Evolutionary diversification of MCM3 genes in Xenopus laevis and Danio rerio

Abstract

Embryonic cell cycles of amphibians are rapid and lack zygotic transcription and checkpoint control. At the mid-blastula transition, zygotic transcription is initiated and cell divisions become asynchronous. Several cell cycle-related amphibian genes retain 2 distinct forms, maternal and zygotic, but little is known about the functional differences between these 2 forms of proteins. The minichromosome maintenance (MCM) 2–7 complex, consisting of 6 MCM proteins, plays a central role in the regulation of eukaryotic DNA replication. Almost all eukaryotes retain just a single MCM gene for each subunit. Here we report that Xenopus and zebrafish have 2 copies of MCM3 genes, one of which shows a maternal and the other a zygotic expression pattern. Phylogenetic analysis shows that the Xenopus and zebrafish zygotic MCM3 genes are more similar to their mammalian MCM3 ortholog, suggesting that maternal MCM3 was lost during evolution in most vertebrate lineages. Maternal MCM3 proteins in these 2 species are functionally different from zygotic MCM3 proteins because zygotic, but not maternal, MCM3 possesses an active nuclear localization signal in its C-terminal region, such as mammalian MCM3 orthologs do. mRNA injection experiments in zebrafish embryos show that overexpression of maternal MCM3 impairs proliferation and causes developmental defects, whereas zygotic MCM3 has a much weaker effect. This difference is brought about by the difference in their C-terminal regions, which contain putative nuclear localization signals; swapping the C-terminal region between maternal and zygotic genes diminishes the developmental defects. This study suggests that evolutionary diversification has occurred in MCM3 genes, leading to distinct functions, possibly as an adaption to the rapid DNA replication required for early development of Xenopus and zebrafish.

Abbreviations

MCM

minichromosome maintenance

NLS

nuclear localization signal

GST

glutathione S-transferase

EGFP

enhanced green fluorescent protein

BrdU

bromodeoxyuridine

c/e defect

convergent extension defect

Introduction

All organisms must duplicate their genomic DNA thoroughly in the cell cycle. Partitioning of the eukaryotic genome through multiple replication origins allows rapid duplication of a large genome within a single cell cycle. At the same time, specialized mechanisms prevent any replication origins from firing more than once in a single cell cycle. Recent studies have clarified that regulation of the minichromosome maintenance (MCM) complex (an MCM2–7 heterohexamer, consisting of 6 subunits, MCM2 to MCM7) is central for the regulation of origin firing.¹ The MCM complex is loaded onto the replication origins by the loading factors ORC, Cdc6 and Cdct1 at the transition of M phase to G1 phase. The loading of the MCM complex initiates the formation of the pre-replicative complex (pre-RC), and by this means, replication origins are “licensed” to fire. Upon the loading of the pre-RC, MCM2–7 forms a double hexamer that encircles dsDNA, which is inactive as a helicase.^2-4 MCM2–7 helicase function is activated by association of the activation factors Cdc45 and GINS, under the control of S-phase cyclin-dependent kinase and Dbf4-dependent kinase (DDK) activity at the G1/S transition. At this transition, the MCM double hexamer is converted into 2 CMG (Cdc45-MCM-GINS) complexes, which function as active replicative helicases.^5,6 Multiple mechanisms prevent MCM reloading at the beginning of S phase until the end of M phase, thus preventing origins from firing multiple times within a single cell cycle.⁷

Origin licensing during G1 and activation of the pre-RC at the onset of S phase is thought to be conserved in all eukaryotes. However, in multicellular organisms, replication programs differ from cell type to cell type and are regulated during development. For example, the first 12 cell cycles in Xenopus laevis (called embryonic cell cycles) are significantly different from somatic cycles.⁸ In embryonic cell cycles, S and M phase alternate in very short periods (less than 30 min). During this period, virtually no S-M checkpoint function is operating, and so mitosis could occur without completion of DNA replication. In addition, little zygotic transcription occurs and the cell cycle is promoted only by the translation of maternal mRNAs. After the first 12 cell cycles, at mid-blastula transition (MBT), the embryonic cell cycle is converted into a somatic cell cycle at which point zygotic transcription starts. The cell cycle duration increases and the synchronicity of the cell cycles of different cells is lost.^9,10 This cell cycle transition is not only observed in Xenopus, but also in other species, such as Drosophila and zebrafish.^11,12 In order to achieve a rapid embryonic cell cycle, DNA replication needs to be fast. Previous studies showed that embryonic replication initiates at intervals of about 10 kb with no sequence specificity, thus suggesting that the number of origins is far greater than the number of origins in somatic cell cycles.¹³ Furthermore, a recent study suggests that limitation of the number of activators of the pre-RC plays a key role in stopping the embryonic cell cycle and entering MBT.¹⁴ This implies that quantitative differences in replication proteins contribute to the different replication programs found in embryonic vs. somatic cell cycles. But it remains to be clarified whether qualitative differences of replication proteins also contribute to the program. For example, it has been reported that Xenopus laevis has 2 forms of MCM6 genes, which are differently expressed during development.¹⁵ In addition, A-type cyclin (cyclin A1 and A2)¹⁶ and subunits of DDK (Drf1 and Dbf4),¹⁷ are both required for the activation of pre-RC and are reported to be differently expressed during development. It is not yet known, however, whether these genes are indeed functionally different, or whether their transitions in expression during development contribute to different cell cycle regulation in embryonic and somatic cell cycles.

Here we report that the 2 forms of MCM3 in Xenopus and zebrafish are differently expressed during development and retain different biological activities in nuclear transport. More intriguingly, the overexpression of the maternal form of MCM3 leads to compromised DNA replication and developmental defects in zebrafish. A possible implication of our results is that maternal MCM3 has adapted to contribute to the replication of relatively large genomes during rapid embryonic cell cycles in these species.

Results

Identification of maternal and zygotic MCM3 and MCM6 in Xenopus and zebrafish

We previously cloned all 6 Xenopus MCM genes from an oocyte cDNA library and raised antibodies against recombinant proteins derived from them.¹⁸ Expression of these genes both in embryonic (oocyte, egg) and somatic cells (Xenopus A6 cells; cloned epithelial cells from kidney) was examined either by immunoblotting or Northern blotting (Supplementary Fig. 1). XMCM2, XMCM4, XMCM5 and XMCM7 were expressed both in embryonic and somatic cells, whereas XMCM3 and XMCM6 were expressed in embryonic cells, but not in somatic cells, which is consistent with a previous report indicating the presence of somatic cycle-specific transcripts of XMCM3 and XMCM6.¹⁵ We searched and isolated the homologous genes of XMCM3 and XMCM6 cDNAs from somatic cells by screening a Xenopus neurula stage cDNA library. The cDNAs encode proteins that show sequence similarity to the previously reported Xenopus MCM3 or MCM6 isolated from the oocyte library, but with differences in their C-terminal regions, in particular in the putative nuclear localization signal (NLS) sequence of MCM3 (Fig. 1A). As for MCM6, the sequence is essentially identical to the previously reported zygotic MCM6.¹⁵ Here, we refer to the genes isolated from the oocyte cDNA library as XMCM3m and XMCM6m (m for maternal) and those from the neurula cDNA library as XMCM3z and XMCM6z (z for zygotic). For both MCM3 and MCM6, maternal and zygotic forms show about 70% amino acid identity, and zygotic forms are more similar to their human orthologs (Fig. 1B). We have cloned a fragment of Xenopus MCM3m with several nucleotide substitutions (data not shown), which is possibly due to the pseudo-tetraploidy of Xenopus laevis. We could identify similar but unique maternal and zygotic MCM3 and MCM6 in the Xenopus tropicalis genome. These results strongly suggest that 2 different forms of MCM3 and MCM6 were expressed during the development of Xenopus embryos.

Figure 1.

Two types of MCM3 genes in Xenopus and zebrafish. (A) Schematic structures of maternal and zygotic MCM3 proteins. (B) Comparison of primary structures of various MCM3 proteins. (C) Expression of Xenopus MCM genes during early development. cDNAs were prepared from oocytes, A6 cells or developing embryos, and expression of Xenopus MCM genes was examined by RT-PCR using gene specific primers. (D) Expression of zebrafish MCM genes during early development. cDNA was prepared from developing embryos, and expression of MCM genes was examined by RT-PCR using gene-specific primers. (E and F) Expression of MCM genes examined by whole mount in situ hybridization. Embryos were fixed at the indicated time points and stained with labeled RNA probe. A sense strand probe was used as a negative control.

In order to know whether other species also have 2 distinct MCM3 and MCM6 paralogs, we searched genome sequences of several model organisms (green anole, coelacanth, medaka, zebrafish, fugu, cod, lamprey, sea squirt). We found that only zebrafish, Danio rerio, retains orthologs to both maternal and zygotic XMCM3 and MCM6 in its genome. Putative maternal and zygotic zebrafish MCM3- and MCM6-like proteins (DrMCM3m and DrMCM3z, DrMCM6m and DrMCM3z, respectively) show similar features to Xenopus counterparts in their amino acid sequences (Fig. 1B), whereas sea squirt and other organisms, including other animals (mammals, birds, flies) and plants, have only single MCM3 and MCM6 genes that are more similar to the Xenopus zygotic forms. So far, we have found only one fragment of newt MCM3 cDNA that encodes protein with significant similarity to the C-terminal regions of Xenopus MCM3m by searching the public sequence databases.¹⁹ These results show that the maternal-specific genes of MCM3 and MCM6 have been retained in only a small fraction of species.

Expression of maternal and zygotic MCM3 and MCM6 genes during development

Developmentally regulated expression of XMCM6m and XMCM6z mRNA has been reported previously.¹⁵ We have examined the expression patterns of XMCM3m and XMCM3z mRNA together with those of XMCM6m and XMCM6z of developing Xenopus embryos by using RT-PCR with gene-specific primers (Fig. 1C). In concordance with a previous report,¹⁵ the expression of XMCM6m slowly declines at later stages of development, whereas that of XMCM6z gradually increases and becomes predominant in A6 cells. Expression of XMCM3 changes in a similar but more prominent way during development. XMCM3m is highly expressed in oocytes, and its expression declines upon fertilization even more conspicuously than that of XMCM6m. XMCM3z is not detectable in oocytes and early blastula, but expression abruptly starts after the MBT, at which time robust zygotic expression initiates. In contrast, XMCM2 is expressed at constant levels during development. With its complete disappearance at 24 hours postfertilization (hpf) and late onset at 8 hpf, XMCM3 shows more prominent regulation than XMCM6 does.

By examining the expression of zebrafish MCM3 and MCM6 during development, we found that the expression patterns of the maternal and zygotic forms of MCM3 and MCM6 show similar features to those of Xenopus counterparts during development (Fig. 1D). In addition, here the shift of expression from the maternal to the zygotic form is more prominent in MCM3 than in MCM6. Again the expression of MCM2 does not change considerably. We were able to confirm the similarity of expression patterns by whole-mount in situ hybridization (WISH) (Fig. 1E). These results support the idea that maternal and zygotic MCM genes found in Xenopus and zebrafish are indeed genuine orthologs. It should be noted that DrMCM3z and DrMCM6z show almost identical expression patterns at later stages, which closely correspond to actively proliferating cells (Fig. 1F and data not shown).

Characterization of the NLS-like motifs found in maternal and zygotic MCM3

Vertebrate MCM3 proteins contain evolutionary conserved NLS-like motifs in their C-terminal region.²⁰ The NLS of metazoan MCM3 consists of 2 stretches of basic amino acids separated by a linker region predominantly containing acidic amino acids (cf. Fig. 2A). NLS-like motifs of zygotic MCM3 retain these features (Fig. 2A). However, basic amino acid stretches in NLS-like motifs of maternal MCM3 are interrupted by other amino acids, and the acidic amino acids in the linker regions are not as prominent as that in zygotic MCM3. We have previously reported that glutathione S-transferase (GST)-fused recombinant Xenopus MCM3 (maternal form) fails to be transported into nuclei formed in Xenopus egg extracts.²¹ To compare the NLS activity of maternal and zygotic forms of MCM3, we examined the nuclear transport of GST-enhanced green fluorescent protein (EGFP) fused recombinant proteins into nuclei formed in the extracts (Fig. 2B). GST-EGFP appears to be excluded from the nuclei and localizes in the cytoplasm, thus suggesting that the molecular size of GST-EGFP is large enough to prevent its entry into nuclei by simple diffusion. When the C-terminal region of XMCM3z was fused to GST-EGFP, the recombinant protein was found to be concentrated in the nuclei, indicating that XMCM3z possesses NLS activity, as in other metazoan MCM3. However, when the C-terminal region of XMCM3m was fused to GST-EGFP, the recombinant protein failed to accumulate in the nucleus and most of the protein localized in the cytoplasm, which is consistent with our previous report.²¹ We obtained similar results by using GST-EGFP recombinant proteins containing only NLS-like motifs of XMCM3, instead of C-terminal regions (Fig. 2B). These results clearly indicate that zygotic but not maternal XMCM3 contains a functional NLS.

Figure 2.

Nuclear transport activities of maternal and zygotic MCM3. (A) Comparison of nuclear localization signal-like sequences of various MCM3 proteins. Basic amino acids and acidic amino acids are shown in red and blue, respectively. Corresponding amino acid numbers are shown for Xenopus laevis and zebrafish proteins. (B) Nuclear transport activities of maternal and zygotic Xenopus MCM3. Recombinant GST-EGFP-MCM3 (either C-terminal region or only NLS-like sequence) fusion proteins were purified from E. coli. The recombinant proteins were added to Xenopus egg extracts after the formation of nuclei from added sperm chromatin, and the localization of the recombinant proteins was observed by fluorescent microscopy. Scale bar 20 μm. (C and E) Interaction of zebrafish importin α with DrMCM3s. An in vitro pull-down assay was performed using recombinant His-tagged zebrafish importin α and recombinant GST-EGFP-MCM3 (either C-terminal region fusion proteins [C] or full-length and maternal/zygotic chimera proteins [E]). GSH agarose-bound proteins were visualized by either anti-His or anti-GST antibodies. (D) Schematic diagram of maternal and zygotic proteins and their chimera proteins. Mz chimera 1, 2 and 3 consist of N-terminal maternal and C-terminal proteins, but transition points are different. Zygotic parts start at 659 a.a. (before NLS), 701 a.a. (after NLS) and 742 a.a. for mz1, mz2 and mz3, respectively.

We next approached the question of whether or not the nuclear transport activities of 2 forms of zebrafish MCM3 differ in the same way as they do for Xenopus MCM3 genes. We first examined the nuclear transport activity of recombinant proteins of maternal and zygotic MCM3 in Xenopus egg extracts. We found that both proteins were transported into the nuclei (data not shown). Since we could not exclude the possibility that these proteins retain different nuclear transport activities in zebrafish embryos, we next examined the interaction of zebrafish MCM3 with its adaptor protein, zebrafish importin α. To examine the physical interaction between DrMCM3 NLS and importin α, we purified recombinant proteins from Escherichia coli, and used them for an in vitro pull-down assay (Fig. 2C and E). Among the zebrafish importin α family proteins, we chose zebrafish importin α2 because it shows the highest homology to Xenopus importin α1, which is the major importin in Xenopus eggs. In an in vitro pull-down assay, using either C-terminal or full-length DrMCM3 recombinant proteins, we found that zebrafish importin α preferentially binds to zygotic DrMCM3 (Fig. 2C and E). This interaction depends on the NLS-like motif of DrMCM3z because chimeric proteins of N-terminal 3-fourths maternal and C-terminal one-fourth zygotic MCM3 retain efficient binding to importin α2, but the substitution of a zygotic NLS-like motif with the maternal version fails to bind importin α (Fig. 2D and E). Thus, these results suggest that the NLS-like motifs of DrMCM3 proteins retain similar properties compared with those of their Xenopus counterparts.

Developmental defects in zebrafish embryos after injection of DrMCM3 mRNAs

As DrMCM3m and DrMCM3z expression is differentially expressed, we questioned whether overexpression of these genes during zebrafish embryogenesis affects their development. To this end, mRNAs were injected into one- or 2-cell-stage zebrafish embryos. In zebrafish, all blastomeres are interconnected by cytoplasmic bridges until the 8-cell stage,²² so that mRNA can disperse from the injected into the non-injected cells. No defects were observed until 7 hpf (data not shown). The first visible defects were observed at 8 hpf, when normally developed embryos have a smooth enveloping layer consisting of flattened epithelial cells (Fig. 3A, normal). After injection of maternal or zygotic DrMCM3 mRNA, some embryos showed the partially rough enveloping layer with a round cell shape (phenotype A). A more severe phenotype observed in the injected embryos was the ejection of the cells from the embryonic cell layer (phenotype B). Those ejected cells were dark, seemingly dead cells. Embryos injected with DrMCM3m showed these phenotypes more frequently and at lower concentrations compared with those injected with DrMCM3z (Fig. 3A). At 10 hpf, a bilaterally symmetric neural plate formed in control embryos. Some embryos showed an unevenly thickened neural plate (phenotype C), or a shorter anterior-posterior body axis, indicating an extension defect (phenotype D) (Fig. 3B). Again, injection of DrMCM3m mRNA led to these phenotypes much more frequently compared with injection of DrMCM3z (Fig. 3B). In order to study MCM3 overexpression effects on replication, we also examined bromodeoxyuridine (BrdU) incorporation into cell nuclei in the injected embryos. Control embryos, which had been incubated with BrdU from 7 to 9 hpf, showed uniform incorporation of BrdU into the dividing cells. Injection of maternal but not zygotic DrMCM3 significantly decreased the incorporation of BrdU, suggesting that ectopic expression of DrMCM3m caused developmental defects through interference with replication (Fig. 3C). Apparent visible developmental defects were observed in 24 hpf embryos (Fig. 4), which could be categorized into 2 groups: phenotype 1, a laterally bent tail (left in Fig. 4A) or a shortened and curled down tail (right in Fig. 4A); and phenotype 2, a severely shortened body axis (left in Fig. 4A) and sometimes a twisted tail (right in Fig. 4A). Phenotype 1 is apparently milder than phenotype 2. The twisted tail in phenotype 1 and the bent tail in phenotype 2 were similar to the dorsalized phenotypes with different severity.^23-25 In addition, embryos with a severely shortened body axis (phenotype 2) resembled those with defects in convergent/extension cell movement.^23,26 This is consistent with the extension defect phenotype observed at 10 hpf (Fig. 3B). Injection of control EGFP mRNA, DrMCM7 (Fig. 4C) and DrMCM2 (data not shown) mRNA at 50 ng/μl induced a developmental defect in only a small fraction of the embryos. Occasionally, dead embryos or embryos showing defects of the head were observed (Supplementary Fig. 3). These phenotypes were also observed in control injected embryos and were not tightly related to the injected mRNA species. Although the injection of DrMCM3m mRNA at 25 ng/μl resulted in more than half of the embryos having developmental defects (phenotypes 1 and 2), almost all of the injected embryos showed the phenotypes at 100 ng/μl (Fig. 4B). On the other hand, the injection of DrMCM3z mRNA at 25 ng/μl led to developmental defects in only about 10% of the embryos (same as the injected control). The fraction of embryos with these phenotypes that were injected with 100 ng/μl appears to be almost the same or even less than that caused by injecting DrMCM3m at 25 ng/μl. In summary, DrMCM3m in later developmental stages (after MBT) affected embryonic development more severely than did DrMCM3z injection, suggesting a difference in molecular activity.

Figure 3.

Developmental defect observed after DrMCM3 mRNA injection. (A) Mid-gastrula embryos. Control (H1M-EGFP), DrMCM3m or DrMCM3z mRNA was microinjected into 2-cell embryos. Embryos were photographed at 8 hours after fertilization. Phenotype A shows a rough cell layer (arrow). Phenotype B shows ejected cells (arrowhead). Scale bar 150 μm. Embryos were categorized and the percentage showing phenotypes is shown. n means number of embryos counted. (B) Tail bud embryos. Control (H1M-EGFP), DrMCM3m or DrMCM3z mRNA was microinjected into 2-cell embryos. Embryos were photographed 10 hours after fertilization. Upper panels show the animal pole view, and lower panels show the lateral view. Phenotype C shows an uneven neural plate (black arrowheads), and phenotype D shows a defect in the extension of the body axis (head: arrow, tail: green arrowhead). Scale bar 150 μm. Embryos were categorized and the percentage showing these phenotypes is shown. (C) Compromised BrdU incorporation into nuclei caused by DrMCM3m mRNA injection. Control (H1M-EGFP), DrMCM3m or DrMCM3z mRNA was microinjected into 2-cell embryos. BrdU was injected at 7 hours after fertilization. Nine hours after fertilization, at the gastrula stage, embryos were fixed and stained with anti-BrdU antibodies. Nuclei that incorporated BrdU are shown as a brown signal. The percentage of embryos showing compromised BrdU incorporation is shown. Scale bar 150 μm.

Figure 4.

Developmental defect caused by DrMCM3 mRNA injection. (A) Representative phenotypes of pharyngula embryos caused by MCM3 mRNA injection. Phenotype 1: non-straight body axis (especially tail); phenotype 2: short body axis. Scale bar 300 μm. (B) Control (H1M-EGFP), DrMCM3m or DrMCM3z mRNA was microinjected into 2-cell embryos. At 24 hours after fertilization, embryos were categorized as in A, and the percentage of embryos showing phenotypes is shown. n means number of embryos. (C) Involvement of C-terminal regions of DrMCM3 in the developmental defect caused by overexpression. Experiments were performed as in B, using mRNA prepared from maternal and zygotic MCM3 and their chimeric genes as presented in Fig. 2D.

C-terminal regions of DrMCM3, containing the putative NLS, showed the least sequence similarity between maternal and zygotic forms (about 40% amino acid identity) and, more important, showed different binding affinity to importin α (Fig. 2). We wondered whether these differences in the C-terminal region might be the sole cause for the different developmental defects observed by DrMCM3 mRNA injections. To test this, we injected various mRNA constructs prepared from a chimera of maternal and zygotic DrMCM3 (Fig. 2D). Again, we observed stronger developmental defects by injecting mRNA of DrMCM3m than that of DrMCM3z. Although the chimeric gene mz1, which consists of an N-terminal 3-fourths maternal and C-terminal one-fourth zygotic gene, showed similar fractions of affected embryos with zygotic MCM3, mRNA injection of mz2 and mz3, which possess NLS of maternal MCM3, induced developmental defects more than that of mz1 or MCM3z (Fig. 4C). These results support the idea that a difference in the C-terminal regions, in particular the putative NLS sequence, contributes to the difference in MCM3 concentration to influence embryonic development.

Discussion

Function of NLS-like motifs of MCM3

MCM3 orthologs retain an evolutionary conserved NLS-like motif in their C-terminal regions. Here we show that maternal and zygotic MCM3 of Xenopus and zebrafish possess different activities in the NLS-like motifs. Xenopus zygotic but not maternal NLS-like motif functions as an active NLS in the egg extracts. In addition, similar motifs in zebrafish show different binding activity to importin α: Importin α prefers to bind the zygotic form. Higher binding activity of zygotic NLS may contribute to the different nuclear transport activity of MCM3 during development in both Xenopus and zebrafish. At present, we do not know the physiological significance of the different nuclear transport activities of maternal and zygotic MCM3. However, the high toxicity of maternal but not zygotic MCM3 overexpression in zebrafish embryo highlights the functional specificities for each of these molecules in developmental events. The results described here are the first experimental indication in vivo that maternal and zygotic forms of proteins have distinct roles in early development.

An open question is whether or not different NLS activities of maternal and zygotic MCM3 have important contributions to embryonic development. Previously, we proposed that the absence of NLS activity in maternal MCM3 contributes to establishing the MCM2–7 complex as a putative licensing factor.^18,21 Recent studies, however, suggest that MCM2–7 is transported into nuclei in Xenopus egg extracts.^27-29 It has also been established that the MCM complex that contains maternal MCM3 is accumulated into germinal vesicles (meiotic nuclei)¹⁹ and that MCM2 orthologs have active NLS.^20,30 Perhaps maternal MCM2–7 can be transported into the nuclei, but presumably at a much slower rate than zygotic MCM2–7 can be transported, depending on the NLS of MCM2.

We speculate that the slower accumulation of MCM2–7 into nuclei during the embryonic cell cycle may be beneficial for preventing re-replication. Egg contains a large amount of replication proteins, sufficient to support DNA replication of more than 4000 nuclei. Although cells use multiple mechanisms to prevent re-replication,⁷ active accumulation of large amounts of MCM complexes into the nuclei may in itself increase the probability of overcoming such mechanisms. In Xenopus egg, geminin, an inhibitor of MCM loading factor Cdt1, plays an important role in preventing re-replication. Interestingly, geminin is activated only after its nuclear transport.²⁸ Thus, it is reasonable to assume that down-regulation of nuclear transport activity of MCM complexes in the embryonic cell cycle contributes to the suppression of re-replication by allowing geminin to be fully active.

Functional NLS in MCM3 is essential for cell viability of budding yeast, which exhibits closed mitosis.³¹ In open mitosis, which includes the embryonic cell cycle, loading of MCM2–7 onto DNA mainly occurs during late anaphase³², i.e., before nuclear formation; thus, active accumulation of MCM2–7 does not play an important role. In the embryonic cell cycle, nuclear envelope breakdown should precede MCM loading. Furthermore, in Xenopus egg extracts, DNA replication occurs even in karyomeres, i.e., before complete nuclear formation,³³ suggesting that there is virtually no time for MCM2–7 to be transported into the nuclei. Thus, nuclear accumulation of MCM2–7 is not apparently required during the embryonic cell cycle. In contrast, MCM2–7 should retain active NLS activity in the somatic cell cycle. It has been reported that MCM proteins are absent in G0 phase cells.³⁴ Thus, newly synthesized MCM2–7 must be transported into the nuclei when cells return to the proliferating cell cycle from G0 phase. Active nuclear transport of MCM2–7 might also contribute to licensing in the normal cell cycle. Although MCM loading occurs during late anaphase in somatic cells,³² many factors involved in the licensing reaction are transcribed in G1 phase under the control of a transcription factor, E2F.³⁵ Thus, it is possible that a continuous supply of nascent MCM2–7 might be necessary to maintain the licensed state during the long G1 phase in the somatic cell cycle.³⁶ Overall, the use of maternal MCM2–7 complexes only during the embryonic cell cycle, which are defective in nuclear transport activity, seems to have a specific role.

Evolutionary adaptation of maternal MCM3

We found maternally expressed paralogs of MCM3 and MCM6 in Xenopus and zebrafish. Phylogenetic analysis (using the resources of Ensembl Compara) suggests that MCM3 and MCM6 were duplicated in the common ancestor of all vertebrates (http://Apr2013.archive.ensembl.org/Xenopus_tropicalis/Gene/Compara_Tree?db=core;g=ENSXETG00000020538;r=GL17 2720.1:703878–726436;t=ENSXETT00000044367;collapse= none; http://Apr2013.archive.ensembl.org/Xenopus_tropicalis/Gene/Compara_Tree?g=ENSXETG00000014891;r=GL1727 13.1:1249507–1257080;t = ENSXETT000000 32564; collapse = none), matching the time for the first (1R) or second round (2R) of whole genome duplication just after the divergence of the Urchordata (tunicates such as Ciona intestinalis). In addition to analyzing the Ensembl tree, in searching various fish genomes, we found that ray-finned fishes—medaka, fugu and Nile tilapia—which belong to Acanthopterygii, and a lobe-finned fish—coelacanth—have only one copy of MCM3 and MCM6 (Supplementary Fig. 1B and C). The Ensembl gene tree, in addition to synteny analysis (Supplementary Fig. 2), suggests that maternal MCM3 was lost in multiple lineages and was only retained in amphibians and some teleost linages such as zebrafish and cod.

Remarkably, the branch containing the zygotic zebrafish and Xenopus paralogs is still vast, whereas the branch with the maternally expressed paralogs has suffered extensive gene loss. The frequent absence of maternal MCM3 and the universal presence of a single zygotic form of the MCM3 gene in vertebrates may imply the presence of a selective advantage to keep the gene in amphibians and zebrafish. In other species, there is no advantage to keeping the maternal MCM3 gene, or there could even be a disadvantage because accidental expression of the maternal MCM3 gene could affect the somatic cell cycle.

A more severe effect being observed after overexpression of maternal MCM3 than of zygotic MCM3 supports the idea that the maternal form of MCM3 has acquired a different effect on proliferating cells during the somatic cell cycle. Compromised incorporation of BrdU into gastrula embryos after the overexpression of DrMCM3m suggests that it impairs DNA replication. The mechanism for that is not yet clear, but one possibility is that overexpressed maternal MCM3 protein forms an MCM2–7 complex with nascent proteins and may sequester them into the cytoplasm as a result of inefficient nuclear transport activity. In addition, it is still not clear whether the defect in DNA replication induced the dorsalized or c/e defect phenotypes, and if so, how a defect in DNA replication can lead to the phenotypes observed in zebrafish embryos. The evidence for a DNA replication defect, together with the observation of the dead cells in the mid-gastrula (Fig. 3A, phenotype B), suggests induction of cell death or less cell proliferation in the embryo; a reduction in the cells with the ventral fate might then cause embryos to dorsalize, or a reduction in the total cell numbers might disturb normal c/e cell movement. To confirm these speculations, we need to analyze the phenotypes of DrMCM3m mRNA-injected embryos more precisely.

Considering that maternal MCM3 has such a negative effect on the somatic cell cycle, why is the maternal MCM3 gene retained in zebrafish and Xenopus? At present we do not know the exact reason for this, but a possible explanation may be found in the difference in the genome size and length of the embryonic cell cycle (2-cell to 4-cell) (Table 1). The ratio of genome size per embryonic cell cycle length roughly indicates the minimal progression rate to complete DNA replication of the genome. Interestingly, Xenopus and zebrafish show much higher values compared with other organisms, which do not have maternal MCM3 and MCM6 genes. The difference is most distinct when we compare zebrafish and medaka, both of which are freshwater fish with similar body size. Zebrafish has a genome that is more than 2 times larger than that of medaka, but the length of the embryonic cell cycle is less than half that of medaka. It is possible that maternal MCM3 helps to quickly replicate the large genome. We also noticed that lungfish, positioned between coelacanths and amphibians in the phylogenetic tree, have larger genomes per embryonic cell cycle length. Although the genome sequence of lungfish is not yet available, our expectation is that lungfish also possess maternal MCM genes.

Table 1.

Minimal DNA replication rate of various organisms. Minimal replication rates were obtained by dividing genome size by cell cycle length from 2-cell to 4-cell

	Genome size (Mbp)	Cell cycle length (2-Cell to 4-Cell)	Minimal replication rate (Mbp/min)	Maternal MCM3 and 6
Human	3000	12 hrs	4.2	No
Mouse	3300	24 hrs	2.3	No
Frog	3100	30 min	103	Yes
Lungfish	51500–73200	4–7 hrs37	123–305	?
Zebrafish	1700	15 min	113	Yes
Medaka	800	40 min38	20	No
Fugu	390	2 hrs39	3.25	No
Lamprey	1900	6 hrs40	5.3	No
Sea squirt	155	30 min	5.2	No

We report that maternal MCM3 and zygotic MCM3 possess different activities in NLS-like motifs. Our results could not exclude the possible role of the motifs other than in nuclear transport activity. Further functional analysis of maternal and zygotic factors, including other factors such as Drf1 and Dbf4, will be crucial for understanding the developmental regulation of DNA replication.

Materials and Methods

Cloning of maternal and zygotic MCM3 and MCM6

Primers and plasmids used for his study are listed in the supplementary materials. XMCM3z was isolated from a lambda-ZAP library of stage 18 Xenopus embryos (Stratagene) using anti-XMCM3m antibody as a probe, which cross-reacts to XMCM3z weakly. XMCM6z was isolated from the same library using XMCM6m cDNA as a probe. Genes encoded by the lamda phage clone were rescued as pBluescript phagemids by in vivo excision following the manufacturer's protocol (Stratagene). DrMCM genes were amplified by PCR using specific primers and cDNA prepared from either 0 hpf (maternal) or 72 hpf (zygotic) embryos with Phusion DNA polymerase (New England Biolabs). DrMCM3z fragments, which contain an AscI site between BamHI and ATG, were cloned into the BamHI-NotI site of pBluescript KS+. The other fragments were subcloned into the plasmids at the AscI-NotI sites. All genes were sequenced using the ABI 3130xl DNA sequencer (Applied Biosystems). Accession numbers for the sequences are as follows: AB973823 for XMCM3m, AB973824 for XMCM3z, AB973825 for XMCM6m, AB973826 for XMCM6z, AB973819 for DrMCM3m, AB973820 for DrMCM3z, AB973821 for DrMCM6m, AB973822 for DrMCM6z.

RT-PCR and WISH

Total RNAs were prepared from developing embryos using TRIzol reagent (Life Technologies) according to the manufacturer's instructions. cDNA was obtained by using the PrimeScript 1^st strand cDNA Synthesis Kit (TaKaRa Bio) according to the manufacturer's instructions. RT-PCR was performed with ExTaq polymerase (TaKaRa Bio) using specific primers, and then amplified PCR products were separated by agarose gel electrophoresis followed by EtBr staining. The amount of DrMCM3 mRNA in each embryo was determined as follows. PCR reactions were performed using either the known amount of PCR products or zebrafish cDNA (2 or 48 hpf) as a template. The PCR products were separated by electrophoresis, and signals were analyzed by Image J (Supplementary Fig. 5). In summary, we estimated that a 2 hpf embryo contains 1.64 amol DrMCM3m mRNA and a 48 hpf embryo contains 0.74 amol DrMCM3z mRNA.

WISH was performed using digoxigenin-labeled probes synthesized by in vitro transcription with T3 and T7 polymerases. After overnight fixation in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS), embryos were transferred to methanol and then rehydrated in PBST (0.1% Tween 20 in PBS). Subsequently, embryos were prehybridized in HYB (50% formamide, 5× saline sodium citrate (SSC), 0.1% Tween 20, 50 μg/ml heparin, 5 mg/ml torula RNA) for a few hours at 65°C. Following an overnight incubation in HYB with a probe at 65°C, embryos were washed for 10 min each in 66% HYB/33% 2 × SSC, 33% HYB/66% 2 × SSC and 2 × SSC at 65°C. After a further 2 washes with 0.2 × SSC for 30 min at 65°C, embryos were replaced gradually transferred in a stepwise fashion to PBST: 10 min incubations at room temperature with 66% 0.2 × SSC/33% PBST, 33% 0.2 × SSC/66% PBST and finally PBST. Embryos were then blocked for 1 hour in blocking buffer (2 mg/ml BSA, 2% goat serum, 0.1% Tween 20 in PBS). Alkaline phosphatase (AP)-coupled anti-digoxigenin Fab fragments (1:8000) were added to fresh blocking buffer. After overnight incubation at 4°C, the antibody solution was washed out with 4 washes of PBST for 15 min at room temperature. Before color reaction, embryos were rinsed with AP buffer (100 mM Tris at pH 9.5, 50 mM MgCl₂, 100 mM NaCl, 0.1% Tween 20, 1 mM levamisol) twice for 15 min. Detection was performed using BM purple (Roche).

Fish maintenance and mRNA injection

Zebrafish were maintained at 27°C and embryos obtained from natural crosses of wild-type fish. Embryos were staged according to hours postfertilization (hpf) at 28.5°C and morphological criteria.²² Capped sense RNAs were synthesized using the mMESSAGE mMACHINE T3 kit (Ambion/Life Technologies). A total of 1 nl of the mRNAs (either 25 or 100 ng/μl) were injected into one- to 2-cell stage embryos. As the length of MCM3 genes is about 2.4 kb, either 30 or 120 amol mRNA was injected per embryo. Injected embryos were cultured in 0.3× Danieau's solution (17.4 mM NaCl, 0.21 mM KCl, 0.12 mM MgSO₄, 0.18 mM Ca(NO₃)₂, 1.5 mM HEPES at pH 7.6) until use.

BrdU labeling and detection

To analyze DNA replication, we injected about 0.2 nl of 3 mg/ml BrdU into the intercellular space of embryos at 60% epiboly stage. After 2 hours’ incubation, injected embryos were fixed with 4% PFA/PBS for 2 hours at room temperature. After washing with PBS, the embryos were treated with 4N HCl for 15 min at room temperature to relax chromatin and facilitate immunodetection of incorporated BrdU. After washing the acid-treated embryos, we performed immunodetection by following the reported protocol.⁴¹

Genome analysis

The presence of MCM3 and MCM6 genes in various organisms was analyzed using the Ensemble, Genomicus, NCBI BLAST, NBRP Medakafish Genome Project, and Fugu Genome Project websites. Genome size of various organisms was examined using Genome Size database.

Assays using recombinant proteins

A PCR fragment of EGFP, which contains the SpeI site just before the stop codon, was inserted into pGEX6P-1 at BamHI-XhoI sites. DNA fragments of MCM3 genes were amplified by PCR and subcloned into the plasmid using the SpeI-XhoI site. BL21-CodonPlus cells were transformed by those plasmids and induced recombinant protein expression by IPTG at 20°C for several hours. Cells were harvested and lysed in lysis buffer (40 mM Tris-HCl at pH 7.5, 300 mM NaCl, 1% TritonX-100, 1 mM PMSF) by sonication following lysozyme (0.1 mg/ml) treatment at 4°C for 15 min. The soluble fraction was separated by centrifugation at 10,000 rpm for 40 min. Proteins were bound to GSH-sepharose beads (GE Healthcare), washed by wash buffer (40 mM Tris-HCl at pH 7.5, 150 mM NaCl, 0.1% TritonX-100) and eluted by elution buffer (100 mM Tris-HCl at pH 7.5, 150 mM NaCl, 0.1% TritonX-100, 20 mM GSH, 5% glycerol). His-tagged Dr importin α2 was cloned by PCR from zebrafish embryo cDNA and subcloned into pET30a (Novagen) with modified MCS containing an AscI site just after the BamHI site at the AscI-NotI site. Proteins were purified similarly to GST-fusion proteins, except that different beads (Ni-NTA agarose [QIAGEN]) and buffers were used (lysis buffer: 40 mM Tris-HCl at pH 7.5, 300 mM NaCl, 1% TritonX-100, 1 mM PMSF, 20 mM imidazole; wash buffer: 40 mM Tris-HCl at pH 7.5, 150 mM NaCl, 0.1% TritonX-100, 20 mM imidazole; elution buffer: 40 mM Tris-HCl at pH 7.5, 150 mM NaCl, 0.1% TritonX-100, 5% glycerol, 300 mM imidazole). For the in vitro pull-down experiment, either 0.25, 0.5 or 1 μg GST-fused proteins were bound to GSH-agarose beads and incubated in TBS (40 mM Tris-HCl at pH 7.5, 150 mM NaCl) with or without 1 μg Dr importin α2 for 60 min at 4°C. Beads were washed with TBS buffer, and bound proteins were separated by SDS-PAGE and detected by immunoblotting using anti-GST antibody and anti-His antibody. For the assay using full-length MCM3, a solubilized fraction of E. coli, instead of purified recombinant proteins, was used because elution of full-length protein resulted in severe degradation (data not shown).

For the nuclear transport assay, sperm nuclei (4000 nuclei/μl) were incubated in interphase Xenopus egg extracts for 30–40 min at 23°C. Nuclear formation was confirmed by phase contrast microscopy. Recombinant EGFP fusion proteins (0.25 μg) were then added and incubated for a further 10 min. Nuclei were fixed with fixation buffer (3.8% formaldehyde, 30 mM KCl, 15 mM NaCl, 15 mM PIPES at pH 7.2, 2 μg/ml Hoechest, 50% glycerol) and observed by fluorescent microscopy.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We are grateful to former laboratory member Yoshitami Hashimoto for cloning of the zygotic MCM6 gene. We thank the Nuclear Function Laboratory members, especially Yon-Soo, Tak, for discussion, critical readings and comments on the manuscript. M.S. performed experiments using zebrafish; DM, YK and SM performed cloning of genes and nuclear transport analysis; TH and SM performed genome analysis; HT and SM designed and managed the study; and MS, TH, HT, and SM wrote the manuscript.

Funding

SM was supported by a Grant-in-Aid for Young Scientists (B) and a Grant-in-Aid for Scientific Research on Priority Areas.