Impact of metabolic pathways and salinity on the hydrogen isotope ratios of haptophyte lipids

Hydrogen isotope ratios of biomarkers have been shown to reflect water isotope ratios, and in some cases correlate significantly with salinity. The δ2H-salinity relationship is best studied for long-chain alkenones, biomarkers for haptophyte 20 algae, and is known to be influenced by a number of different environmental parameters. It is not fully known why δ2H ratios of lipids retain a correlation to salinity, and whether this is a general feature for other lipids produced by haptophyte algae. Here, we analyzed δ2H ratios of three fatty acids, brassicasterol, long-chain C37 alkenones and phytol from three different haptophyte species grown over a range of salinities. Lipids synthesized in the cytosol, or relying on precursors of cytosolic origin, show a correlation between their δ2H ratios and salinity. In contrast, biosynthesis in the chloroplast, or utilizing 25 precursors created in the chloroplast, yields lipids that do not show a correlation between δ2H ratios and salinity. This leads to the conclusion that location of metabolism is the first-order control on the salinity signal retained in δ2H ratios of certain lipids. Additionally, we found that δ2H ratios of alkenones from batch cultures of the Group II haptophyte species Tisochrysis lutea correlate positively with temperature, contrary to findings from cultures of Group III haptophytes, but retain a similar response to nutrient availability in line with other Group III haptophytes. 30 Biogeosciences Discuss., https://doi.org/10.5194/bg-2019-147 Manuscript under review for journal Biogeosciences Discussion started: 29 April 2019 c © Author(s) 2019. CC BY 4.0 License.


Introduction
One of the best studied lipids for hydrogen isotope fractionation are long-chain alkenones, which are ethyl and methyl ketones with chain lengths of 35 -42 carbon atoms and two to four degrees of unsaturation, found globally in lakes and oceans (De Leeuw et al., 1980;Volkman et al., 1980;Cranwell, 1985).Long-chain alkenones are synthesized by specific haptophyte algae from the order Isochrysidales, and are believed to be created from fatty acids via chain elongation and subsequent decarboxylation (Rontani et al., 2006).Haptophytes are genetically divided into three distinct groups: Groups I and II encompass haptophytes preferring lacustrine and coastal environments (although some Group II haptophytes can also be found in hypersaline lakes), and Group III comprises marine alkenone producers (Theroux et al., 2010).Hydrogen isotope ratios of C37 alkenones (δ 2 HC37) and fractionation factor (αC37) values both have a significant positive correlation to salinity in cultures of Group II and Group III haptophytes (Schouten et al., 2006;M'Boule et al., 2014;Chivall et al., 2014;Sachs et al., 2016;Weiss et al., 2017); Group I haptophytes are presently uncultured and thus nothing is known about how they fractionate hydrogen.δ 2 HC37 ratios have been applied to reconstruct salinity across different timescales and marine environments (Pahnke et al., 2007;Leduc et al., 2013;Kasper et al., 2014Kasper et al., , 2015;;Petrick et al., 2015;Simon et al., 2015;Weiss et al., in review).In addition to alkenones, a strong correlation between δ 2 H ratios and salinity has also been shown for fatty acids and sterols from haptophytes and other phytoplankton in laboratory cultures (e.g.Sachs and Kawka, 2015;Sachs et al., 2016;Maloney et al., 2016), and is also found for other biomarkers such as C17 n-alkane, diploptene, and phytene (see compilation in Sachse et al., 2012).
Despite the strong relationship noted between d 2 H ratios and salinity, the precise intracellular mechanisms controlling the salinity relationship are not well constrained.Furthermore, a number of extracellular factors have been shown to affect abiomarker and δ 2 Hbiomarker ratios in culture (Schouten et al., 2006;Wolhowe et al., 2009;Chivall et al., 2014;M'Boule et al., 2014;Sachs and Kawka, 2015;van der Meer et al., 2015;Sachs et al., 2016;Maloney et al., 2016;Weiss et al., 2017).For example, growth rate is shown to negatively correlate with δ 2 HC37 ratios and αC37 values in Group III haptophyte species and is thought to be a limiting factor for reconstructions of paleosalinity (Schouten et al., 2006;Wolhowe et al., 2009;Sachs and Kawka, 2015).
Additionally, the effect of temperature on δ 2 HC37 ratios and αC37 values in culture is not well constrained, with some studies showing an effect (Wolhowe et al., 2009), and others not (Schouten et al., 2006).These extracellular parameters have primarily been investigated for Group III marine haptophytes, with little known about Group II species.
Here, we investigate the relationship between salinity and δ 2 H ratios for a variety of haptophyte lipids to explore the impact of biosynthetic pathways.For this purpose, Group II and Group III haptophytes were grown at different salinities and δ 2 H ratios of long-chain alkenones, C14:0 -C18:1 fatty acids, brassicasterol, and phytol were measured.A secondary aim is to provide the first characterization of δ 2 HC37 ratios from the Group II haptophyte Tisochrysis lutea, which was grown in batch cultures to assess effects of temperature and nutrient concentrations.
Cultures were grown at a temperature of 15° C, and light intensity of 60 µmol photons m -2 s -1 supplied by cool white fluorescent light with a light:dark cycle of 16:8 h.Specific media conditions are outlined in Chivall et al. (2014 -R. lamellosa) andM'Boule et al. (2014 -E. huxleyi andI. galbana).Aliquots from the starter culture were transferred to fresh media of each salinity five times prior to the experiment in order to acclimate algae to the new conditions and remove any potential memory effects.Biomass was filtered onto pre-combusted GF/F filters during exponential growth phase when cell concentrations were above 1 x 10 6 cells / mL.Filters were freeze-dried, and organics were extracted ultrasonically using dichloromethane / methanol 2:1 (v/v) to obtain total lipid extracts (TLE).TLEs were split in two for E. huxleyi and I. galbana.One part of the TLE was separated using an aluminium oxide column into three fractions: apolar (eluted with three column volumes of 9:1 (v/v) hexane / DCM), alkenone (eluted with four column volumes of 1:1 (v/v) hexane / DCM) and polar (eluted with four column volumes of 1:1 (v/v) DCM / MeOH; containing phytol and brassicasterol).
To obtain clean fractions for phytol and brassicasterol analyses, an aliquot of polar fractions from the TLE was saponified using 1N KOH / MeOH (10 mL; 1 h reflux).After cooling, the solvent was neutralised with HCl / MeOH and transferred into a separating funnel with 3 mL of bidistilled water.Lipids were extracted from the aqueous solution into 3 × 3 mL DCM.The solvent was removed under vacuum and the extracts dried over Na2SO4.An aliquot of this saponified polar fraction was cleaned over an aluminium oxide column, eluting with 1:1 (v/v) DCM/MeOH.Phytol and brassicasterol, contained in the saponified polar fraction, were acetylated with 100 µL acetic anhydride (δ 2 H= -126 ‰) at 60° C for 30 min in the presence of a small of amount of Ag(CF3SO3) (Das and Chakraborty, 2011).After cooling, 2 mL of a saturated aqueous solution of NaHCO3 was added and the acetylated lipids extracted into 3 × 1 mL ethyl acetate.The acetylated lipids were then cleaned over an aluminium oxide column, using 1:1 (v/v) hexane / DCM.Fatty acid extraction was conducted as described by Heinzelmann et al. (2015) for both E. huxleyi and I. galbana.Fatty acids were not analyzed for R. lamellosa.

Batch cultures of Tisochrysis lutea
Two strains of T. lutea, CCMP463 and NIES-2590, were grown in artificial seawater with added nutrients at a salinity of 30, light intensity of 100 µmol photons m -2 s -1 and a range of temperatures from 10 to 35° C. Cells were harvested during late linear and early stationary phases.Alkenones were extracted as described in Nakamura et al. (2016).T. lutea strain CCAP 927/14 was grown in triplicate in filtered seawater under N-replete and N-limited conditions with salinity of 34 -35, temperature from 20-23° C, and light intensity of 180 -220 µmol photons m -2 s -1 supplied by cool white fluorescent light.
Cells were harvested after 4 and 10 days of growth.Neutral lipid fractions were extracted as described in da Costa et al. (2017), and ketone fractions were isolated over an aluminum oxide column using 1:1 (v/v) hexane:DCM.

Hydrogen isotope ratios
Hydrogen isotope ratios of growth water were measured prior to starting experiments and after filtration for E. huxleyi, I. galbana, and R. lamellosa.Culture water for the T. lutea nutrient experiment was measured prior to starting the experiments only, and no culture water was available for isotope analysis from the T. lutea temperature cultures.Water isotope ratios were measured following methods outlined in Weiss et al. (2017).Biomarkers were quantified using a gas-chromatograph coupled to a flame ionization detector (GC-FID) before measuring hydrogen isotope ratios on a Thermo Scientific Delta V GC/TC/irMS.For biomarkers from E. huxleyi, I. galbana and R. lamellosa, the GC was equipped with a 25 m CPSil 5 (Agilent, 25 m x 0.32 mm x 0.4 µm) GC column following M' Boule et al. (2014).Due to lack of baseline separation when using a 25 m CPSil 5 GC column, alkenone d 2 H ratios from these three species are the integrated C37:3 and C37:2 values (d 2 HC37).Alkenones from T. lutea were measured by GC/TC/irMS equipped with a RTX-200 60 m column (Restek, 60 m x 0.32 mm x 0.5 µm), allowing for determination of d 2 H ratios of individual alkenones.The GC temperature program is as follows: 70° C to 250° C at 18° C/min, 250° C to 320° C at 1.5° C/min, then kept at 320° C for 25 min.Helium was used as a carrier gas and the flow rate was 1.5 mL/min.The integrated d 2 HC37 ratios, determined by a second peak integration encompassing both the C37:3 and C37:2 alkenone peaks, are also reported for comparison with alkenones from E. huxleyi, I. galbana and R. lamellosa measured on the CPSil 5 GC column.Prior to running samples each day, the H3 + factor was measured and corrected for.Values for the H3 + factor were 2.765 ± 0.468 ppm nA -1 for analytical runs of R. lamellosa lipids, 2.752 ± 0.428 ppm nA -1 for I. galbana lipids, and 4.644 ± 0.648 ppm nA -1 for E. huxleyi lipids and 3.229 ± 0.261 ppm nA -1 for T. lutea lipids.An n-alkane standard, Mix B, supplied by A. Schimmelmann (Indiana University) was measured, and samples were only run once the average deviation from the offline determined value and standard deviation for the Mix B were both less than 5 ‰.H2 gas of known isotopic composition was measured at the beginning and end of each analytical run to further monitor machine stability.Squalane and C30 n-alkane were co-injected with each run as another control on machine accuracy.Values for squalene were -167 ± 4 ‰ and for C30 were -71 ± 4 ‰ for the entire sample set (n = 327).Standard deviations for d 2 H ratios represent the reproducibility between duplicate or triplicate analytical runs, and generally fall within the 3 ‰ precision window for the Thermo Scientific Delta V.

Results
Hydrogen isotope ratios and fractionation of six different lipids were measured and calculated, respectively, from extracts of batch cultures of four different haptophyte species (Supplementary Table 1).The δ 2 HC37 ratios from the Group III haptophyte E. huxleyi and the Group II haptophytes I. galbana and R. lamellosa were previously published by M'Boule et al. ( 2014) and Chivall et al. (2014) and the δ 2 H ratios of C14:0 -C18:1 fatty acids for I. galbana were previously published by Heinzelmann et al. (2015).Here we report the δ 2 HC37 ratios from the Group II haptophyte T. lutea cultivated at different salinities, temperature and nutrient conditions.Furthermore, we analyzed the d 2 H ratios of C14:0 -C18:1 fatty acids from E. huxleyi, and brassicasterol and phytol from E. huxleyi, I. galbana, and R. lamellosa from the same culture material as that of M'Boule al. ( 2014) and Chivall et al. (2014) (Supplementary Table 1).Fractionation factor α values were calculated for all compounds except for alkenones from the T. lutea temperature experiment because culture water was no longer available at the time of analyses.

Temperature experiment
Group II haptophyte T. lutea was grown over a temperature range of 15 -35° C. Combined δ 2 HC37 ratios ranged from -123 ‰ to -152 ‰, while isotope ratios of individual alkenones ranged from -111 ‰ to -153 ‰ for C37:3 and -117 ‰ to -160 ‰ for C37:2 (Table 1, Fig. 1).Growth was monitored by chlorophyll fluorescence, thus growth rates cannot be easily compared with rates calculated from daily cell counts as done for the other species.However, cells appeared to grow fastest at 35° C and slowest at 15° C (Nakamura et al., 2016).Similarly, alkenone concentrations were stable between 20° C and 30° C, and slightly lower at 35° C and 15° C (Nakamura et al., 2016).The hydrogen isotopic composition of the integrated C37 alkenones has a significant positive correlation with temperature (r = 0.80, n = 28, p < 0.001), as does the C37:3 (r = 0.75, p < 0.001), while the C37:2 only has a moderate positive correlation with temperature (r = 0.39, p < 0.05).

Nutrient experiment
To understand the impact of nutrients on hydrogen isotope ratios of alkenones from Group II haptophytes, T. lutea was grown under N-replete (nitrate concentration of 1.2 mM) and N-reduced (nitrate concentration of 0.6 mM) conditions.Cells were harvested in exponential and stationary growth phases (after 4 and 10 days of growth respectively).Growth rates for N-replete conditions were 0.21 div.d -1 , and N-reduced conditions were 0.14 div.d -1 .Under N-replete conditions, δ 2 HC37 ratios ranged from -123 ‰ to -134 ‰, and were more depleted than under N-reduced conditions by around 10 ‰, where δ 2 HC37 ratios ranged from -113 ‰ to -121 ‰ (Table 2; Fig. 2a).There is also a clear distinction between growth phases, with more depleted δ 2 HC37 ratios corresponding to exponential growth phase (Fig. 2a).Fractionation factor αC37 ranged from 0.860 to 0.880, with higher fractionation (lower αC37) for higher nutrient conditions (Fig 2b).Higher fractionation was also observed during exponential growth phase compared to stationary growth phase (Fig. 2b).Higher nutrient concentrations led to faster growth, thus a negative correlation between fractionation and growth rate is noted (Fig. 2c).δ 2 HC37 ratios from both T. lutea experiments align well with other Group II species I. galbana and R. lamellosa, even when temperature and nutrient concentrations caused a 50 ‰ range in δ 2 HC37 ratios (Fig. 3).Hydrogen isotope ratios of individual C37 alkenones follow the same trends as the integrated C37 alkenones (Table 2).Integrated ratios are used for comparison with δ 2 H ratios of I. galbana and R. lamellosa measured on a shorter GC column which prohibited baseline separation of the individual alkenones.

Hydrogen isotope ratios of lipids from E. huxleyi, I. galbana and R. lamellosa
For I. galbana and R. lamellosa, alkenones were the most enriched lipids by an average of 60 ‰ and 110 ‰, respectively, compared to the other lipids (Fig. 4a,b).In E. huxleyi, d 2 H ratios of C14:0 and C16:0 fatty acids were more depleted by 10 to 30 ‰ relative to alkenones, while the C18:1 fatty acid was the most 2 H-enriched compound by around 50 ‰ relative to the other fatty acids and alkenones (Fig. 4c).Fractionation for E. huxleyi fatty acids varied between 0.762 -0.872, and correlates significantly with salinity for C14:0 and C18:1 fatty acids (Fig. 4c), but the correlation between fractionation and salinity is not statistically significant for C16:0 (r = 0.84, n = 4, p = 0.16).However, we treat these significant correlations with caution as they are based on only four data points.The d 2 H ratios of both brassicasterol and phytol are more depleted than d 2 H ratios of alkenones by an average of 120 ‰ and 230 ‰, respectively, for all three species (Fig. 4; Supplementary Table 1).The d 2 H ratios and a of brassicasterol correlate significantly with salinity in all three species (Fig. 4).There is no significant relationship with salinity for phytol from E. huxleyi and I. galbana, but both d 2 Hphytol ratios and aphytol from R. lamellosa do correlate significantly with salinity (d 2 Hphytol: r = 0.90, n= 28, p < 0.001; aphytol: r = 0.54, n = 28, p < 0.05).

Effects of temperature for T. lutea alkenones
The influence of temperature on δ 2 HC37 ratios in haptophyte algae is not well constrained.Previous results from cultures investigating the effect of temperature on δ 2 HC37 ratios in Group III haptophyte species are contradictory (Schouten et al., 2006;Wolhowe et al., 2009), and there has been no characterization of the temperature effect on δ 2 HC37 ratios for Group II haptophytes to our knowledge.Our results show that individual alkenones, as well as integrated C37 alkenones, show a significant positive correlation with temperature in Group II species T. lutea (Fig. 1).The positive linear correlation for the integrated alkenones seems to be driven by the C37:3 alkenone, begging the question whether the observed correlation is really a temperature effect, or primarily a function of relative abundance and therefore an indirect temperature effect (c.f.van der Meer et al., 2013).The relative abundance of the C37:3 and C37:2 alkenones are strongly coupled to temperature, with higher amounts of the C37:2 at higher temperatures (Prahl and Wakeham, 1987).The C37:3 is synthesized from the C37:2 by a desaturation step (Rontani et al., 2006;Endo et al., 2018;Kitamura et al., 2018), which occurs less often at higher temperatures.Desaturation is associated with 2 H depletion (Chikaraishi et al., 2004), thus it follows that the C37:3 alkenone should be more depleted than its C37:2 precursor, as observed here (Fig. 1).At lower temperatures, when there is a higher abundance of C37:3, and much lower relative abundance of C37:2, the C37:2 should be comparatively enriched.For the T. lutea temperature experiment, the C37:3 was depleted relative to the C37:2 from 15 -25° C, but from 30 -35° C, where the C37:2 was much more abundant than the C37:3, a relative depletion of the C37:2 compared to the C37:3 was observed (Fig. 1).It appears that 2 H depletion shifts in favor of the dominant alkenone, and when the difference in abundance of both alkenones is close to zero, the offset between the δ 2 H ratios of the two alkenones is also close to zero.These findings suggest that the temperature correlation observed for alkenones from T. lutea is likely related to relative abundance shifts, which in turn has an effect on δ 2 HC37 ratios, implying an indirect temperature effect on δ 2 HC37 ratios.Growth rate might also influence these correlations, but due to the fact that growth was monitored by chlorophyll fluorescence in these experiments, we cannot compare absolute values with growth rates obtained by daily cell counts.However, alkenone concentrations did not vary substantially over the temperature range (5-10 µg / mL at 15° C and 35° C vs 10-15 µg / mL from 20 -30° C; Fig. 2b from Nakamura et al., 2016), leading us to conclude that growth rates did not significantly impact our δ 2 HC37 ratios.

Effects of nutrients and growth phase for T. lutea alkenones
T. lutea was also grown under two different nutrient concentrations and harvested during exponential and stationary growth phases to assess growth-related effects on αC37 values.N-reduced and N-replete media led to different growth rates: 0.14 µ d - 1 for N-reduced and 0.21 µ d -1 for N-replete.While these two growth rates are not extremely different, they led to a distinct offset in isotope ratios, with more fractionation under higher nutrient concentrations (i.e. higher growth rate) and a relative isotope depletion (Fig. 2).The negative correlation between αC37 values and growth rate was previously recognized in Group III species E. huxleyi and G. oceanica (Schouten et al., 2006;Wolhowe et al., 2009).The same negative relationship between αC37 values and growth rate was noted for our T. lutea experiment (Fig. 2), confirming that similar growth effects on δ 2 HC37 Biogeosciences Discuss., https://doi.org/10.5194/bg-2019-147Manuscript under review for journal Biogeosciences Discussion started: 29 April 2019 c Author(s) 2019.CC BY 4.0 License.ratios also occur in Group II species and are tightly coupled to nutrient availability.Sachs and Kawka (2015) hypothesized that lower growth rate leads to an up-regulation of OPP derived NADPH causing lipids to be relatively enriched; at higher growth rates, photosynthetically derived NADPH becomes the dominant H source, leading to a relative depletion of lipid δ 2 H ratios.This mechanism works in reverse to the salinity mechanism, where increased salinity causes an up-regulation of OPP derived NADPH, and therefore might account for the reduced sensitivity of δ 2 HC37-salinity in the natural environment (e.g.Schwab and Sachs, 2011;Weiss et al., 2019).The δ 2 HC37 ratios from T. lutea (temperature and nutrient experiments) fit well with values noted for other Group II species I. galbana and R. lamellosa (Fig. 3).Three Group II species now show similar δ 2 HC37 ratios which are more enriched than values reported for the Group III species E. huxleyi and G. oceanica, corroborating the hypothesis of a distinct Group II and Group III δ 2 HC37 signal (M' Boule et al., 2014).

Fatty acids and long-chain alkenones
Fatty acids from E. huxleyi in our experiment showed 2 H-enrichment with chain elongation (Supplementary Table 1, Fig. 4).
In general, these observations also hold true for fatty acids from E. huxleyi and I. galbana discussed here (Supplementary Table 1, Fig, 4).Both species show a significant positive correlation between d 2 H ratios of fatty acids and salinity (Fig. 4), except for the a -salinity correlation for C16:0 from E. huxleyi and C18:1 from I. galbana (C16:0 : r = 0.84, p > 0.05; C18:1 : r = -0.34,p > 0.05).However, the correlations for fatty acids from E. huxleyi are based on a low number of data points (n = 4 for each fatty acid), thus we treat these significant correlations with caution.The lack of correlation for C18:1 in I. galbana might partially be explained by the desaturation step since desaturation is associated with 2 H depletion, which has the potential to counteract the 2 H enrichment associated with higher salinities.Table 1, Fig. 4).Interestingly, for E. huxleyi, the C18:1 is the most 2 H-enriched compound out of the six, in contrast to I. galbana, where alkenones are the most 2 Henriched (Supplementary Table 1, Fig. 4).This highlights another potential offset between haptophyte Groups II and III.Species related differences aside, there is a correlation between d 2 H ratios and salinity for both fatty acids and alkenones.In algae, the first two steps of fatty acid biosynthesis (formation of malonyl-CoA and initial elongation by fatty acid synthases) occur in the chloroplast, and additional elongation beyond C16 or C18 relies directly on cytosolic sources (Cook and McMaster, 2002;Huerlimann and Heimann, 2013).Fatty acid elongation occurs via the stepwise addition of two carbon atoms and four hydrogen atoms, and desaturation removes two hydrogen atoms, allowing for substantial isotopic change (Chikaraishi et al., 2004), which can explain the isotopic differences between fatty acids and alkenones.Furthermore, alkenones are thought to be synthesized from these shorter chain fatty acids by elongation and subsequent decarboxylation in the chloroplast (Rontani et al., 2006).Fatty acids, although mainly synthesized in the chloroplast, also rely on acetyl-CoA as a precursor.Acetyl-CoA might be responsible for the salinity signal noted for fatty acids and alkenones, since de novo synthesis of acetyl-CoA comes from pyruvate produced in the cytosol (DeNiro and Epstein, 1977).

Brassicasterol and phytol
The d 2 H ratios and α values of brassicasterol correlate significantly with salinity for all three species, whereas d 2 H ratios of phytol do not correlate with salinity for either E. huxleyi or I. galbana (Fig. 4).Brassicasterol and phytol are isoprenoid lipids formed with either isopentenyl phosphate (IPP) or dimethylallyl pyrophosphate (DMAPP) as precursors.Two pathways exist for the synthesis of IPP and DMAPP.The Mevalonate (MVA) Pathway generates isoprenoid precursors in the cytosol using acetyl-CoA (Schmidt et al., 2003;Eisenreich et al., 2004).In contrast, the Methylerythritol pathway (MEP) creates IPP and DMAPP in the plastids using pyruvate and glyceraldehyde-3-phosphate (G3P) (Schmidt et al., 2003;Guggisberg et al., 2014;Lohr et al., 2012;Sachs et al., 2016).Since phytol is synthesized by precursors generated via MEP and brassicasterol is synthesized by precursors from MVA (Lichtenthaler et al., 1997;Bohlmann et al., 1999;Lichtenthaler, 1999;Chikaraishi et al., 2009), the offset of ~100 ‰ between the two compounds likely stems from differences associated with these two pathways.steps.One of the three reduction steps in the MEP is catalyzed by the IspH enzyme that supplies IPP and/or DMAPP with depleted H (Schmidt et al., 2003), potentially explaining why phytol is more 2 H depleted than brassicasterol.

Synthesis of isoprenoid precursors via the MVA pathway involves two reduction steps, but the MEP involves three reduction
Further evidence to confirm different precursor pathways and cellular compartments for phytol versus sterol comes from a light intensity experiment conducted by Sachs et al. (2017) assessing effects of irradiance on organic compounds from the diatom Thalassiosira pseudonana.The δ 2 H ratios of phytol showed a strong correlation with light intensity (R 2 = 0.90, p < 0.0001), while the δ 2 H ratio of 24-methyl-cholesta-5,24(28)-dien-3b-ol was not affected by light intensity.These results support the idea that phytol is synthesized by precursors in the chloroplast, especially since phytol is a component of chlorophyll (de Souza and Nes, 1969).Sterols, on the other hand, can be synthesized by cytosolic precursors and rely on a H source that is not connected to light availability, but instead potentially correlated to salinity, either directly or indirectly.
Acetyl-CoA used in the MVA pathway is synthesized from pyruvate or b-oxidation of excess fatty acids in the cytosol (Nelson and Cox, 2017).Pyruvate can be synthesized in the chloroplast, and used directly in MEP, but it can also be synthesized in the cytosol, where it can be converted to acetyl-CoA and used in MVA (Williams and Randall, 1979;Disch et al., 1998;Wallace, 2013).Differences in cellular compartment and biosynthetic precursors (acetyl-CoA and pyruvate) associated with metabolic reactions provide an explanation for why δ 2 H ratios of brassicasterol correlate to salinity and δ 2 H ratios of phytol do not.
However, in R. lamellosa, there is a positive linear correlation between phytol δ 2 H and salinity (δ 2 H: r = 0.90, n= 18, p < 0.001), and a much weaker correlation for a (r = 0.54, n = 18, p < 0.05).The correlation with salinity observed for phytol from R. lamellosa is puzzling.It might be related to an addition of pyruvate synthesized in the cytosol mixing with pyruvate synthesized in the chloroplast, since pyruvate is reported to flow between the two cellular compartments (Williams and Randall, 1979).

Mechanisms for salinity impact on hydrogen isotope fractionation
The general trend observed between δ 2 H ratios of organic compounds (excluding phytol) and salinity is δ 2 H enrichment with increasing salinity.As explained above, the absence of a correlation between δ 2 H ratios of phytol and salinity is presumably the result of the location of phytol biosynthesis and biosynthesis of the key intermediate, pyruvate.Fatty acids, alkenones and brassicasterol are all connected more directly to H pools in the cytosol because of their reliance on acetyl-CoA as a precursor, and all show relative 2 H-enrichment at increased salinities (Fig. 4).While the precise mechanism controlling the correlation between δ 2 H ratios and salinity is unknown, certain concepts related to hydrogen isotope fractionation and intracellular pools of hydrogen are known.
NADPH is an essential cofactor that acts as a significant source of H in biosynthesis reactions (Sessions et al., 1999;Zhang et al., 2009 and references cited therein).The hydrogen isotopic composition of NADPH can vary substantially depending on the source or location of reduction within the cell, which can then be evidenced down-stream in the hydrogen isotope ratios of resultant organic compounds (Zhang et al., 2009).NADPH can be reduced via two main sources: light reactions in the chloroplast (photosynthetically-derived) and the oxidative pentose phosphate (OPP) pathway in the cytosol (metabolicallyderived).Photosynthetically-derived NADPH is the product of the reduction of NADP+ by the redox protein ferredoxin (Shin, 2004;Zhang et al., 2009;Cormier et al., 2018).The resultant NADPH is 2 H-depleted compared to growth water.The δ 2 H ratios of NADPH reduced by light reactions of photosynthesis are dependent on light intensity below ~ 115 µ mol photons m - 2 s -1 and large fractionation is associated with lower light intensities (Cormier et al., 2018).In the cytosol, the OPP pathway supplies the cell with NADPH by reducing NADP+ during conversion of glucose-6-phosphate (Wamelink et al., 2008), a reaction that is not linked to photosynthetic activity, but potentially linked to salinity (Sachs et al., 2016).It has been speculated that at lower light intensities, the OPP pathway supplies a larger portion of NADPH for biosynthesis since NADPH from light reactions of photosynthesis is limited (Cormier et al., 2018).Additionally, up-regulation of the OPP pathway is hypothesized at increased salinities, which is associated with relative 2 H-enrichment (Maloney et al., 2016;Sachs et al., 2016).A dominance of metabolically-derived H for biosynthesis would yield more enriched organic compounds relative to a when a photosynthetically-derived H source is dominant, and this enrichment is enhanced at increased salinities.Isotopic enrichment is observed at increased salinities in a majority of the lipids discussed here (Fig. 4), leading to the hypothesis that OPP is likely the dominant H source of NADPH used for synthesis of these lipids.For haptophytes not growing under light-limited conditions, but growing over a range of salinities, a balance between photosynthetically reduced and OPP derived NADPH is expected at lower salinities, and a larger proportion of OPP derived NADPH is predicted at higher salinities.Phytol synthesis occurs in the chloroplast, thus a significant portion of H should come from photosynthetically-derived NADPH, explaining why phytol δ 2 H ratios are significantly depleted relative to other organic compounds, and δ 2 H ratios of phytol might be more sensitive to light rather than salinity.
The amount of H derived from water for biosynthesis is an equally important control on why some lipids retain a correlation to salinity.For example, when pyruvate is formed by the tricarboxylic acid (TCA) cycle, there is more exchange of H with relatively enriched surrounding water, which causes a relative isotopic enrichment (Sachs et al., 2016;Cormier et al., 2018).
In contrast, when pyruvate is formed by glycolysis, only one third of the H comes from water, resulting in a relative depletion (Sachs et al., 2016).However, as explained above, NADPH sources are thought to exert a strong control on 2 H ratios of lipids, thus these differences between pyruvate from TCA compared to glycolysis might be of secondary importance in resultant lipid 2 H ratios.When larger amounts of H derived directly from water is used in lipid synthesis, there is a more direct connection between lipid δ 2 H ratios and growth water δ 2 H ratios, and therefore less fractionation.Less fractionation is observed at higher salinities for the majority of biomarkers presented here, therefore it is possible that up-regulation of certain pathways incorporates proportionally more H directly from intracellular water, or potentially employing more acetyl-CoA from TCAderived pyruvate, at higher salinities causes a closer connection to extracellular δ 2 HH2O ratios.
At higher salinities cells must maintain osmotic pressure to avoid lysis, and create organic compounds known as osmolytes for this purpose (Dickson et al., 1982;Dickson and Kirst, 1987).Indeed, higher amounts of osmolyes (most notably DMSP) are observed at higher salinities in some haptophyte species (Dickson and Kirst, 1987).It has been previously proposed that enhanced production of osmolytes might cause the intracellular pool of H to become more 2 H enriched as a result of increased recycling of intracellular water (Sachse and Sachs, 2008;Maloney et al., 2016), leading to enriched d 2 H ratios of lipids at higher salinities.I. galbana is adapted to grow at a large range of salinities, while E. huxleyi is adapted to grow at a higher, but more restricted range of salinities.Because I. galbana is more cosmopolitan, it might be more metabolically active as a consequence of salinity stress over the much larger range of salinities at which it grows, and could likely be shifting more precursors and reducing power into osmolyte production than E. huxleyi at the same salinity.This would cause relative The above mechanisms suggest that the site of NADPH reduction is the primary driver of the hydrogen isotope composition of lipids, but the source of the salinity signal recorded in haptophyte lipids potentially comes from biosynthetic precursors tied to specific cellular compartments.The hydrogen isotopic composition of compounds formed from cytosol-derived precursors, and those synthesized directly in the cytosol, change with salinity.Lipids that are synthesized in the chloroplast and depend more on chloroplastic precursors in conjunction with photosynthetically-derived NADPH, do not show variation in d 2 H ratios with salinity (e.g.phytol in E. huxleyi and I. galbana), or greatly diminished (e.g.C16:0 in E. huxleyi).

Conclusions
New results presented here showed a positive effect of temperature on δ 2 H ratios of long-chain alkenones for Group II species T. lutea which is likely coupled with changes in relative abundance of individual alkenones.Nutrient concentration showed a similar effect on Group II species T. lutea as previously reported for Group III species E. huxleyi, with depletion of δ 2 H ratios associated with stationary growth phase and higher nutrient concentrations.The effect of nutrients and growth on δ 2 H ratios noted here is opposite to the effect of salinity, and might cause changes in the sensitivity of the δ 2 H-salinity relationship in the natural environment.
Investigation of δ 2 H ratios of multiple lipids produced by the same haptophyte algae showed that biosynthesis occurring in the cytosol, and relying on precursors of cytosolic origin, yields lipids with δ 2 H ratios that are sensitive to salinity.In contrast, the δ 2 H of lipids produced in the chloroplast, and dependent upon precursors partially or whole produced in the chloroplast, did not change with salinity.Therefore, alkenones, brassicasterol and fatty acids with a chain length of at least 16 carbon atoms, or diagenetic products derived from these lipids, should be selected instead of phytol or shorter chain fatty acids for paleosalinity reconstructions.The offset between δ 2 H ratios of lipids synthesized by Group II and III haptophytes is likely a result of differing mechanisms for regulating salt stress.Group III haptophytes are plausibly more adapted to growth at higher salinities than Group II haptophytes, which are adapted for growth at a more variable range of salinities.The offset between species, which can be on the order of 100 ‰, is an essential factor to consider when employing δ 2 H ratios to reconstruct salinity in areas where mixing of species is anticipated.
Biogeosciences Discuss., https://doi.org/10.5194/bg-2019-147Manuscript under review for journal Biogeosciences Discussion started: 29 April 2019 c Author(s) 2019.CC BY 4.0 License.enrichment of the intracellular H pool in I. galbana compared to E. huxleyi, explaining offset in d 2 H ratios observed between these two species.The enrichment associated with increased osmolyte production might not be as strongly evidenced in fatty acids because fatty acids are initially synthesized in the chloroplast, up to C14 or C16, thus are not exposed directly to cytosolic pools of H. Chain elongation leading to long-chain alkenones does take place in the cytosol, thus d 2 H ratios of alkenones reflect changes in salinity.
Contributions D. C and S. K. designed and carried out the E. huxleyi, I. galbana and R. lamellosa experiments and the respective hydrogen isotope analyses.H. N. provided the extracts for the T. lutea temperature experiment.F. D. C. and P. S. provided the extracts for the T. lutea nutrient and growth phase experiment.G. M. W. conducted hydrogen isotope analyses for both T. lutea experiments and prepared the manuscript with contributions from all co-authors.

Figure 2 :
Figure 2: Results from batch cultures of Tisochrysis lutea investigating effect of nutrient concentrations, growth phase and growth rate on hydrogen isotope fractionation of long-chain alkenones (aC37).Hydrogen isotope fractionation factor a C37 shows (a) a negative trend with nutrient concentration, and (b) a negative trend with growth rate.Relative isotopic enrichment is noted for stationary growth phase compared to exponential growth phase.