Subassemblies of MCM form distinct complexes with FACT heterodimer
Our previous work on the transcription elongator FACT has identified and characterized the physical and functional interaction between FACT and a subassembly of the MCM helicase complex, namely MCM2/4/6/7 [29]. To carry out a more thorough examination of FACT-associated complexes, we performed additional Western analysis on different anti-FACT immunoprecipitates isolated from whole cell (Figure 1A, left panel) and nuclear extracts (right panel). The specificity of these antibodies was demonstrated by the lack of cross-reactivity to MCM proteins (Additional file 1 Figure S1A). Intriguingly, by using different antibodies that recognize pan or individual MCM subunits, we discovered that distinct subsets of MCM proteins are associated with the anti-hSpt16p (8D2) or anti-SSRP1 (10D1) precipitates (as summarized in a chart in Figure 1A, bottom right panel). Whereas both MCM2 and MCM4 are commonly shared between these two types of complexes, differential coprecipitation of MCM3/5 (by 8D2) and MCM6/7 (by 10D1) was observed (Figure 1A). These results demonstrate that FACT may interact with discrete subcomplexes of the MCM proteins.
The primary biochemical activity of the MCM complex is the unwinding of the DNA strands [15, 19]. Thus, to examine whether the identified MCM-associated FACT complexes are catalytically active, we performed DNA helicase assay. Immunoprecipitates were incubated with labeled substrate and their abilities to displace annealed oligonucleotide were assayed. We found that, as compared to the control antibody, FACT complexes precipitated by either 8D2 or 10D1 antibody possessed DNA helicase activity (Figure 1B). This suggests that the two distinct FACT immunocomplexes, although differing in MCM constituents, are both competent in unwinding DNA in vitro. Furthermore, such catalysis is an ATP-dependent process, as no such activity was detected in the absence of ATP, nor in the presence of nonhydrolyzable form of ATP (Figure 1C).
To further characterize whether the DNA helicase activity displayed by the FACT immunocomplexes is mediated through the associated MCM complex, we isolated from MCM4RNAi cells (Additional file 1 Figure S1B) the 10D1 immunoprecipitates that are deficient in the MCM complex and found that the DNA helicase activity was greatly reduced (Figure 1D). This reduction was not observed in the 10D1 immunoprecipitates isolated from control or MCM3RNAi cells (Additional file 1 S1B) in which the FACT-MCM association is still present. This result serves as strong evidence that the helicase activity of the 10D1 immunocomplexes can be attributed primarily to the associated MCM complex, but not any non-specifically associated activities.
Next, we subjected HeLa nuclear extracts (Figure 1A) to gel filtration to further verify the presence of FACT-MCM complexes. As shown by Western blot, both FACT and MCM subunits have broad and overlapping distributions in fractions ranging in molecular size from 669 kDa to 2 MDa (Figure 2). To further distinguish the physical association between the FACT and MCM subcomplexes, we subjected gel filtration fractions to immunoprecipitation. Presence of the MCM subunits in the precipitates isolated by the two antibodies (8D2 and 10D1) from fractions 8-32 confirms the coelution of these two complexes and suggests that they combine to form complexes of various sizes (Figure 2). Furthermore, this immunoprecipitation assay approximately resolved three different peaks of MCM copurification (Figure 2, lower section). Interestingly, we could observe a shift in peak elution between peak 2 (10D1) and peak 3 (8D2), indicating a difference in the apparent sizes of FACT-MCM2/4/6/7 and FACT-MCM2/3/4/5. The MCM constituents of these immunocomplexes, as shown by Western blot, are consistent with the above immunoprecipitation experiments that detected two separate types of FACT-MCM association (Figure 1A). Together, these observations strongly corroborate the existence of two distinct FACT-MCM subcomplexes. The true identity of the immunocomplexes in peak 1 fractions cannot be deduced by our experiment. In summary, our biochemical analyses definitely characterize two native, enzymatically active FACT-MCM complexes that contain either the MCM2/4/6/7p or the MCM2/3/4/5p subassembly with the common partner FACT. Our results do not preclude, however, the existence of additional, higher molecular-weight FACT-MCM complexes (in the MDa range) that might include other interacting polypeptides or DNA.
Functional attributes of the two distinct FACT-MCM sub-complexes
Both FACT-MCM immunocomplexes are equally active in unwinding DNA template, indicating that their distinction may lie in other functional aspect. Thus, to further delineate and discriminate the functional attributes of the FACT-MCM complexes, we first sought to probe for the potential association of other pre-replication complex (pre-RC) factor. To this end, we performed co-immunoprecipitation experiments using the 8D2 and 10D1 antibodies. Endogenous ORC1 and Cdc6 were detected in the 8D2 immunocomplexes (Figure 3A). On the other hand, ORC was absent and the signal of Cdc6 was weak in the 10D1 immunoprecipitates. Such preferential association of the FACT-MCM2/3/4/5 complex with the origin-bound pre-RC may imply its role during pre-RC assembly. We also probed for the presence of Cdc45, which is a component of the large replisome progression complex (RPCs) that also includes MCMs and FACT [31]. Interestingly, we were able to detect this factor in both immunocomplexes (Figure 3A), suggesting that identified FACT complexes may be associated with active replication.
To further evaluate the replicative role of these subcomplexes, we examined their origin association. In our previous work, we discovered that the 10D1-targeted FACT complex is present in vivo at a region of known replication origin, namely the replicator associated with the human lamin B2 gene. Using the ChIP assay, chromatin prepared from cells synchronized at different cell cycle stages was precipitated with the 8D2 or 10D1 antibody (Figure 3B). PCR reactions using specific sets of primers were subsequently performed to monitor the existence of the lamin B2 origin sequence (Figure 3B, lanes 1-9). As shown in Figure 3B, at equal loads of chromatin preparations, we observed specific occupancy of 8D2-targeted FACT complex in the ori region in asynchronously growing cells (compare lanes 4 and 7). As a control, sequence of a distant non-transcribed region (Figure 3B, lanes 10-12, see Methods) was not enriched in the immunoprecipitates. Furthermore, no origin binding of either immunocomplexes could be detected in mitotic cells (lanes 3 and 6, and ref. 28). This result is reminiscent of the observed dissociation of FACT from condensed chromatin during mitosis [32]. Interestingly, while the 10D1-complex exhibited an enrichment of such origin binding during the transition from G1 to S phase (compare lanes 1 and 2; Figure 3C, histogram on the right), the 8D2-associated complex seemed to display a relatively greater degree of association in G1 as compared to the point entering S phase (lanes 4 and 5; Figure 3C, histogram on the left). These results reflect the activity of both FACT-MCM complexes locally at origins or replication forks, and imply their differential origin association.
Next, to further delineate the differential origin binding modes of the two FACT-MCM immunocomplexes, we performed additional ChIP assays on MCM3RNAi or MCM4RNAi cells. Chromatin preparations from these two cell lines were subjected to ChIP using either the 8D2 or 10D1 antibody and probed for the presence of the lamin B2 origin sequence (Figure 3D). In the context of downregulated MCM3 or MCM4 protein level, association of the 8D2 immunocomplex with the origin was disrupted (Figure 3D, middle panel). This signals that the origin binding of FACT (and the FACT-MCM2/3/4/5) lies, at least in part, in the presence of an intact MCM2/3/4/5 complex. Furthermore, in the MCM4RNAi cells, we saw a similar reduction of the origin fragment in the 10D1 immunocomplex, demonstrating the importance of MCM4 in this functional regard (top panel). Intriguingly, knockdown of the MCM3 subunit unexpectedly led to a weakened association of FACT-MCM2/4/6/7 with lamin B2 origin. A likely explanation for this phenomenon is that the origin association of the MCM3-associated complexes (including FACT-MCM2/3/4/5) may be preceding and required for the subsequent recruitment of FACT-MCM2/4/6/7. Therefore, there may exist two different temporal modes through which the FACT-MCM complexes are recruited to the replicator (see Discussion). Collectively, results presented in Figure 3 suggest that these two FACT-MCM complexes may operate at distinct stages during the initiation phase of DNA replication.
Functional regulation of the FACT-MCM complex mirrors cell cycle progression
While probing for the presence of MCM4 in the anti-FACT immunocomplexes, we observed smear of immunoreactive signals representative of multiple lower-mobility protein forms (Figures 1A & 4A). Changes in electrophoretic mobility of MCM4 may potentially be induced by phosphorylation, as demonstrated previously [11]. Indeed, treatment of the immunoprecipitates with alkaline phosphatase eliminated the higher protein forms (Figure 4A, compare lanes 5 and 6). However, no phophatase-induced electrophoretic mobility change was seen in the other MCM subunits, implying their lack of modification by phosphate-moiety (Figure 4A, lanes 1-4). Importantly, other groups have reported that, in logarithmically growing or S phase HeLa cells, lower mobility form of MCM4 could be detected largely in the chromatin fraction but not in the Triton-soluble fraction [13, 14]. Our results therefore suggest that the isolated FACT-MCM complex is likely to exist endogenously in a chromatin-bound form.
Based on the previous observation that mouse MCM4 undergoes mitotic hyper-phosphorylation and functional inactivation [11], we wanted to examine whether the FACT-associated MCM4 (and MCM complex) is similarly modulated in a cell cycle-dependent manner (Additional file 1 Figure S1C). First, at the post-translational modification level, lower-mobility forms of FACT-bound MCM4 became increasingly visible starting in G2 phase and peaking in M phase, during which the unmodified form was only marginally present (Figure 4B, middle two panels). While the degree of phosphorylation of MCM4 altered, the profiles of the FACT dimers and the associated MCM subunits remain relatively invariable across cell cycle (Figure 4B, upper two panels). Second, to address whether this alteration in modification status of MCM4 reflects cell cycle-dependent functional changes for the FACT-MCM complex, we performed DNA helicase assay on immunoprecipitates isolated from different stages (Figure 5A). We found that, at the equivalent levels of co-precipitated MCMs (Additional file Figure S1D), the DNA unwinding activity of either the 8D2 or 10D1 immunocomplex peaked at the G1/S junction (Figure 5A). However, both FACT-MCM complexes exhibited significantly lower helicase activity during G2 and mitosis, similarly to the cell cycle regulation of the core MCM helicase activity [11]. Taken together, our observations are in agreement with the previous results on mouse MCM4. Therefore, we conclude based on these data that the association of FACT with the MCM complex preserves MCM helicase activity and its temporal regulation during cell cycle progression.
Further evidence that links the possible role of FACT-MCM to S phase was obtained from immunofluorescence analysis. While examining the subcellular localizations of these two factors at different cell cycle stages, we found a discernible degree of colocalization between FACT and MCM4 at G1/S junction as well as throughout S phase (Figure 5B). However, this colocalization was markedly reduced in early G1 (Figure 5B) or G2/M cells (data not shown). Such manner of interaction between complexes may contribute to their coordinated functions during S phase.
Cell proliferation underlies the interaction between FACT and MCM
Near lack of overlapping signals between FACT and MCM at the G0/early G1 stage of cell cycle led to the speculation that the FACT-MCM interaction may be positively regulated by cell proliferation. To entertain this possibility, we examined the FACT-MCM complex formation in quiescent cells. Interestingly, consistent with the localization data, we detected less MCM proteins (~45% of control) in the FACT immunoprecipitates in K562 cells upon entry into a differentiation/G0-like state (induced by treatment with sodium butyrate). This suggests that the formation/disruption of the FACT-MCM complex is correlated with cell proliferation (Figure 6).