Introduction

Timed and tightly controlled gene expression is crucial in bacterial cells. A variety of mechanisms to control transcription are known, comprising both positive and negative regulators (reviewed in Turnbough, 2019). One of the mechanisms of negative regulation involves transcription attenuation at termination sites (reviewed in Santangelo and Artsimovitch, 2011). The regulation of transcription attenuation is achieved by a diverse class of antiterminators which counteract the attenuation (Santangelo and Artsimovitch, 2011). Antiterminators fall into two categories and can either act specifically at a single, defined terminator site or convert RNA polymerase into a termination-resistant complex that is generally resistant to multiple termination sites over a longer distance (Santangelo and Artsimovitch, 2011; reviewed in Goodson and Winkler, 2018). In the latter case, such antiterminators are referred to as processive antiterminators. Well-known examples of processive antiterminators are Q and N of bacteriophage lambda which control progression of the infection cycle (Goodson and Winkler, 2018).

Peculiar examples of phage evolution are Gene Transfer Agents (GTAs) (reviewed in Québatte and Dehio, 2019; Banks and Tung, 2024; Craske et al., 2024). GTAs are mobile genetic elements, which emerged from defective temperate bacteriophages. GTAs encode bacteriophage-like particles and in contrast to phages transfer random pieces of host bacterial DNA. Despite their phage origin, GTAs are not self-transmitting because the DNA packaging capacity of a GTA particle is too small to package the entire gene cluster encoding it. Therefore, GTAs are transferred vertically (Banks and Tung, 2024; Craske et al., 2024; Kogay et al., 2022; Lang et al., 2012). GTAs are found both in bacteria and archaea, and are one of the non-canonical mechanisms of horizontal gene transfer (HGT) (Arnold et al., 2022).

Bacterial pathogens of the genus Bartonella encode a chimeric GTA known as BaGTA (Québatte and Dehio, 2019; Berglund et al., 2009; Guy et al., 2013). BaGTA encodes the gene clusters bgtA-K, responsible for the synthesis and assembly of GTA particles, bgtL-R, with orphan functions, and bgtS-V, involved in host cell lysis (Fig. 1A). DNA packaging in GTA particles is coupled to a process of site-specific DNA amplification mediated by the ror locus located in between the bgtL-R and bgtS-V gene clusters. This locus, emerging from a different phage domestication event, includes an origin of run-off replication (ROR) and the associated brr gene cluster (brrA-G) (Berglund et al., 2009; Guy et al., 2013; Québatte et al., 2017; Tamarit et al., 2018) (Fig. 1A). The bgt and ror loci are located within a high plasticity zone of the genome that is prone to DNA integrations and is highly enriched for genes encoding virulence properties involved in host interaction and adaptation (Berglund et al., 2009; Guy et al., 2013). This specific genome architecture leads to ROR-driven amplification and packaging of these pathogenicity genes, resulting in elevated transfer rates to other Bartonella bacteria compared to housekeeping genes encoded more distantly to the ROR origin (Québatte et al., 2017). This process is considered to provide the recombination substrates for the duplication and evolutionary diversification of genes encoding the type IV secretion systems (T4SSs) and cognate effector proteins which collectively represent a key innovation associated with the adaptive radiation of these pathogens to diverse mammalian hosts (Guy et al., 2013; Engel et al., 2011). Additionally, it is also considered to facilitate the fast evolution of surface-located autotransporters serving in host interaction and adaptation, such as the recombination-driven antigenic variation of the CFA autotransporter (Siewert et al., 2022). Thus, BaGTA is a crucial factor for the fitness and adaptability of pathogens in the Bartonella genus (Québatte and Dehio, 2019). However, little is known about how the expression of the bgt and brr loci derived from separate phages is regulated in a coordinated manner.

Figure 1.

In this study, we elucidated the regulatory hierarchy between the bgtA-K and ror loci, resolving the mechanism of gene expression control. We discovered that BaGTA operates under a strict regulatory framework, with a protein encoded by the ror locus and controlling the expression of the bgtA-K capsid-production locus. Specifically, our research revealed that in absence of BaGTA activation, transcription of the bgtA-K locus is attenuated by a terminator site. Upon BaGTA activation by a yet unknown process, the ror locus-encoded BrrG overcomes this transcription attenuation of the bgtA-K locus by acting as a processive antiterminator. Furthermore, we propose that antitermination is a common mechanism of GTA regulation in the Alphaproteobacteria. Collectively, these findings offer significant insights into the regulation of BaGTA-mediated gene transfer in pathogens of the genus Bartonella and suggest a broader impact of processive antitermination as a regulatory principle in related GTAs.

Results

BrrG shows structural similarity to the P21 phage-encoded Q21 antiterminator and is essential for GTA transfer and expression of capsid components

Previously, we identified by multiplexed transposon sequencing the genes of BaGTA encoding both for donor transfer and recipient uptake functions (Québatte et al., 2017). However, many of the identified genes had only hypothetical or no functional sequence-based annotation, leaving their contribution to BaGTA function unclear. To address this knowledge gap, we employed the protein structure prediction tool AlphaFold2 to model unannotated proteins of BaGTA genes and searched the Protein Data Bank (PDB) database for similar structures. We discovered that the brrG gene (Fig. 1A), which is located in the ror locus, encodes a protein with structural similarity to the antiterminator Q21 of the lambdoid bacteriophage P21 (Fig. 1B). Q21 and its counterpart Q from phage λ (Qλ) function as processive antiterminators involved in bypassing intrinsic terminators (Guo et al., 1991; Grayhack and Roberts, 1982). This enables uninterrupted expression of the late genes, which encode capsid morphogenesis and lysis cassettes. Therefore, we hypothesized that, similar to Q21 and Qλ, BrrG represents a putative antiterminator controling gene expression within the BaGTA regulatory network.

To test this hypothesis, we assessed the effect of brrG overexpression on the transcription of the bgtA-K locus that encodes components of the capsid synthesis and assembly machinery as essential components of BaGTA production. To address this, we utilized the previously established reporter strain bgtC-gfp that contains a promoterless gfpmut2 inserted in the bgtA-K locus in the intergenic region between bgtC and bgtB (Québatte et al., 2017) (Fig. 1A). This transcriptional reporter enabled us to monitor bgtA-K expression over time using flow cytometry (Fig. 1C, upper panel). We observed that plasmid-based overproduction of BrrG increased the fraction of GFP fluorescent bacteria to up to 75%, compared to about 15% at native BrrG expression (Fig. 1D), thus massively expanding the fraction of cells expressing the bgtA-K locus. To test whether this increase is followed by GTA production and elevated DNA transfer, we performed the GTA transfer assay. For that, we cocultivated donor and recipient cells carrying different resistance markers followed by plating on selective media to enumerate double-resistant bacteria as a read-out for the frequency of GTA-mediated gene transfer (Fig. 1C, lower panel). We observed that - while the proportion of cells expressing the BaGTA locus increased as a result of BrrG overproduction, GTA transfer rates remained similar to native BrrG expression (Fig. 1E). In contrast, deleting brrG led to a significant decrease in the proportion of fluorescent bacteria (Fig. 1F) and caused DNA transfer rates to drop at least 100-fold compared to those observed at native BrrG expression (Fig. 1G). Together, these findings indicate that BrrG plays a crucial role in regulating BaGTA. It is capable of activating the expression of capsid components in a large part of the population that is initially GTA inactive. However, overproduction of BrrG is not sufficient to increase the rate of DNA transfer, indicating that it acts in concert with other regulators controling other crucial steps of this complex process.

BrrG binds in proximity to the P_bgtA-K promoter located upstream of to the putative bgtA-K operon

In lambdoid phages, both Q21 and Qλ are DNA-binding proteins that recognize a motif located in the promoter region of the late operon, which encodes lysis cassettes and proteins required for capsid morphogenesis (Guo et al., 1991; Grayhack and Roberts, 1982; Yin et al., 2019; Shi et al., 2019; Yin et al., 2022). Considering the structural similarity between BrrG and Q21, we hypothesised that BrrG may be functionally conserved and thus able to directly bind to DNA at the yet unidentified promoter of the bgtA-K locus. To test this, we performed chromatin immunoprecipitation of epitope-tagged BrrG followed by DNA sequencing. We found that BrrG binds to DNA, with a single binding site located in the intergenic region far upstream of the bgtK gene representing the first conserved gene of the bgtA-K locus (Fig. 2A) (Guy et al., 2013; Tamarit et al., 2018). Given this remote upstream location of the BrrG binding site, we wondered whether the promoter elements might also be located nearby. First, we tested whether the 999 bp fragment spanning the region upstream of bgtK until the end of clpB could drive expression of a fused promoterless dsRed gene using flow cytometry as readout (Fig. 2A, plasmid pPf3-dsRed). Strikingly, this fragment did not only demonstrate promoter activity but also displayed at the population level a temporal control resembling closely the expression pattern of the chromosomal transcriptional reporter strain bgtC-gfp H1GmR with a characteristic peak in the size of expressing population at the 6 h time-point of the 24 h expression kinetic (Fig. 2B). This finding suggests that the bgtA-K locus is organized as a transcriptional unit (operon) that is driven by a promoter localizing to the 999 bp fragment upstream of bgtK, which we thus named P_bgtA-K (Fig. 2A).

Figure 2.

Next, to narrow down the location of P_bgtA-K, we split the 999 bp fragment into three partially overlapping fragments and tested them with the same approach using a gfpmut2-based promoter-probe plasmid (plasmids psF1, psF2 and psF3, Fig. 2A). We observed that the most upstream fragment spanning nucleotides 594-999 was able to drive broad expression of gfpmut2 at population level (plasmid psF3 in Figure 2C). The fraction of fluorescent bacteria was substantially higher compared to the chromosomal transcriptional reporter strain bgtC-gfp, expanding beyond 40% at each time-point (Fig. 2C). Importantly, the lack of temporal regulation as seen by the chromosomal transcriptional reporter strain bgtC-gfp and likewise the 999 bp fragment in plasmid Pf3 (compare Figs. 2B and 2C) suggests that absence of downstream control elements located in the more proximal region spanning nucleotides 1-593 converted P_bgtA-K into a constitutive state. To map P_bgtA-K based on conserved elements (-10 and -35 boxes) we aligned corresponding sequences from a selection of Bartonella species covering the genomic diversity of this genus. By this comparative approach, we identified a highly conserved sequence motive (TTGACA-N₁₇-TATCAT) that shows almost full identity to the canonical sigma-70 promoter sequence of E. coli (Fig. 2D) (Campbell et al., 2002; Murakami et al., 2002). This suggests that P_bgtA-K functions as a minimal promoter of the bgtA-K locus. Interestingly, the BrrG binding region coincides with P_bgtA-K, suggesting that BrrG controls transcription either by direct binding to P_bgtA-K or in the vicinity of it.

Taken together, we identified P_bgtA-K as a strong constitutive promoter located far upstream of the bgtA-K locus, driving its expression as a transcriptional unit along with more proximal elements required for controlling temporal expression at the population level. Additionally, we found that the putative antiterminator BrrG, similar to Q21 and λQ, is a DNA-binding protein that directly binds to P_bgtA-Kand likely controls bgtA-K operon expression at population level.

A long leader sequence attenuating expression is located downstream of the P_bgtA-K promoter

To confirm that the identified conserved sequence is the promoter and to hint at the function of the long downstream sequence in transcription, we generated a panel of promoter-probe plasmids harboring truncations of the 999 bp fragment bearing promoter activity (pF0-gfp to pF6-gfp) (Fig. 3A). We observed that the 29 bp fragment spanning the predicted P_bgtA-K displayed fluorescence in the vast majority of cells, confirming the promoter prediction (F0 fragment, Fig. 3B). Similar promoter activities were recorded for the fragments F1 and F2 that are extended by 27 bp or 133 bp toward the bgtA-K locus, respectively. A striking difference in fluorescence distribution with an almost full loss of fluorescent bacteria was observed for the F3 fragment containing, relative to F2, an additional 139 bp (Fig. 3B). Fragments F4-F6 with further downstream extensions also displayed pattern of low fluorescence, while a bi-model distribution with low- and intermediate-expressing bacteria was observed for fragments F4 and F6.

Figure 3.

Because fragment F3 is the shortest fragment displaying reduced expression, we hypothesized that a terminator could be present in the sequence extending the highly fluorescent F2 fragment. Indeed, a putative L-shaped terminator with a free energy (ΔG) of -6 kcal/mol is predicted within this stretch of 139 bp of the F3 fragment, which might explain the observed difference in expression levels (Fig. 3C). Altogether, these results demonstrate that the minimal promoter enables uniform and constitutive expression of the BaGTA locus at population level. The observed temporal regulation and heterogeneity of the gene cluster expression appear to be provided by the stretch of 864 bp located downstream of the promoter and upstream to bgtK as first gene of the bgtA-G locus. This stretch of DNA contains the putative terminator and could harbor other terminators or additional regulatory elements.

A -10-like element is required for expression of the bgtA-K locus

In phage lambda, the process of Q-mediated antitermination relies on the engagement of Q into the complex with RNA-polymerase (RNAP) and the conversion of the latter into a termination-resistant form (Perdue and Roberts, 2011). This process begins with RNAP pausing at the -10- like element, a six-nucleotide motif (NANNNT), located just downstream of the -10 promoter element, which allows Q to engage with the transcription complex (Perdue and Roberts, 2011).

Therefore, we set out to determine whether a -10-like element is present in the vicinity of the identified P_bgtA-K minimal promoter. We examined the previously generated alignment and identified a conserved NANNNT site located 5 bp downstream of the identified promoter P_bgtA-K(Fig. 4A). As the conserved nucleotides A and T at positions +2 and +6 in phage lambda and other phages are also conserved in this motive directly downstream of P_bgtA-K, we hypothesized that this site could act as a -10-like element (Fig. 4A). To validate this hypothesis, we used the promoter-probe plasmid containing the F6 fragment (Fig. 3A), and introduced two point mutations, replacing the conserved A in position -2 to G and T in position +6 to C (pF6_(-10mut)-gfp), respectively. Additionally, we introduced a second plasmid directing ATc-inducible BrrG overproduction to test whether these mutations would affect the activity of the putative antiterminator in the flow cytometry assay (Fig. 1C). In contrast to plasmid pF6-gfp encoding the native -10-like element (Fig. 4B, left, -ATc), pF6_(-10mut)-gfp did not display the characteristic kinetics of GFP-expression of wild-type bacteria with a peak at 6 h (Fig. 4B, right, -ATc). Moreover, neither did it display the massive expansion of the GFP-expressing population upon BrrG overproduction (+ATc) (Fig. 4B). Thus, the replacement of two conserved nucleotides at the putative -10-like element completely abolished transcription of the downstream bgtA-K locus under both native BrrG expression and BrrG overproduction conditions. This suggests that the identified conserved site is a functional - 10-like element essential for the BrrG-mediated antitermination.

Figure 4.

BrrG is a processive antiterminator allowing RNAP to bypass the termination site

We showed that BrrG shares structural features with the processive antiterminator Q21 and that its function is dependent on the -10-like element suggesting that BrrG belongs to the class of processive antiterminators. To validate this hypothesis, we tested the antitermination properties of BrrG towards a synthetic terminator. For this, we used the artificially designed strong terminator BBa_B1006 from the iGEM collection and placed it in between the F2 fragment (see Fig. 3A), which contains the native P_bgtA-K promoter and downstream -10-like element, but not the native terminator and the gfp reporter gene (Fig. 5A). The plasmid pF2-gfp containing the terminator-less fragment F2 drives GFP-expression in a large part of the bacterial population at both BrrG native expression and overproduction conditions (Figs. 3B and 5B). In contrast, the derivative pF2-T_{BBa_1006}-gfp containing in addition the artificial terminator showed a strongly reduced population expression of GFP, but a marked increase in GFP-expressing bacteria upon BrrG overproduction (Fig. 5B, right panel). Together, these findings demonstrate that the presence of BrrG allows the RNAP to bypass both the native termination site, as well as an unrelated synthetic terminator, indicating that BrrG is indeed a processive antiterminator.

Figure 5.

Discussion

Processive antitermination is an important mechanism of transcriptional control identified both for bacteria and phages. In temperate phages processive antiterminators, i.e phage lambda N and Q-antiterminators, switch on expression of early and late genes, respectively, and thus enable the progression of the infection cycle (Goodson and Winkler, 2018).

Given the ancestral relationship between GTAs and phages, this study investigates whether transcription attenuation and antitermination are preserved in the chimeric BaGTA and whether they play a role in the regulatory integration of two functional gene clusters of divergent phage origin. We showed that the “late” capsid producing bgtA-K genes belong to a single transcriptional unit derived from one phage, and that their expression is enabled by the antiterminator BrrG, encoded within the ror locus, which originates from a second phage. However, the question remains: Is BrrG a Q-like antiterminator, and does it function in a similar manner?

The canonical Q-dependent regulatory circuit consists of several critical components: Q antiterminator, the promoter P_R’, a termination site and a -10-like element (Perdue and Roberts, 2011; reviewed in Casjens and Hendrix, 2015). The first component is the Q antiterminator which binds to DNA and engages into the complex with RNAP. The Q-RNAP complex is termination-resistant and permits RNAP to bypass the termination signals in front of late genes which results in the synthesis of the capsid components (Guo et al., 1991; Grayhack and Roberts, 1982). In our study we showed that BrrG has structural similarity to the Q21 antiterminator of the P21 lambdoid phage. Moreover, BrrG is able to counteract transcription attenuation imposed by an introduced artificial termination site. Additionally, we demonstrated that BrrG binds in proximity to the P_bgtA-K promoter and upregulates transcription of the bgtA-K cluster – BaGTA’s counterpart of the late gene cluster.

The second component of the canonical Q-dependent regulatory circuit is the promoter P_R’ containing Q-binding site for interaction with Q and engagement with RNAP. In this study, we demonstrated that the P_bgtA-K promoter overlaps with the BrrG binding site. Given that the promoter P_bgtA-K is located 864 bp upstream from the first gene of the bgtA-K cassette and is separated from the former by a gene BH14100, this raises the question of whether BH14100 belongs to the bgtA-K cluster and what functional role it may play. Previous studies have shown that BH14100 is not conserved among bartonellae and thus is likely not involved in the BaGTA regulation or synthesis (Guy et al., 2013; Tamarit et al., 2018). In support of this, our previous study using multiplexed transposon sequencing showed no effect of BH14100 disruption on GTA transfer, indicating that BH14100 has either no or only a very modest effect on GTA synthesis (Québatte et al., 2017).

The third component of the canonical Q-dependent regulatory circuit is the termination site which attenuates expression of the late gene cluster_’. We showed that the putative terminator T_bgtA-K is located downstream of the P_bgtA-K promoter and regulates expression of the bgtA-K operon. Whereas the bgtA-K locus is only active in the GTA producing cells, deleting the putative terminator T_bgtA-K renders broad and constitituve expression of the bgtA-K locus which suggests that the temporal regulation and heterogeneity in expression are granted by T_bgtA-K.

The fourth component of the canonical Q-dependent regulatory circuit is the -10-like element located in proximity to the promoter P_R’ driving expression of late genes. This element resembles the -10-element of the promoter and defines the sigma-dependent pausing of RNAP which is crucial for the engagement of Q into the complex with RNAP (Perdue and Roberts, 2011). We found that expression of the bgtA-K locus is dependent on the conserved -10-like element located downstream of the P_bgtA-K promoter. Altogether, these data suggest that BrrG is an antiterminator found in BaGTA and involved in regulating expression of the bgtA-K cluster. Based on these lines of evidence we propose BrrG as Q-like processive antiterminator.

However, why does GTA-mediated DNA transfer remain unaffected upon brrG overexpression?

A simple explanation could be that bacterial lysis and particle release are controlled independently of BrrG which activates only transcription of the bgtA-K locus. This notion is supported by the apparent absence of lysis of fluorescent bacteria upon BrrG overproduction (Fig. 1D). For the archetypical RcGTA of Rhodobacter capsulatus, particle synthesis and release are controlled differentially by the CckA-ChpT-CtrA phosphorelay system, where the phosphorylation state of CtrA controls the progression from the GTA synthesis to a lytic release (Farrera-Calderon et al., 2021; Fogg, 2019; Westbye et al., 2018; Westbye et al., 2013). We hypothesize that similarly to RcGTA a trigger could be involved in the regulation of bacterial lysis and BaGTA release. BaGTA contains two lysis cassettes – bgtAB and bgtTUV – and only the former one is located within the larger bgt cluster, suggesting that the latter cassette is controlled independently.

Taken together, we demonstrate that the transcriptional control mechanism involving transcription attenuation and antitermination, as described for phage lambda, is functionally conserved in the chimeric BaGTA. This mechanism specifically coordinates the temporal expression of the ROR and bgtA-K loci, which have independent phage origins. The ROR-encoded BrrG acts as a processive antiterminator, facilitating the induction of the bgtA-K transcriptional unit. However, is this regulatory mechanism also conserved in other types of GTA beyond BaGTA of Bartonella? The recent publication by Tran and Le has shown that termination-antitermination is involved in control of gene expression in evolutionary distinct CcGTA of Caulobacter crescentus (Tran and Le, 2024). The authors showed that transcription of the structural genes, which are counterparts of the bgtA-K locus, is co-activated by GafY and IHF, while the GafZ antiterminator prevents expression attenuation at the termination sites. Furthermore, GafYZ regulators of CcGTA are homologous to N’- and C’-domains of GafA – the direct activator of RcGTA (Fogg, 2019; Gozzi et al., 2022; Sherlock and Fogg, 2022), which indicates that RcGTA might also be controlled by processive antitermination. Previous studies of GafA indicated that its C’-domain has similarity to the region 4 of a sigma factor (R4σ) and therefore authors proposed GafA as an alternative sigma factor or a transcription factor (Fogg, 2019; Sherlock and Fogg, 2022). However, the related R4σ-like fold was also found in the resolved structure of Q21 antiterminator indicating that findings of Fogg and Sherlock might, in fact, speak for the implication of GafA in antitermination (Shi et al., 2019). Interestingly, Tran and Le found neither sequence nor structural similarity of GafZ to known antiterminators Qλ and Q21. Our comparison of resolved structure of Q21 and modelled BrrG, GafZ and GafA, clearly indicates similar structural features suggesting both GafZ and GafA being antiterminators (Suppl. Figure 1).

RcGTA and CcGTA, on one hand, and BaGTA on the other, belong to two evolutionary distinct groups of GTA that emerged from independent events of phage domestication (Kogay et al., 2022; Tamarit et al., 2018). The reliance of these groups on the termination-antitermination mechanism of transcription control clearly suggests that processive antitermination is a common regulatory mechanism in GTAs, which evolved independently in two types of GTA within Alphaproteobacteria. To date, five types of GTAs have been identified in both bacteria and archaea. Given their phage heritage, it remains unclear whether other GTA types also depend on processive antitermination for regulation, and this should be explored in future studies.

Materials and Methods

Bacterial Strains and growth conditions

All bacterial strains, plasmids and oligonucleotides used in this study are listed in Supplementary Tables 1, 2 and 3, respectively. B. henselae strains were grown at 35°C on Columbia blood agar (CBA) supplemented with 5% sheep blood, appropriate antibiotics (gentamicin (Gm) – 10 μg/ml, kanamycin (Km) – 30 μg/ml, streptomycin (Sm) – 100 μg/ml, spectinomycin (Sp) – 50 μg/ml) and other supplements (isopropyl β-D-thiogalactosidase (IPTG) – 500 μM, anhydrotetracycline (ATc) – 10 ng/ml or 25 ng/ml) in a humidified atmosphere containing 5% CO2. B. henselae strains were stored as cryostocks at -80°C and prior the experiment were seeded on CBA for 3 days and subsequently expanded and grown on fresh plates for 2 days. E. coli strains were stored as cryostocks at -80°C and were grown at 37°C overnight in lysogeny broth (LB) or on solid LB agar plates (LA) supplemented with appropriate antibiotics (ampicillin (Ap) – 100 μg/ml, Gm – 10 μg/ml, Km – 30 μg/ml) and other supplements (2,6-diaminopimelic acid (DAP) – 1 mM, 1% glucose).

Construction of strains

All bacterial strains used in this study are listed in Table 1.

Table 1.

DNA manipulations were performed according to standard techniques and all cloned inserts were DNA sequenced to confirm sequence integrity. Chromosomal insertions or deletions in B. henselae were generated by a two-step gene replacement procedure as described in Schulein and Dehio (Schulein and Dehio, 2002). Plasmids were introduced into B. henselae strains by conjugation with E. coli strain JKE201 (MFDpir) using two-parental mating as described previously (Harms et al., 2017).

Construction of plasmids

All plasmids used in this study are listed in Table 2 and all oligonucleotides used in this study are listed in Table 3. Plasmids generated in this study were constructed as follows:

Table 2.

Table 3.

pQN011: Plasmid pTR1000 was digested with the restriction enzyme BamHI. Homologous fragments 1 (upstream) and 2 (downstream) were PCR amplified from heat-killed cells B. henselae with primers prQN029 and prQN030, prQN031 and prQN032, respectively. Those two fragments were cloned together by performing overlap extension PCR with primers prQN030 and prQN032. The fragment and the backbone were fused with the 5X In-Fusion® HD Enzyme Premix kit.

pQN018: brrG was PCR amplified from heat-killed cells B. henselae with primers prQN053 and prQN054. tetR was PCR amplified with primer prAK227 and prAK208 from 705-bp tetR-P_tet promoter template ordered as a synthetic fragment from Genescript. The PCR product generated by amplifying pBZ485_a_empty with prAK089 and prAK109 was used as a backbone. Three fragments were combined with the 5X In-Fusion® HD Enzyme Premix kit.

pAK126: brrG was PCR amplified from heat-killed cells B. henselae with primers prAK440 and prAK441. tetR was PCR amplified with primer prAK444 and prAK208 from 705-bp tetR-P_tet promoter template ordered as a synthetic fragment from Genescript. The PCR product generated by amplification of pPG100 with prAK116 and prAK167 was used as a backbone. Three fragments were combined with the 5X In-Fusion® HD Enzyme Premix kit.

pQN003: Fragment Pf3 was PCR amplified from heat-killed cells B. henselae with primers prQN005 and prQN014. dsRed was PCR amplified from pJC44 with primer prAK097 and prAK110. The PCR product generated by amplifying pBZ485_a_empty with prAK089 and prAK109 was used as a backbone. Three fragments were combined with the 5X In-Fusion® HD Enzyme Premix kit.

pQN006: gfpmut2 was PCR amplified from plasmid pCD353 with primers prQN015 and prQN016. The PCR product generated by amplifying pBZ485_a_empty with prAK089 and prAK109 was used as a backbone. Two fragments were combined with the 5X In-Fusion® HD Enzyme Premix kit.

pQN008: Plasmid pQN006 was digested with BamHI. Fragment sF1 was PCR amplified form heat-killed cells B. henselae with primers prQN021 and prQN022. The fragment and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN009: Plasmid pQN006 was digested with BamHI. Fragment sF2 was PCR amplified form heat-killed cells B. henselae with primers prQN018 and prQN017. The fragment and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN010: Plasmid pQN006 was digested with BamHI. Fragment sF3 was PCR amplified form heat-killed cells B. henselae with primers prQN019 and prQN020. The fragment and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN020: The minimal predicted P_GTA promoter was introduced into plasmid pQN006 by PCR with the primers prQN061 and prQN062. The plasmid was circularized with the 5X In-Fusion® HD Enzyme Premix kit.

pQN057: The PCR product generated by amplifying pQN006 with prQN111 and prAK109 was used as a backbone. F1 was PCR amplified from heat-killed cells B. henselae with primers prQ134 and prQN135. F1 and the backbone were combined with the 5X In-Fusion® HD Enzyme Premix kit.

pQN058: The PCR product generated by amplifying pQN006 with prQN111 and prAK109 was used as a backbone. F2 was PCR amplified from heat-killed cells B. henselae with primers prQ134 and prQN136. F2 and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN059: The PCR product generated by amplifying pQN006 with prQN111 and prAK109 was used as a backbone. F3 was PCR amplified from heat-killed cells B. henselae with primers prQ134 and prQN137. F3 and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN060: The PCR product generated by amplifying pQN006 with prQN111 and prAK109 was used as a backbone. F4 was PCR amplified from heat-killed cells B. henselae with primers prQ134 and prQN138. F4 and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN061: The PCR product generated by amplifying pQN006 with prQN111 and prAK109 was used as a backbone. F5 was PCR amplified from heat-killed cells B. henselae with primers prQ134 and prQN139. F5 and the backbone were ligated with the 5X In-Fusion® HD Enzyme Premix kit.

pQN007: Fragment F6 was PCR amplified from heat-killed cells B. henselae with primers prQN005 and prQN014. Plasmid pQN006 and F6 were restricted with BamHI and Xmal and ligated with RAPID DNA Dephos and Ligation Kit.

pQN047: The mutated -10-like sequence was introduced into plasmid pQN007 by PCR with primers prQN120 and prQN121. The plasmid was circularized with the 5X In-Fusion® HD Enzyme Premix kit.

pQN048: N’-3x-FLAG was introduced into pQN018 by PCR with primers prQN114 and prQN115. The plasmid was circularized with the 5X In-Fusion® HD Enzyme Premix kit.

pQN049: C’-3x-FLAG was introduced into pQN018 by PCR with primers prQN116 and prQN117. The plasmid was circularized with the 5X In-Fusion® HD Enzyme Premix kit.

pAK180: Terminator T_{BBa_1006} was introduced into pQN058 by PCR with primers prAK592 and prAK593. The linear PCR product was used for the direct transformation of E. coli.

Measuring expression of the BaGTA locus with flow cytometry

B. henselae strains were grown as described above. Bacteria were resuspended in M199 supplemented with 10% FCS and other additives, e.g inducer, if required (hereafter M199+), diluted to OD_{600 nm} = 0.008 and incubated in 24-well plate for a defined amount of time in a humidified atmosphere at 35°C and 5% CO₂. After a defined amount of time bacteria were collected, centrifuged, and washed once with PBS. Bacteria were resuspended in 4% PFA and fixed for 7 min at RT. The fixation was stopped by adding PBS supplemented with 2% FCS. Samples were kept at 4°C in the dark before flow cytometric analysis. GFP fluorescence in samples was recorded with the BD LSRFortessa™ Cell Analyzer at the FACS Core Facility, Biozentrum, the University of Basel. Data were analysed using software FlowJo (Version 10.8.1). Visualisation of the data was done with GraphPad prism 8.

GTA transfer assay

GTA transfer assays were done as co-cultivations of two B. henselae strains. B. henselae strains were grown as described above. Bacteria were resuspended in M199+ supplemented, if required, with inducer at specified concentrations and diluted to OD_{600 nm} = 0.008. For tested co-cultivation pair 1ml of each strain was mixed and a mixture was incubated in 24-well plate for 24h in a humidified atmosphere at 35°C and 5% CO₂. After 24h bacteria were collected and centrifuged. Bacterial pellets were resuspended in 30ul of M199+ and resuspensions were used for serial dilutions followed by plating onto CBA plates supplemented with SmGm or SmGmSp antibiotics. Bacteria were grown for 6-15 d in a humidified atmosphere at 35°C and 5% CO₂. Bacterial colonies were counted, and CFU/ml was determined for each co-cultivation pair. Frequency of double resistant bacteria (DRB) was calculated as the ratio of CFUs of recipient DRB (SmR GmR SpR) / total number of recipients (SmR GmR), yielding the transfer frequency per recipient cell.

Chromatin immunoprecipitation coupled with high throughput sequencing

B. henselae strains were grown as described above. Bacteria were harvested from plates and resuspended in 5 ml of Schneider’s Insect Medium (24.5 g/l Schneider’s Drosophila Powder Medium Revised, 5% (wt/vol) sucrose, 0.025 M HEPES, adjusted pH 7.25) supplemented with 10% FCS (hereafter Schneider’s Medium) (Guy et al., 2013; Riess et al., 2008). Bacteria were diluted to OD_{600 nm} = 0.01 in 500 ml Schneider’s Medium for each strain and grown at 35°C with shaking 135 rpm for 4d until OD_{600 nm} = 0.6-0.7. Bacteria were pelleted and pellets were resuspended in 50ml PBS. Bacteria were fixed with PFA with final concentration 1% and incubated for 10 min at RT. The fixation was stopped by adding glycine to a final concentration of 125 mM, and mixture was incubated for 20 min on ice. Fixed bacteria were centrifuged for 30 min at 4°C at 2000 x g, and bacterial pellets were frozen at -80°C. Thawed pellets were resuspended in 10 ml of ChIP Buffer (1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8.1), 167 mM NaCl, supplemented with cOmplete™Mini EDTA-free Protease Inhibitor Cocktail) and cells were lysed by 4 passages with French Press. The lysate was sonicated, and DNA size was checked with the 1.5% agarose gel. The lysate was centrifuged with the maximal speed and supernatant was collected. A 50 ul aliquot of supernatant was taken as “Input DNA” for the subsequent sequencing and data analysis. Following manufacturers’ protocols, remaining supernatant was immunoprecipitated ON with anti-FLAG Magnetic Agarose or anti-HA magnetic beads for 3xFLAG-tagged and HA-tagged proteins, respectively. Magnetic beads/agarose were immobilized and after removal of supernatant were sequentially rinsed with a panel of buffers: Low Salt Wash Buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 150 mM NaCl), High Salt Wash Buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 500 mM NaCl), LiCl Wash Buffer (250 mM LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.1)), TBS Buffer (50 mM Tris-HCl, 150 mM NaCl). The thawed Input control and immobilized Magnetic beads/agarose were mixed with 450 ul of TBS buffer, 15 µL of 5 M NaCl and 0.5 ul RNAse A (100mg/ml). The mixtures were incubated at 65°C ON. Then 4 µL of 0.5 M EDTA and 2 µL of 10 mg/mL Proteinase-K were added and the mixtures were incubated at 45°C for 2 h. The resulting mixtures were thoroughly mixed in 1:1 ratio with phenol/chlorophorm/isoamylalcohol combination and centrifuged at >10 000g for 1 min. The upper aqueous phase was transferred to a new tube, mixed in 1:1 ratio with chloroform and centrifuged at >10’000 x g for 1 min. The upper aqueous phase was transferred to a new tube and mixed with: (i) 25 ug of linear polyacrylamide; (ii) NaAc to a final concentration of 0.3 M; (iii) in 1:1 ratio with 100% isopropanol. The solution was centrifuged at >10’000 x g for 30min, washed twice with cold 70% EtOH. Pelleted DNA was dried and resuspend in water ON. Library preparation and sequencing were done at the Genomics Facility Basel, D-BSSE, ETH-Zurich.

Bioinformatic analysis

All alignments were done with the Clustal Omega and MUSCLE implemented in Geneious Prime 2020.2.3 (Edgar, 2004; Sievers et al., 2011). Mapping of ChIP-Seq data was done with bowtie2 (Langmead and Salzberg, 2012). Converting SAM to BAM, extracting unique binding reads and sorting were done with SAMtools (Danecek et al., 2021). Coverage plots were generated with bamCoverage and visualized with the Integrative Genomics Viewer (Ramírez et al., 2016; Robinson et al., 2011). Promoter predictions were done with BProm (Solovyev et al., 2011). Models of protein structures were either done with AlphaFold2 or taken from AlphaFold Protein Structure Database (Jumper et al., 2021; Varadi et al., 2022). Comparing the predicted models to Protein Data Bank (PDB) was done with the Dali server (Holm, 2022). Predictions of the secondary structure of the putative terminator was done with Mfold with the default parameters (Zuker, 2003). Calculations were performed at sciCORE (http://scicore.unibas.ch/) scientific computing center at the University of Basel.

Supplementary Figures

Suppl. Figure 1.