Pheromone Mediated Sexual Reproduction of Pennate Diatom Cylindrotheca closterium

Benthic diatoms dominate primary production in marine subtidal and intertidal environments. Their extraordinary species diversity and ecological success is thought to be linked with their predominantly heterothallic sexual reproduction. Little is known about pheromone involvement during mating of pennate diatoms. Here we describe pheromone guided mating in the coastal raphid diatom Cylindrotheca closterium. We show that the two mating types (mt+ and mt−) have distinct functions. Similar to other benthic diatoms, mt+ cells are searching for the mt− cells to pair. To enhance mating efficiency mt− exudes an attraction pheromone which we proved by establishing a novel capillary assay. Further, two more pheromones produced by mt− promote the sexual events. One arrests the cell cycle progression of mt+ while the other induces gametogenesis of mt+. We suggest that C. closterium shares a functionally similar pheromone system with other pennate diatoms like Seminavis robusta and Pseudostaurosira trainorii which synchronize sexual events and mate attraction. Remarkably, we found no evidence of mt+ producing pheromones, which differentiates C. closterium from other pennates and suggests a less complex pheromone system in C. closterium. Supplementary Information The online version contains supplementary material available at 10.1007/s10886-021-01277-8.


Introduction
Diatoms constitute a diverse group of photosynthetic unicellular microalgae and are key players in carbon cycling and nutrient exchange in marine and freshwater ecosystems (Armbrust 2009;Benoiston et al. 2017;Falkowski et al. 1998;Malviya et al. 2016;Smetacek 1998). Benthic raphid pennate diatoms, however, are often undervalued although they dominate bio lms, contributing tremendously to primary production and the stabilization of coastal marine sediments (Stal et al. 2019;Yallop et al. 1994). They show a high species diversity and have unique adaptations to this highly dynamic and heterogeneous environment (Osuna-Cruz et al. 2020). The evolution of motility and the transition to a heterothallic life cycle are seen as major evolutionary innovations driving their spectacular species diversi cation (Nakov et al. 2018). Reproduction systems of raphid pennate diatoms have been intensively studied in the last decades (Amato 2010;Chepurnov et al. 2004;Poulíčková and Mann 2019). The asexual division of diploid cells is accompanied by cell size reduction and prevails the life cycle of pennate diatoms. Meiosis and subsequent sexual reproduction, however, are vital to restoring original cell size. Typically, sexual reproduction is triggered when cells reach a species-speci c sexual size threshold (SST, 60-70% of their original size) (Chepurnov et al. 2002;Davidovich et al. 2009), following pair formation between sexually compatible mating types. In Seminavis robusta, cells of the two mating types have distinct roles and are referred to as mt¯ cells guiding the more motile mt + cells towards them (Chepurnov et al. 2005;Chepurnov et al. 2002). The mating types are morphologically identical but the sex determination in S. robusta is encoded in its genes (Vanstechelman et al. 2013). In the planktonic diatom Pseudo-nitzschia multistriata the mating type determining gene was even identi ed as MRP3 (Russo et al. 2018). Mate nding and sexual reproduction are energetically costly (Lewis 1987) and pheromones are known to be involved (Basu et al. 2017;Frenkel et al. 2014). These signalling cues are of different chemical nature and function to enhance mating e ciency (Frenkel et al. 2014). In the planktonic diatom P. multistriata a pheromone is suggested to synchronize the sexual events (Scalco et al. 2014) whereas three different consecutively induced pheromones are thought to be involved in the reproduction of the freshwater diatom Pseudostaurosira trainorii to control sexualization of cells and orientate gametes (Sato et al. 2011). Mating of S. robusta was studied intensively and is also promoted by three distinct pheromones (Bilcke et al. 2020;Bondoc et al. 2019;Bondoc et al. 2016;Chepurnov et al. 2002). Two sex inducing pheromones SIP + and SIPare produced by mt + and mt¯, respectively, and synchronize the cell cycle (Moeys et al. 2016). Pheromone structures are poorly studied but the chemoattractant of S. robusta was identi ed as l-diproline, the rst elucidated attraction pheromone of a diatom . mtc ells that are exposed to SIP + release diproline to attract the mt + cells.
The pennate diatom Cylindrotheca closterium is a meroplanktonic species, present in both, planktonic and benthic habitats. This lifestyle allows obtaining high densities in water columns and bio lms. As it is found in marine, brackish, freshwater regions and even inside sea-ice (von Quillfeldt et al. 2003), C. closterium has been used as model in several eco-physiological studies showing its ability to adapt to varying temperatures (Stock et al. 2019) and salinities (Najdek et al. 2005). The unique movement modalities of C. closterium change in response to hypo-or hyper-saline conditions (Apoya-Horton et al. 2006). In recent years a transformation protocol has been established for C. closterium and a gene editing protocol through CRISPR/Cas9 is currently developed, thus providing a toolbox for many molecular studies as well (unpublished data).
The life cycle of C. closterium has also been studied in detail . When cells reach the SST and partners of the two distinct mating types (Cyc1 and Cyc2) meet, successful sexual reproduction takes place. The purpose of this study was to specify the mating behaviour of each mating type and investigate whether there is an attracted and an attracting partner. Further, we aimed to determine if a directed attraction during mate nding occurs. The possibility of pheromone involvement during mating in C. closterium was also investigated and similarities and disparities in the mating systems of pennate benthic diatoms are discussed.
To estimate cell density by microscopically counting and surveying experimental results, photographs of each culture (n = 3) were taken with a Nikon DS-Fi2 CCD camera (Tokyo, Japan) attached to an inverted Leica DM IL LED light microscope (Heerbrugg, Switzerland) (100 x magni cation). The open-source software ImageJ (Rasband WS, ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA. http://imagej.nih.gov/ij/ 1997) and the cell counter plug-in was used for counting cells as well as pairs and clusters of cells in mating experiments.
For experiments cultures were rst grown in 6-well plates (5 mL well -1 , Sarstedt, Nümbrecht, Germany) for 4 days to reach early exponential phase. Thereafter cultures were dark-synchronized for 36 h .
Each well contained 500 µL of spent medium of each strain and 200 µL of cell suspension (30k cells cm -2 ) composed of the above mentioned mating type proportions. Five hours after re-illumination the mating response was evaluated by manually analysing microscopic photographs. Cells of the two mating types were distinguished by the differing apical cell length. The following classes of interacting cells were discriminated: mating-pairs (two cells of different mating type laying against each other with their long axes parallel), multiple pairing cluster (accumulation of cells in which the mating type of the majority is three times more present than the other), as well as the total number of interactions (pairs, clusters and threesomes).

Capillary Assay
For the investigation of cell attraction, dark-synchronized cultures of mt + were split into 12-well plates in new f/2 medium and kept at light for 6 hours.
Two hundred mL of cell free medium of exponentially growing cultures of C2, CA1.15 and CZ1 (mt + , mtā nd MT¯ (>SST) respectively) alike were extracted using HLB cartridges (hydrophilic-lipophilic balanced solid phase extraction, Oasis®, Waters, Eschborn, Germany) following the instructor's manual. MeOH extracts were dried under nitrogen ow and dissolved in 200 µL pure water. Capillaries (l = 30 mm, V = 5 µL, minicaps, Hirschmann, Germany) were prepared directly before use (n = 4). Therefore, 25 µL of medium extract were added (1:1 v/v) to hot agar (2% Agar-Agar, Kobe 1, Roth, Germany), mixed quickly and immediately absorbed into the capillaries by capillary forces. The outside of the capillaries was cleaned carefully and the capillaries dried on air for 5 min. Capillaries were then carefully placed vertically in the wells which were covered by a plastic foil holding the capillaries in place. Pictures of the capillary openings were taken at t = 0 min (directly after inserting the capillary), t = 30 min, and t = 60 min and cells were counted within a de ned area of d = 580 µm around the capillary (ImageJ).

Motility Assay
In 24-well-plates, dark arrested cultures of C2 (mt + ) and CA1.15 (mt¯) were treated with 1.5 µL of medium extract of the opposite mating type (n = 4) (see Capillary Assay for extract preparation) four hours after re-illumination. Control cultures stayed untreated. In C2 experiments three individual videos (30 s) were recorded (1fps) for each well one hour after the treatment to verify the robustness of the assay. In experiments with CA1.15 one video was recorded for each replicate. Cell tracking was done for >300 cell/ replicate using the plug-in TrackMate (http:// ji.sc/TrackMate) for ImageJ.

G1 Cell Cycle Arrest Assay
Five mL of medium of exponentially growing cultures of C2 and CA1.15 (mt + and mt¯) were sterile ltered (Filtropur S 0.2 µm, Sarstedt, Germany) immediately before the experiment. Dark-arrested cultures were split for different treatments 2 hours before re-illumination. In the dark, culture medium of treatments was fully exchanged by ltered medium of the opposite mating type while control cultures stayed untreated.
At t = 0 h (in the dark) and t = 6 h (after re-illumination), 2 mL samples were taken and directly centrifuged (5 min, 3000 rpm). The supernatants were discarded and directly replaced by ice cold 70% ethanol. These xed samples were stored at 4 °C in the dark for at least 24h. The supernatant was discarded and cell pellets were resuspended in 200 µL 70% ethanol and washed twice with phosphate-buffered saline (PBS, 137 mm NaCl, 2.7 mm KCl, 10 mm phosphate, pH 7.4). Samples were treated with 1 µg mL -1 RNAse A for 40 min at 37 °C and afterwards stained with SybrGold (10,000 fold diluted from stock solution, SYBR Gold Nucleic Acid Gel Stain, Thermo Fisher Scienti c, USA) for 10 min in the dark. Cells were transferred to PBS again and ltered over Celltric lters (Sarstedt, USA) to exclude cell clusters. DNA content was measured on a BC Accuri C6 ow cytometer. For each sample, 5000 events were collected and gated in the FL1 and FSC channel. G1 and G2+M peaks were visually selected and statistically analysed.

Gametogenesis
Gametogenesis was induced by crossing dark arrested cultures of mt + and mt¯ (n = 3) in 24 well plates in either 1:9 or 9:1 ratio. Each well contained 500 µL of spent medium of each strain and 100 µL of cell suspension (30k cells cm -2 ) adjusted to the given proportions and cell densities of the pre-cultures. Over a period of 5 days, gamete formation and initial cell formation was evaluated.

Statistical Evaluation
All data are depicted as mean ± standard error of the mean and were statistically analysed using SigmaPlot 13.0 (Sysstat Software GmbH, Erkrath, Germany). After testing for normal distribution and equal variances, effects were evaluated by applying One-Way or Two-Way ANOVA (α = 0.05) and a subsequent Sidak's or Tukey´s multiple comparison post-hoc test. A Mann-Whitney rank sum test was applied to compare two groups.

Results
Designation of Mating Types in Cylindrotheca closterium.
To detect characteristics of mate nding and searching behaviour in Cylindrotheca closterium compatible strains of mating type (mt) Cyc1 and Cyc2  were crossed in different ratios (Fig. 1). The strong accumulation of Cyc1 cells around Cyc2 cells revealed the attracting nature of Cyc2 cells and leads to the assignment of Cyc2 to mating type¯ (mt¯). Cells of Cyc1, on the other hand, seemed to be more motile searching for the opposite partner and thus representing mating type + (mt + ). Pair formation as well as cell cluster formation of distinct mating types were recognized by the differing cell lengths of both mating types. Cell clusters consist of at least three times more cells of the dominant mt ( Fig. 1a). mt + cells that were either smaller or bigger than mt¯ both showed clustering around mt¯ (SI, Fig. S1) which con rms that searching behaviour is independent of cell size concerning mt + . Crosses with an equal number of both mating types (5:5) resulted in most successful mating (11.7k total interactions cm -2 out of 15k possible pairs) comprising formation of pairs, clusters and threesomes (Fig. 1b). As the proportion changed towards one mating type total interactions were limited by the partner present in the lowest density and thus less interactions cm -2 were monitored. In crosses with a low mt + :mt¯ ratio cells mostly formed pairs (1.3-4.8k pairs cm -2 , Fig. 1c). Pair formation also dominated the crosses of equal mating type ratio, comprising two thirds of all interactions. With a descending proportion of mt¯ cells accumulations of mt + cells around mt¯ cells occurred. Clusters were even most prevalent in crosses with a majority of mt + cells. In the 9:1 cross, clusters dominated the culture comprising more than 85% of all total interactions (4.2k clusters cm -2 ). Clusters were still detected in crosses with a low mt + :mt¯ ratio but with comparably low numbers (1.0k clusters cm -2 ). Taken together, these data show that each mating type has a speci c behaviour and a de ned functional purpose that shapes and structures the mate nding process.
Bioassay Reveals Chemical Mediated Attraction of mt + .
We hypothesized that chemical signalling is involved in the mate nding process of C. closterium. To verify this hypothesis, mt + cell attraction towards mt¯ exudates was assessed. Exudates could be extracted via solid phase extraction using methanol as an eluant. A chemotaxis assay comprised of capillaries lled with medium extracts embedded in agar was optimized to meet the requirements of using benthic algae instead of bacteria (Abe et al. 2017). Extracts diffuse from small capillaries that are placed on the bottom of a well (Fig. 2a). Careful optimization included the use of smaller capillaries (5 µL instead of 20 µL) that were placed vertically in the well. A foil, holding the capillaries ensured a stable setup and was required to counteract the extreme variability in initial experiments (SI, Fig. S2) and increased robustness of this delicate procedure dramatically.
Capillaries lled with mt medium extracts led to an accumulation of mt + cells around the capillary opening 60 min after inserting the capillary (Fig. 2b). Cell counting revealed almost three times more mt + cells around target capillaries than around control capillaries. Controls contained extract of the same mating type and showed no activity since cell numbers around the capillaries did not increase over time (Fig. 2c). We therefore suggest that mt + cells must have received a signal from the mt medium extract within the capillary and accumulate at its highest concentration. As expected, extracts of MT cultures (sexually immature; >SST) did not trigger a response of mt + cells.
Further, mt + cells also responded to mt medium extract by changing their motility behaviour. One hour after exposure to mt medium extract, mt + cells employed chemokinesis resulting in an increase of the overall mean speed (Fig. 3a). mt + cells were 2.7 times faster compared to control cultures. Likewise, the travelled net distance of treated mt + cells also increased (Fig. 3b) which enhances mate nding. As expected, mt cell motility was not affected by medium extract of mt + (Fig. 3b, c).
This leads to the assumption that a chemical cue is exuded by mt cells as soon as they fall below the sexual size threshold (SST). The sexually mature cells start to constitutively produce this attraction pheromone without further induction. This pheromone plays an essential role in guiding mt + cells and proves chemical involvement in the mate nding process of C. closterium to increase mating e ciency.
Cell Cycle Regulation of mt + during Mating.
Besides cell attraction we investigated cell cycle regulations of C. closterium during mate nding. DNA content was measured with ow cytometry and re ected the cell cycle phases (G1, S+G2) of the cells.
Initially, cultures of both mating types were successfully synchronized in their cell cycle by dark treatment as described by Vanormelingen et al. (2013). Cells were kept almost solely in G1 phase (>95%) of the cell cycle (Fig. 4a, top) and less than 5% of the cells entered the S+G2 phases of the cell cycle (Fig. 4b, black). As expected at least 20% of the cells proceeded into the S+G2 phases of the cell cycle 6 h after illumination (Fig. 4a, middle, Fig. 4b grey). A comparison of both mating types revealed that mt spent medium caused an arrest of mt + cells in G1 phase that lasted for at least 6 h (6h, bottom). Surprisingly, mt cultures treated with mt + spent medium were not arrested but proceeded in the cell cycle. That contrasts the mt + cultures that were arrested in G1 phase in the presence of mt spent medium. Overall, it con rms that a chemical cue is exuded by mt cells and functions as a cell cycle regulation pheromone in C. closterium. The chemical signal is important upon the mating process only since medium of MT cells (>SST, Fig. 4c) did not lead to an arrest of mt + cells in the cell cycle. Thus, the signal responsible for the cell cycle arrest in mt + is only produced by sexually mature mt cells and is crucial for the sexual reproduction of C. closterium.
Gametogenesis of mt + is Triggered by Pheromone of mt .
Lastly, we studied the mating process regarding gamete formation and subsequent initial cell formation over a period of 5 days (Fig. 5). Crossing cultures with a mt + :mt ratio of either 1:9 or 9:1 resulted in equivalent numbers of initial cells. Compared to crosses with a more balanced mt ratio, these crosses were still e cient with respect to the numbers of compatible partners (SI, Fig. S3). When mt cells were dominant, gametes of each mating type developed equally on the rst day. Their numbers decreased over time while compatible gametes paired effectively and completely evolved into initial cells. Almost no gametes of either mating type were found after 5 days (Fig. 5a). In crosses with a high mt + :mt cell ratio not only mt cells but the majority of mt + cells developed into gametes. This led to an increase of mt + gametes over time. Facing an insu cient number of compatible partners only few gametes developed into initial cells (Fig. 5b). Again, we observed mt + cell clusters around mt cells that resulted in an enhanced number of gametes. Gametes were only formed within those clusters ). However, mt + cells within the cluster, but with no direct contact to mt cells also evolved into gametes. Non-pairing mt + cells were unaffected. This suggests that mt cells exude another short living instable signal that induces gametogenesis in mt + cells. Based on this experiment we suggest a third pheromone to be involved in the sexual reproduction of C. closterium and proved that not only mate nding but sexual reproduction itself is chemically mediated.

Discussion
We investigated the heterothallic mating system of the benthic pennate diatom Cylindrotheca closterium with respect to its distinct mating types (mt) and the pheromones that are involved in the process of mate nding. C. closterium has mating types of opposite sexes Cyc1 and Cyc2 ) that are morphologically indistinguishable. We showed that each mt, when it falls below the sexual size threshold (SST) ful ls a speci c role, either to attract the partner (mt¯) or to purposely search for it (mt + ). The same applies for Pseudo-nitzschia mulistriata but here, the behaviour of opposite mating types is rather caused by different cell sizes. Scalco et al. 2016 suggest that the larger strain, independently of the mating type, actively searches for the smaller, less motile strain . Behaviour in C. closterium, however, is independent of cell size but determined by the respective mating type and suggested to be genetically predetermined. It resembles the mating system of Seminavis robusta (Vanstechelman et al. 2013), which also constitutes of an attracting partner and a more motile searching one as soon as the SST is reached . For most heterothallic diatom species mating types are not categorized in '+' and '¯', e.g. Haslea ostrearia, or Pseudo-nitzschia multistriata. (D'Alelio et al. 2009;Davidovich et al. 2009) and their gametes are morphologically indistinguishable as well. However, assignment of male and female gametes is described frequently in araphid pennate diatoms, e.g. Tabularia fasciculate and Pseudostaurosira trainorii. Female gametes stay attached to the parental valves whereas the more motile male gametes facilitate contact with the females (Davidovich et al. 2010;Sato et al. 2011).
Notably, behaviour of mt + cells in C. closterium is pheromone guided. A signal released by mt¯ cells enables mate nding like in S. robusta . The attraction bead assay used for S. robusta could not be employed since cells attached to the unloaded beads as well . We therefore established a novel capillary assay in which mt + cells were attracted to extracts of mtē mbedded in agar. A long response time of mt + cells resulted, most likely, from an incomplete diffusion of chemical signals from the capillary. We excluded a random encountering like it was shown in P. trainorii. Male gametes of P. trainorii only head directly towards female gametes after an initial "random walk" to come within a close range of opposite sex (Sato et al. 2011). In C. closterium, mt + cells response chemokinetically towards mt¯ exudates with respect to their pace. When mt + cells sense the pheromone, they move almost three times quicker. In S. robusta natural pairing only occurs after 6 h (Chepurnov et al. 2008) but directed attraction was proven (Bondoc et al. 2016). A similar analysis of directed movement is crucial to determine whether chemotaxis is occurring in C. closterium as well.
Other than in S. robusta or P. trainorii the pheromone production in C. closterium is only dependent on cell size. As soon as mt¯ cells get sexualized they secrete the pheromone and unconditioned mt + cells are able to perceive the signal. In contrast, a reciprocal stimulation is known for P. trainorii where three different pheromones are involved in the sexual reproduction (Sato et al. 2011). The production of the attraction pheromone l-Diprolin in S. robusta is also induced by a sex inducing pheromone (SIP) of mt + .
Simultaneously, mt¯ produces SIP¯ that induces the development of a receptor in mt + (Moeys et al. 2016). Compared to S. robusta we con rm a rather short pre-mating induction phase in C. closterium ).
However, induction still takes place in C. closterium. We observed a cell cycle arrest in mt + cells induced by a cytostatic pheromone exuded by mt¯ similar to S. robusta. In S. robusta as well as P. multistriata both mating types are arrested in G1 of the cell cycle (Basu et al. 2017;Moeys et al. 2016). However, mtc ells in P. multistriata initiate the sexual process by undergoing meiosis earlier than mt + whose response is slightly out of phase. Difference in cell cycle progression of mating types is also known in S. robusta. G1 synchronized mt + cells progress in the cell cycle already 3h after light treatment whereas mt¯ only enter the G2/M phase after 9h (Bilcke et al. 2020). mt¯ cells in C. closterium further promote gametogenesis of mt + cells by producing another pheromone. Involvement of pheromones not only in the mate nding process but in the sexual reproduction itself is known for other pennate diatoms as well. In P. trainorii also female cells secrete a pheromone responsible for the sexual stimulation of male cells (Sato et al. 2011) and in T. fasciculata gametogenesis is induced without cell contact (Davidovich et al. 2010).
Overall, the mating type behaviour and pheromone system of C. closterium has a unique complexity but includes aspects observed in other pennate diatoms like S. robusta and P. trainorii. Future experiments will focus on a metabolomics approach to chemically investigate the sexual phases of C. closterium in more detail and identify the attraction pheromone as well as the cytostatic pheromone (Fiorini et al. 2020;Gillard et al. 2013).

Declarations
Funding -Funding by the Friedrich Schiller University Jena is acknowledged Con icts of interest/Competing interests -The authors declare that they have no competing interests.
Availability of data and material -Data will be made available upon request.