The human primosome, a complex of primase and DNA polymerase α (Polα), provides the RNA-DNA primer to initiate synthesis during DNA replication1,2. It is also involved in other DNA transactions, such as telomere maintenance3-5, DNA damage repair6, and innate immunity7-9. The biomedical importance of the primosome is evident with it being an emerging target for anticancer therapy10. The primosome holds a unique functional role in molecular biology due to its ability to de novo synthesize a short RNA primer by primase, comprised of catalytic p49 and large p58 subunits, and then switching it to a two-subunit (catalytic p180 and accessory p70) Pola complex for extension with DNA1,2,11.
Structure-function studies of the primosome have provided much understanding to how the enzyme performs its catalytic actions, but most of these findings are limited to the context of individual domains or subunits, or subcomplexes12-21. Clearly, to understand how primosome makes the RNA-DNA primer, it is important to study the enzyme as a heterotetrametric complex. But it is understandably challenging to elucidate structural details of the flexibly tethered primosome complex22,23. Indeed, till recently, only the apo state structure of the human primosome has been solved20,24. The enzyme apo-state revealed an inhibitory compact structure that possibly explains how the primase domain can act first on a template1. A recent cryogenic-electron microscopy (cryo-EM) structure of the human primosome caught in its preinitiation state with its single-stranded DNA-binding accessory protein, CST, provides critical insights into the enzyme’s RNA primer synthesis mechanism25. Despite these recent advances, the structural details of the primosome in its DNA elongation state remain unclear. In this work, we report the cryo-EM structures of the human primosome trapped in the elongation state, which reveals an architecture different from the enzyme apo20 or preinitiation25 states and provides new insights into how the enzyme rearranges to accommodate RNA-DNA primer elongation and facilitate DNA synthesis termination.
We reconstituted the elongation complex by incubating a recombinant human primosome complex with dATP and a DNA template annealed to the 12-mer RNA-DNA primer that has a dideoxycytosine at the 3'-end to prevent further primer extension (Extended Data Table 1). Single-particle cryo-EM analysis yielded two major elongation complexes; a p180core-p58C binary complex (Elongation complex I, EC-I; Fig. 1a) and an intact four-subunit complex (Elongation complex II, EC-II; Fig. 1b), both bound to the template:primer. They were solved at 3.4 and 3.6 Å global resolution for EC-I and EC-II respectively, allowing us to determine their atomic models (Fig. 1c; Extended Data Figs. 1 and 2 and Supplementary Table 2).
Both complexes bind the template:primer in similar fashion, with the p58C domain binding the primer’s 5'-triphosphate group while the p180core domain engaging the primer 3' end (Fig. 1d). Comparison with the crystal structures of either template:primer-bound p180core21 or template:primer-bound p58C20 revealed that parts of the duplex are structurally well-conserved around areas interacting with the proteins, while some deviations are observed in remote areas (Extended Data Fig. 3). Since EC-I and EC-II shares the same template:primer interactions, it is possible that EC-I is a subcomplex of EC-II that was broken off during cryo-EM sample vitrification.
Our elongation complex structures provide the first direct evidence showing that the p58C domain remains bound to the template:primer after its handover from p49 to p180core, and this engagement persists at least during four cycles of DNA polymerization. This continued primer 5'-end engagement by the p58C domain is consistent with past studies showing that this domain has high affinity to the template:primer (Kd ~ 36 nM)26, is essential for the internal primer handover27,28 and is compatible with primer handover model that was generated without accessory proteins1,20. But then, why would the p58C domain persists in holding onto the template:primer after DNA elongation has started? We think this could be a fail-safe mechanism that prevents the catastrophic loss of the template:primer in an event when p180core releases the molecule.
The EC-II complex revealed that the platform domain, p49-p58N-p180C-p70, serves as a scaffold supporting the p180core-p58C-template:primer complex (Fig. 2a-c). Comparison with the apo state structure20 revealed a large rearrangement in the overall conformation of the primosome as it progressed to an elongation state (Extended Data Fig. 4). Within the platform domain, the disc-like p180C-p70 was rotated 37° relative to its position in the apo state (Extended Data Fig. 5). The thumb subdomain of p180core is wedged between the p49 and p58N domains of primase, fixing the position of p180core relative to the platform (Fig. 2a).
Two new interaction sites, identified from the EC-II structure, seemly support this novel elongation complex conformation (Fig. 2a-c); in one interaction area, interdomain hydrogen bonds are formed with participation of p49’s G97 backbone oxygen, and the sidechains of K95 and Q100 with the side chains of S1127 and E1118 of p180core respectively. In another area, R235 of the p58N domain forms a salt bridge with E1121 of p180core; S113 of p58N interacts with D1242 and T1244 of p180core (Fig. 2b). In addition, the sidechain of L96 from the p49 subunit is inserted into the hydrophobic pocket at surface of p180core (L1087, I1119, V1128 and hydrophobic parts of D1090 and T1091) (Fig. 2c).
To explore the role of the newly-identified platform-p180core interactions in primer DNA elongation, we introduced multiple point mutations in two subunits of primase, p49K95E/L96A and p58R235E, intended to disrupt p180core docking on the platform (Fig. 2d). In our single-turnover trapping assay, the mutations resulted in longer RNA-DNA primers being made by the enzyme (Fig. 2d, lane 3 compared to lane 2 and 2e). For example, the amount of the 41-mer primer is four times higher in the case of the mutated primosome (Fig. 2e). In the absence of a DNA trap, we observed a similar increase in the level of a 41-mer product (Fig. 2d, lane 4 compared to lane 1), which is ~three-fold higher in the case of the mutant (Fig. 2e).
Our structure-guided mutation studies showed that the platform-p180core interaction of EC-II is important for preventing excessive DNA synthesis by Polα. But why would the enzyme evolved to throttle its primer elongation processivity? We believe this early termination behavior works in favor of the replication role of primosome, which is to seed multiple short primers for more processive DNA polymerases to take over. Without proof-reading capability29, extensive DNA elongation by primosome would lead to additional unnecessary mutations.