TLK1B promotes repair of DSBs via its interaction with Rad9 and Asf1
© Canfield et al; licensee BioMed Central Ltd. 2009
Received: 17 August 2009
Accepted: 20 December 2009
Published: 20 December 2009
Skip to main content
© Canfield et al; licensee BioMed Central Ltd. 2009
Received: 17 August 2009
Accepted: 20 December 2009
Published: 20 December 2009
The Tousled-like kinases are involved in chromatin assembly, DNA repair, transcription, and chromosome segregation. Previous evidence indicated that TLK1B can promote repair of plasmids with cohesive ends in vitro, but it was inferred that the mechanism was indirect and via chromatin assembly, mediated by its interaction with the chromatin assembly factor Asf1. We recently identified Rad9 as a substrate of TLK1B, and we presented evidence that the TLK1B-Rad9 interaction plays some role in DSB repair. Hence the relative contribution of Asf1 and Rad9 to the protective effect of TLK1B in DSBs repair is not known. Using an adeno-HO-mediated cleavage system in MM3MG cells, we previously showed that overexpression of either TLK1B or a kinase-dead protein (KD) promoted repair and the assembly of Rad9 in proximity of the DSB at early time points post-infection. This established that it is a chaperone activity of TLK1B and not directly the kinase activity that promotes recruitment of 9-1-1 to the DSB. However, the phosphorylation of Rad9(S328) by TLK1B appeared important for mediating a cell cycle checkpoint, and thus, this phosphorylation of Rad9 may have other effects on 9-1-1 functionality.
Here we present direct evidence that TLK1B can promote repair of linearized plasmids with incompatible ends that require processing prior to ligation. Immunodepletion of Rad9 indicated that Rad9 was important for processing the ends preceding ligation, suggesting that the interaction of TLK1B with Rad9 is a key mediator for this type of repair. Ligation of incompatible ends also required DNA-PK, as addition of wortmannin or immunodepletion of Ku70 abrogated ligation. Depletion of Ku70 prevented the ligation of the plasmid but did not affect stimulation of the fill-in of the ends by added TLK1B, which was attributed to Rad9. From experiments with the HO-cleavage system, we now show that Rad17, a subunit of the "clamp loader", associates normally with the DSB in KD-overexpressing cells. However, the subsequent release of Rad17 and Rad9 upon repair of the DSB was significantly slower in these cells compared to controls or cells expressing wt-TLK1B.
TLKs play important roles in DNA repair, not only by modulation of chromatin assembly via Asf1, but also by a more direct function in processing the ends of a DSB via interaction with Rad9. Inhibition of Rad9 phosphorylation in KD-overexpressing cells may have consequences in signaling completion of the repair and cell cycle re-entry, and could explain a loss of viability from DSBs in these cells.
The gene Tousled of Arabidopsis thaliana encodes a protein kinase which, when mutated, results in abnormal flower development . This was proposed to be linked to a replicative defect during organogenesis, but which may also result from failure to protect the genome from UV damage [2, 3], resulting in mitotic aberrations [4–6]. Two Tousled genes (TLK1 and TLK2) were identified in mammals [7, 8], and were confirmed as encoding kinases. Few physiologic substrates of Tousled like kinases (TLKs) have been identified, namely Asf1 , histone H3 , Aurora B , and more recently Rad9 in mammalian cells  and two mitotic kinesins in Trypanosomes . This suggested a function in chromatin assembly  during transcription [14, 2], DNA repair [3, 15], and condensation of chromosomes at mitosis . Evidence also exists about a link between TLKs and a DNA damage relay . This can be inferred from research that shows that the activity of TLK1 is inhibited by IR and genotoxins . The inhibition is mediated by ATM via Chk1 by direct phosphorylation at S695 . These findings identify a functional cooperation between ATM and Chk1 in propagation of a checkpoint response mediated by transient inhibition of TLK1, which may regulate processes involved in chromatin assembly . A splice variant of TLK1 that is translated upon genotoxic stress , TLK1B, has invoked interest because of its established role in cell survival from DNA damage [3, 10, 15]. TLK1 and TLK1B share identity in the catalytic domain and hence interact with mostly the same substrates, and we often refer to them as TLK1/1B in this respect. Earlier studies showed that elevated expression of TLK1B promotes cell survival after radiation or doxorubicin by facilitating DNA repair . These initial studies suggested that the role of TLK1B in radioprotection could be mediated through Asf1, leading to changes in chromatin disassembly/assembly coupled to repair [3, 15]. In fact, initial studies have explored the role of Asf1 and TLK1/1B by in vitro chromatin assembly on a plasmid [19, 15]; and a possible role of TLKs in repair was reviewed in . More recently however, the identification of Rad9 as a substrate of TLK1B suggests that TLK1B can affect DSB processing through modulation of 9-1-1 activity. Rad9, Rad1, and Hus1 form a trimeric complex (termed 9-1-1) that is structurally similar to the Proliferating Cell Nuclear Antigen  "sliding-clamp", which encircles the DNA conferring processivity to polymerases [22–24]. 9-1-1 assembles in a complex at sites of damage , and it is the genotoxin-activated RFC-Rad17 "clamp loader" that locks the 9-1-1 complex onto DNA . The 9-1-1 then serves as a scaffold for assembly of DNA repair proteins, Flap endonuclease [26, 27], DNA polymerase β , DNA ligase 1 , and DNA glycosylase MutY , in addition to aiding processing of the DNA ends by its own exonucleolytic activity [31–33]. Accordingly, Rad9 is held to be involved in translesion repair synthesis , mismatch repair , removal of 8-oxoguanine , and NHEJ and HRR [37, 38]. We showed that TLK1B phosphorylates Rad9 at S328, and that this phosphorylation appears to play a role in resumption of cell cycle after IR. However, TLK1B had also a function as a chaperone for Rad9 assembly at DSBs that was independent of its kinase function . We now address in more details, in vitro and in intact cells, the role of TLK1B in DSB repair, in particular in relation to Rad9 functionality.
We previously described a system with nuclear extract of MM3MG cells that allows to monitor repair of a plasmid that is cut with EcoRI, which leaves a cohesive 5' overhang . Repair of plasmid was seen on EtBr-stained gels as formation of high mobility forms, which are the result of religation and simultaneous assembly of nucleosomes on the template. The increased mobility of the plasmid compared to the linear form is due to a decrease in the linking number by formation of chromatin, resulting in supercoiling - only covalently closed plasmid can be supercoiled. We have more recently analyzed the predominant repaired junctions obtained from this in vitro repair system, by transformation of bacteria and sequencing of the plasmids rescued from several clones. In most cases examined, the repair had been a high-fidelity ligation which resulted in reconstitution of the EcoRI site by simple annealing and joining of the ends (16/20 clones). Two different clones displayed a deletion of the 4-base overhang, and the two remaining had a 3-base fill-in and deletion of the last base. We had shown that the addition of TLK1B hastened the repair and supercoiling of the plasmid , which at the time was attributed to the known interaction of TLK1/1B with Asf1. However, we had not examined this directly. In particular, it was not known if depletion of Asf1 would result in loss of ligation and supercoiling. This seemed possible since Asf1, by virtue of its activity on nucleosome assembly, could perhaps promote repair indirectly by compacting the plasmid and bringing the ends in juxtaposition. Thus, we needed to test the effect of Asf1 depletion on ligation of cohesive ends.
To test if the repair of incompatible ends was also stimulated by TLK1B, we set out to probe if the ligation of a blunt end generated by EcoRV and one generated by EcoRI (both in the MCS of Bluescript) could be repaired in our system. The predominant form of repair of incompatible ends in mammalian extracts is blunt-end ligation, which is achieved by either fill-in or resection of the ends . In rarer cases, cohesive-ends ligation occurs via more complex processing that results in short stretches of micro-homology . We first determined if this nuclear extract system could repair the plasmid by fill-in of the EcoRI end, followed by blunt ligation to the EcoRV end. Fill-in of the EcoRI end by repair polymerases was monitored via incorporation of [α32P]dATP. In this case, the generation of a flush end at the EcoRI side could either take place by resection of the remaining 5'-ApA overhang or, less likely, by fill-in with residual dTTP that may be present in the nuclear extract. Successful end-joining and simultaneous plasmid supercoiling via formation of nucleosomes was monitored as an increased mobility (Fig. 1C). Since the supercoiled form is labeled, there must have been primarily a blunt-end joining of the filled site, as complete resection of the EcoRI end would have resulted in loss labeling. At least, this assay system is geared toward studying primarily such repair reaction. The addition of recombinant TLK1B resulted in significantly greater amounts of religated/supercoiled molecules than extract alone, as shown by examining the autoradiogram or the EtBr-stained gel. Transformation of bacteria yielded no colonies with linearized plasmid, but dozens with the extract-mediated repair/ligation. Analysis of the repaired junctions in plasmids from 10 clones revealed filling with two A's and the excision of the two remaining 5'A's. Also, under these repair conditions, there was one predominant religated, fast-migrating form of the plasmid. This is in contrast to what we normally see for repair of plasmids singly cut with EcoRI, which typically results in a ladder of topoisomeric forms - the result of the assembly of a dense array of nucleosomes (see panel B). A possible explanation for this different result is that simple ligation of cohesive ends is rapid compared to ligation of incompatible ends, thus allowing more time for the assembly of a dense array of nucleosomes, which are later resolved by endogenous toposiomerases and resulting in supercoiling.
Whereas the addition of TLK1B, the form that is induced in presence of DSBs , stimulated repair/supercoiling, it was not known if TLK1, the constitutively expressed form, is also important for DNA repair. To test this, we immunodepleted the extract of TLK1, and we monitored ligation/supercoiling on a plasmid cut with EcoRI and EcoRV. In Fig. 2A, we show western blots of extracts immunodepleted of TLK1, Rad9, and Ku70, which were then used in this study. In Fig. 2B, we show that depletion of TLK1 resulted in a small effect on plasmid ligation/supercoiling, although it appeared that the labeling of the ends also occurred more slowly (last 3 lanes). In fact, it seems that the reduction in supercoiling is proportional to the reduction in labeling, and likely through an effect of TLK1 on Rad9. Hence, in this case, the fill-in reaction may be stimulated by presence of TLK1, although not absolutely required. Adding back TLK1B restored the efficient labeling of the ends and increased ligation/supercoiling (Fig. 2B, middle lanes). We have failed to produce full-length recombinant TLK1, and hence we have not been able to study the effect of this larger isoform in add-back assays.
We propose that the effect of TLK1B in promoting plasmid repair is mediated by its specific interaction with Asf1  and Rad9 . But the evidence so far is that Rad9 is more directly affecting repair by aiding processing of the ends, seemingly by fill-in via recruitment of repair polymerases [31–33]. Thus, we first tested if the simple repair of cohesive ends was dependent on Rad9. Extract was immunodepleted of Rad9, and EcoRI-cut plasmid was added. Ligation and formation of supercoiled forms was assessed by gel electrophoresis and staining with EtBr. Clearly, the plasmid was rapidly religated in these conditions and assembled in nucleosomes; and presence of Rad9 was not needed for this type of repair/supercoiling (Fig. 2C, left panel). This should not be surprising, since the joining of this type of cohesive ends depends mostly on ligase 4/XRCC4 and the likely association of DNA-PK to the break [42, 41], and not so much on the 9-1-1 complex. We then tested if the repair of incompatible ends does rely on Rad9. The repair reaction was carried on plasmid cut with EcoRI and EcoRV in presence of [α32P]dATP. Labeling of the plasmid and modest ligation and supercoiling was obtained with whole extract, but depletion of Rad9 inhibited both (Fig. 2C, right panel). Adding back Rad9 restored robust labeling of the ends and also more religated/supercoiled plasmid. Overall, this indicates that some repair function of Rad9, likely its aiding of fill-in repair , its own exonucleolytic activity , or its interaction with FEN1  are important to process these types of ends to promote ligation.
Finally, we set out to test if the repair of incompatible ends depends on Ku70 and DNA-PKcs. This was a requirement for the experiment shown in Fig. 1D, and is also a control for the entire work, as end-joining normally requires DNA-PK . In this case, to avoid possible differences in end-labeling efficiency in the different reactions, after cutting with EcoRI and EcoRV, Klenow polymerase was added with [α32P]dATP to pre-label the plasmid. The labeled plasmid was then resin-purified and added to extract immunodepleted of Ku70 or not. Depletion of Ku70 resulted in loss of ligation/supercoiling (Fig. 2D, left panel), indicating that formation of a synaptic complex consisting of the plasmid ends associated with DNA-PK  is a pre-requisite for religation in this system. To further study if DNA-PKcs was important in such conditions, wortmannin (WMN) was included in these reactions. WMN inhibits PI3K members, including DNA-PKcs, ATM and ATR. However, research with more specific inhibitors of DNA-PKcs and the use of extracts from cells deficient in DNA-PKcs have strongly suggested that DNA-PK is the primary enzyme involved in the in vitro repair of these types of ends . The addition of WMN almost completely prevented the formation of religated/supercoiled forms, even when TLK1B was added to the extract (Fig. 2D, right panel). These experiments strongly indicate that the processing of ends prior to ligation requires Rad9 and DNA-PK.
In the past few years, significant evidence has emerged indicating that TLKs are involved in DNA repair [3, 15, 17]. The mechanism of repair was held to be mainly indirect, and modulated by the known interaction of TLK1 with Asf1 . However, more recently Rad9 was identified as a critical interacting partner of TLK1B . This has opened the door for different studies of the function of TLK1/1B as a direct mediator of DSB repair via processing of the ends, which is a critical function attributed to Rad9 (or 9-1-1). Indeed, Rad9 was critical in stimulating processing and repair of cut plasmids with incompatible ends; and we previously demonstrated that Rad9 is essential for the stimulation of DSB repair by TLK1B, since overexpression of TLK1B alone in Rad9-/- ES cells does not restore repair . We should stress, however, that in our in vitro repair conditions, we start the reactions with naked DNA, and that the formation of chromatin on the plasmid is detected only after the plasmid is reclosed and becomes supercoiled. This situation is quite different than that in intact cells, where the repair machinery must contend with chromatin . In such conditions, the role played by Asf1 may be quite different, and likely more important, in disassembly of nucleosomes to promote access of the repair machinery to unencumbered DNA [11, 49]. TLK1B displays two activities on Rad9: phosphorylation, which may affect the checkpoint role of Rad9, and chaperone activity, which may assist its recruitment to DSBs and resulting in ends processing. Although we were not able previously to assign a specific role for the phosphorylation of Rad9 by TLK1B, we have now shown that expression of the KD dominant mutant, which reduces the phosphorylation of Rad9 in vivo , resulted in a delay in the release of Rad9 and Rad17 from the DSB. Although we were not able to test a model for a change in P-Rad9 with the adeno-HO-infected cells, we were able to confirm a pattern of fluctuation during IR and recovery. Hence, the phosphorylation of S328-Rad9 decreases during the initial period after IR and then recovers during repair. This is consistent with the model proposed above. Overall, this pattern of P-Rad9 fluctuation may be important for restoring cell cycle re-entry and release from the checkpoint induced by the DSB, and improving viability. In fact, KD-expressing cells are very radiosensitive .
MM3MG cells (5 × 106) were centrifuged in a swing-out rotor at 150 × g for 10 min at 4°C. The pellet was resuspended in 1 ml ice-cold nuclei isolation buffer [50 mM Tris-HCl (pH 7.5), 0.05 mM spermine, 0.125 mM spermidine, 0.5 mM EDTA, 20 mM KCl, 0.1 mM PMSF, 0.1% (v/v) aprotinin, and 1 mM DTT]. After swelling on ice for 10 min, the cells were broken with a dounce homogenizer using a Wheaton B pestle. Nuclei were pelleted by centrifugation at 150 × g for 10 min at 4°C. Nuclei were lysed by suspending in ice-cold nuclei extraction buffer [10 mM HEPES (pH 7.5), 350 mM KCl, 0.2 mM EDTA, 3 mM MgCl2, 0.1 mM DTT, 0.2 mM PMSF, 0.2% aprotinin (v/v), and 15% (v/v) glycerol]; and incubated on ice for 30 min. The nuclear envelopes were then removed at 70,000 × g for 20 min at 4°C.
Rad9, Asf1, and Ku70 antibodies were purchased from Santa Cruz Biotechnology. TLK1/1B antiserum was made in our lab. Anti-E3-11.6k (fusion protein with HO) was a kind gift of Drs. Tollefson and Wold (St. Louis University). P-S328-Rad9 antiserum was from Abgent. Asf1 siRNAs were from Dharmacon. Recombinant Rad9 and TLK1B were prepared as previously described . The MM3MG cells overexpressing TLK1B or the KD were previously described , as well as conditions for adeno-HO infection and ChIP adjacent the DSB .
The supernatant was retained and dialyzed for 1 h against E buffer [20 mM Tris-HCl (pH 8.0), 0.1 mM KOAc, 20% (v/v) glycerol, 0.5 mM EDTA, and 1 mM DTT], fast frozen, and stored at -80°C.
We are grateful to Ceslovas Venclovas for allowing us to adapt his model published in . This work was supported by grant BCTR131906 from the Susan G. Komen Foundation, and a grant from the LSUHSC Styles Foundation.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.