IST1 insertional inactivation of the resB gene: implications for phenotypic switching in Thiobacillus ferrooxidans

Abstract

Thiobacillus ferrooxidans ATCC 19859 undergoes rapid phenotypic switching between a wild-type state characterized by the ability to oxidize ferrous iron (FeII) and reduced sulfur compounds and a mutant state where it has lost the capacity to oxidize FeII but retains the ability to oxidize sulfur. The mutant has also gained the capacity to swarm. It is proposed that loss of FeII oxidation is due to the reversible transposition of the insertion sequence IST1 into resB encoding a putative cytochrome c-type biogenesis protein. Downstream from resB and co-transcribed with it is resC, encoding another putative cytochrome biogenesis protein. IST1 insertional inactivation of resB could result in the loss of activity of its target c-type cytochrome(s). This putative target cytochrome(s) is proposed to be essential for FeII oxidation but not for sulfur oxidation. Curiously, resB and resC pertain to the proposed system II cytochrome biogenesis pathway whereas γ Proteobacteria, of which T. ferrooxidans is a member, normally use system I. This could represent an example of lateral gene transfer.

Thiobacillus ferrooxidans, Iron oxidation, Insertion sequence, Cytochrome biogenesis, resB, resC, Lateral gene transfer, IST1

1 Introduction

Thiobacillus ferrooxidans is a chemolithoautotrophic, acidophilic, Gram-negative bacterium that participates in the bioleaching of minerals from ores. It can derive the electrons required for growth from the oxidation of ferrous iron (FeII) at a pH optimum of about 2.5 or from sulfur and reduced sulfur compounds at a pH of between about 2 and 4. It uses oxygen as a final electron acceptor for FeII and sulfur oxidations. It fixes carbon dioxide using a Calvin-Bassham-Benson scheme, utilizing NADH as reducing power, and it can fix nitrogen [1].

Biochemical investigations have identified a number of potential proteins involved in FeII oxidation (reviewed in [2]). However, the order of these proteins in the pathway has not been rigorously established, nor is it certain if all the proteins of the pathway have been identified. Genetic analysis of mutants of T. ferrooxidans could greatly aid in the elucidation of the iron oxidizing pathway. Towards this end we isolated a mutant of the strain ATCC 19859 that undergoes rapid phenotypic switching between a wild-type state in which it is capable of oxidizing FeII and/or sulfur and reduced sulfur compounds and a mutant state in which it has lost the capacity to oxidize FeII but retains the ability to oxidize the sulfur compounds [3]. In the mutant state it also exhibits a swarming phenotype. It has been suggested that phenotypic switching is associated with the reversible transposition of a repetitive element into a specific 2.8-kb BamHI fragment of genomic DNA [3, 4]. The repetitive element has been now been identified as an insertion sequence termed IST1 (GenBank accession number 66426) and in this paper we describe the genetic structure of the specific 2.8-kb BamHI restriction fragment referred to above together with a region of the genomic DNA that adjoins this BamHI fragment.

2 Materials and methods

T. ferrooxidans ATCC 19859 was obtained from the American Type Culture Collection. It was maintained in modified 9K-ferrous iron medium [6]. Escherichia coli C600 was grown in LB broth. An iron oxidizing mutant of T. ferrooxidans ATCC 19859 was isolated as described [3] and a specific 2.8-kb BamHI restriction fragment, implicated in phenotypic switching, was cloned as described [4, 5].

DNA cloning, Southern blot hybridization, Northern blot hybridization, primer extension analysis and DNA sequencing and other molecular biology techniques were carried out by standard procedures [7]. RNA was isolated according to the technique of Hagen and Young [8] and was treated an additional two times with DNase 1 (BRL Inc.) at 68 µg ml⁻¹ for 1 h at 37°C. For RT-PCR the following primers were used: 5′-GCA TAC CCA GTA CCC TT (primer 1, Fig. 3) and 5′-TGC TTG CTG ATA CGG TTC (primer 2, Fig. 3). 0.5 µg of each of these primers was mixed with 3 µg of RNA in a total volume of 25 µl according to the manufacturer's instructions (BRL Inc.) and the reaction was incubated at 70°C for 10 min followed by rapid quenching in ice. The reaction was continued at 42°C with 200 U of reverse transcriptase (Superscript II, BRL Inc.) for 50 min followed by inactivation at 70°C for 15 min. PCR amplification was carried out according to the following scheme of cycles: (1) 94°C 2 min, (2) 94°C 30 s, (3) 49°C 30 s, (4) 72°C 1 min, 72°C 10 min. Steps 2, 3 and 4 were repeated 30 times.

3

A: RT-PCR, demonstrating the presence of RNA transcripts from, and the co-transcription of, T. ferrooxidans resB and resC. B: Map illustrating the position of the primers used for the RT-PCR. Lane 2 shows the result after RNaase treatment of the RNA template showing that primer amplification is RNA template dependent. Lane 3 shows a bp standard.

DNA downstream from the 2.8-kb BamHI fragment was cloned by inverse PCR (Elongase, BRL Inc.) using genomic DNA digested with SphI. The primers used were 5′-CATTAGCCTCTTTCAGTACC and 5′-CACCCACTTACCTATCATGG. The expected size of the amplified product was determined by Southern blotting and the resulting fragment was of the predicted size.

DNA and protein data banks were searched using the Blast program [9, 10]. Multiple protein sequence alignments were made using ClustalW and protein transmembrane predictions were made using TMPRED. Both of these programs were accessed through the WIT workstation [11]. Gene organization was searched using the WIT workstation [11].

3 Results

The 2.8-kb BamHI fragment into which IST1 inserts during phenotypic switching, together with flanking downstream DNA, was sequenced. A summary of the features found by an analysis of this sequence is shown in Fig. 1A. Fig. 1B shows part of the sequence of resB into which IST1 inserts in the mutant mode of T. ferrooxidans when it is unable to oxidize FeII. In this location and orientation the IST1 provides an in-frame stop to the translational frame of resB (Fig. 1B). Additional translational stops are found in the IST1 in the other two frames of resB (not shown).

1

A: Diagrammatic representation of the genomic region of T. ferrooxidans ATCC 19859 that is associated with the loss of FeII oxidizing ability during phenotypic switching. The position of insertion of IST1 (in the mutant mode when it is unable to oxidize FeII) is shown. The arrow indicates the proposed direction and co-transcription of the genes resB and resC. The nucleotide sequence from the stop codon of resB to the start codon of resC is shown. The scale is in kilobases. B: Part of the DNA sequence of resB into which IST1 (black rectangle) has inserted when in the mutant mode. The lines over the sequence mark the target site duplication generated when IST1 inserts. Arrow 1 marks the direction of transcription of resB and arrow 2 that of the transposase gene of IST1 [5]. The filled stalked circle marks a translational stop in IST1 in the reading frame of resB. The open stalked circle is the translational stop of the transposase of IST1.

Two open reading frames (ORFs) were detected (Fig. 1A) and the results of a Blast search shows that they are similar to the ResB and ResC families of proteins of Aquifex aeolicus, the cyanobacterium Synechocystis sp., the cyanobacterium Synechococcus PCC7002 and the alga Odontella sinensis (Table 1). The similarities are considered to be statistically significant according to the Blast scores and P values (Table 1). ResB is also known as CcsI or Ycf44 and ResC as CcsA or Ycf5, depending on the organism and initial description [12]. Based on the homologies observed we propose the names resB (GenBank accession number AF089765) and resC (GenBank accession number AF089766) for the two T. ferrooxidans ORFs and ResB and ResC for their respective putative protein products (Table 1).

1

Results of a Blast search [9, 10] of combined data banks using the putative ResB and ResC of T. ferrooxidans

T. ferrooxidans		Homology to T. ferrooxidans
Proposed gene name	Proposed protein name	Organism	Gene identification	Blast score	P value
resB	ResB	Aquifex aeolicus	AE000771	127	7.6e−19
		Synechocystis	D90910	118	3.0e−14
		Synechococcus	AF052290	120	5.2e−06
resC	ResC	Aquifex aeolicus	AE000769	509	7.4e−49
		Odontella	P49523	305	4.6e−45
		Synechosystis	Z72480	321	5.2e−44

Only the three most significant similarities found are shown.

The predicted number of amino acids, molecular masses and isoelectric points of the putative ResB and ResC of T. ferrooxidans are shown in Table 2.

2

Predicted molecular mass and isoelectric points (pI) of the putative protein products of the resB and resC genes of T. ferrooxidans

Putative protein	Number of aa	Predicted Mr (kDa)	pI
ResB	592	65.6	9.48
ResC	382	42.7	9.36

Computer analyses (TMPred) of the putative ResB and ResC of T. ferrooxidans reveal that they are possible integral inner membrane proteins with four and nine predicted transmembrane regions, respectively. Their predicted orientations in the membrane together with the predicted orientations of their orthologs are shown in Fig. 2. These orthologs are also predicted to be integral membrane proteins and it can be seen from inspection of Fig. 2 that, with a few exceptions, there is substantial agreement between the number and position of the proposed transmembrane domains.

2

TMPred predictions of transmembrane domains (numbered solid boxes) and membrane orientations of the putative (A) ResB of T. ferrooxidans and its orthologs and (B) ResC of T. ferrooxidans and its orthologs. Tf=T. ferrooxidans, Aa=Aquifex aeolicus, Ssp=Synechocystis sp., Syn=Synechococcus sp. and O=Odontella sp. Alignments were derived from a ClustalW analysis. The data bank identities of the respective genes are provided in Table 1. Os above the individual alignments indicates the loops oriented towards the outside or P side. In A the asterisk indicates a region in the T. ferrooxidans sequence that apparently does not have a counterpart in the other three sequences. Open boxes indicate regions of minimum homology between the T. ferrooxidans sequence and the other sequences. In B amino acid alignments for a section of the typical tryptophan rich C-terminal loop are given. The scale bars for A and B can be found above the respective figures and are in numbers of amino acids with respect to T. ferrooxidans ResB and C.

The predicted transmembrane domains and membrane orientations of ResB and ResC have not been experimentally verified in any of the above cases. However, the predictions have been experimentally verified for ResC (CcsA) of Mycobacterium tuberculosis[13]. The M. tuberculosis ResC has significant similarity to the proposed T. ferrooxidans ResC (Blast score=315, P value=2.7e⁻²⁸) and the alignment of the transmembrane regions and the membrane orientation coincide substantially with those predicted for the T. ferrooxidans ResC. In addition, the hypothetical ResC of T. ferrooxidans exhibits the typical tryptophan rich region of other characterized ResC proteins (between membrane domains 8 and 9, Fig. 2B), which have been shown to be involved in heme attachment in both apocytochromes c₆ and f[14].

ClustalW [11] was used to generate the alignments presented in Fig. 2 and, in addition to the above mentioned regions of similarity, the respective proteins exhibit substantial similarity over most of their length. Notable exceptions include an additional region (between amino acids 340 and 398) in the predicted outward facing loop between transmembrane domains 3 and 4 in T. ferrooxidans not found in the other ResB proteins (Fig. 2A). Also detected were three regions in the same outward facing loop of T. ferrooxidans with low similarity to the equivalent regions of the other ResB proteins. Two of these regions flank the extra region exhibited by T. ferrooxidans resulting in a total region of about 150 amino acids more specific to T. ferrooxidans than to the other indicated organisms. These differences might be involved in the conformational stabilization and function of ResB in extremely acidic (pH 2) periplasm of T. ferrooxidans.

In the case of ResC, Blast and ClustalW analyses reveal very significant similarity of the proteins over their major length with the exception of two extra domains in the N-terminal region of ResC of T. ferrooxidans (Fig. 2B). Also notable is the apparent lack of a transmembrane region in the T. ferrooxidans sequence between its eighth and ninth transmembrane domains. TMPred identified a weak possible transmembrane domain in this region (score=271, where >500 is considered significant) and the respective scores of this domain in the other proteins are 1120, 620 and 883, which might be considered marginal. Thus it remains ambiguous if T. ferrooxidans contains a transmembrane domain in this region or if the others lack one.

Codon usage analysis of the T. ferrooxidans resB and resC reveals significant differences from the average codon usage of T. ferrooxidans genes. For example, the two codons for Phe, UUU, and UUC, are used 34% and 66% on average in T. ferrooxidans whereas in T. ferrooxidans resB and resC these ratios are approximately reversed (66% and 33% and 64% and 36% respectively). Several other marked differences in codon usage were observed and can be determined from an inspection of the T. ferrooxidans codon usage table which we have deposited elsewhere [15]. Analyses of the GC content of resB and resC show that they are 49 and 54% G+C respectively whereas T. ferrooxidans is thought to be about 54% G+C.

Northern blot analysis of RNA extracted from wild-type T. ferrooxidans failed to reveal discrete transcriptional products of resB and/or resC and primer extension failed to demonstrate unique transcriptional start sites of these genes (data not shown). However, RT-PCR experiments of RNA isolated from wild-type T. ferrooxidans confirm the presence of RNA transcripts from both genes, supporting the idea that they are functional genes and not merely hypothetical ORFs. RT-PCR also demonstrates that resB and resC are co-transcribed (Fig. 3).

The distance between the end of the ORF of resB and the start of the ORF of resC is only 34 nucleotides (see Fig. 1) and does not appear to contain an E. coli type ρ independent transcription terminator nor an E. coli type σ₇₀ promoter consensus. Unfortunately, the lack of information regarding T. ferrooxidans genetic control regions does not allow a definitive evaluation of the presence or absence of equivalent T. ferrooxidans transcriptional stops and promoters in this region, but the RT-PCR experiments and the close proximity of the genes argue that ResB and ResC are cotranscribed.

4 Discussion

The phenotypic switching of T. ferrooxidans ATCC 19859 involves the reversible change between a wild-type phenotype in which both FeII and sulfur (and various forms of reduced sulfur) can be oxidized and a mutant form that can only oxidize sulfur and reduced sulfur. The wild-type state is also associated with the formation of relatively compact colonies on solid media whereas, in the mutant state, the microorganism exhibits swarming [3]. The rate of phenotypic switching is too high to be accounted for by normal mutation rates and it was proposed that it was correlated with the reversible transposition of a repetitive element into a specific 2.8-kb BamHI fragment of the genome [3]. The repetitive element has been identified as an insertion sequence termed IST1 [4, 5].

In this paper it is demonstrated that this 2.8-kb BamHI fragment and its downstream DNA contain two ORFs whose hypothetical products have substantial similarity to ResB and ResC, proteins that are involved in the maturation of c-type cytochromes. The putative ResB and ResC are predicted to be integral inner membrane proteins like their orthologs. In the case of ResC, the homology includes a characteristic region rich in tryptophan that has been proposed to be involved in the attachment of heme to both apocytochrome c₆ and f in Chlamydomonas reinhardtii[14].

Insertion of IST1 during phenotypic switching is shown to occur in the resB gene in a position that places an in-frame translational stop with respect to the resB ORF. This suggests that the loss of ability to oxidize FeII could be due to the absence of a functional ResB which, in turn, leads to improper processing of a c-type cytochrome that is involved in the FeII oxidation pathway. The nature of this target cytochrome remains to be established but it is predicted to be essential for FeII oxidation and to be dispensable, or not involved, in the oxidation of sulfur compounds.

ResB and ResC are thought to be involved in the transport and correct insertion of the heme group in c-type cytochromes [12]. Heme addition to c-type cytochromes does not take place in the cytoplasm, rather the apocytochrome c is first passed through the membrane by the Sec transport system and correct folding and heme addition take place on the periplasmic side aided by various cytochrome biogenesis proteins. The heme is passed separately through the membrane and this is achieved using additional cytochrome biogenesis proteins. However, the number and type of cytochrome biogenesis proteins involved in these processes differ between organisms.

Kranz et al. [16] have proposed that there are three different systems involved in cytochrome c-type biogenesis. System 1 organisms, including α and γ Proteobacteria, plant mitochondria and Archaea, use the Hel family of proteins (and other proteins), whereas system 2 organisms, including Gram-positive bacteria, cyanobacteria and plastids, use the Res family of proteins. System 3 organisms, including fungal, invertebrate and vertebrate mitochondria, have a much reduced method of cytochrome c-type biogenesis that involves neither the Hel nor the Res proteins. It is curious that T. ferrooxidans, which is a member of the γ Proteobacteria, appears to use the Res protein family pertaining to system 2 organisms. It remains to be seen if T. ferrooxidans also has copies of the hel genes in addition to the res genes, although all the entirely sequenced organisms and organelles that have the Res family do not have the Hel family and vice versa. Perhaps the presence of the res genes in T. ferrooxidans is a result of lateral gene transfer and the codon usage of its putative resB and resC do not conform to the general codon usage of T. ferrooxidans consistent with this idea. However, the base compositions of resB and resC are close to the average composition of T. ferrooxidans. The possible lateral transfer of the system II cytochrome biogenesis genes during evolution has been postulated to have occurred in other instances [17].

In addition to the above mentioned ResB and ResC, system II organisms use ResA and CcdA for cytochrome-c type maturation. These latter proteins are thought to be involved in the maintenance of the cytochrome in a reduced state during transport and maturation. It remains to be determined if T. ferrooxidans has orthologs of these proteins.

Inspection of the data base in WIT [11] suggests that resB and resC are organized in operons in Bacillus subtilis, Campylobacter, Neisseria gonorrhoeae and Neisseria meningitidis but not in A. aeolicus or Helicobacter pylori. This postulation is based on the short intergenic space found between resB and resC and has not been experimentally verified except in the case of B. subtilis[18]. The short intergenic region between resB and resC coupled with the detection of a cotranscribed RNA argues that these genes are part of an operon in T. ferrooxidans. Whether the insertion of IST1 into resB could have a polar effect on the transcription of resC awaits experimental investigation.

With respect to the phenotype observed in phenotypic switching, it is not unreasonable to postulate that the loss of ability to oxidize FeII could be due to an improperly matured cytochrome c-type protein(s). Such a cytochrome is postulated to be essential in an electron transfer pathway during FeII oxidation but not to be used or at least to be disposable for sulfur type oxidations. However, the cause of the other characteristic feature of phenotypic switching, namely the change to a swarming mode of movement, remains to be explained.

Biochemical studies [2] have identified several candidate proteins involved in the oxidation of FeII including at least one FeS protein, a small copper-containing protein called rusticyanin that has similarity to the plastid protein plastocyanin, various soluble cytochromes and two terminal cytochrome oxidases. However, the order of these proteins in the electron transfer chain in vivo remains to be rigorously proved and it is not certain if all the components of such a chain have yet been identified. Genetic analysis of mutants involved in FeII oxidation could greatly aid in verifying what has been suggested by these biochemical studies and in revealing novel aspects of the pathway as exemplified in the present study. Unfortunately, the genetic manipulation of T. ferrooxidans is still at a rudimentary stage of development. Although techniques for introducing DNA into T. ferrooxidans have been described [19, 20] they have not found widespread application. In the absence of such techniques it becomes difficult to generate genetic support for models of FeII oxidation and the use of native insertion sequences such as IST1 offers a promising alternative route for the elucidation of not only the FeII oxidizing pathway of T. ferrooxidans but also of other biochemical functions. Such an option may be especially important when the full sequence of T. ferrooxidans becomes available.

Acknowledgements

This work was supported by Fondecyt Grant 1980665. The part of the work carried out at Lawrence Livermore National Laboratory was performed under the auspices of the US Department of Energy, Contract W-7405-Eng-48.