Analysis of structural requirements for thermo-adaptation from orthologs in microbial genomes

Abstract

A comprehensive survey was carried out to identify orthologs of proteins from 526 bacterial and archaeal genomes, with the aim of investigating the mechanism of thermal adaptation of protein sequences. A large number of orthologs were distributed only in thermophiles/hyperthermophiles (HT-only group) and mesophiles (M-only group). A significant relationship between amino acid composition and optimal growth temperature (OGT) was observed. There were significantly higher proportions of charged, basic and acidic amino acids in hyperthermophilic and thermophilic genomes than in mesophilic and psychrophilic genomes. The orthologs distributed in all the four temperature ranges (Top-90 group) were also investigated, and a similar correlation between amino acid composition and OGT was found. The composition of the cluster of orthologous groups of proteins (COG) of the above three groups was analyzed; the composition of ‘information storage and processing’ in the HT-only group and Top-90 groups was much higher than that of M-only group.

Introduction

The thermal adaptation of hyperthermophiles and their protein thermo-stability are attractive and complex topics that have drawn considerable attention (Burra et al. 2010; Dehouck et al. 2008; De Vendittis et al. 2008; Dutta and Chaudhuri 2010; Farias and Bonato 2003; Karlin and Altschul 1990; Zeldovich et al. 2007). More than 100 hyperthermophilic or thermophilic bacterial and archaeal genomes have been sequenced and stored in public databases, thereby providing an unprecedented opportunity for studying the genetics, biochemistry, and evolution of these species, as well as for exploring the mechanisms of thermal adaptation. Many studies have been performed based on DNA sequence, protein sequence, certain protein families, and protein structure to investigate the mechanism of protein thermo-stability (Bae and Phillips 2004; Basak et al. 2007; Berezovsky and Shakhnovich 2005; Cambillau and Claverie 2000; De Vendittis et al. 2008). G + C content is not correlated significantly with optimal growth temperature (OGT). However, the increase of A + G content in coding genes can stabilize their thermo-stability, owing to the stacking effect of purines (Lao and Forsdyke 2000; Zeldovich et al. 2007). The thermo-stability of proteins is determined by a fine balance between many contributing factors, such as increment of hydrogen bonds, ion pairs, disulfide bridges or hydrophobic and aromatic interactions, changes in surface charge distribution, helix dipole stabilization, packing and reduction in solvent-accessible hydrophobic surface, contribution of specific chaperones, more compactness native conformation, variations of secondary structures, and so on (Basak and Ghosh 2005; De Vendittis and Bocchini 1996; Di Giulio 2000; Dong et al. 2008; Robb and Clark 1999; Tekaia and Yeramian 2006; Zeldovich et al. 2007). High-throughput comparative analysis of structures and complete genomes of several hyperthermophilic archaea has revealed that these organisms develop diverse strategies of thermophilic adaptation using two fundamental physical mechanisms (Berezovsky and Shakhnovich 2005). One is a ‘structure-based’ mechanism, i.e., some hyperthermostable proteins are significantly more compact than their mesophilic homologs, whereas no particular interaction type appears to cause stabilization (Bae and Phillips 2004; Berezovsky and Shakhnovich 2005; Dutta and Chaudhuri 2010). The other mechanism is a ‘sequence-based’ mechanism, i.e., hyperthermostable proteins do not show distinct structural differences from mesophilic homologs, whereas some apparently strong interactions, such as ionic interactions or additional salt bridges, are responsible for the high thermal stability of thermostable proteins (Berezovsky and Shakhnovich 2005; Makarova et al. 2003; Vetriani et al. 1998). Structure-stabilized proteins originate mostly from archaea, whereas sequence-stabilized proteins are mostly from bacteria (Berezovsky and Shakhnovich 2005).

Most previous studies have focused on certain protein families in different bacterial and archaeal genomes, ranging from psychrophiles to mesophiles to thermophiles/hyperthermophiles (Haney et al 1999; Georlette et al 2003; Bae and Phillips 2004). Although these studies show that the amino acid composition of many protein families is correlated with OGT, the proteomics characteristic of thermal adaptation remains unclear. The benefit of studying whole genomic sequences is that more fundamental properties of thermal adaptation can be obtained. Whole protein sequences of completely sequenced bacterial and archaeal genomes were compared instead of focusing on a single protein family. In total, more than 48,000 homologs from 526 bacterial and archaeal genomes were identified, and comprehensive comparison was carried out among the homologs distributed in four temperature range environments, i.e., hyperthermophilic (H, the organism grows above 75°C), thermophilic (T, the organism grows within the range 46°C to 75°C), mesophilic (M, the organism grows within the range 11°C to 45°C) and psychrophilic (P, the organism grows below 10°C). Analysis of these orthologs revealed a significant relationship between amino acid composition and thermal adaptation. The cluster of orthologous groups of proteins (COG) categories varied in the different temperature ranges (Tatusov et al. 2001). The current results strongly support the presence of a correlation between amino acid composition and thermostability, not only in a certain protein family, but also at the whole protein sequence level in prokaryotic genomes.

Materials and methods

Whole genomics sequences of 844 bacteria and archaea collected before the end of 2009 were downloaded from the NCBI Reference Sequence (RefSeq) project (ftp://ftp.ncbi.nih.gov/genomes/Bacteria/). Among these, 526 genomes were retained by excluding the substrains in the same species. The classification, general features, and optimal growth temperatures of these species are listed in Supplemental Table 1. These genomes were classified into four temperature ranges according to the optimal growth temperature. Based on the information from the NCBI Entrez Genome Project (http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi), 30, 51, 431, and 14 genomes were classified into hyperthermophiles (H, >75°C), thermophiles (T, 46°C to 75°C), mesophiles (M, 11°C to 45°C), and psychrophiles (P, <10°C).

Reciprocal-BLAST (E value <10⁻¹⁰) was performed to obtain the orthologs of 526 bacterial and archaeal genomes (Altschul et al. 1990, 1994; Karlin and Altschul 1990). The distributions of each ortholog in the four temperature ranges were calculated. Three groups of datasets (HT-only group, M-only group, and Top-90 group) were picked out for further analysis. Orthologs containing less than ten protein sequences were excluded from the analysis. The HT-only group was defined as orthologs distributed either in thermophiles/hyperthermophiles that are not found in mesophiles and psychrophiles. The M-only group stands for orthologs distributed only in mesophiles and not found in thermophiles/hyperthermophiles and psychrophiles. The Top-90 group indicates orthologs distributed in at least 90% of the microbial genomes in the four temperature ranges. The program used to classify orthologs in the four temperature ranges was an in-house Perl script. The statistics of protein properties in orthologs were calculated with the PEPSTATS program in the EMBOSS package (Harrison 2000).

Results and discussion

Orthologs identified from 526 bacteria

General statistical results of the identified orthologs in the 526 bacterial and archaeal genomes are shown in Table 1. A total of 48,313 orthologs were identified, 18,042 of which contain more than ten proteins in their orthologous groups. Of these, 95 orthologs were found exclusively in the HT-only group, 1,659 in the M-only group, and 115 in the Top-90 group. The HT-only, M-only, and Top-90 group orthologs represent the proteins unique in thermophiles/hyperthermophiles, mesophiles and in species that live in the four temperature ranges, respectively.

Table 1 General statistics of orthologs in HT-only, M-only, and Top-90 groups

Functional analysis of the HT-only group and COG composition of the three ortholog groups

The functions of the 95 orthologs in the HT-only group are listed in Table 2. According to the COG functional category (http://www.ncbi.nlm.nih.gov/COG/old/palox.cgi?fun=all), 19, 11, 8, and 57 orthologs fall under ‘information storage and processing’, ‘cellular processes and signalling’, ‘metabolism’, and ‘poorly characterized’. Sixty percent of orthologs in the HT-only group belong to the ‘poorly characterized’ category, which indicates that most of the HT-only orthologs have poor annotation information. The second largest category of HT-only orthologs is a group of 19 proteins classified under ‘information storage and processing’, of which 6 are ribosomal proteins and 7 are involved in transcriptional regulation. Although there are no particular functions ascribed to HT-only orthologs, their protein sequences are vastly different from those of mesophiles and psychrophiles. The HT-only orthologs play important roles in the thermal adaptation of thermophilic/hyperthermophilic species, which should be further investigated.

Table 2 Functions and cluster of orthologous groups of proteins (COG) categories of HT-only orthologs

The COG functional categories in the three groups were compared. Except for the orthologs with no COG information or categorized under ‘poorly characterized’, the COG categories of orthologs in the HT-only group, M-only group and Top-90 groups were collected. The composition in each group is shown in Fig. 1. Bars in the first cluster represent the COG composition of the HT-only group, and the second and third clusters represent the composition of the M-only group and Top-90 group, respectively. Nearly half of the orthologs in the HT-only group and more than half of those in the Top-90 group were categorized under ‘information storage and processing’, whereas only 28% of those in the M-only group were classified as such. The proportion of orthologs classified under ‘metabolism’ was similar in the three groups, which accounted for about one-third of each ortholog group. The proportion falling under ‘cellular processes and signalling’ varies in the three groups. Orthologs in the M-only group have significantly higher proportion (35%), followed by the HT-only group (20%) and Top-90 group (16%). It has been hypothesized that in order to maintain the survival of organisms in high-temperature environments, thermophiles/hyperthermophiles evolved more functional proteins for ‘information storage and processing’.

Fig. 1

Composition of cluster of orthologous groups of proteins (COG) of orthologs in HT-only, M-only, and Top-90 groups. The first, second, and third cluster of bars represent COG composition in HT-only, M-only, and Top-90 group, respectively. COG functional categories are as indicated in the figure

Amino acid composition and protein property analysis

Amino acid composition in the three groups was calculated, and the corresponding comparison between the HT-only and M-only groups is listed in Table 3. Significant differences were found. More charged (D + E + H + K + R, 24.95% vs. 20.25%; P < 0.001), acidic (D + E, 9.67% vs. 7.48%; P < 0.001) and basic (H + K + R, 15.28% vs. 12.77%; P < 0.001) amino acid residues were found in the HT-only orthologs. Other research groups have also observed that thermostable proteins have an amino acid composition biased for enhancing electrostatic or hydrophobic interactions (De Farias and Bonato 2002; De Vendittis et al. 2008; Di Giulio 2000; Farias and Bonato 2003; Zeldovich et al. 2007). The M-only set has significantly higher proportions of tiny (A + C + G + S + T, 29.44% vs 24.51%; P < 0.001) and small (A + C + D + G + N + P + S + T + V, 48.87% vs 43.96%; P < 0.001) amino acid residues. Differences found in proportions of aliphatic, aromatic, non-polar, and polar amino acids are not significant. The average length of HT-only orthologs tends to be shorter than those of M-only orthologs, with an average length of 236 vs 274 amino acids, respectively (P < 0.05). The protein properties can be attributed to the maintenance of the thermo-stability of thermophilic/hyperthermophilic genomes in high-temperature environments.

Table 3 Comparison of protein properties between HT-only and M-only groups

Many protein properties in the Top-90 group of orthologs in the four temperature ranges were also calculated and compared (Fig. 2a–h). The protein properties investigated include the number of amino acid residues (Fig. 2a), average residue weight (Fig. 2b), isoelectric point (Fig. 2c), tiny amino acids (A + C + G + S + T, Fig. 2d), small amino acids (A + C + D + G + N + P + S + T + V, Fig. 2e), charged amino acids (D + E + H + K + R, Fig. 2f), basic amino acids (H + K + R, Fig. 2g), and acidic amino acids (D + E, Fig. 2h). In each histogram, the columns indicate average value and the error bars indicates standard deviation. Most of the above mentioned protein properties differ significantly among hyperthermophiles, thermophiles, and mesophiles, except isoelectric point. However, the differences between mesophiles and psychrophiles are not significant. In the protein properties in the HT-only group, there are significantly higher proportions of charged, basic, and acidic amino acids in hyperthermophilic and thermophilic genomes than in mesophilic and psychrophilic genomes, as well as lower proportions of tiny and small amino acids. The same trends are also observed between the HT-only group and M-only group of orthologs. The average protein sequence length of the orthologs also differs among the different temperature ranges, and the bacterial and archaeal genomes of organisms that live in higher-temperature environments tend to have the shortest average protein sequence lengths.

Fig. 2

Comparison of the protein properties among orthologs in the Top-90 group in hyperthermophiles, thermophiles, mesophiles, and psychrophiles. The calculated protein properties include number of a residues, b average residue weight, c isoelectric point, d tiny amino acids, e small amino acids, f charged amino acids, g basic amino acids, and h acidic amino acids. H Hyperthermophiles, T thermophiles, M mesophiles, P psychrophiles

Conclusion

The current research confirmed the previous findings of a significant relationship between amino acid composition and protein thermostability at the level of whole prokaryotes protein sequences. The proteins from thermophiles/hyperthermophiles tend to use more charged (D + E + H + K + R) amino acids to adapt to high temperature environments. The average length of protein sequences in thermophiles/hyperthermophiles is much shorter than those in mesophiles. The differences of amino acid composition between thermophiles/hyperthermophiles and mesophiles might be beneficial for the maintenance and stability of proteins in high temperatures. Furthermore, the differences of COG composition showed that more functional proteins participate in ‘information storage and processing’ in thermophiles/hyperthermophiles, thus enabling the organisms to adapt to the high-temperature environments.