Proteome-Based Comparative Analyses of Growth Stages Reveal New Cell Cycle-Dependent Functions in the Predatory Bacterium Bdellovibrio bacteriovorus

Abstract

Bdellovibrio and like organisms are obligate predators of bacteria that are ubiquitously found in the environment. Most exhibit a peculiar dimorphic life cycle during which free-swimming attack-phase (AP) cells search for and invade bacterial prey cells. The invader develops in the prey as a filamentous polynucleoid-containing cell that finally splits into progeny cells. Therapeutic and biocontrol applications of Bdellovibrio in human and animal health and plant health, respectively, have been proposed, but more knowledge of this peculiar cell cycle is needed to develop such applications. A proteomic approach was applied to study cell cycle-dependent expression of the Bdellovibrio bacteriovorus proteome in synchronous cultures of a facultative host-independent (HI) strain able to grow in the absence of prey. Results from two-dimensional gel electrophoresis, mass spectrometry, and temporal expression of selected genes in predicted operons were analyzed. In total, about 21% of the in silico predicted proteome was covered. One hundred ninety-six proteins were identified, including 63 hitherto unknown proteins and 140 life stage-dependent spots. Of those, 47 were differentially expressed, including chemotaxis, attachment, growth- and replication-related, cell wall, and regulatory proteins. Novel cell cycle-dependent adhesion, gliding, mechanosensing, signaling, and hydrolytic functions were assigned. The HI model was further studied by comparing HI and wild-type AP cells, revealing that proteins involved in DNA replication and signaling were deregulated in the former. A complementary analysis of the secreted proteome identified 59 polypeptides, including cell contact proteins and hydrolytic enzymes specific to predatory bacteria.

Bdellovibrio and like organisms (BALOs) are gram-negative bacteria that couple predation upon other gram-negative cells with cell growth and replication and are therefore obligate predators (32). They are common in the environment and are phylogenetically diverse (14). BALOs exhibit a dimorphic cell cycle that has been divided into seven separate event-based periods based on microscopy (I to VII) (40). The Bdellovibrio bacteriovorus life cycle is composed of a free-living, small and highly motile attack-phase (AP) (event I) cell possessing a long sheathed flagellum (64). This phase ends with attachment to a prey cell (event II). A few strains perform predation attached to the outer membrane of their prey and divide by binary fission (32, 36). However, most penetrate (event III) and settle within the prey's periplasm, forming a bdelloplast (event IV). During events I to IV the predatory cell remains morphologically unaltered besides the usual loss of flagellum upon entry (59). Thereafter, a growth phase becomes apparent, during which the predatory cell elongates and its DNA is replicated multiple times (event V). Septa and flagella are finally synthesized by the filamentous cell, to generate a number of progeny AP cells (event VI) that lyse the remains of the prey cell wall (event VII) (35, 40). Soon after a successful BALO attack, the prey's metabolism is abolished, and the prey's content is degraded in a highly concerted manner and used to build the growing predatory cell (54). Their peculiar life cycles make the BALOs interesting models for the study of predator-prey interactions, parasitism, and cell differentiation (40; for detailed reviews, see reference 32). The ability of BALOs to prey on other gram-negative bacteria including many pathogens, their presence in animal and human guts, and their aptitude for preying on biofilms make them potential therapeutic agents for the control of human, animal, and plant pathogens (18, 33, 58, 60, 61). Skillful applications require a better understanding of the BALOs' cell cycle and of predatory functions.

The species Bdellovibrio bacteriovorus is the most studied of the BALOs, and the genome of the type strain 100^T has been sequenced (52). The genome of strain 100^T encodes many potential lytic enzymes (estimated as 293 proteins), such as lipases, glycanases, and proteases/peptidases (52). Nevertheless, the mechanism of Bdellovibrio predation is still shrouded in mystery, although some required traits have been elucidated by molecular (4, 17, 40, 57) and physiological (53) studies. Accordingly, the cellular processes driving the peculiar cell cycle of B. bacteriovorus have not been investigated.

Although BALOs are obligate predators, host-independent (HI) mutants able to grow in the absence of prey in a rich medium can be isolated in the laboratory (5). Strikingly, the cell cycle of HI strains still shows the dimorphic pattern of wild-type (wt) BALOs, with an AP followed by filamentous growth, division, and differentiation. These mutant strains maintain their predatory capabilities when regularly grown in the presence of prey. Thus, the use of HI strains is appealing since they provide a relatively simple model, alleviating the complexity of two-membered cultures. Such model strains are especially valuable for proteomic analyses, as they avoid the complications and the interference stemming from the presence of prey proteins and of their degradation products.

In this study we investigated the developmental sequence expressed during the life cycle of B. bacteriovorus, as reflected by its proteome. We present a differential proteomic analysis of synchronized cultures of an HI mutant that spans the whole cell cycle. This analysis is substantiated by a differential examination of wt and HI AP cells and includes a study of the secreted proteins. In addition, we present data on the expression patterns of selected individual and cotranslated genes. Based on temporal, spatial, and differential expression data, we assign functional roles to hitherto hypothetical proteins and link peculiar functions to the dimorphic, predatory cell cycle. These data also underline the relevance of HI mutants as models for the study of obligate predators while at the same time revealing underlying differences between the wt and HI phenotype.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

wt B. bacteriovorus 109J (a classical strain for Bdellovibrio studies [53]) was grown at 30°C in HEPES buffer (25 mM HEPES, 2 mM CaCl₂·2H₂O, 3 mM MgCl₂·6H₂O, pH 7.8) in two-membered cultures with Escherichia coli ML35 as a prey. The HI strain HI-6 is a spontaneous mutant able to grow axenically in a rich medium such as PYE (1% Bacto peptone, 0.3% yeast extract, 2 mM CaCl₂·2H₂O, and 3 mM MgCl₂·6H₂O [pH 7.6]) (5).

Fresh AP cells of the wt and HI-6 strains (about 10⁷/ml) were inoculated from overnight cultures with 2 × 10⁹ CFU/ml of E. coli ML35 prey. AP cells were separated from residual prey and debris on a Percoll-sucrose cushion by ultracentrifugation (4).

Synchronization of HI cultures.

Synchronized growth of the HI-6 strain was obtained as in the work of Gray and Ruby (23) with modifications. A loopful of bacteria from an overnight growth on a PYE plate was inoculated into liquid PYE. After two consecutive overnight subcultures (1:10 dilution in fresh PYE), the culture was amended with 0.5 mg/ml of an autoclaved E. coli cell extract. This resulted in growth synchronization: cells started to elongate, reached similar sizes, and divided into progeny simultaneously. Synchronous growth lasted for about 11 h. Three samples were taken at approximately 4 (T1, growth initiation), 8 (T2, midgrowth) and 10 (T3, septation, just before fragmentation) h after the initiation of growth, corresponding to optical density values at 570 nm (OD₅₇₀) of 0.4, 1.2, and 2.1, respectively. These samples are defined as stages of the growth phase (GP). In this work, T1, T2, and T3 in HI cultures correspond to events IV to V, V, and VI to VII in a wt culture, respectively.

Synchronization of prey-dependent cultures.

Synchronization of wt B. bacteriovorus 109J cultures was obtained by mixing 2 × 10⁹ CFU/ml of an E. coli ML35 prey in exponential phase with 10¹⁰ PFU/ml of the predator in HEPES buffer (31). Under these conditions prey cells are attacked and penetrated synchronously as confirmed by phase microscopy. This stage is followed by synchronous periplasmic growth of the invading cells. Cultures were sampled at 1, 2, and 3 h after mixing. A full cycle lasted for about 3.5 h, culminating with the release of progeny cells from lysed bdelloplasts. Bdelloplasts were harvested on a Percoll-sucrose cushion as described above.

Protein extraction.

The extracellular protein fraction of an overnight HI-6 culture grown in PYE was centrifuged (12,000 × g, 15 min), filtered (0.45-μm-pore-size filtration followed by an 0.2-μm-pore-size filtration), and concentrated (10-kDa cutoff; Vivascience). Proteins were precipitated in 15% trichloroacetic acid, washed with cold acetone, and dissolved in 8 M urea, 2% CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}, 0.5% isopropylthiogalactopyranoside, 0.28% dithiothreitol. This protein fraction was run on two-dimensional gel electrophoresis (2DE) gels (see below). About 100 of the resulting gel spots were systematically grouped according to their molecular weight and intensity and then analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) (see below). For the differential analysis, total cellular proteins from AP cells of wt and HI-6 strains and from synchronized GP stages of HI-6 were extracted with phenol according to the method of Barel et al. (4).

2DE.

2DE was performed both in our lab and at the Smoler Proteomics Center, The Technion, Haifa, Israel. Two hundred fifty micrograms of protein was separated on a 13-cm nonlinear pH gradient (3 to 10) strip (Immobiline DryStrip; Pharmacia) and run at 500, 1,000, and 2,000 V (1 h each) followed by 8,000 V until 48,900 V/h was reached, in an Ettan IPGphor isoelectric focusing system (Pharmacia). Each strip was then equilibrated twice in buffer according to the manufacturer's recommendations. Electrophoresis in the second dimension was performed in a 12% polyacrylamide gel at 35 mA per gel in a Protean II XI cell (Bio-Rad). The gels were stained with Coomassie brilliant blue G-250 (0.04% [wt/vol] Coomassie brilliant blue G-250, 3.5% [wt/vol] perchloric acid) and scanned with an ImageScanner (Pharmacia).

Alternatively, 200 μg of protein was loaded on 11-cm nonlinear pH gradient 3 to 10 strips or on 11-cm pH gradient 4 to 7 strips (Immobiline DryStrip; Bio-Rad). The strips were rehydrated with the protein samples for 1 h at zero voltage and then for 10 h at 50 V. Isoelectric focusing was performed at 200, 500, and 1,000 V (1 h each) followed by 8,000 V until 75,000 V/h was reached using a Protean IEF cell (Bio-Rad). Equilibration was performed as described above. Electrophoresis in the second dimension was performed in a 4 to 20% gradient polyacrylamide gel (Bio-Rad) at 100 V for 2.5 h using a Criterion Dodeca cell (Bio-Rad). Gels were stained with SeeBand Forte (GeBA) and scanned with ScanMaker 9800XL (Microtek). For comparative analysis between phases, one master gel was constructed for each of the two pH ranges investigated. This was done by combining all the 2DE gel data for the relevant range. Spot intensities were quantified, and differential spots were selected using Student's t test within PDquest (Bio-Rad) with a 95% confidence level. Within this set, only spots differing at least fivefold were selected for further analysis. Statistical significance of differential protein expression was established using analysis of variance on log₁₀-transformed spot intensity values with Tukey's highly significant difference test (P < 0.05) for multiple pairwise comparisons, using the JMP software (JMP). Comparative analyses of the expressed proteome in the various cultures and time points were performed with at least three independent cultures.

MS analysis.

Spots from the 2DE analyses were submitted to in-gel proteolysis and LC-MS/MS (Smoler Proteomics Center, The Technion, Haifa, Israel). Proteins were dehydrated in gel with acetonitrile and rehydrated with 10% acetonitrile in 10 mM ammonium bicarbonate containing modified trypsin (Promega) at a 1:100 enzyme-to-substrate ratio. Gel pieces were incubated overnight at 37°C, and the resulting peptides were recovered and analyzed. Tryptic peptides were resolved by reverse-phase chromatography on 0.1- × 200-mm fused silica capillaries (J&W; 100-μm inside diameter) packed with Everest reversed-phase material (Grace Vydac). The peptides were eluted with linear 50-min gradients of 5 to 95% acetonitrile with 0.1% formic acid in water at flow rates of 0.4 μl/min. MS was performed with an ion-trap mass spectrometer (LCQ-DecaXP; Finnigan) in a positive mode using repetitively full MS scanning followed by collision-induced dissociation of the three most dominant ions selected from the first MS scan. The MS data were clustered and analyzed using a Pep-Miner search of the bacterial database (7).

Sequence analyses.

Predictions of protein domains were conducted using PFAM and the BioInfoBank server (http://meta.bioinfo.pl/submit_wizard.pl). Theoretical molecular weights and pIs were calculated with http://www.expasy.ch/tools/pi_tool.html. Protein functional category classification was obtained from the COG database (http://www.ncbi.nlm.nih.gov/sutils/coxik.cgi?gi=384). Operons were predicted with the Softberry fgenesb program using the B. bacteriovorus 100^T genome sequence (http://www.softberry.com/berry.phtml?topic=fgenesb&group=programs&subgroup=gfindb).

To identify proteins specific to predatory deltaproteobacteria, we performed the following analysis. First, proteins from each predatory deltaproteobacterium were compared to proteins from other predatory and nonpredatory deltaproteobacteria. Proteins common to at least three of the four predatory deltaproteobacteria and not present in any representative nonpredatory deltaproteobacteria were kept. The predatory bacteria belong to two distinct orders: Bdellovibrionales (B. bacteriovorus HD100 and Bacteriovorax marinus) and Myxococcales (Myxococcus xanthus DK 1622 and Stigmatella aurantiaca DW4/3-1). The representative nonpredatory bacteria belonged to the orders Desulfobacterales (Desulfotalea psychrophila LSv54), Desulfuromonadales (Geobacter sulfurreducens PCA), Syntrophobacterales (Syntrophobacter fumaroxidans MPOB), Myxococcales (Anaeromyxobacter dehalogenans 2CP-C), and Desulfovibrionales (Desulfovibrio vulgaris Hildenborough). MLMS-1 was unclassified. Protein-toprotein comparisons (except for protein/DNA comparison in B. marinus) were performed with the BLAST program version 2.2.8 (2), at this step using a 1e⁻³ e-value threshold. The resulting group of protein sequences was then compared to the proteins and genomes of all deltaproteobacteria available in November 2006 using the more permissive BLAST default e-value (e = 10). Genomic and protein sequences were retrieved from the NCBI databases for all species except B. marinus, for which the sequence was retrieved from the Sanger Center.

RNA extraction.

Total RNA was extracted from wt AP cells and from synchronized wt GP cells at 1, 2, and 3 h after mixing predator and prey cells; from HI-6 AP cells and from synchronized HI-6 GP cells at initiation (T1), midgrowth (T2), and septation stage (T3); and from an exponentially growing culture of E. coli ML35, using the Total RNA extraction kit (Real Biotech Corporation). To eliminate DNA contamination, samples were concomitantly treated with DNase I (Fermentas). DNA traces were removed with Turbo DNA-free (Ambion), as confirmed by PCR without the reverse transcription step.

cDNA synthesis and semiquantitative RT-PCR.

The primers used in this study are presented in Table S1 in the supplemental material. cDNA was synthesized from RNA using the ImProm-II reverse transcription system (Promega) with random hexameric primers. Semiquantitative reverse transcription-PCR (RT-PCR) was performed on samples of the wt and HI-6 strains in a 10-μl mixture with the following protocol: 96°C for 2 min, 95°C for 20 s, annealing (58°C for all primer sets, except for Bd2944 [63°C]) for 20 s, 72°C for 30 s, for x cycles (see Table S1 in the supplemental material for cycle numbers). Negative controls included E. coli ML35 and a sample with no cDNA added. Band intensity was normalized relative to the internal reference Bd2400 gene using the Quantity One software (Bio-Rad). This gene exhibits constant gene expression in both the HI-6 and wt strains (M. Dori-Bachash, B. Dassa, O. Peleg, S. A. Pineiro, E. Jurkevitch, and S. Pietrokovski, unpublished data). Semiquantitative RT-PCR procedures were performed with two independent biological experiments.

Skim milk degradation assay.

The proteolytic activities of the wt and HI-6 strains were examined using skim milk as a substrate. The supernatant of an overnight culture (12 ml) was filtered (0.45-μm filtration followed by an 0.2-μm filtration), lyophilized, and resuspended in 1.5 ml HEPES buffer without Ca²⁺ and Mg²⁺. Samples (83 μl) were introduced into tubes containing 250 μl of 10% skim milk (dissolved in double-distilled water) and 167 μl phosphate-buffered saline and incubated at 30°C. To identify specific activities, phenylmethylsulfonyl fluoride, a serine protease inhibitor, was added at the final concentrations of 1, 2, 3, 4, and 5 mM. Clearance was monitored after 24 h of incubation.

RESULTS

Biological system.

Synchronized growth of the HI-6 strain mimicked events of a synchronized wt culture of the parental strain B. bacteriovorus 109J (23). While our unsynchronized HI-6 inoculum was composed of cells of various sizes, our synchronized cultures exhibited coordinated growth, starting with short elongating cells that developed into long filaments that finally fragmented into progeny AP cells (Fig. 1). Although the HI cycle is slower than the synchronized growth cycle of the wt strain, which lasts for about 3.5 h, the two strains exhibited strikingly parallel morphological features (1, 4).

FIG. 1.

Time course of a synchronous culture of the HI strain HI-6. An unsynchronized culture was obtained by overnight cultivation and used as an inoculum for synchronization. T1, T2, and T3 are time points at which samples were taken for analysis during the synchronous growth phase. T1, growth initiation, OD₅₇₀ = 0.4; T2, midgrowth phase, OD₅₇₀ = 1.2; T3, septation, OD₅₇₀ = 2.1. Fragmentation: the filamentous cell splits into progeny AP cells. Pictures were taken using a BX40 phase-contrast microscope equipped with an SC35 camera at an ×1,000 magnification (Olympus).

Expressed proteome of B. bacteriovorus.

Protein extracts from the HI-6 strain were obtained from AP and from GP cells at the growth initiation (T1), midgrowth (T2), and septation (T3) stages. A wide-range (pH 3 to 10) 2DE analysis of total cellular proteins revealed about 490 spots per gel, representing 13.6% of the estimated B. bacteriovorus 100^T proteome, which contains 3,587 open reading frames (see Fig. S1 in the supplemental material). Since the genomes of strains 109J and type strain 100^T may be approximately 98% identical (4, 14), we assume that the proteomes of the two strains are very similar. Supporting that claim was the fact that the MS data of proteins from strain 109J always matched the genome data of strain 100^T. Thirty-nine potentially cell cycle-differential spots were detected, most appearing in the pH 4 to 7 range. Thus, the proteome was accordingly reanalyzed in that range, yielding about 485 spots per gel, representing 27.4% of the in silico proteome (1,769 putative proteins in this range). Taken together from all pH ranges, approximately 645 nonredundant spots were detected, representing 18% of the estimated proteome. About 20% of these proteins were found to be differentially expressed during the cell cycle (see below). In addition, examination of the secreted complement in the extracellular milieu of the HI-6 strain revealed another 100 spots. In total, 745 spots representing 21% of the estimated proteome were detected, including isoforms. From these spots, about 200 proteins were further identified by LC-MS/MS (see Table S2 in the supplemental material). Among these proteins, 63 were previously annotated as hypothetical proteins.

Identification of differentially expressed proteins during the HI-6 strain's cell cycle.

Our proteomic analyses disclosed 39 and 98 differential spots in the pH 3 to 10 and pH 4 to 7 ranges, respectively, that could be classified according to their expression during the various stages of the bacterium's life cycle. In the pH 3 to 10 range, 24 spots were differentially expressed between the AP and GP stages and 12 spots were expressed within the GP stages. Three spots were differential between other combinations of stages. In the pH 4 to 7 range, 31 and 38 differentially expressed spots were detected between the AP and the GP stages and within the GP stages, respectively. Another 44 spots were differential between wt and HI-6 AP cells, some overlapping the other stages. Forty-seven of the differential spots were identified with certainty by LC-MS/MS (Table 1). These proteins belonged to various COG protein functional categories, with about half of them involved in signaling/information processing or of unknown functions (see Fig. S2 in the supplemental material). Thirty-nine of the identified differential proteins could be associated with 11 expression profiles (expression clusters) (Fig. 2) in which expression levels varied at least fivefold. Clusters 1 and 2 represent proteins strongly enhanced or repressed in the AP, respectively, and encompassed about half of the differential set. The full list of differentially expressed proteins and their expression profiles is presented in Table 1 and Fig. 2. Eleven of these are predicted as highly expressed genes in B. bacteriovorus (Table 1) (34).

TABLE 1.

Proteins differentially expressed during the life cycle of the HI-6 strain listed according to COG functional categories^a

Function and gene no.	Cluster	Identity	Ratio
Replication, recombination, and repair
Bd1564	2	MutL, DNA mismatch repair protein	22
Nucleotide metabolism
Bd1553		dUTP nucleotidohydrolase (dUTPase)	109e
Transcription
Bd2950b	2	DNA-directed RNA polymerase, alpha subunit	55
Bd2950b	2	DNA-directed RNA polymerase, alpha subunit	92
Translation
Bd1766c	5	Putative translation initiation inhibitor	27
Bd2994b		Translation factor EF-Tu	181e
Bd2994b		Translation factor EF-Tu	82f
Bd2994b		Translation factor EF-Tu	566f
Bd2994b	3	Translation factor EF-Tu	173
Bd2491b	7	Translation factor EF-P	68
Cell cycle control
Bd1158b,c	2	SMC, chromosome segregation protein	91
Bd3189b	5	FtsZ, cell division protein	58
Bd3189b	11	FtsZ, cell division protein	10
Bd3190	3	FtsA, cell division protein	76
Posttranslational modification, protein turnover, chaperones
Bd1298b,c	1	DnaK, chaperone protein	43
Bd0336c	2	FKBP-type peptidyl-prolyl cis-trans isomerase	66
Bd2518	2	Alkyl hydroperoxide reductase, subunit F	109
Bd0099b	2	60-kDa chaperonin (GroEL)	7
Bd0099b	2	60-kDa chaperonin (GroEL)	218
Bd0099b	2	60-kDa chaperonin (GroEL)	84 and 107f
Signal transduction mechanisms
Bd1513c	1	PilR, regulator protein	5
Bd1300	6	Conserved hypothetical protein	102
Cell motility
Bd1823	1	CheX, putative chemotaxis protein	143
Bd0864	10	PilN, fimbrial assembly membrane protein	57
Bd0408b	9	FlaA, flagellin	153
Bd0604b	4	Flagellin	175
Cell wall/membrane biogenesis
Bd2801c	1	Stage V sporulation protein G	5
GI 56387480d	9	Omp-like protein	149
GI 56387480d	9	Omp-like protein	13
Bd1334	1	Cell wall surface anchor family protein	26
Intracellular trafficking and secretion
Bd1597c	10	General secretion pathway protein D	22
Energy production and conversion
Bd0026	1	Succinate dehydrogenase subunit B	88 and 88e
Function unknown
Bd0231		Putative sporulation protein R	120f
Bd0163c	1	YdiY	71
Bd0223b,c	2	Putative regulator for granule-associated protein	215
Bd0223b	2	Putative regulator for granule-associated protein	7 and 43e
Bd0108	4	hit locus orf4	93
Bd0543	2	Putative transmembrane protein	85
Bd0991	1	Conserved hypothetical protein	124
Bd0790	1	Conserved hypothetical protein	5 and 116e
Bd2464	1	Conserved hypothetical protein	53 and 7f
Bd0801	3	Hypothetical protein	63
Bd2573		Hypothetical protein	118f
Bd2573		Hypothetical protein	128f
Bd1542b		Hypothetical protein	108e
Bd1558	8	Hypothetical protein	14
Bd1100	10	Hypothetical protein	52 and 52e

^aThe gene number is according to the B. bacteriovorus 100^T genome (52). Cluster refers to the temporal expression patterns (Fig. 2). The ratio presented is the highest ratio of the differential spot intensities obtained between life stages.

^bPredicted as highly expressed genes according to the work of Karlin et al. (34).

^cData from 2DE gels in the pH 3 to 10 range. All other proteins were identified from 2DE gels in the pH 4 to 7 range.

^dGI number is given when no identity was obtained from the annotated genome.

^eAP cell protein expression is higher in the HI-6 strain than in the wt strain.

^fAP cell protein expression is higher in the wt strain than in the HI-6 strain.

FIG. 2.

Temporal patterns of proteins differentially expressed during the life cycle of the HI-6 strain. (A) Example of a differentially expressed protein: the CheX protein (circle) is strongly expressed (from left to right) in the AP but not during growth initiation (T1), midgrowth (T2), and septation (T3). (B) Differentially expressed proteins are clustered according to their temporal expression. Each cluster contains proteins exhibiting the same pattern of expression. However, values in each cluster may vary, although a ratio of at least fivefold is conserved between the lowest and the highest expression levels. Different letters represent statistically significant values (Tukey's highly significant difference test [P < 0.05] for multiple pairwise comparisons).

Comparison of wt and HI-6 AP cells.

Forty-four differentiating spots were detected when AP cells of the wt and the HI-6 strains were compared, and among them 14 were identified (Table 1). These latter proteins varied in expression patterns: some differed between the APs of the two strains and not during the cell cycle of the HI-6 strain (e.g., a deoxyuridine 5′-triphosphate nucleotidohydrolase [Bd1553], three elongation factor Tu [EF-Tu, Bd2994] variants, and a putative sporulation protein R [Bd0231]); six were further differentially expressed during the HI-6 strain's life cycle (e.g., succinate dehydrogenase subunit B [Bd0026] and DNA-directed RNA polymerase, alpha subunit [Bd2950]).

Extracellular complement of strain HI-6.

Two hundred ninety-three hydrolases are annotated in strain 100^T, but only 15 were mentioned as potentially extracellular (52). We identified 59 polypeptides among the secreted complement (Table 2). All but nine contained a signal peptide, suggesting that they use the type II pathway to be secreted. Fourteen of these proteins were similar to known enzymes, including 10 serine proteases, two carboxypeptidases, one glycerophosphoryl diester phosphodiesterase (UgpQ), and one putative polysaccharide deacetylase. A milk protein degradation assay using a lyophilized sample of the secreted complement confirmed a major serine protease activity (see Fig. S3 in the supplemental material). A comparative genomic analysis further determined that three of the serine proteases (two trypsins [Bd0035 and Bd1391] and one V8-like Glu-specific endopeptidase [Bd1962]) and one of the carboxypeptidases (Bd3234) were specific to the four deltaproteobacterial predatory bacteria Myxococcus xanthus, Stigmatella aurantiaca, Bacteriovorax marinus, and B. bacteriovorus and were not found in the other deltaproteobacteria with determined genomes.

TABLE 2.

Extracellular proteins secreted by the HI-6 strain^a

Protein group and gene no.(s)	Identity
Serine proteases
Bd0035, Bd2630, Bd1391	Trypsin
Bd2428, Bd2269	Serine protease
Bd2321	Subtilisin-like serine protease
Bd0029	Subtilase
Bd1962	Putative V8-like Glu-specific endopeptidase
Bd3087, Bd2692	Protease
Other proteases
Bd0306, Bd3234	Carboxypeptidase
Miscellaneous enzymes
Bd2817	UgpQ, glycerophosphoryl diester phosphodiesterase
Bd3279	Putative polysaccharide deacetylase
Transport
Bd1658	Periplasmic phosphate-binding protein of ABC transporter
Bd1074	Putative iron ABC transporter
Flagellar motion
Bd3052, Bd0606, Bd0604, Bd0408, Bd0410	Flagellin
Bd0540	FlgK, flagellar hook-associated protein
Bd3395	FlgE, flagellar hook protein
Bd3407	FlgB, flagellar basal-body rod protein
Membrane associated
GI 56387480 in 109J	Omp-like protein
Bd2400	Putative cell surface protein
Function unknown
Bd2410	Conserved hypothetical iron-regulated protein
Bd0875, Bd0782, Bd1174, Bd0485, Bd2683, Bd1483, Bd1874, Bd2052, Bd0625, Bd3099, Bd0042, Bd2368, Bd1377, Bd2947, Bd0756, Bd0520, Bd0459, Bd2035, Bd1355, Bd1447, Bd1815, Bd3518, Bd1579, Bd2723, Bd2043, Bd0293, Bd3719, Bd2361, Bd0735, Bd3844, Bd1148, Bd0767	Hypothetical protein

^aThe gene number is according to the B. bacteriovorus 100^T genome (52). The GI number is given when no identity was obtained from the annotated genome.

In addition, the expression of 32 hypothetical proteins (ORFans) was confirmed. Seven of these (Bd0625, Bd0767, Bd0875, Bd1483, Bd2947, Bd2368, and Bd3099) possess a von Willebrand factor domain, a feature that mediates adhesion via metal ion-dependent adhesion sites and is mostly found in eukaryotic extracellular proteins (62). This domain was also exclusively detected in the four predatory deltaproteobacteria mentioned above.

Isoforms and protein modifications.

Fifteen isoforms of six proteins were detected, suggesting posttranslational protein modifications. These isoforms varied in molecular mass and pI (most certainly resulting from cleavage) or were detected as adjacent spots (“pearls-on-string”) or focused on more-acidic pIs than expected. The two latter patterns are typical of phosphorylated polypeptides (29, 55). Noticeably, EF-Tu, a GTPase (37) that is one of the most abundant proteins in 2DE gels (10), was detected in at least four spots that were differentially expressed in the various stages of growth or between wt and HI-6 AP cells (see Table S4 in the supplemental material). An Omp-like, differentially expressed protein that may be inserted in prey membranes (GI 56387480) (4) was identified as two isoforms that exhibited similar temporal patterns (Table 1). Based on in-gel molecular weight determinations, both were dimers in the 2DE gels, suggesting that these dimers are stabilized by strong protein-protein interactions (56) (more details on isoforms are provided in Table S3 in the supplemental material).

Operon prediction and confirmation and semiquantitative RT-PCR.

The genomic neighborhood of our set of identified differential proteins as well as of the secreted proteins was systematically explored, resulting in the prediction of 45 operons (see Table S4 in the supplemental material). Seven genes and operons were examined by semiquantitative RT-PCR (Table 1; Fig. 3). Two predicted polycistrons spanning the Bd0991 to Bd0992 and the Bd2573 to Bd2577 genes were confirmed by RT-PCR (Fig. 3), substantiating our detection method. Two genes, mutL and the dUTPase gene, exhibited gene expression patterns that matched those detected for their encoded proteins, being correspondingly transcribed at the GP stages in both HI and wt strains (Table 1; Fig. 3). Although the expression of other genes did not fully reflect protein expression patterns, uncoupling between transcription and translation appears to be a common feature in bacteria that can be explained by translational regulation and posttranscriptional regulatory control (30, 72).

FIG. 3.

Semiquantitative RT-PCR of genes encoding differentially expressed proteins. (A) RNA was purified from all the life cycle stages of both the wt and the HI-6 strains. Negative controls (−) included an E. coli template or no addition of cDNA. The RT-PCR amplicons were run on 2% agarose gels and stained with ethidium bromide. T1, T2, and T3 are growth initiation, midgrowth, and septation, respectively. The Bd2400 gene served as a reference gene. The Bd2400 gene is constitutively expressed in both the HI-6 and wt strains. In the latter, a dilution effect due to the presence of prey RNA in the bdelloplast formation has to be accounted for (Dori-Bachash et al., unpublished). (B) Visual grading of the gene expression analysis results according to the Bd2400 reference gene. Gene numbering is according to the B. bacteriovorus 100^T genome (52).

In contrast to the strong regulation observed at the protein level, the ftsA, cheX, Bd0991, and Bd2573 genes were constitutively expressed in the HI-6 strain, suggesting that gene regulation may be affected in the mutant. EF-Tu exhibited a similar gene expression pattern in both the wt and the HI-6 strains, but the expression pattern of the protein was complex and included at least four isoforms.

DISCUSSION

In this work, we provide the first detailed description of the dynamics of the proteome of an obligate predatory bacterium. The data acquired here provide a temporal frame of protein expression, lead to functional assignments for unknown proteins, and suggest hitherto unknown links between the cell cycle and particular functions. The possibility of synchronizing HI and wt cultures makes B. bacteriovorus a potent system for proteomics as well as for other “-omics” and cellular biology studies. These data can further serve as a scaffold to decipher protein expression in the wt strain and patterns of prey protein degradation during the predatory interaction.

Cell cycle-dependent proteome of B. bacteriovorus.

Few predation-required or cell cycle-related functions are known in B. bacteriovorus. A previous investigation (45) revealed that at least 30 polypeptides were differentially expressed during the bacterium's life cycle. However, none were identified. Here, we expanded this set to 140 detected differential spots, representing approximately 20% of the detected cellular proteome, and 47 were identified. Proteins that were expressed only during the HI-6 GP, were absent from the AP cells of both the HI and wt strains, and filled functions that could be assigned to the GP could confidently be considered real differentials (Fig. 2; Table 1). Differentially expressed proteins with no known functions or with functions with no obvious need for a temporal regulation may either be true differentials or result from the HI phenotype. Yet, a significant fraction of the identified differentially expressed proteins in the HI-6 strain validated the use of the HI model, serving as landmarks of the biological system, as these proteins were expressed during the expected phase (see below).

(i) Prey search, docking, and invasion (events I, II, and III).

A large proportion of AP-expressed proteins were related to cell movement, cell-to-cell contact, and ancillary systems. During the AP stage, the predator seeks its prey by chemotaxis (38, 41, 65). Accordingly, the putative chemotaxis protein CheX was expressed in the AP stage. CheX may regulate CheY phosphorylation (48), and both proteins are encoded in the same operon in B. bacteriovorus, further supporting a functional association between them.

The long and dampened wave-like sheathed flagellum of Bdellovibrio enables it to swim at speeds of up to 100 body lengths per second (64). It is made of six flagellin proteins (52) five of which were detected in this proteomic analysis, along with three hook proteins. FlaA deletion (FliC3, Bd0408) was shown to totally impair swimming and to abolish predation in liquid cultures (39). We found that FlaA and FliC1 (Bd0604) were strongly expressed during the AP but exhibited different expression patterns (Fig. 2). Consistent with the proposed role of FlaA as a core flagellum protein or as a linker between the hook and the filament (39), FlaA expression was low during T1 but strongly increased in T2 and reached AP levels in T3.

Fimbriae are required for predation (17). Accordingly, the regulatory and assembly proteins PilN (Bd0864) and PilR (Bd1513) and the general secretion pathway protein D (Bd1597) were strongly expressed during AP. The latter is coexpressed in an operon along with general secretion pathway protein F, a predicted ATPase involved in pilus biogenesis, and a PilB homolog. Based on their genomic context and expression patterns, we propose that the hitherto hypothetical proteins Bd0108 and Bd1483 are also involved in predation-related adhesion processes. The former is strongly expressed during the AP and is part of the hit locus, a large cluster of adhesion-related genes (52), and the latter, a secreted protein, is found in an operon encoding gliding proteins. A recent screen for predatory mutants showed that when the Bd1483 gene was mutated, predation in biofilms was impaired but the formation of plaques on prey lawns was not prevented (46).

During attachment to and entry into the prey, the predator's membranes are certainly exposed to osmotic and other stresses (1). YdiY (Bd016), a putative salt-induced outer membrane protein, was exclusively expressed at the AP stage. Its gene is polycistronic with mscL (Bd0162), a high-conductance, mechanosensitive, cytoplasmic membrane channel. The MscL protein transduces physical stresses in the cell membrane into an electrochemical response, participating in the regulation of osmotic pressure changes (12). Mechanosensing through YdiY and MscL may form a signaling mechanism that is activated when the predator attaches to the prey's cell wall and penetrates the prey cell.

During prey cell penetration, the prey's peptidoglycan undergoes several modifications and the following proteins are putatively involved in this process: Bd3279, a secreted polysaccharide deacetylase which can perform N deacetylation of the peptidoglycan amino sugars (68); Bd1334, an AP-expressed cell wall anchor protein that contains a d-Ala-d-Ala carboxypeptidase domain characteristic of a class of peptidoglycan binding proteins; and the AP-expressed uncharacterized protein Bd0991 and its coexpressed Bd0992 gene-encoded cell wall hydrolase.

(ii) Cell growth and metabolic control (IV to VII).

Proteins involved in transcription, translation, translation termination, and gene regulation were sequentially expressed during the GP stages. The DNA mismatch repair protein MutL (Bd1564) (44) and the chromosome segregation protein SMC (Bd1158) (21) were strongly expressed from the onset of the growth phase, suggesting that DNA replication is rapidly upregulated, in accordance with the unbalanced growth pattern of B. bacteriovorus (23), and that DNA condensation, segregation, and proper positioning in the expanding cell occur as chromosome copies are synthesized. In longer prey cells, the elongated BALO predator exhibits a corkscrew shape (1). This conformation may help optimize contact with the prey's cytoplasmic membrane and the use of the restricted space available for growth within the bdelloplast. This morphology is conserved in some HI mutants, including the strain used in this study (5) (Fig. 1), strongly suggesting that it does not stem from physical constraints in the bdelloplast but rather results from a regulated cell plan. A recent study has shown that the Bdellovibrio cell is unusually flexible and contains numerous filamentous structures (8). Notably, MreB, an actin-like homolog, is present as two copies in B. bacteriovorus. MreB regulates cell shape and may also be implicated in chromosome segregation (19).

RNA polymerase subunits (Bd2950 and rnpO, data not shown), some isoforms of EF-Tu (Bd2994) (37), and translation elongation factor P (EF-P) (Bd2491) were strongly enhanced in GP, consistent with an increased rate of protein production during growth. EF-Tu modifications regulate its activity: methylation occurs in response to nutrient deprivation (74), reducing GTP hydrolysis and thus slowing translation and increasing accuracy (70). A putative EF-Tu-GTP isoform was exclusively expressed in wt AP cells, when translation occurs at lower rates than it does during the GP stages (23). EF-P is involved in the formation of the first peptide bond of nascent proteins (26), and it is therefore surprising that it was detected only during the midgrowth stage (at T2). An undetected isoform may be present, or a different protein may play a similar role at other stages.

Another unusual feature is the strong expression during GP of a granule-associated protein (Bd0223) that contains a polyhydroxyalkanoate (PHA) accumulation regulator DNA-binding domain, as reserve material accumulation commonly starts during the late growth phase (3). Moreover, the granules observed in B. bacteriovorus resemble phosphate-rich acidocalcisomes but not polyalkanoate granules (8). Nevertheless, B. bacteriovorus possesses most of the butanoate metabolism pathway as well as a poly-β-hydroxybutyrate depolymerase (Bd2637) and a lipase A (Bd0664) that may be involved in extracellular PHA degradation (49), as both contain signal peptides. Whether these enzymes degrade prey PHAs or have other roles in Bdellovibrio remains to be elucidated. Although in BALOs prey cell size determines the number of progeny cells (35), the influence of prey cell content on progeny yield or survival is unknown.

(iii) Termination of cell growth and septation (VI to VII).

Both Bd1766 and Bd1300 proteins, which were respectively expressed and repressed during T3, may be involved in GP termination. Bd1766 possesses an endoribonuclease L-PSP domain, known to inhibit protein synthesis through mRNA cleavage (47), thereby decreasing translation rates. Bd1300, the expression of which is strongly reduced at the transition between growth phase and cell differentiation, is a candidate for a central regulatory function linked to the metabolic state of the cell: it contains a cNMP binding domain and a Crp domain, both found in cyclic AMP receptor proteins, prokaryotic catabolite gene activators/repressors (9, 28). This regulatory activity may be connected to putative endogenous or prey-derived compounds that signal phase transition (16, 22, 24). Since the predator cannot anticipate the amount of resources available within a prey, the utilization of prey content for cell growth until the resource is exhausted is a very sensible growth strategy, more so in the oligotrophic habitats where BALOs are found.

Fragmentation of the Bdellovibrio filamentous cell requires that multiple septation events occur concomitantly. The bacterial Z ring is formed at the plane of cell division by FtsZ polymerization (43). We suggest that, in the absence of FtsA, the isoform detected in AP and in T1 forms an increasingly large cytosolic pool of FtsZ short protofilaments (51) toward T3 (cluster 11; 10×). Since FtsA has been shown to tether such FtsZ protofilaments to the cell's membrane (50, 51), its increased expression from T2 onwards (cluster 3; 76×) may bring about membrane anchoring of the FtsZ protofilaments and modification of the FtsZ protein (cluster 5; 58×). Septation of the long Bdellovibrio cell occurs at sites so that each progeny cell inherits a copy of the genome. The determination of these sites may rely on a mechanism different than the one described for E. coli, as the proteins MinC, MinD, and MinE, which oscillate between cell poles (42), are not found in the B. bacteriovorus genome.

Interestingly, the two chaperones DnaK and GroEL were differentially regulated in the HI mutant, and at least GroEL expression differs between the wt and the HI-6 strains. The data presented here, their distinct roles in stress responses (55) during cell cycle progression, and their effect on cell morphology in Caulobacter crescentus (25, 67) suggest that these chaperones may also be important components of the AP-GP transition in B. bacteriovorus.

Finally, upregulation of AP traits before AP cells are released (i.e., during T3; Bd0408, Bd0864, Bd1100, and Bd1597) may be advantageous, as the newly released cells do not need to mature and can immediately engage in predation, thereby minimizing energy losses. Moreover, AP and differentiating cells may require some common functions (e.g., prey cell wall hydrolysis).

Variations between wt and HI-6 AP cells may reflect deregulated pathways in the HI mutant.

By definition, our mutant model is different from the wt strain. These differences are informative and can suggest pathways affected in the mutant phenotype that can be relevant to the transition from obligate to facultative predation (32) or vice versa. HI mutants behave as facultative predators, i.e., in contrast to wt BALOs, they are not “locked” in the AP but readily enter the growth phase when resources are available in the growth medium. This suggests that in HI AP cells, gene regulation and/or the processing and turnover of some proteins are altered, as seen with dUTPase, an enzyme that prevents uracil incorporation into the replicating DNA molecule (11) (Table 1; Fig. 3). Interestingly, half of the differentially expressed proteins had no affiliation whatsoever, suggesting that unknown functions may be involved in this transition. HI AP cells are clearly not “true” AP cells. Since they do not predate as well as wt AP cells (reference 71 and data not shown), a trade-off may exist between predation efficiency and a facultative life-style. A genome-wide characterization of this HI mutant is under way that may shed light on this transition.

Extracellular proteins.

There are as many as 293 annotated hydrolytic enzymes in the 100^T genome (52). In wt strains, few proteins appear to be secreted from the bdelloplast to the outer medium (reference 13 and data not shown), implying that at least a fraction of the extracellular complement of proteins detected in this study could, in a bdelloplast, be targeted to the periplasm or to the cytoplasm of the prey cell.

Serine proteases and carboxypeptidases, the activities of which have been detected in Bdellovibrio (20), were identified within the set of secreted proteins. Serine proteases form a varied family of endopeptidases (27), which constitute the bulk of the proteolytic activity in the extracellular fraction. The two carboxypeptidase enzymes that we have identified possess a zinc-carboxypeptidase M14 domain typical of metallopeptidases (73). These enzymes may be used by the predator to penetrate the prey and/or to consume its internal proteins as a nutrient source. Such activities have previously been observed in the degradation of prey outer membrane proteins (4, 6) and peptidoglycan breakdown (69). The genes encoding the unaffiliated proteins Bd0520 and Bd0756 are located in operons containing a serine protease (Bd0521) and a putative aminopeptidase (Bd0755), respectively, and may participate in prey cell lysis. Strikingly, 11 secreted proteins were exclusive to deltaproteobacterial predators. These might be predation-specific hydrolytic and adhesion proteins, as they are found only in the arsenals of bacterial predators and may endow them with the ability to crack uncompromised cell walls. Whether these capabilities were acquired through lateral gene transfer is a pertinent question that remains to be answered.

Conclusions.

This work provides empirical data to answer some of the open questions regarding the predatory life cycle of B. bacteriovorus (66). It shows that HI strains constitute a convenient model for the analysis of BALO proteomes. A large number of proteins implicated in the predatory phenotype and in the dimorphic cell cycle of B. bacteriovorus were identified, providing new, nonintuitive targets for mutagenesis. Novel technologies for protein analysis (15) and enhanced methods for targeted mutagenesis in B. bacteriovorus (63) will certainly greatly improve our ability to decipher what turns this bacterium into an obligate predator and how its peculiar dimorphic cell cycle is organized.