Dissemination of Tn916-Related Integrative and Conjugative Elements in Streptococcus pneumoniae Occurs by Transformation and Homologous Recombination in Nasopharyngeal Biofilms

ABSTRACT Multidrug resistance in Streptococcus pneumoniae (or pneumococcus) continues to be a global challenge. An important class of antibiotic resistance determinants disseminating in S. pneumoniae are >20-kb Tn916-related integrative and conjugative elements (ICEs), such as Tn2009, Tn6002, and Tn2010. Although conjugation has been implicated as the transfer mechanism for ICEs in several bacteria, including S. pneumoniae, the molecular basis for widespread dissemination of pneumococcal Tn916-related ICEs remains to be fully elucidated. We found that Tn2009 acquisition was not detectable via in vitro transformation nor conjugative mating with donor GA16833, yielding a transfer frequency of <10−7. GA16833 Tn2009 conjugative gene expression was not significantly induced, and ICE circular intermediate formation was not detected in biofilms. Consistently, Tn2009 transfer efficiency in biofilms was not affected by deletion of the ICE conjugative gene ftsK. However, GA16833 Tn2009 transfer occurred efficiently at a recombination frequency (rF) of 10−4 in dual-strain biofilms formed in a human nasopharyngeal cell bioreactor. DNase I addition and deletions of the early competence gene comE or transformation apparatus genes comEA and comEC in the D39 recipient strain prevented Tn2009 acquisition (rF of <10−7). Genome sequencing and single nucleotide polymorphism analyses of independent recombinants of recipient genotype identified ~33- to ~55-kb donor DNAs containing intact Tn2009, supporting homologous recombination. Additional pneumococcal donor and recipient combinations were demonstrated to efficiently transfer Tn916-related ICEs at a rF of 10−4 in the biofilms. Tn916-related ICEs horizontally disseminate at high frequency in human nasopharyngeal S. pneumoniae biofilms by transformation and homologous recombination of >30-kb DNA fragments into the pneumococcal genome. IMPORTANCE The World Health Organization has designated Streptococcus pneumoniae as a priority pathogen for research and development of new drug treatments due to extensive multidrug resistance. Multiple strains of S. pneumoniae colonize and form mixed biofilms in the human nasopharynx, which could enable exchange of antibiotic resistance determinants. Tn916-related integrative and conjugative elements (ICEs) are largely responsible for the widespread presence of macrolide and tetracycline resistance in S. pneumoniae. Utilizing a system that simulates colonization of donor and recipient S. pneumoniae strains in the human nasopharynx, efficient transfer of Tn916-related ICEs occurred in human nasopharyngeal biofilms, in contrast to in vitro conditions of planktonic cells with exogenous DNA. This high-frequency Tn916-related ICE transfer between S. pneumoniae strains in biofilms was due to transformation and homologous recombination, not conjugation. Understanding the molecular mechanism for dissemination of Tn916-related ICEs can facilitate the design of new strategies to combat antibiotic resistance.

S treptococcus pneumoniae (or pneumococcus) is a Gram-positive bacterial human pathogen and inhabitant of the human nasopharynx. Colonization with pneumococci occurs during early childhood, with carriage rates ranging from 20 to 93.4% and persisting into adulthood at lower levels (1)(2)(3). S. pneumoniae may remain as a commensal in the nasopharynx, spread to other mucosal sites, and cause otitis media or pneumonia, or it may invade the bloodstream, resulting in bacteremia or meningitis (4). Individuals at risk for invasive pneumococcal disease are children less than 2 years of age, those over 65 years of age, and immunocompromised individuals with other underlying chronic conditions (5). Although new, effective pneumococcal conjugate vaccines are available, invasive pneumococcal disease still poses as a global public health concern. In 2017, the World Health Organization (WHO) recognized S. pneumoniae as a priority pathogen due to extensive multidrug resistance (6). The widespread use of antibiotics, such as macrolides for treatment of community-acquired pneumonia and other upper respiratory infections, has led to marked increases in S. pneumoniae antibiotic resistance (7,8). The horizontal transfer of resistance determinants followed by the selection of resistant strains is the major driver for the development of antibiotic resistance in S. pneumoniae, yet the molecular mechanisms of resistance determinant dissemination have not been clearly defined.
Large, mobile genetic elements known as integrative and conjugative elements (ICEs) of the Tn916 family are responsible for tetracycline and macrolide resistance in S. pneumoniae. Tn916-related ICEs that have been widely identified in S. pneumoniae include Tn2009 (23.5 kb), Tn6002 (20.8 kb), and Tn2010 (26.3 kb) (9)(10)(11)(12)(13). Although prototype Tn916 (18.0 kb) and other classes of ICEs are disseminated via conjugation in several Gram-positive species, such as Bacillus subtilis, Enterococcus faecalis, Clostridium difficile, and many streptococcal species (14)(15)(16), the role of conjugation as an efficient horizontal exchange process for Tn916-related ICEs in S. pneumoniae is unclear. While conjugation of S. pneumoniae Tn6002 into Streptococcus pyogenes has been demonstrated by mating experiments at the low frequency of 10 28 , Tn2009 and Tn2010 did not conjugate into S. pneumoniae (10,11). Another common horizontal genetic exchange strategy is transformation as S. pneumoniae naturally becomes competent for extracellular DNA uptake during growth (17). Transformation has long been associated with S. pneumoniae genome evolution (18,19), particularly for acquisition of point mutations and small ;1to ;5-kb determinants conferring antibiotic resistance.
To investigate the molecular mechanism for horizontal dissemination of Tn916related ICEs among S. pneumoniae strains, we utilized a bioreactor system consisting of a dual-strain pneumococcal biofilm on human nasopharyngeal cells (20) with Tn2009 as a model. Gene expression, mutation studies in key conjugative and transformation genes, ICE circular intermediate quantification, genome sequencing, and single nucleotide polymorphism (SNP) analyses revealed that transformation and homologous recombination are responsible for efficient horizontal transfer of S. pneumoniae Tn2009. These mechanistic observations were applicable to additional Tn916-related ICEs and S. pneumoniae strains.

RESULTS
Classic in vitro transformation failed to demonstrate acquisition of Tn916-related ICEs by S. pneumoniae. S. pneumoniae is naturally competent for extracellular DNA (eDNA) uptake. Classic in vitro transformation assays were used to examine uptake of genomic DNA carrying ICEs. Planktonic cells of D39 Str or TIGR4, treated with the cognate synthetic competence-stimulating peptide (CSP) to initiate competence development, were incubated with various amounts of genomic DNA. Control reactions for competency demonstrated that D39 Str acquired trimethoprim (Tmp) resistance (folA/ I100L) (21) at a recombination frequency (rF) of 4.08 Â 10 25 6 4.33 Â 10 25 per mg DNA, whereas TIGR4 obtained streptomycin (Str) resistance (rpsL/K56T) (22) at a rF of 1.03 Â 10 25 6 7.88 Â 10 26 per mg DNA. However, in vitro transformation of D39 Str under analogous conditions with genomic DNA harboring Tn2009 (23.5 kb) or Tn2010 (26.3 kb) yielded no recombinants with tetracycline (Tet) resistance, and the rFs were ,1.13 Â 10 27 6 1.04 Â 10 27 and ,2.68 Â 10 28 for Tn2009 and Tn2010, respectively. Similarly, no Tet-resistant TIGR4 recombinants were recovered with genomic DNA carrying Tn2009 (rF of ,3.75 Â 10 28 6 3.63 Â 10 29 ). Agarose gel electrophoresis confirmed that purified genomic DNA preparations contained fragments significantly larger than the ICEs (data not shown), and increasing the CSP concentrations from 100 ng/mL to 5000 ng/mL in in vitro transformations did not enhance recombination frequencies for uptake of Str resistance. Thus, classic in vitro transformation aided by exogenous CSP did not result in uptake of Tn916-related ICEs by D39 nor TIGR4.
Efficient transfer of pneumococcal Tn916-related ICEs occurred in dual-strain biofilms formed on human nasopharyngeal cells. A continuous flow bioreactor system, characterized by formation of dual-strain biofilms on a confluent monolayer of human pharyngeal Detroit 562 cells at 35°C has previously been shown to yield high rF for transfer of Str, Tmp, and Tet (tetM-mediated) resistance between D39 and TIGR4 (20). Thus, we investigated pneumococcal Tn916-related ICE dissemination using the bioreactor system with a 6-h incubation, which was previously found to be sufficient for efficient recombination. The ICE donor clinical isolates GA16833 Tet/Ery (serotype 19F with Tn2009) or GA47281 Tet/Ery (serotype 19F with Tn2010), were coinoculated in the bioreactor with designated recipient D39 (serotype 2) derivatives containing either Str (D39 Str ) or dual Str and Tmp (D39 Str/Tmp ) resistance.
When GA16833 Tet/Ery served as the Tn2009 donor, the rFs of Tet1Str resistance were 2.60 Â 10 24 6 2.08 Â 10 24 for recipient D39 Str/Tmp (Fig. 1) and 1.34 Â 10 24 6 7.02 Â 10 25 for recipient D39 Str . The rF of 10 24 obtained in the bioreactor system represented an ;1,000-fold enhancement over the in vitro rF of , 10 27 where no recombinants were recovered. Similarly, with coinoculation of the recipient D39 Str/Tmp and the Tn2010 donor GA47281 Tet/Ery , we obtained a rF of 1.34 Â 10 24 6 1.62 Â 10 24 for Tet-resistant recombinants of D39 Str/Tmp , once again demonstrating an ;1,000-fold enhancement over the in vitro transformation (Fig. 1). Additional selection combinations, Ery1Str, Tet1Ery1Str, and Ery1Str1Tmp, were also tested and yielded similar rFs (see Table S3 in FIG 1 Efficient transfer of pneumococcal Tn916-related ICEs occurs in dual-strain nasopharyngeal biofilms. The wild-type recipient D39 Str/Tmp (serotype 2) and GA16833 Tet/Ery (with Tn2009, serotype 19F) or GA47281 Tet/Ery (with Tn2010, serotype 19F) were coinoculated in a bioreactor at 35°C on a confluent monolayer of human nasopharyngeal Detroit cells such that dual-strain biofilms formed. After a 6-h total incubation, recombination frequencies for D39 uptake of the tetracycline (Tet) resistance marker were calculated. the supplemental material). When Tet was used for recombinant selection, Tn2009 and Tn2010 yielded comparable rFs (10 24 ). However, when Ery was included in the selection, higher rF values were observed for Tn2010 relative to those obtained with Tn2009 (Table  S3). This difference could be due to the constitutively expressed ermB on Tn2010, whereas Tn2009 carries inducible Ery resistance on the macrolide efflux genetic assembly (Mega) element.
Conjugation was not involved in the highly efficient transfer of Tn916-related ICEs (>20 kb) between pneumococci in bioreactor biofilms. Transcriptional regulation of conjugation genes and the molecular mechanism resulting in Tn916 conjugative transfer have been well characterized (23). Conjugation of Tn916 is stimulated by tetracycline at the tetM promoter, which mediates antisense mRNA derepression of downstream regulatory genes (23,24) and subsequently results in expression of the xis and int genes encoding essential enzymes for the excision of Tn916 (23,24) as well as the formation of the Tn916 circular intermediate (CI). CI formation occurs via binding of coupling sequences on the 59 and 39 ends of the ICE (23,25,26). This physical association of the 59 and 39 ends of the circularized Tn916 allows for upregulated transcription at the 39 end of Tn916 to extend through to the conjugative genes at the 59 end, encoding the necessary conjugative machinery, such as the type IV secretion system. Subsequent passage of a single-stranded Tn916 DNA from a donor to the recipient cell occurs (25), resulting in the integration of Tn916 into the recipient genome via precise site-specific recombination (27). While conjugation of Tn916 and other ICEs has been demonstrated in Gram-positive bacteria, we obtained several lines of evidence to exclude conjugation as the mechanism responsible for Tn2009 dissemination among S. pneumoniae as described below.
Conjugative gene expression in Tn2009 was limited. We performed BLASTN analysis and demonstrated that all three pneumococcal Tn916-related ICEs contained genes encoding the excisionase, Xis, and the integrase, Int, which are required for Tn916 excision and site-specific recombination (26,28,29). Tn916-related ICEs also contained open reading frames (ORFs) that had .99% sequence identity to the corresponding ORFs in Tn916 involved in conjugative transfer (Fig. 2).
Expression of xis and int is coupled with tetracycline-inducible tetM transcription and subinhibitory concentrations of tetracycline, which result in enhanced conjugative transfer of Tn916 in E. faecalis and B. subtilis (23,30). We first compared the basal expression of conjugative genes in pneumococcal Tn2009 with that of the prototype Tn916 in a B. subtilis donor strain, CMJ253 (31). Expression of tetM, int, orf20 (relaxase), FIG 2 Sequence alignments of prototype Tn916 with pneumococcal Tn916-related ICEs. EasyFig 2.2.2 was utilized to produce alignments of ICE elements using comparison files generated by BLASTN. Tetracycline resistance is conferred by the tetM gene (purple arrow) found in prototype Tn916 as well as pneumococcal ICEs. Pneumococcal Tn916-related ICEs harbor macrolide resistance via the ermB element (blue arrows) found on Tn6002 or Tn2010 as well as the macrolide efflux genetic assembly (Mega) element (yellow arrows) on Tn2009 or Tn2010. Conjugative genes shared by prototype Tn916 and pneumococcal Tn916-related ICEs are denoted by magenta arrows.
Dissemination of S. pneumoniae Tn916-Related ICEs Microbiology Spectrum and ftsK in Tn2009 was significantly lower: tetM and int were expressed 19.7-and 32.4fold lower, respectively, while orf20 and ftsK were expressed about 300-fold lower than the corresponding genes in Tn916. Tetracycline induces transcriptional upregulation of conjugative genes at the 59 end of Tn916 (23). To investigate if such a regulatory coupling was present in pneumococcal Tn2009, we grew the Tn2009-containing GA16833 strain (Tet MIC of 24 mg/mL) and the Tn916-containing CMJ253 strain in the presence or absence of 2.5 mg/mL tetracycline for 2.5 h (27). As expected for Tn916, there was a 4-fold induction of tetM, and similar levels of induction were detected for orf20 and ftsK (Table 1). There was no induction of int expression in Tn916 (Table 1), consistent with the observation by Celli et al. using Northern blotting (25). In contrast, while tetM expression in Tn2009 was induced nearly 28-fold by a sublethal concentration of tetracycline, no concerted upregulation of conjugative genes was detected and only ;2.3to ;3.7-fold changes were seen in int, orf20, and ftsK (Table 1) relative to no-tetracycline controls. Thus, unlike Tn916, conjugative gene expression was not coupled to tetM induction in Tn2009.
We also investigated if conjugative gene expression in the Tn2009 donor was induced in the dual-strain biofilm of recipient D39 Str and donor GA16833 DcomCDE::cat Cm/Tet/Ery (BASP1), where Tn2009 transfer was detected at high frequency. When normalized to broth cultures of the single donor strain BASP1 or a 1:1 mixture of donor and recipient strains, there was no induction of conjugative genes in biofilms, with a fold change of 1.13 or 1.04 for int and 0.28 or 0.94 for orf20, respectively. These data suggested that conjugative gene expression from pneumococcal Tn2009 was minimal and not induced under the bioreactor biofilm conditions, and thus was unlikely to support efficient Tn2009 transfer.
There was no detectable circular intermediate formation of Tn2009. Circular intermediate (CI) formation resulting from excision from the donor chromosome and circularization of Tn916, mediated by Xis and Int proteins, is a prerequisite for Tn916 conjugative transfer and is induced by tetracycline (25,26). The circular junction is the same between the prototype Tn916 and Tn2009. Thus using quantitative PCR (qPCR), we quantified CIs derived from Tn2009 in GA16833 and Tn916 in CMJ253 with tetracycline induction. The CI copy number was normalized to the chromosomal copy number of ftsK. The positive control, B. subtilis donor CMJ253, produced a Tn916 CI/chromosome ratio of 2.50 Â 10 22 6 1.42 Â 10 22 , while the S. pneumoniae donor GA16833 yielded no detectable Tn2009 CIs, with a ratio of ,2.48 Â 10 27 6 1.47 Â 10 27 .
Mating reactions did not support conjugative transfer of Tn2009. Transfer of large mobile genetic elements via conjugation has been observed in mating experiments  3B).
Mutation of ftsK had no effect on the transfer of Tn2009 in mixed nasopharyngeal biofilms. During conjugation, Orf21 (FtsK) of Tn916 serves as a coupling protein for translocating the DNA through the secretion system from the donor to recipient cell (33)(34)(35). Given the critical role of a coupling protein in the conjugative machinery, an ftsK mutant was created in the donor strain (GA16833 DftsK::cat Tet/Ery [BASP3]). Coinoculation of donor BASP3 with recipient D39 Str in the bioreactor yielded a rF of ;10 24 , similar to that obtained when wild-type GA16833 served as the Tn2009 donor. Thus, disruption of conjugation via the ftsK mutation had no impact on the transfer of Tn2009 (Fig. 3C), supporting that conjugation played no significant role in the highly efficient dissemination of Tn2009 under the bioreactor biofilm conditions.
Competence development and transformation machinery in the pneumococcal recipient were required for efficient ICE uptake in human nasopharyngeal biofilms. Natural competence for DNA uptake via transformation is an important mechanism for horizontal gene transfer in S. pneumoniae. Early competence development depends on expression of the comCDE operon, encoding a competence-stimulating peptide (CSP), ComC, a histidine kinase, ComD, and a response regulator, ComE (19,36). Recipient BASP2 (D39 DcomE), confirmed by in vitro transformation to be incompetent in acquiring point mutations, was examined in the bioreactor with donor GA16833. As shown in Fig. 4A, no Tet-resistant recombinants were observed (rF of ,1.21 Â 10 27 6 6.35 Â 10 28 ), suggesting that recipient competence initiation was critical for Tn2009 uptake.
The involvement of two late competence proteins, ComEA and ComEC, was also examined. ComEA is a DNA receptor that binds double-stranded DNA captured by the type IV pilus as it retracts, and the ComEC protein channel subsequently imports single-stranded DNA fragments (37). The comEA and comEC genes, which are transcribed in tandem, were deleted and replaced with the cat cassette in strain D39 Str . The recipient D39 DcomEA/EC::cat Str double mutant (BASP4), when coinoculated with donor GA16833 in the bioreactor, yielded no detectable Tet-resistant recombinants and a rF of ,8.34 Â 10 28 6 4.16 Â 10 28 (Fig. 4A). When recovered from bioreactor experiments, BASP2 and BASP4 recipients were at similar cell counts to that of donor GA16833, indicating that rF reductions were not caused by growth defects or lower mutant densities relative to the donor strain (Fig. 4B).
As extracellular DNA (eDNA) levels influence transformation efficiency, we recovered eDNA in spent medium from the bioreactor experiments (20) and quantified eDNA concentrations of each strain using qPCR with serogroup-specific primers. Comparable eDNA concentrations were recovered from both wild-type donor GA16833 and wild-type recipient D39 Str (Fig. 4C) as well as for wild-type donor GA16833 and recipient BASP2 (Fig. S1). Additionally, the consequence of eDNA degradation was investigated by treating the biofilm with exogenous DNase I during a 6-h incubation. Compared to the parallel positive control of no treatment (rF of 2.60 Â 10 24 6 2.08 Â 10 24 ), DNase I treatment resulted in no recombinant recovery, with an estimated rF of ,5.34 Â 10 27 6 4.34 Â 10 27 (Fig. 4D). Thus, eDNA degradation eliminated Tn2009 transfer.
The increase in rF of Tn2009 transfer in the bioreactor relative to that observed by in vitro transformation implied differential competence development under these two conditions, and we hypothesized that competence gene expression would be higher in the bioreactor environment. As no synthetic CSP was added in the bioreactor, the comparison was made with in vitro transformation reactions without CSP added.
Coinoculating in the bioreactor a D39 Ery/Str recipient and the BASP1 donor with a DcomCDE deletion to avoid detecting donor com gene expression, comD and comE expression in the D39 recipient was about 120-fold greater in the bioreactor biofilm than under the in vitro transformation condition (Fig. 4E). Interestingly, this recipient upregulation of early com genes in the bioreactor biofilm was also much greater than that of classic in vitro transformation reactions, where 100 ng/mL synthetic CSP induced 83-and 57-fold increases in comD and comE expression, respectively, compared to reactions without CSP addition. The higher recipient competence gene expression in the donor GA16833 and recipient D39 mixed biofilm formed in the absence of exogenously added CSP supported a more robust competent state of D39 recipient cells relative to in vitro transformation conditions, which corroborated the efficient biofilm-mediated transfer of Tn2009.
Integration of intact Tn916-related ICEs into the recipient pneumococcal genome occurred by homologous recombination. The assumption that the Tet-resistant recombinants with the recipient genotype had integrated the entire Tn2009 was confirmed by whole-genome sequencing (WGS). To analyze the extent of Tn2009 integration in the recipient genome, four Tn2009 recombinants independently recovered from selections on Tet1Str, Ery1Str, Tet1Ery1Str, and Ery1Str1Tmp were first characterized in detail. Quellung reactions and serotype-specific conventional PCRs confirmed that the recombinants were serotype 2 of the D39 recipient. Multilocus sequence typing (MLST) analysis of seven housekeeping genetic loci (38) showed all recombinants were sequence type 595 (ST595) of recipient D39 and not ST236 of donor GA16833 ( Table 2), confirming that the recombinants were of the D39 genetic lineage and not a result of GA16833 undergoing capsule switching. WGS of D39 recombinants was performed to probe the extent of recombination. WGS data confirmed that all four recombinants indeed harbored the entire 23.5-kb Tn2009 integrated at the same genomic locus as the GA16833 donor. There was an ;9.5-kb genome sequence of D39 replaced by Tn2009 in the GA16833 genome, and this expected deletion was also confirmed in the recombinants. Based on donor-specific SNP distributions, donor DNA fragments of various sizes flanking Tn2009 were detected in the recombinant genomes. As shown in Table 2 (Table 2). Thus, these Tn2009-containing D39 recombinants incorporated donor DNA fragments that ranged from ;33 kb to ;55 kb. Genomes of two Tn2010-containing D39 recombinants, independently selected on Tet1Str and Ery1Str after the coinoculation with donor GA47281, were also examined. Donor DNA fragments of ;38.4 and ;34.5 kb containing the intact Tn2010 were identified, respectively (Table  S4). Conjugative transfer of prototype Tn916 is expected to result in site-specific recombination with a precise excision and integration of Tn916 flanked by coupling sequences, thus incapable of transferring additional flanking SNPs from the donor genome (39). The variable donor DNA lengths indicated that integration of Tn2009 and Tn2010 into pneumococcal genomes occurred by homologous recombination.
Additionally, we identified two clinical isolates, GA47179 (serotype 15A) and GA44194 (serotype 19A), each containing an incomplete copy of Tn6002 (;17.0 kb), that retained both ermB and tetM. Tn6002 in these isolates was truncated ;100 bp downstream of tetM, thus missing critical conjugative xis and int genes. Bioreactor experiments with donor GA44194 recovered only serotype 19A recombinants, indicating that GA44194 acquired Str resistance more efficiently from D39 Str leading to a rF of ,4.09 Â 10 28 6 3.40 Â 10 28 for uptake of Tet resistance by the designated D39 recipient (Fig. 5). Utilizing recipient D39 Ery/Str and donor GA44194 DcomCDE::cat Str (BASP8), we detected that comD expression in the D39 recipient was about 10.8-fold greater, while comE expression showed no changes (0.7-fold) in the biofilm compared to the in vitro transformation condition without synthetic CSP (Table 3). Thus, when coinoculated with donor BASP8, the induction of com gene expression in the D39 Ery/Str recipient was significantly lower than that obtained with donor BASP1 (Fig. 4E). These data suggested a less competent D39 recipient with the GA44194 donor strain, potentially contributing to the lower rF. When examined with the bioreactor system, Tet resistance of another partial Tn6002 donor, GA47179, was transferred to recipient D39 Str at a rF of 3.44 Â 10 25 6 1.69 Â 10 25 (Fig. 5). WGS and SNP analyses of a Tet1Str-selected D39 recombinant found that Tn6002 was incorporated within an estimated donor length of ;101.0 kb (Table S4). The strain-dependent, efficient transfer of an incomplete Tn6002 lacking critical conjugation genes again supported the horizontal transfer of large ICE-containing DNA fragments between pneumococcal strains via transformation and homologous recombination. Congression, the cotransformation of distinct unlinked fragments of DNA into the same cell, can occur during transformation. Although rare, congression has been previously demonstrated in S. pneumoniae (40). To detect additional independent recombination events distant from integration of the ICE-containing fragments, we conducted whole-genome variant analyses. SNPs transferred from the donor GA16833 genome into the recombinants were identified, and the presence of multiple consecutive GA16833 SNPs flanked by extensive recipient D39 sequence was considered incorporation of a donor DNA fragment and, thus, a possible transformation event. The outermost 59 and 39 donor SNPs were then used to estimate the minimal donor DNA length recombined into the recipient genome. Multiple possible cotransformation events at various genomic locations in each of the four Tn2009 recombinant genomes were detected (Fig. 6).
Efficient transfer of Tn916-related ICEs occurred between other S. pneumoniae donors and recipients in human nasopharyngeal biofilms. To assess whether Tn916related ICE dissemination via transformation was widely applicable in S. pneumoniae, additional Tn2009-and Tn2010-containing S. pneumoniae clinical isolates were studied in the bioreactor using the Tet1Str selection. Tn2009 was integrated in SP_1638 (TIGR4 annotation) in GA49542 (serotype 9V), distinct from GA16833 with Tn2009 integrated in SP_1947 (TIGR4 annotation). When donor GA49542 was coinoculated with recipient D39 Str in the bioreactor, we observed a rF of 4.71 Â 10 24 6 1.77 Â 10 24 (Fig. 7A) for Tn2009 transfer, similar to the rF obtained with donor GA16833. We also investigated another strain, GA44288 (serotype 19A), with Tn2010 incorporated into the same genomic locus as in GA47281 (serotype 19F). The rF for recipient D39 Str uptake of Tn2010 from donor GA44288 was 1.02 Â 10 24 6 5.96 Â 10 25 , comparable to that obtained from donor GA47281 (Fig. 7B). WGS analysis of three independent Tet1Str recombi-FIG 5 Partial pneumococcal Tn6002 lacking conjugative regulatory mechanism transfers via transformation and homologous recombination in a strain-dependent manner. Wild-type recipient D39 Str and GA44194 Tet/Ery (with partial Tn6002, serotype 19A) or GA47179 Tet/Ery (with partial Tn6002, serotype 15A) were coinoculated in the bioreactor at 35°C. After a 6-h total incubation, recombination frequencies for D39 uptake of the tetracycline (Tet) resistance marker were calculated.  (Table S4). Thus, Tn2009 and Tn2010 were transferred from multiple donor strains to recipient D39 efficiently (rF of ;10 24 ), indicating that the dissemination of Tn916-related pneumococcal ICEs in biofilms was independent of donor genomic lineages and the genomic location of ICEs.
We also sought to determine if the two serotype 19A Tmp-resistant clinical isolates, GA40410 and GA43265, examined as recipients in mating experiments, could acquire ICEs as efficiently as D39. To prevent trimethoprim resistance marker uptake by the donor, we used the competence-deficient donor BASP1. Tn2009 was efficiently taken up by recipient GA40410 at a recombination frequency of 9.87 Â 10 24 6 2.80 Â 10 24 , while recipient GA43265 incorporated Tn2009 at a rF of 6.92 Â 10 24 6 2.41 Â 10 25  (Table S4). Together, these data demonstrated efficient Tn916-related ICE transfer between multiple S. pneumoniae strains in a dual-strain, nasopharyngeal biofilm via transformation and homologous recombination.

DISCUSSION
The continued emergence of multidrug resistance in S. pneumoniae limits treatment options for invasive pneumococcal disease. Thus, a WHO initiative, launched in 2017, emphasizes research into pneumococcal antibiotic resistance mechanisms and development of new antipneumococcal agents (6). The mechanism of high-frequency horizontal transfer of .20-kb ICEs of the Tn916 family, such as Tn2009, Tn6002, and Tn2010, which carry the important resistance genes tetM, mefE/mel, and/or ermB in S. pneumoniae, was addressed in the present study.
During colonization, which can last for months, the pneumococcus forms highly organized biofilms on the epithelial surface of the human nasopharynx (41). Approximately 50% of children carrying S. pneumoniae can be colonized by two different pneumococcal serotypes, and up to five cocolonizing pneumococcal serotypes can be detected at any one time (42)(43)(44). We found that .20-kb Tn916-related ICEs horizontally disseminate at high frequencies between S. pneumoniae in dual-strain biofilms (rF of 10 24 ) established on a human nasopharyngeal cell monolayer and in a continuous flow bioreactor system, which mimics the microenvironment of the human respiratory epithelium. Previous work using the bioreactor has shown high rFs for the exchange of smaller antibiotic resistance determinants between S. pneumoniae strains (20,45).
Conjugation is a transfer mechanism for ICEs in several bacterial species, including S. pneumoniae. Examples of interspecies conjugation include those of pneumococcal ICEs Tn6002 (20.8 kb) into Streptococcus pyogenes (cF of ;10 28 ) (10), Tn6003 (25.1 kb) into Enterococcus faecalis (cF of ;10 27 ) (10), and Tn1207.3 (52.5 kb) into S. pyogenes (cF of ;10 23 ) or Streptococcus gordonii (cF ;10 24 ) (46). Additionally, the large composite Tn5253 (64.5 kb) conjugates to pneumococcal recipients at cFs ranging from ;10 27 to ;10 24 (47), and Tn5251, usually found within Tn5253, has been shown to conjugate independently from a pneumococcal Tn5253 donor to S. pneumoniae TIGR4 (cF of ;10 25 ) (15). While we demonstrated Tn916 conjugation between B. subtilis strains at frequencies of 10 25 , conjugative transfer of a pneumococcal Tn916-related ICE, Tn2009, was Tetracycline induces expression of conjugative genes on Tn916, resulting in formation of circular intermediates and subsequent transfer of Tn916 to recipient cells (24,25). However, conjugative gene expression in Tn2009 was not induced by tetracycline, likely due to insertion of the Mega element downstream of tetM, disrupting the regulatory coupling between tetM and conjugative genes. In addition, we did not detect Tn2009 CI formation and Tn2009 transfer in nasopharyngeal biofilms was not affected by a deletion of the critical conjugative gene ftsK. Collectively, these data demonstrated that conjugation is not the underlying mechanism for the high-frequency transfer of pneumococcal ICEs in biofilms on human nasopharyngeal cells.
Another potential horizontal gene transfer method is transduction mediated by bacteriophages. Prophages are abundantly present in pneumococcal genomes and contribute to virulence and pneumococcal physiology (48,49). However, the importance of phage transduction in the spread of antibiotic resistance in S. pneumoniae is much less clear. Wyres et al. identified a pneumococcal 1968 isolate, 18C/3, carrying a Tn916-related ICE that is associated with a streptococcal phage similar to Streptococcus phage 040922 (50). We found no evidence that Tn2009, Tn6002, and Tn2010 carried by the donor strains examined in this study were part of or were associated with phage elements. Furthermore, the variable sizes of donor DNA fragments containing ICEs identified in the recombinant genomes did not support phage-mediated transfer of Tn916-related ICEs.
Transformation is the major mechanism of horizontal genetic exchange in naturally competent S. pneumoniae. In vitro transformation experiments using mixtures of planktonic pneumococci, exogenous synthetic CSP, and purified DNA generally yield frequencies of 10 24 to 10 26 with the uptake of ;2to ;6-kb DNA fragments (51)(52)(53)(54). Our data reproduced these observations, in which recipients D39 and TIGR4 were transformed with point mutation-mediated Tmp or Str resistance at frequencies of 10 25 per mg of DNA. However, analogous experiments using purified ICE-containing genomic DNA did not yield Tet-resistant transformants (rF , 10 27 ), and this was not caused by a DNA fragment size limitation.
Previously, Cowley et al. noted that an environment of close cell-to-cell contact, which consisted of pneumococcal coincubation on filters or mixed cultures forming mature biofilms, led to transformation of 8-to 30-kb DNA fragments (40). Additionally, serotype switching events were a result of transformation of 22-to 39-kb DNA fragments (55). We found that within biofilms on human nasopharyngeal cells, Tn916related ICEs transferred efficiently to pneumococcal recipients, resulting in acquisition of these large antibiotic resistance elements at a rF of ;10 24 . Similar results were shown with multiple S. pneumoniae donors, including two donors of Tn2009, two of Tn2010, and one of Tn6002, as well as with four recipient strains, D39 Str/Tmp , D39 Str , GA40410 Tmp , and GA43265 Tmp .
Efficient transfer of Tn916-related ICEs likely requires close contact between donor and recipient S. pneumoniae strains in mixed biofilms (40), an environment likely found during nasopharyngeal colonization (56). The temperature maintained for the bioreactor (35°C) is lower than the in vitro transformation experiments (37°C). In vitro transformation conducted at 35°C resulted in a lower rF for streptomycin resistance uptake than that at 37°C, while no uptake of Tn2009-encoded tetracycline resistance was detected at either temperature (see Fig. S3 in the supplemental material). Thus, the difference in temperatures is unlikely to account for the efficient ICE transfer within dualstrain nasopharyngeal biofilms. Preliminary data from single-strain recipient D39 Tmp or D39 Str biofilms formed on nasopharyngeal cells and exposed to extracellular genomic DNA supplied in the flow medium resulted in the uptake of streptomycin resistance at a rF of 1.66 Â 10 26 per mg DNA, but not of Tn2009-mediated tetracycline resistance (rF of ,7.68 Â 10 28 per mg DNA) (Fig. S2). These data further support that close interaction between different donor and recipient strains within the dual-strain biofilms is needed for efficient transformation of Tn916-related ICEs. Induction of the competent state in S. pneumoniae initiates DNA release from a subfraction of different S. pneumoniae populations, potentially via cell lysis (57), and fratricide is critical for efficient gene transfer between donor and recipient pneumococci in biofilms (58). Overall, this close contact, donor-recipient interactive environment is absent in the classic transformation of planktonic cells.
Whole-genome sequencing and minimum recombination junctions defined by SNP analysis confirmed the integration of very large donor DNA fragments containing intact ICE elements. For Tn2009, donor fragments were estimated to range from ;33 to ;55 kb in size, while Tn2010 was transferred on ;34to ;45-kb DNA from two donor strains. Finally, we observed that the partial Tn6002 ICE missing critical conjugation genes recombined on a donor fragment size of ;101 kb. Evidence of multiple homologous recombination events, or congression, was also detected in several Tn2009-containing recombinants.
S. pneumoniae develops a naturally competent state through a positive feedback mechanism involving the ComCDE signaling cascade (19,36). Upon pneumococcal colonization of the upper respiratory tract, biofilms are formed with upregulation of competence genes (59)(60)(61). We detected 120-fold increases of comD and comE expression in the D39 recipient recovered from the mixed nasopharyngeal biofilm compared to planktonic D39 cultures. Deletions of comE or comEA and comEC in the D39 recipient eliminated recovery of Tn2009-containing D39 recombinants. The elimination of recombinants after DNase I treatment further supported transformation as the mechanism responsible for efficient ICE transfer in the biofilms.
In conclusion, efficient Tn916-related ICE dissemination in S. pneumoniae was demonstrated in human nasopharyngeal cell biofilms via transformation and homologous recombination of large DNA fragments. High-frequency transfer of Tn916-related ICEs in biofilms occurred in multiple combinations of pneumococcal donors and recipients. A biofilm environment with close contact of pneumococcal cells and the consequent upregulation of the competence pathway were critical for supporting horizontal dissemination of .20-kb Tn916-related ICEs and the antibiotic resistance determinants of these elements. Efficient transformation in mixed biofilms that mimic a natural nasopharyngeal colonization environment supports the epidemiological observations of widespread dissemination of S. pneumoniae ICEs in the pneumococcal population.

MATERIALS AND METHODS
Bacterial strains, culture media, and antibiotics. The strains of S. pneumoniae used in this study are listed in Table S1 in the supplemental material. Clinical isolates were provided by the Georgia Emerging Infections Program. All S. pneumoniae strains were cultured on blood agar plates or with Todd-Hewitt broth with yeast extract (THY broth) and grown at 37°C with 5% CO 2 . B. subtilis strains were cultured on LB plates or LB broth at 37°C. Where indicated, the following antibiotics were utilized for preparation of antibiotic agar plates: tetracycline (1 or 2 mg/mL), streptomycin (100, 200, and 220 mg/ mL), trimethoprim (14 mg/mL), chloramphenicol (4.5 mg/mL), erythromycin (0.5 mg/mL), and spectinomycin (100 mg/mL). All antibiotics were purchased from Millipore-Sigma (Saint Louis, MO).
Mutant construction. Mutation constructs were created by sequential overlapping PCR or splicing by overlap extension (SOE) PCR. GA16833 Tet/Ery served as the parental strain for the conjugation mutant (DftsK), while D39 Str and D39 Str/Tmp were the parental strains for competence (DcomE or DcomCDE) and transformation (DcomEA/EC) mutants. The primers utilized are listed in Table S2. The chloramphenicol resistance gene, cat, was amplified from pEVP3 (62). Purified genomic DNA was utilized as the template to amplify the upstream (59-end) and downstream (39-end) regions of the target gene using the corresponding primers carrying the necessary overlapping sequences. These individual DNA fragments were amplified using Q5 high-fidelity polymerase (New England Biolabs, Ipswich, MA). Mixtures of two fragments (59 end plus cat or cat plus 39 end) were used as the template for the first round of overlapping PCRs. The third fragment was then linked by a second PCR using either One Taq (New England Biolabs) or Taq (Roche Diagnostics, Indianapolis, IN) DNA polymerase plus Taq Extender PCR additive (Agilent, Santa Clara, CA) to generate the final construct. PCR products purified with the Zymo DNA Clean and Concentrator kit (Irvine, CA) were sequenced to confirm the desired mutation. Purified PCR constructs (0.5 to 1 mg) were used to transform precompetent cells by in vitro transformation. Transformants selected on 4.5 mg/mL chloramphenicol were sequenced to confirm the presence of the intended mutation.
To generate the DftsK mutant, an upstream region was amplified with BSA17 and BSA18 and a downstream region was amplified with BSA19.1 and BSA20. The chloramphenicol resistance marker cat was amplified with primers EVP3_CmF and EVP3_CmR. SOE PCR was performed using three individual In vitro transformation assay. D39 and TIGR4 pneumococcal strains were made precompetent using standard methods (63). Briefly, an overnight plate culture was used to inoculate complete transformation medium (CTM) and grown to an optical density at 600 nm (OD 600 ) of 0.6 to 0.7. This primary culture was used to make a 1:20-diluted secondary culture, which was grown to OD 600 of 0.35 to 0.45 (midlog phase). Glycerol was then added to the competent cell aliquots at a final concentration of 10% (vol/ vol) and stored at 280°C. These precompetent pneumococcal cells were transformed in CTM using 500 ng of purified genomic DNA and 100 ng/mL of synthetic CSP1 or CSP2 in 200-or 300-mL total reaction volumes. The CSPs were synthesized by Millipore-Sigma (Saint Louis, MO), the Emory University Microchemical Facility (64), or GenScript (Piscataway, NJ). Recombination frequencies for in vitro transformation reactions were calculated as number of transformants on antibiotic selection plate divided by the total population of S. pneumoniae cells recovered on nonselective blood agar plate and normalized to micrograms of transforming DNA.
Human nasopharyngeal biofilm bioreactor coinoculations. A confluent monolayer of Detroit 562 cells (ATCC CCL-138) was grown on a Corning Snapwell with a 0.4-mm-pore polyester membrane (Corning, NY). These Snapwells were placed inside a sterile vertical diffusion chamber from the bioreactor, allowing the Detroit 562 cells to rest on the apical side (inner chamber) and to be perfused with flowing bioreactor medium composed of 1Â MEM supplemented with 5% FBS, 1% nonessential amino acids (100Â), 1% L-glutamine (200 mM), and 1% HEPES buffer (1 M). The flow of bioreactor medium at a rate of ;0.2 mL/min was generated by a Cole Parmer Master Flex L/S peristaltic pump (Vernon Hills, IL). Where indicated, 20 U/mL of DNase I was added to the bioreactor medium. Spent medium (collected for eDNA quantification as described below) and planktonic cells exited the bioreactor chamber through a parallel outlet located at the top of the chamber (20).
For the coinoculation, overnight plate cultures of each pneumococcal strain were washed three times with 1Â Dulbecco's phosphate-buffered saline (DPBS) and resuspended in 500 mL of 1Â DPBS, and the OD 600 was measured. Appropriate volumes of bacteria were mixed to make a suspension of equal densities at OD 600 of 0.1 (10 6 CFU/mL each) and inoculated through the apical perfusion path of the bioreactor chamber. After a static 1-h incubation at 35°C to allow for S. pneumoniae adherence to Detroit 562 cells, the flow of medium was initiated and continued for another 5 h. At the conclusion of the incubation period, the Snapwells were removed and the dual-strain biofilms formed on the Detroit 562 cells were gently washed and sonicated for 20 s in a Branson ultrasonic water bath (Danbury, CT). The bacteria were suspended by extensive pipetting and vortexing, and serial dilutions in 1Â DPBS were performed. To obtain the total population for each strain in the coinoculations, serial dilutions were plated on antibiotic plates unique to each individual strain. The cells of the total recombinant population were plated on dual-or triple-antibiotic selection plates. The recombination frequencies (rFs) for bioreactor experiments were calculated as follows: (total number of recombinants Â serotype proportion of recombinants)/total population for specific serotype (20).
DNA extraction and serotype-specific qPCR. Recombinants recovered from bioreactor experiments were pooled in 200 mL of sterile 1Â DPBS. A lysis buffer containing 100 mL of TE buffer (10 mM Tris, 1 mM EDTA [pH 8.0]), 40 mg/mL of lysozyme, and 75 U/mL of mutanolysin was added to each of the samples and incubated at 37°C for 1 h. Two hundred microliters of Qiagen buffer AL and 20 mL of Qiagen proteinase K were added, followed by incubation at 56°C for 30 min. DNA was extracted using the Qiagen QIAamp DNA Mini Kit (Valencia, CA) per the manufacturer's instructions. DNA was eluted in 100 mL of elution buffer and used for serotype-specific qPCR with the primers and probes listed in Table  S2 as well as the Bio-Rad SsoAdvanced Universal Probes supermix (Hercules, CA). These reactions targeted the capsule (cps) locus of specific pneumococcal serotypes. The cycling conditions were as follows: 1 cycle at 95°C for 2 min, followed by 39 cycles of 95°C for 15 s and 60°C for 30 s. A standard curve for final genome equivalents per milliliter of each serotype was performed alongside the samples and consisted of serially diluted DNA standards corresponding to the following genome equivalents: 8.58 Â 10 6 , 4.29 Â 10 6 , 2.14 Â 10 6 , 4.29 Â 10 5 , 4.29 Â 10 4 , 4.29 Â 10 3 , 4.29 Â 10 2 , 4.29 Â 10 1 , 2.14 Â 10 1 , and 2.14 (42,65). The proportion of recombinants that belong to each serotype was calculated by dividing the number of genome equivalents for a specific serotype by the total sum of genome equivalents for both strains in the bioreactor coinoculations, and this value was utilized to calculate recombination frequencies (rFs).
Alignments of pneumococcal Tn916-related ICEs with prototype Tn916. BLASTN and -outfmt 6 were utilized for generating comparison files between ICEs. Along with GenBank (.gbk) files, the comparison files were inputted into EasyFig 2.2.2 (66) to produce sequence alignments as well as to determine percentages of identity.
Gene expression studies with qRT-PCR. For tetM and conjugative gene expression studies, singlestrain broth cultures of S. pneumoniae strains and B. subtilis CMJ253 were grown in a primary culture to an OD 600 of 0.50 to 0.65. The primary cultures were then diluted 1:20 to inoculate a secondary culture. The secondary cultures were incubated in the presence or absence of 2.5 mg/mL tetracycline for 2.5 h at 37°C. One milliliter of cultures (;10 8 CFU) was pelleted at 16,400 rcf for 8 min at room temperature. Five hundred microliters of supernatant was discarded, and the remaining 500 mL supernatant was used to resuspend the bacterial pellets. One milliliter of Qiagen's RNAprotect reagent was added, and the mixture was incubated at room temperature for 5 min. For single-donor strain or mixed donor-recipient broth cultures, S. pneumoniae strains were first grown separately in broth to OD 600 of about 0.4. A mixture of 10 8 CFU of both strains or a single-strain culture was incubated for another 2.5 h and then treated with 2 vol of RNAprotect reagent. For bioreactor experiments that underwent a 1-h static incubation and 5-h continuous flow incubation at 35°C, total biofilm bacteria suspended in 1Â DPBS were collected at the end of the 6-h incubation period and treated with 2 vol of RNAprotect reagent, while in vitro transformation reactions with or without synthetic CSP were incubated at 37°C for 2 h and then treated with 2 vol of RNAprotect reagent. Following incubation, the samples were centrifuged at 12,000 rcf for 10 min, the supernatant was discarded, and the pellets were stored at 280°C.
Qiagen RNeasy Mini Kit's protocol was followed, with the exception that 40 mg/mL of lysozyme and 75 U/mL of mutanolysin were used in TE buffer for bacterial cell lysis. RNA samples were treated with the TURBO DNA-free kit (Gaithersburg, MD) and subsequently purified and concentrated with the Zymo RNA Clean and Concentrator kit (Irvine, CA). All RNA samples were verified to be free of DNA contamination via conventional PCR using primers targeting S. pneumoniae recA or B. subtilis orf20 (relaxase) (BSA25 and BSA26). cDNA was prepared using the Bio-Rad iScript reverse transcription supermix (Hercules, CA) and 350 ng, 500 ng, 800 ng, or 1 mg of purified RNA. Quantitative reverse transcription-PCR (RT-PCR) was conducted with the Bio-Rad iQ SYBR green supermix (Hercules, CA). All primers for the target and internal control genes were validated by the threshold cycle (2 2DDCt ) method (67). The following cycle conditions were utilized on a Bio-Rad CFX96 Touch real-time PCR machine (Hercules, CA): 1 cycle at 95°C for 3 min, 40 cycles of 95°C for 15 s, 57°C or 60°C for 15 s, and 72°C for 30 s.
Fold change in expression was calculated using the 2 2DDCt method (67). For the tetracycline-induced and uninduced broth cultures of S. pneumoniae strains and B. subtilis CMJ253, 16S rRNA was utilized as the internal control and fold change was normalized to the uninduced condition. Due to 16S rRNA being present in both wild-type donor and recipient S. pneumoniae strains in the bioreactor, the internal control for conjugative gene expression was the donor-specific cat gene from BASP1. When analyzing expression of com genes specifically from the recipient strain, expression of ermB inserted in the nonessential bgaA locus of the D39 Ery/Str recipient served as the internal control for expression analysis of bioreactor biofilms.
Quantification of ICE circular intermediates. The formation of circular intermediates was examined under the following conditions: (i) broth cultures treated with or without 2.5 mg/mL tetracycline for 2.5 h at 37°C, (ii) 4-h mating reaction mixtures consisting of GA16833:BASP2 or CMJ253:CAL419 on a blood agar plate, and (iii) the bioreactor experiment including D39 Str and GA16833. One milliliter of each broth culture was pelleted down at 16,400 rcf for 10 min at room temperature, and the supernatants were discarded. The bacterial pellets were then resuspended in 200 mL of 1Â DPBS for DNA extraction. For the mating experiment, bacteria on the blood agar plate were collected in 400 mL of 1Â DPBS and aliquots of 200 mL were utilized for DNA extraction. Finally, dual-strain biofilms were collected from the bioreactor and underwent DNA extraction. DNA was extracted with the Qiagen QIAamp DNA Mini Kit (Valencia, CA). qPCR of the circular junctions was performed with the Bio-Rad iQ SYBR green supermix (Hercules, CA). A standard curve using genomic DNA from B. subtilis strain LDW737 (27), containing a cloned copy of the circular junction sequence that is shared between the Tn916 prototype and the pneumococcal Tn916related ICEs, was performed with the following genome equivalents: 8.58 Â 10 6 , 8.58 Â 10 5 , 8.58 Â 10 4 , 8.58 Â 10 3 , 8.58 Â 10 2 , 8.58 Â 10 1 , 4.29 Â 10 1 , and 4.29. For the broth samples, 1 mL of 30 ng/mL DNA was added to the reaction mixtures. For the mating and bioreactor biofilm samples, 1 mL of 20 ng/mL DNA was utilized as the template for the qPCRs with the following cycling conditions: 1 cycle at 95°C for 3 min, followed by 40 cycles of 95°C for 15 s, 53°C for 15 s, and 72°C for 30 s. The genome copy numbers were based on ftsK (1 chromosomal copy) quantification, and the data were calculated as copy number of circular junction per genome, or CI/chromosome.
Mating experiments. Strains were inoculated in LB broth or THY broth from overnight plate cultures and grown to the late-log phase. Donor and recipient strains were mixed at a 1:10 ratio of 10 8 /10 9 CFU and centrifuged for 15 min at 3,000 rcf to pellet bacteria. The pellet was resuspended in 100 mL of LB or THY broth, and DNase I was added to the mating mixture at 10 mg/mL. The mixture was plated on blood agar plates and incubated at 37°C for 4 h. The mating mixture was collected and resuspended in 1 mL LB or THY broth with a 10% glycerol final concentration. Serial dilutions were performed, and multilayer plating was carried out as described previously (32). S. pneumoniae was selected with 2 mg/mL Tet and/ or 220 mg/mL Str or 14 mg/mL Tmp, while B. subtilis was selected with 10 mg/mL Tet and/or 100 mg/mL spectinomycin or 100 mg/mL Str.
Quantification of extracellular DNA. Spent medium was collected from the outlets of the bioreactor chambers at h 1, 2, 4, and 6 of incubation for 1 h. The samples were centrifuged for 10 min at 15,000 rpm at 4°C (Eppendorf, Hauppauge, NY) and subsequently sterilized with a 0.45-mm-pore filter. Extracellular DNA was extracted from 400 mL of the sterile supernatant samples using the Qiagen QIAamp DNA Mini Kit following the instructions starting from addition of ethanol to the samples. The eDNA was eluted in 100 mL of elution buffer and stored at 280°C for further use. The purified eDNA was utilized as the templates for serotype-specific qPCR as described above, and the standard curve was built using 1 Â 10 3 , 1 Â 10 2 , 1 Â 10 1 , 1 Â 10°, 1 Â 10 21 , 5 Â 10 22 , and 5 Â 10 23 pg of chromosomal DNA corresponding to the appropriate serotype (20). The standard curve was used to calculate the eDNA concentrations for each time point sample utilizing the Bio-Rad CFX Manager software.
Whole-genome sequencing and variant analysis. Genomic DNA from bioreactor recombinants was purified using the Qiagen QIAamp DNA Mini Kit as instructed. Libraries were prepared utilizing the Illumina Nextera XT DNA library preparation kit (San Diego, CA) and sequenced by SeqCenter (Pittsburgh, PA) using the NextSeq 2000 platform. The paired-end read data were assembled and annotated using tools available on PATRIC's Bacterial and Viral Bioinformatics Resource Center. For variant analysis, the Illumina reads of the D39 Str/Tmp or D39 Str recipient strains were first mapped onto the closed genome sequence of reference strain D39 (NC_008533.2) using the DNAStar NGen program, and single nucleotide polymorphisms already present in the D39 Str/Tmp or D39 Str recipients were identified relative to the reference D39. These SNPs of D39 Str/Tmp or D39 Str were then discounted from the reference-guided assemblies of the recombinants to identify the remaining SNPs introduced by the donor strain. For the GA40410 and GA43265 recipient strains, the recombinants were assembled using individual WGS data as references to identify donor SNPs. Recombination blocks were estimated as clustering of consecutive donor SNPs flanked by recipient sequence. The outermost 59 and 39 donor SNPs were used to calculate the minimal length of the recombined donor DNA fragment.
Statistical analysis. All frequency, ratio, bacterial density, and eDNA concentration data were analyzed using two-tailed unpaired t tests on GraphPad Prism8.
Data availability. All data supporting the research findings of this study are included within the article and in the supplemental material. Annotated whole-genome sequences have been deposited in NCBI GenBank under BioProject no. PRJNA933161.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.3 MB.