Our experiments tested the hypothesis that MutS homologs are capable of recognizing non-B form DNA structures, and that such binding is independent of the classically defined mismatch repair pathway. We show here that E. coli
MutS specifically recognizes G4 DNA with apparent affinity higher to that of G-T mismatched binding. Structural analysis and binding studies of the MutS homodimer previously demonstrated that recognition of mismatched bases is facilitated by one highly conserved Phe at position 36 which facilitates mismatch recognition by stacking with one of the mispaired bases [47
]. The utility of such a mechanism is a capacity to recognize multiple types of different base mismatches [49
]. Substitution of this residue completely blocks mismatch recognition and repair [41
]. Nevertheless, MutS F36A retains moderate affinity for G4 structuresas measured by mobility shift assay (Figure ). Further, expression of a MutS F36A transgene rescued the ability of G4-capable phage to efficiently infect MutS deficient cells (Figure ). Our findings indicate that the Phe at position 36, required for recognition of heteroduplexes, is not required for structure recognition. This strongly points toward a distinct mechanism for structure recognition by MutS. Even so, it seems most likely that the G4 DNA binding domain of MutS overlaps with that of the heteroduplex domain because G4 binding was reduced by ~20% (Figure C) for purified MutS F36A compared to wild type protein. We cannot, however, exclude the possibility that an alternate binding domain is collaterally perturbed by substitution of Phe36 in a way that influences structure binding affinity, or that the MutS F36A protein has lower activity overall.
The licensing of mismatch repair activities by MutS depends upon conformational changes and communication between two distant MutS domains, one for DNA recognition and the other for ATP binding and hydrolysis. In binding experiments nucleotide status modulates MutS complex formation with heteroduplex DNA, and the addition of ATP results in movement of MutS away from the mismatch [50
]. Consistent with that activity, we did not observe stable MutS-bound G-T mismatched oligonucleotides at ATPγS concentrations tested in mobility shift assays (Figure A). In contrast, and under identical reaction conditions, both MutS and MutS F36A remained associated with G4 DNA in the presence of ATPγS (Figure B), although the binding pattern appears different for MutS F36A compared to wild-type MutS (compare Figures B, B and B). Therefore it appears that ATP-induced MutS conformational changes that promote heteroduplex release are inadequate to dissociate MutS from G4 DNA. This is a deviation from well-defined binding properties of the MutS homologs, and supports the notion that G4 recognition is not affiliated with mismatch repair as currently defined. Indeed, ATPγS-independent binding to G4 DNA is not confined to E. coli
MutS as this binding mode is shared with the human MutS homologs. Both Holliday Junctions and G4 DNA are bound with high affinity by human MutSα in the presence of ATP [9
]. In contrast with E. coli
MutS, the human complex may have additional responses to DNA structures because synthetic four-way junctions do not appear to be specific binding substrates (not shown, and [54
]). Regardless, it seems likely that the MutS homologs have at least two binding activities; one is affiliated with heteroduplex recognition and ATP-induced mismatch repair, and the other responsive to G4 DNA and independent of methyl-directed mismatch repair.
Genomic regions rich in repetitive guanine are common in higher eukaryotes, but rarer in prokaryotes. G-rich introns located at the 5’ end of expressed loci have been correlated with transcriptional pausing, providing a gene regulation rational for why genetically unstable G-rich DNA may be retained in mammals [24
]. However, and relevant to recombination, many hot-spots for genetic rearrangement in mammals contain guanine repeats, such as the immunoglobulin switch region, telomeres, and the rDNA (recently reviewed in [55
]). This is in striking contrast to prokaryote genomes because, with the exception of the pillin genes in Neisseria
], few recombination-associated sites with strong G4-forming potential have been described. Generally, the prokaryotic loci with G4 potential are short in sequence and located within promoter regions [23
]. In other words, G4-capable sequences found within extensive non-coding repetitive elements are rare in prokaryotes, and this is consistent with the more minimalist genomes of single-celled organisms. It is possible that the genetic instability inherent to large sequences with G4 potential is not well tolerated in prokaryotes, and any gene regulatory benefits attributed to G4 structure formation are not sufficient to outweigh the negative consequences associated with the higher potential for genome instability.
Considering the paucity of extensive G4 DNA in prokaryotes, our findings raise an interesting question regarding the reason E. coli
MutS has such a robust G4 binding ability. We consider it unlikely that the high affinity G4-binding activity of E. coli
MutS (Figure ), or even that the mammalian homologs [9
], has been evolutionarily retained for gene regulation activities through binding promoter-proximal G4 structures. This notion is based exclusively on well-established roles for the complex in repair and recombination. Instead, we find it more plausible that the MutS homologs play yet to be defined roles in alternative DNA structure resolution or site-specific recombination. In humans, MutSα directly interacts with the BLM helicase [58
], and BLM has a G4 DNA unwinding activity [32
]. Further, human MutSα was shown to inhibit FANCJ unwinding of G4 DNA structures [29
] suggesting MutS homologs may play a role in pathway selection for G4 resolution. However, additional studies will be required to determine G4-specific functions in the cell. Nevertheless, the G4 binding we observe with E. coli MutS may reflect a mechanism for discouraging domestication of repetitive DNA elements.
The specific pathways remain to be identified, but it is feasible that MutS helps facilitate replication when structures are present or allows unstable DNA elements to be removed by recombination. Either way, it is clear that the function of MutS in G4 DNA metabolism is not associated with the methyl-directed mismatch repair as currently defined. Mismatch repair factors in E. coli mediated instability at non-B form structure sites within a plasmid-encoded intron from the PKD1 gene [59
]. This possibly reflects structure-dependent responses. Such activities may explain the reduced plaques upon M13-G infection of JW1. However, such functions cannot be attributed to mismatch recognition because return of MutS F36A to JW1 recovered M13-G infection success to M13mp18 levels (Figure ). MutS homologs may contribute to genomic stability at non-B form DNA by rearrangement at DNA structures or another pathway that permits replication through difficult templates. These pathways are not defined for G4 DNA, and further experimentation will be required to determine how MutS proteins participate in cellular responses to non-B form DNA.