Interactions between Thermal Acclimation, Growth Rate, and Phylogeny Influence Prochlorococcus Elemental Stoichiometry

Variability in plankton elemental requirements can be important for global ocean biogeochemistry but we currently have a limited understanding of how ocean temperature influences the plankton C/N/P ratio. Multiple studies have put forward a ‘translation-compensation’ hypothesis to describe the positive relationship between temperature and plankton N/P or C/P as cells should have lower demand for P-rich ribosomes and associated depressed QP when growing at higher temperature. However, temperature affects many cellular processes beyond translation with unknown outcomes on cellular elemental composition. In addition, the impact of temperature on growth and elemental composition of phytoplankton is likely modulated by the life history and growth rate of the organism. To test the direct and indirect (via growth rate changes) effect of temperature, we here analyzed the elemental composition and ratios in six strains affiliated with the globally abundant marine Cyanobacteria Prochlorococcus. We found that temperature had a significant positive effect on the carbon and nitrogen cell quota, whereas no clear trend was observed for the phosphorus cell quota. The effect on N/P and C/P were marginally significantly positive across Prochlorococcus. The elemental composition and ratios of individual strains were also affected but we found complex interactions between the strain identity, temperature, and growth rate in controlling the individual elemental ratios in Prochlorococcus and no common trends emerged. Thus, the observations presented here does not support the ‘translation-compensation’ theory and instead suggest unique cellular elemental effects as a result of rising temperature among closely related phytoplankton lineages. Thus, the biodiversity context should be considered when predicting future elemental ratios and how cycles of carbon, nitrogen, and phosphorus may change in a future ocean.


Introduction
The cellular contents of carbon (C), nitrogen (N), phosphorus (P), and other elements in marine phytoplankton are emerging as important features of ocean biogeochemistry. For a long time, C/N/P was assumed static at Redfield proportions (106/16/1) [1]. However, variability in plankton elemental requirements can influence nutrient limitation patterns and stress [2,3], nitrogen fixation rates [4,5], the link between nutrient supply and C export [6], and atmospheric CO 2 levels [7]. Recent work has demonstrated extensive differences in the elemental content and ratios of marine communities across regions or seasons [8][9][10][11][12]. However, the exact mechanisms controlling the observed regional differences are still uncertain as key environmental factors strongly co-vary in the ocean.
Multiple biological mechanisms controlling the elemental composition of marine phytoplankton have been proposed. The main suggested controls include nutrient availability, growth rate, temperature, and life history. Extensive experimental and model studies have demonstrated a strong effect of nutrient availability, whereby a low supply of nitrogen or phosphorus leads to a low cell quota (Q) of the corresponding element [13][14][15][16]. Another important factor is the cellular allocation towards P rich ribosomes at elevated growth rates. Coined the 'Growth Rate Hypothesis' [17], fast growth is hypothesized to result in high Q P and corresponding low C/P and N/P ratios. However, this growth effect on stoichiometry appears to vary extensively by organism and environmental conditions [16,18,19]. Thus, the genetic and environmental contexts (and possible interactions) for changes in growth rate may be important to consider.
Temperature has also been proposed as a relevant factor for setting the elemental allocation in marine phytoplankton but we currently have limited understanding and data for the quantitative effect [20][21][22]. Toseland and co-workers showed that phytoplankton produce more P-rich ribosomes at lower temperature; putatively to compensate for lower translational efficiency. Hence, temperature was hypothesized to influence the elemental ratios in phytoplankton such that a future warming of the oceans would lead to increasing N/P ratios of marine communities [20]. Supported by a meta-analysis of eukaryotic phytoplankton lineages, Yvon-Durocher and co-workers detected an increase in C/P and N/P (but not C/N) for cells growing at higher temperature [22]. However, temperature affects many cellular processes beyond translation with unknown outcomes on cellular elemental composition. In addition, the impact of temperature on growth and elemental composition of phytoplankton is likely modulated by the life history of the organism. Important life history traits include the thermal growth optimum and more broadly adaptation of individual cellular processes to various temperature conditions. For example, an increase in temperature may have very different physiological effects depending on whether the rise occurs below or above the thermal growth optimum. Thus, the organismal context should be considered for understanding the influence of temperature on the elemental composition of phytoplankton.
The most abundant phytoplankton lineage in the ocean is the marine Cyanobacteria Prochlorococcus [23]. The lineage is responsible for a substantial fraction of ocean primary productivity and thus central to ocean biogeochemical functioning. Most studies of phytoplankton elemental stoichiometry are done using eukaryotic lineages with a large cell size that are either rare or absent in the ocean. In contrast, we currently know little about what regulates the elemental composition of Prochlorococcus but it appears that changes in growth rate could affect C/N/P [24]. Further, a prior study of Prochlorococcus strain MED4 found that concomitant with an increase in growth rate and cell size, C,N, and P quotas increased with temperature, maintaining the same stoichiometry [25]. The Prochlorococcus clade also harbors extensive genetic diversity including clades adapted to different ocean temperature regimes [26]. The HLII clade dominates in warm tropical waters, whereas the HLI clade is more common in higher latitude, cooler waters [27]. These distributions are consistent with the growth responses of representative strains, with the HLI strains growing faster than HLII at low temperature, and the HLII strains growing faster than HLI at high temperature. However, it is unknown how adaptations to different ocean regimes and temperature will modulate a thermal effect on the elemental composition.
Here, we investigated the sensitivity of the elemental quotas of Prochlorococcus to changes in temperature, with the hypothesis that their N/P and C/P ratios are positively related to temperature. As a possible temperature effect will be modulated by changes in growth rate as well as the life history (i.e., genotype) of the organisms, we quantified the effect of temperature on the growth rate and elemental composition of three strains of the high-temperature-adapted HLII clade and three of the low-temperature-adapted HLI clade. This study contributes fundamental information on how temperature influences the elemental composition of this key, abundant lineage and its contribution to global biogeochemical cycles.

Strains and growth conditions
Six axenic Prochlorococcus strains affiliated with the HLI and HLII clades were analyzed in this study (Table 1). All strains except VOL29 were previously rendered axenic (Table 1), while VOL29 isolation and purification is described presently. VOL29 was isolated during the POWOW1 cruise in the N. Pacific Ocean (29.6˚N, 125.07˚W) on March 9 th , 2012 at a depth of 3 m using Instant Ocean Sea Salt media (Spectrum Brands, CA) media to grow under ambient conditions (20-24˚C, 40 μmol quanta m −2 s −1 light) [28]. VOL29 was rendered axenic using the established helper method on agarose plates [29]. A spontaneous streptomycin-resistant derivative was obtained, and plated for colonies on AMP-J agarose medium pre-seeded with the streptomycin-sensitive helper strain EZ55. Prochlorococcus colonies were inoculated in AMP-J liquid and then rendered axenic by the addition of streptomycin to eliminate the helper and subsequently verified as axenic [29,30].
Prochlorococcus strains were cultured in filtered (0.2 μm polycarbonate filter, pressure <10 mm Hg) artificial seawater AMP-J medium [29] (per L, 28.1 g NaCl, 6.9 g MgSO 4 Ã 7H 2 O, 5.49 g MgCl 2 Ã 6H2O, 0.67 g KCl. 1.47g CaCl 2 , 0.504 g NaHCO 3 with 2 ml 0.5 M TAPS, pH 8.0, 1 ml 0.4 M (NH 4 ) 2 SO 4 , 2 ml 0.025 M NaH 2 PO 4 pH7.5, 100 μl 10,000 X Pro99 Trace Metal Mix) with 40 μmol quanta m −2 s −1 light on a 12:12 light:dark cycle using cool white fluorescent bulbs at temperatures from 16˚C to 26˚C (Table 1). Cultures were acclimated to the test temperature for at least three transfers (~20 generations) at high cell concentration (> 10 7 cells ml -1 ), before transferring at 10 6 cells ml -1 . The purity of strains was tested before and after strains were inoculated to culture. Prochlorococcus was inoculated into YTSS and 1/10 ProAC purity test broths in dark and monitored for visible signs of heterotrophic growth [29,30]. Samples were strictly taken during exponential growth. All collected data is listed in S1 Table. Cell counting Concentration of Prochlorococcus was measured by flow cytometry using a Guava EasyCyte 8HT cytometer (Millipore, Billerica, MA) and growth rates were estimated. Particulate organic matter Particulate organic carbon (POC), nitrogen (PON) and phosphorus (POP) samples were each collected in duplicate from each of three biological replicates (6 total) by filtration of 50 ml of culture onto precombusted (5 h, 500˚C) GF/F filters (Whatman, Florham Park, New Jersey) and stored at -20˚C. To quantify POC and PON, filter samples were thawed and allowed to dry overnight at 65˚C. Filters were then packed into a 30 mm tin capsule (CE Elantech, Lakewood, New Jersey) and analyzed for C and N content on a FlashEA 1112 nitrogen and carbon analyzer (Thermo Scientific, Waltham, Massachusetts) [31]. POC and PON concentrations were calibrated using known quantities of atropine and peach leaves in each run. The amount of POP was determined in each sample using a modified ash-hydrolysis method [15,32]. We also analyzed multiple blank controls.

Data analysis
All data was plotted using Matlab. Statistical analyses were done using linear models in R. To account for non-linear effects of T on the elemental content of Prochlorococcus strains, T was treated as a factor with four levels.

Phylogenetic analysis
Prochlorococcus ITS nucleotide sequences from each strain were aligned using ClustalW [33]. Pair-wise DNA distance matrix (w. F84 substitution matrix) and neighbor-joining tree were calculated using Phylip v. 3.69 [34] using ITS sequences from Prochlorococcus assemblies HNLC1 and HNLC2 as outgroup [35]. Next, we found the linear contribution of temperature, growth rate and strain identity on cell quotas and rates. To evaluate if the strain identity effects were phylogenetically structured, we then compared an Euclidian distance matrix of the strain identity effects to the pair-wise DNA distance matrix using a Mantel test in the R package 'vegan' [36].

Results
To identify the impact of temperature on the elemental composition of Prochlorococcus, we quantified the carbon, nitrogen, and phosphorus cell quota as well as growth rate of six axenic strains (Table 1). HLI and HLII clades, adapted to different temperatures, were represented by 3 strains each, and to facilitate comparisons between clades, the temperatures assayed were within the permissive range for growth of all strains. Median cell quotas across all strains of 0.44 fg P, 6.4 fg N, and 33 fg C were similar to previously measured levels [24,37]. Temperature had a significant linear positive effect on Q N and Q C across all strains but no direct effect on Q P ( Table 2 and Fig 1A-1C). Over the 10˚C increase in temperature, Q N and Q C rose by 40% and 35%, respectively. We also examined the elemental ratios. C/N showed little variability and was close to Redfield proportions (median C/N = 6.1) (Fig 1D). In contrast, C/P and N/P were above Redfield proportions (median C/P = 174, median N/P = 29) (Fig 1E and 1F). Both ratios showed some effect of temperature and there was a marginally significant positive linear trends across all strains (Table 2).
We also quantified growth rates of all the isolates to determine how changes in growth rate in conjunction with temperature affected the elemental composition of Prochlorococcus (Fig  2). At a light level of 40 μmol quanta m −2 s −1 , the growth ranged between 0.13 d -1 and 0.39 d -1 . Temperature affected the growth of HLI and HLII isolates slightly different whereby several HLI isolates sustained growth a lower T whereas HLII isolates were less inhibited at high T. Relating growth rate and elemental quotas and ratios, we detected a negative effect of growth rate on Q P , whereas the other cell quotas and ratios did not display any linear trends (Fig 1 and Table 2).
We next examined the influence of temperature on the cell quotas in the context of each strain as well as indirectly via changes in growth rate (Table 2 and Fig 3). We observed some similarities as well as difference in the response across the six strains. As seen in the aggregated response for all strains, individual strains displayed negative relationships between growth rate and Q P . In addition, temperature also influenced Q P on a per strain basis (Fig 3A), but there were no systematic differences between strains nor interactions between factors ( Table 2). The HLI strains VOL8 and VOL29 had higher overall Q N and Q C and temperature plus growth rate influenced Q N and Q C across all strains (Fig 3B and 3C). Thus, there was evidence for direct influences of strain identity, temperature, and growth rate-as well as some interactions-in setting the overall elemental composition ( Table 2).
Temperature and growth rate also affected the elemental ratios of each strain in unique ways (Fig 4). For C/N, we observed differences in the overall level across the strains, whereby strain VOL8 showed the highest and MIT9312 the lowest level ( Fig 4A). The strain specific C/N was also marginally affected by growth rate but not temperature ( Table 2). The strain specific C/P and N/P varied considerably between strains (Fig 4B and 4C) and in particular, MIT9215 had considerably higher ratios compared to the other strains. Temperature had a significant impact on C/P and N/P but the direction varied between strains. C/P and N/P in strains VOL8, VOL29, and MIT9215 were positively affected. In contrast, VOL7 showed high variability with lower ratios at 16˚C as well as 26˚C and higher at the intermediate temperature, UH81301 showed no response, and a negative response was observed in MIT9312. Thus, there were complex interactions between the strain identity and temperature in controlling the elemental ratios in Prochlorococcus.
We next found little phylogenetic structuring of elemental changes across strains. After subtracting the overall influence of temperature and growth rates on cell quotas or elemental ratios, we then identified additional variation attributed to each strain. These 'strain-specific' contributions were then compared to the phylogenetic distance between each strain (Mantel test, Table 2). This comparison revealed that neither cell quotas nor ratios were phylogenetically structured.

Discussion
Multiple studies have put forward a 'translation-compensation' hypothesis for a positive relationship between temperature vs. N/P or C/P. Cells should have lower demand for P-rich ribosomes and associated depressed Q P when growing at higher temperature [20,22]. A lower Q P will cause elevated C/P and N/P and such an acclimation mechanism should further explain the high elemental ratios observed in cells growing in the hot, oligotrophic gyres [8,9]. However, we see little support for this hypothesis in Prochlorococcus. Instead, the thermal effect leads to increasing Q N and Q C , whereas Q P shows little systematic change. This points towards other physiological acclimation mechanisms as the primary drivers of elemental changes in Prochlorococcus. The observed elemental changes are likely associated with a cell size increase as Q N and Q C increase in tandem. The underlying mechanism for this increase in Q N and Q C in Prochlorococcus is not known but the response was opposite to Scenedesmus and Asterionella [38]. Based on studies in heterotrophic organisms, it is likely associated with an increase in cellular macromolecules and especially protein content [39]. Such a change in cell size means that you cannot simply extrapolate from an increase or a decrease in an individual cell quota (like Q P ) to the stoichiometric ratio. Thus, our study adds to an emerging concept, whereby changes in cell size due to physiological responses to different environmental conditions are important for regulating the elemental composition and ratios in marine Cyanobacteria [16]. Q P appears linked to changes in thermally induced growth rate but not temperature itself. This would indicate support for the growth rate hypothesis [17] but Q P is actually decreasing at elevated growth rates across all strains as well as for most individual strains. As seen in other marine Cyanobacteria [16], it is clear that the growth rate hypothesis alone cannot explain differences in elemental composition across Prochlorococcus strains. However, there is a lot of variation, which suggests individual strain Q p responses to temperature and growth physiology. The composition of P containing macromolecules underlying the overall cellular P content is poorly constrained in marine Cyanobacteria [15,16] as the sum of phospholipids and nucleic acids does not get close to Q p . Thus, it is currently uncertain, which biochemical mechanism will lead to the observed changes in Q p .
We observe overall high C/P and N/P, whereas C/N is close to Redfield proportions in Prochlorococcus. The cells are growing under nutrient replete conditions, which should lead to C/P and N/P at the lower end of the range for an organism [3,40]. Our observations of above Redfield ratios in Prochlorococcus are consistent with past observations [8,24,41] and suggest this lineage has overall high C/P and N/P. As such, the presence of Prochlorococcus in low latitude marine communities will contribute to elevated elemental ratios independently of thermal and nutrient conditions. We do not observe a direct phylogenetic structuring of cell quotas and ratios within Prochlorococcus. However, we observed a significant influence on the elemental composition via changes in growth rate along the temperature gradient. The temperature effect on growth in Prochlorococcus strains have been shown to be strongly phylogenetically structured, whereby the HLI and HLII clades are adapted to lower and higher temperature regimes, respectively [26,27]. Thus, we see an indirect phylogenetic structuring through the effect of growth rate on the cell quotas and ratios. In addition, we see extensive strain variability in the elemental content and ratios due to thermal acclimation. Thus, the organismal context and potentially growth optimum appear important for the individual response. This is consistent with the thermal response in other phytoplankton lineages and strain specific variability in quotas and ratios of Gyrodinium species [42,43]. In an analysis across nine eukaryotic phytoplankton lineages, Yvon-Durocher and co-workers observed substantial variability in the link between thermal changes and elemental cellular composition [22]. Furthermore, this meta-analysis as well as our study found little thermal effect on C/N, suggesting C/N being fairly invariant to temperature changes.
The broader environmental growth conditions are important to consider when evaluating the elemental outcome in Prochlorococcus to thermal changes. In this study, the cells were growing under nutrient replete conditions and Prochlorococcus may store large reserves of P overwhelming any contributions from ribosomal RNA. Multiple studies have shown the possibility for interactions between factors including interactions between nutrient limitation and temperature [22,38]. Scenedesmus showed stronger thermal responses under nutrient limited vs. replete conditions. Hence, future work studying the interaction between nutrient limitation and thermal conditions would enhance our understanding for how changes in ocean temperature would affect Prochlorococcus elemental stoichiometry.
Our study has implications for understanding both present day and future biogeochemical functioning. The oceans are projected to undergo substantial changes in temperature due to rising CO 2 in the atmosphere. Such environmental changes will likely have a large impact on phytoplankton community structure and physiology [23,44]. This has been predicted to lead to an increase in N/P ratios in phytoplankton communities [20,22]. However, the observations presented here suggest unique cellular elemental effects as a result of rising temperature among closely related phytoplankton lineages. Thus, the biodiversity context should be considered when predicting future elemental ratios and how the link between the cycles of carbon, nitrogen, and phosphorus may change in a future ocean.
Supporting Information S1 Table. Data associated with this study. (CSV)