Abstract
Pyrrolobenzodiazepines (PBDs) are naturally occurring DNA binding compounds that possess anti-tumor and anti-bacterial activity. Chemical modifications of PBDs can result in improved DNA binding, sequence specificity and enhanced efficacy. More recently, synthetic PBD monomers have shown promise as payloads for antibody drug conjugates and antibacterial agents. The precise mechanism of action of these PBD monomers and their role in causing DNA damage remains to be elucidated. Here we characterized the damage-inducing potential of two C8-linked PBD bi-aryl monomers in Caulobacter crescentus and investigated the strategies employed by cells to repair the same. We show that these compounds cause DNA damage and efficiently kill bacteria, in a manner comparable to the extensively used DNA cross-linking agent mitomycin-C (MMC). However, in stark contrast to MMC which employs a mutagenic lesion tolerance pathway, we implicate essential functions for error-free mechanisms in repairing PBD monomer-mediated damage. We find that survival is severely compromised in cells lacking nucleotide excision repair and to a lesser extent, in cells with impaired recombination-based repair. Loss of nucleotide excision repair leads to significant increase in double-strand breaks, underscoring the critical role of this pathway in mediating repair of PBD-induced DNA lesions. Together, our study provides comprehensive insights into how mono-alkylating DNA-targeting therapeutic compounds like PBD monomers challenge cell growth, and identifies the specific mechanisms employed by the cell to counter the same.
Introduction
Persistent DNA damage can be problematic to cells across domains of life, from unicellular bacteria to multicellular eukaryotes. It can have deleterious effects on basic cellular processes as well as organismal functions, and subsequently lead to cell death (Surova & Zhivotovsky, 2013). However, when designed and used appropriately, DNA damage can also work as tools to eliminate hazardous microorganisms or malignant tissues (de Almeida et al., 2021).
DNA damage can be mediated by endogenous or exogenous agents, leading to broad spectrum of DNA modifications (Chatterjee & Walker, 2017). For example, methlymethansulfonate (MMS) leads to methylation on a single base (G, A or C) causing mono-alkylated adducts such as 7-MeG, 1-MeA and 3-MeC (Beranek, 1990). UV exposure induces covalent linkages between adjacent pyrimidines, creating intra-strand crosslinks including cyclobutane pyrimidine dimers (CPDs) and pyrimidine-(6-4)-pyrimidone photoproducts (Chatterjee & Walker, 2017). On the contrary, mitomycin-C (MMC) causes inter-strand crosslinks in addition to mono-alkylated adducts and intra-strand crosslinks (Bargonetti et al., 2010; Tomasz, 1995).
Pyrrolobenzodiazepines (PBDs) are a group of DNA damaging agents that bind to the minor groove of DNA and alkylate DNA in a sequence-specific manner (Gerratana, 2012; Mantaj et al., 2017). PBDs typically contain an aromatic A-ring, a diazepine B-ring and a pyrrolidone C-ring, a structure that fits into the minor groove of DNA. Once secured within the minor grove, an electrophilic imine moiety in the B-ring establishes a covalent link with the C2-NH2 group of guanine base on the DNA, preferably within Pu-G-Pu sequences. Several PBDs including anthramycin, tomaymycin and sibiromycin, naturally produced by actinomycetes are mono-alkylators of DNA and exhibit strong anti-microbial and anti-tumor properties (Gerratana, 2012; Hurley, 1977; Leimgruber et al., 1965)
Diversity in PBDs can be typically brought about by the variations in the A- and C-rings (Mantaj et al., 2017). For instance, the A ring can be functionalized with electron-donating sidechains at the C7- and C8-positions resulting in increased alkylating potential and DNA binding capability of PBDs (Mantaj et al., 2017; Thurston et al., 1999). Research in the last three decades has led to development of several synthetic PBDs by engineering modifications to the basic PBD structure resulting in enhanced sequence specificity, stability and functionality (Bose et al., 1992; Gregson et al., 2001; Mantaj et al., 2017; Rahman et al., 2012).
Of these, synthetic PBD dimers formed by linking two PBD monomers via their aromatic A-ring has the potential to form cross-links on DNA, similar to cross-links formed by mitomycin C (MMC) and have been extensively studied for their chemotherapeutic value (Kung Sutherland et al., 2013; Puzanov et al., 2011; Rahman et al., 2011). Few of these dimers are being evaluated as payloads for antibody drug conjugates (ADCs) in clinical trials to specifically target and kill tumor cells (Kung Sutherland et al., 2013; Morgensztern et al., 2019). However, there have been a number of failures in the clinical development of PBD-dimer containing ADCs due to their toxicity (Jackson et al., 2018). This has resulted in significant interest in PBD-pseudodimer containing only one N10-C11 imine group, PBD-monomers and structurally related pyridinobenzodiazepines (PDD) monomers (Hoffmann et al., 2020; Kovtun et al., 2018), which are considered less toxic as they can only mono-alkylate DNA. C8-linked PBD bi-aryl monomers are a group of synthetic PBD monomers where a heteroaromatic group like pyrrole or imidazole is directly linked to a phenyl group using C-C coupling or fused with a phenyl ring and these bi-aryl units are attached to the C8 position of the PBD A-ring. These PBD monomers exhibit a strong preference for GC-rich DNA, leading to formation of mono-alkylated adducts on guanine. One such monomer, KMR-28-39 has shown low nanomolar to picomolar in vitro cytotoxicity against a panel of cancer cell lines and in vivo anti-tumor activity to breast cancer and pancreatic cancer xenografts in mouse models (Rahman et al., 2013). Interestingly, many of these PBD monomers also exhibited anti-bacterial activity against a range of Gram-positive bacteria, including methicillin-resistant Staphylococcus (Rahman et al., 2012). Another set of PBDs containing a terminal heteroaliphatic ring has similarly shown excellent activity against a panel of multidrug resistant Gram-negative bacteria (Picconi et al., 2020), without noticeable toxicity against eukaryotic cells. The therapeutic potential exhibited by PBD bi-aryl monomers (Andriollo et al., 2018; Brucoli et al., 2016; Picconi et al., 2020; Rahman et al., 2012; Rosado et al., 2011) is encouraging as they appear to possess significant anticancer and antibacterial activity and their eukaryotic toxicity can be tuned by altering the C8-side chain to make them selective against either eukaryotic or prokaryotic cells.
While induction of cross-links and DSBs by PBD dimers (Arnould et al., 2006; Jenkins et al., 1994) and their repair via endonuclease ERCC1 and homologous recombination (Hartley et al., 2010; Xing et al., 2019; Zhong et al., 2019) are well characterized, there has been little research in elucidating the mechanism of action of PBD monomers beyond their ability to inhibit transcription factors (Corcoran et al., 2019; Kotecha et al., 2008; Rahman et al., 2013). The contribution of DNA-damage mediated effect on their overall cytotoxicity or antibacterial activity needs to be studied to properly evaluate their potential as chemotherapeutic agents. Furthermore, as efforts progress towards developing DNA-interactive PBD monomers as payloads for antibody drug conjugate, it is also important to elucidate the mechanism(s) that lead to repair of such PBD lesions, and identify the outcome of their repair on both survival and mutagenesis.
The main objective of this work was to identify the pathways(s) involved in the repair or tolerance of lesions induced by PBD monomers and assess the possible involvement of error-prone repair or tolerance mechanisms (such as translesion synthesis) that can impact damage-induced mutagenesis, contributing to development of resistance. We used the non-pathogenic Gram-negative bacteria Caulobacter crescentus as our model system. Unlike E. coli, this GC-rich organism shares several key genome maintenance features, including error-prone lesion tolerance mechanisms, with pathogenic bacteria such as Pseudomonas aeruginosa and Mycobacterium tuberculosis (Alves et al., 2017; Boshoff et al., 2003; Galhardo, 2005; Jatsenko et al., 2017; Warner et al., 2010). Using two prototypes (described in Fig. 1A and below), we found that C8-linked PBD bi-aryl monomers induced DNA damage and efficiently killed Caulobacter crescentus. Repair of these lesions was predominantly mediated by nucleotide excision repair; lack of repair led to generation of double-strand breaks and severely compromised survival. In contrast to MMC we found that mutagenic translesion synthesis was not essential for PBD monomer-mediated damage tolerance or repair. Taken together, our study uncovers, for the first time, the mechanisms involved in repair of DNA-monoalkylations induced by PBD monomers and their overall impact on genome integrity and survival of bacterial cells.
Results
Chemically synthesized C8-linked PBD bi-aryl monomers KMR-28-33 and KMR-28-35 cause DNA damage in Caulobacter crescentus
To assess the DNA damaging potential of PBD monomers, we chose two C8-linked PBD bi-aryl monomers (KMR-28-33 and KMR-28-35) (Fig. 1A) as prototypes. These compounds, which have propensity to bind to DNA (preferentially to GC-rich tracts (Rahman et al., 2013)), showed strong cytotoxic (Rahman et al., 2013) and antibacterial activity towards Gram-positive bacteria in the earlier studies (Rahman et al., 2012). As a comparison, we evaluated the repair of lesions caused by a conventional and well-characterized DNA cross-linking agent (mitomycin-C, MMC).
Indeed, molecular modelling of KMR-28-33 and KMR-28-35 binding to DNA also lent strong support to the possibility that these PBD monomers could result in DNA lesions, without distorting the DNA helix itself, as they snugly fit within the DNA minor groove (Fig. 1B, S2A. S2B). Our in vitro FRET-based DNA melting assays corroborated the binding of these PBD monomers to DNA (Fig. S2C). Given these observations, we investigate whether these PBD monomers cause DNA damage and if so, what mechanisms are employed by cells to repair the same. For this, we use the GC-rich model system Caulobacter crescentus, where we deleted genes involved in repair of specific types of DNA damage (Fig. S2E).
We first assessed cell survival upon treatment with KMR-28-33 and KMR-28-35 (synthesis described in Fig. S1A-B) in a wild type background. As a reference, we also exposed cells to well-characterized DNA damaging agents, that are known to induce specific types of damage (Fig. 2A and S2D). In particular, we compared the effects of KMR-28-33 and KMR-28-35 to mitomycin-C (MMC). Both KMR-28-33 and KMR-28-35 can only form DNA monoadducts while MMC can form intra-strand and inter-strand crosslinks in addition to monoadducts (Rahman et al., 2012; Warren et al., 1998).
We observed that both C8-linked PBD bi-aryl monomers caused cell death in a dose-dependent manner (Fig. 2A). We next asked whether these PBD monomers resulted in DNA damage. For this, we measured SOS response induction in wild type Caulobacter after treatment with the KMR-28-33 and KMR-28-35. We specifically quantified the expression of a fluorescence marker (YFP) induced under an SOS promoter (PsidA) integrated on the Caulobacter chromosome at the xyl locus (Badrinarayanan et al., 2015; Joseph et al., 2021). We found that exposure of cells to both PBD monomers resulted in significant accumulation of SOS-induced YFP (comparable to that observed in case of MMC, at concentrations that similarly affected cell growth in all three damages), suggesting that KMR-28-33 and KMR-28-35 caused DNA damage (Fig. 2B).
Requirement for RecA, but not the SOS response, in C8-linked PBD bi-aryl monomer-treated cells
Based on the above observations we wondered whether specific pathways under the SOS response were required to repair KMR-28-33 and KMR-28-35-mediated damage. For example, in the case of MMC, the error-prone translesion synthesis polymerase, DnaE2, has been found to be essential (Boshoff et al., 2003; Galhardo, 2005; Joseph et al., 2021). In support of the possibility that these PBD monomers indeed induce DNA damage, we found that cells lacking recA were compromised for survival upon treatment with both PBD-monomers (Fig. 3A). The difference in sensitivity for ΔrecA cells across KMR-28-33, KMR-38-35 and MMC suggested that there may be distinct repair mechanisms at play. We thus uncoupled key DNA damage-specific functions of RecA (recombination and SOS induction), to assess the contribution of the two towards survival. For this, we generated a strain where the SOS repressor, lexA, is deleted. To circumvent the problem of cell length elongation in this background, previous studies have additionally deleted the SOS-induced division inhibitor, sidA (Modell et al., 2011). In this constitutive ‘SOS-ON’ background, deletion of recA would predominantly eliminate its function in recombination. Thus, this triple deletion of lexAsidArecA can be used as a genetic read-out to test requirement of SOS vs recombination functions of RecA upon DNA damage treatment.
In case of MMC-damage, SOS-ON cells performed better than those deleted for recA (Fig. 3A), suggesting that a pathway regulated under the SOS response contributed significantly to cell survival under MMC damage. Indeed, this phenotype can be attributed to the expression of the TLS pathway (including the error-prone polymerase DnaE2) in SOS-ON cells, but not in cells deleted for recA. Previous studies have implicated an important role for this mechanism in tolerance of MMC-induced lesions, independent of recA (Galhardo, 2005; Joseph et al., 2021) (Fig. 3B). In contrast to MMC, we found that ΔrecA or ΔlexAsidArecA cells were similarly compromised in growth when treated with the C8-linked PBD bi-aryl monomers, suggesting that SOS function was not required for combatting KMR-28-33 and KMR-28-35-mediated damage (Fig. 3A). In line with this observation, we also found that TLS polymerase DnaE2 was not essential to tolerate C8-linked PBD bi-aryl monomer-mediated damage (Fig. 3B).
Given the sensitivity of ΔrecA cells (independent of SOS) to treatment with the KMR compounds, we asked whether recombination-mediated repair contributed to cell survival under PBD monomer-mediated damage. For this we deleted genes involved in specific recombination-based repair: a. recF, recO and recR that function in single-strand gap (SSG) repair and b. addAB and recN that function in double-strand break (DSB) repair (Rocha et al., 2005; Spies & Kowalczykowski, 2014). It is important to note that although we categorize the genes in this manner, there is also evidence to suggest that they may have overlapping functions (Pages, 2003).
In case of MMC damage, we found that cells deleted for recF, recO or recR were similarly sensitive to damage, and comparable with a recA deletion (Fig. 4A-B). Given the ΔrecA-like sensitivity in these backgrounds, it is tempting to speculate that these proteins may play a role in loading RecA at ssDNA gaps to enable SOS induction, apart from contribution to recombination-based repair. In line with this possibility, addAB and recN deleted cells were less compromised in growth when compared to the ΔrecA cells (Fig. 4A). In contrast to MMC damage, cells treated with KMR-28-33 or KMR-28-35 were similarly compromised in growth in the absence of SSG or DSB repair (Fig. 4A-B). Importantly, the sensitivity observed in case of these deletions mirrored that of recA, suggesting that both recombination pathways may contribute to repairing PBD bi-aryl monomer-induced damage.
Nucleotide excision repair (NER) is essential for survival under KMR-28-33 and KMR-28-35-induced DNA damage
Although we implicated a role for recombination in repair of C8-linked PBD bi-aryl monomer-mediated damage, these PBD monomers are thought to predominantly form DNA mono-alkylations and are not known to directly induce double-strand breaks. To estimate the DSB-inducing potential of KMR-28-33 and KMR-28-35, we adapted the Gam-GFP reporter system previously described in E. coli (Shee et al., 2013) to mark DSB ends in vivo in Caulobacter. Using this system, we estimated the percentage of cells with Gam localization in the presence and absence of the lesion-inducing damaging agents (Fig. S3A). In the absence of any damage, <1% cells had detectable foci. As anticipated, in the presence of the KMR compounds, this number increased only nominally to ∼5% after 2 h of treatment with the compounds. This was similar to observations made for MMC-induced damage as well (Fig. S3A). Interestingly loss of recN, required for recombination repair, only resulted in a modest increase in DSBs under KMR compounds or MMC (Fig. S3A). Together, this suggested to us that recombination may only be a minor repair pathway, with some other mechanism(s) likely enabling lesion repair or tolerance.
We thus wondered which lesion repair or tolerance pathways were required to repair damage induced by the C8-linked PBD bi-aryl monomers. As shown earlier, unlike MMC, we had ruled out a role for TLS polymerase DnaE2 (Fig. 3A-3B). Indeed, the lack of SOS response essentiality also eliminated a role for the other TLS polymerase, DinB (Galhardo, 2005; Joseph & Badrinarayanan, 2020), in this case (Fig. 3A). We next assessed survival in cells compromised for alkylation repair (alkB) (Colombi & Gomes, 1997) or mismatch repair (mutL) (Chai et al., 2021) and found that these pathways also did not contribute to repair of KMR-induced damage (Fig. S3B).
We thus turned to Nucleotide Excision Repair (NER). NER is an important mechanism of DNA lesion repair, that predominantly acts on helix-distorting lesions (Jia et al., 2009; Liu et al., 2011). Although damage induced by these PBD monomers is thought to not cause significant distortion to the DNA helix, we found that cells lacking uvrA (the lesion scanning component of the NER pathway) were severely compromised in survival under KMR-28-33 or KMR-28-35 damage (Fig. 5A). This was in contrast to MMC, where UvrA is required but not as essential, as seen in case of the C8-linked PBD bi-aryl monomers (Fig. 5A). Indeed, differential sensitivity of ΔuvrA strains to UV, MMS and norfloxacin damage further underscored the specificity of this repair pathway (Fig. S4A).
NER primarily functions via two ways: transcription-coupled repair (TCR) and global genomic repair (GGR). In case of TCR, Mfd plays a central role in recruiting Uvr components to the site of lesion for excision, followed by gap filling (C. Selby & Sancar, 1993; Strick & Portman, 2019). On the other hand, in case of GGR, UvrA is thought to scan and recognize lesions across the genome, and subsequently initiate repair (Kisker et al., 2013). We thus deleted the mfd homolog in Caulobacter and found that cells lacking the ability to engage in TCR were not as severely compromised in survival, when compared to ΔuvrA deleted cells (Fig. S4B). These data suggest that the GGR arm of NER is primarily responsible for repair of DNA lesions induced by KMR-28-33 and KMR-28-35. Indeed, alternate pathways for repair coupled to transcription, independent of Mfd, have also been proposed (C. P. Selby, 2017). We have not investigated the role of such mechanisms, which are currently not well-characterized or identified in Caulobacter.
Importantly, the absence of NER resulted in severe genome instability in cells treated with KMR-28-33 or KMR-28-35. As shown above, only 5-6% wild type cells treated with KMR compounds or with MMC had Gam localizations, indicating DSBs. In contrast, the lack of uvrA resulted in a significant increase in cells with Gam localizations in case of the PBD monomers, with 56% cells having DSBs upon KMR-28-33 treatment and 40% cells with DSBs on KMR-28-35 treatment (Fig. 5B). This was not found to be the case for MMC-treated cells, where uvrA deletion only led to modest increase in percentage cells with DSBs (6% in wild type to 10% in NER-compromised cells) (Fig. 5B). Such marked genome instability was observed only when NER action was compromised, as deletion of recN did not result in increase in localization of Gam-GFP (Fig. S3A).
Together these results highlight: a. the essentiality of NER in repairing C8-linked PBD bi-aryl monomer-mediated damage and b. the distinct mechanisms of DNA damage repair in case of the PBD monomers, when compared to an extensively studied DNA cross-linking agent, MMC (Fig. S5).
Discussion
Earlier studies have affirmed the specificity of action of pathways for repair of DNA damage. For example, MrfA and MrfB in Bacillus subtilis, and MmcA and MmcB in Caulobacter crescentus are essential for repair of only MMC-induced lesions (Burby & Simmons, 2019; Lopes-Kulishev et al., 2015). Similarly, NER has been implicated as the primary repair pathway in case of nitrofurazone damage in E. coli (Ona et al., 2009). In many instances, the requirement for different repair components is likely driven the structural variations in lesions induced by specific damaging agents (Cole et al., 2018; Ona et al., 2009; Williams et al., 2013). Such difference in function can be observed even within a pathway, when the type of lesion differs. Both prokaryotic and eukaryotic TLS polymerases exhibit substrate specificity which defines their lesion bypass properties including efficiency of bypass and fidelity (Inomata et al., 2021; Ippoliti et al., 2012; Prakash et al., 2005; Waters et al., 2009). In case of Caulobacter, requirement for TLS polymerase DnaE2 is relatively higher for repair of MMC lesions than UV lesions (Galhardo, 2005; Joseph et al., 2021).
Harnessing this specificity in function of repair pathways, in this study, we determined the damage-inducing potential of two C8-linked PBD bi-aryl monomers and delineated the strategies employed by bacterial cells to repair the same. Our results indicate that base modifications (in the form of mono-alkylated adducts) caused by the KMR compounds are predominantly repaired by NER and do not employ error-prone TLS mechanisms. Indeed, it is tempting to attribute the difference between MMC and the KMR compounds to repair/ tolerance of monoalkylations vs inter and intra-strand DNA crosslinks. While the C8-linked PBD bi-aryl monomers can only form mono-adducts, MMC treatment can result in mono-adducts as well as both inter and intra-strand DNA crosslinks. Thus, the relative contribution of various repair components could differ between lesions that are structurally and chemically distinct, but mechanistically similar.
In addition to NER, we find contribution of recombination-mediated repair to cell survival in case of the KMR compounds. The exact sequence of event(s) that lead to conversion of a mono-adduct into a DSB remains elusive. It is speculated that cellular processes like transcription, replication and even incomplete repair can lead to generation of single stranded gaps as well as double stranded breaks (Aguilera & Gaillard, 2014; Mehta & Haber, 2014). The observation that the NER mutant is far more compromised in survival and generates much higher proportion of DSBs than a recombination mutant supports the idea that recombination may act secondary to lesion repair via NER. Indeed, we did consider making a strain impaired in both NER and recombination to test this possibility. However, the very high sensitivity of the ΔuvrA strain to KMR compounds precludes our ability to do so with reliability. Future work aimed at quantitative estimation of the levels and types of DNA damage induced in vivo in all three cases (KMR compounds and MMC) will enable us to further discern the hierarchy of requirement and action of repair pathways.
In sum, our work highlights the importance of studying the mechanism of action of potential DNA-interactive therapeutics like PBD monomers in depth, to understand how they may affect cell growth and what strategies may be employed by the cell to respond to the same. For example, when considering a DNA damaging agent for therapeutic purposes, it is important to understand the fidelity of repair mechanisms that could be employed by the cells. Mutagenic repair can be a major source of stress-induced mutagenesis and subsequent development of resistance (Fitzgerald et al., 2017; Ippoliti et al., 2012; Joseph & Badrinarayanan, 2020). Our findings suggest that C8-linked PBD bi-aryl monomer-induced lesions are likely non-mutagenic, and are predominantly repaired by nucleotide excision repair, thus negating an important driver for development of chemoresistance and antimicrobial resistance. This contrasts with PBD dimers which are known to cause mutagenesis and non-selective toxicity, and makes the case for using PBD monomers as ADC payloads to overcome recent clinical failures observed with PBD dimers (Jackson et al., 2018). Identifying the dependency on NER (specifically Uvr components, which are restricted to bacteria) further opens up possibilities for considering inhibitors for Uvr components to use in combination with the PBD monomers. Indeed, chemical inhibitors for specific repair pathways have been identified previously, including for E. coli RecBCD, H. pylori AddAB and M. smegmatis Uvr proteins (Amundsen et al., 2012; Mazloum et al., 2011). Combining a DNA damaging drug with a small molecule inhibitor capable of dampening damage repair in the pathogen can potentiate the efficacy of the drug and reduce pleotropic cytotoxicity (Lim et al., 2019).
Materials and methods
Synthesis of KMR-28-33 and KMR-28-35
The PBD component of the hybrids was synthesized from vanillin as previously described in the literature (Rahman et al., 2013) and is summarized in Fig. S1A. A four-carbon linker was used to connect the PBD component with the non-covalently interactive subunits, as chains of this length had proven optimal in previous hybrid SAR studies (Wells et al., 2006). The linker was located at the C8 position of the molecule to allow an isohelical fit of the non-covalent component of the hybrid along the minor groove upon covalent PBD binding. After the synthesis of the PBD core, non-covalently interactive side chains were constructed using combinations of benzofused (benzothiophene, KMR-28-33), five membered heterocyclic structures (N-methyl pyrrole/N-methyl imidazole) and MPB (4-(1-methyl-1H-pyrrol-3-yl)benzenamine, KMR-28-35) moieties. The MPB subunit was synthesized using Suzuki-Miyuara conditions previously described (Rahman et al., 2013) (Fig. S1B). These moieties were linked via Steglich amide bond formation at positions which maintained the overall fit of the hybrid for the DNA minor groove (C2/C5 for benzofused, C1/C4 for heterocyclic components), and finally the N10/C11 imine component of the molecule was activated using tetrakis palladium and pyrrolidine.
Bacterial strains and growth conditions
Bacterial strains, plasmids and primers used in the study are listed in Table S1-S3. Chromosomal deletions and integrations were performed using either a two-step recombination method with a sacB counter-selection marker (Skerker et al., 2005) or using integrating vectors from Thanbichler et al. (Thanbichler et al., 2007). Transductions were carried out with FCR30 (Ely, 1991). Caulobacter crescentus cultures were routinely grown at 30°C in PYE (0.2% peptone, 0.1% yeast extract and 0.06% MgSO4). For strains expressing Gam-GFP under Pxyl, 0.3% xylose was added 3h prior to imaging.
Survival assay
Caulobacter cultures were grown in PYE to O.D600 of 0.3. Serial dilutions in 10-fold increments were made and 6 µl of each dilution (10−1 to 10−8) were spotted on PYE agar containing appropriate concentrations of different DNA damaging agents. For UV damage, serial dilutions of the culture were spotted on PYE agar plates and exposed to specific energy settings in a UV Stratalinker 1800 (STRATAGENE). Growth was assessed from the number of spots on the plates after two days of incubation at 30°C. Percentage survival for each strain was calculated by normalizing growth of that specific strain on different doses of DNA damage to that on media without DNA damage.
Fluorescence microscopy and image analysis
Saturated overnight cultures were back diluted in fresh PYE and allowed to grow at least for two generations (approx. 3h) until OD600 was 0.1. Images were taken without damage treatment (no damage control) and after treatment with specified doses of DNA damage. 1 ml aliquots of cultures were taken at specified time points, pelleted and resuspended in 100 µl of growth medium. 2 µl of cell suspension was spotted on 1% agarose pads (prepared in water). Imaging was performed on a wide-field epifluorescence microscope (Eclipse Ti-2E, Nikon) equipped with a 60X oil immersion objective (plan apochromat objective with NA 1.41) and pE4000 light source (CoolLED). Images were acquired with Hamamatsu Orca Flash 4.0 camera using NIS-elements software (version 5.1). For quantifying YFP induction under PsidA promoter, cells were segmented using Oufti (Paintdakhi et al., 2016) in MatLab, and florescence intensities normalized to cell lengths were extracted. Percentage cells with DSBs were quantified by counting cells with Gam-GFP foci using the Cell Counter plugin in ImageJ. Graphs were plotted in GraphPad Prism 7.
Molecular Modelling
The 3D structures of desired B-form DNA sequences from the NA1000 (Caulobacter crescentus) genome sequence (15 bp from dnaE ORF: 5’-ATCGGCAAGCTGGCC-3’, LexA box within the promoter of recA: 5’-GTTCGCAAGATGTTC-3’ and CCNA_RR0074 sRNA: 5’-CCCCTTCGCCCTCCT-3’,) were generated using PyMOL 2.5 structure Builder. For small molecule ligands used in this study, 3D structures were generated using Chem3D 20.0 program. The DNA structures were processed (energy minimization and addition of polar hydrogens) using MGLTools v1.5.7 (https://autodock.scripps.edu/). The grid box was configured for each DNA macromolecule to cover the whole length of the structure so that the ligand was able to find best possible binding sites along with the DNA structures including both the major and minor grooves. The small molecular ligands were also processed with the same tools. Finally, the molecular docking was performed using opensource AutoDock Vina v.1.2.0 (https://vina.scripps.edu/) (Trott & Olson, 2009). The default flexible docking parameters were kept for docking. The post processing of the output files was curated using PyMOL 2.5 and the molecular interactions were visualized using BIOVIA Discovery Studio Visualizer.
FRET-based DNA melting
All FRET duplexes and hairpins were purchased as pairs of complimentary or self-complimentary single-stranded oligonucleotides in lyophilised form from Eurogentec Ltd. The oligonucleotides were fluoro-tagged at the 5’ position with TAM and 3’ position with TAMRA. Sequences used were as follows; AT-rich sequence (seq-1): 5’-FAM-TAT-ATA-TAG-ATA-TTT-TTT-TAT-CTA-TAT-ATA-TAMRA-3’; GC-rich sequence (seq-2): 5’-FAM-TAT-AGG-GAC-AGC-CCT-ATA-3’, 3’-TAMRA-ATA-TCC-CTG-TCG-GGA-TAT-5’. Nuclease-free water was used to prepare stock solutions (20 μM) of the oligonucleotide hairpins/duplex strands. These stock solutions were diluted to concentrations of 400 nM using FRET buffer (50 mM potassium cacodylate, pH 7.4). The solutions were then heated to 85 °C/80 °C for five/ten minutes (hairpin/duplex solutions, respectively) using a heating block (Grant-Bio). The solutions were allowed to cool to room temperature overnight and cooled to -20 °C to complete the annealing process. Annealed stock solutions were diluted to concentrations of 100 and 10 nM using FRET buffer to prepare working solutions. PBD monomers, GWL-78 (Wells et al., 2006) and mitomycin C to be incubated with the DNA duplexes were dissolved in DMSO to form 5 mM solutions. Working solutions of PBD monomers and mitomycin C (5 μM and 1 μM) were prepared using FRET buffer. The working solutions of the compounds and DNA hairpins/duplexes were mixed (1:1 ratio, 25 μL of each solution) in the wells of a 96 well plate (Bio-Rad). The wells were covered and placed in a DNA Engine Opticon system for melting. The samples were heated over a range of 30-100 °C, with fluorescence readings (incident radiation 450-495 nm, detection 515-545 nm) taken at intervals of 0.5 °C. Experimental data was imported into Origin (OriginLab Corp.), where the curves were smoothed and normalised. Using a script, the point of inflection of the first derivative of the melting point for each sample on the plate was calculated. The difference between the melting temperature of each sample and that of the blank (i.e., the ΔTm) was used for comparative purposes. Mean is shown from three independent repeats.
Author contributions
AJ led the project, generated tools and reagents, carried out in vivo experiments in Caulobacter and conducted data analysis. SD contributed tools and reagents, and carried out in vivo experiments. KN and MMH carried out experiments pertaining to the KMR compounds synthesis and in vitro characterization. RL and TL contributed tools and reagents. KMR and AB conceived and supervised the project. AJ, KMR and AB procured funding and wrote the manuscript, with feedback from all authors.
Declaration of interests
None declared.
Supplementary Information
Supplementary Figures
Acknowledgements
AJ and AB thank members of the AB lab for feedback on the work. AJ acknowledges support from DST N-PDF SERB. AB acknowledges support from the DBT-IYBA grant and intra-mural funding from NCBS-TIFR. TL acknowledges support from the Royal Society University Research Fellowship Renewal (URF\R\201020) and BBSRC (BBS/E/J/000PR9791).