Abstract
Homologous recombination (HR) is a crucial mechanism of DNA strand exchange that promotes genetic repair and diversity in all kingdoms of life. Bacterial HR is driven by the universal recombinase RecA, assisted by dedicated mediators that promote its polymerization on single-stranded DNA (ssDNA). In bacteria, natural transformation is a prominent HR-driven mechanism of horizontal gene transfer specifically dependent on the conserved DprA recombination mediator. Transformation involves internalisation of exogenous DNA as ssDNA, followed by its integration into the chromosome by RecA-directed HR. How DprA-mediated RecA filamentation on transforming ssDNA is spatiotemporally coordinated with other cellular processes remains unknown. Here, we tracked the localisation of functional fluorescent fusions to DprA and RecA in Streptococcus pneumoniae and revealed that both accumulate in an interdependent manner with internalised ssDNA at replication forks. In addition, dynamic RecA filaments were observed emanating from replication forks, even with heterologous transforming DNA, which probably represent chromosomal homology search. In conclusion, this unveiled interaction between HR transformation and replication machineries highlights an unprecedented role for replisomes in anchoring transforming ssDNA to the chromosome, which would define a pivotal early HR step for its chromosomal integration.
Introduction
Homologous recombination (HR) is a universal DNA strand exchange mechanism, which is vital to genome biology via its implication in specific pathways of DNA repair and genetic diversification1–4. The widely conserved recombinases of the RecA/Rad51 family are core HR effectors that form dynamic nucleofilaments to promote exchange between complementary DNA sequences5. These reactions are controlled and assisted by specific effectors, which define different HR pathways across all kingdoms of life. Any dysfunction in these HR assistants can alter cell development, threaten the integrity or adaptive capacity of the genome and endanger cell survival1, 6, 7.
Natural transformation is a programmed HR-directed horizontal gene transfer mechanism that is widespread in bacteria and promotes the shuffling of chromosomally-encoded genetic information8. As such, transformation facilitates adaptive responses to stresses, including the acquisition of new genetic traits such as antibiotic resistance and vaccine escape 8–10, as well as limiting the genetic drift of species by curing genomes of mobile genetic elements 11–14.
Transformation is a multistep DNA processing mechanism directed by proteins encoded by the recipient cell (Figure 1A). Most of these are expressed during a distinct physiological state, defined as competence, which is triggered and regulated in different ways depending on the species8. Transformation proteins first direct the uptake of exogenous double-stranded DNA (dsDNA) through the cell envelope to the periplasmic space 15–18; next, they couple transport of a linear single-stranded DNA (ssDNA) strand across the cell membrane with degradation of its complementary strand 16, 19–22; internalised ssDNA is then integrated into the genome by RecA-directed HR at homology sites. A key conserved early effector of the HR pathway of transformation is the ssDNA-binding protein DprA, which specifically interacts with RecA to mediate its loading onto ssDNA 23–25. Next, as in all HR pathways, RecA polymerises on the ssDNA to form a nucleofilament, referred to as the presynaptic filament, and promotes homology search in chromosomal DNA and pairing with a complementary DNA strand to generate a 3-stranded DNA molecule, defined as the HR heteroduplex, synapse or D-loop. Next, an helicase involved in extending ssDNA incorporation in the genome from the D-loop, which differs from one species to another. In firmicutes, this HR motor is RadA, a protein which also acts with RecA in pathways of DNA repair 26, 27. In contrast, a transformation-dedicated helicase ComM is conserved in all other bacterial species28. The final reactions of transformation, including covalent linkage and integration of the paired ssDNA molecule with the recipient chromosomal dsDNA, remain uncharacterised. Ultimately, a replication cycle generates a wildtype and a transformed chromosome, each segregated into a daughter cell.
Our current understanding of the transformation mechanism results from studies conducted in a dozen distinct species, including the historical models Bacillus subtilis and Streptococcus pneumoniae (the pneumococcus), as well as many other human pathogens such as Haemophilus influenzae, V. cholerae and Helicobacter pylori (for reviews, see 8, 29). These studies highlighted important general features of transformation, including the remarkable speed at which transforming DNA (tDNA) is captured, internalised and integrated into the chromosome. This was shown to occur in a minute time frame in S. pneumoniae and V. cholerae30, 31. How the HR system of transformation achieves such efficiency is unexplained. Pioneering studies in B. subtilis reported the gradual and stable accumulation of GFP fusions to transformation proteins involved in tDNA uptake and ssDNA transport, as well as RecA, at one pole of competent cells independent of tDNA addition 32, 33. In the presence of tDNA, polar RecA evolved into filaments proposed to represent presynaptic filaments formed during the polar entry of ssDNA, which next scan chromosomal DNA for homology 33, 34.
Here, we investigated DprA and RecA localisation dynamics during transformation in S. pneumoniae. In stark contrast to B. subtilis, in which competence occurs in non-replicating cells and lasts for several hours, pneumococcal competence occurs during the exponential phase of growth for a short period of about 30 minutes35. In addition, tDNA is captured and enters competent S. pneumoniae cells not at the pole but at midcell36, 37. Using functional fluorescent fusions of DprA and RecA, we tracked the early HR intermediates of transformation in actively transforming pneumococcal cells. Both proteins formed distinct foci at midcell in transforming cells, dependent on their physical interaction, showing that these nucleoprotein assemblies represent early HR intermediates of transformation. Furthermore, DprA and RecA foci were proven to localise to chromosomal replication forks. Importantly, RecA was observed to form short, dynamic filaments emanating from this replisomal accumulation point, possibly revealing homology search on the chromosome. These results represent an unprecedented link between the HR machinery of natural transformation and the chromosomal replication apparatus, shedding light on the mechanism of targeted homology search on the chromosome during pneumococcal transformation.
Results
DprA accumulates at midcell during transformation in S. pneumoniae
To observe the early DprA-mediated HR steps of natural transformation in individual living competent pneumococcal cells, we tracked the localisation of a fluorescent DprA-GFP fusion proven to be fully functional in transformation assays38. Purified DprA-GFP was as efficient as DprA in assisting RecA-directed HR in an in vitro D-loop assay (Extended Figure 1), validating use of this fusion for analysing DprA localisation dynamics during transformation. We previously showed that DprA-GFP accumulated at a single cell pole in competent cells38. This localisation was related to an additional role for DprA in shut-off of pneumococcal competence. This negative feedback loop is independent of the ability of competent cells to uptake tDNA but dependent on a high cellular concentration of DprA38, 39. Here, we analysed DprA-GFP localisation upon addition of tDNA to competent, transformable pneumococcal cells. Competence was induced by incubating cells with saturating levels (100 ng mL-1) of synthetic competence-stimulating peptide (CSP) for 10 minutes40, ensuring all cells in the population were competent. Addition of saturating levels of tDNA (250 ng µL-1) then ensured all cells were engaging in transformation. Cells were visualised 5 minutes after tDNA addition and compared with cells without tDNA. DprA-GFP formed foci in competent cells, irrespective of the addition of tDNA (Figure 1B). The frequency and cellular localisation of DprA-GFP foci, presented as focus density maps ordered by cell length (Figure 1C), showed that addition of tDNA did not modify the frequency of foci in competent cells but slightly altered their localisation (Figure 1D), with a significant increase of midcell foci from 13 % to 25 % (Figure 1E). These results suggested that DprA-GFP may interact with internalised ssDNA to generate midcell foci.
We showed previously using a strain expressing dprA under the control of an IPTG-inducible Plac promoter (CEPlac-dprA) and lacking native dprA that reducing cellular levels of DprA in competent cells prevented competence shut-off from occurring whilst maintaining optimal transformation efficiency39. Using a CEPlac-dprA-gfp fusion, we also showed that in similar conditions (6 µM IPTG), no polar foci of DprA-GFP were observed38. We took advantage of this strain and these conditions to visualise DprA-GFP during transformation as described above, with 48% of cells possessing DprA-GFP foci, mostly at midcell (Figures 2AB). By contrast, only 7% of cells possessed foci in the absence of tDNA, in agreement with previous results38. Most cells present DprA-GFP foci at midcell, while late dividing cells present foci at the ¼ and ¾ positions, future sites of midcell in daughter cells (Figure 2B). To explore where these tDNA-dependent DprA-GFP foci localise along the lateral axis of the cells, data was represented as heatmaps split into six cell categories. tDNA-dependent DprA-GFP foci were present near the centre of the longitudinal axis in all cell types (Figure 2C). In non- constricted cells they were either side of the central axis, while in constricted cells they appeared more central. Thus, DprA was found to accumulate at midcell in a tDNA dependent manner. We next analysed this localisation at different time-points after competence induction and tDNA addition. The results showed that the highest number of cells possessing tDNA-dependent DprA-GFP foci were observed 20 minutes after CSP addition, and that the majority of cells with foci possess a single focus (Extended Figure 2A). In addition, the localisation profile of these foci remains similar over time (Extended Figure 2B). Transforming cells with the same concentration of heterologous chromosomal DNA from Escherichia coli resulted in the formation of DprA-GFP foci at a similar frequency and localisation (Figure 2BC), showing that homology between tDNA and chromosomal DNA is not required for focus formation. We next examined how exogenous tDNA concentration impacted DprA-GFP focus formation. Results showed that a 1,000-fold reduction in tDNA concentration, starting from saturating conditions (250 ng µL-1), reduced the frequency of cells exhibiting midcell DprA- GFP foci from 47 % to 17 % (Extended Figure 2C). This suggests that the more ssDNA enters each cell, the more DprA-GFP molecules accumulate at midcell to generate detectable fluorescent foci. In conclusion, pneumococcal DprA accumulates at two distinct locations in competent cells, correlating with its two roles in competence and transformation. First, as reported previously38, the majority of DprA accumulates at one cell pole to mediate competence shut-off. Second, as observed here, a minority of DprA accumulates at midcell in a tDNA-dependent manner. This clustering of DprA at midcell appears therefore to be related to its role in transformation.
DprA anchoring at midcell is dependent on RecA
To gain further insight into the formation of tDNA-dependent DprA-GFP foci at midcell in competent cells, we reproduced these localisation experiments in strains disrupted in three genes involved in different stages of the transformation process, i.e. comEC, ssbB and radA. ComEC is proposed to form a channel in the cell membrane enabling ssDNA transfer into the cytoplasm41. Only 2 % of comEC-cells exhibited tDNA-dependent DprA-GFP foci, demonstrating that assembly of these foci depends on tDNA internalisation (Extended Figure 2D). Of note, it can be inferred from this result that DprA-GFP foci in competent cells grown without tDNA (Figure 2) result from the internalisation of DNA released in the medium from lysed cells. SsbB contributes to the protection and storage of internalised ssDNA to foster multiple chromosomal recombination events42, 43, and RadA is a helicase that extends ssDNA integration at RecA-directed D-loop intermediates26. Despite these key roles in transformation, neither was found to be involved in midcell DprA-GFP foci formation (Extended Figure 3).
Next, we further explored DprA-GFP localisation with two DprA point mutants, both severely impaired in transformation and differentially altered in DprA properties: DprAAR, defective in dimerisation and cooperative interaction with ssDNA and DprAQNQ, disrupted in RecA interaction24. Only 2 % of transforming cells expressing the DprAAR-GFP fusion from the ectopic Plac-dprAAR-gfp construct possessed foci (Extended Figure 2D). In contrast, the DprAQNQ-GFP fusion still formed tDNA-dependent midcell foci. However, these were observed in fewer cells than in an isogenic wildtype DprA-GFP fusion and their localisation appeared markedly different (Figure 2DE and Extended Figure 2E). This difference can be clearly seen when the data is represented as heatmaps, with cells split into six size categories. DprAQNQ- GFP accumulated at the extremities of the lateral cellular axis in small and medium sized cells, and at cell poles in large cells or at the constriction site and/or at the pole in constricted cells (Extended Figure 2F). Thus, the localisation patterns of DprA-GFP and DprAQNQ-GFP foci markedly differ: the later appear to be excluded from the cellular areas where the former form. This result strongly suggested that DprA interaction with RecA is key for the tDNA- dependent midcell accumulation of DprA-GFP. To confirm this, we repeated the experiment with the wildtype DprA-GFP fusion in a recA- mutant, and results were similar to those observed with DprAQNQ-GFP (Figure 2DEF and Extended Figure 2E). These results show that the accumulation of DprA-GFP at midcell depends on DprA interaction with RecA and tDNA. Thus, this midcell accumulation point appears to attract a trio of cross-interacting partners, i.e. DprA, RecA and tDNA. Importantly, when RecA or DprA are absent, most internalised ssDNA molecules are rapidly degraded30. Transforming cells of a recA comEC double mutant showed almost no DprA-GFP foci (Extended Figure 2FG). This suggested that sufficient internalised ssDNA remains protected by DprA within recA- competent cells to enable DprA- GFP foci formation. Together, these results show that RecA drives midcell localisation of DprA- GFP foci.
RecA accumulates at midcell during transformation
The dependency on RecA for the midcell localisation of DprA-GFP foci during transformation led us to analyse RecA localisation in competent cells. However, a recA- mturquoise fluorescent gene fusion generated at the native recA locus, despite being produced at wild type levels, was only partially functional in directing transformation and recombination repair of chromosomal damages (Extended Figure 4). In addition, this RecA- mTurquoise fusion accumulated at the cell poles during competence and generated DprA foci at this cell location when dprA-yfp was expressed at low concentrations, which were not formed in a RecA+ background (Extended Figure 5). In an attempt to visualise RecA localisation under fully functional recombination conditions, we placed the recA-mturquoise construct under the control of IPTG-inducible Plac promoter at the ectopic chromosomal CEP locus, allowing the production of a mixture of RecA and RecA-mTurquoise proteins in the cells, a strategy successfully used in various species44–46. This merodiploid strain, referred to as recA/recA-mturquoise, was equally as proficient in transformation and genome maintenance as the wildtype strain (Extended Figure 4). In this context, we found that RecA-mTurquoise accumulates into fluorescent foci at midcell in competent cells, dependent on tDNA, displaying the same foci localisation profile as cells expressing low-level DprA-GFP in the same conditions (Figure 3ABC). This result strongly suggests that midcell foci represented a functional cellular localisation for RecA and DprA during transformation. Indeed, formation of RecA-mTurquoise midcell foci was found to be dependent on tDNA and DprA (Figure 3D). Repeating this experiment in an IPTG gradient revealed that reducing the cellular levels of RecA-mTurquoise reduced the number of transformed competent cells with foci (Extended Figure 6). In addition, reducing the length of tDNA fragment reduced the number of cells presenting RecA-mTurquoise foci (Figure 3E). This interdependency between DprA and RecA for their midcell accumulation highlights the role of DprA in HR as a mediator of RecA loading on ssDNA at this precise cell location. Finally, we also attempted to directly visualise internalised ssDNA by fluorescent labelling as previously described in B. subtilis47. However, we were unable to conclusively visualise fluorescently labelled tDNA internalised in pneumococcal competent cells, which was essentially found randomly retained into multiple patches on the cell surface or in the periplasmic space as reported recently with B. subtilis19 (See Supplementary Results and Extended Figure 7). Together, these localisation studies of DprA and RecA in transforming competent cells revealed their interdependent assembly into foci at midcell, which are functionally linked to their concerted role in directing the early HR steps of transformation.
DprA and RecA colocalise with chromosomal DNA replication forks in transforming cells
The localisation of midcell tDNA-dependent DprA-GFP and RecA-mTurquoise foci was very similar to the localisation of the replication forks of the chromosome tracked by using fusions to proteins of the replisome48. To explore whether these foci colocalised with chromosomal replication forks, a strain was generated allowing controlled, ectopic expression of both the replisomal DnaX protein fused to YFP (CEPM-yfp-dnaX; induced by maltose) and DprA-mTurquoise (CEPIIlac-dprA-mturquoise; induced by IPTG). Firstly, 81 % (+/- 3,9 %) of non-competent cells possessed at least one YFP-DnaX focus, while this was slightly reduced to 73% (+/- 1,6 %) in competent cells 20 minutes after CSP addition (Extended Figure 8AB), showing that most replisomes remain intact in competent pneumococcal cells. Next, repeating the transformation experiment in this strain revealed that the cellular distribution of foci was almost identical for both fusions (Figure 4AB). 83,4 % of DprA-mTurquoise foci colocalised with YFP-DnaX foci (Figure 4C), irrespective of the cell cycle stage of the cell. The replisome protein DnaX moves dynamically around midcell48. Time-lapse microscopy of both DprA- mTurquoise and YFP-DnaX in competent, transforming cells showed that their midcell foci exhibited the same dynamics (Movie 1, Figure 4D). In all, this demonstrated that early DprA- mediated transformation HR intermediates navigate with the replisome. To strengthen this conclusion, ChIP-PCR experiments were carried out to explore whether YFP-DnaX was in close proximity to tDNA in transforming cells. First, results showed that a heterologous tDNA PCR fragment was copurified with DprA-GFP and DprA-YFP at 10-fold higher levels than with the DprAAR-GFP dimerization mutant or the unfused GFP used as a negative control (Figure 4E). Second, this tDNA was co-purified with YFP-DnaX at a similar level to DprA-GFP (Figure 4E), suggesting close proximity between early DprA-mediated HR intermediates and chromosomal replication forks during transformation. We also used a NanoBit assay49, 50 to explore the proximity of DprA engaged in transformation with the replisome. This system employs a luciferase separated into a large bit (LgBit) and a small bit (SmBit). Fusion of each part to different proteins that interact or are in close proximity in cells can restore luciferase activity and produce light in the presence of a furimazine-based substrate51. A strain possessing DprA- LgBit and an ectopic DprA-SmBit was used as a positive control and competent cells demonstrated strong luminescence irrespective of tDNA addition, due to dimerization of DprA (Figure 4F). The LgBit tag was fused to the Cter of DprA or DprAAR and the SmBit was fused to the Cter of DnaX. Addition of tDNA to competent cells increased luminescence in cells coexpressing DnaX-SmBit and DprA-LgBit but not DnaX-SmBit and DprAAR-LgBit (Figure 4F). This result further demonstrated a close proximity between the replisome and DprA, dependent on tDNA.. Then, to formally demonstrate that RecA is also targeted to chromosomal replication forks during transformation, we analysed RecA-mTurquoise localisation in recA/recA-mturquoise competent cells co-expressing YFP-DnaX. As expected, RecA-mTurquoise foci were found to strongly colocalise with YFP-DnaX at midcell in transforming cells (Figure 5ABC).
RecA forms dynamic tDNA-dependent filaments at chromosomal replication forks
In experiments investigating localisation of RecA-mTurquoise in the recA/recA-mturquoise strain, we observed filamentous fluorescent structures in a minority of non-competent cells (8%). These long polymers (0,82 µm +/- 0,28 µm) appear similar to RecA filaments reported as HR filaments involved in recombinational repair of double-strand breaks (DSB) in other bacteria (Badrinarayanan et al., 2015; Amarh et al., 2018; Wiktor et al., 2021). Similarly,, exposure of non-competent pneumococcal cells to DNA damaging agent norfloxacin increased the number of cells presenting long, dynamic RecA polymers, averaging 0,84 µm (+/- 0,31 µm) in length, from 8% to 69.8% (Movies 2 and 3, Extended Figure 9). Notably, in transforming cells of this strain, we observed that 59 % (+/- 5 %) of RecA-mTurquoise foci colocalising with replisomes exhibited dynamic filaments emanating from these foci (Figure 5D). We used time-lapse microfluidics to track RecA-mTurquoise and YFP-DnaX localisation in real time in transforming cells52. Results showed rapid formation of RecA-mTurquoise foci in the vicinity of the replisome as little as 2 minutes after tDNA addition, and subsequent
Then, to test whether blocking DNA replication altered the capacity of DprA to mediate tDNA dependent RecA filamentation at chromosomal replication forks, we reproduced these localisation experiments in the presence of HpUra, a nucleotide analogue that selectively inhibits the essential PolC DNA polymerase of the pneumococcal replisome30, 53. We first analysed RecA-mTurquoise localisation in non-competent recA/recA- mturquoise cells following addition of saturating amount of HpUra that fully blocks chromosomal DNA replication and cell growth30, 54 (Extended Figure 10A). We observed the formation of long RecA-mTurquoise filaments (0.94 µm +/- 0.41 µm) 5 minutes after HpUra addition (Extended Figure 10B), reproducing what was observed previously in B. subtilis55. Importantly, these filaments were lost in cells lacking recO, showing that they depend on the RecFOR recombinase loading system52, 53 (Extended Figure 10B). We previously demonstrated that transformation is RecO independent56. Thus, we analysed RecA-mTurquoise and YFP- DnaX localisations in recO-, recA/recAmTurquoise, yfp-dnaX competent cells, to prevent formation of HpUra-dependent and RecO-mediated RecA-mTurquoise filaments. Cells were exposed to HpUra for 5 minutes, then CSP was added to induce competence, and tDNA was added 10 minutes later. Cells were visualised after a further 5 minute incubation to allow tDNA internalisation. Results showed that firstly, YFP-DnaX still accumulated into midcell foci even after PolC-directed replication was blocked, showing that replisomes remained intact, although stalled (Extended Figure 10CDEF). In transforming cells with stalled replisomes, RecA still accumulated into midcell foci, which strongly colocalised with DnaX-YFP (Extended Figure 10CDEF). In conclusion, these results show that active replication is not required for RecA access to chromosomal forks during transformation and that RecO is not involved in tDNA- dependent RecA filamentation at that precise chromosomal location.
Discussion
In this study, we reveal that the dedicated DprA-mediated and RecA-directed HR pathway of natural genetic transformation is spatiotemporally orchestrated at chromosomal replication forks in S. pneumoniae (Figure 6A). First, by using functional GFP fusions, we demonstrate that both DprA and RecA accumulate at midcell and colocalise with the replisome protein DnaX in a tDNA-dependent manner. These colocalisations are observed in ∼70 % of a competent, transforming population, roughly equivalent to the number of cells undergoing chromosomal replication at a given time in these growth conditions (Figures 4 and 5). Second, we found that DnaX is in physical proximity to tDNA in transforming cells (Figure 4E) and that tDNA addition promotes interaction between DnaX and DprA (Figure 4F). Interdependent DprA and RecA accumulation at chromosomal replication forks following tDNA internalisation matches the interplay between DprA, RecA and ssDNA previously uncovered by combining biochemical and genetic analyses, and proven to promote HR during transformation via the formation of the presynaptic HR filament23, 24. Strong evidence supporting this conclusion is the formation of tDNA-dependent and DprA-mediated RecA filaments emanating from the replisome (Figure 5DE). In addition, midcell DprA-GFP localisation is not observed in a comEC mutant (Extended Figure 2D), demonstrating the need for tDNA internalisation to generate DprA and RecA foci by interaction with the internalised ssDNA template. Third, both DprA and RecA foci and RecA filaments are observed in a minute time frame in transforming cells. These kinetics are as rapid as that of tDNA integration into the pneumococcal chromosome tracked over time with the use of a short radiolabeled tDNA fragment homologous to a specific chromosomal locus30. Fourth, both DprA and RecA foci and RecA filaments are formed at replication forks either with homologous or heterologous tDNA, in the same short time frame (Figure 5 and Extended Figure S8). Thus, we conclude that the RecA filaments emanating from the replication forks in transforming pneumococcal cells are presynaptic HR intermediates engaged in the search of homology on the chromosome (Figure 6B). Altogether, these findings link the HR machinery of transformation and the replisome for the first time in a naturally transformable bacterial species.
Dynamic RecA filaments formed at replication forks in transforming pneumococcal cells are similar to those of genome maintenance HR pathways visualised in single cells in several bacterial species, and demonstrated to be presynaptic filaments actively searching for homology and promoting recombinational DNA repair44–46. However, these RecA filaments exhibit marked differences compared to those of pneumococcal transformation reported here. Notably, the dynamic RecA filaments assembled at a single double-strand break on one copy of the neoreplicated chromosome of Escherichia coli or Caulobacter crescentus extend across the cell length, which is proposed to correspond to the bidirectional search for the uncleaved homologous DNA on the second copy of the chromosome segregated to the opposite cell pole44–46. We observed such long RecA filaments in growing pneumococcal cells suffering genome damages, including specific replication fork arrest caused by the HpUra PolC inhibitor as previously reported in B. subtilis55 (Extended Figure 10). In contrast, RecA filaments formed during pneumococcal transformation at replication forks are shorter (Figure 5DE). One reason for this difference may be the length of the tDNA entering the cell, evaluated to be in the range of 3 to 7 kb and gradually reduced over the competence window43. Another marked difference between RecA filaments in HR pathways of pneumococcal transformation and genome maintenance is their assembly site in the cell. In the latter case, RecA filaments are formed on the chromosome at the site of DNA damage, in conjunction with the formation of ssDNA template. In contrast, in the case of pneumococcal transformation, ssDNA is formed and enters the cell at the cytoplasmic membrane through ComEC, and must reach the replication fork where DprA-mediated RecA filamentation occurs. Therefore, ssDNA formation and presynaptic RecA filamentation appear to be spatially separated during transformation in the pneumococcus. Interestingly, however, tDNA capture and uptake was previously found to also occur at midcell in the pneumococcus36, 37. Thus, transformation appears to proceed via a midcell channel coupling DNA capture and internalisation with chromosome access and HR, which may underpin the speed at which transformation occurs in a minute time frame in the pneumococcus30.
Previous analysis of RecA localization during transformation in B. subtilis depicted a different choreography than the one reported here for S. pneumoniae. Interestingly, transformation occurs in non-replicating B. subtilis cells, as proven by the lack of DnaX-GFP foci in competent cells57, and GFP-RecA has been found to localise at one cell pole where the proteins directing tDNA capture and uptake accumulate33. Upon transformation, GFP-RecA has been found to generate long filaments from the cell pole, proposed to represent homology search on chromosomal DNA33. However, B. subtilis DprA was not found to follow the same choreography as it accumulates at midcell in transforming cells34. These marked deviations in RecA filamentation dynamics during transformation between S. pneumoniae and B. subtilis run parallel to the difference in the timing of competence development between these two species. Pneumococcal competence is triggered in actively replicating cells in response to a large panel of stresses35, including genome damage54, 58, 59, and lasts for a short period of time of 30 minutes60. In contrast, competence in B. subtilis occurs during nutrient starvation when cells stop replicating, and lasts for several hours8. Thus, transformable bacterial species have evolved distinct strategies to mediate HR-mediated chromosomal integration of tDNA, depending on how competence is integrated into their cell cycle. It will be interesting to explore how other transformable species integrate the early HR steps of transformation into their varied cell cycles. The anchoring of the presynaptic HR filaments of transformation to the chromosomal replication forks of S. pneumoniae not only provides them immediate access to chromosomal DNA for homology search, but also to the potential actions of the large set of proteins acting at the forks, either directly in DNA replication or occasionally to repair the damaged forks. This toolbox of DNA effectors are ideally located to assist the whole HR process of transformation up to covalent linkage of tDNA to the chromosome, many steps of which remain uncharacterised. Of note, we demonstrate with HpUra-treated competent cells that the replication forks do not need to be active to act as molecular anchors for the early step of HR of transformation (Extended Figure 10). This mirrors a previous study showing that HpUra-treated competent pneumococcal cells integrate tDNA as efficiently as non-treated cells59. This indicates that RecA filaments spread over the genome for homology search, emanating from replication forks.
A major perspective of this study is to identify how early HR intermediates, composed of DprA and RecA bound to tDNA, are driven to the chromosomal replication forks. Many proteins are concentrated at these vital chromosomal sites, either essential or accessory to the DNA replication process. We show that RecA drives early HR intermediates to midcell (Figure 2DEF), opening up the possibility of an interaction between RecA and such a replication protein partner. One of these known accessory effectors is the RecO protein, which is known to mediate RecA loading on ssDNA gaps. However, we demonstrate RecO is not needed for replisome access of early HR intermediates or RecA filamentation at replication forks (Extended Figure 10). In addition, transformation HR effectors SsbB and RadA also played no role in this chromosome access mechanism (Extended Figure 3).
In conclusion, this study revealed that early HR intermediates of pneumococcal transformation accumulate at chromosomal replication forks. By doing so, replication forks could provide a landing pad for presynaptic filaments of HR to access the recipient chromosome and carry out homology search, optimising the speed and efficiency of transformation.
Materials and Methods
Bacterial strains, competence and transformation
The pneumococcal strains, primers and plasmids used in this study can be found in Table S1. Standard procedures for transformation and growth media were used61. In this study, cells were prevented from spontaneously developing competence by deletion of the comC gene (comC0)62. Pre-competent cultures were prepared and transformation carried out as previously described38. Antibiotic concentrations (μg mL-1) used for the selection of S. pneumoniae transformants were: chloramphenicol (Cm), 4.5; erythromycin, 0.05; kanamycin (Kan), 250; spectinomycin (Spc), 100; streptomycin (Sm), 200; trimethoprim (Trim), 20. GraphPad Prism was used for statistical analyses. Detailed information regarding the construction of new plasmids and strains can be found in the Supplementary Information. To compare protein expression profiles, Western blots were carried out as previously described38. Secondary polyclonal antibodies raised against RecA and SsbB were used at 1/10,000 dilution.
Fluorescence microscopy and image analysis
To visualise cells by epifluorescence microscopy, pneumococcal precultures grown in C+Y medium at 37 °C to an OD550 of 0.1 were induced with CSP (100 ng mL-1). Cells were incubated for 10 minutes at 37°C before addition of transforming DNA. Transforming DNA we either homologous (S. pneumoniae R1501 genomic DNA) or heterologous (Escherichia coli genomic DNA) prepared using QIAGEN 500/g Genomic tips. Cells were then incubated at 37 °C for 5 minutes unless stated. After this incubation, 2 µL samples were spotted onto a warmed microscope slide containing a slab of 1.2 % C+Y agarose as previously described 37 before imaging. To generate movies, images were taken of the same fields of vision at varying time points during incubation at 37 °C. Phase contrast and fluorescence microscopy was performed as previously described63. Images were processed using the Nis-Elements AR software (Nikon). Images were analysed using MicrobeJ, a plug-in of ImageJ64. Data was analysed in R and represented in two distinct ways. Firstly, focus density maps were plotted on the longitudinal axis of half cells ordered by cell length. Each spot represents the localisation of an individual focus, and spot colour represents focus density at a specific location on the half cell. Only cells with > 0 foci shown. In cells possessing > 1 foci, foci were represented adjacently on cells of the same length. Secondly, cells were separated into six categories based on cell size and presence or absence of constriction, and heatmaps were generated for each category. The six cell categories were defined in MicrobeJ to reflect those determined previously for pneumococci63. End of constriction (cons. end); septum = 1, circularity < 0.7; middle of constriction (cons. middle), septum = 1, 0.8 > circularity > 0.7; start of constriction (cons. start), all other cells with septum = 1; Large cells, septum = 0, cell length > 1.4 µm, circularity < 0.9; medium cells, septum = 0, 1.4 µm > cell length > 1.2 µm,0.94 > circularity < 0.9; small cells, all other cells with septum = 0. The proportions of cells found in each category were consistent with those previously observed in these conditions63, validating the parameters used to define the categories.
Chromatin immunoprecipitation PCR (ChIP-PCR)
Chromatin immunoprecipitation (ChIP) was done using magnetic GFP-Trap beads as per manufacturer’s instructions (Chromotek). Briefly, cells were inoculated 1/50 in 30 mL of C+Y medium pH 7 and grown to OD550 0.1. Competence was induced by addition of 100 ng mL-1 CSP, and cells were incubated for 10 min at 37 °C. Transforming DNA (1 kb capsule fragment absent from recipient strains amplified from D39 using primer pair DDL35-DDL36) was added at a final concentration of 1 ng µL-1 and cells were incubated at 37 °C for 5 minutes to allow internalisation. Cells were then fixed by addition of 3 mL Fixation solution F (50 mM Tris pH 8.0, 100 mM NaCl, 0.5 mM EGTA, 1 mM EDTA, 10 % formaldehyde) and incubation for 30 min at room temperature. Cultures were then centrifuged for 10 min at 5,000 g, 4 °C and supernantants were discarded. Pellets were washed twice in 30 mL cold PBS with centrifugation at 5,000 g, 4 °C for 10 min in between. Cells were then washed in 1 mL cold PBS and centrifuged at 10,000 g, 4 °C for 2 min before being frozen with liquid nitrogen and storage at −80 °C until use. Pellets were defrosted and resuspended in 2 mL cold Lysis L buffer (50 mM Hepes-KOH pH 7.55, 140 mM NaCl, 1 mM EDTA, 1 % triton X-100, 0.1 % Sodium deoxycholate, 100 µg mL-1 RNase A) before sonication in a Diagenode Bioruptor Plus sonication bath (29 cycles, 30 s sonication, 30 s rest). Resulting samples were centrifuged for 5 min at 16,000 g, 4 °C and supernatants were transferred into fresh 2 mL tubes and centrifuged for 5 min at 16,000 g, 4 °C. After transfer into fresh 2 mL tubes, 200 µL of each sample was taken and stored at −80 °C to act as a whole cell extract prior to immunoprecipitation. 25 µL of GFP-TRAP magnetic beads was then added to each sample, which was subsequently tumbled gently at 4°C for 3h 30 min. Magnetic beads were recovered by magnetism, supernatants were discarded and beads were resuspended in 1 mL cold Lysis L buffer before being centrifuged at 800 g for 5 min. Magnetic beads were recovered by magnetism, supernatants were discarded and beads were resuspended in 1 mL cold Lysis L5 buffer (50 mM Hepes-KOH pH 7.55, 500 mM NaCl, 1 mM EDTA, 1 % triton X-100, 0.1 % Sodium deoxycholate, 100 µg mL-1 RNase A). Magnetic beads were recovered by magnetism, supernatants were discarded and beads were resuspended in 1 mL cold Wash buffer W (10 mM Tris/HCL pH 8.0, 250 mM LiCl, 1 mM EDTA, 0.5 % NP-40, 0.5 % DOC) and beads were then recovered by magnetism and resuspended in 520 µL TES buffer (50 mM Tris/HCl pH 8.0, 10 mM EDTA, 1 % SDS). At this stage, the WCE samples were defrosted and supplemented with 300 µL TES buffer and 20 µL SDS 10 %. All samples were then incubated at 65 °C with vigorous shaking overnight. Magnetic beads were removed by magnetism and 12.5 µL proteinase K (20 mg mL-1) was added before incubation of the samples at 37 °C for 2 h. DNA was purified from the samples by sequential phenol:chloroform extraction and 1 µL glycogen, 40 µL sodium acetate 3M pH 5.3 and 1 mL ethanol was added before incubation at −20 °C to precipitate the DNA. Samples were then centrifuged at 16,000 g, 4 °C for 15 min and the supernatant was carefully removed before the pellets were resuspended in 100 µL TE buffer pH 8 and incubated at 65 °C for 20 min. DNA was then purified using GFX PCR purification columns (GE Healthcare). DNA samples were diluted (1/200 for WCE, 1/20 for IP samples) and probed in triplicate by qPCR for the presence of the 1 kb capsule fragment using iTaq DNA polymerase (BIO-RAD) and primer pair DDL34-DDL35, which amplify a 115 bp region within the 1 kb fragment. Specific amplifications were confirmed by single peaks in melting curve analysis. Cycle threshold (CT) values were obtained according to the software instructions. Relative quantification was performed with the 2-ΔΔCT method65. Each PCR reaction, run in duplicate for each sample, was repeated for at least two independent times. Data are represented as mean ± s.e.m calculated from triplicate repeats, with individual data points plotted.
Split-luciferase assay
Split luciferase assays were carried out as previously described49, 50, with modifications. Briefly, pneumococcal cells were grown in C+Y medium (with 50 µM IPTG where required) at 37 °C until OD550 0.1 and competence was induced by addition of 100 ng mL-1 CSP. Cells were then incubated for 10 min at 37 °C before addition of R1501 chromosomal DNA (250 ng µL-1) where noted, followed by a further 5 min incubation at 37 °C. Cells were then washed in fresh C+Y medium and 1 % NanoGlo substrate (Promega) was added and luminescence was measured 20 times every 1 min in a plate reader (VarioSkan luminometer, ThermoFisher). Data are represented as mean ± s.e.m calculated from nine independent repeats, with individual data points plotted.
Author contributions
C. J. and P. P. wrote the paper. C. J., R. H., A-L. S., M. D. and D. D. L. performed the experiments. C. J. and P. P. designed and analysed the experiments and interpreted the data.
Extended Figure Legends
Movie Legends
Movie 1: Early HR intermediates visualised via DprA-mTurquoise navigate with the dynamic replisome around midcell during transformation of competent pneumococci. Time-lapse microscopy of strain R4631 (comC0, CEPM-yfp-dnaX, CEPII-Plac-dprA-mturquoise, dprA::spc). Images taken at two minute intervals starting 10 min after competence induction and 5 min after DNA addition. Still images from Movie used to make Figure 4D.
Movie 2 – RecA-mTurquoise does not accumulate into filaments in most RecA/RecA- mTurquoise cells in the absence of norfloxacin exposure. Time-lapse microscopy of individual cell of strain R4848 (comC0, CEPlac-recA-mturquoise) in the absence of norfloxacin. Images taken at 1 min intervals. Still image from movie used in Extended Figure 9A.
Movie 3 – RecA-mTurquoise accumulates into filaments in most RecA/RecA-mTurquoise cells after norfloxacin exposure. Time-lapse microscopy of individual cell of strain R4848 (comC0, CEPlac-recA-mturquoise) in the presence of norfloxacin (100 ng µL-1). Images taken at 1 min intervals. Still image from movie used in Extended Figure 9B.
Movie 4: Early HR intermediates visualised via RecA-mTurquoise navigate with the dynamic replisome around midcell during transformation of competent pneumococci. Time-lapse microscopy of strain R4840 (comC0, CEPM-yfp-dnaX, CEPII-Plac-recA-mturquoise) taken using microfluidics. Images taken at one min intervals starting 5 min after competence induction and immediately upon DNA addition. Still images from Movie used to make Figure 5E.
Supplementary information
Supplementary Materials and Methods
Protein purification
To purify DprA-GFP, the dprA-gfp sequence was amplified from R372838 using primer pair oALS12 and oALS13; The resulting DNA fragment was digested with EcoRI and EagI restriction enzymes and ligated into a pET21 vector digested with the same enzymes to generate the pALS1 plasmid. This plasmid was transformed into Escherichia coli Rosetta cells and cells were grown at 37 °C to OD550 0.8 with 0.5 mM IPTG to stimulate DprA-GFP expression. Purification was achieved by sequential passage through three columns as follows; HiTrap Heparin HP 1 mL, gel filtration Superdex 200 Hiload 16/60, HiTrap Q HP 1 mL.
In vitro HR assays
HR assays were carried out as follows. 75 nM of DprA or DprA-GFP were incubated at 37 °C for 10 min with varying concentrations of wild type RecA (150, 300, 600 nM) in the presence of 10 nM of Cy3-tagged ovio54 primer (70 nt, fully homologous sequence to pUC18), 10 mM MgOAc and 2 mM ATP. 5 mM of pUC18 plasmid was then added followed by incubation at 37 °C for 10 min. A 1/20 volume of xylene cyanol was added and samples were then denatured by addition of 0.1 % SDS and 10 mM EDTA followed by 3 min incubation at 37°C. Samples were then run on a 1.25 % TBE gel for 60 min at 50 V and DNA was then directly detected on the gel using the Typhoon Trio. Quantification of HR was carried out using MultiGauge software.
Plasmid and strain construction
Here we describe how the new plasmids and mutant strains used in this study were generated. Previously published constructs and mutants were simply transferred from published strains by transformation with appropriate selection. The pCJ1 plasmid was generated by removing the MCS from the pUC57-CEPIIR-comX plasmid 66. To achieve this, the plasmid was digested with EcoRV enzyme, and the insert side recovered. The pUC57 side of the plasmid was amplified with primer pair CJ735-CJ736, each possessing EcoRV enzyme sites, removing the MCS site in the process. The insert and PCR were ligated together to generate pCJ1. The pCJ2 plasmid was generated by amplifying a lacI-Plac PCR fragment from R3833 with primer pair CJ567-CJ730 and a dprA-mturquoise PCR fragment from R4062 with primer pair CJ731-CJ595. pCJ1 was digested by SalI and KpnI enyzmes, lacI-Plac by SalI and NcoI enzymes and dprA-mturquoise by NcoI and KpnI, and the three fragments were ligated together to generate pCJ2. The pCJ3 plasmid was generated by digesting the pMB42 plasmid with the XhoI and HindIII enzymes to remove gfp and ligating in an mTurquoise PCR fragment amplified from the R4011 strain67 with CJ455-CJ456 primer pair, digested with the same restriction enzymes. The pCJ4 plasmid was generated by amplifying two adjacent DNA fragments by PCR on the R3728 strain66 around dprA-gfp construct using primer pairs CJ391-CJ465 and CJ466- CJ378 respectively. The three base mutations required to alter gfp to yfp present in both primers CJ465 and CJ466. Splicing overlap extension (SOE) PCR on these two fragments with the CJ391-CJ378 primer pair generated a DNA fragment with the yfp mutation. This DNA fragment was transformed without selection into R3728 66 with a 3 h 30 min phenotypic expression phase in liquid culture to introduce the yfp mutation, and positive clones were determined by PCR amplification with the CJ391-CJ378 primer pair and sequencing with the CJ378 primer. The pCJ5 plasmid was generated by digesting the pMB42 plasmid 66 with the EcoRI and XhoI enzymes to remove ‘dprA and ligating in a ‘recA PCR fragment amplified from the R1501 strain with CJ764-CJ765 primer pair, digested with the same restriction enzymes. The pCJ6 plasmid was generated by amplifying a PCR fragment consisting of the Plac promoter and upstream lacI gene from R4261 66 using primer pair CJ567-CJ615 and digesting it with SalI and NcoI enzymes. A dprAQNQ-gfp DNA fragment was amplified from R4046 66 using primer pair CJ411-CJ616 and digested with NcoI and BamHI enzymes. The pCEPR-luc plasmid was digested with SalI and BamHI enzymes and these three fragments were ligated together to generate pCJ6. The pCJ7 plasmid was generated in the same manner but with an amplification of a dprAAR-gfp DNA fragment was amplified from R4047 66 using primer pair CJ411-CJ616. The R2546 strain (comC0, CEPX-gfp) was constructed by transforming R1501 with the pCN35 plasmid 68 and selecting for kanamycin resistance. The R3406 strain (comC0, ssbB-luc, CEPM- yfp-dnaX) was generated by making four DNA fragments by PCR; a fragment of the upstream CEP platform sequence from pCEP 69 with primer pair OVK53-OVK54; the yfp sequence from R4404 with primer pair OVK55-OVK56; the dnaX sequence from R1501 with primer pair OVK61-OVK62 and the downstream CEP platform sequence from pCEP 69 with primer pair OVK57-OVK73. A SOE PCR fragment was generated using these four fragments with primer pair OVK53-OVK73, and this was transformed into R1502, with transformants selected with kanamycin. The R4062 strain (comC0, dprA-mturquoise) was generated by transforming R1501 with the pCJ3 plasmid and selecting for spectinomycin resistance. The R4400 strain (comC0, CEPlac-dprA-gfp, ssbB::cat) was generated by transforming R4262 with genomic DNA from the R4812 strain and selecting for kanamycin resistance. The R4401 strain (comC0, CEPlac-dprA-gfp, comEC::ery) was generated by transforming R4262 with genomic DNA from the R2586 strain 37 and selecting for erythromycin resistance. The R4404 strain (comC0, dprA- yfp) was generated by transforming R1501 with the pCJ4 plasmid and selecting for spectinomycin resistance. The R4412 strain (comC0, CEPlac-dprAQNQ-gfp) was generated by transforming R1501 with pCJ6 and selecting transformants with kanamycin. The R4413 strain (comC0, CEPlac-dprAAR-gfp) was generated by transforming R1501 with pCJ7 and selecting transformants with kanamycin. The R4415 strain (comC0, CEPlac-dprAQNQ-gfp, dprA::spc) was generated by transforming R4412 with genomic DNA from the R751 strain 70 and selecting for spectinomycin resistance. The R4416 strain (comC0, CEPlac-dprAAR-gfp, dprA::spc) was generated by transforming R4413 with genomic DNA from the R751 strain 70 and selecting for spectinomycin resistance. The R4429 strain (comC0, CEPlac-dprA-gfp, dprA::spc, recA::cat) was generated by transforming R4262 with genomic DNA from the R209 strain 71 in presence of 50 µM IPTG and selecting for chloramphenicol resistance. To generate the R4618 strain (comC0, CEPlac-dprA-gfp, dprA::spc, comEC::trim, recA::cat), a comEC::trim DNA fragment was created by initial amplification of the regions upstream and downstream of the comEC gene using primer pairs CJ720-721 and CJ724-725 and R1501 gDNA as template. The trimethoprim resistance cassette was amplified using the primer pair CJ722-723 and the R4107 strain 66 as template. SOE PCR on these three fragments with the primer pair CJ720-725 generated a DNA fragment with the comEC gene replaced with the trimethoprim resistance cassette, which was co-transformed into R4262 with a recA::cat DNA fragment amplified from R209 71 using primer pair CJ726-CJ727. Transformants were selected with trimethoprim and chloramphenicol to integrate both comEC::trim and recA::cat at the same time, since both abrogate transformation. To generate the R4625 strain (comC0, CEPlac-dprA-gfp, dprA::spc, radA::trim), a radA::trim DNA fragment was created by initial amplification of the regions upstream and downstream of the radA gene using primer pairs CJ748-CJ749 and CJ752-oIM58 and R1501 gDNA as template. The trimethoprim resistance cassette was amplified using the primer pair CJ750-751 and the R4107 strain 66 as template. SOE PCR on these three fragments with the primer pair CJ748-oIM58 generated a DNA fragment with the radA gene replaced with the trimethoprim resistance cassette, which was transformed into R4262 66 in presence of 50 µM IPTG and transformants were selected with trimethoprim. To generate the R4626 strain (comC0, ssbB-luc, CEPM-yfp-dnaX, CEPIIlac-dprA-mTurquoise), R3406 was transformed with the pCJ2 plasmid, and transformants were selected with erythromycin. To generate the R4631 strain (comC0, ssbB-luc, CEPM-yfp-dnaX, CEPIIlac-dprA-mTurquoise, dprA::spc), R4626 was transformed with genomic DNA from strain R751 and transformants were selected with spectinomycin. To generate strain R4664, a fragment of CEPlac was amplified using primer pair CJ588-CJ680 and R3833 as template, and the recA gene was amplified using primer pair CJ681- CJ682 and R1501 as template. The pCEPlac-dprA-gfp plasmid was digested with SalI and BamHI restriction enzymes, while the DNA fragments were digested with SalI/NcoI and NcoI/BamHI respectively. These three fragments were ligated together and transformed into R1501, with transformants selected with kanamycin. To generate strain R4712 (comC0, recA- mTurquoise), R1501 was transformed with pCJ5 and transformants were selected with spectinomycin. To generate strain R4716 (comC0, CEPlac-dprA-gfp, recA-mTurquoise), R426166 was transformed with pCJ5 and transformants were selected with spectinomycin. To generate strain R4731 (comC0, CEPlac-dprA-yfp, recA-mTurquoise), two adjacent DNA fragments were amplified by PCR on the R4262 strain 66 around CEPlac-dprA-gfp construct using primer pairs CJ114-CJ465 and CJ466-kan1 respectively. The three base mutations required to alter gfp to yfp present in both primers CJ465 and CJ466. SOE PCR on these two fragments with the CJ114-kan1 primer pair generated a DNA fragment with the yfp mutation. This DNA fragment was transformed without selection into R4716 with a 3 h 30 min phenotypic expression phase in liquid culture to introduce the yfp mutation, and positive clones were determined by PCR amplification with the CJ114-kan1 primer pair and sequencing with the CJ114 primer. To generate the R4742 strain (comC0, CEPlac-dprA-yfp, recA-mTurquoise, dprA::trim), a dprA::trim DNA fragment was created by initial amplification of the regions upstream and downstream of the dprA gene using primer pairs CJ373-CJ770 and CJ773-CJ378 and R1501 gDNA as template. The trimethoprim resistance cassette was amplified using the primer pair CJ771-CJ772 and the R4107 strain 66 as template. SOE PCR on these three fragments with the primer pair CJ373-CJ378 generated a DNA fragment with the dprA gene replaced with the trimethoprim resistance cassette, which was transformed into R4731 and transformants were selected with trimethoprim. The R4812 strain (comC0, ssbB::cat) was generated by transforming R2294 43 with the pEMcat plasmid and transformants were selected with chloramphenicol. To generate strain R4840, the regions upstream and downstream of dprA-mturquoise in the CEPII platform were amplified from R4631 using primer pairs CJ662-CJ793 and CJ667-CJ796 respectively, and recA-mturquoise was amplified from R4712 using primer pair CJ794-CJ795. SOE PCR using there three fragments and primer pair CJ662-CJ667 generated a CEPII-Plac-recA-mturquoise DNA fragment which was transformed into R3406, with transformants selected with erythromycin. To generate strain R4848 (comC0, CEPlac-recA-mturquoise), 5’ and 3’ fragments of CEPlac were amplified from R4262 with primer pairs CJ574-CJ799 and CJ802-CJ575 respectively, and recA- mturquoise was amplified from R4712 using primer pair CJ800-CJ801. SOE PCR with these DNA fragments and primer pair CJ574-CJ575 generated a CEPlac-recA-mturquoise fragment which was transformed into R1501 and transformants were selected with kanamycin. To generate strain R4849 (comC0, dprA-lgbit), 5’ and 3’ fragments of dprA were amplified from R1501 with primer pairs CJ689-CJ690 and CJ693-CJ694, and the lgbit tag with appropriate linker and over hang sequences for SOE PCR was synthesized based on previously-published sequence optimized for the pneumococcus using gBlocks (Intergrated DNA technologies). SOE PCR with these DNA fragments and primer pair CJ689-CJ694 generated a dprA-lgbit fragment which was transformed into R1501 without selection and transformants were screened by PCR for integration using primer pair CJ689-CJ694. To generate strains R4851 (comC0, CEPlac- recA-mturquoise, dprA::spc), R4848 was transformed with chromosomal DNA from R751 (rpsL41, dprA::spc) 70 and transformants were selected with spectinomycin. To generate strain R4856, a DNA fragment containing the smbit tag fused to the 3’ end of the dnaX gene with a linker, flanked by 5’ and 3’ sequences of dnaX was generated using gBlocks (Integrated DNA technologies). 5’ and 3’ fragments of dnaX were amplified from R1501 with primer pairs CJ809-CJ810 and CJ813-CJ814, and SOE PCR using these three DNA fragments and primer pair CJ809-CJ814 generated a dnaX-smbit DNA fragment which was transformed into R1501 without selection and transformants were screened by PCR for integration using primer pair CJ811-CJ812. To generate strain R4857 (comC0, ΔrecA::trim), upstream and downstream sequences around the recA gene were amplified using primer pairs CJ829-CJ830 and CJ833- CJ808 respectively, and the TrimR resistance cassette was amplified from strain R4107 66 using primer pair CJ831-CJ832. A ΔrecA::trim DNA fragment was generated by SOE PCR using these three DNA fragments and transformed into R1501, with transformants selected with trimethoprim. To generate strain R4858 (comC0, dprA-lgbit, CEPlac-dprA-smbit), two DNA fragments containing upstream and downstream sequences around the 3’ end of dprA in CEPlac-dprA were amplified from strain R4262 using primer pairs CJ574-CJ827 and CJ575- CJ828 respectively, where primers CJ827 and CJ828 include the linker-smbit sequence. SOE PCR using these two DNA fragments generated a CEPlac-dprA-smbit fragment, which was transformed into R4849 and transformants were selected with kanamycin. To generate strain R4859 (comC0, dprA-lgbit, dnaX-smbit, hexA::ermAM), R4856 cells were transformed with genomic DNA from strain R246 (hexA::ermAM) and transformants were selected with erythromycin. To generate strain R4861, R4859 cells were transformed with a dprAAR PCR fragment amplified from strain R2585 24 using primer pair CJ311-CJ391, and transformants were screened by PCR and sequencing (Eurofins MWG) for insertion of the two independent mutations conferring the dprAAR phenotype 24.
Time-lapse microfluidics experiments
Time-lapse microfluidics experiments were carried out using a CellASIC ONIX Microfluidic platform and B04A microfluidic plates (Merck-Millipore, Billerica, MA, U.S.A) as previously described 72, with modifications. Briefly, exponentially growing cultures (OD550 0,3) of R4840 (comC0, CEPM-yfp-dnaX, CEPII-Plac-recA-mturquoise) were diluted 50-fold in C+Y medium (supplemented with 300 U/mL catalase, 0.3 % maltose and 50 µM IPTG) and incubated at 37 °C to an OD550 of 0,1. Cells were then loaded into the microfluidic chamber and maintained at 37 °C in a thermostated chamber with a constant flow rate of 0,3 µL/h (0,25 psi). Competence induction was achieved by injecting CSP (1 µg mL-1 in C+Y medium with catalase, maltose and IPTG) for 3 min at 6 psi. DNA (250 ng µL -1, diluted in C+Y medium with catalase, maltose and IPTG) was then injected for 6 min at 3 psi, followed by 1 h at 0.25 psi. Images were captured every minute throughout using the same microscope set-up as described above, with a thermostated chamber at 37 °C.
Sensitivity to DNA damage assays
Survival assays were performed as previously described 56, with modifications. Briefly, cells were grown to OD550 0.1 in C+Y medium (with 50 µM IPTG where appropriate) before serial dilution and spotting of 10 µL volumes onto pre-dried plates containing 0.02 % MMS and 50 µM IPTG where appropriate. After spot drying, plates were incubated overnight at 37°C in a bell jar with Anaerocult A (Merck) to promote anaerobic conditions.
Fluorescent DNA microscopy experiments
Two independent DNA fragments (rpsL1 and radA::spc) were used to test internalization of transforming DNA, rpsL1, conferring streptomycin resistant via point mutation, or radA::spc, conferring spectinomycin resistance by integration of a heterologous cassette. A 2,008 bp DNA fragment containing rpsL1 was amplified from R2980 (dpnMAB, rpsL1) using primer pair MB83-MB84. A 4,649 bp DNA fragment containing radA::spc was amplified from R3255 (dpnMAB, hexA::ermAM, radA::spc, ssbB::kan) using primer pair CJ338- CJ368. Strains possessing the DpnII restriction system were used as templates to allow template removal by DpnI digestion. PCR fragments were labelled with either fluorescein, Dylight 550 or Dylight 650. For fluorescein labelling, 1 µL of 1 mM fluorescein-12-dUTP (Thermo Fisher Scientific), 2 µL dNTP mix (1 mM dATP, dCTP, and dGTP and 0.5 mM dTTP (Thermo Fisher Scientific)), 0.5 µL DreamTaq DNA polymerase (Thermo Fisher Scientific), 5 µL DreamTaq buffer, 1 µM of each primer and 2 µL of genomic DNA were used in a total reaction mixture volume of 50 µL. The reaction conditions for Dylight labelling were the same as for fluorescein, but with 1 µL of dNTP mixture (10 mM dGTP, dCTP, and dATP and 5 mM dTTP and aminoallyl-dUTP (Thermo Fisher Scientific)). After labelling, samples were protected from light throughout. Samples were then mixed 4:1 with Dylight 550 or 650 (10 mg mL-1, Thermo Fisher Scientific) and incubated at room temperature for 3 hours. Samples were then incubated for 2 h with 0.5 µL DpnI (20 U µL-1, FastDigest, Thermo Fisher Scientific) per 50 µL PCR sample. Label incorporation was calculated using a NanoDrop (Thermo Fisher Scientific) as follows: fluorescein, 0,3-1,6 pmol µL-1; Dylight 550 and 650, 0,9-4,4 pmol µL-1. Microscopy images were captured as described above using FITC (fluorescein), Cy3 (Dylight 550) and Cy5 (Dylight 650) filters respectively.
Supplementary Results
Exploring the visualization of fluorescent tDNA during pneumococcal transformation
In this study, fluorescent fusions of DprA and RecA were used to visualize the early HR intermediates in actively growing pneumococci. The other main actor of early HR intermediates is ssDNA, and the possibility of visualizing fluorescent transforming ssDNA in B. subtilis and S. pneumoniae has previously been explored, showing fluorescent foci on cells after DNase I treatment 47. However, a further study in B. subtilis showed that resistance to DNase I does not necessarily indicate entry into the cytoplasm, but rather the periplasm 19. In light of this, the potential of using such fluorescent DNA to visualize early HR intermediates during transformation was explored in S. pneumoniae. To begin, the transformation efficiency of labelled DNA fragments possessing an rpsL41 point mutation 73 was compared to unlabelled controls with dUTP/dTTP mix or dTTP alone. Results showed that the dUTP/dTTP unlabelled mix showed reduced transformation efficiency compared to dTTP alone, while labelled DNA fragments showed similar or slightly reduced transformation efficiency compared to the dUTP/dTTP unlabelled mix (Extended Figure 7A). This suggested that fluorescent DNA could be internalised, however, internalization and integration of short unlabelled fragments around the point mutation could not be excluded. Fluorescence microscopy on competent cells transforming with labelled DNA fragments showed distinct foci associated to a minority of cells (Extended Figure 7B). To further explore whether fluorescently labelled DNA could be internalized by competent cells, transformation experiments were carried out using DNA fragments possessing a point mutation as above (rpsL1, otherwise homologous DNA) or a heterologous antibiotic resistance cassette flanked by homologous sequences (radA::spc), labelled or not with Dylight 500 or 650, at varying DNA concentrations. Unlike a point mutation, integration of a heterologous cassette requires transfer of the entire cassette plus flanking sequences, making the presence of entirely unlabelled fragments of transformable DNA much less likely. Results show that while reducing the concentration of rpsL1 DNA below saturating levels did not alter the ratio between transformation efficiency of labelled and unlabelled DNA donors, reducing the concentration of radA::spc donor DNA specifically reduced the transformation efficiency of labelled DNA, increasing the ratio of transformation efficiency between labelled and unlabelled DNA (Extended Figure 7CD). This result suggests that high levels of fluorescent labelling negatively impacted the transformation of a donor DNA molecule, but that nonetheless, less labelled donor DNA can be internalized and integrated into the recipient chromosome by transformation. To explore whether it was possible to visualize this subpopulation of transforming DNA, fluorescence microscopy was carried out on wildtype cells in the presence or absence of CSP, as well as in several transformasome mutant strains. Results show that in wildtype cells, although foci are observed in non-competent cells, competence specific foci are observed (Extended Figure 7E). The absence of the pilus (comGA-), the EndA nuclease (endA-) or the transformation-dedicated recombinase loader (dprA-) did not alter the number of competent cells possessing foci despite being key for capture, processing and protection of transforming DNA, respectively. However, a reducing in cells possessing foci was observed in the absence of the DNA transformation pore (comEC-), while removing the DNA receptor (comEA-) reduced the number of cells possessing foci to non-competent levels (Extended Figure 7E). The localization of foci did not vary significantly in all of these strains (Extended Figure 7F). In conclusion, although it appears DNA molecules possessing fewer fluorescent tags can be transformed, visualisation of these is not possible due to the pollution from DNA molecules with greater numbers of fluorescent tags, which are not transformable and thus remain on the outside of the cell.
Acknowledgements
We thank Isabelle Mortier-Barriere for support with microfluidics experiments. We thank Jérome Rech for support with microscopy and video assembly. We thank the LITC imaging platform of Toulouse TRI for their assistance in microscopy. This work was funded by the Centre National de la Recherche Scientifique, University Paul Sabatier and the Agence Nationale de la Recherche (grants ANR-10-BLAN-1331 and ANR-17-CE13-0031).
References
- 1.↵
- 2.
- 3.
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.
- 10.↵
- 11.↵
- 12.
- 13.
- 14.↵
- 15.↵
- 16.↵
- 17.
- 18.↵
- 19.↵
- 20.
- 21.
- 22.
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.
- 75.
- 76.
- 77.
- 78.
- 79.
- 80.
- 81.
- 82.