Development and Use of a Gene Deletion Strategy for Flavobacterium johnsoniae To Identify the Redundant Gliding Motility Genes remF, remG, remH, and remI ^{▿ †}

Bacterial strains, bacteriophages, plasmids, and growth conditions.

F. johnsoniae ATCC 17061 (UW101) was the wild-type strain used in this study (21, 24). F. johnsoniae strains were grown in Casitone-yeast extract (CYE) medium at 30°C, as previously described (23). To observe colony spreading, F. johnsoniae was grown on PY2 agar medium (1) at 25°C. Motility medium (MM) was used to observe movement of individual cells in wet mounts (18). The bacteriophages active against F. johnsoniae that were used in this study were φCj1, φCj13, φCj23, φCj28, φCj29, φCj42, φCj48, and φCj54 (6, 27, 34). Sensitivity to bacteriophages was determined essentially as previously described, by spotting 5 μl of phage lysate (10⁹ PFU/ml) onto lawns of cells in CYE overlay agar (14). The plates were incubated for 24 h at 25°C to observe lysis. Strains and plasmids used in this study are listed in Table 1. The plasmids used for complementation were all derived from pCP1 and have copy numbers of approximately 10 in F. johnsoniae (1, 16, 23). Primers used in this study are listed in Table S1 of the supplemental material. Antibiotics were used at the following concentrations when needed: ampicillin, 100 μg/ml; cefoxitin, 100 μg/ml; erythromycin, 100 μg/ml; kanamycin, 35 μg/ml; streptomycin, 100 μg/ml; tetracycline, 20 μg/ml.

Isolation of the streptomycin-resistant rpsL mutant CJ1827.

Streptomycin-resistant F. johnsoniae cells were obtained by plating 10⁹ wild-type cells on CYE agar containing streptomycin. Streptomycin-resistant clones were streaked for isolation on fresh medium, and the rpsL gene from each clone was PCR amplified using primers 964 and 965 and sequenced. Point mutations in the rpsL gene conferring streptomycin resistance were identified by comparison to the wild-type rpsL gene sequence.

To determine if the wild-type rpsL gene was dominant to the mutant rpsL alleles, the wild-type rpsL gene was cloned into the shuttle vector pCP23. Specifically, a 711-bp fragment spanning the wild-type rpsL gene and its presumed promoter region was amplified using primers 979 and 980, which were designed with engineered SphI and XbaI restriction sites, respectively. The fragment was digested with SphI and XbaI and cloned into the corresponding sites of pCP23 to generate pRR50. pRR50 was introduced into streptomycin-resistant F. johnsoniae by triparental conjugation as previously described (14), except that the helper strain carried pRK2013 instead of R702.

Construction of the rpsL-containing suicide vector pRR51.

To generate unmarked mutations using rpsL as a counterselectable marker, the wild-type rpsL gene was cloned into the suicide vector pLYL03. Primers 981 and 982 were designed with engineered EcoRI restriction sites and used to amplify a 681-bp fragment spanning rpsL and its putative promoter. The fragment was digested with EcoRI and cloned into the EcoRI site of pLYL03 to generate pRR51 (Fig. 1). Orientation of the rpsL fragment in pRR51 was determined by sequencing.

Construction of the sprB deletion mutant CJ1922.

A 2.1-kbp fragment spanning the 3′ end of sprD and including the first 168 bp of sprB was amplified using primers 1032 (introducing a BamHI site) and 1033 (introducing a SalI site). The fragment was digested with BamHI and SalI and ligated into pRR51, which had been digested with the same enzymes, to generate pRR66. A 2.8-kbp fragment spanning sprF, pgk, and the final 222 bp of sprB was amplified with primers 720 (introducing a SalI site) and 727 (introducing an SphI site). The fragment was digested with SalI and SphI and fused to the upstream region of sprB by ligation with pRR66, which had been digested with the same enzymes, to generate the deletion construct pRR67 (Fig. 2).

Plasmid pRR67 was introduced into the streptomycin-resistant wild-type F. johnsoniae strain CJ1827 by triparental conjugation followed by growth on CYE agar containing erythromycin to select for integration of the plasmid into the genome by homologous recombination. An erythromycin-resistant (streptomycin-sensitive) clone was grown overnight in CYE in the absence of antibiotics, and loss of the plasmid by a second recombination event was selected by growth on PY2 agar medium containing streptomycin. Deletion of sprB in CJ1922 was confirmed by PCR amplification using primers 848 and 947, which flank the sprB coding sequence, and sequencing the resulting 1.6-kbp product.

Construction of CJ1883, which has a deletion spanning remF, remG, and fjoh_0982.

To generate a deletion in the three proximal genes of the sprB operon, a 2.5-kbp fragment spanning fjoh_0987, fjoh_0986, fjoh_0985, and the first 50 bp of fjoh_0984 (remF) was amplified using primers 983 (introducing a BamHI site) and 984 (introducing an XbaI site). The PCR fragment was digested with BamHI and XbaI and ligated into pRR51, which had been digested with the same enzymes, to generate pRR52. The 2.4-kbp region spanning the final 191 bp of fjoh_0982, sprC, and the 5′ end of sprD was amplified with primers 985 (introducing an XbaI site) and 986 (introducing a PstI site). The amplified product was digested with XbaI and PstI and ligated into pRR52, which had been digested with the same enzymes, to generate the deletion construct pRR53 (Fig. 2).

Plasmid pRR53 was introduced into CJ1827, and the double recombination event resulting in the deletion of remF, remG, and fjoh_0982 was selected for as described above, except that CYE agar medium containing streptomycin was used for the counterselection step. Streptomycin-resistant clones were tested for the deletion by PCR using primers 831 and 842, which flanked the region of interest and resulted in either a 3.4-kbp PCR product for clones retaining remF, remG, and fjoh_0982 or a 1.2-kbp PCR product for clones in which these three genes had been deleted. The deletion in strain CJ1883 was confirmed by sequencing the 1.2-kbp PCR product.

Deletion of remH and remI.

Separate constructs were generated to make deletions of the remF paralog, remH (fjoh_3206), and the remG paralog, remI (fjoh_3194). For remH, a 2.7-kbp fragment spanning the 5′ end of fjoh_3203, fjoh_3204, fjoh_3205, and the final 89 bp of remH was amplified with primers 999 (introducing a BamHI site) and 1000 (introducing a SalI site). The PCR fragment was digested with BamHI and SalI and ligated into pRR51, which had been digested with the same enzymes, to generate pRR55. The 2.6-kbp region upstream of remH spanning the 5′ end of fjoh_3209, fjoh_3208, fjoh_3207, and the first 60 bp of remH was amplified with primers 1001 (introducing a SalI site) and 1002 (introducing an SphI site). The upstream fragment was digested with SalI and SphI and ligated into pRR55, which had been digested with the same enzymes, to generate the remH deletion construct pRR61 (Fig. 3 A). Plasmid pRR61 was used to generate remH deletions in various backgrounds as described above. Primers 1022 and 1023 flanking remH were used to identify streptomycin-resistant clones carrying the deletion. A 0.6-kbp PCR product was obtained for clones carrying the deletion, and a 0.9-kbp PCR product was obtained for clones retaining the wild-type remH allele.

To generate a deletion in remI, a 2.7-kbp fragment spanning the 5′ end of fjoh_3191, fjoh_3192, fjoh3193, and the first 112 bp of remI was amplified with primers 995 (introducing a BamHI site) and 996 (introducing a SalI site). The PCR fragment was digested with BamHI and SalI and ligated into pRR51, which had been digested with the same enzymes, to generate pRR54. The 2.6-kbp region downstream of remI spanning the 3′ end of fjoh_3197, fjoh_3196, fjoh_3195, and the final 149 bp of remI was amplified with primers 997 (introducing a SalI site) and 998 (introducing an SphI site). The downstream fragment was digested with SalI and SphI and ligated into pRR54, which had been digested with the same enzymes, to generate the remI deletion construct pRR60 (Fig. 3B). Plasmid pRR60 was used to generate remI deletions in various backgrounds as described above. Primers 1020 and 1021 flanking remI were used to identify streptomycin-resistant clones carrying the remI deletion. A 0.7-kbp product was obtained for clones carrying the deletion, and a 1.8-kbp product was obtained for clones retaining the wild-type remI allele.

Construction of complementation plasmids.

Plasmids carrying remF and remG were constructed using shuttle vectors pCP23 and pCP29 and transferred to various F. johnsoniae strains by triparental conjugation for complementation studies. We previously constructed pRR68, which carries remF, remG, and fjoh_0982 (28). To construct a plasmid containing only remF and remG, primers 718 (introducing a BamHI site) and 1039 (introducing an SphI site) were used to amplify a 2.3-kbp fragment spanning remF and remG and the experimentally determined promoter region (28). The fragment was digested with BamHI and SphI and ligated into pCP23, which had been digested with the same enzymes, to generate pRR72 (Fig. 2). Plasmids containing only remF or remG were also constructed. An 881-bp fragment spanning remF and its promoter region was amplified using primers 718 (introducing a BamHI site) and 1044 (introducing an SphI site). The fragment was digested with BamHI and SphI and ligated into pCP23, which had been digested with the same enzymes, to generate pRR73 (Fig. 2). To construct a plasmid containing remG, a 1.9-kbp fragment spanning remG was amplified using primers 1038 and 1039, which were designed with engineered BamHI and SphI restriction sites, respectively. The fragment was digested with BamHI and SphI and ligated into pCP23 to generate pRR69 (Fig. 2). The remG fragment in pRR69 did not carry its own promoter but was oriented in the same direction as orf1 in pCP23 (2) and resulted in expression from the vector promoter.

Plasmids containing the remF and remG paralogs, remH and remI, were also constructed. To construct a plasmid containing remH, a 0.9-kbp fragment spanning remH and its putative promoter was amplified with primers 1042 and 1043, which were designed with engineered SalI and XmaI restriction sites, respectively. The fragment was digested with SalI and XmaI and ligated into pCP29, which had been digested with the same enzymes, to generate pRR75 (Fig. 3A). To generate a plasmid carrying remI, primers 1036 (introducing an SphI site) and 1037 (introducing a BamHI site) were used to amplify a 2.0-kbp fragment spanning remI and its putative promoter region. The fragment was digested with SphI and BamHI and inserted into pCP23, which had been digested with the same enzymes, resulting in pRR70 (Fig. 3B). The 2.0-kbp KpnI-SphI fragment of pRR70 spanning remI was ligated into pCP29, which had been digested with the same enzymes, to generate pRR71.

Microscopic observations of cell movement.

Wild-type and mutant cells of F. johnsoniae were examined for movement over glass by using phase-contrast microscopy. Cells in MM were spotted onto a glass microscope slide and were covered with a glass coverslip, incubated for 1 min, and observed for motility using an Olympus BH-2 phase-contrast microscope with a heated stage set at 25°C. For analyses of cell tracks, cell movements were observed using Palmer counting cells (Wildlife Supply Company, Saginaw, MI) as previously described (25), except that 22-mm² glass coverslips were used. Images were recorded using a Photometrics CoolSNAP_cf² camera and were analyzed using MetaMorph software (Molecular Devices, Downingtown, PA). Tracks illustrating the movements of cells were obtained by superimposing individual digital video frames using the logical AND operation of MetaMorph.

Binding and movement of protein G-coated polystyrene spheres carrying antibodies against SprB.

Movement of surface-localized SprB was detected as previously described (25). Cells were grown overnight at 25°C in MM without shaking. Purified anti-SprB (1 μl of a 1:10 dilution of a 300-mg/liter stock), 0.5-μm-diameter protein G-coated polystyrene spheres (1 μl of a 0.1% stock preparation; Spherotech Inc., Libertyville, IL), and bovine serum albumin (1 μl of a 1% solution) were added to 7 μl of cells (approximately 5 × 10⁸ cells per ml) in MM. The cells were spotted on a glass slide, covered with a glass coverslip, and examined by phase-contrast microscopy at 25°C. Samples were examined 1 min after spotting, and images were captured for 30 s.

Western blot analyses.

Wild-type and mutant F. johnsoniae cells were grown to mid-log phase in MM at 25°C without shaking (100-ml cultures in 500-ml Erlenmeyer flasks). Cells were harvested by centrifugation at 4,000 × g, and pellets were concentrated in 1 ml of lysis buffer (20 mM sodium phosphate [pH 7.4], 10 mM EDTA). Ten microliters of 100× HALT protease inhibitor (Thermo Fisher Scientific, Waltham, MA) was added, and cells were lysed with a French pressure cell. Cell lysates were solubilized in SDS-PAGE loading buffer by boiling for 5 min, and proteins were separated by SDS-PAGE on 3-to-8% Criterion XT Tris-acetate-acrylamide gradient gels (Bio-Rad, Hercules, CA). Detection of SprB was carried out as previously described (29).

Measurement of chitin digestion.

Chitin utilization on agar plates was detected as previously described (29). Briefly, an MYA agar (0.5 mM MgSO₄, 0.05 mM FeSO₄, 0.04 mM EDTA, 20 mM potassium phosphate, [pH 7.25], 0.1 g of yeast extract per liter, 15 g of agar per liter) plate was overlaid with 3 ml of a 2% chitin slurry and allowed to dry overnight at room temperature. Cells were grown overnight in MM, and 3 μl (approximately 10⁶ cells) was spotted onto the plate, allowed to dry, and incubated at 25°C for 4 days.

Development of rpsL as a counterselectable marker for F. johnsoniae and construction of an sprB deletion mutant.

Five spontaneous streptomycin-resistant mutants of wild-type F. johnsoniae were isolated, and the rpsL genes were amplified and sequenced. Four of the mutants had an adenine-to-guanine point mutation at bp 263 of the rpsL coding sequence (rpsL1) that resulted in a K88R substitution in RpsL. This is the same lysine residue altered in streptomycin-resistant Escherichia coli (10) and Borrelia burgdorferi (8). The fifth mutant, CJ1827, had an adenine-to-guanine point mutation at bp 128 (rpsL2), resulting in a K43R substitution in RpsL. The mutants producing K88R RpsL formed colonies that spread slightly less well than the wild type, whereas CJ1827 (rpsL2) was indistinguishable from the wild type except for its resistance to streptomycin. In E. coli, wild-type rpsL is dominant to the mutant allele, and merodiploids are sensitive to streptomycin (30). We tested whether this was also true for F. johnsoniae. pRR50, which carries wild-type F. johnsoniae rpsL inserted into the shuttle vector pCP23, was introduced into CJ1827. CJ1827 without the plasmid grew well in the presence of 100 μg per ml streptomycin, whereas CJ1827 carrying pRR50 failed to grow under these conditions.

Wild-type rpsL was introduced into the suicide vector pLYL03 to generate pRR51 (Fig. 1), which was designed to facilitate construction of unmarked mutations. As a test case, an sprB deletion mutant was constructed. Regions upstream and downstream of sprB were cloned in pRR51, and the plasmid was introduced into CJ1827 by conjugation. Erythromycin-resistant colonies arose as a result of integration of the plasmid into the chromosome. Cells were plated on a medium containing streptomycin to select for loss of the plasmid by a second recombination event, and thousands of streptomycin-resistant colonies were obtained. Depending on the site of the second recombination event, the resulting colonies were expected to have either a wild-type sprB locus or an in-frame deletion spanning most of sprB, and colonies of each type were identified. Deletion of sprB in CJ1922 was verified by PCR and sequencing as described in Materials and Methods. CJ1922 formed nonspreading colonies (Fig. 4 B), as previously reported for other sprB mutants (25, 28). CJ1922 also exhibited resistance to the same bacteriophages (see Fig. S1 in the supplemental material), as do other sprB mutants (25, 28). Complementation with pSN60 (which carries sprB) restored the ability to form spreading colonies (Fig. 4C), verifying the lack of polarity of the unmarked in-frame deletion. In contrast, polar mutations in sprB are not complemented by pSN60 but rather require plasmids expressing both sprB and the downstream gene, sprF, for complementation (28).

Deletion of the region spanning remF, remG, and fjoh_0982 does not disrupt motility.

sprB is part of a seven-gene operon, and each of the distal genes (sprC, sprD, sprB, and sprF) is required for normal motility and for the formation of spreading colonies. The situation was less clear for the proximal genes, remF, remG, and fjoh_0982. CJ1708, which has an insertion in remG, formed nonspreading colonies (Fig. 4D), but the insertion resulted in polar effects on the downstream genes that could have been responsible for this phenotype (28). We used pRR53 to construct CJ1883, which has a deletion spanning remF, remG, and fjoh_0982. Western blot analysis revealed that the deletion in CJ1883 did not exhibit polar effects on expression of the downstream gene, sprB (Fig. 5). Cells of CJ1883 formed spreading colonies on agar that were similar to those formed by wild-type cells (Fig. 4E), indicating that remF, remG, and fjoh_0982 are not required for the formation of spreading colonies.

Deletion of remH and remI cause motility defects in CJ1883 [Δ(remF remG fjoh_0982)].

Analysis of the F. johnsoniae genome (24) revealed a paralog of remF (fjoh_3206, which we named remH) and a paralog of remG (fjoh_3194, which we named remI) (Fig. 3). remF and remG are cotranscribed with sprB (28), but remH and remI appear to be transcribed as single genes, and they are not located near sprB paralogs or near known motility genes. RemF and RemH exhibit 92% identity over their entire length (150 amino acids each), and RemG (430 amino acids in length) and RemI (429 amino acids in length) exhibit 80% identity with each other over 428 amino acids. Several additional remG paralogs (fjoh_0546, fjoh_3856, and fjoh_4889) were also identified, but the proteins encoded by these genes were less similar to RemG (31% to 47% amino acid identity), and they were not considered further in this study. RemF, RemG, RemH, and RemI are each predicted to have N-terminal signal peptides targeting them for export across the cytoplasmic membrane by the sec system. PSORT and CELLO analyses predict that RemF and RemH are periplasmic proteins and that RemG and RemI are outer membrane or extracellular proteins. RemF and RemH are similar in sequence to predicted proteins of unknown function from several members of the phylum Bacteroidetes, but they do not share significant sequence similarity with predicted proteins from other phyla of bacteria or from members of the domains Archaea and Eukarya. RemG and RemI exhibit sequence similarity to many bacteroidete proteins of unknown function, and they exhibit more limited sequence similarity to predicted proteins of unknown function from members of other bacterial phyla.

The ability to make unmarked deletions allowed us to repeatedly use the same selectable marker (erythromycin resistance) and counterselectable marker (streptomycin resistance) to construct strains with multiple deletions. A collection of strains with multiple deletions was constructed to determine the effect of elimination of either set of paralogs (remF and remH or remG and remI). Strains with single deletions of remH or remI or strains missing both remH and remI formed spreading colonies that were similar to those of the wild type or of CJ1883 [Δ(remF remG fjoh_0982)] (Fig. 6 A to E). However, introduction of either a remH deletion or a remI deletion into CJ1883 resulted in motility defects and the formation of nonspreading colonies (Fig. 6F and K). The ability to form spreading colonies was restored by introduction of remF and remG on pRR68 or pRR72, or by the introduction of remH and remI on pRR75 and pRR70, verifying that the mutations in remF, remG, remH, and remI were responsible for the nonspreading phenotypes. fjoh_0982 is not required for formation of spreading colonies, since CJ1913 [Δ(remF remG fjoh_0982) ΔremH] and CJ1912 [Δ(remF remG fjoh_0982) ΔremI] were complemented by either pRR68 (which carries remF, remG, and fjoh_0982) or by pRR72 (which carries just remF and remG). The phenotypes associated with mutations in the paralogous genes suggest that the encoded proteins perform redundant or semiredundant roles associated with motility. Surprisingly, introduction of pRR75, which carries remH, into CJ1913 [Δ(remF remG fjoh_0982) ΔremH] resulted in the formation of colonies that had a few tiny flares (Fig. 6I) but spread much less well than wild-type colonies or than colonies of CJ1883 [Δ(remF remG fjoh_0982)]. Further analysis revealed the likely reason for the poor complementation, since expression of remH from pRR75 in wild-type cells also resulted in a dramatic reduction of colony spreading (Fig. 7). The motility defect caused by pRR75 could be corrected by introducing remI on pRR70 (Fig. 6J and 7), suggesting the possibility that overexpression of remH causes a motility defect and overexpression of remI in the same cells counteracts the deleterious effects of high levels of RemH.

To determine which combination of genes allows formation of spreading colonies, we constructed CJ1939, which is missing remF, remG, fjoh_0982, remH, and remI, and introduced combinations of genes on plasmids (Fig. 8). As expected, introduction of the individual genes (remF, remG, remH, and remI) failed to restore colony spreading. Introduction of remF and remG, remH and remI, or remF and remI resulted in formation of spreading colonies, whereas introduction of remG and remH did not. Apparently RemF can function with either RemG or RemI, whereas RemH (the paralog of RemF) can function with RemI (paralog of RemG) but not with RemG. Further evidence that the presence of RemH and RemG is not sufficient for formation of spreading colonies was obtained from complementation experiments performed with CJ1912 [Δ(remF remG fjoh_0982) ΔremI]. Introduction of remI on pRR70 restored formation of spreading colonies, whereas introduction of remG on pRR69 did not (Fig. 6N and O).

Effect of mutations in remF, remG, remH, and remI on gliding of cells on glass.

Wild-type and mutant cells were examined for movement on glass as previously described (25). Wild-type cells attached to glass and glided at speeds of approximately 2 μm/s. Cells of CJ1883 [Δ(remF remG fjoh_0982)], CJ1899 (ΔremH), CJ1898 (ΔremI), and CJ1988 (ΔremH ΔremI) attached to and moved on glass as well as the wild-type cells did (Fig. 9; see also Movie S1 in the supplemental material). In contrast, cells of CJ1922 (ΔsprB), CJ1913 [Δ(remF remG fjoh_0982) ΔremH], CJ1912 [Δ(remF remG fjoh_0982) ΔremI], and CJ1939 [Δ(remF remG fjoh_0982) ΔremH ΔremI] attached to glass as well as wild-type cells did but were partially defective in movement (Fig. 9 and 10; see also Movies S1, S2, and S3 in the supplemental material). Cells of these mutants changed direction of movement more frequently than did wild-type cells and thus made less net progress. Changes of direction occurred either when the leading pole became the lagging pole or when a cell pivoted or flipped, thus changing direction while maintaining the same leading pole. This behavior is similar to that previously reported for cells with mutations in sprC, sprD, sprB, and sprF (28). Complementation with plasmids that restored colony spreading (Fig. 6 and 8) also restored normal gliding on glass (Fig. 9 and 10; see also Movies S1, S2, and S3 in the supplemental material).

RemF, RemG, RemH, and RemI are not required for surface localization or movement of SprB.

A mutation in sprF, another gene of the sprB operon, results in lack of secretion of SprB across the outer membrane (28). We examined wild-type and mutant cells to determine whether RemF, RemG, RemH, and RemI have roles in secretion of SprB. The presence of SprB on the cell surface was monitored by adding protein G-coated polystyrene spheres and anti-SprB to cells. As previously reported (25), antibody-coated spheres attached specifically to wild-type cells expressing SprB and were rapidly propelled along the cell surface (Table 2; see also Movie S4 in the supplemental material). Such spheres failed to attach to cells of the sprB deletion mutant CJ1922, and protein G-coated spheres without antibodies failed to bind to wild-type cells. Cells of CJ1883 [Δ(remF remG fjoh_0982)] and CJ1939 [Δ(remF remG fjoh_0982) ΔremH ΔremI] produced SprB (Fig. 5). These cells bound to anti-SprB-coated spheres and propelled them similar to wild-type cells (Table 2; see also Movie S4 in the supplemental material), indicating that SprB was present on the cell surface. RemF, RemG, RemH, and RemI are clearly involved in motility, but they are not required for expression, secretion, or movement of the cell surface motility protein SprB.

RemF, RemG, RemH, and RemI are not required for chitin utilization or for bacteriophage sensitivity.

Many motility mutants of F. johnsoniae display defects in chitin utilization and resistance to bacteriophages that infect wild-type cells (5, 12, 13, 22). The defects in chitin utilization are thought to arise because disruption of the motility genes results in defects in the protein secretion system (PorSS), which is involved in secretion of surface components of the motility machinery and of extracellular chitinase (29, 32). Strains with mutations in remF, remG, remH, and remI, including CJ1939, which lacks each of these genes, retained the ability to digest chitin (see Fig. S2 in the supplemental material). This is similar to the phenotype of cells with mutations in sprB, sprC, sprD, and sprF (28) and suggests that RemF, RemG, RemH, and RemI are not components of the PorSS required for secretion of extracellular chitinase.

It has been suggested that F. johnsoniae bacteriophages may use SprB and other cell surface proteins secreted by the PorSS as receptors to infect cells (25, 28, 29). This may explain why sprB mutants are resistant to some bacteriophages and why cells with mutations in genes that are thought to encode core components of the PorSS (gldK, gldL, gldM, gldN, and sprA) are resistant to most or all bacteriophages that infect F. johnsoniae (4, 26, 29). Strains with mutations in remF, remG, remH, and remI, including CJ1939, which lacks each of these genes, were sensitive to infection by each of the bacteriophages tested (see Fig. S1 in the supplemental material). This suggests that RemF, RemG, RemH, and RemI are not receptors for these bacteriophages and are not required for secretion of the receptors.