Together with poly-ethylene (bioPE), poly-ethyleneterephthalate (bioPET) and poly-lactic acid (PLA), Polyhydroxyalkanoates (PHAs) represent a class of naturally bio-degradable and environmental friendly polyesters. These compounds cover a wide range of possible applications and result in an advantage for the life in the Western countries [1,2]. Among other biomaterials, polyhydroxyalkanoates (PHAs) have attractive physical properties such as thermoplasticity, low crystallinity and high UV-stability [3,4]. These characteristics can be tuned for tailor-made applications like elastic coatings for disposable items [5]. Moreover, biodegradability ensures lower disposal costs and brings environmental advantages [1,2,6-8]. In contrast to synthetic plastics, biopolymers can be entirely produced from renewable sources, such as solar energy, sugars, other carbohydrates, lipids and CO₂. Accordingly, PHAs production via microbial cell factories acquired a significant interest with the ultimate goal of replacing oil-derived synthetic plastic materials, even if the information at the genetic and enzymatic level are still limited [9].

Taking into account the market demand of a “green” PHA production, many biotechnological processes are evolving toward plant-based bioplastic production, (e.g. in Arabidopsis thaliana and Nicotiana tabacum) with the disadvantage of long time process [10-12]. At present, the major industrial process for bioplastic production efficiently utilizes heterotrophic bacteria fermentation, even if with high production costs and the utilization of chemical compounds [13]. The heterologous expression of PHA bacterial operons in microalgae, e.g. the diatom Phaeodactylum tricornutum [14], is an attempt to overcome the chemical supplementation but with difficulties in genetic manipulations.

Cyanobacteria represent one of the most promising microbial cell factories [15-18]. These are phototrophic organisms able to convert carbon dioxide into PHAs via the Calvin-Benson cycle. Synthesized PHAs accumulate in storage granules as carbon source, when cyanobacteria growth occurs upon nutrient starvation (e.g. nitrogen limitation) or osmotic stress [19,20].

In the model cyanobacterium Synechocystis sp. PCC6803, PHAs synthesis occurs in the presence of light and starts with the condensation of two Acetyl-CoA molecules by PhaA (acetyl-CoA-acetyltransferase), generating Acetoacetyl-CoA, then reduced by PhaB [3-oxacyl-(acylcarrier- protein) reductase 2] to (R)-3-hydroxybutyryl-CoA. At this stage, a heterodimer complex composed of PhaE [poly(3-hydroxyalkanoate) synthase component] and PhaC [poly(3-hydroxyalkanoate) synthase] polymerizes the Hydroxybutyryl-CoA to Polyhydroxybutyrate [21]. Recently, the activity of the co-expressed PhaE and PhaC in a cell-free system was determined and the values obtained were comparable to those of PHA synthases belonging to class I [22]. The PhaE-C complex activity is essential for PHA polymerization but not crucial for the PHA yield, suggesting the involvement of significant regulative mechanisms combined to photosynthetic activity and glycogen biosynthesis [22-24]. Several attempts of genetic modification were performed mainly through the transfer of genes belonging to heterotrophic bacteria [25]. Interesting amounts of data were obtained for Synechococcus PCC7942, of which the genetic improvement allowed to an increment of PHA cell content up to 60% in two-weeks fermentation [26,27]. Furthermore, genetically modified transconjugants of Synechocystis sp. PCC6803 produce PHB up to 7% per dry cell weight (12-fold higher than the control), when heterologous PHA genes from Mycrocystis aeruginosa are expressed [28].

A bigger hurdle is the amount of PHA that cyanobacteria can accumulate natively. In our laboratory, we screened multiple cyanobacteria strains for their capability to convert CO₂ in PHAs and the productivity of Synechocystis cf. salina PCC6909 remained the most promising as it achieved, natively, up to 9 g/L cell mass and 12% PHA content after 21 days of autotrophic growth (see Figure S1 as indication). On the basis of the data obtained from S. salina, we oriented our studies toward the genetic strain improvement for PHA production. As the genome data of the latter are not yet available, as first we identified and sequenced the S. salina pha genes and we compared them to the pha genes of other cyanobacteria strains. This analysis is complemented with simulations of Pha proteins topology and their tridimensional structure. We are confident that the presented data will contribute to the comprehension of the PHA biosynthesis at genetic and enzymatic level, essential for future applications in the biotechnology of PHA production via cyanobacteria as “green” microbial cell factories.

Strains and cultivation conditions

Synechocystis cf. salina strain PCC6909 (CCALA 192, sub Gloeothece membranacea) was cultivated in Erlenmeyer flasks on a rotary shaker at 30°C in BG11 liquid medium. During growth, the culture was illuminated by a high-pressure gas discharge bulb (Philips HPI-T, 250W) achieving an illumination intensity of 5000 lux at 4500K colour temperature with an artificial day to night ratio of 16 to 8 hours. JM109 Escherichia coli strain (Sigma-Aldrich) was used for routine DNA analysis, grown at 37°C in Luria-Bertani medium containing 50 mg/ml of ampicillin as selective antibiotic.

Polymerase chain reaction for identification of pha genes

The identification of S. salina pha genes was performed by PCR reaction using genomic DNA as template, obtained from heating treatment of biomass resuspended in sterile distilled water. The suspension was heated at 95°C for 20 min and cooled down on ice. 2 μL of supernatant were used as template in the PCR reaction performed by Phusion^® High-Fidelity Taq polymerase (Thermo Scientific) following the manufacturer protocol. Primer used to identify S. salina pha target sequences were designed on the bases of Synechocystis PCC6803 pha gene sequences annotated in CYORF (Cyanobacteria Annotation Database, http://cyano.genome.ad.jp). Primer sequences are indicated in Table S1.

Sequences isolation

PCR products were purified by MinElute gel extraction kit (QIAGEN) and an A-tailing reaction was performed for each amplified product in accord to Kobs [29] for cloning into pGEM^®-T vector (Promega) following the manufacturer instructions. E. coli JM109 cells (Invitrogen) were transformed with the ligation reaction and plasmids extracted from positive colonies utilizing MiniPrep kit (QIAGEN). Plasmids were used as templates for amplification of isolated fragments, adopting PHA primer sets listed in Table S1.

Sequencing, codon usage and phylogeny

Sequencing of purified DNA fragments obtained from PCR amplifications and of plasmids containing the insert of interest was performed by GATC Biotech AG (European Genome and Diagnostics Center, Constance, Germany) and LGC Genomics (http://www.lgcgroup.com/). Sequences similarity searches were performed in silico by nucleotide BLASTn tool from NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and BLASTn/BLASTp from CyanoBase (http://genome.microbedb.jp/blast/blast_search/cyanobase/genes). Obtained nucleotide and amino acid sequences were deposited in GenBankTM with the following accession numbers: phaE, # KR231685; phaC, # KR231684; phaA, # KR231686; phaB, # KR231687. Amino acid sequence analysis was carried out using ClustalW algorithm [30] and GeneDoc software [31]. Protein domains were detected by Prosite Tool (http://prosite.expasy.org/).

Codon preferences of pha sequences were determined by GCUA (Graphical Codon Usage Analyser) analyser at gcua.schoedl.de [32].

Neighbour-joining phylogenetic trees were generated from multiple sequence alignments using ClustalW2 (http://www.ebi.ac.uk/Tools/phylogeny/clustalw2_phylogeny/) and displayed by iTOL tool (http://itol.embl.de; [33]).

Protein structure determination

The 3D modelling of S. salina Pha proteins was achieved using iTASSER server (http://zhanglab.ccmb.med.umich.edu/I-TASSER/; [34]). PDBsum tool (http://www.ebi.ac.uk/pdbsum/; [35,36]) was used to analyse the secondary structure, the topology and the predicted protein clefts. The 2D membrane topology was predicted by PRED-TMBB (Prediction of Transmembrane Beta-Barrel; http://bioinformatics.biol.uoa.gr//PRED-TMBB/input.jsp) server. Image manipulations were performed by using GNU Image Manipulation Program (Table S2).

Isolation of phaA-B_Syn6909 and phaE-C_Syn6909 gene clusters

The entire genome (3,957 Mbp) of Synechocystis sp. PCC6803 was sequenced in 1996 [37-40], allowing identification of genes responsible for the natural production of PHA in cyanobacteria. In contrast, genomic data of Synechocystis cf. salina PCC6909 are not available yet. We therefore adopted the annotated PHA sequences of Synechocystis sp. PCC6803 (slr1992, 1436699-1437163; sll1906, 1439487-1440941; slr1828, 931639-931959; sll1736, 934324-934707) to amplify the genes involved in S. salina PHA biosynthesis (Figure 1A). Two fragments of ca. 2300 bp and ca. 2100 bp were obtained and sequenced. A similarity search analysis performed by BLASTn tool recognized two open reading frames of 993 bp and 1137 bp merged into the DNA fragment of 2300 bp, showing identity of 89% and 88% to phaE_Syn6803 (slr1829) and phaC_Syn6803 (slr1830) genes, respectively. Interestingly, significant similarities were also detected for phaC ORFs of Arthrospira platensis (73%) and Microcystis aeruginosa (74%). A second similarity search of the S. salina PCC6909 2100 bp-fragment identified two ORF candidates with a similarity of 92% and 90% respectively to phaA_Syn6803 (slr1993) and phaB_Syn6803 (slr1994) genes (Figures S2 and S3). Interestingly, phaA_Syn6909 harbours two HIP1D sequences, also found in the bacterium Haemophilus influenzae (highly iterated palindromic decamer sequence, GGCGATCGCC; [38,40]). The intergenic regions, of 153 bp between genes phaE_Syn6909 and phaC_Syn6909 and 99 bp between genes phaA_Syn6909 and phaB_Syn6909 do not show any peculiar signal, indicating it to be a gene linker. Different primers combinations were tested in order to detect whether phaA-B_Syn6909 and phaE-C_Syn6909 gene clusters were co-linear or located in different genomic loci. The genome co-linearity was demonstrated for phaA and phaB and for phaE and phaC genes but taken together pha A-B and pha E-C were located in different genomic loci, in accord with the PHA gene distribution in Synechocystis sp. PCC6803 [41].

Figure 1:Genomic organization and location of genes involved in polyhydroxyalkanoates synthesis. Synechocystis cf. salina PCC6909 pha gene organization reported in this work (panel A) is compared to the pha gene organization in Synechocystis sp. PCC6803 (panel B), Mycrocystis aeruginosa NIES-843 (panel C), Arthrospira platensis NIES-39 (panel D) and Ralstonia eutropha H16 (panel E). phaA, PHA-specific beta-ketothiolase; phaB, PHA-specific acetoacetyl-CoA reductase; phaE, putative poly(3-hydroxyalkanoate) synthase component; phaC, poly(3-hydroxyalkanoate) synthase. A. In Synechocystis sp. PCC6909 pha genes (black arrows) are pair-grouped. The clusters phaA-B_Syn6909 and phaE-C_Syn6909 are located in two different genomic regions, flanked by unknown genes (gray arrows with question marks). The exact gene location in the genome is unknown (XXX). B. Distribution of pha genes in Synechocystis sp. PCC6803 as annotated in CyanoBase (http://genome.microbedb.jp/cyanobase/) and in CYORF (http://cyano.genome.ad.jp/). Genomic available data were used as reference for our investigation in S. salina PCC6909. Gpx2, glutathione peroxidase; sll1906, hypothetical protein; petF, ferredoxin; sll1736, hypothetical protein. C. Organization of the pha gene cluster in Microcystis aeruginosa NIES-843, as annotated in CyanoBase. Pha genes are grouped in the same genomic region (4581257-4585628). ChlP, geranylgeranyl hydrogenase; MAE_50070, selenide water dikinase. D. Genomic location of pha genes in Arthrospira platensis NIES-39. As in Synechocystis sp., pha genes are pair-grouped in different genomic positions (4469214-4471155 and 6284046-6286220). L000340, SNF2 helicase homolog; L000370, reverse transcriptase homolog; Q00050, hypothetical protein; Q00080, hypothetical protein. E. Pha genes location in Ralstonia eutropha genome. Three copies of phaB (phaB1, phaB2, phaB3) and two of phaC (phaC1, phaC2) genes are present; phaA exists in a single copy. A complete pha cluster composed by phaC1, phaA and phaB1 is located between genomic locations 1557353 and 156203. A second cluster is composed by phaB2 and phaC2 (position 2174303-2176821). The third copy of phaB gene (phaB3) is the sole located between positions 2364912 and 2365622. A1436, hypothetical protein; phaR, transcriptional regulator of phasin expression; A2001, hypothetical protein; A2004, universal stress protein; A2170, ABC transporter ATPase/permease; phaP3, phasin/PHA-granule associated protein.

In silico analysis of Synechocystis cf. salina PCC6909 Pha proteins

As we theorize that PhaE-C synthase complex represents the key enzyme for PHA synthesis, we here focus our attention on the proteins composing that complex (see infra and [42,43]). Additional data related to the reductase (PhaB) and the thiolase (PhaA) are displayed in the Supporting Material.

PhaE_Syn6909: S. salina PhaE_Syn6909 gene is predicted to codify a protein of 330 amino acids and 38 kDa, with an isoelectric point of 5,61. A BLASTp analysis of PhaE_Syn6909 deduced amino acids sequence detected identity of 93% to PhaE_Syn6803, and of 47% and 42% for the corresponding protein in Cyanothece PCC7425 and Synechocystis PCC7424. A Prosite scan analysis recognized two putative coiled-coil domains. The latter are indicated in Figure 2 (grey boxes) as coiled-coil domain 1 (Cc1), harbouring conserved Tyr94 and Gln101, and coiled-coil domain 2 (Cc2), with numerous conserved residues. It is reported that these domains are important in protein-protein interactions for the assembly of protein complexes [44,45]. As expected, a PhaE box (PTRSE; in Figure 2, dark grey box of Synechocystis group), usually conserved among cyanobacteria, is also detected in S. salina PhaE (residues 296-300) [46] and it appears identical to that one of Synechocystis sp. PCC6803, Synechocystis sp. PCC6714 and Microcystis aeruginosa. In the PhaE box of other genera, the Thr297 is replaced by a Leu (Cyanothece, Arthrospira and Chlorogloeopsis) or a Val (Pleurocapsa) residue. The mentioned amino acids string was also found in PhaE proteins of the sulfur bacteria Allocromatium vinosum, Thiocystis violacea and Thiococcus pfennigii [46], that also have PHA-granule binding strings [47], absent in S. salina PhaE protein.

Figure 2: Alignment of PhaE protein. The identified in Synechocystis cf. salina PCC6909 (GenBank^TM accession n. KR231685) is aligned with PhaE proteins found in cyanobacteria. Highly conserved amino acids are highlighted (light gray columns and letters beneath). The coiled-coil domain 1 (Cc1) and coiled-coil domain 2 (Cc2), respectively comprised between residues Thr94, Met116 and Val302, Glu326 are indicated by the gray blocks. The location of PhaE box (PEb), peculiar of cyanobacterial PhaE proteins, is located between residues Pro297 and Glu301. Cyanobacteria harboring the PhaE aligned sequences are in order: Synechocystis sp. salina PCC6909 (Syn6909); Synechocystis sp. PCC6803 (Syn6803); Synechocystis sp. PCC6714 (Syn6714); Microcystis aeruginosa NIES-843; Cyanothece PCC7425; Arthrospira platensis NIES-39; Chlorogloeopsis fritschii; Pleurocapsa minor; Spirulina subsalsa.

PhaC_Syn6909: S. salina PhaC_Syn6909 nucleotide sequence encodes for a protein of 378 amino acids and 43 kDa, with an isoelectric point of 4, 79. A BLASTp analysis indicates an identity of 95% to PhaC_Syn6803 while it is not higher than 74% and 73% for the corresponding protein of Arthrospira platensis and Microcystis aeruginosa respectively. The amino acids sequence harbours a conserved substrate-binding site (SBs), consisting of 18 residues (aa 157-173; SBs box in Figure 3) and displaying the conserved Cys164 and Thr159, usually involved in the protein-substrate interaction [46]. The other synthases analysed show the Thr159Asp substitution, as also detected in the PhaC1 synthase of Ralstonia eutropha. A second conserved cyanobacterial box (CYb, Figure 3, grey box, aa 203-212) harbouring a cysteine residue at position 206 [46], was also recognized. A high sensitive Prosite scan of PhaC_Syn6909 detects a leucine-zipper domain (bZIP) at C-terminal end (aa 311-332; Figure 3), commonly involved in gene regulation of eukaryotic systems and promoting the protein dimerization though coiled-coil domains [48]. A leucine-rich repeat (LRR) profile, also implicated in macromolecular interactions, is recognized at the N-terminal end (aa 26-48; data not shown), even though it shows a low confidence level [49,50]. The N-terminal end is characterized by 13 additional amino acids conserved in Synechocystis PhaC proteins (Figure 3). The amino acids His, Trp and Lys, at positions 9, 10 and 12 respectively, are important target amino acids for post-translational modifications, not well understood in prokaryotes [51-53]. A codon usage analysis of PhaC_Syn6909 (Figure S4A, red bars) based on the Synechocystis PCC6803 codontable (black bars) resulted in a mean difference of only 9% in codon preference frequency, revealing high accuracy in codon selection [54].

Figure 3: Alignment of PhaC proteins. Synechocystis cf. salina PCC6909 polyhydroxyalkanoate synthase (GenBank^TM accession n. KR231684) is compared to known cyanobacterial PhaC proteins and to the two isoforms (PhaC1 and PhaC2) of the bacterium Ralstonia eutropha. Highly conserved amino acids are highlighted (light gray columns and letters beneath). The substrate-binding site (SBs; aa 157-174) is indicated in dark gray for Synechocystis group and the crucial residue Cys164 for substrate binding is highlighted. A cyanobacterial box (CYb) between amino acids 203-212 is highlighted in Synechocystis group. A leucine-zipper domain (bZIP) comprises the residues 311-332. At the N-terminus an additional amino acids string is conserved in the Synechocystis genus only. Details are described in the text.

PhaA_Syn6909: The predicted molecular weight of PhaA_Syn6909 protein (409 aa) is 43.2 kDa with an isoelectric point of 5.79. A BLASTp similarities search resulted of 96% amino acids identity to PhaA_Syn6803 and not higher than 75% and 69% to the corresponding protein of A. platensis and Cyanothece PCC6425. A Prosite analysis detects the typical thiolase 2 signature (Ts2 box, Figure S2, aa 355-371), frequently found in prokaryotes and involved in the thiolysis of acetoacetyl-CoA and a thiolase 3 signature (Ts3, Figure S2, aa 390-403). Here the conserved Cys395 is recognized as proton acceptor for the substrate activation (Figure S2) and a leucine zipper domain (bZIP) is located between amino acids 63-84 (Figure S2). Residues Glu331, Asn333 and Arg373, involved in the hydrogen bonding network within the active site, are also conserved (Figures S2 and S5). A typical hydrophobic-rich box (Figure S2, aa 133-158), is most probably a tetramerization domain [51]. As for PhaC, the PhaA_Syn6909 annotated sequence exhibits 13 additional amino acids at the N-terminal end. Also here, some residues (Pro, Asn, Lys) are probably subjected to post-translational modification events [51-53]. Panel B of Figure S4 shows the frequency of preferred codons which differ of ca. 10% to the reference codontable, indicating an accurate codon selection in the translation process.

PhaB_Syn6909: S. salina PhaB_Syn6909 is a protein of 240 amino acids and 23.31 kDa with an isoelectric point of 6.18. A BLASTp analysis detected high identity up to 99% to PhaB_Syn6803 and of 77% and 74% to the corresponding protein in M. aeruginosa and Synechocystis PCC7424 respectively. A Prosite analysis of PhaB_Syn6909 protein detects a short-chain dehydrogenase/reductase (SDRs) family signature (SDRs, Figure S3, aa 134-162) and the protein is biochemically classified as ‘classical’ as the length does not exceed 250 residues [55,56]. In Synechocystis genera, the SDR coenzyme binding motif involves three glycine residues (Gly15, 19 and 21 in L24 box in Figure S3), that together with Thr14 form the NADP binding-specific sequence; Ser134, Tyr147 and Lys151 represent the catalytic residues (SDRs box of Figure S3, [51,57]). The N-terminal end highlights an amino acid string (Figure S3, aa 7-25) similar to the ribosomal protein L24 signature, peculiar of proteins located in the large ribosomal subunit.

Phylogeny of Pha proteins

The evolutionary relationships among PHA synthesizing cyanobacteria and the out-group bacterium Ralstonia eutropha were examined on the basis of the four amino acid sequences involved in the S. salina PHA biosynthesis. The inter-species relationships here reported provide information on the phylogenetic evolution of the PHA biosynthetic pathway.

In Figure 4, S. salina PHA synthase component (PhaE) and PHA synthase (PhaC) were investigated for their phylogenetic relationships with representative species of diverse cyanobacteria genera (Synechocystis, Microcystis, Arthrospira, Chlorogloeopsis, Fischerella, Pleurocapsa and Cyanothece) and the non-cyanobacterium Ralstonia. On the basis of PhaE aminoacid sequences (Figure 4A), S. salina PCC6909 and Synechocystis sp. PCC6803 form a single auto-collapsed clade with an average phylogenetic distance less than 0.05. A close relation with Synechocystis PCC6714 is also evident. Gloeocapsa sp., Cyanothece PCC7425 and Pleurocapsa minor form a second clade, more distant to the Synechocystis group. Here, Pleurocapsa and Cyanothece exhibit a closer phylogenetic relation conversely than Gloeocapsa, even if all three organisms originate from a common ancestor. A third clade is represented by Arthrospira platensis NIES-39 and Microcystis aeruginosa NIES-843. At variance with PhaE, the PhaC phylogenetic tree (panel B) exhibits a common ancestor between the Synechocystis and Microcystis PHA synthases. A separate monophyletic group is composed of all the other cyanobacteria species here considered. Ralstonia eutropha is included in the tree as out-group strain, as it possesses two isoforms of phaC gene in the genome (see Figure 1, panel E, phaC1 and phaC2).

Figure 4: Phylogenetic analysis of PhaE and PhaC proteins in cyanobacteria. A. On the basis of the PhaE synthase protein sequences, Synechocystis sp. salina PCC6909 (this work) and Synechocystis sp. PCC6803 (P73389/Slr1829) constitute an auto-collapsed clade (gray triangle) and are close related to Synechocystis sp. PCC6714 (TIGR01834). Gloeocapsa sp. PCC73106 (L8LSA0) Pleurocapsa minor sp. PCC7324 (K9TAX6) and Cyanothece PCC7425 (B8HVQ8) share a common ancestor, as Microcystis aeruginosa NIES-843 (B0JWT9) and Arthrospira platensis NIES-39 (D4ZNW7). B. Phylogenetic tree based on PhaC protein sequences of Ralstonia eutropha H16 (PhaC1, P23608; PhaC2, Q0KA68), Microcystis aeruginosa NIES-843 (B0JWT8), Synechocystis sp. PCC6803 (P73390), Synechocystis sp. PCC6714 (A0A068MZI4) Arthrospira platensis NIES-39 (D4ZNVW6), Chlorogloeopsis fritschii (Q8RTL8), Spirulina subsalsa (WP_026079979.1), Cyanothece sp. PCC7425 (B8HVQ9), Fischerella sp. PCC9605 (WP_026732742), Pleurocapsa minor (K9T9D7). The branch lengths and the reference scale are indicated.

Supplementary Figure 6 illustrates the evolutionary lineages calculated on the basis of PhaA (panel A) and PhaB (panel B) proteins of S. salina PCC6909 and of other PHA-synthesizing species. In panel A, the evolutionary relationship among the different PhaA proteins is mainly based on the common thiolase signatures (Figure S2; [58]). The phylogenetic tree is separated in two clades, generated by an ultimate common ancestor. S. salina PhaA appears phylogenetically related (distance not higher than 0.04) to the corresponding proteins of Synechocystis sp. PCC6803 and Synechocystis PCC6714, represented in the tree as an auto-collapsed clade (leaves distance <0.05). Moreover, the thiolases of the Synechocystis group, of Arthrospira platensis and Spirulina subsalsa originate from a common ancestor. A second clade is represented by Microcystis aeruginosa, Pleurocapsa minor, Cyanothece PCC7425, Chlorogloeopsis fritschii and Fischerella sp. Interestingly, in the clade, only Fischerella belongs to a different order, namely Stigonematales. As expected, Ralstonia possesses the highest phylogenetic distance (0.41). The phylogeny based on PhaB proteins again exhibits a short distance between Synechocystis and Microcystis genera (Figure S6B). The reductases of the genera Synechocystis, Microcystis, Arthrospira, Pleurocapsa, Chlorogloeopsis, and Fischerella, originate from a common ancestral population a situation similar to that one described for PhaA. Interestingly, also the three PhaB isoforms of Ralstonia belong to the latter group (see Figure 1, phaB1, phaB2 and phaB3) while Spirulina and Cyanothece establish an out-group.

Figure 5: Secondary structures and topology diagrams of PhaE and PhaC proteins in Synechocystis sp. salina PCC6909. Data are represented as predicted by PDBsum tool. A,B. “Wiring diagrams” of the PhaE and PhaC secondary structures, showing α-helices and β-sheets plus additional motifs as β- and γ-turns and the corresponding amino acid strings. In panel A, PhaE does not show β-strands otherwise PhaC exhibits seven β-strands (indicated by arrows) belonging to the same β-sheet (pointed out as red “A”). Helices are counted and indicated (H1-H20 for PhaE; H1-H15 for PhaC). C,D. Topology diagrams of PhaE (panel C) and PhaC (panel D) proteins. They represent how the structural elements building the secondary structure (β-sheets and α-helices) are organized within the space and how these elements are linked to each other. Red cylinders represent the α-helices location. Each large arrow indicates a single β-strand, forming a β-sheet in PhaC. The N- to C-terminal orientation of the protein is indicated by thin arrows.

Figure 6: 3D modeling and volumetric clefts analysis of Synechocystis sp. salina PCC6909 PhaE and PhaC proteins. Structures are represented as predicted by iTASSER software and PDBsum tool. A,B. Comparison between PhaE 3D model (A) and the predicted location of protein cavities (B). A common Cartesian laboratory reference system was chosen for both drawings. The red cleft is the cavity (volume of 1635.61 Å3) with higher probability of being the interaction site. Additional smaller clefts are indicated in the Figure by different colours. C,D. 3D model and clefts analysis of PhaC protein. The red cavity (largest volume, 2305.55 Å3) represents the predicted interaction site. The model manipulation was done using Jmol software. Detailed values about the protein folding and clefts are described in supporting table S2 and S3.

Prediction of Pha_Syn6909 protein structures

To shed light on S. salina Pha proteins, we focused our attention on the structural data obtained by the sequence analysis. Using PDBsum tool [36] and iTASSER server [34], we investigated the secondary structures and topologies of PHA proteins, together with the 3D models and the cleft distributions.

PhaE_Syn6909: The secondary structure organization and the topology of the S. salina PhaE synthase component are illustrated in Figure 5 (panels A and C). The secondary structure displays 18 helices (Figure 5A, H1-H18) involved in 25 helix-helix interactions, while β-sheets and β-hairpins are absent. The protein topology represented in panel C shows a α-helices organization of which N- and C- terminal ends are oriented on the same side. In Figure 6A, the PhaE_Syn6909 3D structure is compared to the location of the predicted protein clefts. A cavity with an estimated volume of 1635.61 Å3 and an average depth of 12.21 Å is indicated as the putative active site (red cleft in Figure 6B and Table S3). The accessible (buried) vertices are 55.36 Å (9.e negative (Asp115, Asp117 and Glu126) and three positive (Arg113, Lys125 and Lys129) residues.

PhaC_Syn6909: A PDBsum analysis of the secondary structure and topology of PhaC_Syn6909 synthase is represented in Figure 5 (panels B and D). Our investigation detected 15 α-helices, involved in 15 helix-helix interactions, and one β-sheet motif, composed of 7 β-strands. Interestingly, the protein harbours 4 β-α-β motifs, where an α-helix usually connects two β-strands. Interestingly, PhaC_Syn6909 contains quite a rare ψ-loop motif, involving residues Leu70/Phe73 in the first strand and Pro84/Val88 in the second strand [59-61]. Figure 6 compares the protein 3D model (panel C) to the clefts location (panel D). A major cavity with a volume of 2305 ų and an average depth of 13.22 Å is indicated (Figure 6D, TableS3). The distance between accessible (buried) vertices measures 72.94 Å (13.22 Å). Of the 49 residues composing the cleft, 18 are aliphatic and cysteines are absent.

PhaA_Syn6909: S. salina acetyl-CoA acetyltransferase is a member of the thiolase type II family, exerting the first step of PHA biosynthesis. In Figure S7, PhaA_Syn6909 secondary structure (panel A) is compared to the protein topology (panel C). The secondary structure shows 20 helices, 14 of which involved in helix-helix interactions, and 3 β-sheets composed by 14 β-strands (Figure S7A and C). The protein contains also 3 β-hairpins, one of which belongs to the class 19:19 and exhibits interaction of Phe136/Tyr137 with Asp158/Thr157 residues (Figure S8). Also for PhaA_Syn6909, the 3D model (Figure S9, panel A) is compared to the predicted distribution of protein clefts (panel B). The major cleft has a predicted volume of 1360 ų (Figure S9B) and it contains 2 cysteines. A prediction of the 2D transmembrane topology detects two putative transmembrane domains, corresponding to residues 9-19 and 40-56 (Figure S10A).

PhaB_Syn6909: The analysis of PhaB_Syn6909 secondary structure (Figure S7, panel B) and topology (panel D) identifies a Rossmann fold motif, which is peculiar of nucleotide binding proteins and cofactors (Figure S6D). The secondary structure counts 10 α-helices (H1-H10), 6 of which show helix-helix interactions. As reported by Kim et al. [62] for Ralstonia eutropha, RePhaB harbours a clamp domain, involved in acetoacetyl-CoA binding, which is difficult to detect in PhaB_Syn6909 even if an indication is given by the amino acids string 183-201 (Figure S3). A deep cleft with a volume of 2814 ų and an average depth of 14.65 Å, harboring the alleged active site is indicated in Figure S9D. As in PhaA_Syn6909, also PhaB_Syn6909 exhibits two putative transmembrane domains (residues 2-12 and 25-37) at the N-terminal end (Figure S10B). It is worth noting that the presence of a transmembrane domain in the 3-oxacyl-(acyl-carrier-protein) reductase 2 is also reported in Nostoc sp. PCC7524 (gene9, BGA database, A Comparative Genomic Resource for Cyanobacteria, unpublished data).

In this work, we focused our studies on Synechocystis cf. salina PCC6909, a promising natural PHA producer. We investigated the origin of pha genes, with the perspective of gaining insights into the key biochemical features that make this organism quite attractive for a strain improvement. As the genome data of S. salina are not yet available, we referred to the related organism Synechocystis sp. PCC6803. We found pair-grouped pha genes in S. salina PCC6909, similar to Synechocystis sp. PCC6803 (Figure 1B) and to Arthrospira platensis (Figure 1D), even though we still do not know their exact location. The genome dissemination of pha genes shows similarities to α-Proteobacteria, probably due to random insertions of exogenous DNA or to fragment transposition.

We isolated two fragments corresponding to phaA-B_Syn6909 and phaE-C_Syn6909 operons which encode the enzymes responsible for PHA synthesis in S. salina. We deduced the protein sequences and investigated the amino acid conservation, predicting the protein domains composition and analyzing the codon preferences. In silico modelling of S. salina PCC6909 PHA enzymes provides sterical information, determining protein structure and function. In particular, the cleft analysis provides a base for the comprehension of protein-protein interaction. In PhaE_Syn6909 we detected two coiled-coil domains that, together with the 3D protein modelling, putatively represent the interacting sites with PhaC_Syn6909 synthase. Observing the 3D architecture of PhaC_Syn6909 and PhaE_Syn6909, we speculate that the peculiar structure of PhaE associates with the major cavity of PhaC, allowing a synthase complex assembly. Our hypothesis is in compliance with previous studies on similar organisms [63]. We further exclude disulphide bridge formation in the phaE-C_Syn6909 complex because of cysteine absence in the PhaC cavity. If proven true, these data can portray one of the key regulatory mechanisms of PHA production. Moreover, the substrate binding box of PhaC_Syn6909 is probably involved in one of the following: a) in the (R)-3-hydroxybutyryl-CoA recognition, b) in the nucleophilic attack and c) in the catalysis of PHA polymerization within granules, as reported for other PHA synthases of class III. The finding that PhaA_Syn6909 contains iterated palyndromic sequences (HIPD1) could facilitate the spontaneous uptake of exogenous DNA in S. salina, as observed in other cyanobacteria. The presence of super-secondary structures such as β-hairpins (e.g. the 19:19 class β-hairpin in PhaA_Syn6909), most probably representing the nucleation sites for the protein folding [64], is a good target for point mutations with the scope of improved enzyme efficiency. Interestingly, the sequence alignment detects 13 additional amino acids at the N-terminal end of S. salina PhaA and PhaC which most probably represent the sites of post-translational modifications [52,53]. Moreover, the presence of extra amino acids can confer translational robustness to the protein sequence, against missense errors. The translational accuracy is also supported by the results of the codon usage analysis of PhaA and PhaC where the proteins differ only with a frequency of ca. 10% to the reference (Figure S4).

The L24 motif of S. salina PhaB is also found in eubacteria, plant chloroplasts and red algae indicating the close relation between cyanobacteria and these organisms [65]. Additionally, the detection of transmembrane domains in PhaA_Syn6909 and PhaB_Syn6909 provides an indication of their sub-cellular organization (Figure S10).

A close relationship between S. salina PCC6909 and the model Synechocystis sp. PCC6803, together with Synechocystis sp. PCC6714 arises from our phylogenetic analysis (Figure 4 and Figure S6). As the mentioned strains belong to the same order, namely Synechoccales, only few amino acids differ in PHA enzymes (Figures 2 and 3; Figures S2 and S3). This evidence suggests that small differences in the amino acid string of PHA proteins do not influence the strain monophyly, resulting from a horizontal gene transfer occurred in several speciation events. On the other hand, the phylogenetic relation of S. salina with other genera of cyanobacteria varies with the protein or protein part investigated. For example the PhaA protein shows a highly conserved thiolase domain in the phylogenetically distant Microcystis aeruginosa and in all analysed Synechocystis. The same facts still hold true when the non-cyanobacterium Ralstonia eutropha is considered. Interestingly, based on PhaB sequence, Spirulina and Cyanothece form a closely related out-group, although they belong to different orders (Chroococcales and Oscillatoriales respectively) (Figures S11-S13).

The identification of pha genes and the description of the predicted protein in S. salina PCC6909 provides important information for the upcoming strain improvement work. In the long term, the knowledge of gene sequences paves the way towards the design of a ‘green’ PHA-production. Accordingly, the results reported in this work represent the base of an ongoing applied research project aimed to the conversion of waste CO₂ into PHAs through photo-autotrophic growth of S. salina PCC6909, currently passing through the biochemical optimization for a production in pilot scale.

This work was supported by the Austrian Climate and Energy Fund and Austrian Research Promotion Agency (FFG). We want to thank to our industrial partners EVN and Andritz for their support.