Compensatory Transcriptional Response of Fischerella thermalis to Thermal Damage of the Photosynthetic Electron Transfer Chain

Key organisms in the environment, such as oxygenic photosynthetic primary producers (photosynthetic eukaryotes and cyanobacteria), are responsible for fixing most of the carbon globally. However, they are affected by environmental conditions, such as temperature, which in turn affect their distribution. Globally, the cyanobacterium Fischerella thermalis is one of the main primary producers in terrestrial hot springs with thermal gradients up to 60 °C, but the mechanisms by which F. thermalis maintains its photosynthetic activity at these high temperatures are not known. In this study, we used molecular approaches and bioinformatics, in addition to photophysiological analyses, to determine the genetic activity associated with the energy metabolism of F. thermalis both in situ and in high-temperature (40 °C to 65 °C) cultures. Our results show that photosynthesis of F. thermalis decays with temperature, while increased transcriptional activity of genes encoding photosystem II reaction center proteins, such as PsbA (D1), could help overcome thermal damage at up to 60 °C. We observed that F. thermalis tends to lose copies of the standard G4 D1 isoform while maintaining the recently described D1INT isoform, suggesting a preference for photoresistant isoforms in response to the thermal gradient. The transcriptional activity and metabolic characteristics of F. thermalis, as measured by metatranscriptomics, further suggest that carbon metabolism occurs in parallel with photosynthesis, thereby assisting in energy acquisition under high temperatures at which other photosynthetic organisms cannot survive. This study reveals that, to cope with the harsh conditions of hot springs, F. thermalis has several compensatory adaptations, and provides emerging evidence for mixotrophic metabolism as being potentially relevant to the thermotolerance of this species. Ultimately, this work increases our knowledge about thermal adaptation strategies of cyanobacteria.

Fischerella thermalis (also known as Mastigocladus laminosus) was first described in a hot spring in the Czech Republic [16]. It is an oxygenic photosynthetic organism with high metabolic versatility that can grow autotrophically or mixotrophically [17,18] using different nitrogen sources, including N 2 [8,[18][19][20]. Environmental and laboratory data on F. thermalis have defined its optimal growth range between 45 • C and 55 • C, with an estimated photosynthetic limit near~58 • C [8,15,[19][20][21][22][23][24]. Moreover, sequences belonging to F. thermalis have been detected at temperatures as high as 66 • C [8,19], but its photosynthetic activity at and above this temperature has not yet been demonstrated, neither in the environment nor in the laboratory.
Of the cyanobacterial photosynthesis components, the least thermoresistant is photosystem II (PSII) [14], a macromolecular complex composed of several protein subunits and cofactors, including the water-oxidizing complex (WOC). The WOC is mainly coordinated by the proteins D1 (PsbA) and D2 (PsbD), with D1 being directly involved in the photochemical reaction and requiring the highest turnover rates of the complex [36,37]. In cyanobacteria, there are several currently known distinct D1 isoforms that have been grouped as G0 to G4, of which G3 and G4 are the most common [38]. Recently, two new D1 forms belonging to G4 have been proposed: D1 INT and D1 FR , which are present in most Nostocales cyanobacteria [39], including F. thermalis. In other cyanobacteria, such as Thermosynechococcus elongatus BP-1, it has been reported that the D1 isoforms are expressed differently in response to temperature [40]; however, their relationship to the thermal-adaptation process is still not well understood.
In cyanobacteria, photosynthesis is connected to aerobic respiration and carbon catabolism by participating in photorespiration and regulating photosynthesis through redox modulation in thylakoid membranes [41,42]. It has been suggested that these alternative pathways assist photosynthesis under stress conditions [43], but whether these energy pathways are involved in the thermal adaptation of F. thermalis is unknown. Understanding how these pathways are connected to F. thermalis photosynthesis could help to reveal how this organism adapted to high-temperature hot springs conditions. Furthermore, the genus Fischerella harbors both thermophilic and mesophilic species [5]; thus, studying their evolutive history and physiological processes could help us better undestand the underlying adaptation of photosynthesis and its components to high temperatures. In this study, we analyzed how temperature affects the energy metabolism of F. thermalis using both environmental data and an isolated strain. The transcriptional activity in samples obtained from a thermal gradient of the Porcelana hot spring (Northern Patagonia, Chile) was evaluated, and gene coexpression in samples obtained from the El Tatio geothermal field (Atacama, Chile) was investigated. Additionally, the expression of key energy metabolism and photosynthetic efficiency genes was evaluated in temperature-treated axenic F. thermalis PCC 7521 cultures. We found that the transcriptional response of F. thermalis is consistent with a photosynthetic compensational response model, where increased transcription of energy metabolism genes helps to sustain F. thermalis photosynthesis at higher temperatures.

Prevalence of the D1 Isoform in the Fischerella thermalis PSII Reaction Center
Most cyanobacteria have several copies and isoforms of the D1 protein [38]. These isoforms respond to environmental changes [36,40]; however, their relationship to high temperature adaptation is still poorly described. To determine the main D1 isoforms in the genus Fischerella, and particularly in F. thermalis, we reconstructed a phylogenetic tree of the psbA gene that encodes the D1 protein using 48 Fischerella genomes available in the NCBI RefSeq and GenBank databases [44,45] (Figure 1A). Previously identified D1 sequences from Gloeobacter violaceus, Nostoc sp., and Chlorogloeopsis fritschii [38,39,46] were used here as isoform outgroups, along with 30 psbD sequences. Our results show that the D1 isoforms G1-G4, including D1 INT and D1 FR , are widespread within the genus Fischerella. Differences in amino acids have been previously reported between D1 INT and G4 associated with PSII internal electron acceptors [39]. Consequently, we found that the amino acid residues Trp126 and Trp260 (corresponding to Tyr126 and Phe260, respectively, in G4) [39] are conserved in all D1 INT sequences of the genus Fischerella. For D1 FR , we found that amino acid residues related to chlorophyll f binding, such as Tyr120 and Thr155 [47,48], are also conserved. The gene content of F. thermalis is known to vary compared with closely related cyanobacteria, thereby generating phenotypic plasticity related to specific niches in a temperature gradient [44,45]. Thus, we analyzed the gene copy number of each isoform within the genus Fischerella ( Figure 1B). Genomes belonging to the genus Fischerella were classified using the GTDB-tk software R202 and by phylogenomic reconstruction, being consistent with the 16S rRNA phylogeny ( Figure S1). As a result, we found that the number of D1 INT -psbA and D1 FR -psbA remained stable within the genus, with~1 copy per genome ( Figure 1B and Table S1). As most of the 48 Fischerella genomes were fragmented, we focused on the only F. thermalis closed genome (strain NIES-3754, GCF_001548455.1 [49]), which showed that F. thermalis has three G4-psbA copies (Table S1). In contrast, the only complete genome from a mesophilic Fischerella (strain NIES-4106, GCF_002368315.1 [50]) has five copies of G4-psbA (Table S1). Differences in the G4-psbA copy number between F. thermalis and the mesophilic Fischerella were also observed in the high-quality open genomes of the genus Fischerella (Table S1), supporting that F. thermalis has fewer copies of G4-psbA than mesophilic Fischerella. Additionally, we compared the genomic context of G4-psbA and D1 INT -psbA genes in the F. thermalis NIES-3754 and mesophilic Fischerella sp. NIES-4106 genomes ( Figure 1C,D). The analysis showed that one of the G4-psbA copies in F. thermalis was close to D1 INT -psbA, while another was close to genes encoding phycobilisome rod proteins ( Figure 1C), suggesting some degree of G4 gene clustering in F. thermalis. Conversely, some G4-psbA of Fischerella sp. NIES-4106 were found near transposase genes ( Figure 1D), possibly as part of mobile elements of the Fischerella sp. NIES-4106 genome. The genomic context of D1 INT -psbA was analyzed in 45 genomes of the genus Fischerella where D1 INT -psbA was detected (summarized in Table S2). The results show that D1 INT -psbA tends to be next to a G4-psbA or a G4-psbA pseudogene in F. thermalis, except when D1 INT -psbA is located at the end of the contig. In contrast, G4-psbA was not found next to D1 INT -psbA in the mesophilic Fischerella genomes (Table S2). The genomic context for the other isoforms and sequence alignment files can be found in the Supplemental Material. These results suggest that some isoforms and gene copies of D1/psbA are selected by high-temperature conditions; thus, we then investigated whether the transcription of photosynthesis-associated genes was also affected by temperature. GCF_000009705.1 and GTDB R202 species Trichormus sp000009705), and Chlorogloeopsis fritschii PCC 9212 (NCBI genome ID: GCF_000317265.1), as well 30 psbD sequences. Branches are colored according to the D1 isoform, while labels are colored according to GTDB R202 Fischerella species, sequences from outgroup genomes, or identical sequences present in more than one species. For collapsed clades, the labels show the number of sequences, number of non-redundant sequences (i.e., identical NCBI protein ID), and number of genomes. Bootstrap support values over 90% are shown as black dots in the tree nodes. (B) Phylogenomic reconstruction of 48 Fischerella genomes and one outgroup (Nostoc sp. PCC 7120). Tree leaves are collapsed at species level and colored according to (A). Bootstrap support values over 95% are shown as black dots in the tree nodes. Theisoform count was performed according to the distribution of psbA sequences in (A), where the outer, middle, and inner circles represent the maximum, average, and minimum number of psbA sequences copies, respectively. (C,D) To analyze the genomic context of the G4 (green) and D1 INT (brown) isoforms of the psbA gene from F. thermalis NIES-3754 (C) and Fischerella sp. NIES-4106 (D), a region consisting of 5 genes upstream of the psbA gene, the respective psbA, and five genes downstream of the psbA gene was selected as the genome fragment. Genes are represented by boxed colored as per the legend.

Effect of Temperature on the Transcription of Photosynthesis-Related Genes in Fischerella thermalis
To determine the effect of temperature on photosynthetic transcriptional activity in the cyanobacterium F. thermalis, we recovered and analyzed the transcriptional levels of photosynthesis-related genes along the thermal gradient (66, 58, 48 • C) of Porcelana hot spring ( Figure 2A) using metatranscriptomes already available in the literature [8]. The results show that most of the genes associated with photosynthesis, such as the subunits of PSII (e.g., psbA and psbD) and PSI (e.g., psaA and psaB), exhibited higher levels at 48 • C or 58 • C, while some genes also related to respiration, such as those encoding ATP synthase or electron carriers, had higher levels at 58 • C or 66 • C. In the case of reaction center genes, the transcript levels of psbA and psbD (encoding for D1 and D2, respectively) were likely to be lower at 58-66 • C ( Figure 2A). However, owing to a lack of data that prevented more robust and specific analyses, we decided to evaluate the transcriptional levels of the D1 INT -encoded psbA and psbD under controlled laboratory conditions using cultures of strain F. thermalis PCC 7521 grown at 45 • C (Section 4.2). The subsequent cultures were subjected to different temperature treatments from 40 • C to 65 • C lasting from 30 min to 6 h. The heat-shock responsive gene groEL [51] showed increased transcript levels at higher temperatures ( Figure S2), suggesting that high-temperature treatments were affecting the stability of cellular components and triggering a heat-shock response [52][53][54][55], which probably affected the photosynthetic apparatus [55,56]. In the case of the D1 INT -psbA gene ( Figure 2B), increasing transcript levels were observed from 40 • C to 55 • C, with a 15.28-fold increase after 1 h of treatment, whereas transcript levels decreased at 60 • C to 65 • C. This indicates that D1 INT -psbA transcription probably peaks near the photosynthetic limit of F. thermalis (~58 • C), but is reduced above this temperature. In contrast to D1 INT -psbA, the maximum psbD transcript levels were observed at 55 • C and 60 • C ( Figure 2C), suggesting that D1 (specifically D1 INT ) and D2 respond differently to temperature. In addition, the increase in psbD transcripts was lower, showing less than a threefold change at the peak levels compared with that at 40 • C, and its transcripts were overall 10-to 60-fold lower than those of the D1 INT -psbA gene. We also found that the transcript levels of genes associated with PSI (e.g., psaA, psaB and psaF; Figure 2A) were reduced under high temperatures, suggesting that cyclic electron flow in F. thermalis is also affected by temperature. Because carbon-based energy pathways, especially those related to electron transfer, are suggested as auxiliary pathways in cyanobacteria under stress [43] and because we found indications that respiration is transcriptionally induced by temperature (Figure 2A), we decided to evaluate alternative energy pathways for photosynthesis and the co-expression of energy metabolism genes in F. thermalis.  [19], the transcriptional levels of photosynthesisrelated genes were analyzed at 48 °C, 58 °C , and 66 °C. Genes with multiple copies or isoforms and their respective transcript levels were grouped as a single gene. The Z-score color-scale indicates transcriptional levels from low (blue) to high (red). (B,C) The transcriptional levels of psbA (B) and psbD (C) in F. thermalis cultures treated with temperatures from 40 °C to 65 °C (n = 3) were analyzed by RT-qPCR. Relative transcript levels were adjusted to 1 using 40 °C as a reference for each analyzed time. Error bars denote the standard error of the mean (SEM). Barplots are colored for aesthetic purposes. Welch's ANOVA was used for statistical analysis. Different lowercase characters above each bar denotes significant differences between temperature treatments.

Alternative Energy Pathways and Co-Expression of Energy Metabolism Genes in Fischerella thermalis
To analyze whether alternative energy pathways for photosynthesis are involved in the high-temperature response of F. thermalis, and, because cyanobacteria are reported to store carbon and energy as carbohydrates [57], we analyzed in silico the transcriptional levels of genes related to glycolysis in the metatranscriptomes of the thermal gradient (48, 58, 66 °C ) at Porcelana hot spring ( Figure 3a). As a result, we found a heterogeneous transcriptional pattern, consistent with the fact that some of these genes are involved in both carbon fixation and carbon consumption processes. Key genes related to glycolysis, such as pfk and pyk (encoding glycolysis-associated enzymes 6-phosphofructokinase and pyruvate kinase, respectively), presented higher transcriptional levels at 58 °C or 66 °C in Porcelana hot spring ( Figure 3a). Furthermore, analysis of these same genes in F. thermalis PCC 7521 cultures demonstrated higher pfk levels at 55 °C and 60 °C for most evaluated time points (Figure 3b), with a 7.12-fold peak at 60 °C after 1 h of treatment, which is similar to what was previously found for the psbD gene ( Figure 2c). In the case of pyk (Figure 3c), transcriptional levels in the F. thermalis PCC 7521 cultures were more variable over the treatment time. Specifically, higher levels were observed at low temperatures  [19], the transcriptional levels of photosynthesisrelated genes were analyzed at 48 • C, 58 • C, and 66 • C. Genes with multiple copies or isoforms and their respective transcript levels were grouped as a single gene. The Z-score color-scale indicates transcriptional levels from low (blue) to high (red). (B,C) The transcriptional levels of psbA (B) and psbD (C) in F. thermalis cultures treated with temperatures from 40 • C to 65 • C (n = 3) were analyzed by RT-qPCR. Relative transcript levels were adjusted to 1 using 40 • C as a reference for each analyzed time. Error bars denote the standard error of the mean (SEM). Barplots are colored for aesthetic purposes. Welch's ANOVA was used for statistical analysis. Different lowercase characters above each bar denotes significant differences between temperature treatments.

Alternative Energy Pathways and Co-Expression of Energy Metabolism Genes in Fischerella thermalis
To analyze whether alternative energy pathways for photosynthesis are involved in the high-temperature response of F. thermalis, and, because cyanobacteria are reported to store carbon and energy as carbohydrates [57], we analyzed in silico the transcriptional levels of genes related to glycolysis in the metatranscriptomes of the thermal gradient (48,58,66 • C) at Porcelana hot spring ( Figure 3A). As a result, we found a heterogeneous transcriptional pattern, consistent with the fact that some of these genes are involved in both carbon fixation and carbon consumption processes. Key genes related to glycolysis, such as pfk and pyk (encoding glycolysis-associated enzymes 6-phosphofructokinase and pyruvate kinase, respectively), presented higher transcriptional levels at 58 • C or 66 • C in Porcelana hot spring ( Figure 3A). Furthermore, analysis of these same genes in F. thermalis PCC 7521 cultures demonstrated higher pfk levels at 55 • C and 60 • C for most evaluated time points ( Figure 3B), with a 7.12-fold peak at 60 • C after 1 h of treatment, which is similar to what was previously found for the psbD gene ( Figure 2C). In the case of pyk ( Figure 3C), transcriptional levels in the F. thermalis PCC 7521 cultures were more variable over the treatment time. Specifically, higher levels were observed at low temperatures during the 30 min treatment; the levels then increased at 50-60 • C after 1 h of treatment; and finally, no statistical differences were found after 3 or 6 h of treatment. As transcriptional levels of glycolysisrelated genes are likely to increase at high temperatures and, as in cyanobacteria, pathways associated with carbon usage are connected to photosynthesis [43,58], we explored if this induction is part of a photosynthesis-related response in F. thermalis.
during the 30 min treatment; the levels then increased at 50-60 °C after 1 h of treatment; and finally, no statistical differences were found after 3 or 6 h of treatment. As transcriptional levels of glycolysis-related genes are likely to increase at high temperatures and, as in cyanobacteria, pathways associated with carbon usage are connected to photosynthesis [43,58], we explored if this induction is part of a photosynthesis-related response in F. thermalis.  [19], the transcriptional levels of photosynthesis-related genes were analyzed at 48 °C, 58 °C , and 66 °C. Genes with multiple copies or isoforms and their respective transcript levels were grouped as a single gene. The Z-score color-scale indicates transcriptional levels from low (blue) to high (red). (B,C) The transcriptional levels of pfk (B) and pyk (C) in F. thermalis cultures treated with temperatures from 40 °C to 65 °C (n = 3) were analyzed by RT-qPCR. Relative transcript levels were adjusted to 1 using 40 °C as reference for each analyzed time. Error bars denote the SEM. Barplots are colored for aesthetic purposes. Welch's ANOVA was used for statistical analysis [19]. Different lowercase characters above each bar denotes significant differences between temperature treatments.
We also evaluated whether genes for energetic metabolism are co-expressed in F. thermalis within the context of the thermal gradient. To that end, we analyzed metatranscriptomic sequences associated with F. thermalis from 11 samples collected within the El Tatio geothermal field (Atacama, Chile) at hot springs representing temperatures between 45 °C and 57 °C (Table S3). The abundance of F. thermalis in the El Tatio hot springs was more variable than that in the Porcelana hot spring (Patagonia, Chile), and the number of retrieved sequences varied among samples (1 × 10 5 to 1 × 10 6 ). For this reason, we did not analyze a direct association with temperature, but instead measured transcriptional correlation levels of the genes. Because we were only interested in metabolic genes, we restricted the analysis to those genes affiliated with enzyme-associated proteins in the  [19], the transcriptional levels of photosynthesis-related genes were analyzed at 48 • C, 58 • C, and 66 • C. Genes with multiple copies or isoforms and their respective transcript levels were grouped as a single gene. The Z-score color-scale indicates transcriptional levels from low (blue) to high (red). (B,C) The transcriptional levels of pfk (B) and pyk (C) in F. thermalis cultures treated with temperatures from 40 • C to 65 • C (n = 3) were analyzed by RT-qPCR. Relative transcript levels were adjusted to 1 using 40 • C as reference for each analyzed time. Error bars denote the SEM. Barplots are colored for aesthetic purposes. Welch's ANOVA was used for statistical analysis [19]. Different lowercase characters above each bar denotes significant differences between temperature treatments.
We also evaluated whether genes for energetic metabolism are co-expressed in F. thermalis within the context of the thermal gradient. To that end, we analyzed metatranscriptomic sequences associated with F. thermalis from 11 samples collected within the El Tatio geothermal field (Atacama, Chile) at hot springs representing temperatures between 45 • C and 57 • C (Table S3). The abundance of F. thermalis in the El Tatio hot springs was more variable than that in the Porcelana hot spring (Patagonia, Chile), and the number of retrieved sequences varied among samples (1 × 10 5 to 1 × 10 6 ). For this reason, we did not analyze a direct association with temperature, but instead measured transcriptional correlation levels of the genes. Because we were only interested in metabolic genes, we restricted the analysis to those genes affiliated with enzyme-associated proteins in the BioCyc database [59]. As a result, we identified 710 genes that had an r value ≥ 0.85 with at least one other gene. Based on the correlation levels, we built a correlation network that exhibited a total of 3679 connections forming a total of 19 clusters with at least 5 genes ( Figure 4A). Using the BioCyc platform [59], we analyzed which metabolic pathways were overrepresented in clusters 1, 6, 7, 8, and 12 (Table S4), corresponding to the five clusters with higher number of nodes. We found that clusters 1, 7, and 8 were associated with the energy metabolism of F. thermalis ( Figure 4B and Table S4). In general, cluster 1 was associated with amino acid metabolism and glycolysis, suggesting an association with carbon metabolism, whereas clusters 7 and 8 were strongly associated electron transfer processes. Cluster 7 contained genes associated with the PSII reaction center and cytochrome b 6 f subunits, indicating a primary association with photosynthesis. Finally, cluster 8 harbored genes associated with NDH-1 and cytochrome C oxidase, as well as some structural PSII genes, indicating an association with both photosynthesis and respiration. BioCyc database [59]. As a result, we identified 710 genes that had an r value ≥ 0.85 with at least one other gene. Based on the correlation levels, we built a correlation network that exhibited a total of 3679 connections forming a total of 19 clusters with at least 5 genes (Figure 4a). Using the BioCyc platform [59], we analyzed which metabolic pathways were overrepresented in clusters 1, 6, 7, 8, and 12 (Table S4), corresponding to the five clusters with higher number of nodes. We found that clusters 1, 7, and 8 were associated with the energy metabolism of F. thermalis (Figure 4b and Table S4). In general, cluster 1 was associated with amino acid metabolism and glycolysis, suggesting an association with carbon metabolism, whereas clusters 7 and 8 were strongly associated electron transfer processes. Cluster 7 contained genes associated with the PSII reaction center and cytochrome b6f subunits, indicating a primary association with photosynthesis. Finally, cluster 8 harbored genes associated with NDH-1 and cytochrome C oxidase, as well as some structural PSII genes, indicating an association with both photosynthesis and respiration. Additionally, to gain a better understanding of how the energy metabolism of F. thermalis is connected to the thermal gradient of the El Tatio hot springs, we reconstructed a metabolic network of F. thermalis based on the BioCyc database ( Figure S3). Some steps of the TCA-cycle were curated manually based on the literature (See Section 4.8). Subsequently, we merged our correlation data with the energy metabolism network of F. thermalis ( Figure 5). As expected, PSII was mostly associated with clusters 7 and 8, as well as with the cytochrome C oxidase and NDH-1 complexes. Conversely, glycolysis was mostly associated with cluster 12, which is related to sugar metabolism (Table S4). These results Additionally, to gain a better understanding of how the energy metabolism of F. thermalis is connected to the thermal gradient of the El Tatio hot springs, we reconstructed a metabolic network of F. thermalis based on the BioCyc database ( Figure S3). Some steps of the TCA-cycle were curated manually based on the literature (See Section 4.8). Subsequently, we merged our correlation data with the energy metabolism network of F. thermalis ( Figure 5). As expected, PSII was mostly associated with clusters 7 and 8, as well as with the cytochrome C oxidase and NDH-1 complexes. Conversely, glycolysis was mostly associated with cluster 12, which is related to sugar metabolism (Table S4).
These results show a strong association between photosynthesis and respiration, at both the transcriptional and functional levels in F. thermalis. show a strong association between photosynthesis and respiration, at both the transcriptional and functional levels in F. thermalis.

Figure 5.
Co-expression of energy metabolism genes in F. thermalis. Gene co-expression clustering ( Figure 4) was integrated with the energy metabolism network of F. thermalis ( Figure S3). The node color denotes the corresponding gene co-expression cluster. The link color denotes the corresponding metabolic pathway, with intermedial colors indicating an association with more than one pathway. Node labels were omitted to prevent text overlapping. Respiration and photosynthesis complexes are indicated with white labels. Network Cytoscape file can be found in the Supplemental Material.

Effect of Temperature on the Photosynthetic Efficiency of Fischerella thermalis
Thus far, our results have shown that photosynthesis and carbon-related pathways are probably overexpressed near the reported photosynthetic limit of F. thermalis. However, it was necessary to understand how this is related to the photochemical activity of PSII. To that end, we evaluated the effect of temperature on PSII efficiency using cultures of the F. thermalis strain PCC 7521. Our preliminary assays showed that the Fm values obtained were lower than those obtained of Fp (data not shown), so we opted to measure the Kautsky effect under continuous light conditions. The resulting Fv values were calculated using Fp-F0 instead of Fm-F0 [60]. Quenching-associated measurements were discarded. As in previous F. thermalis PCC 7521 gene transcription experiments performed in this study, the cultures were subjected to different temperature treatments from 40 °C to 65 °C for 2 h, after which chlorophyll fluorescence was measured. To obtain reliable  Figure S3). The node color denotes the corresponding gene co-expression cluster. The link color denotes the corresponding metabolic pathway, with intermedial colors indicating an association with more than one pathway. Node labels were omitted to prevent text overlapping. Respiration and photosynthesis complexes are indicated with white labels. Network Cytoscape file can be found in the Supplemental Material.

Effect of Temperature on the Photosynthetic Efficiency of Fischerella thermalis
Thus far, our results have shown that photosynthesis and carbon-related pathways are probably overexpressed near the reported photosynthetic limit of F. thermalis. However, it was necessary to understand how this is related to the photochemical activity of PSII. To that end, we evaluated the effect of temperature on PSII efficiency using cultures of the F. thermalis strain PCC 7521. Our preliminary assays showed that the Fm values obtained were lower than those obtained of Fp (data not shown), so we opted to measure the Kautsky effect under continuous light conditions. The resulting Fv values were calculated using Fp-F 0 instead of Fm-F 0 [60]. Quenching-associated measurements were discarded. As in previous F. thermalis PCC 7521 gene transcription experiments performed in this study, the cultures were subjected to different temperature treatments from 40 • C to 65 • C for 2 h, after which chlorophyll fluorescence was measured. To obtain reliable Fv/Fp measurements, F. thermalis PCC 7521 was kept under dark conditions and pretreated with the PSII inhibitor DCMU for 2 h before temperature treatment, as has been reported in studies with other cyanobacteria [61,62].
As a result, we found that the Fv/Fp of F. thermalis was significantly affected by temperature after 2 h of treatment ( Figure 6 and Tables 1 and 2). Fischerella thermalis was most efficient at 40 • C and 45 • C (Fv/Fp = 0.27). The Fv/Fp value then decreased with increasing temperature to 0.18 and 0.17 at 50 • C and 55 • C, respectively, and finally to 0.11 at 60 • C, while no fluorescence was detected at 65 • C. The loss of fluorescence is consistent with the transcriptional decrease in the psbA and psbD genes at 60-65 • C, but not with the levels observed at lower temperatures ( Figure 2B,C). The highest Fv/Fp value of F. thermalis was 0.27, which was lower than the maximum photosynthetic yield reported for other cyanobacteria (usually about 0.6 [61]). As our results were based on Fp and not Fm, further studies based on Fm are needed to compare the photosynthetic efficiency of F. thermalis to other cyanobacteria.
Fv/Fp measurements, F. thermalis PCC 7521 was kept under dark conditions and pretreated with the PSII inhibitor DCMU for 2 h before temperature treatment, as has been reported in studies with other cyanobacteria [61,62].
As a result, we found that the Fv/Fp of F. thermalis was significantly affected by temperature after 2 h of treatment ( Figure 6 and Tables 1 and 2). Fischerella thermalis was most efficient at 40 °C and 45 °C (Fv/Fp = 0.27). The Fv/Fp value then decreased with increasing temperature to 0.18 and 0.17 at 50 °C and 55 °C, respectively, and finally to 0.11 at 60 °C, while no fluorescence was detected at 65 °C. The loss of fluorescence is consistent with the transcriptional decrease in the psbA and psbD genes at 60-65 °C, but not with the levels observed at lower temperatures (Figure 2b,c). The highest Fv/Fp value of F. thermalis was 0.27, which was lower than the maximum photosynthetic yield reported for other cyanobacteria (usually about 0.6 [61]). As our results were based on Fp and not Fm, further studies based on Fm are needed to compare the photosynthetic efficiency of F. thermalis to other cyanobacteria.

Possible Loss of D1 Gene Copies and Retention of the Photoresistant D1 INT Isoform in Fischerella thermalis
Photosynthesis is the main energy metabolism mechanism of cyanobacteria. Our results indicate that photosynthesis in F. thermalis is negatively affected by temperature. We found that the copy number of the classical G4 D1 isoform of the PSII reaction center in F. thermalis was lower than other non-thermal cyanobacteria of the genus Fischerella ( Figure 1B). This may suggest a process of genomic optimization, as genome reduction has previously been associated with adaptation to thermal environments in other bacteria [63]. Most genomes belonging to the genus Fischerella were highly fragmented (Table S1), thus it is also possible than the lower G4 copy number in F. thermalis is because of assembly related issues, especially considering the high similarity of the psbA-related sequences. However, we found that some copies of G4-psbA found only in mesophilic Fischerella were located near transposase-coding genes ( Figure 1D), suggesting that these potential transposon-associated G4-psbA copies were either lost during the F. thermalis speciation process or originated as part of a genome expansion exclusive to mesophilic Fischerella. Conversely, two of the three psbA genes encoding the G4 isoform in F. thermalis were located near other photosynthesis-related genes, such as phycobilisome-rod-encoding genes and the D1 INT -psbA gene ( Figure 1C). This suggests that high-temperatures conditions may have favored the clustering of photosynthesis-related genes rather than the transposonassociated G4-psbA copies. The synteny of the G4-psbA copies was consistent between the genomes of the mesophilic Fischerella and the same was observed for the F. thermalis genomes (Table S2 and Supplemental Information). Thus, this strongly suggests that the observed differences are due to differences between the genomes of F. thermalis and the mesophilic Fischerella. Still, the low availability of unfragmented Fischerella genomes prevented us from fully confirming loss of the G4-psbA copies. As such, further studies on the evolution of the genus Fischerella are now required to fully understand the genome optimization process of F. thermalis. Interestingly, D1 INT and D1 FR isoforms were previously recognized as part of the G4 group [38], but have been more recently suggested as new groups positioned between G3 and G4 [39]. Previous studies indicate that F. thermalis can use far red-light for photosynthesis by undergoing a complex acclimation process known as far-red light photoacclimation (FaRLiP), which leads to the formation of the red-shifted chlorophylls d and f and the incorporation of these pigments into newly synthesized photosystems [64][65][66]. Therefore, the presence of D1 FR (Figure 1), which is associated with the FaRLiP gene cluster [65], is expected in combination with G1-D1 encoding the chlorophyll f synthase [67] and the phycobilisome-linker protein ApcE2 [5,39,65]. Relative to the standard G4 isoform, D1 INT [39] possesses specific amino acid changes that interact with electron acceptors within the PSII reaction center [39,[68][69][70]. Hence, the increased D1 INT -psbA transcription could be associated with changes in the electron transfer activity, as we found that electron transfer chain activity and carbon metabolism also respond to high temperatures (Figures 2-5). Although the physiological characteristics of D1 INT require further study, it is noteworthy that this potentially photoresistant D1 isoform is retained in the thermal cyanobacterium F. thermalis. As temperature can aggravate PSII photoinhibition in photosynthetic organisms [71], we hypothesize that retention of the D1 INT isoform may be associated with a strategy to overcome adverse thermal conditions by reducing the associated negative consequences rather than increasing PSII thermotolerance.

Photosynthesis-Related Genes Are Differentially Regulated by Temperature in Fischerella thermalis
The photosynthesis-related genes in F. thermalis seem to be largely down-regulated by temperature (Figure 2A), except for psbA and psbD, whose transcript levels increased with increasing temperature up to~60 • C ( Figure 2B,C). In this study, we found that Fv/Fp, which can be associated with the light used by PSII under non-saturation conditions, decreases with temperature in F. thermalis (Figure 6), being consistent with previously reported photosynthetic analyses for F. thermalis [72,73]. Although the measured Fv/Fm in cyanobacteria is usually lower than the effective Fv/Fm [62], our Fp-based results show a clear effect of high temperature over Fv/Fp. Given the fact that PSII is highly susceptible to high-temperature stress [14], our results suggest that PSII complexes in F. thermalis are being destabilized by high temperature. Both the Fv/Fp measurements and the overall transcriptional pattern observed for photosynthesis-related genes are consistent with a decreasing number of PSII complexes as temperature increases. However, the increased transcription of psbA and psbD is not consistent with a decrease in the photosynthetic efficiency. Furthermore, the transcriptional levels of groEL suggest an active heat shock response at high temperatures ( Figure S2), which probably affects the stability of cellular components and the photosynthetic process [52][53][54][55][56]. A possible explanation for this is that the transcription of psbA and psbD is related to specific D1 and D2 temperature-associated turnover requirements rather than the photosynthetic efficiency. It is known that D1 participates in photochemical reactions, thereby requiring higher turnover rates than D2 to maintain PSII [37]. Thus, the transcription of psbA and psbD could be increased to compensate for the temperature-induced destabilization of PSII, which would fulfill their turnover requirements and maintain PSII activity over a wide temperature range. At temperatures above 60 • C, thermal damage to PSII may be irreparable or the metabolic cost of maintaining functional photosynthesis may be higher than the energy produced, thus leading to loss of photosynthetic activity. Another possibility is that photosynthesis-related genes are post-transcriptionally regulated by non-coding RNA (ncRNA) [74]. Although it has been reported that photosynthesis-related genes in cyanobacteria are mostly transcriptionally regulated [75], the inconsistency uncovered in the present study between F v /F m and psbA activity could also be explained by post-transcriptional regulation. Further experiments focusing on, for example, ncRNA and the stability of cellular components (such as mRNA and proteins) should be conducted to understand the role of post-transcriptional regulation on the high-temperature-response of F. thermalis.

Emerging Evidence of Mixotrophic Metabolism by Fischerella thermalis in Relation to Its Thermotolerance
Transcriptional data have revealed that genes related to glycolysis are induced by high-temperature stress in cyanobacteria, such as Synechocystis sp. PCC 6803, [14], while auxiliary energy pathways have been proposed to be relevant for regulating photosynthesis under stress conditions [43]. The mixotrophic behavior of F. thermalis has been previously reported [17,18], but how this relates to its thermotolerant capacities is unknown. In the present study, we found emerging evidence that the mixotrophic metabolism of F. thermalis could be relevant to its thermotolerance. First, we found that, although the transcriptional pattern of glycolysis is variable among genes, the pfk gene, which directs carbohydrates for glycolysis, was induced by high temperature (Figure 3B), suggesting that temperature is associated with organic carbon usage. Furthermore, we corroborated that respiration and photosynthesis are not only metabolically connected, but are also coexpressed processes (Figures 4 and 5). As respiration plays a major role in thylakoid redox regulation and photorespiration, thus preventing photoinhibition [41,42], this co-expression suggests that it is likely related to the regulation and/or optimization of photosynthesis at high temperatures. However, the transcriptional levels of other genes, such as pyk, were not found to be related to temperature; thus, there is still a lack of evidence to support the hypothesis that sugars are being consumed by F. thermalis in response to high temperatures. Further studies are thus required to uncover whether the carbon metabolismrelated response is related only to regulation and optimization of photosynthesis or whether there is a genuine mixotrophic response in F. thermalis.
Overall, our results are consistent with a compensatory photosynthetic response to high temperatures in F. thermalis. Increased transcription of reaction center genes in this organism would allow it to repair its temperature-damaged PSII reaction centers, while respiration prevents damage to the photosynthetic apparatus. Furthermore, even though more information is needed in this regard, the results of this work suggest that respiration and carbon metabolism may help obtain energy from organic carbon sources in the environment. Finally, the selection of potentially photoresistant PSII reaction centers in F. thermalis could help mitigate the negative consequences of living at high temperatures. This thermal adaptation strategy by F. thermalis conceptually relies on transcriptional and metabolic regulation rather than on major changes in protein sequences (although it is compatible with them), which could be especially relevant for highly conserved complexes like PSII. This study provides novel information on the response of F. thermalis to the high temperatures of hot springs and proposes a molecular model of photosynthesis in high-temperature ecological niches. While there are several aspects of this model that remain to be demonstrated (including the biochemical properties of D1 INT and the posttranscriptional regulation of D1 isoforms), we expect that future studies will help us to better understand the photosynthetic response of F. thermalis to high temperatures.

Fischerella thermalis PCC 7521 Culture Conditions and Temperature Treatments
Axenic cultures of F. thermalis PCC 7521 were grown in BG11 0 media at 45 • C and 15.5 µmol m −2 s −1 (6500K LED light, 12/12 h light/dark cycle) until they reached~24 mg chlorophyll/L. Next, 15 mL of culture was placed in individual 25 mL flasks 3 h after turning on the lights to ensure active photosynthesis. Cultures were then immediately subjected to temperature treatments at 40 • C, 45 • C, 50 • C, 55 • C, 60 • C, or 65 • C for 30 min, 1 h, 3 h, or 6 h. Three biological replicates per time/temperature were used. The cultures were subjected to ultracentrifugation at 8000× g for 15 min, after which the supernatant was discarded, and then the pellets were stored in liquid N 2 until analysis.

RNA Extraction
Eleven samples from El Tatio geothermal field (45 • C to 57 • C) were collected (Table  S3) and stored in RNALater™ (Thermo Fisher, Waltham, MA, USA). RNA from F. thermalis PCC 7521 cultures and the El Tatio environmental samples was extracted with 1 mL of TRIzol™ (Thermo Fisher, Waltham, MA, USA) and 250 µL of chloroform, which was added for phase-separation, followed by centrifugation at 12,000× g for 15 min. RNALater™ from the El Tatio samples was diluted by washing with sterile double distilled water prior to TRIzol™ addition. After phase-separation of the TRIzol™-embedded samples, the aqueous phase was transferred to a 2 mL tube and amended with 1 volume of molecular-grade ethanol. The mixtures were subjected to the RNA Clean and Concentrator kit (Zymo Research, Irvine, CA, USA), according to the manufacturer's protocol. To prevent genomic DNA contamination, the RNA samples were incubated for 40 min with DNAse from the "Turbo DNAse kit" (Invitrogen, Waltham, MA, USA), as per the manufacturer's protocol.

cDNA Synthesis
cDNA was synthetized from 1 µg of DNA-free RNA of F. thermalis PCC 7521 using the ImProm-II Reverse Transcription System kit (Promega, Madison, WI, USA) according to the manufacturer's protocol. To improve the specific cDNA synthesis, we used a mix of reverse primers targeting the genes of interest [84]. Primers were designed using Primer3 [85] based on the F. thermalis NIES-3754 genome. Primers are listed in Table S6.

Metatranscriptomic Analysis of Porcelana Hot Spring and El Tatio Environmental Samples
RNA-seq libraries of eleven metatranscriptomes from El Tatio geothermal field samples representing temperatures of 45 • C to 57 • C (Table S3) were prepared and sequenced at the Roy J. Carver Biotechnology Center (University of Illinois at Urbana-Champaign, Champaign, IL, USA). The KAPA HyperPrep kit (Roche, Basel, Switzerland) was used for library preparation and sequencing was performed on the Illumina NovaSeq 6000 on flowcell S1 (2 × 150 bp). Metatranscriptomes representing temperatures of 48 • C, 58 • C, and 66 • C from Porcelana hot spring were retrieved from the NCBI SRA database (SRA ID: SRP104009 [19]). Sequences belonging to F. thermalis were retrieved by alignment with previously assembled bacterial metagenome-assembled genomes (MAGs) from hot spring samples [5,8] using DIAMOND (e-value ≤ 1 × 10 −7 ) [77]. Reads assigned to non-Fischerella species were discarded. Remaining sequences were double-checked by aligning them to the gene set of the recovered F. thermalis MAGs, as well as the F. thermalis NIES 3754 and F. thermalis PCC 7521 proteins using DIAMOND (e-value ≤ 1 × 10 −7 ). Proteins were functionally assigned using eggNOG mapper [88]. Gene counts were summarized according to their respective KEGG ortholog (KO) ID. Gene counts were normalized by quantiles using the command normalize.quantiles() from the R package preprocessCore with default parameters [89].

Co-Expression Network Analysis and Pathway Overrepresentation
Fischerella thermalis gene count correlation was analyzed using the R package WGCNA [90]. Metabolic genes were identified using the BioCyc database [59], with F. thermalis PCC 7521 as a reference. Gene pairs with an r ≥ 0.85 were selected as co-expressed genes. The co-expression network was reconstructed in Cytoscape [91] and submitted for cluster analysis with the Cytoscape package ClusterMaker2 [92]. Isolated gene clusters comprising less than five genes were not considered as clustered. Pathway overrepresentation was then analyzed for the main clusters (see results). Overrepresented pathways for the selected clusters were depicted as a word cloud plot using the R packages ggplot2 [93] and ggwordcloud [94].

Metabolic Network of Fischerella thermalis Reconstruction
Metabolic pathways of F. thermalis PCC 7521 were obtained from the BioCyc database [59]. As cyanobacteria use succinic semialdehyde as an alternative to succinyl-CoA [95], SSADH and 2-OGDC coding genes were identified in F. thermalis PCC 7521 by sequence similarity to Synechococcus sp. PCC 7002 orthologs (SynPCC7002_A2771 and SynPCC7002_A2772) using BLASTp (-k 1000 -e 1e-7 -query-cover 70). Metabolic pathways were parsed as a table using R. The metabolic network was visualized using Cytoscape [91]. To improve network visualization, ATP, GTP, ADP, GDP, and H 2 O nodes were removed from the network owing to their extensive connectivity.

Chlorophyll Fluorescence Measurement of Temperature-Treated Fischerella thermalis PCC 7521 Cultures under Continuous Light Conditions
Axenic cultures (15 mL) from F. thermalis PCC 7521 grown under the same conditions as those used for the temperature treatments were placed in 25 mL flasks under dark conditions overnight. Cultures were pretreated for 2 h with either dimethyl sulfoxide (DSMO; Sigma-Aldrich, St. Louis, MO, USA)) as the control or 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU; Thermo Fisher; 1 × 10 −5 M (10 mM stock solution)). The cultures were then subjected to a 2 h temperature treatment at 40 • C, 45 • C, 50 • C, 55 • C, 60 • C, or 65 • C. Cultures were under dark conditions until fluorometry. Chlorophyll fluorometry was measured using the FluorCam HFC 1000-H fluorometer (Photon Systems Instruments, Drásov, Czech Republic) adjusted to 30% sensitivity with the FluorCam 7 software (Photon Systems Instruments). The default F 0 Kautsky protocol of the FluorCam 7 software (available in the Supplemental Material) was used to measure Kautsky kinetics under continuous light conditions. The F 0 from F. thermalis PCC 7521 cultures acclimated in darkness was measured for 5 s, then the Fp was measured by irradiating the cultures with actinic light (light irradiance 1000 µmol m −2 s −1 ) for 30 s. Then, the Fv/Fp value was calculated as Fp-F 0 /Fp [60]. The results were analyzed by two-way ANOVA in R [96] and graphed with Graphpad Prism 9 (https://www.graphpad.com/; GraphPad Software).

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/molecules27238515/s1, D1 isoform sequence alignments. Genomic contexts for G1-D1 FR , G2, and G4-D1 INT . F 0 Kautsky script for FluorCam 7 software. Cytoscape file containing Figures 4 and 5 networks. Table S1. psbA gene copy numbers summarized by D1 isoform and genome fragmentation. Table S2. Genomic context of D1 INT in the genus Fischerella genomes. Table S3. Summary of environmental samples from El Tatio. Table S4. Overrepresented pathways in the gene co-expression network of F. thermalis from El Tatio samples. Table S5: Summary of detected D1 proteins in the genus Fischerella. Table S6. Primers used for RT-qPCR analyzes. Figure S1. Phylogenetic reconstruction of the genus Fischerella by phylogenomics and 16S rRNA. Figure S2. Transcriptional activity of groEL in F. thermalis at different temperatures. Figure S3. Energy metabolism network of F. thermalis.