![[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 
Journal of Bacteriology, June 2004, p. 3687-3694, Vol. 186, No. 12
0021-9193/04/$08.00+0 DOI: 10.1128/JB.186.12.3687-3694.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Department of Biochemistry, University of Missouri-Columbia, Columbia, Missouri 65211
Received 20 February 2004/ Accepted 11 March 2004
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The core signaling complex consists of a ternary complex of chemoreceptors, the chemotaxis-specific, autophosphorylating histidine kinase CheA, and an SH3-related coupling protein, CheW. This core complex is stable over durations relevant to sensory excitation and adaptation (13). Other Che proteins bind to the core complex components, with apparent dissociation constants in the micromolar range. Two response regulators, CheY and CheB, bind to the P2 domain of CheA (26) and interact with the P1 domain of CheA to capture a phosphoryl group carried on a histidinyl side chain. Phospho-CheY is inherently unstable, undergoing hydrolysis to release the phosphate with a half time of a few seconds (20). Hydrolysis is substantially enhanced by CheZ (20), a protein that binds to phospho-CheY (50). CheZ also binds to CheAS (46), the "short" form of kinase CheA that results from an alternative translational start site in E. coli approximately 90 codons interior to the start site that generates the "long" form, CheAL (24). CheAS lacks the phosphorylated histidine, but truncation results in a binding site for CheZ not available in the intact protein. Receptors are methylated and demethylated at specific methyl-accepting glutamates in reactions that are crucial to sensory adaptation and catalyzed by methyltransferase CheR and methylesterase-deamidase CheB. Each enzyme interacts at multiple substrate sites on each receptor monomer, the four to six methyl-accepting glutamyl residues in their unmodified and methylester-amide forms, respectively. In addition, both enzymes interact with a specific pentapeptide sequence present at the extreme carboxyl terminus of high-abundance receptors (5, 49). In E. coli, the high-abundance chemoreceptors Tsr and Tar are present at higher dosages than the three low-abundance receptors, Trg, Tap, and Aer, a difference commonly estimated as 
10-fold (17, 18, 45, 48).
To determine the cellular content and stoichiometry of the components of the signaling complex, we used immunoblotting to quantify amounts of chemoreceptors and Che proteins.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Purified proteins. Purified CheZ was a gift from Birgit Scharf (Universität Regensburg). Purification of CheA, CheW, and CheY followed the method described by Barnakov et al. (4). CheR and CheB were purified using a pentapeptide affinity column (5). Histidine-tagged receptors were purified using a Ni2+-nitrilotriacetic acid-agarose column (QIAGEN). A. N. Barnakov provided Trg-pd purified by solubilizing washed inclusion bodies with urea and renaturing by dialysis.
Concentrations of proteins used as immunoblotting standards were determined by quantitative amino acid analysis, using means of determinations for 7 to 10 amino acids, adjusted for the yield of the specific run determined by recovery of an internal standard, ornithine. Yields were 90 to 111% of an ornithine-alone sample. Standard deviations for mean values of different quantifications of a specific protein derived from determinations of individual amino acids were 3 to 10%, with the exception of 16% for Trg-6H. Concentrations of the protein standards were adjusted for the proportion of intact protein determined by densitometry of overloaded gels. This proportion was 
87% for Che proteins and 59 to 85% for chemoreceptors.
Antisera and antibodies.
Covance Research Products (Richmond, Calif.) raised polyclonal antisera to CheA, CheW, Tar-6H, and Trg-pd in rabbits, using purified proteins that we provided. Generous colleagues gave us mouse monoclonal anti-CheY (Birgit Scharf, Universität Regensburg), rabbit polyclonal anti-CheB and anti-CheR (Ann Stock, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School and Howard Hughes Medical Institute), and rabbit polyclonal anti-CheZ (Philip Matsumura, University of Illinois-Chicago). To reduce nonspecific background in immunoblots, some antisera were adsorbed with acetone powder (14) of a culture of RP3098 or, for anti-Trg-pd, of CP177 (30) by diluting antiserum 1:10 in 20 mM Tris-HCl (pH 7.5), 500 mM NaCl, adding acetone powder to 1% (wt/vol), incubating at room temperature 20 to 30 min, and spinning for 5 min at 16,000 x g in a table-top centrifuge. Secondary antibodies were goat anti-rabbit immunoglobulin G (heavy and light) horseradish peroxidase conjugate (Bio-Rad) diluted 1:2,000 or 1:4,000 for colorimetric or chemiluminescent detection, respectively, and goat anti-mouse immunoglobulin G (heavy and light) horseradish peroxidase conjugate (Molecular Probes, Eugene, Oreg.) diluted 1:3,000 from a 1-mg/ml stock for chemiluminescence detection.
Cell growth.
Cultures were inoculated to an optical density at 560 nm (OD560) of 0.05 (3.75 x 107 cells/ml) with highly motile cells grown in the same medium to an OD560 of 0.5 and grown at 35°C with good aeration. At OD560 = 0.5, cells were harvested by adding 0.2 ml of 50% (wt/wt) ice-cold trichloroacetic acid (TCA) to 1-ml samples, incubating 15 min on ice, centrifuging for 15 min at 16,000 x g in a table-top centrifuge at 4°C, adding 1 ml of ice-cold acetone to the pellet, centrifuging as before, air drying the pellet, and storing at –20°C. TCA precipitation was important to avoid proteolysis of chemoreceptors. In a control experiment, we compared the amount of protein detected in samples harvested by TCA precipitation to parallel samples harvested by centrifugation, washing, and solubilization of cells in sodium dodecyl sulfate (SDS) electrophoresis sample buffer. For Che proteins, values determined by direct sampling were within 9% of values determined by TCA precipitation. For Tsr, Tar, and Trg, direct sampling values were 25% lower than for TCA precipitation, presumably because of proteolysis during sample preparation. Thus, we used TCA precipitation for all samples. Tryptone broth (TB) contained 1% Difco Bacto-Tryptone (Becton Dickinson and Company, Sparks, Md.), and 0.4% NaCl, pH 7.0 (with NaOH), and H1 minimal salts medium (19) contained 50 mM potassium phosphate (pH 7.0), 7.6 mM (NH4)2SO4, 0.5 mM MgSO4, 1.25 µM Fe2(SO4)3, 0.01% vitamin B1, and 0.4% ribose and required 1 mM amino acids (histidine, leucine, and threonine for OW1 and those plus methionine for the other strains).
For each strain and growth condition, we determined the concentration of cellular protein at the optical density of harvest, OD560 = 0.5, by the bicinchoninic acid assay (Pierce, Rockford, Ill.) with bovine serum albumin as the standard, and the concentration of cells was determined by plating dilutions on Luria-Bertani agar plates. Protein concentrations, in micrograms of protein per milliliter of culture, were as follows: RP437 in TB, 150; RP437 in H1, 120; OW1 in TB, 140; OW1 in H1, 125. Since the amount of protein is a measure of cell volume (10), these values indicated only modest differences in cell volume per milliliter as strain and growth conditions varied. However, cell concentrations expressed as 108 cells/ml were as follows: RP437 in TB, 2.9; RP437 in H1, 5.35; OW1 in TB, 4.1; OW1 in H1, 2.9. These findings implied a greater percent variation in cell size. This was confirmed under a microscope: RP437 cells were significantly smaller in H1 cultures than in TB cultures, and OW1 were smaller in TB cultures than in H1 cultures. Thus, the most informative parameter was not molecules per cell, but molecules per cell volume, measured by determining total cellular protein. Yet, expressing cellular amounts in units of molecules per total cellular protein would not have provided ready comparison to literature values of molecules per cell. Thus, we defined the TB-grown RP437 value, 0.51 pg of total protein/cell, as a standard cell volume and have expressed cellular contents as molecules per standard cell volume.
Quantitative immunoblotting. SDS electrophoresis sample buffer was added to tubes containing TCA-precipitated cells and boiled for 5 min. Solubilized samples were submitted to SDS-polyacrylamide gel electrophoresis and immunoblotting, and the intensity of bands was converted to the protein amount by using a calibration curve of pure protein present on each blot. Each 0.5-mm gel was loaded with nine different dilutions of a protein standard (usually comprising a ninefold concentration range in equal steps), three serial dilutions of three experimental samples (usually representing independent cultures of the same strain in the same growth medium and adjusted after pilot runs for amounts that would fall within the range of the standard), and a negative control (a sample of cells lacking the protein of interest at a dilution equivalent to the most concentrated of experimental samples). Both standards and experimental samples were adjusted to contain the same amount of total cellular protein by addition of solubilized, TCA-precipitated RP3098 or CP177.
Each protein was quantified using gels with a polyacrylamide concentration that provided a well-focused band, as follows: CheA and chemoreceptors, 12%; CheB, CheR, and CheZ, 13%; CheW, 14%; CheY, a gradient from 12 to 18%. Proteins were transferred electrophoretically from gels to 0.45-µm-pore-size nitrocellulose (Hybond ECL; Amersham Pharmacia Biotech) for 2 h at 12 V. Nitrocellulose sheets were immersed in 2.5% instant nonfat powdered milk in 20 mM Tris (pH 7.0) and 0.5 M NaCl (PM-TBS) and placed on a rocking platform at room temperature for 1 h, immersed in a dilution of primary antiserum in PM-TBS, placed on a rocking platform at room temperature overnight, washed with TBS-Tween (0.05% [vol/vol]) three times for 5 min (colorimetric detection) or four times for 15 min (chemiluminescent detection), incubated for 2 h with a secondary antibody, washed as above, and immersed for 30 min in a solution of 4-chloro-1-naphthol created by dissolving 15 mg in 5 ml of methanol, mixing with 25 ml of TBS, and adding 15 µl of 30% H2O2 immediately before use (colorimetric detection) or treated with ECL Western blotting detection reagents (Amersham Pharmacia Biotech) according to the manufacturer's protocol and exposed to Kodak scientific imaging film (chemiluminescent detection). CheA, CheW, and chemoreceptors were quantified by colorimetric detection, CheB, CheR, and CheY were quantified by chemiluminescent detection (ECL Western blotting detection reagents; Amersham Pharmacia Biotech), and CheZ was quantified by both. Colorimetric and chemiluminescent images were captured by using a digital camera, and intensities were quantified using TotalLab software (version 2.01; Nonlinear Dynamics Ltd.). CheAL and CheAS were quantified separately, and the values were summed to determine total CheA. Antisera varied in avidity and degree of specificity, and thus the following dilutions, preabsorptions with acetone powders, and amounts of protein standards were used: anti-CheA, 1:5,000, 2 preabsorptions, 0.1 to 0.9 ng standard; anti-CheB, 1:500, 3, 0.1 to 0.9 ng; anti-CheR, 1:500, 3, 0.2 to 1.8 ng; anti-CheW, 1:50, 0, 1 to 9 ng; anti-CheY, 1:1,250 (0.01 mg/ml), 0, 0.5 to 4.5 ng; anti-CheZ, 1:250 (colorimetric) or 1:500 (chemiluminescent), 0, 0.5 to 4.5 ng (colorimetric) or 0.1 to 0.9 ng (chemiluminescent); anti-Tar, 1:2,500, 0, 0.5 to 4.5 ng (Tar-6H or Tsr-6H); anti-Trg-pd, 1:40, 5, 1 to 8 ng Trg-6H.
High-abundance receptors Tsr and Tar were quantified together (Fig. 1C), using an anti-Tar serum that reacted essentially equally with the two proteins (Fig. 1E). The low-abundance receptor Trg was quantified (Fig. 1D) with a Trg-specific antiserum raised to its periplasmic domain, the sequence of which is essentially unrelated to other receptors (7). With the aim of quantifying Tsr and Tar individually, we used RP2867, a derivative of RP437 lacking CheR and CheB (32) and thus producing only one electrophoretic form of each receptor (Fig. 1E). However, in this strain Tsr/Tar ratios were close to 1:1 (56:44 in TB and 52:48 in H1), values inconsistent with many published observations indicating 2:1 ratios in wild-type cells (e.g., references 11, 16, 39, and 42). Also, the cellular contents of high-abundance receptors, CheA and CheW in RP2867, were all 25 to 50% lower than in the parent RP437 (see Fig. 2). We surmise that insertion and excision of a lambda transducing phage that created RP2867 from RP437 (32) altered expression of chemosensory genes and may well have altered the cellular ratio of Tsr to Tar. Thus, we have not used Tsr/Tar ratios determined in RP2867 to deduce the ratio in wild-type cells.
|
|
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
To quantify each Che protein, we performed separate immunoblot assays specific for that protein. To quantify the sum of Tsr and Tar, we used an antiserum that was raised to Tar but recognized the two high-abundance chemoreceptors with essentially the same efficiency (Fig. 1E). Under immunoblotting conditions that yielded a linear relation between intensity and amount of the high-abundance receptors, low-abundance receptors were not detected. However, it was possible to quantify specifically the low-abundance receptor Trg in samples containing all chemoreceptors by using a Trg-specific antiserum, raised to the periplasmic domain of that receptor, that did not react with the other receptors (Fig. 1D). Receptor-specific antisera were not available for the other low-abundance receptors, Tap and Aer. However, a measurement for Aer cited in a review (45) and the intensity on a two-dimensional gel of radiolabeled protein likely to be Tap (18) indicate that these two receptors are present at levels no higher, and likely lower, than that of Trg. To include these low-abundance receptors into our values for cellular content of chemoreceptors, we estimated that the content of the two together would be approximately the content of Trg. Values for total chemoreceptors are thus the sum of experimental values for total high-abundance receptors and Trg plus the estimate for Tap plus Aer.
Cell size varied almost twofold as a function of strain and growth condition. To make comparisons among different strains and growth conditions that would reflect the concentrations of the sensory components and be independent of differences in cell size, we expressed cellular contents in units of molecules per standard cell volume, defining TB-grown RP437 as the standard cell (see Materials and Methods).
Cellular content as a function of strain and growth condition. The cellular content of chemoreceptors and Che proteins varied as much as 10-fold as a function of strain and growth medium (Table 1 and Fig. 2). However, these variations occurred in concert for all components (Fig. 2) and, thus, hardly changed the relative amounts of the proteins. This is illustrated by pair-wise ratios (Fig. 3). These ratios were relatively constant as a function of strain and growth condition, with standard deviations of 4 to 31%. These standard deviations were comparable to or only modestly larger than those observed for protein ratios determined for independent cultures of the same strain grown under the same condition. Thus, within the limits of our assays, ratios varied little or not at all as amounts of components varied as much as 10-fold. This indicates a rather constant cellular stoichiometry among the proteins of the chemosensory system.
|
2.5-fold difference in content of the two proteins. Thus, we were concerned when our values for CheY and CheZ in TB-grown RP437 were approximately threefold higher than theirs, even though we observed essentially the same ratio between these two proteins. However, Scharf et al. had analyzed cells growing in TB at a density equivalent to 1.5 additional generations, near the end of the exponential phase of growth in that medium. When we performed quantifications of CheY and CheZ in RP437 at that cell density, we obtained essentially the same values as Scharf et al. Thus, modest differences in growth phase can have significant effects on cellular content of chemotaxis components, as previously documented by Wang and Matsumura for CheAL and CheAS (47). |
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The two strains we characterized are wild type for chemotaxis, exhibiting effective responses to many compounds when grown in rich or minimal media. Thus, the chemosensory system functions effectively over at least a 10-fold range of cellular dosage. This situation reflects robustness in the bacterial chemosensory system (3). Importantly, it also provides a test for mathematical models of chemotaxis. In addition, maintenance of relative amounts, as strain and growth medium vary, must involve yet-unstudied mechanisms for maintaining relative expression in four different operons and for relative protein amounts that vary from 1:1 to 50:1.
As described in the introduction, independently determined values for cellular content of specific chemotaxis components published previously have differed by as much as 10-fold, and the origin of this variation has been unclear. Since our data demonstrate that the cellular content of each component can vary as much as 10-fold as a function of strain, growth medium, and growth phase, variation in published values could well reflect one or more of these factors.
Cellular ratios and the stoichiometry of the core signaling complex. We observed cellular ratios of 2.9 ± 0.7 receptor monomers per CheA monomer, 2.4 ± 0.7 receptor monomers per CheW, and 1.2 ± 0.3 CheW per CheA monomer. Both chemoreceptors and CheA are native dimers, whereas CheW is considered a monomer, so these values suggest interaction of three receptor dimers, one CheA dimer and two CheW monomers. This is tantalizing, because X-ray crystallography showed that a fragment of much of the cytoplasmic domain of chemoreceptor Tsr formed trimers of dimers in vitro (23), and biochemical studies have provided evidence that intact chemoreceptors form trimers in vivo (44). In addition, in vitro reconstitution studies of core signaling complexes have observed maximal kinase activation at ratios of receptors or receptor fragments to CheA and CheW significantly greater than one receptor to one kinase (2, 25, 27). The ratio of receptors to CheA required for effective activation in those in vitro studies was higher than the stoichiometry we observed in vivo, perhaps because of reduced activities for isolated proteins that were used or because receptor participation in core complexes is not identical to kinase activation.
In any case, extrapolation of cellular stoichiometry to stoichiometry of the ternary core signaling complex in vivo requires knowing the proportion of each protein in a cell that is in such complexes. Values have been provided by immunogold studies of RP437 cells, which identified 79% of CheA and 53% of CheW, both water-soluble proteins, as membrane associated in cells containing all three components of the core complex (28). The CheW value may be artificially low because of the small number of particles in analyses using anti-CheW (J. Maddock, personal communication). CheA, CheW, and receptors appear in clusters, and effective clustering requires all three (28), implying that clustering is linked to formation of ternary complexes. An assessment of chemoreceptor clustering in wild-type cells, using original and subsequent data, indicates that at least 93% of receptors are in clusters (J. Maddock, personal communication) and, thus, participants in ternary complexes. We used the cellular stoichiometry we determined and the values for participation in signaling complexes to calculate a deduced in vivo stoichiometry of core signaling complexes. Expressed per CheA dimer, the stoichiometry was 1 CheA dimer per 1.6 ± 0.5 CheW monomers and 3.4 ± 0.8 receptor dimers. Like the cellular stoichiometry, these values are within 1 standard deviation of an integral stoichiometry of three receptor dimers, 1 CheA dimer (or two CheA monomers from different dimers [36]), and two CheW monomers. However, in the cell, stoichiometries may not be integral. For instance, if the components of the core complex form an extended array (36, 43), not every trimer of receptor dimers would necessarily associate with exactly two monomers of CheA and of CheW. A finding of 3.4 receptor dimers per CheA dimer could reflect an array in which 13% of receptor trimers interacted with one monomer of a CheA dimer instead of two, perhaps at borders of arrays in which CheA dimers bridged trimers (36). A finding of 1.6 CheW per CheA dimer could reflect an array in which 40% of CheA dimer-receptor trimer units bound 1 CheW monomer, not 2. Alternatively, multiple dynamic equilibria across a signaling array could result in nonintegral steady-state stoichiometries.
Low-abundance chemoreceptor Trg was present in cells at a ratio to total receptors of 0.036 ± 0.007, a value close to the early estimate that Trg was present at 1/10 the level of a high-abundance receptor (16). If different receptor dimers have equal probabilities of being in a trimer, then 10% of signaling complexes would contain Trg.
Loosely bound components of the signaling complex. Most clusters of core signaling components are at cell poles (28). CheY, CheZ, and CheR, loosely bound components of signaling complexes, are also localized there, dependent on presence of intact core components (9, 37, 40). The localization of CheY is consistent with persistent interaction with core complexes in plasmon resonance studies (35). The notion of molecular brachiation (25) implies that CheB, as well as CheR, could be retained in extended receptor clusters. Thus, cellular stoichiometries are relevant to steady-state interactions of loosely bound components in the signaling complex.
In vivo ratios of 2.9 ± 0.9 receptors per CheY, 1.2 ± 0.2 CheY per CheA, and 0.9 ± 0.2 CheY per CheW indicated that essentially all 70% of cellular CheY that is not phosphorylated (1) could be bound to P2 domains of CheA contained in signaling complexes and, conversely, most signaling complexes would be occupied with at least one unphosphorylated CheY. As we varied the strain and growth condition, the ratio of CheAL to CheAS was notably constant at 2.0 ± 0.2. However, this ratio has been observed to vary fourfold as a function of growth phase (47). In any case, at the growth phase we tested, cells contained two molecules of CheAL for every molecule of CheAS, even though the absolute amounts of the two components varied close to 10-fold. If probabilities of heterologous and homologous dimerization of CheAL and CheAS were equal, as might be expected since the dimerization domain is structurally distinct (6) from the P1 region truncated in CheAS, then 45% of core signaling units would have only CheAL, 44% would have both CheAL and CheAS, and 11% would have only CheAS. Thus, 11% of core signaling units could not donate phosphoryl groups to CheY and CheB. There would be CheAS binding sites for CheZ in approximately half the core signaling units that could donate phosphoryl groups to CheY: the 44% that contained both CheAL and CheAS, but not the 45% containing only CheAL. The ratio of CheZ monomer to CheAS was 1.4 ± 0.3, corresponding to a ratio of 0.7 for native CheZ dimer to CheAS. Thus, there would be sufficient sites for all CheZ to interact with signaling complexes, a situation consistent with localization of a CheZ-green fluorescent protein fusion to cell poles in the presence of signaling complexes but not in their absence (9, 40).
We found that methyltransferase CheR and methylesterase-deamidase CheB were present in the cell at lower levels than their protein substrates, a common cellular relationship of enzyme to substrate. Ratios of chemoreceptor monomers to CheR and CheB were 120 ± 30 and 80 ± 15, respectively. Since there are four and six potential methylation and demethylation sites on each monomer of high-abundance receptors Tar and Tsr, respectively (21, 33), the ratio of cellular substrate to enzyme is several hundred to one, depending on the adaptational state of the receptor population. Considering stoichiometry in terms of core signaling complexes, cells would contain 1 CheR molecule for every 20 signaling complexes and 1 CheB for every 13. Previously published values for cellular content of the two enzymes, determined independently, had implied that CheB was present at substantially higher levels than CheR (www.anat.cam.ac.uk/comp-cell). This was consistently not the case for the strains and conditions we investigated.
Conclusions. It is striking that variation as large as 10-fold in cellular content of the components of the chemotaxis signaling complex did not substantially alter cellular stoichiometry. It is equally striking that the cellular stoichiometry of the core signaling complex suggests an organization based on trimers of receptor dimers. Our data imply that signaling clusters are built from core units of three receptor dimers, two monomers of the dimeric kinase, and two monomers of CheW. Figure 4 provides a cartoon of the complete signaling complex, labeled with the specific stoichiometry deduced from our studies. The ratios of CheY and CheZ to components of the core complex mean there are sufficient copies of the two proteins to interact with all (CheY) or half (CheZ) of the core complexes. These substantial presences in complexes could be important in sensory mechanisms and ought to be considered experimentally and theoretically. One CheB for every 80 receptor monomers and approximately every 13 signaling complexes means that the role of CheB in signal amplification (22, 41), which is not yet understood, cannot involve mechanisms that require close to stoichiometric interaction with the population of receptors or signaling complexes. Although there is no unique set of cellular amounts for the chemotaxis components of E. coli, it is useful to define representative values. If a cell contained 7,000 CheA monomers, the average ratios we determined (Table 1; Fig. 3) suggest a cellular content, expressed as monomers, of 7,000 CheA, 8,400 CheW, 20,000 receptors, 8,400 CheY, 3,300 CheZ, 280 CheB, and 180 CheR.
|
|
ACKNOWLEDGMENTS |
|---|
This work was supported by GM29963 from NIGMS.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Appl. Environ. Microbiol. | Infect. Immun. | Eukaryot. Cell |
|---|---|---|
| Mol. Cell. Biol. | J. Virol. | Microbiol. Mol. Biol. Rev. |
| ALL ASM JOURNALS |
