TLK1B protects the cells from UV
We have previously shown that stable overexpression of TLK1B protects a normal cell line (MM3MG) from IR [11]. Therefore, we investigated if TLK1B could protect these cells from other types of DNA damage, or if it was specific for DSBs. Studies of its role in other types of DNA damage would help us to get a clearer picture of the position of this kinase in the DNA damage response pathway. We tested the effect of UV, which induces formation of pyrimidine dimers, on control cells (MM3MG), cells overexpressing TLK1B (MM3-TLK1B) and cells expressing the kinase dead mutant of TLK1B (MM3-KD) by a clonogenic assay (Figure 1). For making the kinase-dead mutant expressing cells the highly conserved D386 and K389 residues in the ATP-binding pocket of the kinase were replaced with Alanine. In vitro, the mutant kinase, expressed as a GST-fusion in bacteria, lost the capacity to phosphorylate itself and histone H3 [9]. The kinase-dead (KD) protein was then expressed from the BK-Shuttle vector as done before for the wt TLK1B. MM3-KD cells may be expected to show increased sensitivity to UV damage, if the KD mutant acts as a dominant negative for TLK1 and TLK2 substrates. The BK-Shuttle vector replicates as an episome with a very consistent copy number in each cell and hence, the transfected population gives homogeneous expression of the transgene, making it unnecessary to select individual G418-resistant clones [32]. This is an important point since the results obtained thereafter cannot be simply attributed to the selection of an aberrant clone.
We found that 21% of the MM3MG cells survive at 2 J/m2, 10% of the MM3-KD cells survive, while 63% of the MM3-TLK1B cells survive at the same dose. For the dose of 4 J/m2 also, the survival fraction is considerably higher in MM3-TLK1B cells as compared to MM3MG and MM3-KD cells (19.3% vs. 1.4% vs. 0.7%); (P < 0.01 in both cases). This indicates that overexpression of TLK1B significantly increases the resistance to UV radiation.
In vivo repair of UV damage in MM3MG and MM3-TLK1B cells
We next wanted to determine the basis behind the increased survival of MM3-TLK1B cells after exposure to UV. Expression of TLK1B did not appreciably alter the transcriptome in MM3MG cell lines (microarray analyses, unpublished data), indicating that the protective effect was likely post-transcriptional. We recently showed that TLK1B promotes the repair of double-stranded breaks (DSBs), most likely by promoting chromatin remodeling at the sites of damage [24]. Therefore we hypothesized that a similar mechanism of more efficient DNA repair might be responsible for the increased survival of MM3-TLK1B cells after exposure to UV radiation. In order to investigate this, the repair of genomic DNA in MM3MG and MM3-TLK1B cells was determined using an immunoblot assay and monoclonal antibodies specific to CPDs, over a time course. As shown in figures 2A and 2B, the MM3-TLK1B cells efficiently repaired almost all the CPDs by 12 hours. On the other hand, MM3MG cells show very poor levels of CPD removal.
Another method was used to compare the repair abilities of the two cell lines. For this purpose, we assessed the repair of episomal vectors in cells over a time course following exposure to UV. MM3MG cells (that were stably transfected with the empty BK-shuttle vector) and the MM3-TLK1B cells were exposed to 5 J/m2 of UV radiation, and allowed to recover for 0 to12 hours (Fig. 3). Episomes were extracted at various time points (0, 2, 4, 8 and 12 hours) and were either mock-treated or digested with T4 endonuclease V enzyme, which specifically cuts DNA at CPD-sites. Since the episomes are damaged on both strands at multiple sites, digestion with T4 endonuclease results in extensive cleavage of the plasmid DNA, whereas in unirradiated cells T4 endonuclease leaves the plasmids intact (see control lanes). Figure 3 shows the repair ability of MM3MG and MM3-TLK1B cell lines by this method. On inspection of the two gels it was seen that the majority of low molecular weight episomal DNA was recovered into repaired high molecular weight DNA in the MM3-TLK1B cells between 8 and 12 hours, while the T4 endonuclease treated episomal DNA still appeared as a low molecular weight smear in the MM3MG cells. These two experiments strongly suggest that the MM3-TLK1B cells repaired DNA faster than the control MM3MG cells.
The phosphorylation of H3 recovers faster in TLK1B-expressing cells
Genotoxins cause a temporary, ATM-dependent, inactivation of TLK1 [6] that can best be monitored through a loss of phosphorylation of histone H3-S10 [9]. We set out to test whether UV would also be a treatment that caused dephosphorylation followed by re-phosphorylation of H3 when the cells are allowed to recover post treatment. Indeed, there was a dramatic loss of H3P-S10 following UV damage, which persisted for 8 hours in the MM3MG cells, but only for 6 hours in the cells expressing TLK1B (Figure 4). Therefore, if the phosphorylation of H3 can be taken as an indication of repair activity in UV-damaged cells, then there is a much faster recovery/repair in cells expressing TLK1B. The phosphorylation of H3 was taken as an indicator of TLK1 activity/inactivation, since no change in mobility was observed for the other substrate of TLK1, Asf1 (data not shown) following UV. We should stress that the change in H3 phosphorylation is most likely through an ATM (or more likely ATR in the case of UV)-mediated inhibition of TLK1 through Chk1 [6] and not a change in the cell cycle (like a mitotic arrest) since the loss of H3 phosphorylation after UV was immediate and reversible.
In vitro assays to study the repair of UV-induced DNA damage
We elected to use an in vitro reconstituted system to study the role of TLK1B in the repair process. For these assays we used UV-damaged Bluescript plasmid which was preassembled into a chromatin template. Nucleosome assembly on supercoiled plasmid was done by the deposition of histones using salt dialysis, and efficient assembly was confirmed by observation of a regularly protected DNA ladder on digestion with micrococcal nuclease (MNase), as shown in Figure 5A. As a control, Bluescript plasmid not assembled into nucleosomes was digested completely by MNase, resulting in a smear (Figure 5B). Repair on pre-assembled nucleosomal templates was measured by the incorporation of radiolabeled nucleotides into plasmid DNA during repair synthesis catalyzed by nuclear extracts of MM3MG and MM3-TLK1B cell lines. Unirradiated plasmid DNA was used as a negative control, and only minimal background incorporation of label was observed in this mock NER assay. Figures 6A and 6B show that only reactions with irradiated plasmid DNA show incorporation of radiolabeled nucleotide, which suggests that the label is incorporated at sites of repair of UV damage, and the fact that more than twice as much repair is seen in the reactions containing extracts from MM3-TLK1B cells indicates that the presence of TLK1B stimulates repair. Similar results were obtained when plasmids not preassembled into nucleosomes were used for the repair assay (data not shown), but we should point out that this is likely due to the fact that the plasmid was getting assembled into chromatin by extract-mediated chaperones and endogenous histones (data not shown).
TLK1B stimulates chromatin assembly in vitro
From the results obtained in Figure 6, it can be inferred that there is more efficient DNA repair of the plasmids incubated with MM3-TLK1B extracts. The only currently known substrates of TLK1 (or TLK1B) are histone H3 [11] and Asf1 [1], which suggest that perhaps chromatin remodeling is involved in the more efficient repair process. The Bluescript DNA was fully competent as a substrate for repair synthesis, but the previous assay yielded little evidence about the chromatin remodeling capacity of the MM3MG and MM3-TLK1B extracts. Therefore, we investigated the ability of TLK1B to enhance the assembly of chromatin by an in vitro plasmid supercoiling assay (Figure 7A). In this assay, the Bluescript plasmid (nearly 100% supercoiled, input, lane 1) was used as a template for the deposition of core histones in the presence of MM3MG nuclear extract and an energy mix. In the absence of exogenous histones, the extract causes the bacterially supercoiled form to convert to mostly the relaxed form due to endogenous topoisomerases (lane 2). However, after incubation in the presence of histones, the plasmid migrates as a series of discrete supercoiled forms due to the formation of nucleosomes, which decrease the linking number by one integer (i.e., one negative supercoil) per nucleosome. The addition of recombinant TLK1B stimulated the formation of the more highly supercoiled forms, particularly a form that comigrates with the bacterially supercoiled plasmid (lane 5, band 7), and thus maybe fully assembled into nucleosomes, i.e. one nucleosome per 200 bp, mimicking the superhelical density (-0.05) of the bacterially propagated plasmid. Asf1 has been shown to stimulate chromatin assembly in vitro [16]. When recombinant Asf1b protein was added to the in vitro reaction (lane 4), it also increased the supercoiling. Strikingly, maximum supercoiling was seen in lane 6 when both Asf1 and TLK1B were included. This increase in supercoiling is possibly due to the interaction of TLK1B with Asf1, wherein TLK1B increases the chromatin assembly activity of Asf1. The results of the supercoiling assay are further confirmed by the densitometric analysis of each lane (Figure 7A). In comparison to lanes 2 and 3, lanes 4, 5 and 6 show an additional band (band 7), which, as discussed above, may represent the fully supercoiled topoisomer that runs like the most supercoiled band in lane 1. However, the intensity of this band is 5–6 fold higher in lane 6 (when both Asf1b and TLK1B are included) as compared to lanes 4 and 5.
The importance of TLK1 (or TLK1B) in the chromatin assembly is further demonstrated when an antibody that inactivates TLK1 is added to the in vitro chromatin assembly reaction prior to the addition of the plasmid. As is seen in Figure 7B, addition of anti-TLK1 antibody decreased the supercoiling (lane 3), while no appreciable effect was seen when a non-specific antibody was used (lane 4, pre-immune serum). At the same time, addition of recombinant TLK1B protein to the TLK-depleted extract could restore the supercoiling activity considerably, indicating that the recombinant protein could complement the depletion by the antibody (lane 5). Both extracts showed very similar results in this case.
To analyze the supercoiling activity of the two extracts during repair synthesis (which is dependent on CAF-1) a similar assay was performed, but using completely relaxed UV-damaged plasmid as a substrate. In such an assay, NER nicking of the plasmid at CPDs will tend to maintain the plasmid in a relaxed form, whereas at the same time extract-mediated chromatin assembly will decrease the superhelicity of the plasmid generating faster migrating forms. As shown in Figure 8, completely relaxed UV-damaged plasmid was incubated with either MM3MG or MM3-TLK1B extract. As a control, undamaged plasmid was used for comparison. From the figure, it is clear that the MM3-TLK1B extract induces more supercoiling of the UV-damaged plasmid, while the plasmid remains in its relaxed form when incubated with the MM3-MG extract (compare lanes 3 and 5). This confirmed that the TLK1B extract has better chromatin remodeling activity for UV-damaged plasmids.
Role of TLK1-Asf1 relationship in DNA repair
The role of Asf1 in DNA repair and maintenance of genomic integrity has been well documented. Cells lacking CAF-1 and RCAF (a complex of Asf1, and acetylated histones H3 and H4) [14] are hypersensitive to DNA-damaging agents, such as camptothecin, suggesting a possible defect in double-strand break (DSB) repair [19]. In budding yeast lacking Asf1, there is activation of the DNA damage checkpoint [21]. Asf1 has also been shown to act synergistically with CAF-1 to assemble nucleosomes during nucleotide excision repair in vitro [20]. It is possible that increased repair synthesis by the MM3-TLK1B extracts can be attributed to more efficient chromatin remodeling coupled DNA repair because of a more active Asf1. If TLK1B increases the activity of Asf1, then addition of Asf1 should give a comparable result as addition of TLK1B. In order to determine this, the in vitro repair assay was repeated on the UV-damaged nucleosomal Bluescript template, but with increasing amounts of recombinant Asf1b protein. Since the effect of MM3-TLK1B on repair synthesis was evident as early as 10 minutes (Figure 5, compare lanes 2 and 6 of the right panel), this assay was performed for 10 minutes. The results are shown in Figure 9A. As seen before, the repair synthesis by the MM3-TLK1B extracts was several fold higher than that by the MM3MG extracts (compare lanes 1 and 5). However with increase in the amount of Asf1b protein (100 and 200 ng), there was a considerably higher amount of dATP incorporation by the MM3MG cell extracts (compare lanes 1, 2 and 3). Addition of recombinant Asf1b protein also increased the repair synthesis in case of the plasmid treated with the MM3-TLK1B extract. In fact, addition of recombinant Asf1b resulted in a similar increase in dATP incorporation in both the cases (1.5 times increase with 100 ng and 2 times increase with 200 ng). A graphical representation with three repeats of this experiment is shown in Figure 9B.
When a western blot for Asf1b was done on whole cell extracts of both the cell lines, we clearly detected higher migrating forms of the protein in the MM3-TLK1B extracts as compared to the MM3MG extracts (Figure 9C, compare lanes 1 and 2), indicating that it is more phosphorylated. That the slower migrating forms were due to phosphorylation was confirmed when treatment with calf-intestinal alkaline phosphatase (lane 3) caused the Asf1b in the TLK1B cells to migrate like the Asf1b in the MM3MG cells. The most direct interpretation of these results is that TLK1B extracts have increased activity of Asf1 due to phosphorylation.
The role of Asf1 was further demonstrated by the results shown in Figure 10A. In this case the repair assay was done in the same way as described before (incubating the UV-damaged nucleosomal DNA template with nuclear extracts, energy mix and [α]32P-dATP for 10–60 minutes), but Asf1-antiserum was added to the reaction mix, prior to addition of the plasmid (lanes 2–5). That the Asf1-antibodies bind the native proteins was determined by immunoprecipitation (data not shown). As a control, the reaction was also done in the presence of a non-specific antibody (pre-immune serum, lane 1). There was a significant decrease in the repair synthesis in the presence of Asf1 antiserum (compare lane 1 with lanes 2, 3 and 4). Only when the reaction was allowed to continue for a longer time (60 minutes, lane 5), did the repair synthesis reach levels similar to that seen in lane1. Similar results were obtained with both the MM3MG and MM3-TLK1B extracts. This shows that Asf1 is important for the DNA repair process, even in the presence of TLK1B in the extract. Interestingly, Asf-1 depletion did not show any appreciable effect in case of repair assays that were done on non-chromatin UV-damaged templates (data not shown). This indicates that Asf1 only affects repair during assembly/disassembly of chromatin, and it is not intrinsically increasing the overall repair reaction on UV damaged templates.
The effect of the Asf1-antiserum could also be complemented by adding back recombinant Asf1b protein, indicating that it was specifically interfering with the Asf1 function during the repair process (Figure 10B).
Role of TLK1B in the DNA repair process
Next, we tested the role of TLK1B itself using the in vitro assays (Figure 11). The in vitro repair assay was repeated as before, however, in the presence (lane 3) or absence (lane 1) of TLK-1 antiserum. As a control, a non-specific antibody (pre-immune serum, lane 2) was also used. We saw that the repair synthesis diminished by half (lane 3) as compared to the control extracts (lanes 1 and 2), as indicated by the amount of radiolabel incorporated. However, when increasing amounts of recombinant TLK1B were added to the in vitro reaction, it could restore repair synthesis in the TLK-depleted extracts. Here too, similar results were seen in case of both MM3MG and MM3-TLK1B extracts.