Active RNAP backtracking exposes DNA lesions to repair

We have shown that the core TFIIH can push the stalled RNAP backwards in the presence of XPA and XPG (Fig. 30c). However, it is unknown if the backtracking is efficient enough to expose DNA lesions to repair. RNAP arrested by a CPD sequesters ~18 nucleotides of the DNA downstream of the lesion^51,233. In addition, XPG cleaves the DNA 5 nucleotides 3’ of the lesion during NER^107,111. Thus, RNAP has to be pushed backwards for more than 23 nucleotides to fully expose the lesion to repair enzymes.

We designed a biochemical assay which allowed us to determine the RNAP position on the DNA after backtracking at single-nucleotide precision (Fig 32a). We assembled the elongation complex using a CPD-containing DNA scaffold and a fluorescently labeled RNA.

Upon the addition of NTPs, the RNAP transcribed into the CPD lesion, as we observed the appearance of the 53 nucleotides long RNA products (Fig 32b).

Figure 32 | Active RNAP backtracking by XPB exposes CPD lesions to repair.

(a) Schematic representation of the experimental design used for measuring the extent of RNAP backtracking at a single nucleotide resolution. (b) In vitro transcription reaction (2.8 pmol RNAP) on a CPD containing DNA scaffold produced 53 nucleotides long RNA products, as in Fig.25. After the incubation with increasing amounts of the core TFIIH, XPA and XPG (2.2, 4.4 and 8.8 pmol) and RNA cleavage induction by TFIIS (0.5 pmol) we observed the 24 nucleotides long cleavage products, suggesting that RNAP was backtracked for 29 nucleotides. (c) Backtracking was also observed when core TFIIH containing XPD inactive mutant⁷³ was used, but not in the presence of triptolide which inhibits XPB^200,201. Thus, XPB is responsible for pushing the RNAP backwards.

In the next step, we added the backtracking machinery composed of the core TFIIH, XPA and XPG to the lesion-arrested RNAP and incubated the reaction with ATP to enable backtracking. Finally, after the RNAP endonuclease activity was triggered by TFIIS, we observed the appearance of 24 nucleotides long RNA cleavage products. Since the RNA cleavage is mediated by RNAP, we can conclude that the RNAP active site was pushed backwards for 29 nucleotides with increasing amount of the backtracking machinery (Fig.

32b). Thus, the backtracking is sufficient for the complete removal of RNAP from the lesion site. The backtracking was also observed when the XPD inactive mutant was used, but the effect was abolished when XPB translocase was inhibited by triptolide (Fig 32c).

Figure 33 | Crosslinking mass-spectrometry network of the RNAP-core TFIIH–XPA–XPG complex and the effect of DNA lesions on XPB translocase activity.

(a) Core TFIIH subunits are color-coded as in Fig. 14, XPG is in brown, XPA in purple and RNAP subunits in gray, except Rpb5 is in green and Rpb9 in orange. XPA and XPG crosslinks to the core TFIIH are shown with purple and brown lines, respectively. Crosslinks between XPA and RNAP are shown with thick purple lines and between XPB and RNAP with thick pink lines. The list of inter-subunit crosslinks is provided in the Supplemental Table 4. Domain abbreviations as in Fig. 9. (b) Selected crosslink sites for XPA (purple spheres) and XPB (pink spheres) were mapped onto the RNAP structure (PDB code: 5FLM)¹²³ and cluster around the downstream DNA (dashed black circle). RNAP subunits are shown in gray, except Rpb5 is in green and Rpb9 in orange, DNA is in blue. (c) XPB translocase assay performed as in Fig. 11c. (left) Schematic representation of scaffolds used to check if a lesion mimic (internal biotin) placed in the 5’-3’ or the 3’-5’ DNA strand can inhibit the XPB translocase activity. Position of the internal biotin is indicated with red squares. (middle) Representative fluorescence traces of the core TFIIH translocation on corresponding scaffolds (the color of the trace matches the color of the frame around the corresponding scaffold). XPB cannot translocate efficiently when biotin is installed in the 5’-3’ DNA strand and the biotin located in the 3’-5’ DNA strand has no effect on XPB translocation. (right) Bar graph shows the percentage of triplex disrupted after 4000 s for 2 independent replicates. Error bars represent s.d. from the mean values.

To gain insight into the architecture of the RNAP backtracking complex we obtained a protein interaction network within the purified RNAP-core TFIIH-XPA-XPG complex by chemical crosslinking and mass-spectrometry (Fig. 33a). The complex was assembled by a large-scale in vitro transcription and backtracking reaction on a CPD containing DNA scaffold, followed by the removal of excess NER factors and NTPs via size-exclusion chromatography (Methods). Many crosslinks were detected within the core TFIIH–XPA–

XPG complex and between the RNAP subunits, however, crosslinks between the repair proteins and RNAP were rather scarce. It may be that the interface between RNAP and the backtracking machinery is quite small and largely mediated by DNA. Also, the complex with RNAP might be unstable and accompanied by a high rate of TFIIH dissociation, which could lead to more inefficient crosslinking. XPA crosslinks to the core TFIIH ATPases, TTDA and the N-terminus of p52 subunit, which is consistent with the structure we presented here (Fig. 14). XPG crosslinks to XPB and XPD, which is similar to other crosslinking data we acquired (Fig. 23). Interestingly, the XPB and XPA crosslink sites map to Rpb1, Rpb2, Rpb5 and Rpb9 RNAP subunits, mostly around the downstream DNA (Fig.

33b). This suggests that XPB binds the downstream DNA during TCR and pushes the lesion-arrested RNAP backwards by translocating on the DNA duplex towards RNAP.

Since we observed that the backtracking of lesion arrested RNAP proceeds for exactly 29 nucleotides (Fig. 32b), it is tempting to speculate that XPB pushes RNAP backwards until it encounters the lesion. This would suggest that a lesion in the 5’-3’ DNA strand inhibits the XPB translocase activity while XPB migrates on the downstream DNA towards RNAP. To test this, we performed a double-strand DNA translocase assay using DNA templates which contain an internal biotin as a DNA lesion mimic placed in the 5’-3’

or the 3’-5’ DNA strand (Fig. 33c). Indeed, we observed a strong inhibition of XPB translocase activity when the lesion mimic was located in the 5’-3’ DNA strand but not when it was located in the 3’-5’ DNA strand, as also shown for the yeast XPB homologue Ssl2⁶⁹. Thus, during TCR, the XPB translocase activity might be employed for the removal of RNAP from the lesion site and for the subsequent verification of the lesion in the template strand. After the lesion verification, XPB may dissociate from the DNA and provide the space for lesion processing by the canonical NER pathway.