|Home | About | Journals | Submit | Contact Us | Français|
The ubiquitin binding zinc finger (UBZ) domain in the C-terminal portion of Polη has been found to interact with ubiquitin. However, the affinity between the Polη UBZ and ubiquitin was shown to be low with a previously reported Kd of 73~81 μM. This low-affinity binding between Polη UBZ and ubiquitin has been difficult to reconcile with its presumed role in translesion synthesis as suggested by genetic and cell biology studies. In this work, we constructed a minimal S. cerevisiae Polη UBZ domain and probed the Polη UBZ-ubiquitin interaction using surface plasmon resonance (SPR) technique. Our quantitative binding data between the wild-type or mutant Polη UBZ and ubiquitin revealed an interesting divergence between the Polη UBZ from S. cerevisiae and humans. Moreover, we found that the C-terminal portion of yeast Polη (amino acid 515~632) binds ubiquitin with a much higher affinity than the minimal UBZ domain. Further, distinct ubiquitin-binding kinetics were observed for the C-terminal portion of Polη and the isolated UBZ domain. This observation raised the interesting possibility that the Polη C-terminal portion binds ubiquitin in a novel mode that affords higher affinity. Our findings have broader implication in understanding the generally weak interaction between the known ubiquitin-binding domains and ubiquitin.
Ubiquitin, an 8.6 kDa protein, covalently modifies a target protein via an isopeptide bond between the C-terminal carboxylate of ubiquitin and the lysine ε-amino group in the target protein. Ubiquitination was initially discovered as a signaling mechanism for proteasome-mediated protein degradation1. More recently, it was realized that ubiquitin is a highly versatile modifier that participates in a diverse array of cellular processes, including DNA damage response, transcription, nuclear transport, autophagy and immune response2. Adding to the versatility of ubiquitin, a group of ubiquitin-like proteins (Ubls) also exists in eukaryotic cells and play essential roles in processes not related to protein degradation3, 4.
Ubiquitination, like phosphorylation, is a reversible process, catalyzed by ubiquitin ligase and deubiquitinating enzyme. However, ubiquitination possesses a higher degree of complexity intrinsic to ubiquitin as a modifier. It has been shown that ubiquitin can be conjugated to proteins in different forms, either as a monoubiquitin moiety or as polyubiquitin chains. Remarkably, polyubiquitin chains linked through one of the seven lysine residues on ubiquitin have been identified in cells5. The K48-linked polyubiquitin chains are known to target protein for proteasomal degradation. Monoubiquitination and polyubiquitination linked through several other ubiquitin lysine residues play nonproteolytic functions in endocytosis, vesicular trafficking and DNA damage response.
Central to the ubiquitin’s diverse function is its ability to interact with proteins that contain ubiquitin-binding domain (UBD). To date at least sixteen ubiquitin-binding domains (or motifs) have been characterized. UBDs are usually small in size (20~150 a. a.) and many adopt an α-helical structure. Earlier studies found that many characterized UBDs interact with ubiquitin on a hydrophobic patch surrounding Ile44 with a low to moderate affinity (10–500 μM)6. To rationalize the low-affinity interaction between most UBDs and ubiquitin, it has been proposed that multivalent interaction between UBD and multi-monoubiquitinated or polyubiquitinated proteins likely overcome the low affinity between UBD and a single ubiquitin moiety7. While it has been demonstrated that some UBDs bind polyubiquitin chain with higher affinity compared to monoubiquitin8, direct evidence for high-affinity multivalent interaction between UBD and monoubiquitinated protein has not been reported. It is therefore still not clear how UBDs achieve high-affinity interaction with monoubiquitinated proteins.
In eukaryotes, ubiquitination of proliferating cell nuclear antigen (PCNA) is essential for the normal cellular response to genotoxic stress. PCNA undergoes both monoubiquitination and polyubiquitination following UV-induced DNA damage9, 10. Strikingly, K164 of PCNA is modified by either a single ubiquitin or by a K63-linked polyubiquitin chain. Available evidence suggests that monoubiquitination and polyubiquitination of PCNA likely channels the DNA damage response pathway to different branches by recruiting specific protein factors11, 12. It is believed that the stalling of a replication fork by a DNA lesion leads to monoubiquitination of PCNA at Lys164, which in turn contributes to the recruitment of a specialized DNA polymerase to effect translesion synthesis (TLS)13–17. Remarkably, many specialized polymerases contain either ubiquitin-binding zinc finger (UBZ) domain or ubiquitin-binding motif (UBM) that helps to recruit the specialized polymerase to the DNA damage site 18–20. Polyubiquitination of PCNA, on the other hand, is thought to trigger the error-free lesion bypass process and acts to restart the stalled replication fork 9, 21–25. The Rad6/Rad18 complex is responsible for monoubiquitination of PCNA on Lys164 9, 10. Polyubiquitination of PCNA requires Mms2-Ubc13 and the ubiquitin ligase Rad59. More recently, monoubiquitination of PCNA at Lys107 was reported in the DNA ligase I deficient yeast cells 26. Instead of activating translesion synthesis, Lys107 ubiquitination might provide a signal for DNA damage on newly synthesized DNA strands. Besides ubiquitination, PCNA also undergoes SUMOylation at Lys164, and to a less extent at Lys1279. SUMOylation of PCNA recruits Srs2, a DNA helicase, and exerts an anti-recombinogenic effect 27–29.
Compared to many other ubiquitin binding domains, the function of UBZ domain is less well understood6, 7. The UBZ domain was identified in TLS polymerase η from a number of eukaryotic organisms, ranging from humans to budding yeast S. cerevisiae. UBZ is on average 35 amino acids in size. A multiple sequence alignment of the Polη UBZ domains from different eukaryotic organisms revealed that the sequences are highly diverged in the N-terminal 20 amino acids in UBZ (Fig. 1). In contrast, the C-terminal 15 residues are more conserved. Most Polη UBZs contain a canonical C2H2 zinc-finger motif. A notable exception is the S. cerevisiae Polη (scPolη) UBZ. Instead of existing in a CXXC motif, the two cysteines in scPolη UBZ are juxtaposed next to each other. The second cysteine in the CXXC motif is replaced by Y555 in scPolη UBZ. Several residues, including D570 and A574 in scPolη, are conserved among the Polη UBZ sequences analyzed.
The structure of human Polη (hPolη) UBZ was solved by solution NMR technique30. It contains two short antiparallel β-strands and a 12-amino acid α helix. Like the classical C2H2 zinc finger, a zinc ion was found sandwiched between the major α helix and fingertip formed by the antiparallel β-strands through coordination with the conserved residues, C635, C638, H650 and H654. One zinc ion per hPolη UBZ was determined by electrospray ionization (ESI) mass spectrometry30. NMR titration experiment revealed that hPolη UBZ binds ubiquitin in the exposed surface of the α helix. A binding affinity of 73~81 μM for ubiquitin was determined for hPolη UBZ by NMR titration and isothermal titration calorimetry (ITC)30. The structure of scPolη UBZ has not been reported. However, a recent study revealed surprising divergence between the budding yeast and the human Polη UBZ in the requirement of zinc ion for UBZ function31. It was found that the scPolη UBZ binds ubiquitin in a zinc-independent manner, which is in sharp contrast to the absolute requirement of zinc binding for ubiquitin interaction in its human counterpart. This observation raised the intriguing possibility that the Polη UBZ from S. cerevisiae and humans are evolutionarily diverged and may possess different modes of ubiquitin binding.
In this study we carried out detailed interaction studies between the scPolη UBZ and ubiquitin. Further, we investigated the interaction of the C-terminal portion of scPolη with ubiquitin. Our quantitative approach revealed interesting difference in the ubiquitin binding mode between the isolated UBZ domain and the C-terminal portion of scPolη. Our results suggest that other element(s) in the scPolη C-terminal region may contribute to the ubiquitin binding in conjunction with the UBZ domain.
The S. cerevisiae Polη (632 a.a.) is constituted of a catalytic core domain (a.a. 1 to 514) and a C-terminal extension (a.a. 515 to 632), which is dispensable for Polη’s polymerase activity. Compared to the well-characterized catalytic core domain of Polη, detailed structural and functional information of the C-terminal portion of Polη is still lacking. So far two motifs have been identified in the Polη’s C-terminal sequence, i.e. the PCNA-interacting protein (PIP) motif and the ubiquitin-binding zinc finger (UBZ) domain (see Fig. 1). PIP resides at the very C-terminus of Polη and is responsible for a high affinity interaction with PCNA32. The UBZ domain is located between the PAD (polymerase associated domain) in the catalytic core and the PIP motif. Although the high-resolution structure of the scPolη UBZ is not available, it was predicted that the scPolη UBZ likely adopts a zinc-finger motif similar to its human counterpart33.
To better define the interaction between scPolη UBZ and ubiquitin, we introduced a series of point mutations into the potential ubiquitin-binding α helix of scPolη UBZ and determined UBZ’s binding affinity to ubiquitin. We generated a minimal scPolη UBZ domain (a.a. 545–580) with a N-terminal GST tag. The scPolη UBZ was purified by sequential chromatography on glutathione-Sepharose and HiTrap Q columns. To ensure that the minimal scPolη UBZ domain is properly folded, we removed the GST tag by thrombin cleavage. Circular dichroism (CD) spectrum of the wild-type scPolη UBZ domain is indicative of a properly-folded protein domain (Fig. 2). The spectrum is consistent with that reported for a larger yeast Polη UBZ construct (538–609)31. By NMR perturbation experiments, we demonstrated that the scPolη UBZ (a.a. 545–580) is functional in binding ubiquitin. We determined the 2-D TROSY spectrum of N15-labeled ubiquitin alone or in complex with the GST-tagged scPolη UBZ (a.a. 545–580) (Supplementary Fig. 1). The chemical shift of the ubiquitin spectra was perturbed or attenuated at a number of residues, similar to that reported for the human Polη UBZ and the larger scPolη UBZ construct 30, 31. Therefore, we conclude that the minimal scPolη UBZ domain retains normal ability in binding ubiquitin.
To probe the scPolη UBZ-ubiquitin binding, we selected several strictly conserved residues in the UBZ α helix for site-directed mutagenesis (see Fig. 1). We replaced the conserved A574 with a bulky residue phenylalanine. We also mutated the highly conserved D570 to alanine. Further, we introduced a proline residue at three different positions in the UBZ α helix (Q566P, A569P and A574P). We expect that the proline residue will disrupt the UBZ-Ub interaction by perturbing the structure of the UBZ α helix. All scPolη UBZ mutants were expressed as a GST fusion and purified to homogeneity as described in Experimental.
A recent report suggested that the scPolη UBZ interacts with ubiquitin independent of zinc binding31. We generated the scPolη UBZ mutants, HH568,572AA and CC552,553AA, with the potential zinc coordinating residues replaced with alanine. To confirm that both mutants no longer bind zinc, we used a colorimetric method to determine the content of zinc bound to the wild-type and mutant UBZ as described in Experimental. We observed approximately one zinc ion bound to the wild-type UBZ, in accord with the recent report for a larger scPolη UBZ construct31. In contrast, the CC552,553AA and HH568,572AA scPolη UBZ completely lost the zinc-binding ability (Fig. 3). Interestingly, the A574P scPolη UBZ showed substoichiometric zinc binding (0.4) (Fig. 3). This result indicates that disruption of the UBZ α-helix has an indirect effect on scPolη UBZ’s zinc binding ability.
We used surface plasmon resonance (SPR) to obtain quantitative binding information on wild-type and mutant scPolη UBZ interacting with ubiquitin by following a published protocol34. scPolη UBZ with a N-terminal GST tag was flowed across the SPR surface immobilized with ubiquitin and the sensorgram was recorded. A fast binding and dissociation of GST-UBZ to immobilized ubiquitin was observed (Fig. 4a). Plotting the steady-state SPR response units against the different concentrations of scPolη UBZ revealed a hyperbolic binding isotherm. The Kd for scPolη UBZ binding to ubiquitin was obtained by fitting the binding isotherm to a one-site binding equation34, 35. An equilibrium dissociation constant (Kd) of 14.6 ± 2.5 μM was obtained for the wild-type scPolη UBZ (Fig. 4b).
Binding affinity of the scPolη UBZ mutants to ubiquitin was also determined by SPR following the same protocol. The Kd for UBZ mutants, CC552,553AA (22.3 ± 4.1 μM) and HH568,572AA (22.2 ± 4.8 μM), are only slightly larger than that of the wild-type scPolη UBZ. This observation suggests that zinc binding in scPolη UBZ is not required for ubiquitin interaction.
Next we determined the ubiquitin-binding affinity of the D570A and A574F scPolη UBZ mutants. The Kd of D570A UBZ (23.6 ± 2.4 μM) is only slightly higher than that of the wild-type UBZ. More significant effect was observed for the A574F UBZ mutant with a Kd of 38.6 ± 8.6 μM, which is 2.6-fold higher than that of the wild-type UBZ.
The most profound effect on ubiquitin binding was observed for the UBZ mutants with a proline introduced in the α helix. As summarized in Table 1, a large increase in Kd was observed for A569P (43.1 ± 10.2 μM), Q566P (48.4 ± 7.2 μM), and A574P (55.3 ± 5.5 μM), corresponding to a 2.9- to 3.8-fold decrease in binding affinity. Notably A574P mutation had the largest impact on the ubiquitin-binding affinity among all mutants analyzed.
Given the low-affinity interaction between the scPolη UBZ and ubiquitin (Kd = 14.6 μM), we are interested in determining whether other sequence motif(s) in the Polη C-terminal portion contributes to the binding of ubiquitin. To address this question, we generated a truncated scPolη containing the C-terminal portion (a.a. 515 to 632) of Polη with a N-terminal GST tag. Remarkably, we obtained a Kd of 0.44 ± 0.05 μM for ubiquitin binding by the C-term scPolη following a similar SPR binding protocol (Fig. 4c & d). This represents a 33-fold increase in binding affinity for ubiquitin compared to the isolated scPolη UBZ domain.
Moreover, an inspection of the SPR sensorgram revealed marked differences in the binding and dissociation kinetics compared to the minimal scPolη UBZ domain (see Fig. 4c). While the scPolη UBZ domain binds to ubiquitin instantaneously, the C-term Polη binds to ubiquitin with a much slower rate. Moreover, the dissociation of C-term Polη from ubiquitin is extremely slow, which is in sharp contrast to the UBZ domain’s fast dissociation from ubiquitin. A global fitting of the sensorgram to a simple Langmuir binding model afforded a satisfactory fit of both association and dissociation phases (Supplementary Fig. 2). The fitted kinetic constants of C-term Polη-ubiquitin association and dissociation, kon=9.9 × 103 M−1s−1 and koff =1.7 × 10−3 s−1, allowed us to calculate the equilibrium dissociation constant of 0.17 μM, which is close to the Kd of 0.44 μM obtained from fitting the steady-state SPR binding data.
Next, we introduced the same set of UBZ mutations into the C-term scPolη and determined their affinity for ubiquitin by SPR. As summarized in Table 2, the HH568,572AA and CC552,553AA C-term scPolη mutants demonstrated a Kd of 0.59 ± 0.14 μM and 0.30 ± 0.10 μM, respectively, which are close to that of the wild-type C-term scPolη. No significant change in Kd was observed for D570A C-term scPolη (Kd = 0.30 ± 0.08 μM). In contrast, a significant (2.2-fold) increase in Kd was observed for the A574F C-term scPolη mutant (Kd = 0.97 ± 0.28 μM), similar to the change observed for the A574F UBZ domain. We also observed a 2- to 4- fold increase in Kd for A574P, A569P and Q566P C-term scPolη mutants. The measured equilibrium dissociation constants for the three mutants are 1.65 ± 0.88 μM, 0.92 ± 0.26 μM and 0.77 ± 0.18 μM, respectively.
To probe the effect of the various UBZ mutations on the function of Polη, we utilized the polymerase exchange assay to assess the ability of Polη to switch with Polη under stalled condition15, 36. We introduced the mutations, CC552, 553AA, HH568, 572AA, D570A, A574F, Q566P, A569P, and A574P into the full-length scPolη. The mutations in UBZ did not affect scPolη’s DNA synthesis activity when assayed on a singly primed ssM13 DNA template in the presence of PCNA and RFC (Supplementary Fig. 3).
In the polymerase exchange assay, we observed a 34 ± 5% decrease in DNA synthesis in the presence of 160 nM wild-type Polη (Fig. 5). In accord with the interaction studies, mutations of CC552, 553AA, and HH568, 572AA in the zinc-binding residues did not adversely affect the function of Polη in polymerase exchange. A 41 ± 3% and 39 ± 2% decrease in DNA synthesis by Polδ was observed when the CC552, 553AA and HH568, 572AA scPolη were assayed, respectively (Fig. 5). In contrast, all three UBZ proline mutants, Q566P, A569P, A574P, differed significantly from the wild-type Polη in effecting polymerase exchange. The extent of polymerase exchange observed for Q566P, A569P, and A574P Polη (20 ± 3%, 18 ± 4%, and 19 ± 6% respectively) was approximately 1.5- to 2-fold lower compared to the wild-type Polη, which agrees well with the fold of difference observed for the wild-type and mutant C-term Polη binding to ubiquitin. A574F mutation in Polη also resulted in close to two-fold decrease in polymerase exchange efficiency (18 ± 2% versus 34 ± 5% as observed for the wild-type Polη). Interestingly, D570A Polη showed no significant decrease in polymerase exchange assay (30 ± 6%), agreeing with the ubiquitin binding data.
Prompted by the observation of a large difference in ubiquitin binding affinity between the UBZ domain and the entire C-terminal portion of scPolη, we designed a series of scPolη mutants with different truncation in the C-terminal region (Fig. 6). We tested the sequences that flank the minimal scPolη UBZ domain in ubiquitin binding. The equilibrium dissociation constants (Kd) of the various constructs are summarized in Table 3. UBZ-C (545–608), which contains the 28 amino acid sequence C-terminal to the UBZ domain, binds ubiquitin with a Kd of 15.8 ± 3.7 μM, similar to the minimal UBZ domain (545–580). UBZ-N (515–580), which contains the 30 amino acids N-terminal to the UBZ binds ubiquitin with a Kd of 18.2 ± 2.2 μM, again close to that of the minimal UBZ domain. We then generated a construct, UBZ-CN (515–615), which contains sequences flanking both the N- and C-terminus of UBZ, but lacks the PIP motif. Surprisingly, UBZ-CN binds ubiquitin with much reduced affinity (Kd = 11.1 ± 1.8 μM), 25-fold lower compared to the entire Polη C-terminal portion (515–632). The large impact of removing the C-terminal 17 amino acids on the ubiquitin-binding affinity was unexpected. This observation suggests that the C-terminal 17 amino acid sequence of yeast Polη may directly interact with ubiquitin, thus contribute to the higher affinity of Ub binding observed in the complete Polη C-terminal region. Alternatively, the deletion of the very C-terminal 17 amino acid sequence in Polη may have an indirect impact on ubiquitin binding.
To date a number of ubiquitin-binding domains or motifs have been described. The large number of ubiquitin-binding domains and motifs existing in eukaryotic cells underlies the versatility of ubiquitin as a signaling molecule in many cellular processes. Despite extensive structural and functional characterization of UBD, it is not yet clear how the low-affinity interaction between most UBDs and ubiquitin can mediate the formation of functional protein complexes. One attractive proposal to account for this conundrum is that the known low-affinity UBDs may cooperate with other domain/motif to achieve higher affinity interaction with ubiquitin. Direct evidence for such a notion has been scarce, partly because many reported binding studies for UBDs utilized isolated ubiquitin-binding domain. Notably, a previous study showed that the CUE domain in Vps9 can form a domain-swap dimer that binds ubiquitin at two different sites, the canonical Ile-44 patch and the region defined by Ile-36/Leu-73 37. As a result, a higher affinity of Ub binding (Kd = 1.2 μM) was observed for Vps9 CUE than a monomeric CUE (Kd = 155 μM)38. In the case of Vps9 CUE, the two UBDs are contributed by separate polypeptides. It is possible that multiple ubiquitin-binding domains or motifs in a single ubiquitin receptor protein contribute to Ub binding by interacting with different surface patches on ubiquitin. This binding mode results in larger buried surface between ubiquitin and the ubiquitin receptor protein, thus leading to higher binding affinity. We addressed this possibility by investigating the S. cerevisiae Polη-ubiquitin interaction.
We first carried out quantitative ubiquitin-binding studies of a minimal scPolη UBZ. Our data showed that the minimal scPolη UBZ retains the ability to bind zinc ion despite that it contains a uncanonical zinc finger, agreeing with a recent report on a larger Polη UBZ construct 31. Our quantitative binding study of CC552, 553AA and HH568, 572AA scPolη UBZ provided further support that zinc binding by UBZ is dispensable for its interaction with ubiquitin. This observation underlines the difference between the hPolη and scPolη UBZ in the mode of zinc binding.
Given the marked divergence between the scPolη and hPolη UBZ in their sequences and the zinc-binding mode, the question remains whether scPolη UBZ binds ubiquitin similarly to its human counterpart. We found that the affinity of isolated scPolη UBZ for ubiquitin is comparable to that reported for hPolη30. Our mutational analyses of the scPolη UBZ in binding ubiquitin support a model proposed for hPolη UBZ with the α helix in UBZ being the major ubiquitin-binding site30.
Remarkably, when we determined the binding affinity of the C-terminal portion of scPolη (a.a. 515–632) to ubiquitina submicromolar Kd of 0.44 μM was obtained, representing a 33-fold increase in ubiquitin-binding affinity compared to the UBZ domain alone. Abolishment in zinc binding had little effect on the C-term Polη’s affinity for ubiquitin, similar to the minimal scPolη UBZ domain. Mutation in the UBZ α helix either by introducing proline at position A574, A569 and Q566, or by introducing a bulky group at A574 in the C-term Polη had significant effect on ubiquitin binding. Our result agrees with the study on human Polη UBZ suggesting that A574 and its equivalent residue is located at the interface between UBZ and ubiquitin. An unexpected observation was made for D570 in scPolη UBZ. Although the equivalent residue, D652, in human Polη UBZ was shown to be important for ubiquitin binding18, mutation of D570 to alanine in the scPolη UBZ had little effect on ubiquitin binding. The normal function of D570A scPolη in polymerase exchange assay provided further support that the ubiquitin binding ability of D570A scPolη is not adversely affected.
We further investigated the possibility that the sequence flanking the UBZ domain may contribute to ubiquitin binding. After testing several truncations of the C-terminal portion of scPolη, we found that adding flanking sequences to either the N- or C-terminus of the minimal UBZ domain did not restore the ubiquitin binding affinity observed for the complete C-term Polη. Strikingly, even a truncation of the C-terminal seventeen residues (a.a. 616–632) of scPolη resulted in a large decrease (25 folds) in its affinity for ubiquitin when compared to the complete C-terminal portion of Polη. It is not clear whether the scPolη C-terminal seventeen amino acids directly binds ubiquitin, or it contributes to ubiquitin binding indirectly. Combined with the observation that the C-term Polη binds to ubiquitin with a slower ubiquitin-binding kinetics than the minimal UBZ domain, it is possible that a conformational/structural rearrangement in the C-term Polη precedes ubiquitin binding. This rearrangement may expose the UBZ α-helix for interaction with ubiquitin and lead to increase in the buried surface area between C-term Polη and ubiquitin. High-resolution structural information on the C-terminal portion of scPolη will be required to address this possibility.
The Polη UBZ domain has been implicated in polymerase switch between the high and low fidelity polymerases, such as Polδ and Polη39. However, biochemical studies addressing the functional and structural details of UBZ’s role in polymerase switch have been lacking. We investigated the effect of mutation in the scPolη UBZ on polymerase switch. We found that mutation of the zinc-binding residues (CC552,553AA and HH568,572AA) had little effect on Polη’s ability to exchange with Polδ. This observation is in accord with the normal ability of both mutants in binding ubiquitin. More pronounced effect was observed for several mutants with mutation introduced in the UBZ α helix. The approximately two-fold decrease in polymerase exchange efficiency is in accord with the decrease in the ubiquitin-binding affinity observed for the C-term Polη mutants. Notably, the impact of UBZ mutations on Polη’s function is less significant compared to the Polη PIP mutation. It has been shown that mutation of the conserved PIP Phe residues to Ala or truncation of the PIP motif largely abolished yeast Polη’s ability to exchange with Polδ15. Therefore, the PIP motif provides the primary interaction between Polη and PCNA in polymerase switch, consistent with the recent genetic studies on the human Polη40.
Another interesting observation centers on Asp570 in scPolη. We found that D570A scPolη showed no significant decrease in the polymerase exchange efficiency, consistent with a minimal decrease in its affinity for ubiquitin. However, the D570A mutation resulted in elevated level of UV sensitivity and UV mutagenesis in yeast cells33. This points to the possibility that D570 in scPolη UBZ may be required for interaction with proteins involved in DNA damage response other than ubiquitin. Indeed, a recent study found that PDIP38 interacts with Polη through the UBZ domain and contributes to the localization of Polη to the repair foci41. Our results support the notion that scPolη may interact with more than one proteins through its UBZ domain. Different sets of residues on UBZ may be responsible for the interaction with specific partners.
A high-resolution structure of the scPolη C-terminal portion is yet to be determined. Our attempt to identify the fold of the scPolη C-terminal region failed to retrieve any reliable structural homolog. Our limited proteolysis study revealed that multiple sites in the scPolη C-terminal sequence are vulnerable to proteolytic cleavage (J. Wang and Z. Zhuang, unpublished work). This result suggests that the C-terminal region of the scPolη likely adopts a flexible conformation. The flexibility of scPolη helps to explain the previous observation that monoubiquitination at several different positions on PCNA can lead to functional Polη-(Ub)PCNA complex36.
Our secondary structure analysis revealed that the region between UBZ and PIP motif is largely disordered, and likely acts as a hinge that allows movement of the UBZ relative to PIP motif once Polη binds to Ub-PCNA. The available structural information of archael Dpo4 bound to PCNA1-PCNA2 dimer suggests a potentially flexible conformation because of the modular structure of Dpo4 connected by two peptide hinges42. It should be pointed out that in Dpo4, the little finger domain (or PAD) is juxtaposed to the PIP motif. In comparison, there are more than 100 amino acids embedded between the scPolη PAD and the PIP motif. The extra sequence likely contributes to a higher degree of flexibility in its conformation. Among the available eukaryotic Polη sequences, the C-terminal region of Polη is highly diverged compared to the more conserved catalytic core domain. Our observation that removing the C-terminal seventeen residues led to drastic decrease in scPolη’s ubiquitin binding affinity warrants further structural and functional elucidation of the C-terminal region of Polη.
The genes encoding the S. cerevisiae Polη UBZ domain and the truncated Polη C-terminal region were cloned into the pGEX-4T3 vector (Amersham Biosciences) with a N-terminal GST-tag. The Polη UBZ (a.a. 545–580) was cloned using primer UBZ+ (5′-CTACCAGCTCGAAAGGATCCGAGAAAACTCCGAAGTTGG-3′) and UBZ− (5′-GGATGATTCTTCAGCCTCGAGTCAGCCTTCCGACAGTTTC -3′). The entire C-terminal portion of Polη (C-term Polη, a.a. 515~632) was cloned using primers, C-term Polη + (5′-GTGCCGCGCGGCAGCGGATCCGTTGTAGATATGTTTGGC- 3′) and C-term Polη − (5′-TAATTGTCTATTTGGCTCGAGTCATTTTTTTCTTGTAAAAAATG- 3′). UBZ-C (a.a. 545~608) was cloned using primers UBZ-C+ (5′-ACCAGCTCGAAAGGATCCGAGAAAACTCCGAAGTTGG-3′) and UBZ-C− (5′-CTTCTGAGGTGTCTCGAGTCATTGTGAGTTTGGTCTTTTCCG-3′). UBZ-N (a.a. 515~580) was cloned using primers UBZ-N+ (5′-GTGCCGCGCGGCAGCGGATCCGTTGTAGATATGTTTGGC- 3′) and UBZ-N− (5′-GGATGATTCTTCAGCCTCGAGTCAGCCTTCCGACAGTTTC -3′). UBZ-CN (a.a. 515~615) was constructed by introducing a stop codon at amino acid position 616 of the C-term Polη-pGEX4T3 construct using QuikChange PCR primers UBZ-CN+ (5′-CTGCCACACCTCAGAAGTGACAAGTTACATCTTCC -3′) and UBZ-CN− (5′-GGAAGATGTAACTTGTCACTTCTGAGGTGTGGCAG -3′). QuikChange PCR was used to introduce mutation (CC552,553AA, HH568,572AA, D570A, A574F, Q566P, A569P and A574P) into the UBZ domain and the C-term Polη, respectively. The GST fusion protein was overexpressed in Rosetta(DE3) cells (Novagen) upon induction with 0.2 mM IPTG. Cell pellet was resuspended in lysis buffer (25 mM Tris•HCl, pH 6.9, 10% glycerol, 1 mM DTT, 1 mM EDTA) supplemented with protease inhibitor cocktail (Roche). Cells were lysed by sonication and cleared by centrifugation. Following incubation with glutathione resin (GE Biosciences), the resin was washed extensively with the lysis buffer. The target protein was eluted in elution buffer (25 mM Tris•HCl, pH 6.9, 10% glycerol, 33 mM Glutathione). The pooled protein fraction was further purified by 5 ml HiTrap SP-FF (GE Biosciences) for C-term Polη (WT and mutants), UBZ-C, and UBZ-CN or by 5 ml HiTrap Q-FF (GE Biosciences) for UBZ (WT and mutants) and UBZ-N. The target protein was eluted with a salt gradient in 25 mM Tris•HCl, 10% glycerol, pH 6.9.
The 15N-labeled S. cerevisiae ubiquitin was prepared using a bacterial expression system as previously described36 except in M9 minimal media containing 15NH4Cl as the sole nitrogen source (Cambridge Isotope Laboratories, Andover, MA, USA). All samples for NMR experiments were prepared in a buffer containing 50 mM Tris·HCl, pH 7.0, 100 mM NaCl, 2 mM DTT and 10% D2O. The two-dimensional 1H-15N TROSY NMR spectra were acquired with a Bruker Avance AV600 spectrometer equipped with a 5 mm inverse CryoProbe at 25 °C. NMR data were processed by NMRpipe43 and analyzed with Sparky 3. The residues in ubiquitin were assigned based on the previously published HSQC spectra30, 44. Chemical shift perturbation experiments were performed by mixing a 2-fold molar excess of unlabeled GST-UBZ with 15N-labeled ubiquitin. The changes in chemical shift (ΔδH and ΔδN) were calculated for each identified residue, except for those residues severely attenuated in the scPolη UBZ-ubiquitin mixture.
For CD experiment, the UBZ domain was obtained by removing the GST tag from GST-UBZ with thrombin (GE Healthcare). The purified UBZ domain was dialyzed in KH2PO4 buffer (pH 7.5). Far-UV CD spectra (190~250 nm) were collected in a Jasco J-810 spectropolarimeter (Jasco Inc.) using in a rectangular quartz cuvette of 1.0 mm path length at 20 °C. Spectra were acquired at 50 nm/sec with a time constant of 4 sec and a band width equal to 1 sec. The KH2PO4 buffer was used as a blank. The mean residue molar ellipticity [θ] was determined using equation 1:
where θ is the measured ellipticity in millidegree, C is the molar concentration of protein, N is the number of amino acids, I is the cell path length in centimeter.
Wild-type and mutant UBZ were obtained after the removal of GST tag. The minimal UBZ domain was treated with DTT and zinc sulfate, and then dialyzed against HNG buffer (30 mM HEPES, pH 8.0, 10% glycerol, 350 mM NaCl, treated with Chelex 100 resin). 1 μM wild-type or mutant scPolη UBZ was added in HNG buffer containing 0.1 mM 4-(2-pyridylazo)-resorcinol (PAR). The mixture was titrated with p-hydroxymercuribenzoic acid sodium salt (PHMB) to release the zinc ion from the scPolη UBZ. The equivalent of released zinc ion was determined by the absorbance at 500 nm using an ε500 value of 66,000 M−1 cm−1 for the (PAR)2·Zn2+ complex. The HNG buffer was used as a control.
The real-time surface plasmon resonance (SPR) binding measurements were performed on the Biacore 3000 system (Uppsala, Sweden). The running buffer used for analysis contains 10 mM HEPES, pH 7.4, 150 mM NaCl, 0.005% P20 detergent (GE Healthcare). All buffer solutions were freshly prepared, degassed, and passed through a 0.22 μm pore size filter. Ubiquitin was immobilized on the CM5 chip surface following the standard EDC/NHS coupling chemistry as previously described34, 45. The 1:1 mixture of 0.05 M N-hydroxysuccinimide (NHS) and 0.2 M 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) was flowed through the flow cell for 20 min at a flow rate of 5 μl/min. Then ubiquitin (10 μM) in sodium acetate buffer (10 mM, pH 5.5) were passed over the surface at 5 μl/min for 40 min, followed by a blocking step using ethanolamine (1 M, pH 8.5) at 5 μl/min for 20 min. The SPR binding experiments were performed at 25 °C using a flow rate of 20 μl/min. For each experiment, a underivatized flow cell was used as the reference surface. Six different concentrations of each protein were injected over the experimental and reference flow cell. The NaCl solution (1 M) was used to regenerate the surface after each injection cycle. The sensorgrams were processed using the double-referencing technique as described by Myszka et al. 46 to correct against both the underivatized surface and the buffer component that could generate signal. To obtain the equilibrium dissociation constant (Kd), the steady-state SPR response of the wild-type or mutant Polη UBZ binding to ubiquitin were fitted to equation 2:
where Req is the steady-state response, C is the analyte concentration, Rmax is the maximum binding response and Kd is the equilibrium dissociation constant. For C-term Polη binding to ubiquitin, the sensorgram did not reach saturation due to the intrinsically slow binding kinetics. Thus Req was obtained by fitting the sensorgrams to a 1:1 binding model using software BIAevaluation 3.2. Kd was obtained by fitting Req to equation 2 as described above. The binding and dissociation rate constants (konand koff) were fitted using the software BIAevaluation 3.2 to a 1:1 binding with mass transfer model.
A reaction mixture containing 12 nM Polδ, 150 nM monoubiquitinated PCNA, 60 nM RFC, 2.3 nM singly primed ssM13 DNA, 1.4 μM single-stranded DNA binding protein (E. coli SSB), 1 mM ATP, 50 μM dATP and 50 μM dTTP was incubated in a reaction buffer (25 mM Tris, pH 7.5, 5 mM MgCl2, 10% glycerol, 50 mM NaCl, and 0.1 mg/ml BSA) at 37 °C for 1 minute to allow the formation of Polη-PCNA holoenzyme on DNA. Then 160 nM wild-type or mutant full-length Polη was added into the mixture and incubated for 1 min. DNA synthesis was initiated by adding 50 μM dNTPs containing [α-32P]dGTP. The reaction was allowed to proceed for 30 seconds before being quenched with equal volume of 500 mM EDTA (pH 8.0). The DNA synthesis product was separated on a 1.2% alkaline agarose gel and quantified by PhosphorImager (Storm, GE Healthcare Bioscience) as previously described15, 36.
A reaction mixture containing 5 nM singly primed ssM13 DNA, 2.5 nM Polη, and 300 nM RPA were incubated in the presence or absence of 90 nM PCNA and 5 nM RFC in the assay buffer containing 40 mM Tris, pH 7.5, 8 mM MgCl2, 150 mM NaCl, 1mM DTT, 100 μg/ml BSA, 0.5 mM ATP at 30 °C for 1 min. The reactions were then initiated by addition of 100 μM dNTP and allowed to proceed for 10 min before quenched by equal volume of 500 mM EDTA (pH 8.0) solution. The reaction solution were treated with phenol-chloroform and precipitated with 100% ethanol. The DNA synthesis products were separated on a 10% urea-polyacrylamide gel and imaged by PhosphorImager (Storm, GE Healthcare Bioscience)
Ubiquitin-binding domains or motifs are essential mediators in a number of cellular processes. We demonstrated that the translesion synthesis DNA polymerase η in S. cerevisiae binds ubiquitin in a novel mode that has not been observed for the isolated UBZ domain. Quantitative binding study revealed high-affinity interaction between the C-terminal portion of Polη and ubiquitin. Given the widespread presence of ubiquitin-binding zinc finger (UBZ) and ubiquitin-interacting motif (UIM) in many newly identified ubiquitin receptor proteins, our findings provided useful insights into how the high-affinity ubiquitin binding can be achieved through the otherwise low-affinity ubiquitin-binding domains/motifs.
This work was supported by US National Science Foundation Grant MCB 0953764 (to Z. Z.) and National Institute of Health Grant CA107650 (to L.P.). L.H. acknowledges funding from the Hungarian Science Foundation (OTKA 77495-TAMOP-4.2.2/08/1).