Comparative analysis of structural and functional aspects of phytoene synthase from Meiothermus taiwanensis strain RP

The aim of this study is the in silico characterization of the structure and function of the phytoene synthase (PSY) of a red carotenoid producing thermophile Meiothermus taiwanensis strain RP with a comparative approach. PSYs from M. taiwanensis strain RP and other groups of thermophilic, mesophilic and psychrophilic bacteria, plants, protozoa, and algae were analyzed by ExPASy ProtParam, NCBI Conserved Domain Search, SOPMA, PSIPRED, Robetta server, ProQ, and QMEAN, with the superposition of 3-D structures in PyMOL. RP PSY shows the highest (97.5 %) similarity with M. ruber and the lowest with the psychrophile Gelidibacter algens (36.7 %). The amino acid sequence of RP PSY is one of the shortest, with 275 residues. The instability index of RP PSY is much lower compared with plant sequences. Alanine, arginine, glycine, and leucine residues are the highest in Meiothermus sp., and they have a high amount of alpha-helix. Most of the 32 active site residues are conserved in all the sequences. However, some residues are more prone to substitutions in other PSYs except M. ruber. The three-dimensional structures of M. taiwanensis strain RP, Gelidibacter algens, Thermus thermophiles, Meiothermus ruber, and Brassica napus PSYs were homology modeled, validated, and submitted to Protein Model Databank. The superposition of the 3-D structures shows that their active site region structure is identical. RP PSY is one of the most stable PSYs and knowledge of its individual properties, similarities, and dissimilarities with other PSYs may be useful for genetic engineering and purification of the protein for improved carotenoid production.


Introduction
Carotenoids are a subfamily of isoprenoids with an extensive conjugated double bond system, acting as chromophores. They are red, yellow, or orange colored and are found in plants, fungi, bacteria, protozoa, algae, and archaea. Higher order animals do not synthesize carotenoids but may obtain them through diet. Carotenoids have a wide variety of functions such as coloring property, anti-oxidant property, conferring membrane stability, and protection against heat and light (Britton et al. 2004;Britton 2008a). The carotenoid beta-carotene acts as the precursor for vitamin A, which is a vital nutrient component of human health (Britton 2008b). Phytoene, the first 40 carbon intermediate carotenoid formed in the carotenoid biosynthetic pathway, is colorless in nature (Paniagua-Michel et al. 2012). However, phytoene is the precursor for the synthesis of colorful carotenoids through structural modifications and the addition of functional groups. Phytoene synthase (PSY) is the enzyme which catalyzes the synthesis of phytoene from the condensation of two molecules of geranylgeranyl diphosphate using manganese as co-factor. This reaction is the rate limiting and first committed step of carotenoid biosynthesis. Thus, it is an essential regulatory enzyme controlling carotenoid pigment formation (López-Emparán et al. 2014). This enzyme is so important that when suppressed it leads to a decreased production of carotenoids (Kato et al. 2017). An in-depth study of this enzyme is crucial to understand, regulate, and engineer carotenoid biosynthesis for increased production of carotenoids. Phytoene synthases have been studied in silico, directly from natural sources and by cloning and expression in host cells (Iwata-Reuyl et al. 2003;Fu et al. 2014;Agarwal et al. 2015). Most of the previous studies on phytoene synthases have focused on plant origin enzymes; studies on phytoene synthases of bacterial origin are few (Iwata-Reuyl et al. 2003). Clearly, we need to focus more on bacterial origin PSYs, especially thermostable ones.
Meiothermus taiwanensis strain RP, belonging to the Deinococcus-Thermus phylum, produces red carotenoid pigments having a high anti-oxidative capacity (Mukherjee et al. 2017). We isolated this thermophilic organism from the Paniphala hot spring of India (Mukherjee et al. 2016). Detailed analysis of phytoene synthases from the Deinococcus-Thermus group of bacteria have not been reported until now. Meiothermus sp. forms red and yellow pigments related to carotenoid glycoside esters. One carotenoid pigment has been identified from M. ruber as 1′-β-glucopyranosyl-3,4,3′,4′-tetradehydro-1′,2′-dihydro-β,ψ-caroten-2-one glucose acetylated at position 6 (Burgess et al. 1999). In our study, we took the sequence of phytoene synthase of Meiothermus taiwanensis strain RP from its whole genome sequence database, which was previously sequenced by us (Mukherjee et al. 2016) and analyzed in silico. We analyzed the protein structurally on three levels-primary, secondary, and tertiary, along with phylogenetic analysis and study of active site residues. Simultaneously, we also carried out the structural, functional, and phylogenetic analysis of phytoene synthases from other organisms for comparison. This comparative analysis revealed crucial facts regarding structural, functional, and phylogenetic similarity and dissimilarity among different phytoene synthases.

Retrieval of phytoene synthase (PSY) sequences
The FASTA format of the phytoene synthase protein sequence of Meiothermus taiwanensis strain RP (KZK15231.1) (Mukherjee et al. 2016) was retrieved from its whole genome database. Other phytoene synthase sequences used in different sections of this study were retrieved from the NCBI protein database (https://www.ncbi.nlm.nih.gov/protein/). After downloading a large number of PSY sequences from NCBI, they were initially checked in NCBI conserved domain search (Marchler-Bauer et al. 2011) for removal of sequences which lacked some or all of the PSY functional domains, or contained multiple unrelated domains indicating unspliced sequences from translated genomic data (observed in case of some eukaryotic sequences). Details of all the phytoene synthase protein sequences finally selected for comparison are given in Table 1. SI 1-6 contain sequences from 6 different species of Meiothermus sp. for intra-genus comparison. SI 7-13 contain sequences from Deinococcus-Thermus phylum for intra-phylum analysis. SI 1-13 formed a comparative entity for analyzing any patterns arising in thermophilic PSYs. SI 14-22 show some sequences selected from mesophilic bacteria, psychrophilic bacteria, and cyanobacteria and were selected for intra-domain comparison among various taxonomic groups of bacteria. SI 20 contains the PSY sequence from the psychrophilic bacteria Gelidibacter algens for thermo-psychro PSY analysis. SI 21-22 contain sequences from cyanobacteria, SI 23 contains sequence from protozoa, and SI 24-28 contain sequences of plant origin. The percentage of similarity and identity of the sequence with other phytoene synthase sequences was calculated using Water Pairwise Sequence Alignment (EM-BOSS) (Rice et al. 2000) (http://www.ebi.ac.uk/Tools/psa/ emboss_water/) following the Smith-Waterman algorithm (Smith and Waterman 1981).

Phylogenetic analysis of phytoene synthase
Multiple sequence alignment (MSA) of RP PSY with all the sequences given in Table 1 was done using MUSCLE (Edgar 2004) in MEGA7 . Using this MSA, the phylogenetic tree was also constructed in MEGA 7 ) by the maximum likelihood method (Kishino and Hasegawa 2001) based on the Jones, Taylor, and Thornton (JTT) model (Jones et al. 1992) with a gamma distribution (Nei and Zhang 2006). The scale of the branch lengths of the tree corresponds to the number of substitutions per site.

Primary structure analysis and physicochemical characterization
For primary structure analysis, the ExPASy ProtParam tool (Gasteiger et al. 2005) was used. The molecular weight, instability index, pI value (isoelectric point), aliphatic index, and grand average of hydropathicity (GRAVY) of the phytoene synthase of Meiothermus taiwanensis strain RP and all the proteins of Table 1 were calculated and compared. Proteins with low pI values were considered as acidic, and those with high pI values were considered as basic (Moldoveanu et al. 2017). The instability index predicted the stability of the protein in a test tube and values above 40 indicated that the protein might be unstable (Guruprasad et al. 1990). The aliphatic index was the relative volume occupancy of the protein by side chains, which are aliphatic and was considered to be greater in the case of thermophilic proteins (Ikai 1980). Positive GRAVY values indicated the hydrophobic nature of the protein (Kyte and Doolittle 1982). The percentage amino-acid composition of the phytoene synthase sequences of Meiothermus taiwanensis, M. ruber, Cyanobacterium aponinum, Euglena gracilis, Gelidibacter algens and Arabidopsis thaliana was determined and compared.
Analysis of the secondary structure of the proteins PSIPRED v3.3 (McGuffin et al. 2000) was used to determine the secondary structure of the phytoene synthase of M. taiwanensis strain RP. The percentages of the different secondary structures found in the phytoene synthase of RP and all the sequences from Table 1 were determined with the help of SOPMA (Geourjon and Deléage 1995) and compared with RP.

Active site analysis of the phytoene synthases
The functional analysis of the phytoene synthase protein from Meiothermus taiwanensis strain RP was performed and compared with other phytoene synthases. For this study, the substrate binding pocket, substrate Mg 2+ binding site, active lid-residues, catalytic residues, and aspartate-rich regions of the phytoene synthases from Meiothermus and compared. We detected significant variations and similarities between the active sites of this protein from different organisms.
Comparative analysis of three-dimensional structures of phytoene synthases The three-dimensional structure of the phytoene synthase sequences of Meiothermus taiwanensis strain RP, M. ruber, Thermus thermophilus, Brassica napus, and Gelidibacter algens was elucidated by homology modeling in Robetta server (Kim et al. 2004) using multiple sequence alignment of related models. The PDB files of the models were visualized using the PyMOL Molecular Graphics System, version 2.0 Schrödinger, LLC. The models were energy minimized using the Chiron server (Ramachandran et al. 2011). The servers which were used for validation and model quality interpretation were QMEAN server (Benkert et al. 2009), ProQ (Wallner and Elofsson 2003), PROCHECK (Laskowski et al. 1993), PROVE (Pontius et al. 1996), ERRAT (Colovos and Yeates 1993), and Verify3D (Bowie et al. 1991). The structures were compared with each other by superposition studies using the align method of alignment in PyMOL. The PDB files of the predicted structures were submitted to Protein Model Databank (Castrignanò et al. 2006).

Results and discussion
Retrieval of phytoene synthase sequences and phylogenetic analysis The percentage similarity and identity of Meiothermus taiwanensis strain RP PSY with other phytoene sequences is shown in Table 1. Even with its closest relative M. ruber, its percentage similarity is not more than 97.5% percent. It shows the lowest intra-genus similarity with M. silvanus (83.9 %). Meiothermus belongs to the Deinococcus-Thermus phylum, comprising mostly of microbes living under thermophilic conditions (Griffiths and Gupta 2007). Strain RP shows minimum intra-phylum similarity with Deinococcus radiodurans (60. 7 %). The similarity of strain RP PSY varies between 55.7 and 43.5% with other mesophilic or thermotolerant bacteria and 55.4 to 51% with phytoene synthase sequences from cyanobacteria, protozoa, and plants given in the table. The phytoene synthase sequence from Gelidibacter algens (Bowman et al. 1997) is of psychrophilic origin in contrast to the thermophilic RP PSY. Interestingly, this sequence has the least similarity with RP, with a percentage of 36.7%. The phylogenetic analysis of the sequence of RP is shown in Fig. 1. Here the phytoene synthase sequence of strain RP forms a phylogenetic group with other members of the Deinococcus-Thermus family, initially with Meiothermus sp. and then with Thermus sp. and Deinococcus sp. The previously reported phylogenetic trees constructed from analysis of phytoene synthase genes (Phadwal 2005;Klassen 2010) also showed phytoene synthase genes of Deinococcus-Thermus phylum forming a cluster. The PSY genes from eukaryotes, actinobacteria, proteobacteria, cyanobacteria, and bacteroidetes formed separate clusters in previous studies (Phadwal 2005;Klassen 2010). The phytoene synthase amino-acid sequences used in this study also show similar topologies.

Primary structure analysis and physicochemical characterization
The number of amino acids, pI value (isoelectric point), aliphatic index, and grand average of hydropathicity (GRAVY) of all the PSY proteins from Table 1 are given in Additional file 1: Table S1. The molecular weight and instability index of the proteins are represented in Fig. 2.

Number of amino acids and molecular weight
The amino acid sequence of M. taiwanensis strain RP is 275 residues long. All the sequences of Meiothermus sp. are between 275 and 280 amino acid (a.a.) residues long. The molecular weight of strain RP is 30.6 kDa, and the Meiothermus sequences vary between 30.5 and 31.5 kDa. The phytoene synthases from Meiothermus sp. are some of the smallest PSY proteins among the tested sequences used in this study. From the Brenda database (https://www.brenda-enzymes.org/enzyme.php ?ecno=2.5.1.32&onlyTable=Sequence), it can be observed that phytoene synthases from bacteria have a wide range of size and sequence length, but most of them are found towards the lower range of the molecular weight of phytoene synthases (< 36 kDa), except cyanobacteria. Phytoene synthase of the lowest sequence length having all the necessary conserved domains has been derived from previously reported whole genome sequence of the bacteria Cronobacter sakazakii 696 (240 amino acid residues and 26.5 kDa molecular weight) (https://www.brenda-enzymes.org/ sequences.php?ID=4201498&fasta=1) (Joseph et al. 2012). The PSY sequences from plant origin have shown the highest molecular weights in this study. The results are similar to previous reports of other phytoene synthases of plant origin such as Oryza sativa subsp. japonica (444 amino acid residues and 49.3 kDa molecular weight) (Welsch et al. 2008), Triticum aestivum (425 amino acid residues and 47.5 kDa molecular weight) (Crawford et al. 2011) and Zea mays (426 amino acid residues and 47.3 kDa molecular weight) (Li et al. 2008). Phytoene synthases have directly evolved from bacteria to plants through various bacterial taxonomic groups and algae (Sandmann 2002  inter-domain similarity concerning pI values which is similar to previous reports of plant origin phytoene synthases showing a wide range of pI values (Iwata-Reuyl et al. 2003;Lao et al. 2011;Han et al. 2015).

Aliphatic index
The aliphatic index of all the Meiothermus sp. sequences is greater than 90. The aliphatic index of M. taiwanensis is 96 and that of M. cerbereus is 98.07. The high value of the aliphatic index shows the thermophilic nature of the proteins from Meiothermus. The aliphatic index of the psychrophilic Gelidibacter sp. is 88.46, lower than the thermophilic proteins, though higher than many of the other sequences. All the phytoene synthases have aliphatic values of greater than 80, which is similar to some previously reported phytoene synthases of plant origin (Leng et al. 2017).

GRAVY value
The

Amino acid composition
The comparative analysis of the percentage amino-acid composition of the phytoene synthase sequences of Meiothermus taiwanensis, M. ruber, Cyanobacterium aponinum, Euglena gracilis, Gelidibacter algens, and Arabidopsis thaliana is given in Fig. 3. Meiothermus taiwanensis strain RP and M. ruber show an abundance of alanine, arginine, glutamic acid, tryptophan, glycine, and leucine residues with percentages higher than all the other phytoene synthase sequences including Gelidibacter. Cysteine and histidine are present in low numbers in all the sequences. The psychrophilic Gelidibacter sp. sequence has higher percentages of asparagine, aspartic acid, lysine, and phenylalanine as compared with other sequences and higher percentages of histidine,

Analysis of the secondary structure of the proteins
The secondary structure of M. taiwanensis strain RP PSY and its comparative analysis with secondary structures of the other PSY sequences are shown in Fig. 4. The major secondary structure present in the tested phytoene synthases is alpha-helix and no β-strands are present. Some of the previously reported protozoan and algal phytoene synthases do not have β-strands (Agarwal et al. 2015;Li et al. 2015), while some have a very short single stretch of β-strand (Lao et al. 2011;Agarwal et al. 2015). Extended strands are present in the phytoene synthases used in our study, which are not part of βstrands. These extended strands contain irregular hydrogen bonding and might form loops (Eswar et al. 2003). β-turns are present at the ends. Random coils are present in the secondary structure of all the sequences, though the proportions vary. Previously reported phytoene synthases also contained extended strands, βturns, and random coils (Lao et al. 2011;Agarwal et al. 2015;Li et al. 2015). Random coils are non-native protein states which have no structure except some inherent local interactions (Smith et al. 1996 (Flowerika et al. 2016). However, there were some differences in the amino acid residues present. The only exception was found in the case of Chlorella PSY which contained three aspartate rich DXXXD regions instead of two (Li et al. 2015). All the active site residues of Meiothermus taiwanensis are 100% similar in Meiothermus ruber. In M. silvanus, an alanine residue replaces a glycine residue in RP PSY (227 RP) which is a similar amino acid. This residue is involved in substrate binding and catalysis. In Thermus thermophilus, a total of 5 active site residues are different. The methionine (153), glycine (227), aspartic acid (46), serine (256), and isoleucine (132) residues in RP PSY are replaced by leucine, alanine, glutamate, histidine, and methionine residues respectively in T. thermophiles PSY. A glutamate residue replaces the aspartic acid (46)

Comparative analysis of the three-dimensional structures of phytoene synthases
For homology modeling of the three-dimensional (3-D) structures of phytoene synthases, PDB files given in Additional file 1: Table S2 containing crystallographic data of 3-D structures of proteins from the Protein Data Bank (Parasuraman 2012) were used for multiple sequence alignment. The three-dimensional structure of Meiothermus taiwanensis strain RP is given in Fig. 5 and that of Gelidibacter algens, Thermus thermophiles, Meiothermus ruber, and Brassica napus are given in Additional file 1: Fig. S1. The parameters used for quality analysis of the predicted 3-D structures of the phytoene synthases were Z score analysis from QMEAN4, error analysis and confidence of the models from Robetta, Ramachandran plot analysis from PROCHECK, QMEANDisCo Global values, LGscore, MaxSub values, Verify3D, and Errat values; and the Z value mean  (45)

Meiothermus silvanus
Ala (17) (168) His (15), Ser (16), Ala (17), Thr (18) , Phe (19), Arg (257), Ala (258), Ser (259), Leu (260) Phe (19), Tyr (38) S2, S3, S4, S5, S6, S7, S8, S9, S10, S11, S12, S13, S14, S 15, and S16, and Table S3). The 3-D structures of the phytoene synthases of Meiothermus taiwanensis strain RP, M. ruber, Thermus thermophilus, Brassica napus, and Gelidibacter algens were submitted to the Protein Model Databank with the accession numbers PM0082187, PM0082192, PM0082193, PM0082188, and PM0082190 respectively. The three-dimensional structures of the phytoene synthases from Meiothermus ruber, Thermus thermophilus, Gelidibacter algens, and Brassica napus were superimposed on Meiothermus taiwanensis strain RP PSY with RMSD (Root Mean Square Deviation) values of 0.596, 0.859, 1.562, and 1.026 respectively (Fig. 6). The phytoene synthase of the thermophiles Meiothermus ruber and Thermus thermophilus are structurally most similar to RP. The superposition image with Meiothermus ruber (Fig. 6a) is the most perfect. In the case of Thermus thermophilus (Fig. 6b), a part of the structure was not superposed as the sequence is greater in length than RP. However, the nonsuperimposed part does not have any active site residues. Huge areas of the phytoene synthase of Brassica napus (Fig. 6e) could not be superimposed; those areas did not contain any active site residues. The sequence length and molecular weight of B. napus phytoene synthase are much higher than that of RP. However, the superimposed areas are almost similar in structure. The psychrophilic phytoene synthase of Gelidibacter algens (Fig. 6c,  d) is most dissimilar to RP structurally. In Fig. 6d, the area covered by the active site residues is marked with pink spots. Though there are a lot of structural differences between G. algens and M. taiwanensis strain RP, most of them occur outside the active site region. Overall, despite having differences in structures, the enzyme phytoene synthase tries to retain its active site region structure, irrespective of its taxonomic affiliation or the growth temperature range of its source organism. Superposition studies of the three-dimensional structures of Plasmodium falciparum and Thermotoga maritima PSY (Agarwal et al. 2015) showed that amino acid differences affected 3-D structures but did not affect active sites. The regions with differences in structure are regions outside the catalytic region, and they may play a role in maintaining the properly folded state under various internal and external environmental conditions, such as differences in growth temperature of the organisms, or localization within the cell.

Conclusion
Meiothermus taiwanensis RP PSY was found to be one of the smallest PSY among the tested sequences with low instability index and high aliphatic index, which indicate its stability and thermophilicity. In some cases, intragenus patterns were found, for example, in sequence size and molecular weight, active sites, and structural aspects, while in some cases like pI values and GRAVY values, no patterns were found. There is a marked difference in the molecular weight and instability index of RP PSY and plant-origin PSYs, which lie at two extremities of the graph with the forming having one of the smallest values and the latter having the highest value. There is a significant difference in the percentages of some amino acid residues between the psychrophilic G. algens PSY and the thermophilic M. taiwanensis. Amino acids like alanine, arginine, glutamine, and glycine may be required in higher numbers for the stability of phytoene synthases at high temperatures, whereas, amino acids like asparagine, aspartic acid, histidine, lysine, and phenyl alanine may be required for maintenance of structural flexibility at low temperatures. Among the two temperature groups (thermophilic and psychrophilic) of PSY, the differences in some of the properties like amino acid percentage composition, aliphatic index, and 3-D structures may be clearly attributed to the differences of their growth temperatures, while other properties like sequence dissimilarity and differences in active site composition can also be due to factors other than their temperature differences. Interestingly, the differences among structures at the secondary structure level are negligible, as all of them contain high amounts of alphahelix. Most of the PSY active site residues are conserved across all the sequences with slightly more differences between G. algens and RP PSY. In the folded state, the active site regions of the PSYs are almost similar, though there are dissimilarities outside the region. Overall, this study provides structural and functional information which can be used in purification, genetic engineering, and expression of RP PSY with improved stability under different temperature conditions and in different organisms for carotenoid production without compromising the activity of the enzyme.
Additional file 1: Fig. S1 The three-dimensional structure of the phytoene synthase of a) Gelidibacter algens, b) Thermus thermophilus, c) Meiothermus ruber and d) Brassica napus. The helix chains are indicated by red color and the coils are indicated by green color. Fig. S2 The error estimates of the three dimensional model of Meiothermus taiwanensis strain RP phytoene synthase. Fig. S3 Z score value (<1) of the predicted 3D model of the phytoene synthase of Meiothermus taiwanensis strain RP shown as a red star. (If Z score value is >5, the quality analysis fails, if the value is <1 the quality of the structure is good.) Graph is obtained from Qmean server. Fig. S4 Plot statistics of the Ramachandran plot of the predicted 3D model of the phytoene synthase of Meiothermus taiwanensis strain RP as obtained from PROCHECK, based on an analysis of 118 structures of resolution of at least 2.0 Angstroms (residues in most favoured regions [A,B,L]: 94.2%, residues in additional allowed regions: [a, b, l, p] 5.4%, residues in generously allowed regions [~a,~b,~l,~p]: 0.4% and residues in disallowed regions: 0.0%). Glycine residues are shown as triangles. A good quality model has more than 90 % residues in favoured regions. Fig. S5 The error estimates of the three dimensional model of Meiothermus ruber phytoene synthase. Fig. S6 Z score value (<1) of the predicted 3D model of the phytoene synthase of Meiothermus ruber shown as a red star. (If Z score value is >5, the quality analysis fails, if the value is <1 the quality of the structure is good.) Graph is obtained from Qmean server. Fig. S7  . Glycine residues are shown as triangles. A good quality model has more than 90 % residues in favoured regions. Fig. S11 The error estimates of the three dimensional model of Gelidibacter algens phytoene synthase. Fig. S12 Z score value (<1) of the predicted 3D model of the phytoene synthase of Gelidibacter algens shown as a red star. (If Z score value is >5, the quality analysis fails, if the value is <1 the quality of the structure is good.) Graph is obtained from Qmean server. Fig. S13 Plot statistics of the Ramachandran plot of the predicted 3D model of the phytoene synthase of Gelidibacter algens as obtained from PROCHECK ,based on an analysis of 118 structures of resolution of at least 2.0 Angstroms (residues in most favoured regions [A,B,L]: 93.9 %, residues in additional allowed regions [a, b, l, p]: 5.7 %, residues in generously allowed regions [~a,~b,~l,~p]: 0.0% and residues in disallowed regions: 0.4 %). Glycine residues are shown as triangles. A good quality model has more than 90 % residues in favoured regions. Fig. S14 The error estimates of the three dimensional model of Brassica napus phytoene synthase. . Glycine residues are shown as triangles. A good quality model has more than 90 % residues in favoured regions. Table S1 The molecular weight, instability index, pI value, aliphatic index, and grand average of hydropathicity (GRAVY) of the all the phytoene synthase sequences. Table S2 Details of PDB files containing crystallographic data of 3-D structures of proteins from the Protein Data Bank used for multiple sequence alignment during 3-D structure modelling. Table S3 The evaluation of the quality of the three dimensional models of the phytoene synthases from different organisms. The quality of the models is considered as worst when the value in the first two columns (confidence of the models from Robetta and QMEAN-DisCo Global values) is 0.0 and becomes better with increase in value up to 1.0. The LGscore of the models is obtained from ProQ (>1.5 fairly good model, >2.5 very good model, >4 extremely good model). The fourth column contains the MaxSub values of the models obtained from ProQ (>0.1: fairly good model, >0.5: very good model, >0.8: extremely good model). The fifth column contains the Verify3D values of the models, where the percentage of amino acids scoring >=0.2 in their 3D/1D profile is determined. The model is valid if this percentage is >=80 %. The Errat quality factor is considered as good when the values are >=95 %. The Z value mean is obtained from PROVE (>5: the quality analysis fails, <1: the quality of the structure is good). (DOCX 1865 kb)

Research involving human participants and/or animals N/A
Authors' contributions TM and SKM made substantial contributions to design of research, experiments, acquisition of data and analysis, interpretation of data and drafting the manuscript. The authors read and approved the final manuscript. Ethics approval and consent to participate All prevailing local, national and international regulations and conventions, and normal ethical scientific practices have been respected.