPS production and aggregation.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

MK designed the experiment, conducted the enzyme measurements and data analysis, and wrote the manuscript. CX helped in conducting the EPS and carbon analysis and manuscript preparation. KS helped in designing the experiment and writing of the manuscript. LB helped in measurements of the particle size and manuscript preparation. MB helped in conducting the EPS and carbon analysis. SD conducted the genomic and associated data analysis and manuscript preparation. JG and JH helped in conducting the experiment. JS helped in genomic data analysis and manuscript preparation. PS helped in designing the experiment and manuscript preparation. AQ mentored the study and helped in designing the experiment and manuscript preparation.

### FUNDING

This research was made possible by a grant from The Gulf of Mexico Research Initiative to support consortium research entitled ADDOMEx (Aggregation and Degradation of Dispersants and Oil by Microbial Exopolymers) Consortium. Data are publicly available through the Gulf of Mexico Research Initiative Information and Data Cooperative (GRIIDC) at http://data.gulfresearchinitiative.org (doi: 10.7266/N7348HTQ, doi: 10.7266/N7C53JDP, and doi: 10.7266/N77D2SPZ).

#### ACKNOWLEDGMENTS

We thank Terry Wade, Gopal Bera, and Tony Knap for making the CEWAF, Kendra Dean in assisting in the collection of the water from the Gulf of Mexico, Julia Sweet for collecting the aggregates, Gen Mei Lin for cell counts, and Jocelyn Simmons for help with enzyme activity measurements.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00798/full#supplementary-material



biofilms from two contrasting rivers. Hydrobiologia 257, 1–16. doi: 10.1007/ BF00013991



Biovar II). Bioresour. Technol. 100, 6010–6021. doi: 10.1016/j.biortech.2009. 06.008


**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 Kamalanathan, Xu, Schwehr, Bretherton, Beaver, Doyle, Genzer, Hillhouse, Sylvan, Santschi and Quigg. 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.

# Methodological Considerations and Comparisons of Measurement Results for Extracellular Proteolytic Enzyme Activities in Seawater

Yumiko Obayashi <sup>1</sup> \*, Chui Wei Bong1, 2, 3 and Satoru Suzuki <sup>1</sup>

*<sup>1</sup> Center for Marine Environmental Studies, Ehime University, Matsuyama, Japan, <sup>2</sup> Institute of Biological Sciences, Faculty of Science, University of Malaya, Kuala Lumpur, Malaysia, <sup>3</sup> Institute of Ocean and Earth Sciences (IOES), University of Malaya, Kuala Lumpur, Malaysia*

Microbial extracellular hydrolytic enzymes that degrade organic matter in aquatic ecosystems play key roles in the biogeochemical carbon cycle. To provide linkages between hydrolytic enzyme activities and genomic or metabolomic studies in aquatic environments, reliable measurements are required for many samples at one time. Extracellular proteases are one of the most important classes of enzymes in aquatic microbial ecosystems, and protease activities in seawater are commonly measured using fluorogenic model substrates. Here, we examined several concerns for measurements of extracellular protease activities (aminopeptidases, and trypsin-type, and chymotrypsin-type activities) in seawater. Using a fluorometric microplate reader with low protein binding, 96-well microplates produced reliable enzymatic activity readings, while use of regular polystyrene microplates produced readings that showed significant underestimation, especially for trypsin-type proteases. From the results of kinetic experiments, this underestimation was thought to be attributable to the adsorption of both enzymes and substrates onto the microplate. We also examined solvent type and concentration in the working solution of oligopeptide-analog fluorogenic substrates using dimethyl sulfoxide (DMSO) and 2-methoxyethanol (MTXE). The results showed that both 2% (final concentration of solvent in the mixture of seawater sample and substrate working solution) DMSO and 2% MTXE provide similarly reliable data for most of the tested substrates, except for some substrates which did not dissolve completely in these assay conditions. Sample containers are also important to maintain the level of enzyme activity in natural seawater samples. In a small polypropylene containers (e.g., standard 50-mL centrifugal tube), protease activities in seawater sample rapidly decreased, and it caused underestimation of natural activities, especially for trypsin-type and chymotrypsin-type proteases. In conclusion, the materials and method for measurements should be carefully selected in order to accurately determine the activities of microbial extracellular hydrolytic enzymes in aquatic ecosystems; especially, low protein binding materials should be chosen to use at overall processes of the measurement.

Keywords: extracellular hydrolytic enzyme, protease, activity measurement, microbial loop, organic matter degradation, low protein binding microplate, MCA substrate

#### Edited by:

*Andrew Decker Steen, University of Tennessee, United States*

#### Reviewed by:

*Zhanfei Liu, University of Texas at Austin, United States Jason B. Sylvan, Texas A&M University, United States*

\*Correspondence: *Yumiko Obayashi obayashi.yumiko.nn@ehime-u.ac.jp*

#### Specialty section:

*This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology*

Received: *31 March 2017* Accepted: *22 September 2017* Published: *10 October 2017*

#### Citation:

*Obayashi Y, Wei Bong C and Suzuki S (2017) Methodological Considerations and Comparisons of Measurement Results for Extracellular Proteolytic Enzyme Activities in Seawater. Front. Microbiol. 8:1952. doi: 10.3389/fmicb.2017.01952*

## INTRODUCTION

In aquatic ecosystems, heterotrophic prokaryotes play important roles in organic matter cycling, including the transformation and remineralization of organic molecules and in its transfer to other organisms via trophic interactions. In order for heterotrophic bacteria that are osmotrophs to obtain nutrients from polymeric biomolecules such as proteins, these high molecular weight organic molecules must be hydrolyzed extracellularly to smaller sizes (approx. <600 Da, Nikaido and Vaara, 1985) prior to their transport across the bacterial outer membrane (Weiss et al., 1991). Thus, hydrolytic activities of extracellular enzymes in aquatic environment are investigated from the standpoint of microbial ecology, biogeochemistry, and organic geochemistry (Arnosti, 2011).

Hydrolytic enzyme activities, such as protease, glucosidase, phosphatase, and chitinase, have been detected and estimated in natural seawaters (reviewed in Hoppe et al., 2002; Arnosti, 2003) using model substrates as proxies of natural substrates. Model substrates added to the sample for measuring potential hydrolytic activities in seawater may be unlabeled oligomers (Liu et al., 2010; Liu and Liu, 2015) or labeled molecules that can be detected as hydrolytic derivatives (e.g., Arnosti, 1996; Pantoja et al., 1997; Steen et al., 2006). Among fluorogenic model substrates, which have fluorophores liberated by enzymatic hydrolysis, 4-methylumbelliferyl (MUF) substrates for α-glucosidase, βglucosidase, and alkaline phosphatase, and 4-methylcoumaryl-7-amide (MCA) substrate for leucine-aminopeptidase are the most commonly used proxies of natural substrates for measuring individual enzymatic activities in seawater samples (Hoppe, 1993). Although, using these proxies to assess natural hydrolytic activities of enzymes in environmental samples results in some theoretical limitations and uncertainties (e.g., Steen et al., 2015), important information on biogeochemical processes in aquatic ecosystems can be obtained.

Estimating more than two hydrolytic activities in the same sample permits consideration of the nutritional mode of the bacteria and the biochemical composition of available polymeric substrates in marine systems (Nagata, 2008). For example, Fukuda et al. (2000) investigated the ratio of activities by leucine-aminopeptidase and β-glucosidase along the east-west transect of the North Pacific and suggested that there is a difference in microbial biochemical conditions between the eastern and western parts of the northern North Pacific. Sala et al. (2001) suggested that the ratio of alkaline phosphatase and aminopeptidase activities could be an indicator of nitrogen and phosphate limitation in the microbial community.

Enzyme activity in bulk seawater samples are often operationally divided into fractions such as "particle-associated" and "dissolved (cell-free)" activities by taking measurements separately on seawater that passes through filters of a specified pore size, as well as unfiltered samples. These fractionations may provide insights into the natural forms of hydrolytic enzymes in seawater and the ecological roles that each play (Arnosti, 2003). Smith et al. (1992) and Karner and Herndl (1992) showed that particle (marine snow)-associated hydrolytic activities were much higher than those in the surrounding seawater at least for their tested enzyme types. Meanwhile, dissolved (free) enzymes have been reported to make substantial contributions to the total activity (e.g., Keith and Arnosti, 2001; Obayashi and Suzuki, 2008a; Baltar et al., 2016), although their ratios vary depending on the sampling conditions and enzyme type.

Proteins and peptides should be good nutrition for heterotrophic prokaryotes after suitable hydrolysis by extracellular enzymes, and thus extracellular proteases are one of the most important hydrolytic enzymes in aquatic microbial ecosystems. Measurements of extracellular proteolytic enzyme activity in aquatic samples using fluorogenic model substrates have been popular since their introduction (e.g., Hoppe, 1993), especially for leucine aminopeptidase, because of their high sensitivity and easiness of the method. Not only aminopeptidase activity but a number of diverse proteolytic enzymes and the importance of trypsin-type endopeptidases, which cleave peptide bonds within a peptide, in natural seawater were also reported using 16 different MCA substrates (Obayashi and Suzuki, 2005). To discuss the relationships or interactions with physical and chemical environmental conditions, or to link those with genomic or metabolomic information on microbial communities, high-resolution reliable measurement data on extracellular proteolytic enzyme activities are required. Highresolution analysis requires many samples to be measured at one time, such as collecting samples at many depth layers at each sampling site, many size fractionations, or testing with different kinds of model substrates. To get reliable activity data, many samples should be analyzed as soon as possible after sampling (German et al., 2011) and each sample measurement should be performed in replicate. Moreover, for reliable estimation of the potential activity of microbial extracellular hydrolytic enzymes in the aquatic environment, many factors should be considered. For example, Obayashi and Suzuki (2008b) pointed out that adsorption effects to the filter for size fractionation could lead underestimation of the enzyme activity in the filtrates depending on the type of filter material and that the effects were different among the type of enzymes. Recently, not only a standard spectrofluorometer with a cuvette, a fluorometric microplate reader with a micro-well plate has been also applied as a device to read fluorescent intensity during a measurement of activity (e.g., Baltar et al., 2010). Microplate reader offers considerable advantage to get measurement data for many samples at one time, however, using 96-well microplates, water sample volume is small and the ratio of the area touching plate material (the wall and bottom of each well) to the sample volume is relatively large. Considering a kind of adsorption effect likely to filters for size fractionation, samples in a microplate might be more susceptible to some kinds of artifact than in a larger volume cuvette. However, many researches might be performed with microplate not always taking care about the types of microplate materials, and to our knowledge, systematic comparison between data obtained from microplates and cuvettes has not been reported so far regarding the measurement of extracellular enzyme activity in a natural seawater sample. Here, we examine several concerns regarding to achieve high-resolution reliable estimations of extracellular proteolytic enzyme activities in seawater using many kinds of MCA substrates and a fluorometric microplate reader.

### MATERIALS AND METHODS

#### Seawater Samples

Seawater samples for each test were collected by bucket or Van Dorn water sampler along the coast of Ehime Prefecture, Japan, and filtered through nylon mesh (150 or 50µm) into bottles to remove large particles. Polycarbonate 500 mL bottles were used as a sample container, except for the experiment to compare several types of containers. Samples were immediately placed on ice for transport to the laboratory where the samples were refrigerated at 4◦C until proceeding with the assays within several hours. In general, to assess the extracellular enzyme activities in natural aquatic samples the activities should be measured as soon as possible after sampling, to minimize the possible alteration after sampling such as degradation of dissolved enzymes and changing of microbial community and their activity. Even though the main purpose of each experiment in this study was to compare the data for the same water sample using different methodological conditions, we conducted each experiment as soon as possible after seawater sampling (within several hours), except for the preservation test in different sample containers. When we kept water samples for a short time before measurement, samples were kept cool and in the dark for the least degradation of enzymes possible, although there were still unavoidable possibilities of microbial cell lysis at 4◦C.

For some experiments, an aliquot of sample was filtered through a 0.2µm pore size polycarbonate Nuclepore filter (Whatman), then both unfiltered and filtered (<0.2µm) seawater samples were used.

#### Enzyme Activities Measurement

The potential activities of extracellular proteolytic enzymes in seawater samples were measured using 17 MCA substrates (**Table 1**, Peptide Institute): 5 for aminopeptidase, 10 for trypsin, and 2 for chymotrypsin. Enzyme activities measurement was conducted as follows with modifications for the different methods tested, as noted below.

MCA substrates were dissolved in solvents dimethyl sulfoxide (DMSO) or 2-methoxyethanol (MTXE) to prepare stock solutions (10 or 20 mM). For assay, 10× substrate working solutions were prepared from the stock solutions in autoclaved artificial seawater and solvent to control for solvent concentration in the solution, with substrate and solvent concentrations 10 times higher than the target final concentrations in assay. Seawater samples and 10× substrate solutions were mixed in disposable cuvettes or 96-well microplates and incubated at 25◦C in the dark to measure potential enzyme activities. The fluorescence of the hydrolytic product, 7-amino-4-methylcoumarin (AMC), was measured several times at intervals (t0, t1, t2, t3; typically 1 h interval) during the incubation. The excitation/emission wavelengths for fluorescence measurements were 380/460 nm on a spectrofluorometer (Hitachi F-2500) or 380/440 nm on a microplate reader (Corona SH8100Lab). A solvent blank (seawater sample with solvent but without substrate) was also prepared and subjected to fluorescence measurements along with samples. After subtracting the solvent fluorescence blank, the concentration of AMC generated during the incubation was calculated using a calibration curve prepared by measuring fluorescence intensity of AMC solutions at seven concentrations (0–1µM) under the same conditions as the sample measurements. To measure the non-enzymatic produced AMC during incubation, collected seawater was autoclaved and prepared and assayed as an inactivated control under the same conditions as the intact sample. The hydrolysis rate of the substrate in the seawater sample, namely, extracellular enzyme activity in seawater, was calculated by determining the increase in AMC concentration with time after subtracting the concentration of non-enzymatic produced AMC estimated in autoclaved seawater.

Every assay was performed in triplicate using three cuvettes or three wells in a microplate for each sample and substrate pair.

### Comparison Methods for Protease Activities Measurement

Extracellular proteolytic enzyme activities in natural seawater samples were measured using different assay methods (Methods A, B, and C) simultaneously with the same working solutions. In Method A, we used a spectrofluorometer with disposable cuvettes, while Methods B and C were conducted on a fluorometric microplate reader with regular and low protein binding microplates, respectively. The excitation/emission wavelengths recommended by the supplier of MCA substrates (Peptide Institute) for the assay were 380/460 nm; however, to obtain a higher intensity fluorescence signal from smaller sample volumes used on microplates (Methods B and C), we set the emission wavelength at 440 nm, the wavelength of maximum fluorescence intensity for AMC. We confirmed that readings taken at 440 and 460 nm provided equivalent results when the corresponding calibration curve was applied. Following are brief overviews of Methods A, B, and C:

Method A (cuvette) was conducted with a reaction volume of 1 mL in a disposable cuvette made from polymethylmethacrylate (PMMA). Fluorescence was measured by a spectrofluorometer at excitation/emission wavelengths of 380/460 nm.

Method B (regular microplate) was conducted with a reaction volume of 300 µL in regular 96-well black microplates made from polystyrene (Nunc #237107). Fluorescence was measured by a microplate reader in fluorescence mode at an excitation/emission wavelength of 380/440 nm.

Method C (low protein binding microplate) was conducted as for Method B except with a low protein binding, black, 96-well microplate made from polystyrene coated with methacryloyloxyethyl phosphorylcholine (MPC) (Nunc #245393).

Pearson's correlation coefficient and simple regression were used to compare data obtained from different methods in pairwise comparisons (Methods A vs. B, Methods A vs. C, Methods C vs. B). Regression analyses of pairs of methods were performed with combined dataset from unfiltered and TABLE 1 | List of fluorogenic substrates used in the present study.


+*Substrate used in experiment;* \**Substrate not soluble in one or both concentrations of solvent. MCA, 4-methylcoumaryl-7-amide; Bz, Benzoyl; Z, Carbobenzoxy; Boc, t-Butyloxycarbonyl; Suc, Succinyl; MUF, 4-Methylumbelliferyl.*

filtered seawater samples for "all estimated activities" and for the "aminopeptidase activities," "trypsin-type activities," and "chymotrypsin-type activities."

#### Kinetic Experiments

Using selected MCA substrates for proteases (**Table 1**), kinetic experiments were performed, and the Michaelis plots obtained by Method B (regular microplate) and Method C (low protein binding microplate) were compared. For the kinetic experiments, MCA substrates were added to samples at final concentrations of 0, 10, 20, 50, 100, 150, 200, and 250µM. Hydrolysis rates of substrates were measured by Methods B and C with the protocol given above.

To determine whether differences in results between Methods B and C are proteolytic enzyme-specific, the same experiment was conducted using MUF substrate for phosphatase (MUF-phosphate, Wako Chemicals) at a final concentration of 0, 10, 30, 60, 100, 160, and 200 µM. Measurement of the hydrolysis rate of the MUF substrates was conducted by the same method as for the MCA substrates, except that the excitation/emission wavelengths of 365/445 nm were used to detect the 4-methylumbelliferon product.

Based on the measured hydrolytic activities, V, and the substrate concentrations, [S], the theoretical maximum activity, Vmax, and the Michaelis constant, Km, were estimated by curve fitting using software OriginPro 9.1 to the Michaelis–Menten equation:

$$V = V\_{\text{max}} \times \text{[S]} / (K\_m + \text{[S]}) \tag{1}$$

#### Comparison of Low Protein Binding Microplates from Different Suppliers

Low protein binding 96-well black microplates from different suppliers were tested using seawater collected by bucket and strained through 50µm nylon mesh into polycarbonate bottles. An aliquot of seawater (180 µL) was mixed with 20 µL of each of 15 MCA substrate solutions (**Table 1**; final concentration, 200µM substrate, 2% DMSO) in three different microplates. Nunc low protein binding plate (Nunc #245393, as described above), Greiner Bio-one No-binding plate (Greiner 655900), and SUMILON Proteosave plate (Sumitomo Bakelite MS-8296K) were used to measure proteases activities in the same seawater sample with the same substrate solution, and the results were compared. Enzyme activities were measured using the protocol described above.

#### Examination of Solvent

The manufacturer of the MCA substrates (Peptide Institute) recommends using DMSO to prepare stock solutions. Considering the possibility of solvent bias in activities measurement, a lower concentration of solvent in assay is better. However, lower solvent concentration may result in reduced solubility of the substrate in seawater. To examine the effect of DMSO, two experiments were conducted with different DMSO concentrations: (1) kinetic experiments comparing 10% (v/v) and 2% (v/v) DMSO, and (2) comparison of the activities measurement with 2 and 1% DMSO with 200µM substrate final concentration.

Substrate working solution (10×) were prepared for each substrate concentration with 100% DMSO for final 10% DMSO in the assay, while those were prepared with autoclaved artificial seawater containing 20% DMSO for final 2% in the assay. Using low protein binding, 96-well microplates (Nunc), natural seawater sample 270 µL and 10× substrate working solution 30 µL were mixed in each well of the microplate. For this experiment, five MCA substrates (2 for aminopeptidase, 2 for trypsin, and 1 for chymotrypsin) were used (**Table 1**). Calibration curves of AMC were generated for each assays conducted with 10 and 2% DMSO.

Activity measurements were compared between 2 and 1% DMSO concentrations with 200µM final substrate concentration. For this experiment, 15 substrates (**Table 1**) and low protein binding microplates from three suppliers (Nunc, Greiner, Sumitomo, as described above) were used. The seawater sample (180 µL) and 20 µL of 2 mM substrate working solution with 20 or 10% DMSO were mixed in each well of the microplate. Calibration curves of AMC were generated for each the assays conducted with 2 and 1% DMSO.

Previous studies used 2-methoxyethanol (MTXE, methylcellosolve) as a solvent of substrates to measure potential activities of extracellular hydrolytic enzymes in seawater (e.g., Fukuda et al., 2000). We also tested MTXE instead of DMSO as a solvent to dissolve the MCA substrates for protease assay. Substrate stock solutions (20 mM) were prepared with MTXE, and working solutions (10×) of each substrate, which contain 2 mM substrate and 20% MTXE, were prepared from the stock solution and autoclaved artificial seawater. A calibration curve of AMC with 2% MTXE was also prepared, and other procedures for fluorescence measurement and activities estimations were conducted as described above.

Significance of differences between 1 and 2% DMSO concentrations and different solvents (2% MTXE and 2% DMSO) were tested by Student's t-test.

#### Sample Containers for Seawater Collection

To test the effect of sampling container material, the following five types of containers were used: 500 mL polycarbonate bottles (PC500) (Nalgene), 50 mL polypropylene tubes supplied by Corning (PPC) and Eppendorf (PPE), 50 mL polyethylene terephthalate tubes (PET) (Corning), and low protein binding ProteosaveSS 50 mL tubes (SS) (Sumitomo Bakelite Co., Ltd.). The four 50 mL tubes had a similar shape of ordinary plastic centrifugal tubes. Natural seawater samples were collected by a bucket and immediately transferred to these containers through 50µm nylon mesh. All seawater samples were kept cool until measurements of proteases activities. Proteases activities were measured using five MCA substrates (2 for aminopeptidase, 2 for trypsin, and 1 for chymotrypsin) at several hours after sampling, and at 1 day (26 h) and 2 days (47 h) after sampling. Reaction volume for the assay was 200 µL (180 µL seawater sample + 20 µL substrate solution) in low protein binding microplate (Nunc), and the final concentrations in the mixture were 200µM substrate with 2% DMSO. Other procedures for fluorescence measurement and enzyme activities estimation were the same as described above.

### RESULTS AND DISCUSSION

### Microplates for Enzyme Activities Measurement in Seawater

#### Comparison of Activity Estimation by Methods A (Cuvette), B (Regular Microplate), and C (Low Protein Binding Microplate)

**Figure 1** shows the relationship hydrolytic activities measurements obtained by each method for the same samples. Data from unfiltered and 0.2µm filtered seawater samples were indicated as filled and opened symbols, respectively. Different shapes of the symbols in **Figure 1** refer to different types of enzyme activities estimated by using different substrates.

The protease activities measured by Method B (microplate reader with regular polystyrene microplates) were systematically lower than those by Method A (spectrofluorometer with cuvette; **Figure 1A**) and Method C (microplate reader with low protein binding microplates; **Figure 1C**). For the comparison of Method B to Method A, the linear relationship between measured activity for the two methods on all samples was significant [n = 64 (unfiltered and filtered samples were combined), r = 0.953, p < 0.0001] with a regression coefficient of 0.394 ± 0.015 (regression line for all data not shown in **Figure 1A**) indicated that the measured values obtained by Method B were only about 39% of those measured by Method A. The linear relationships between the results of Methods B and A were significant for both aminopeptidase activity (n = 20, r = 0.988, p < 0.0001) and trypsin-type activity (n = 36, r = 0.972, p < 0.0001). Regression line was not shown for chymotrypsin-type activity in **Figure 1A** because of the limited number of data points comparing with aminopeptidase and trypsin-type activity. The regression coefficient for trypsin-type activity (0.361 ± 0.014) was significantly smaller (t = 4.12, p < 0.0005) than that for aminopeptidase activity (0.643 ± 0.023). Although, there is not absolute evidence that higher estimation is more accurate, these results seem to suggest that Method B (microplate reader with regular polystyrene microplates) resulted in an underestimation of extracellular enzyme activity in natural seawater, and this effect was greater for trypsin-type activity than for aminopeptidase activity. Similarly, for the comparison of activities measurements by Method B and Method C (**Figure 1C**) shows that aminopeptidase activity and trypsin-type activities measured by Method B are 85 and 41%, respectively, of the measured values by Method C. For this comparison, the analytical methods of Methods B and C are identical and the

the standard deviations of triplicate sample preparations. Regression lines for aminopeptidase and trypsin-type endopeptidase and their equations are also shown. These regression analyses were performed with combined dataset from unfiltered and filtered samples. Dashed line indicates 1:1.

differences in measurement are attributable to the material of the microplates.

The linear relationship between the enzyme activities measurements obtained by Methods A and C was significant (n = 64, r = 0.968, p < 0.0001) and the regression coefficient was near 1 (1.015 ± 0.031), indicating that proteases activities measurements by these two methods are equivalent. Separate analyses for each of the enzyme types also were significant with regression coefficients for aminopeptidase and trypsintype activity of 0.805 ± 0.039 and 1.038 ± 0.044, respectively (**Figure 1B**); the difference between these two coefficients was not significant (t = 0.95, p > 0.35). These results indicated that the systematic underestimation of enzyme activities using Method B can be avoided by using low protein binding microplates.

#### Factors Causing Reduced Enzyme Activities Measurements with Regular Polystyrene Microplates

To clarify the discrepancies between the results obtained using regular microplates and low protein binding microplates, kinetic experiments were conducted based on the following hypotheses. If the observed underestimation is due solely to the adsorption of the artificial substrate and not due

(substrate for phosphatase) measured by Method C (low protein binding microplate) and Method B (regular polystyrene microplate). Error bars are the standard deviation of triplicate sample preparations (each sample and substrate pair was incubated and measured in 3 wells separately). Solid lines and dashed lines indicate curves fitting data from the low protein binding microplates and regular polystyrene microplates, respectively, to the Michaelis-Menten equation.

to the adsorption and deactivation of enzymes onto the surface of the polystyrene microplates, the measured activity by Method B should become saturated at higher level of the substrate than by Method C, and the differences between the measurements by Methods B and C should diminish at higher concentrations of substrate. On the other hand, if the reduced measurements are due to adsorption/deactivation of enzymes, the activities measurement by Method B should become saturated at almost the same level of substrate concentration as for Method C; namely, the Michaelis constant,

Km, should be at a similar value, and the maximum activity, Vmax, estimated by Method B should be smaller than that for Method C.

Our results show that aminopeptidase and trypsin-type activities estimated by Method B were lower than those by Method C even at higher substrate concentrations (**Figure 2**), although not much differences in chymotrypsin-type activity at the highest substrate concentration for Methods B and C. By curve fitting to the Michaelis–Menten equation, Vmax of aminopeptidase by Methods B and C was estimated to be 38.0 ± 5.6 and 50.1 ± 3.9 nmol L−<sup>1</sup> h −1 , respectively, for hydrolysis of Leu-MCA and 68.7 ± 3.0 and 89.3 ± 2.0 nmol L−<sup>1</sup> h −1 , respectively, for Ala-MCA. Vmax estimation of trypsin-type activity by Methods B and C was 58.2 ± 1.9 and 131.8 ± 4.6 nmol L−<sup>1</sup> h −1 , respectively, for hydrolysis of Boc-Phe-Ser-Arg-MCA and 68.3 ± 3.0 and 147.4 ± 2.5 nmol L−<sup>1</sup> h −1 , respectively, for hydrolysis of Boc-Leu-Ser-Thr-Arg-MCA. Thus, for aminopeptidase (**Figures 2A,B**) and trypsin-type activity (**Figures 2C,D**), Vmax by Method B were 76–77% and 44–46% of those by Method C, respectively. These results indicate that significant adsorption of enzyme itself onto the surface of the regular polystyrene microplate occurred and that it could result in the underestimation of protease activity in seawater, especially for trypsin-type enzymes.

If the reason for lower enzyme activity is due solely to adsorption of the enzyme, the Michaelis constant (half-saturation constant) should be the same for a sample by both assay methods. However, the K<sup>m</sup> estimated by curve fitting the trypsin-type activity data obtained by Methods B and C to the Michaelis– Menten equation were different: 30.1 ± 3.8 and 17.9 ± 1.6µM, respectively, for hydrolysis of Boc-Phe-Ser-Arg-MCA and 18.1 ± 2.6 and 8.4 ± 0.4µM, respectively, for hydrolysis of Boc-Leu-Ser-Thr-Arg-MCA. These differences could be explained by the adsorption of the substrates onto the regular polystyrene microplates and the available substrate concentration becoming lower than expected.

We also conducted the same kinetic experiment for phosphatase activity in terms of hydrolysis of MUF-phosphate to determine whether the discrepancy in measurement results between Methods B and C was specific to proteolytic enzymes, which is estimated by hydrolysis of oligopeptide analog MCA substrates. The results of phosphatase activity assays were almost the same as those of trypsin-type proteases activity tests with Boc-Phe-Ser-Arg-MCA and Boc-Leu-Ser-Thr-Arg-MCA. The phosphatase activity, Vmax of hydrolysis of MUF-phosphate, was 5.5 ± 0.4 nmol L−<sup>1</sup> h <sup>−</sup><sup>1</sup> by Method B, which is 45% of that measured by Method C (12.3 ± 0.4 nmol L−<sup>1</sup> h −1 ) for the same seawater sample (**Figure 2F**). The K<sup>m</sup> estimated by curve fitting to the data obtained from a regular plate and a low protein binding plate was 25.2 ± 6.2 and 12.8 ± 1.3µM, respectively.

From these results, we conclude that the reduced enzyme activities measurements with the use of regular polystyrene microplates (Method B) are attributable to the adsorption of both enzymes and substrates to the microplate surface. Although, the relationship between enzyme adsorption (binding) onto the polystyrene surface and deactivation of the adsorbed enzymes was not determined, Calliou et al. (2008) reported that adsorbed enzymes could be deactivated based on their model experiments. Among the proteases tested in our study, adsorption effects appeared to be more severe for trypsin-type enzymes than for aminopeptidases. A similar suppression of enzymatic activities measurements in seawater was previously reported during filtration (Obayashi and Suzuki, 2008b). In that case, trypsin-type enzyme appeared to be much more readily adsorbed onto the mixed cellulose esters filter (0.22µm pore size) than aminopeptidase; as a result, not only particles in the sample but also much of the dissolved trypsin were removed by filtration. Taking the results of previous and present studies together, we suppose that trypsin-type enzymes in seawater are more easily adsorbed on some kinds of solid surfaces and/or deactivated on solid surfaces than are aminopeptidases. These results imply that enzyme behaviors and characteristics in natural environment are different for extracellular aminopeptidases and trypsin-type endopeptidases.

#### Low Protein Binding 96-Well Microplates

**Figure 3** shows a comparison of estimated proteases activities in the same seawater sample using low protein binding, black, 96-well microplates from three suppliers. All three low protein binding microplates tested here gave similar measurements of hydrolytic activities of all tested substrates. Although, most experiments in present study were conducted using Nunc low binding microplates, Greiner Bio-one No-binding black 96-well microplates and SUMILON ProteosaveSS black microplates can provide equivalent results.

#### Solvent for Substrates DMSO Concentration in the Assay

Most MCA substrates need to be dissolved in organic solvent prior to mixing with the seawater sample. DMSO is a good solvent for dissolving MCA substrates; however, toxic effects to microbial cells in seawater and other unexpected effects could occur during the incubation and affect the measurement of extracellular enzyme activities in seawater samples. To minimize these types of artifacts, a lower concentration of organic solvent in the assay mixture is preferred. However, too low of a solvent concentration with a high concentration of substrate might result in solubility difficulties during the assay.

Michaelis plots for assays conducted with 10 and 2% DMSO with various substrate concentrations of five selected substrates (2 for aminopeptidase, 2 for trypsin, and 1 for chymotrypsin) are shown in **Figure 4**. Aminopeptidase and chymotrypsin-type activities were estimated lower in 10% DMSO than in 2% DMSO, irrespective of substrate type. In the case of the trypsin-type activity, hydrolysis rates in 2% DMSO were higher than in 10% DMSO for lower concentration of substrate, while rates of both were at the same level (hydrolysis of Boc-Leu-Ser-Thr-Arg-MCA, **Figure 4D**), or the rate in 2% DMSO was lower than that in 10% DMSO condition (hydrolysis of Boc-Phe-Ser-Arg-MCA, **Figure 4C**) at higher concentration of substrate. Previous studies have reported that a large proportion of aminopeptidase activity in seawater was detected from the bacterial cell size fraction as ectoenzymes, while trypsin-type activities were mostly detected in the dissolved (<0.2µm filtered) fraction (Karner and Rassoulzadegan, 1995; Hoppe et al., 2002; Obayashi and Suzuki, 2008a; Bong et al., 2013). Higher contributions of bacterial cellassociated fractions of chymotrypsin-type activity were reported in some cases (Bong et al., 2013). The difference in the apparent DMSO effects between trypsin-type and other enzymes could be due to predominant state of existence in seawater: Lower activity of aminopeptidase and chymotrypsin-type enzymes in 10% DMSO in this study might reflect the toxic effects of DMSO to microbial cells during the incubation. Taking the Michaelis plot for the assay with Boc-Leu-Ser-Thr-Arg-MCA as a substrate (**Figure 4D**), 10% DMSO appears to act as a competitive inhibitor for dissolved trypsin-type enzyme. The reason for the discrepancy between the results of hydrolysis of two substrates for trypsin (Boc-Phe-Ser-Arg-MCA and Boc-Leu-Ser-Thr-Arg-MCA) at higher concentration of substrates was not clear; however, it might be related to the solubility of Boc-Phe-Ser-Arg-MCA as explored below.

Among the 17 tested MCA substrates listed in **Table 1**, Z-Phe-Arg-MCA (substrate for trypsin) and Suc-Leu-Leu-Val-Tyr-MCA (substrate for chymotrypsin) were not soluble enough to use with 2% DMSO, and additionally, Arg-MCA (substrate for aminopeptidase), and Boc-Phe-Ser-Arg-MCA (substrate for trypsin) could not be used with 1% DMSO. These substrates produced visible precipitates or aggregates in the 10× working solution or upon mixing with seawater samples for assay.

Excluding these four substrates, which were not sufficiently soluble, we compared the enzyme activities measurements in assays with 2 and 1% DMSO (**Figure 5**). For this test, the final concentration of each substrate was set to 200µM, which is the saturation level for most of the tested substrates. Hydrolytic activity results for 2 and 1% DMSO were similar and the differences were not significant (p = 0.295 for all pairs, n = 39; p = 0.274 for aminopeptidase, n = 12; p = 0.256 for trypsin-type, n = 24; p = 0.965 for chymotrypsin-type, n = 3). Steen et al. (2015) examined the effect of DMSO on aminopeptidase (Leu-MCA hydrolysis) kinetics in river water samples and reported that DMSO at 4% or less did not influence the estimation of K<sup>m</sup> but use of 5% DMSO produced different results.

Although, the actual effects of DMSO might differ among the enzymes, estimation of hydrolysis for many MCA substrates with 2% DMSO seemed to provide reliable estimation of protease activity in environmental seawater samples. For substrates with high solubility throughout the assay, a lower percentage of DMSO should be acceptable for measuring enzyme activity in seawater samples.

#### Comparison of Using 2% DMSO and 2% MTXE in Assay

In previous studies, MTXE was used to prepare stock solutions of hydrolytic substrates. Here, we compared the enzyme activities measurements in assays with the only difference being the use of solvents 2% DMSO and 2% MTXE. Among the MCA substrates tested, two could not be used with MTXE: Suc-Leu-Leu-Val-Tyr-MCA (for chymotrypsin) did not dissolve in MTXE for preparation of the stock solution, and Arg-MCA (for aminopeptidase) appeared as a suspension in the 10× working solution (2 mM substrate, 20% MTXE in autoclaved artificial seawater). **Figure 6** shows a comparison of hydrolytic activity for 14 MCA substrates (**Table 1**) in seawater samples with assay solutions of 2% DMSO and 2% MTXE and a 200µM final substrate concentration. Hydrolytic activity was nearly the same by both methods, and the differences were not significant (p = 0.948 for all pairs, n = 42; p = 0.905 for aminopeptidase, n = 12; p = 0.963 for trypsin-type, n = 27; p = 0.323 for chymotrypsin-type, n = 3).

To assess the potential extracellular protease activities in natural seawater, the use of both 2% DMSO and 2% MTXE in assay give similar results for the hydrolysis of most MCA substrates with a final substrate concentration of 200µM.

#### Assessment of Sample Container Material

Underestimation of hydrolytic enzyme activity in seawater using regular polystyrene microplates in the assay implies that similar concern is needed for the water sample container used to store seawater samples until assay. In general, adsorption effects are less for larger volume containers. While keeping seawater samples in smaller volume containers is convenient, the sample can be easily affected by differences among materials of the containers. For samples stored five types of container materials (**Figure 7**), the estimated activities in seawater were different among the type of sample containers. At the first measurement, several hours after seawater collection, activity in

the samples kept in 50 mL regular polypropylene tubes (PPC and PPE) was already lower than in the others, and the activities continued to decrease greatly after 1 day (at 26 h) and 2 days (at 47 h). Differences in the estimated activities of trypsintype and chymotrypsin-type enzymes among the different tubes were greater than those of the aminopeptidase, and these trends corresponded with the effects observed for microplates (**Figure 1**) and from filters for size fractionation reported in Obayashi and Suzuki (2008b). The seawater sample stored in the ProteosaveSS 50 mL tube (SS), which has a hydrophilic polymer coating designed to reduce nonspecific adsorption of protein and peptide to the inside of the tube, showed similar results with the sample stored in 500 mL polycarbonate bottle (PC500), while the sample kept in the 50 mL polyethylene terephthalate tube (PET) showed lower trypsin- and chymotrypsin-type enzyme activities than water samples stored in PC500 and SS but higher than those in polypropylene tubes (PPC and PPE).

These results show that protease activity in seawater samples stored in small polypropylene tubes decreased rapidly, causing the underestimation of natural activities, especially for trypsintype and chymotrypsin-type endopeptidases. The choice of containers for water samples is an important consideration, even

FIGURE 5 | Scatter plots of hydrolytic activities measured in 2 and 1% DMSO on (A) Nunc low binding microplate, (B) Greiner Bio-one No-binding microplate, and (C) SUMILON Proteosave microplate. Error bars are the standard deviation of triplicate sample preparations (same sample incubated in 3 wells). Dashed line indicates 1:1.

when assays are conducted as soon after sampling as possible, for assessing the level of enzyme activity in natural environmental seawater samples. In general, adsorption of organic molecules onto glassware is thought to be less than that on plastics, especially if the glass surfaces are silanized. We did not test glassware in this study, however, glass vials could be thought as a good sample container, depending on the research purposes.

## CONCLUSIONS

Dashed line indicates 1:1.

It is thought that various enzymes in nature have a range of characteristics and behaviors in natural aquatic environments. Some enzymes in seawater are easily adsorbed onto the surfaces of some materials, and that may cause artificial effects or biases in the enzyme activities measurement. To assess the actual natural activities of microbial extracellular hydrolytic enzymes in aquatic ecosystems, materials used for both sampling and measurement assays should be carefully selected.

Water sample containers must maintain enzyme activity in the natural seawater samples for the short-term; protease activities in seawater decreased rapidly in small volume polypropylene tubes. Using fluorogenic substrates and a fluorometric microplate reader with low protein binding, black, 96-well microplates was effective for obtaining high-resolution and reliable measurements of hydrolytic enzyme activities in small volume (180 µL) of seawater samples, while regular polystyrene microplates showed significant underestimation of activities, especially for trypsin-type proteases.

For measuring the potential activities of extracellular proteases in seawater, a final substrate concentration at 200µM in the assay (seawater sample with substrate solution) appeared to be a good saturation level for most of the tested oligopeptide analog MCA substrates for aminopeptidase, trypsin, and chymotrypsin. Stock solutions of MCA substrate are usually dissolved in a solvent, and both 1 or 2% DMSO and 2% MTXE in assay provided similar and reliable activities measurements, except for some

substrates which were not sufficiently soluble. Calibration curves of AMC (product of substrate hydrolysis) should be generated under the same conditions as are used for the sample measurement. A solvent blank (sample with solvent without substrate) and an inactivated control (autoclaved seawater) should be also prepared and assayed with the samples for reliable calculation of hydrolytic enzyme activity in the sample.

Substrate concentrations at saturation level for assay of potential activity and substrate solubility may depend on sample type, targeted enzyme, and its substrate. It is important to optimize these factors for the sample types and target enzymes to obtain data that are as reliable as possible.

### AUTHOR CONTRIBUTIONS

YO designed and conducted all experiments, analyzed data, constructed discussion, and prepared the manuscript; CWB

#### REFERENCES


conducted portions of the experiments with YO; and SS supported all stages of the research and contributed to preparation of the manuscript.

#### ACKNOWLEDGMENTS

We thank H. Onishi, S. Nakano, and his colleagues for their kind cooperation in the sample collection. This work was supported in part by JSPS KAKENHI Grant Numbers JP26450245 and JP24710005, and grants from the government to national university corporations for Joint Usage/Research Center, MEXT, Japan.

Gulf of Mexico Mississippi River plume. Mar. Chem. 177, 398–407. doi: 10.1016/j.marchem.2015.06.021


**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 Obayashi, Wei Bong and Suzuki. 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.

# An Experimental Insight into Extracellular Phosphatases – Differential Induction of Cell-Specific Activity in Green Algae Cultured under Various Phosphorus Conditions

Jaroslav Vrba1,2 \*, Markéta Macholdová<sup>3</sup> , Linda Nedbalová<sup>3</sup> , Jirí Nedoma ˇ <sup>2</sup> and Michal Šorf1,4

<sup>1</sup> Department of Ecosystem Biology, Faculty of Science, University of South Bohemia, Ceské Bud ˇ ejovice, Czechia, ˇ 2 Institute of Hydrobiology, Biology Centre CAS, Ceské Bud ˇ ejovice, Czechia, ˇ <sup>3</sup> Department of Ecology, Faculty of Science, Charles University, Prague, Czechia, <sup>4</sup> Department of Zoology, Fisheries, Hydrobiology and Apiculture, Faculty of AgriSciences, Mendel University, Brno, Czechia

#### Edited by:

Sonja Endres, Max Planck Institute for Chemistry (MPG), Germany

#### Reviewed by:

Monika Nausch, Leibniz Institute for Baltic Sea Research (LG), Germany Michael R. Twiss, Clarkson University, United States

> \*Correspondence: Jaroslav Vrba jaroslav.vrba@prf.jcu.cz

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 12 October 2017 Accepted: 06 February 2018 Published: 21 February 2018

#### Citation:

Vrba J, Macholdová M, Nedbalová L, Nedoma J and Šorf M (2018) An Experimental Insight into Extracellular Phosphatases – Differential Induction of Cell-Specific Activity in Green Algae Cultured under Various Phosphorus Conditions. Front. Microbiol. 9:271. doi: 10.3389/fmicb.2018.00271 Extracellular phosphatase activity (PA) has been used as an overall indicator of P depletion in lake phytoplankton. However, detailed insights into the mechanisms of PA regulation are still limited, especially in the case of acid phosphatases. The novel substrate ELF97 phosphate allows for tagging PA on single cells in an epifluorescence microscope. This fluorescence-labeled enzyme activity (FLEA) assay enables for autecological studies in natural phytoplankton and algal cultures. We combined the FLEA assay with image analysis to measure cell-specific acid PA in two closely related species of the genus Coccomyxa (Trebouxiophyceae, Chlorophyta) isolated from two acidic lakes with distinct P availability. The strains were cultured in a mineral medium supplied with organic (beta-glycerol phosphate) or inorganic (orthophosphate) P at three concentrations. Both strains responded to experimental conditions in a similar way, suggesting that acid extracellular phosphatases were regulated irrespectively of the origin and history of the strains. We found an increase in cell-specific PA at low P concentration and the cultures grown with organic P produced significantly higher (ca. 10-fold) PA than those cultured with the same concentrations of inorganic P. The cell-specific PA measured in the cultures grown with the lowest organic P concentration roughly corresponded to those of the original Coccomyxa population from an acidic lake with impaired P availability. The ability of Coccomyxa strains to produce extracellular phosphatases, together with tolerance for both low pH and metals can be one of the factors enabling the dominance of the genus in extreme conditions of acidic lakes. The analysis of frequency distribution of the single-cell PA documented that simple visual counting of 'active' (labeled) and 'non-active' (non-labeled) cells can lead to biased conclusions regarding algal P status because the actual PA of the 'active' cells can vary from negligible to very high values. The FLEA assay using image cytometry offers a strong tool in plankton ecology for exploring P metabolism.

Keywords: acid phosphatase, Coccomyxa, ELF97 phosphate, FLEA technique, image cytometry, inorganic phosphorus, organic phosphorus, phosphorus limitation

## INTRODUCTION

fmicb-09-00271 February 19, 2018 Time: 14:51 # 2

Phosphorus (P) has been proven to be a limiting resource in many aquatic ecosystems (Schindler, 2012; Schindler et al., 2016). Aquatic microorganisms, except for phagotropic protists, can only assimilate dissolved inorganic P, i.e., dissolved orthophosphate (Pi) (Reynolds, 1997). Yet P<sup>i</sup> also reacts with and adsorbs to various compounds or seston particles (e.g., clay) that may sediment and ultimately reduce the availability of P in the epilimnion and euphotic zone. Therefore, P<sup>i</sup> is a subject of more or less severe competition in the planktonic microbial community, encompassing not only individual phytoplankton species (Sommer, 1981, 1985), but also bacterioplankton (Currie and Kalff, 1984; Cotner and Wetzel, 1992). On the other hand, plankton consumers may regenerate substantial amounts of P<sup>i</sup> into the water column (e.g., Knoll et al., 2016). Such a consumer driven nutrient recycling often results in dissolved organic P (DOP) forms that are not readily available to microorganisms. The DOP compounds need to be cleaved by extracellular enzymes before they can be taken up by microbial cells (Cembella et al., 1984; Jansson et al., 1988; Cotner and Wetzel, 1991).

In the light of this, several artificial chromogenic or fluorogenic substrates have been used for regular measurements of the extracellular phosphatase activity (Jones, 1972; Healey and Hendzel, 1979; Hoppe, 1983), increased level of which in lake water was proposed to indicate P deficiency in lake phytoplankton (Healey and Hendzel, 1980). By adding an artificial DOP substrate to a water sample and to its cellfree filtrate, total and free (dissolved) phosphatase activities are measured, respectively. The free, or dissolved, activity represents the bulk activity of all free (dissolved) enzymes, both of microbial and metazoan origin (Boavida and Heath, 1984; Carr and Goulder, 1990). The particulate activity, calculated as the difference of total and free activity, represents the bulk activity of both microbial ectoenzymes and free enzymes adsorbed to particles (Wetzel, 1991; Nagata and Kirchman, 1992). It is sometimes possible to estimate proportions of 'bacterial' and 'algal' ectoenzymes by using a more detailed size fractionation of water samples (Vrba et al., 1993; Nedoma et al., 2006); however, none of the widely-used substrates allows for enzyme localization or detection of phosphatase producers, which is the serious methodological drawback of the bulk phosphatase assay.

A new generation of fluorogenic substrates, such as ELF <sup>R</sup> 97 phosphate (ELFP) based on 2-(2<sup>0</sup> -phosphoryloxyphenyl)-4- (3H)-quinazolinone (Huang et al., 1992), can overcome most disadvantages. Insoluble precipitates of the hydrolysis product (ELF <sup>R</sup> 97 alcohol, ELFA) at the sites of hydrolysis (under certain conditions – see below) allow for direct visualization of the active enzymes in organisms (cells) by epifluorescence microscopy. An early application of the ELF <sup>R</sup> 97 Endogenous Phosphatase Detection Kit successfully visualized phosphatase-positive cells in both algal cultures and natural phytoplankton, i.e., directly tagged the P-limited algal species (González-Gil et al., 1998; Dyhrman and Palenik, 1999; Rengefors et al., 2001), although further studies revealed some uncertainties and/or potential misinterpretations (e.g., Rengefors et al., 2003; Dignum et al., 2004; Ou et al., 2010). The principle shortcoming of the ELF <sup>R</sup> 97 method as it is applied in recent studies, including the original paper by González-Gil et al. (1998), is that only the occurrence of ELFA-labeling, i.e., merely qualitative estimates of phosphatasepositive algal cells and/or species, presence/absence of tagged phytoplankton, etc., could be reported (Štrojsová et al., 2003; Cao et al., 2005; Rychtecký et al., 2015; Ren et al., 2017). The only quantification of phosphatase activity accessible within the limits of this original method is to score a percentage of ELFA-labeled cells (e.g., Rengefors et al., 2003; Dyhrman and Ruttenberg, 2006; Litchman and Nguyen, 2008; Young et al., 2010).

This problem has been largely solved by using ELF <sup>R</sup> 97 phosphate (ELFP) according to a modified protocol (Nedoma et al., 2003b; Štrojsová et al., 2003) derived from a common fluorescence assay (e.g., Hoppe, 1983) for extracellular activity in plankton and further standardized by buffering the samples (Štrojsová and Vrba, 2006). Inhibition experiments suggested that both substrates (i.e., ELFP and 4-methylumbelliferyl phosphate) were hydrolyzed by the same extracellular phosphatases (Štrojsová et al., 2003). This protocol, referred to as fluorescence-labeled enzyme activity (FLEA) assay, allows not only for distinguishing between enzymatically active and inactive specimens in a sample, but, most essentially, for the quantification of the ELFA fluorescence at single cell or species level using image cytometry (Nedoma et al., 2003b; Nedoma and Vrba, 2006; Novotná et al., 2010). This cell-specific fluorescence intensity can be further converted to a specific rate of enzymatic ELFP hydrolysis by the particular producers.

Since decades ago, realistic interpretation and sometimes contradictory results of various phosphatase assays remains a subject of discussions in plankton ecology (e.g., Berman et al., 1990; Nedoma et al., 2003a). There is no doubt, at present, that the bulk extracellular phosphatase activity must not be interpreted as exclusively algal activity (e.g., Hoppe, 2003; Cao et al., 2005; Nedoma et al., 2006). The paradigm of phosphatase expression only under P deficiency is overly simplistic as algae may constitutively express some phosphatase activity, but also may not efficiently regulate it in response to P availability (Young et al., 2010). There is also increasing awareness that all phytoplankton species do not react uniformly to P depletion (e.g., Litchman and Nguyen, 2008) and many species indeed do not produce extracellular phosphatases at all under such circumstances, while other species can exhibit constitutive activity (e.g., Rengefors et al., 2001, 2003; Štrojsová et al., 2003, 2005, 2008; Rychtecký et al., 2015). For instance, ELFA-labeled, i.e., phosphatasepositive phytoplankton species were reported from eutrophic lakes under high concentrations of soluble reactive P (SRP) (e.g., Cao et al., 2005, 2009). Most data, however, have been obtained from field studies. Laboratory experiments focused on the influence of P form and concentration on cell-specific phosphatase activity under controlled conditions are scarce and performed entirely in batch cultures (Huang et al., 2000; Young et al., 2010; Ren et al., 2017).

In this study, we examined two closely related algal species isolated from two acidic lakes differing in their P concentrations. Each algal population had been exposed to distinct environmental conditions for decades. We tested the ability of both species to grow on inorganic or organic P in a

semi-continuous system, and the response of single algal cells to various degree of P depletion. Cell-specific acid phosphatase activity was measured using the FLEA assay according to the protocol, which enabled us to quantify more accurately its variability in individual experimental treatments. The main objectives of this study were to determine (i) if expression of algal acid phosphatases is under environmental control, (ii) if the manner of the control differs in the isolates originating from environments contrasting in P availability, and (iii) if acid phosphatase activity reflects actual needs of algal cells given by their growth rate and source of P.

#### MATERIALS AND METHODS

#### Algal Cultures

We isolated two unialgal cultures of Coccomyxa strains (Trebouxiophyceae, Chlorophyta) by serial dilution from the plankton of two acidic lakes of distinct trophic status in Czechia. Lake Plešné (48◦ 460 35<sup>00</sup> N, 13◦ 510 55<sup>00</sup> E; 1087 m a.s.l.) is of glacial origin and it was strongly acidified due to atmospheric sulfur and nitrogen deposition that peaked in the 1980s (Vrba et al., 2003). In this acidic (pH = 4.8–5.5) mesotrophic lake, P availability remains largely impaired by reactive aluminum (Vrba et al., 2006), with mean epilimnetic SRP concentrations as low as ∼40 nmol L−<sup>1</sup> (Novotná et al., 2010). Its P-limited phytoplankton are dominated by coccoid green algae (formerly misidentified as Monoraphidium dybowskii; cf. Štrojsová and Vrba, 2006, 2009) that was recently described as a new species, Coccomyxa silvae-gabretae (Barcyte and Nedbalová, 2017 ˙ ). Eutrophic Lake Hromnice (49◦ 510 03<sup>00</sup> N, 13◦ 260 39<sup>00</sup> E; 330 m a.s.l.) is a former pyritic shale mine. Its lake water is characterized by extremely low pH (2.3−2.9), high concentrations of P (1−52 µmol L−<sup>1</sup> SRP) and several metals (Al, Fe, Mn, Ni, Cu, Co, and Pb) (Hrdinka et al., 2013). A common phytoplankton species in Lake Hromnice is Coccomyxa elongata (Barcyte and ˙ Nedbalová, 2017). We maintained non-axenic cultures of the two strains in an acidified BBM medium (Bischoff and Bold, 1963), with the pH adjusted to 4, at room temperature and daylight.

For all phosphatase experiments, we cultivated both Coccomyxa strains in semi-continuous, turbidostatic systems, in the acidified BBM medium supplied with distinct P sources at three concentrations (see below), at room temperature and permanent light provided by fluorescent tubes (photosynthetically active radiation ∼40 µmol s−<sup>1</sup> m−<sup>2</sup> ). We used 0.5-L conical vessels (separatory funnels with stopcock for easy sampling), filled with 200 ml of medium and inoculated with 0.5 ml of stock culture at the beginning of each experiment. The medium as well as cultivation vessels were sterilized. Continuous aeration by sterile air bubbling into the bottom of each vessel ensured both CO<sup>2</sup> saturation and mixing of algal suspension. To provide merely inorganic (hereafter referred as I) or organic (hereafter referred as O) P sources, we supplied the BBM medium with P<sup>i</sup> or β-glycerol phosphate (β-GP) as the single source of P, respectively. For either source, we used one P-replete (variants I1 and O1) and two P-depleted (variants I2−I3 or O2−O3) media with the original concentrations adjusted to 858, 16 and 10 µmol L−<sup>1</sup> of P.

We ran all experimental variants in triplicates for 3 weeks. We regularly screened all variants for chlorophyll a concentration using a fluorometer (TD-700 Laboratory Flurometer, Turner Designs, San Jose, CA, United States) and diluted the cultures by the corresponding fresh medium at regular intervals, i.e., three times during the cultivation, to maintain chlorophyll a concentrations close to ∼10 µg L−<sup>1</sup> in P-depleted (I2/O2 and I3/O3) and to 50 µg L−<sup>1</sup> in P-replete (I1/O1) variants. In addition, in the middle and at the end of each experiment, we checked all variants for residual P concentrations in cultures – SRP was determined by the molybdate method after filtering the samples through glass fiber filters (0.7 µm, Macherey-Nagel, Düren, Germany). After the 3-week cultivation, we sampled all replicates to estimate cell-specific phosphatase activity of individual Coccomyxa populations in each experimental variant.

For each cultivation, we further calculated a specific growth rate (µ, day−<sup>1</sup> ) of the individual Coccomyxa population for the period between the second and third dilutions according to the equation:

$$\mu = \frac{\ln \text{N}\_{\text{f}} - \ln \text{N}\_{\text{i}}}{\text{t}\_{\text{f}} - \text{t}\_{\text{i}}}$$

where N is the final (f) or initial (i) cell density at time (t). A conversion curve was used to calculate N from chlorophyll fluorescence values.

#### Cell-Specific Phosphatase Activity (FLEA Assay)

After the 3-week cultivations, we employed the protocol for FLEA assay (Nedoma et al., 2003b) to estimate extracellular cellspecific phosphatase activity of the Coccomyxa strains grown on different P sources. We incubated 5-ml samples with fluorogenic substrate ELFP (Molecular Probes; Invitrogen, Eugene, OR, United States). The incubation started by the addition of ELFP solution (final concentration of 20 µmol L−<sup>1</sup> ) and lasted 3 h at room temperature and daylight. Then, each incubation was terminated by filtering 1-ml subsamples over mild vacuum (<20 kPa) through polycarbonate membrane filters (pore size 2 µm; Osmonics, Minnetonka, MN, United States). The filter with retained algae was placed on a microscopic slide, embedded with immersion oil, covered with a coverslip, and preserved in a freezer at −20◦C until the image cytometry analysis (cf. Nedoma et al., 2003b).

#### Image Cytometry

The image analysis system used for ELFA fluorescence quantification included the fluorescence microscope Nikon Eclipse 90i (Nikon, Tokyo, Japan; Nikon Plan Fluor 60×), monochromatic digital camera (Andor Clara, Andor Technology, Ltd., Belfast, United Kingdom), and the software NIS-Elements 4.12 (Laboratory Imaging, s.r.o., Prague, Czechia). From every slide, 30 image files corresponding to 30 randomly selected microscope fields were made. Each image file contained two types of images from two channels (**Figure 1**) – one was captured with ELFA-fluorescence-specific filter cube

(excitation/emission: 360–370 nm/520–540 nm) and served for the measurement of cell-associated ELFA fluorescence; the second image was captured with chlorophyll-autofluorescencespecific filter cube (excitation/emission: 510–550 nm/>590 nm) and served for cell localization and sizing (see below). In NIS-Elements software, 3−6 randomly chosen cells (90−180 from one slide) were demarcated (segmented) manually on the chlorophyll-fluorescence image. The system then measured cell dimensions and the mean gray level of the cell and of the ELFA image background. Cell-associated ELFA fluorescence [FELFA, in relative fluorescence units (FUs) cell−1h −1 ] was then calculated using the following equation (Nedoma et al., 2003b):

$$\text{FELFA} = \frac{\text{Area} \times (\text{MGrey} - \text{BgMGrey})}{\text{T}\_{\text{exp}}} \times \text{F}\_{\text{cal}}$$

where Area (µm<sup>2</sup> ) is projected area of the cell, MGrey (dimensionless) is mean gray of the cell, BgMGrey (dimensionless) is mean gray of the background, Fcal (dimensionless) is fluorescence calibration factor, and Texp (ms) is exposure time. For rough estimation of the cell-specific phosphatase activity (in the units of fmol cell−1h −1 ), we used the conversion factor of 0.1 fmol FU−<sup>1</sup> , based on experiments with Plešné Lake natural plankton (for details see Nedoma et al., 2003b).

Mean cell volumes of individual Coccomyxa populations in each replicate were calculated using the measured cell dimensions, i.e., cell length and area from the chlorophyllfluorescence images, by approximation of cell shape to an ellipsoid.

#### Statistical Analyses

A three-way ANOVA with a post hoc Tukey HSD test of differences among experimental variants were performed to test the effects of P sources, P concentrations, species of Coccomyxa, and their interactions on algal growth rate, mean cell size, and cell-specific phosphatase activity. All data were transformed by log (x+1) to meet the assumptions of ANOVA. All analyses were performed using Statistica 13.2 (Dell Inc., 2016).

#### RESULTS

The two strains of Coccomyxa species revealed very similar results and generally responded in a consistent way to all experimental treatments. No significant differences in either of the treatments were detected between both species (**Tables 1**–**3**). Final residual concentrations averaged at around 600 µmol L−<sup>1</sup> of SRP in P-replete cultures grown on inorganic medium (P<sup>i</sup> , I1), whereas they leveled at ∼30 µmol L−<sup>1</sup> of SRP in those grown on organic medium (β-GP, O1). Yet, in all P-depleted variants, these concentrations were very similar, on average 2–6 µmol L−<sup>1</sup> of SRP. In the organic media, β-GP was obviously transformed into SRP in all treatments.

We found the highest growth rates (0.17–0.18 day−<sup>1</sup> ) in P-replete P<sup>i</sup> medium (variants I1), whereas they were significantly lower (0.06–0.10 day−<sup>1</sup> ) in both P-depleted variants

TABLE 1 | Results of three-way ANOVA testing the effects of species (C. elongata vs. C. silvae-gabretae), media (inorganic vs. organic), and three P concentrations on growth rate.


Significant differences (P < 0.05) are given in bold; ×, interaction of factors.

TABLE 2 | Results of three-way ANOVA testing the effects of species (C. elongata vs. C. silvae-gabretae), media (inorganic vs. organic), and three P concentrations on cell volume.


Significant differences (P < 0.05) are given in bold; ×, interaction of factors.

TABLE 3 | Results of three-way ANOVA testing the effects of species (C. elongata vs. C. silvae-gabretae), media (inorganic vs. organic), and three P concentrations on cell-specific phosphatase activity.


Significant differences (P < 0.05) are given in bold; ×, interaction of factors.

(I2 and I3; **Figure 2**). Moreover, both the species reached significantly lower growth rates in media with β-GP (0.13–0.15 and 0.03–0.10 day−<sup>1</sup> , respectively) compared to their P<sup>i</sup> counterparts (**Table 1**), while keeping the same descending trends from P-replete to depleted variants (O1–O3; **Figure 2B**). Notwithstanding the P source and species, all P-depleted cultures revealed significantly larger mean cell volumes (38–52 µm<sup>3</sup> ) than those that were P-replete (23–31 µm<sup>3</sup> ; **Figure 3** and **Table 2**).

In both P-replete media, we detected negligible ELFA labeling (**Figures 4A,B**) in both Cocomyxa cultures. We therefore estimated close to zero cell-specific phosphatase activity (relative FU cell−1h −1 ) in every replicate of the I1 and O1 variants (**Figure 5**). On the contrary, we detected substantial phosphatase activity in all P-depleted cultures. While its increase, compared to the corresponding P-replete variant, was lower in P-depleted P<sup>i</sup> media and significant only in the I3 variant, all P-depleted cultures grown with β-GP exhibited bright fluorescence (**Figures 4C–F**). Moreover, the cell-specific phosphatase activities in O2 and O3 treatments exceeded those in I2 by one order of magnitude (**Figure 5**). Three-way ANOVA confirmed highly significant effects of both P source and P concentration, as well as their interaction (see, respectively, factors M and P in **Table 3**). This interaction reflected different responses to the P source concertation in the P<sup>i</sup> and β-GP cultures (cf. **Figures 5A,B**).

We further analyzed the frequency distributions of the cellspecific phosphatase activities measured in all replicates of each treatment. In general, we did not find any remarkable

FIGURE 2 | Comparison of growth rates of the Coccomyxa cultures in P-replete (1) and P-depleted (2 and 3) media with either inorganic (I; top: A) or organic (O; bottom: B) P source. Columns are means, bars represent SDs; note that scales are the same. Differences among treatments were tested using three-way ANOVA with a post hoc Tukey HSD test; lower case letters above the columns (a–f) indicate significant differences among treatments (see Table 1 for summary statistics).

difference in the distribution patterns among the two Coccomyxa species tested (**Figure 6**). In the P-replete cultures, most of the algal cells (nearly 100% in I1 and almost 80% in O1) exhibited negligible activity (<0.02 FU cell−1h −1 ). Unlike in I1 variants, up to ∼20% of algae in the O1 cultures exhibited low activity (<0.64 FU cell−1h −1 ). On the contrary, we observed low percentage of such weakly ELFAlabeled and/or inactive cells with very similar distribution patterns in all P-depleted β-GP cultures (cf. O2 and O3 in **Figures 6C,D**). The P-depleted P<sup>i</sup> and β-GP cultures, however, showed very different distribution patterns in two aspects: (i) the maximum in the histogram of single-cell phosphatase activities was notably shifted toward higher activities in β-GP compared to P<sup>i</sup> cultures (peaking around 0.6 and 2.5 FU cell−1h −1 , respectively), and (ii) in P<sup>i</sup> cultures, the moderate P-depletion (I2) resulted in a flat and uniform frequency distribution limited to the region of low activities (<1.26 FU cell−1h −1 ), whereas the high P-depletion (I3) induced clear maximum between 0.32 and 1.26 FU cell−1h −1 (**Figures 6A,B**).

At comparable growth rates, the cell-specific phosphatase activities were roughly 5–10 times higher in the variants with β-GP compared to P<sup>i</sup> as phosphorus source. The relationship between growth rate and phosphatase activity was similar in both Coccomyxa species examined (**Figure 7**).

FIGURE 3 | Comparison of mean cell volume of the Coccomyxa cultures in P-replete (1) and P-depleted (2 and 3) media with either inorganic (I; top: A) or organic (O; bottom: B) P source. Box and whisker plots show medians (bar), 25 and 75% quartiles (box), and 10–90% percentiles (whiskers); note same scales. Differences among treatments were tested using three-way ANOVA with a post hoc Tukey HSD test; lower case letters above the columns (a, b) indicate significant differences among treatments (see Table 2 for summary statistics).

#### DISCUSSION

Our results clearly suggest that both tested Coccomyxa species, although their original populations had lived in acidic lakes with the contrasting P availability for decades, possessed the same ability to produce acid extracellular phosphatases. We can speculate that the absence of a genomic adaptation to high P concentrations in C. elongata indicates that the production of acid phosphatases represent an evolutionarily conservative trait of vital importance for the acidotolerant algae. These phosphatases were inducible ectoenzymes (cf. Chróst, 1991), exclusively produced in all P-depleted cultures, while their production in the P-replete variants (I1 or O1) was negligible. Some early studies considered alkaline phosphatases as inducible and acid phosphatases as constitutive (Cembella et al., 1984; Jansson et al., 1988). In contrast, individual phytoplankton species exhibited, depending on circumstances, zero to extreme acid phosphatase activity per cell in chronically P limited acidic lakes, indicating that these ectoenzymes were inducible too (Štrojsová and Vrba, 2009; Novotná et al., 2010). Our results in this study suggested that acid phosphatases in Coccomyxa species were regulated in the same manner as it is known for alkaline phosphatases (e.g., Jansson et al., 1988).

Surprisingly, the relatively well-growing O1 cultures, grown entirely with organic P source, produced very little phosphatases. Most likely, some non-enzymatic hydrolysis of β-GP could liberate enough P<sup>i</sup> for algal growth, as suggested by the residual SRP concentrations (∼30 µmol L−<sup>1</sup> ) observed in this P-replete medium (O1). The high β-GP concentration could also saturate the enzymes to such a degree that the ELFP substrate was outcompeted during the assay (likewise glucose-6-phosphate and 4-methylumbelliferyl phosphate inhibited the ELFP hydrolysis; cf. Figure 1 in Štrojsová et al., 2003). Consequently, just <20% of algal cells showed weak ELFA labeling (**Figure 6**). Moreover, we could not exclude some β-GP hydrolysis by bacterial extracellular enzymes (cf. Siuda and Chróst, 2001) as the algal cultures were not axenic. Such an enzymatic activity, however, would not interfere with the FLEA assay, which specifically quantifies relative fluorescence of individual algae (**Figure 1**). Furthermore, hardly any bacterial or free activity would be retained on the filter used (2-µm pore size); indeed very few such ELFA precipitates were observed by epifluorescence microscopy (**Figure 4B**).

Cell-specific phosphatase activities were almost an order of magnitude higher with β-GP (O2 and O3) compared to those with P<sup>i</sup> (I2 and I3) (cf. the different scales in **Figures 5A,B**) and the growth rates with P<sup>i</sup> were slightly but significantly higher than those with β-GP (**Figure 2** and **Table 1**). Similar responses were recently reported also by Ren et al. (2017), who cultured algal (Chlorella pyrenoidosa and Pseudokirchneriella subcapitata) or cyanobacterial (Microcystis aeruginosa) species with various P sources in axenic batch cultures. Similarly to our study, both green algae (C. pyrenoidosa and P. subcapitata) and cyanobacteria grew faster with P<sup>i</sup> than with β-GP or glucose-6 phosphate (Ren et al., 2017). At the same P concentrations, P<sup>i</sup> provided apparently better support for growth than organic P sources (**Figure 7**). Hence, the production of phosphatases might represent additional investment of energy (Novotná et al., 2010) and/or the phosphatases were not able to liberate enough P<sup>i</sup> for growth. The substantially higher phosphatase activity in the cultures grown with organic P could reflect stronger P deficiency in these cultures compared to those grown with P<sup>i</sup> . Besides, not only the lack of P<sup>i</sup> but also the presence of organic P could contribute to phosphatase upregulation. In other words, both Coccomyxa species maintained approximately twofold higher phosphatase activity to perform the growth rates lower or equal to 0.1 day−<sup>1</sup> as shown in **Figure 7**.

Our results suggested fully inducible nature of acid phosphatases in the studied algae, because ELFA labeling was negligible in either P<sup>i</sup> or β-GP excess. On the contrary, Young et al. (2010) observed certain P-insensitive component of alkaline phosphatase activity in the benthic Cladophoraepiphyte assemblage from Lake Michigan, cultured with P<sup>i</sup> and α-glycerol phosphate supply, as well. Their conclusions, however, were based on an experimental study on the benthic assemblage, i.e., neither planktonic nor unialgal populations (Young et al., 2010). Moreover, their conclusions were based only on qualitative evidence (i.e., presence of ELFA-labeling) and not on quantification of the cell-specific phosphatase activity as was the case in this and other studies on C. silvae-gabretae (Štrojsová and Vrba, 2006, 2009; Novotná et al., 2010).

In our study, all P-depleted Coccomyxa cultures had significantly higher mean cell volumes compared to those that

were P-replete (**Figure 3**). Such larger cells of P-limited algae were reported in batch cultures of Scenedesmus quadricauda and Asterionella formosa (Litchman and Nguyen, 2008), as well as in continuous cultures of Cryptomonas phaseolus (Mindl et al., 2005). Cells in the P-limited cultures likely divided less often due to the lack of P and, at the same time, enlarged their volume via storing photosynthates. The larger cells indeed should result in a relatively low P content, hence, in other words, in the low cellular P quota and, consequently, in high phosphatase expression (Litchman and Nguyen, 2008).

Unlike in the short-term experiments with gradually exhausting sources in the batch cultures (e.g., Ou et al., 2010; Ren et al., 2017), in the present study, we employed semicontinuous cultures, better adjusted to keep the (limiting) P concentration tested. This setup allowed us to maintain the algal cultures at a more stable P deficiency or P sufficiency

for longer periods of time, to maintain more close-to-natural conditions of both original Coccomyxa populations (Barcyte and ˙ Nedbalová, 2017), and to allow for their cross-comparison. While C. elongata dominated in the phytoplankton of the eutrophic Lake Hromnice (Hrdinka et al., 2013), C. silvae-gabretae (previously misdetermined as Monoraphidium dybowskii) prevailed in the phytoplankton biomass of the mesotrophic Lake Plešné, with enormous bulk activity of extracellular phosphatases due to impaired P availability (Vrba et al., 2006).

Our results on C. silvae-gabretae cultures in this study are consistent with the observed in situ response of this species to increasing P availability. The in situ phosphatase activity of C. silvae-gabretae gradually decreased with the progressing lake recovery from acid stress during past decades (Novotná et al., 2010). The first application of FLEA assay in Lake Plešné in 2001 showed remarkable ELFA-labeling of all bacterioplankton, as well as of most phytoplankton species, including the entire population of C. silvae-gabretae (Nedoma et al., 2003b). The cell-specific phosphatase activity of this species varied within four orders of magnitude (0.5–500 fmol cell−1h −1 ) in 2003 (Štrojsová et al., 2005), but averaged within a closer range (0.4–12 fmol cell−1h −1 ) in 2005 (Štrojsová and Vrba, 2009), whereas hardly any production of extracellular phosphatases, i.e., weak ELFA-labeling of C. silvae-gabretae cells, was detected in 2007 (Novotná et al., 2010). Using a conversion factor of 0.1 fmol FU−<sup>1</sup> for our experimental results (in relative FU cell−1h −1 ), medians of phosphatase activity in the P-depleted variants I3 and O3 (**Figure 5**) were 0.04 and 0.16 fmol cell−1h −1 , respectively, and the most frequent classes in both O2 and O3 variants (**Figure 6D**) fell between 1.3 and 5.1 fmol cell−1h −1 for the strain of C. silvae-gabretae. These estimates of cell-specific phosphatase activity corresponded well to those determined for growing native populations of C. silvae-gabretae in Lake Plešné during the period of severe P limitation (Nedoma et al., 2003b; Štrojsová et al., 2005; Štrojsová and Vrba, 2009). The same estimates were calculated for the strain of C. elongata as well, although its native population in Lake Hromnice likely never experienced serious P depletion (Hrdinka et al., 2013).

The residual SRP concentrations in all P-depleted experimental variants (2–6 µmol L−<sup>1</sup> ) were at least by one order of magnitude higher than the threshold indicating P limitation in freshwaters (∼0.15 µmol L−<sup>1</sup> of SRP; Nedoma et al., 1993; Vrba et al., 1993), whereas the epilimnetic SRP concentrations in Lake Plešné averaged seasonally as low as 0.04 µmol L−<sup>1</sup> (Novotná et al., 2010). Yet in this lake, Štrojsová and Vrba (2009) documented remarkable diurnal variations in cell-specific phosphatase activity within the native population of C. silvae-gabretae in 2005, while Novotná et al. (2010) could not detect any measurable activity in this species at all in 2007. Both field studies suggested that single cells in the phytoplankton populations may differ remarkably in their individual cellspecific phosphatase activities due to the asynchronous character of the populations. Single algal cells likely reflected their internal needs in P, i.e., individual cellular P quota, as also suggested by Litchman and Nguyen (2008) or Ren et al. (2017). Štrojsová and Vrba (2009) proposed that distinct sub-populations (such as epilimnetic and metalimnetic) of C. silvae-gabretae with different life history characteristics could occur in the lake phytoplankton (e.g., due to strong mixing). Novotná et al. (2010) explained the absence of ELFA-labeling in the C. silvae-gabretae population by a pronounced P regeneration by grazing of abundant zooplankton, which re-colonized the lake between 2005 and 2007.

In our opinion, there is a common, but serious, methodological limitation in many recent studies employing ELFP (Rengefors et al., 2003; Dyhrman and Ruttenberg, 2006; Litchman and Nguyen, 2008; Young et al., 2010). Our analysis of frequency distribution of cell-associated ELFA fluorescences measured in this study (**Figure 6**) clearly illustrates the weakness of many studies employing ELFP that were based just on the scoring of ELFA-labeled cells (Rengefors et al., 2003; Dyhrman and Ruttenberg, 2006; Litchman and Nguyen, 2008; Young et al., 2010). For instance, our analysis showed that some ELFA-labeled cells could be found even in the P-replete culture (I1) grown on P<sup>i</sup> (about some 40% of cells were 'positive,' i.e., with non-zero fluorescence; **Figure 6**), which could lead to a false conclusion that the population was P-deficient. Yet their actual phosphatase activity was negligible (in the range of 0–0.08 FU cell−1h −1 ; cf. **Figures 5A** and **6A** or **6B**). At

media with either inorganic (I; top: A,B) or organic (O; bottom: C,D) P source. Except for zero classes (no activity), all FU classes are expressed in a geometric progression to cover broad range of ELFA fluorescence. All symbols are means of triplicates, bars represent SDs; note same scales; 100% is the sum of means in a variant. The total cell number (in a variant) measured for C. elongata: 300 (I1), 360 (I2), 420 (I3), 357 (O1), 360 (O2), 360 (O3), and C. silvae-gabretae: 300 (I1), 360 (I2), 355 (I3), 355 (O1), 359 (O2), 360 (O3).

present, the Fluorescently-Labeled Enzyme Activity assay (the true FLEA assay – Štrojsová and Vrba, 2006) is the only method available for adequate quantification of phosphatase activity at the level of individual cells or populations. The data on the percentages of ELFA-labeled cells should be interpreted with caution and considered, at the best, as semi-quantitative estimates of phosphatase activity in the assemblages (cf. Young et al., 2010).

Our experimental study suggested that the conclusions of our former field studies were plausible and confirmed the FLEA assay as a strong tool in phytoplankton ecology to explore P metabolism. To obtain reliable results, however, one should keep the following methodological recommendations: (i) Using ELF <sup>R</sup> 97 phosphate as the substrate for algal extracellular phosphatases, because an application of the ELF <sup>R</sup> 97 Endogenous Phosphatase Detection Kit likely may cause permeability of cell membranes and result in tagging of intracellular enzymes in some species (cf. Dyhrman and Palenik, 1999). (ii) Buffering phytoplankton samples, if pH in situ exceeds 8, to ensure the precipitation of ELFA molecules (Štrojsová and Vrba, 2006). (iii) Terminating sample incubation by gentle filtration (without application of phosphate buffered saline); only preserving samples with HgCl<sup>2</sup> before filtration is recommended for fragile flagellates (Nedoma et al., 2003b; Štrojsová et al., 2003). (iv) The ELFA-fluorescence-specific filter cube (see section "Materials and Methods") should be used for acquiring images; another chlorophyll-autofluorescence-specific filter cube is recommended to localize algal cells (**Figure 1**). A monochromatic rather than color camera is optimal for convenient image cytometry.

### CONCLUSION

fmicb-09-00271 February 19, 2018 Time: 14:51 # 10

The application of the FLEA assay in this experimental study confirmed the existence of environmental control of extracellular phosphatase expression in some acidotolerant algae and provided an insight into the impact of different P forms and concentrations on phosphatase activity in phytoplankton. This study represents the first evidence of inducible nature of acid phosphatases in algae. Our results further stress the importance of careful application of the FLEA method to gain reliable quantification of phosphatase activity at the single cell level.

### REFERENCES


#### AUTHOR CONTRIBUTIONS

JV, MM, and LN designed the experiment. MM and JN performed the image analysis. LN and MŠ performed the statistical analyses. All authors contributed to the manuscript.

### FUNDING

This study was partly supported from the specific research funding of the Charles University and University of South Bohemia, as well as from the institutional funding of the Biology Centre CAS.

#### ACKNOWLEDGMENTS

Dagmara Sirová did a language revision of the final text.



**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 Vrba, Macholdová, Nedbalová, Nedoma and Šorf. 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.

# Identification and Characterization of a Novel Salt-Tolerant Esterase from the Deep-Sea Sediment of the South China Sea

Yi Zhang<sup>1</sup> , Jie Hao<sup>1</sup> , Yan-Qi Zhang<sup>1</sup> , Xiu-Lan Chen<sup>1</sup> , Bin-Bin Xie<sup>1</sup> , Mei Shi<sup>1</sup> , Bai-Cheng Zhou<sup>1</sup> , Yu-Zhong Zhang1,2 and Ping-Yi Li<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Microbial Technology, Marine Biotechnology Research Center, Institute of Marine Science and Technology, Shandong University, Jinan, China, <sup>2</sup> Laboratory for Marine Biology and Biotechnology, Qingdao National Laboratory for Marine Science and Technology, Qingdao, China

Marine esterases play an important role in marine organic carbon degradation and cycling. Halotolerant esterases from the sea may have good potentials in industrial processes requiring high salts. Although a large number of marine esterases have been characterized, reports on halotolerant esterases are only a few. Here, a fosmid library containing 7,200 clones was constructed from a deep-sea sediment sample from the South China Sea. A gene H8 encoding an esterase was identified from this library by functional screening and expressed in Escherichia coli. Phylogenetic analysis showed that H8 is a new member of family V of bacterial lipolytic enzymes. H8 could effectively hydrolyze short-chain monoesters (C4–C10), with the highest activity toward p-nitrophenyl hexanoate. The optimal temperature and pH for H8 activity were 35◦C and pH 10.0, respectively. H8 had high salt tolerance, remaining stable in 4.5 M NaCl, which suggests that H8 is well adapted to the marine saline environment and that H8 may have industrial potentials. Unlike reported halophilic/halotolerant enzymes with high acidic/basic residue ratios and low pI values, H8 contains a large number of basic residues, leading to its high basic/acidic residue ratio and high predicted pI (9.09). Moreover, more than 10 homologous sequences with similar basic/acidic residue ratios and predicted pI values were found in database, suggesting that H8 and its homologs represent a new group of halotolerant esterases. We also investigated the role of basic residues in H8 halotolerance by site-directed mutation. Mutation of Arg195, Arg203 or Arg236 to acidic Glu significantly decreased the activity and/or stability of H8 under high salts, suggesting that these basic residues play a role in the salt tolerance of H8. These results shed light on marine bacterial esterases and halotolerant enzymes.

Keywords: esterase, salt-tolerance, deep-sea sediment, metagenomics, basic residues

### INTRODUCTION

Lipolytic enzymes, including esterases and lipases, are involved in catalyzing the hydrolysis and synthesis of esters. Esterases usually hydrolyze water-soluble short-chain monoesters, while lipases prefer water-insoluble long-chain triglycerides (Jaeger et al., 1999). Marine lipolytic enzymes play an important role in marine organic carbon degradation and cycling. A large number of active

#### Edited by:

Andrew Decker Steen, University of Tennessee, Knoxville, USA

#### Reviewed by:

Amy Michele Grunden, North Carolina State University, USA Wei Xie, Tongji University, China

> \*Correspondence: Ping-Yi Li lipingyipeace@sdu.edu.cn

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 05 December 2016 Accepted: 03 March 2017 Published: 23 March 2017

#### Citation:

Zhang Y, Hao J, Zhang Y -Q, Chen X -L, Xie B -B, Shi M, Zhou B -C, Zhang Y -Z and Li P -Y (2017) Identification and Characterization of a Novel Salt-Tolerant Esterase from the Deep-Sea Sediment of the South China Sea. Front. Microbiol. 8:441. doi: 10.3389/fmicb.2017.00441

microbial lipolytic enzymes have been discovered from surface and deep-sea seawater (Chu et al., 2008; Fang et al., 2014), hydrothermal vents (Placido et al., 2015), and marine sediments (Li et al., 2014), suggesting their potential roles in marine ecosystems.

Marine environments usually contain ∼3.5% (w/v) NaCl, and in some salterns, the salinity can even reach as high as 37% (w/v) (Ventosa et al., 2014). Many microbial enzymes of marine origin have evolved to be halotolerant or halophilic. Several halotolerant or halophilic lipolytic enzymes have been discovered from marine environments, including a halotolerant esterase (Est10) from Psychrobacter pacificensis (Wu et al., 2013), a halophilic esterase (LipC) from Haloarcula marismortui (Rao et al., 2009), a halotolerant esterase (ThaEst2349) from Thalassospira sp. (De Santi et al., 2016), and a halophilic lipase (LipBL) from Marinobacter lipolyticus (Perez et al., 2011). Studies on halotolerant/halophilic lipolytic enzymes and other halophilic enzymes show that these proteins have a significant increase in negatively charged acidic amino acid residues over their surfaces, which may form protective hydrated ion network and promote the adaption of the protein to salinity (Oren et al., 2005). The increase in acidity over the surface also prevents the aggregation of proteins (Elcock and Mccammon, 1998). However, there is also a report showing that an increase in positively charged basic residues on the enzyme surface may contribute to the adaption of an endonuclease VsEndA from Vibrio salmonicida to saline habitat (Altermark et al., 2008). Thus, halotolerant and halophilic enzymes may have diverse salt-adapted strategies. The halotolerance of lipolytic enzymes can help themselves and the strains producing them well adapt to the saline environments and play a role in marine organic carbon degradation and cycling. It has been reported that halotolerant/halophilic lipolytic enzymes have potentials in industrial processes requiring high salts, low water activity, and the presence of organic solvents.

Based on amino acid sequences and biochemical properties, microbial lipolytic enzymes have been classified into eight families (families I-VIII) (Arpigny and Jaeger, 1999). Enzymes grouped in family V originates from a wide variety of bacteria, including mesophilic, psychrophilic, and thermophilic organisms (Arpigny and Jaeger, 1999). Recently, many members of family V lipolytic enzymes have been discovered (Prive et al., 2013; Sumby et al., 2013; Tchigvintsev et al., 2015). This family contains lipases and esterases, displaying diverse substrate specificities and characteristics (Peng et al., 2011; Chen et al., 2013; Pereira et al., 2015). However, studies on the salt tolerance of this family are still scarce.

Marine environments benefit the discovery of novel enzymes with special characteristics. Because more than 99% of marine microorganisms are still uncultured (Schloss and Handelsman, 2003), metagenomics, a cultivation-independent method, has been developed to discover new functional genes from both cultured and uncultured microorganisms (Handelsman, 2004). The application of functional metagenomics has led to the discovery of several new lipases and esterases from diverse marine environments, such as intertidal flat (Kim et al., 2009), tidal flat sediment (Jeon et al., 2012), and marine surface water (Chu et al., 2008).

To identify novel esterases from marine sediments, in this study, a fosmid library of a deep-sea sediment sample from the South China Sea was constructed, and functional metagenomic screening was performed to obtain novel esterases. A lipolytic enzyme gene H8 was identified from the library, and the encoding esterase H8 was expressed and characterized. The result showed that H8 was a new member of family V of bacterial lipolytic enzymes with a substrate preference toward short-chain monoesters (C4–C10). H8 displayed high halotolerance. The sequence of H8 contains a large number of basic residues, leading to a high basic/acidic residue ratio and a high predicted isoelectric point (pI). The roles of basic residues in H8 halotolerance were investigated by site-directed mutagenesis. Sequence analysis suggests that H8 together with its homologs represent a new group of halotolerant esterases. These results shed light on marine bacterial esterases and halotolerant enzymes.

### MATERIALS AND METHODS

### Sample Collection and DNA Extraction

Marine sediment sample S100 was collected from the South China Sea (13.5◦N, 118◦E) at a water depth of 3,939 m in September 2011. Temperature and salinity of bottom water in this area was 2.4◦C and 3.46% (w/v), respectively. The sample was stored at −20◦C until processing. Environmental genomic DNA was extracted from the sample by following the SDS-based extraction procedure described by Zhou et al. (1996).

### Metagenomic Library Construction and Screening of Lipolytic Enzymes

The DNA extract was separated by pulsed-filed gel electrophoresis (PFGE), and DNA bands of ∼35 kbp in the gel were extracted by gelase enzymolysis and ethanol precipitation. A metagenomic DNA library was constructed using the Copy-Control Fosmid Library Production Kit (Epicentre Biotechnologies, Madison, WI, USA) by following the manufacturer's instructions. A total of 7,200 fosmid clones were obtained, which were spread onto Luria-Bertani (LB) agar plates supplemented with 1% (v/v) emulsified tributyrin to screen clones with lipolytic enzyme activity. Clones exhibiting a clear zone around the colony were selected to construct their respective subcloning libraries. Fosmid DNA was extracted from positive clones and partially digested by the restriction enzyme Sau3AI. The DNA fragments of 1.5–5 kbp were recovered from an agarose gel, end-repaired and ligated into the pUC19 vector that had been digested by BamHI and pretreated by bacterial alkaline phosphatase. The ligated products were transformed into E. coli TOP10 cells, and the transformants were spread onto LB agar plates supplemented with 100 µg/ml ampicillin and 1% (v/v) tributyrin. Transformants forming clear zones were sequenced. The open reading frames (ORFs) in the sequenced

DNA fragments were predicted by the GeneMark program<sup>1</sup> and the genes encoding potential lipolytic enzymes were identified by using the blastx program against the NCBI nonredundant protein database (nr). Multiple sequence alignment was performed using MUSCLE (Edgar, 2004). Phylogenetic analysis was carried out with the neighbor-joining method using MEGA 6.0 (Tamura et al., 2013). The potential signal peptide sequence was predicted by SignalP 4.0 (Petersen et al., 2011).

### Gene Cloning and Protein Expression and Purification

The gene H8 encoding a lipolytic enzyme was amplified from the fosmid DNA using the primer pair of H8\_F (5<sup>0</sup> -GGGAATTC CATATGCAGTCTGGCACGGTGAG-3<sup>0</sup> , NdeI digestion site was underlined) and H8\_R (5<sup>0</sup> -CCGCTCGAGCGCCACCGCCGGT TGCGCC-3<sup>0</sup> , XhoI digestion site was underlined), and cloned into the expression vector pET-22b.

The constructed plasmid pET-22b-H8 was transformed into E. coli BL21 (DE3). Transformants were cultured at 37◦C and 180 rpm in LB liquid medium containing 100 µg/mL ampicillin. When the OD<sup>600</sup> of cells reached approximately 0.6, 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) was added for the induction of protein expression. Then, the culture was incubated at 20◦C and 110 rpm for 20 h. After incubation, the cells in the culture were harvested, resuspended in lysis buffer (50 mM Tris-HCl, 100 mM NaCl, pH 8.0) and disrupted by pressure. The recombinant His-tagged protein in the extract was first purified by Ni affinity chromatography (Qiagen, USA), and further purified by gel filtration chromatography on a Superdex 200 column (GE healthcare, Sweden). Protein concentrations were determined by using the Pierce BCA Protein Assay Kit (Thermo Scientific, USA).

### Enzyme Assays

The esterase activity was measured by monitoring the hydrolysis of p-nitrophenyl (pNP) esters (Sigma, USA) using a spectrophotometric method (Shirai and Jackson, 1982). The reaction mixture contained 50 mM Tris-HCl buffer (pH 8.0), 0.02 ml of 10 mM substrate, and 0.02 ml enzyme with appropriate concentration in a final volume of 1 m1. After incubation at an indicated temperature for 5 min, the reaction was terminated by an addition of 0.1 ml 20% (w/v) SDS. The absorbance of the reaction mixture at 405 nm was measured to detect the amount of released p-nitrophenol (Li et al., 2014). The background hydrolysis of the substrate was determined by using a blank control with a composition identical with the reaction mixture except that the enzyme was replaced by buffer. One unit of enzyme (U) is defined as the amount of enzyme required to liberate 1 µmol p-nitrophenol per minute.

### Biochemical Characterization of H8

The substrate specificity of H8 was investigated using the substrates pNP acetate (C2), pNP butyrate (C4), pNP caproate (C6), pNP caprylate (C8), pNP decanoate (C10), pNP laurate (C12), pNP myristate (C14), and pNP palmitate (C16) (Sigma, USA). The optimum temperature for H8 activity was measured at temperatures ranging from 0 to 60◦C at pH 8.0. For thermostability assay, the enzyme was incubated at temperatures ranging from 0 to 60◦C for 1 h, and then the residual activity was measured at 35◦C. The optimum pH of H8 was determined at 35◦C in the Britton–Robinson buffers ranging from pH 4.0 to 13.0. For pH stability assay, the enzyme was incubated in buffers with a pH range of 4.0–13.0 at 25◦C for 1 h, and then the residual activity was measured at pH 8.0 and 35◦C. The effect of NaCl on H8 activity was determined at NaCl concentrations ranging from 0 to 4.8 M. For salt tolerance assay, the enzyme was incubated at 0 ◦C for 1 h in buffers containing NaCl ranging from 0 to 4.6 M before the residual activity was measured at 35◦C.

The effects of metal ions (K+, Li+, Ba2+, Ca2+, Co2+, Cu2+, Fe2+, Mg2+, Mn2+, Ni2+, and Zn2+) and potential inhibitors (β-mercaptoethanol, DTT, Thiourea, Urea, EDTA, and PMSF) on H8 activity were examined at pH 8.0 and 35◦C in a final concentration of 1 mM or 10 mM. The effects of organic solvents on H8 activity were examined using methanol, ethanol, isopropanol, acetone, acetonitrile, dimethylsulfoxide (DMSO), and dimethylformamide (DMF) at final concentrations of 10% and 20% (v/v). The effects of Tween 20, Tween 80, and Triton X-100 on H8 activity were examined at final concentrations of 0.001–0.1% (v/v). The effect of SDS on enzyme activity was measured at final concentrations of 0.001–0.1% (w/v).

### Site-Directed Mutagenesis of H8

Using plasmid pET-22b-H8 as the template, site-directed mutagenesis on H8 was performed with the QuikChange <sup>R</sup> 146 mutagenesis kit II (Agilent technologies, USA) according to the method of QuikChange site-directed mutagenesis (Liu and Naismith, 2008). After verified by DNA sequencing, mutated plasmids were transformed into E. coli BL21 (DE3) for protein expression. The purification of H8 mutants was performed under the same conditions as those of the wild type (WT) H8.

#### Enzyme Kinetic Assays

Enzyme kinetic assays of H8 and its mutants were carried out at pH 7.5 (50 mM Tris-HCl) using pNPC6 at concentrations from 0.02 to 2.0 mM. Kinetic parameters were calculated by non-linear regression fit directly to the Michaelis–Menten equation using the Origin8.5 software.

#### Circular Dichroism Spectroscopy

Circular dichroism (CD) spectra of WT H8 and its mutants were recorded at 25◦C on a J-810 spectropolarimeter (JASCO, Japan). All the spectra were collected from 200 to 250 nm at a scanning speed of 200 nm/min with a path length of 0.1 cm. Proteins for CD spectroscopy assays were at a concentration of 0.3 mg/ml in 50 mM Tris-HCl buffer (pH 8.0).

#### Nucleotide Sequence Accession Number

The nucleotide sequence of H8 has been deposited in the GenBank database under accession number KY273927.

<sup>1</sup>http://topaz.gatech.edu/GeneMark/

branch length is shown below the tree. Lipases from family I were used as outgroups.


FIGURE 2 | Multiple sequence alignment of H8 and its homologs. Identical and similar amino acids are shaded in black and gray, respectively. Stars indicate amino acid residues belonging to the catalytic triad, and circles indicate basic amino acid residues (Arg195, Arg203, Arg216, Arg236, and Arg263) selected for site-directed mutation. Sequence analysis suggested that residues Arg203, Arg216, and Arg236 are highly conserved, and residues Arg195 and Arg263 are partially conserved.

FIGURE 3 | Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of purified esterase H8 and the substrate specificity of H8. (A) SDS-PAGE analysis of purified H8. Lane 1, purified H8; lane M, protein mass markers. (B) Substrate specificity of H8 evaluated with pNP esters. The graph shows data from triplicate experiments (mean ± SD).

ranging from pH 4.0 to 13.0. The highest activity at pH 10.0 (70.2 U/mg) was taken as 100%. (B) Effect of pH on the stability of H8. The enzyme was incubated in buffers ranging from pH 4.0 to 13.0 at 25◦C for 1 h. The remaining activity was measured under optimal conditions. The highest activity at pH 8.0 (54.4 U/mg) was taken as 100%. The graphs show data from triplicate experiments (mean ± SD).

## RESULTS

fmicb-08-00441 March 21, 2017 Time: 15:52 # 6

### Functional Metagenomic Screening of Lipolytic Enzymes

A metagenomic library was constructed from a deep-sea sediment sample from the South China Sea, which contained a total of 7,200 fosmid clones. The metagenomic library represented approximately 252 Mbp environmental DNA assuming an average insert size of 35 kbp. By functional assays, 10 fosmid clones showing lipolytic enzyme activities were screened from this library. The sequences of putative lipolytic enzymeencoding genes in the identified fosmids were determined by subcloning library construction and subsequent sequencing. Among these genes, the gene H8 containing 918 bp was predicted to encode a lipolytic enzyme and chosen for further analysis.

### Sequence Analysis of the Lipolytic Enzyme H8

The gene H8 encodes a lipolytic enzyme of 305 amino acid residues with a predicted molecular weight of 32.8 kDa and a predicted pI of 9.09. Prediction by SignalP 4.0 suggested that H8 may lack an N-terminal signal sequence. Among the characterized lipolytic enzymes, H8 showed the highest sequence identity (46%) to a family V esterase (Est16) from a microbial consortium specialized for diesel oil degradation (Pereira et al., 2015). Phylogenetic tree also showed that H8 belongs to the family V of bacterial lipolytic enzymes (**Figure 1**). Based on sequence alignments with other proteins from family V, the catalytic triad of H8 was identified, which is composed of Ser120, Asp247, and His275 (**Figure 2**). The catalytic Ser120 is located in the conserved GASMGGMI motif, Asp247 in the conserved DPL motif, and His275 in the conserved MG/AHD motif.

#### Expression and Characterization of the Esterase H8

H8 was over-expressed in E. coli BL21 (DE3), and the recombinant H8 protein was first purified by Ni affinity chromatography and then by gel filtration chromatography. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis showed that the purified H8 displayed an apparent molecular weight of approximately 33 kDa, accordant to that predicted from its sequence (32.8 kDa) (**Figure 3A**). H8 could efficiently hydrolyze short-chain pNP esters (C4–C10), with the maximal activity toward pNPC6 (69.0 U/mg) (**Figure 3B** and Supplementary Figure 1). H8 showed a limited ability to degrade pNP esters longer than 10 carbon atoms, indicating that H8 is an esterase. H8 showed the highest activity at 35◦C and retained 30% of its highest activity at 0◦C (**Figure 4A**). H8 retained more than 80% of its highest activity after 1 h incubation at temperatures lower than 40◦C, but lost all the activity after 1 h incubation at 50◦C (**Figure 4B**). H8 had the highest activity at pH 10.0 (**Figure 5A**) and showed good tolerance in a range of pH 6.0– 9.0, retaining over 80% of its highest activity after 1 h incubation in the buffers of pH 6.0–9.0 at 25◦C (**Figure 5B**).

The effect of metal ions on the activity of H8 was also investigated (**Table 1**). H8 activity was almost unaffected by TABLE 1 | Effects of metal ions and potential inhibitors on H8 activity.


<sup>a</sup> LD indicates that the value was less than the limit of detection.

#### TABLE 2 | Effects of detergents on H8 activity.


<sup>a</sup> The concentrations of SDS used were presented in w/v.

<sup>b</sup> LD indicates that the value was less than the limit of detection.

TABLE 3 | Effects of organic solvents on H8 activity.


K <sup>+</sup>, Li+, Ca2+, Co2<sup>+</sup> or Mg2<sup>+</sup> at 1∼10 mM, but significantly inhibited by Zn2<sup>+</sup> at 1∼10 mM, and Ba2+, Mn2+, and Ni2<sup>+</sup> at 10 mM, and fully inhibited by Cu2<sup>+</sup> and Fe2<sup>+</sup> at 10 mM. EDTA had no effect on H8 activity, suggesting that the catalysis by H8 may not require metal ions. H8 activity was significantly inhibited by 10 mM PMSF, indicating that H8 is most likely a serine hydrolase. H8 activity was also severely inhibited by

TABLE 4 | Comparison of the amino acid composition of H8 and its homologs and reported halotolerant enzymes.


<sup>a</sup> –, no sequence identity detected.

reductants DTT and β-mercaptoethanol. However, H8 showed high resistance to chaotropic agents urea and thiourea (**Table 1**). H8 activity was slightly increased by 0.001–0.1% (v/v) Tween 20, but fully inhibited by 0.01% (w/v) SDS (**Table 2**). Among all the tested organic solvents at 10% (v/v) concentration, methanol, ethanol, DMF, and DMSO slightly increased H8 activity, and other detergents slightly reduced H8 activity. At 20% (v/v) concentration, DMF and DMSO had nearly no effect on H8 activity, whereas other detergents significantly inhibited H8 activity (**Table 3**).

#### High Salt Tolerance of H8

Because the H8 gene is isolated from a deep-sea sediment, we investigated the effect of NaCl of different concentrations on the activity and stability of H8. H8 still had full activity in NaCl with a concentration as high as 4.0 M (**Figure 6A**), indicating that H8 has high salt tolerance. Moreover, after 1 h incubation in 4.6 M NaCl, H8 still retained 80% activity (**Figure 6B**). These results show that H8 is a halotolerant enzyme.

#### High Contents of Basic Residues in H8

Unlike most reported halophilic/halotolerant enzymes that have high acidic/basic residue ratios and relatively low pI values ranging from 4.3 to 6.8 (Ng et al., 2000; Kennedy et al., 2001; Dassarma et al., 2013), H8 contains more basic residues (10.49%) than acidic residues (9.18%) (**Table 4**), leading to a high predicted pI value of 9.09. By searching NCBI nr database using the H8 sequence as a query, more than 10 homologs of H8 are found to have more basic residues than acidic residues and high pI values (**Table 4**), suggesting that H8 and its homologs may represent an uncharacterized group of lipolytic enzymes rich in basic residues.

### The Role of Basic Residues in the Salt Tolerance of H8

Until now, only one halophilic enzyme, the endonuclease VsEndA from V. salmonicida, is found to have an overwhelming number of basic residues distributed on the protein surface for its haloadaption (Altermark et al., 2008). To study the roles of the basic residues, especially the surface basic residues, in the salt tolerance of H8, we tried to obtain its crystal structure or modeled structure. Unfortunately, the crystal structure of H8 was unable to be solved due to the low resolution of the H8 crystals we obtained. In addition, due to the low sequence identity (lower than 28%) between H8 and reported proteins with resolved structures, no modeled structure of H8 could be constructed. Finally, according to multiple sequence alignment (**Figure 2**), we selected five basic residues (Arg195, Arg203, Arg216, Arg236, and Arg263) for site-directed mutation to acidic Glu to investigate their roles in H8 halotolerance. Residues Arg195 and Arg263 are partially conserved and residues Arg203, Arg216, and Arg236 are highly conserved in H8 homologs (**Figure 2**).

The effect of NaCl on the activities and stabilities of the mutants was measured and compared to WT H8 (**Figure 7**). Under their respective optimum temperatures, the effect of NaCl

FIGURE 7 | Effect of NaCl on the activity and stability of the mutants of H8. (A) Effect of NaCl on the activity of WT H8 and its mutants. The activities of H8 and its mutants were determined at different NaCl concentrations at their respective optimum temperatures. The activities of WT H8 (70.9 U/mg), R195E (65.6 U/mg), R203E (7.5 U/mg), R216E (82.6 U/mg), R236E (53.0 U/mg), and R263E (62.0 U/mg) in 0 M NaCl were taken as 100%, respectively. (B) Effect of NaCl on the stability of WT H8 and its mutants. The enzymes were incubated in buffers containing different NaCl concentrations at 0◦C for 1 h, and the residual activity was measured at their optimum temperatures, respectively. The activities of WT H8 (68.6 U/mg), R195E (64.5 U/mg), R203E (6.7 U/mg), R216E (72.8 U/mg), R236E (48.8 U/mg), and R263E (61.0 U/mg) in 0 M NaCl were taken as 100%, respectively. The graphs show data from triplicate experiments (mean ± SD).

FIGURE 8 | Relative specific activities and circular dichroism (CD) spectra of WT H8 and its mutants. (A) Relative specific activities of WT H8 and its mutants. The specific activity of WT H8 (69.7 U/mg) was defined as 100%. (B) CD spectra of WT H8 and its mutants. All the spectra were collected from 200 to 250 nm at a scanning speed of 200 nm/min with a path length of 0.1 cm. The graphs show data from triplicate experiments (mean ± SD).



on the activities and stabilities of mutants R216E and R263E was similar to that of the WT (**Figures 7A,B**), suggesting that these two residues Arg216 and Arg263 may not be related to the salt tolerance of H8. In addition, mutants R216E and R263E had similar K<sup>m</sup> and kcat values and specific activities to the WT (**Figure 8A** and **Table 5**), indicating that these two mutations had little effect on the substrate binding and catalysis of H8 and that Arg216 and Arg263 are potentially surface residues.

Although the activity of mutant R203E was slightly stimulated by 1.2- to 1.3-fold in NaCl ranging from 2.0 to 4.0 M, its stability was significantly reduced in increased concentrations of NaCl. After incubated in 4.0 M NaCl for 1 h, H8 retained 80% activity, whereas R203E retained only 30% activity, indicating that mutant R203E is less tolerant than the WT under high salts. Mutation R203E had no impact on the K<sup>m</sup> of H8, but significantly reduced its kcat and specific activity (**Figure 8A** and **Table 5**). These data

suggest that Arg203 might be directly or indirectly involved in the catalysis of H8.

The effect of NaCl on the stability of mutant R195E was similar to that of the WT, whereas its activity was significantly reduced by NaCl. Compared to WT H8, the activity of mutant R195E in 4 M NaCl was reduced by 51%, suggesting its potential role in the salt tolerance of H8. Mutation R195E had no impact on the specific activity of H8 and small impact on the K<sup>m</sup> and kcat values (**Figure 8A** and **Table 5**), indicating that Arg195 is potentially located on the surface of H8 protein.

For mutant R236E, both the activity and stability was significantly reduced in increased concentrations of NaCl, suggesting that Arg236 may play an important role in the salt tolerance of H8. Mutant R236E had only small effect on the substrate binding and catalysis of H8 (**Table 5**), suggesting that residue Arg236 is potentially a surface residue.

Circular dichroism spectral analysis showed that these mutations caused no visible changes in H8 structure (**Figure 8B**), indicating that the decrease in the enzymatic activity and stability of the mutants resulted from residue substitution rather than structural changes.

#### DISCUSSION

In this study, a metagenomic library containing 7,200 fosmid clones was constructed from a deep-sea sediment sample from the South China Sea to functionally screen lipolytic enzymeencoding genes, and a gene encoding an esterase H8 was identified, cloned and over-expressed. Phylogenetic analysis showed that H8 belongs to the family V of bacterial lipolytic enzymes. Among the characterized lipolytic enzymes, H8 has the highest identity (46%) to Est16 from a microbial consortium specialized for diesel oil degradation (Pereira et al., 2015). However, H8 shows different substrate specificities from Est16 (Pereira et al., 2015). H8 can efficiently hydrolyze short-chain monoesters (C4–C10), especially for pNPC6 and pNPC8, while Est16 prefers to hydrolyze pNPC4 and pNPC5 (Pereira et al., 2015). Therefore, H8 is a new family V esterase.

A few halotolerant/halophilic lipolytic enzymes have been reported. Consistent with other known halotolerant/halophilic enzymes, halotolerant/halophilic lipolytic enzymes contain more acidic residues (Asp and Glu) than basic residues in their sequences, leading to their low pI values (Ng et al., 2000; Kennedy et al., 2001; Dassarma et al., 2013). Moreover, modeled structural analysis suggests that a large number of negatively charged acidic residues are distributed over the protein surface of halotolerant/halophilic lipolytic enzymes, which may form a protective solvation shell to keep the protein surface hydrated and promote the adaption of the protein to salinity (De Santi et al., 2016; Wang et al., 2016). Until now, reports on the salt tolerance of family V lipolytic enzymes are still scarce. We studied the effect of NaCl on the activity and stability of H8. The result showed that H8 has high halotolerance. However, unlike most reported halotolerant/halophilic enzymes, H8 contains a large number of basic residues in sequence, leading to its high basic/acidic residue ratio and high pI value (9.09). In addition, more than 10 homologous sequences with similar basic/acidic residue ratios and predicted pI values to H8 are found in NCBI nr database. Therefore, H8 and its homologs may represent a new subgroup of family V halotolerant lipolytic enzymes rich in basic residues.

Up to date, only a halophilic endonuclease VsEndA from V. salmonicida is reported to contain more basic residues than acidic residues and have high pI value (9.57) (Altermark et al., 2008). Moreover, structural analysis shows that the protein surface of VsEndA is populated with positively charged basic residues, which may result in the haloadaption of VsEndA (Altermark et al., 2008). We studied the roles of five conserved basic residues in the salt tolerance of H8 by residue replacement. The results suggested that Arg195, Arg203, and Arg236 may play a role in the salt tolerance of H8, but Arg216 and Arg263 have little effect on the salt tolerance of H8. However, due to the lack of H8 structure, it is difficult to determine the exact positions and roles of these basic residues in H8, which still need further study.

In addition, H8 is also a cold-adapted enzyme. H8 had a low optimal temperature (35◦C) for activity and still remained 30% of the maximal activity at 0◦C. The halotolerant and coldadapted characteristics indicate that H8 is well adapted to deepsea sediment and may play a role in marine organic degradation and carbon cycling. Moreover, the good halotolerance of H8 implies its potentials in harsh industrial processes requiring high salts, low water activity, and the presence of organic solvents (such as DMSO).

### AUTHOR CONTRIBUTIONS

YZ, JH, and Y-QZ performed all experiments. P-YL and X-LC directed the experiments. YZ and P-YL wrote the manuscript. B-BX and MS helped in data analysis. Y-ZZ and B-CZ designed the research.

#### ACKNOWLEDGMENTS

This work was supported by the National Science Foundation of China (grants 31290231, 91328208, 31670497, and 41676180), and the Program of Shandong for Taishan Scholars (TS200 90803).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.00441/full#supplementary-material

#### REFERENCES

fmicb-08-00441 March 21, 2017 Time: 15:52 # 10


**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 Zhang, Hao, Zhang, Chen, Xie, Shi, Zhou, Zhang and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) 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.

# Characterization of a New S8 serine Protease from Marine Sedimentary Photobacterium sp. A5–7 and the Function of Its Protease-Associated Domain

Hui-Juan Li1,2† , Bai-Lu Tang<sup>1</sup>† , Xuan Shao<sup>1</sup> , Bai-Xue Liu<sup>1</sup> , Xiao-Yu Zheng<sup>1</sup> , Xiao-Xu Han<sup>1</sup> , Ping-Yi Li<sup>1</sup> , Xi-Ying Zhang<sup>1</sup> , Xiao-Yan Song<sup>1</sup> and Xiu-Lan Chen<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Microbial Technology, Marine Biotechnology Research Center, Institute of Marine Science and Technology, Shandong University, Jinan, China, <sup>2</sup> College of Chemical and Environmental Engineering, Shandong University of Science and Technology, Qingdao, China

#### Edited by:

Andrew Decker Steen, University of Tennessee, USA

#### Reviewed by:

Nina Dombrowski, Texas State University System, USA Karolina Michalska, Argonne National Laboratory, USA

#### \*Correspondence:

Xiu-Lan Chen cxl0423@sdu.edu.cn †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Received: 08 September 2016 Accepted: 01 December 2016 Published: 22 December 2016

#### Citation:

Li H-J, Tang B-L, Shao X, Liu B-X, Zheng X-Y, Han X-X, Li P-Y, Zhang X-Y, Song X-Y and Chen X-L (2016) Characterization of a New S8 serine Protease from Marine Sedimentary Photobacterium sp. A5–7 and the Function of Its Protease-Associated Domain. Front. Microbiol. 7:2016. doi: 10.3389/fmicb.2016.02016 Bacterial extracellular proteases are important for bacterial nutrition and marine sedimentary organic nitrogen degradation. However, only a few proteases from marine sedimentary bacteria have been characterized. Some subtilases have a proteaseassociated (PA) domain inserted in the catalytic domain. Although structural analysis and deletion mutation suggests that the PA domain in subtilases is involved in substrate binding, direct evidence to support this function is still absent. Here, a protease, P57, secreted by Photobacterium sp. A5-7 isolated from marine sediment was characterized. P57 could hydrolyze casein, gelatin and collagen. It showed the highest activity at 40◦C and pH 8.0. P57 is a new subtilase, with 63% sequence identity to the closest characterized protease. Mature P57 contains a catalytic domain and an inserted PA domain. The recombinant PA domain from P57 was shown to have collagen-binding ability, and Phe349 and Tyr432 were revealed to be key residues for collagen binding in the PA domain. This study first shows direct evidence that the PA domain of a subtilase can bind substrate, which provides a better understanding of the function of the PA domain of subtilases and bacterial extracellular proteases from marine sediment.

#### Keywords: subtilase, protease-associated domain, marine sediment, collagen-binding, aromatic residues

#### INTRODUCTION

Organic nitrogen degradation is an important part of marine nitrogen cycle (Aluwihare et al., 2005). Particulate organic nitrogen (PON) that deposits to marine sediments is mainly decomposed by bacterial extracellular proteases, which is generally considered to be the initial and rate-limiting step of nitrogen cycle in marine sediments (Talbot and Bianchi, 1997; Brunnegård et al., 2004). It has been found that protease-producing bacteria and their extracellular proteases are rich and diverse in marine sediments (Olivera et al., 2007; Zhou et al., 2009, 2013). Some proteases from marine sedimentary bacteria have been characterized, most of which are shown to have special properties, such as cold adaptation (Chen et al., 2007; Yan et al., 2009; Kurata et al., 2010; Yang et al., 2013), salt tolerance (Yan et al., 2009), distinct substrate specificity and catalytic mechanism (Ran et al., 2014). Therefore, marine sediments are good resources for exploring novel proteases.

Peptidase family S8, also known as the subtilisin or subtilase family, is the second-largest family of serine proteases (Rawlings et al., 2010). Proteases in this family

are all characterized by an Asp/His/Ser catalytic triad and an α/β fold catalytic center containing a seven-stranded parallel β-sheet (Rawlings et al., 2006). Some proteases in the S8 family contain a protease-associated (PA) domain, which is inserted in the catalytic domain. This kind of proteases are reported from both plants and bacteria, such as tomato SBT3 (Ottmann et al., 2009), soybean protease C1 (Tan-Wilson et al., 2012), streptococcal C5a peptidase (Brown et al., 2005; Kagawa et al., 2009), lactococcal cell-envelope protease (Bruinenberg et al., 1994; Sadat-Mekmene et al., 2011), VapT from Vibrio metschnikovii RH530 (Kwon et al., 1995), Apa1 from Pseudoalteromonas sp. AS-11 (Dong et al., 2001), SapSh from Shewanella sp. Ac10 (Kulakova et al., 1999) and AcpII from Alkalimonas collagenimarina AC40 (Kurata et al., 2010).

The PA domains in subtilases were reported to have diverse functions. For plant subtilases, the PA domain of tomato SBT3 was suggested to be required for enzyme maturation, secretion, dimerization and activation (Cedzich et al., 2009; Ottmann et al., 2009); homology modeling and molecular simulation indicated that the PA domain of soybean protease C1 was crucial for determining the optimum length of peptide substrate (Tan-Wilson et al., 2012). For bacterial subtilases, deletion experiment indicated that the PA domain of lactococcal cell-envelope protease influenced the enzyme substrate specificity, but was unnecessary for enzyme folding or autoprocessing (Bruinenberg et al., 1994); the PA domains of AcpII from Alkalimonas collagenimarina AC40 and of streptococcal C5a peptidase SCPA from Streptococcus pyogenes were reported to restrict substrate access to the active site (Kagawa et al., 2009; Kurata et al., 2010). Although structural analyses and deletion experiments suggest that the PA domains in subtilases are involved in substrate binding, there has been no evidence showing that the PA domain of subtilase can bind substrate directly.

In a previous study, 66 protease-producing strains were screened from 6 sediment samples from Jiaozhou Bay, China, in which Photobacterium (39.4%), Bacillus (25.8%), Vibrio (19.7%) and Shewanella (7.6%) were the major groups and Photobacterium strains were distributed in all sediment samples. Among the strains, Photobacterium sp. A5–7 had the highest protease activity, which was isolated from the sediment sample from the A5 station site where the depth, temperature, pH and carbon/nitrogen ratio were 5.9 m, 24.7◦C, 8.11 and 7.0, respectively (Zhang et al., 2015). In this study, we aimed to purify and characterize the protease secreted by Photobacterium sp. A5–7. The result showed that the protease P57 secreted by Photobacterium sp. A5–7 could hydrolyze casein, gelatin and collagen. We then cloned the gene encoding P57. Sequence analysis showed that P57 is a subtilase of the S8 family, containing a PA domain inserted in its catalytic domain. To study its function, the PA domain of P57 was expressed in Escherichia coli as a EGFP-fused protein, and its collagen-binding ability was determined by using fluorescent technology. The recombinant PA domain was shown to have collagen-binding ability, suggesting that the PA domain is likely involved in substrate binding during the hydrolysis of insoluble collagen by P57. Site-directed mutations were also performed to analyze the key residues for collagen binding in the PA domain of P57. Our results reveal a new S8 subtilase from a marine sedimentary bacterium and show direct evidence that the PA domain of a subtilase can bind substrate, which shed light on marine bacterial proteases and marine PON degradation.

#### MATERIALS AND METHODS

#### Phylogenetic Analysis of Strain A5-7

Genomic DNA of strain A5-7 was extracted using a bacterial genomic DNA isolation kit (BioTeke, China). Using the genomic DNA as template, the 16S rRNA gene of strain A5–7 was amplified by PCR with primers 27F and 1492R (Lane, 1991) and sequenced at Shanghai Biosune Biotechnology Corp. (China). The 16S rRNA gene sequence of strain A5–7, deposited in GenBank database under the accession number JX134463, was compared with those in GenBank and EzTaxon<sup>1</sup> (Kim et al., 2012) databases using BLASTN (Altschul et al., 1997) to determine the approximate phylogenetic affiliation and select reference sequences of related species for subsequent phylogenetic analysis. The 16S rRNA gene sequence of strain A5–7 was aligned with those of type strains of closely related species using the Clustal W program (Thompson et al., 1994) in the MEGA 5 (Tamura et al., 2011). The alignment obtained was manually trimmed to remove the uneven 5<sup>0</sup> and 3<sup>0</sup> ends. Phylogenetic trees were constructed using the neighbor-joining (Saitou and Nei, 1987) and maximum-likelihood (Felsenstein, 1981) methods with MEGA 5 (Tamura et al., 2011). Bootstrap analyses (1000 replications) were performed to evaluate tree topologies. Evolutionary distances were calculated using the model of Jukes and Cantor (1969).

#### Purification of the Protease P57 Secreted by Strain A5–7

Strain A5–7 was cultivated at 15◦C for 72 h in a marine LB medium supplemented with 0.3% (w/v) casein and 0.5% (w/v) gelatin in a rotatory shaker at 180 rpm. The culture was centrifuged at 12,000 g and 4◦C for 15 min. The cell-free supernatant was concentrated against polyethylene glycol (PEG) 20,000, and then dialyzed against 20 mM phosphate buffer (pH 7.0). The sample was loaded onto a DEAE-Sepharose Fast Flow column (GE Healthcare, USA) equilibrated with the same buffer. Bound proteins were eluted with a linear gradient of 0-1 M NaCl in 20 mM phosphate buffer (pH 7.0). The protease activity of every fraction (4 mL) was measured with casein as substrate. Fractions having protease activity were further subjected to 12.5% SDS-PAGE. Then, fractions having protease activity and displaying a single band in SDS-PAGE gel were collected for further use. The purified protease was named P57.

#### Analysis of the N-Terminal Amino Acid Sequence of Protease P57

The purified P57 was transferred from SDS-PAGE gel to a Sequi-Blot polyvinylidene difluoride membrane (PVDF

<sup>1</sup>http://eztaxon-e.ezbiocloud.net/

membrane, Bio-Rad, USA). Its N-terminal amino acid sequence was determined by the Edman degradation method with a PROCISE491 sequencer (Applied Biosystems, USA) at Peking University, China. The obtained N-terminal sequence of P57 was aligned with those of proteases in GenBank using BLASTP to determine the protease type.

#### Characterization of Protease P57

fmicb-07-02016 December 20, 2016 Time: 17:27 # 3

The activity of P57 was measured as described by Chen et al. (2003) with 2% (w/v) casein as substrate. The optimal temperature of P57 was determined over the range from 0 to 70◦C in 20 mM phosphate buffer (pH 7.0). The effect of temperature on protease stability was evaluated by measuring the residual activity at 40◦C after P57 was incubated at 30◦C, 40◦C, or 45◦C for different time intervals (5, 10, 15, 20, 30, 40, or 60 min). The optimum pH of P57 was assayed at 40◦C in the following buffers (20 mM): Na2HPO4-citric acid (pH 4.0– 6.0), Na2HPO4-NaH2PO<sup>4</sup> (pH 6.0–8.0), Tris-HCl (pH 8.0–10.0), and Na2CO3-NaHCO<sup>3</sup> (pH 9.0–11.0). To evaluate the effect of NaCl on enzyme activity, the assay was carried out at 40◦C and pH 8.0 with different salt concentrations from 0 to 3 M. The effects of enzyme inhibitors (PMSF (phenylmethylsulfonyl fluoride), EDTA (ethylene diamine tetraacetic acid), EGTA (ethylene glycol tetraacetic acid), o-P (o-phenanthroline), and IA (iodoacetic acid)) and metal ions (K+, Li+, Ba2+, Ca2+, Co2+, Cu2+, Mg2+, Mn2+, Ni2+, Sr2+, Zn2+, and Fe3+) on the activity of P57 were evaluated by measuring the enzyme activity at 40◦C and pH 8.0 after the enzyme was pre-incubated with each inhibitor or metal ion for 1 h at 4 ◦C.

Substrate specificity of protease P57 was determined by measuring its activities toward casein, gelatin, collagen (Bovineinsoluble type I collagen fiber, Worthington Biochemical Corp., USA), elastin and synthetic peptides. The activities of protease P57 toward collagen and gelatin were measured by the method as described by Worthington Biochemical Corp. (Freehold, 1972). For collagen, the mixture of 1 mL enzyme solution (0.2 mg/mL) and 5 mg collagen was stirred at 37◦C for 5 h. One unit is defined as the amount of enzyme that released 1 µmol leucine from collagen in 1 h. For gelatin, the mixture of 100 µL enzyme solution (0.04 mg/mL) and 100 µL of 2% (w/v) gelatin were incubated at 40◦C for 10 min. One unit is defined as the amount of enzyme that released 1 µmol leucine from gelatin in 1 min. The activity of protease P57 toward elastin was assayed by the method of Chen et al. (2009). The activities of protease P57 toward synthetic peptides were determined in 20 mM phosphate buffer (pH 8.0) at 40◦C according to Peek's method (Peek et al., 1993). One unit was defined as the amount of enzyme that catalyzed the formation of 1 µmol p-nitroaniline in 1 min (Chen et al., 2007). The synthetic peptides used as substrate (0.2%, w/v) were as follows, N-succinyl-Ala-Ala-Pro-Leu-p-nitroanilide (AAPL), N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (AAPF), N-succinyl-Phe-Ala-Ala-Phe-p-nitroanilide (FAAF), N-succinyl-Ala-Ala-Pro-Arg-p -nitroanilide (AAPR), N-succinyl-Ala-Ala-Pro-Lys-pnitroanilide (AAPK) and N-succinyl-Ala-Ala-Val-Ala-pnitroanilide (AAVA). Protein concentrations were determined by the Bradford method (Bradford, 1976) with bovine serum albumin (BSA) as the standard.

#### Gene Cloning and Sequence Analysis of P57

Based on the N-terminal amino acid sequence of P57 and the conserved sequence of the catalytic center of serine proteases (Maciver et al., 1994), two degenerated primers (P57N and RM6) were designed (**Table 1**). With the degenerated primers and the genomic DNA of strain A5–7 as template, a part of the gene encoding P57 was amplified by PCR and sequenced at Shanghai Biosune Biotechnology Corp. (China). By using specific primers and general primers (**Table 1**), the neighboring sequences of the obtained gene fragment were amplified by thermal asymmetric interlaced PCR (TAIL-PCR) (Liu and Whittier, 1995). Through assembly, an entire gene sequence containing an ATG start codon and a TAA stop codon was obtained. Two specific primers (A5–7N and A5–7C) were designed (**Table 1**) according to this ORF, and the full gene encoding P57 was amplified by PCR from the genomic DNA of strain A5–7, and verified by sequencing. The gene sequence of P57 was deposited in GenBank database under the accession number KT923662.

The domain architectures of P57 were analyzed by the Conserved Domain Database (CDD) of NCBI<sup>2</sup> (Marchlerbauer et al., 2007). Homologous sequences to P57 were searched using the BLAST against NCBI nr database and the MEROPS database<sup>3</sup> (Rawlings et al., 2006) with default parameters. The representative homologs to P57 were aligned by using DNAMAN software.

#### Expression and Purification of the PA Domain and Its Site-directed Mutants

The DNA fragment encoding the PA domain of protease P57 with an overlapping sequence of EGFP was amplified by PCR using the genomic DNA of strain A5–7 as the template and two primers (PA-N, PA-GFP1) (**Table 1**). The DNA fragment encoding EGFP with the same overlapping sequence was also amplified by PCR using the vector pEGFP-N1 (Clontech, USA) as the template and the primers PA-GFP2 and PA-GFP(C) (**Table 1**). The two fragments were concatenated by the overlapping extension PCR (Liu and Whittier, 1995). The chimeric gene was sub-cloned into the NdeI-XhoI site of pET-22b (+) to construct the expression vector pET-22b-PA-EGFP, which was then transformed into E. coli BL21 (DE3). The recombinant PA-EGFP was expressed as a C-terminal His6-tagged protein. The transformant cells were cultivated in a 100 mL LB medium supplemented with 100 µg/mg ampicillin at 37◦C and 180 rpm, until the absorbance at 600 nm reached 1.5. Then, isopropyl β-D-thiogalactopyranoside (IPTG) was added to a final concentration of 0.1 mM. After induction for 24 h at 15◦C and 150 rpm, cells were harvested by centrifugation. Cell pellets were suspended in a 35 mL lysis buffer (50 mM Tris-HCl pH 9.0) and disrupted by sonication. After centrifugation at

<sup>2</sup>http://www.ncbi.nlm.nih.gov/Structure/cdd

<sup>3</sup>http://www.merops.ac.uk

#### TABLE 1 | Primers used in this study.

fmicb-07-02016 December 20, 2016 Time: 17:27 # 4


Note, the underlined sequences indicate restricted enzyme sites.

4 ◦C and 12,000 g for 30 min, the supernatant was collected and the recombinant PA-EGFP protein was purified with a HisBind metal chelating column.

Alignment of the PA domains from P57 and other reported S8 subtilases were performed by using Clustal W. The conserved aromatic residues and positively charged residues were chosen for mutagensis. Site-directed mutagenesis on the PA domain was carried out by overlapping extension PCR (An et al., 2005) using the vector pET-22b-PA-EGFP as template. Mutated sites were introduced by the primers with singlepoint mutations. The mutated genes were sub-cloned into pET-22b (+) and transformed into E. coli BL21 (DE3). All mutations were confirmed by enzyme digestion and nucleotide sequencing. The expression and purification of the mutants were performed under the same conditions as those of PA-EGFP.

### Analysis of the Binding Ability of Wild P57, the PA Domain and Its Mutants to Bovine-Insoluble Type I Collagen Fiber

To analyze the binding ability of the PA domain and its mutants toward collagen, 5 mg collagen fibers were mixed with 500 µL PA-EGFP, EGFP or the mutants (0.5 mg/mL) in 50 mM Tris-HCl buffer (pH 8.0). The mixture was incubated at 37◦C for 2 h, and then centrifuged at 12,000 g and 4◦C for 10 min. The free fluorescence intensity in the supernatant before and after incubation was measured on a FP-6500 spectrofluorometer (Jasco, Japan). Fraction of fusion protein bound (%) = (B-A)/B\*100%, B and A represent the free fluorescence intensity in the solution before and after a fusion protein binds to collagen, respectively.

The collagen-binding ability of wild P57 was assayed as described by Itoi et al. (2006). Purified P57 (0.2 mg/mL, 0.5 mL) in 20 mM Na2HPO4-NaH2PO<sup>4</sup> (pH 7.0) containing 10 mM Fe3<sup>+</sup> was incubated at 4◦C for 1 h to inhibit the enzyme activity. Then collagen fibers (1, 5, or 10 mg) was added and the mixture was incubated at 37◦C for 2 h with stirring. After incubation, the mixture was centrifuged for 10 min at 12,000 g and 4◦C. The supernatant was analyzed by 12.5% SDS-PAGE. BSA as a negative control was treated as P57.

### Analysis of the PA Domain and Its Mutants by Circular Dichroism

Circular dichroism (CD) spectra of the purified PA-EGFP and its mutants F349A-EGFP and Y432A-EGFP in 50 mM Tris-HCl buffer (0.1 mg/mL) were measured on a Jasco J810 spectropolarimeter (Japan) according to the method described by Zhao et al. (2008).

### RESULTS

#### Phylogenetic Analysis of Strain A5–7

Strain A5–7 was isolated from marine sediment collected from A5 station in Jiaozhou Bay, China, and the nearly complete 16S rRNA gene sequence (1536 bp) of the strain was determined. Sequence comparison showed that strain A5– 7 shared the highest 16S rRNA gene sequence identities with known Photobacterium species, suggesting an affiliation with the genus Photobacterium. In the neighbor-joining and maximumlikelihood trees (**Figures 1A,B**) based on the 16S rRNA gene sequences, strain A5–7 fell within the clade of the genus Photobacterium and formed a distinct intra-branch with type strain of Photobacterium aplysiae supported by high bootstrap values (>85%), indicating its close phylogenetic relationship to the latter. However, considering that strain A5–7 has only 96.8% 16S rRNA gene sequence identity to the type strain of P. aplysiae, strain A5–7 may represent a potential new species of the genus Photobacterium, which merits further study.

### Purification and Characterization of Protease P57 Secreted by Photobacterium sp. A5–7

A protease was purified from the culture of strain A5–7 by anion exchange chromatography with a yield of 15.9%. SDS-PAGE analysis showed that the purified protease had high purity, with an apparent molecular mass of approximately 45 kDa (**Figure 2A**). This protease was named P57 in this study.

With casein as substrate, the optimum temperature of P57 was 40◦C, and 10% of the maximum activity was retained at 0 ◦C (**Figure 2B**). P57 was stable at 30◦C for at least 1 h, but unstable at temperatures higher than 30◦C. The half time of its activity at 40◦C and 45◦C was 40 min and 7 min, respectively (**Figure 2C**). P57 was active in a wide range of pH from 5.0 to 11.0, with the maximum activity at pH 8.0 (**Figure 2D**). In the buffers containing 0–3 M NaCl, the activity of P57 peaked at 0.25 M NaCl, then declined with the increase of NaCl concentration, but still retained 50% of the maximum activity at 1.5 M NaCl (**Figure 2E**).

Effects of metal ions and inhibitors on P57 activity were shown in **Table 2**. P57 activity was not affected by Li+, K+, Ba2+, Co2+, or Sr2+. At the final concentration of 8 mM, Cu2+, Zn2+, and Ni2<sup>+</sup> obviously inhibited P57 activity, and Fe3<sup>+</sup> severely inhibited P57 activity by 96.4%. While Ca2<sup>+</sup> and Mg2<sup>+</sup> slightly activated the enzyme activity, Mn2<sup>+</sup> significantly increased P57 activity to 194.1%, showing an activating effect on P57 activity. In addition, P57 activity was strongly inhibited by PMSF, a serine protease inhibitor, indicating that P57 is likely a serine protease. The activity of P57 was also inhibited by metal chelators EDTA and EGTA, but not affected by o-P or IA.

Substrate specificity analysis showed that P57 could hydrolyze casein, gelatin and collagen, with the highest activity toward gelatin, the degenerated form of collagen (**Table 3**). P57 showed no activity toward elastin. Among the synthetic peptides, P57 had higher activities toward AAPL and AAPF and lower activities toward FAAF, AAPR, AAPK and AAVA (**Table 3**).

### Gene Cloning and Sequence Analysis of P57

The N-terminal sequence of mature P57 was determined by protein sequencing to be Ser-Gln-Ser-Leu-Pro-Trp-Gly-Gln-Thr-Phe-Val-Gly-Ala-Thr-Leu, which shows identities to some serine proteases of the S8 family, such as SapSh (66.7%) from Shewanella sp. Ac10 (Kulakova et al., 1999), Apa1 (66.7%) from Pseudoalteromonas sp. AS-11 (Dong et al., 2001), VapT (40%) from V. metschnikovii RH530 (Kwon et al., 1995) and AcpII (40%) from A. collagenimarina AC40 (Kurata et al., 2010). This suggests that P57 is a serine protease of the S8 family. Based on the N-terminal amino acid sequence of P57 and the highly conserved sequence of the catalytic domains of the S8 serine proteases, the gene encoding protease P57 was cloned from the genomic DNA of strain A5–7 by a combination of PCR and TAIL-PCR.

The ORF of P57 is 2,037 bp in length, including an ATG start codon and a TAA stop codon. It encodes a protein of 678 amino acid residues with a calculated molecular weight of 71,788 Da. According to BLAST analysis against CDD database (Marchlerbauer et al., 2007), the precursor of P57 contains an N-terminal pre-sequence (Met1-Leu123), an S8 catalytic domain (Asn146-Ala565), a PA domain (Asp340-Ser479) inserted in the catalytic domain, a linker (Glu566-Leu599) and a C-terminal P-proprotein (Thr600-Lys678) (**Figure 3A**). SignalP 3.0 (Bendtsen et al., 2004) prediction suggested that the pre-sequence of P57 contains an N-terminal signal peptide sequence (Met1-Ala29). According to the result of N-terminal sequencing, the first N-terminal residue of mature P57 is Ser124. The molecular mass of mature P57 was determined to be 46,016 Da by MALDI-TOF mass spectrometry. Thus, based on the sequence and the molecular mass of P57, it was

(B) Effect of temperature on P57 activity. Enzyme activity was measured at 0–70◦C in 20 mM phosphate buffer (pH 7.0). (C) Thermostability of P57. The enzyme was incubated at 30◦C, 40◦C, or 45◦C for different time periods, and then residual activity was measured at 40◦C. (D) Effect of pH on the activity of P57. Enzyme activity was measured at 40◦C in different pH buffers (20 mM): Na2HPO4-citric acid (pH 4.0–6.0), Na2HPO4-NaH2PO<sup>4</sup> (pH 6.0–8.0), Tris-HCl (pH 8.0–10.0), and Na2CO3-NaHCO<sup>3</sup> (pH 9.0–11.0). (E) Effect of NaCl on the activity of P57. Enzyme activity was measured at 40◦C in different concentrations of NaCl. The graph shows data from triplicate experiments (mean ± SD).



<sup>a</sup>The activity of P57 without any metal ion or inhibitor was taken as control. <sup>b</sup>The final concentration of metal ion in the reaction mixture. <sup>c</sup>The final concentration of inhibitor in the reaction mixture.

∗ , ∗∗ stand for different significance at p < 0.05 and p < 0.01, respectively. The data shown are the means of three repeats with standard deviations. PMSF, phenylmethylsulfonyl fluoride; EDTA, ethylene diamine tetraacetic acid; EGTA, ethylene glycol tetraacetic acid; o-P, o-phenanthroline; IA, iodoacetic acid.

predicted that mature P57 contains 442 residues from Ser124 to Ala565. Therefore, both the N-terminal pre-sequence and the C-terminal P-proprotein are cleaved off during enzyme maturation, and mature P57 only contains the catalytic domain and the inserted PA domain (**Figure 3A**). Multiple sequence alignment suggests that the catalytic triad of P57 is composed of Asp153, His188 and Ser492 (**Figure 3B**). The highest sequence identity of P57 with reported S8 serine proteases is 63% according to BLASTP analysis against NCBI non-redundant protein database.

### Collagen Binding Ability of the PA Domain

To study the function of the PA domain in P57 catalysis, the PA domain fused with EGFP (PA-EGFP) was expressed in E. coli BL21 (DE3), and purified by HisBind metal chelating column with a yield of 51.7%. Because P57 could hydrolyze insoluble collagen, we investigated the binding ability of the recombinant PA domain to insoluble collagen by fluorescence analysis. After PA-EGFP was mixed with insoluble collagen fibers for 2 h at 37◦C,

TABLE 3 | Substrate specificity of P57 toward various proteins and synthetic peptides<sup>a</sup> .


<sup>a</sup>The data shown are the means of three repeats with standard deviations. –, No detectable activity. AAPL, N-succinyl-Ala-Ala-Pro-Leu-p-nitroanilide; AAPF, N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide; FAAF, N-succinyl-Phe-Ala-Ala-Phe-p-nitroanilide; AAPR, N-succinyl-Ala-Ala-Pro-Arg-p-nitroanilide; AAPK, N-succinyl-Ala-Ala-Pro-Lys-p-nitroanilide; AAVA, N-succinyl-Ala-Ala-Val-Ala-p-nitroanilide.

the precipitated collagen fibers displayed a bright green color, whereas the collagen fibers mixed with EGFP still displayed its own white color (**Figure 4A**). This result suggests that the PA domain of P57 has collagen-binding ability. To further confirm this, we quantified the fluorescence intensity changes of the supernatant with different amounts of collagen in the mixture. As shown in **Figure 4B**, the fluorescence intensity in the supernatant decreased from 500 to 360 with the increase of collagen amount from 0 to 10 mg in the mixture, which indicated that the amount of PA-EGFP bound to collagen fibers increased with the increase of collagen amount in the mixture, thereby leading to a successive decrease of PA-EGFP amount in the solution. Taken together, these results indicate that the PA domain of P57 has collagenbinding ability, implying that the PA domain in P57 may be involved in substrate binding in the catalysis of P57 toward collagen.

In addition, to estimate the contribution of the PA domain to the interaction of P57 with collagen, the collagen-binding ability of P57 was assayed after the enzyme activity was inhibited by 10 mM Fe3+. As shown in **Figure 4C**, after P57 was incubated with collagen at 37◦C for 2 h in the presence of 10 mM Fe3+, the amount of P57 protein in the supernatant significantly decreased, whereas the amount of BSA in the supernatant showed little change after incubation with collagen (**Figure 4C**). This result indicated that P57 can bind to collagen at 37◦C. The PA domains probably play an important role in the interaction of P57 with collagen because it has collagen-binding ability (**Figures 4A,B**).

### Key Residues for Collagen Binding in the PA Domain

To determine the key amino acid residues in the PA domain of P57 for collagen binding, site-directed mutagenesis was performed. It has been reported that aromatic residues and charged residues usually play a key role in the binding of binding domains to insoluble substrates (Bhaskaran et al., 2008; Philominathan et al., 2009). Based on sequence alignment of the PA domains from P57 and other reported proteases (**Figure 3B**), conserved aromatic residues (Phe349, Tyr412, Tyr432, Phe444 and Tyr452) and conserved positively charged residues (Lys397, Arg403 and Lys419) in the PA domain of P57 were mutated to Ala, and all the mutants were expressed as EGFP-fused proteins. The mutants R403A-EGFP and F444A-EGFP could not be expressed as soluble proteins, probably because these two residues are important for the correct folding of the proteins. The binding abilities of the mutants to collagen were measured and compared with that of PA-EGFP. While the collagen-binding abilities of mutants K397A-EGFP, Y412A-EGFP, K419A-EGFP and Y452A-EGFP were partly reduced, the collagen-binding abilities of mutants F349A-EGFP and Y432A-EGFP were severely destroyed, implying that Phe349 and Tyr432 may be essential for the PA domain to bind collagen (**Figure 5A**). In addition, CD spectra of PA-EGFP, F349A-EGFP and Y432A-EGFP were collected to analyze the structural changes that residue substitution mutation may cause in the mutants. There were few differences in the spectra of PA-EGFP, F349A-EGFP and Y432A-EGFP (**Figure 5B**), suggesting that the loss of the collagen-binding ability of F349A-EGFP and Y432A-EGFP should be caused by amino acid substitution. Altogether, these data indicate that Phe349 and Tyr432 are likely two key residues for collagen binding in the PA domain of P57. The data also suggest that hydrophobic interactions likely play an important role in the binding of the PA domain to collagen because Phe349 and Tyr432 are both aromatic amino acids.

### DISCUSSION

In this study, the extracellular protease P57 from Photobacterium sp. A5–7 isolated from the sea sediment of Jiaozhou Bay in China was characterized. P57 is a serine protease of the S8 family. Among reported S8 serine proteases, P57 has the highest identity (63%) to VapT from V. metschnikovii RH530 (Kwon et al., 1995), which suggests that P57 is a new member of the S8 family. Although the precursor of P57 contains an N-terminal presequence, an S8 catalytic domain, a PA domain and a C-terminal P-proprotein domain, mature P57 only contains the catalytic domain and the PA domain that is inserted in the catalytic domain. The N-terminal pre-sequence, which is required for protease folding and secretion (Chen and Inouye, 2008), and the C-terminal P-proprotein domain, which is also important in enzyme secretion through the cell membrane (Kurata et al., 2007), are all cleaved off during P57 maturation.

Our results show that P57 is a subtilase with an inserted PA domain in its catalytic domain. Some subtilases with an inserted PA domain in the catalytic domain have been reported from plants and bacteria, and structural and deletion mutational analyses have suggested that the PA domains in some subtilases are involved in substrate binding. For example, the crystal structures of C5a peptidases from S. agalactiae and S. pyogenes show that the PA domain is located near the active-site cleft of the catalytic domain (Brown et al., 2005; Kagawa et al., 2009). Three out of four residues that form the S4 subsite belong to the PA domain in soybean protease C1 (Tan-Wilson et al., 2012). Cleavage specificity toward natural casein substrate was

changed in PrtP from Lactococcus lactis without the PA domain (Bruinenberg et al., 1994). The activity of AcpII-1PA toward gelatin, casein and collagen was remarkably increased (Kurata et al., 2010). Despite these structural and deletion mutational analyses, there is no direct evidence that the PA domain from a subtilase has substrate-binding ability. In this study, the results showed that the PA domain from the subtilase P57 had collagenbinding ability, and that residues Phe349 and Tyr432 are two key residues responsible for collagen binding in the PA domain, showing direct evidence that the PA domain of a subtilase has substrate-binding ability. These results also suggest that the function of the PA domain in P57 is most likely to participate in the binding of insoluble substrate such as collagen during P57 catalysis.

While most subtilases have no collagenolytic activity, some subtilases have been reported to be serine collagenolytic

FIGURE 4 | Analysis of the collagen-binding ability of the recombinant PA domain. (A) The binding ability of EGFP (left tube) and PA-EGFP (right tube) on bovine-insoluble type I collagen. After the mixture of EGFP or PA-EGFP with 5 mg collagen was incubated at 37◦C for 2 h with agitating and subsequently centrifuged, the precipitated collagen fibers were washed two times with distilled water. (B) Fluorescence analysis of the collagen-binding ability of the PA domain. The fusion protein PA-EGFP was incubated with various amounts of collagen at 37◦C for 2 h. After centrifugation, the fluorescence intensities of the supernatants were determined at 510 nm. PA-EGFP incubated with no collagen at the same conditions was served as control. The graph shows data from triplicate experiments (mean ± SD). (C) SDS-PAGE analysis of the binding ability of P57 to bovine insoluble type I collagen in the presence of 10 mM Fe3+. P57 in 20 mM Na2HPO4-NaH2PO<sup>4</sup> (pH 7.0) containing 10 mM Fe3<sup>+</sup> was incubated at 4◦C for 1 h to inhibit the enzyme activity, and then was mixed with 0, 1, 5, or 10 mg collagen. The mixtures were incubated at 37◦C for 2 h, and the unbound P57 fractions in the mixtures were analyzed by 12.5% SDS-PAGE. BSA was used as a negative control. M, protein molecular mass marker.

proteases that can degrade insoluble collagen, such as MCP-01 from Pseudoalteromonas sp. SM9913 (Zhao et al., 2008; Ran et al., 2013), the thermostable protease from Geobacillus collagenovorans MO-1 (Okamoto et al., 2001), AcpII from Alkalimonas collagenimarina AC40 (Kurata et al., 2010), and myroicolsin from Myroides profundi D25 (Ran et al., 2014). P57 has collagenolytic activity and its PA domain has collagenbinding ability, indicating that P57 is an S8 serine collagenolytic protease. Subtilases, such as SapSh (Kulakova et al., 1999), VapT (Kwon et al., 1995) and AcpII (Kurata et al., 2010), usually have activity toward synthetic peptides AAPL and AAPF. Consistent with this, P57 has obvious activity toward these two peptides. Metal ions have different effects on the activity of different enzymes. It was reported that Ca2<sup>+</sup> could increase the enzyme activity or the thermal stability of some subtilases (Kurata et al., 2007, 2010; Zhao et al., 2008; Ran et al., 2014). However, Ca2<sup>+</sup> only had a little effect on P57 activity. Previous reports showed that Mn2<sup>+</sup> did not affect the enzyme activity of subtilase significantly (Kurata et al., 2007) or slightly inhibited the enzyme activity of subtilase (Zhao et al., 2008; Ran et al., 2014). Our result showed that 8 mM Mn2<sup>+</sup> significantly increased P57 activity. EDTA and EGTA are ion chelators, which usually inactivate subtilase that contains metal ion and/or lower their stability by depriving the metal ion (Kurata et al., 2010). The activity of P57 was significantly reduced by EDTA and EGTA, implying that P57 may contain metal ion.

Extracellular proteases from marine sedimentary bacteria usually have some environment-adapted characters, including

cold-adaptation, salt tolerance/activation, and an optimal pH equal or near that of seawater, such as proteases MCP-01 and MCP-03 from Pseudoalteromonas sp. SM9913 (Chen et al., 2003; Zhao et al., 2008; Yan et al., 2009) and pseudoalterin from Pseudoalteromonas sp. CF6-2 (Zhao et al., 2012). Consistent with this, P57 also shows some characters adapted to marine sediment environment. Photobacterium sp. A5–7 was isolated from a marine sediment sample from the A5 station site in Jiaozhou Bay, China, where the depth, temperature, pH and C/N ration were 5.9 m, 24.7◦C, 8.11 and 7.0, respectively (Zhang et al., 2015). P57 had low optimal temperature (40◦C), and low thermostability at moderate temperatures (unstable at temperatures higher than 30◦C), and displayed high activity at alkaline pH (pH 7.0– 9.0) with the optimum pH of 8.0. P57 showed the highest activity at 0.25 M NaCl, and retained 50% of the maximum activity at 1.5 M NaCl. These characters reflect the adaptation of P57 to the marine sedimentary environment. Extracellular proteases of marine sedimentary bacteria play important roles in PON degradation and nitrogen cycling in marine sediments. As a bacterial extracellular protease from marine sediment can degrade proteins and peptides, P57 should be actively involved in sedimentary PON degradation.

In summary, protease P57 secreted by marine sedimentary Photobacterium sp. A5–7 was characterized in this study. P57 is a new S8 subtilase that can degrade casein, collagen and gelatin. P57 has a PA domain inserted in its catalytic domain. The PA domain

#### REFERENCES


from P57 has collagen-binding ability, which represents the first direct evidence that the PA domain of a subtilase has substratebinding ability. Phe349 and Tyr432 were further revealed to be key residues responsible for collagen binding in the PA domain of P57 by site-directed mutational analysis. Our results provide more insight into the function of PA domain of subtilases and shed light on marine sedimentary bacterial proteases and PON degradation.

#### AUTHOR CONTRIBUTIONS

H-JL, B-LT, and X-XH performed the biochemical experiments. XS, B-XL, and X-YZ helped in protein purification. X-LC designed and directed the research. H-JL and X-LC wrote the manuscript. P-YL and X-YZ helped in data analysis and manuscript editing.

#### ACKNOWLEDGMENTS

This work was supported by the National Science Foundation of China (31290230, 31290231, 91228210, 41276149, 31270117, 31270064), the Hi-Tech Research and Development Program of China (2014AA093509), and the Fundamental Research Funds of Shandong University (2014QY006).

activity of tomato subtilase 3 (SlSBT3). J. Biol. Chem. 284, 14068–14078. doi: 10.1074/jbc.M900370200


Streptococcus pyogenes: a 1.9 Å crystal structure of the active form of ScpA. J. Mol. Biol. 386, 754–772. doi: 10.1016/j.jmb.2008.12.074


**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 Li, Tang, Shao, Liu, Zheng, Han, Li, Zhang, Song and Chen. 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.

fmicb-07-02016 December 20, 2016 Time: 17:27 # 11