![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Applied and Environmental Microbiology, July 2005, p. 4004-4013, Vol. 71, No. 7
0099-2240/05/$08.00+0 doi:10.1128/AEM.71.7.4004-4013.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.
London School of Hygiene and Tropical Medicine, Keppel St., London WC1E 7HT, United Kingdom
Received 2 September 2004/ Accepted 24 January 2005
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although insertional mutagenesis is now routine, the introduction and expression of exogenous genes into Campylobacter cells remain a major problem, as many strains, including the sequenced strain NCTC 11168, are refractive to complementation and gene expression. Although some reports described shuttle vectors for the complementation of mutations in trans and for introduction and expression of a gfp gene, their use is often inefficient and is limited to certain strains (23, 24). For example we have been unable to introduce shuttle vectors pRY112 (34), pMW10 (32), pGUO0202 (1), and pMEK91 (24) into various C. jejuni strains, including the sequenced strain NCTC 11168 and its derivative 11168H.
In this study we describe an efficient procedure for gene delivery and expression, which overcomes shuttle vector-related limitations. The method has been successfully tested in a number of applications that have facilitated our understanding of this important pathogen.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Electroporation.
The electroporation procedure was carried out as follows. A 2-day bacterial plate culture was resuspended in buffer containing 272 mM sucrose and 15% glycerol at 0oC and washed three times with the same buffer, and 50-µl aliquots were used for each transformation. After addition of DNA (0.5 µg in 5 µl), the mixture was transferred to an ice-cold electroporation cuvette. Electroporation was performed at 2.5 kV, 200 
, and 25 µF, after which 100 µl of SOC buffer were added to the cuvette, and the bacterial suspension was transferred onto a nonselective blood agar plate. After overnight incubation at 37°C under microaerobic conditions, bacteria were spread onto a selective plate and incubated for a further 3 to 4 days.
Growth competition test.
The growth competition index (CI) was determined as described previously (8). Two-day agar cultures of the Campylobacter Camr derivative 11168H/pRRC4 and the wild-type recipient 11168H, grown with or without chloramphenicol (10 µg/ml), respectively, were inoculated into separate tissue culture flasks containing 10 ml of brucella broth without chloramphenicol. After incubation for a further 2 days on a rotary platform at 90 rpm under microaerobic conditions, the optical density at 600 nm was measured and serial dilutions were plated onto agar plates with (11168H/pRRC4) or without (11168H) chloramphenicol. A 100-µl aliquot of liquid culture of the Camr derivative (2.8 x 104 CFU) was mixed with an equal volume of wild-type strain 11168H (3.5 x 104 CFU) in a flask with fresh brucella broth (10 ml), which corresponded to an input Camr/Cams ratio of 1.25. After incubation for two days, serial dilutions were plated onto blood agar plates with and without chloramphenicol in triplicates. The colonies were counted, and the output Camr/Cams ratio was calculated. The CI was determined as output Camr/Cams ratio divided by input Camr/Cams ratio.
General cloning techniques.
Plasmids used in this study are listed in Table 1. Restriction enzymes were purchased from either Promega (Southampton, United Kingdom) or New England Biolabs (Hitchin, United Kingdom). T4 DNA ligase, T4 DNA polymerase, and calf thymus alkaline phosphatase were purchased from Promega (Southampton, United Kingdom). Oligonucleotides (Table 2) were from Sigma-Genosys (Pampisford, United Kingdom). Standard restriction digestion, DNA ligation, and cloning procedures were essentially as described elsewhere (26). T4 DNA polymerase treatment for generation of blunt ends was performed according to the manufacturer's protocol (Promega, Southampton, United Kingdom).
|
|
|
(ii) maf5 gene.
The maf5 gene was PCR amplified using primers ak244 M and ak245 M, and the product was digested with XbaI enzyme and cloned into XbaI-digested pRRC plasmid to produce pRMAF in such a way that the maf5 gene was transcribed in the same orientation as the Camr gene, which was verified via restriction analysis. The forward primer (ak244 M) contained a Shine-Dalgarno (SD) region optimized for Campylobacter (Fig. 1; Table 1).
(iii) Wild-type gfp gene.
Plasmid pMW10-13 (Table 1) containing the wild-type gfp gene was digested with XbaI and ClaI. The fragments were made blunt ended using T4 DNA polymerase and ligated to blunt-ended vector pRRC after digestion with XbaI (Fig. 1; Table 1). A recombinant clone containing pRRC with the wild-type gfp gene transcribed in the same direction as the Camr gene cassette was selected. A plasmid with the correct orientation of the wild-type gfp gene was selected via restriction analysis. Expression of GFP was confirmed via visualization of colonies with a fluorescence microscope.
(iv) egfp-SD gene.
Primers ak261 and ak262 were used to PCR amplify the enhanced GFP gene (egfp) from the pEGFP plasmid (Fig. 1; Table 1). The primer ak261 contained a Campylobacter SD sequence. The PCR product was digested with XbaI enzyme and cloned into vector pRRC so that the gene encoding the enhanced green fluorescent protein with the Campylobacter-specific SD sequence (egfp-SD) was under the control of Camr gene promoter as verified by restriction analysis.
(v) egfp-pMEK gene.
The EcoRI fragment of plasmid pMEK91 (24) containing the efgp gene (23) under the control of the PompE promoter of C. jejuni was blunt ended and inserted into XbaI-digested and blunt-ended vector pRRC (Fig. 1; Table 1). By using restriction analysis, a recombinant plasmid containing the gene encoding enhanced green fluorescent protein from plasmid pMEK91 (egfp-pMEK) under the control of two promoters, Pcam and PompE, was selected. E. coli colonies harboring this plasmid appeared yellow when visualized by eye under daylight and appeared bright green under the fluorescence microscope (excitation at 488 nm).
Construction of xerD mutant.
Plasmid cam71a9, constructed during the course of the C. jejuni genome sequencing project (25), contains a 1.5-kb DNA insert with a fragment of gene xerD (Cj0863). After insertion of a blunt-ended BamHI fragment of pJMK30 (29) containing a Kanr gene cassette into the unique SwaI site within the xerD gene on plasmid cam71a9 and transformation into 11168H, Kanr clones were selected. Insertion of the Kanr gene cassette in a nonpolar orientation was confirmed by PCR with primers ak238 and DL3 (Table 2).
Gel electrophoresis and Western blotting.
Bacteria were resuspended in sample buffer and incubated at 100°C for 10 min, and the lysate was analyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis in 12.5% precast polyacrylamide gels (Invitrogen, Paisley, United Kingdom). Gels were blotted onto a polyvinylidene difluoride membrane (Millipore, Watford, United Kingdom), blocked in phosphate-buffered saline containing 0.5% Tween 20 (PBST) for at least 30 min, and incubated with biotinylated soybean agglutinin (Vector Laboratories, Burlingame, CA) at a concentration of 10 to 20 µg ml–1 in PBST for 1 hour. Following three brief washes in PBST, blots were incubated with extravidin peroxidase (Sigma-Aldrich, Poole, United Kingdom) diluted 1 in 1,000 in PBST for 30 min. Following a further three brief washes in PBST, blots were developed using a diaminobenzidine staining kit with nickel enhancement according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA). Alternatively, the blots were probed with Penner 6 typing antiserum (1:100 dilution), followed by treatment with anti-rabbit immunoglobulin G peroxidase conjugate (Sigma-Aldrich, Poole, United Kingdom) at a 1:1000 dilution. All antibody dilutions were made using Tris-buffered saline containing 0.01% Tween 20 and supplemented with 1% bovine serum albumin (Sigma). The blots were developed using the diaminobenzidine staining kit with nickel enhancement according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA). Broad-range molecular weight standards were supplied by New England Biolabs (Hitchin, United Kingdom).
Confocal microscopy.
Bacterial cells were visualized using a Carl Zeiss LSM 510 confocal microscope according to the manufacturer's manual. The laser excitation wavelength was 488 nm, and the detection wavelength was 505 nm.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Construction of the delivery vector.
In order to minimize the disruption of rRNA genes, a spacer region between the 16S and 28S rRNA genes was selected. The genome of C. jejuni strain NCTC 11168 carries three identical clusters of rRNA genes. An XbaI site located immediately downstream from 16S rRNA genes was selected as an insertion site (Fig. 1A). A region containing this site with long flanking regions was PCR amplified and cloned into pGEM-T Easy vector to produce plasmid pRR (Fig. 1B). Insertion of the blunt-ended BamHI fragment containing the Camr gene into the blunt-ended XbaI site of pRR resulted in plasmid pRRC (Fig. 1C). A unique XbaI site of pRRC downstream from the Camr gene was used for cloning other genes, so that, if inserted in the correct orientation, these genes would be under the control of the constitutively expressed Camr gene promoter.
The Camr gene can be inserted into any of the three rRNA clusters.
The pRRC vector was transformed into C. jejuni strain 11168H via electroporation, and Camr colonies were selected. The transformants were analyzed using the Camr gene-specific PCR primer ak237 and three other primers (ak233, ak234, and ak235) corresponding to the regions adjacent to the three potential insertion sites on the chromosome (Fig. 1D). The results of the analysis demonstrated random recombination with any of the three rRNA clusters (Fig. 2).
|
The subclones of isolate A (Fig. 3) produced the same band patterns as the original strains, with one major band with the primer pair ak234/ak237 and minor bands of variable yields with the two other primer pairs. The presence of the minor bands suggested occasional migration of the Camr gene cassette from the primary insertion site into one of the two other rRNA clusters after subcloning.
|
In contrast, all subclones of isolate C produced strong bands with two primer pairs, ak234/ak235 and ak235/ak237, indicating that both the original isolate and its subclones contained the cassette integrated into two rRNA clusters on the same chromosome. The presence of minor bands with the ak233/ak237 primer pair also suggested migration of the cassette into another rRNA cluster (Fig. 1) in a subpopulation of bacterial cells, similarly to what was observed with derivatives A and B (see above). The authenticity of these minor bands to the respective rRNA cluster was confirmed by gel extraction followed by sequencing with Camr primer ak290 (Table 2) (data not shown).
Allelic replacement in C. jejuni is not affected by the product of the xerD gene.
Many strains of C. jejuni are capable of highly efficient double recombination events; the mechanism involved is unknown. One of the candidate genes that might be involved in such recombination was the xerD (Cj0863) gene. In E. coli XerD acts in concert with XerC and is involved in segregation of chromosome strands after replication (12). It was also shown that the XerC and XerD proteins of E. coli are required for filamentous phage integration (14).
The xerD mutant of 11168H was transformed via electroporation by plasmid pRRC, and Kanr clones were selected and allelic replacement confirmed by PCR. The mutant was then transformed with pRRC and, following selection of Camr Kanr clones, the chromosomal DNA was analyzed for the presence of Camr gene- and vector-specific sequences using PCR. The Camr gene was found to be inserted into an rRNA gene cluster near gene Cj0029 (Fig. 1D). However, no vector-derived sequences could be detected (data not shown), indicating that the allelic replacement does not require expression of the XerD protein.
Integration of the Camr gene cassette into the spacer region of the rRNA gene cluster does not affect growth in liquid culture.
Preliminary experiments demonstrated no difference in growth rates in liquid cultures (brucella broth) or in colony sizes on solid medium (blood agar plates) between the recipient strain 11168H and the Camr derivatives. For a more detailed comparative analysis of growth rates, in vitro competition studies were carried out as described elsewhere (8). One randomly selected Camr derivative, 11168H/pRRC4, with the Camr gene inserted into cluster III (Fig. 1) was grown in a mixture with the wild-type strain 11168H. The CI determined after 2 days of incubation (CI = 0.89) indicated that the insertion of the Camr gene cassette into the XbaI site of the rRNA cluster does not have a dramatic effect on cell viability under the conditions used. Similar results were obtained with the derivatives containing the Camr gene cassette inserted into clusters I and III (Fig. 1) and even with the derivative containing a Camr-gfp (wild-type gfp) fusion inserted into two clusters (II and III) simultaneously (data not shown).
Complementation of the pglH mutation in strain 11168H.
It was shown previously that insertional inactivation of the pglH gene affects protein glycosylation (21). In order to check whether the integration system described in this study can restore protein glycosylation in this mutant, the pglH gene was PCR amplified and inserted into the pRRC vector to produce pRPGL (Table 1). After transformation of the 11168H/pglH::Kanr mutant with plasmid pRPGL, Kanr Camr clones were selected. PCR analysis confirmed integration of the pglH gene into one of the rRNA gene clusters. Protein glycosylation in the derivatives was found to be completely restored (Fig. 4), indicating expression of a functionally active PglH protein.
|
Complementation of mutants in motility studies is especially important because of a large number of genes involved in flagellar formation and motility (33). Spontaneous mutations in these genes may also result in reduced motility in a specific mutant, thus skewing the effect of a primary mutation. In our previous study we overcame this problem by analyzing several independent clones of the maf5 mutation in various strains (15). However, a possibility of direct complementation of the mutation would be advantageous.
The maf5 gene was cloned into the pRRC vector, and the resulting plasmid pRMAF (Table 1) was transformed into the maf5 mutant described previously (15). The integration of the maf5 gene into an rRNA cluster was confirmed by PCR. The motility of the Kanr Camr derivatives was restored (although partially [see Discussion]) compared with that of the mutant (Fig. 5), indicating that the cells produced a functionally active Maf5 protein.
|
Expression of wild-type GFP.
The gene encoding wild-type GFP was PCR amplified and cloned into pRRC to produce pRGW (Table 1). Integration of the Camr-wild-type gfp cassette into various rRNA gene clusters was confirmed by PCR. No fluorescence of the Camr colonies could be detected with the fluorescence microscope (excitation at 488 nm). Similarly, only very weakly fluorescent cells could be seen using a confocal microscope. Importantly, this fluorescence was the same as that in the recipient strain 11168H, indicating some background fluorescence. Since both the clones and cells of the control strain E. coli/pRGW were strongly fluorescent, the lack of fluorescence in C. jejuni could be explained by a low level of expression of the wild-type GFP. This could be because in E. coli the wild-type gfp gene is present on a multicopy vector, whereas in C. jejuni it is present at only one copy per cell due to integration into the chromosome. In order to circumvent the problem, we tried expressing an enhanced protein (EGFP) in C. jejuni cells.
Expression of the egfp-SD gene.
The commercially available plasmid pEGFP was used as a source of the egfp gene (Table 1). Although both E. coli/pEGFP clones and cells were highly fluorescent, the expression vector had to be modified for expression in Campylobacter. The egfp gene was PCR amplified using a primer containing an SD sequence optimal for expression in Campylobacter, which resulted in generation of the vector pRED. The E. coli cells and clones containing pRED (Table 1) revealed bright fluorescence. After transformation into C. jejuni, the integration of the Camr-egfp-SD cluster into an rRNA cluster was confirmed by PCR as described above. Compared with wild-type gfp derivatives, the egfp-SD derivatives of C. jejuni revealed enhanced fluorescence under the confocal microscope. However, no fluorescence could be detected using a conventional fluorescence microscope. A relatively low fluorescence level in this case could be explained by a suboptimal codon usage in the egfp gene. Indeed, pEGFP was designed for expression in eukaryotic cells, particularly in human cell lines. Due to a low AT content in the Campylobacter genome, its codon preference is dramatically different from that in E. coli and human cells. Therefore, we decided to use the same technique for expression of the egfp gene from plasmid pMEK91, which as a codon content more optimal for C. jejuni.
Expression of EGFP from pMEK91.
The source of this type of egfp gene was plasmid pMEK91 (24). The plasmid contains a version of the egfp gene (egfp-pMEK) under the control of the Campylobacter ompE promoter (PompE) (Fig. 1). The EcoRI fragment containing the egfp-pMEK gene was cloned into pRRC, and fluorescence of E. coli transformants was tested. Two types of colonies with different levels of fluorescence were detected. The highest level of fluorescence corresponded to both the Camr and egfp-pMEK genes transcribed in the same direction. A plasmid from one such clone was used in transformation into C. jejuni, and integration of the Camr-egfp-pMEK cassette into an rRNA cluster was confirmed as described above. The colonies of the Campylobacter derivatives were fluorescent but were not as bright as E. coli colonies containing the delivery plasmid. However, very strong fluorescence of the C. jejuni cells could be detected using a confocal microscope. Therefore, a better codon composition and the use of an additional promoter allowed efficient expression of GFP in C. jejuni even when the gene was present as just a single copy per cell.
Use of the GFP derivative of C. jejuni for visualization of bacterial forms in situ.
The GFP derivative of C. jejuni strain 11168H (11168H/egfp-pMEK) was used for investigation of sessile bacteria attached to a coverslip in a stationary liquid culture. The sessile bacteria grew on glass coverslips immersed in flasks with stationary liquid cultures as indicated in Materials and Methods. Figure 6 demonstrates intense fluorescence of E. coli/pREM donor cells (Fig. 6A, panel 1) and 11168H/egfp-pMEK derivatives (Fig. 6A, panel 2). Similarly to the recipient 11168H cells (16), prolonged incubation (e.g., 4 days) of 11168H/egfp-pMEK on the solid medium resulted in conversion of spiral cells to coccoid forms (Fig. 6A, panel 3). Such a conversion was significantly reduced in bacteria attached to a coverslip in the stationary liquid culture (Fig. 6B, panels 1 to 3). The aggregates of sessile bacteria were found to contain extensive channels with highly motile darting rod- and spiral-shaped cells (Fig. 6B, panel 3).
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Despite routine use of allelic replacement in Campylobacter, at present little is known about its mechanism and the genes involved. One of the candidate genes that could be responsible for allelic replacement was xerC (Cj0863). Indeed, the gene shows extensive similarity with other genes involved in resolution of cointegrants (data not shown). In order to check the possibility that this gene is also involved in allelic replacement in C. jejuni, we used our integration system as a model in the investigation of recombination in the 11168H/xerC::Kanr mutant. Transformation of the pRRC delivery plasmid into this mutant resulted in insertion of the Camr gene cassette into different rRNA clusters. However, in all cases no vector-derived sequences could be detected, indicating double recombination and elimination of the vector sequences even in the xerC-negative strain. Therefore, genes other than xerC are likely to be responsible for allelic replacement.
The number of rRNA clusters in bacterial genomes varies significantly (18). The reason for such variation is not completely clear. There is no strong correlation between the rRNA gene copy number per genome and the bacterial growth rate. The reduction of the number of rRNA genes in E. coli, which is normally seven, has a moderate effect on the growth rate (2, 9). Overexpression of the rRNA gene may actually decrease the bacterial growth rate (28). A link between the rRNA gene copy number and ecological strategies of bacteria has been reported (17). Those authors showed that depending on environmental conditions, either a higher or a lower copy number of rRNA gene clusters may be preferable. It was suggested that the multiple copies of rRNA clusters may in some circumstances provide an advantage when a quick response to varying environmental conditions is required (9). However, for slowly growing bacteria, usually isolated from a low-nutrient environment, a low copy number of rRNA clusters is adequate (11). Our results indicated little or no effect of insertion of the Camr gene cassette on growth of C. jejuni. One should bear in mind, however, that our integration system was designed in such a way that only a minimal disruption (if any) of the rRNA cluster affected would occur.
Analysis of the insertion products revealed that the Camr gene cassette could be inserted randomly into any of the rRNA gene clusters. Interestingly, in some cases the PCR analysis indicated integration into two rRNA clusters simultaneously. This was not due to the presence of two different kinds of cells in the same sample, since subcloning resulted in colonies still producing PCR products with different primers. Simultaneous integration of the Camr gene into two rRNA clusters did not have a significant effect on the colony sizes or on the bacterial growth in a liquid culture. The additional minor PCR products produced by some derivatives suggest a possibility of migration of the Camr gene cassette between the rRNA clusters.
Various Campylobacter shuttle vectors have been described previously. However, these vectors have not been proven versatile. In our hands, none of the available shuttle vectors could be introduced into the sequenced strain C. jejuni strain NCTC 11168 or its derivative 11168H. In addition, even in a few cases of successful introduction of shuttle vectors into a Campylobacter strain as described by other authors, additional steps were required to overcome host-specific restriction. For example, in order to introduce pMW10-based vectors into C. jejuni RM1221 via electroporation, Miller and colleagues had to first transform these vectors into an Strr derivative of this strain via conjugation (23). Only plasmids extracted from C. jejuni could then transform the same strain of C. jejuni. However, strains carrying other restriction-modification systems would be refractory to acquiring this DNA. Another limitation of the shuttle vectors may be a requirement for the presence of residential plasmids in the recipient strains, which may be required for plasmid rescue (30). An additional disadvantage of the shuttle vectors used in other studies is the requirement for using a complementing gene with its own promoter (19, 22, 27). However, this is not always feasible, since a promoter is often located at quite a significant distance from a gene. In our study we overcame the shuttle vector-related limitations by putting an exogenous gene under the control of a constitutive Camr promoter, followed by integration of this cassette into a spacer region within an rRNA gene cluster.
The use of the rRNA gene cluster as a gene insertion target has a number of other advantages. First, due to redundancy of the rRNA genes in Campylobacter, even if a certain rRNA gene cluster is affected by insertion, it would result in only a minor effect on cell functioning. Indeed, we could detect no significant difference in growth rates even after simultaneous insertion of the Camr cassette into two different rRNA clusters. Second, due to the very high conservation of rRNA genes between Campylobacter spp., a recombination vector designed for one strain can be effectively used for most other Campylobacter strains. The vector contains long conserved flanking regions of DNA, ensuring high efficiency of recombination. Using a total DNA preparation extracted from an already-constructed derivative, an inserted gene can be transferred to other Campylobacter strains with even higher transformation efficiency.
In this study we demonstrated the complementation of two knockout mutants. However, while the complementation completely restored the glycosylation phenotype in the pglH mutant, the complementation of the maf5 mutation only partially restored motility. This could be due do tight requirements for the regulation, timing, and level of expression of the genes required for motility. However, even partial phenotype restoration confirms that the phenotype changes are really associated with a particular gene. Moreover, partial phenotype restoration after expressing a His tag fusion protein would provide a tool for isolating functional protein complexes, thus assisting functional analysis. We have demonstrated the same restoration of motility in a 11168H/maf5::Kanr mutant complemented with a His-tagged Maf5 protein (data not shown).
We also investigated expression of three genes of non-Campylobacter origin (gfp) in C. jejuni. Despite extensive differences among the three gfp genes, the derived amino acid sequences are almost identical. Fluorescence properties of the EGFP-SD and EGFP-pMEK proteins are enhanced compared with those of wild-type GFP due to the presence of two mutations, F64L and S65T. The EGFP-pMEK protein also contains a Q80R replacement compared with EGFP. E. coli cells containing either pRED (EGFP-SD) or pREM (EGFP-pMEK) were highly fluorescent. However, the fluorescence level of C. jejuni cells expressing EGFP-pMEK was much higher than that of cells expressing EGFP-SD, which is due to the presence of two tandem promoters and more optimal codon usage (Table 3) in the latter construct.
|
In summary, we demonstrated an efficient procedure of introduction and expression of various genes in Campylobacter cells. It is noteworthy that the method is applicable to the sequenced strain NCTC 111168, which is used by the majority of researchers. The fact that this strain was particularly difficult to complement was hindering functional genomics studies on this pathogen. This study therefore is a breakthrough for the research community. The method allows complementation of mutations and expression of other genes of non-Campylobacter origin, assisting better understanding of the physiology of this microorganism, including the mechanisms of survival of bacteria in the environment, which may be important for the design of intervention strategies to reduce the presence of Campylobacter in the food chain.
|
ACKNOWLEDGMENTS |
|---|
Financial support for this research was from the BBSRC and the Leverhulme Trust.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| J. Bacteriol. | Microbiol. Mol. Biol. Rev. | Eukaryot. Cell | All ASM Journals |
|---|
