Morphological and transcriptomic evidence for ammonium induction of sexual reproduction in Thalassiosira pseudonana and other centric diatoms

The reproductive strategy of diatoms includes asexual and sexual phases, but in many species, including the model centric diatom Thalassiosira pseudonana, sexual reproduction has never been observed. Furthermore, the environmental factors that trigger sexual reproduction in diatoms are not understood. Although genome sequences of a few diatoms are available, little is known about the molecular basis for sexual reproduction. Here we show that ammonium reliably induces the key sexual morphologies, including oogonia, auxospores, and spermatogonia, in two strains of T. pseudonana, T. weissflogii, and Cyclotella cryptica. RNA sequencing revealed 1,274 genes whose expression patterns changed when T. pseudonana was induced into sexual reproduction by ammonium. Some of the induced genes are linked to meiosis or encode flagellar structures of heterokont and cryptophyte algae. The identification of ammonium as an environmental trigger suggests an unexpected link between diatom bloom dynamics and strategies for enhancing population genetic diversity.


Introduction
Diatoms are protists that form massive annual spring and fall blooms in aquatic environments and are estimated to be responsible for about half of photosynthesis in the global oceans [1]. This predictable annual bloom dynamic fuels higher trophic levels and initiates delivery of carbon into the deep ocean biome. Diatoms have complex life history strategies that are presumed to have contributed to their rapid genetic diversification into~200,000 species [2] that are distributed between the two major diatom groups: centrics and pennates [3]. A defining characteristic of all diatoms is their restrictive and bipartite silica cell wall that causes them to progressively shrink during asexual cell division. At a critically small cell size and under certain conditions, auxosporulation restitutes cell size and prevents clonal death [4][5][6]. The entire lifecycles of only a few diatoms have been described and rarely have sexual events been captured in the environment [7][8][9]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 concentration, with up to 39% of the population identified as oogonia or auxospores in cultures supplemented with 800 μM NH 4 Cl (Fig 2A). Oogonia and auxospores were first observed at the onset of stationary phase and reached maximum population proportions in late stationary phase (Fig 2A). Ammonium also induced oogonia and auxospore production in T. pseudonana CCMP1015 (collected from the North Pacific Ocean), T. weissflogii, and Cyclotella cryptica (S1 Fig). A few oogonia and auxospores were observed in nitrate grown cultures with no added ammonium (Fig 2A and S1 Fig). However, with the exception of nitrate grown C. cryptica cultures, which generated oogonia and auxospores constituting 11% of the total population, oogonia and auxospores were only a small percentage of the total population in nitrategrown T. pseudonana and T. weissflogii. Even though auxospores can have diameters 3-4 times that of asexual cells, such small population proportions do not lead to discernable shifts in cell size distributions obtained by particle size analysis (e.g., Coulter counter, a commonly used method to assess population size). We initially observed auxospores when performing visual inspections using a light microscope of our cultures that were growing in ammonium. For the The average cell size of a population of asexually dividing diatoms decreases as a result of differential thecae inheritance. At a critically small size, cells can initiate sexual reproduction and differentiate into male and female cells. Meiosis in the male spermatogonangium produces multinucleate spermatogonia that divide into individual haploid spermatocytes. Meiosis in the female oogonia produces a single functional haploid nucleus that is fertilized by a flagellated spermatocyte through an opening in the oogonia thecae. Fertilized oogonia expand into a large auxospore where new, large thecae are formed for the new initial cell.
https://doi.org/10.1371/journal.pone.0181098.g001  data reported here, oogonia and auxospores were quantified by manually counting the cell types using a hemocytometer. We suspect that reliance on laboratory instruments such as particle counters and flow cytometers in place of microscopic analysis is one reason that sexual morphologies in these well-studied diatom species have gone undetected until now. Laboratory stock cultures are typically maintained in media with low concentrations of nitrogen, especially when ammonium is supplied as the nitrogen source because it has been considered to be toxic to diatoms in high concentrations [27]. Therefore, it may not be surprising that sexual cells have gone un-noticed due to the low rates of sexual induction in the presence of low ammonium concentrations (Fig 2A).
Cells differentiated into oogonia and auxospores regardless of whether ammonium was supplied at inoculation or at the onset of stationary phase ( Fig 1B). Thus, it appears that stationary phase and ammonium availability are key factors that trigger formation of sexual cells in centric diatoms. Resource depletion can arrest the cell cycle [14], and the presence of ammonium at the onset of stationary phase appears to activate the sexual cycle. Auxospore formation was observed in Cyclotella meneghiniana during stationary phase [28], and other protists initiate sex under stress in response to nutrient depletion or oxidative DNA damage [29]. Ammonium can inhibit photosynthesis [27]; however, diatoms, including T. pseudonana, can acclimate to millimolar ammonium concentrations [30]. It is possible that high ammonium concentrations intensify the stress condition required for the sexual pathway. Nevertheless, ammonium consistently caused formation of ten-fold more sexualized cells than the same concentrations of nitrate (Fig 2 and S1 Fig).
Our results showing that ammonium induced formation of sexual cells in several centric diatom species suggests that it may serve as a key environmental factor regulating the sexual lifecycle across centric diatoms. Ammonium is typically present in very low concentrations in aquatic ecosystems. However, ammonium reached 12.6 mM in a eutrophic lake where the centric diatom, Aulacoseira subarctica, was observed undergoing sexual reproduction [8]. Clearly, ammonium was not the growth-limiting nutrient under those conditions or in our laboratory cultures (Fig 2). Pseudo-nitzschia auxospore formation was positively correlated with ammonium, which was measured to be 14 μM during a major bloom event off the coast of Washington [7]. Thus, the formation of sexual cells appears to be triggered by the presence of ammonium while at least one other growth factor becomes limiting, such as light (discussed below), phosphorous, silica [7], vitamins, or trace elements.
Cell differentiation in T. pseudonana was induced irrespective of growth rate in exponential phase, light intensity, or light regime. However, the growth parameters did affect the proportion of differentiated cells. Oogonia and auxospores were only 0.5% of the population when grown under very low light (5 μE) with 200 μM NH 4 Cl and increased to 39% when grown under moderate light (100 μE) and 800 μM NH 4 Cl ( Table 1). The proportions of oogonia and auxospores increased with light up to moderate intensities (70-100 μE) but decreased at high (220 μE) intensities ( Table 1), suggesting that photon flux has an important role in meeting the energetic demands of sexual reproduction. Other work showed sexualization was more prevalent at light <50 μE [31] or with the addition of a dark period [13,32,33]. Likely, the optimum light intensity or need for a dark period to precede sexual induction [34] is species-specific and linked to adaptive life histories [5].

Visualization of sexual morphologies
Confocal, light, and scanning electron microscopy were used to document the cell morphologies at various stages in the life history cycle of T. pseudonana (Fig 3 and S2 Fig). Oogonia were elongated relative to asexually growing vegetative cells and exhibited a bent morphology and swelling of the plasma membrane at the junction of the hypotheca and epitheca ( In oogonia, cellular contents became localized to the ends of the cell resulting in an apparent empty space near the area of membrane swelling where fertilization likely occurs [5]. T. pseudonana spermatogonium harbored at least eight nuclei (Fig 3D), suggesting that a depauperating mitosis preceded meiosis [11,35]. Sperm released were very small, about 1 μm, and flagellated ( Fig 3E, S2E and S2F Fig), but they often became entangled with other cells and debris [8,36,37]. Sperm cells attached to oogonia at the junction of the thecae for fertilization (Fig 3F and 3G, S2I and S2J Fig) as shown in T. punctigera [38]. Also similar to T. punctigera, flagella were not visible at that stage, possibly because flagella are abandoned upon attachment to the oogonia [36].
Changes in DNA content in T. pseudonana cells induced into sexuality by ammonium were observed using flow cytometry-based analysis. Fluorescence-activated cell sorting (FACS) analysis showed that as the culture progressed from exponential into stationary growth phases, the diploid population ( Fig 4B) expanded to include DNA fluorescence intensities that were consistent with the presence of spermatogonangia and spermatogonia containing multiple gametes ( Fig 4C). In late stationary phase, a new population was observed that had DNA fluorescence signals consistent with haploid sperm cells with little to no chlorophyll [13] (Fig 4D).
Cell size restitution via auxosporulation produced progeny cells that were considerably larger than the parent cells from nitrate stock cultures. To induce a high proportion of the eligible cells into the sexual pathway we repeatedly propagated cultures in 800 μM ammonium with inoculum of 25%. This strategy raises the ratio between the exposure of cells to ammonium in stationary phase and the total number of cell divisions. Our findings can be explained Table 1. Effects of growth parameters on induction of sexual reproduction in T. pseudonana CCMP1335, T. weissflogii and C. cryptica. Oogonia and auxospores always appeared in stationary phase. The percentage of the total population (at least 300 cells counted per replicate) differentiated into oogonia or auxospores when grown in nitrate or ammonium is shown for the day they were at their maximum number; data are mean values ± s.d., biological replicates n = 3.

Species
Light intensity (μE) by assuming that cells in nitrate stock cultures are already at or below the critical size threshold for induction into sexuality, but with each passage through growth and stationary phases in batch culture, only a fraction of the eligible cells are induced into the sexual cycle. The average cell diameters of the resulting cultures were larger relative to stock cultures maintained in nitrate ( Fig 4E, S3 Fig). The T. pseudonana initial cells were 7-12 μm, the largest size reported for this species (Fig 3J and 3K, S2K Fig). Presuming that cell size reduction during vegetative growth occurs in T. pseudonana, this process of cell size reduction and cell size restitution via ammonium induction have opposing influences on the average size of populations. These processes confound the ability to observe the impacts of sexual induction without experimental designs that maximize the percentage of the population induced into the sexual pathway and minimize the number of vegetative replications between episodes of induction. We identified oogonia, male gametes, auxospores, and initial cells in cultures of the model centric diatom, T. pseudonana providing new evidence for sexuality in this species that was previously assumed to be asexual [10]. Although cell enlargement through asexual/apomictic mechanisms has been recorded in other species [42][43][44], the presence of all sexual cell types, and the expression of meiotic genes (discussed below), suggest apomixis is not the mechanism being used by T. pseudonana for cell enlargement. Furthermore, apomixis typically occurs in species that also undergo sexual reproduction [5]. Only spermatogenesis had previously been reported in T. weissflogii [13,45], but we have now also documented induction of oogonia and auxospores by ammonium and subsequent formation of initial cells in this species. A major challenge in visualizing the morphological characteristics of these species is their smaller cell sizes compared to other species for which morphological details have been documented. Now Highlighted in red are the four strains induced into sexual reproduction in this study. Species for which some evidence already exists for sexual reproduction are starred [9,13,16,28,33,38]. B-D: Changes in DNA and chlorophyll fluorescence in exponential (EXP), stationary (STA) and late stationary (L-STA) growth phases of T. pseudonana induced by ammonium; 30,000 events recorded, representative of two biological replicates. E: Coulter Counter distributions of cell diameter for T. pseudonana cultures in exponential phase of growth and maintained in NaNO 3 (red) and after six successive 25% transfers to medium with ammonium (blue). Each new culture was allowed to remain in stationary phase for three days before the next 25% transfer was made. Single replicates of cultures with cell densities of 2.4 x 10 6 ml -1 (NaNO 3 ) and 2.3 x 10 6 ml -1 (ammonium). Dashed lines are the mode for each peak.
https://doi.org/10.1371/journal.pone.0181098.g004 that we have determined a reliable method for inducing the sexual morphologies, future studies will dissect additional details associated with the sexual pathways in these, and perhaps other species inducible by ammonium, to determine their variation from other centric diatoms. For example, the presence of auxospore scales, precise timing of fertilization and meiotic activity, repeated auxosporulation, and polyspermy events (e.g., [38]). The case of T. pseudonana also presents interesting questions about whether this species has retained the ability to reduce in cell size. It appears that T. pseudonana has the capacity to avoid clonal death by maintaining a relatively constant cell size (3-9 μM) [46][47][48]. Our experiments show that cells in this size range are inducible into the sexual pathway. Nevertheless, the question remains whether the progeny of induced small cells of T. pseudonana are capable of cell size reduction.

Gene expression analysis of ammonium induced sexual morphologies
We used RNAseq to identify genes that were differentially expressed in conditions that triggered cell differentiation into sexual morphologies. We compared the transcriptomes of T. pseudonana harvested in exponential (EXP), stationary (STA), and late stationary phases (L-STA). Cells were grown in 100 μM NaNO 3 or, to capture a dose-dependent change in gene expression, either 100 or 800 μM NH 4 Cl (S4 Fig). We identified genes that were significantly differentially expressed in multiple pairwise comparisons of growth phases and nitrogen sources (S1A-S1K Table). Next, we examined the statistical interactions of pairwise condition comparisons to identify genes with significantly greater or lesser magnitude changes in expression between growth stages in the presence of ammonium relative to 100 μM NaNO 3 (Fig 5A, S5 Fig).
This conservative approach yielded a total of 1,274 genes in the four analyses of statistical interactions (S1L-S1O Table). A total of 89 of the genes have an annotated function (Fig 5A,  S1P Table). The set of 89 genes includes four meiotic genes (mcm2, mcm8, mcm9, and mlh1, Fig 4B, S6 Fig) that were also up-regulated in pennate diatoms during sexual reproduction [26]. The Mcm family of DNA helicases function in DNA replication, with mcm8, mcm9, and mlh1 having roles in double stranded break repair [49][50][51]. Mcm8 was one of the four genes related to meiosis that were upregulated in the pennate diatom, Seminavis robusta during treatment with a sex-inducing pheromone [17]. Eight genes in our list are homologous to yeast genes involved in meiosis [52] (Fig 5B, S1Q Table), including genes that regulate the meiotic anaphase promoting complex (cdc16, cdc23, ama1) [53,54] and rad54, a motor protein that regulates branch migration of Holliday junctions during homologous recombination [55]. Expression of genes encoding other RAD proteins (rad50 and rad 51) increased in pennate diatoms induced into meiosis [17,26].
Of three 'sexually induced genes' that were up-regulated in T. weissflogii at the initiation of gametogenesis [45] and associated with sperm flagella mastigonemes [56], one, sig3, was significantly up-regulated in stationary phase compared to exponential phase (Fig 5B). In addition, a gene encoding an intraflagellar transport protein (IFT88) was also up-regulated in ammonium induced cells during stationary phase (Fig 5B). An IFT system is required for flagellar assembly [57] and five genes encoding IFT particle proteins, including IFT88, and a kinesinassociated protein involved in anterograde transport were found in the T. pseudonana genome [58]. The genes encoding Sig3 and IFT88 are unique to flagellar structures, and their differential expression in ammonium induced T. pseudonana compared to the nitrate-grown control treatments provide additional evidence that ammonium induced spermatogenesis in this species.
The MYB factor and bZIP families of proteins are transcriptional regulators that control a variety of cell processes including stress responses, development, and differentiation in plants [59]. Expression of two genes having the characteristic R2R3 MYB DNA binding domains common in plants, myb24 and myb16, was generally lower in ammonium induced cultures compared to the nitrate grown controls (Fig 5B). In Arabidopsis thaliana, hormonal activation of myb24 is required for stamen development and male fertility [60]. Whether myb24, myb16, and bzip2 play roles in regulating gamete development or sex differentiation in diatoms remains to be determined. The 1,274 genes provide new avenues to understand the evolution of sexuality in the Heterokont eukaryotic lineage. Diatoms emerged~200 Mya, about 800 My after a eukaryotic heterotroph engulfed a red alga in the secondary endosymbiosis event that gave rise to the SAR eukaryotic supergroup [61]. Of 171 diatom genes of red algal origin [61], 17 were identified as differentially expressed in conditions that induced sexual reproduction (S1R Table). None of these genes are annotated in the T. pseudonana genome, but in red algae they are predicted to function in transport and plastid-targeted processes [23].

Conclusions
That some of the most well studied centric diatoms were never observed undergoing sexual reproduction was a mystery. Possibly even more elusive was the ability to reliably control or induce the sexual pathway of centric diatoms in the laboratory [10] despite a myriad of efforts that ranged from sweet-talk to torture. Factors that have limited progress in this field center on the problem that even under 'favorable environmental conditions' that result in the sexual lifecycle, only a fraction of cells undergo sexual reproduction (Fig 2). Thus, capturing the sexual event requires near constant visual observation because (a) only cells that have become sufficiently small and reach the critical size threshold can undergo sexual reproduction [62], (b) only a fraction of those size-eligible cells may undergo sexual reproduction [8,15,16,31], (c) there has been a lack of understanding about what constitutes conditions that are 'favorable' for triggering diatom sex [5,7], and (d) morphological changes indicative of sex may not be recognized by untrained scientists [63].
Our results provide strong evidence that T. pseudonana is a sexual organism, expressing the major morphologies associated with the sexual pathway that result in enlarged initial cells. Furthermore, the sexual pathway was reliably induced in T. pseudonana, and two other centric diatom species by exposure of size-eligible cells to ammonium. Ammonium triggered formation of sexual cells in a dose-dependent manner and significant changes in expression of genes involved in meiosis, spermatocyte flagellar structures and assembly, and sex differentiation. RNAseq analysis revealed many more genes with unknown functions that were expressed under conditions of sexual differentiation. Other genes involved in sex are likely to have been missed by our analysis because their changes in expression were masked by the mixed population of asexual and sexual cells, or they were not captured in the coarse time-resolution of sampling used in this study. Nevertheless, our discoveries resolve two persistent mysteries that have plagued diatom researchers. Furthermore, the RNAseq data provide a subset of genes that can be used to study the molecular ecology of diatoms.
The ecology of centric diatom sexual reproduction that can be inferred from our findings appears best described as synchronous sexuality [12] triggered by ammonium in cells experiencing growth stress. Asexual cell cycle arrest appears to be prerequisite to activation of the diatom sexual life cycle [13,14,28,29]. In the environment, diatoms bloom following elevated nutrient concentrations driven by vertical mixing, coastal upwelling, or river inputs and the bloom reaches its peak biomass when essential nutrients are depleted. Within a week, the bulk of a bloom can be consumed by heterotrophic protists [64] that excrete ammonium to maintain homeostatic elemental composition [65]. We propose that ammonium released by grazers at bloom climax may be a principal ecological trigger for sexual morphologies in centric diatoms. Synchronization of sexuality at the onset of resource depletion (stationary phase) increases the chances for successful fertilization because cell density is at its maximum [12]. Environmental concentrations of ammonium in the environment rarely reach the concentrations used in this study to demonstrate the dose response effect on sexuality. Other methods that have sometimes successfully triggered sexual reproduction in other species are similarly unusual compared to environmental conditions. For example, the magnitude of the salinity shifts used to induce sexual reproduction in Skeletonema marioni in the laboratory do not occur in the Baltic sea [16]. Nevertheless, pulses of ammonium, shifts in salinity, and other environmental fluctuations do occur in aquatic ecosystems, and provided the other conditions for sexuality are met (e.g., cell size threshold, stress, population density), are likely to induce sexuality in at least a small fraction of a population. The presence of ammonium and the onset of stationary phase also point to involvement of another growth factor whose depletion triggers sexual reproduction. The specific collection of factors that lead to sexual reproduction in diatoms in the environment is not yet known and neither is whether ammonium is a direct or indirect trigger of sexuality [4]. Nevertheless, this work suggests an intriguing ecological role for ammonium in the mechanisms underlying sexuality in centric diatoms and will certainly be a valuable tool to control sexuality in the laboratory.
The identification of ammonium as a reliable inducer of sexuality in T. pseudonana and other centric diatoms has the potential to shift perspectives on diatom ecology, open avenues for the experimental investigation of diatom reproductive mechanisms, and provide tools for genetic manipulation of centric diatoms that have not heretofore been available. Diatom blooms have a global impact but the factors that control these blooms and their demise are complex and a consensus has not been reached about these processes. Our evidence suggests that induction of sexuality may play a vital role in diatom bloom conclusion and the production of genetic diversity that seeds future blooms [66]. Our analysis suggests an involvement of genes of red-algal origin, providing new lines of evolutionary enquiry. Interest in diatoms for biotechnological applications is high due to their uses in biofuels, materials chemistry and medicine. Our work will likely propel this exploration by enabling improved breeding and genetic modification to control and understand unique diatom traits.  Table 1. Cell populations were quantified using a Coulter counter (Beckman-Coulter, Indianapolis, Indiana). Oogonia and auxospores were counted using a hemocytometer.

Materials and methods
We found that a modified f/2 medium yielded better cell images using light and confocal microscopy.  For RNAseq analysis, 1.61x10 8 -1.10x10 9 cells from triplicate independent cultures were filtered onto 0.8 μm 47 mm polycarbonate filters during exponential, stationary and late stationary phases and flash frozen in liquid nitrogen. RNA was extracted using a Qiagen RNeasy midi kit according to modified manufacturer's instructions [68]. Silica beads (0.5 mm) were added to the cells and lysis buffer and vortexed until homogeneous before being filtered through Qiashredder columns to remove large particles. Eluted RNA was subjected to off column RNase free DNase I treatment and secondary purification according to manufacturer's recommendations. Total RNA was prepared and sequenced as a 150 bp single end library on an Illumina HiSeq 3000 at the Center for Genome Research and Biocomputing at Oregon State University. Sequencing data/interaction analyses were conducted using the Ballgown pipeline [69]. Sequencing reads were trimmed to remove sequencing adapters using BBDuk v. with the parameters " ktrim = r k = 23 mink = 9 hdist = 1 minlength = 100 tpe tbo" [70]. Reads aligned to the T. pseudonana reference genome (NCBI accession GCA_000149405.2) using HISAT2 v. 2.0.4 with the parameters "-min-intronlen 20-max-intronlen 1500-rna-strandness F-dtacufflinks" [71]. Transcripts were assembled for each dataset and merged using Stringtie v 1.2.4 [72]. Pairwise differential expression analyses for genes were performed using the "stattest" function in Ballgown version 2.2.0 [73]. Interaction effects were tested by comparing the models with (timepoint + treatment + timepoint Ã treatment) and without (timepoint + treatment) the interaction term using the custom model option in the "stattest" function.
All RNAseq data have been deposited to NCBI under BioProject ID PRJNA391000. cultures in exponential phases of growth and maintained in NaNO 3 (red) and after two successive 25% transfers to media with ammonium (blue), Each new culture was allowed to remain in stationary phase for three days before the next 25% transfer was made. Single replicates.  Table. Pairwise comparisons, interaction analyses, genes with annotated functions, homologues to yeast meiotic genes, and genes of red algal origin. A-K: Genes that were significantly differentially expressed in multiple pairwise comparisons of growth phases and nitrogen sources. L-O: Genes with significantly greater or lesser magnitude changes in expression between growth stages in the presence of ammonium relative to 100 μM NaNO 3 identified via statistical interactions of pairwise condition comparisons. P: Genes that were identified as differentially expressed in conditions that triggered cell differentiation and sexual reproduction and have annotated functions. Q: Genes that were identified as differentially expressed in conditions that triggered cell differentiation and sexual reproduction and are homologous to yeast genes involved in meiosis. R: Genes that were identified as differentially expressed in conditions that triggered cell differentiation and sexual reproduction and are of red algal origin.