The DNA transporter ComEC has metal‐dependent nuclease activity that is important for natural transformation

Abstract In the process of natural transformation bacteria import extracellular DNA molecules for integration into their genome. One strand of the incoming DNA molecule is degraded, whereas the remaining strand is transported across the cytoplasmic membrane. The DNA transport channel is provided by the protein ComEC. Many ComEC proteins have an extracellular C‐terminal domain (CTD) with homology to the metallo‐β‐lactamase fold. Here we show that this CTD binds Mn2+ ions and exhibits Mn2+‐dependent phosphodiesterase and nuclease activities. Inactivation of the enzymatic activity of the CTD severely inhibits natural transformation in Bacillus subtilis. These data suggest that the ComEC CTD is a nuclease responsible for degrading the nontransforming DNA strand during natural transformation and that this process is important for efficient DNA import.


| INTRODUC TI ON
Natural transformation is a mechanism of horizontal gene transfer in which bacteria take up DNA from their environment and integrate it into their genome (Dubnau & Blokesch, 2019;Johnston et al., 2014).
Transformation allows bacteria to acquire adaptively useful genes, to operate a distributed gene pool, and to remove parasitic mobile genetic elements from their genomes (Carvalho et al., 2020;Croucher et al., 2016;Power et al., 2021). In pathogenic bacteria, natural transformation is important for the spread of virulence traits including capsule variation and antibiotic resistance (Chi et al., 2007;Croucher et al., 2011;Domingues et al., 2012;Griffith, 1928). Specific physiological conditions are normally required for expression of the transformation machinery, with those cells developing the ability to import DNA being said to have reached a state of competence. Only about 80 bacterial species have been experimentally shown to be transformable (Johnston et al., 2014). However, homologues of core competence genes are found in most bacterial genomes suggesting that the majority of bacteria undergo transformation in their natural habitats (Carvalho et al., 2020;Pimentel & Zhang, 2018).
DNA uptake during transformation occurs in two stages. In the first step extracellular DNA is transported across the outer membrane (where present) and cell wall. This step is normally mediated by a retractile type IV pilus (Ellison et al., 2018) with subsequent | 417 SILALE Et AL.
During transformation with double-stranded DNA only one of the strands is transported into the cytoplasm through ComEC and the other strand is degraded (Dubnau & Blokesch, 2019;Piechowska & Fox, 1971). The reason why the incoming double-stranded DNA is converted to single-stranded DNA is not known. However, singlestranded DNA has been speculated to provide a better substrate for recombination, to reduce susceptibility of the imported DNA to restriction enzymes, and to permit cytoplasmic single-stranded DNA binding proteins to assist transport through a Brownian ratchet mechanism (Dubnau & Blokesch, 2019). It is also possible that the greater flexibility or smaller cross-sectional area of single-stranded DNA is necessary to achieve transport through the ComEC channel. In the Gram-positive bacterium Streptococcus pneumoniae, the nontransforming strand is hydrolyzed at the exterior face of the cytoplasmic membrane by the endonuclease EndA (Berge et al., 2002;Mejean & Claverys, 1993;Puyet et al., 1990). Other transformable bacteria lack EndA homologues, and no alternative nucleases involved in degrading the nontransforming strand have been identified, although a nuclease that introduces double-strand breaks in the incoming DNA (NucA) is found in Bacillus subtilis (Provvedi et al., 2001). Intriguingly, in a B. subtilis comEC mutant the cellassociated DNA remains double-stranded (Provvedi et al., 2001) raising the possibility that in this organism ComEC is itself the nuclease responsible for degrading the nontransforming strand (Inamine & Dubnau, 1995). More specifically, it has been proposed that the nuclease is the MBL-like CTD of ComEC that is located on the external side of the cytoplasmic membrane (Baker et al., 2016;Dubnau & Blokesch, 2019). However, the proposed nuclease activity of the ComEC CTD has not been experimentally tested.
The catalytic activity of MBLs arises from activation of a water molecule bound by the metal ions (Palzkill, 2013;Pettinati et al., 2016).
Canonical MBLs hydrolyse the amide bond of the β-lactam ring in β-lactam class antibiotics. Other members of the MBL family include nucleases involved in DNA repair and RNA maturation (Dominski, 2007). The suggestion that the ComEC CTD has nuclease activity is, thus, consistent with the known catalytic range of the MBL family. The ComEC CTD conserves most of the metal ligands found in the closest homologue of known structure, the Zn 2+binding teichoic acid phosphorylcholine esterase of S. pneumoniae (Hermoso et al., 2005), with the exception that an aspartate replaces one of the metal-co-ordinating histidine ligands provided by motif II and an asparagine replaces the histidine of motif III (Figure 1a,b).
Here we have investigated the metal-binding properties and catalytic activity of the ComEC CTD. We find that the CTD binds Mn 2+ ions and has general phosphodiesterase and nuclease activities.
Disruption of the CTD metal binding site abolishes nuclease activity and strongly impairs the ability of ComEC to support transformation in B. subtilis. These data support the hypothesis that the ComEC CTD is involved in degrading the nontransforming strand of dsDNA and that this activity is important for successful DNA uptake.

| Metal ion binding by the ComEC C-terminal domain
The MBL-like CTD of the DNA translocator ComEC is proposed to degrade the nontransforming DNA strand during natural transformation (Baker et al., 2016;Dubnau & Blokesch, 2019). We decided to directly test this hypothesis by isolating the CTD and assessing whether it has metal-dependent nuclease activity. Note that the fulllength ComEC protein has so far proved refractory to purification (Diallo et al., 2017;Inamine & Dubnau, 1995;Yeh et al., 2003).
We screened the CTDs of ComEC from 11 different organisms for soluble heterologous expression in Escherichia coli. Following this initial screening we were able to successfully overproduce and purify the ComEC CTD of the thermophilic Gram-positive bacterium The ComEC CTD is predicted to bind metal ions. To identify the metal ions involved, we took advantage of the fact that the bound metal ions are likely to stabilize the protein fold. Thermal shift assays were used to measure the stability of the CTD to unfolding in the presence of different metal ions. Initially, the assays were carried out with protein that retained a His 6 -tag used for purification. Of and Cu 2+ on CTD stability could not be tested because these metal ions interfered with the assay.
To exclude the possibility that the observed Mn 2+ -dependent increase in MthCTD stability was mediated by metal ion binding to the His-tag present on the recombinant CTD protein, the thermal shift assays were repeated on samples from which the tag had been removed. The untagged CTD was more thermostable than the tagged protein in the absence of metal ions (T m of 63°C vs 55°C) ( Figure 2b).
Nevertheless, the addition of Mn 2+ , but not Zn 2+ , still increased the stability of the protein (T m increase of 5°C) ( Figure 2b). Again, removal of the motif II aspartate residues blocked Mn 2+ -dependent stabilization of the untagged CTD. Thus, the Mn 2+ -dependent stabilization of the CTD requires the motif II aspartate residues but not the His 6 -tag. but that this is distinct from the Mn 2+ -dependent stabilization of the wild-type protein mediated by the aspartate residues.
Taken together, the stability assays suggest that MthCTD binds only Mn 2+ out of the metal ions tested and that this binding requires the motif II aspartate residues.

| Enzymatic activities of the ComEC Cterminal domain
The structural homology of ComEC CTD to the MBL family suggests that it has hydrolytic activity, most likely acting as a DNA-specific phosphodiesterase (Baker et al., 2016). Consistent with this hypothesis, we found that the purified MthCTD exhibited phosphodiesterase activity in the presence of Mn 2+ with the synthetic phosphodiester substrate bis-(p-nitrophenyl) phosphate (bpNPP) (Figure 3a). Within the pH range 4.2-9.5, this activity was highest at pH = 9.5. Measurements at higher pH values were not possible due to precipitation of the Mn 2+ ions. MthCTD exhibited negligible phosphomonoesterase activity against the phosphate monoester p-nitrophenyl phosphate (pNPP) across the same pH range (Figure 3a).
The absence of a phosphomonoesterase activity is consistent with the proposed nuclease role of the CTD because a phosphodiesterase activity is necessary for DNA degradation. The absence of phosphomonoesterase activity also demonstrates that the observed phosphodiesterase activity of the CTD is a specific catalytic reaction rather than arising from a generalized nonspecific hydrolytic activity of the domain.
The metal ion specificity of the observed phosphodiesterase activity of the MthCTD was assessed with a range of divalent metal cations, including some that could not be used in the earlier thermal shift stability assays. Significant phosphodiesterase activity was only observed in the presence of Mn 2+ or Co 2+ (Figure 3b). This activity was dependent on the metal-binding aspartates in motif II of the CTD because the MthCTD-DADA variant was catalytically inactive under the same conditions. Since the phosphodiesterase activity of MthCTD was higher with Mn 2+ than with Co 2+ , and because the use of non-corrin Co 2+ is extremely rare in biology (Cracan & Banerjee, 2013), it is likely that Mn 2+ is the physiologically relevant catalytic cofactor of ComEC.
To directly test the hypothesis that the ComEC CTD is a nuclease, we assessed the ability of purified MthCTD and MthCTD-DADA to degrade DNA. For these experiments the CTD proteins were purified without addition of DNase to the lysis buffer to minimize contaminating nuclease activity. MthCTD was able to degrade supercoiled plasmid DNA, linear double-stranded DNA, and single-stranded DNA when assayed at a temperature (50°C) approximating the M. thermoacetica growth optimum ( Figure 3c). The CTD had this nuclease activity in the presence of Mn 2+ ions but not without added metal ions with the exception of trace activity with linear double-stranded DNA. The nuclease activity was also dependent on the metal-ion-binding aspartate residues since MthCTD-DADA had only minor nuclease activity compared with the WT variant in the presence of Mn 2+ ions (we cannot eliminate the possibility that even this residual activity is due to trace nuclease contamination from the E. coli expression host). These data show that MthCTD is a Mn 2+ -dependent DNA nuclease. Because the CTD is able to degrade circular double-stranded plasmid DNA, it must be able to function as an endonuclease.

| The nuclease activity of the ComEC CTD is important for natural transformation
The experiments above show that MthCTD has a metal-dependent nuclease activity. To examine the importance of this nuclease activity in natural transformation, we constructed a strain of B. subtilis that produces a variant ComEC (ComEC DADA) in which the metal-binding aspartate residues of the CTD motif II are replaced with alanine residues (the B. subtilis ComEC CTD retains all the predicted metal ion ligands found in the M. thermoacetica protein and the two protein domains are 31% identical in amino acid sequence). By analogy to the in vitro experiments with the equivalent MthCTD variant, these substitutions are expected to abolish the nuclease activity of the CTD. The ComEC protein was also modified by addition of an epitope tag to allow immunological detection.
Competence was induced either by nutrient limitation or by overproduction of the competence regulator ComK from a xyloseregulated promoter during growth on a rich medium. In both cases, the strain expressing the ComEC DADA variant could still be transformed with genomic DNA (Figure 4a,b). However, the transformation efficiency of the mutant was approximately 10-fold lower than that of the parental strain when competence was induced by nutrient limitation ( Figure 4a) and approximately 100-fold lower than that of the parental strain when competence was induced by ComK overproduction (Figure 4b), the difference presumably reflecting the degree to which ComEC is the limiting step in transformation under the two competence conditions. No transformation was detected for a strain in which F I G U R E 2 Metal ion binding by the CTD of MthComEC. (a) Size exclusion chromatography on a Superdex 75 10/300 GL column of His 6 -tagged wild-type MthCTD (WT) and a variant with D611A and D613A substitutions of the predicted metal-ion-binding residues (DADA). The arrows indicate elution volumes of reference proteins. The inset shows the peak fraction of each preparation analyzed by SDS-PAGE and Coomassie Blue staining. The expected molecular mass of MthCTD is 30.4 kDa. (b) Stability of the MthComEC CTD in the presence of different divalent metal cations as assessed by thermal shift assay. The CTD proteins possessed a His 6 tag except where indicated (No His-tag). The metal ions were added as 1 mM of the chloride salt with the exception of Zn 2+ where the sulfate salt was used. Error bars represent the standard deviation, *denotes a statistically significant difference (p < .05) compared with the buffer condition in a Dunnett's test, and # denotes a statistically significant difference (p < .05) between the WT and DADA variant in a Welch's t-test, n = 3. (c) Stabilization of the His 6 -tagged MthComEC CTD by 1 mM Mn 2+ ions as assessed by differential scanning calorimetry In conclusion, our data show that the nuclease activity of the ComEC CTD is important for natural transformation in B. subtilis.

| D ISCUSS I ON
The ComEC protein of the canonical competence system of B. subtilis contains an extracellular MBL-like domain (the CTD). An analogous domain is present in approximately half of the ComEC proteins encoded in bacterial genomes (Pimentel & Zhang, 2018). The function of this domain has never been investigated, but it has been suggested to be a metal-dependent nuclease involved in degrading the nontransforming DNA strand during DNA uptake across the cell envelope (Baker et al., 2016;Dubnau & Blokesch, 2019). Here we have investigated the biochemical properties of the CTD and its role in natural transformation.
The ComEC CTD has been predicted to bind two Zn 2+ ions based on the known metal content of the most closely related MBL fold proteins (Baker et al., 2016). However, our biochemical analysis of the iso-  (Christianson, 1997) and fall within the range of co-ordinating side chain atoms found in structurally characterized dinuclear Mn 2+ binding proteins (e.g., the thiosulfohydrolase SoxB with 5 × N, 3 × O (Sauve et al., 2009) and arginase with 2 × N, 6 × O (Kanyo et al., 1996)). The metal ion ligands are conserved in most other ComEC CTDs, including that found in the B. subtilis ComEC, leading us to expect that Mn 2+ is the metal ion that is normally bound to this domain.
The enzyme responsible for digesting the nontransforming DNA strand would be anticipated to be an exonuclease because it is expected to act processively on the transporting DNA molecule and also F I G U R E 3 The ComEC CTD has phosphodiesterase and nuclease activities. (a) pH dependence of the phosphomonoesterase (with pNPP as substrate) and phosphodiesterase (with bpNPP as substrate) activities of the His 6 -tagged MthComEC CTD in the presence of 1 mM Mn 2+ . The error bars show the standard deviation of three technical repeats. (b) Metal ion dependence of the phosphodiesterase activity of His 6tagged MthComEC CTD at pH 9.0 with bpNPP as substrate. The metal ions were added as 1 mM of the chloride salt. The activities of both the wild-type protein (WT) and of a variant (DADA) with D611A and D613A substitutions of the predicted metal-ion-binding residues are compared. The error bars show the standard deviation (three technical repeats). *denotes a statistically significant difference (p < .05) compared with the buffer condition in a Dunnett's test, and # denotes a statistically significant difference (p < .05) between the WT and DADA variant in a Welch's t-test, n = 3. (c) The MthComEC CTD has nuclease activity. Supercoiled double-stranded plasmid DNA, linearized plasmid DNA, or closed circle M13 phage single-stranded DNA, were incubated at 50°C for 30 min either with no protein additions (buffer), or with 10 μM of either the WT or DADA variant of the His 6 -tagged MthCTD. Where indicated the samples were supplemented with 5 mM MnCl 2 . The samples were subject to electrophoresis in an agarose gel and DNA detected with SYBR Gold stain. The experiment was repeated three times with similar results. The high-molecular-weight band in the condition where linearized plasmid DNA is incubated with WT protein in the absence of Mn 2+ likely represents protein-bound but uncleaved DNA  (Provvedi et al., 2001). Nevertheless, we find that the isolated ComEC CTD is able to digest circular double-and single-stranded DNA molecules ( Figure 3c) and so must also possess some endonuclease ability. In our in vitro experiments, the ComEC CTD was ultimately able to digest both strands of doublestranded DNA molecules (Figure 3c). Presumably, digestion of the transforming strand is prevented in the physiological context because it is transported immediately after it leaves the CTD active site.
B. subtilis takes up DNA during natural transformation at a rate of 80 bp/s (Maier et al., 2004). Assuming that the rate of degradation of the nontransforming strand is similar to the rate of DNA uptake in B. subtilis, as is the case in S. pneumoniae (Mejean & Claverys, 1993), then the rate of DNA hydrolysis by the CTD should be ~80 phosphodiester bonds per second. The rate of bpNPP hydrolysis by MthCTD in the presence of Mn 2+ was 3.4 mol/min (mol protein) −1 (Figure 3b), which corresponds to ~0.06 molecules of bpNPP molecules hydrolyzed per second per molecule of MthCTD. Thus, the rate of bpNPP hydrolysis by MthCTD is three orders of magnitude lower than that would be required for degradation of the nontransforming strand in B. subtilis. A number of factors could contribute to this discrepancy. First, the test substrate bpNPP does not closely resemble a DNA molecule. Second, the activity of the CTD may be affected by being taken out of its native context. For example, other domains of ComEC, or other components of the transformation machinery, may orient or tension the DNA molecule appropriately for hydrolysis and may be required to actively move the strand through the ComEC active site. Third, the CTD tested comes from an organism that grows optimally at 55°C, but for technical reasons, the phosphodiesterase activity of the isolated CTD was assayed at 30°C. Thus, the measured phosphodiesterase activity is likely to significantly underestimate the activity of the enzyme under physiological conditions. Our data show that the metal-dependent nuclease activity of the ComEC CTD is important for transformation in B. subtilis since the efficiency of transformation dropped between 10-and 100fold (depending on the experimental system) when the CTD could no longer bind metal ions (a D573A, D575A variant) but did not block transformation completely (Figure 4a,b). However, we note that a contrary recent study reports that substitution of one of the predicted metal-binding residues replaced here (D573) completely prevents transformation in B. subtilis (bioRxiv preprint https://doi. However, if the S. pneumoniae CTD is catalytically inactive, it may be retained to act as a guide for the transported DNA molecule. Almost half the ComEC proteins encoded in sequenced genomes lack a CTD of the type characterized here (Pimentel & Zhang, 2018).
How the nontransforming DNA strand is degraded (or perhaps displaced) in these organisms remains an open and interesting question.

| Construction of expression plasmids
All primers used in this work are listed in Table S1. All constructs were verified by sequencing.

| Biophysical methods
Reactions were started by addition of the substrate to a final concentration of 0.8 mM and monitored at 405 nm for 10 min. A linear regression fit was performed in MARS Data Analysis software (BMG Labtech) to extract the slopes from the absorbance versus time curves and the amount of p-nitrophenol released was calculated using ε 405 (p-nitrophenol) = 18,000 M −1 cm −1 .
Induction of competence through induction of P xylA comK used the protocol of (Zhang & Zhang, 2011

| Construction of unmarked B. subtilis chromosomal integrants
The pMAD suicide plasmid system (Arnaud et al., 2004)  primers AS63 and AS64. The fragment containing the mutated codons was then amplified using primers AS65 and AS66 and cloned into pMAD as above to yield plasmid pMAD-ComEC-DADA.
B. subtilis cells were grown to competence under starvation conditions as described above. 3 μg of the appropriate pMAD plasmid were added to 500 μl competent cells in glass test tubes and incubated for 1 hr at 30°C with 180 rpm shaking. Cells were plated on LB agar plates containing MLS (25 μg/ml lincomycin and 1 μg/ ml erythromycin) and incubated at 30°C for 2 days. Two colonies were picked into separate 5 ml LB-MLS cultures and incubated overnight at 37°C with 180 rpm shaking to promote integration of the temperature-sensitive plasmid. The following morning the culture was serially diluted 10 −4 to 10 −6 , plated on LB-MLS agar plates containing 100 μg/ml 5-bromo-4-chloro-3-indolylβ-D-galactopyra noside (X-gal), and then incubated overnight at 37°C. The following day, four blue colonies were picked and used to inoculate separate 5 ml LB cultures. These cells were then cultured at 30°C overnight without shaking to promote excision of the pMAD plasmid from the chromosome. The following morning, the cultures were incubated for a further 4 hr at 30°C with 180 rpm shaking. The cultures were serially diluted 10 −6 to 10 −8 and plated on LB-X-gal agar plates, which were incubated at 37°C overnight. The following day, white colonies were replica-patched on LB-X-gal and LB-MLS agar plates followed by overnight incubation at 37°C. Patched colonies that were white and MLS-sensitive were screened by PCR and sequencing to confirm the desired alterations to the chromosome. Strains were stored as glycerol stocks at −80°C.

| Isolation of membrane fractions and Western blotting
Strains were cultured and induced for competence as for the transformation assays above. Cell pellets were harvested and stored at −20°C.
Frozen pellets were resuspended in 2 ml ice-cold phosphate-buffered saline (Sigma) with 1 mM EDTA (PBSE), lysozyme and DNase I, and incubated on ice for 30 min. Cells were lysed by sonication using a 130 W Vibra-Cell sonicator at 100% amplitude for 1 min on ice.
Cellular debris and unbroken cells were removed by centrifugation at 15,000 ×g for 10 min at 4°C. The membrane fraction was pelleted by ultracentrifugation at 220,000 ×g for 1 hr, at 4°C. The membrane fraction was resuspended in 200 μl PBSE, and the total protein content was estimated using the DC Assay (Bio-Rad). 90 μg of membrane protein from each strain was subjected to SDS-PAGE and analyzed by immunoblotting with a mouse anti-Strep-tag II primary antibody (QIAGEN) (1:2,000 dilution) and a polyclonal goat anti-mouse antibody coupled to horseradish peroxidase (Sigma) (1:10,000 dilution).

ACK N OWLED G M ENTS
We thank Nicola Stanley-Wall and Richard Daniel for providing advice and reagents for the genetic manipulation of B. subtilis. We thank David Staunton for carrying out the differential scanning calorimetry experiments. We acknowledge the use of the Oxford Molecular Biophysics Suite. This work was supported by Wellcome Trust studentship 109137/Z/15/Z to A.S. and Wellcome Trust Investigator Awards 107929/Z/15/Z and 219477/Z/19/Z.

AUTH O R CO NTR I B UTI O N S
A. Silale carried out all experiments. B.C. Berks and S.M. Lea conceived the project. All authors interpreted data and wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding authors on reasonable request.