hMCPH1 inhibits chromosomal binding of condensin II in Xenopus egg extracts
To dissect the molecular action of hMCPH1, we sought to develop a cell-free assay by using extracts prepared from unfertilized Xenopus
eggs. The use of these egg extracts has two big advantages. First, the cell cycle of the unfertilized eggs is naturally arrested at metaphase of meiosis II, at which the activity of cyclin B–Cdk1 is kept high (Morgan, 2007
). Second, the dynamics and function of many chromosomal proteins, including condensin I and II, have been characterized extensively in these extracts (e.g., Hirano et al., 1997
; Ono et al., 2003
). These properties allow us to study the specific role of hMCPH1 in condensin regulation, if any, without being disturbed by other parameters, such as cell cycle progression.
We started with a very simple experiment. Full-length hMCPH1 was produced in a reticulocyte lysate and preincubated with a metaphase extract for 30 min. Sperm chromatin was added into the mixture, and after incubation for another 120 min, chromosome fractions were isolated and analyzed by immunoblotting. Strikingly, we found that exogenously added hMCPH1 effectively inhibited loading of condensin II onto chromosomes. The effect of hMCPH1 was highly specific to condensin II: loading of condensin I and topoisomerase IIα was hardly affected in the same reaction (, lanes 7 and 8). We then tested whether two point mutations that had been identified in MCPH1 patients (T27R [Trimborn et al., 2005
] and W75R [Gavvovidis et al., 2010
]) might affect the activity of hMCPH1 to inhibit condensin II in this cell-free assay. Remarkably, again, both of the mutant proteins barely blocked loading of condensin II (, lanes 9 and 10), suggesting that this cell-free assay could indeed recapitulate an aspect of the physiological functions of hMCPH1.
The biochemical data were fully supported by morphological analysis. In this experiment, metaphase chromosomes were assembled in the presence or absence of hMCPH1, fixed, and stained simultaneously with antibodies against condensin I– and condensin II–specific subunits (XCAP-G and -H2, respectively). In a mock-treated extract, both condensin I and II were readily detectable along the entire length of chromatids (, a–d) as had been reported previously (Ono et al., 2003
). In an hMCPH1-treated extract, however, the signal for condensin II was lost almost completely, whereas the signal for condensin I was unaffected (, e–h). We also noticed that the axes of chromosomes assembled in the presence of hMCPH1 were structurally distorted (, g′); this zigzag morphology was reminiscent of that observed in chromosomes assembled in an extract depleted of condensin II (Ono et al., 2003
). As expected, extracts supplemented with the two mutant forms of hMCPH1 produced chromosomes that were indistinguishable from those assembled in the mock-treated extract (, i–p).
We then tested whether hMCPH1 might have an ability to strip condensin II from chromosomes that had already been assembled in the extracts. To this end, metaphase chromosomes were assembled in a control extract, and hMCPH1 was added and incubated for another 60 min. Immunoblotting analysis of chromosome fractions revealed a significant reduction of the level of condensin II but not of condensin I (). As judged by morphological analysis, again, hMCPH1-treated chromosomes displayed a zigzag and fragile appearance (). Taking these results together, we conclude that exogenously added hMCPH1 blocks the action of condensin II present in the Xenopus egg extracts in a highly specific manner.
An N-terminal domain of hMCPH1 is sufficient to specifically inhibit condensin II in the cell-free assay
Next, we asked which domains of hMCPH1 might be responsible for the activity to inhibit condensin II observed in the cell-free assay. Selected examples of deletion constructs of hMCPH1 tested are shown in . In short, we found that an N-terminal fragment (amino acids 1–195) containing the first BRCT domain (BRCT1) is required and sufficient to inhibit loading of condensin II onto chromosomes (, lanes 8 and 9). This N-terminal fragment itself bound to chromosomes, but it does so only weakly when compared with a C-terminal fragment (amino acids 638–835) containing the second and third BRCT domains (, lane 10). Introduction of the T27R mutation into the N-terminal domain (amino acids 1–195) not only abolished its condensin II inhibitory activity () but also weakened its own chromosome-binding activity (, lane 6). None of the shorter N-terminal constructs tested (amino acids 1–95, 1–138, 1–162, and 1–183) inhibited condensin II loading in this assay.
Figure 2. An N-terminal domain of hMCPH1 is sufficient to specifically inhibit chromosomal loading of condensin II in Xenopus egg extracts. (A) Deletion constructs of hMCPH1 used in this paper. (B) Full-length hMCPH1 and three different deletion constructs were (more ...)
A recent study had reported that okadaic acid, a potent inhibitor of PP2A (protein phosphatase 2A), inhibits chromosomal loading of condensin II in Xenopus
egg extracts and in human cells (Takemoto et al., 2009
). Therefore, we wished to compare the impacts of hMCPH1 and okadaic acid in parallel in our cell-free assay. We found that okadaic acid blocked loading of condensin II, XCAP-D (KIF4a), and aurora B but not of condensin I (, lanes 11 and 12) as had been reported by Takemoto et al. (2009)
. In contrast, hMCPH1 hardly affected loading of XCAP-D and aurora B (, lanes 8 and 9), indicating that hMCPH1’s ability to inhibit chromosomal loading of condensin II in Xenopus
egg extracts is far more specific than okadaic acid.
The N-terminal domain of hMCPH1 competes for chromosome binding of condensin II in Xenopus egg extracts
To understand the mechanism by which hMCPH1 inhibits the action of condensin II, we wished to test the possibility that the N-terminal domain of hMCPH1 might compete for binding sites of condensin II on chromosomes. Consistent with this idea, the T27R mutant not only poorly inhibited condensin II but also poorly associated with chromosomes as judged by immunoblotting of chromosomal factions (). In experiments shown in , we attempted double immunofluorescent labeling of condensin II and the N-terminal domain of hMCPH1 (a GFP-tagged version) on chromosomes. As expected, GFP alone or the T27R mutant of GFP-hMCPH1 (amino acids 1–195) produced only a background level of its own signals on chromosomes and did not affect loading of condensin II (, a–d and m–p). On the other hand, GFP-hMCPH1 (amino acids 1–195) associated with chromosomes and inhibited loading of condensin II in a dose-dependent manner (, e–h and i–l). Remarkably, we noticed that condensin II and GFP-hMCPH1 (amino acids 1–195) apparently distributed along chromosomes in a mutually exclusive manner (, i′–l′). Under the same condition, we found virtually no sign for displacement of condensin I by GFP-hMCPH1 (amino acids 1–195) exogenously added into the extract (). These results suggest that the N-terminal domain of hMCPH1 inhibits loading of condensin II by competing for its binding sites on chromosomes.
Figure 3. The N-terminal domain of hMCPH1 competes for chromosome binding of condensin II in Xenopus egg extracts. (A) A reticulocyte lysate containing GFP or GFP-hMCPH1 (amino acids 1–195; wild type and T27R) was mixed with 10 vol metaphase egg extracts (more ...)
Characterization of Xenopus MCPH1 (xMCPH1) and mouse MCPH1 (mMCPH1) in the cell-free assay
The specific inhibition of condensin II by exogenously added hMCPH1 in the cell-free assay raises the question of whether Xenopus
egg extracts might contain endogenous MCPH1. To explore this possibility, we prepared specific antibodies against recombinant fragments of xMCPH1 and found that the extracts indeed contained an appreciable level of endogenous xMCPH1, which was estimated to be comparable with the standard dose of hMCPH1 added exogenously (Fig. S2
). We reasoned that, if xMCPH1 also had an activity to inhibit condensin II, depletion of endogenous xMCPH1 from an extract might cause hyperactivation and overloading of condensin II in the cell-free assay. It was found, however, that the presence or absence of endogenous xMCPH1 made little, if any, difference in the morphology of chromosomes or the level of condensin II loaded on them ().
Figure 4. Characterization of xMCPH1 and mMCPH1 in the cell-free assay. (A) Sperm chromatin was added to metaphase egg extracts that had been depleted with control IgG (Δmock; lanes 1 and 3) or anti-xMCPH1 (ΔxMCPH1; lanes 2 and 4). After incubation (more ...)
MCPH1 is a rapidly evolving protein and is implicated in the expansion of brain size during vertebrate evolution (Ponting and Jackson, 2005
). It would therefore be of considerable interest if MCPH1 proteins from different vertebrate species were subjected to a comparative study in this particular cell-free assay. To this end, we added the same dose of full-length MCPH1 from human, mouse, and Xenopus
into a control extract and an extract depleted of endogenous xMCPH1 (Fig. S2). Intriguingly, we found that, compared with hMCPH1, mMCPH1 or xMCPH1 displayed a much weaker activity to inhibit chromosomal loading of condensin II either in the presence (, rows a–d) or absence (, rows e–h) of endogenous xMCPH1. Similar results were obtained when the N-terminal domains of hMCPH1, mMCPH1, and xMCPH1 were used. In the experiment shown in , the dose of the N-terminal domains of MCPH1 added into metaphase extracts was titrated down starting from the standard dose. The N-terminal domain of hMCPH1 displayed a dose-dependent inhibition of condensin II loading (, lanes 15–18), and substantial inhibition was detectable with as low as 25% of the standard dose (, lane 16). In contrast, a very low level of inhibition was observed even with the highest dose of the mouse or Xenopus
counterpart (, lane 22 or lane 26). Although the observation that xMCPH1 has a poor activity to inhibit condensin II in the cell-free assay was somewhat puzzling, it provides a reasonable explanation for why depletion of endogenous xMCPH1 from the extract had little impact on the behavior of condensin II ().
mMCPH1 can be “humanized” by specific amino acid substitutions
To substantiate the findings described in the previous section, we sought to identify the determinants that would differentiate the activities between hMCPH1 and mMCPH1. Because the N-terminal domains of hMCPH1 and mMCPH1 were relatively divergent in their last quarters, each domain was divided into two subdomains, A (Ah
, amino acids 1–145 in hMCPH1; Am
, amino acids 1–151 in mMCPH1) and B (Bh
, amino acids 146–195 in hMCPH1; Bm
, amino acids 152–194 in mMCPH1), and a chimera protein, referred to as Am
, was created ( and Fig. S1). We found that Am
bound to chromosomes slightly better than the authentic mouse sequence (Am
) but failed to acquire an appreciable level of activity to inhibit condensin II loading (, lane 16). Because the simple substitution of the subdomain B was not enough to confer the hMCPH1-like activity on mMCPH1, we took a brute force mutagenesis approach by focusing on nonconserved residues in the subdomain A. A list of substituted residues (singly or in combinations) is shown in Fig. S3 A
. In short, we found that mMCPH1 acquired an activity comparable with that of hMCPH1 when two mouse to human substitutions, T72R and E142K, were combined with the subdomain B substitution (, compare lanes 12 and 13). A double substitution alone (T72R-E142K) displayed little, if any, effect (, lane 17), indicating that successful conversion of the mouse sequence into one with a humanlike activity requires the combination of all three substitutions. Conversely, when two human to mouse substitutions were introduced into the corresponding residues of hMCPH1 (i.e., R66T-K136E), the activity associated with the human protein was greatly reduced (). Thus, the cell-free assay successfully allowed us to identify a subset of specific amino acid substitutions that contributes to the functional difference between hMCPH1 and mMCPH1.
Figure 5. mMCPH1 can be converted into a humanlike, competent form by specific amino acid substitutions. (A) Schematic representation of the constructs used in B. The N-terminal domain of hMCPH1 was divided into two subdomains, Ah (amino acids 1–145) and (more ...)
Robust interaction with condensin II through its central domain is not required for hMCPH1’s inhibitory activity in the cell-free assay
We then tested for physical interactions between hMCPH1 and condensin II. In HeLa cell nuclear extract, endogenous hMCPH1 and human condensin II specifically interacted with each other (Fig. S4, A and B
). By extending the previous work by Wood et al. (2008)
, we found that a central domain of hMCPH1 (amino acids 381–435) was primarily responsible for the condensin II–hMCPH1 interaction (Fig. S4, C and D) and further identified a pair of point mutations (Y412A and F416A) in the central domain that disrupts this interaction (Fig. S4 E). Consistently, full-length hMCPH1 added into Xenopus
extracts interacted with endogenous condensin II, and such an interaction was largely abolished when the F416A mutation was introduced into hMCPH1 (). The F416A mutant protein, however, was found to retain a full activity to inhibit condensin II loading in the cell-free assay (). This observation was surprising at first glance, yet it was consistent with our result that the N-terminal region (amino acids 1–195) of hMCPH1 is sufficient for condensin II inhibition ().
Figure 6. hMCPH1 physically interacts with condensin II through its N-terminal and central domains. (A) A reticulocyte lysate containing no hMCPH1 (mock) or FLAG-tagged hMCPH1 was mixed with 10 vol metaphase egg extracts and incubated for 60 min. Anti-FLAG beads (more ...)
We then sought the possibility that the N-terminal domain of hMCPH1 might possess a minor condensin II–interacting activity independently of its central region. We also wished to identify which subunits of condensin II might be responsible for interactions with hMCPH1. To test this, the three regulatory subunits of human condensin II (CAP-D3, -G2, and -H2) were individually translated in vitro, unmixed or mixed in all possible pairwise combinations, and processed for subunit–subunit interaction assays with full-length or truncated versions of hMCPH1 (). In short, we found that the central domain (amino acids 381–435) of hMCPH1 coprecipitated predominantly with CAP-G2, whereas the N-terminal domain (amino acids 1–195) did primarily with the D3 subunit (Fig. S4 F). Together with our previous data showing no direct interaction between CAP-G2 and -D3 subunits (Onn et al., 2007
), we conclude that two separate domains of hMCPH1 are capable of interacting with different HEAT subunits of condensin II (, cartoon). It should be noted, however, that the interaction between the N-terminal domain of hMCPH1 and CAP-D3 is rather cryptic and not readily detectable in the context of the condensin II holocomplex ( and Fig. S4 D).
Complementation assay reveals a contribution of the central domain of hMCPH1 to shaping metaphase chromosomes
Finally, we wished to understand to what extent the information obtained from the cell-free assay might be relevant to the in vivo function of hMCPH1. To this end, we set up a complementation assay in which various constructs of hMCPH1 were expressed in MCPH1 patient cells bearing a homozygous truncating mutation by means of a lentivirus vector. In the first set of experiments, GFP-tagged versions of full-length hMCPH1 were introduced into the patient cells, and their expression was confirmed by immunoblotting against total cell lysates (). The frequency of prophaselike cells (PLCs) in each cell population was scored (; also see Materials and methods). We found that wild-type hMCPH1 rescued the PCC phenotype very efficiently ( and see Fig. S5 [A and B]
for representative images). On the other hand, an hMCPH1 construct with the mild allele T27R failed to fully restore the defective phenotype (Trimborn et al., 2005
). Notably, a construct with the F416A mutation displayed a full activity to restore the PCC phenotype, suggesting that the interaction with condensin II mediated through the central domain of hMCPH1 is dispensable for rescuing the PCC phenotype.
Figure 7. The N-terminal domain of hMCPH1 is sufficient to rescue the PCC phenotype, whereas its central domain is additionally required for shaping metaphase chromosomes in patient cells. (A) MCPH1 patient cells were transduced with GFP-tagged full-length hMCPH1 (more ...)
To study whether hMCPH1 might impact the morphology of metaphase chromosomes, chromosome spreads were prepared and stained with an antibody against Smc2, a core subunit shared by both condensin I and II. In MCPH1 patient cells (either mock treated or transduced with GFP alone), most of metaphase chromosomes displayed an abnormal morphology with very short and thick chromatids. When these so-called “dumpy” chromosomes were stained with anti-Smc2, chromatid axes with a characteristic wavy appearance could be visualized (, a and graph, bars 1 and 2). When the patient cells expressing GFP-hMCPH1 were analyzed, the dumpy phenotype was barely observed; ~80% of chromosomes in the sample displayed a more typical morphology with long, thin, and straight chromatids (, c and graph, bar 3). As expected, expression of the T27R mutant protein failed to rescue the dumpy phenotype effectively (, bar 4). In contrast, somewhat unexpected was the observation that the F416A protein rescued the phenotype only partially (, bar 5).
We then asked whether expression of the N-terminal domain (amino acids 1–195) of hMCPH1 was sufficient to rescue the PCC and/or dumpy phenotypes (). The wild-type N-terminal domain efficiently rescued the PCC phenotype, whereas the corresponding domain containing the T27R mutation displayed a partial rescue (), an observation similar to that made with the full-length proteins (). Two additional constructs (amino acids 196–835 and 638–835) lacking the N-terminal domain failed to rescue the PCC phenotype (Fig. S5, C–E). We noticed, however, that expression of the N-terminal domain alone was not sufficient to fully rescue the dumpy phenotype in patient cells (). Collectively, the current results suggest that the N-terminal domain of hMCPH1 is sufficient to rescue the PCC phenotype but that both the N-terminal and central domains are required for properly shaping metaphase chromosomes.