MCM2-7 proteins exist abundantly in proliferating cells and are bound to chromatin in amounts exceeding that required to license all replication origins that initiate DNA synthesis 
. The role of excess chromatin-bound MCM2-7 has been a mystery referred to as the “MCM paradox” 
, perpetuated by observations that drastic MCM reductions in certain systems can be compatible with normal DNA replication or cell proliferation 
. However, these circumstances are not universal, and reductions are not entirely without consequences. Early studies showed that a reduction in MCMs resulted in decreased usage of certain ARSs 
and conferred genome instability 
in yeasts. In cell culture systems, depletion of certain MCMs have been found to cause cell cycle defects, checkpoint abberations and GIN 
Recent work has shed light on aspects of the MCM paradox. Using Xenopus
egg extracts attenuated for licensing by addition of geminin (an inhibitor of CDT1, which is required for MCM loading onto origins), one study proposed that excess chromatin-bound MCM2-7 complexes license “dormant” origins that can be activated to rescue stalled or damaged replication forks, a situation that can become important under conditions of replication stress 
. Similar results were subsequently reported for human cells depleted of MCMs by siRNA 
, and for replication stressed MCM2-deficient MEFs 
. Our finding that nuclear MCM2 levels decrease as S-phase progresses, and moreso in WT than in Mcm4Chaos3/Chaos3
MEFs, is consistent with the dormant origin hypothesis. The decrease may reflect displacement of dormant hexamers by active replisomes, followed by subsequent degradation or nuclear export. If WT nuclei have more dormant licensed origins than Chaos3 mutants, then WT cells would be expected show a greater loss of MCMs.
The isolation of Mcm4Chaos3
provided the first demonstration that mutant alleles of essential replication licensing proteins can cause GIN and cancer 
. Diploid budding yeast containing the same amino acid change in scMcm4 as the mouse Mcm4Chaos3
exhibited Rad9-dependent G2/M delay (Rad9 is a DNA damage checkpoint protein), elevated mitotic recombination, chromosome rearrangements, and intralocus mutations 
(Li, X. and Tye, B., personal communication). One explanation for these outcomes is that the Chaos3
mutation impairs MCM4 biochemically in a manner leading to elevated replication fork defects, and that these defects lead to the GIN and cancer phenotypes. Alternatively, and/or in addition, the observed associated pan-reductions of MCMs in mouse cells 
raised the possibility that decreased replication licensing might be the primary or ancillary cause for the mouse phenotypes.
The subsequent finding that mice (Mcm2IRES-CreERT
) containing ~1/3 the normal level of MCM2 had GIN and and cancer lent support for the idea that reductions in MCMs contribute to the Chaos3 phenotypes 
. Although amounts of all MCMs were not investigated in Mcm2IRES-CreERT/IRES-CreERT
mice, 65% reduction of MCM2 caused a reduction of dormant replication origins in MEFs that were replication stressed by hydroxyurea 
. In Mcm4Chaos3/Chaos3
mice, we hypothesize that in the context of Mcm2
heterozygosity, which further reduces overall and chromatin-bound MCM levels below that already caused by Mcm4Chaos3
(measured to be <20% of WT mRNA levels for Mcm2
), MCMs are reduced to a degree that compromises cell proliferation. This then translates into the various developmental defects and increased cancer susceptibility we observed. Whatever the exact mechanistic cause of these phenotypes, it is clear that the phenotypes are related to reduction of one or more MCMs below a threshold level that is <50%. The severe developmental consequences of MCM depletion in mice suggests that certain cell types in the developing embryo are highly sensitive to the effects of replicative stress, and/or that relatively minor cell growth perturbations of such cells are not well-tolerated in the context of complex, rapidly-occuring developmental events. The molecular basis for these phenotypes does not appear to be directly related to GIN, because whereas Mcm3
hemizygosity rescued several phenotypes, and delayed cancer latency in Mcm4Chaos3/Chaos3
mice, it did not concommitantly decrease MN. This suggests that phenotypes such as decreased proliferation and embryonic death are caused by genetically-induced replication stress, moreso (or in addition to) than GIN alone.
Our genetic studies indicate that there is a quantitative MCM threshold required for embryonic viability, as demonstrated by the synthetic lethalities we observed when combining homozygosity of Mcm4Chaos3
, Mcm6Gt or Mcm7Gt
heterozygosity, but not in the heterozygous single mutants. Additionally, the Mcm4Chaos3/Gt
genotype, which reduced MCM levels below 50%, caused embryonic and neonatal lethality 
. Underscoring the exquisite sensitivity of whole animals to subtle perturbations in the DNA replication machinery were the remarkable phenotypic rescues (viability, growth, iPS efficiency, etc.) by Mcm3
hemizygosity. The decreased MCM dosage led to increases in S phase nuclear MCMs and chromatin-bound MCMs, presumably reflecting increased replication origin formation. The various single and compound mutants described here and elsewhere 
, which show that 50% reductions of any one MCM is well-tolerated but decreases of ~2/3 are not, supports the idea of a threshold effect, and suggests that the threshold lies somewhere between 1/3 and 1/2 of normal MCM levels (at least in the cases of MCM2, MCM6 and MCM7).
These results also emphasize the importance of relevant physiological models, both in general and with respect to the MCMs. RNAi knockdown of MCM3 in human cells to ~3% normal levels was still compatible with normal short-term proliferation, although the cells had GIN and high sensitivity to replication stress 
. It is doubtful such a drastic situation would be recapitulated in vivo
(it would likely result in embryonic lethality as in Mcm3Gt/Gt
mice). Nevertheless, it is noteworthy in that study that MCM3 depletion was better tolerated than knockdowns of any other member of the replicative helicase.
The finding that reductions in MCM3 rescued MCM2/4/6 depletion phenotypes lends insight into dynamics and regulation of mammalian DNA replication. In budding yeast, MCMs shuttle between the nucleus and cytoplasm during the cell cycle. MCM2-7 multimers must be assembled in the cytoplasm before being imported into the nucleus during G1 phase 
. The MCM2-7 importation is dependent upon synergistic nuclear localization signals (NLS) on Mcm2 and Mcm3 
. In order to prevent over-replication of the genome, MCMs are exported from the nucleus during S, G2 and M 
. This export is dependent upon Mcm3, which has a nuclear export signal (NES) that is recognized by Cdc28 to promote MCM2-7 export in a Crm1-dependent manner 
In contrast to budding yeast, MCMs that have been studied (MCM2/3/7) are primarily nuclear-localized throughout the cell cycle in metazoans and in fission yeast 
. Upon dissociation from chromatin during S phase, MCM2-7 complexes are reported to remain in the nucleus but are sequestered via attachement to the nuclear envelope or other nuclear structures 
. Interestingly, mcm
mutations in fission yeast that disrupt intact MCM2-7 heterohexamers triggers active redistribution of MCMs to the cytoplasm 
. Additionally, re-distribution of MCMs between the cytoplasmic and nuclear compartments has been observed in hormonally-treated mouse uterine cells 
Our observations support the idea that intracellular re-distribution of MCMs also occurs in mammals, and that it is an important regulatory process. Staining of MCM2 in intact nuclei of normal NIH 3T3 fibroblasts and MEFs show a steady decline (but not elimination) as S phase progresses. Furthermore, it appears that the process of nuclear MCM2 elimination during S phase is regulated, since in situations of decreased MCMs (as in the Mcm4Chaos3/Chaos3 mutant), there is decreased loss of nuclear MCM2 during S phase.
Three lines of experimentation implicate MCM3 as playing a key role in regulating intracellular MCM localization: 1) Rescue of reduced-MCM phenotypes by genetic reduction of MCM3; 2) Increased S-phase nuclear MCM2 by Mcm3
hemizygosity in MCM-depleted cells (); and increased chromatin-bound MCM2/4 by Mcm3
hemizygosity in MCM-depleted cells. Our data suggests that MCM3 acts as a negative regulator that prevents re-assembly or reloading of MCM complexes as they dissociate from DNA during replication. As described earlier, mouse and human MCM3 have predicted NESs in similar positions of their primary amino acid sequences as do the yeast genes. Thus, one explanation for these phenomena is that decreased MCM3 suppresses MCM2-7 nuclear export, which occurs normally and which may be accentuated by the Chaos3
mutation in a fashion analogous to mcm
mutant fission yeast discussed above 
. This would effectively increase the amounts of MCMs available for replication licensing. More work is required to determine if the rescue mechanism is indeed related to a decrease in MCMs export, as opposed to direct or indirect involvement in other events such as increased nuclear import or enhanced chromatin loading.
With respect to the early lymphoma susceptibility phenotype in Mcm4Chaos3/Chaos3 Mcm2Gt/+
mice, it is unclear whether the type of tumor is dictated primarily by the particular Mcm depletion (in this case MCM2, thus resembling Mcm2IRES-CreERT2/IRES-CreERT2
animals), the genetic background, or the age of particular cancer onset (if animals die of thymic lymphoma at an early age, they will be unable to manifest later-arising mammary tumors). The compound mutant mice used for the aging aspects of this study were bred to at least the N3 generation in strain C3H. Mcm4Chaos3/Chaos3
mice congenic in this background are predisposed exclusively to mammary tumors, whereas lymphomas were observed in mutants of mixed background 
. Presently, we favor the idea that genetic background and age of tumor type onset are primary determinants of the cancers that arise in the mice we have studied thus far. Genetic background has also been reported to influence tumor latency in MCM2-deficient mice 
The MCM2-7 pan-reduction in Chaos3 cells is consistent with other studies involving mutation or knockdown of a single MCM in mammalian cells 
. In these examples of parallel MCM decreases, the general assumption is that there is hexamer destabilization or impaired MCM chromatin loading followed by degradation of monomers. However, we found that the protein decreases are related to decreased mRNA levels. These large (~40%) decreases do not appear to be attributable to transcriptional alterations from cell cycle disruptions (these cells have a small elevation in the G2/M population), but rather occur at the post-transcriptional level (unpublished observations). Since we also found that MEFs carrying only 1 functional Mcm2
allele caused ~20% decreases of Mcm3-7
mRNAs, it is possible that mRNA downregulation drove MCM reductions in these other model systems. However, the mechanism for coordinated mRNA regulation, and what triggers it, is a mystery that we are currently investigating.
Our data contribute to a growing body of data that replication stress, which can occur via perturbations of the DNA replication machinery, plays a significant role in driving cancer 
. While the Mcm4Chaos3
mutation is an unique case, the deleterious consequences of MCM reductions suggest that genetically-based variability in DNA replication factors can have physiological consequences. Such variability in functions or levels may be caused by Mendelian mutations or multigenic allele interactions. Mutations affecting transcriptional activity of one or more Mcms, which might occur in non-coding cis
-linked sequences or unlinked transcription factors, could have such effects. This has implications for cancer genome resequencing projects, whereby such mutations would not be obviously associated with MCM expression. The allelic collection we generated, when used alone or in combination with each other or Mcm4Chaos3/Chaos3
mice, allow the generation of mouse models with a graded range of MCM levels. These should be valuable for investigations into the impact of replication stress on animal development, cancer formation, and cellular homeostasis.