Regulation of mixotrophy in Synechocystis by a rhomboid protease

Abstract

The intramembrane ‘rhomboid’ protease family is almost ubiquitous across evolution, with its well-conserved transmembrane domains typified in crystal structures of bacterial representatives, such as the Escherichia coli GlpG. In contrast with accumulating data on rhomboid function in higher organisms, where roles in human disease are an incentive for study, findings remain sparse about the functions and substrates of the prokaryotic enzymes, even though these provided the earliest protein structures. In particular, nothing at all is known about the rhomboid proteases of photosynthetic prokaryotes despite the importance of cyanobacteria as relatives of the progenitor of chloroplasts. Findings relating to the cyanobacterial enzymes would complement data on plant plastid rhomboids from work in Arabidopsis thaliana. Synechocystis sp. PCC 6803 was used, therefore, to investigate conserved photosynthetic functions across evolution for this protein family. Reverse-genetics studies using Slr1461, the single rhomboid protease of Synechocystis 6803, did not reveal a non-photochemical quenching phenotype as observed for the Arabidopsis RBL10 null mutant, which lacked a chloroplast outer membrane rhomboid. The Slr1461 mutant exhibited a marginal change in pigment composition, and its growth rate was only slightly different from that of WT under optimal light intensity. The most dramatic effect of the inactivation of Slr1461 was the mutant’s distinct inability to reduce photosynthetic activity under mixotrophic conditions. Concurrent phototrophy and heterotrophy in mixotrophic growth aids survival and competitiveness in phytoplankton, allowing conservation of energy by reducing the need for uptake and fixing of CO₂ when an organic carbon source is available. It was notable, therefore, that, in the absence of the Slr1461 rhomboid, the steady-state mRNA levels were reduced for a subset of genes encoding facilitators of high-affinity CO₂ import and of transcriptional regulators of the carbon-concentrating mechanism (CCM). Slr1461 activity was also linked with that of another membrane protease, the AAA protease FtsH2, which was likewise observed to act within regulatory networks for the cyanobacterial carbon uptake mechanism. Aberrant transcript levels were most evident specifically under high CO₂ conditions, when the impact of Slr1461 enzymatic activity appeared to be upstream of NdhR, a central, controlling transcription factor of the CCM.

Introduction

The membrane-located families were the most recently discovered group of proteases, and investigations of the large ‘rhomboid’ class quickly proved their importance across diverse cellular processes in all kingdoms of life. They are ubiquitous across evolution, rhomboid proteases being found in prokaryotic to eukaryotic microbes, as organellar members in mitochondria and plastids, and also in the metazoa [1]. Although this early paradigm has become less fixed with accumulating research, a typical rhomboid substrate is typically a single transmembrane protein located in the lipid bilayer, from which proteolysis releases an activated peptide. A well-researched example of this is Drosophila Rhomboid (Rho)-1, which releases epidermal growth factor (EGF) receptor for signalling [2]. Substrate and enzyme are initially localized separately, on the endoplasmic reticulum (ER) and the Golgi apparatus, respectively [3], a mechanism that prevents cleavage of the substrate, Spitz, until it is trafficked to the ER.

Accumulating evidence for a more diverse repertoire for rhomboid action was presaged by an atypical mode of action observed when an activated intermediate was found to be retained in the membrane [4] (rather than released as an activated peptide) in the case of the Providencia stuartii AarA rhomboid protease. In that case, quorum signalling was permitted following AarA activation of the membrane-located TatA component of the Tat transport channel by cleaving the N-terminal six amino acids [5].

Rhomboid proteases are atypical serine proteases whose conserved structure of a six- or seven-transmembrane domain core contains only a catalytic dyad, comprising Ser and His without the usual third active-site residue, Asp [6,7]. Whereas crystal structures were resolved for the Haemophilus influenzae and Escherichia coli bacterial enzymes, the roles of prokaryotic rhomboids are only slowly being elucidated [8]. A hypothesis that E. coli GlpG may facilitate fatty acid utilization, enhancing survival in intestinal mucus [9], does correspond with other links between rhomboids and lipids. For example, both the Corynebacterium glutamicum [10] and Shigella sonnei rhomboid pairs were suggested to affect membrane quality control [11], whereas two organellar rhomboid proteases were hypothesized to regulate membrane-fusion dynamics, in the mitochondria of Saccharomyces cerevisiae and Drosophila melanogaster [12,13].

Another eukaryotic, organellar representative is the presenilin-associated rhomboid-like protease of the inner mitochondrial membrane, an enzyme that operates within human mitochondrial-disease networks [14]. Rhomboid proteins also have medical significance in their regulation of host-cell infection in the apicomplexan genera Toxoplasma and Plasmodium [15,16].

A catalytically inactive subset of rhomboids (iRhoms, or pseudoproteases), which lack protease active-site residues, can still play roles in signal transduction. For example, overexpression of Drosophila iRhom inhibited EGF receptor-mediated signalling [17], whereas human iRhom2 is required for trafficking and activation of the TNF-α-converting enzyme for TNF signalling [18].

One or two rhomboid genes, only, tend to be found within prokaryotic genomes, whereas plant genomes encode multigene families of the proteases [19], but few, again, have been studied. Given the prokaryotic origin of eukaryotic organelles, investigating the chloroplast or mitochondrial rhomboid proteases of plants might allow evolutionary links in the enzyme family to be revealed. This type of parallel study was useful for chloroplast versus cyanobacterium FtsH-family membrane proteases, with Arabidopsis thaliana VAR2 and Synechocystis sp. PCC 6803 Slr0228 (FtsH2) enzymes, both discovered to be operating in the photosystem II (PSII) D1 protein replacement cycle [20,21]. Compared with four of 12 FtsH enzymes thought to function in A. thaliana plastids [22,23], 3 of about 20 rhomboid proteins in the plant have been demonstrated to be targeted to organelles. In transient assays, Arabidopsis At1g18600 (RBL12) was observed in the mitochondria, although this protein did not complement yeast mitochondrial rhomboid mutant function [24]. Arabidopsis At5g25752 (RBL11) was directed to the chloroplast [25], where it was hypothesized to function in plastid-protein translocation [26]. A stably transformed RBL10-GFP reporter was visualized in the chloroplast outer membrane, and rbl10 null-mutants demonstrated floral abnormalities, reduced fertility and a photosynthetic phenotype correlated with increasing light intensity [27].

Identifying cyanobacterial rhomboid proteases was of interest because much of their photosynthetic machinery is conserved in plant chloroplasts. As above, these oxygenic photoautotrophs were already found to share some proteolytic mechanisms, but aquatic cyanobacteria encounter distinct environmental challenges. One such is the acquisition of carbon dioxide, which diffuses poorly in water [28]. Cyanobacteria have overcome this constraint, which coexists alongside the imperfect affinity of ribulose bisphosphate carboxylase-oxygenase (RubisCO) for CO₂ and its wasteful side reaction with oxygen [29], through the evolution of a specialized carbon-concentrating mechanism (CCM). Carboxysome ‘organelles’ house the cyanobacterial RubisCO [30,32] and enrich the CO₂ environment. Because the CCM is energy-dependent, it is regulated at multiple levels, including transcriptional control of CO₂-uptake genes (for review, see [33]). Regulatory components include both activator and repressor transcription factors [34] and, notably, proteolysis of the central NdhR transcription factor (TF) by FtsH2 [35].

Here, mutant cells lacking the Slr1461 rhomboid protease were observed to aberrantly regulate oxygen evolution during mixotrophic growth, when the need for carbon fixation from CO₂ was reduced. The possibility of this rhomboid protease affecting the regulation of the CCM, in addition to the activity of the FtsH protease, was considered by investigating transcription of CCM TF genes, including NdhR, CmpR and an AbrB family member in Δslr1461 Synechocystis. Whereas there was no indication of transporters being misregulated at the level of steady-state mRNA, the Δslr1461 rhomboid-null mutant exhibited altered levels of TF transcripts when CO₂ concentrations were raised. There is increasing interest in TF activated by proteolytic cleavage in other systems [36], and transcriptional links also found between slr1461 and ftsH2 were intriguing when considered alongside a recent report of a Bacillus bacterial rhomboid cooperating with FtsH in degradation of a transporter protein [37]. This study of the Synechocystis rhomboid, therefore, adds a further protein regulator to control networks for bacterial oxygenic photosynthesis and provides a route for further study of coordinated proteolysis events.

Methods

Inactivation of Slr1461

Glucose-tolerant (GT) Synechocystis sp. PCC 6803 cells from 25 to 50 ml cultures in the exponential growth phase (OD₇₅₀ ~0.5) were harvested by centrifugation, resuspended in 100 µl fresh BG11 medium [38] and mixed with 1–2 µg of plasmid DNA for insertional inactivation by homologous recombination of an antibiotic-resistance gene-interrupted slr1461 ORF. Following incubation at 30 °C for 4 h at 8 µmol photons m⁻² s⁻¹ in a stationary illuminated incubator, with occasional inversion of the tube, cells were plated onto non-selective BG11-agar and incubated overnight. After 24 h, plates were overlaid with 3 ml of 0.6 % agar containing selection antibiotic (i.e. 15 µl of 100 mg ml⁻¹ kanamycin per 3 ml of 0.6 % agar). Antibiotic-resistant transformants were picked as single colonies after ~2 weeks and plated on fresh selective agar plates at least twice to ensure homoplasmicity of the polyploid genome before confirming this by PCR (Fig. S1, available in the online Supplementary Material). The pEERM4 vector (Addgene plasmid #64026), which integrates into the chromosomal neutral site 2, was used to construct the Slr1461 complementation plasmid. pEERM4 plasmid contains a nickel-inducible rnsB promoter [39]. The full-length slr1461 coding sequence was cloned into pEERM4 using XbaI and PstI restriction endonucleases to generate pEERM_Slr1461. This construct was subsequently transformed into the Δslr1461 mutant to enable integration at neutral site 2.

Oxygen evolution

Synechocystis 6803 [glucose-tolerant (GT)] cells were harvested once they reached an OD₇₅₀ of ~0.4–0.5 by centrifuging at 2,000 g for 5 min at room temperature (22 °C) and were then adjusted to an OD₇₅₀ of ~0.5 with fresh BG11 medium. For mixotrophic cultures, the pellet was resuspended in fresh BG11 medium containing 5 mM glucose. NaHCO₃ was added to a final concentration of 10 mM, 5 min before the start of measurements. The rates of net O₂ evolution and dark respiration were then measured [40] at 30 °C using a Clark-type electrode (Hansatech, Kings Lynn, UK [41]). Sodium dithionite was used to calibrate the electrode. Actinic light was provided using a 650-nm red LED light. Variable light intensity was set for 0–2300 µmol photons m⁻² s⁻¹. After an initial 30-min dark adaptation, O₂ evolution was measured for 5 min, followed by dark respiration for 20 min. The mean net rate of photosynthesis was then determined from the oxygen gradient over 5 min. Dark respiration was determined by following the same procedure, except that it was calculated with data from the last 5 min of the 20 min experiment.

Room temperature fluorescence

Maximum quantum yield of PSII was measured using a Walz (Effeltrich, Germany) pulse-amplitude modulation (PAM) 101 fluorimeter with the 101-ED emitter-detector unit. A Walz liquid cell adaptor maintained at 30 °C was used to contain the sample. Actinic light was white light (50 µmol photons m⁻² s⁻¹) behind a 420-nm Corning 4–96 glass filter (cut-on wavelength at 600 nm). Far-red light was provided by PAM-102. Prior to measurement, cells were harvested as before and concentrated to 10 µg ml⁻¹ chlorophyll and dark-adapted for 5 min. Minimum fluorescence (F₀) was determined by illuminating with measuring light (0.01 µmol photons m⁻² s⁻¹). Maximum fluorescence (F_m) for the dark-adapted state was determined with a 0.8 s long saturation (4500 µmol photons m⁻² s⁻¹) pulse delivered from a Schott (Elmsford, USA) KL 1500 white light source. The actinic light was then switched on and a second saturation pulse was applied after 5 min (F’_m). Finally, maximum fluorescence (F_m) was determined following the injection of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) at a 10 µM final concentration to inhibit electron transport from PSII while actinic light was on [42].

Low-temperature fluorescence

77 K fluorescence emission spectra from the WT and Δslr1461 insertional inactivation mutant were recorded in liquid nitrogen using a Perkin Elmer (Waltham, MA, USA) LS 55 luminescence spectrometer. All samples were dark-adapted for 5 min before freezing. The excitation was 440 nm and 580 nm (5 nm slit width), and the fluorescence emission was scanned in a wavelength range of 600–750 nm.

Growth assays

Synechocystis 6803 (GT) cultures were treated with 2.5 µM NiCl₂ and incubated at 30 °C for 3 h with shaking to induce slr1461 gene expression in the complemented strain. For consistent growth conditions, the four replicates of WT and Δslr1461 cells were also treated with 2.5 µM NiCl₂ as above. The cells were then diluted ~22-fold to an OD₇₅₀ of 0.1 in 50 ml BG11 medium containing 5 mM glucose and incubated at a light intensity of 16, 81 or 150 µmol photons m⁻² s⁻¹ for ~6 days. A 1 ml sample was removed every 24 h for analysis. Specific growth rate (µ) and doubling time (Td) of all cultures were calculated according to the recorded OD₇₅₀ by the photospectrometer (Lambda U3900; Hitachi) according to:

$${\displaystyle {\displaystyle \mu =\frac{\ln O{D}_t-\ln O{D}_0}{\mathrm{\Delta }t}}}$$

$${\displaystyle {\displaystyle Td=\frac{\ln 2}{\mu }}}$$

where OD_t and OD₀ refer to optical density at time t (h) and time zero, respectively. Chlorophyll content was determined by subtracting the OD₇₅₀ nm value from the OD₆₈₀ nm value and multiplying the result by 10.854, as previously described [43].

RNA isolation and quantitative real-time PCR

Four independent cultures of each strain were grown at a light intensity of 30 µmol photons m⁻² s⁻¹ in ‘low’ CO₂ (i.e. normal atmospheric levels) conditions to an OD₇₅₀ of 0.4–0.5. Samples were either treated with 5 mM glucose (for mixotrophic growth), bubbled with 5 % CO₂ (raised CO₂ condition) or left untreated (low CO₂ condition). After 6 h, 40 ml samples were collected and pelleted at 3,000 g before being stored at −80 °C for RNA extraction. Total RNA was extracted with QIAGEN (Hilden, Germany) RNeasy Plant Mini Kit. RNA was treated with DNase I, RNase-free (Merck Life Sciences, Gillingham, UK) to eliminate possible genomic DNA contamination. First-strand cDNA was synthesized from 1 µg of RNA with the RevertAid First Strand cDNA Synthesis Kit (Fisher Scientific, Loughborough, UK). The cDNA was diluted 100-fold, and 1 µl was used in quantitative real-time PCR, performed with PowerUp (Applied Biosystems, Waltham, USA) SYBR green kit. The values for target genes’ transcript level were normalized to both total RNA and to housekeeping gene 16S rRNA (for primer pairs, see Table 1); the relative changes in transcript abundance were analysed using the 2^−ΔCt method [44,45].

Table 1. Primer pairs used in quantitative real-time PCR and slr1461 complementation.

For quantitative real-time PCR
slr1461_F: TCTTTGCACCCTTTCTCCAT slr1461_R: TTTGGAACCAACCTCGAAAG
ftsH2_F: GCAAGTTAGCGACCCAGAAG ftsH2_R: CCCACAAAACCCCAAACCAT
sbtA_F: TGGTGCTCTGTATCCCTTTATG sbtA_R: TGCTGAACGATTCCTCAATACT
ndhF3_F: TTTGGCCTTAATTGCTGGAC ndhF3_R: GCGATCGCAATTAAAGAAGC
16S rRNA_F: CACACTGGGACTGAGACAC 16S rRNA_R: CTGCTGGCACGGAGTTAG
abrB2_F: CTGCCCCGGGTCAATATGAT abrB2_R: TTGGATTGCACGCTAATGCG
cmpR_F: GTGATTGCTGACCTCCAGGG cmpR_R: CCCAAAAGTCGTGGCACAAA
For complementation
slr1461_F: CTCTAGACCACCATGAGCCAAAATTCC slr1461_R: CTGCAGTGCTAATTTTTGCTGTAGTAATGC

Sequence analysis

Sequence similarity searching was carried out with blastp and blastn using the National Center for Biotechnology Information public database (https://www.ncbi.nlm.nih.gov/) [46]. Multiple alignments were generated with clustal omega [47], and alignments were edited with Jalview [48]. Active site and expected transmembrane architecture were predicted using hydropathy plots (TMHMM) [49] and AlphaFold3 [50].

Results

Identification of rhomboid proteases in cyanobacterial lineages

A single rhomboid protease encoded by the Synechocystis 6803 genome was identified from sequence searches, and active site and expected transmembrane architecture were predicted (Fig. 1a–c). Almost all sequenced cyanobacterial genomes were found to encode a single rhomboid (Fig. 1d), but there was some variation, notably within the reduced genomes of the Prochlorococcus genus. In contrast, the genome of Acaryochloris marina MBIC 11017 includes six rhomboid protease genes, four of which were predicted for Gloeothece citriformis PCC 7424 (also known as Cyanothece sp. PCC 7424) and Leptolyngbya sp. CCNP 1308 and three in Trichormus variabilis ATCC 29413 (previously named Anabaena varibilis ATCC 29413), Nostoc punctiforme PCC 73102, Crocosphaera subtropica ATCC 51142 and Picosynechococcus sp. PCC 7002 (previously named Synechococcus sp. PCC 7002/Agmenellum quadruplicatum PR-6). In contrast, Prochlorococcus marinus str. MIT 9313 and Rippkaea orientalis PCC 8801 were more typical of a fairly common subset predicted to have duplicated rather than single rhomboid representatives. Examples of cyanobacterial genomes that have evolved rapidly over evolution and lost their rhomboid gene were found to include Thermosynechococcus elongatus BP-1, Trichodesmium erythraeum IMS101 and P. marinus str. MIT 9312 and MIT 9211 [51], in tandem with a reduced complement of genes for regulatory components such as sigma factors, required for transcription initiation, light sensors/transducers and two-component sensor-kinases [51].

Fig. 1. Slr1461 motifs for rhomboid catalytic activity. (a) Slr1461 aligned with plant and human rhomboid aa sequences. Colour, aa chemical properties and degree of conservation (standard clustalx scheme [47]: purple, acidic amino acids; blue, hydrophobic; green, polar and neutral; brown, glycine; khaki, proline). Segments of sequences aligned are from predicted full-length proteins as follows: Synechocystis Slr1461 aa27–196; A. thaliana enzymatically active rhomboids RBL2, aa64−252/plastid RBL9, aa275−375/plastid RBL10, 128–310/plastid RBL11, 85–272; H. sapiens catalytically active RHBDL2, aa73–273; H. sapiens catalytically inactive iRhom2, aa605–796; and E. coli GlpG, aa97–276. Bar above sequence alignment indicates conserved aa W/YR, S and H. (b) Slr1461, Predicted transmembrane domains (TMHMM) [49]. (c) AlphaFold 3-predicted structure of full-length Synechocystis sp. PCC 6803 Slr1461 protein [50]. (d) Distribution of rhomboid proteases in cyanobacteria: unrooted phylogenetic tree for 16S rRNA generated using the neighbour-joining method. The presence or absence of rhomboid (Rho) protein in each cyanobacterial strain is shown on the right. ‘No Rho’, absence of rhomboid protein.

Synechocystis rhomboid sequence features

The Synechocystis 6803 rhomboid gene, ORF slr1461, translates to an amino acid sequence with 37 % and 26 % identity to the A. thaliana chloroplast rhomboids At5g25752/RBL11 and At1g25290/RBL10, respectively. It also shows 28 % identity with the endoplasmic reticulum–localized At1g63120/RBL2, but no significant similarity was found with the mitochondrial At5g38510/RBL9.

To predict whether the Slr1461 protein was likely to be enzymatically active or was an inactive iRhom-type rhomboid, its sequence was aligned with representative enzymes already experimentally verified to be catalytically active from A. thaliana [52], Homo sapiens and E. coli [53] and against the catalytically inactive H. sapiens iRhom2 rhomboid protease [54,55] (Fig. 1a). The Slr1461 sequence aligned well with catalytically active rhomboids across transmembrane domains, with clearly conserved active-site serine and histidine residues (Fig. 1a) [56]. These residues were similarly conserved in the active A. thaliana rhomboid family (At1g63120/RBL2, mitochondrial At5g38510/RBL9 and chloroplast-located At1g25290/RBL10), H. sapiens RHBDL2 and E. coli GlpG. Although the overall sequence of human iRhom2 also aligned reasonably well (25 % identity between Slr1461 and iRhom2), its corresponding positions lack these catalytic residues, consistent with iRhom2’s known enzymatic inactivity [18]. Slr1461, therefore, displays features of the catalytically active rhomboids. It is worthy of note, however, that it lacks a commonly conserved rhomboid N-terminal tryptophan–tyrosine/arginine (W/YR) motif (Figs 1a and S2) [55,57], Which was hypothesized at one time to influence the rate of proteolysis [58].

Photosynthesis phenotype

The A. thaliana chloroplast RBL10 mutant’s significantly higher NPQ values than in WT plants, correlated with increasing light intensity [27], prompted exploration of a range of photosynthetic parameters in Δslr1461 Synechocystis. Whole-cell absorbance spectra of Δslr1461 and WT cells were broadly similar under low-intensity light (Fig. 2), but spectra normalized to the phycocyanin peak (635 nm) revealed a slightly higher absorbance at 687 nm in Δslr1461, consistent with an increased chlorophyll a contribution relative to phycobiliproteins. Quantification of pigment confirmed that Δslr1461 cultures contained significantly higher total chlorophyll a than WT (Δslr1461 mean±sem, 8.7±0.51 µM vs. WT 6.1±0.08 µM; t-test P=0.024, 7 d.f.), supported by ANOVA (see below, Table 2). Oxygen evolution in low light-grown Synechocystis WT vs. Δslr1461 cells was not significantly different if cultures were grown in standard BG11 autotrophic medium with saturating bicarbonate (Fig. 3a). Likewise, the PAM fluorometry assays of Δslr1461 Fv/Fm, PSII efficiency and NPQ produced measurements not significantly different from WT under low-intensity light conditions (Table 3).

Fig. 2. Photosystems and pigments in 77K fluorescence emission and absorption spectra of mixotrophic WT vs Δslr1461 cells. (a) 77K spectra normalized to the emission band at 650 nm (580 nm excitation); (b) 77K spectra normalized to the emission band at 725 nm (440 nm excitation); (c) absorption spectra normalized at 687 nm. Maximum absorbance at 448 nm, 635 nm and 687 nm corresponds to carotenoid, phycocyanin and chlorophyll peaks, respectively. Cultures for A-C grown in BG11 medium supplemented with 5 mM glucose at 8 µmol photons m⁻² s⁻¹ light. Green line, Δslr1461; purple line, WT Synechocystis sp. PCC 6803.

Table 2. Two-way ANOVA to analyse the effects of WT vs. Δslr1461 genotype and light intensity, and their interaction, on doubling time and chlorophyll a (Chl a) content.

	Genotype			Light intensity			Genotype vs light intensity
	Df	F	p	Df	F	p	Df	F	p
Doubling time	2	3.2	0.0551	2	30	<0.0001	4	1.9	0.1287
Chl a	2	51	<0.0001	2	17	<0.0001	4	1.0	0.3927

Data were transformed (log, square root or reciprocal) as needed to meet linear model assumptions. Df, degrees of freedom; F, variation between sample means; p, significance levels.

Fig. 3. Light-response curve for WT vs. Δslr1461 cells. Oxygen evolution rate versus photon flux density (PFD) for autotrophic (a) vs. mixotrophic (b) cultures grown at 8 µmol photons m⁻² s⁻¹ light. Cells grown mixotrophically were supplemented with 5 mM glucose. Net photosynthesis rate (measured as in the ‘Methods’ section) datapoints are the mean of three measurements±sem. For statistics, an unpaired Student’s t-test was performed. Asterisks, Statistically significant changes (P-value <0.05).

Table 3. Photosynthesis in the Synechocystis Δslr1461 rhomboid protease mutant.

	Autotrophic				Mixotrophic				High light autotrophic
	Fv/Fm	Fv/Fm DCMU	ΦPSII	NPQ	Fv/Fm	Fv/Fm DCMU	ΦPSII	NPQ	Fv/Fm	Fv/Fm DCMU	ΦPSII	NPQ
WT	0.38(±0.04)	0.56(±0.01)	0.35(±0.02)	0.25(±0.02)	0.30(±0.03)	0.49(±0.01)	0.20(±0.06)	0.13(±0.06)	0.26(±0.02)	0.32(±0.02)	0.14(±0.03)	0.09(±0.01)
Δslr1461	0.37(±0.02)	0.58(±0.0)	0.37(±0.04)	0.26(±0.02)	0.32(±0.04)	0.49(±0.03)	0.17(±0.06)	0.13(±0.06)	0.31(±0.00)	0.36(±0.01)	0.13(±0.01)	0.10(±0.02)

Pulse-modulated chlorophyll fluorescence measurements of quantum yield of photosystem II (PSII): photochemical quantum yield F_v/F_m for dark-adapted cells was calculated from (F_m-F₀)/F_m. Maximum quantum yield of PSII (F_v/F_m DCMU) was measured using a protocol employing continuous illumination in the presence of 10 µM DCMU. The effective quantum yield of PSII (Φ_PSII) was calculated as (F_m-F_t)/F_m, and non-photochemical quenching (NPQ) was calculated from (F_m-F_m)/F_m. Results are means from three biological measurements±sd.

GT Synechocystis strains are useful for mutant studies because genes encoding key photosynthetic components can be knocked out, but cultures are able to remain viable [59] if the BG11 defined medium is supplemented to permit heterotrophy. GT cells can, therefore, acquire inorganic carbon either from CO₂ during autotrophic growth or, alternatively, from both glucose and CO₂ in mixotrophic growth. The measurement of oxygen evolution rate under such mixotrophic conditions revealed the most distinct phenotype for glucose-tolerant Δslr1461 Synechocystis. The rate of photosynthesis, previously identical in autotrophic cultures of Δslr1461 and WT (Fig. 3a), was regulated down by about a third in WT cells when grown mixotrophically, as expected. This reduction, however, did not occur in mixotrophic Δslr1461 cells (Fig. 3b). Importantly, dark respiration rates were comparable between the mutant and WT strains (Fig. S5), indicating that the observed differences in O₂ evolution under mixotrophy reflected changes in photosynthetic regulation rather than respiratory activity.

As cyanobacteria divide more rapidly in the presence of glucose [60], a marked difference like this in photosynthetic activity (Fig. 3b) might arise if there was a defect in glucose import in the absence of the Slr1461 rhomboid. Growth rates of the strains did not suggest altered nutrient acquisition; however, statistical analysis using two-way ANOVA (Table 2, Fig. S4) confirmed that, whereas increasing light intensity significantly affected both doubling time (F=30, P≤0.0001) and chlorophyll a content (F=17, P≤0.0001), the genotype (i.e. mutant vs WT) only influenced chlorophyll a (F=51, P=0.0001), with the lack of Slr1461 having little effect on doubling time (F=3.2, P=0.0551). A check on experimental conditions showed there was no significant interaction between genotype and light intensity (P>0.12).

Genetic regulation of CO₂ concentration components

Mutant Δslr1461 cells failed to downregulate photosynthesis under mixotrophic growth, when a decrease in photosynthetic activity would normally be observed in Synechocystis strains that are capable of heterotrophy. The reduced photosynthesis in WT GT cells follows a decrease in the rate of bicarbonate uptake [61] and metabolic regulation that reduces the activity of the carbon-fixing enzyme, RuBisCO [62]. A hypothesis that lack of Slr1461 resulted in altered behaviour of the CCM was, therefore, tested by quantifying the steady-state mRNA of slr1461 and of key CCM genes.

Whether the cyanobacterial rhomboid protease itself was regulated at the level of transcript abundance under high CO₂ conditions was investigated by quantifying slr1461 mRNA level in WT Synechocystis (Fig. 4a). slr1461 transcript level did indeed increase when cultures were bubbled with 5 % CO₂, when cells also had higher chlorophyll a content than did WT, as noted previously (Table 2). These data indicated that the Slr1461 protease may be involved in the regulation of carbon assimilation.

Fig. 4. Effect of high CO₂ on CCM gene transcript level in (a) WT or (b, c) Δslr1461 compared with WT. Strains grown in atmospheric-level CO₂ (low CO₂) at 30 µmol photons m⁻² s⁻¹ light intensity until OD₇₅₀ of 0.4–0.5, after which they were bubbled with 5 % CO₂ (high CO₂) or left untreated before further 6 h incubation. Quantitative real-time PCR on extracted RNA permitted calculation of expression levels relative to 16S rRNA reference gene, with further normalization to the WT (Fig. 4b–c, Table S2). Statistical significance (P-value≤0.05), denoted by *≤ 0.05, **0.01 and ****0.0001. For a complete list of P-values, see Table S3.

Genes encoding components of the CCM were tested, therefore, during adaptation of Δslr1461 cultures to high CO₂ or to mixotrophic conditions, to assess their transcription levels when carbon fixation demand was reduced (Fig. 4, Table S2). The sbtA gene, encoding the Na⁺-dependent HCO₃^- transporter, was aberrantly low under all conditions in Δslr1461: for example, the mutant had a log₂-fold sbtA level of −0.63 (P=0.014) vs WT under standard (lower) CO₂ conditions, moving to a −1.01 log₂-fold value (P=0.015) under 5 % (raised) CO₂ (Fig. 4b, Tables S2 and S3).

Also aberrant was the transcript for the NdhF3 protein, a component of the cyanobacterial high-affinity CO₂ uptake system (comprising a complex of NdhD3/D4 and NdhF3/F4), which catalyses CO₂ uptake by converting CO₂ into HCO₃^-. In this case, in mixotrophic Δslr1461 cultures, ndhF3 mRNA levels in the mutant were similar to those in WT (Table S2). In autotrophy, however, under normal atmospheric CO₂ conditions, ndhF3 mRNA log₂-fold levels were −5.90 and in raised CO₂ conditions were −4.57 (Fig. 4a, Table S2), i.e. lower than in Δslr1461 than in WT in both conditions. Likewise, slr1461 itself had been observed to accumulate only under raised CO₂, and not under mixotrophic glucose-supplemented conditions (Fig. 4a, Table S2), a CO₂-dependent regulation of slr1461 was reported before [33]. Therefore, the lack of difference in ndhF3 transcript level between WT and Δslr1461 mutant in mixotrophic growth can be attributed to the absence of slr1461 transcripts. When Slr1461 is not expressed (i.e. in mixotrophy), its regulatory influence on NdhF3 expression is lost.

Other CCM components examined were CmpR and an AbrB family member. Both abrB2 and cmpR were also aberrantly regulated when Slr1461 was lacking, being significantly lower in the Δslr1461 mutant than WT when cultures were subject to increased CO₂ (Fig. 4b, Table S2), although their levels were little different in Δslr1461 to WT under lower CO₂ levels.

Coordinate control of CCM by proteases

It was previously reported that induction of ndhF3 and sbtA requires the FtsH2 protease as these CCM transcripts were not upregulated under low CO₂ conditions in a null mutant with inactivated ftsh2 [35]. Meanwhile, increased ftsh2 transcript levels have been reported either in response to oxidative stress; not to change in a shift from high to low CO₂ [35]; or with a very minor upregulation in transcript under CO₂ downshift in an earlier study [33]. The presence of the FtsH2 protease in networks connected with CCM regulation was further monitored in WT compared with Δslr1461 cells, comparing ftsh2 transcript levels in standard or raised CO₂ conditions in this mutant. Notably, after increasing CO₂ levels, the ftsh2 transcript level in Δslr1461 remained lower than in WT cells (Fig. 4b, Table S2), indicating that Slr1461 presence or proteolytic activity may affect induction of the ftsh2 gene in WT cyanobacteria.

To confirm that the decrease in CCM genes’ transcript levels resulted from the absence of Slr1461 activity, the slr1461 gene was used to complement Δslr1461 cells. In the complemented line, slr1461, ftsh2 and ndhF3 transcript levels were rescued (Fig. S6). The regulatory changes here suggested that further exploration is warranted to elucidate precisely the role of Slr1461 in connection with CCM processes.

Discussion

Given their evolutionary relationship with plastids, cyanobacterial predicted rhomboid proteases were examined (see Fig. S3) to compare features and conserved photosynthesis-related roles. This first experimental investigation into the function of a cyanobacterial rhomboid protease utilized the single rhomboid identified in the genome of the model oxygenic photosynthetic prokaryote Synechocystis, with reference to the homologous RBL10 protease of the A. thaliana chloroplast. Early literature upon discovery of this protein family reported the most common occurrence in prokaryotic genomes was of a single rhomboid enzyme, as identified in Synechocystis, but, subsequently, it has become clear that there are many cyanobacterial genera whose genomes encode multiple copies [63]. blast searches of more recently sequenced genomes in fact found many bacteria with duplicated rhomboid genes: the cyanobacterial genera documented here include those encoding as many as six copies (Fig. 1d). This mirrors another group of membrane proteases, where a single FtsH protease regulates many processes in E. coli [7] in contrast with the four FtsH copies of both Synechococcus sp. PCC 7502 and Synechocystis sp. PCC 6803 [20,64, 65], whose shared and specific roles are integral to oxygenic photosynthesis [20,21]. Phylogenetic analysis of rhomboid proteins in cyanobacteria showed that species encoding multiple representatives have highly similar sequences indicative of gene duplication, except for the Acaryochloris marina genome, which, despite the presence of six rhomboids, contains divergent genes that appear more distinct from one another (Fig. S3). Further dissection of the cyanobacterial single-copy versus multiple-copy rhomboids’ roles might provide insight into how a handful of organisms (such as P. marinus subspecies) have been able to function without their rhomboid protease (Fig. 1d) despite this protein family’s near-ubiquity across evolution.

The Synechocystis Slr1461 amino-acid sequence shows good overall alignment with other organisms’ rhomboid proteases that are known to be catalytically active. Slr1461 contains the nucleophile Ser and acceptor His required for proteolytic attack of substrate (Fig. 1), but it lacks the W/YR motif often conserved near rhomboid proteases’ N-termini (Figs 1a and S2). This motif was initially postulated to play a role in substrate gating but has since been ruled out as the mechanism of capture [58], and its link with enzymatic activity is now questioned. As the presence of conserved active-site serine and histidine suggests a catalytically active protease, this rhomboid may in fact be an interesting candidate for investigations of rhomboid proteolysis in the presence or absence of the W/YR motif [66,67]. The observed phenotypes following the Δslr1461 insertional inactivation mutant also support a hypothesis of Slr1461 being a proteolytically active enzyme, but it must be noted that regulatory roles for inactive rhomboids are plausible in bacteria [68], since an ‘adaptor’ function was reported for the Bacillus enzyme when acting in concert with an FtsH [37].

The constructed Synechocystis Δslr1461 mutant line was investigated for a range of photosynthetic parameters because of the plastid membrane location of the homologous A. thaliana rhomboid RBL10 and the RBL10 null mutant’s light-intensity-dependent raised NPQ [27]. No dramatic difference was identified, however, in Δslr1461 cells’ pigment–photosystem composition (Fig. 2). Furthermore, there was no difference in NPQ under light conditions tested (Table 3), despite that aspect of the A. thaliana RBL10-null phenotype. The most dramatic difference observed with the lack of Slr1461 was higher oxygen evolution compared with WT when cells were grown mixotrophically, i.e. with insufficient glucose for full heterotrophy (Fig. 3b). Given that the Bacillus rhomboid-FtsH cooperative function was reported to regulate a transporter, namely MgtE of Bacillus subtilis [37], the question was prompted over possible defective transport of glucose in the Δslr1461 cells. Since glucose typically increases growth rates and alters photosystem composition, the absence of any significant difference in these parameters between mutant and WT cultures under mixotrophic conditions (Table 2, Fig. S4) ruled out this possibility.

Since rhomboid proteases have often been reported to activate or release peptides involved in signalling, the misregulation of Δslr1461 carbon utilization could result from a role for the Slr1461 protease in activating the regulatory pathways that control the complex variations needed for cyanobacterial CO₂ uptake and C fixing. It was reported before that numerous CCM components’ transcript levels respond to CO₂ [33,35], which proved to be the case here: several CCM-gene transcript levels were increased in WT cultures moved to high-CO₂ (Fig. 4a), amongst them the slr1461 transcript. Changes in levels tallied with those reported previously when cells were moved from higher to lower CO₂ and with a small slr1461 transcript-level change found when the ftsh2 transcript was upregulated [33]. To integrate data from literature reports, further investigation of the magnitude and direction of changes is needed, ensuring consistent and duplicated culture conditions and CO₂ levels.

Studies in prokaryotes to date only once implicated rhomboid proteases in the regulation of transcript levels [10], namely for the transcription factor (RipA, cg1120) involved in iron homeostasis in C. glutamicum [10]. Certainly, the release of eukaryotic ER-bound transcription factors is an attractive hypothesis for the roles of this protease family, for which some experimental evidence is accumulating [1,69,71]. A relevant observation about cyanobacterial carbon assimilation, then, was that the cAMP-regulated TF SyCRP1, which affects carboxysome formation (as well as many other pathways), can be released from a membrane location [72]. That possible addition to the expanding CCM regulatory map also supports a recent statement that, although the CCM is well characterized, its regulation is complex and not yet fully understood [73].

A study in Synechocystis already showed that the FtsH2 protease was involved in the CCM [35], in addition to its well-known role in turnover of the PSII D1 protein [20]. It was proposed that this FtsH degrades the TF, NdhR, that is required for the induction of CCM genes [35]. It is notable how functional photosynthesis was required for the induction of CCM [34] in WT cells cultured in low CO₂ conditions, with the suggestion that oxidative stress in the cells might enhance the expression of inducible CCM genes [35]. Conversely, Haimovich-Dayan et al. (2011) reported that CCM genes were induced by glucose [74], possibly through the previously mentioned cAMP-dependent TF SyCRP1 [72].

Quantitative PCR assays here showed that loss of Slr1461 affected transcript levels of the high-affinity CO₂ import system (Fig. 4, Table S2), with CCM transport component and ftsH2 genes all misregulated. Whereas raising the CO₂ levels in cultures mildly repressed CCM transcription in WT cells, a more marked reduction occurred in the Δslr1461 mutant (Fig. 4, Table S2). A key observation is how transcript levels of the ftsh2 gene were affected by the absence of Slr1461, which raises the possibility that the role of the rhomboid protease is to modulate the input of FtsH2 into CCM regulation. A hypothesis then could be that Slr1461 action (its cleaved substrate or its own adaptor function) would permit coordination of the cell’s responses to mixotrophic growth and oxidative stress, i.e. an upstream rhomboid protease fine-tunes FtsH regulation of the CCM. Results here (Fig. 4, Table S2) suggested that this proposed Slr1461 activity would specifically affect transcript levels under raised CO₂ conditions, potentially preventing excessive downregulation of CCM genes. The relative position of the proteases and any rhomboid substrate to CCM regulators requires further investigation (Fig. 5). As raised CO₂ levels regulated the slr1461 transcript level itself, that abiotic factor might likewise regulate the availability of a Slr1461-proteolysed substrate(s). Increased CO₂ in the scheme here could increase either accessibility, degradation or activation of the unidentified Slr1461 substrate, which could have upregulatory or downregulatory functions downstream. A scheme could be that increased CO₂ results in increased transcript levels of slr1461 and in Slr1461 cleaving its substrate: this also increases the level of ftsh2. Notably, FtsH2 metalloprotease upregulation can then increase transcript levels from CCM promoters in higher CO₂ because FtsH2 degrades the NdhR repressor protein [35]. Finally, it is worth mentioning that, since the ftsh2 gene is induced by oxidative stress in WT Synechocystis but not in the stress response histidine kinase hik33^- mutant [35,75], it would be of interest to explore how the kinase-response regulator Hik33-RpaB abiotic stress mechanism interlinks with this extended CCM regulatory network.

Fig. 5. Proposed further investigation of Slr1461 roles in adaptation of Synechocystis to the environment. (a) Under low CO₂ conditions, the oxygenase activity of RuBisCO leads to the accumulation of 2-phosphoglycolate (2-PG) in the cells [81]. The binding of 2-PG to NdhR causes it to be released from the DNA, resulting in an activated transcript level of CCM genes [82]. Meanwhile, NdhR is not degraded in the absence of FtsH2 [35]. (b) High CO₂ levels were correlated with increased transcript level of the slr1461 gene, which is hypothesized to enhance proteolysis of the rhomboid’s as-yet unidentified, activated substrate. This cleaving of the Slr1461-substrate(s) by Slr1461 protease is proposed as a possible mechanism in the pathway for the observed upregulation of ftsh2 transcript levels. At the same time, 2-oxoglutarate (2-OG) accumulates under high CO₂-low nitrogen conditions: binding of 2-OG to NdhR leads to NdhR adapting to a repressor conformation [82], resulting in repression of CCM genes. (c) Resulting network of coordinated regulation of NdhR, CmpR and AbrB2 by carbon and nitrogen metabolism. The 2-PG and 2-OG levels have opposite effects on the activity of NdhR. Under low CO₂ conditions, the activity of RuBisCO shifts predominantly to the oxygenation of ribulose-1,5-bisphosphate. This, in turn, causes the cell to accumulate 2-PG. 2-OG serves as an intermediate between carbon and nitrogen assimilation, but under high CO₂ conditions, the conversion of 2-OG with ammonium (NH₄⁺) into glutamate is low, resulting in 2-OG accumulation in the cell. The 2-OG and 2-PG molecules are denoted with and respectively. Fig. 5C was adapted from [82].

Better knowledge of how photosynthetic organisms respond to CO₂ increases may be useful, underpinning future applications in the capture and storage of the rising CO₂ levels in the atmosphere and oceans resulting from human activity. It has also been suggested that utilizing a cyanobacterial CCM in plants to improve Rubisco efficiency and hence crop yield might not require changes to leaf anatomy [76]. This addition of control by Slr1461 to the FtsH2 release-from-repression of select CCM-related genes [35] would add a mechanism for crosstalk and step-specific feedback. This echoes how proteases of the Clp and Deg families supplement FtsH hetero-oligomers’ maintenance of functional PSII in varying environments, which is augmented by phosphorylation controls from kinases and phosphatases [77,78]. A multilayered response mechanism might likewise be investigated for the CCM. Additionally, as FtsH copies of type ‘A’ and ‘B’ can operate as heterodimers (namely FtsH2/3 and FtsH1/3 in Synechocystis [23,79, 80], another avenue of work would be to identify if upstream Slr1461 rhomboid activity affects more than just the FtsH2 protease. As noted above, in the case of B. subtilis, the N-terminal cytosolic domain of rhomboid YqgP facilitates MgtE transporter cleavage by FtsH, by an adaptor function that is separate from any catalytic abilities [37]. Such united protease and pseudoprotease functions add to an increasing body of evidence for proteases’ ability to regulate cellular activities by a suite of mechanisms. The diverse rhomboid functions reported to date lack a unifying and evolutionarily conserved framework for their operation, but findings here for the CCM once again support linked roles for rhomboid and FtsH membrane proteases. Further study of this aspect of the cyanobacterial carbon-concentrating system may prove a useful focus for developing our understanding.

Abbreviations

CCM

carbon-concentrating mechanism

DCMU

3-(3,4-dichlorophenyl)-1,1-dimethylurea

EGF

epidermal growth factor

ER

endoplasmic reticulum

GT

glucose-tolerant

iRhoms

inactive subset of rhomboids

PAM

pulse-amplitude modulation

PSII

photosystem II

TF

transcription factor