Specific host translation perturbations promote TnSmu1 loss independently of the canonical ImmA regulator

Abstract

Integrative and conjugative elements (ICEs) are major drivers of horizontal gene transfer and bacterial genome evolution. Although ICE-encoded regulatory circuits have been extensively characterized, the impact of host physiology on the stability of integrated ICEs remains poorly understood. Here, we identify a host-dependent pathway that links specific host translation perturbations to loss of the ICE TnSmu1 in Streptococcus mutans. Analysis of host-gene deletion mutants revealed that disruption of fmt, rnjA, or rnjB—three translation-associated host genes—reproducibly promoted TnSmu1 loss through a mechanism that bypasses the canonical ICE-encoded metalloprotease ImmA but remains dependent on the native attachment site attR. This phenotype was selective, as mutations affecting other essential cellular functions, including protein folding, tRNA modification, cell division, and fatty acid biosynthesis, failed to destabilize TnSmu1 despite undergoing the same experimental evolution and accumulating adaptive genomic changes. Preventing TnSmu1 loss in these translation-associated mutants markedly reduced bacterial growth, whereas loss of the element improved fitness, indicating that ICE elimination alleviates the cost associated with TnSmu1 retention under these conditions. Finally, we show that the relationship between host translation and TnSmu1 stability extends to a genetically distinct S. mutans clinical isolate, although with strain-dependent penetrance. Together, these findings identify host translational state as an important physiological determinant of TnSmu1 stability and reveal that bacterial hosts can influence the maintenance of integrated mobile genetic elements through mechanisms that extend beyond element-encoded regulatory circuits.

Introduction

Horizontal gene transfer (HGT) is a major driver of bacterial evolution, enabling the rapid dissemination of antibiotic resistance, virulence determinants, and metabolic traits(1–3). HGT occurs through three major mechanisms—transformation, transduction, and conjugation—each relying on distinct molecular processes and genetic determinants(4). Among the mobile genetic elements that mediate HGT, integrative and conjugative elements (ICEs) are particularly widespread and play an important role in bacterial genome diversification(2, 5–8). ICEs reside stably integrated within the bacterial chromosome, where they replicate passively with the host genome. Upon activation, they excise from the chromosome, circularize, replicate in the donor cells, are transferred to recipient cells by conjugation, and integrate into the recipient genome, thereby completing their horizontal transmission cycle(2, 7).

The life cycle of ICEs is tightly regulated because it determines both the persistence of the element within its current host and its capacity for horizontal dissemination. Current models have focused primarily on regulatory circuits encoded by the ICE itself. For example, the ImmR/ImmA module controls excision of TnSmu1 and ICEBs1(2, 9, 10), whereas Tn916 employs an Orf9/antisense RNA regulatory system(7). However, host-encoded factors can also influence ICE dynamics. For example, RecA, together with ICE-encoded recombination proteins, promotes inter-ICE recombination and contributes to the formation of mosaic SXT/R391 ICE genomes(11). In addition, ICE excision and replication can engage the host SOS response, demonstrating that ICE activity can be coupled to broader host stress pathways(12). These findings indicate that ICE dynamics are shaped by both element-encoded regulatory circuits and host cellular processes. Whether host physiological pathways actively influence the stability and loss of integrated ICEs, however, remains poorly understood.

Translation is a central and highly regulated component of bacterial physiology (13–15). Changes in the translational machinery, ribosome function, and RNA metabolism can rapidly reshape global protein synthesis and cellular physiology, particularly during adaptation to environmental or metabolic stress(13). Because translation is tightly integrated with cellular growth and homeostasis, perturbation of translation-associated processes could provide a physiological interface through which the host influences the maintenance of integrated mobile genetic elements. However, whether host translational state affects ICE stability or loss remains unclear.

Streptococcus mutans, a primary etiologic agent of dental caries and an opportunistic cause of infective endocarditis, provides an excellent model for addressing this question(9, 10, 16). The oral microbiome is recognized as a hotspot for horizontal gene transfer, and S. mutans harbors multiple mobile genetic elements that contribute to genome diversification(9, 10, 16) . Among these is the ICE TnSmu1. Excision of TnSmu1 is regulated by the conserved ImmR/ImmA repressor system(9, 10). Whether host translation-associated processes also contribute to TnSmu1 stability has not been examined. Building on our previous success in generating viable mutants of genes previously classified as essential(17), we asked whether disruption of core cellular functions(18) affects TnSmu1 maintenance. Working in strain UA159, we found that disruption of three translation-associated host genes—fmt, rnjA, and rnjB—reproducibly promoted adaptive loss of TnSmu1, whereas mutations affecting other essential cellular processes did not. We further show that this host-mediated response requires the attR recombination site but bypasses the canonical ImmA protease, and that preventing TnSmu1 loss under these translation-associated perturbations substantially reduces bacterial fitness. Finally, we demonstrate that the host translation–ICE stability relationship extends to a genetically distinct clinical isolate of S. mutans, although with strain-dependent penetrance. Together, these findings identify host translational state as an important physiological determinant of ICE stability and reveal that core host processes, in addition to ICE-encoded regulatory circuits, govern the maintenance of an integrated mobile genetic element.

Results

Specific host translation perturbations promote TnSmu1 loss

TnSmu1 is a ~20-kb ICE comprising at least 30 predicted open reading frames in S. mutans UA159 (Figure 1A). Under the canonical regulatory model, TnSmu1 activation is controlled by the ImmR/ImmA module, in which the metalloprotease ImmA cleaves the ImmR repressor to initiate excision (9, 10, 16). Excision occurs through site-specific recombination between the left and right attachment sites (attL and attR)(19), which share a 17-bp direct repeat(9). The attL site lies within the 3′ end of the tRNA-Leu gene, whereas attR is located in an intergenic region (Figure 1A) (9).

Figure 1. Analysis of host-gene deletion mutants identifies translation-associated host factors linked to TnSmu1 stability.

(A) Genetic organization of the ICE TnSmu1 in S. mutans UA159. The 17-bp direct repeats corresponding to attL and attR are underlined and shown in uppercase letters, whereas the flanking chromosomal sequences are shown in lowercase. attL is located within the 3′ end of the tRNA-Leu gene, with the tRNA-derived nucleotides highlighted in red, whereas attR resides in an intergenic region. The integrase (int) and excisionase (xis) genes are shown in orange, and the regulatory genes immR and immA are shown in yellow.

(B) Experimental workflow. Ninety-four targeted deletion mutants were constructed. For each mutant, 1–10 independent experimental evolution lineages were established, yielding a total of 207 evolved lineages. Evolved populations were analyzed by whole-genome sequencing. The image was created in BioRender. Bao, L. (2027) https://BioRender.com/61n2m6z

(C–H) Representative whole-genome sequencing read coverage across the S. mutans UA159 chromosome after mapping to the reference genome (NC_004350). Regions marked by asterisks indicate loss of the ~20-kb TnSmu1 element in the canonical control (ΔimmR) and in mutants affecting translation-associated host processes (Δfmt, ΔrnjA, ΔrnjB, and ΔrbfA). Red arrows indicate the locations of the deleted genes. The ΔimmR lineage (C and D) was mapped to a modified reference genome in which immR was replaced by a kanamycin resistance cassette (coordinates 1–2,033,530 bp), whereas the Δfmt (E), ΔrnjA (F), ΔrnjB (G), and ΔrbfA (H) lineages were mapped to the original NC_004350 reference genome (coordinates 1–2,032,925 bp). Vertical red arrow, site of original gene deletion; the height of the blue segments indicates the number of sequence reads mapped to the reference sequence at the coordinates shown on the X axis.

(I) Summary of TnSmu1 retention and loss frequencies across independently evolved mutant lineages. Gene names are shown below each bar, with the corresponding S. mutans UA159 locus tags indicated in parentheses. Blue indicates TnSmu1 retention and orange indicates TnSmu1 loss. TnSmu1 loss was observed in 4/4 ΔimmR (SMU218), 7/7 ΔrnjA (SMU368), 1/5 ΔrbfA (SMU422), 8/8 Δfmt (SMU481), and 3/3 ΔrnjB (SMU1444) lineages, whereas ΔdnaK (SMU82; 0/8), ΔtrmD (SMU868; 0/5), ΔftsX (SMU1324; 0/8), and ΔfabF (SMU1739; 0/6) retained TnSmu1 in all lineages examined.

To identify host factors associated with ICE stability, we targeted 309 genes previously classified as essential in S. mutans UA159 and successfully constructed deletion mutants for 88 of them, together with six nonessential-gene deletion mutants, yielding 94 mutants in total (Figure 1B; Table S1)(18). Notably, many of these viable mutants, including Δfmt, represent, to our knowledge, the first reported deletions in S. mutans of genes previously considered essential (Table S1)(18). For each mutant, one to ten independent lineages were established, generating a total of 207 experimental-evolution populations that were serially passaged and analyzed by whole-genome sequencing (Figure 1B; Table S1).

As expected, deletion of the ICE-encoded repressor gene, immR, resulted in TnSmu1 loss in four independently evolved lineages (Figure 1C–D; Table S1). Remarkably, apart from this positive control, reproducible TnSmu1 loss was confined to a subset of mutants affecting translation-associated host processes (Figure 1E–I; Table S1). Deletion of fmt, encoding methionyl-tRNA formyltransferase(20), resulted in TnSmu1 loss from all eight independently evolved lineages (Figure 1E and 1I; Table S1). Likewise, deletion of rnjA or rnjB, which encode the two RNase J paralogs involved in mRNA maturation (21, 22), resulted in TnSmu1 loss in all seven and three independently evolved lineages, respectively (Figure 1F, G, and I; Table S1). In contrast, deletion of rbfA, encoding a 30S ribosome maturation factor(23), produced only partial penetrance, with TnSmu1 loss occurring in one of five independently evolved populations (Figure 1H and I; Table S1).

To verify that TnSmu1 destabilization resulted specifically from disruption of fmt, we took advantage of a Δfmt mutant that retained a wild-type copy of fmt. This strain, designated Δfmt-WTCR, (for “wild type copy retained” (17)), arose spontaneously during construction of the Δfmt mutant, when the chromosomal region containing fmt duplicated before allelic replacement of one copy of fmt with a kanamycin-resistance cassette. Thus, Δfmt-WTCR retained one wild-type copy of fmt while otherwise undergoing the same allelic-replacement procedure and subsequent strain handling as the Δfmt mutant. It therefore provided an internal genetic control for distinguishing effects caused by loss of fmt from those associated with strain construction or subsequent passage (Fig. S1A–B). Unlike the complete Δfmt mutant, the Δfmt-WTCR retained TnSmu1 throughout serial passage (Fig. S1A–B), supporting the conclusion that TnSmu1 destabilization was associated with loss of fmt rather than with the strain-construction or passage history alone. To exclude potential polar effects on neighboring genes, we analyzed four different constructions of deletion mutants. Both a Δfmt mutant that also removed the 5′ portion of the downstream gene SMU482 (Fig. S1C–D; N=7, Table S1) and a newly generated Δfmt mutant that left SMU482 intact (Fig. S1E–F; N=1, Table S1) exhibited TnSmu1 loss. In contrast, deletion of the downstream gene SMU482 alone (Fig. S1G–H; N=1, Table S1) or the upstream gene SMU480 alone (Fig. S1I–J; N=2, Table S1) did not destabilize TnSmu1. Together, these results support a specific association between fmt disruption and TnSmu1 destabilization and argue against polar effects on adjacent genes as the underlying cause.

To determine whether TnSmu1 loss emerged progressively during serial passage, we examined TnSmu1 abundance, based on sequencing reads aligned to the reference genome, in three independent Δfmt lineages (line_2, line_11, and line_14) at passage 1. Two of the three lineages, line_11 and line_14 (line_11 shown in Fig. S1C–D), exhibited >50% reductions in TnSmu1 coverage relative to the chromosomal background, whereas line_2 showed a <50% reduction (Fig. S1K–L). During subsequent passages, TnSmu1 coverage declined further, with >50% reductions observed by passage 10 in line_11 and line_14 (line_11 shown in Fig. 1E) and by passage 3 in line_2 (Fig. S1E–F). These results indicate that TnSmu1 loss emerged progressively during experimental evolution rather than being uniformly established at the outset of the experiment.

To determine whether TnSmu1 loss simply reflected prolonged serial passage or adaptive genome evolution, we analyzed mutants affecting several additional cellular processes as negative controls, including protein folding (ΔdnaK), tRNA methylation (ΔtrmD), cell division (ΔftsX), and fatty acid synthesis II (ΔfabF). Although these mutants underwent the same experimental evolution protocol and accumulated adaptive genomic changes (Fig. S2A–J), none exhibited TnSmu1 loss (0/8, 0/5, 0/8, and 0/6 evolved lineages, respectively; Figure 1I; Table S1). Whole-genome sequencing revealed multiple adaptive mutations, including recurrent chromosomal copy-number variations outside the TnSmu1 locus (Fig. S2A–J). For example, independently evolved ΔtrmD lineages repeatedly acquired chromosomal amplifications ranging from 3.8 to 185.8 kb that encompassed proS, which encodes prolyl-tRNA synthetase (Fig S2C–J). Similar amplification of chromosomal regions encompassing proS has been implicated in adaptation to impaired tRNA methylation in E. coli(24). Likewise, in our previous study, mutants deleted for any of four genes in the aromatic amino acid synthase pathway consistently acquired chromosomal duplications encompassing a peptide transporter gene cluster (SMU255–SMU259)(25). Despite the obvious selective pressure that led to these adaptive genomic changes, TnSmu1 remained stably integrated in every lineage.

Collectively, these findings demonstrate that TnSmu1 destabilization is not a general consequence of serial passage or adaptive genome evolution. Instead, reproducible TnSmu1 loss is associated with specific perturbations of host translation-associated processes. The affected genes function in distinct processes, including tRNA formylation, RNA metabolism, and ribosome maturation, suggesting that TnSmu1 stability is sensitive to the translational state of the host rather than to disruption of a single molecular step. The absence of TnSmu1 loss in ΔtrmD lineages further indicates that not all perturbations affecting translation-associated functions are sufficient to destabilize the element.

Loss of kanamycin-marked TnSmu1 during serial passage of ΔimmR lineages

Because the ΔimmR allele was marked with a kanamycin resistance cassette in our deletion construction, we independently verified TnSmu1 loss by monitoring the frequency of kanamycin-resistant cells during serial passage. By passages 3 and 7, the wild-type retained strain (ΔimmR-WTCR), which carries a wild-type copy of immR, retained both TnSmu1 and kanamycin resistance (Figure 2A–B; Fig. S3A–D; Table S2). In contrast, all four independently evolved ΔimmR lineages underwent rapid population turnover, with kanamycin-resistant cells declining to 2.8–17.4% by passage 3 and becoming nearly undetectable by passage 7 (below the limit of detection to 0.0008%; Figure 2A–B; Table S2). Whole-genome sequencing of kanamycin-sensitive colonies isolated from passage 7 confirmed loss of the ~20-kb TnSmu1 locus (Fig. S3E–H, N=4). The stable retention of both TnSmu1 and kanamycin resistance in the ΔimmR-WTCR strain supports the conclusion that loss of immR promotes TnSmu1 excision and destabilization. Together, these findings indicate that disruption of ImmR-mediated repression imposes a substantial fitness burden, resulting in strong positive selection for TnSmu1 loss.

Figure 2. immR deletion imposes strong selection for TnSmu1 loss.

(A) Growth of ΔimmR::kan and the ΔimmR::kan-WTCR strain on selective and nonselective media. At passage 3 and 7, cultures from four independently evolved ΔimmR lineages (_11, _12, _15 and _16) and one ΔimmR-WTCR lineage (_17) were serially diluted. Two microliters of each dilution were spotted onto BHI agar with or without kanamycin and incubated anaerobically at 37°C for 48 h. Dilution factors were indicated in the figure.

(B) Quantification of the proportion of kanamycin-resistant and kanamycin-sensitive cells in the evolved ΔimmR and ΔimmR-WTCR lineages at passage 3 and 7. Values represent the mean ± SD from three biological replicates. Statistical significance was determined using one-way ANOVA with Tukey's HSD test; groups sharing a common letter are not significantly different (P > 0.05). Note that the CFU for two ΔimmR lineages (_11 and _15) at P7 was set at the limit of detection (LOD) since no colonies were recovered by assuming that one colony formed in the most concentrated sample that was plated.

TnSmu1 loss following immR deletion requires attR-dependent recombination

Because the ΔimmR populations became heterogeneous for TnSmu1 retention during serial passage, and because ΔimmR cells exhibited smaller colonies than the ΔimmR-WTCR strain on kanamycin-containing plates (Figure 2A) and reduced fitness relative to ΔTnSmu1 cells (Figure 2B), we next quantified the growth of the ΔimmR populations under kanamycin selection. Passage 2 lineages were cultured overnight in BHI containing kanamycin to maintain selection for ΔimmR or ΔimmR-WTCR, then diluted into fresh, selective BHI. Growth monitoring over 12 hours revealed that the ΔimmR lineage grew significantly slower than the ΔimmR-WTCR control (Figure 3A; Table S3), confirming that the disruption of immR causes a distinct fitness defect.

Figure 3. Blocking TnSmu1 loss through attR deletion suppresses the growth defect associated with immR deletion.

(A) Growth kinetics of ΔimmR (line_15), ΔimmR-WTCR (line_17), and ΔimmR ΔattR (line_11) strains. Passage-derived cultures were grown in BHI supplemented with kanamycin, and growth was monitored for 12 h by measuring OD₆₀₀ with a microplate reader. Solid lines represent the mean optical density of triplicate wells (n = 3), and shaded areas indicate the standard deviation (± SD).

(B) Spot-dilution analysis of strains retaining or lacking TnSmu1. Twenty-four-hour cultures were serially diluted, and 2 μL of each dilution was spotted onto BHI agar without antibiotics and incubated anaerobically at 37°C for 48 h. The strains tested were wild-type S. mutans UA159 (WT), a TnSmu1-free derivative isolated from the ΔimmR lineage (ΔTnSmu1; line_11P7_kanfree; Fig. S3E–F), ΔattR, ΔimmR-WTCR (line_17), and two independently passaged ΔimmR ΔattR lineages (line_11 and line_12). Representative whole-genome sequencing profiles of the ΔimmR-WTCR and ΔimmR ΔattR lineages are shown in Fig. S3A–D and Fig. S3I–L, respectively.

(C) Growth curves of WT, ΔTnSmu1, ΔattR, ΔimmR-WTCR, and two lineages of ΔimmR ΔattR. Viable cell numbers were determined by CFU enumeration at 0, 24, and 48 h. Data represent the mean ± SD from three biological replicates.

To determine whether TnSmu1 loss following immR deletion requires attR-dependent recombination, we replaced the attR attachment site with an erythromycin-resistance cassette before introducing the ΔimmR mutation, generating the ΔimmR ΔattR strain. Whole-genome sequencing of two independently passaged lineages confirmed retention of TnSmu1 in the ΔimmR ΔattR background, demonstrating that deletion of attR prevents the loss of the integrated ICE that occurs following immR deletion (Fig. S3I–L). Thus, the canonical ImmR regulatory pathway requires the attR recombination site to eliminate the integrated ICE.

Unexpectedly, preventing TnSmu1 excision also eliminated the severe growth defect associated with ΔimmR. The ΔimmR ΔattR strain exhibited growth comparable to that of the wild type under standard laboratory conditions (Figure 3A–C; Table S3–S4). These findings place immR genetically upstream of attR-mediated recombination and indicate that the pronounced fitness defect caused by loss of immR depends on successful ICE excision rather than on immR deletion alone.

Adaptive TnSmu1 loss improves bacterial fitness under translation-associated perturbations

To determine whether TnSmu1 loss provides a selective advantage under host translation perturbation, we genetically prevented ICE excision by deleting the attR site. In this excision-deficient background (Δfmt ΔattR), TnSmu1 remained stably integrated throughout serial passage (Fig. S4A; Fig. S5A–B). We then assessed bacterial fitness by measuring viable cell counts (CFU mL⁻¹) at 0, 24, and 48 h (Figure 4 A–B).

Figure 4. Retention of TnSmu1 imposes a substantial fitness cost in translation-associated mutant backgrounds.

(A) Cultures of the strains indicated grown for 24 h were serially diluted, and 2 μL of each dilution were spotted onto BHI agar and incubated anaerobically at 37°C for 48 h. Dilution factors are indicated in the figure. Additional details are provided in the Materials and Methods.

(B) Growth curves of WT, ΔTnSmu1, ΔattR, Δfmt ΔTnSmu1, Δfmt-WTCR, Δfmt ΔattR, and Δfmt-WTCR ΔattR. Viable cell numbers were determined by CFU enumeration at 0, 24, and 48 h. Data represent the mean ± SD of three biological replicates.

(C) Cultures of the strains indicated grown for 24 h were serially diluted, and 2 μL of each dilution were spotted onto BHI agar and incubated anaerobically at 37°C for 48 h. Dilution factors are indicated in the figure.

(D) Growth curves of WT, ΔTnSmu1, ΔattR, ΔrnjA ΔTnSmu1, ΔrnjA ΔattR, ΔrnjB ΔTnSmu1, and ΔrnjB ΔattR. Viable cell numbers were determined by CFU enumeration at 0, 24, and 48 h. Data represent the mean ± SD of three biological replicates.

The Δfmt ΔattR strain, in which the host translation defect persisted but TnSmu1 could no longer be eliminated, exhibited the most severe growth defect and the lowest final cell density of all strains examined (Figure 4A–B; Table S5). In contrast, independently evolved Δfmt lineages that had naturally lost TnSmu1 (Figure 1I) showed substantially improved growth (Figure 4A–B; Table S5), indicating that elimination of the ICE partially rescued the fitness defect associated with loss of fmt. As controls, the TnSmu1-free strain (ΔTnSmu1; derived from the ΔimmR mutant) (Fig. S3E–F), the ΔattR single mutant, the Δfmt-WTCR strain (Fig. S1A–B), and the Δfmt-WTCR ΔattR strain (Fig. S4C–D) all exhibited growth comparable to that of the wild-type strain (Figure 4A–B; Table S5).

To determine whether this fitness advantage extended to other translation-associated mutants, we performed the same analysis in ΔrnjA and ΔrnjB backgrounds. Similar to the Δfmt mutant, the excision-deficient strains ΔrnjA ΔattR and ΔrnjB ΔattR exhibited substantially greater growth defects and lower final cell densities than the corresponding passaged ΔrnjA ΔTnSmu1 and ΔrnjB ΔTnSmu1 strains, which had spontaneously eliminated TnSmu1 during experimental evolution (Figure 4C–D; Table S6).

Collectively, these findings demonstrate that TnSmu1 retention imposes a substantial fitness burden in these translation-associated mutant backgrounds, whereas adaptive loss of the ICE alleviates this burden and provides a clear selective advantage to the host.

Host-mediated TnSmu1 loss bypasses ImmA but requires attR-dependent recombination

Because the growth analyses in Figures 3 and 4 indicated that attR is required for both canonical ImmR derepression and host-mediated TnSmu1 loss, we next asked whether these phenotypes were reproducible across independently evolved lineages. Whole-genome sequencing of two to three independently passaged ΔattR lineages confirmed retention of TnSmu1 in ΔimmR ΔattR, Δfmt ΔattR, ΔrnjA ΔattR, and ΔrnjB ΔattR strains (Figure 5A; Fig. S3I–L; Fig. S5A–F; Table S7). These results independently validate that TnSmu1 loss driven by either ImmR disruption or specific host translation-associated perturbations requires the native attR site, consistent with site-specific recombination.

Figure 5. Host-mediated TnSmu1 loss requires attR but bypasses the canonical ImmA pathway.

Whole-genome sequencing was performed on all evolved strains.

(A) TnSmu1 status in the ΔattR background following serial passage.

(B) TnSmu1 status in the ΔimmA background following serial passage.

We next examined whether host-mediated TnSmu1 loss requires the canonical ImmA metalloprotease. As expected, deletion of immA alone did not trigger TnSmu1 loss during serial passage. As a control, we examined a ΔSMU277 ΔimmA strain because disruption of SMU277 did not cause TnSmu1 loss in our initial analysis (Table S1). SMU277 encodes a hypothetical protein, and its deletion provided a control for the effects of introducing a second host-gene mutation into the ΔimmA background. ΔSMU277 ΔimmA likewise retained the integrated TnSmu1. In contrast, introduction of Δfmt, ΔrnjA, or ΔrnjB into the ΔimmA background still resulted in reproducible TnSmu1 elimination following experimental evolution (Figure 5B; Fig. S5G–P; Table S8).

Collectively, these findings confirm across multiple independently evolved lineages that TnSmu1 loss associated with specific host translation perturbations requires the native attR site but bypasses the canonical ImmA pathway. Together with the growth analyses in Figures 3 and 4, these results support a model in which specific host translation perturbations promote TnSmu1 loss through an ImmA-independent but attR-dependent mechanism (Figure 6).

Figure 6. Model for host-mediated destabilization of integrated TnSmu1.

Under the canonical pathway, the TnSmu1-encoded repressor ImmR maintains the element in an integrated state, whereas ImmA-mediated proteolysis of ImmR relieves repression and promotes excision (left). In contrast, disruption of translation-associated host genes (fmt, rnjA, and rnjB) promotes loss of chromosomally integrated TnSmu1 through a pathway that does not require ImmA. This host-mediated pathway may involve reduced ImmR synthesis or an alternative mechanism that destabilizes ImmR, although the molecular mechanism remains to be determined. TnSmu1 loss remains dependent on the native attR site, consistent with a requirement for site-specific excision. The subsequent fate of the excised TnSmu1 molecule was not determined in this study.

Host translation influences TnSmu1 stability across genetically distinct S. mutans strains

To determine whether the relationship between host translation and TnSmu1 stability extends beyond the laboratory strain UA159, we examined fmt, rnjA, rnjB, and rbfA mutants in the genetically distinct bloodstream isolate S. mutans V403, which harbors an endogenous plasmid and causes both caries and endocarditis(27). Deletion of fmt or rnjB resulted in TnSmu1 loss in all independently evolved V403 lineages (3/3 and 4/4, respectively). Deletion of rnjA produced a less penetrant phenotype, with TnSmu1 loss detected in 2 of 5 lineages after extended evolution, whereas no TnSmu1 loss was observed in any of the three rbfA lineages examined (0/3; Fig. S6A–E; Table S9). Thus, disruption of fmt and rnjB reproducibly promotes TnSmu1 loss in two genetically distinct S. mutans backgrounds, while the effects of rnjA and rbfA are more strain- and context-dependent. The delayed and partial phenotype of ΔrnjA in V403, together with the absence of TnSmu1 loss following rbfA deletion in V403 despite its partial phenotype in UA159, indicates that the host translation–ICE stability relationship extends across strains but exhibits strain-dependent differences in penetrance. These differences may reflect variation in the genetic or physiological responses to specific translation-associated perturbations between the two strains.

Discussion

Integrative and conjugative elements have traditionally been viewed as self-regulated mobile genetic elements whose maintenance and dissemination are governed primarily by element-encoded regulatory circuits (5, 6). In this study, we demonstrate that the bacterial host also plays an active role in determining ICE stability. By analyzing a broad collection of host-gene deletion mutants during experimental evolution, we identified host translational state as an important physiological determinant of the maintenance of the endogenous ICE TnSmu1 in S. mutans (Figure 1A–B). Disruption of three translation-associated host genes, fmt, rnjA, and rnjB, reproducibly promoted adaptive TnSmu1 loss, revealing an additional layer of host control over integrated mobile genetic elements.

A notable feature of this response is its specificity. Although mutations in fmt, rnjA, and rnjB perturb translation-associated processes through distinct molecular mechanisms, they all destabilized TnSmu1 (Figure 1C–I). In contrast, control mutants affecting protein folding, tRNA modification, cell division, and fatty acid biosynthesis failed to induce ICE loss despite undergoing the same experimental evolution and accumulating adaptive genomic changes (Fig. S2A–L). These findings argue against a nonspecific stress response and instead suggest that TnSmu1 responds, directly or indirectly, to physiological changes associated with specific perturbations of host translation. Whether this response reflects altered tRNA formylation, ribosome assembly, RNA metabolism, translation efficiency, or another consequence of altered host translational state remains to be determined.

Our genetic analyses further demonstrate that host-mediated TnSmu1 loss occurs through a mechanism distinct from the canonical regulatory pathway. In previously described models, activation of TnSmu1 depends on ImmA-mediated cleavage of the ImmR repressor. In contrast, deletion of fmt, rnjA, or rnjB resulted in TnSmu1 loss even in the absence of ImmA, demonstrating that the canonical metalloprotease is dispensable for host-mediated TnSmu1 loss (Figure 5B). Nevertheless, deletion of the attR recombination site completely abolished TnSmu1 loss in all three translation-associated mutants tested, indicating that host-mediated loss requires the native attR site and is consistent with site-specific recombination rather than nonspecific chromosomal instability or direct deletion of the ICE locus (Figure 5A). Together, these findings support a model in which specific host translation-associated perturbations engage an alternative host-dependent route that bypasses the canonical ImmA regulator while remaining dependent on attR-mediated recombination (Figure 6).

Interestingly, preventing excision by deleting attR produced strikingly different phenotypic outcomes in the two conditions. In the canonical pathway, the ΔimmR ΔattR mutant retained TnSmu1 yet exhibited nearly wild-type growth, indicating that constitutive derepression of the ICE alone imposes little measurable fitness cost when excision is blocked (Figure 3). In contrast, the Δfmt ΔattR, ΔrnjA ΔattR and ΔrnjB ΔattR mutants displayed the most severe growth defects. These defects were substantially alleviated in the corresponding evolved Δfmt, ΔrnjA or ΔrnjB populations that had eliminated TnSmu1, which showed markedly improved fitness (Figure 4). These contrasting phenotypes indicate that the physiological consequences of ICE retention depend on the physiological context in which TnSmu1 excision occurs. Rather than simply derepressing ICE gene expression, these translation-associated perturbations appear to create a host physiological state in which maintaining integrated TnSmu1 becomes particularly costly. Thus, TnSmu1 loss under these conditions provides a fitness benefit by relieving this physiological burden, whereas ImmR derepression in the canonical pathway primarily promotes the normal ICE excision cycle.

The molecular basis of this host-dependent regulatory pathway remains to be elucidated. Altered host translational state could influence the abundance, stability, or activity of ICE regulatory components independently of the canonical ImmA pathway. Changes in protein synthesis or RNA metabolism could affect ImmR abundance or alter the production of the recombination proteins Int and Xis. Alternatively, broader physiological changes associated with these translation-associated perturbations could modify the cellular environment required for efficient ICE recombination. Distinguishing among these possibilities will require direct measurements of ICE transcript abundance, regulatory-protein levels, and recombination activity.

Our results further indicate that the relationship between host translation and TnSmu1 stability extends across genetically distinct S. mutans backgrounds but varies in penetrance. Deletion of fmt or rnjB reproducibly promoted TnSmu1 loss in both the laboratory strain UA159 and the bloodstream isolate V403 (Figure 1; Fig. S6), demonstrating that this host–ICE relationship is not restricted to a single genetic background. In contrast, the effects of rnjA and rbfA were more context dependent. In V403, ΔrnjA produced delayed and incompletely penetrant TnSmu1 loss (Table S9), whereas ΔrbfA failed to promote detectable element loss despite exhibiting a partially penetrant phenotype in UA159. These differences suggest that the effects of specific translation-associated perturbations on ICE stability are influenced by the broader genetic and physiological background of the host. Variation in compensatory pathways, RNA metabolism, ribosome biogenesis, or other strain-specific regulatory features could alter the magnitude or timing of this response. Thus, although the link between host translation and TnSmu1 stability is maintained across distinct S. mutans backgrounds, its penetrance is modulated by host context, further emphasizing that ICE maintenance reflects an interaction between element-encoded functions and host physiology.

More broadly, because the core translation machinery is highly conserved across bacteria and ImmR/ImmA-like regulatory components occur in diverse mobile genetic elements in Gram-positive bacteria, including ICEBs1 in Bacillus subtilis(28), Streptococcus thermophilus (ICESt1 and ICESt3)(29), and Staphylococcus aureus (SaPI3)(30), related host-dependent mechanisms may contribute to the stability and maintenance of mobile genetic elements in other bacterial species. Whether similar perturbations of host translation influence these elements remains to be determined, but our findings raise the broader possibility that core host physiology constitutes an additional layer of mobile-element regulation.

In conclusion, our findings identify host translational state as an important physiological determinant of TnSmu1 stability and demonstrate that bacterial physiology contributes directly to the maintenance of integrated mobile genetic elements. Rather than acting solely as passive carriers of ICEs, bacterial hosts actively influence the balance between ICE maintenance and elimination through core physiological pathways. These results expand current models of ICE regulation and establish a new framework for understanding how host physiology shapes horizontal gene transfer and bacterial genome evolution. Elucidating this host-dependent regulatory pathway may ultimately provide new strategies to limit horizontal gene transfer and reduce the spread of bacterial virulence and antimicrobial resistance(31).

Materials and Methods

Wild-type parental strains and growth conditions

In this study, we primarily utilized Streptococcus mutans strain UA159, obtained from Ann Progulske-Fox (University of Florida). S. mutans strain V403 was provided by R. Facklam, Centers for Disease Control and Prevention and has been maintained as a frozen stock(32). S. mutans strains created in this study are described in the text. All bacterial cultures were routinely cultured in a BHI medium at 37°C under anaerobic conditions (0.2% O₂, 9.9% H₂, 9.9% CO₂, and 80% N₂), using a Coy anaerobic chamber, or stored at −80 °C for long-term storage.

Transformation and mutant isolation:

To prepare competent cells, S. mutans was inoculated into 1 mL of Todd–Hewitt broth supplemented with horse serum (TH+HS) and incubated anaerobically at 37°C overnight. The following day, 100 μL of the overnight culture was transferred into 1 mL of fresh TH+HS medium and incubated for an additional 3 h. Subsequently, 20 μL of this culture was inoculated into 1 mL of fresh TH+HS medium and incubated for 15 min to induce competence prior to the addition of transforming DNA.

Gene deletion constructs were generated by overlap extension PCR(33) using approximately 1-kb upstream and downstream homologous regions flanking the target coding sequence (CDS) together with either a kanamycin (kan) or erythromycin (erm) resistance cassette (primer sequences are listed in Table S10). For transformation, 50–500 ng of purified PCR product was mixed with 200 ng of competence-stimulating peptide (CSP; SGSLSTFFRLFNRSFTQALGK) and 150 μL of competent cells in either 96-well plates or 1.5-mL microcentrifuge tubes. Transformation reactions were incubated anaerobically at 37°C for 24 h. Following incubation, 2–5 μL of each transformation mixture was spotted onto BHI agar containing the appropriate selective antibiotic (kanamycin, 500 μg/mL; erythromycin, 10 μg/mL). After the inoculum had dried, plates were overlaid with 1 mL of molten BHI agar containing the corresponding antibiotic and allowed to solidify. Plates were incubated anaerobically at 37°C for 4–5 days to allow transformant colonies to develop.

Transformants were confirmed by PCR and, when indicated, by whole-genome sequencing. Genomic DNA was isolated from 2 mL of saturated cultures. Cells were harvested by centrifugation at 9,000 × g for 10 min at room temperature and resuspended in 200 μL of resuspension buffer (20 mM EDTA, 200 mM Tris-HCl, and 2% Triton X-100). Cell lysis was performed by adding 200 μL of Buffer AL (Qiagen, Cat. No. 19075). DNA was precipitated with 1 mL of 100% ethanol containing 100 mM sodium acetate, washed, air-dried, and resuspended in 100 μL of nuclease-free water for PCR analysis or whole-genome sequencing.

Strain passage:

Antibiotic-resistant transformants were inoculated into 1 mL of BHI supplemented with the appropriate antibiotic and incubated anaerobically at 37°C for 2–8 days until the culture reached an optical density at 600 nm (OD₆₀₀) of 0.1–0.5. This initial culture was designated passage 0 (P0). A 300-μL aliquot of the P0 culture was then transferred into 3 mL of fresh BHI containing the same antibiotic and grown to saturation to generate passage 1 (P1).

To initiate subsequent passages, the P1 culture was mixed thoroughly by pipetting five times with a P1000 pipette, and 50 μL was transferred into 1 mL of fresh BHI containing the appropriate antibiotic. Cultures were incubated anaerobically at 37°C until reaching an OD₆₀₀ 0.1 to 0.5, at which point the culture was designated as the next passage. This serial passaging procedure was repeated for the indicated number of passages, with each passage maintained in 1 mL of medium. For lineages requiring a larger volume at the final passage, 50 μL of the preceding passage was transferred into 3 mL of fresh BHI containing the appropriate antibiotic and grown to saturation. Passage cultures were preserved at −80°C in BHI containing 20% glycerol. Genomic DNA was isolated from the indicated passages for PCR analysis and/or whole-genome sequencing.

Kanamycin resistance assay of ΔimmR mutant strains.

ΔimmR or ΔimmR-WTCR mutant strains were serially passaged in antibiotic-free BHI to monitor spontaneous loss of the kanamycin-marked TnSmu1 element. Briefly, passage 1 (P1) cultures were inoculated into 1 mL of BHI without antibiotics and incubated for 24 h to generate passage 2 (P2). P2 cultures were stored at −80°C in BHI containing 20% glycerol. Subsequently, 50 μL of each passage was transferred into 1 mL of fresh antibiotic-free BHI and incubated for 24 h to generate the next passage. This serial passaging was continued through passage 6 (P6), after which cultures were also stored at −80°C in 20% glycerol.

To determine the frequency of kanamycin-resistant cells, frozen stocks of P2 and P6 were inoculated into 1 mL of antibiotic-free BHI in triplicate and incubated for 24 h, generating P3 and P7 cultures, respectively. Cultures were serially diluted in phosphate-buffered saline (PBS), and 2 μL of each dilution was spotted onto BHI agar plates with or without kanamycin (500 μg/mL). Plates were incubated anaerobically at 37°C for 48 h, and CFUs were enumerated. The frequency of kanamycin-resistant cells was calculated as the ratio of CFUs recovered on kanamycin-containing plates to those recovered on antibiotic-free plates.

Growth assay of mutant strains

To quantify the growth of ΔimmR and ΔimmR-WTCR strains, passage 2 lineages from the serial-passage experiment described above were cultured overnight in BHI containing kanamycin (500 μg/mL) under anaerobic conditions at 37°C. The resulting overnight cultures (P3) were diluted 1:10 into 150 μL of fresh BHI in 96-well plates containing kanamycin (500 μg/mL). Growth was monitored under aerobic conditions by measuring OD₆₀₀ using an Agilent BioTek Synergy H1 microplate reader at 37°C for 12 h.

For experiments involving translation-associated mutants, cultures were prepared to achieve comparable cell densities before growth assays and spot-dilution analyses because the strains differed substantially in growth rate. Faster-growing strains, including the Δfmt ΔTnSmu1, ΔrnjA ΔTnSmu1, and ΔrnjB ΔTnSmu1 strains, were prepared by inoculating 100 μL of overnight culture into 900 μL of fresh BHI and incubating overnight. Slower-growing ΔattR derivatives, including Δfmt ΔattR, ΔrnjA ΔattR, and ΔrnjB ΔattR, were cultured for 2 days, harvested by centrifugation at 800 × g for 1 min, resuspended in 1 mL of fresh BHI, and incubated for an additional 2 days. For growth assays, cultures were inoculated into 150 μL of fresh BHI in 96-well plates and incubated anaerobically at 37°C. WT, ΔTnSmu1, and TnSmu1-loss derivatives were inoculated at a 1:1,000 dilution, whereas the slower-growing ΔattR derivatives were inoculated at a 1:10 dilution to compensate for their severe growth defects. Viable cell numbers were determined at 0, 24, and 48 h by colony-forming unit (CFU) enumeration.

Whole-genome sequencing and variant calling:

Whole-genome sequencing was performed by SeqCenter (https://www.seqcenter.com/) using 2 × 150-bp paired-end Illumina sequencing. Only genomes with an average sequencing depth of at least 100× were included in downstream analyses.

Raw FASTQ reads were quality trimmed using BBDuk and aligned to the S. mutans UA159 reference genome (NC_004350) using Geneious Prime (version 2025.1.2) (https://www.geneious.com/). Structural variants, including copy-number variations and deletions, were identified using the Geneious Prime variant-calling pipeline and manually inspected.

Data and materials availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary files. The genome sequence data were deposited to BioProject (PRJNA1504854). All bacterial strains generated in this study, including the essential-gene deletion mutants, are available from the authors upon request.