Specificity and Selective Advantage of an Exclusion System in the Integrative and Conjugative Element ICEBs1 of Bacillus subtilis

Horizontal gene transfer is a driving force in bacterial evolution, responsible for the spread of many traits, including antibiotic and heavy metal resistance. Conjugation, one type of horizontal gene transfer, involves DNA transfer from donor to recipient cells through conjugation machinery and direct cell-cell contact.

ABSTRACT

Integrative and conjugative elements (ICEs) are mobile genetic elements capable of transferring their own and other DNA. They contribute to the spread of antibiotic resistance and other important traits for bacterial evolution. Exclusion is a mechanism used by many conjugative plasmids and a few ICEs to prevent their host cell from acquiring a second copy of the cognate element. ICEBs1 of Bacillus subtilis has an exclusion mechanism whereby the exclusion protein YddJ in a potential recipient inhibits the activity of the ICEBs1-encoded conjugation machinery in a potential donor. The target of YddJ-mediated exclusion is the conjugation protein ConG (a VirB6 homolog). Here, we defined the regions of YddJ and ConG that confer exclusion specificity and determined the importance of exclusion to host cells. Using chimeras that had parts of ConG from ICEBs1 and the closely related ICEBat1, we identified a putative extracellular loop of ConG that conferred specificity for exclusion by the cognate YddJ. Using chimeras of YddJ from ICEBs1 and ICEBat1, we identified two regions in YddJ needed for exclusion specificity. We also found that YddJ-mediated exclusion reduced the death of donor cells following conjugation into recipients. Donor death was dependent on the ability of transconjugants to themselves become donors and was reduced under osmoprotective conditions, indicating that death was likely due to alterations in the donor cell envelope caused by excessive conjugation. We postulate that elements that can have high frequencies of transfer likely evolved exclusion mechanisms to protect the host cells from excessive death.

IMPORTANCE Horizontal gene transfer is a driving force in bacterial evolution, responsible for the spread of many traits, including antibiotic and heavy metal resistance. Conjugation, one type of horizontal gene transfer, involves DNA transfer from donor to recipient cells through conjugation machinery and direct cell-cell contact. Exclusion mechanisms allow conjugative elements to prevent their host from acquiring additional copies of the element and are highly specific, enabling hosts to acquire heterologous elements. We defined regions of the exclusion protein and its target in the conjugation machinery that convey high specificity of exclusion. We found that exclusion protects donors from cell death during periods of high transfer. This is likely important for the element to enter new populations of cells.

INTRODUCTION

Integrative and conjugative elements (ICEs) (also called conjugative transposons) play a major role in bacterial evolution by contributing to the spread of genetic material, including genes for antibiotic resistance, pathogenesis, symbiosis, and metabolic functions (1–3). ICEs are typically found integrated into the host chromosome. Under certain conditions, they can excise and transfer to a new host through conjugation machinery encoded by the element (4, 5), thus enabling their spread through a population of bacterial cells. The conjugation machinery encoded by most ICEs is a type 4 secretion system (T4SS) (1), and the genes that confer various phenotypes to the host cells are typically not required for conjugation and are called cargo genes. The conjugation machineries from many ICEs are also capable of transferring (mobilizing) other elements, notably plasmids, to new host cells, allowing the dissemination of elements that do not encode their own conjugation machinery (6–8).

ICEBs1 is relatively small (∼20 kb) and present in a unique site (in trnS-leu2) in most strains of Bacillus subtilis (9, 10). DNA damage to its host cell and crowding by B. subtilis cells that do not contain ICEBs1 both lead to the derepression of transcription of ICEBs1 genes and the subsequent excision and potential transfer of the element. ICEBs1 can be activated in >90% of cells in a population by the overproduction of the element-encoded activator protein RapI, making the element readily amenable to population-based studies (9, 11, 12). ICEBs1 has three known mechanisms for inhibiting its host cell from receiving an additional copy of the element: (i) inhibition of ICEBs1 activation by cell-cell signaling from neighboring cells that already contain a copy of the element (9), (ii) repressor-mediated immunity (13), and (iii) exclusion (12). Exclusion is a key part of conjugative plasmid biology, and most conjugative plasmids appear to have an exclusion system (14). In the F plasmid of Escherichia coli, exclusion protects host cells against lethal zygosis, a phenomenon in which host cells that serve as recipients during excessive transfer events die, likely due to cell wall damage (15–18). In addition, exclusion prevents cells from having recombination events that result in deletions and defective plasmid copies (19–21).

In general, exclusion systems are mediated by a single protein encoded by the element that is localized to the membrane of the host cell, where it is in a position to inhibit cognate conjugation machinery (14). Identified exclusion proteins tend to be fairly small, and membrane attachment is in the form of one or more transmembrane domains, lipid modification, or both. The target protein in the donor has been identified for a variety of exclusion systems, including those from the F/R100 family of plasmids (22, 23), the R64/R62Ia plasmids (24), the SXT/R391 ICEs (25, 26), the IncC plasmids (27, 28), and ICEBs1 (12). ICEBs1 is the only ICE from Gram-positive bacteria that is known to have an exclusion system.

In ICEBs1, the element-encoded exclusion protein YddJ specifically inhibits its cognate conjugation machinery by targeting the conjugation protein ConG in would-be donor cells, thereby inhibiting the transfer of DNA into a cell that already contains ICEBs1 (12). ConG, a homolog of VirB6 in the pTI conjugation system from Agrobacterium tumefaciens, is a membrane protein with seven predicted transmembrane segments and is essential for the function of the ICEBs1 conjugation system (29). Exclusion protects the viability of ICEBs1 host cells under conditions that promote conjugation, although it was not clear whether ICEBs1 donors, recipients, or both were being protected (12).

Here, we identify the regions in YddJ and ConG that determine the specificity of exclusion. We found that exclusion promotes the viability of ICEBs1 donor cells by limiting ICEBs1 transfer from new transconjugants back into the original donors.

RESULTS

Rationale and experimental approach.

Exclusion specificity in ICEBs1 was established using conG and yddJ from ICEBat1 in place of their homologs in ICEBs1 (12). Here, to define the regions of each gene needed to confer specificity, we made chimeras between conG or yddJ from ICEBs1 and conG or yddJ from ICEBat1. We expressed the conG chimeras ectopically in donor strains that contained ICEBs1 ΔconG. To study the effects of exclusion, we used experimental conditions that bypass both cell-cell signaling and immunity. Cell-cell signaling is bypassed by overexpressing rapI from an inducible promoter (Pxyl-rapI) in ICE-containing donor cells (9). Repressor-mediated immunity in potential recipient cells is bypassed by expressing yddJ from an exogenous locus under the control of a strong promoter [Pspank(hy)-yddJ] in the absence of ICEBs1 (12).

Identification of regions of ConG that are essential for exclusion specificity.

Regions and residues of ConG and YddJ that are needed for exclusion specificity must be divergent between the proteins from ICEBs1 and ICEBat1. There are two main regions of divergence in ConG (12). One includes residues 276 to 295 of both ConG_Bs1 and ConG_Bat1 (Fig. 1A) and is predicted to be a loop between the putative third and fourth transmembrane regions. Exclusion-resistant mutations in conG are in this loop region (12). The C-terminal region of ConG is also divergent between the two elements. This region is predicted to be a large extracellular domain.

FIG 1.

Regions of ConG_Bs1 and ConG_Bat1 that confer specificity of exclusion. (A, top) Predicted topology of ConG. The N terminus is predicted to be cytoplasmic. The region of ConG known to confer specificity for exclusion is indicated by the rectangle. (Bottom) Comparison of the indicated regions of ConG from ICEBs1 and ICEBat1. Amino acids that differ between the two proteins in this region are indicated. The two circled residues are sites of mutation that make ConG insensitive to exclusion (12). The bars below the sequence indicate the regions in the hybrid proteins from ICEBs1 (black) and ICEBat1 (gray). (B to E, left) Cartoon of ConG present in the donor strains. Regions of ConG from ICEBs1 (black) and ICEBat1 (gray) are indicated. (B) ConG_Bs1 (KPD225); (C) ConG_Bs1-H1 (KPD136); (D) ConG_Bat1-H2 (KPD135); (E) ConG_Bat1 (KPD224). (Right) Efficiencies of conjugation of the indicated donors (left) into recipients with no YddJ (ICE⁰ [strain CAL89]), YddJ from ICEBs1 (YddJ_Bs1 [strain MA982]), and YddJ from ICEBat1 (YddJ_Bat1 [strain KPD219]). Conjugation efficiency is calculated as the CFU per milliliter of transconjugants divided by the CFU per milliliter of donors at the start of mating, multiplied by 100% to convert to a percentage. Data bars represent averages from three independent experiments. Error bars indicate standard errors.

We found that amino acids 276 to 295 in ConG were sufficient to confer specificity. We replaced amino acids 276 to 295 in ConG_Bs1 with the corresponding residues from ConG_Bat1, generating ConG_Bs1-Bat1(276–295), referred to as ConG_Bs1-H1 (Fig. 1A to C). This hybrid protein was functional in conjugation with the ICEBs1 conjugation machinery. The conjugation efficiency was ∼5% (∼5 transconjugants per 100 initial donors) into recipients that did not contain YddJ (or ICEBs1). When recipients produced YddJ_Bat1, the conjugation efficiency was reduced by a factor of ∼10³, which we refer to as ∼1,000-fold exclusion (Fig. 1C). Exclusion is the ratio of transconjugants into recipients without (no exclusion) versus those with yddJ. This level of exclusion is similar to that observed when both ConG and YddJ were from ICEBat1 (Fig. 1E). In contrast, when recipients produced YddJ_Bs1, there was no detectable change in the conjugation efficiency, giving an exclusion of ∼1 (no exclusion) (Fig. 1C). Based on these results, we conclude that amino acids 276 to 295 of ConG from ICEBat1 are sufficient to confer exclusion specificity to YddJ from ICEBat1.

We also made the reciprocal replacement, replacing residues 276 to 295 from ConG_Bat1 with those from ConG_Bs1 (Fig. 1B and D). This hybrid, ConG_Bat1-Bs1(276–295), referred to as ConG_Bat1-H2, was functional in conjugation with the ICEBs1 conjugation machinery but less so than the wild type (WT) or the other hybrid. The reduced transfer efficiency was expected based on previous analyses substituting ConG_Bat1 for ConG_Bs1 in the context of the ICEBs1 conjugation machinery (12). The efficiency of conjugation was ∼0.1% transconjugants per donor into recipients that did not contain YddJ. When recipients produced YddJ from ICEBs1, exclusion was ∼1,000-fold, similar to that when both ConG and YddJ were from ICEBs1 (Fig. 1D). In contrast, when recipients produced YddJ_Bat1, there was no detectable exclusion (Fig. 1D). Based on these results, we conclude that amino acids 276 to 295 of ConG from ICEBs1 are sufficient to confer exclusion specificity to YddJ from ICEBs1.

Together, the results described above indicate that residues 276 to 295 of both ConG_Bs1 and ConG_Bat1 confer specificity of exclusion. These might not be the only residues that contribute to specificity of exclusion, but without these key residues from the cognate ConG, no exclusion by YddJ is observed. With these key residues, exclusion is virtually indistinguishable from that for donors expressing the wild-type cognate ConG protein.

Identification of YddJ regions essential for exclusion specificity.

To identify regions of YddJ_Bs1 and YddJ_Bat1 that confer exclusion specificity, we used a similar approach of generating hybrids and testing whether recipient strains expressing these hybrid proteins could exclude ICEBs1 using ConG_Bs1 or ConG_Bat1. There are two main regions of sequence dissimilarity between the YddJ homologs (Fig. 2A). We made and tested three functional hybrid proteins, focusing on these regions (Fig. 2B).

FIG 2.

Regions of YddJ_Bs1 and YddJ_Bat1 that confer specificity of exclusion. (A) Protein sequences of the indicated regions of YddJ from ICEBs1 and ICEBat1. Amino acids that differ between the two proteins in these regions are indicated. The bars below the sequence compare the proteins across their entire lengths. Black and white indicate sequence identity and divergence, respectively. (B, left) YddJ in recipients. Portions of YddJ from ICE_Bs1 are indicated in black, and portions from ICE_Bat1 are indicated in gray. (Right) Exclusion of the ICEBs1 conjugation machinery that contains ConG_Bs1 (KPD225) or ConG_Bat1 (KPD224) by the indicated YddJ proteins (left): hybrid J1 (KPD132), hybrid J2 (KPD131), hybrid J3 (KPD128), hybrid J4 (KPD137), YddJ_Bat1 (KPD219), and YddJ_Bs1 (MA982). Exclusion was calculated as the efficiency of conjugation into a recipient without YddJ (no exclusion) (CAL89) divided by that into a recipient expressing the indicated YddJ. Data bars represent averages from three independent experiments, and error bars indicate standard errors.

(i) Hybrid J1, YddJ_Bat1-Bs1(30–48)(65–81)(86–95).

We made constructs that replaced amino acids from regions 1, 2a, and 2b (amino acids 30 to 50, 67 to 82, and 87 to 96) in YddJ_Bat1 with the corresponding residues (amino acids 30 to 48, 65 to 81, and 86 to 95 [note the slightly different numbering]) from YddJ_Bs1 to make hybrid J1, formally known as YddJ_Bat1-Bs1(30–48)(65–81)(86–95) (Fig. 2B). Hybrid J1 was able to exclude an element that had conjugation machinery with ConG_Bs1 (exclusion, ∼1,600) but was unable to exclude conjugation machinery with ConG_Bat1 (Fig. 2B). These results indicate that the specificity of YddJ_Bat1 had been switched to that of YddJ_Bs1 and that these three regions were sufficient to confer specificity.

(ii) Hybrid J2, YddJ_Bat1-Bs1(30–48)(65–81).

We made a construct similar to hybrid J1 but replaced only amino acids in regions 1 and 2a (amino acids 30 to 50 and 67 to 82) in YddJ_Bat1 with the corresponding residues (amino acids 30 to 48 and 65 to 81, respectively) from YddJ_Bs1 to generate hybrid J2, formally known as YddJ_Bat1-Bs1(30–48)(65–81) (Fig. 2B). Hybrid J2 excluded a donor with ConG_Bs1 (exclusion, ∼1,000) but did not exclude a donor with ConG_Bat1 (Fig. 2B). These results indicate that the specificity of YddJ_Bat1 had been switched to that of YddJ_Bs1 and that these two regions were sufficient to change the specificity. They also indicate that amino acids in region 2b (residues 86 to 95) of YddJ from ICEBs1 contributed little if anything to specificity in this context.

(iii) Hybrid J3, YddJ_Bs1-Bat1(30–50)(67–82)(87–96).

We made a construct that replaced amino acid regions 1, 2a, and 2b (amino acids 30 to 50, 67 to 82, and 87 to 96) in YddJ from ICEBs1 with the corresponding residues from YddJ from ICEBat1 to make hybrid J3, formally known as YddJ_Bs1-Bat1(30–50)(67–82)(87–96) (Fig. 2B). This hybrid is essentially the reciprocal of hybrid J1. Hybrid J3 was able to exclude an element that had conjugation machinery with ConG_Bat1 but was unable to exclude conjugation machinery with ConG_Bs1 (Fig. 2B). These results indicate that the specificity of YddJ_Bs1 had been switched to that of YddJ_Bat1 and that these three regions were sufficient to confer specificity.

We made another hybrid (J4) similar to hybrid J3 but that replaced only regions 1 and 2a (not region 2b) of YddJ_Bs1 with the corresponding two regions from YddJ_Bat1. Formally, this hybrid is known as YddJ_Bs1-Bat1(30–50)(67–82). This hybrid had about 100-fold exclusion of a donor with ConG_Bat1 (Fig. 2B), significantly less than that of hybrid J3, and no detectable exclusion of a donor with ConG_Bs1. These results indicate that hybrid J4 is not fully functional for the exclusion of ConG from either element. We believe that these results are largely uninformative about exclusion specificity. The results indicate that hybrid J4 is less functional than hybrid J3. The size of region 2b is different in YddJ_Bs1 than in YddJ_Bat1 (Fig. 2A), and these differences could affect protein folding or stability and highlight the risks with making hybrid proteins. What makes the results with the other hybrids robust and interpretable is that they have activity against ConG from one of the two ICEs. Together, our results indicate that regions 1, 2a, and 2b, and, in the context of YddJ_Bat1, regions 1 and 2a from YddJ_Bs1, are sufficient to confer exclusion specificity.

YddJ hybrid proteins can exclude conjugation machinery with ConG hybrid proteins.

As a final demonstration of specificity, we tested the ability of hybrids J1 and J3 to exclude conjugation machinery containing the ConG hybrids H1 and H2. Hybrid J1 (YddJ_Bat1 with regions 1, 2a, and 2b from YddJ_Bs1) was able to inhibit ConG_Bat1-H2 (ConG_Bat1 with amino acids 276 to 295 from ConG_Bs1) but not ConG_Bat1 (Fig. 3). Likewise, hybrid J3 (YddJ_Bs1 with regions 1, 2a, and 2b from YddJ_Bat1) was able to inhibit ConG_Bat1-H1 but not ConG_Bs1 (Fig. 3). Together, our results demonstrate that the key residues in ConG_Bs1 and ConG_Bat1, and in YddJ_Bs1 and YddJ_Bat1, are sufficient to generate the exclusion specificity of their counterpart wild-type proteins.

FIG 3.

YddJ hybrid proteins can exclude conjugation machinery with ConG hybrid proteins. Conjugation machinery in the donor contained ConG_Bs1-H1 (KPD136), ConG_Bat1-H2 (KPD135), ConG_Bs1 (KPD225), and ConG_Bat1 (KPD224). Recipients expressed YddJ_Bs1 (MA982), YddJ_Bat1 (KPD219), hybrid J3 (KPD128), and hybrid J1 (KPD132). Exclusion was calculated as described in the legend to Fig. 2, with results from matings into recipients that did not contain yddJ (CAL89) that were done in parallel. Data bars represent averages from three independent experiments, and error bars indicate standard errors.

Death of ICEBs1 donors due to excessive mating.

Previous work found that the loss of exclusion leads to a drop in cell viability under conditions that support mating (12). However, it is not known if the drop in viability was due to the killing of donors, recipients, or both. The experiments described below demonstrate that decreased viability of exclusion-defective mutants occurs when cells function concurrently as both donors and recipients.

There was considerable death of ICEBs1 host cells (initial donors) when these cells were surrounded by an excess of recipient cells. We mixed ICEBs1 donors that had exclusion (KPD154) with recipient cells at a ratio of 1 donor to ∼100 recipients. After mating on filters, there was a dramatic drop in the viability of the original donor cells such that only ∼5% (4.6% ± 2.0%) of the original donors survived postmating (Fig. 4). In the absence of exclusion (ΔyddJ [strain KPD155]), only ∼2% (1.8% ± 0.5%) of the original donors survived (Fig. 4), a significant difference based on the one-tailed t test (P = 0.0174).

FIG 4.

Death of donors by excessive mating is exacerbated by loss of exclusion and largely alleviated under osmoprotective conditions. WT ICEBs1 donors were mixed with recipients that lacked an ICE (CAL419) at a ratio of 1 donor to ∼100 recipients and put through mating conditions as described in the legend of Fig. 1. In this case, mating filters were placed on agar plates as described above (regular mating) or with osmoprotection. After incubation, mating mixtures were resuspended either without (regular mating) or with osmoprotection. Percent donor survival was determined by measuring CFU per milliliter after mating compared to that prior to mating. Donors are indicated on the x axis and included ICEBs1 (WT) (KPD154); ICEBs1 ΔyddJ (KPD155); ICEBs1 ΔyddJ overexpressing yddJ from an ectopic locus (KPD156); and ICEBs1 ΔyddJ conGE288K (resistant to exclusion), also overexpressing yddJ from an ectopic locus (KPD157). Data bars represent averages from the three replicate mating assays for each donor, and error bars indicate standard errors. P values from a one-tailed t test were 0.0174 for the wild type compared to the ΔyddJ strain, 0.0052 for the ΔyddJ strain with overexpressed yddJ compared to the conGE288K (exclusion-resistant) strain with overexpressed yddJ, and 8.43 × 10⁻⁴ for the pair compared under osmoprotective conditions.

The decrease in donor survival was due to the loss of exclusion and not the absence of YddJ per se. We analyzed the survival of donors that express YddJ but contain a missense mutation in conG that makes them insensitive to exclusion (12). This mutant also had decreased survival (1.7% ± 0.5%), similar to that of cells without yddJ (Fig. 4). Together, these results indicate that (i) there is significant donor death when the donors are surrounded by a vast excess of recipients and likely to be transferring to multiple cells and (ii) exclusion provides ∼2- to 3-fold protection from this death.

The protection conferred by YddJ was due to its function in the original donors. We expressed yddJ from an ectopic locus in donor cells that contained ICEBs1 ΔyddJ (KPD156). The original donor has exclusion, but a transconjugant will not because the transferred element (ICEBs1 ΔyddJ) is missing yddJ. The survival of donors ectopically expressing yddJ was ∼4% (3.6% ± 0.9%) (Fig. 4) in mating experiments analogous to those described above, with 1 donor to ∼100 recipients. These results indicate that in the absence of exclusion, some of the donor death is likely due to donors acting as recipients in conjugation and that the transconjugants are likely transferring DNA back to the original donors. In donors capable of excluding the entry of a second copy of ICEBs1, donor death is likely from mating events with many recipients.

Exclusion provided a 2- to 3-fold benefit in the survival of donors under the experimental conditions used here. This seemingly small increase in survival could allow ICEBs1 to spread through a population of B. subtilis cells from a few initial donors. Given the high rate of transfer of ICEBs1 (9), particularly in cell chains (30), even a small increase in donor survival should translate to a substantial increase in transfer events. Furthermore, there should be strong selective pressure for anything with this type of survival advantage in nature, providing tremendous selection for ICEBs1 that has exclusion (or that would not cause cell death).

The drop in the viability of donor cells was due to the presence of the mating machinery and the proximity of recipients. Donor death was dependent on the overexpression of rapI to induce ICEBs1. Without rapI induction, there was no detectable drop in donor cell viability under conditions that mimic the mating described above. Furthermore, donor death was not simply due to the overexpression of rapI and the activation of ICEBs1. Matings done with both wild-type ICEBs1 donors (KPD154) and ΔyddJ ICEBs1 donors (KPD155) at a ratio of 1 donor to ∼100 recipients, with overexpression of rapI, but at a cell concentration low enough to reduce mating (∼4 × 10⁵ rather than ∼8 × 10⁸ total cells for mating) had 63% ± 16% and 89% ± 7% survival of wild-type and yddJ donors, respectively. Thus, the large drop in viability was dependent on the activation of ICEBs1 and conditions that support multiple mating events. We postulate that excessive mating, and serving as both a donor and recipient (in the absence of YddJ-mediated exclusion), likely causes cell wall damage that leads to cell death.

Osmoprotective conditions increase survival during excessive mating.

We found that osmoprotective conditions increased donor survival under conditions of excessive mating, both with and without exclusion. Mating assays were done using donors with (KPD154) and without (ICEBs1 ΔyddJ [strain KPD155]) exclusion at a ratio of 1 donor to ∼100 recipients (CAL419) under osmoprotective conditions. These osmoprotective conditions consisted of replacing 1× Spizizen’s salts, used as the support for mating filters and to recover cells from the mating filters, with a solution containing 20 mM MgCl₂ and 0.5 M sucrose buffered with 20 mM maleic acid (see Materials and Methods). This solution (called MSM) has been used for the propagation of B. subtilis cells with no cell wall, so-called L forms (31).

In matings under osmoprotective conditions, the survival of wild-type donors was 51.9% ± 11.0%, and the survival of donors without exclusion (ICEBs1 ΔyddJ) was 17.7% ± 6.3%, compared to 4.6% ± 2.0% and 1.8% ± 0.5%, respectively, in matings without osmoprotection (Fig. 4). Thus, a significant amount of donor death for both wild-type and exclusion-deficient donors was eliminated by osmoprotection. These results indicate that donors surrounded by an excess of recipients are likely dying from cell wall damage due to excessive mating, either into a single recipient or into multiple recipients. Nonetheless, there was still lower survival of donors without exclusion, even under the osmoprotective conditions used here. We suspect that the osmoprotection is not complete because cells were shifted to LB agar (no osmoprotection) from the osmoprotective conditions. We postulate that the donors without exclusion have more envelope damage than donors with exclusion and that the cells with excessive envelope damage did not recover sufficiently to survive the shift to conditions without osmoprotection.

DISCUSSION

The results presented here show that B. subtilis cells that transfer ICEBs1 can die from excessive transfer. This death is exacerbated by the loss of exclusion, which likely enables transfer from a transconjugant back into the original donor. Death of the donors is largely relieved under osmoprotective conditions, indicating that death is likely due to alterations in the integrity of the cell envelope. If there is death of recipients, we would not have detected it in our assays.

Death of ICEBs1 donors compared to lethal zygosis in E. coli.

Death of ICEBs1 donors and protection by exclusion are different from the previously characterized phenomenon of lethal zygosis in E. coli. In lethal zygosis, cells are killed when they serve as recipients during multiple conjugation events. This killing occurs when recipients lacking the F plasmid (F⁻) are mixed with an excess of either Hfr donors or F⁺ exclusion-deficient donors. Recipient death by lethal zygosis also occurs when F⁺ exclusion-deficient recipients are mixed with an excess of Hfr donors (15–18). This killing is probably caused by increased permeability of the cell wall due to multiple matings (15). During Hfr transfer, the transconjugants do not become donors because the entire conjugative element is not transferred. This is in contrast to the situation with ICEBs1, in which transconjugants acquire the entire element and quickly become donors (30). It is not known if E. coli donors also die under conditions of excess mating.

Benefits of ICEBs1 exclusion.

The protective benefit of ICEBs1 exclusion probably serves an important role when ICEBs1 is breaking into a new population of host cells, a situation that is likely mimicked by matings with 1 donor to ∼100 recipients. Once a cell receives an ICE, it is ready to quickly donate it to other cells, and this feature gives ICEBs1 some distinct advantages, like being able to move quickly through cell chains via conjugation (30) and spread in a biofilm (32, 33). That transconjugants quickly become donors indicates that if a mating pair is reasonably stable, then there is the likelihood that a transconjugant could transfer the ICE back to the original donor. Our results indicate that it is taxing for ICEBs1 host cells to serve as donors. There is a considerable amount of donor death with multiple transfer opportunities (1 donor per 100 recipients at high cell concentrations), and there is more death in the absence of exclusion. Whatever the mechanism of killing, it seems that exclusion can protect ICEBs1 donor cells when they are already in the vulnerable state of serving or having just served as donors.

Comparison of exclusion proteins.

Exclusion proteins from different families of conjugative elements have limited sequence similarity but still have some common features. In general, exclusion proteins are relatively small and found on the surface of the host cell, where they are in a position to inhibit cognate conjugation machinery in a potential donor (14). Exclusion proteins are usually not required for conjugative transfer, except for that of R27 (34); often function in a dose-dependent manner; and target a VirB6 (ConG) homolog or analog in the cognate secretion system, as discussed above.

In the case of the F/R100 family of plasmids (E. coli and Shigella flexneri), the exclusion protein TraS is a small hydrophobic protein except for a short hydrophilic region (35) that is predicted to be localized to the inner membrane (23). In the SXT/R391 family of ICEs (Vibrio cholerae and Providencia rettgeri), the exclusion protein Eex is in the inner membrane (26), and paradoxically, the regions essential for exclusion specificity are in a cytoplasm region of the protein (36). For ICEBs1 (B. subtilis), the exclusion protein YddJ is predicted to be extracellular and attached to the cell surface via a lipid modification (12).

Targets of exclusion proteins.

The targets of exclusion proteins from five different families of conjugative elements have been identified. In each case, the exclusion protein targets its cognate VirB6 homolog or analog. The region conferring specificity of exclusion appears to be either periplasmic (23) or cytoplasmic (24, 26, 36). Likewise, the region of ConG of ICEBs1 (and ICEBat1) that confers exclusion specificity is predicted to be cytoplasmic. Together, these analyses indicate that either this normally periplasmic or cytoplasmic region can be present on the cell surface or the cognate exclusion protein has access to part of the periplasm or cytoplasm.

Specificity of exclusion and contributions to ICE biology.

The identification of key regions for exclusion specificity in ConG and YddJ also highlights important aspects of ICEBs1 biology and how ICEs contribute to bacterial evolution by spreading genetic material. The fact that exclusion (or a lack thereof) can be based on differences in a few residues in the exclusion protein or target protein demonstrates that exclusion by a copy of ICEBs1 can very selectively allow slightly different elements (such as ICEBat1) to enter the host cell while significantly reducing the number of conjugation attempts by other would-be ICEBs1 donors.

ICEs play an important role in bacterial evolution by contributing to the spread of genetic material, and one way in which an ICE gains or loses genetic material (which it can then transfer along with itself) is through genetic rearrangement events with other ICEs and plasmids (5). It has been theorized (14) that the lack of exclusion systems in some ICEs allows for the more rapid evolution of the ICE, but this could be harmful to ICEBs1 given its strict requirement for an integration site. Having an exclusion system that allows for as much exposure to other elements as possible, while limiting the number of identical elements that enter, would allow ICEBs1 to have the chance to be exposed to as many other ICEs as possible and benefit from the genetic diversity while avoiding suffering the ill effects.

MATERIALS AND METHODS

Media and growth conditions.

Cells were typically grown in S7₅₀ defined medium (37) supplemented as needed for auxotrophic requirements (40 μg/ml tryptophan, 40 μg/ml phenylalanine, and 200 μg/ml threonine for strains containing alleles inserted at thrC). Isopropyl-β-d-thiogalactopyranoside (IPTG; Sigma) was used at a final concentration of 1 mM to induce expression from the promoter Pspank(hy). LB plates contained the following antibiotics, where indicated: kanamycin (5 μg/ml), spectinomycin (100 μg/ml), streptomycin (100 μg/ml), and a combination of erythromycin (0.5 μg/ml) and lincomycin (12.5 μg/ml) to select for macrolide-lincosamide-streptogramin (MLS) resistance.

Strains and alleles.

B. subtilis strains used in this study are listed in Table 1. The cloning and generation of strains were done according to standard techniques (38). All strains (KPD219, CAL89, MA982, KPD128, KPD131, KPD132, KPD137, and CAL419) used as recipients in mating experiments did not contain ICEBs1 (ICE⁰), contained null mutations in comK or comC (described below), and were streptomycin resistant (str-84) (9, 39). Streptomycin was used as a counterselective marker in mating assays (see more on mating assays below).

TABLE 1.

B. subtilis strains used^a

Strain	Relevant genotype (reference)
CAL419	ICEBs10 str-84 comK::cat (13)
CAL89	ICEBs10 str-84 comK::spc (9)
KPD128	ICEBs10 lacA::{Pspank(hy)-yddJBs1[Bat1(30–50)(67–82)(87–96)] mls} str-84 comK::spc (hybrid J3)
KPD131	ICEBs10 lacA::{Pspank(hy)-yddJBat1[Bs1(30–48)(65–81)] mls} str-84 comK::spc (hybrid J2)
KPD132	ICEBs10 lacA::{Pspank(hy)-yddJBat1[Bs1(30–48)(65–81)(86–95)] mls} str-84 comK::spc (hybrid J1)
KPD135	ICEBs1 ΔconG(5–805) Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI cat) thrC::{Pspank(hy)-conGBat1[Bs1(276–295)] mls} (ConGBat1-H2)
KPD136	ICEBs1 ΔconG(5–805) Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI cat) thrC::{Pspank(hy)-conGBs1[Bat1(276–295)] mls} (ConGBs1-H1)
KPD137	ICEBs10 lacA::{Pspank(hy)-yddJBs1[Bat1(30–50)(67–82)] mls} str-84 comK::spc (hybrid J4)
KPD154	ICEBs1 Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI spc) comC::tet
KPD155	ICEBs1 ΔyddJ Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI spc) comC::tet
KPD156	ICEBs1 ΔyddJ Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI spc) lacA::[Pspank(hy)-yddJBs1 mls] comC::tet
KPD157	ICEBs1 E288K-conG ΔyddJ Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI spc) lacA::[Pspank(hy)-yddJBs1 mls] comC::tet
KPD219	ICEBs10 lacA::[Pspank(hy)-yddJBat1 mls] str-84 comK::spc (12)
KPD224	ICEBs1 ΔconG(5–805) Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI cat) thrC::[Pspank(hy)-conGBat1 mls] (12)
KPD225	ICEBs1 ΔconG(5–805) Δ(rapI-phrI)342::kan amyE::(Pxyl-rapI cat) thrC::[Pspank(hy)-conGBs1 mls] (12)
MA982	ICEBs10 lacA::[Pspank(hy)-yddJBs1 mls] str-84 comK::spc (12)

^aAll strains are derived from our laboratory strain AG174 (JH642) and contain pheA1 and trpC2 mutations (not listed) (43, 44).

All ICEBs1 donor strains contain a version of ICEBs1 that has a kanamycin resistance gene inserted in place of rapI-phrI: Δ(rapI-phrI)342::kan (9). rapI was overexpressed from Pxyl-rapI in donor cells to achieve inducible ICEBs1 gene expression and excision. Pxyl-rapI alleles were integrated into amyE with spc or cat antibiotic resistance genes: amyE::(Pxyl-rapI spc) (40) or amyE::(Pxyl-rapI cat) (40, 41). Any ICEBs1 donor strains containing a deletion of conG, ΔconG(5–805), were derived from MMB1283 (29). KPD210, a donor strain containing a complete deletion of yddJ, was derived from MA11 (12).

(i) Construction of comK- and comC-null mutations.

Null mutations in comK and comC were used to prevent transformation in all recipient strains and in donor strains (KPD154, KPD155, KPD156, and KPD157) used in experiments where even low levels of transformation could significantly alter donor CFU counts. comK::cat was from CAL419 (39), and comK::spc (9) and comC::mls (12) were also previously described.

(ii) Construction of Pspank(hy)-yddJ and Pspank(hy)-yddJ chimeras at lacA.

All yddJ overexpression constructs consist of yddJ fused to the LacI-repressible IPTG-inducible promoter Pspank(hy) and integrated at lacA with an mls resistance gene. Pspank(hy)-yddJ_Bs1 (yddJ from ICEBs1), present in strain MA982, and Pspank(hy)-yddJ_Bat1 (yddJ from ICEBat1), present in strain KPD219, were described previously (12).

To make the yddJ hybrids, yddJ_Bat1 DNA was amplified by PCR from genomic DNA from Bacillus atrophaeus strain 11A1 (from the Bacillus Genetic Stock Center [www.bgsc.org]), and yddJ_Bs1 DNA was amplified by PCR from genomic DNA from B. subtilis strain AG174. Fragments of each yddJ gene were amplified and fused by isothermal assembly as necessary to make four chimeric constructs: (i) hybrid J1, formally known as YddJ_Bat1[Bs1(30–48)(65–81)(86–95)], in which YddJ_Bs1 residues 30 to 48, 65 to 81, and 86 to 95 were substituted for their corresponding YddJ_Bat1 residues (amino acids 30 to 50, 67 to 82, and 87 to 96, respectively); (ii) hybrid J2, formally known as YddJ_Bat1[Bs1(30–48)(65–81)], in which YddJ_Bs1 residues 30 to 48 and 65 to 81 were substituted for their corresponding YddJ_Bat1 residues (amino acids 30 to 50 and 67 to 82, respectively); (iii) hybrid J3, formally known as YddJ_Bs1[Bat1(30–50)(67–82)(87–96)], in which YddJ_Bat1 amino acids 30 to 50, 67 to 82, and 87 to 96 were substituted for their corresponding YddJ_Bs1 amino acids (residues 30 to 48, 65 to 81, and 86 to 95, respectively); and (iv) hybrid J4, formally known as YddJ_Bs1[Bat1(30–50)(67–82)], in which YddJ_Bat1 residues 30 to 50 and 67 to 82 were substituted for residues 30 to 48 and 65 to 81, respectively, in YddJ_Bs1.

For all Pspank(hy)-yddJ constructs, the yddJ PCR fragments were joined together and then joined to two fragments amplified from pCJ80 [a vector for making fusions to Pspank(hy) and integration at lacA] (42) by isothermal assembly. One fragment from pCJ80 included the pCJ80 SphI cut site and 2,409 bp upstream of this cut site, including homology to the 5′ end of lacA. The other fragment included the pCJ80 SacI cut site and 2,299 bp downstream of this cut site, including homology to the 3′ end of lacA. These two fragments were digested with SphI and SacI, respectively, before isothermal assembly with the yddJ PCR DNA. The primers used to amplify yddJ contained sequences complementary to sequences in the primers used to amplify regions of lacA, thereby enabling joining by isothermal assembly. The resulting isothermal assembly product was integrated by double crossover into the chromosome by transformation and selection for MLS resistance to generate the yddJ overexpression alleles.

(iii) Construction of Pspank(hy)-conG and Pspank(hy)-conG chimeras at thrC.

conG was expressed ectopically from the LacI-repressible IPTG-inducible promoter Pspank(hy) from constructs integrated at thrC with an mls resistance gene. The Pspank(hy)-conG alleles were used to complement the ΔconG(5–805) deletion in ICEBs1. Pspank(hy)-conG_Bs1 (conG from ICEBs1), present in strain KPD225, and Pspank(hy)-conG_Bat1 (conG from ICEBat1), present in strain KPD224, have been described previously (12, 29).

To make the conG hybrids, conG_Bat1 DNA was amplified by PCR from genomic DNA from B. atrophaeus strain 11A1 (from the Bacillus Genetic Stock Center [www.bgsc.org]), and conG_Bs1 DNA was amplified by PCR from genomic DNA from B. subtilis strain AG174. Fragments of each conG gene were amplified and fused by isothermal assembly as necessary to make two chimeric constructs: (i) ConG_Bs1-H1, formally known as ConG_Bs1[Bat1(276–295)], in which ConG_Bat1 residues 276 to 295 were substituted for residues 276 to 295 in ConG_Bs1, and (ii) ConG_Bat1-H2, formally known as ConG_Bat1[Bs1(276–295)], in which ConG_Bs1 residues 276 to 295 were substituted for residues 276 to 295 in ConG_Bat1.

For all Pspank(hy)-conG constructs, the conG PCR fragments were joined together and then joined to two fragments amplified from pMMB1341 (29) by isothermal assembly. One fragment from pMMB1341 included the HindIII cut site and the adjacent 2,330 bp upstream of this cut site, which includes sequences from the 3′ end of thrC. The other fragment included the SphI cut site and the adjacent 1,867 bp downstream, which includes sequences from the 5′ end of thrC. These two fragments were digested with HindIII and SphI, respectively, before isothermal assembly with the conG PCR DNA. The primers used to amplify conG contained sequences complementary to sequences in the primers used to amplify regions of thrC, thereby enabling joining by isothermal assembly. The resulting isothermal assembly product was integrated by double crossover into the chromosome by transformation and selection for MLS resistance to generate the conG overexpression alleles.

Mating assays.

Mating assays were performed essentially as described previously (9, 39). Donor and recipient cultures were grown in S7₅₀ defined minimal medium supplemented with 0.1% glutamate and 1% arabinose until they reached the mid-exponential growth phase and then diluted back to an optical density at 600 nm (OD₆₀₀) of 0.1. At this point, 1% xylose was added to donor cultures to induce the expression of Pxyl-rapI. After 2 h of xylose induction, donor and recipient cells were mixed and poured over a nitrocellulose filter under vacuum filtration. Unless otherwise indicated, equal numbers of donor and recipient cells were used (∼4 × 10⁸ cells each). Filters were incubated for 3 h at 37°C on 1.5% agar plates containing 1× Spizizen’s salts [2 g/liter (NH₄)SO₄, 14 g/liter K₂HPO₄, 6 g/liter KH₂PO₄, 1 g/liter Na₃ citrate·2H₂O, 0.2 g/liter MgSO₄·7H₂O] (38). Cells were resuspended from the filters, serially diluted in 1× Spizizen’s salts, and plated on LB agar plates containing kanamycin and streptomycin to select for transconjugants. For matings done under osmoprotective conditions, 1× Spizizen’s salts (in mating plates and resuspension media) was replaced with MSM (20 mM MgCl₂ plus 0.5 M sucrose buffered with 20 mM maleic acid) (31). The number of viable ICEBs1 donor cells (CFU per milliliter) was determined at the time of donor and recipient cell mixing by serial dilution plating on LB agar plates containing kanamycin. The mating efficiency was calculated as the percent transconjugant CFU per milliliter per donor CFU per milliliter (at the time of mixing donors with recipients). The fold exclusion was calculated as the percent transfer into an ICE⁰ recipient divided by the percent transfer into an ICE⁰ recipient that was expressing yddJ.