OUP user menu

Genetic analysis of the gas vesicle gene cluster in haloarchaea

Shiladitya DasSarma, Priya Arora
DOI: http://dx.doi.org/10.1111/j.1574-6968.1997.tb10456.x 1-10 First published online: 1 August 1997

Abstract

Gas vesicles are buoyant intracellular organelles composed of a rigid proteinaceous membrane surrounding a gas-filled space. Many prokaryotic microorganisms including photosynthetic and heterotrophic bacteria and halophilic and methanogenic archaea produce gas vesicles. In the majority of cases gas vesicles function in providing vertical motility to cells in aquatic environments. Recent genetic analysis of several halophilic archaeal (haloarchaeal) species has shown that a large cluster of genes [gvp MLKJIHGFEDACN(O)] is necessary for gas vesicle formation.

Keywords
  • Buoyancy
  • Flotation
  • Organelle
  • Motility
  • Halobacteria
  • Cyanobacteria
  • Plasmid

1 Introduction

Gas vesicles were first described as phase-bright vacuole-like inclusions in prokaryotic cells by Winogradsky in 1888[1]. Seven years later Klebahn showed that these inclusions release gas when subjected to hydrostatic pressure, and they came to be known as ‘gas vacuoles’[2]. It was not until 1965 that Bowen and Jensen found using electron microscopy that the vacuoles were in fact vesicular aggregates of smaller, individual organelles that they termed gas vesicles[3]. Detailed electron microscopic examination in a wide variety of prokaryotic species has shown that gas vesicles are membrane-bounded structures which are generally cylindrical-shaped with conical ends ([4], most recently reviewed by Walsby in ref.[5]).

Gas vesicles are commonly found in many prokaryotic species inhabiting aquatic environments and have been reported in both archaea (halophiles and methanogens) and bacteria (phototrophs and heterotrophs)[5]. Ecological studies have shown that gas vesicles function in cell buoyancy for vertical motility, e.g. in stratification of cyanobacterial species, increasing their ability to respire and photosynthesize[6]. However, their function in some obligately anaerobic nonphotosynthetic microorganisms such as methanogens is unclear [7, 8].

Gas vesicles usually have a species-characteristic shape and morphology. They range in length from 50 nm to over 1 μm and in width from 30 to 250 nm [5, 9]. Narrower gas vesicles are more resistant to collapse when exposed to hydrostatic pressure and are generally found in microorganisms growing at greater depths[10]. Those microorganisms which grow in shallow pools, such as many halophilic archaea, contain predominantly wider lemon-shaped gas vesicles (Fig. 1Fig. 2) which are more susceptible to collapse. Some strains of Halobacterium are capable of synthesizing both wide lemon-shaped and narrow cylindrical-shaped vesicles [4, 1113], suggesting that they may inhabit both shallow and deep brine pools.

Figure 1

Micrographs of Halobacterium NRC-1 containing gas vesicles. A: Freeze-fracture micrograph of a cell pellet with a large number of gas vesicles visible where the fracture plane has broken through cells. The arrowhead indicates the direction of shadowing. B: Negatively stained cells containing large numbers of gas vesicles. Vesicles are visible as distinct particles within cell ghosts.

Figure 2

Micrographs of lemon-shaped gas vesicles of Halobacterium NRC-1. The upper panel is taken from a freeze-fracture specimen of cells. Rings or shallow spirals of protein subunits, probably GvpA and GvpC, are visible. The lower panel shows a negatively stained gas vesicle. The gas vesicle contains two large holes but has not accumulated stain internally, presumably because of hydrophobic forces, and has maintained a turgid structure, likely due to the rigidity of the membrane.

Gas vesicles are easily purified by flotation after cell lysis. Analysis of purified vesicles has shown that they have a thin (20 Å) lipid-free membrane composed solely of protein surrounding a gas-filled space [14, 15]. However, the membrane is gas-permeable and allows the free diffusion of many dissolved gases such as nitrogen, oxygen, carbon dioxide, and methane[16]. As a result, the gases inside vesicles are probably in equilibrium with those in the cytoplasm. Gas vesicle synthesis proceeds by the growth of small bicones to progressively larger cylindrical structures[17]. During this process, water is thought to be excluded by hydrophobicity at the inner surface of the membrane, while gases accumulate through passive diffusion[18].

Detailed biochemical analysis of purified gas vesicles suffered from technical limitations resulting from the extreme resistance of the vesicle membrane to solubilization by detergents and chaotropic agents[15]. The lack of biochemical information prevented the identification of gas vesicle proteins and for many years led to the incorrect conclusion that the organelle was composed of only a single self-assembling protein[19]. Thus far, only two proteins, GvpA and GvpC, have been demonstrated to be present in gas vesicles [20, 21], although genetic analysis has shown the involvement of many more genes in gas vesicle formation.

Genetic studies of gas vesicle formation have been carried out largely in haloarchaeal microorganisms such as Halobacterium which flourish in hypersaline brine containing 3–5 M NaCl. This has been the result of a combination of factors, including the availability of a gene probe for the major gas vesicle protein (GvpA) gene, easily identifiable spontaneous mutants in gas vesicle genes, and methodology for conducting genetic analysis in haloarchaea. As reviewed below, in the last 10 years, a cluster of over a dozen genes on large plasmids of Halobacterium have been discovered and shown to encode proteins with likely functions in structure, assembly, and regulation of gas vesicle formation.

2 Cloning and sequencing of gvpA

Initially, the major protein in gas vesicles, GvpA, was identified by N-terminal sequencing of gas vesicle preparations of two haloarchaeal and four cyanobacterial species by Walker et al.[22]. This protein was found to be highly conserved in all six species examined. The N-terminal sequence of the Anabaena flos-aquae GvpA protein was used to synthesize an oligonucleotide probe and clone the corresponding gene from Calothrix PCC 7601 by Tandeau de Marsac et al. [23, 24]. The Halobacterium gvp A gene was cloned subsequently by heterologous probing from strains NRC-1[25] and NRC817 (renamed PHH1)[26]. The sequences of the genes confirmed that all of the GvpA proteins are small, highly conserved acidic proteins of ∼8000 molecular mass with a significant hydrophobic character.

In Halobacterium strains NRC-1 and PHH1, the gvp A gene mapped to large plasmid replicons of 150–200 kb [25, 26]. These results explained the earlier observations of a correlation between high-frequency gas vesicle-deficient mutations (Fig. 3) to plasmid rearrangements [27, 28]. In addition, a second copy of the gvp A gene, named gvpB (also referred to as c-vac), differing from gvp A by two single amino acid substitutions and a small insertion near the C-terminus, was identified at another genomic site[26]. The location of the gvpB gene was reported to be on the chromosome in strain PHH1[26], and on a second large plasmid replicon, pNRC200, in strain NRC-1[29]. The gvpB gene was silent in some strains such as NRC-1, but active in other isolates, e.g. PHH1, particularly when the gvp A gene was inactivated by insertion or deletion. An interesting observation reported was that in strains expressing gvpB, the morphology of gas vesicles was altered, with a prevalence of narrower cylindrical vesicles as opposed to the wider lemon-shaped vesicles usually present in wild-type Halobacterium species [11, 12].

Figure 3

Colonies of Halobacterium producing or lacking gas vesicles on agar plates. The left panel shows colonies of strain NRC-1 which have Vac+ (milky, opaque), Vac (orange, translucent), and sectored Vac+/Vac phenotypes. Here, sectoring is the result of recombinational activities promoted by transposable elements. The colonies on the right panel are formed by strain SD109 containing the entire pNRC100 gvp gene cluster on a recombinant plasmid (pFL2[38]) plated under non-selective conditions. In this case, sectoring is the result of frequent loss of the recombinant plasmid during cell division.

3 Discovery of the gas vesicle gene cluster

Isolation of the gvp A gene was followed by the analysis of gas vesicle-deficient mutants (Vac) of Halobacterium, which occur spontaneously at a frequency of about 1%[30]. Initially, Southern hybridization analysis of mutants of strain NRC-1 showed that a 200 kb replicon, pNRC100, had undergone rearrangements in the Vac mutants, including both insertions and deletions near the gvp A gene while the gvpB gene region was unchanged[30]. In one class of completely Vac mutants, the gvp A gene was found to be deleted as a result of IS element-mediated rearrangements in pNRC100[31], while in another class of mutants, which have less than 10% of the wild-type gas vesicle content, insertions of IS elements were observed in a 2 kb region upstream of gvp A[32]. Sequencing of the upstream region at first revealed an operon containing two genes, named gvp D and gvp E, which were transcribed divergently from gvp A[32]. The phenotype of these mutants was partially Vac, suggesting their involvement in functions not absolutely required for gas vesicle formation, e.g. the regulation of gvp gene expression.

Further sequencing and transcript analysis surrounding the gvp EDA region of pNRC100 revealed the presence of 13 open reading frames, named gvp M, L, K, J, I, H, G, F, E, D, A, C and N which were organized into two divergent operons (Fig. 4) [21, 33]. The leftward operon was found to contain 10 genes, gvp D, E, F, G, H, I, J, K, L and M, while the rightward operon contained three genes gvp A, C and N. A nearly identical arrangement of genes was observed in a 150 kb plasmid of Halobacterium strain PHH1 (pHH1), except that a fourteenth open reading frame, gvp O, was found downstream of gvp N [34, 35]. Gas vesicle gene clusters have also been sequenced from the gvpB region of strain PHH1, and the chromosome of Haloferax mediterranei[35]. Conservation of gene arrangement and close sequence homology suggested that these regions are likely to be necessary for gas vesicle formation.

Figure 4

The gvp gene cluster of Halobacterium. The gvp genes are designated by boxes. The approximate locations of divergent promoters in the gvp A–D intergenic region are indicated by arrows. The leftward operon contains putative regulatory genes (gvp D and E) and minor structural or assembly genes (gvp F, G, H, I, J, K, L and M), and the rightward operon contains structural genes (gvp A and C, and also gvp N). Another open reading frame, gvp O, with a possible regulatory role is present downstream of gvp N.

4 Analysis of mutations in the gas vesicle gene cluster

A large number of both natural and in vitro constructed gvp mutants of Halobacterium have been characterized. Of the several dozen natural mutants, one-half contained IS elements in the gvp gene cluster of pNRC100 and pHH1 [3234], while the other half had lost the gene cluster by deletion[31]. Insertions were found in the gvp A promoter as well as in gvp C, D, E, J and L or M resulting in partial or complete loss of gas vesicle formation. These findings suggested that most if not all of the interrupted genes in the gene cluster are probably necessary for gas vesicle formation; however, the operon structure made it difficult to distinguish between the direct effects of gene interruption and polar effects on downstream genes.

In order to carry out systematic mutagenesis of the gvp gene cluster, complementation systems were established [36, 37]. For the pNRC100 gas vesicle genes, the entire gene cluster was cloned in an Escherichia coli-Halobacterium shuttle plasmid and introduced into Halobacterium strain SD109[31], a mutant of NRC-1 lacking the whole gene cluster. A series of mutations were constructed by the insertion of a kanamycin resistance cassette into random sites in the gvp gene cluster by linker scanning mutagenesis [37, 38]. Insertions were introduced into gvp M, L, K, J, I, H, F, E, D, C, N, and the N/O boundary. While the unmutated gvp gene cluster produced Vac+ transformants (Fig. 4), all of the mutated derivatives, except for gvp M and the N/O boundary insertions, produced phenotypically Vac transformants with less than 10% of wild-type gas vesicle content[38].

To rule out polar effects of the insertion mutants, an internal portion of the kanamycin cassette was deleted using rare restriction sites which had been incorporated near its ends, resulting in gvp mutants with small insertions[38]. Based on the phenotypes resulting from these mutations, it was concluded that four genes, gvp D, H, M and O may or may not be required for gas vesicle formation, while the other 10 genes were clearly required for wild-type gas vesicle formation. Insertions into gvp C and N produced small gas vesicles about one-quarter the length of the wild-type, suggesting that these genes act in the latter stages of gas vesicle formation. In contrast, insertions in the gvp E, F, G, I, J, K and L genes produced few if any gas vesicles, suggesting that they are necessary for earlier stages of gas vesicle formation. In several of these mutants, intracellular inclusions were observed, suggesting the accumulation of side products of gas vesicle synthesis (DasSarma, S. and Yin, L.R.-S., unpublished).

Three other gene clusters, two from Halobacterium PHH1 (on pHH1 and the chromosome), and another from Haloferax mediterranei, have also been used for mutagenic analyses. In these cases, deletion mutagenesis was done using either a single cloning vector [36, 39] or two different vectors to introduce different subsets of genes [4042], with the related halophile Haloferax volcanii, which is naturally gas vesicle-deficient, as recipient. Plasmid constructs containing the entire Haloferax mediterranei gene cluster, including gvp M at the left end and gvp O at the right end produced Vac+ transformants, while constructs lacking either of the terminal genes produced Vac transformants[39]. These results suggested that, contrary to the pNRC100 gene cluster[38], the terminal genes are required for gas vesicle formation. However, secondary effects of the terminal deletions on destabilization of upstream regions of long messenger RNAs was not ruled out in these investigations.

Deletion of the Halobacterium pHH1 gvp A, F, or a region containing the three genes, gvp G, H and I, resulted in Vac transformants, indicating the requirement of these genes for gas vesicle formation[40], and corroborating the studies using the pNRC100 gene cluster[38]. A deletion in the pHH1 gvp C gene resulted in irregularly shaped gas vesicles, suggesting the involvement of GvpC in maintaining shape[41], in addition to assisting vesicle growth[38] and increasing stability (see below). Deletions of gvp D and E in both pHH1 and the Halobacterium chromosome resulted in partially Vac+ phenotypes [40, 41], confirming that these genes are not absolutely necessary for gas vesicle formation[38]. Interestingly, in the Haloferax mediterranei gene cluster, a small internal in-frame deletion of gvp D resulted in the overproduction of gas vesicles, suggesting that this gene may function in negative regulation of gas vesicle synthesis[39].

5 Transcription and regulation

Transcript mapping of the pNRC100 gas vesicle gene cluster indicated the presence of divergent promoters in the 200 bp gvp D–A intergenic region [25, 32]. The promoter sequences were of the archaeal type, with AT-rich box A sequences located about 25 nucleotides upstream of the putative transcription start sites. Most rightward transcripts terminated between gvp A and C but longer transcripts covering gvp A and C, and gvp A, C and N were also found[21]. A still longer transcript encoding gvp ACN(O) was reported in strain PHH1[40] but was absent in NRC-1[21]. In PHH1, two additional transcripts, one beginning at the start codon of gvp O and a second encoding the 3′ terminal half of gvp O were also reported[40]. In addition to leftward transcription starting in the gvp A–D intergenic region, a second leftward transcript starting near the 3′ end of the gvp E gene and covering gvp F-M was detected for the pHH1 gene cluster[40].

The gvp A transcript was found to be the most abundant of the gvp gene cluster mRNAs detected by Northern hybridization analysis. In Halobacterium NRC-1, gvp A transcription was induced in early exponential phase, and plateaued at mid-exponential phase[43]. Transcription was inhibited by aeration and the addition of DNA gyrase inhibitor to cultures, suggesting the involvement of DNA supercoiling in microaerobic activation of gvp A transcription[43]. In Haloferax mediterranei, the gvp A transcript level was found to be higher in media containing greater salinity[44]. However, since salt reduces the solubility of oxygen in brine, its effect may have been partly mediated via the changes in oxygen availability.

Several studies explored the possible regulatory roles of gvp D, gvp E and gvp O using Northern blotting analysis [3942]. A correlation between gvp D and gvp ACN(O) transcription was reported in Haloferax mediterranei and for the chromosomal genes in Halobacterium PHH1 [39, 40]. Moreover, in the absence of gvp D, gvp A transcription was elevated in early exponential phase (and gas vesicles were overproduced). In the pHH1 gene cluster, however, deletion of gvp D and E, together and separately, reduced transcription of gvp F-M and gas vesicle synthesis, but not gvp A transcription[40]. A role for pairing between the leader regions of the gvp DE and gvp F-M transcripts was proposed to account for the observations. In the PHH1 chromosomal gene cluster, a single mRNA species was observed for leftward transcription (gvp D–M) but higher levels of GvpE protein was detected in early growth phase compared to GvpD protein by Western blotting[42]. Several antisense transcripts were also detected from the gvp D, E, and F regions, and a role for these transcripts in differential regulation of the two genes was proposed. An additional finding was that the deletion of the chromosomal gvp O gene resulted in greatly decreased levels of rightward mRNAs[41]. Conversely, deletion of gvp N resulted in reduced gvp O mRNA. The interpretation of the above results is complex, and most likely reflects polar effects, copy number effects, and differences between the gene clusters studied. Although the evidence thus far is consistent with regulatory roles for gvp D and E, additional studies are necessary to clearly establish the regulatory mechanisms utilized.

6 Gas vesicle proteins

Biochemical analysis of gas vesicles has been limited by the difficulty in solubilization of the gas vesicle membrane. However, highly denaturing phenol-acetic acid-urea-polyacrylamide gels[11] have been used to identify the GvpA protein in Halobacterium by staining and Western blotting using rabbit sera produced against whole gas vesicles and LacZ-GvpA fusion proteins expressed in E. coli[21]. Detectable amounts of GvpB protein were also observed in gas vesicles purified from a gvp A mutant strain[12]. Analysis of the sequence of the gvp gene cluster indicated that gvp J and gvp M encode small, acidic polypeptides similar to GvpA (Table 1)[33]. Although the GvpJ and GvpM proteins have not yet been detected, it has been proposed that GvpA, GvpJ, and GvpM may constitute a small family of gas vesicle proteins which are used in different ratios for construction of gas vesicle membranes with varying geometry, e.g. cones or cylinders.

View this table:
1
Gene productApproximate molecular weight (×103)Isoelectric pointKnown or hypothetical functions
GvpA84.0major structural protein
GvpB84.0structural protein associated with minor cylindrical form
GvpC423.6structural protein involved in strengthening, assisting in growth, and determining shape of gas vesicles; contains internal repeats
GvpD594.2possible negative regulator; contains nucleotide-binding motif
GvpE214.1possible regulator
GvpF244.0possible minor structural protein or assembly protein
GvpG104.1possible minor structural protein or assembly protein
GvpH203.9possible minor structural protein or assembly protein
GvpI1610.8possible minor structural protein or assembly (nucleation) protein
GvpJ123.7likely minor structural protein
GvpK133.9possible minor structural protein or assembly protein
GvpL324.2possible minor structural protein or assembly protein
GvpM94.1possible minor structural protein
GvpN394.9possible minor structural protein or assembly protein, assisting in growth; contains nucleotide-binding motif
GvpO12–154.0–4.2possible regulator
  • Data from [3235].

The Halobacterium GvpC protein was observed in gas vesicles by Western blotting using antiserum against LacZ-GvpC fusion protein[21]. Studies using gas vesicles from the cyanobacterium Anabaena flos-aquae showed that GvpC may be released from gas vesicles under mildly denaturing conditions[20]. Absence of GvpC reduced the stability of gas vesicles to collapse while addition of GvpC restored stability, suggesting its involvement in strengthening gas vesicles by binding to the external surface of the membrane[20]. The GvpC proteins contain a motif which is repeated 4–5 times for the cyanobacterial proteins [24, 45] and 7–8 times for the halobacterial proteins [33, 34], suggesting that they bind to multiple copies of GvpA in the gas vesicle membrane (Table 1).

Western blotting analysis using antisera directed against gvp J, M, G, and K gene products produced in E. coli, did not detect the presence of these proteins in gas vesicles, although GvpG and GvpK were observed in Halobacterium lysates[40]. These results were consistent with the presence of very small quantities or absence of these Gvp proteins in gas vesicles. Both GvpG and K putative proteins are mainly hydrophilic, with hydrophobic N-terminal parts, suggesting that they could be minor structural components or chaperons for GvpA, J and/or M (Table 1). The predicted sequences of GvpD and GvpN suggested the presence of nucleotide binding motifs in both, consistent with a role in an energy requiring step in gas vesicle formation [32, 33]. GvpI was the only predicted basic protein, suggesting a possible function in nucleation of gas vesicle formation[33].

7 Conclusions and future studies

Genetic analysis of the gas vesicle gene cluster in several haloarchaeal species has resulted in a better understanding of this complex, unique, and interesting organelle. Analysis of natural mutants led to the discovery of the large gene clusters containing 13 or 14 genes, gvp MLKJIHGFEDACN(O). Based on mutagenic analysis of these gene clusters, the involvement of nearly all of the gvp genes in gas vesicle formation has been confirmed. The rightward operon was found to encode the major structural proteins, GvpA and GvpC, and GvpN, which have either structural or assembly roles. The leftward operon encoded likely regulatory proteins, GvpD and GvpE, and minor structural or assembly proteins, GvpF, GvpG, GvpI, GvpJ, GvpK, GvpL and possibly GvpH and GvpM. Consistent with the involvement of many gene products in gas vesicle formation, several gvp gene homologs have been found in other microorganisms, such as Calothrix PCC 7601 (gvp A and C)[24], Anabaena flos-aquae (gvp A, C, N, J, K, F and L)[5], and Bacillus megaterium (gvp A, N, F, G, L, and J) (M.C. Cannon and N. Li, personal communication). Future genetic and biochemical analyses of gvp genes and gene products in these and other microorganisms is likely to contribute to a fuller understanding of gas vesicle formation in both archaea and bacteria.

The occurrence of gas vesicles in both archaeal and bacterial microorganisms leads to some interesting evolutionary questions. The wide distribution of gas vesicles in nature and the location of their genes on plasmids suggest that the gvp genes may have been transferred among microorganisms by lateral gene transfer. The conservation of genetic organization of the gene cluster in haloarchaea, has raised the speculation that the cluster may have originated from a transposable element or perhaps even a phage. A survey of gvp genes in a wider diversity of microorganisms should provide insights into the evolution of gas vesicles. Finally, our recent finding that gas vesicle production can be programmed in E. coli by a suitably genetically engineered gvp gene cluster is also likely to help advance our understanding of these interesting organelles and lends itself to their use in biotechnology (DasSarma, S. and Halladay, J.T., unpublished).

Acknowledgements

The work in the authors’ laboratory is supported by the National Science Foundation (MCB-9221144 and MCB-9604443) and the Department of Energy (97ER623239) We acknowledge our present and former students and colleagues, Drs. J.T. Halladay, W.-L. Ng, C.-F. Yang, and V. Rangaswamy, Ms. F. Lin, E. Molinari, and F. Morshed, and Mr. J.G. Jones, D.C. Young, J. Perkel, S. Pyarajan, and S. Kennedy, for their experimental contributions, Ms. L.R.-S. Yin and Dr. D. Callahan for expert assistance in electron microscopy, Dr. M.C. Cannon and Mr. N. Li for communicating results prior to publication, and Drs. E.S. Stuart, N. Tandeau de Marsac, and S.H. Zinder for stimulating discussions.

Footnotes

  • 1 Halobacterium halobium, salinarium (salinarum), and cutirubrum have recently been reclassified as a single species. Since no consensus on the species designation exists at this time, we will use the genus and strain designations in this review, omitting the species designation for clarity.

  • 2 In the literature, the Halobacterium PHH1 plasmid and chromosomal gas vesicle genes are referred to as p-vac or p-gvp, and c-vac or c-gvp, respectively, while the Haloferax mediterranei genes are designated as mc-vac or mc-gvp.

References

View Abstract