Bacterioplankton Groups Involved in the Uptake of Phosphate and Dissolved Organic Phosphorus in a Mesocosm Experiment with Pstar

bs_bs_banner

Environmental Microbiology (2012) doi:10.1111/j.1462-2920.2012.02772.x

Bacterioplankton groups involved in the uptake of phosphate and dissolved organic phosphorus in a mesocosm experiment with P-starved Mediterranean

watersemi_2772 1..14

Marta Sebastián,1* Paraskevi Pitta,2 whereas Gammaproteobacteria, Roseobacter and José M. González,3 T. Frede Thingstad4 and Bacteroidetes showed similar percentages to the Josep M. Gasol1 ones in the control mesocosms and (ii) on day 4 of the 1Departament de Biologia Marina i Oceanografia, Institut experiment, probably when the bacterial community de Ciències del Mar, CSIC, Pg Marítim de la had fully responded to the P input, all the probe- Barceloneta 37-49, E08003 Barcelona, Catalunya, identified groups showed low percentages of cells Spain. taking up the substrate as compared with the control 2Hellenic Centre for Marine Research, Oceanography mesocosms. These differences are likely related to Institute, P.O. Box 2214, 71003 Heraklion, Crete, different P requirements among the bacterial groups Greece. and point out to the existence of two contrasting 3Department of Microbiology, University of La Laguna, strategies in P use. ES-38206 La Laguna, Tenerife, Spain. 4Department of Biology, University of Bergen, P.O. Box Introduction 7803, 5020 Bergen, Norway. Phosphorus (P) limits microbial growth and production in several areas of the ocean, at least during part of the year Summary (Cotner et al., 1997; Rivkin and Anderson, 1997; Caron The use of inorganic phosphate (Pi) and dissolved et al., 2000; Pinhassi et al., 2006). This limitation appears organic phosphorus (DOP) by different bacterial to be particularly pronounced in the Mediterranean Sea groups was studied in experimental mesocosms of (Krom et al., 1991; van Wambeke et al., 2002; Thingstad P-starved eastern Mediterranean waters in the et al., 2005; Pinhassi et al., 2006), where inorganic phos- absence (control mesocosms) and presence of addi- phate (Pi) concentrations are usually in the low nanomolar tional Pi (P-amended mesocosms). The low Pi turn- range (Krom et al., 2005; Pulido-Villena et al., 2010). over times in the control mesocosms and the When Pi is depleted, both phytoplankton and bacteria increase in heterotrophic prokaryotic abundance and can meet their phosphorus demand using dissolved production upon Pi addition confirmed that the bac- organic compounds (Bentzen et al., 1992), as long as terial community was originally P-limited. The bacte- they are able to split off the phosphate moiety (Lugten- rioplankton groups taking up Pi and DOP were berg, 1987; Lovdal et al., 2007). Although Pi is the pre- identified by means of microautoradiography com- ferred source of P, previous studies have demonstrated bined with catalysed reporter deposition fluores- that the available fraction of the dissolved organic phos- cence in situ hybridization. Incubations with leucine phorus (DOP) pool is also actively utilized (e.g. Bentzen were also performed for comparative purposes. All et al., 1992; Lovdal et al., 2007; Casey et al., 2009; Mich- the probe-identified groups showed a high percent- elou et al., 2011). age of cells taking up Pi and DOP in the control, The contribution of the components of the microbial P-limited, mesocosms throughout the experiment. community to Pi and DOP uptake has traditionally been However, in response to Pi addition two contrasting evaluated in size-fractionation studies (Currie and Kalff, scenarios in Pi use were observed: (i) on day 1 of 1984; Bentzen et al., 1992; Thingstad et al., 1993; Lovdal the experiment Pi addition caused a clear reduction et al., 2007). These studies have indicated that under in the percentage of SAR11 cells taking up Pi, P-limiting conditions bacteria are superior competitors for Pi and DOP uptake, whereas large phytoplankton can take advantage of sudden pulses of nutrients (e.g. Suttle Received 9 September, 2011; revised 28 March, 2012; accepted 11 April, 2012. *For correspondence. E-mail: [email protected]; et al., 1990; Thingstad et al., 1993; Zohary and Robarts, Tel. +34 93 2309500; Fax +34 93 2309555. 1998; Lovdal et al., 2007). Nevertheless, in this approach

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd 2 M. Sebastián et al. the contribution of heterotrophic bacteria is commonly cell quotas in this oligotrophic system. But, do all the estimated as the fraction of radiolabelled substrate taken bacterioplankton groups use DOP? up by the plankton fraction that passes through filters with In this study a mesocosm experiment was carried out to a pore size Յ 1 mm, which also includes picocyanobacte- confirm the results obtained during CYCLOPS. This ria. On the other hand, the fraction 0.6 to 2 mm is fre- experiment consisted of three parallel controls and three quently attributed to Synechococcus (e.g. Moutin et al., P-amended bags as experimental units. We used that 2002) but the majority of picoeukaryotes are smaller than setup to answer the aforementioned questions by charac- 2 mm (Vaulot et al., 1996). Furthermore, the most active terizing the bacterial community taking up 33P-Pi and fraction of heterotrophic bacterial biomass is often larger AT-33P (as a model molecule for bioavailable low molecu- than 0.6 mm (Thingstad et al., 1996). lar weight DOP) using MARFISH. For comparison, we The problems of the size fractionation approach have also studied the uptake of 3H-leucine, which is used as a been circumvented in recent years by flow cytometric universal substrate to estimate bacterial production. To sorting of microbial populations after incubation with date there is only one marine study using MARFISH with radiolabelled P (Zubkov et al., 2007; Larsen et al., 2008; 33P-Pi as tracer, which showed that there is large variabil- Michelou et al., 2011). These studies confirmed that ity in the uptake of this substrate by different bacterial the heterotrophic component of the bacterioplankton groups (Longnecker et al., 2010). Our mesocosm experi- plays a major role in P cycling, accounting for 45% of ment constitutes one step forward because it provides a total Pi uptake in north-eastern Atlantic (Zubkov et al., unique framework to investigate the effect of Pi addition 2007) and > 90% of total Pi and ATP uptake in north- on Pi and DOP use by the different components of western Atlantic (Michelou et al., 2011). However, the bacterial community in a system likely limited by whereas this approach allows distinction between differ- P-availability. ent picoplankton groups, it still considers heterotrophic bacteria as a homogeneous black box. In the last few Results years single-cell methods like microautoradiography combined with fluorescence in situ hybridization Turnover time of phosphate (Pi) and ATP, and (MARFISH) or halogen in situ hybridization-secondary prokaryotic heterotrophic abundance and production ion mass spectroscopy (HISH-SIMS) are revealing high Pi turnover times in the control mesocosms were around heterogeneity in the uptake of various organic substrates 1 h throughout the experiment. On the contrary, the by different bacterial phylogenetic groups (Cottrell and P-amended mesocosms displayed a pronounced Kirchman, 2003; Alonso-Sáez and Gasol, 2007; del increase in Pi turnover times (Fig. 1), which went from 1 to Giorgio and Gasol, 2008; Musat et al., 2008). Likewise, 2 h before Pi addition to around 60 h on day 1 and the number of heterotrophic bacterial cells actively par- increased along the experiment reaching values up to two ticipating in the uptake of Pi and organic phosphorus orders of magnitude higher than in the control meso- substrates may vary markedly between phylogenetic cosms. ATP turnover time in the control mesocosms was groups, depending on their uptake capacities and their ~ 6 h on day 0, decreased to 3 h on day 1 and remained nutrient quotas. constant throughout the experiment. In the P-amended The eastern Mediterranean is considered one of the mesocosms the turnover time of ATP increased when most oligotrophic and P-starved marine systems on earth (e.g. Tanaka et al., 2007), where P limitation of bacterioplankton has been shown to be a generalized phenom- 1000 enon (van Wambeke et al., 2002; Thingstad et al., 2005), with few exceptions (Tanaka et al., 2011). This system 100 Pi was the site for CYCLOPS, a large-scale Pi-addition ATP experiment, in which the phytoplankton community was 10 found to be N- and P-co-limited, and the heterotrophic ATP

rnover time (h) 1 Pi

community P-limited (Krom et al., 2005; Thingstad et al., u 2005). However, many questions remained unanswered T 0.1 after this experiment: did P equally stimulate all the com- 02468 ponents of the bacterial community? Which phylogenetic day groups contributed to Pi uptake? In phosphate depleted waters the majority (> 80%) of total dissolved P exists as Fig. 1. Turnover times of phosphate (Pi, circles) and ATP (squares) DOP (Ammerman et al., 2003; Lomas et al., 2010; Lagh- in the control (open symbols) and P-amended mesocosms (ﬁlled symbols). Each data point represents the average of two dass et al., 2011; Tanaka et al., 2011). Hence, DOP likely mesocosms with two independent measurements for each plays an important role in supplying P to satisfy bacterial mesocosm. Error bars represent the standard deviation.

A 107 the remaining cells probably represent a variable contribution of non-nucleoid containing bacteria (Zweifel and Hagstrom, 1995) or decaying cells still stainable with DAPI but lacking sufficient RNA (del Giorgio and Gasol, –1 6 2008). The increase in prokaryotic abundance was mir- 10 rored by an increase in prokaryotic heterotrophic produc- cell ml tion, with values 1.5-fold to twofold higher in the P-amended mesocosms than in the controls (Fig. 2B). 105 B 10 Percentage of total bacterial cells taking up Pi, ATP and leucine ) 8 –1 Based on the Pi turnover time results, three time points

day 6 were chosen for MARFISH analyses: day 0, before Pi –1 addition; day 1, when the most drastic change in Pi turn-

g C l 4 over time was observed in the P-amended mesocosms in μ 2 relation to the controls, and day 4, when we expected the BP ( bacterial assemblage to have fully responded to the Pi 0 input. On average around 60% of the eubacterial cells 02468 took up Pi and ATP in the control mesocosms (Fig. 3), day whereas ~ 50% of the cells took up leucine at the concentration used (0.5 nM). The percentage of cells taking up Fig. 2. Time course of (A) prokaryotic abundance estimated by Pi in the P-amended mesocosms decreased along the flow cytometry (circles) and eubacterial cell numbers estimated by CARDFISH (bars), and (B) prokaryotic heterotrophic production, in experiment, reaching a value of 26% on day 4. In contrast, both the control (open symbols) and P-amended mesocosm (filled the percentage of cells taking up ATP dropped to 26% on symbols). day 1 of the experiment but increased back to 40% on day 4. On the other hand, there was no difference in the compared with the control reaching a maximum of 40 h, percentage of cells taking up leucine in the control and the which was one order of magnitude lower than the corre- P-amended mesocosms (ANOVA test, P = 0.86, Fig. 3). sponding increase in Pi turnover times (Fig. 1). Initial prokaryotic abundance was on average Community structure 5.9 ¥ 105 cells ml-1. Pi addition caused an increase in abundance, which reached values threefold higher than in The percent contribution of the different groups to the the control mesocosms at the end of the experiment prokaryotic assemblage did not change after Pi addition (Fig. 2A). Eubacterial cell numbers (estimated by cataly- and remained stable throughout the experiment (Fig. 4). sed reporter deposition fluorescence in situ hybridization, In this study we did not quantify the abundance of CARDFISH) accounted for ~ 70% of the prokaryotic cell Alphaproteobacteria but, instead, focused on two relevant counts throughout the experiment. Considering the alphaproteobacterial subgroups, SAR11 and Roseo- almost complete coverage of all the bacterial groups with bacter, which show contrasting life styles and are present the EUB338 probe mix (Amann and Fuchs, 2008), and the in significant numbers in diverse marine habitats (Morris low abundance of Archaea in Mediterranean surface et al., 2002; Buchan et al., 2005). SAR11 cells were the waters (Alonso-Sáez et al. 2007, De Corte et al., 2008) most abundant, comprising 40% and 48% of the DAPI

Phosphate ATP Leucine Fig. 3. Percentage of eubacterial cells 100 actively taking up phosphate (33P-Pi), AT33P 3 a C and H-leucine in the control (open bars) and 80 P-amended mesocosms (ﬁlled bars). Each cteri +P data point represents the average of two 60 mesocosms. Error bars represent the uba standard deviation. 40 (MAR+) ctive E a

20 % 0 014 014 014 day day day

and significant decrease in the percentage of cells taking up 33Pi in Synechococcus and SAR11 (ANOVA, P < 0.05), the percentage of labelled cells in the Roseobacter, Bacteroidetes and Gammaproteobacteria groups 1 day after Pi addition was similar in the control and P-amended mesocosms. However, within 4 days of Pi amendment the percentage of cells taking up 33Pi in all the groups had significantly dropped when compared with the control (ANOVA, P < 0.05, Fig. 6). In contrast to Pi uptake, the percentage of cells active in the uptake of ATP dropped significantly in all the probe-identified groups in the P-amended mesocosms on day 1 (Fig. 6, middle panel)

Fig. 4. Community composition in the control and P-amended but on day 4 it increased back considerably in the Gam- mesocosms. Data presented as percent contribution of the maproteobacteria, Roseobacter and Bacteroidetes. probe-identified groups to DAPI counts on day 4 of the experiment. Nonetheless, the percentage of cells taking up ATP on Very similar community composition was observed on day 0 and day 1. day 4 was still significantly lower when compared with the control in Gammaproteobacteria and SAR11. It is note- worthy that incubation times with 33Pi and AT-33P in the counts in the control and P-amended mesocosms respectively. The second most abundant group was the Gam- 100 maproteobacteria, which accounted for c. 15% of the Gammaproteobacteria 75 DAPI counts in both mesocosms. Then the Bacteroidetes 50 group, which comprised on average 10% of the DAPI counts, and Synechococcus and Roseobacter-related 25 bacteria, which represented around 3% and 2% respec- 0 tively (Fig. 4). The percentage of cells that hybridized with 100 SAR11 the EUB probes that could not be assigned to any of these 75 groups was negligible. 50 25 Percentage of cells within the probe-identified groups 0 taking up Pi, ATP and leucine 100 Roseobacter We next investigated the percentage of cells within the 75 probe-identified groups that were active in the uptake of 50 Pi, ATP and leucine. In this study cells were scored as 25 active (MAR+) when they were in contact with at least one silver grain. Although we are aware that a high number of 0 100 active cells does not necessarily imply high uptake rates, Bacteroidetes we considered that a group actively participated in the 75 uptake of a given substrate when a high percentage of 50 cells were labelled. At the beginning of the experiment, 25 before Pi addition, all the groups studied were relatively > 0 active ( 40% of the cells) in the uptake of Pi and ATP % of cells associated with silver grains (MAR+) 100 (Fig. 5). In contrast, Bacteroidetes and Synechococcus Synechococcus cells were relatively inactive in the uptake of leucine 75 (< 15% and < 10% of the cells respectively). The largest 50 percentage of cells that took up Pi and ATP was found in 25 the Gammaproteobacteria and Roseobacter groups (over 0 60% of the cells in these groups were MAR+), and both Pi ATP Leu these groups and SAR11 were the most active in the uptake of leucine. Fig. 5. Percentage of cells actively taking up phosphate (33P-Pi), Phosphate addition had contrasting effects on the AT33P and 3H-leucine (Leu) within each probe-identified group in the control mesocosms at the beginning of the experiment. Each uptake of the substrates by the different phylogenetic data point represents the average of two mesocosms. Error bars groups (Fig. 6). Whereas Pi amendment caused a rapid represent the standard deviation.

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology Bacterioplankton groups using Pi and DOP 5 Phosphate Fig. 6. Percentage of cells actively taking up 33P-Pi, AT33P and 3H-leucine within each 100 probe-identiﬁed group in the P-amended Synechococcus mesocosm plotted against the corresponding SAR11 80 Roseobacter percentage in the control mesocosms. The Bacteroidetes diagonal line indicates a 1:1 relationship. The Gammaprot. 60 circle highlights the difference in behaviour of

+P Gammaproteobacteria, Roseobacter and 40 Bacteroidetes from SAR11 and Synechococcus. 20 Day 1 Day 4 0 Control Control ATP 100

+P 40

20 Day 1 Day 4 0 Control Control Leucine 100

60 +P 40

20 Day 1 Day 4 0 0 204060801000 20406080100 Control Control

P-amended mesocosms were longer than in the controls were the second highest contributors, accounting for (see Experimental procedures) to ensure we could detect ~ 30% of the cells taking up Pi and ATP, and 15% of the differences in the uptake patterns of different bacterial cells taking up leucine. Bacteroidetes also contributed groups. Therefore, the non-significant differences that we substantially to the pool of cells taking up Pi and ATP in observed between the control and the P-treatments could the control mesocosms (~ 16%), but its contribution to the have been significant if we had kept the same incubation pool of cells taking up leucine was negligible. times in both mesocosms. Pi addition did not have any In the P-amended mesocosms, on day 1 of the experi- significant effect in the percentage of cells within the dif- ment, Gammaproteobacteria dominated the population ferent bacterial groups that was active in the uptake of taking up Pi (~ 40%), followed by Bacteroidetes and leucine (Fig. 6, lower panel, P > 0.05). SAR11 that showed similar contributions (27%) (Fig. 7). In contrast, on day 4, even though Gammaproteobacteria still contributed substantially to the cell pool taking up Pi Contribution of each group to cells taking up the (26%), the majority of the cells taking up Pi belonged to different substrates the SAR11 group (60%). The distribution of the percent SAR11 bacteria represented on average ~ 50% and 45% contribution of the bacterial groups to the population of the cells taking up Pi and ATP in the control meso- actively taking up ATP in the P-amended mesocosms was cosms, and completely dominated the pool of cells taking similar to the one observed in the control mesocosms, but up leucine in both mesocosms (~ 80% of the cells) with slightly higher contributions of Gammaproteobacteria (Fig. 7). In the control mesocosms Gammaproteobacteria (on average 37%) and Bacteroidetes (on average 22%).

SAR11 represented on average 37% of the ATP- Lugtenberg, 1987; Sebastian and Ammerman, 2011). Our consuming bacteria in these mesocosms. values in the control mesocosms were one to two orders Even though the percent contribution of the bacterial of magnitude lower than other ATP turnover time values groups to the pool of cells taking up leucine was unaf- measured also in the summer in the eastern Mediterra- fected by Pi addition, the total abundance of SAR11 cells nean Sea (Tanaka et al., 2011), indicating that ATP was taking up leucine did increase in the P-amended meso- being rapidly utilized in our system. ATP has been used in cosms in relation to the control mesocosms (Fig. S1). several studies as a model molecule to assess the dynamics of bioavailable low molecular weight DOP (Lovdal et al., 2007; Zubkov et al., 2007; Casey et al., Discussion 2009; Michelou et al., 2011) because it is ubiquitously found in seawater and its uptake is inhibited by a wide Turnover times of Pi and ATP, and prokaryotic variety of organic P compounds (Azam and Hodson, heterotrophic abundance and production 1977). In our study, the ratio between the ATP and Pi To evaluate the P-status of the initial seawater, we first turnover rates in the control mesocosms throughout the measured the turnover times of Pi and ATP, which indicate experiment indicated that ATP uptake was ~ 30% of Pi how fast these substrates are being utilized. Pi turnover uptake (details not shown), suggesting DOP played a times in the control bags were in the lower range of those substantial role in supplying P to satisfy the P require- observed in eastern Mediterranean waters (1–28 h, ments of the microbial cells in these P-starved waters. Tanaka et al., 2007). Zohary and Robarts (1998) estab- The fact that ATP turnover time in the P-amended meso- lished that Pi turnover times < 7 h in the eastern Mediter- cosms did not increase as much as the turnover time of Pi ranean were indicative of P-deficiency, suggesting that the (Fig. 1) might seem intriguing, since in the presence of microbial community in our control bags was very likely sufficient Pi we would not expect bacteria to use ATP as a P-starved. Indeed, our observed Pi turnover time values source of P, which would result in longer ATP turnover were similar to those measured in the western Mediterra- times. However, similar patterns were obtained in a recent nean when both phytoplankton and bacteria were P-limited study in the Mediterranean Sea (Tanaka et al., 2011). ATP (Thingstad et al., 1998) and similar to the values observed is mainly hydrolysed by two types of enzymes: alkaline during the CYCLOPS experiment before addition of Pi phosphatases (APases) and 5′-nucleotidases (5PN) (2–4 h) (Fonnes Flaten et al., 2005), when the bacterial (Ammerman and Azam, 1991). APases are non-specific community was P-limited. In these studies P-limitation was phosphomonoesterases that recognize terminal phos- defined as the limitation of the rates of heterotrophic phates of many phosphomonoesters and even short poly- prokaryotic activity. In our study, the observed increase in phosphate chains (McComb et al., 1979) whereas 5PN heterotrophic prokaryotic abundance and production in the recognizes the carbon moiety of 5′-nucleotides (Bengis- P-amended mesocosm supports the hypothesis that the Garber and Kushner, 1982). APases are induced upon P microbial community was originally limited by P availability. stress (e.g. Ammerman and Azam, 1991; Sebastian and An increase in heterotrophic prokaryotic production upon Ammerman, 2009), but in contrast, 5PN are not regulated Pi supply was also observed during the CYCLOPS experi- by Pi concentration (Ammerman and Azam, 1985; 1991; ment, but in contrast to what we observe here, prokaryotic Tamminen, 1989). Therefore, 5PN were probably respon- numbers were unaffected by the increase in P (Pitta et al., sible for most of ATP hydrolysis in the P-amended meso- 2005). In the P-amended mesocosms Pi turnover time cosms, although it is also possible that the concentration of increased, as it has been observed in other studies when Pi Pi was not enough to completely shut down APase activity was added (Thingstad et al., 2005; Tanaka et al., 2011) or and both enzymes were contributing to the hydrolysis of when there is an injection of Pi in the system due to winter ATP. The reason we observed uptake of the hydrolysed Pi mixing (Tanaka et al., 2004) or due to local convective from ATP in the P-amended mesocosms might be that 5PN mixing events, as it occurs in the Rhodes Gyre (Zohary and and APases are membrane bound or periplasmic enzymes Robarts, 1998). (e.g. Bengis-Garber and Kushner, 1982; Martinez and Pi is the preferred source of P for the microbial commu- Azam, 1993) and therefore the Pi released upon hydrolysis nity. When Pi is not sufficient to meet their requirements, could be taken up directly without intermediate mixing with bacteria induce the Pho regulon, a suite of genes that the pool of ambient Pi (Tamminen, 1989; Ammerman and code for proteins required for scavenging Pi and the use Azam, 1991; Thingstad et al., 1993). of DOP compounds, like ATP (Wanner, 1996; Sebastian and Ammerman, 2009). The turnover times of ATP in the Single cell analyses control mesocosms were longer than the Pi turnover times (Fig. 1) because ATP needs to be hydrolysed prior MARFISH has been used in environmental studies in an to uptake by the cell (Bengis-Garber and Kushner, 1982; attempt to understand the ecophysiological roles that

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology Bacterioplankton groups using Pi and DOP 7 microbes play in complex natural systems (e.g. Cottrell P-amended mesocosm the radiolabelled Pi was diluted and Kirchman, 2000; Vila-Costa et al., 2007; Salcher more than 300-fold with ambient Pi (Fig. S2). et al., 2011). This technique provides qualitative informa- There might be considerable heterogeneity in the actual tion about the amount of cells in a probe-identified group ‘stress’ different bacteria experience under the same envi- that is actively taking up a radiolabelled substrate (e.g. del ronmental conditions, depending on their different cellular Giorgio and Gasol, 2008). Because of the fact that larger quotas, nutrient acquisition capacities and metabolism. silver grain clusters corresponds to higher uptake per Our results show that the percentage of cells taking cell than smaller clusters (Nielsen et al., 2003), some up 33Pi in Gammaproteobacteria, Roseobacter and researchers have quantified the uptake by different bac- Bacteroidetes in the P-amended mesocosm 1 day after Pi terial groups by measuring the silver grain area around addition was similar to the ones observed in the control the cells (e.g. Cottrell and Kirchman, 2003; Sintes and mesocosms, whereas the percentages in SAR11 and Herndl, 2006). However, relative grain densities around Synechococcus had considerably decreased. This obser- organisms depend not only on the cell but also on the vation suggests that Gammaproteobacteria, Roseobacter energy of the tracer and the thickness and sensitivity of and Bacteroidetes cells by day 1 had not yet completely the photographic emulsion (Perry, 1964). High energy replenished their P quotas. However, on day 4 of the isotopes like 33P provide high sensitivity but also the high experiment the percentage of cells taking up 33Pi in these energy particles produced during decay of the isotope groups decreased to values similar to the ones observed travel further through the autoradiographic emulsion and in SAR11 and Synechococcus, suggesting that all the produce a wide scatter of silver grains, and therefore, less groups had by then filled up their internal cell quotas. Fast resolution (Okabe et al., 2004). accumulation of particulate phosphorus in the bacterial Due to difficulties in accurately estimating the silver size fraction due to luxury consumption (i.e. Pi uptake in grain area when using 33P, in this study cells were scored excess of the immediate requirement for growth) upon a as active (MAR+) when they were in contact with at least Pi pulse is a recurrent phenomenon in P-starved waters one silver grain. Even though the scattering of silver (Thingstad et al., 1993; 1996; 2005). Indeed, culture grains could yield errors in the quantification of MAR+ studies have shown that one of the components of the cells, we obtained very consistent results even when rep- Pho-regulon is the enzyme polyphosphate kinase (Ppk) licates were processed separately, in different hybridiza- (Geissdorfer et al., 1998; Jahid et al., 2006), which cataly- tions and microautoradiography batches. All the bacterial ses the formation of polyphosphates (PolyP) in P-starved phylogenetic groups contributed to the uptake of both Pi cells in response to a sudden Pi supply. Despite their and ATP in the P-starved waters of the control meso- potential importance for survival in P-depleted systems, cosms (Fig. 5). Despite the MARFISH approach does not studies about the role of PolyP in oligotrophic marine enable us estimating uptake rates, we can make some environments are scarce, and limited only to Cyanobac- inferences based on our observations and knowledge of P teria (Orchard et al., 2010) or to in silico studies (Temper- metabolism. When phytoplankton and bacterial cells are ton et al., 2011). By searching for the ppk gene in the P-starved, a sudden supply of phosphate results in surge genomes of marine cultured representatives, we investi- uptake, followed by lower uptake rates as the internal cell gated whether the different probe-identified groups in this quotas get filled up (Jansson, 1993; Ducobu et al., 1998). study are potentially able to accumulate phosphorus This observation has been made with cultures but is sup- inside the cells. We found that most of the cultured rep- ported by field observations. For instance, higher uptake resentatives in the Gammaproteobacteria, Roseobacter, rates are observed under P stress in aquatic systems (e.g. Bacteroidetes and Synechococcus lineages are poten- Vadstein and Olsen, 1989; Fonnes Flaten et al., 2005) tially able to form PolyP (Table S1). In contrast, only one and Suttle and colleagues (1988) observed that Pi addi- out of the five isolates of SAR11 sequenced to date pos- tion to P-starved waters resulted in rapid Pi uptake, but sesses this enzyme, indicating that this group likely plays the uptake rates gradually declined when followed during a limited role in PolyP accumulation. Using staining pro- a time course. In our experiment, the percentage of cells cedures (see Experimental procedures) we confirmed taking up leucine was similar between the P-amended that Synechococcus and some heterotrophic bacterial and the control mesocosms throughout the experiment cells accumulated PolyP in the P-amended mesocosms, (Fig. 6), thereby indicating that the cells were similarly while PolyP formation did not occur in the control meso- active. Therefore, we posit that the decrease in the per- cosms (Fig. 8 and Fig. S3). We found that the percentage centage of 33P-labelled cells in the P-amended meso- of PolyP-accumulating heterotrophic bacterial cells in the cosms is mainly due to a reduction in Pi uptake rates P-amended mesocosms increased from day 2 to day 3, (Fig. 3 and Fig. 6, upper panels). A lower percentage of whereas the percentage of Synechococcus cells contain- labelled cells does not imply that non-labelled cells ing PolyP was already notably high on day 2 and stopped completely the uptake of Pi, because in the remained constant on day 3 (Fig. 8). Although unfortu-

Fig. 7. Contribution of each analysed bacterial phylogenetic group to the population of cells taking up 33P-Pi, AT33P and 3H-leucine in both the control and P-amended mesocosms. Percentages were calculated relative to the total number of cells identiﬁed by the different probes. Asterisks highlight those samples in the P-amended mesocosms where the percent contribution was notably different from the ones observed in the rest of the samples. nately we do not have data for day 1 and 4 of the experi- et al., 2011), resulting in further lower P requirements. In ment, these results are consistent with our hypothesis that contrast, the opportunitrophs Gammaproteobacteria and Synechococcus cells had replenished their P quota by Roseobacter frequently display growth rates well above day 1. Flow cytometric sorting of microbial populations the average community growth (Yokokawa and Nagata, has shown that Synechococcus cells display per cell 2005; Teira et al., 2009; Ferrera et al., 2011), and these phosphate uptake rates higher than heterotrophic bacte- high growth rates are reﬂected in a high P demand. Strik- ria (Zubkov et al., 2007; Michelou et al., 2011), which may ingly, cells belonging to the fast growers NOR5/OM60 explain why they replenished their P quotas faster. On the clade, which represented around 40% of the Gammapro- other hand, the increase in heterotrophic prokaryotes con- teobacteria in our experiment (data not shown), were taining PolyP granules is likely due to gradual PolyP accu- elongated in the control mesocosms and upon P addition mulation in Gammaproteobacterial, Bacteroidetes and cells returned to coccoid or rod-shaped morphologies Roseobacter cells. (Fig. S4). Enlargement of cells under P starvation condi- Our results support the notion of the occurrence of two tions has been attributed to inhibition of DNA synthesis contrasting ecological lifestyle strategies in heterotrophic and, as a consequence, of cell division (Seeger and bacteria, as it has been hitherto suggested: the oligotro- Jerez, 1993), suggesting that this clade may be strongly phs and the opportunitrophs or copiotrophs (which are affected by P availability. able to exploit temporally variable resources) (e.g. Polz SAR11 bacteria dominated the population of cells taking et al., 2006; Lauro et al., 2009). Most of the phosphorus up Pi, ATP and leucine in the control mesocosms, but demands in the bacterial cells are allocated to the synthe- in the P-amended mesocosms Gammaproteobacteria sis of phospholipids or nucleic acids (Van Mooy et al., increased considerably their relative contribution and 2006), being the synthesis of RNA the largest biochemical even dominated the pool of cells taking up Pi on day 1 of sink for Pi in the ocean (Bossard and Karl; 1986 Van Mooy the experiment (Fig. 7). Large number of cells taking up a and Devol, 2008). The strict oligotroph SAR11, which substrate does not imply a higher contribution to the total dominates P-depleted environments, have gone through uptake rate, because the least abundant species may a genome streamlining process that greatly reduces their contribute the most to substrate uptake if they have high P demands (Giovannoni et al., 2005). Likewise, picocy- per cell rates (Musat et al., 2008). Nevertheless, gam- anobacteria like Synechococcus and Prochlorococcus maproteobacterial cells in all cases showed many per are able to synthesize non-phosphorus membrane lipids cell-associated silver grains (Fig. S5). Because silver under P-stress conditions, which can spare them between grain area is proportional to uptake rate (e.g. Sintes and 5–43% of their cellular P requirements (Van Mooy et al., Herndl, 2006), Gammaproteobacteria seem to have 2009). Furthermore, RNA synthesis is related to growth higher per cell rates than the rest of the groups consid- rate (Kemp et al., 1993) and SAR11 are slow growers ered. As Gammaproteobacteria was the second most (Yokokawa and Nagata, 2005; Teira et al., 2009; Ferrera abundant group in both mesocosms throughout the

submersible water pump. All equipment used had been aged 15 A for at least 1 week with tap water, washed with HCL (10%) and rinsed three times with deionized water. Once ﬁlled, the 10 tanks were taken to the Heraklion harbour and transported by truck to the Hellenic Centre for Marine Research facilities, 5 15 km east of Heraklion, on the north coast of Crete. The

containing PolyP containing PolyP water was gravity poured into six mesocosms (polyethylene 0 bags, 1.32 m diameter) of 3 m3 each, which were incubated 3 % of heterotrophic prokaryotes 100 in a large land-based tank (350 m ) with running water that B controlled temperature. Water was allowed to sit overnight. 75 The following day, time zero was sampled and immediately afterwards three of the bags were amended with 100 nM 50 3- PO4 (ﬁnal concentration, P-amended bags), and the other

Synechococcus 25 three were left as a control (Control bags). All the bags were containing PolyP

% of continuously mixed by light air bubbling. Samples were taken 0 daily. For MARFISH, and Pi and ATP uptake, samples were Control day 2 day 3 taken only from two bags of each treatment, in order to be able to process a higher amount of replicates and time points. P-amended

Fig. 8. Polyphosphate (PolyP) staining results. 33 3- 33 A. Percentage of heterotrophic prokaryotes containing PolyP PO4 and AT P uptake measurements granules. 33 -1 B. Percentage of Synechococcus cells with PolyP. H3 PO4 (40–158 Ci mg ; Perkin Elmer) was diluted in dis- Control data represent the average percentage of PolyP-containing tilled water and 15 ml of aliquot was added to duplicate 9 ml of cells on day 2 and day 3 in the control mesocosms. Error bars subsample to give final concentrations ranging between 25 represent the standard deviation. and 108 pM. Adenosine 5′-[g-33P]triphosphate (AT33P; Perkin Elmer) was diluted in distilled water, and 3 ml were added to 9 ml of subsample to give a final concentration of 100 pM. experiment, they likely played a major role in Pi and ATP Duplicate killed controls were included with each set of uptake in our system. samples. Killed controls were amended with paraformalde- In summary, based on the Pi turnover time results and hyde (2% final concentration) 30 min before the addition of the observed increase in prokaryotic heterotrophic abun- the isotopic tracer. Incubations were done in 15 ml Falcon dance and production upon Pi addition, heterotrophic bac- tubes at room temperature and subdued light. The duration of each incubation varied depending on the expected turnover teria were originally P limited in this study. All the bacterial time, ranging from 20 min to 2 h. Incubations were terminated groups actively utilized DOP under P limiting conditions, by the addition of paraformaldehyde (2% final concentration) but the percentage of cells taking up DOP decreased and filtered, within 30 min, onto a 0.2 mm polycarbonate filter, considerably under higher Pi concentrations. The patterns which was placed on top of a Whatman (GF/C) glass fibre -1 of Pi utilization in the control and P-amended mesocosms filter saturated with 100 mmol l KH2PO4. To stop incuba- among the bacterial groups supports the coexistence of tions, fixation was chosen over cold-chase addition of cold 3- two ecological life strategies, the oligotrophs like SAR11, PO4 and ATP to be consistent with the methodology employed during the MARFISH analyses (see below). which are frugal in their P requirements, and opportuni- However, phosphate turnover time estimations using the tra- trophs like Gammaproteobacteria and Roseobacter, ditional cold-chase method were also performed to facilitate which have high P requirements and are able to accumu- comparison with previous studies (see P. Pitta, J.C. Nejst- late phosphate inside the cell. Different P requirements gaard, T.M. Tsagaraki, J. Egge, C. Frangoulis, A. Lagaria, I. are probably reflected into different degrees of in situ P Magiopoulos, S. Psarra, R.A. Sandaa, E.F. Skjoldal, T. stress. Further analyses with molecular markers of P Tanaka, R. Thyrhaug, S. Zervoudaki and T.F. Thingstad, in stress are on the way to elucidate the degree of hetero- preparation). Fixation has an experimental limitation, which is that some of the accumulated 33P could leak out of fixed cells geneity in the levels of stress of the different bacterial after the cell membranes become compromised. Talarmin groups. and colleagues (2011) recently reported that up to 40% of the 33Pi label is lost from fixed heterotrophic bacterial cells immediately after fixation and Casey and colleagues (2009) esti- Experimental procedures mated that ~ 25% and ~ 13% of the isotope in Pi and ATP Mesocosms set-up and sampling incubations respectively, could leak out from the cells within 24 h. To minimize this leakage, samples were filtered within Water was collected in September 2009 from the eastern 1 h after stopping the incubation. After filtration, the filters Mediterranean Sea offshore Crete, aboard the R/V Philia, were rinsed twice with sterile Milli-Q water and transferred to from a site 170 m deep and about 5 nautical miles north of scintillation vials with 1 ml of Ultima Gold scintillation cocktail. 33 3- Heraklion (35°24.957′N, 25°14.441′E). Surface water was Aliquots (50 ml) from the subsamples incubated with PO4 pumped into high density polyethylene tanks of 1 m3 using a and AT33P were transferred directly to scintillation vials and

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology 10 M. Sebastián et al. mixed with 1 ml of scintillation cocktail to measure the total time. This equation follows an asymptotic curve, and incuba- added radioactivity. Samples were radioassayed in a Packard tion times were always chosen so that the uptake of the Tri-Carb 4000 scintillation counter. Turnover times were cal- substrate was well in the linear phase, but long enough to be culated using the equation T = t/[-ln (1 - R)], where able to appreciate differences between phylogenetic groups. t = incubation time and R = consumed fraction of added One replicate (for each substrate and treatment) was killed tracer (Thingstad et al., 1993). Killed controls were sub- with formaldehyde before the addition of the radiolabelled tracted prior to the calculations. substrates and was used as a control. At the end of the incubation, samples were fixed with paraformaldehyde, allowed to sit in the dark for at least 1 h, and then portions of Prokaryotic abundance 10 ml were filtered onto three different 0.2 mm polycarbonate Samples for prokaryotic abundance determination (2 ml) filters. Filters were washed twice with sterile Milli-Q water and were fixed with a 0.2 mm filtered 25% glutaraldehyde solution, frozen at -80°C until processing in the lab. Filters were then to give a final concentration of 0.5%. After fixation (20 min, hybridized following the CARDFISH protocol (Pernthaler 4°C), the samples were fast frozen in liquid nitrogen and et al., 2002) to identify the different bacterial groups. After finally stored in -80°C. Once in the laboratory, the samples thawing, the filters were dipped in 0.1% agarose, dried at were thawed at room temperature, stained with SYBRGreen 37°C, and then dehydrated with 95% ethanol. This allowed I (final dilution: 4 ¥ 10-4 of the commercial dilution) and incu- attachment of the cells to the filters. Then, cell walls were bated for 10 min in the dark. Prokaryotic abundance was permeabilized with lysozyme (1 h) and achromopeptidase determined on a daily basis with a Becton-Dickinson FACS- (30 min) at 37°C. Filters were cut into multiple pieces and Calibur flow cytometer using Milli-Q water as sheath fluid. hybridized with one of six horseradish peroxidase (HRP)- Abundances were calculated from the measured flow rate. labelled probes: EUB338 I-II and –III (targets most Eubacte- Flow cytometry data were acquired with the CellQuest ria, Daims et al., 1999), GAM42a together with its unlabelled and analysed with the Paint-A-Gate software packages competitor probe (targets most Gammaproteobacteria, Manz (Becton-Dickinson). et al., 1992), CF319a (targets many members of the Bacteroidetes group, Manz et al., 1996), ROS537 (targets members of the Alphaproteobacteria Roseobacter- Prokaryotic heterotrophic production Sulfitobacter-Silicibacter group, Eilers et al., 2000), SYN405 (targets Synechococcus, West et al., 2001), SAR11-441R Prokaryotic heterotrophic production was determined in all (targets the Alphaproteobacteria SAR11, Morris et al., 2002). mesocosms every day by uptake of tritium-labelled leucine Specific hybridization conditions were established by addition (Kirchman et al., 1985) using the centrifugation procedure. of formamide to the hybridization buffers (45% formamide for Triplicate samples and one prefixed control sample were the SAR11 probe, 60% for the SYN405 probe, 55% for the 3 -1 incubated with H-Leucine (4.27 TBq mmol , Perkin Elmer, other probes). Hybridization was performed overnight at Boston, USA) at a final concentration of 60 nM. Incubation 35°C. For amplification, we used tyramide labelled with Alexa was performed in the dark at in situ temperature for 1 h and 488. After processing, a small portion of the filter was cut and stopped with 5% TCA, final concentration. The samples were stained with DAPI (final concentration 1 mgml-1) to quantify then centrifuged at 16 000 g for 10 min before removal of the the abundance of the different phylogenetic groups in relation supernatant. Then the samples were washed twice by adding to total prokaryotic counts. The rest of the filter was glued 5% TCA, vortexed, centrifuged and the supernatant removed. onto a glass slide and subsequently processed for microau- Counting cocktail (Ecoscint A, National Diagnostics, Atlanta, toradiography as described in detail in Alonso-Sáez and USA) was added and the uptake of radioactive leucine mea- Gasol (2007), which is a modification of the protocol sured by liquid scintillation counting. Conversion of leucine described by Alonso and Pernthaler (2005). Exposure times to carbon units was done with the theoretical factor were determined empirically for each compound by following -1 1.5 kg C mol Leu . changes in number of cells taking up the substrate over time. Optimal exposure times were selected once the number of MAR-(CARD)-FISH cells taking up the substrate reached a plateau but accumulation of silver grains still allowed visualization of the cells Incubations for microautoradiography were similar to those to associated to them. Cells were counted in an Olympus BX61 measure bulk uptake rates. Samples (30 ml) were spiked epifluorescence microscope. Cells touching or overlapping 33 33 3 with H3 PO4,AT P and H-leucine (Perkin Elmer) to yield silver grains after developing of the emulsion were consid- 25–108 pM, 100 pM and 0.5 nM respectively. The samples ered as active cells or MAR+ cells. For abundance of probe- 33 33 were incubated for 1 h (H3 PO4) and 3 h (AT P) in the positive cells, between 500 and 1000 DAPI-positive cells 33 33 control mesocosms and 4 h (H3 PO4,ATP) in the were counted manually in a minimum of 10 fields. Killed P-amended mesocosms. Incubations with 3H-leucine always controls were evaluated with the probe EUB338 I-II and –III. lasted 2.5 h. Because of the expected decrease in the uptake The proportion of labelled cells in the killed controls was 6% 33 33 3 of Pi and ATP in the P-amended mesocosms due to substrate for H3 PO4,4%forAT P and 2% for H-leucine. This propor- dilution with ambient phosphate, longer incubations were tion was not subtracted from the percent of cells taking up the carried out to ensure detection. Uptake of Pi and ATP is well substrates in the live incubations. Even though Synechococ- described by the theoretical equation R(t) = (1 - e-t/T) (e.g. cus is not a heterotroph we included it in our analyses Thingstad et al., 1993), where R(t) is the fraction of added because it had been hypothesized to be the major player in Pi label consumed after incubation time t and T is the turnover uptake in eastern Mediterranean (Moutin et al., 2002).

Polyphosphate staining waters: characterization of enzyme activity. Limnol Ocean- ogr 36: 1427–1436. Polyphosphate (PolyP) granules were stained with DAPI. Ammerman, J.W., Hood, R.R., Case, D.A., and Cotner, J.B. This stain is commonly used for DNA detection but it also (2003) Phosphorus deficiency in the Atlantic: an emerging interacts with PolyP. When excited at 360 nm DAPI-DNA paradigm in oceanography. Eos 84: 165–170. fluorescence is blue-white, but DAPI-PolyP emits a typical Azam, F., and Hodson, R.E. (1977) Size distribution and green yellow fluorescence signal in organisms that are able activity of marine microheterotrophs. Limnol Oceanogr 22: to accumulate high levels of PolyP (Tijssen et al., 1982). 492–501. Paraformaldehyde-fixed cells, collected onto 0.2 mm polycar- Bengis-Garber, C., and Kushner, D.J. (1982) Role of bonate filters were stored frozen at -80°C until analyses. For membrane-bound 5′-nucleotidase in nucleotide uptake by the staining, filters were submerged in 5 ml of PBS and 50 ml the moderate halophile Vibrio costicola. J Bacteriol 149: -1 of DAPI solution (0.5 mg ml ) were added. After incubating 808–815. for 3 min in the dark, filters were washed shortly with distilled Bentzen, E., Taylor, W.D., and Millard, E.S. (1992) The impor- water and dried at room temperature. Cells were counted in tance of dissolved organic phosphorus to phosphorus an Olympus BX61 epifluorescence microscope. For the uptake by limnetic plankton. Limnol Oceanogr 37: 217– determination of PolyP-accumulating heterotrophic bacteria, 231. PolyP granules associated to pigmented picoplankton were Bossard, P., and Karl, D.M. (1986) The direct measurement substracted from each frame (see Fig. S3). of ATP and adenine nucleotide pool turnover in microorganisms: a new method for environmental assessment of metabolism, energy flux and phosphorus dynamics. Acknowledgements J Plankton Res 8: 1–13. The mesocosm experiment in the Cretacosmos facility was Buchan, A., Gonzalez, J.M., and Moran, M.A. (2005) Over- arranged in cooperation by project Nutritunnel financed by view of the marine Roseobacter lineage. Appl Environ the Research Council of Norway and the MESOAQUA EU Microbiol 71: 5665–5677. network. The processing of the samples was supported Caron, D.A., Lim, E.L., Sanders, R.W., Dennett, M.R., and by the grants FOSMICRO (CTM2009-07679-E), STORM Berninger, U.G. (2000) Responses of bacterioplankton and (CTM2009-09352/MAR), SUMMER (CTM2008-03309/MAR) phytoplankton to organic carbon and inorganic nutrient and MALASPINA (CSD2008 – 00077) funded by the Spanish additions in contrasting oceanic ecosystems. Aquat Microb Ministry of Science and Innovation. We thank Irene Forn and Ecol 22: 175–184. Clara Ruiz for their help with the CARDFISH and MARFISH Casey, J.R., Lomas, M.W., Michelou, V.K., Dyhrman, S.T., techniques, Iordanis Magiopoulos for performing the flow Orchard, E.D., Ammerman, J.W., and Sylvan, J.B. (2009) cytometer analyses and Tatiana Tsagaraki for performing the Phytoplankton taxon-specific orthophosphate (Pi) and ATP inorganic nutrient analyses. Marta Sebastián was supported utilization in the western subtropical North Atlantic. Aquat by the Spanish Ministry of Science and Innovation through a Microb Ecol 58: 31–44. ‘Juan de la Cierva’ award. Cotner, J.B., Ammerman, J.W., Peele, E.R., and Bentzen, E. (1997) Phosphorus-limited bacterioplankton growth in the Sargasso Sea. Aquat Microb Ecol 13: 141–149. References Cottrell, M.T., and Kirchman, D.L. (2000) Natural assem- blages of marine proteobacteria and members of the Alonso, C., and Pernthaler, J. (2005) Incorporation of glucose Cytophaga-Flavobacter cluster consuming low-and high- under anoxic conditions by bacterioplankton from coastal molecular-weight dissolved organic matter. Appl Environ North Sea surface waters. Appl Environ Microbiol 71: Microbiol 66: 1692–1697. 1709–1716. Cottrell, M.T., and Kirchman, D.L. (2003) Contribution of Alonso-Sáez, L., and Gasol, J.M. (2007) Seasonal variation major bacterial groups to bacterial biomass production in the contribution of different bacterial groups to the (thymidine and leucine incorporation) in the Delaware uptake of low molecular weight-compounds in NW Medi- estuary. Limnol Oceanogr 48: 168–178. terranean coastal waters. Appl Environ Microbiol 73: Currie, D.J., and Kalff, J. (1984) The relative importance of 3528–3535. phytoplankton and bacterioplankton in phosphorus uptake Alonso-Sáez, L., Balague, V., Sa, E., Sanchez, O., Gonzalez, in freshwater. Limnol Oceanogr 29: 311–321. J.M., Pinhassi, J., et al. (2007) Seasonality in bacterial Daims, H., Brühl, A., Amann, R., Schleifer, K.H., and Wagner, diversity in north-west Mediterranean coastal waters: M. (1999) The domain-specific probe EUB338 is insuffi- assessment through clone libraries, fingerprinting and cient for the detection of all bacteria: development and FISH. FEMS Microbiol Ecol 60: 98–112. evaluation of a more comprehensive probe set. Syst Appl Amann, R., and Fuchs, B.M. (2008) Single-cell identification Microbiol 22: 434–444. in microbial communities by improved fluorescence in situ De Corte, D., Yokokawa, T., Varela, M.M., Agogué, H., and hybridization techniques. Nat Rev Microbiol 6: 339–348. Herndl, G.J. (2008) Spatial distribution of Bacteria and Ammerman, J.W., and Azam, F. (1985) Bacterial Archaea and amoA gene copy numbers throughout the 5-nucleotidase in aquatic ecosystems: a novel mechanism water column of the Eastern Mediterranean Sea. ISME J 3: of phosphorus regeneration. Science 227: 1338–1340. 147–158. Ammerman, J.W., and Azam, F. (1991) Bacterial Ducobu, H., Huisman, J., Jonker, R.R., and Mur, L.R. (1998) 5′-nucleotidase activity in estuarine and coastal marine Competition between a prochlorophyte and a cyanobacte-

rium under various phosphorus regimes: comparison with strategy in marine bacteria. Proc Natl Acad Sci USA 106: the Droop model. J Phycol 34: 467–476. 15527. Eilers, H., Pernthaler, J., Glockner, F.O., and Amann, R. Lomas, M.W., Burke, A.L., Lomas, D.A., Bell, D.W., Shen, C., (2000) Culturability and in situ abundance of pelagic bac- Dyhrman, S.T., and Ammerman, J.W. (2010) Sargasso teria from the North Sea. Appl Environ Microbiol 66: 3044. Sea phosphorus biogeochemistry: an important role for Ferrera, I., Gasol, J.M., Sebastián, M., Hojerová, E., and dissolved organic phosphorus (DOP). Biogeosciences 7: Koblí\vzek, M. (2011) Comparison of growth rates of 695–710. aerobic anoxygenic phototrophic bacteria and other bacte- Longnecker, K., Lomas, M.W., and Van Mooy, B.A.S. (2010) rioplankton groups in coastal Mediterranean waters. Appl Abundance and diversity of heterotrophic bacterial cells Environ Microbiol 77: 7451–7458. assimilating phosphate in the subtropical North Atlantic Fonnes Flaten, G.A., Skjoldal, E.F., Krom, M.D., Law, C.S., Ocean. Environ Microbiol 12: 2773–2782. Mantoura, R.F.C., Pitta, P., et al. (2005) Studies of the Lovdal, T., Tanaka, T., and Thingstad, T.F. (2007) Algal- microbial P-cycle during a Lagrangian phosphate-addition bacterial competition for phosphorus from dissolved DNA, experiment in the Eastern Mediterranean. Deep Sea Res ATP, and orthophosphate in a mesocosm experiment. Part II Top Stud Oceanogr 52: 2928–2943. Limnol Oceanogr 52: 1407–1419. Geissdorfer, W., Ratajczak, A., and Hillen, W. (1998) Tran- Lugtenberg, B. (1987) The pho regulon in Escherichia coli.In scription of ppk from Acinetobacter sp. strain ADP1, encod- Phosphate Metabolism and Cellular Regulation in Microor- ing a putative polyphosphate kinase, is induced by ganisms. Torriani-Gorini, A., Rothman, F.G., Silver, S., phosphate starvation. Appl Environ Microbiol 64: 896. Wright, A., and Yagil, E. (eds). Washington, DC, USA: Am. del Giorgio, P.A., and Gasol, J.M. (2008) Physiological struc- Soc. Microbiol., pp. 1–2. ture and single-cell activity in marine bacterioplankton. In McComb, R.B., Bowers, G.N., and Posen, S. (1979) Alkaline Microbial Ecology of the Oceans, Second Edition. Kirch- Phosphatase. New York, NY, USA: Plenum Publishing Cor- man, D.L. (ed.). Hoboken, NJ, USA: John Wiley & Sons, poration. pp. 243–298. Manz, W., Amann, R., Ludwig, W., Wagner, M., and Schleifer, Giovannoni, S.J., Tripp, H.J., Givan, S., Podar, M., Vergin, K.H. (1992) Phylogenetic oligodeoxynucleotide probes for K.L., Baptista, D., et al. (2005) Genome streamlining in a the major subclasses of proteobacteria: problems and cosmopolitan oceanic bacterium. Science 309: 1242. solutions. Syst Appl Microbiol 15: 593–600. Jahid, I.K., Silva, A.J., and Benitez, J.A. (2006) Polyphos- Manz, W., Amann, R., Ludwig, W., Vancanneyt, M., and phate stores enhance the ability of Vibrio cholerae to over- Schleifer, K.H. (1996) Application of a suite of 16S rRNA- come environmental stresses in a low-phosphate specific oligonucleotide probes designed to investigate environment. Appl Environ Microbiol 72: 7043. bacteria of the phylum Cytophaga-Flavobacter- Jansson, M. (1993) Uptake, exchange, and excretion of Bacteroides in the natural environment. Microbiology 142: orthophosphate in phosphate-starved Scenedesmus quad- 1097–1106. ricauda and Pseudomonas K7. Limnol Oceanogr 38: Martinez, J., and Azam, F. (1993) Periplasmic aminopepti- 1162–1178. dase and alkaline phosphatase activities in a marine bac- Kemp, P.F., Lee, S., and LaRoche, J. (1993) Estimating the terium: implications for substrate processing in the sea. growth rate of slowly growing marine bacteria from RNA Mar Ecol Prog Ser 92: 89–89. content. Appl Environ Microbiol 59: 2594–2601. Michelou, V.K., Lomas, M.W., and Kirchman, D.L. (2011) Kirchman, D., K’nees, E., and Hodson, R. (1985) Leucine Phosphate and adenosine-5′-triphosphate uptake by incorporation and its potential as a measure of protein cyanobacteria and heterotrophic bacteria in the Sargasso synthesis by bacteria in natural aquatic systems. Appl Sea. Limnol Oceanogr 56: 323–332. Environ Microbiol 49: 599. Morris, R.M., Rappé, M.S., Connon, S.A., Vergin, K.L., Krom, M.D., Kress, N., Brenner, S., and Gordon, L.I. (1991) Siebold, W.A., Carlson, C.A., and Giovannoni, S.J. (2002) Phosphorus limitation of primary productivity in the eastern SAR11 clade dominates ocean surface bacterioplankton Mediterranean Sea. Limnol Oceanogr 36: 424–432. communities. Nature 420: 806–810. Krom, M.D., Thingstad, T.F., Brenner, S., Carbo, P., Drako- Moutin, T., Thingstad, T.F., Van Wambeke, F., Marie, D., poulos, P., Fileman, T.W., et al. (2005) Summary and over- Slawyk, G., Raimbault, P., and Claustre, H. (2002) Does view of the CYCLOPS P addition Lagrangian experiment in competition for nanomolar phosphate supply explain the the Eastern Mediterranean. Deep Sea Res Part II Top Stud predominance of the cyanobacterium Synechococcus? Oceanogr 52: 3090–3108. Limnol Oceanogr 47: 1562–1567. Laghdass, M., Catala, P., Caparros, J., Oriol, L., Lebaron, P., Musat, N., Halm, H., Winterholler, B., Hoppe, P., Peduzzi, S., and Obernosterer, I. (2011) High Contribution of SAR11 Hillion, F., et al. (2008) A single-cell view on the ecophysi- to Microbial Activity in the North West Mediterranean ology of anaerobic phototrophic bacteria. Proc Natl Acad Sea. Microb Ecol 63: 1–10. DOI: 10.1007/s00248-011- Sci USA 105: 17861. 9915-7. Nielsen, J.L., Christensen, D., Kloppenborg, M., and Nielsen, Larsen, A., Tanaka, T., Zubkov, M.V., and Thingstad, T.F. P.H. (2003) Quantification of cell-specific substrate uptake (2008) P-affinity measurements of specific osmotroph by probe-defined bacteria under in situ conditions by populations using cell-sorting flow cytometry. Limnol microautoradiography and fluorescence in situ hybridiza- Oceanogr Methods 6: 355–363. tion. Environ Microbiol 5: 202–211. Lauro, F.M., McDougald, D., Thomas, T., Williams, T.J., Egan, Okabe, S., Kindaichi, T., and Ito, T. (2004) MAR-FISH – an S., Rice, S., et al. (2009) The genomic basis of trophic ecophysiological approach to link phylogenetic affiliation

and in situ metabolic activity of microorganisms at a single- ing of ammonium and phosphate: implications for carbon cell resolution. Microbes Environ 19: 83–98. transfer in planktonic communities. Limnol Oceanogr 35: Orchard, E.D., Benitez-Nelson, C.R., Pellechia, P.J., Lomas, 424–433. M.W., and Dyhrmane, S.T. (2010) Polyphosphate in Tri- Talarmin, A., Van Wambeke, F., Duhamel, S., Catala, P., chodesmium from the low-phosphorus Sargasso Sea. Moutin, T., and Lebaron, P. (2011) Improved methodology Limnol Oceanogr 55: 2161–2169. to measure taxon-specific phosphate uptake in live and Pernthaler, A., Pernthaler, J., and Amann, R. (2002) Fluores- unfiltered samples. Limnol Oceanogr Methods 9: 443–453. cence in situ hybridization and catalyzed reporter deposi- Tamminen, T. (1989) Dissolved organic phosphorus regen- tion for the identification of marine bacteria. Appl Environ eration by bacterioplankton: 5′-nucleotidase activity and Microbiol 68: 3094–3101. subsequent phosphate uptake in a mesocosm enrichment Perry, R.P. (1964) Quantitative autoradiography. In Methods experiment. Mar Ecol Prog Ser 58: 89–100. in Cell Physiology. Prescott, D.M. (ed.). New York, NY, Tanaka, T., Rassoulzadegan, F., and Thingstad, T.F. (2004) USA: Academic Press, pp. 305–326. Orthophosphate uptake by heterotrophic bacteria, cyano- Pinhassi, J., Gómez-Consarnau, L., Alonso-Sáez, L., Sala, bacteria, and autotrophic nanoflagellates in Villefranche M.M., Vidal, M., Pedrós-Alió, C., and Gasol, J.M. (2006) Bay, northwestern Mediterranean: Vertical, seasonal, and Seasonal changes in bacterioplankton nutrient limitation short-term variations of the competitive relationship for and their effects on bacterial community composition in the phosphorus. Limnol Oceanogr 49: 1063–1072. NW Mediterranean Sea. Aquat Microb Ecol 44: 241–252. Tanaka, T., Zohary, T., Krom, M.D., Law, C.S., Pitta, P., Pitta, P., Stambler, N., Tanaka, T., Zohary, T., Tselepides, A., Psarra, S., et al. (2007) Microbial community structure and and Rassoulzadegan, F. (2005) Biological response to P function in the Levantine Basin of the eastern Mediterra- addition in the Eastern Mediterranean Sea. The microbial nean. Deep Sea Res A 54: 1721–1743. race against time. Deep Sea Res Part II Top Stud Ocean- Tanaka, T., Thingstad, T.F., Christaki, U., Colombet, J., ogr 52: 2961–2974. Cornet-Barthaux, V., Courties, C., et al. (2011) Lack of Polz, M.F., Hunt, D.E., Preheim, S.P., and Weinreich, D.M. P-limitation of phytoplankton and heterotrophic prokary- (2006) Patterns and mechanisms of genetic and pheno- otes in surface waters of three anticyclonic eddies in the typic differentiation in marine microbes. Philos Trans R Soc stratified Mediterranean Sea. Biogeosciences 8: 525–538. Lond B Biol Sci 361: 2009–2021. Teira, E., Martínez-García, S., Lønborg, C., and Álvarez- Pulido-Villena, E., Rérolle, V., and Guieu, C. (2010) Transient Salgado, X.A. (2009) Growth rates of different phylogenetic fertilizing effect of dust in P-deficient LNLC surface ocean. bacterioplankton groups in a coastal upwelling system. Geophys Res Lett 37: L01603. Environ Microbiol Rep 1: 545–554. Rivkin, R.B., and Anderson, M.R. (1997) Inorganic nutrient Temperton, B., Gilbert, J.A., Quinn, J.P., McGrath, J.W., and limitation of oceanic bacterioplankton. Limnol Oceanogr Friedberg, I. (2011) Novel analysis of oceanic surface 42: 730–740. water metagenomes suggests importance of polyphos- Salcher, M.M., Pernthaler, J., and Posch, T. (2011) Seasonal phate metabolism in oligotrophic environments. PLoS ONE bloom dynamics and ecophysiology of the freshwater sister 6: e16499. clade of SAR11 bacteria ‘that rule the waves’ (LD12). ISME Thingstad, T.F., Skjoldal, E.F., and Bohne, R.A. (1993) Phos- J 5: 1242–1252. phorus cycling and algal-bacterial competition in Sandsf- Sebastian, M., and Ammerman, J.W. (2009) The alkaline jord, western Norway. Mar Ecol Prog Ser 99: 239–239. phosphatase PhoX is more widely distributed in marine Thingstad, T.F., Dolan, J.R., and Fuhrman, J.A. (1996) Loss bacteria than the classical PhoA. ISME J 3: 563–572. rate of an oligotrophic bacterial assemblage as measured Sebastian, M., and Ammerman, J.W. (2011) Role of the phos- by 3H-thymidine and 32P04: good agreement and near- phatase PhoX in the phosphorus metabolism of the marine balance with production. Aquat Microb Ecol 10: 29–36. bacterium Ruegeria pomeroyi DSS-3. Environ Microbiol Thingstad, T.F., Zweifel, U.L., and Rassoulzadegan, F. (1998) Rep 3: 535–542. DOI:10.1111/j.1758-2229.2011.00253.x. P limitation of heterotrophic bacteria and phytoplankton in Seeger, M., and Jerez, C.A. (1993) Phosphate-starvation the northwest Mediterranean. Limnol Oceanogr 43: 88–94. induced changes in Thiobacillus ferrooxidans. FEMS Thingstad, T.F., Krom, M.D., Mantoura, R.F.C., Flaten, G.A.F., Microbiol Lett 108: 35–41. Groom, S., Herut, B., et al. (2005) Nature of phosphorus Sintes, E., and Herndl, G.J. (2006) Quantifying substrate limitation in the ultraoligotrophic eastern Mediterranean. uptake by individual cells of marine bacterioplankton by Science 309: 1068–1071. catalyzed reporter deposition fluorescence in situ hybrid- Tijssen, J.P., Beekes, H.W., and Van, S.J. (1982) Localization ization combined with microautoradiography. Appl Environ of polyphosphates in Saccharomyces fragilis, as revealed Microbiol 72: 7022–7028. by 4′,6-diamidino-2-phenylindole fluorescence. Biochim Strickland, J.D., and Parsons, T.R. (1972) A Practical Hand- Biophys Acta 721: 394–398. book of Seawater Analysis, Bulletin 167. Ottawa, ON, Vadstein, O., and Olsen, Y. (1989) Chemical compositon and Canada: Fisheries Research Board of Canada, p. 310. phosphate uptake kinetics of limnetic bacterial communi- Suttle, C.A., Stockner, J.G., Shortreed, K.S., and Harrison, ties cultured in chemostats under phosphorus limitation. P.J. (1988) Time-courses of size-fractionated phosphate Limnol Oceanogr 34: 939–946. uptake: are larger cells better competitors for pulses of Van Mooy, B.A.S., and Devol, A.H. (2008) Assessing nutrient phosphate than smaller cells? Oecologia 74: 571–576. limitation of Prochlorococcus in the North Pacific subtropi- Suttle, C.A., Fuhrman, J.A., and Capone, D.G. (1990) Rapid cal gyre by using an RNA capture method. Limnol Ocean- ammonium cycling and concentration-dependent partition- ogr 53: 78–88.

Van Mooy, B.A.S., Rocap, G., Fredricks, H.F., Evans, C.T., Fig. S1. Abundance of cells within each probe-identified and Devol, A.H. (2006) Sulfolipids dramatically decrease group actively incorporating 33P-Pi, AT33P and 3H-leucine in phosphorus demand by picocyanobacteria in oligotrophic the control (left panel) and P-amended mesocosms (right marine environments. Proc Natl Acad Sci USA 103: 8607. panel). Van Mooy, B.A.S., Fredricks, H.F., Pedler, B.E., Dyhrman, Fig. S2. Phosphate (Pi) concentration estimated by conven- S.T., Karl, D.M., Koblízek, M., et al (2009) Phytoplankton in tional methods (Strickland and Parsons, 1972) in the control the ocean use non-phosphorus lipids in response to phos- (open symbols) and P-amended mesocosms (filled symbols). phorus scarcity. Nature 458: 69–72. Pi concentration in the P-amended mesocosms immediately Van Wambeke, F., Christaki, U., Giannakourou, A., Moutin, after Pi addition was not measured, so the theoretical con- T., and Souvemerzoglou, K. (2002) Longitudinal and verti- centration (sum of existing Pi + 100 nM added Pi) is plotted cal trends of bacterial limitation by phosphorus and carbon instead. The 33P-Pi added ranged between 25 and 108 pM. in the Mediterranean Sea. Microb Ecol 43: 119–133. Fig. S3. Epifluorescence microscopy images of (A) DAPI- Vaulot, D., LeBot, N., Marie, D., and Fukai, E. (1996) Effect of stained cells showing the presence of polyphosphates phosphorus on the Synechococcus cell cycle in surface (yellow granules) on day 3 of the experiment in the Mediterranean waters during summer. Appl Environ Micro- P-amended mesocosm; (B) same field of view showing autof- biol 62: 2527–2533. luorescence of cyanobacterial cells. Note that heterotrophic Vila-Costa, M., Pinhassi, J., Alonso, C., Pernthaler, J., bacterial cells also contain polyphosphate granules. No poly- and Simó, R. (2007) An annual cycle of phosphate granules were observed in the control mesocosm, dimethylsulfoniopropionate-sulfur and leucine assimilating where microbial cells where limited by phosphorus bacterioplankton in the coastal NW Mediterranean. Environ availability. Microbiol 9: 2451–2463. Fig. S4. Morphology of NOR5/OM60 cells on day 4 of the Wanner, B.L. (1996) Phosphorus assimilation and control of experiment in (A) the control mesocosm, (B) P-amended the phosphate regulon. In Escherichia and Salmonella: mesocosm. NOR5/OM60 cells in the control mesocosm are Cellular and Molecular Biology. Neidhardt, F.C. (ed.). elongated probably as a response to phosphorus stress. Four Washington, DC, USA: ASM Press, pp. 1357–1381. days after P addition NOR5/OM60 cells had decreased con- West, N.J., Schonhuber, W.A., Fuller, N.J., Amann, R.I., siderably in size. This phenomenon was not observed with Rippka, R., Post, A.F., and Scanlan, D.J. (2001) Closely other Gammaproteobacteria like Alteromonadales. Other related Prochlorococcus genotypes show remarkably dif- elongated non-NOR5/OM60 cells can as well be seen in the ferent depth distributions in two oceanic regions as control mesocosms, but we were not able to identify them revealed by in situ hybridization using 16S rRNA-targeted with any of the probes used in this study. Abundance of the oligonucleotides. Microbiology 147: 1731. NOR5/OM60 clade was determined with the probes NOR5- Yokokawa, T., and Nagata, T. (2005) Growth and grazing 730, as described elsewhere (Ferrera et al., 2011). mortality rates of phylogenetic groups of bacterioplankton Fig. S5. Microautoradiogram showing uptake of 33P- in coastal marine environments. Appl Environ Microbiol 71: phosphate by Gammaproteobacteria cells in the control 6799–6807. mesocosms. Blue dots represent DAPI-stained bacteria. Zohary, T., and Robarts, R.D. (1998) Experimental study of Green dots represent gammaproteobacterial cells hybridized microbial P limitation in the eastern Mediterranean. Limnol with the CARD-FISH Gam42a probe. Black spots are silver Oceanogr 43: 387–395. grains deposited in the photographic emulsion. Cells were Zubkov, M.V., Mary, I., Woodward, E.M., Warwick, P.E., scored as active (MAR+) when they were in contact with at Fuchs, B.M., Scanlan, D.J., and Burkill, P.H. (2007) Micro- least one silver grain. bial control of phosphate in the nutrient-depleted North Table S1. Presence of homologues to the ppk gene in Atlantic subtropical gyre. Environ Microbiol 9: 2079–2089. genomes of cultured marine bacteria that belong to the phy- Zweifel, U.L., and Hagstrom, A. (1995) Total counts of marine logenetic groups investigated in this study. bacteria include a large fraction of non-nucleoid-containing Please note: Wiley-Blackwell are not responsible for the bacteria (ghosts). Appl Environ Microbiol 61: 2180. content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) Supporting information should be directed to the corresponding author for the article.

Additional Supporting Information may be found in the online version of this article:

Figure S1. Abundance of cells within each probe-identified group actively incorporating 33P-Pi, AT33P and 3H-leucine in the control (left panel) and P-amended mesocosms (right panel).

0.12

0.08 M) µ

Pi ( 0.04

0 02468 Day Figure S2. Phosphate (Pi) concentration estimated by conventional methods (Strickland and Parsons, 1972) in the control (open symbols) and P-amended mesocosms (filled symbols). Pi concentration in the P-amended mesocosms immediately after Pi addition was not measured, so the theoretical concentration (sum of existing Pi + 100 nM added Pi) is plotted instead. The 33P-Pi added ranged between 25 and 108 pM.

20 µm

Figure S3. Epifluorescence microscopy images of a) DAPI-stained cells showing the presence of polyphosphates (yellow granules) on day 3 of the experiment in the P-amended mesocosm; b) same field of view showing autofluorescence of cyanobacterial cells. Note that heterotrophic bacterial cells also contain polyphosphate granules. No polyphosphate granules were observed in the control mesocosm, where microbial cells where limited by phosphorus availability. a)

Figure S4. Morphology of NOR5/OM60 cells on day 4 of the experiment in a) the control mesocosm, b) P-amended mesocosm. NOR5/OM60 cells in the control mesocosm are elongated probably as a response to phosphorus stress. Four days after P addition NOR5/0M60 cells had decreased considerably in size. This phenomenon was not observed with other Gammaproteobacteria like Alteromonadales. Other elongated non-NOR5/OM60 cells can as well be seen in the control mesocosms, but we were not be able to identify them with any of the probes used in this study. Abundance of the NOR5/OM60 clade was determined with the probes NOR5-730, as described elsewhere (Ferrera et al., 2011). Figure S5. Microautoradiogram showing uptake of 33P-phosphate by Gammaproteobacteria cells in the control mesocosms. Blue dots represent DAPI-stained bacteria. Green dots represent gammaproteobacterial cells hybridized with the CARD-FISH Gam42a probe. Black spots are silver grains deposited in the photographic emulsion. Cells were scored as active (MAR+) when they were in contact with at least one silver grain. Table S1. Presence of homologs to the ppk gene in genomes of cultured marine bacteria that belong to the phylogenetic groups investigated in this study.

Phylogenetic group # genomes considered* % genomes with ppk gene**

Gammaproteobacteria 60 87

Bacteroidetes 19 84

Rhodobacteraceae 38 76

Synechococcus 9 100

SAR11 5 20

* Marine genomes were those in the CAMERA database (http://camera.calit2.net/microgenome) ** ppk was searched for by hits of the predicted peptides to the corresponding protein family (PF02503)