Coevolution of aah: A dps-Like Gene with the Host Bacterium Revealed by Comparative Genomic Analysis
A protein named AAH was isolated from the bacterium Microbacterium arborescens SE14, a gut commensal of the lepidopteran larvae. It showed not only a high sequence similarity to Dps-like proteins (DNA-binding proteins from starved cell) but also reversible hydrolase activity. A comparative genomic analysis was performed to gain more insights into its evolution. The GC profile of the aah gene indicated that it was evolved from a low GC ancestor. Its stop codon usage was also different from the general pattern of Actinobacterial genomes. The phylogeny of dps-like proteins showed strong correlation with the phylogeny of host bacteria. A conserved genomic synteny was identified in some taxonomically related Actinobacteria, suggesting that the ancestor genes had incorporated into the genome before the divergence of Micrococcineae from other families. The aah gene had evolved new function but still retained the typical dodecameric structure.
A protein isolated in 1992 from Escherichia coli cells can bind DNA in vitro and was proposed to protect DNA from oxidative damage . Its name DNA-binding protein from starved cell (Dps) has been widely accepted. However, some proteins belonging to the same family, which was isolated earlier, do not bind DNA, that is, the 4D antigen of Treponema pallidum isolated in 1987 , which was also known as TpF1  and C1-5 . They were proposed to play an important structural role in the outer membrane of the bacteria. Immunoelectron microscopy confirmed its surface localization , but it precipitated differently from other membrane proteins in ultracentrifugation .
The DNA-binding activity of the E. coli Dps protein is due to a nonspecific interaction between the positively charged N-terminus and the negatively charged phosphate group of DNA backbone (Figure 1(a)). The protein isolated from Agrobacterium tumefaciens lacks such an N-terminal and hence does not bind DNA . The protein from Mycobacterium smegmatis can bind DNA, but duet to a positively charged C-terminus . It lost the capability as soon as the C-terminus was truncated.
All of the known Dps-like proteins are composed of small subunits of about 20 kD (Table 1). They usually form an aggregate of 12 subunits with a central hollow cavity. Ferrous ions bind to the dinuclear ferroxidase sites at dimer interfaces [8–11]. Their proposed functions include oxidative detoxification, resistance to toxic electrophile and acid stress and so forth [12–14]. Many Dps proteins contain a large amount of iron. The iron sequestration of the E. coli Dps has also been detected . A Dps-like protein, MrgA, was identified in Bacillus subtilis in a search for metallo-regulated genes . MrgA is also inducible by starvation and protects the cell against oxidative stress . Another Dps-like protein was isolated from Listeria innocua, and named ferritin-like protein (Flp), because it can sequester a large amount of iron in its central cavity . The expression of flp genes in L. innocua and Listeria monocytogenes is not only inducible by both stationary phase and low iron availability  but also by cold shock . Another Dps-like protein, the Neutrophil-activating protein (Nap), was isolated from Helicobacter pylori . Nap was thought to mediate cell adhesion . The positively charged surface makes it highly soluble. It selectively bind to acidic glycosphingolipids on human neutrophils, as well as sulphated oligosaccharides on mucin and Lewis x blood group antigen . The alkaline protein surface has also been proposed to mediate DNA binding . While, the Fine tangle pili (Ftp) from Haemophilus ducreyi has also been proposed to mediate cell adhesion .
We isolated a protein from M. arborescens SE14, and named it AAH for N-acyl amino acid hydrolase. N-acyl glutamines are plant defense elicitors in the insect saliva. AAH catalyzes the hydrolysis of the amide bond and, less efficient, the formation of the elicitor . AAH also belongs to the Dps-like protein family. Since Dps-like proteins are such a structurally similar, but functionally heterogeneous group, a comparative genomic analysis was carried out to further understand its evolution, with special emphasis on the GC-rich Gram-positive Actinobacteria.
The GC content of the aah gene was 67.7%, a little lower than the average value of the whole contig (69.3%) (Supplementary Table S1). The GC content of the third base of codons (GC3) of aah was very high (94.4%), compared to the surrounding genes. It is even higher than the GC3 of the whole Streptomyces coelicolor genome (92%), while the GC content of S. coelicolor genome is 72.1%. The GC content of the first base of codons (GC1) and of the second base of the codons (GC2) of aah were extremely low (Supplementary Figure S1). aah was the only gene on this contig using “TAA” as stop codon. When the usage of stop codons was plotted against the GC content of the genomes, a correlation emerged (Figure 2). TAA was used more often in low-GC genomes; TGA became more dominant in high-GC genomes. TAG remained at low level showed little correlation to genomic GC contents. There are exceptions, for example, the GC content of Tropheryma whipplei was low, but it did not use much TAA. Among the studied 14 Actinobacterial genomes, 5 dps-like genes used TAA, 3 used TAG, and 6 used TGA (Figure 3).
A conserved genomic synteny surrounding dps-like genes was found in Micrococcineae species (Figure 4(a)). Another type of synteny was found in three Corynebacterium species. One gene encoding the DNA-formamidopyrimidine glycosylase presented in both syntenies (Figure 4(b)). However, the glycosylase gene and the dps-like gene were back to back in corynebacteria, but face to face in the Micrococcineae species. There were more insertion and deletion in this region on the Corynebacterium genomes than on the Micrococcineae genomes. Similar genomic context was not detected in other Actinobacteria (Figure 4(c)). AAH showed highest sequence similarity to a protein on the genome of Leifsonia xyli (Figure 4). Its GC content was also similar to that of this plant pathogen (Table S1).
Coincident with the conserved genomic synteny, the three dps-like proteins of the Micrococcineae species clustered within the Actinomycetales clade (Figure 1(b)). Those proteins from the genera of the suborder Corynebacterineae and Streptomycineae formed two large clusters in the Actinomycetales group and intermingled with each other. Surprisingly, FtpA and two other proteins from γ-proteobacteria show high similarity to the Actinomycetales clade. This was reproducible by all three computer algorithms, namely, Neibour-joining of clustalX (NJ), NJ of Vector NTI, and maximum-likelihood of Phylip (ML). What remained to be explained is that a protein from Bifidobacterium, which belongs to another order, and the one from a Planctomycetes, Rhodopirellula baltica were also clustered to this clade.
The Dps proteins from enterobacteria grouped perfectly into one clade (Figure 1(b)). The positive N-termini were conserved in all of these proteins (data not shown). All of Dsp-like proteins from the α-proteobacteria had a short N-terminus (data not shown). The protein from a cyanobacterium, Nostoc sp. PCC 7120, was grouped into this α- and γ-proteobacteria super family. The Bacilli proteins showed higher sequence similarity than others to ferritin, which was used for rooting the tree. Proteins from spiral-shaped bacteria formed an isolated cluster. The protein from an anaerobe, Bacteroides fragilis, was also found in this group. Some bacteria contained two copies of dps-like genes, most of them showed very high sequence similarity except those two from M. smegmatis. On the other hand, dps homologs were not detected in other Mycobacterium species.
Dps-like proteins widely exist in eubacteria (Table 1). The genes were proposed to have evolved from a common ancestor with bacterioferritins and eukaryotic ferritins . Their reported properties can be correlated, to some extent, to their phylogeny (Figure 1(b)). The proteins from enterobacteriaceae all contain positively charged N-termini, while in α-proteobacteria, the termini are much shorter. Those from the spiral-shaped bacteria were surface proteins. In the Gram-positive bacilli, they mainly associated with iron sequestration. The AAH from M. arborescens (Actinomycetales) is, as yet, unique on the reversible hydrolase activity .
Genomic analysis revealed that the GC3 of the aah gene is significantly higher than surrounding and the S. coelicolor genome, though the average GC contain of the later two was much higher. The codon usage of GC-rich bacteria is always biased, resulting in a high GC3 . The Sueoka theory suggests that GC1 and GC2 are under strong selection pressure, while this control is much weaker on GC3 . It is, therefore, possible that the ancestor of the aah gene evolved from a low GC origin. It almost saturated the third position of its codons in GC content to match the host genomic environment.
Furthermore, aah is the only gene on the contig using TAA as stop codon. A previous study showed the high-GC thermophiles favoring TAG over TAA . We analyzed 30 genomes (Figure 2) and confirmed the preference of TAA in low-GC genomes, and dominance of TGA in high-GC genomes. The slightly deviation in some species is very easy to explain, for example, T. whipplei that use less TAA was probably due to its evolution from a GC-rich species to a low GC environment in eukaryotic cells . Other atypical species are thermophiles, halophiles, and so forth (Supplementary Table S1). They have undoubtedly encountered additional selection pressure. Among the 14 Actinobacterial dps-like genes studied, the stop codon TAA and TGA were equally used, which also indicates the acquisition was not an old event.
The ancestor gene might have incorporated into the Micrococcineae genome before the divergence of the families. The aah gene showed highest similarity to the proteins from L. xyli and T. whipplei (Figure 1). None of them uses TGA as stop codon (Figure 3). The genomic signature surrounding dps-like genes (Figure 4) seems to be a relic of the ancient incorporation event. Such conserved gene clustering has also been observed in some closely related plasmids . Analysis of the dps gene from B. fragilis and Porphyromonas gingivalis suggests that the dps-like genes are present prior to the divergence of anaerobic bacterium from aerobics . The ubiquity of dps-like genes in bacteria and the variability on functions supported the theory that the emergence of dps-like genes was an ancient event, and the observed sequences similarity is simply due to a common ancestry rather than common function .
Dps-like proteins could be grouped into four major branches. It is not a surprise to find the Gram-negative anaerobe B. fragilis in the Gram-positive branch, since its ancestor is the Gram-positive Flavobacterium/Cytophaga . Furthermore, each major branch contains some proteobacteria. If proteobacteria is the vector dispersing the dps gene, the Actinomycetales would have acquired the dps-like gene from γ-proteobacteria, while ε-proteobacteria donated the dps-like gene to spirochetes. It is also possible that the intermingling of Dsp-type proteins in the dendrograph was a consequence of convergent evolution, although our data do not favour this possibility.
An earlier research revealed 50 out of 254 bacteria have more than two copies of dps genes . We found that four species contained two copies of dps genes. The high similarity between the two copies suggests a recent duplication. While in M. smegmatis, the two copies were very different. They might have been acquired independently. On the contrary, in the other two Mycobacterium species, no dps-like gene was detected (Figure 3). Horizontal gene transfer is very limited in these bacteria , the dps-like gene might have never been acquired. After coevolution with their bacterial host over a long time, Dps proteins clearly had undergone subfunctionalization and neofunctionalization. Despite sharing a few highly conserved amino acid residues, even the ferroxidase centers vary considerably . The ferroxidase centers are formed by amino acids brought together from adjacent subunits, other amino acids could also meet there to form other types of catalytic centers.
4. Material and Methods
Genomic DNA from M. arborescens SE14 was isolated as described elsewhere . A cosmid library was constructed using the pWEB vector (Epicentre) according to the manufacturer’s instruction. 8 positive colonies have been selected from a 1,121 colony library by PCR screening with primers afpup2: ACAGCTCGCCGATGGTCACA and afprc3: CCGTCGGCGCCGGGTATTAC. DNA insert was sheared using Standard Nebulizer (Octurno). Fragments were polished by T4 DNA and Klenow polymerases. Fragments were ligated into a pUC18 vector, and end-sequencing was achieved using BigDye Terminator Ready Reaction Kit (Applied Biosystems). Sequence data were assembled by GAP4 software (http://staden.sourceforge.net) and deposited in GenBank (accession number AY993941). Putative protein-coding sequences were predicted by the softwares Lasergene (DNAStar) and VectorNTI (Invitrogen) and then analyzed by BLAST search at NCBI (http://www.ncbi.nlm.nih.gov). Web-based software tRNAscanSE v.1.1 was used to predict tRNA .
The GC content was calculated with VectorNTI using a window size of 500 bp and 40 bp for the entire sequence and the aah gene, respectively. GC1, GC2, and GC3 were calculated by FramePlot 3.0 beta* (http://watson.nih.go.jp/) with a window size of 10 codons and step size of 1 codon. The genomic GC content and stop codon usage were calculated at website TIGR-Comprehensive Microbial Resource (http://cmr.tigr.org/tigr-scripts/CMR/CmrHomePage.cgi). To compare the genomic context, 14 Actinobacteria genomes were retrieved from NCBI and aligned manually.
The protein sequences of AAH, Dps from E. coli, and MrgA from B. subtilis were employed to BLAST at NCBI separately. Homology analysis of retrieved sequences was performed with three methods independently: (i) NJ of Vector NTI. The Guide Tree is calculated by AlignX with distance values. (ii) NJ of ClustalX 1.8 . Bootstrap values were based on 1,000 replications. (iii) ML of Phylip3.67 (http://evolution.genetics.washington.edu/phylip.html). The final evolutionary tree was displayed in TreeView 1.6.6 (http://taxonomy.zoology.gla.ac.uk/rod/rod.html) and modified.
The authors thank Professor J. Piel from the University of Bonn and Professor R. Kolter from the Harvard Medical School for fruitful discussion, and Ms. Emily R. Wheeler for English editing.
Enclosed are the list of IDs of proteins used in Figure 1, FIGURE S1 that shows the GC profiles of the aah gene, and TABLE S1 that provides the information on the stop codon usage and genome GC content in some bacteria.
M. Almiron, A. J. Link, D. Furlong, and R. Kolter, “A novel DNA-binding protein with regulatory and protective roles in starved Escherichia coli,” Genes & Development, vol. 6, pp. 2646–2654, 1992.View at: Google Scholar
J. D. Radolf, L. A. Borenstein, and J. Y. Kim, “Role of disulfide bonds in the oligomeric structure and protease resistance of recombinant and native Treponema pallidum surface antigen 4D,” Journal of Bacteriology, vol. 169, no. 4, pp. 1365–1371, 1987.View at: Google Scholar
G. T. Noordhoek, P. W. M. Hermans, A. N. Paul, L. M. Schouls, J. J. van der Sluis, and J. D. A. van Embden, “Treponema pallidum subspecies pallidum (Nichols) and Treponema pallidum subspecies pertenue (CDC 2575) differ in at least one nucleotide: comparison of two homologous antigens,” Microbial Pathogenesis, vol. 6, no. 1, pp. 29–42, 1989.View at: Google Scholar
A. M. Walfield, E. S. Roche, M. C. Zounes et al., “Primary structure of an oligomeric antigen of Treponema pallidum,” Infection and Immunity, vol. 57, no. 2, pp. 633–635, 1989.View at: Google Scholar
S. J. Norris, “Polypeptides of Treponema pallidum: progress toward understanding their structural, functional, and immunologic roles. Treponema Pallidum Polypeptide Research Group,” Microbiological Reviews, vol. 57, no. 3, pp. 750–779, 1993.View at: Google Scholar
P. Ceci, A. Ilari, E. Falvo, and E. Chiancone, “The Dps protein of Agrobacterium tumefaciens does not bind to DNA but protects it toward oxidative cleavage: x-ray crystal structure, iron binding, and hydroxyl-radical scavenging properties,” The Journal of Biological Chemistry, vol. 278, no. 22, pp. 20319–20326, 2003.View at: Publisher Site | Google Scholar
S. Roy, R. Saraswathi, S. Gupta, K. Sekar, D. Chatterji, and M. Vijayan, “Role of N and C-terminal tails in DNA binding and assembly in Dps: structural studies of Mycobacterium smegmatis Dps deletion mutants,” Journal of Molecular Biology, vol. 370, no. 4, pp. 752–767, 2007.View at: Publisher Site | Google Scholar
G. Zhao, P. Ceci, A. Ilari et al., “Iron and hydrogen peroxide detoxification properties of DNA-binding protein from starved cells. A ferritin-like DNA-binding protein of Escherichia coli,” The Journal of Biological Chemistry, vol. 277, no. 31, pp. 27689–27696, 2002.View at: Publisher Site | Google Scholar
G. P. Ferguson, R. I. Creighton, Y. Nikolaev, and I. R. Booth, “Importance of RpoS and Dps in survival of exposure of both exponential- and stationary-phase Escherichia coli cells to the electrophile N- ethylmaleimide,” Journal of Bacteriology, vol. 180, no. 5, pp. 1030–1036, 1998.View at: Google Scholar
L. Chen, L. P. James, and J. D. Helmann, “Metalloregulation in Bacillus subtilis: isolation and characterization of two genes differentially repressed by metal ions,” Journal of Bacteriology, vol. 175, no. 17, pp. 5428–5437, 1993.View at: Google Scholar
L. Chen and J. D. Helmann, “Bacillus subtilis MrgA is a Dps(PexB) homologue: evidence for metalloregulation of an oxidative-stress gene,” Molecular Microbiology, vol. 18, no. 2, pp. 295–300, 1995.View at: Google Scholar
D. J. Evans Jr., D. G. Evans, T. Takemura et al., “Characterization of a Helicobacter pylori neutrophil-activating protein,” Infection and Immunity, vol. 63, no. 6, pp. 2213–2220, 1995.View at: Google Scholar
F. Namavar, M. Sparrius, E. C. I. Veerman, B. J. Appelmelk, and C. M. J. E. Vandenbroucke-Grauls, “Neutrophil-activating protein mediates adhesion of Helicobacter pylori to sulfated carbohydrates on high-molecular-weight salivary mucin,” Infection and Immunity, vol. 66, no. 2, pp. 444–447, 1998.View at: Google Scholar
P. Ceci, L. Mangiarotti, C. Rivetti, and E. Chiancone, “The neutrophil-activating Dps protein of Helicobacter pylori, HP-NAP, adopts a mechanism different from Escherichia coli Dps to bind and condense DNA,” Nucleic Acids Research, vol. 35, no. 7, pp. 2247–2256, 2007.View at: Publisher Site | Google Scholar
R. J. Brentjens, M. Ketterer, M. A. Apicella, and S. M. Spinola, “Fine tangled pili expressed by Haemophilus ducreyi are a novel class of pili,” Journal of Bacteriology, vol. 178, no. 3, pp. 808–816, 1996.View at: Google Scholar
A. T. Pulliainen, S. Haataja, S. Kähkönen, and J. Finne, “Molecular basis of H2O2 resistance mediated by streptococcal Dpr: demonstration of the functional involvement of the putative ferroxidase center by site-directed mutagenesis in Streptococcus suis,” The Journal of Biological Chemistry, vol. 278, no. 10, pp. 7996–8005, 2003.View at: Publisher Site | Google Scholar
M. J. Horsburgh, M. O. Clements, H. Crossley, E. Ingham, and S. J. Foster, “PerR controls oxidative stress resistance and iron storage proteins and is required for virulence in Staphylococcus aureus,” Infection and Immunity, vol. 69, no. 6, pp. 3744–3754, 2001.View at: Publisher Site | Google Scholar
S. Gupta, S. B. Pandit, N. Srinivasan, and D. Chatterji, “Proteomics analysis of carbon-starved Mycobacterium smegmatis: induction of Dps-like protein,” Protein Engineering, vol. 15, no. 6, pp. 503–511, 2002.View at: Google Scholar
M. G. Cuypers, E. P. Mitchell, C. V. Romão, and S. M. McSweeney, “The crystal structure of the Dps2 from Deinococcus radiodurans reveals an unusual pore profile with a non-specific metal binding site,” Journal of Molecular Biology, vol. 371, no. 3, pp. 787–799, 2007.View at: Publisher Site | Google Scholar
N. Sueoka, “Directional mutation pressure and neutral molecular evolution,” Proceedings of the National Academy of Sciences of the United States of America, vol. 85, no. 8, pp. 2653–2657, 1988.View at: Google Scholar
J. R. Lobry and D. Chessel, “Internal correspondence analysis of codon and amino-acid usage in thermophilic bacteria,” Journal of Applied Genetics, vol. 44, no. 2, pp. 235–261, 2003.View at: Google Scholar
S. Casjens, N. Palmer, R. van Vugt et al., “A bacterial genome in flux: the twelve linear and nine circular extrachromosomal DNAs in an infectious isolate of the Lyme disease spirochete Borrelia burgdorferi,” Molecular Microbiology, vol. 35, no. 3, pp. 490–516, 2000.View at: Publisher Site | Google Scholar
D. J. Evans Jr., D. G. Evans, H. C. Lampert, and H. Nakano, “Identification of four new prokaryotic bacterioferritins, from Helicobacter pylori, Anabaena variabilis, Bacillus subtilis and Treponema pallidum, by analysis of gene sequences,” Gene, vol. 153, no. 1, pp. 123–127, 1995.View at: Publisher Site | Google Scholar
W. G. Weisburg, Y. Oyaizu, H. Oyaizu, and C. R. Woese, “Natural relationship between bacteroides and flavobacteria,” Journal of Bacteriology, vol. 164, no. 1, pp. 230–236, 1985.View at: Google Scholar