Detecting Protein-Protein Interactions in the Intact Cell of Bacillus subtilis (ATCC 6633)

Abstract

The salt bridge, paired group-specific reagent cyanogen (ethanedinitrile; C₂N₂) converts naturally occurring pairs of functional groups into covalently linked products. Cyanogen readily permeates cell walls and membranes. When the paired groups are shared between associated proteins, isolation of the covalently linked proteins allows their identity to be assigned. Examination of organisms of known genome sequence permits identification of the linked proteins by mass spectrometric techniques applied to peptides derived from them. The cyanogen-linked proteins were isolated by polyacrylamide gel electrophoresis. Digestion of the isolated proteins with proteases of known specificity afforded sets of peptides that could be analyzed by mass spectrometry. These data were compared with those derived theoretically from the Swiss Protein Database by computer-based comparisons (Protein Prospector; http://prospector.ucsf.edu). Identification of associated proteins in the ribosome of Bacillus subtilis strain ATCC 6633 showed that there is an association homology with the association patterns of the ribosomal proteins of Haloarcula marismortui and Thermus thermophilus. In addition, other proteins involved in protein biosynthesis were shown to be associated with ribosomal proteins.

Determination of protein-protein interactions is fundamental to understanding the metabolism, life cycle, and regulatory aspects of a cell. A protein's function is defined to a great extent by its interaction with other proteins. The first step in the determination is the identification of which interactions occur. While this concept has long been known and the list of interactions grows steadily, recent developments in strategies to map the potentially myriad interactions that appear to occur underscore the complex nature of the problem (8, 20, 21, 24, 26). Much of this mapping is based on an extension of the yeast two-hybrid system (2, 17). A constant requirement has been specificity. Chemical cross-linking has been one of the methods used. One of these methods uses the only known reagent specific for naturally occurring paired functional groups in protein complexes. That reagent, cyanogen (ethanedinitrile; C₂N₂), appears to react only with paired groups, the most important of which are the salt bridges. These occur both within monomeric proteins (13, 14) and between subunits of oligomeric proteins and associating proteins of complexes (1, 3, 9-11, 27, 29, 30).

In the presence of cyanogen, salt bridges between associating protein molecules become covalently linked through a carbodiimide-like reaction. Hemoglobin with extensive intersubunit salt bridges is converted into a covalently linked tetramer by cyanogen in tens of seconds (9). Covalent links are introduced into monomeric proteins (13, 14). Cyanogen was used to covalently stabilize bacterial cell wall-synthesizing complexes, penicillin-binding proteins (PBPs), in a number of bacteria. Eight PBPs were covalently linked into two complexes in Escherichia coli (3), 10 PBPs were covalently linked into two complexes in Bacillus subtilis (27), and 7 PBPs were linked into two complexes in Haemophilus influenzae (1). Cyanogen is the only known reagent that can permeate the membrane of an intact, genetically unmodified cell and react with salt-bridged pairs of functional groups.

At room temperature, proteins and model peptides are modified by cyanogen on a time scale of seconds to minutes (9, 10). The side chains of aspartic acid, glutamic acid, arginine, lysine, and histidine are the primary targets of this reaction. The C-terminal carboxyl and N-terminal amino groups are also possible cyanogen modification sites. Carboxylates and basic groups that are not associated do not get linked. When applied to cells of known genome sequence and coupled with electrophoretic and mass spectrometric (MS) techniques, data can be generated that provide a basis for the facile identification of native associations. It should be noted that these techniques have a high level of reliability (18, 19).

MATERIALS AND METHODS

Bacterial growth and cyanogen treatment.

B. subtilis from a subtilis spore suspension (Difco Laboratories; ATCC 6633) was grown to an optical density of 0.45 at 600 nm. Aliquots (1.5 ml) were treated with cyanogen (1 atm) for 10 s, 30 s, 90 s, 2 min, and 5 min. An untreated aliquot provided the control. After treatment, the aliquots were spun down and washed three times with a solution of Tris-HCl (10 mM [pH 7.4], 1 mg of MgCl₂ · 6H₂O per ml). The precipitates were picked up in 0.15 ml of the same solution followed by the addition of 0.1 ml of hen's egg white lysozyme (1 mg/ml). The suspended cells were chilled at 0°C for 30 min and then sonicated six times for 30 s each with 30-s breaks (Fisher model FS6). The cellular debris was removed, and the supernatant was collected.

Cyanogen was obtained from Tex La Gases (Sulfur, La.). It is a toxic gas and should be handled with care.

Electrophoresis.

Polyacrylamide gel electrophoresis was carried out as described previously (30).

In-gel digestion.

Immediately following the gel staining and water washing, the protein bands of interest (∼34-kDa molecular mass region) were cut out of the gel and minced. One band was cut from the control, and one band each was cut from the cyanogen-treated bands. Acetonitrile (0.1 ml, 50% 25 mM NH₄HCO₃) was added to each sample, which was then vortexed for 10 min. The supernatant was discarded. This step was repeated three times. The gel pieces were dried in a Centri-Vap at 45°C for 20 min. Dithiothreitol (0.1 ml, with 10 or 25 mM NH₄HCO₃) was added to the dried gel pieces, and they were vortexed briefly. They were allowed to stand at 56°C for 1 h. The supernatant was removed, and iodoacetamide (0.1 ml with 50 or 25 mM NH₄HCO₃) was added to the gel pieces, which were then vortexed. These suspensions were allowed to stand at room temperature in the dark for 45 min. The supernatant was removed and discarded. The gel pieces were washed with aqueous NH₄HCO₃ (0.l ml, 25 mM) by vortexing for 10 min. The supernatant was discarded, and the gel pieces were dehydrated by treatment with acetonitrile (0.1 ml, 50% 25 mM NH₄HCO₃) two times. The gel pieces were dried as described above. To each gel, 0.1 ml of trypsin solution (13 mg/ml, 25 mM NH₄HCO₃) was added. Excess solution was removed, 0.1 ml of 25 mM NH₄HCO₃ was added, and the gel pieces were incubated at 37°C for 19 h.

Extraction of peptides for MALDI-MS analysis.

Matrix-assisted laser desorption mass spectrometry (MALDI-MS) analysis was performed as follows. After 19 h of tryptic digestion, water (0.1 ml) was added. Each sample was vortexed for 10 min and sonicated for 5 min. The digest solution from each sample was transferred to a centrifuge tube containing 5 μl of 50% acetonitrile-5% aqueous trifluoroacetic acid. To the remaining gel pieces, 0.05 ml of the same solution was added, followed by 10 min of vortexing and 5 min of sonication. This process was repeated twice, and all of the extracts from each were pooled, giving ∼0.5 ml. Each was concentrated to 0.025 ml for MS analysis.

MALDI-TOF MS analysis.

MALDI-time-of-flight (TOF) analyses were carried out with a Micromass QTOF-2 mass spectrometer as described previously (12, 29, 30).

HPLC ESI-Q-TOF MS (Micromass TOF SPEC 2E) analysis (25).

For high-performance liquid chromatography (HPLC) electrospray ionization quadrupole (ESI-Q)-TOF MS, three aliquots analyzed by MALDI-TOF MS (10 s, 90 s, and 5 min) were pooled and concentrated to ∼0.05 ml. Each was injected onto a Vydac C₁₈ column coupled to a Micromass ESI-Q-TOF2 mass spectrometer. For the chromatography, a gradient of 2 to 98% acetonitrile-0.09% aqueous formic acid and a flow rate of 0.2 ml/min were used. Fragmentation data were collected from the m/z values of interest by using an electron impact energy of 65 eV. All mass spectra were calibrated externally with a polyalanine standard (Sigma). Elution times correspond to peak fractions from the chromatogram.

Computer-based searching.

The MS-FIT program at Protein Prospector (http://prospector.ucsf.edu) was used for matching of MS data with information in the Swiss Protein Databank. The mass tolerance for the monoisotopic precursor ion was set at ± 250 ppm. The mass range was set to select only proteins with molecular masses of between 1 and 35 kDa.

RESULTS

Treatment of B. subtilis with cyanogen led to protein linking in a manner dependent on time of exposure to the coupling reagent. B. subtilis was grown to an optical density of 0.45 and treated with cyanogen for various periods. Electrophoresis revealed new protein bands not present in the untreated culture (Fig. 1). The most intense of these bands were at molecular masses of 34 and 80 kDa. The more intense 34-kDa band was excised from the gel, digested with trypsin, and analyzed by MS. Figures 1 and 2 show the MALDI-TOF spectra acquired from the 34-kDa band from the samples treated with cyanogen for10 s, 90 s, and 5 min. The m/z values are of quasimolecular ions (MH⁺) of tryptic peptides labeled with their corresponding database matches made through the MS-FIT program. To ensure that the m/z values are indeed B. subtilis proteins, the 34-kDa region of the electrophoretogram corresponding to the proteins from the cyanogen-untreated aliquot was analyzed. The m/z values in this control originate from acrylamide matrix, trypsin autodigestion, and instrument noise.

FIG. 1.

Electrophoretogram shown here is a photograph of the electrophoretogram of B. subtilis proteins used to separate the cyanogen cross-linked products. Lanes 1 and 9 are molecular mass markers. Lanes 2 and 3 represent both of the proteins from the untreated cells. Lanes 4 to 8 represent progressively longer C₂N₂ treatment times: 10 s, 20 s, 90 s, 2.5 min, and 5 min, respectively.

FIG. 2.

(a) MALDI-TOF mass spectrum acquired from an in-gel tryptic digestion of a protein band removed from the 34-kDa molecular mass region. (b) Fig. 1a rescaled. Each m/z value is labeled with its corresponding protein database match.

Aliquots from the samples yielding the spectra shown in Fig. 2 and 3 were combined and analyzed by reverse HPLC. Shown in Fig. 4, 5, and 6 are the ESI-Q-TOF spectra obtained from the species eluting at 13.52, 13.54, and 59.20 min, respectively. Each m/z value is labeled with its corresponding protein database match. The m/z of 977.1 in the bottom spectrum of Fig. 7 corresponds to a tryptic fragment from the 50S ribosomal protein L14. The amino acid sequence match of this doubly charged species (MH²⁺) is KGEVVKAVIVRTKSGARR. MS/MS (mass spectrum of selected ions from a previous mass spectrum) analysis on the ESI-Q-TOF machine confirmed the sequence adduced from the protein database. Other tryptic peptides arising from the L14 protein matched the m/z values 659.4 and 1,614.5.

FIG. 3.

(a) MALDI-TOF spectrum acquired from an in-gel tryptic digestion of a protein band removed from the 34-kDa molecular mass region taken over a more extended range than in Fig. 1. (b and c) Fig. 1a rescaled. Each m/z value is labeled with its corresponding protein database match.

FIG. 4.

ESI-Q-TOF mass spectrum of a tryptic digest of the protein eluting at 13.52 min. Each m/z value is labeled with its corresponding database match. The bottom spectrum is a rescaling of the top spectrum.

FIG. 5.

ESI-Q-TOF mass spectrum of a tryptic digest of the protein eluting at 33.64 min. Each m/z value is labeled with its corresponding database match. The bottom spectrum is a rescaling of the top spectrum.

FIG. 6.

ESI-Q-TOF mass spectrum of a tryptic digest of the protein eluting at 59.20 min. Each m/z value is labeled with its corresponding database match. The middle and bottom spectra are rescalings of the top one.

FIG. 7.

ESI-Q-TOF MS/MS spectrum of the species with the m/z = 977.1 (z = +2) in Fig. 5. A collision energy of 65 eV was used for this analysis.

Four protein families were found to contribute m/z values to the cyanogen cross-linked proteins. They were matched with values derived from the Swiss Protein Database. There were 78 m/z values matching ribosomal proteins, 10 matching RNA polymerase sigma factors, 16 matching transcriptional regulators, and 16 matching sporulation proteins. The preliminary results were obtained from the search of all of the mass spectral data shown in Fig. 2 to 7, where a minimum of three tryptic peptide matches were required. The number of ribosomal protein matches outnumbers those in the other three families. Assuming that we are involved mainly with ribosomal proteins, the final mass assignments from the data shown in Fig. 2 to 7 were based on the selection of the most accurate molecular masses compared with the theoretical ribosomal protein values from the Swiss Protein Database. Masses that could not be assigned to ribosomal proteins were found to match RNA polymerase sigma factors, transcriptional regulators, or sporulation proteins.

DISCUSSION

Six cross-linked proteins were identified from five bands excised from the 34-kDa region of the electrophoretogram. The proteins detected could not have migrated into this region unless they were covalently linked to other proteins due to their low molecular mass (4 to 26 kDa). Table 1 lists the predicted cross-linked species identified in part by their theoretical molecular mass, prior cross-linking studies, and the proximity of the coupled proteins to one another in the ribosome crystal structure from another species (21). It should be noted that the complexes yielded 65% of the total m/z values observed. It had previously been shown that the rates of reaction of different types of salt bridges varied by orders of magnitude (10); therefore, it was important to allow the slower-reacting salt bridges adequate time for reaction.

TABLE 1.

List of the predicted cyanogen cross-linked species that migrated to the 34-kDa molecular mass region of the electrophoretogram

Cross-linked proteinsa	Mol mass of cross-linked proteins (kDa)	Additional supporting evidence (reference)
L18-L4	35	Proximity in three-dimensional structure (20)
L14-L31-L19	35	Proximity in three-dimensional structure (20), cross-linking (28), L14 sequence data (this study)
S11-IF3	35	Cross-linking (4)
S4-S13	36	Cross-linking (15)
S12-L16	32	Cross-linking (6, 15, 16)
Pseudouridine synthase with L35 or L33	34	No prior knowledge

^aThe proteins are unique to cyanogen-treated B. subtilis.

In addition to low-molecular-mass ribosomal proteins cross-linked with one another, we detected the presence of two proteins that are not a part of the ribosome assembly, but interact closely with it during the cell cycle. They are initiation factor 3 (IF3) (4, 6) and pseudouridine synthase (uracil hydrolase) (5, 22, 23). We conclude that both are involved in the cyanogen cross-linking due to the large number of tryptic peptides detected. Fifteen proteins were obtained from IF3, and 11 were obtained from pseudouridine synthase.

Bacterial translation initiation requires the participation of three protein initiation factors: IF1, IF2, and IF3. Initiation factors bind to the 30S subunit and prevent the smaller subunit from binding the 50S subunit but facilitate its binding to mRNA. When the 30S subunit binds to mRNA, the initiation factors are released, and the 50S subunit can associate with the 30S subunit. IF3 is known to increase the selectivity of initiator tRNA both in vitro and in vivo (7).

Prior cross-linking studies of purified ribosomes have shown that IF3 and ribosomal protein S11 are in sufficiently close proximity to be cross-linked (4). A proposed model places a portion of one of the domains of IF3 between S7 and S11 (6). Tryptic peptides from S11 and IF3 were detected in the same 34-kDa protein band excised from the gel. Evidence from prior studies suggested that IF3 interacted with S11 (4, 6). Our data suggest that when B. subtilis was treated with cyanogen, IF3 and S11 became covalently linked. The linked IF3-S11 combination would have a molecular mass of 34 kDa, as observed.

Also detected in this study were high numbers of tryptic peptides from pseudouridine synthase, another nonribosomal protein found coupled to a ribosomal protein after B. subtilis was treated with cyanogen. This protein is a 26-kDa enzyme that catalyzes isomerization of uridine to pseudouridine. Pseudouridine is the most common single posttranscriptionally modified nucleoside in rRNA. While there is only one pseudouridine in the ribosome of B. subtilis, there are as many as 30 to 40 per ribosome in some eukaryotes (23). Enzymatic recognition of the correct uridine to be isomerized is not well understood. There have been no prior data pertaining to the specific ribosomal proteins with which pseudouridine synthase interacts during isomerization.

The detection of pseudouridine synthase, a 26-kDa protein, in the 35-kDa region of the electrophoretogram suggests that it was coupled with a protein with a molecular mass of approximately 8,674 kDa. Any ribosomal proteins matched against the database with molecular masses close to this value become candidates for the cyanogen-induced linkage. The matches were tryptic peptides found in this region from L33 (7,557 kDa) and L35 (7,249 kDa).

Six cross-linked proteins were isolated and identified after treatment of B. subtilis with cyanogen and analysis. As long as the genome sequence is known for the organism under study, the methodology presented here provides a way to detect protein-protein interactions in the intact cell.

In conclusion, use of the salt bridge-specific, paired group-specific reagent cyanogen allows examination of genetically unmodified cells for (i) unknown protein-protein interactions and (ii) comparison of patterns of interactions against those of defined systems, such as the ribosome (21). Extension of this facile methodology by the use of two-dimensional gels will undoubtedly allow identification of a much larger fraction of the cross-links introduced by cyanogen.