Integrative analysis of the salt stress response in cyanobacteria

Microorganisms evolved specific acclimation strategies to thrive in environments of high or fluctuating salinities. Here, salt acclimation in the model cyanobacterium Synechocystis sp. PCC 6803 was analyzed by integrating transcriptomic, proteomic and metabolomic data. A dynamic reorganization of the transcriptome occurred during the first hours after salt shock, e.g. involving the upregulation of genes to activate compatible solute biochemistry balancing osmotic pressure. The massive accumulation of glucosylglycerol then had a measurable impact on the overall carbon and nitrogen metabolism. In addition, we observed the coordinated induction of putative regulatory RNAs and of several proteins known for their involvement in other stress responses. Overall, salt-induced changes in the proteome and transcriptome showed good correlations, especially among the stably up-regulated proteins and their transcripts. We define an extended salt stimulon comprising proteins directly or indirectly related to compatible solute metabolism, ion and water movements, and a distinct set of regulatory RNAs involved in post-transcriptional regulation. Our comprehensive data set provides the basis for engineering cyanobacterial salt tolerance and to further understand its regulation. Supplementary Information The online version contains supplementary material available at 10.1186/s13062-021-00316-4.


50
Salinity is a prominent environmental factor determining the natural distribution of 51 microorganisms, as approximately 97% of the water resources contain more than 30 g salt 52 (mainly sodium chloride) per liter. Accordingly, the capability of microorganisms to cope with 53 high or changing salinities is crucial, not only in aquatic environments but also in terrestrial 54 habitats, in which the alternation between evaporation and rainfall can rapidly change salt 55 concentrations. In a hypersaline environment, in which the external salt concentration 56 exceeds the cellular ion content, microorganisms have to manage two major challenges: (i) 57 the low external water potential results in water loss from the cell and collapse of turgor 58 pressure, and (ii) inorganic ions permeate into cells along the electrochemical gradient, 59 which could compromise the structure of critical macromolecules. Accordingly, most 60 microorganisms feature acclimation strategies aiming at maintaining a high water and low 61 inorganic ion content in the cell. This so-called "salt-out" strategy is based on the active 62 extrusion of inorganic ions accompanied by the accumulation of compatible solutes, i.e. 63 In total, 1816 proteins, approximately 52% of the entire proteome of Synechocystis 6803 in 242 the UniProt database, were identified by at least two tryptic peptides per protein (Suppl. 243 Table S6). In particular, 1253 proteins were found in the total extracts and 823 in the soluble 244 fraction, while 1608 and 1421 proteins were identified in the membrane-enriched and debris 245 fractions, respectively (Fig. 4). Similar to previous reports on the Synechocystis 6803 246 proteome (Gao et al., 2015a), many proteins (79%) were found in both, the soluble as well as 247 the membrane fractions, although the membrane pellet was washed with high salt and high 248 pH buffers. The occurrence of proteins in different fractions clearly correlates with their 249 numbers of transmembrane helices (Suppl. Fig. S2). In the present study, we also 250 investigated the cell debris that is usually removed from proteome analyses if cellular 251 fractionation is performed. The vast majority (96.3%) of proteins in this fraction was also 252 identified in the membrane-enriched fraction. However, 29 proteins were exclusively found in 253 the debris fraction, for example the large Slr1567 protein, a putative outer membrane protein 254 (Suppl . Table S7). Many other proteins enriched in the debris fraction are annotated as 255 components of the cell envelope or of the outer membrane-bound periplasmic space. 256 Salt-dependent changes in protein abundances were evaluated in total protein extracts as 257 well as three subcellular fractions (Fig. 4). For the selection of differentially expressed 258 proteins cut-off criteria were defined by a fold change of ≥│1.5│ and a corresponding p-value 259 < 0.05. However, for a large number of proteins less pronounced fold change values were 260 also statistically significant. The volcano plots indicate that the group of stably up-regulated 261 proteins (Suppl. Fig. S3) displayed larger fold changes than down-regulated proteins, 262 resembling the observations made with the transcriptomic data (Suppl. Fig. S1). 263 Next, we analyzed whether or not particular proteins showed consistent salt-related changes 264 in the different proteome fractions. For most proteins, well-matching values were found in the 265 different fractions and in the total extract. However, in some cases the relative abundances 266 showed an inverse relation in membrane and soluble fractions. For example, many ribosomal 267 proteins occurred in lower amounts in the soluble fraction but were elevated in the 268 membrane and debris fractions of salt-acclimated cells compared to control cells (Suppl. fraction. To deal with this situation, we calculated a weighted fold change from the 276 subcellular fractions data of each protein and showed that it was highly concordant with the 277 corresponding fold change from the total protein extract (Suppl. Fig. S5; more detailed 278 description is given in the supplementary material). Finally, the mean value of the weighted 279 fold change from the subcellular fractions and the fold change from the total protein extract 280 was used. If only one of these two values was available, it was used directly as final fold 281 change. This led to a list of 1803 proteins, to which two membrane proteins were appended 282 that are known (Slr0531) or suspected (Sll1037) to be important for salt acclimation, but were 283 identified by individual peptides only. As a final result, 190 proteins were up-regulated and 284 189 protein down-regulated 7 days after salt shock among the 1805 quantified proteins 285 (Suppl . Table S6). 286

Correlation analysis of salt-stimulated transcriptome and proteome 287
A correlation analysis was performed to analyze the overall relation between transcriptomic 288 (log 2 fold changes of mRNA after 24 h) and proteomic changes (log 2 fold changes of protein 289 abundances after 7 d). 1749 transcript/protein pairs could be matched ( Fig. 5; Suppl. Table  290 S8). The Pearson correlation coefficient for the proteomic and transcriptomic data sets was r 291 = 0.58 indicating a quite good relationship, especially taking into consideration that sampling 292 was done in different laboratories and at different time points. To grade the correlation 293 between the newly acquired transcriptomic and proteomic data sets, we also compared our 294 proteome data to the previously published transcriptome data sets by Marin et al. using the same subset of mRNAs as for the comparison with the proteome. Surprisingly, the 299 obtained coefficients of r = 0.49 and r = 0.50 showed a slightly lower correlation between 300 different transcriptomic data sets than correlation between the proteomic and the present 301 transcriptomic data set (Fig. 5C). These findings indicate that culture conditions significantly 302 influence the comparability of different data. Nevertheless, in all cases a close correlation 303 was observed for the expression of genes that are of direct importance for salt acclimation, 304 while the expression of other genes can vary depending on small differences between the 305 culture conditions leading to relatively low Pearson correlation coefficients. 306

Compatible solute metabolism and transport constitutes a salt-specific stimulon 308
Central to salt acclimation of Synechocystis 6803 is the accumulation of the main compatible 309 solute GG, because mutants affected in the genes ggpS and ggpP (stpA) encoding the GG 310 synthesis enzymes showed the highest degree of salt sensitivity compared to wild type 311 (Hagemann et al., 1997;Marin et al., 1998). As initially found by Reed and Stewart (1985), 312 salt-acclimated cells accumulate high amounts of the compatible solute GG, which 313 represents the by far largest pool of low molecular mass organic compounds. The amount of 314 GG is approximately 2000times higher in salt-grown cells compared to the trace amounts of 315 GG in control cells (Fig. 6C). The second compatible solute sucrose is approximately 1000-316 fold less abundant (0.28 nmol OD 750 -1 ml -1 ) than GG in salt-acclimated cells (Suppl. Corresponding to the high GG accumulation, the GgpS and GgpP proteins and mRNAs 320 showed significantly elevated levels (Table 2), which is supported for GgpS by Northern-and 321 Western-blotting (Fig. 6AB). The sucrose synthesis enzymes Sps and Spp also exceeded 322 the threshold for significant protein changes ( on the opposite strand, the genes for the GG hydrolase (GghA) and glycerol kinase (GlpK), 330 the latter is involved in synthesis of the GG precursor G3P from glycerol, are located. They 331 show similar expression pattern as ggpS. In contrast to the salt induction of genes for G3P 332 synthesis, the glgC gene encoding the enzyme for ADP-glucose synthesis, the second 333 precursor for GG, is not salt-regulated on RNA or protein levels. 334 Immediately downstream of ggpP, the salt-induced GgtA protein is encoded that acts as the 335 ATP-binding subunit of the GG transporter (Ggt). The genes for the other Ggt subunits, 336 GgtB, C and D (two of them were identified among the salt-stimulated proteins) form a 337 separate salt-induced operon. The co-regulation of genes and proteins for GG synthesis and 338 the ABC-type osmolyte transporter Ggt, which all belong to the cluster 2 (Table 2) The occurrence of all genes related to GG biochemistry, which are proven to be involved in 360 salt acclimation, make it very probable that several of the co-regulated genes encode 361 proteins also specific for this stress acclimation. This assumption is supported by our finding 362 that many of these proteins also accumulated to higher levels in salt-acclimated cells. In 363 cases where the protein levels were not significantly elevated, the corresponding gene 364 showed only transient stimulation at the earliest time points after salt addition (Suppl. Fig.  365 S6). 366

Transporters and channels belonging to the salt-specific stimulon 367
Clustering and functional enrichment analyses clearly indicated that differential regulation of 368 proteins related to membrane transport is generally an important mechanism for salt 369 acclimation (Suppl . Table S9 for six Na + /H + antiporters, five of them were identified in the proteome. Among them, the 384 protein abundance of NhaS2 and NhaS5 was significantly enhanced in salt-acclimated cells 385 (Fig. 7), whereas the amount of NhaS3, which is essential for cell viability and was discussed 386 to be mainly responsible for Na + export (Wang et al., 2002;Elanskaya et al., 2002), remained 387 unchanged (Suppl . Table S6). However, the fact that the abundance of a protein remains 388 unchanged does not exclude an essential function of this protein for salt acclimation, 389 because it could be regulated on biochemical level according to cellular demands. While 390 crucial roles were assigned to NhaS2 under low Na + /K + ratios (Mikkat et al., 2000) or growth 391 at different pH values (Wang et al., 2002), the nhaS5 mutant did not show any changes 392 compared to wild type (Wang et al., 2002;Elanskaya et al., 2002). Unfortunately, none of the 393 previously discussed candidates for chloride exporters (Sll1864, Slr0753, Sll0855; 394 Hagemann, 2011) could be identified in our proteome data set. Among them, only the gene 395 sll1864 was transiently stronger expressed on mRNA level in salt-shocked cells (Suppl. 396 Table S1). 397 In addition, several other proteins potentially involved in ion transport were found in higher 398 abundances in salt-acclimated cells (Fig. 7). For example, two of the 7 annotated cation-399 transporting ATPases (Wang et al., 2002) were elevated in the proteome (Slr1950 with FC 400 1.8 and Sll1614 with FC 1.68) and two hours after salt shock in the transcriptome as well. control mRNA as well as protein levels after long-term salt acclimation (Fig. 7). 411

Many salt-induced proteins are involved in general stress response 412
Two proteins of unknown function, Sll1862 and sll1863 showed 9.9-and 13.3-fold, 413 respectively, higher abundances in salt-acclimated cells (Table 3) Several heat-shock proteins that are involved in protein folding and repair have been 424 previously identified among the salt-induced, general stress proteins (Fulda et al., 2006). In 425 the present study, only the 33 kDa chaperone (Sll1988) was more than 2-fold accumulated 426 while its mRNA showed only a slight increase, whereas the amounts of DnaK, GroEL, or 427 DnaJ proteins were not significantly changed. Moreover, the small, 16.6 kDa heat shock 428 protein (Sll1514) was decreased despite its RNA accumulated after the salt addition (Table  429 3), whereas it was found before in significant enhanced amounts in salt-acclimated cells 430  Table S1). However, in 440 long-term acclimated cells only two iron-regulated proteins are significantly elevated (Table  441 3). The general stress protein Slr1894, which is annotated as MrgA or Dps like protein, was 442 found in higher abundances in salt-acclimated cells. It has been shown to be involved in lyase (CpcF, sll1051), a protein involved in phycobiliprotein assembly into phycobilisomes, 462 increased 1.7-fold. Another phycocyanobilin lyase (CpcE, slr1878) was also increased. 463 Finally, many ribosomal proteins showed a tendency to slightly lowered amounts (Fig. 8 Fig. S9). In addition to glpK four other genes 536 (three encode subunits of protochlorophyllide synthase, e.g. ChlN, slr0750; see Table 3) are 537 at least transiently salt-regulated but have not yet shown to code for proteins directly involved 538 in salt acclimation. 539 Another salt-stimulated gene cluster was found on the plasmid pSYSA. The genes sll7063-540 sll7067 (for one example see Table 3 Metabolome analysis of salt-acclimated cells 565 The presence of high NaCl amounts in the medium induces a massive GG accumulation 566 (Fig. 6C), which likely triggers a strong redistribution of organic carbon in Synechocystis 567 6803. The large impact of GG synthesis on overall carbon metabolism is also consistent with 568 the observation that many proteins (and their genes) involved in glycogen metabolism as well 569 as glycolysis showed significant changes in their abundances (Table 4). For example, the 570 neopullulanase, glycogen phosphorylase GlgP2 (Slr1367) and one debranching enzyme 571 GlgX1 (Slr0237, the other one Sll1857 is decreased) showed higher abundances on protein 572 as well as RNA levels in salt-acclimated cells. The different response of the two GlgX 573 proteins towards salt stress has been previously shown with Western-blotting (Iijima et al., 574 2015). Hence, the demand of organic carbon for the synthesis of GG precursors is at least 575 partly supported by an enhanced glycogen breakdown and reduced glycogen build up, 576 because glycogen and GG synthesis are competing for the same precursor, ADP glucose. 577 The relatively low carbon/nitrogen state in salt-acclimated cells is also reflected by the 578 lowered amount of 2-oxoglutarate (2OG, Fig. 9), which is the key metabolic signal reporting 579 changes of the cellular carbon/nitrogen ratio in cyanobacteria (Hagemann et al., 2021). 580 To obtain a snapshot on metabolites of the central carbon and nitrogen metabolism, LC-581 MS/MS was used (Suppl. Table S10). The relative levels of the RubisCO carboxylation and 582 oxygenation products 3PGA and 2PG, respectively, showed opposite behavior (Fig. 9). 583 3PGA accumulated approximately 3-fold less in salt-acclimated cells, while 2PG was clearly 584 enhanced. This could indicate a decreased CO 2 /O 2 ratio at the active site of Rubisco, since 585 the amounts of these gasses mainly regulate its relative carboxylation/oxygenation activity. 586 For example, it might be possible that due to the higher content of inorganic ions inside the 587 salt-exposed cells carboxysomes are less gas tight in high salt-grown cells, thereby 588 promoting a better diffusion of oxygen into carboxysomes reducing the CO 2 /O 2 ratio. The 589 observed changes in the 3PGA and 2PG levels are consistent with the reported lower 590 photosynthetic activity and growth rate in salt-acclimated cells of Synechocystis 6803 (e.g., 591 Hagemann et al., 1994), which certainly also reduce Calvin-Benson-cycle activity. 592 Consistently, the protein abundances of photosynthetic complexes, Calvin-Benson-cycle 593 enzymes including RubisCO and components of the cyanobacterial inorganic carbon-594 concentrating mechanism were found at 10-40% lower levels, which is below our significance 595 threshold but might contribute to lower photosynthetic activity in salt-acclimated cells. 596 Furthermore, more organic carbon could be taken out from the Calvin-Benson-cycle, which is 597 for example seen in the increased amount of pyruvate and organic acids in the reductive 598 branch of the TCA cycle, such as malate and fumarate. This interpretation is also supported 599 by the finding that Gap1, the glyceraldehyde dehydrogenase 1 (Slr0884) involved in sugar 600 catabolism (Koksharova et al., 1998) is also up-regulated (Table 4), while Gap2 involved in 601 photosynthetic carbon assimilation did not change (Suppl . Table S1). 602 In addition to the carbon fixation and allocation, nitrogen assimilation is altered in salt-603 acclimated cells, which is reflected by enhanced glutamine and glutamate levels while 2OG, 604 the carbon skeleton used for ammonia assimilation decreased (Fig. 9). Increased glutamate 605 levels have been often reported in salt-exposed bacteria (Hagemann, 2011), because this 606 negatively charged amino acid is compensating the positive charge of cations, especially K + . 607 The enzymes involved in the GS/GOGAT cycle for assimilation of NH 4 + into 2OG did not 608 significantly change their expression in long-term salt acclimated cells, but were significantly 609 lowered immediately after salt addition. However, the glutamate decarboxylase Gad 610 (Sll1641) showed increased expression in salt-grown cells. Proline, which is often used in 611 heterotrophic bacteria as compatible solute and can be synthesized from glutamate, is not 612 changed during salt acclimation of Synechocystis 6803 (Fig. 9) The salt acclimation response goes way beyond the induction of gene expression required 667 for the compatible solute machinery, it has clearly a great impact on general metabolism. 668 Along the temporal axis, the reprogramming of gene expression can be differentiated 669 between an early response with the respective minima and maxima leading to the four 670 different clusters (Fig. 2). The metabolic response included the rapid repression of the 671 ammonia assimilation system, detected by decreased transcript abundances for amt1 and 672 glnA encoding the ammonium transporter and the primary enzyme for ammonium 673 incorporation, glutamine synthetase (GS). Consistently, increased transcript levels were 674 found for genes gifA and gifB, which encode inhibitory proteins for GS, thereby blocking 675 ammonium assimilation (for an overview see Bolay et al., 2018). After 24 h the transcript 676 levels of these genes were at the initial levels. 677 The question arises, which processes are responsible for the staggered reshaping of the 678 transcriptome. It has been shown that shortly after salt addition the cytoplasmic composition 679 underwent rapid changes due to ion and water movements, whereby the early high internal 680 ion contents, especially of Na + are discussed to inhibit metabolic activities but also to trigger 681 acclimation responses such as GG synthesis activation (reviewed in Hagemann, 2011). The 682 transporters responsible for the rapid ion movements are largely unknown, especially verified 683 candidates for Clexport are still missing. In the present study we did not find marked 684 expression changes for genes encoding potential anion exporters. It can be assumed that 685 these transporters are mainly regulated on their activity levels to manage the ion regulation 686 within the first minutes to hours after salt shock, because de novo protein synthesis is one candidates for a salt-stress specific gene regulation. First, the transcription factor PrqR 735 (Slr0895) represents an interesting candidate, because the gene sl0895 is clearly co-736 regulated with many genes coding proteins involved on GG metabolism (Suppl. Fig. S6). It 737 has been recently shown that PrqR is involved in the acclimation to oxidative stress in 738 Synechocystis 6803 (Khan et al., 2016). Salt stress is also inducing oxidative stress in 739 cyanobacterial cells, hence, the finding of the role of PrqR in this stress acclimation process 740 might of secondary importance. Second, the gene ssl1326 that possibly encodes a CopG 741 family transcription factor has been found strongly induced after salt shock in the DNA 742 microarray data set (Suppl . Table S1). Further work is necessary to validate whether PrqR or 743 CopG are somehow acting as salt-specific gene expression regulators. In the moment, it 744 might well be possible that ion-mediated changes in the DNA structure, RNA-polymerase 745 affinity, and enzyme activities might be the main and sufficient mechanism to acclimate 746 towards different salt conditions in euryhaline bacteria such as Synechocystis 6803. 747   with reported fold changes could be mapped to corresponding changes in protein 1263 abundances. Matches with ratio differences below log 2 1.5-fold changes were considered to 1264 be similar. C.  (0.5, 2, and 24 h) and proteome measurements (7 d) for the top ranked transport related 1285 genes. Highlighted in Red = compatible solute transport (ggtABCD); Blue = related to iron 1286 transport, Green = nhaS genes (Na + /H + antiporter). Detailed information is provided in Suppl. 1287 Table S9. mass compounds were isolated from cells of Synechocystis 6803 grown in NaCl-free BG11 1299 medium or medium supplemented with 4% NaCl for 7 days. LC-MS/MS was used to estimate 1300 the relative levels (Y axis show fold changes, amount in cells from 0% NaCl cultivation (-) set 1301 to 1 and relative level at 4% NaCl (+) is shown) of central metabolites as part of primary 1302 carbon and nitrogen metabolism. Shown are mean values from three biological replicates 1303 (details in Suppl.