Samples were collected on a zonal transect between Guadeloupe and Tenerife at approx. ∼ 22° N between 26 June and 12 August 2017 onboard the RRS James Cook (JC150, Fig. 1a). Sea surface temperature (SST) was measured via the underway seawater system using Seabird sensors. Using a trace-metal clean towed FISH and a Teflon diaphragm pump (Almatec A-15), seawater samples were collected every 2 h, at a resolution of ∼ 25 km, from ∼ 3 m below the surface (Fig. 1a), with seawater flow terminating into a class-100 clean air-laboratory.

Using unfiltered seawater samples from the towed FISH, concentrations of nitrate plus nitrite (Brewer and Riley, 1965), phosphate (Kirkwood 1989) and ammonium (Jones, 1991) were analysed onboard according to GO-SHIP nutrient protocols (Becker et al., 2020). Using filtered seawater from the towed FISH (Sartobran, Sartorius, 0.8/0.2 µm polyethersulfone membrane), concentrations of dissolved iron (Kunde et al., 2019) were measured onboard while concentrations of dissolved zinc (Nowicki et al., 1994) were determined at the University of Southampton. Concentrations of DOP were determined at the University of Liverpool using a modified version of Lomas et al. (2010) as described by Davis et al. (2019). Using unfiltered seawater from the towed FISH, rates of alkaline phosphatase were determined onboard every 4 h or ∼ 50 km (Davis et al., 2019). Enzyme kinetic parameters were determined at each station by incubating unfiltered surface seawater with various concentrations of the synthetic fluorogenic substrate 4-methylumbeliferyll-phosphate (MUFP, Sigma Aldrich) and measuring the change in fluorescence for 8 hours (as described by Davis et al., 2019). The maximum hydrolysis rates (Vmax) and the half saturation constant (Km) were determined using the Hanes-Woolf plot graphical linearization of the Michaelis-Menten equation following Duhamel et al. (2011). Prochlorococcus, Synechococcus (or Parasynechococcus, (Coutinho et al., 2016) and high and low nucleic acid bacteria (HNA and LNA, respectively) were enumerated every 2 h at Plymouth Marine Laboratory using flow cytometry (Tarran et al., 2006). Surface ocean concentrations of chlorophyll a (on GF/F) were determined on every sample (Welschmeyer, 1994). Concentrations of dissolved cobalt were measured in separate samples collected from 40 m from 4 stations only using high resolution inductively coupled plasma mass spectrometry (HR-ICP-MS), preceded by UV-digestion and off-line preconcentration into a chelating resin (WAKO) at the University of Southampton (Lough et al., 2019; Rapp et al., 2017).

At 7 stations, McLane pumps were deployed to 15 m (see Table S1 for deployment details). Data from station 1 was omitted from this study due to significant riverine influence (Kunde et al., 2019). Pumps were fitted with a trace metal clean mini-MULVS filter head. Between 17 and 359 L of seawater was filtered through a 51 µm (Nitex), 3 µm (Versapor) and 0.2 µm (Supor) filter stack. Filters were immediately frozen at -80 °C, with subsequent transportation and storage at -80 °C. Protein biomarker analysis was conducted on the 0.2 µm filter, representing the 0.2 to 3 µm particle fraction. Briefly, upon return to the laboratory, the total microbial protein was extracted using a detergent based method. The filter was unfolded and placed in an ethanol rinsed tube, then covered in 1 % SDS extraction buffer (1 % SDS, 0.1 M Tris HCl pH 7.5, 10 mM EDTA), incubated at room temperature for 10 min, then at 95 °C for 10 min, and then shaken at room temperature for 1 h. The extract was decanted and clarified by centrifugation before being concentrated by 5 kD membrane centrifugation to a small volume, washed in extraction buffer, and concentrated again. The total protein concentration was determined by BCA assay (kit) at this time. The proteins were precipitated in cold 50 % methanol 50 % acetone 0.5 mM HCl at 20 °C for one week, collected by centrifugation at 4 °C, and dried by vacuum. Purified protein pellets were resuspended in 1 % SDS extraction buffer and redissolved for 1 h at room temperature. Total protein was again quantified by BCA assay to assess recovery of the purification.

Extracted proteins were immobilized in a small volume polyacrylamide tube gel using a previously published method (Lu and Zhu, 2005; Saito et al., 2014). LC-MS/MS grade reagents were used and all tubes were ethanol rinsed. The gels were fixed in 50 % ethanol, 10 % acetic acid, then cut into 1 mm cubes and washed in 50:50 acetonitrile: 25 mM ammonium bicarbonate for 1 h at room temperature, then washed again in the same solution overnight. Next, the gels were dehydrated by acetonitrile treatment before protein reduction by 10 mM dithiothreitol treatment at 56 °C for 1 h with shaking. Gel pieces were rinsed in 50:50 acetonitrile: ammonium bicarbonate solution, then proteins were alkylated by treatment with 55 mM iodacetamide at room temperature for 1 h with shaking. Gels were again dehydrated by acetonitrile treatment and dried by vacuum. Finally, proteins were digested by treatment with trypsin gold (Promega) prepared in 25 mM ammonium bicarbonate at the ratio of 1:20 µg trypsin: ug total protein overnight at 37 °C with shaking. The next morning, any supernatant was decanted into a clean microfuge tube, and 50 µL protein extraction buffer (50 % acetonitrile, 5 % formic acid in water) was added to the gels, incubated for 20 min, centrifuged and collected. The extraction was repeated and combined with the original supernatant. Peptides were concentrated to approximately 1 µg total protein per µL solution by vacuum at room temperature. 10 µL or 10 µg were injected per analysis.

Global metaproteome analysis, which is conducted with no prior determined targets, was performed in Data-Dependent-Acquisition (DDA) mode using Reverse Phase Liquid Chromatography – active modulation – Reverse Phase Liquid Chromatography Mass Spectrometry (RPLC-am-RPLC-MS) (McIlvin and Saito, 2021). RPLC-am-RPLC-MS involves two orthogonal chromatography steps, which are performed in-line on a Thermo Dionex Ultimate 3000 LC system equipped with two pumps. The first separation was on a PLRP-S column (200 µm × 150 mm, 3 µm bead size, 300 Å pore size, NanoLCMS Solutions) using an 8 h pH 10 gradient (10 mM ammonium formate and 10 mM ammonium formate in 90 % acetonitrile), with trapping and elution every 30 min onto the second column. The second separation occurred in 30 min intervals on a C18 column (100 m × 150 mm, 3 µm particle size, 120 Å pore size, C18 Reprosil-God, Maisch, packed in a New Objective PicoFrit column) using 0.1 % formic acid and a 0.1 % formic acid in 99.9 % acetonitrile. The eluent was analyzed on a Thermo Orbitrap Fusion mass spectrometer with a Thermo Flex ion source. MS1 scans were monitored between m/z 380 and 1580, with an m/z 1.6 MS2 isolation window (CID mode), 50 ms maximum injection time and 5 s dynamic exclusion time.

Resulting spectra were searched in Proteome Discoverer 2.2 with SequestHT using a custom DNA sequence database consisting of over 30 genomes from cyanobacteria isolates and metagenomic data from the Pacific and Atlantic oceans (including metagenomes from Metzyme and Geotraces cruise GA03). Annotations were derived using BLASTp against the NCBI non-redundant protein database. The corresponding protein FASTA file is available with the raw mass spectra files (see Supplement S1). SequestHT parameters were set to 10 ppm for the parent ion, 0.6 Da for the fragment, with cysteine modification (+57.022) and variable methionine (+16.0) and cysteine oxidation allowed. Protein identifications were made using Protein Prophet in Scaffold (Proteome Software) at the 95 % peptide confidence level, resulting in < 1 % protein and peptide FDRs. Details of the peptides identified relative to protein name and organism can be found in Table S2 and the protein report and analytical details can be found in Supplement S2.

Global metaproteome protein abundances are reported in normalized spectral counts. The normalization is performed by summing the total number of spectra in each sample, calculating the average number of spectra across all the samples, and then multiplying each spectrum count by the average count over the sample's total spectral count. This is done to control for small differences in the amount of sample injected into the mass spectrometer.

A small number of tryptic peptides were selected for absolute quantitative analysis in the samples from nutrient addition experiments (see Sect. 2.5 for details) and were analysed as described in detail by Held et al. (2026) (companion manuscript). The amino acid sequence for the protein biomarkers quantified in this study (PstS, PhoA, PhoX) for Prochlorococcus and Synechococcus are summarised in Table S3 and peptide report and analytical details are found in Supplement S3.

Trace-metal clean sampling and incubation protocols used to setup onboard bioassays are described in detail in the Supplement S4. Aliquots of Fe, Zn and Co solutions were added to unfiltered seawater to investigate metal limitation of alkaline phosphatase and results are reported in Held et al. (2026). Alongside these experiments, we added DOP alone or with Fe and Zn to investigate the potential for organic P availability to influence AP activity at stations 2 and 3 only, where concentrations of DOP were low (< 80 nM, Fig. 1a and e, Table S4), and the results are reported here. Trace-metal clean 20 L carboys were triple rinsed with unfiltered seawater collected from 40 m (to avoid contamination from the ship) via the FISH and filled and amended accordingly (Table S4). At the start and end of 48 h, we measured phytoplankton biomass (chlorophyll a, abundance of Prochlorococcus, Synechococcus) and AP activity. After 48 h, we collected samples to quantify protein concentration (PstS, PhoA and PhoX) as described in Sect. 2.4 (Table S3). Incubations were conducted in triplicate. However, due to the biomass (therefore volume) required for protein analysis, we were unable to collect samples from three incubation bottles for further analyses. Instead, all measurements were collected from two incubation bottles, except aliquots for determination of AP, which was collected from three incubation bottles. To compare the change in states or rates in treatments relative to the control, we considered a significant change in a property to occur when the mean of the property in the amended incubation was 2-times higher (or lower) than the mean control incubation. Incubations were conducted in a temperature controlled container set to a temperature measured at 40m (between 25 and 27 °C) and with 12:12 h light : dark cycle simulated by LED light panels (Part no: LED-PANEL-300-1200-DW and LED-PANEL-200-6-DW, Daylight White, supplier Power Pax UK Limited).

Strong zonal gradients were evident in surface temperature and phosphorus concentrations. From west to east, SST decreased by ∼ 3 °C (Fig. 1b), phosphate increased by ∼ 15 nM (Fig. 1c) and DOP increased 3-fold (from ∼ 50 nM to ∼ 150 nM, Fig. 1d). By comparison, there were no clear zonal trends in fixed nitrogen, with concentrations of nitrate plus nitrate (N + N, herein nitrate) ranging from < 10 nM to ∼ 40 nM (Fig. 1e) and ammonium, which ranged from 3 to 21 nM, being highest at stations 5 and 6 (Fig. 1f).

There was a clear zonal trend in dissolved Fe concentrations, which decreased west to east by ∼ 1.0 nM (Fig. 2a) owing to enhanced Saharan dust deposition in the western Atlantic Ocean (Kunde et al., 2019). In contrast, Zn concentrations were variable throughout the transect (ranging from 0.04 to 0.8 nM, Fig. 2b) and cobalt was constant (∼ 11 to 14 pM, data not shown).

Microbial biomass, picocyanobacteria abundance and biological rates exhibited strong zonal gradients. From west to east, there were increases in chlorophyll aconcentration (Fig. 3a) and Prochlorococcus cell abundance (Fig. 3b) whereas Synechococcus cell abundance decreased (Fig. 3c). HNA and LNA bacterial abundance (Fig. 3d and e, respectively) also increased from east to west. There was also a 4-fold decrease in chlorophyll corrected APA (from > 2000 nmol P µg chl a d⁻¹ to < 500 nmol P µg chl a d⁻¹, Fig. 3f), likely in response to the observed gradient in P/DOP availability (Mahaffey et al., 2014 , Fig. S1a, S1b).

In addition, the abundance of key diazotrophs Trichodesmium and UCYN-A increased from west to east (Cerdan-Garcia et al., 2022). Although rates of N₂ fixation in the east exceeded those in the west (3 to 10 nM d⁻¹ and < 3 nM d⁻¹, respectively), the highest rates were in the central transect between stations 4 and 5 (12 to 18 nM d⁻¹, Fig. 3g).

These zonal gradients in hydrography, nutrients, biological rates and picocyanobacteria create two contrasting regions – one in the west (west of 46° W or west of station 4) and one in the east (east of 46° W or east of station 4). Thus, quantitative comparisons of key characteristics can be drawn between station 2 at 54° W and station 7 at 31° E (Fig. 1a, Table 2). Compared to the east, conditions in the west were characterized by notably higher dissolved Fe and ammonium concentrations (3 to 4-fold higher), APA (4-fold higher) and Synechococcus abundance (2-fold higher). In contrast, the east was characterized by relatively high phosphate, DOP, chlorophyll a, Prochlorococcus abundance, rates of N₂ fixation, Trichodesmium and UCYN-A abundances (Table 2).

Based on these biogeochemical parameters, the phosphate-binding protein PstS, which is expressed under P-limiting conditions, would be expected to be prevalent throughout the transect consistent with low phosphate concentrations across the entire transect. Protein biomarkers would also be expected to indicate higher alkaline phosphatase (AP) abundances in the west, corresponding with the observed trends in APA. In addition, PhoX would be expected to be prevalent in the Fe-rich west, with greater prevalence of Fe-stress biomarkers in the east.

Higher PstS and PhoA in the west compared to the east, alongside the positive relationship between Pro-PstS, Pro-PhoA and Syn-PhoA with AP activity and negative relationship with DOP corroborate that these protein biomarkers are P-stress biomarkers in both Prochlorococcus (Martiny et al., 2006; Moore et al., 2005; Reistetter et al., 2013) and Synechococcus (Scanlan et al., 1993; Tetu et al., 2009). However, DOP addition stimulated distinct biomarker responses in Prochlorococcus and Synechococcus (Fig. 6, Table S5). For Prochlorococcus, DOP addition reduced Pro-PstS and Pro-PhoA but increased PhoX after 48 h relative to the control (Fig. 6, Table S5). In contrast, for Synechococcus DOP addition increased Syn-PstS and Syn-PhoA, with no change in PhoX (Fig. 6, Table S5). We intuit that the protein biomarkers changed due to a physiological response rather than change in cell abundance because the per cell protein content (Fig. S5) showed the same pattern (with the caveat that the protein is clade specific yet likely targeted a major ecotype, whereas cell abundance represents all cells). DOP addition stimulated a decrease in Pro-PstS and Pro-PhoA per cell and increase Pro-PhoA per cell relative to the control (Fig. S5a), whereas DOP addition stimulated an increase in Syn-PstS, Syn-PhoA and Syn-PhoX per cell (Fig. S5b).

Either the protein regulatory pathway differs between Prochlorococcus and Synechococcus, and/or the strain specific differences in quantified proteins is complicating our interpretation response of proteins across different strains. Here we describe evidence for the former hypothesis. In Prochlorococcus, the Pho regulon controls P-acquisition genes such as pstS (phosphate transporter) and phoA and includes the two-component regulatory genes, phoB and phoR (Martiny et al., 2006). The Pro-phoX gene is controlled by the pho regulon as it sits within a genomic island with other P stress responsive genes (Kathuria and Martiny, 2011). In contrast, Synechococcus (WH8102) has a two-tiered phosphate response system , where the PhoBR regulator controls pstS using a Pho box (Cox and Saito, 2013; Tetu et al., 2009) and second regulator, PtrA, controls one of the phoA phosphatase copies, Zn transport, and various other cellular processes (Ostrowski et al., 2010). The gene neighbourhood containing phoA (SYNW2391) in Synechococcus is also located near an efflux transporter and close to the ferric uptake regulator, Fur. To our knowledge, regulation of PhoX and its interaction with PhoA regulation in the marine picocyanobacteria is not well understood, but analysis of the gene neighbourhood in the model organism Prochlorococcus sp. NATL1A reveals that phoX is not within the phoA neighbourhood and is in the vicinity of a putative manganese transporter. For Synechococcus (WH8102), the position of phoX (SYN1799) is like Prochlorococcus and is located directly next to the futAB iron ABC transport system, consistent with the iron requirement of this enzyme. The separation of phoA and phoX within the genome in both Prochlorococcus and Synechococcus (at least in the representative strains described above) implies their regulation may be distinct in the different organisms. Consistent with prior observations (Browning et al., 2017; Mahaffey et al., 2014; Rouco et al., 2018), PhoA and PhoX may be regulated by both metals and phosphorus availability but the specific regulatory system in picocyanobacteria is complex and still unknown.

The alternative interpretation is that differences in strain specificity among the identified proteins explains the differences in the Prochlorococcus vs. Synechococcus response patterns. Proteins concentrations are measured by detecting peptides, and biological specificity (e.g. to strains or species) is determined by comparing the amino acid sequence of the peptide to isolate genomes or annotated MAGs. A least-common-ancestor (LCA) analysis can then be performed to assess the level of biological specificity that is represented by that peptide (Saunders et al., 2020) (Table S3, Supplement S3). The targeted peptides were selected based on their abundance in a preliminary metaproteomics analysis, suggesting that these were the most abundant proteins within the microbial community (see Held et al., 2026 for a more complete description of how targeted peptides were selected for this study region). Peptide sequences for up to 5 strains of Prochlorococcus were targeted with a focus on the HLII clade, particularly strain MIT9314, allowing comparison between proteins of the same strain. In contrast, protein sequences for Synechococcus were compared across clades because the peptide sequence for PhoA and PhoX targeted WH8102 (clade III) but the peptide sequence for PstS targeted RCC307 (clade X, Table S3). While co-occurring clades III and X are geographically positively correlated in warm oligotrophic waters (Sohm et al., 2008), RCC307 possesses a different putative alkaline phosphatase gene compared to WH8102 (likely PhoA, see Tetu et al., 2009). This mismatch in targeted strains and clades means that interpretation of the response of Synechococcus (and perhaps Prochlorococcus) to nutrient addition needs to be treated with some caution until the physiology and regulatory pathways of protein production are better understood. However, based on current knowledge of phosphate acquisition genes in marine Synechococcus and Prochlorococcus, we do expect that the targeted proteins/strains are major players in our study region.

Unlike other cyanobacteria, where the trace metal availability aligns well with the biogeography of PhoA and PhoX (e.g. Trichodesmium, Rouco et al., 2018), there were no consistent trends between iron or zinc concentrations and proteins PhoX and PhoA respectively in this study. The distribution of Pro-PhoX from metaproteomes (Fig. 4a) did not reflect iron availability (Fig. 2a). Despite elevated iron in the west, Pro-PhoA concentrations were 2.7 to 4.7-fold higher than Pro-PhoX (reported fmol L⁻¹, Table 3), with Pro-PhoX being greater than Pro-PhoA in the east where Fe was lowest (Fig. 2a). PhoX was not detected for Synechococcus in metaproteome analysis but was detected quantitatively at the start of bioassay experiments (Table 3). Syn-PhoX concentrations also did not reflect iron availability (Table 3) and there was no consistent trend in the ratio between Syn-PhoA and Syn-PhoX (Table 3). Zonal trends in quantitative versus metaproteome-derived Syn-PhoA were different and likely driven by differences in depth horizons sampled (40 m for experiments, 15 m for metaproteome analysis) as well as the different Synechococcus populations captured using quantitative peptide analysis (see Table S3, clade III and X) compared to metaproteomes.

The systems biology of the PhoX enzyme is poorly understood compared to that of PstS and PhoA, where the latter is known to be regulated by phosphate and zinc (Cox and Saito, 2013; Martiny et al., 2006; Ostrowski et al., 2010; Tetu et al., 2009). Despite the lack of correlation between trace metal availability, PhoA and PhoX in the surface ocean in this study, results from bioassays conducted during the same expedition (Held et al., 2026; companion manuscript) lends support to the potential for a direct metal control on APA (Browning et al., 2017; Jakuba et al., 2008; Mahaffey et al., 2014; Saito et al., 2017). In the west, Zn addition stimulated a 6-fold increase in Syn-PhoA relative to the control. Cobalt addition simulated a 7-fold increase in Syn-PhoA and 8-fold increase in Pro-PhoX. Finally, iron addition stimulated a 2-fold increase in Pro-PhoX in the iron-deplete eastern Atlantic (Held et al., 2026; companion manuscript).

Proteins relating to iron, zinc and B₁₂ metabolism in Prochlorococcus increased in the east alongside the increase in Prochlorococcus cell abundance. Ferredoxin and zinc transporters increased eastward by 3 to 9-fold and protein annotated as CobW, a member of the COG0523 family implicated in metal chaperone functions (Edmonds et al., 2021) and Co chaperone for B₁₂ synthesis (Young et al., 2021) also increased 3 to 10-fold (Fig. 4c). The eastward increase in three independent proteins (up to 10-fold) was greater than the increase in total protein for Prochlorococcus (∼ 1.6-fold) implying a regulated molecular increase in response to resource limitation or competition, rather than reflecting a change in biomass only. However, zinc protein annotations in Prochlorococcus are putative and alignment-based transporter annotations are unable to discern cognate metal use. In addition, the role of zinc in Prochlorococcus physiology is uncertain. Prochlorococcus does not have an obligate Zn requirement when phosphate is available (Saito et al., 2002), and Zn is highly toxic to a Pacific Ocean strain of Prochlorococcus (Hawco and Saito, 2018). In addition, while CobW is an abundant protein among the ∼ 20 genes involved in cobalamin biosynthesis, there are currently no known biomarkers for cobalt or zinc metabolism in Prochlorococcus, with studies producing negative results (Hawco et al., 2020). For Synechococcus, there were no clear trends in ferredoxin (Fig. 4c) and flavodoxin was infrequently detected. These findings highlight the challenge when predicting the direct metal requirement alongside metals controlling multiple APs in-situ, and makes a strong case for continued biochemical characterization of cyanobacterial trace metal physiology and enzymes.

The spatial patterns in nitrogen stress biomarkers gleaned from non-targeted metaproteomics provided insight into how fixed nitrogen availability contributed to shaping the biogeography of Prochlorococcus and Synechococcus across the subtropical Atlantic. Surface ocean gradients in fixed nitrogen are established via a combination of upwelling in the eastern Atlantic (Menna et al., 2016), nitrogen fixation (Fig. 3g, Cerdan-Garcia et al., 2022) and dust deposition (Powell et al., 2015) delivering nitrate, ammonium and urea to the surface subtropical Atlantic Ocean alongside microbial demand consuming fixed nitrogen. In summer 2017, concentrations of nitrate (< 40 nM) and ammonium (< 20 nM) were relatively low across the transect (Fig. 1e and f). The lowest spectral counts of the N-stress proteins, AmtB and UrtA in Prochlorococcus (Fig. 4c) coincided with a region of elevated nitrogen fixation rates (Fig 3g), reflecting alleviation of N stress in Prochlorococcus. The combination of efficient uptake of nitrogen derived from nitrogen fixers (Caffin et al., 2018), alongside the small cell size of Prochlorococcus provides a strong competitive advantage under oligotrophic conditions, as observed in the North Pacific gyre (Saito et al., 2014, 2015). Otherwise, Prochlorococcus was subjected to increasing N stress towards the eastern Atlantic, evidenced by eastward increases in protein biomarkers in Prochlorococcus, specifically P-II, ammonium transporter AmtB and urea transporter UrtA (30 %–70 %, Fig. 4b and Table 2).

In contrast to Prochlorococcus, consistently elevated UrtA in Synechococcus suggests chronic N stress throughout the transect, with UrtA in Synechococcus being more than 5 times higher than for Prochlorococcus (Fig. 4b). This is consistent with the physiological disadvantage of Synechococcus, with its larger cell size and less efficient surface-area to volume ratio for nutrient acquisition (Chisholm et al., 1992). P-II and AmtB (or NtcA) was not detected in the metaproteome of Synechococcus, perhaps because Synechococcus was 5 to 10 times less abundant in the metaproteomes compared to Prochlorococcus. The dominance of proteins for ammonium and urea acquisition of Synechococcus and Prochlorococcus are consistent with the premise that while marine Synechococcus and some Prochlorococcus strains have the genetic makeup to assimilate nitrate (Berube et al., 2015; Domínguez-Martín et al., 2022; Martiny et al., 2009, it accounts for < 5 % of their total N demand, and instead ammonium and urea are the dominant N sources (Berthelot et al., 2019; Casey et al., 2016; Painter et al., 2008). The patterns observed align with established biogeographical trends in which Prochlorococcus dominates in the nutrient-deplete surface ocean due to its competitive advantage as a small cell, whereas Synechococcus persists in regions where fixed N is available. The proteomic data indicate that nitrogen acquisition traits are one of the key determinants of population dynamics, driving spatial partitioning between Prochlorococcus and Synechococcus and ultimately influencing primary productivity and nutrient cycling across the subtropical Atlantic.