Occurrence of chlorophyll allomers during virus‐induced mortality and population decline in the ubiquitous picoeukaryote Ostreococcus tauri

During viral infection and growth limitation of the picoeukaryote Ostreococcus tauri, we examined the relationship between membrane permeability, oxidative stress and chlorophyll allomers (oxidation products). Chlorophyll allomers were measured in batch-cultures of O. tauri in parallel with maximum quantum efficiency of photosystem II photochemistry (Fv /Fm ), carotenoids, and reactive oxygen species and membrane permeability using fluorescent probes (CM-H2 DCFDA and SYTOX-Green). Viral infection led to mass cell lysis of the O. tauri cells within 48 h. The concentration of the allomer hydroxychlorophyll a peaked with a 16-fold increase (relative to chlorophyll-a) just after the major lysis event. In contrast, cell death due to growth limitation resulted in a twofold increase in allomer production, relative to chl-a. Two allomers were detected solely in association with O. tauri debris after viral lysis, and unlike other allomers were not observed before viral lysis, or during cell death due to growth limitation. Conversely, the component chl-aP276 was found in the highest concentrations relative to chl-a, in exponentially growing O. tauri. The components described have potential as indicators of mode of phytoplankton mortality, and of population growth.

4 Seaton et al., 1995;Juneau et al., 2003). Hence the effect of viral infection on host photosynthetic capacity is variable. Pigment transformations during viral lysis of Emiliania huxleyi have been described (Llewellyn et al., 2007;Bale et al., 2013), and there is some evidence for production of chlorophyll allomers (Bale et al., 2013).
These studies however, lack concurrent physiological measurements that describe the average phytoplankton population cell state during virus infection and lysis.
To assess the ubiquity of allomer formation and its potential co-occurrence with loss of algal cell function, it is important to expand the study of chlorophyll transformation during viral lysis beyond one genus.
The cosmopolitan picoeukaryote Ostreococcus tauri was chosen for this study as it is an environmentally relevant primary producer for which a lytic virus has previously been isolated (thus, a model host:virus system was available, Derelle et al, 2008). Typically, picoeukaryotes are abundant in the world's oceans, have a relatively high rate of carbon fixation (Li, 1994;Worden et al., 2004) and are particularly important in the open ocean (Grob et al., 2011). Furthermore, the eukaryotic picoplankton genus Ostreococcus has a global distribution (coastal seas, the oligotrophic North Atlantic, the Mediterranean, the Indian and Pacific Oceans (Worden et al., 2004;Zhu et al., 2005;Countway and Caron, 2006), and a wide depth profile (surface waters to 120 m depth, Worden et al., 2004). Here we used a well characterised strain of Ostrecoccus tauri (OTH95 RCC745), which has undergone complete genomic sequencing (Derelle et al., 2006;Robbens et al., 2007), to analyse the relationship between cell demise and chlorophyll allomers.
This study describes the formation of chlorophyll allomers in Ostreococcus tauri during two mortality pathways: viral infection and growth limitation, with particular focus on the period directly before, and at the onset of, population decline. We performed concurrent measurements of O. tauri growth, OtV5 virus abundance, membrane permeability, cellular un-scavenged reactive oxygen species, chlorophyll a (chl-a), chlorophyll b (chl-b) and their allomers, the chl-a precursor (chl-a p276 ), maximum quantum efficiency of photosystem II photochemistry (F v /F m, ), and the carotenoid components of the non-photochemical quenching xanthophyll cycle, to describe in detail the timing of allomer formation in relation to population physiological state.

Virus-infected cultures; physiological indicators-
The virus OtV5, which infects O. tauri, was added to triplicate O. tauri cultures after 3 days of growth. The observed cycle of infection was consistent with previous work by Derelle et al. (2008). Infected cultures began to decline between 8 and 24 hours post-infection (hpi, days 3.3 and 4, Fig. 1A). At this time the O. tauri population density and F v /F m decreased, while simultaneously This article is protected by copyright. All rights reserved.
SYTOX-Green staining for membrane permeability, increased (Figs. 1A-C). By 32 hpi (day 4.3), 93.4±2.0% (mean±SD) of the O. tauri population had lysed, and of the remaining cells, 52.5±4.5% stained positively with SYTOX (Fig. 1C). The cytograms of O. tauri (Fig. 2) show cellular material which have retained a degree of red (chlorophyll-type) fluorescence (in the lower left hand corner). For flasks inoculated with OtV5, these cell fragments, labelled as debris, show maxima on days 5 and 5.3, just after the main lysis event between days 4 and 4.3 (Fig. 1A). This debris may have also contributed to the increased F v /F m values. In the control flasks, the relative proportion of cell debris remained relatively constant over the same period.
After viral inoculation CM-H 2 DCFDA staining for detection of reactive oxygen species (ROS) was significantly elevated. The percentage of cells stained CM-H 2 DCFDA-positive increased 0.6-fold relative to the noninoculated control flasks 24 hpi (Fig. 1D Fig. 2 ), as identified previously in this O. tauri strain (Thomas et al., 2011), with the specific growth rate (µ) of 0.07±0.01 h -1 . This residual population was potentially resistant to infection by OTV5, and continued growing exponentially (µ≥0.69) for at least 5 days. This residual population may account for the recovering photosynthetic capacity evident from F v /F m values (Fig. 1B).  This article is protected by copyright. All rights reserved.
6 Virus-infected cultures; chlorophylls, allomers, and chl-a precursor-For chemical assignment of components, see below. All chlorophyll allomers reached their maximum concentration relative to their parent chlorophylls on day 5 ( Fig. 3B-G), coinciding with the maximum proportion of cell debris observed in the cytograms (Fig. 2). The ratio of hydroxychlorophyll b (HO-chl-b) increased relative to chl-b above levels of the control flasks 24 hpi (day 4, 0.076±0.003, mean±SE, 0.9-fold increase), and reached a maximum of 0.42±0.04 at 48 hpi (day 5, Fig. 3B, 9.9-fold increase. This timing of chl-b allomer production was matched by the dominant chl-a allomer produced during OtV5-infection, hydroxychlorophyll a (HO-chl-a) and its epimer. These HO-chla allomers, relative to chl-a, increased above levels of the control flasks 24 hpi, and reached their maxima of 0.110±0.008 and 0.08±0.009 respectively (equal to 26-fold and 78-fold increases) at 48 hpi (day 5, Figs. 3F&3G), when 99.0±2.8% of the O. tauri population had lysed. At this time, non-infected cultures maintained low levels of HO-chl-a of 0.004±0.000.
After the major lysis event between 32 and 48 hpi, pigment samples contained methoxychlorophyll alike (MeO-chl-a-like) and hydroxychlorophyll a-like (HO-chl-a-like) allomers, with average allomer to chl-a ratios of 0.015±0.003 and 0.011±0.002 respectively (n=6, Figs. 3D&3E). Unlike the other allomers, these components were not detected in the cultures before viral lysis, nor were they detected in the control cultures.
The timing of the detection of methoxychlorophyll a-like and hydroxychlorophyll a-like coincided with high levels cell debris that showed high red fluorescence in the cytograms (days 4.3 and 5, Fig. 2).
Both chlorophyll b, and the chlorophyll a precursor chl-a P276 showed similar profiles relative to chl-a in the infected cultures (Fig. 3A&H). Both components showed small relative increases on days 4.3 and 5, and maximum abundance relative to chl-a on day 5.3 ( Fig. 3A&H), coincident with the regrowth of the residual O. tauri population (Fig. 2). Chl-a P276 is a precursor in chlorophyll a biosynthesis, and chlorophyll b is used as a light harvesting chlorophyll in the peripheral antenna. Their small increases relative to chlorophyll a on day 4.3, during the main lysis event, may be due to their location in the peripheral antenna, whereas chlorophyll a is located in the core antenna and reaction centre as well as the peripheral antenna (Scheer, 2006) and therefore could be subject to a different rate of destruction during cell lysis.

Virus-infected cultures; carotenoids-
In O. tauri (as well as higher plants and green-algae) the xanthophyll cycle consists of the conversion of violaxanthin to zeaxanthin via antheraxanthin. During viral infection and lysis, zeaxanthin increased relative to the total abundance of the xanthophyll cycle components, 24 hpi to 48 hpi (day 4 to 5, Fig. 3K). This coincided with a decrease in violaxanthin relative to the xanthophyll cycle pigments over the same period (Fig. 3I), indicating active conversion of violaxanthin to zeaxanthin during and after viral lysis. β-carotene, neoxanthin and dihydrolutein all decreased relative to chl-a between 24 hpi and 7 48 hpi (day 4 and day 5, Figs. 3L-N), and β-carotene and dihydrolutein then increased relative to chl-a at day 5.3, coincident with population regrowth. Assuming cells were using nutrients in the Redfield ratio; by calculation, phosphate was potentially the initial nutrient to become limiting, as F/2 media has an N:P ratio of 24:1. The concentration of chl-a increased  Growth limited cultures; chlorophylls, allomers, and chl-a precursor-All allomers showed increases relative to their parent chlorophyll during the death phase, between days 20 and 29 (Figs. 5B-E). The HO-chl-b to chl-b ratio increased during the death phase to a maximum of 0.075±0.013 on day 29 (Figs. 5B&C, a 1.1-fold increase relative to day 5). HO-chl-a and HO-chl-a` reached maxima of 0.0074±0.0005 and 0.0072±0.0007 (increases of 2-fold and 7-fold respectively from day 5, Figs. 5D&E). These increases were small compared to maximum levels observed after viral infection.
Chlorophyll b also increased during the death phase ( Fig. 5A), and also reached a lower maximum than that observed after viral infection.
The component chl-a P276 was found in the highest concentration, relative to chl-a, in exponentially growing O. tauri, (days 0 to 3, when µ≥0.69, Fig. 5F), consistent with its assignment as a precursor to chl-a.
Like chlorophyll b, it also increased relative to chl-a between days 20 and 29.  This article is protected by copyright. All rights reserved.

Growth-limited cultures; carotenoids-
Photosynthetic pigments; chemical assignment-Pigment extracts from OtV5-infected Ostreococcus tauri cultures and non-infected cultures were analysed for chlorophyll alteration products (Fig. 6). Assignments of chlorophyll a and chlorophyll b were made by comparison of their retention times to standards, comparison of UV/vis absorption spectra (Table 1) and major ions observed during LC/MS n analysis (Table 2), to published data (Airs et al., 2001;Bale et al., 2011;Franklin et al., 2012). The extracts gave rise to several peaks which eluted in the region expected for chlorophyll allomers , prior to chlorophyll a (peak IX, Fig.   6), with UV/vis absorption spectra (Table 1) consistent with chl-a allomers (Franklin et al., 2012). An array of peaks exhibiting chlorin-like UV/vis spectra were also detected eluting prior to chl-b (Peak V, Fig. 6).
Peak I was present only in pigment extracts from the OtV5-infected cultures; it was not detected at any time in the non-infected cultures. During LC/MS/MS with post column addition of acid (Airs et al., 2001), peak I gave rise to precursor and daughter ions in the same pattern as methoxychlorophyll a (Table 2; Franklin et al., 2012), but with all components exhibiting an increased mass of 4 Da. The component is therefore assigned as methoxychlorophyll a-like allomer.
Peak II gave rise to a UV/vis absorption spectrum, as expected for a chl-a allomer but with an additional absorption band at 481 nm (Table 1). From LC/MS analysis with post-column acidification, peak II gave rise to one major ion at m/z 887 [M+H-Mg] + . The MS 2 spectra contained a major ion at m/z 593 (Table 2) Da. It is possible therefore, that peak II was a chlorin esterified by phytol, but in such a stereochemical configuration to promote fragmentation via the loss of 294 Da instead of 278 Da. Notably, the MS 2 spectrum shows a less abundant ion at m/z 609, arising from the loss of 278 Da from the parent ion, providing support for the assignment of phytol as the esterifying alcohol, and indicating that both fragmentation mechanisms were taking place. It is important to note that the relative intensity of the ions in the MS 2 spectrum was low, and therefore the spectrum must be interpreted cautiously. Other fragmentations indicate the presence of a CO 2 Me group (60 Da loss to produce an ion at m/z 533), and HO-substituent (18 Da loss to produce an ion at m/z 575), consistent with a hydroxychlorophyll a (HO-chl-a) structure. The component is therefore tentatively assigned as a hydroxychlorophyll a-type structure with unusual stereochemistry (HO-chl-a-like).
Peaks III and IV eluted in the region expected for chlorophyll b allomers, and were assigned as hydroxychlorophyll b and it's epimer by comparison to published MS/MS data (Hyvärinen and Hynninen, This article is protected by copyright. All rights reserved. 1999). Similarly, components VII and VIII were assigned as hydroxychlorophyll a and it's epimer based on MS n data (Table 2; Walker et al 2002).
Peak VI had a UV/vis absorption spectrum similar to chl-a (Airs et al., 2001) (Table 1). From LC/MS/MS with post-column addition of acid, peak VI gave rise to a major ion at m/z 869 corresponding to [M+H-Mg] + (Table 2). On resonance induced fragmentation, the ion at m/z 869, gave rise to m/z 593 in the MS 2 spectrum, equating to a loss of 276 Da indicating an extra double bond in the phytyl chain. The component at peak VI is therefore identified as a biosynthetic precursor to chlorophyll a (Rüdiger, 2006), chlorophyll a P276 (chl-a P276 ). This component has been detected previously in Thalassiosira pseudonana, Emiliania huxleyi (Franklin et al., 2012) and Pavlova gyrans (Bale et al., 2011).  Table 1.

Discussion
Chlorophyll a allomers were produced in Ostreococcus tauri cultures 24 h post-infection (hpi) with OtV5, with an increase in the ratio of total chl-a allomers to chl-a of approximately 28-fold by 48 hpi (day 5).
Allomers were also produced during O. tauri growth-limitation but in smaller amounts, with approximately a 2fold increase in the ratio of total allomers to chl-a, 14 d after the onset of population decline. In flasks inoculated with OtV5, and in growth-limited flasks, the maximum ratio of allomers to chl-a coincided with the maximum proportion of cell debris detected in cytograms (day 5 and day 29 respectively, Fig. 2). Although the proportion of cell debris retained in the cell pellet during pigment sample collection via centrifugation is unknown, it can be assumed to be consistent between days 4.3 and 5.3, as the size spectrum of the debris, indicated by the forward scatter of the cytograms, did not change significantly during this period (Fig. 2). To be detected by flow cytometry (in this case), the cell debris must have exhibited red fluorescence, which arises from chlorophylltype structures. Notably the debris showed maximum fluorescence on days 4.3 and 5 in samples from the infected flasks, and day 29 in the growth-limited flask (Fig. 2), coincident with the maximum increase in This article is protected by copyright. All rights reserved.
chlorophyll allomers relative to parent chlorophylls. The cell debris is likely to comprise a "soup" of cellular components, including fragments of membrane and chloroplasts, and likely to be a prime source of reactive oxygen species, due to the presence of illuminated chlorophyll, and the disintegration of cellular machinery to prevent ROS formation and provide effective scavenging. Indeed, staining of the cell debris by CM-H 2 DCFDA was observed, but not quantified (data not shown). Therefore, we propose that the increase in chlorophyll allomers observed after viral lysis and growth-limitation were formed in the cell debris, rather than in unlysed cells. Two allomers have been found solely in association with O. tauri lysis by OtV5; a methoxychlorophyll alike allomer and a hydroxychlorophyll a-like allomer. These components were not detected prior to viral infection, or during growth limitation, and therefore may be specific to viral infection and lysis.

Measurement of ROS within the cells showed a response in OtV5-infected culturesa 7.3-fold
increase in the percentage of CM-H 2 DCFDA-positive cells, but this was due to the small residual population. In the growth-limited population cellular ROS increased as growth limitation progressed, however, a peak in CM-H 2 DCFDA-positive cells also occurred just prior to population decline (day 15). A peak in SYTOX-positive cells (with compromised membranes) occurred prior to this (day 12). Cell cycle arrest in a part of, or in the whole population during stationary phase may have allowed this build-up of ROS. A peak in ROS before population decline has also been detected in the diatom Thalassiosira oceanica (D. J. Steele, unpublished). The proportion of cells detected with compromised membranes during OtV5-infection was much greater (maximum 51.8%), although as the lysis process happens quickly, there was only a brief window when membranes were compromised but cells were still relatively intact. This level of SYTOX staining is comparable to that of natural picoplankton (Veldhuis et al., 2001;Baudoux et al., 2008), and small eukaryote (Veldhuis et al., 1997) populations in mixed natural assemblages, where between 3% and 75% of the cells stained SYTOX-positive. The xanthophyll cycle dissipates excess excitation energy which would otherwise lead to the formation of destructive singlet oxygen ( 1 O 2 ). The product of this xanthophyll cycle, zeaxanthin, deactivates excited singlet chlorophyll (Niyogi et al., 1998) and is also an antioxidant in the lipid phase of the thylakoid membrane (Havaux et al., 2007). The increase of zeaxanthin 24 hpi indicates that the xanthophyll cycle, a nonphotochemical quenching (NPQ) process, increased in rate, during decreased photochemical quenching, reflected by decreased F v /F m 24 hpi (Fig 1B). β-carotene is a precursor to the xanthophylls, and is also a ROS scavenger (Fiedor et al., 2001(Fiedor et al., , 2005, specifically of 1 O 2 (Telfer, 2002). β-carotene decreased relative to chl-a 24 hpi. As β-carotene is situated in the reaction centres (Young, 1993;Fiedor et al., 2001Fiedor et al., , 2005Telfer, 2002), it also may be more prone to photooxidation (Llewellyn et al., 2007). Therefore its decrease may be due to 1 O 2 scavenging and photodegredation of the photosystem by ROS. This, along with the increased rate of nonphotochemical quenching, prevented the build up of cellular ROS, hence chlorophyll allomers remained low until after the major lysis events, due to OtV5-infection, and growth-limitation.
This capacity of O. tauri to buffer ROS accounts for the minimal chlorophyll allomer production of cultures before cell lysis. Increased formation of cellular HO-chl-a during cell lysis by viral action has been observed previously in E. huxleyi CCMP 1516 (Bale et al., 2013); an increase of approx. 4-fold in absolute mass per cell, where the ratio of HO-chl-a to chl-a increased from approx. 0.004 at the beginning of population decline (4 days post-infection), increasing to approx. 0.02 after the loss of 94% of the population (16 days postinfection). Notably this also represents cellular HO-chl-a plus HO-chl-a in cell debris, which would have been collected by filtration. Here, hydroxychlorophyll a production in growth-limited O. tauri cultures, was consistent with previous observations during senescence of Isochrysis galbana (Prymnesiophyte, Bale et al., 2011) and Thalassiosira pseudonana (Diatom, Franklin et al., 2012). Chlorophyll a allomers have been observed not to increase during population decline of E. huxleyi (Franklin et al., 2012), however it was hypothesised by Franklin et al. (2012) that the E. huxleyi population, under N-limitation, may have undergone a physiological change, possibly in preparation for meiosis, rather than undergone mortality (Franklin et al., 2012). Also E. huxleyi can maintain PSII repair through periods of nitrogen depletion (Loebl et al., 2010), limiting PSII photo-inactivation and ROS production (Holt et al., 2004;Key et al., 2010), which may decrease the formation of the oxidation product HO-chl-a. Rates of PSII repair are lower in O. tauri and lower still in T.
pseudonana (Six et al., 2009;Key et al., 2010), which is reflected in the higher ratios of HO-chl-a to chl-a in T.
This article is protected by copyright. All rights reserved.
To date, in the published studies of chlorophyll allomers in phytoplankton cultures, HO-chl-a has been detected ubiquitously across taxa (Bale et al., 2011(Bale et al., , 2013Franklin et al., 2012), but production during cell death varies with species and mode of death. Therefore the total HO-chl-a in a mixed, natural population will vary according to community composition and the modes of mortality taking place. In natural phytoplankton populations the main causes of mortality are grazing (Walsh, 1983), viral lysis (Suttle et al., 1990;Brussaard, 2004) and senescence due to environmental factors (Walsh, 1983). The contribution of each of these factors varies greatly with season and region. HO-chl-a has been detected in various natural waters (Walker and Keely, 2004;Steele, 2014;Bale et al., 2015;Steele et al., 2015). Increased concentrations of HO-chl-a have been reported in sinking particles i.e. during a diatom bloom terminated by nutrient-limitation (Bale et al., 2015), and after declines of phytoplankton blooms, I.e. a Phaeocystis spp. crash likely due to nitrate limitation; a Guinardia delicatula decline due to parasitic infection; and a Chaetoceros socialis reduction due to an unknown cause (Steele et al., 2015). Unfortunately, none of the reports of HO-chl-a in natural waters (to date) have included viral enumeration. In O. tauri, the total amount of allomers (relative to chl-a) varied depending on the mode of death, from 13.8±2.1% during viral-infection to 3.1±0.8% during growth limitation. Hence when a bulk measurement of total chlorophyll a is taken from a population dominated by O. tauri, the contribution to allomers of this total will be greatly increased if viral-infection is occurring.
Here, the observed cycle of infection and lysis of O. tauri by OtV5 within 2 d of the maximum population density, with rates of population decline from 0.06 d -1 (24 hpi) to 3.69 d -1 (48 hpi), was comparable to the decline of two consecutive O. tauri-like (picoalga resembling O. tauri) algal blooms in West Neck Bay (Maine). The O. tauri-like algal cells contained virus-like particles, but were also grazed by heterotrophic nanoflagellates (O'Kelly et al., 2003). The blooms had maximum population densities of 2 × 10 5 and 5 × 10 5 cells mL -1 , and collapsed within 4 d and 6 d respectively. Following the bloom declines, the population of O.
tauri-like cells began to regrow. The longer decline period was probably due to a lower maximum population density, lower density of viruses, and competition between viruses and grazers. This time-frame of bloom termination by a combination of viral lysis and grazing has been recorded for other phytoplankton taxa, with similar lysis rates e.g. Emiliania huxleyi (Prymnesiophyte, Bratbak et al., 1993Bratbak et al., , 1995Castberg et al., 2001;Vardi et al., 2012); Micromonas spp (Prasinophyte, Evans et al., 2003); and Phaeocystis globosa (Prymnesiophyte, Baudoux et al., 2006). If the different rates of cell mortality caused by grazing and viral lysis changed, favouring viral lysis, our study suggests that allomer production would increase. For example, if the O.
tauri-like blooms observed by O' Kelly et al. (2003) had declined with the maximum rate observed here (3.69 d -production of chl-a allomers (HO-chl-a + HO-chl-a' + HO-chl-a-like + MeO-chl-a-like) at a ratio of approx.
0.2 with chl-a. If a bulk measurement of chl-a were taken at this time, the total chl-a measurement coming from the O. tauri population, would consist of 20% allomers.
Given previous (Bale et al., 2011;Franklin et al., 2012) and present observations; elevated hydroxychlorophyll a levels should be considered an indicator of phytoplankton death. The occurrence of the methoxychlorophyll-a-like and hydroxychlorophyll a-like allomers during this study (present after viral lysis of O. tauri), provides evidence that these allomers occur during termination by viral lysis. The ratio of allomers and chl-a precursor can be used to determine if the dominant population at a particular location is growing or declining -providing physiological context.

Experimental Procedures
Cultures and viral infection-Unialgal triplicate cultures of Ostreococcus tauri (OTH95 RCC745) were grown in 5 L conical flasks containing 1 L artificial seawater base media ESAW (Harrison and Berges, 2005) enriched with F/2 nutrients (Guillard and Ryther, 1962). Bacterial contamination was minimised by regular subculturing (every 3 days) prior to the study. Illumination from cool white fluorescent tubes was provided at 100 - Algal staining and enumeration-SYTOX-Green (Invitrogen S7020) used to measure changes in membrane permeability (Veldhuis et al., 1997); was applied at 0.5 µmol L -1 final concentration with incubation in the dark at 20 o C for 15 min. CM-H 2 DCFDA (5-and 6-chloromethyl-2`,7`-dichlorodihydrofluorescein diacetate, Invitrogen C6827), used to measure the relative amount of reactive oxygen species within a cell, was applied at 5 µmol L -1 final concentration (dark, 20 o C for 30 min). Incubation conditions were optimised prior to This article is protected by copyright. All rights reserved.
the study using positive controls (Peperzak and Brussaard, 2011); for SYTOX, heat-killed cells (80 o C, 5 min) and for CM-H 2 DCFDA, hydrogen peroxide treated cells (10 µmol L -1 final concentration). Uptake of the stain was compared with unstained cells by flow cytometry, using an excitation laser of 488 nm, on a green (530/30 nm) vs. red fluorescence (670 nm) plot. O. tauri population density was also quantified by flow cytometry (Accuri C6, BD Biosciences): Milli-Q was used as sheath fluid and analysis was triggered on forward scatter and red fluorescence. A core size of 22 µm was used and the event rate was kept below 5000 events s -1 to avoid coincidence; when necessary samples were diluted with filtered media (0.2 µm). Flow rate was set to 66 µL min -1 and measured daily by uptake of Milli-Q over 5 min (by mass).  (Brussaard et al., 2000) and incubated for 10 min at 80 o C,and then 5 mins at room temperature. Samples were run on a FACScan flow cytometer (Becton Dickinson) triggered on green fluorescence and set to "low" flow (~20 μL min -1 ) for 1 min, with event rates between 100 and 500 cells s -

.
Photosynthetic pigments-The sample volume for photosynthetic pigments varied over the course of the study. For the OtV5 treatment flasks sample volumes were as follows: 50 mL on days 0, 1 and 2; 30 mL on day 3; 15 mL on days 3.3 and 4; 10 mL on day 4.3 and 7 mL thereafter. The same volumes were collected from the control flasks from day 0 to day 4, 15 mL was sampled from control flasks on day 4.3; and 50 mL from day 5 onwards until the termination of the study. The culture samples were centrifuged (at 5300 × g for 20 min at 8 o C), and the pellet was flash frozen in liquid nitrogen and stored at -80 o C until analysis. Exhaustive extraction of pigments used acetone (90% in Milli-Q) under dim light by sonication (at 40 W, Vibra Cell Probe; Sonics) for 40 s. The extract was clarified by centrifugation at 17000 × g (Thermo Scientific). A 200 µL aliquot of extract was mixed with 80 µL water in the autosampler, and 25 µL of this mix was injected onto the column. Reversedphase high performance liquid chromatography (HPLC) was carried out using an Accela system (Thermo Scientific) with photodiode array detector, controlled using ChromQuest software. Chromatography was performed using a Waters Symmetry 3.5 µm C 8 column (2.1 × 150 mm) with pre-column of the same phase.
Elution used a mobile phase gradient composed of methanol, acetonitrile, aqueous pyridine (0.25 mol L -1 ) and acetone (all HPLC grade) at a flow rate of 0.2 mL min -1 (Method B in Zapata et al., 2000). Assignment of chlorophyll allomers was carried out by LC/MS n using an Agilent 1200 HPLC with photodiode array detector, coupled via an atmospheric pressure chemical ionization (APCI) source to an Agilent 6330 ion trap mass This article is protected by copyright. All rights reserved.

17
spectrometer. HPLC conditions were as described above with instrument control and analysis performed using         Table 1.
This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved.