Structure of the WYL-domain containing transcription activator, DriD, in complex with ssDNA effector and DNA target site

Abstract

Transcription regulators play central roles in orchestrating responses to changing environmental conditions. Recently the Caulobacter crescentus transcription activator DriD, which belongs to the newly defined WYL-domain family, was shown to regulate DNA damage responses independent of the canonical SOS pathway. However, the molecular mechanisms by which DriD and other WYL-regulators sense environmental signals and recognize DNA are not well understood. We showed DriD DNA-binding is triggered by its interaction with ssDNA, which is produced during DNA damage. Here we describe the structure of the full-length C. crescentus DriD bound to both target DNA and effector ssDNA. DriD consists of an N-terminal winged-HTH (wHTH) domain, linker region, three-helix bundle, WYL-domain and C-terminal WCX-dimer domain. Strikingly, DriD binds DNA using a novel, asymmetric DNA-binding mechanism that results from different conformations adopted by the linker. Although the linker does not touch DNA, our data show that contacts it makes with the wHTH are key for specific DNA binding. The structure indicates how ssDNA-effector binding to the WYL-domain impacts wHTH DNA binding. In conclusion, we present the first structure of a WYL-activator bound to both effector and target DNA. The structure unveils a unique, asymmetric DNA binding mode that is likely conserved among WYL-activators.

Graphical Abstract

Graphical Abstract.

Introduction

Bacteria are continually exposed to environmental and metabolic stresses that require rapid responses to ensure survival. Transcription factors that integrate metabolic cues to either repress or activate transcription of particular genes play key roles in mediating these responses. One potentially deadly stress that bacteria encounter is DNA mutation and damage. Such damage requires the bacteria to express genes that not only repair DNA damage but also transiently halt cell division until the damage can be resolved (1,2). This tight coordination is necessary to prevent genomic instability. The best characterized DNA damage response system in bacteria is the SOS pathway, which is initiated by RecA upon its interaction with ssDNA (3–9). ssDNA is generated during DNA damage and hence acts as an environmental signal for such damage (7,10). RecA forms filaments upon ssDNA binding, which stimulates autocleavage of the global SOS repressor, LexA, allowing for the expression of cell cycle control and DNA repair genes (7–9). Despite the near universal presence of the bacterial SOS response, studies in Escherichia coli, Bacillus subtilis and Caulobacter crescentus have shown that cells unable to induce an SOS response can still become filamentous after DNA damage, indicating the presence of alternative pathways for blocking cell division (11–16). Most of these systems have not been characterized. However, recent investigations in C. crescentus identified an SOS independent DNA damage pathway activated by a protein called DriD (16).

Studies on DriD showed it functions as a transcription activator and drives the expression of a unique C. crescentus cell division inhibitor, DidA (16). Unlike many cell division inhibitors in bacteria, which target the FtsZ Z-ring forming protein, DidA halts the cell cycle by binding the essential late-arriving divisome component, FtsN, disrupting the assembly of the division machinery (16). The DNA target site bound by DriD within the didA promoter was mapped to a 20 bp site containing a pseudo-palindromic site (17). Based on this binding sequence, additional DriD binding sites within promoters were identified and subsequent studies showed that DriD binds these sites and activates transcription of genes expressing recA, recJ, bapE and dnaG (17). DriD also was shown to bind a site within the promoter controlling its own expression (17). Comparison of all the DriD binding sites revealed a DNA binding consensus of ATACGAC(X)₇GTCGTAT, where underlined nucleotides are highly conserved among DriD DNA targets. These data thus indicate that DriD coordinately regulates multiple genes involved in the DNA damage response and cell division. DriD is a 327-residue protein with a ∼70-residue predicted N-terminal winged helix-turn-helix (wHTH) motif and larger, C-terminal domain. Within the C-terminal domain of DriD is a WYL-motif, which is named for a conserved Trp-Tyr-Leu sequence that was first found in CRISPR-Cas associated proteins (17–28). Later bioinformatic analyses showed that WYL-containing proteins are widespread in bacteria and act as regulators in multiple cellular contexts (22,24). The few structures that have been determined of WYL-domain containing proteins thus far reveal that the WYL-domain is composed of an Sm-like fold arranged into a β-sandwich preceded by a helix (29). Most WYL-domain proteins function as integrators of cellular processes, however, the specific cellular cues or ligands utilized by most of these proteins remain unknown.

Analyses of the co-occurrence of specific domains and topology in WYL-proteins led to their classification into nine groups (22,24). The majority of WYL-domain proteins are grouped into the A and C categories (22,24). In addition to predicted wHTH motifs, the A/C class proteins contain WYL protein C-terminal (WCX) domains (22,24). Proteins in the second largest group, class B, also contain N-terminal wHTH motifs linked to WYL domains. But these proteins lack WCX domains (22,24). The D, E and F class proteins feature the same wHTH-WYL-WCX architecture as in the A/C classes but differ in sequence and connecting regions. The remaining classes, which account for less than 10% of identified WYL-domain proteins, lack wHTH motifs. The predominance of wHTH domains in WYL-domain proteins suggests that they function primarily as DNA-binding proteins. Indeed, recent progress towards understanding WYL-domain proteins has revealed that many are transcription regulators (17–25,27,28). To date, these investigations have revealed two major roles for these WYL-domain regulators, which is in bacterial immunity and DNA damage responses (17–25,27,28). Interestingly, the data indicate that WYL regulators of bacterial immunity processes function as transcription repressors (22,27,28) while those playing roles in DNA damage responses act as transcription activators (16,17,22–24). Like other WYL proteins, both classes of WYL transcription regulators are proposed to sense specific environmental signals via their WYL-motifs. Effectors have yet to be illuminated for WYL-containing repressors. However, ssDNA has been shown to bind and activate the functions of the class A WYL-activators DriD and PafBC (17,23,24). Structures were recently reported for a WYL-domain containing fragment of DriD, DriD(126–327), bound to ssDNA, representing the first structure of an effector bound to a WYL-transcription regulator (17).

To date, structures of a full length (FL) WYL-domain transcription activator bound to both effector and DNA are lacking. Hence, the role of effector binding to the WYL-domain in cognate DNA binding remains unclear as does the mechanism by which WYL-domain containing activators, such as DriD, bind specifically to target DNA. To gain insight into these questions, we performed functional and biochemical studies on the C. crescentus DriD protein and solved the crystal structure of the ternary complex of FL DriD bound to ssDNA and target DNA. The structure shows that FL DriD is composed of five regions, a wHTH, linker region, three-helix bundle, WYL and WCX domain. The structure also revealed a novel asymmetric mode of DNA binding and shows how DriD achieves specific recognition of its target DNA site. Residues involved in ssDNA binding are conserved in WYL transcription activators, suggesting that the mechanism of effector binding is conserved among these proteins. Thus, these studies provide key molecular insight into effector and cognate DNA binding by a WYL-activator.

Materials and methods

Growth conditions

All Caulobacter strains were grown in rich medium (PYE) at 30°C. Antibiotics were used at the following concentrations for strain construction and plasmid maintenance in Caulobacter strains in liquid:plates—kanamycin 5 μg/ml:25 μg/ml; oxytetracycline 1 μg/ml:2 μg/ml. Transformations and transductions were performed as previously described (30).

Plasmid and strain construction

Plasmids and Caulobacter strains used in this study are listed in Supplementary Tables S1 and S2, respectively. DNA oligonucleotides used in strain construction are listed in Supplementary Table S3. C. crescentus CB15N genomic DNA was used as template for PCR amplifications unless noted otherwise. PCR was performed with Phusion HF DNA polymerase with 5 × Phusion GC reaction buffer (NEB). Each reaction contained 10 μl of buffer, 4 μl of dNTPs (final concentration of 200 μM), 5 μl of a 10 μM forward and reverse primer mix, 50 ng of template, 10 μl of 3 M betaine monohydrate (Sigma), 1 μl of DMSO, 0.5 μl (1 unit) of polymerase, and nuclease-free water to 50 μl. Two-step cycling was performed as follows: 30 s at 98°C, 34 times for 10 s at 98°C and 30 s/kb at 72°C, and 5 min at 72°C. Fusion PCR was performed similarly with 50 ng of the largest template fragment and equimolar amounts of the shorter template fragments, with an additional annealing step of 20 sec at 60°C. All PCR products were digested with the noted restriction enzymes and ligated into the double-enzyme cut corresponding plasmid. Five microliters of ligation reactions was subsequently transformed into chemically competent Escherichia coli DH5α cells. All resulting plasmids were verified by Sanger sequencing.

To generate the plasmids pRVMCS-2::P_driD-driD(R37A), pRVMCS-2::P_driD-driD(R41A), pRVMCS-2::P_driD-driD(K62E), pRVMCS-2::P_driD-driD(T61A), pRVMCS-2::P_driD-driD(E40A), pRVMCS-2::P_driD-driD(E40K), pRVMCS-2::P_driD-driD(R8A) and pRVMCS-2::P_driD-driD(R8E), the region spanning the driD promoter to 15 bp after the stop codon of driD was amplified in two parts, using oKRG317 paired with either oKRG559, oKRG561, oKRG563, oKRG565, oKRG567, oKRG567, oKRG570 or oKRG570, respectively, for the first half and oKRG316 paired with either oKRG558. oKRG560, oKRG562, oKRG564, oKRG566, oKRG568, oKRG569 or oKRG571, for the second half. The upstream and downstream halves of driD were combined using fusion PCR with primers oKRG316 and oKRG317, gel-purified, digested and ligated into an SacI/SacII cut pRVMCS-2(kan^R).

To generate plasmids pMT687-P_didA(C₄/G₁₂:T/A)-yfp::kan^R, pMT687-P_didA(G₂/C₁₄:A/T)-yfp::kan^R, pMT687-P_didA(T₃/G₁₃:G/C)-yfp::kan^R, pMT687-P_didA(N₇:GC-rich)-yfp::kan^R and pMT687-P_didA(C₁/G₁₅:T/A)-yfp::kan^R, mutations were introduced into the vector pMT687-P_didA-yfp::kan^R with the Q5^® Site-Directed Mutagenesis Kit (NEB, cat # E0554S) following the manufacturer's protocol, using the primer pairs oKRG580/oKRG581, oKRG582/oKRG583, oKRG586/oKRG587, oKRG588/oKRG589 and oKRG592/oKRG593, respectively.

To generate strain KG570, the insert in plasmid pNPTS138-ΔdriD::tet^R was introduced by two-step recombination into CB15N. To generate strains KG571-578, the plasmids pRVMCS-2::P_driD-driD(R37A), pRVMCS-2::P_driD-driD(R41A), pRVMCS-2::P_driD-driD(K62E), pRVMCS-2::P_driD-driD(T61A), pRVMCS-2::P_driD-driD(E40A), pRVMCS-2::P_driD-driD(E40K), pRVMCS-2::P_driD-driD(R8A) and pRVMCS-2::P_driD-driD(R8E) were introduced into strain KG570 via electroporation, respectively.

To generate strains KG615 and KG616, the plasmid pMT687-P_didA-yfp::kan^Rwas introduced into CB15N and ML3757 via electroporation, respectively. To generate strains KG617-621, the plasmids pMT687-P_didA(C₄/G₁₂:T/A)-yfp::kan^R, pMT687-P_didA(G₂/C₁₄:A/T)-yfp::kan^R, pMT687-P_didA(T₃/G₁₃:G/C)-yfp::kan^R, pMT687-P_didA(N₇:GC-rich)-yfp::kan^R and pMT687-P_didA(C₁/G₁₅:T/A)-yfp::kan^R were introduced into CB15N via electroporation, respectively.

Expression and purification of C. crescentus full length (FL) DriD and DriD mutants

The gene encoding C. crescentus DriD was purchased from Genscript Corporation and subcloned into pET15b such that an N-terminal His-tag was expressed on the protein for purification (Piscataway, NJ, USA:http://www.genscript.com). E. coli C41(DE3) cells were transformed with the expression vector. Cells with the expression construct were grown at 37°C in LB medium with 0.17 mg/ml ampicillin to an OD₆₀₀ of 0.6, then induced with 1.0 mM isopropyl β-d-thiogalactopyranoside (IPTG) at 15°C overnight. Cells were harvested by centrifugation and resuspended in a buffer composed of 50 mM Tris-Cl pH 7.5, 800 mM NaCl, 5% (v/v) glycerol, 0.5 mM β-mercaptoethanol (βME), with 1X protease inhibitor cocktail and DNase I (∼10 μl of 100 mg/ml DNase I per reconstitution). The resuspended cells were then disrupted with a microfluidizer and cell debris was removed by centrifugation (15 000 rpm, 4°C, 45 min). The supernatant was loaded onto a cobalt NTA column and the column was washed with 300–500 ml of 5 mM imidazole in the resuspension buffer. The protein was eluted in steps of increasing imidazole (10, 20, 30, 50, 100, 200, 500 mM imidazole in a buffer consisting of 50 mM Tris–Cl pH 7.5, 300 mM NaCl, 5% (v/v) glycerol, 0.5 mM β-mercaptoethanol (βME)). Fractions were analyzed by SDS-PAGE and those containing the protein were combined. Proteins were further purified by size exclusion using an S75 column. His-tags were cleaved by thrombin digestion overnight at 37°C using a thrombin cleavage capture kit. The cleaved His-tags were removed by loading cleavage reactions onto a Ni-NTA column and collecting the flow through. Tag-free proteins were concentrated using centricons with a 30 kDa MW cutoff. Fractions were analyzed by SDS-PAGE and those containing the protein were combined. DriD mutants were obtained from Genscript. These mutants were DriD(L10E), DriD(R37A), DriD(R41A), DriD(E40A), DriD(K62E), DriD(E40K), DriD(T61A-K62A) and DriD(V73P-F74P) and all were expressed and purified as per the WT protein.

Crystallization and structure determination of C. crescentus DriD-ssDNA-21 bp DNA complex

For crystallization, tag-free FL DriD was concentrated to ∼30 mg/ml and was mixed with 1 mM 7mer ssDNA (5′-TGTCTAT-3′). The complex was then mixed at a ratio of 1 dimer of DriD to 1 DNA duplex of a double stranded DNA containing a DriD consensus DNA site (top strand: 5′-ATACGACAGTTACTGTCGTAT-3′). Wizard screens I to IV and the Natrix screen were used for screening at room temperature (rt) by the hanging drop vapor diffusion method. Crystals were produced of DriD-ssDNA-DNA by mixing the complex 1 to 1 with a solution consisting of 0.1 M Tris pH 7.0, 1. 0 M sodium citrate and 200 mM NaCl. The crystals grew at rt and obtained their maximum size in 3 weeks. Crystals were cryopreserved by dipping them in a solution consisting of the crystallization reagent supplemented with 20% (v/v) glycerol before plunging into liquid nitrogen. X-ray intensity data were collected at ALS beamline 5.0.2 and processed with XDS (31) (Table 1). A second crystal form of the complex was obtained by mixing it 1:1 with a crystallization solution consisting of 1.5 M ammonium sulphate, 0.1 M Tris pH 7.5. These crystals took several months to grow and adopted a hexagonal appearance. The crystals were cryo-preserved straight from the drop but only diffracted to 5–6 Å. Hence, the crystals were subjected to dehydration wherein the drops containing crystals were placed over nearly dry wells and let sit for several days before harvesting for data collection. This improved the diffraction, allowing data to be collected to 3.54 Å (Table 1).

Table 1.

Data collection and refinement statistics: C. crescentus DriD-ssDNA-DNA complexes

	DriD-ssDNA-DNA complex/crystal form 1	DriD-ssDNA-DNA complex/crystal form 2
Data collection
Space group	P21	P6522
PDB code	8TP8	8TPK
Cell dimensions
a, b, c (Å)	69.3, 178.9, 92.2	115.1, 115.1, 300.3
α, β, γ (°)	90.0, 95.2, 90.0	90.0, 90.0, 120.0
Resolution (Å)	91.77–2.74  (2.89–2.74)*	49.16–3.54  (3.67–3.54)
R sym or Rmerge	0.092 (0.657)	0.137 (1.480)
R pim	0.060 (0.446)	0.038 (0.408)
I/σI	7.6 (1.5)	9.3 (1.3)
Completeness (%)	96.1 (87.0)	92.5 (94.6)
Redundancy	3.3 (3.0)	10.1 (10.1)
CC(1/2)	0.990 (0.635)	0.999 (0.716)
Refinement
Resolution (Å)	69.0–2.74	47.30–3.54
No. reflections	56088 (3276)	14675 (1434)
R work/Rfree (%)	20.1/24.7	25.9/29.8
R.m.s. deviations
Bond lengths (Å)	0.009	0.004
Bond angles (°)	1.15	0.825
Ramachandran analyses
Favored (%)	92.5	91.9
Disallowed (%)	0.00	0.00

*Values in parentheses are for highest-resolution shell.

The structure of the P2₁ DriD-ssDNA-DNA complex was solved first by molecular replacement using the structure of the DriD C-domain (pdb code: 7TZV) as a search model. The DriD(135–327) dimer located two solutions consistent with the presence of two DriD-ssDNA-DNA complexes in the crystallographic asymmetric unit (ASU). Refinement with the partial model revealed density for the central helical domain, which was then added in Coot (32,33). Following refinement with this model, electron density became evident for the DNA binding domains and DNA. After multiple rounds of refitting in Coot (33) and refinement in Phenix (32), the ssDNA was visible. The structure was refined to convergence in Phenix (32). Molecular replacement was then used to solve the structure of the P6₅22 crystal form. This crystal form contains one DriD–ssDNA–DNA complex in the ASU. One DriD–DNA complex (the ssDNA was removed from the starting model) from the P2₁ crystal form was used as a search model and produced a molecular replacement solution containing DriD and the cognate DNA. After multiple rounds of refitting and refinement, clear electron density for three nucleotides of ssDNA bound to each DriD subunit was revealed (see Table 1 for final refinement statistics).

Fluorescence polarization (FP) binding experiments

Fluorescence polarization was used to determine the binding affinity of WT or mutant C. crescentus DriD and double stranded 5′ fluorescein tagged driD promoter DNA (top strand: 5′-ATACGACAGTTACTGTCGTAT-3′). Experiments were performed using the Panvera BEACON-2000 with the excitation wavelength at 490 nm and fluorescence measured at 530 nm. WT or mutant C. crescentus DriD were serially diluted into binding buffer containing 12.5 mM Tris pH 7.5, 100 mM NaCl, 5 mM MgCl₂, 5% (v/v) glycerol and 1 mM βME. Experiments were blanked with binding buffer plus 10 μM nonfluoresceinated 7mer ssDNA (5′-TGTCTAT-3′). An average of five reads of binding buffer plus 10 μM ssDNA and 1 nM of the 5′ fluorescein tagged driD promoter DNA was used as the baseline, subtracted from subsequent reads. Increasing concentrations of WT or DriD(L10E), DriD(R37A), DriD(R41A), DriD(E40A), DriD(E40K), DriD(K62E), DriD(T61A-K62A) or DriD(V73P-F74P) were titrated into the tube containing binding buffer, 10 μM ssDNA, and 1 nM 5′ fluorescein tagged driD promoter DNA and polarization was recorded. Three technical replicates were performed for each sample. Normalized change in polarization was plotted against increasing protein concentration and binding affinities were calculated by fitting the curves with GraphPad Prism. The error in K_d was determined as the SD between the calculated K_ds for three runs.

To measure the binding affinity of WT DriD for mutant DNAs, the same experiments were performed as done above, except 1 nM of different fluorescein tagged DNAs were used. To test the importance of DriD contacts to specific bases the following fluoresceinated DNA mutants were analyzed in FP. Binding assays (top strands: 5′-ATACAATAGTTACTATTGTAT-3′, 5′-ATACGATAGTTACTATCGTAT-3′, 5′-ATACGGCAGTTACTGCCGTAT-3′ and 5′-ATATGACAGTTACTGTCATAT-3′. with a 5′ fluorescein tag. To test the importance of the cognate DNA AT rich spacer region in DriD binding, the GC rich mutant, top strand: 5′-ATACGACGGCCGCCGTCGTAT-3′, with a 5′ fluorescein tag was used. These experiments included 10 μM ssDNA. The error in K_d was determined as the SD between the calculated K_ds for three runs.

To analyze ssDNA binding by DriD, the following ssDNA sites with 5′ fluorescein tags were used (F-TGTCTATGTC, F-ATATATAATA, F-CGCGCGCGCG, F-TTTTTTTTTT and F-CCCCCCCCCC). FL DriD was titrated into the reaction cell with 1 nM of each fluoresceinated DNA in a buffer composed of 12.5 mM Tris pH 7.5, 100 mM NaCl, 5 mM MgCl₂, 5% (v/v) glycerol and 1 mM βME. The change in polarization was plotted against increasing protein concentration and binding affinities were calculated by fitting the curves with KaleidaGraph. The error in K_d was determined as the SD between the calculated K_ds for three runs.

BS3 crosslinking experiments

BS3 crosslinking experiments were performed with the BS3 (bis(sulfosuccinimidyl)suberate), No-Weigh Format (Thermo Scientific A39266) crosslinker. For these experiments, 100 mM of BS3 was prepared at rt by adding 35 ul of crosslinking buffer (12.5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 7.5, 100 mM NaCl, 5% (v/v) glycerol, 5 mM MgCl₂) to the appropriate BS3. Dilutions were made to produce 50, 25, 10, 5 and 2.5 mM BS3 crosslinking solutions. Purified C. crescentus DriD was buffer exchanged 3 times into the crosslinking buffer, then 1 mg/ml of DriD was mixed with 1 ul of each BS3 dilution, resulting in final concentration of BS3 of 5, 2.5, 1, 0.5 and 0.25 mM. 1 ul of crosslinking buffer was added to the tube containing 0 mM of BS3 crosslinker. Crosslinking reactions were incubated at rt for 1 h then the reactions were quenched with 2 ul of 1 M Tris pH 7.5. After thorough mixing with the quenching reagent, the crosslinking reactions were incubated at rt for 15 min. Then 4 ul of 4× SDS loading buffer was added to the crosslinking reactions and boiled at 98°C for 5 min. Crosslinking reactions were run on a NuPAGE 7% Tris-Acetate gel (Invitrogen EA03552BOX) for 1 h and 15 min at 150V. HiMark Pre-stained Protein Standard (Invitrogen LC5699) was used to assess MW.

Mass photometry (MP) experiments

MP analysis was done using a Refeyn OneMP instrument (Oxford, UK). Contrast-to-mass calibration was achieved by measuring the contrast of proteins in the native marker protein standard mixture (NativeMark Unstained Protein Standard, Thermo Fisher) to generate a standard calibration curve. The experiments were performed using glass coverslips, which were thoroughly washed several times with Milli-Q water and isopropyl alcohol. Silicone gaskets were positioned on the glass surface for sample loading. FL C. crescentus DriD in 50 mM Tris pH 7.5, 300 mM NaCl, 5% glycerol was diluted in the buffer to final concentrations of 20 nM prior to sample analyses. 6000 frame movies were recorded using AcquireMP software and large field-of-view acquisition settings. The MP data was analyzed with DiscoverMP software to produce mass values.

β-Galactosidase assay

Indicated strains were grown in biological triplicate up to mid-log (OD₆₀₀ = 0.2–0.3) and treated with (or without) 15 μg/ml zeocin for 45 min. Harvested cells were permeabilized by adding 100 μl of chloroform to 900 μl of cells followed by vortexing. Cells were incubated for 15 min at 30°C prior to addition of ortho-nitrophenyl-β-galactoside. The assay and subsequent activity calculations were done as previously described (34).

Size exclusion chromatography (SEC) experiments

SEC studies were performed using a Hiload 26/600 SUPERDEX™ 75 pg column and an AKTA prime plus. The buffer used for the runs was 25 mM Tris pH 7.5, 300 mM NaCl, 5% (v/v) glycerol and 0.5 mM βME. Fractions containing WT DriD, DriD(L10E) and DriD(V73P-F74P), all with His-tags, were concentrated using Sigma-Millipore concentrators (Amicon) prior to column application. Samples were loaded using a 1 ml (final volume) syringe. The SEC studies were carried out on samples at 20–24 mg/ml. The elution volumes for each sample was compared with a series of protein standards to determine molecular weights. The protein standards used for calculation of the standard curve were aprotinin (6.5 kDa), cytochrome c (12.4 kDa), carbonic anhydrase (29.0 kDa), and bovine serum albumin (66.0 kDa). Elution parameter Kav was calculated by Kav = (elution volume for the standard-void volume)/(column volume-void volume).

Fluorescent microscopy of DriD motif mutants

Bacterial cultures were grown up overnight in PYE supplemented with kanamycin and back diluted to an OD₆₀₀ of 0.05. Then cultures were grown to mid-exponential phase (OD₆₀₀ = 0.2–0.3) and treated with 20 μg/ml zeocin for 45 min. Cells were washed in 1× PBS prior to imaging. 1 μl of resuspended cells were spotted on PBS + 1% agarose pads in a 35 mm Glass bottom dish. Phase contrast and epifluorescence images were taken on a Zeiss Axios Observer 7 with a 100×/1.4 oil immersion objective and a Prime 95B BI CMOS camera with Zen Imaging software. Fluorescent intensity for 200 cells per sample were measured using MicrobeJ software.

Bioinformatic search for putative DriD motifs

Homologs of DriD from Caulobacter crescentus NA1000 were identified across bacteria using the NCBI blastn homology search tool. Positive hits were retained if the e-value was less than 1e-20 and the query coverage length was greater than 25%. DNA sequences of each of the top putative homologs were extracted from each genome. Potential DriD DNA binding motifs were identified in each promoter region of the respective DriD homolog by searching for CGACNNNNNNNGTCG, allowing for up to two mismatches and a gap of either six or seven nucleotides between the palindromic sites. For unannotated genomes, putative start sites for the DriD homologs were individually predicted based on similarity in gene length and adjacency to potential ribosome binding sites.

Results and discussion

Overall crystal structure of C. crescentus DriD bound to ssDNA and DNA site

Our previous studies showed that the WYL-protein, DriD, binds its cognate DNA site with high affinity when ssDNA is present (17). To elucidate the molecular mechanism for these interactions, we crystallized FL C. crescentus DriD in the presence of a seven nucleotide ssDNA (5′-TGTCTAT-3′) and a palindromic cognate DNA site (top strand: 5′-ATACGACAGTTACTGTCGTAT-3′) (17). Two crystal forms were obtained that diffracted to 2.74 and 3.54 Å resolution (Table 1). The 2.74 Å structure was solved first and contains two DriD-ssDNA-DNA complexes in the crystallographic asymmetric unit (ASU) (Materials and Methods). The structure was refined to final R_work/R_free values of 20.1%/24.7% (Table 1; Figure 1A). The 3.54 Å structure contains one DriD-ssDNA-DNA complex in the ASU. This structure was solved using a complex of the 2.74 Å structure as a search model. The structure shows DriD binds its palindromic DNA site as a dimer. Formation of the DriD dimer buries a significant 4300 Ų of protein surface from solvent. That DriD functions as a dimer was supported by size exclusion chromatography (SEC) experiments, which revealed a calculated molecular weight (MW) of 71.8 kDa (compared to the theoretical MW of 72 kDa and 76 kDa with the His-tag) as well as BS3 (bis(sulfosuccinimidyl)suberate (BS3) crosslinking experiments (Figure 1B; Supplementary Figure S1A). Mass photometry, which can be performed at close to physiological concentrations, was also employed and revealed one distribution (accounting for 92% of all counts) with a molecular weight of 76 ± 8.4 kDa, also consistent with a dimer (Supplementary Figure S1B).

Figure 1.

C. crescentus full length DriD-ssDNA-DNA complex. (A) Cartoon diagram showing two views of the DriD–ssDNA–DNA complex. One DriD subunit is colored cyan and the other slate. The ssDNA bound by the WYL-domains of each subunit are shown as sticks and the DNA as a cartoon. The figure underscores the asymmetric binding mode utilized by DriD. (B) Size exclusion chromatography (SEC) analysis of FL DriD. The x and y axes are Log MW and elution parameter (K_av). DriD eluted at a calculated MW of 71.8 kDa (pink diamond) consistent with a dimer. The standards used for calculation of the standard curve are (indicated by slate circles) aprotinin (6.5 kDa), cytochrome c (12.4 kDa), carbonic anhydrase (29.0 kDa), and bovine serum albumin (66.0 kDa). (C) Domain organization of DriD. The DriD domains are colored differently and labeled for one subunit. The other subunit is shown as a white surface. The wHTH domain is cyan, the linker is gray, the three-helix bundle is yellow, the WYL-domain is magenta and the WCX domain is green. Also labeled in this subunit are the secondary structural elements.

In addition to the two DriD subunits, electron density was evident for three nucleotides of ssDNA bound to each DriD protomer and the 21 bp of the target DNA site for each DriD-ssDNA-DNA complex. Each DriD subunit consists of five regions, an N-terminal wHTH DNA binding domain (residues 1–70), linker region (residues 71–78), three helix bundle domain (residues 79–133), WYL-domain (residues 134–245) and C-terminal dimer domain, which has also been termed the WYL C-terminal extension (WCX) (residues 246–327) (Figure 1C). α3, the recognition helices, from each DriD dimer dock into successive major grooves of the DNA target site, while the wing regions, β1–β2, interact with the minor groove at each half site (Figure 1A, C). An electrostatic surface representation of DriD shows that both the cognate DNA-binding surface as well as the ssDNA binding pockets are strikingly electropositive, indicating charge complementation as a contributing factor to ssDNA and target DNA binding (Figure 2A, B).

Figure 2.

DriD binds DNA asymmetrically. (A) Electrostatic surface representation underscoring that the DNA binding surface of DriD is highly electropositive. Electropositive and electronegative regions are colored blue and red, respectively. (B) Surface electrostatic potential showing that the ssDNA binding pockets of DriD are also positively charged. (C) Superimposition of the three-helix bundle-WYL-WCX unit of all six DriD crystallographically independent subunits. The domains are labeled. Notable is the finding that the DNA binding domains adopt only one of two orientations that are asymmetrically arranged relative to the WCX-WYL-three helix bundle units. (D) Shown are the overlays in (C) rotated by ∼90º. This view reveals the two distinct conformations dictated by the linker region between the wHTH domain and the three-helix bundle domain.

DALI searches revealed several WYL-domain proteins show structural similarity to DriD. The structure with the strongest homology outside our previously solved DriD(126–327) fragment, is the WYL-domain CRISPR CAS-related protein (pdb id code: 6OAW) (15), of which 175 similar Cα atoms superimpose onto DriD with a root mean squared deviation (rmsd) of 4.0 Å. The A. aurescens PafBC structure (pdb id code: 6SJ9) superimposes onto DriD, with an rmsd of 4.4 Å (for an overlay of 146 corresponding Cα atoms). The recently solved WYL-domain containing transcription repressor, BrxR (BREX regulator), also showed structural homology, whereby 167 corresponding Cα atoms of the BrxR (pdb id code: 7T8K) and DriD structures superimpose with an rmsd of 4.7 Å.

DriD binds asymmetrically to cognate DNA

Our two crystal forms provide three crystallographically independent views of the DriD-ssDNA-DNA complex. Comparison of these complexes revealed that the three-helix bundle-WYL-WCX units of these structures are essentially identical and superimpose with rmsds of 0.5–0.8 Å for 185 Cα atoms (Figure 2C, D). The two-fold related WCX domains generate an interlinked dimer and each WYL-domain is positioned perpendicular to the WCX dimer axis. Like the WYL-domain, the three-helix bundle region is symmetrically oriented around the WCX dimer axis. This domain also contributes dimer contacts. Strikingly, however, the DriD DNA-binding domains are asymmetrically positioned and oriented at an angle of ∼90º with respect to the three-helix bundle-WYL-WCX unit (Figure 2C, D). Inspection of the structures shows that the asymmetric positioning of the DNA-binding domains is caused by distinct conformations adopted by the linker region, residues 71–78, that connects the DNA-binding domain to the three-helix bundle (Figure 3A, B). Superimposition of all six crystallographically independent DriD subunits shows the linker region and the attached DNA-binding domains adopt one of two orientations (Figure 3A, B). In one conformation residues 72–75-fold into a helix (α3b) while in the other conformation these residues are disordered and the attached DNA-binding domain adopts a different position relative to the other subunit in the dimer (Figure 3A, B). Because the wHTH domains of each DriD dimer adopt one of two orientations, the three DriD-ssDNA-DNA complexes, despite being in different crystallographic environments, all adopt the same overall asymmetric structures (Supplementary Figure S2).

Figure 3.

Linker region contacts that determine wHTH domain orientation. (A) The DriD linker regions adopt two distinct conformations. The linker regions are either disordered or adopt a short helical region that makes intimate contacts with the wHTH domains. The DriD dimer is colored as in Figure 1A and the structural elements of the three-helix bundle are labeled. (B) Contacts between the three-helix bundle and linker conformations are shown with residues involved shown as sticks and labeled. The weakly dimeric wHTH is labeled underscoring the interface between the two α1 helices, which is anchored by contacts from linker helix, α3b. Colored red are residues that appear key in the wHTH dimer (Leu10) and the stabilizing interaction between the linker α3b and the wHTH dimer (Val73 and Phe74), which were mutated to test impacts on DNA binding. (C) Fluorescence polarization (FP) binding assay measuring the affinity of WT DriD (cyan triangles) for its cognate DNA and the DriD(L10E) mutant (violet circles). The experiments were performed in the presence of 10 μM ssDNA. (D) FP binding assay measuring the affinity of WT DriD (cyan triangles) for its cognate DNA and the linker mutant, DriD(V73P-F74P) (light_blue squares). The experiments were performed in the presence of 10 μM ssDNA. For Figure 3 C and D, the x and y axes are concentration of FL C. crescentus DriD in nM and normalized millipolarization units (mP) ((A – A₀)/(A_max– A₀)), respectively. A is change in mP reading, A₀ is the initial mP value before addition and A_max is the maximal mP reading upon binding saturation. Error bars represent range in values per each measurement. The error in K_d was determined as the SD between the calculated K_ds for three runs.

The role of the linker region between the wHTH and three-helix bundle in cognate DNA binding

The DriD DNA-binding domains docked on the DNA form a dimer between the two wHTH-domains that appears key in permitting proper DNA contacts. The wHTH-wHTH dimer is weak, burying only 380 Ų of protein surface area. Strikingly, however the linker region that folds into the α3b helix, contributes contacts to the wHTH interface when DriD is bound to DNA, leading to an increase in the buried surface area of 560 Ų. In this interaction, the side chain of α3b helix residue Phe74 interacts with Phe50 and Phe50′ from the C-terminus of the α3 helices (where ′ indicates other subunit in the dimer) as well as Leu10 and Leu10′ from the α1 helices (Figure 3A, B). In addition, Val73 from the α3b helix interacts with Ile66 from a DNA binding wing and Ala17 from the α1 helix and α3b helix Asp71 makes anchoring hydrogen bonds with α1 residue Arg14 (Figure 3B). The positioning of the DNA-binding domains is also facilitated by interactions between the three-helix bundle region and the DNA-binding domain. Specifically, Arg120 and Arg121 from α6 of the three-helix bundle interacts with the C-terminus of α1 and carbonyl oxygens from the loop following β2 (Figure 3A, B). The other DNA-binding domain makes a different set of interactions whereby α6 residue Arg122 interacts with carbonyl oxygens in the loop after β2. These analyses suggest that the contacts to the wHTHs from both the linker and the three-helix bundle domain are critical for the orientation and proper docking of the wHTHs on cognate DNA. The residues in the linker that mediate these contacts are well conserved among DriD homologs, bolstering this hypothesis (Supplementary Figure S3; Supplementary Figure S4).

To test the importance of wHTH dimerization and the role of the linker in DNA binding, we generated mutations in α1 of the DNA-binding domain and the α3b helix and performed binding assays. Specifically, a DriD(L10E) mutant was constructed based on the prediction that the L to E substitution would prevent dimer contacts between the α1 helices and a DriD(V73P-F74P) mutant was made to impair formation of the α3b helix and prevent its interactions with the DNA-binding domains. Fluorescence polarization (FP) DNA-binding binding assays done in the presence of excess ssDNA showed that the L10E mutation resulted in a more than 75-fold reduction in DNA binding with a K_d of 119.5 ± 23.6 nM compared to 1.6 ± 0.04 nM for the WT protein. The V73P-F74P mutation also led to a significant decrease in DNA binding, resulting in a K_d of 133.2 ± 10.4 nM (Figure 3C, D). To ensure that the linker and DriD(L10E) mutants were still folded properly we performed SEC analyses. These data showed, that indeed both proteins eluted as dimers (with calculated MWs of 69.0 and 70.3 kDa, respectively (Supplementary Figure S5), compared the theoretical value of 72.0 kDa) as would be expected as the main dimer contributions in FL DriD are from the three-helix bundle and WCX domains. Thus, these data support the structural model, which indicates that the wHTH dimer, as enforced by linker region interactions, is key for cognate DNA binding by DriD.

DriD contacts to ssDNA effector

A central question regarding WYL-motif containing regulators has been what molecules bind these domains to act as effectors? Our recent studies revealed that ssDNA not only binds DriD, but also activates its ability to interact with cognate DNA and activate transcription (17). As ssDNA is produced during DNA damage, this is consistent with it serving as a signaling molecule for DriD. We previously reported a structures of the DriD fragment, DriD(126–327) bound to ssDNA (17). However, the structures revealed two DriD-ssDNA interaction modes that, although similar, showed different base stacking interactions, with one involving base intercalation by Tyr240 and the other, by His241. These structures also showed from three to five nucleotides binding to DriD and slightly different conformations of the WYL-domain relative to the WCX domain. Hence, these data left in question the specific mechanism of ssDNA binding and the minimal ssDNA sequence needed for DriD interaction (17). Our DriD-ssDNA-DNA structures resolve these questions as all six crystallographically independent DriD subunits reveal clear and identical contacts for three nucleotides of ssDNA bound to each DriD protomer, despite the fact that a 7-mer ssDNA was added for crystallization (Figure 4A-B). This binding mode is similar to one observed in our DriD(126–327)-ssDNA complex (17) involving ssDNA stacking by Tyr240 and indicates that the other binding mode, was likely influenced by crystal packing.

Figure 4.

DriD–ssDNA contacts. (A) Structure of the FL DriD-ssDNA-DNA complex shown superimposed with an mF_o-DF_c omit map contoured at 4 σ, in which the ssDNA had not been included in the refinement. The map is shown as a magenta mesh and the ssDNA is shown as sticks. One DriD subunit is cyan and the other slate. (B) Close up of the ssDNA binding pocket with omit mF_o-DF_c map from (A) included (contoured at 4σ). Shown as sticks are the ssDNA and the DriD side chains that contact the ssDNA. The WYL motif is also highlighted with residues in the motif underlined.

In the ssDNA binding mode observed in the FL DriD-ssDNA structures, the DNA bases insert into a pocket between the DriD WYL and WCX domains. The DriD WYL motif is degenerate (containing a NYL instead of WYL). However, of the three residues of the motif, only tyrosine (Tyr192 in DriD) is within the ssDNA binding pocket. In the complex, Tyr192 hydrogen bonds with the ssDNA phosphate and sugar moieties. Multiple DriD residues proximal to the WYL motif also contribute to ssDNA binding. Tyr168, Ser172 and Arg178, from strands β3 and β4 and the connecting loop and Arg189 from the loop between β4 and β5, make phosphate contacts to the ssDNA. Arg204 and Arg207 from β6, stack with a base and make hydrogen bonds with the ssDNA phosphate backbone, respectively (Figure 4A, B). Trp206, also from β6, makes van der Waals interactions with the phosphate backbone. As noted, in all DriD subunits from the DriD-ssDNA-DNA structures, the side chain of Tyr240 intercalates between two of the bases in the bound ssDNA (Figure 4B). Arg288 from the DriD WCX-domain also participates in ssDNA binding by making a stacking interaction with a DNA base. Previous experiments showed that substitutions in residues observed to contact ssDNA in the structure, Y192A, Y240A and R207A, significantly impaired the ability of DriD to activate transcription and Y168A and R176A-R178A mutations led to complete loss of DriD transcriptional activation activity (17). Moreover, substitutions of exposed arginines in the WYL-domain of Mycobacterium smegmatis PafBC, corresponding to DriD residues Arg204 and Arg208, led to lack of complementation in M. smegmatis strains (22,24).

None of the DriD contacts to ssDNA are base specific. This indicates that DriD binds in a non-sequence specific manner to ssDNA, as would be expected if its function is to recognize any ssDNA sequence present in resected DNA generated during DNA damage. To investigate this further we performed FP experiments on 10 bp ssDNAs, including the sequence encompassed in the ssDNA site used for crystallization, AT-rich, T-rich, C-rich and GC-rich oligonucleotides (Supplementary Figure S6). These oligonucleotides bound DriD with K_ds ranging from 180 nM to 1.2 μM. Hence, in these studies while the T and A rich ssDNAs bound with somewhat reduced affinity, all the ssDNAs bound DriD. Our studies have demonstrated that ssDNA binding to DriD is critical for high affinity binding of DriD to cognate DNA. Yet ssDNA binds to a site on DriD that is 34 Å from the DNA-binding domain, suggesting that ssDNA-mediated activation of DriD DNA binding involves a long-range allosteric mechanism.

Mechanism of ssDNA mediated activation of DriD

While we lack a structure of the FL apo DriD, a structure has been obtained of the apo form of the FL WYL-activator, PafBC, from A. aurescens (24). PafBC proteins are composed of two proteins, PafB and PafC but the A. aurescens protein is a natural fusion. Interestingly, in the apo A. aurescens PafBC structure one of the wHTH domains is sequestered by binding to the region corresponding to the three-helix bundle in DriD. As a result, this wHTH would be unable to contact DNA. Possibly, ssDNA binding could lead to structural reorientation and ejection of the buried wHTH and such a mechanism may be at play for DriD. As noted, the wHTH domains of DriD interact with three-helix bundle domain in the DriD–DNA complex and hence may employ a similar mechanism. Arguing against such a mechanism, however, is that the inhibitory interaction in PafBC relies on a structure in the three-helix bundle domain that is different from that observed in DriD. In particular, while the PafC region of the PafBC fusion harbors a three-helix bundle domain similar to DriD, in the PafB protein this region is composed of a long helix. This helix mediates key contacts to the wHTH leading to inhibition. Thus, for the same mechanism to be at play for DriD, one of its three-helix bundles would need to undergo a large structural change involving the unlikely generation of one helix from its three-helix region. While a FL DriD structure would be needed to assess if such a conformation exists in one of its three-helix bundle regions in its apo form, structural modeling programs (i.e. AlphaFold2) do not predict the formation of such a structure in apo DriD. By contrast, the corresponding domain in PafBC that is involved in inhibition of the wHTH, is predicted to be somewhat disordered and form two or one helix, not three, by AlphaFold2 (35–37).

As noted, the DriD wHTH regions interact with residues in the three-helix bundle in the DNA-bound form and hence could possibly shift in the apo form to a different set of autoinhibitory interactions that would be relieved by ssDNA binding. However, another mechanism for ssDNA mediated activation of DriD is suggested by comparing our truncated apo DriD structure (17) with our ssDNA and DNA bound form. The ssDNA bound form of DriD shows that ssDNA binding orders residues in the linker between the DriD WYL and WCX domains and the three helix bundle region as well as affixing the WYL-domain in a specific orientation relative to the rest of the structure. As a result, the WYL-domain is poised to make contacts to the adjacent three-helix bundle region that positions and orders the latter domain, enabling the correct folding and orientation of the wHTH domains in way that stabilizes its DNA-bound form (Supplementary Figure S7). The end result of such a conformational domino effect would be the anchoring of the wHTHs in positions that favor their dimerization and DNA docking.

DriD–DNA contacts

The DriD-ssDNA-DNA complexes reveal specific contacts with the cognate DNA that explain its sequence-specific interaction with its target sites. In both the DriD–ssDNA–DNA crystal structures, the wHTH domains dock in an identical manner on the DNA. Hence, the interactions from the high resolution 2.74 Å structure will be described here. DriD residues from α1, the loop between β1-β2 and the recognition helix, α3, provide all the contacts to the DNA. Curves+ analyses show that the DriD-bound DNA target site displays an overall bend angle of ∼30° (38). The significant bend in the DriD bound DNA permits the extensive sets of interactions to each DNA half site. Interestingly, however, there are few contacts from DriD to the central region of the DNA, suggesting that the interactions at each end of the DNA, possibly in combination with intrinsic conformational preferences of the central DNA sequence, favors the DNA bend. While the overall twist and rise of the DriD-bound DNA are the same as B-DNA, the bases in the central region of the DNA display high propeller twist angles, which may enable plasticity of the central region. The majority of DriD DNA binding sites that have been mapped contain AT-rich regions in the center of the palindrome. As AT-rich sequences favor high propeller twist, such sequences might favor DNA distortion and DriD binding. Phosphate contacts (Figure 5A–C) at the ends of the DNA site, from residues Thr61 and Lys62 from the wing, Arg36, Arg43 and Thr38 from the recognition helix and Arg2 and Arg8 from α1, also likely favor the DriD-bound DNA conformation.

Figure 5.

DriD contacts to target DNA. (A) Schematic showing contacts from DriD subunits (slate and magenta) to the target DNA. Hydrogen bonds are indicated by arrows. The arrows are colored according to subunit for phosphate contacts and colored black for base specific hydrogen bonds. Hydrophobic contacts are shown as green dashed lines. (B) Close up of DriD-DNA contacts from one half site (contacts to each half site are essentially identical). Base contacting residues are shown as sticks and labeled. (C) Base specific contacts shown as sticks with superimposed mFo-DFc omit map in which these DNA contacting resides were not included in the refinement. The map is colored magenta and contoured at 2.5 σ. Hydrogen bonds are shown as black dashed lines and the hydrophobic contact from Arg37 to thymine methyl group is shown as a green dashed line.

The DriD binding site, 5′-ATACGTCCGTTTCTGGCGCA-3′, was first identified in the didA promoter via EMSA (17). Fluorescence polarization (FP) binding assays confirmed that DriD binds this site with high affinity and specificity (17). Subsequent searches of the C. crescentus genome revealed additional, DriD binding sites, which were confirmed via EMSA analyses, located in the promoters for bapE, recA, dnaG,gene 02375, driD and recJ (17). Comparison of these DNA elements reveal they all contain CG-rich half-site motifs, XC₁G₂X₃C₄(X)₇G₁₂X₁₃C₁₄G₁₅X, where the bases corresponding to G₂ and C₄ and G₁₂ and C₁₄ are highly conserved. The bapE site is the only exception in which C₄ in the first half site is substituted to T₄ (Supplementary Figure S8). Given that DriD has a DriD binding site in its own promoter, we searched for putative DriD motifs in the promoters of DriD homologs across α-proteobacteria. In 68 out of the 69 homologs found, a CGXC(X)₇GXCG motif was present in each DriD’s promoter, at an average of ∼17 nucleotides upstream of the start codon (Supplementary Table S4). Of note, in 41/68 motifs both half sites were perfectly palindromic matches (27/68 had 1 or 2 palindrome mismatches). In 66/68 motifs, the gap between the half-sites was notably AT-rich.

In the DriD-ssDNA-DNA structures, the conserved bases are contacted specifically by DriD residues (Figure 5A-C). In particular, DriD residue Arg37 makes bidentate hydrogen bonds via its side chain NH2 and Nϵ to the guanine N7 and O6 atoms, respectively, of G₂. The side chain of DriD residue Arg41 similarly reads both the N7 and O6 atoms of the guanine paired with C₄ via its side chain NH1 and NH2 atoms. By providing hydrogen bonds to both O6 and N7 atoms of guanine bases in the binding motif, Arg37 and Arg41 provide exquisite DNA binding specificity. Although not as conserved, the 3^rd position in most DriD DNA sites is typically a thymine (Supplementary Figure S8). This is also explained by the structure which shows that the side chain of Arg37 contacts the thymine methyl group of T₃ using its side chain Cγ moiety (Figure 5A-C). Somewhat conserved in the DriD DNA targets are positions 1 and 15, which are typically C and G. The structure shows that the C₁ N4 atom is contacted by the side chain of Glu40.

Probing the DriD-DNA structural model

The base contacts by DriD revealed in our DriD-ssDNA-DNA complex are consistent with and explain the conservation of the CG-rich bases in positions 1–4 and 12–15 in each half site of the DriD target DNA, in particular the nucleobases contacted by Arg37 and Arg41. However, to test our structural model in more detail we analyzed DriD binding by FP to DNA sites with specific mutations. These experiments were performed in the presence of excess ssDNA, which our previous studies showed was necessary for high affinity DNA binding (17). These analyses showed that DriD-ssDNA bound the fluoresceinated 21-bp driD promoter site (Materials and Methods) with a K_d of 1.60 ± 0.04 nM (17). Mutation of bases C₄/G₁₂ to T/A and G₂/C₁₄ to A/T, contacted by Arg41 and Arg37, led to significant reductions in binding; K_ds of 962.4 ± 182.1 and 511.7 ± 226.6 nM, respectively (Figure 6A). Substitutions of T₃/A₁₃ to C/G, which the Cγ of Arg41 interacts with, and C₁/G₁₅, contacted by Glu40, to T/A had moderate effects on binding, resulting in K_ds of 7.1 ± 0.9 and 6.9 ± 1.7 nM, respectively. Although the AT-rich center of the DNA site is involved in few DriD contacts, we postulated that it may play a role in indirect readout, by favoring the bent or distorted conformation of the DNA as it is also conserved among DriD DNA binding sites. To test this hypothesis, we analyzed DriD-ssDNA binding to a site with a GC-rich sequence in place of the AT-rich central region (5′-ATACGACGGCCGCCGTCGTAT-3′) to increase the rigidity of the DNA site. Indeed, nearest neighbor melting temperatures, which accounts for base stacking energies, were calculated using the Northwestern University OligoCalc model (39) described by (40) with parameters set by Sugimoto et al. (41). According to these calculations, the sequence with the GC-rich center was found to have a nearest neighbor T_m of 60°C which is 11°C higher than the AT-rich sequence. Our FP experiments showed that DriD binding to the sequence with the GC-rich center region led to a 20-fold reduction in binding, suggesting that an AT-rich linker between half sites plays a role in high affinity binding by DriD, which was further supported by the AT-rich nature of the central regions in the DriD motifs in the promoters of DriD homologs we identified across α-proteobacteria (Supplementary Table S4).

Figure 6.

Functional and biochemical tests of structure. (A) FP binding assays analyzing the effects of mutating key bases for binding to WT DriD. The mutations are indicated in the key. (B) FP binding isotherms for DriD WT and DriD mutants binding to the WT DNA site. (indicated by the key). For 6A and B, the x and y axes are concentration of FL C. crescentus DriD in nM and normalized millipolarization units (mP) ((A – A₀)/(A_max- A₀)), respectively. A is change in mP reading, A₀ is the initial mP value before addition and A_max is the maximal mP reading upon binding saturation. Error bars represent range in values per each measurement. The error in K_d was determined as the SD between the calculated K_ds for three runs. (C) β-Galactosidase assay measuring transcriptional activity of a P_didA-lacZ reporter in cells with the native copy of driD replaced by a plasmid-borne copy of either WT or mutant driD constructs (n = 3; error bars indicate SD). Exponential phase cells were assayed after 45 min of treatment with or without zeocin.

To test the role of the DriD DNA mutations in vivo, we generated a panel of equivalent mutations in the DriD DNA binding site in the didA promoter of a plasmid-borne P_didA-yfp transcriptional reporter. In these assays the C₄/G₁₂:T/A and G₂/C₁₄:A/T mutations fully disrupted the ability of the didA reporter to respond to zeocin treatment (Supplementary Figure S9), in agreement with the strong disruption of DriD’s binding in vitro (Figure 6A). The remaining mutations (T₃/A₁₃:G/C, N₇:GC-rich, and C₁/G₁₅:T/A) lowered overall expression of the reporter but still maintained the ability to be induced, also consistent with the weaker, yet still disrupted, binding affinities observed in the FP analysis.

Examination of a multiple sequence alignment of DriD homologs revealed that residues Arg37, Glu40 and Arg41, which contact bases, are highly conserved (Supplementary Figure S3). Hence, we next tested the effects of mutating these residues on DriD-ssDNA-DNA binding. Both the R37A and R41A mutations essentially abrogated high affinity DNA binding, while the E40A substitution led to a 7-fold reduction in binding (Figure 6B). However, mutation of Glu40 to lysine, resulted in a 83-fold reduction in binding. These data support that the Arg37 and Arg41 contacts are central for not only DriD binding specificity but also binding affinity, while Glu40 plays a less critical role. We also predicted that phosphate contacts may be important in DNA binding by DriD-ssDNA, in particular the contacts at the ends of the DNA by wing residues Thr61 and Lys62. The Lys62 is sometimes substituted to an arginine in DriD homologs, which could also make phosphate contacts. To test the importance of these residues in DNA binding we generated a DriD(T61A-K62A) mutant. This mutant, indeed, was significantly impaired in DNA binding, resulting in a K_d of 1.2 ± 0.2 μM (Figure 6B).

To correlate DriD DNA binding to its transcription activity, we next investigated the effects of mutating key DNA-binding residues in an in vivo transcriptional reporter system (16). Our system contains a transcriptional fusion of the didA promoter to lacZ integrated at a distal locus on the genome, with didA and driD deletions. The strain was transformed with a vector containing either no driD, a WT copy of driD driven by the driD promoter, or various point mutations of driD (driven by the driD promoter) bearing changes to key residues involved in target DNA binding. To induce DriD-mediated activation of the didA reporter, we treated each of these strains with the DNA-damaging agent zeocin known to activate DriD’s function as a transcriptional activator (16,17). As a control, plasmid-borne WT driD copies were first analyzed and revealed didA expression upon induction with zeocin, as expected. While the E40A mutant retained near WT activation, the E40K substitution led to complete inactivation of DriD activity in vivo. Substitutions in key DriD DNA binding residues, such as R37A, R41A and K62E, essentially abrogated transcriptional activation by DriD, consistent with our structures and biochemical data (Figure 6C). Indeed, the transcription activity of these mutants mirrored that observed in the driD deletion and empty vector strains.

Comparison of DriD–DNA to other WYL regulator complexes

To date, few structures have been solved of WYL-transcription regulators. Of the WYL repressors, only the Bacteriophage exclusion regulator (BrxR) and CBASS-associated protein with WYL domain (CapW) proteins have been structurally characterized (27,28) and only the Acinetobacter BrxR was solved bound to DNA (28). BrxR and CapW show similar overall structures indicating a common fold amongst WYL-repressors (28). However, comparison of the structure of DriD with that of the WYL repressors shows that they have significantly different overall folds. Indeed, while both proteins have wHTH and WYL domains, the remainder of their structures are distinct. For example, the domain connecting the wHTH and WYL regions is a three-helix bundle in DriD, but in BrxR/CapW, consists of a helix-β-hairpin motif. This domain contributes to dimerization in both DriD and BrxR/CapW, however, in the repressors it mediates the majority of dimer contacts. By contrast, in DriD, the WCX domain is the main region involved in dimerization (Supplementary Figure S10). Consistent with these marked structural differences, comparison of the DriD-DNA and BrxR-DNA complexes reveal that although both utilize N-terminal wHTH domains for DNA binding, their DNA-interaction modes are distinct; BrxR is symmetrically arranged on its DNA site while DriD binds asymmetrically (Supplementary Figure S10). And while the wHTH domains of these proteins are structurally similar they utilize completely different modes of docking on the DNA (Supplementary Figure S10). These differences likely reflect the distinct forms of allosterism in each; in DriD effector binding to the WYL-domain stimulates DNA binding while in WYL-repressors interaction of signaling molecules induces or removes it from its operator DNA. We have demonstrated that ssDNA acts as an WYL-binding effector for DriD activation (17). However, effectors that bind WYL-repressors, while also thought to be oligonucleotide in nature, are still unknown. These combined data indicate that WYL-repressors and WYL-activators represent different classes of DNA-binding proteins within the WYL-regulator family.

Aside from DriD, the only structure available for a WYL-activator is that of PafBC. Like DriD, PafBC functions in DNA damage and activates transcription of ∼150 genes during the damage response (42,43). PafBC harbors a more similar domain organization to DriD than the WYL-repressors, with an N-terminal wHTH domain connected to a helical bundle domain, followed by a WYL-domain and a WCX domain utilized in dimerization. Data also suggest that, like DriD, PafBC is activated by ssDNA binding (23). However, PafBC is composed of two different subunits, PafB and PafC and while DriD binds a palindromic DNA site with high affinity and specificity in the absence of RNA Polymerase (RNAP), PafBC interacts with a nonpalindromic sequence and is unable to bind DNA in the absence of RNAP (Supplementary Figure S11). This is explained by the cryo-EM structure of PafBC with RNAP and promoter DNA, which while not revealing well resolved density for the domains following the wHTH, showed that only the DNA-binding domain of PafB interacts with DNA; the PafC wHTH instead binds domain 4 of σ70 (23). Interestingly, however, the weak density that was observed for the C-terminal domains of PafBC appears to be asymmetrically posed relative to its wHTHs, similar to the asymmetric binding mode observed in our DriD structure. Such an interaction mode may permit DriD to contact RNAP subunits via its exposed wHTHs. However, structures will be needed of DriD bound to RNAP and promoter DNA to elucidate the mechanism by which this homodimeric WYL-regulator activates transcription.

As noted, PafBC and DriD appear to also utilize distinct modes of effector mediated conformational activation as the apo PafBC structure showed an auto-inhibitory interaction between a wHTH and the helical domain, which in PafB consists of a single helix (24). The corresponding region in DriD is instead comprised of a three-helix bundle. To ascertain if the structural distinction in this region is conserved among PafB homologs, we utilized AlphaFold2 to predict structures of several PafB homologs with a range of sequence conservation (Supplementary Figure S12) (35–37). Strikingly, these analyses showed that this region in PafB homologs is, first of all, not predicted with a high level of confidence. In addition, in all cases this region was not predicted to contain a three-helix bundle, but rather two helices and flexible loops. This was also the case for the PafB regions contained within PafBC fusion proteins (Supplementary Figure S13). By contrast, a similar survey of several WYL-activators predicted to form homodimers, revealed that all are modeled to contain a three-helix bundle in their active folded form similar to that observed in our DriD structure (Supplementary Figure S14). This was the case even for homologs with low sequence homology to DriD (Supplementary Figure S14). We suggest that DriD is representative of a large subclass of WYL-activators that may be considered distinct from the Paf proteins subclass. Hence, our findings indicate that our DriD structure can serve as a model for the large group of homologous WYL-activators.

In summary, recent work has unveiled a newly described family of transcription activators with WYL-domains. How these regulators function has been unclear due to the lack of structural information on these proteins in complex with their target DNA sites and effectors. Our structures of FL DriD bound to effector and target DNA indicates how effector binding mediates downstream target DNA interactions through base specific contacts from a unique DNA binding homodimer. Our data further support that DriD can serve as a structural model for the large group of homologous WYL-activators.

Data availability

The atomic coordinates and structure factors for the DriD-ssDNA-DNA structures have been deposited in the Protein Data Bank (https://www.rcsb.org) and are publicly available as of the date of publication. Accession codes are PDB: 8TP8 (the P2₁ crystal form of DriD-ssDNA-DNA complex), PDB: 8TPK (the P6₅22 crystal form of DriD-ssDNA-DNA complex). Any additional information required to reanalyse the data reported in the paper is available from the corresponding author upon request.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

Nanaline H Duke Endowed Chair and National Institutes of Health [R35GM130290 to M.A.S.]. Funding for open access charge: NIH MIRA [R35GM130290].

Conflicts of interest statement

None declared.