WRN belongs to the RECQ helicase family whose members include E.coli
RECQ, Saccharomyces cerevisiae
SGS1, and human RECQL, BLM, WRN, RECQL4 and RECQL5. WRN is unique among these helicases in that it is also an exonuclease (18
). This activity, which hydrolyzes double-stranded DNA with 3′→5′ directionality, is located in the N-terminal region of the protein.
We cloned and expressed the N-terminal region of the mouse WRN protein, and showed that it is also an exonuclease with 3′→5′ polarity. The substrate specificity of mWRN was markedly similar to that of hWRN. The only difference we observed between the hWRN and mWRN exonucleases was that the mouse protein appeared to be somewhat less active than the human protein. The N-terminal region of mWRN is 30% divergent from the corresponding region of hWRN. This sequence divergence could be responsible for the reduced activity of the mWRN exonuclease. Alternatively, the mWRN exonuclease may have slightly different salt, pH or other requirements for optimal activity. Recently, exonuclease activity with the opposite polarity (5′→3′) was reported for hWRN (28
). We (20
) and others (21
) have not observed 5′→3′ activity intrinsic to WRN. In this study, the N-terminal fragments were purified to >90% homogeneity, and both the human and mouse fragments showed 3′→5′ exonuclease activity. We introduced a Glu→Ala substitution at amino acid 84 in the human fragment. This residue is one of five predicted to be critical for exonuclease activity (18
). The mutation abolished the 3′→5′ exonuclease activity of the fragment, as it did in the full-length protein (20
). An independent point mutation (D82A, Asp→Ala at amino acid 82) also inactivated the 3′→5′ exonuclease activity of full-length hWRN (20
). Thus, biochemical and genetic data indicate that WRN is a 3′→5′ exonuclease. Moreover, this activity exists in both the human and mouse proteins.
Despite slightly different levels of activity, the hWRN and mWRN exonucleases displayed strikingly similar substrate specificities. First, both enzymes preferred a 3′ recessed terminus. Little or no activity was observed on blunt-ended double-stranded DNA, or a protruding 3′ strand. Similarly, little activity was observed on single-stranded DNA. Second, both enzymes were capable of removing a terminal mismatched nucleotide as efficiently as they removed a terminal matched nucleotide. However, the efficiency of terminal nucleotide removal decreased dramatically as the number of mismatches increased. The enzyme was barely active on a DNA duplex with three terminal mismatches, and no activity was detected on a duplex with six terminal mismatches. Third, both enzymes were capable of initiating degradation from gapped or nicked double-stranded DNA. Finally, neither the mouse nor human N-terminal fragments displayed endonuclease activity on blunt-ended substrates, suggesting that they are devoid of endonuclease activity. Identical substrate specificities were observed for full-length hWRN (not shown), suggesting that the N-terminal fragments contained all the amino acids needed for exonuclease activity. The substrate preferences of the WRN exonuclease are similar to those of E.coli
exo III (29
). However, unlike exo III, WRN has no endonuclease activity. One difference between the mWRN and hWRN exonucleases is the degradation pattern of the 5′ labeled substrates, which could indicate a difference in processivity between the two enzymes.
We do not as yet know the exact function(s) of the WRN exonuclease (or helicase) in vivo
. One possibility is that WRN participates in one or more aspect of DNA replication, since abnormalities in S phase initiation or transit have been reported (6
). The efficient removal of a terminal mismatched nucleotide raises the possibility that WRN may provide 3′→5′ proofreading activity for DNA polymerases that lack such activity. WS cells are known to be hypermutable (9
), accumulating cytogenetic abnormalities consistent with deletions and rearrangements. However, it is not known whether WS cells also accumulate point mutations. WRN is homologous to FFA-1, a Xenopus laevis
protein required for replication focus formation during DNA replication (32
). Finally, the findings that WRN interacts with PCNA, and is associated with a mammalian DNA replication complex (27
), suggest that WRN may function in DNA replication.
The WRN exonuclease may also function in DNA repair, since PCNA also functions in DNA repair, and WRN can initiate DNA degradation from a nick. Mismatch repair involves an incision on the damaged DNA strand followed by removal by an exonuclease (33
). WRN can potentially fulfill the exonuclease requirement in this process. However, evidence for a deficiency in mismatch repair in WS is limited. Such a deficiency was reported for SV40-transformed WS cells (34
), but the repair deficiency could not be unambiguously attributed to a defect in WRN. However, WRN may participate in the repair of other types of DNA damage. WS cells are hypersensitive to 4NQO, which leads to elevated rates of chromosome breaks and exchanges (35
). This hypersensitivity suggests that WRN may participate in the repair of such damage. Finally, the homology to RECQ suggests that the WRN exonuclease, together with the WRN or another helicase, may act in a recombinational pathway, which can repair double strand DNA breaks.
Most helicases characterized to date form either dimers or hexamers (36
). For example, E.coli
UvrD is dimeric (37
) and BLM is hexameric (26
). Interestingly, recombinant hWRN eluted as a trimer in gel filtration assays. The oligomerization state of WRN could be important for its function. The human genome encodes more than 30 helicases, a number that may increase upon complete characterization of the genome (16
). Thus far, five are members of the RECQ family, and three of these are associated with hereditary disorders (13
). In addition to WS, defects in BLM cause Bloom’s syndrome (41
), and defects in RECQL4 cause a subset of Rothmund–Thomson syndrome (42
). All three syndromes are cancer-prone, but differ markedly in other associated pathophysiology. Differential expression may contribute to the distinct phenotypes caused by defects in these helicases (17
). Our finding that hWRN forms trimers, in contrast to the ability of BLM to form hexamers, suggests another level of distinction between these helicases. Although oligomerization of a helicase may not be essential for activity (43
), oligomeric structure may influence substrate specificity and/or ability to interact with different regulatory proteins, and thus lead to different functions. It is interesting, therefore, that WRN interacts with PCNA, which also forms trimers (44
). The trimerization of WRN and PCNA may not be a coincidence. WRN trimers may provide the necessary quaternary structure for interaction with trimeric PCNA. Such an interaction might enhance WRN exonuclease and/or helicase activities, possibilities that remain to be explored in future studies.