Whereas the structure of a number of viral and bacterial DNA polymerases have been determined by X-ray crystallography, the structure of yeast DNA polymerase epsilon is the first one of a more complex, multi-subunit eukaryotic DNA polymerase. Taken together, the structure and functional characterization of DNA polymerase epsilon have led us to propose a model for interaction of the polymerase with DNA that might explain its intrinsic processivity.

A Cryo-EM reconstruction of DNA polymerase epsilon was refined starting from an initial model calculated (using the random conical tilt method) from images of Pol epsilon particles preserved in stain. The structure of Pol epsilon, the first of a multi-subunit DNA polymerase, reveals that the enzyme comprises two domains, an extended "tail-like" structure, and a globular portion.

A previously-determined reconstruction of the Pol2 subunit was filtered to 4 nm resolution (a) and used as reference for alignment of Pol2-Dpb2 images. Additional density (shown in red) corresponding to the Dpb2 subunit was identified (b). The additional mass corresponds to only a fraction of the Dpb2 MW, indicating that Dpb2 is mobile and likely constitutes a flexible connection between the Pol2 subunit and the extended tail formed by the Dpb2, Dpb3, Dpb4 subunits.

Multivariate statistical analysis and classification of Pol e images preserved in stain revealed the flexibility of the structure suggested by the structure of the Pol2-Dpb2 subcomplex. The relative orientation of the globular and extended tail portions of the Pol e structure can vary signficantly, and may play a role in mediating the interaction of the enzyme with DNA.

Consideration of the Pol e structure suggested that the extended tail portion of the structure might mediate interaction with DNA. Based on the dimensions of the tail, a double-stranded primer/template region with a minimal length of ~ 40bp would be required for full engagement of the tail domain. Accordingly, elongation assays were set up using a number of specifically designed primer-templates. The results of these elongation assays reveal a strong correlation between the length of the double stranded region and Pol e processivity. Therefore, the Pol e tail appears to act as a built-in DNA clamp, which would explain the capacity of Pol e to commit to its substrate in the absence of an auxiliary clamp.

The structure of Pol epsilon suggest an explanation for the intrinsic processivity of the enzyme, which, unlike other DNA polymerases, can remain commited to its template without an accessory DNA clamp. The extended tail formed by the Dpb2/Dpb3/Dpb4 subunits could interact with a double-stranded primer/template region. This double stranded region would have to be at least 40bp long to allow full inteaction with the tail domain. In agreement with this model, eongation assays show that the processivity of Pol2 alone is independent of the length of the primer/template duplex, whereas Pol epsilon becomes fully processive only when the length of the primer/template region is at least 40 bp.