INTRODUCTION

Top Previous Next References

The genus Paracoccus is one of the most distantly related of the Proteobacteria to Escherichia coli (178) as judged by 16S rRNA sequence. For many years, the sole representative of the genus was Paracoccus denitrificans, first isolated in 1908 by Beijerinck (13) as Micrococcus denitrificans. The original selection of this species was based on its ability to convert nitrate into molecular nitrogen. Improved molecular phylogenetics have led to the inclusion of Thiobacillus versutus (as Paracoccus versutus [145]) and Thiosphaera pantotropha (101, 178, 233) into the genus and to the addition of P. kocurii (203), P. alcaliphilus (301), P. aminophilus (300), P. aminovorans (300), P. thiocyanatus (145), and P. solventivorans (264). More recently, two other species have been characterized by using 16S rRNA (P. marcusii [112] and P. alkenifer [170]), but no other properties of these species have been published.

These newer species were isolated by using a range of organic and inorganic compounds, including acetone (P. solventivorans), dimethylformamide (P. aminovorans and P. aminophilus) and thiocyanate (P. thiocyanatus), as growth substrates. Recently, it has been shown that some strains of P. denitrificans can use carbon disulfide (139, 233). These properties raise the possibility of using Paracoccus species for bioremediation, particularly since most species in the genus can use nitrate and its reduction products as an alternative electron acceptor to oxygen during anaerobic respiratory growth (except P. aminovorans, P. aminophilus, and P. alcaliphilus [145]). Unifying characteristics of the species include an obligately respiratory mode of growth and the use of ribulose bisphosphate carboxylase/oxygenase to fix carbon during methylotrophic or chemolithotrophic growth. All these organisms are characterized by a high genomic guanine-plus-cytosine (G+C) content (63.8 to 70.2% [145]).

The electron transport chain used for aerobic growth by P. denitrificans has long been used as a model for the mitochondrial electron transport chain (137, 280), since it possesses a full complement of proteins with counterparts in mitochondria: electron transport flavoproteins, NADH-ubiquinone oxidoreductase, bc₁ complex, c-type cytochromes, and an aa₃-type terminal cytochrome oxidase (Fig. 1). This is in contrast to the usual bacterial model organism, Escherichia coli, which does not possess some of these complexes. Branches of the "conventional" electron transport chain (75a) allow the obligately respiratory members of Paracoccus to grow under different oxygen concentrations, to use N-oxides as alternative electron acceptors, and to use a variety of carbon sources, including amines and alcohols (Fig. 1). The interest in P. denitrificans electron transport has led to the striking achievement of the determination of the crystal structure of the terminal aa₃-type cytochrome c oxidase (136). Other redox proteins isolated and structurally characterized from species within the genus include methylamine dehydrogenase, amicyanin, cytochrome c₅₅₁ (45, 46, 73), cytochrome c₅₅₀ (19), pseudoazurin (321), electron transfer flavoprotein (245), and cytochrome cd₁ nitrite reductase (6, 87).

View larger version (45K): [in this window] [in a new window]   FIG. 1.   Branched electron transport chains of Paracoccus species. Enzyme complexes colored black indicate a periplasmic location. Electron transfer between cytochromes c552 and c550 has not been demonstrated experimentally but is possible, given the redox potential of the proteins. The exact nature of the roles of cytochrome c550 and pseudoazurin is currently being studied. UQ = ubiquinone, UQH2 = reduced ubiquinone. TOMES, thiosulfate oxidation multienzyme system.

Organisms such as Paracoccus species often have to face large fluctuations in the free-oxygen concentration. The adaptive responses of P. denitrificans to changing environmental conditions sometimes resemble that of E. coli, and parts of the signal transduction cascades appear to be common between these organisms. A typical example concerns the switch from aerobic to nitrate respiration. Optimal synthesis of nitrate reductase in both organisms requires a coordinated reaction to two different types of environmental trigger: the absence of oxygen and the presence of nitrate. The molecular basis for this type of regulation in E. coli is now well understood (165, 299). To date, two proteins similar to the fumarate/nitrate respiration (FNR) regulatory protein family (important in E. coli for the sensing of oxygen and the induction and repression of several operons) have been discovered in P. denitrificans. However, many other types of regulation must occur in order to account for the diversity of electron transport shown in Fig. 1.

The derivation of the strains of P. denitrificans from the first strain isolated by Beijerinck has been well reviewed by Goodhew et al. (101), who found that the type strain (ATCC 17741) was a direct subculture of the original strain isolated by Beijerinck. Relatively few new isolates of P. denitrificans have been found, and differences between these strains of P. denitrificans have been described (101, 199, 312). However, most of the molecular biology has been performed on a single strain, P. denitrificans Pd1222. The strains used and the various loci discussed in this review are listed in Table 1. The position of the strain initially named Thiosphaera pantotropha (247) has recently been the subject of some controversy. Although Ludwig et al. (178) reclassified T. pantotropha as P. denitrificans, there is doubt (233) about this reclassification (101, 139, 281). The strain of P. denitrificans used by Ludwig et al. itself appears to be not entirely a typical P. denitrificans strain as judged by analysis of either c-type cytochromes (101) or methyl fatty acids (5). More recently, an extensive survey of various P. denitrificans strains, based on a 16S rRNA analysis, has been undertaken (233). The outcome is a proposal to name Thiosphaera pantotropha as Paracoccus pantotrophus, a species to which several strains of P. denitrificans held in culture collections for many years may be transferred. These changes in nomenclature are likely to cause confusion. It is important to note that since 1993 some research groups have continued to use the name Thiosphaera pantotropha while others have adopted P. denitrificans GB-17. A careful reading of the literature is needed to identify which strain of P. denitrificans has been used in a particular study. At the time of writing, the proposal to revive Thiosphaera pantotropha under the name Paracoccus pantotrophus seems destined for acceptance (233); we have used the name P. denitrificans GB-17 in this review. It has been our experience that apart from the necessary considerations that must be taken into account as far as antibiotic resistance are concerned, molecular genetic techniques described for P. denitrificans may equally be applied to P. denitrificans GB-17 (P. pantotrophus). The commonly used P. denitrificans Pd1222 is intrinsically resistant to spectinomycin but sensitive to streptomycin, whereas the reverse is true for P. denitrificans GB-17 (P. pantrophus). The latter is resistant to lead and arsenic, but other P. denitrificans strains are not. This review focuses mainly on P. denitrificans, the species on which the majority of molecular biological work has been performed: where necessary, we distinguish between strains of P. denitrificans, but note that the species designated P. denitrificans GB-17 also has the strain numbers LMD 92.62 and LMD 82.5 in the literature. We would refer the reader to the forthcoming article by Rainey et al. (233) for more information.

The genus Paracoccus is member of a part of the alpha Proteobacteria known as the Rhodobacter group. Paracoccus is closely related to the physiologically well-studied photosynthetic species Rhodobacter sphaeroides and Rhodobacter capsulatus, but species of Rhodovulum (147), Sagittula (100), Amaricoccus (184), Octadecobacter (103), Roseobacteria (78), and Tetracoccus (27) are also members of the group.

The molecular biology of the genus has developed considerably since de Vries et al. (68) first obtained a mutant of P. denitrificans that was amenable to genetic techniques. The review by Steinrücke and Ludwig (277) considered a number of aspects of the molecular biology of P. denitrificans, including a proposed promoter structure (unique to the genus) and related aspects of gene regulation. A considerable amount of new information has become available, which in part does not confirm the earlier proposals regarding promoter structure and otherwise is of general interest in the context of bacterial respiration, for which P. denitrificans is a model organism. Thus, a new review is timely. We consider the molecular genetics of the commonly used strains of P. denitrificans and, to a lesser extent P. versutus, the organisms on which the majority of structural, biochemical, and genetic work has been done.

(The sequences referred to in this review and their annotations are from GenBank 106.0 (released March 1998) and EMBL 54.0 (released March 1998), plus their cumulative updates until 1 May 1998, held at the Oxford University Molecular Biology Data Centre, Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford OX1 3RE, United Kingdom.)

Relatively little work has been done on Paracoccus species to determine the overall genetic makeup of these bacteria, apart from their relatively high G+C content, which has been determined during classification studies (see, e.g., reference 199). However, since 82 loci have been sequenced (listed in Table 1), it is now possible to derive valuable information about the genomics of the genus.

The G+C content of the fragments of the P. denitrificans genome that have been sequenced to date is 65.63%. This compares well with the published genomic G+C content of 66.5% (145). The difference most probably arises because sequences submitted to the databases are biased toward coding regions of DNA. The total amount of P. denitrificans DNA sequenced (by 31 December 1997) is 166,864 bp, which can be estimated to be about 4% of the total genome (Table 2). This DNA contains 162 open reading frames (ORFs), of which 129 have had a function assigned to them, either by biochemical demonstration or by inference from closely related genes from other organisms. A compilation of this information for P. versutus is presented in Table 2.

Within the ORFs from P. denitrificans identified so far, there is a bias at the third codon to guanine or cytosine. These nucleotides occur in the third position in 84.03% of codons. The GC bias is also reflected in the frequency at the first (61.13%) but not the second (52.73%) position (Table 3). This codon usage is slightly different from that previously reported (277) because of the larger number of ORFs considered in this study. The additional ORFs included here also contain very rare codons (CTA and TTA for leucine for example), so that all combinations are represented. Hence, the tRNA composition of P. denitrificans cannot be deduced confidently from codon usage. A similar codon bias is also seen in P. versutus (data not shown), but guanine or cytosine occurs in the first position in 67.13% of codons. The codon usage tables used in this review (Genetics Computer Group format) are available on request from the authors for both P. denitrificans and P. versutus.

Bacterial genomes are generally believed to be a single circular DNA molecule, with the model being E. coli. The alpha subgroup of the Proteobacteria, a division which includes Paracoccus, contains notable exceptions to this: Rhodobacter sphaeroides has two circular chromosomes (282), and Rhizobium meliloti has three (267). Agrobacterium tumefaciens C58 also has two chromosomes, but one is linear and the other is circular (2). The reason why these bacteria have multiple replicons is obscure: although some genes are duplicated (for example, the two copies of the carbon dioxide fixation genes in R. sphaeroides on separate chromosomes [93]), genes forming enzymes for a complete pathway are scattered over all the replicons (see http://capsulapedia.uchicago.edu for emerging results on R. capsulatus). P. denitrificans has also proved to have an unusual genomic structure.

When chromosomal DNA from P. denitrificans Pd1222 was separated by pulsed-field gel electrophoresis, it became apparent that the genome consists of three distinct DNA molecules of 1.83, 1.16, and 0.67 Mbp (323), designated molecules I, II and III, respectively. The behavior of the molecules under various electrophoretic conditions suggested that at least the two smaller ones were linear. To determine if the molecules were large plasmids conferring specific properties to P. denitrificans, probes to respiratory genes were used to gain an insight into gene distribution. Genes coding for the aa₃-type oxidase were spread between molecules I and II, while ubiquinol oxidase genes were found on molecule III. The genes encoding specific pathways, if transcribed from separate loci, appeared to be randomly distributed: for example, a methanol oxidation gene (mxa) was found on molecule I but the cytochrome c₅₅₀ structural gene (cycA) and the S-formylglutathione hydrolase gene (fghA) were found on molecule II. It thus seems likely that these three replicons comprise the P. denitrificans genome and will not be replicated independently of one another (323). However, the presence or absence of rRNA genes was not investigated, and so it was not possible to say which, if any, of these molecules were true chromosomes.

The composition of the genomes of other strains and species of Paracoccus varies. P. denitrificans GB-17 and DSM 65 both possess four DNA molecules of 2.2, 1.5, 0.71, and 0.5 Mbp (323) and are proposed P. pantotrophus strains (233). The electrophoretic characteristics of the 0.71-Mbp molecule indicate that this molecule is in closed-circular rather than linear form. Additionally, a much smaller molecule of less than 1 Mbp was seen in some preparations (323). A plasmid (pTAV1) of 107 kbp has been isolated from P. versutus and has been used to construct minireplicons (8, 9). P. versutus cured of the plasmid retained wild-type growth characteristics, except with respect to cesium and barium resistance (8). A second linear replicon (pTAV2) has also been found in P. versutus (201).

The possession of an efficient means of ameliorating the effects of the introduction of foreign DNA into a cell is an important trait for a microorganism living in environments where mixed cultures occur. However, when these bacteria are transferred from their environment to the laboratory, DNA restriction and modification systems present a problem to the molecular geneticist. Studies of regulation in P. denitrificans NCIMB 8944 (traditionally used for biochemical studies) were hampered by the lack of a mutant suitable for the maintenance of plasmids without significant recombination into the genome. Stable inheritance of extrachromosomal material does, however, occur in P. versutus, as well as in P. denitrificans GB-17. Furthermore, the type culture of P. denitrificans (ATCC 17741) will maintain plasmids in the wild-type form of the strain (143).

An undefined P. denitrificans N-methyl-N'-nitro-N-nitrosoguanidine chemical mutant (Pd1222) which had a recombination-minus phenotype and an enhanced frequency of conjugation was isolated from DSM 413 (68). The useful property of resistance to rifampin was subsequently introduced, and this antibiotic resistance can be used to select against E. coli strains present in bi- or triparental mating experiments. However, the apparatus for recombination of plasmid DNA with the genome still remained in this mutant, and to manufacture a truly recombinant-deficient organism, Fernandez de Henestrosa et al. (75) isolated and mutated the recA gene of P. denitrificans Pd1222. This new derivative should prove valuable in future work.

Despite the high identity of recA proteins within the Proteobacteria (144) and the high identity within the coding regions (P. denitrificans recA is 88.6% identical to the Rhodobacter sphaeroides gene and 64.3% identical to that from E. coli [75]), regulation of P. denitrificans recA differed not only from that of E. coli but also from that of the phylogenetic near neighbor R. sphaeroides. No LexA binding site could be seen in the putative promoter region of P. denitrificans recA, but the use of a plasmid containing the promoter translationally fused to a reporter suggested that conditions for repression and activation of the gene in P. denitrificans were similar to those required by E. coli. Further evidence for differences in the details of control of the recA gene in P. denitrificans were obtained when the reporter gene was fused to recA promoters of Rhizobium etli, R. sphaeroides, and R. capsulatus (75). When these fusions were introduced into P. denitrificans, the reporter was induced (on the addition of mitomycin C, which induces the SOS response) only from the Rhizobium etli promoter. This was not the expected result in view of the closer phylogenetic relationship of Paracoccus and Rhodobacter than of Paracoccus and Rhizobium. Examination of the promoter sequences revealed little similarity between the Rhodobacter promoters and that of Paracoccus, but the Rhizobium etli promoter contained a similar region of dyad symmetry (5'-TTGN₁₀CAA-3' in P. denitrificans and in R. etli, N = 11). Interruption of this inverted repeat in Rhizobium etli led to inactivation of the recA promoter (284). It would thus appear that P. denitrificans possesses a recA system more like that found in the rhizobia than in Rhodobacter species.

rRNA functions in the assembly of the ribosome but has assumed new significance with the realization that it can be regarded as a molecular clock (204, 329). The 5S, 16S, and 23S genes of P. denitrificans have been sequenced (Table 1), as have the 16S genes from all of the other species of Paracoccus. Unfortunately, due to the use of thermal polymerase amplification involving primers to conserved sequences within the genes, little information can be obtained about the promoters, which would be expected to be of the ⁷⁰ RNA polymerase (RNAP) type. The derivation of a consensus ribosome binding site from the 16S rRNA sequence has been discussed previously (277).

The transcript from the 23S gene is unusual in that it seems to be unstable in some preparations when isolated with total RNA from P. denitrificans GB-17 (P. pantotrophus), appearing on formamide-agarose gels as two smaller molecules (one the same size as 16S rRNA) cleaved at a distinct site (252). This phenomenon has also been noted in R. capsulatus (343). Since the integrity of the 16S rRNA transcript is often used as an indicator of the state of degradation of RNA, this may give a misleading result when considering the quality of a P. denitrificans GB-17 total RNA preparation. The instability of the 23S rRNA might indicate the presence of an intervening sequence (usually originating from an insertion sequence or other mobile genetic element, appearing as inverted repeats and/or an ORF[s] in the middle of some rRNA genes). Such intervening sequences have been found in several bacteria, including Salmonella typhimurium, and result in no apparent intact 23S rRNA in the cell (105).

Bacterial insertion sequence (IS) elements are small, discrete elements of DNA that are integrated into the host genome or, more frequently, into naturally occurring plasmids in bacteria. The coding capacity of these elements is often limited to the synthesis of transposase, the protein which drives the transpositional event and allows the element to jump along the host DNA. The IS element IS1248, which was characterized in P. denitrificans, belongs to a larger family of elements that are found in strains belonging to different clusters of gram-positive as well as gram-negative bacteria. This family includes IS869 and IS427 of Agrobacterium tumefaciens (67, 224), IS402 of Pseudomonas cepacia (77), ISmyco of Mycobacterium tuberculosis (183), IS1106 of Neisseria meningitidis (156), Tn4811 of Streptomyces lividans (44), ISRm4 and a similar element from Rhizobium meliloti (84, 227, 268), and IS1031 of Acetobacter xylinum (50). Trapping of IS1248 occurred during plasmid transfer experiments with derivatives of suicide vector pRVS3, which appeared to be integrated into the genome via IS1248-mediated cointegrate formation (311). The finding that the vector was flanked by identical copies of the transposed IS element as well as of the target site, 5'-CTAG-3', even suggested that integration had occurred via replicative transposition, an event which is preceded by a staggered cleavage of the IS target site, resulting in duplication of it. IS1248 is 830 bp long and has 13-bp imperfect inverted repeats at the borders. Two of the five ORFs identified in IS1248 correspond to counterparts from the other members of this IS family. Since these putative genes have the potential to encode proteins that are hydrophilic overall and have relatively high isoelectric points, they might be the candidates for the transposase function. Two sequences are found in the inverted repeats of IS1248, which have been suggested to be involved in the transpositional event. The first sequence, 5'-GANNNNTTGAT-3', resembles the binding site for the integration host factor, which is involved in stimulation of transposition of a number of IS elements (90). The second sequence, 5'-GNNTCATAA-3', is identical to that found in related elements and may be a recognition site for their transposases. IS1248 is present in multiple (four to six) copies in the genome of many strains of Paracoccus (312), and the pattern of IS1248-hybridizing fragments appeared to be different in P. denitrificans Pd1222 and P. denitrificans GB-17 (P. pantotrophus). IS1248 is not present in P. versutus, suggesting that it invaded P. denitrificans after these two species had branched from a common ancestor. This suggestion would support the idea of horizontal gene transfer (312).

Apart from the IS1248-mediated integration mechanism, P. denitrificans has a second mechanism involved in the integration of heterologous DNA into its genome (312). The result of the latter type of integration is different from that observed for IS1248, in that the integrated DNA is not flanked by two identical sequences. Furthermore, the DNA sequences of the donor backbone and the target DNA at the integration site were found to be similar and to resemble the res site found in transposons belonging to the Tn3 family (90, 152). These res sites are an essential part of the transposon-mediated site-specific recombination system involved in cointegrate resolution. At least two copies of this integrative element are present in the genome of P. denitrificans (312).

When considering how and when a particular gene from Paracoccus is transcribed, researchers find themselves in an unusual position. It is possible to define transcript start sites and some regulatory protein binding sites (such as FNR-type proteins [310]) but not to determine where RNAP might bind or even which type of RNAP is effective. Consensus sequences that have been proposed previously (277) for Paracoccus are, as discussed below, unsuitable. In the absence of any direct biochemical or genetic evidence for the presence of an RNAP of the ⁷⁰ type in Paracoccus, it is difficult to define clearly the elements of promoters from this genus that may be involved in transcription. The promoter regions that have been sequenced rarely contain the typical 10 or 35 motifs, and workers studying Paracoccus frequently note that its promoters rarely function in aerobically grown E. coli.

An obligately respiratory organism such as P. denitrificans achieves metabolic flexibility by having many alternative electron transport chains. The bacterium must have some overall control of these branched electron transport pathways it possesses: in many cases, the concentration of more than one respiratory enzyme is either elevated or diminished under a particular growth condition, suggesting that a single regulatory protein has pleiotropic control over the expression of their allocated genes. Comparison of the promoter regions in front of the known respiratory genes and gene clusters revealed a number of sequences with a minimum of 8 bases conserved in two or more of the putative promoter regions. Palindromic sequences, which may be probable candidates for binding transcriptional activators or DNA binding proteins, can be selected from these conserved sequences. A list of these sequences is presented in Table 4. Whether these sequences indeed act as regulatory elements is speculative at the moment, but these palindromes do not resemble those noted in R. capsulatus (5'-GTGTAART-N₆-TTACAC-3' [1]), nor, in most cases, do they conform to the consensus sequence for E. coli transcriptional regulatory factors (5'-TGTGT-N_6-10-ACACA-3' [95]).

Promoter structure in Paracoccus and the Rhodobacter group of the alpha Proteobacteria. There has been no further review of Paracoccus promoter sequences since Steinrücke and Ludwig (277) deduced a consensus sequence (5'-TCGGGGN-N_(18 ± 2)-GATNGS-3') based on promoters from Paracoccus, Rhodobacter, and Bradyrhizobium. Surprisingly, little attention has been paid in general to promoters in the alpha-Proteobacteria, the division of the Eubacteria to which Paracoccus belongs. Although alternative polymerases (e.g., RpoN [37] of R. capsulatus) have been isolated, purified, and characterized, work on the binding of housekeeping holopolymerase to constitutively induced promoters is just beginning (54, 180), with Rhodobacter as the model organism. However, alignment of Paracoccus and Rhodobacter promoters (Table 5) indicates that most constitutive promoters have some sequences in common.

Termination of transcription. To date, no direct experimental evidence exists for any termination event in Paracoccus. However, mRNA analysis and other indirect evidence suggests that Paracoccus possesses both factor-dependent and factor-independent pathways for termination of transcription. A truncated form of the Rho-dependent terminator gene (rho') of R. sphaeroides 2.4.1 was lethal in the wild-type organism but partially interfered with the transcription termination machinery of E. coli. When rho' was introduced into P. denitrificans ATCC 17741, the construct was also found to be toxic (99). This suggests not only that a Rho-like system exists in the genus Paracoccus but also that the mechanism of termination is the same for R. sphaeroides and perhaps that the structure of Rho in these organisms is similar as well.

A number of respiratory systems and their underlying biosynthesis genes from Paracoccus have been characterized by molecular biological methods. The regulation of these genes has some features in common with the regulation of the genes in the more intensively studied organisms such as E. coli. The loci sequenced, their accession numbers, and the strain of origin are listed in Table 1. The respiratory pathways of Paracoccus are dependent on many metalloproteins, the best characterized of which are c-type cytochromes (Table 6). Although the biochemical mechanisms by which these proteins are synthesized are only just becoming understood, P. denitrificans has proved to be a good model organism for these studies, producing c-type cytochromes aerobically as well as under oxygen limitation.

Cytochromes c are distinguished from cytochromes of other classes by covalent attachment of the heme moiety to the cytochrome polypeptide via thioether links between the two protoporphyrin IX vinyl groups and the thiol groups of two cysteine residues in the conserved motif Cys-X-Y-Cys-His. The process of c-type cytochrome biosynthesis thus includes posttranslational modification of the apocytochrome polypeptide. A number of lines of evidence indicate that in gram-negative bacteria this process takes place in the periplasm, although this has yet to be rigorously demonstrated experimentally. Eight genes required for c-type cytochrome maturation have now been identified in P. denitrificans; all are clearly homologous to genes found in a number of other gram-negative bacteria including Bradyrhizobium japonicum, R. capsulatus, and E. coli (for comprehensive reviews, see references 218 and 285). The organization of the known P. denitrificans c-type cytochrome biosynthetic genes resembles that in R. capsulatus, in that they are distributed over at least three loci, but differs from that in E. coli, in which genes are clustered at a single locus (ccmABCDEFGH), and in the Rhizobiaceae, in which the genes are present at two loci (cycHJKL and cycVWZXY). Southern blotting of a cosmid library suggests that the three loci are separated by at least 20 kbp in the P. denitrificans genome (200). The E. coli nomenclature has been adopted for the P. denitrificans genes; an exception is cycH, which has no clear equivalent in E. coli (although it exhibits some similarity to the C-terminal region of ccmH). No gene corresponding to ccmE/cycJ has been identified in P. denitrificans but has for R. capsulatus (see http://capsulapedia.uchicago.edu).

ccmA, ccmB, ccmC, ccmD, and ccmG. ccmA, ccmB, and ccmC appear to encode the components of a membrane transporter of the ABC (ATP-binding cassette) superfamily. The corresponding hypothetical transporters in B. japonicum and R. capsulatus have been suggested to translocate heme or apocytochromes to the periplasm; however, sequence analysis indicates no similarity between CcmB and CcmC (or their homologues) and the membrane-integral components of transporters mediating the uptake of heme or other iron complexes. Supplementation of growth media with heme did not stimulate c-type cytochrome formation in mutants disrupted in ccmA or ccmB, although it elevated the levels of soluble hemoproteins and membrane-bound cytochromes b, suggesting that exogenous heme can traverse both outer and inner membranes in P. denitrificans. Expression of an apocytochrome c₅₅₀-alkaline phosphatase fusion protein and of apocytochrome cd₁ was unaffected in a ccmB::Tn5 mutant. These results suggest that the substrate for the putative CcmABC transporter may be neither heme nor c-type apocytochromes (217).

cycH. Disruption of cycH (ccmI has also been suggested as a suitable name [218]) results in loss of soluble c-type cytochromes, but low levels of membrane cytochromes c (estimated at 5 to 10% of wild-type levels) remain. Thus, CycH is not absolutely required for c-type cytochrome assembly in P. denitrificans, but it clearly increases the efficiency of the process manyfold (215). Analysis of a cycH-lacZ fusion indicates that it is expressed during aerobic growth but is induced fourfold under anaerobic growth conditions and that this induction is mediated by the transcriptional activator FnrP but not by the closely related protein Nnr (210). FnrP and Nnr are discussed further in the context of the regulation of denitrification (see below).

ccmF and ccmH. The P. denitrificans ccmF and ccmH have recently been established (225). CcmF is predicted to be a membrane-integral protein with 11 or more membrane-spanning -helices, and, as such, it is potentially a transporter; however, supplementation of growth media with heme did not stimulate c-type cytochrome formation in a mutant disrupted in ccmF. CcmH has a Cys-X-X-Cys motif and thus may be a protein-disulfide oxidoreductase, but a ccmH mutant has yet to be constructed and characterized.

hemA. While not sensu stricto a c-type cytochrome biogenesis gene, the P. denitrificans hemA gene (coding for 5-aminolevulinic acid [5-ALA] synthase) was identified during screening for mutants defective in c-type cytochrome assembly. A transposon mutant in which Tn5::phoA had integrated in the hemA promoter region, reducing but not eliminating hemA expression, was obtained. This had the effect of reducing the levels of a- and b-type cytochromes and membrane-bound c-type cytochromes in the mutant strain to about 50% of those in Pd1222 and virtually eliminating the formation of soluble periplasmic cytochromes c. Disruption of the hemA structural gene led to 5-ALA auxotrophy, indicating that P. denitrificans, like R. capsulatus but unlike R. sphaeroides (127, 195), possesses only one 5-ALA synthase (confirmed by Southern blotting) and that no 5-ALA synthase-independent route of 5-ALA synthesis exists in P. denitrificans (214).

NADH-ubiquinone oxidoreductase. The NADH-ubiquinone oxidoreductase holoenzyme from P. denitrificans is thought to contain at least 14 subunits, whereas that from mitochondria is considerably more complex, with 28 additional subunits (126, 258). Despite the difference in subunit composition, the function of the two enzymes is the same and there is considerable protein sequence homology between equivalent subunits (331, 332). Therefore, the Paracoccus proteins are named after their mitochondrial counterparts (Nqo1, Nqo2 etc.). The genes coding for these subunits are found in an operon between an ORF possibly coding for the Paracoccus UvrA (a DNA repair enzyme), and another gene (ORF240) similar to birA (biotin [acetyl coenzyme A (CoA) carboxylase] ligase). P. denitrificans UvrA has 74 and 71% identity to the equivalent E. coli (131) and Micrococcus luteus (263) proteins, respectively, while P. denitrificans BirA is 31% identical to the equivalent E. coli protein. The proposal that a bacterium such as Paracoccus is the forerunner of the eukaryotic mitochondrion (137) receives little support from the gene order of this nqo operon: the arrangement of the genes is more similar to that of chloroplasts (e.g., liverwort [202]) than to that of the bovine mitochondrion.

Succinate dehydrogenase. Succinate dehydrogenase of P. denitrificans has been purified (226) and shown to have four subunits, and the genes for these subunits appear to be in an operon (69). The enzyme contains covalently bound flavin, iron-sulfur centers, and cytochrome b, thus showing considerable amino acid sequence similarity to its mitochondrial counterpart. The promoter of the P. denitrificans sdhCDAB operon has been characterized (69), but sequence analysis alone does not provide much information on this constitutively expressed cluster (160, 226). The presence of ⁷⁰ hexamers at 10 with respect to the transcription start site (Table 5) suggests that the genes are transcribed with a ⁷⁰-like RNA polymerase.

The cytochrome bc₁ complex. After an early report on the purification and characterization of cytochrome c₁ from P. denitrificans as a polypeptide with an unusually high molecular mass (177), the bc₁ complex was isolated initially as a "supercomplex" along with cytochrome c oxidase and a membrane-bound cytochrome c₅₅₂, yielding high quinol-oxidizing activity (26). Subsequently, its subunit composition was confirmed unequivocally (340), showing that only the three subunits carrying redox centers make up a complex that is also fully competent in free energy transduction (341). The cloning of the corresponding genes (162) revealed an operon structure, fbcFBC, coding for the Rieske FeS subunit, the cytochrome b, and the cytochrome c₁ subunit. The latter is unique in having an additional N-terminal domain of around 150 amino acids (compared to the eukaryotic mitochondrial proteins, explaining the higher molecular weight of the P. denitrificans protein), with a characteristic composition (40% alanine, 38% acidic residues, and no basic residues [162]); its function is still not understood. While cytochrome b shows an amazingly high degree of sequence identity to other bacterial and mitochondrial subunits (see references 155, 285, and 290 for reviews), the existence of an additional transmembrane helix in the N terminus of the protein has been suggested (153) on the basis of using monoclonal antibody fragments in conjunction with electron microscopy. Gene and operon deletion studies, as well as expression of the fbc operon from a multicopy plasmid (92), resulted in a considerable overexpression of the complex in the homologous host. Once again, an E. coli ⁷⁰-like 10 region (TAGAAC; Table 5) can be found in the promoter region.

Cytochrome aa₃. P. denitrificans can use several terminal electron acceptors during aerobic growth, the best characterized of which is the aa₃-type cytochrome c oxidase (cytochrome aa₃). The biochemistry of this multisubunit enzyme has been reviewed extensively (62, 88, 109, 206, 291). The structure of the holoenzyme has been determined to a resolution of 2.8 Å by X-ray crystallography (136).

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Branched electron transport pathway of P. denitrificans under various conditions of oxygen limitation. FeS, iron-sulfur center; NDH, NADH-ubiquinone oxidoreductase; SDH, succinate dehydrogenase; Q, ubiquinone pool.

The cbb₃-type oxidase. Bacteria that rely entirely on respiration for the liberation of free energy are challenged under near-anoxic conditions since the terminal oxidases of the aa₃ and bo₃ or ba₃ type are unable to function at exceptionally low oxygen concentrations. One of the strategies for survival under these conditions is the recruitment of a cbb₃-type oxidase, which has a relatively high affinity for oxygen. This type of oxidase was first encountered in endosymbiotic rhizobia, which use it during nitrogen fixation in the root nodule (142, 182, 230, 231). The finding that its derived K_m value for oxygen is 7 nM may explain why the cbb₃-type oxidase supports the growth of the bacteroids in these nodules, where the free-oxygen concentration is only 3 to 22 nM (230).

Quinol oxidase. Initially it was speculated that quinol oxidase activity in Paracoccus could be ascribed to cytochrome o (51), based mainly on spectroscopic data. Less than a decade ago, more detailed studies indicated the presence of a proton-pumping quinol oxidase in whole cells (232). The complex thought to be responsible was purified shortly afterward from membranes of P. denitrificans (175). Several characteristics of this complex, such as its function and subunit pattern, suggested some similarities to the bo₃ quinol oxidase studied extensively in E. coli (reference 47 and references therein). This was substantiated by sequencing of the genes for the Paracoccus quinol oxidase (62).

Cytochrome c₅₅₀. Cytochrome c₅₅₀ is believed to function as an electron donor in several respiratory pathways including those for denitrification (30, 185), methanol oxidation and methylamine oxidation (57, 62, 74). The P. denitrificans locus coding for cytochrome c₅₅₀ is found just upstream of ctaDII (235, 315), separated by a putative Rho-dependent terminator. A similar gene order is found in the P. versutus locus, and the two species have 89% DNA identity over comparable regions. Truncated versions of the cycA-ctaDII locus from P. versutus fused to a reporter and transferred to E. coli showed that the two genes are separated by a strong terminator. No activity from the P. versutus ctaDII promoter region could be demonstrated in E. coli (296).

Cytochrome c₅₅₂. Among the many c-type cytochromes found in P. denitrificans (Table 6), a membrane-bound polypeptide, cytochrome c₅₅₂ (CycM), is the most likely mediator between the cytochrome bc₁ complex and the heme aa₃ cytochrome c oxidase, as suggested by a number of arguments. (i) Under certain solubilization conditions, a complex of this cytochrome either with the oxidase or with oxidase and the bc₁ complex has been isolated from membranes (26, 109); a high rate of electron transfer points at its efficient role as a redox link. (ii) Specific antibodies obtained against the purified protein (294) inhibited electron transport in membranes between NADH and oxygen but had no effect when partial reactions were assayed with mitochondrial cytochrome c. (iii) Cloning of the gene encoding cytochrome c₅₅₂ (294) showed a tripartite structure of the polypeptide with an N-terminal membrane anchor; membranes isolated from gene deletion mutants (293) were blocked in their electron transport from NADH to oxygen via complexes I, III, and IV, and this inhibition could be partially overcome by the addition of mitochondrial cytochrome c.

Electron transport flavoprotein. Electron transport flavoproteins (ETF) of bacteria transfer electrons between flavoprotein dehydrogenases and the respiratory chain. The electron acceptor of ETF is presumably ubiquinone in P. denitrificans. Although the mitochondrial ETF accepts electrons from a wide variety of dehydrogenases, the P. denitrificans protein shows a more restricted substrate range. ETF has been isolated in abundance from trimethylamine-grown cells but has also be shown to transfer electrons from glutyryl coenzyme A (CoA) dehydrogenase (135). However, it does not accept electrons from periplasmic dehydrogenases such as methanol dehydrogenase and methylamine dehydrogenase (56, 273).