Helicases have been divided into several superfamilies on the basis of primary sequence 6. Superfamily 1 and 2 structures show a single DNA strand associated along a similar path (fig 1).
However, the entry path of DNA (fig 1) and location of accessory protein domains in PcrA 1 (fig 2) suggest the overall mechanism of translocation by Snf2 family proteins may not be closely related to that of the SF1 helicases.
The only SF2 structure other than SSO1653 containing nucleic acid associated with the recA-like domains is the Hepatitis C virus RNA helicase NS3 bound to a deoxyuridine octamer (fig 1). In NS3 helicase a 5 nucleotide span of the phosphate backbone associates with residues from structurally related positions contributed by the TxGx and Ia motifs of recA-like domain 1 and by motifs IV and V of recA-like domain 2.
The recA-like domain 1 of SSO1653 shows the expected pattern of interactions with the 3’-5’ DNA strand of conserved block A (ie TxGx) and motif Ia (fig 1). Contacts are also made with the second strand of the DNA duplex via residues in conserved block G which are situated in a loop immediately downstream of motif II. Binding by these residues is suggested to act as a switch directing a conformational change in motif II which activates the ATPase activity 2, which nicely explains the DNA dependence of ATPase activity for most Snf2 family enzymes studied so far.
Several basic residues on one side of protrusion 1 in the SSO1653 structure also face towards the DNA at the next minor groove (fig 3), but surprisingly they do not appear to be conserved in Snf2 family alignments.
Based on the alignment with NS3 and the known direction of NS3 movement 3, 4, 1, it can be predicted that Snf2 family enzymes would translocate with protrusion 1 ahead of recA-like domain 1 2 (ie right to left in fig 1). The overall pattern of DNA contacts is restricted to one face of the DNA duplex and occurs predominantly in the minor groove (fig 3).
Interestingly, widening of this groove in the SSO1653 structure appears to induce bend away from the enzyme 2. If this were to occur for eukaryotic Snf2 family proteins, it might be consistent with an interaction with the bent DNA on a nucleosome surface.
RecA-like domain 2 is in a flipped conformation in the SSO1653 structure, making contact with DNA via non-conserved residues from the major insertion region (fig 3). This orientation of recA-like domain 2 has never previously been observed so validation of its functional relevance would make this a major insight into the mechanics of Snf2 family.
The Rad54 structure contains recA-like domains in the conventional orientation seen in all other SF2 enzymes. If DNA is docked relative to recA-like domain 1 of Rad54 in the way observed for SSO1653, a reasonable model for how DNA might bind to recA-like domains in the conventional orientation can be generated with only a few clashes (not shown).
In such a model, motifs IV and V are very near the position predicted by NS3 and could interact with the 3’-5’ DNA strand as expected (fig 3). Conserved blocks D and L are also in relatively close proximity of DNA. Mutations of the highly conserved arginine in conserved block D interfere with DNA dependent ATPase activity 2 and are linked to genetic diseases 5. Ahead of these contacts, both protrusion 2 and the brace reach across the minor groove as far as the open major groove beyond (fig 4).