Chloroplast–mitochondria synergy modulates responses to iron limitation in two Thalassiosira diatom species

Abstract

Iron is naturally present at low levels in modern oceans, where it is essential for life. Ocean-dwelling organisms such as oceanic phytoplankton must therefore adapt to the available levels. Among phytoplankton, diatoms are a highly diverse and successful taxon that includes the Thalassiosira genus. As a group, diatoms contribute around 20% of global primary photosynthetic productivity; they have also developed specific resources allowing them to thrive in low-iron regions. However, the major biological factors underlying their success in these ocean environments remain unknown. Here, we compared two Thalassiosira species: Thalassiosira oceanica from iron-poor open-ocean; and Thalassiosira pseudonana, from iron-rich coastal waters. Since iron is essential for both photosynthesis and respiration, we examined the specificities of the bioenergetic machineries in these organisms using a combination of photo-physiological, proteomics, and focused ion beam scanning electron microscopy methods. We particularly focused on chloroplast-mitochondrial coupling, a mechanism deployed by diatoms to ensure optimal transfer of photosynthetic products to promote cell growth. This study of this mechanism in the context of iron limitation reveals that the two diatoms differentially remodel chloroplast compartments in response to iron limitation. Their tolerance to these conditions is also linked to distinct constitutive mitochondrial architectures.

Graphical Abstract

Introduction

Diatoms, unicellular photosynthetic marine heterokonts, are major contributors to the oceanic food chain and to carbon dioxide storage, accounting for ~20% of primary production worldwide [1]. Their genomic and metabolic complexity [1–3], linked to their origins in secondary endosymbiosis, is characterized by frequent horizontal gene transfers [4, 5]. This complexity provides them with the resilience and adaptability needed to survive in diverse and fluctuating environments, where light intensity may vary [6, 7], darkness may be prolonged [8], mineral nutrient availability may fluctuate [9], and nutrient-stimulated blooms may occur [10]. The growth of diatom species and their distribution between turbid coastal and open-ocean waters often follow dissolved iron gradients [11, 12]. Indeed, low iron bioavailability chronically limits photosynthetic carbon fixation in open-ocean areas [13, 14]. Conversely, iron levels in coastal waters tend to be higher due to supply from the land [15, 16]. To rationalize the ecological success of diatoms in oceans, we need to understand how they adapt and/or acclimate to distinct environments [17].

Iron-containing proteins—such as iron–sulfur centers, cytochromes—are highly represented in chloroplasts and mitochondria, the organelles responsible for photosynthesis and respiration, respectively [18, 19]. These two processes are thus potentially impacted by iron limitation. The iron-containing proteins act as electron carriers, transducing reducing power across cell compartments and contributing to the generation of the transmembrane proton gradient required for cellular bioenergetic processes. The photosynthetic capacity of diatoms tends to be limited by iron restriction [20, 21]. Diatoms have developed mechanisms to allow energetic cross-talk between chloroplasts and mitochondria [22, 23] as a strategy to ensure optimal carbon fixation and cell growth [22–24]. However, the interactions between the subcellular compartments and their consequences for both cellular organization and bioenergetic processes have yet to be characterized in the context of iron limitation. Previous proteomics and biochemical data suggest increased coupling between photosynthetic and respiratory functions in diatoms grown under low iron levels [23, 25]. However, this cross-talk has not been fully elucidated, and we lack information on essential physiological and morphological adaptations or acclimations in the context of iron limitation.

Thalassiosira oceanica is a low-iron-tolerant diatom. Its responses to limited iron availability involve specific acclimation, such as constitutive substitution in chloroplasts of iron-containing cytochrome c₆ by copper-containing plastocyanin [26] and of ferredoxin by flavodoxin [27]. No equivalent changes to mitochondrial proteins have been observed in this species, and the relative abundance of iron-rich mitochondrial complexes is preserved under iron starvation [3]. Iron starvation is known to trigger an adaptive transcriptional activation of specific proteins in other Thalassiosira species: the iron reductase in Thalassiosira pseudonana or a flavodoxin in Thalassiosira weissflogii [28]. In Phaedactylum tricornutum, high-affinity carrier proteins, such as (Iron Starvation-Induced Protein 2a ISIP2a, are expressed to enhance iron supply [29]. In addition to these iron-uptake and -storage traits, iron starvation triggers metabolic and physiological adaptive responses, such as modified lipid metabolism, increased degradation of cell components [3], as well as recycling and reallocation of carbon and nitrogen [3, 23, 30].

To complete this picture and assess the overall relevance of these responses, we conducted a comparative analysis of laboratory-cultured T. oceanica and its close relative species T. pseudonana. The two diatoms were grown in media containing high and low iron concentrations, typical of their respective natural environments [12, 31]. The iron-rich and iron-poor media represented challenging conditions in terms of iron availability for each species. We then combined morphological, physiological, and proteomic approaches to shed light on the strategies deployed by the two organisms to cope with low iron availability. Our data revealed species-specific traits of subcellular morphology and physiological acclimation to sustain chloroplast–mitochondrial coupling.

Materials and methods

Cultures, growth, and establishing iron limitation

The coastal T. pseudonana [Center for Culture of Marine Phytoplankton (CCMP) 1335] was originally isolated from Long Island Sound, NY, USA. Thalassiosira oceanica (CCMP 1005) was originally isolated from the Sargasso Sea. Cells were grown under controlled light (120 μmol m^–2 s^–1; with a 12:12 photoperiod, Supplementary Fig. S1) at 18°C, in artificial seawater (AQUIL [32]) supplemented with major nutrients, trace elements, and vitamins, in line with the recipe for L1 medium [33] published by the Culture Collection of Algae and Protozoa (www.ccap.ac.uk). Silicate and copper concentrations were adapted based on the National Center for Marine Algae and Microbiota L1 media recipe (https://ncma.bigelow.org/). Culture media were supplemented with FeCl₃ to produce iron-rich (113 nM) or iron-poor (18 nM) conditions ([25, 31]; Supplementary Fig. S2). Cells were serially diluted until a constant growth rate was attained in each condition (~7 generations, 3 passages), to ensure that a stable status was established [25, 31]. For all experiments, cells were harvested during the exponential growth phase, 2–3 h after initiating light illumination in the photoperiod. Each biological replicate corresponds to a culture obtained from one initial independent cell culture prepared following this protocol.

Quantification of elements by inductively coupled plasma mass spectrometry

Cells were collected by centrifugation (15 min, 4000× g, 20°C). Pellets were washed three times in 0.3 M NaCl, 10 mM KCl, 20 mM HEPES pH 7.0 [29, 34]. Cell pellets were dehydrated to dryness before acid digestion at 90°C for 4 h in 65% (w/v) nitric acid. Mineralized samples were analyzed on an iCAP^TM RQ quadrupole mass spectrometer (Thermo Fisher Scientific) operating in helium collision mode. Magnesium (mean of ²⁴Mg and ²⁵Mg), phosphorus (³¹P), potassium (³⁹K), manganese (⁵⁵Mn), iron (mean of ⁵⁶Fe and ⁵⁷Fe), copper (mean of ⁶³Cu and ⁶⁵Cu), and zinc (mean of ⁶⁴Zn and ⁶⁶Zn) concentrations were determined by comparing signals from samples to standard curves produced using serial dilutions of a multi-element solution [inductively coupled plasma (ICP) multi-element standard IV, Merck]. Values were corrected thanks to an internal standard solution containing ⁴⁵Sc and ¹⁰³Rh, added online.

Oxygen measurements

Oxygen was measured as described in Yee et al. [35]. Briefly, 700 μl of cell culture at 10 million cells/ml was maintained under gentle stirring at 18°C in a WALZ KS-2500 water-jacketed chamber (Heinz Walz GmbH) paired with an FSO2-1 oxygen meter and optical microsensor (PyroScience GmbH). Before measurement, samples were bubbled with N₂ for ~20 s to lower oxygen levels. NaHCO₃ 5 mM was added to avoid carbon limitation during measurements. The gross oxygen evolution rate was calculated after ~8 min of light exposure [120 μmol photons m⁻² s⁻¹ photosynthetic photon flux density (PPFD)]. Oxygen consumption rate in the dark was measured just after extinction of the light source. This sequence of illumination was achieved thanks to a portable chlorophyll fluorometer based on the pulse-amplitude modulation (PAM) controlled by WinControl-3 software (MINI-PAM II, Heinz Walz GmbH).

Pigment extraction and quantification

A total of 10–50 million cells were collected by centrifugation (4000× g, 5 min, 4°C) and supernatant was removed. Protocols from previous studies [36–38] were adapted as follows: pellets were resuspended in 1 ml of 96% (v/v) ethanol, then incubated for 30 min at 4°C to ensure complete solubilization. After incubation, supernatants were collected by centrifugation (1 min, 10 000× g). Total chlorophyll and fucoxanthin concentrations were calculated from absorbance measurements as follows:

(1)

(2)

Chlorophyll fluorescence measurements

Cell cultures were centrifuged (4000× g, 5 min, 20°C) and pellets were re-suspended in culture media at the appropriate density (10 million cells/ml). The cell suspension (100 μl) was supplemented with NaHCO₃ 5 mM to prevent carbon limitation during measurements. After dark acclimation (10 min), the maximum photochemical quantum yield of Photosystem II (PSII), Fv/Fm, and the PSII operating efficiency, Φ_PSII, were measured as described in Baker [39]. An imaging fluorometer (Speedzen 3, JBeamBio, France) equipped with a highly sensitive and fast camera (Orca 4-Flash, Hamamatsu, Japan) was used. Tunable-intensity red light illumination was used to generate different actinic light irradiance levels (PPFD from 0 to 250 μmol photons m⁻² s⁻¹) and to produce a saturating pulse (3000 μmol photons m⁻² s⁻¹). Chlorophyll fluorescence was induced by exposing the sample to blue light detection pulses. Experiments were performed at room temperature. Photosynthetic electron transfer rate (ETR) through PSII was deduced from Φ_PSII using the following equation:

(3)

where I is incident PPFD, A is the fraction of PPFD absorbed by the cells (assumed to be 0.84, i.e. 84%), fraction PSII is the fraction of absorbed light driving PSII photochemistry (assumed to be 0.5 for PSII, with the other half absorbed by PSI), and Φ_PSII is PSII operating efficiency [39].

Photosynthetic ETR was measured using the same setup, but adding respiratory inhibitors antimycin A 1.2 μM [40] and salicylhydroxamic acid 250 μM [41]. These compounds respectively inhibit mitochondrial cytochrome c reductase (complex III) and alternative oxidase (AOX), thus targeting both the cyanide-sensitive and -insensitive respiratory pathways [22].

FIB-SEM sample preparation and data acquisition

Cells were harvested by centrifugation at 2000× g for 5 min. The culture medium was removed, and pelleted cells were submitted to high-pressure freezing (HPM100, Leica) and freeze substitution (EM ASF2, Leica), as described previously [42]. Focused ion beam scanning electron microscopy (FIB-SEM) tomography was then performed on either a Zeiss NVision 40 or a Zeiss CrossBeam 550 instrument (Zeiss, Germany). The resin block containing the cells after substitution was mounted on a stub using carbon paste. It was then surface-abraded with a diamond knife in a microtome to obtain a clean, flat surface. Samples were metallized with 4 nm of platinum to avoid charging during observations. Inside the microscope, a second platinum layer (1–2 μm) was deposited locally on the area to be analyzed to mitigate possible curtaining artifacts. The sample was then scraped one slice at a time with a Ga + ion beam (700-pA current at 30 kV), and each exposed layer was captured by SEM at 1.5 kV and a current of around 1 nA using an in-lens EsB backscatter detector. Individual SEM images, 1000–2000 frames, were recorded to provide an image stack with an isotropic voxel size of 6 nm. Image stacks were binned two-fold leading to a final isotropic voxel size of 12 nm for downstream analysis. Multiple cells from each culture were imaged. The number of cells used for the morphometric analysis is indicated in the legends of the figures.

FIB-SEM data analysis

The FIB-SEM image stack was pre-processed to facilitate subsequent segmentation [43]. Fiji software was then used to crop individual cells from each pre-processed stack, thus reducing the number of frames to be analyzed. The segmentation method was adapted from previous studies [44] and was conducted using 3D Slicer 5.2.2 software tools (https://www.slicer.org/). Pixels were classified within each region of interest (ROI) as corresponding to cellular, subcellular, or organelle structures either manually or semi-automatically. For this step, a voxel volume of 1 (relative units) was set for image treatment and further corrected in line with the true voxel size (12 nm) during morphometric analysis. 3D representations of the segments were generated for subsequent smoothing and visual representation using MeshLab v2022.02-64-bit software (www.meshlab.net) and Paraview 5.11.0 software (https://www.paraview.org/). Morphometric analyses of volume and surface information for each ROI were performed using the quantification/segments statistics module in 3D Slicer Results and the quality measures and computations module of MeshLab for comparison. The two methods returned comparable results and data were therefore averaged. The raw data set, including the segmented volume for each cell (species and condition) is publicly available on BioImage Archive (accession number S-BIAD3307).

Proximity surface measurements

Simplified meshes of the 3D reconstructions (STL files) of chloroplasts and mitochondria were generated using MeshLab software’s Quadric Edge Collapse Decimation simplification algorithm. A reduction of 30% of the total number of mesh faces was applied to facilitate analysis by the minsurf.py Python script (from https://gitlab.com/clariaddy/mindist; [43]). This script was used to compute the total surface of a mesh present in a region within a threshold distance from a second mesh.

Sample preparation for proteomics analysis

Cells (50 million) were harvested by centrifugation at 4000× g, 4°C, for 90 s to remove culture media. The pellet was then re-suspended in 50 μl of Lyse reagent (PREOMICS iST^TM sample preparation kit, www.preomics.com). Samples were incubated in line with the PREOMICS instructions and subjected to three cycles of sonication for 30 s to shear DNA. The resulting total protein extract was analyzed. Protein concentrations were determined using the Pierce™ BCA protein assay kit (Thermo Scientific) according to the manufacturer’s instructions, using bovine serum albumin to create a standard curve. For analyses, 20–35 μg of protein was digested and purified according to the instructions provided with the PREOMICS iST sample preparation kit.

Mass spectrometry-based proteomics and data analysis

Approximately 500 ng of peptide mixture was analyzed by online nanoliquid chromatography (Ultimate 3000 RSLCnano) coupled to a Q-Exactive HF (Thermo Fisher Scientific). Peptides were concentrated on a precolumn (300 μm × 5 mm PepMap C18, Thermo Scientific) before separation on a 75 μm × 250 mm C18 column (Reprosil-Pur 120 C18-AQ, 1.9 μm; Dr Maisch GmbH) by application of a 120-min gradient with the following profiles: 0%–10% solvent B [0.1% (v/v) formic acid in 80% (v/v) acetonitrile] over 8 min, 12%–35% solvent B over 80 min, 36%–44% solvent B over 10 min, 45%–100% solvent B over 10 min, and finally 100% B for 10 min. Solvent A was 0.1% formic acid in water. Full-scan mass spectrometry (MS) spectra were collected in a mass range of m/z 350–1300 Th in the Orbitrap at a resolution of 60 000 at m/z = 200 Th after accumulation to an Automatic Gain Control (AGC) target value of 1e6 over a maximum injection time of 50 ms. In-source fragmentation was activated and set to 15 eV. The cycle time for the acquisition of MS/MS fragmentation scans was set to 2 s. Charge states accepted for MS/MS fragmentation was set to 3–8. Dynamic exclusion properties were set to n = 1 with an exclusion duration of 15 s. Stepped higher-energy collision dissociation fragmentation (MS/MS) was performed with increasing normalized collision energy (27%). Mass spectra were acquired in the Orbitrap at a resolution of 30 000 at m/z = 200 Th after accumulation to an AGC target value of 1e5 with an isolation window of m/z = 1.4 Th, and maximum injection time of 55 ms. Acquisition of MS and MS/MS data was controlled using Xcalibur (version 2.9; Thermo Fisher Scientific). Raw LC–MS/MS data were processed using MS-Fragger (v4.0) [45] integrated into FragPipe Graphical User Interface (v21.1). Sequences were identified by searching against a database populated with the T. pseudonana genome (taxonomy id 35128 downloaded from UniprotKB on 3 September 2024, 11 934 sequences), the T. oceanica genome (taxonomy id 159749 downloaded from UniprotKB on 4 October 2024, 34 461 sequences) or the genome for the whole Thalassiosira genus (taxonomy id 35127 downloaded from UniprotKB on 14 October 2024, 47 418 sequences). The latter database was used to retrieve a more complete list of proteins to compensate for the lack of a fully annotated T. pseudonana database. An integrated contaminant database was used to filter out artifacts due to common contaminant proteins. For all searches, enzyme specificity was set to Trypsin/P (C-terminal cleavage of lysine and arginine) with up to two missed cleavages allowed. The minimum peptide length was set to seven amino acids, and the maximum peptide mass was set to 6000 Da (~55 amino acids). For peptide searches, fixed carbamidomethylation of cysteine and up to two variable modifications (methionine oxidation and protein N-terminal acetylation) were allowed per peptide. For database searches, precursor and fragment ion mass errors were set to 10 and 20 ppm, respectively. Match between runs was allowed within a retention time window of 0.5 min. A false discovery rate (FDR) of 1% for peptide spectrum matches and proteins was applied. Statistical analysis was performed using ProStaR software [46] based on the quantitative data obtained for the three biological replicates analyzed per condition. Proteins identified in the contaminant database, proteins identified by MS/MS in less than two replicates of one condition, and proteins quantified in less than two replicates of each condition were discarded. After log2 transformation, abundance values were normalized using variance-stabilizing normalization before partially observed values were imputed using the Structured Least Squares Algorithm (SLSA). Proteins for which no values were obtained in one condition were discarded. Differentially expressed individual proteins (volcano plots) were selected using a log2 (fold change) cutoff of 0.65 and a P-value cutoff of .006 (T. pseudonana) or .01 (T. oceanica). These parameters provided an FDR of <1% for both species according to the Benjamini–Hochberg estimator. When considering all the proteins linked to a single biological process (e.g. Calvin cycle, Krebs cycle), significant differences were selected based on a P-value cutoff of .01, regardless of the fold change (Supplementary Dataset S1).

Statistical treatments

All statistical analyses were performed using the Rstatix and Rsmic packages in RStudio. P-values are indicated in the different figures and tables.

Data availability

R Codes for data analyses (proteomic, physiology, and proteomics) are available at https://zenodo.org/records/17748824.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [47] partner repository with the dataset identifier PXD071344.

The raw dataset and the corresponding segmented volumes have been deposited in the BioImage Archive (accession number S-BIAD3307, https://www.ebi.ac.uk/biostudies/bioimages/studies/S-BIAD3307).

Results

Mitochondrial architecture differs between Thalassiosira oceanica and Thalassiosira pseudonana

We first compared cell morphologies in T. oceanica and T. pseudonana cells grown in an iron-rich medium (Fig. 1 and Supplementary Fig. S3). Using FIB-SEM, we imaged thin lamellae from fixed cells and derived 3D structures. This analysis revealed that T. oceanica cells are larger than T. pseudonana cells (Fig. 1A–C). In both species, the frustule and ribs of the valves were visible at the working voxel size of 12 nm (Fig. 1A and B). All cells contained one large vacuole, occupying ~40% of the total cell volume, consistent with [12, 48] (Fig. 1D–F). The proportion of the cell volume occupied by all organelles—vacuole plus chloroplasts, mitochondria, and nucleus—was equivalent in the two species, at ~60% of the total cell volume (Fig. 1F). Chloroplasts and mitochondria were located at the cell periphery (Fig. 1G and H). The sub-organelle organization of chloroplasts was equivalent in both diatoms. Conversely, their mitochondrial architecture was distinct. Thus, the relative volume of the mitochondrial compartment and its surface-to-volume ratio were higher in T. pseudonana than in T. oceanica (Fig. 1I and L). Moreover, the two species had distinct mitochondrial shapes: spherical in T. oceanica (Fig. 1J), and elongated in T. pseudonana (Fig. 1K). However, the inner mitochondrial structures appeared to be the same in the two diatoms at the resolution used in the study (Fig. 1J and K).

Figure 1.

Mitochondrial morphology differs between T. oceanica and T. pseudonana; T. oceanica and T. pseudonana cells were observed after growth in an iron-rich medium; the 3D models shown are representative of at least nine cells from two to three independent biological replicates for each condition ( A and B) 3D scan view of cell morphology; voxel size = 12 nm, (c) absolute cell volume, derived from quantitative analysis of the 3D models; (Dand E) 3D models of the main subcellular compartments; green: chloroplast; yellow: mitochondria; blue: nucleus; pink: vacuole; gray: other cell structures and silica frustule; (F) volumes of the main subcellular compartments, derived from quantitative analysis of occupancy of the 3D models; error bars: SD; color code as in d and e; (G and H) 3D models highlighting chloroplasts (green), mitochondria (yellow), and cell frustule (gray); (I) total cell volume occupied by chloroplasts (green) and mitochondria (yellow), derived from quantitative analysis of the 3D models; error bars: SD Jj and K manually-annotated segmentation of mitochondrial and cristae volumes; (L) surface-to-volume ratio of total mitochondrial compartment in a single cell; (A, B, C, D, E, G, and H) scale bar: 2 μm; (J and K) scale bar = 0.5 μm; (C, F, I, and L); T.o.: T. oceanica; open circles; n = 10 cells; T.p.: T. pseudonana; open triangles; n = 11 cells; statistical analysis based on paired Welch’s t-test; ^***: P ≤ .001; ns: non-significant.

In addition to differences in shape, the subcellular location of mitochondria differed between T. pseudonana, where they formed a network appressed to the chloroplast envelope, and T. oceanica, where they were dispersed in the cytoplasm (Fig. 2A and B). Based on this organization, the total chloroplast surface in proximity to mitochondria, calculated at a maximal distance of 30 nm, as in Uwizeye et al. [43], was higher in T. pseudonana than in T. oceanica (Fig. 2C–F). This distinct subcellular arrangement of the two organelles may affect the physical interactions between chloroplasts and mitochondria.

Figure 2.

Chloroplast-to-mitochondria proximity differs between T. oceanica and T. pseudonana; (A and B) sections through cellular 3D volumes, segmented from FIB-SEM micrographs of whole cells, chloroplasts (green), mitochondria (yellow); scale bar: 1 μm; (C and D) 3D models of the cells from A, B showing chloroplasts (light gray) and mitochondria (dark gray), and their proximity (points at ≤30 nm are indicated by dark orange shading); scale bar = 2 μm; sections and models are representative of at least nine cells from nine to three independent biological replicates for each condition; (E) total chloroplast surface at <30 nm from mitochondria (%); T.o.: T. oceanica; open circles; n = 10 cells; T.p.: T. pseudonana; open triangles; n = 11 cells; statistical analysis by paired Welch’s t-test; ^***: P ≤ .001.

Iron limitation has distinct effects on cell physiology in Thalassiosira oceanica and Thalassiosira pseudonana

We next compared how the two diatom species cope with differences in iron availability during growth (Table 1). We selected the iron concentrations of iron-poor and iron-rich media based on previous reports [25, 28, 31, 49] and on the outcome of dissolved iron in the global ocean [35]. Selected iron concentrations allowed significant biomass accumulation (Supplementary Fig. S2). In iron-limited conditions, the growth rate of T. pseudonana was significantly reduced compared to that of T. oceanica (Table 1). The elemental composition (Mg, P, K, Mn, Fe, Cu, and Zn) of cells grown in iron-rich and iron-poor conditions was measured by ICP-MS. The iron content was normalized to Mg content, a relatively abundant element whose quantification is robust and reproducible (Table 1 and Supplementary Table S1). A similar trend was observed upon normalization relative to P content (Supplementary Table S1, [49, 50]).

Table 1.

Iron limitation effects on T. oceanica and T. pseudonana physiology.

	T. oceanica					T. pseudonana
	+Fe	–Fe	Absolute –Fe/+Fe variation (%)	(n)	P	+Fe	–Fe	Absolute –Fe/+Fe variation (%)	(n)	P
Intracellular Fe (mmol mol −1  Mg)	37.9 (4.9)	39.4 (3.8)	3.9	(3)	ns	61.9 (8.5)	39.7 (8.3)	36	(3)	*
Total chlorophylls a and c (fg cell −1 )	358 (120)	224 (79)	35	(7)	* *	240 (66)	169 (55)	31	(7)	* * *
Fucoxanthin (fg cell –1 )	320 (98)	227 (76)	27	(7)	*	153 (33)	117 (36)	23	(7)	* *
Maximum quantum efficiency F v /F m  (ru)	0.63 (0.06)	0.57 (0.04)	8.4	(8)	* *	0.56 (0.06)	0.49 (0.06)	10.6	(9)	* *
Photosynthetic ETR  (μmol photons m−2  s−1)	19.6 (3.3)	15.4 (2.2)	25.6	(8)	* *	16.3 (2.9)	12.0 (2.2)	30.5	(9)	* * * *
O 2  evolution in the light  (fmol min−1  cell−1)	2.39 (0.9)	1.75 (0.65)	25	(5)	.09	2.02 (0.31)	0.72 (0.3)	63	(4)	*
O 2  consumption in the dark (fmol min−1  cell−1)	0.50 (0.18)	0.4 (0.2)	24	(5)	.07	0.38 (0.02)	0.19 (0.02)	50	(5)	* * *
Growth rate (day −1 )	0.87 (0.12)	0.82 (0.12)	5.6	(11)	ns	1.02 (0.15)	0.66 (0.17)	35	(12)	* * *

T. oceanica and T. pseudonana cells were grown under iron-rich (+Fe) and iron-poor (–Fe) conditions. Absolute –Fe/+Fe variations were calculated as follows: 1 – (–Fe)/(+Fe) ^* 100. Values in parentheses correspond to SD. Light intensity during measurements: 120 to 133 μmol photons m^–2 s⁻¹. (n): number of cells imaged from independent biological replicates. (c) P: statistical significance as determined by paired Welch’s t-test;

^**** P ≤ .0001.

^*** P ≤ .001.

^** P ≤ .01.

^* P ≤ .05.

ns: non-significant.

In iron-rich conditions, the total iron quota was significantly higher in T. pseudonana compared to T. oceanica. In iron-poor conditions, conversely, the total iron quota decreased by 36% in T. pseudonana cells but not in T. oceanica cells (Table 1). These results suggest that the iron accumulation capacity of T. oceanica does not depend on the iron concentration in the growth medium, at least over the mild range tested in this study, conversely to Southern Ocean species in some natural conditions [36].

In iron-poor conditions, both strains had decreased total chlorophyll and fucoxanthin content compared to the iron-rich condition (~30% lower in T. pseudonana and ~25% lower in T. oceanica; Table 1). The maximum photochemical quantum yield of PSII (Fv/Fm) and the photosynthetic ETR were reduced to a similar extent. These observations reveal the impact of iron limitation on the photosynthetic machinery to be similar in both species. Nevertheless, O₂ evolution and consumption rates were differentially impaired in the two species studied. Thus, in the presence of limited iron, we calculated a 25% reduction in both evolution and consumption in T. oceanica, contrasting with a 63% reduction in evolution and a 50% reduction in consumption in T. pseudonana (Table 1). Differences in photosynthetic electron flow activity, as determined from the electron transfer rate (ETR) and O₂ evolution measurements, were also observed between species, suggesting possible alternative electron sinks.

Thus, overall, moderate iron limitation led to distinct physiological responses between the open-ocean and coastal Thalassiosira species.

In iron-limited Thalassiosira oceanica, functional and protein changes are associated with rearranged organelle morphology

From segmented FIB-SEM micrographs of iron-limited cells, we derived 3D models of chloroplasts and mitochondria, and determined their volumes (Fig. 3). These data revealed differences in chloroplast plasticity between the two species in response to limited iron availability. In T. oceanica, the average number of chloroplasts per cell was reduced from four to two under iron-poor conditions (Fig. 3A and B). The chloroplast-occupied cell volume was also decreased, whereas the volume of the mitochondrial compartment remained unaltered (Fig. 3C). Thus, based on organelle morphology, iron limitation has a more drastic effect on chloroplasts than on mitochondria in T. oceanica. In T. pseudonana, the number of chloroplasts remained constant at two per cell (Fig. 3D and E) and the volume occupancy of chloroplasts and mitochondria (Fig. 3F) was unaffected under iron-poor conditions. Moreover, the peculiar elongated mitochondrial architecture, and the proximity of chloroplasts and mitochondria were similar in all culture conditions, whatever the iron levels (Supplementary Fig. S4). Differences are visible in the subcellular architecture of the mitochondrial network between T. oceanica and T. pseudonana, likely reflecting an evolutionary response to the environment.

Figure 3.

Comparative effects of iron limitation on the subcellular morphology of T. oceanica and T. pseudonana; (A and D) 3D models from segmented FIB-SEM micrographs of chloroplasts (green), mitochondria (yellow); scale bar 2 μm; 3D models are representative of at least nine cells from two to three independent biological replicates for each condition; (B and E) distribution of cell population by number of chloroplasts per cell under iron-rich (+Fe) and iron-limited (-Fe) conditions; 2 chloroplasts: light green, 3 chloroplasts: medium green, 4 chloroplasts: dark green; n: total number of cells investigated; (C and F) total cell volume occupancy by chloroplasts (green) and mitochondria (yellow), as derived from quantitative analysis of the 3D models; bars indicate SD; T.o.:T. oceanica; circles; n = 10 cells; T.p.: T. pseudonana; triangles; n = 11 cells; statistical significance tested using paired Welch’s t-test; ^***: P ≤ .001; ns: non-significant.

To investigate possible whole-cell effects induced by iron limitation in the two cell types, we conducted a comparative proteomics analysis of proteins isolated from cells following growth in iron-poor or iron-rich conditions (Supplementary Dataset S1). This study led to the identification of 6133 proteins in T. oceanica and 5533 proteins in T. pseudonana (Supplementary Dataset S1).

After statistical evaluation, a subset of proteins (1628 in T. oceanica and 2263 in T. pseudonana) was reliably quantified and the comparison of the relative protein abundance was conducted (Fig. 4 and Supplementary Fig. S5). Overall, the relative abundances of the proteins detected were not extensively modified by iron limitation (88% in T. oceanica and 96% in T. pseudonana). In T. oceanica, 73 proteins increased in abundance (4.5% of total identified proteins) and 121 decreased (7.4%) (Supplementary Fig. S5 and Supplementary Dataset S1). In T. pseudonana, 30 proteins increased in abundance (1.3%) and 58 decreased (2.6%). In order to identify function-related patterns among these differentially expressed proteins, we clustered them according to their roles in chloroplasts and mitochondria, as detailed by Thimm et al. [51] (Fig. 4) and manually annotated the different pathways (Supplementary Dataset S1). In T. oceanica, in line with the physiological changes reported above, most of the iron-limitation-sensitive proteins were chloroplast proteins, with a specific reduction of the light-harvesting complexes (LHC) antenna and of proteins from the photosynthetic electron transport chain (ETC) including subunits of the iron-requiring photosystems I and II, and of cytochromes (Fig. 4). The iron-free plastocyanin electron carrier [26] was detected in T. oceanica, but its abundance was unaffected by iron availability during growth (Supplementary Fig. S5 and Supplementary Dataset S1). Expression of mitochondrial proteins (i.e. iron-rich ETC complexes and Krebs cycle enzymes) tended to be unchanged under iron limitation (Fig. 4 and Supplementary Fig. S5). Conversely, in T. pseudonana, the relative abundances of most of the photosynthetic ETC and LHC subunits and of mitochondrial ETC complexes were not significantly affected by iron limitation (Fig. 4).

Figure 4.

Proteome rearrangement in T. oceanica and T. pseudonana under iron limitation; the total proteomes of (A) T. oceanica (circles) and (B) T. pseudonana (triangles) in iron-poor and iron-rich conditions were analyzed by MS-based label-free quantitative proteomics (Fig. S5); the proteomic analysis was performed as single-shot whole-cell LC–MS/MS without offline fractionation; the plot displays the differential abundance (log₂FC iron-poor vs. iron-rich, y-axis) of proteins identified, clustered by their functional categories, as in Thimm et al. [51]; photo ETC: photosynthesis-specific electron transport chain proteins; LHC: light-harvesting complex proteins; CBB cycle: Calvin–Benson–Bassham cycle proteins; Resp ETC: respiration-specific electron transport chain proteins; TCA cycle: mitochondrial tricarboxylic acid cycle proteins; Fe metabolism: iron-metabolism-related proteins.; blue and red dots highlight proteins that were significantly depleted or enriched, respectively, in samples grown in iron-poor conditions as compared to samples from iron-rich cultures; thresholds applied were: a log₂FC cutoff of 0.3 (delimited by horizontal lines on plot) and a P-value cutoff of .01 (Limma P-value) for the analysis of proteins clustered by process; results presented are representative of four independent biological replicates; see Supplementary Dataset S1 for accession numbers of the proteins listed.

Photosynthesis and respiration are coupled differently in Thalassiosira oceanica and Thalassiosira pseudonana

The readjustment of the architecture of the chloroplast and mitochondrial networks, and the differences in physiological responses under iron limitation suggests differences in coupling between the photosynthetic and respiratory functions between the two species. We reasoned that, in the presence of mitochondrial dysfunction, the coupling between the organelles might result in reduced photosynthesis, e.g. lower electron transfer rate (ETR). We tested this possibility by exposing cells to a cocktail of mitochondrial inhibitors, including inhibitors of mitochondrial AOX (salycylhydroxamic acid) and complex III (antimycin A), to target both the cyanide-sensitive and cyanide-insensitive mitochondrial electron transfer pathways [22, 52]. We then measured the photosynthetic ETR, based on chloroplast fluorescence. In iron-rich conditions, in both species, the photosynthetic ETR was reduced in the presence of the respiratory inhibitors (Fig. 5). This suggests that both organisms have mechanisms to ensure metabolic exchange between chloroplasts and mitochondria. However, the effect was more significant in T. pseudonana than in T. oceanica. When iron availability was limited, mitochondrial inhibitors had an even more drastic effect in T. pseudonana (Fig. 5).

Figure 5.

Differences in functional plastid–mitochondria coupling between species could determine tolerance to Fe limitation; photosynthetic ETR estimated from the PSII operating efficiency in the presence of respiratory inhibitors (see materials and methods); light intensity during measurements: 45 μmol photons M⁻² s⁻¹ PPFD; data shown are representative of eight independent biological replicates in each condition; statistical significance was determined by applying a paired Welch’s t-test; ^**: P ≤ .01; ns: non-significant.

Discussion

In this study, we compared two closely related diatom species (T. oceanica and T. pseudonana) that occupy ecological niches distinguished by their iron bioavailability. In laboratory conditions, the two species were grown in media containing iron concentrations reflecting the levels measured at the two sites where the diatoms were isolated. To understand how they acclimate to the different environments, we characterized their morphology, their physiological responses, and the relative abundances of their proteins under conditions where iron was the only variable parameter.

Morphological assessment revealed distinct organelle architectures in the two diatoms: spherical mitochondria located independently of chloroplasts in T. oceanica, contrasting with elongated mitochondria appressed against chloroplasts in T. pseudonana (Figs 1 and 2). These features were unaffected by iron levels in culture media (Fig. 1 and Supplementary Fig. S4). Thus, they likely represent evolutionary adaptations to their distinct natural habitats (i.e. open-ocean and coastal waters, respectively), rather than reflecting acclimation responses.

In both diatoms under iron limitation, we observed a very similar sustained chlorosis and a decrease in photosynthetic and respiratory rates, although the effect on photosynthetic and respiratory rates was more pronounced in T. pseudonana (Table 1). The apparent contradiction between fluorescence- and oxygen-based measurements of photosynthetic activity under iron limitation can be explained by the activity of the plastoquinol terminal oxidase (PTOX), which catalyzes the flow of electrons from plastoquinol to O₂ during the chlororespiration process [53, 54]. This process contributes to the ETR but not to O₂ evolution, since PTOX converts oxygen to H₂O in stoichiometric amounts [55]. However, the PTOX protein was not detected in our study.

Despite the physiological responses observed, the global cellular iron content and growth rate were unchanged in T. oceanica (Table 1), suggesting that this species is better equipped to cope with iron limitation than T. pseudonana. Several strategies may be involved and likely combined: (i) maintenance of iron uptake capacities, (ii) optimization of the intracellular iron storage, possibly using specific Fe-containing proteins, and (iii) moderation of Fe requirements [12, 56, 57].

Analysis of whole-cell protein relative abundances indicated that iron limitation in T. oceanica leads to down-regulation of the photosynthetic electron transfer chain and light-harvesting components (Fig. 4). This finding is consistent with the decreased chloroplast number observed when iron availability is limited (Fig. 3), and could contribute to moderate iron requirements upon limitation. Our analysis also revealed that the abundance of the chloroplast and mitochondrial ETC components was unchanged by iron limitation in T. pseudonana (Fig. 4).

Altogether, our data suggest that, unlike T. pseudonana, T. oceanica favors respiratory processes over photosynthetic processes in conditions with moderate iron limitation. This hypothesis is in line with earlier findings in deeply-starved T. oceanica [3, 58] and other diatoms [21, 59, 60], but contrasts with observations on Chlamydomonas reinhardtii [18, 61].

The differences observed between the two species in response to iron limitation cannot be attributed solely to variations in the abundance of molecular components of the photosynthetic chain. Rather, the cells adapt to control their energy use. As chloroplasts were distinctly positioned relative to mitochondria in the two species (Figs 1 and 2), with the relative positions of the organelles likely influencing their metabolic crosstalk (Fig. 5), we therefore propose that some of the dynamic differences reported here could be linked to the constitutive intracellular topology. Indeed, the adaptive structural differences between the two species are conserved upon iron limitation (Fig. 1 and Supplementary Fig. S4). Overall, it appears that the phenotype is the outcome of both structural and metabolic differences.

In T. pseudonana, photosynthetic activity in the chloroplast is closely linked to mitochondrial activity (Fig. 5). This strong link is more than likely related to the specific subcellular organization of these cells, with the two organelles sharing a physical proximity that is not affected by iron limitation (Fig. 2 and Supplementary Fig. S4). The elongated shape of the mitochondria, a feature that is commonly observed in diatoms [62, 63] and other phototrophs [43], could be a determinant for this interaction. In cells grown with limited iron, the decrease in the total iron content tracks the decrease in total photosynthetic pigment content (Table 1). Both photosynthesis and respiration are severely decreased, and growth slows (Table 1). However, the energetic coupling between chloroplasts and mitochondria in iron-limited T. pseudonana cells (Fig. 5) could maintain a “high energy budget” in a natural environment where iron is sufficiently abundant. The increase in expression of mitochondrial Adenosine Tri Phosphate synthase components (Fig. 4) could be part of this response, boosting energy exchanges by providing more “exchangeable” mitochondrial-derived ATP [64]. In T. pseudonana cells grown in iron-poor medium, where the iron-demanding respiration activity is extensively diminished (Table 1), the exchanges between chloroplasts and mitochondria would only partially sustain the energy demands of photosynthesis. Therefore, the growth rate, which is mostly based on this energetic interaction, would ultimately collapse (Table 1).

In T. oceanica, unlike any other diatom species studied so far [62], mitochondria are spherical and have minimal physical contact with chloroplasts, with their shape creating geometric constraints on any potential interactions (Fig. 2 and Supplementary Fig. S4). As a result of the lack of physical contact, metabolic synergy is hampered, and energy exchanges between chloroplasts and mitochondria must occur by less efficient processes, such as diffusion or indirect exchanges (Fig. 5). In line with this hypothesis, our results indicated that photosynthesis depends less on mitochondrial function and thus on iron bioavailability in this species (Fig. 5 and Table 1). In T. oceanica, limited iron availability triggers a reduction in the chloroplast compartment and its related components (Figs 3 and 4), whereas the rate of respiration is minimally reduced, prioritizing maintenance of peak growth rates. The conservation of intracellular iron levels and iron-independent compensatory mechanisms, including constitutive expression of plastocyanin and reduced PSI to PSII ratios [25, 26] (Supplementary Dataset S1), appears to provide further compensatory adaptations. Overall, the limited interdependence of chloroplasts and mitochondria and their reduced reliance on iron for metabolic exchanges appear to reinforce these constitutive modifications. This functional adaptation strategy allows T. oceanica to maintain the relative homeostasis of its cellular energy processes, and, by reducing the energy demands of growth, is apparently sufficient to promote growth in the open-ocean environment [65, 66].

Based on our results, we conclude that: (a) the photosynthesis rate is more dependent on mitochondrial function in T. pseudonana than in T. oceanica, and (b) this interdependency is linked to iron availability in T. pseudonana, consistent with enhanced chloroplast–mitochondria synergy in this diatom. Because the iron concentrations used in our comparative analysis were chosen based on those observed in natural conditions, we believe that the conclusions of our in-laboratory study also apply in these natural oceanic environments.

Supplementary material

Supplementary material is available at ISME Communications online.

Conflicts of interest

None declared.

Funding

We acknowledge funding support from the European Research Council (ERC) through advanced grant “ChloroMito” (833184), and through the European Plankt-ON (101099192) project. This work used the platforms of the Grenoble Instruct-ERIC center (ISBG; UAR 3518 CNRS-CEA-UGA-EMBL) within the Grenoble Partnership for Structural Biology (PSB), supported by FRISBI (ANR-10-INBS-0005-02) and Labex GRAL (ANR-10-LABX-49-01), financed within the Université Grenoble Alpes graduate school (École Universitaire de Recherche) CBH-EUR-GS (ANR-17-EURE-0003). The IBS Electron Microscope facility is supported by the Auvergne Rhône-Alpes Region, the Fonds Feder, the Fondation pour la Recherche Médicale, and GIS-IBiSA.