Valdes-2008-Hydrometallurgy Acidithiobacillus ferrooxidans

Valdes-2008-Hydrometallurgy Acidithiobacillus ferrooxidans

(Parte 1 de 2)

Comparative genome analysis of Acidithiobacillus ferrooxidans, A. thiooxidans and A. caldus: Insights into their metabolism and ecophysiology

Jorge Valdés 1, Inti Pedroso, Raquel Quatrini, David S. Holmes ⁎ Center for Bioinformatics and Genome Biology, Life Science Foundation, MIFAB and Andrés Bello University, Santiago, Chile

Keywords: Acidithiobacillus Bioinformatics Comparative genomics Metabolic integration

Draft genome sequences of Acidthiobacillus thiooxidans and A. caldus have been annotated and compared to the previously annotated genome of A. ferrooxidans. This has allowed the prediction of metabolic and regulatory models for each species and has provided a unique opportunity to undertake comparative genomic studies of this group of related bioleaching bacteria. In this paper, the presence or absence of predicted genes for eleven metabolic processes, electron transfer pathways and other phenotypic characteristics are reported for the three acidithiobacilli: CO2 fixation, the TCA cycle, sulfur oxidation, sulfur reduction, iron oxidation, iron assimilation, quorum sensing via the acyl homoserine lactone mechanism, hydrogen oxidation, flagella formation, Che signaling (chemotaxis) and nitrogen fixation. Predicted transcriptional and metabolic interplay between pathways pinpoints possible coordinated responses to environmental signals such as energy source, oxygen and nutrient limitations. The predicted pathway for nitrogen fixation in A. ferrooxidans will be described as an example of such an integrated response. Several responses appear to be especially characteristic of autotrophic microorganisms and may have direct implications for metabolic processes of critical relevance to the understanding of how these microorganisms survive and proliferate in extreme environments, including industrial bioleaching operations. © 2008 Elsevier B.V. All rights reserved.

1. Introduction

Acidophilic prokaryotes involved in bioleaching process have been the subject of active research from the viewpoints of microbiology, biophysics, biochemistry and genetics. With the advent of genome sequencing and the continuously decreasing costs and time needed to sequence new genomes, novel opportunities and challenges have arisen for developing integrative metabolic and regulatory models that couple traditional experimental data sources with information obtained from computationally-derived comparative genome information. Advanced bioinformatic tools, developed in the last few years, have allowed relatively rapid ways to generate initial models of metabolic and regulatory features potentially encoded within a genome and to pinpoint key features that can be highlighted for subsequent experimental validation.

Chemolithoautotrophic microorganismsin general, and the extreme acidophilic ones in particular, are difficult to handle experimentally and can be recalcitrant to genetic manipulation. These hurdles can be compensated for, in part, by the enormous amount of information that can be extracted by careful genome analysis, allowing the experimental biologist to focus on key issues suggested by the bioinformatic predictions, thus saving considerable time and resources (Quatrini et al., 2007a).

The bioinformatic analysis of the genome sequence of the γproteobacterium A. ferrooxidans has led to the generation of several metabolic and regulatory models some of whichhave been experimentally validated in part, such as sulfur uptake and assimilation (Valdés et al., 2003), iron uptake and assimilation (Quatrini et al., 2005; Osorio et al., 2008-this issue), Fur regulatory gene circuits (Quatrini et al., 2007b), quorum sensing (Rivas et al., 2005; Farah et al., 2005; Rivas et al., 2007), biofilm formation (Barreto et al., 2005), small regulatory RNA gene prediction (Shmaryahu and Holmes, unpublished results) and carbon metabolism(Barretoetal.,2003;Appia-Aymeetal.,2006).Ananalysisof the genome of A. ferrooxidans has also supported and extended earlier models of iron and sulfur oxidation in this microorganism (Quatrini et al., 2006; Valdés et al., unpublished results).

Large-scale comparisons of genomes address basic questions, such as the number of functional genes, identification of species-specific genes, distribution of genes among functional families, gene density, preservation of gene order, mechanisms of genome reshuffling, the rate of sequence divergence, etc. (Hardison, 2003). In general, the choice of species that mark evolutionary distances for comparative genomics depends on the aim of the analysis.

In the present study, the genomes of three closely related gramnegative, chemoautotrophic bioleaching microorganisms, A. ferrooxidans, A. thiooxidans and A. caldus, were compared, anticipating that

Contents lists available at ScienceDirect Hydrometallurgy journal hom epage: w .else vier .com /locate/ hydr omet such a study would help to consolidate the prediction and identification of genes and conserved gene clusters involved in various predicted metabolic models and that the identification of conserved intergenic regions between the three species would facilitate the prediction of small regulatory RNA genes and transcriptional regulatory regions. It was also hoped that it would allow the identification of genes involved in iron oxidation specifict o A. ferrooxidans and might, in the long term, suggest the genetic basis of other potential differences between the three species such as the ability of A. caldus to grow at 45–5 °C compared to the mesophilic temperatures of the other two species (Hallberg and Lindstrom,1994). Most importantly, it was hoped that a comparative genomics study could identify novel genetic differences between the three species, suggest models of the evolution of their respective genomes and allows the prediction of mutual interactions between the three acidithiobacilli that might occur in their natural environment and in biomining operations (ecophysiology).

2. Methods

ThecompletegenomesequenceofAcidithiobacillusferrooxidansATCC 23270 was obtained from the Institute for Genomic Research database (TIGR, Valdés et al., unpublished results). Draft genome sequences of A. thiooxidansATCC 19377 andA. caldus ATCC 51756 were obtained from the Center for Bioinformatics and Genome Biology (CBGB), Santiago, Chile. Candidate protein coding genes were identified in genome sequences using Glimmer (Salzberg et al., 1998), Critica (Badger and Olsen,1999)andBlastX(Altschuletal.,1997).The5′and3′regionsofeach ORF were inspected to define initiation codons using homologies, position of ribosomal binding sites, and transcriptional terminators.

The following bioinformatic programs were used to further characterize candidate genes and their predicted protein products: BlastP and PsiBlast (Altschul et al., 1997), the suite of protein comparison and classification programs available in InterproScan (Mulder and Apweiler, 2007). Model metabolic pathways were reconstructed using PRIAM (Claudel-Renard et al., 2003)a nd compared to those obtained from BIOCYC (Caspi et al., 2008), KEGG (Kanehisa et al., 2008) and ERGO (Overbeek et al., 2003). Amino acid sequences derived for the genes identified in the genome sequence of A. ferrooxidans ATCC 23270 and draft genome sequences for A. caldus ATCC 51756 and A. thiooxidans ATCC 19377 were annotated, analyzed and compared to known metabolic models using perl scripts developed in our laboratory. The annotated genomes were displayed in the interactive format of Artemis (Rutherford et al., 2000).

The genomes were compared at the amino acid level using tblastx

(Altschul et al., 1997) using an e-value of 10−5 as cutoff to identify syntenic regions. Syntenic regions were displayed and analyzed using the Artemis comparison tool (ACT) (Carver et al., 2005)a nd Genomeviz (Ghai and Chakraborty, 2007).

3. Results and discussion 3.1. Comparative genomics pipeline

A bioinformatic pipeline was developed to undertake the genetic and metabolic comparison of the three acidothiobacilli (Fig. 1). The pipeline is composed of two initial bioinformatics steps that cooperate to generate the searchable Structural Query Language (SQL) server termed the “Acidthiobacillus database” that is at the core of the analysis. These steps are i) open reading frame (ORF) prediction using Glimmer and Critica where predicted ORFs are then compared to annotated genes deposited in NR-DB and MG-DB using InterproScan, PRIAM and Blast in order to assign potential function and i) BlastX of the genomes against the NR-DB to improve the annotation developed in (i). The SQL database was then mined and results visualized using four tools: Artemis, ACT, GenomeViz and KEGG maps.

3.2. Whole genome comparison

To identify genomic regions with significant similarity, predicted protein comparisons were performed between the A. ferrooxidans complete genome sequence and the two draft genomes sequences of A. caldus and A. thiooxidans. Fig. 2 shows a similarity map that facilitates the identification of regions in the A. ferrooxidans genome that have no counterparts in the other two genomes. This first step is required for a rapid identification of potential functional modules present in the model organism A. ferrooxidans and their respective locations in the genome sequence. Most of the exclusive regions of the A. ferrooxidans genome are characterized by abnormal GC content and GC skew signatures. Fig. 2 has four highlighted regions that correspond to examples of conserved/non-conserved genomic loci:

(A)pet-Iandrus operonsinvolved in ironoxidationinA.ferrooxidans.

These loci are absent in the other two acidithiobacilli. They are located close to, and on either side of, the origin of replication; a positionwhich provides duplicated copies early in the DNA replication cycle. It is not knownwhether this locationis important to augmentthe copy number of the iron oxidation genes to provide increased transcriptional activity; B) nitrogen fixation gene cluster in A. ferrooxidans absent in the other two acidithiobacilli; (C) large region present in A. ferrooxidans that is absent in the other two acidithiobacilli. This region is rich in predicted hypothetical genes and potential mobile sequences including predicted plasmid and phage elements suggesting that this is a region acquired in A. ferrooxidansafteritsdivergence fromthelastcommonancestorofthe three acidithiobacilli. However, this hypothesis requires substantiation by phylogenic analysis. It also remains to be investigated if the hypothetical genes in this region provide A. ferrooxidans with speciesspecific properties; (D) a region that is highly conserved between the three acidithiobacilli that is mainly composed of genes predicted to encode rRNAs, ribosomal proteins and other genes related to protein synthesis functions.

3.3. Function predictions

A comparative analysis of the genomes of A. ferrooxidans and the other acidithiobacilli revealed fundamental similarities and differences that reflect their specific metabolic capacities. Table 1 lists a number of predicted functions for each of the three acidithiobacilli. All

Fig. 1. Schematic representation of the bioinformatics pipeline used in this study. Abbreviations: ACT, Artemis comparison tool; NR-DB, Non-redundant database; MGDB, microbial genomes database. Additional information can be found in the Methods.

three have genes for the Calvin-Benson scheme for CO2 fixation and lack the gene for α-keto glutarate dehydrogenase suggesting that they have an incomplete TCA cycle, as has been found in all obligate chemoautotrophs to date. However, they differ in the genes predicted to be involved in sulfur oxidation. A. thiooxidans and A. caldus have a predicted suite of essential soxABXYZ genes with similarity to the well studied SOX (sulfur oxidases) system (Friedrich et al., 2001). In contrast, A. ferrooxidans lacks genes encoding the SOX system but has genes with similarity to those encoding SQR (sulfide quinone oxidoreductases), TQR (thiosulfate quinone oxidoreducatase) and TetH (tetrathionate hydrolase), suggesting that it has evolved a different mechanism(s) for sulfur oxidation.

As predict ed from experimental evidence, A. caldus and

A. thiooxidans lack the genes that permit A. ferrooxidans to oxidize ferrous iron. In A. ferrooxidans the genes coding for the iron oxidation system are organized in two transcriptional units, the petI (petC-1, petB-1, petA-1, sdrA-1 and cycA-1) and rus (cyc2, cyc1, hyp, coxB, coxA coxC, coxD and rus) operons (Holmes and Bonnefoy, 2006 and references therein) that were not found by similarity searches.

Regarding iron assimilation and homeostasis, A. ferrooxidans has more predicted outer membrane receptors (OMRs) for Fe(I) siderophores than either of the other two acidithiobacilli (Table 1). There are also differences in the number of TonB and ABC transporters involved in Fe(I) siderophore uptake between the three microorganisms as well as Fe(I) uptake systems (data not shown). These differences point to distinctive functional roles in iron management between the three acidithiobacilli and the future study of these distinctions will contribute to our understanding of the ecophysiology and general survival strategies in acidic andiron loaded environments.

Table 1 Presence (Yes) or absence (No = not detected) of predicted genes for several metabolic and phenotypic characteristics of A. ferrooxidans, A. caldus and A. thiooxidans based on bioinformatics genome comparisons

Microorganism A. ferrooxidans A. thiooxidans A. caldus

Predicted genes for: CO fixation Calvin-Benson Calvin-Benson Calvin-Benson TCA cycle Incomplete Incomplete Incomplete S oxidation SQR system SOX system SOX system S reduction Yes No No Fe(I) oxidation Yes No No No. of Fe(I) uptake OMRs 1 8 6 Quorum sensing by the AHL system Yes No No Hydrogen oxidation Yes NI Yes Flagella formation No Yes Yes Che signaling No Yes Yes Nitrogen fixation Yes No No

SQR = sulfide quinone oxidoreductase; SOX = sulfur oxidase; APS = adenosine-5′- phosphosulfate; PAPS = 3′ phosphoadenosine-5′-phosphosulfate; OMRs = outer membrane Fe(I) siderophore receptors; AHL = acyl homoserine lactone; NI = no information.

Fig. 2. Whole genome comparison of the A. ferrooxidans genome and the two draft genomes of A. caldus and A. thiooxidans. First circle, examples of conserved/non-conserved regions in the three acidithiobacilli. A–D, information discussed in the text; second outer circle (green), A. caldus blast hits mapped against the A. ferrooxidans genome; third circle (orange), A. thiooxidans blast hits mapped against the A. ferrooxidans genome; fourth and fifth circles, A. ferrooxidans protein coding genes; sixth circle, A. ferrooxidans GC content; seventh circle, A. ferrooxidans GC skew. The predicted origin of replication is located at “12 o'clock”.

For example, the significantly larger number of Fe(I) siderophore OMRs found in A. ferrooxidans might help to explain its greater sensitivity to Fe(I) rich environments.

Some strains of A. ferrooxidans exhibit chemotaxis to thiosulfate

(Chakraborty and Roy, 1992) and have single coiled flagellum and pili (DiSpirito et al., 1982; Gonzalez and Cotoras, 1987; Kelly and Wood, 2000) that can mediate the attachment to sulfur (Ohmura et al.,1996). However, the genome of the type strain of A. ferrooxidans lacks genes for both flagella formation and Che two-component signaling transduction. This makes it unlikely that it can carry out chemotaxis, at least by the well-studied pathways and suggests that the species A. ferrooxidans has a diverse phenotype. The other two acidithiobacilli contain genes for both properties. Aword of caution is needed to point out that all three sequenced genomes come from strains grown in laboratory conditions prior to being deposited in culture banks, meaning that original properties of the environmental founders could have been lost during passage in the laboratory. This is unlikely in the case of the Che genes because they are typically distributed throughout the genome, meaning that multiple excision events would be necessary for their elimination. The flagella genes, on the other hand, are often located in one large gene cluster (about 30 genes) and so could be lost by one excision event.

Only A. ferrooxidans has genes for homoserine lactone-based quorum sensing and these been experimentally demonstrated to occur through the classic LuxIR system (Rivas et al., 2005; Farah et al., 2005). A. ferrooxidans also has a potential quorum sensing system based on act gene cluster activity; alternatively this system may be involved in cell membrane formation and the two possibilities remain to be explored experimentally (Rivas et al., 2007).

A. ferrooxidans is the only one of the three to have predicted nitrogenase genes and so is the only one to be able to fix atmospheric nitrogen as experimentally demonstrated for A. ferrooxidans by Mackintosh (Mackintosh, 1978). This means that, in a consortium of the three bacteria and in the absence of fixed nitrogen such as ammonia, A. ferrooxidans would have to support the entire community's requirements for nitrogen. This prediction remains to be tested but could potentially have important consequences in bioleaching operations and natural communities where sources exogenous fixed nitrogen might be limiting.

3.4. Interplay between assimilatory pathways in acidithiobacilli using nitrogen fixation in a. ferrooxidans as an example

In this study we have identified several regulatory components that provide potential for crosstalk between assimilatory and anabolic pathways. Most of the assimilatory processes detected have homeostatically controlled systems to detect the bioavailability of a specific element. Nitrogen metabolism is mainly regulated by oxygen concentration at the transcriptional (by the repressor activity of NifL over NifA) and post-translational level (by the reversible modification of the nitrogenase enzyme by the Drag/DraT proteins) and ammonia availability in many bacteria (Dixon and Kahn, 2004).

Nitrogen metabolism has a critical signal transduction component, the P-I protein, that is encoded by the genes glnK and glnB in bacteria and archaea. These genes are predicted in the A. ferrooxidans genome. The active form of P-I is a trimer that can inhibit the activity of NtrB and represses activity of the ammonia transporter encoded by amtB (Dixon and Kahn, 2004). NtrB is a positive regulator that, via NifA, activates the operon used to fix nitrogen (N2-ase in Fig. 3). NifL is a negative regulator of the N2-ase operon that transduces the signal generated by O2 concentration. Hence, in the presence of P-I and/or oxygen, nitrogen cannot be fixed. However, in the presence of ATP and alpha-ketoglutarate P-I is inactivated (Ninfa and Jiang, 2005). Thus when energy is abundant and there is not enough reduced nitrogen in the form of ammonia and glutamine, information can be channeled from the CO2 fixation machinery via the TCA cycle and energy status via ATP synthesis to inactivate P-I and allow N2 fixation in the absence of O2. However, in leaner times P-I is not inactivated and so the energy that could have been used for N2 fixation can be used by other metabolic processes. A. ferrooxidans exhibits candidate genes for the abovementioned activities and can thus balance the fixation of N2 and importation of NH4 according to metabolic circumstances and oxygen concentration, whereas A. thiooxidans and A. caldus lack the N2-ase operon and must rely on A. ferrooxidans for fixed nitrogen, presumably supplied as NH4+.

4. Conclusions

Comparative genomics provides a rich source of information for improving gene annotations, assigning gene functions, identifying potential regulatory regions and for building metabolic models. It also provides insight into gene and genome evolution. Comparison of the closely related A. ferrooxidans, A. thiooxidans and A. caldus genomes has identified sets of genes that could be responsible for differences in iron and sulfur oxidation in these organisms. It also suggests that fixed nitrogen can only be supplied by A. ferrooxidans, suggesting that it is the major contributor of nitrogen in a consortium of the three bacteria such as might be found in a bioleaching heap. Surprisingly, the genome of the type strain of A. ferrooxidans does not encode flagella or the che signaling genes associated with chemotaxis whereas the other two acidithiobacilli have these capacities.


UNAB 34-06, DI-UNAB 15-06/I and a Microsoft Sponsored Research Award.

Fig. 3. Predicted metabolic and regulatory interplay between nitrogen and carbon metabolism in A. ferrooxidans. Proposed connections to sulfur, hydrogen and energy are not shown for simplicity.


Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389–3402.

J., Cárdenas, J.P., Esparza, M., Orellana, O., Jedlicki, E., Bonnefoy, V., Holmes, D.S., 2006. Microarray and bioinformatic analyses suggest models for carbon metabolism in the autotroph Acidithiobacillus ferrooxidans. Hydrometallurgy 83, 273–280.

(Parte 1 de 2)