The MCM8/9 complex: A recent recruit to the roster of helicases involved in genome maintenance

Abstract

There are several DNA helicases involved in seemingly overlapping aspects of homologous and homoeologous recombination. Mutations of many of these helicases are directly implicated in genetic diseases including cancer, rapid aging, and infertility. MCM8/9 are recent additions to the catalog of helicases involved in recombination, and so far, the evidence is sparse, making assignment of function difficult. Mutations in MCM8/9 correlate principally with primary ovarian failure/insufficiency (POF/POI) and infertility indicating a meiotic defect. However, they also act when replication forks collapse/break shuttling products into mitotic recombination and several mutations are found in various somatic cancers. This review puts MCM8/9 in context with other replication and recombination helicases to narrow down its genomic maintenance role. We discuss the known structure/function relationship, the mutational spectrum, and dissect the available cellular and organismal data to better define its role in recombination.

1. Introduction

Cellular survival and propagation of an uncorrupted genome by the replisome is dependent on the ability of the cell to efficiently recognize and remove genetic insults which impede routine nuclear processes and threaten genomic integrity. Further, the origins, whether endogenous or exogenous, and biochemical outcomes of DNA damage are fraught with chemical and physical diversity. These genomic insults can result in replisome stalling and fork collapse giving rise to DNA double-strand breaks (DSBs). The total number of DSBs suggested to form is staggering. Up to ten DSBs are projected to form in any given cell per day, and based on current estimates of cellular abundance (3.72 × 10¹³) in the human body, 372 trillion DSBs are predicted to form every day [1]. Fortunately, there are a multitude of pathways that the cell has to prevent DSB formation or repair a DSB once formed. A comprehensive assessment of these pathways and mechanisms are beyond the scope of this review, but should the reader seek more information, multiple recent reviews can be easily found [2–5].

Prolonged replisome stalling represents a serious threat to cellular survival due to replication fork collapse and potential DSB formation. In response, the cell has evolved several mechanisms to mediate replisome progression, prevent replication fork collapse, and initiate fork restart to restore replication progression. Should DSBs happen, the global pathways of non-homologous end joining (NHEJ) and homologous recombination (HR), each with its own menagerie of sub-pathways are the most general mechanisms of repair. Choice between NHEJ and HR is mainly influenced by the cell cycle phase. NHEJ serves as the default pathway and predominates during G₁. As cells progress into S and then G₂, NHEJ is inhibited and the balance shifts toward HR repair [6]. Mitotic HR requires that a homologous sister chromatid be present to serve as a template from which to restore genetic information. The processed 3’ end of the DSB coated with the Rad51 recombinase invades the sister chromatid and searches for homology for downstream synthesis. A single Holliday junction (HJ) intermediate is formed and dissolved after sufficient synthesis to produce primarily non-crossovers (NCO) products.

Meiotic recombination shares many similarities and components to that discussed for mitotic HR with some specific differences [7, 8]. First, meiotic DSB formation is a more highly controlled process catalyzed by Spo11 and other accessory proteins acting primarily during the leptotene stage of meiosis. Instead of Rad51, the meiosis-specific recombinase, Dmc1, catalyzes strand invasion, synapsis, and facilitates more extensive synthesis to promote capture of the free DSB end to create a joint double Holliday junction (dHJ) intermediate. Structure and meiosis-specific nucleases then cleave (or resolve) dHJs to facilitate crossover formation (CO) to promote genetic recombination.

During all these processes, there are a multitude of DNA helicases that act in specific ways and on specific intermediates to prevent replication fork collapse, restart stalled forks, facilitate strand resection, mediate homology search, and promote dissolution of HJs. A newly identified helicase complex that is emerging as an important mediator of DNA repair and guardian of genomic integrity is formed between the minichromosome maintenance (MCM) 8 and 9 proteins [9–12].

2. Roles of MCM8/9 in meiosis, infertility, and cancer

Initial experiments by Rhoda Grell in Drosophila melanogaster in 1984 identified a temperature sensitive recombination gene termed recombination defective (REC) [13]. REC null organisms displayed higher levels of nondisjunction and infertility despite undisrupted homologous chromosome pairing, and so, it was proposed that REC functions to create CO during meiosis. Subsequent deletion, restriction mapping, and sequencing work localized the REC gene to chromosome 3 and was identified as a homolog of MCM8 [14, 15]. Further, MCM8 deficiency reduced CO events by ~95%, were abnormally distributed, and gene conversion tract lengths were ~2-fold shorter. Based on these data, it was proposed that dmMCM8 functions after SPN-A (Rad51 homolog) but before the MEI-9 endonuclease to aid in strand invasion and DNA synthesis to give rise to successful CO events.

Links connecting MCM8 and 9 to infertility and cancer did not come until several years after their initial discovery when Hartford and colleagues, prompted by a report that MCM9 interacts with Cdt1 to assist MCM2–7 loading prior to replication firing, created MCM9 knockout mice using gene trap insertions. MCM9^−/− mice were viable and grossly normal with no discernible MCM2–7 loading defects. However, MCM9-mutated mouse embryonic fibroblasts (MEFs) exposed to aphidicolin (a B-family DNA polymerase inhibitor) were delayed in their progression from G0/G1 to S-phase [16]. Male and female MCM9-mutated mice also exhibited reproductive system abnormalities and germ cell depletion. Removal of p53 and p21 through cross-breeding further depleted germ cell abundance and indicated that loss likely occurs through an apoptotic-independent pathway. Female mice also displayed fewer primordial follicles, were infertile, and presented hyperplasias and tubulostromal adenomas consistent with pre-mature ovarian failure (POF). Further, an increase, though minimal, in genome instability was observed along with an increased incidence of sex-specific tumor formation with a higher incidence of hepatocellular carcinoma in males, while females were prone to ovarian tumor formation. Based on these data, it was proposed that MCM8/9 function in genome maintenance during meiotic replication-induced stress.

Currently, multiple reports have corroborated mutations in MCM8 or 9 with POF leading to infertility in humans [17–22]. The genetic basis of POF is complex with ~1% of women under 40 years-of-age affected. Many of the molecular mechanisms which cause POF are unknown, however the disruption of a multitude of genes involved in all stages of oogenesis, folliculogenesis, and DNA repair in establishing and maintaining the ovarian reserve have been implicated [23].

The first report outlining MCM8/9 contributions to POF in humans came in 2014/15 when our colleagues examined daughters from three separate consanguineous families that exhibited primary amenorrhea, short stature, low weight, and genomic instability [17, 18]. A combination of single nucleotide polymorphism (SNP) arrays, comparative genomic hybridization, and whole-exome sequencing identified a C>T nonsense mutation that likely results in an early truncated form of MCM9 and a T>C mutation that disrupted normal splicing between exons 9 and 10 of MCM9. Within the affected MCM8 family, a missense mutation resulted in the amino acid change P149R within the canonical DNA binding domain. For all these mutants, the ability to form damage foci upon treatment with the DNA cross-linker mitomycin C (MMC) was severely compromised and affected patient lymphocytes showed a significant increase in chromosomal breaks and rearrangements.

Since these initial discoveries, multiple mutations in MCM8 and MCM9 have been associated with POF (Figure 1A–C) and direct links to cancer are less clear. Only a single report links mutations in MCM9 with mixed polyposis and early-onset colorectal cancer [22], although another study found it was more rare in hereditary Lynch syndrome patients in a broader Australian population [24]. Even so, there are more than 460 mutations in MCM8 or MCM9 cataloged in The Cancer Genome Atlas and cBioPortal for Cancer Genomics implicating MCM8/9 in cancer transformation or progression that are awaiting characterization. The MCM8/9 mutations identified in POF patients (Figure 1A–C) and the ever expanding database of cancer genomes with MCM8 or 9 mutations all lack sufficient biochemical characterization to detail what molecular and cellular effects these mutations have on contributing to POF or cancer progression.

Figure 1:

POF implicated mutations mapped onto the A) primary sequences of MCM8 or MCM9 with indicated domains and motifs for each. Structural homology models shown in ribbon of B) MCM8 or C) MCM9 modelled from the most closely related yeast MCM2 subunit (PDBID:3JA8) with POF mutations mapped as indicated. D) Predicted quaternary homology model of the heterohexameric structure of MCM8/9 built upon the yeast MCM2–7 complex.

3. Identification, conservation, and structural characterization of MCM8/9

Though MCM8 and 9 form a functional complex in vivo, the two enzymes and alternatively spliced isoforms were discovered separately through bioinformatics analyses by multiple groups [25–29]. MCM8 and 9 arose early during eukaryotic evolution and are conserved among many eukaryotic organisms, with yeast and C. elegans as exceptions, which is suggestive of a co-evolutionary function [30]. However, D. melanogaster and related species possess only the MCM8 protein (i.e. REC) in the absence of an MCM9 partner, suggestive of a divergence point during evolution in which MCM8 assumed the role of MCM9 or MCM9 presence was redundant and selected against.

MCM8 and 9 are homologs of the MCM2–7 hetero-hexameric helicase complex, which together with Cdt1 and GINS make up the CMG complex becoming the primary DNA helicase involved in eukaryotic replication. MCM8/9 was first thought to aid in replication initiation [31–33], however subsequent studies have argued instead for a role during replication elongation or recombination [34, 35], although direct interactions with the other MCM2–7 subunits has not been found. At the time of writing this review, there has been no structural information (X-ray or EM) for MCM8 or 9. However, MCM structures from several species (including eukaryotic) have been solved [36–39]. It is likely that the highly conserved helicase domains of MCM8 (Figure 1B) and MCM9 (Figure 1C) adopt similar heterohexameric structures to that of the other MCM complexes (Figure 1D). The precise stoichiometry of the MCM8/9 complex also remains to be definitively addressed. Initial differential centrifugation and gel filtration experiments suggest that MCM8/9 can exist within a range of dimeric to hexameric species [11, 34]. With this data, in combination with structural and stoichiometric information from the primary replicative MCM complexes and other AAA⁺ proteins from yeast and Archaea, it is attractive to propose a structural model whereby MCM8 and 9 subunits alternate to form the hetero-hexameric ring (Figure 1D) with a central channel that can accommodate DNA. If this is the case, it will be interesting to learn how the ringed hexamer is loaded onto DNA and whether its loading is as tightly controlled as it is for MCM2–7.

There are over 100 mutations in both MCM8 and MCM9 which are implicated in POF and cancer. Mapping all reported mutations to their respective domains has revealed that there is a clear bias in mutational abundance within the ATPase domain (61%) of MCM8 with the remaining 39% localized to the DNA binding domain. The opposite trend was observed for MCM9 with 43% of total reported mutations localized to the DNA binding domain with a nearly even distribution of the remaining mutations at 30% and 27% between the ATPase and C-terminal domains (CTD), respectively. It is expected that most point mutations within these conserved DNA binding and ATPase domains will negatively affect their respective activities (DNA binding and ATPase/helicase) concomitant with replication or repair activity. Additionally, point mutations within the long CTD of MCM9, which likely functions as a regulatory domain for protein-protein interactions (discussed in more detail below), would severely perturb MCM8/9 activity. It is possible that mutations within these domains might also affect the formation of the MCM8/9 complex and subsequent localization to sites of DNA damage.

The defining feature separating MCM9 from MCM8 and other MCM proteins is the presence of a unique and long CTD (~530 amino acid residues) (Figure 1A). At the time of writing this review, no structural information for MCM9 CTD has been reported. It is hypothesized that the majority of this CTD is unstructured and disordered. However, structure prediction with the RaptorX-Property software has suggested and number of α-helical and β-sheet regions present within amino acids 643–862 [40]. This CTD is important for MCM9 function as nonsense mutations that result in MCM9 truncation or complete deletion are linked primarily to infertility but also cancer. Strikingly, the layout of MCM9 bears resemblance to that of the yeast Srs2 helicase (despite having low sequence homology) which also contains a long C-terminal extension with multiple motifs that mediate the interaction and regulation of other proteins including Rad51, PCNA, and SUMO [41] (Figure 2). Interestingly, it has been previously demonstrated through immunofluorescence confocal microscopy that MCM9 possesses Rad51 recruitment activity in the presence of DNA damage [12]. These overlapping domain architectures and activities of Srs2 and MCM9 likely point to a function whereby the MCM9 CTD serves as an interaction “hub” to mediate protein-protein interactions.

Figure 2:

Linear representations shown with conserved DNA helicase domains (blue) and extended and uncharacterized C-terminal domains (yellow) from a variety of related helicases with indicated motifs.

The presence of this C-terminal domain is not confined to MCM9 or Srs2. In fact, multiple human helicases involved in various DNA repair pathways possess CTDs which contain protein-protein interaction motifs or domains (Figure 2). For example, the nuclear isoform of RecQL5 (RecQL5β) also contains a C-terminal extension of ~500 amino acids that houses a triple helix bundle termed a KIX domain and a Set2 Rpb 1-interacting (SRI) domain which interact with the large subunit of RNA pol II [42]. The KIX domain along with the helicase function of RecQL5 is suggested to play an important role in regulation of HR and sister chromatid exchange by disrupting Rad51 nucleoprotein filaments [43]. The FancJ helicase, which is involved in interstrand crosslink repair possesses a BRCA1 binding domain and associates with BRCA1 during replication to protect the replication fork [44]. ZRANB3, like Srs2, also has a PCNA-interacting peptide (PIP) motif along with a secondary PCNA interacting motif termed an AlkB homolog 2 PCNA interacting motif (APIM) and an Npl4 Zinc-finger (NZF) which recognizes polyubiquitinated (pUB) PCNA [45]. Translocase activity and interaction with PCNA by ZRANB3 are important for fork protection and reversal to effect error-free DNA repair [46]. Lastly, the CTD of RTEL1 contains a PIP motif, Harmonin-N-like, and Really interesting new gene (RING) finger domains which are suggested to facilitate replication and stress responses through protein-protein interactions [47–50]. Any sequence motifs or specific protein partners which interact with the CTD of MCM9 remain to be identified, but it would not be surprising to find them.

4. Helicases bridging DNA replication and recombination

MCM8/9 have been linked with DNA elongation processes, upregulated with specific DNA damaging agents, localized with several HR proteins, and mutations are implicated in both infertility and cancer. However, a specific enzymatic role or substrate specificity has not been defined, making assignment of precise function difficult. In fact, there are several analogous DNA helicases which appear to share similar or overlapping roles in various aspects of DNA processing linking replication with interstrand crosslink repair or directly with recombination (Table 1). Specific activities for helicases in these arenas include aiding in fork protection as accessory helicases to the CMG complex, being directly involved in fork reversal for replication restart, promoting resection of reversed forks or double-strand breaks, acting as antirecombinases to remove homeologous synapses formed from microhomologies, or branch migrating and dissolving specific HR intermediates. The RecQ family of helicases (RecQL1, WRN, BLM, RecQL4, and RecQL5) are conserved from bacteria to humans and are critical for many of these aspects of genomic maintenance [51], however they are not alone. Many of these DNA helicases have gene defects linked directly to cancer and other aging related diseases, including Werner syndrome (WRN), Rothmond-Thomson (RecQL4), Bloom syndrome (BLM), and Fanconi anemia (FancJ) [52, 53]. Loss of function in any DNA repair helicase has the potential to lead to disease [54].

Table 1:

Replication/Recombination Helicases

	E. coli	S. cerevisiae	X. laevis / H. sapiens
Fork Progression	DnaB	MCM2–7	MCM2–7
	Rep	Pif1, Rrm3	Pif1
	RecQ		RecQL4, RecQL5
	PriA		FancJ, RTEL
		Mph1	FancM
			MCM8, MCM9
Fork Reversal			SMARCAL1*, ZRANB3*
			HTLF*
	RecG	Sgs1	BLM, WRN
	UvrD	Srs2	FBH1
		Mph1	FancM
			MCM8, MCM9
Fork Protection			FancJ, FancM
Fork Restart	RecQ	Sgs1	RecQL1
			FancJ
Resection	RecQ	Sgs1	BLM, WRN
	UvrD	Srs2	FBH1
		DNA2	DNA2
	RecBCD
			FancJ
			MCM8, MCM9
Synapsis		Sgs1	RecQL1, BLM, RecQL5
			FancJ, RTEL1
	UvrD	Mph1	FBH1
			MCM8, MCM9
		Mer3	HFM1, HelQ
Branch Migration	RecG, RecQ	Sgs1	RecQL1, BLM, WRN
	RuvAB
Dissolution	RecQ	Sgs1	BLM

^*translocase

A variety of factors including genotoxic agents that cause DNA damage, DNA-protein crosslinks, protein complex collisions on chromosomes, and difficult to replicate sequences will all stall replication forks [55–58]. Analysis of cellular viability, protein localization/colocalization fluorescence microscopy, and coimmunoprecipitations using a variety of genotoxic agents in both animal and cell models indicate that MCM8/9 are likely involved in only a sub-set of DNA repair pathways dependent on the type of damage (Figure 3). For example, cells deficient for MCM8/9 show increased sensitivity to DNA cross-linking agents MMC and cisplatin (cis-Pt) in addition to the DNA alkylating agent methyl methanesulfonate (MMS) [10–12, 34]. These data indicate that MCM8/9 are important for crosslink repair preferentially encountered during replication and likely interact with members of the Fanconi anemia pathway. A less dramatic effect was observed for the DNA polymerase inhibitor aphidicolin (APH) and the topoisomerase poisons: camptothecin (CPT) and etoposide (ETO). Lastly, little to no effect was observed for ionizing radiation (IR), doxorubicin (DXR), bleomycin (BLE), and hydroxyurea (HU) arguing against a role in direct DSB repair or fork stalling. Instead, MCM8/9 seem to be present after severe replication induced stalling induced primarily by crosslinks or other large lesions that shuttles repair towards HR.

Figure 3:

Cumulative impact of various DNA damage agents on perceived activity, regulation, localization, interactions, and viability of MCM8/9 from organismal and cellular studies.

4.1. Fork Progression.

Besides the CMG helicase, other accessory helicases are associated with the replisome and thought to aid in fork progression (Figure 4) over difficult to replicate regions. Pif1 is known to bind to and unwind structures relating to stalled replication forks and is active during replication stress where replication fork slowing is observed [59]. Besides, CMG, RecQL5 and MCM8/9 are the only other DNA helicases enriched at active replication forks over those on general chromatin in the absence of stress [43, 60]. Cells deficient in RecQL5 (either knockout or knockdown) have an increase in spontaneous γ-H2AX foci positive for PCNA and are strongly sensitive to topoisomerase inhibition, but are proficient in HR [61, 62]. This result suggests that RecQL5 is more important in maintaining active DNA replication and preventing fork collapse [42]. FancM, and recently RTEL, have been implicated in aiding MCM2–7 progression past ICLs and unwinding difficult hairpin structures or fragile sites [63, 64], although neither are known to be normally associated with the replisome. Interestingly, a genetic truncating mutation of FancM has also found to be associated with POI and secondary amenorrhea with slightly milder developmental phenotypes than for MCM8 or MCM9 affected patients but with sensitivities to MMC and decreased monoubiquitination (mUb) of FancD2 [65].

Figure 4:

Pathways bridging DNA replication, restart, and recombination pathways where helicase and translocases will be required. Numbers and titles correspond to the text, and red arrows indicate possible steps for MCM8/9 contribution.

MCM8/9 do not interact directly with MCM2–7 nor are necessary for pre-replication complex (pre-RC) initiation [34], but rather associate with replisomes later during S-phase. In fact, targeted inactivation of MCM2-delpeted cells maintains DNA synthesis activity primarily through the presumed accessory helicase ability of MCM8/9 [66].

4.2. Fork Reversal.

Replication forks that are stalled or have built up significant ssDNA from unregulated DNA unwinding can be reversed to form ‘chicken foot’-like structures that anneals the nascent DNA strands and resembles a Holliday junction with a one-sided DSB (Figure 4). Fork reversal is fueled primarily by three DNA translocases with certain specificities, HLTF, ZRANB3, and SMARCAL1 [67]. SMARCAL1 seems to target stalled forks with stable RPA- bound ssDNA gaps at the fork junction specifically on the leading strand for direct reversal and annealing. This specificity prevents fork reversal of normal RPA coated lagging-strand gaps that occur normally during replication. ZRANB3 interacts specifically with and stimulates pUB-PCNA found at severely stalled forks [46]. Another translocase, HLTF, also promotes pUB-PCNA and can directly reverse forks via its HIRAN domain [68]. Finally, the major recombinase, Rad51, can participate in the initial phases of fork reversal, particularly when there is significant ssDNA in the reversed end [69, 70]. ssDNA at stalled forks can be engaged by Rad51 to promote pairing with the complementary strand to essentially strand exchange, forming a reversed fork. Other helicases including WRN, BLM, FancM, and FBH1 have also been linked with fork reversal processes [71–73].

Although no direct role for MCM8/9 has been implicated in fork reversal, both are present at the replication fork, known to interact with a subset of Rad51 (that is not bound to BRCA2), and are important for recruitment of Rad51 to DSBs [12]. Therefore, it is plausible that MCM8/9 may have a role in preparing the fork for reversal or recruiting Rad51 to a subset of reversed forks in concert with other translocases. In fact, patient cells deficient in MCM8 or MCM9 show large radial chromosome aberrations [17, 18] indicative of either stalled or reversed forks or other downstream HR intermediates that resulted in translocated and NHEJ processed chromosomes.

4.3. Reversed Fork Protection.

Excessive fork reversal can lead to uncontrolled resection by the various nucleases creating large genomic deletions. When replication forks are remodeled into alternative structures (i.e. reversed forks), they become excellent substrates for a variety of nucleases that can shuttle them into the DSBR pathways. Knockdown of FancJ causes an increase in DNA damage associated with induction of γ-H2AX, slower growth, and microsatellite instability [74, 75]. More recent evidence points to a fork protection role (Figure 4) for FancJ that suppresses HLTF and limits fork degradation [76]. BRCA1 and BRCA2 are now included as fork protectors that stabilize Rad51 filaments on the regressed arms, and when they are absent, forks are robustly degraded by Mre11 [77]. Rad51 can also stabilize the reversed fork by preventing branch migration or preventing access from nucleases [78]. Under situations where G-quadruplex structures are present at reversed/broken forks, the Pif1 helicase can stimulate Exo1, DNA2/BLM, or Mre11 degradation [79].

No role for MCM8/9 in fork protection has been described, but like Pif1, MCM8/9 can stimulate Mre11 resection (discussed below).

4.4. Fork Restart.

The mechanism of restart will likely require a DNA translocase/helicase to convert the four-way one-ended HJ to a three-way replication fork (Figure 4). In the presence of Toposiomerase I (TOP1) inhibited replication forks, RecQ1 can restart forks once PARP1 is degraded [80]. DNA2/WRN degrades reversed replication forks directly to restart unresolved forks in an aberrant manner [81]. ATR phosphorylates BLM to aid in restart of forks after stalling with hydroxyurea (HU) or aphidicolin (APH) [82]. Recently, FancJ was shown to counteract fork reversal by HLTF to reestablish replication forks [76]. Irreversibly damaged forks can only be rescued by endonuclease cleavage to provide substrates for HR. The helicase/nuclease activity of WRN [83] and the BLM dissolvasome complex (BLM/TopIIIα/Rmi1/Rmi2 - BTRR) are probable actors in this pathway. If all else fails, the endonucleases Mus81 or SLX4 can recognize extremely stalled forks and cause DSBs that feed into HR [84, 85].

The evidence for MCM8/9 in fork restart is not direct. Targeted degradation of the MCM2 does not fully halt DNA unwinding. Instead, MCM8/9 is thought to resume some DNA unwinding to process downstream HR intermediates [66]. Whether this unwinding by MCM8/9 occurs from a reversed or regular fork is unknown.

4.5. Resection.

Resection of DSBs or reversed forks that resemble DSBs are substrates for a variety of exonucleases (Figure 4). BLM can stimulate resection of DNA ends by either the helicase/nuclease, DNA2, or the MRN complex in the presence of RPA [86]. In another report, DNA2 was found to cooperate with either BLM or WRN to promote long range end resection [87]. In this role, DNA2 helps to recruit BLM and stimulates unwinding processivity, favoring 5’ strand cleavage by DNA2 resulting in the formation of 3’ ssDNA tails [88]. BLM/DNA2 also interacts with CtIP and/or the Topo IIIα–RMI1-RMI2 (TRR) complex enhancing DNA unwinding processivity for long range end resection [89]. Finally, FancJ (5’−3’) is thought to work together with BLM (3’−5’), each translocating on opposite strands to processively unwind DNA in the presence of RPA for resection by structure specific nucleases DNA2 or Exo1 providing a 3’-end of HR [90–92].

Interestingly, MCM8/9 are also found to be required for efficient end resection of DSBs in concert with the MRN complex [10]. Immunoprecipitation of MCM9 detected the MRN components (and vice versa) showing a direct interaction of these proteins. It may be that the exonuclease activity is stimulated when DNA helicases can provide ssDNA substrates or by enhancing cleaved DNA product release during resection or that the presence of MCM8/9 at a reversed fork can counteract fork protection by recruiting MRN. These possibilities remain to be distinguished.

4.6. Control of Synapsis.

Invasion of duplex DNA by a 3’ ssDNA overhang created after resection is facilitated primarily by Rad51 in combination with BRCA2. The formation of a stable Rad51 (or Dmc1 in meiosis) invaded D-loop structure is a trigger for HR and strand extension by various DNA polymerases (Figure 4). Antagonists to this process will include DNA helicases that can act as antirecombinases or checkpoints for maintaining homology over homeology in mitotic repair. Homeologous synapses are beneficial in creating genetic diversity required in meiosis, however they contribute to genomic instability in mitotic cells. Strand invasion intermediates can either dissociate after extension by a synthesis dependent strand annealing (SDSA) mechanism to create NCO or can be stabilized by more downstream DNA unwinding and subsequent dHJ formation promoting CO. Therefore, there will be a delicate balance between D-loop formation/dissolution, homeology/homology checkpoints, meiotic/mitotic recombination, and SDSA/dHJ. All of these processes will be regulated by various recombinogenic/antirecombinogenic DNA helicases with overlapping and unknown specificities to various intermediates and situations.

Therefore, and not surprisingly, many DNA helicases are involved in regulating synapsis. RecQL5 interacts with Rad51 to displace it from filaments acting as a antirecombinase to suppress D-loops and promote SDSA [61, 93]. Overlapping activities from FancJ, RTEL1, and FBH1 also all strip Rad51 from DNA to disrupt D-loops and suppress HR as an antirecombinase to favor SDSA [44, 94, 95].

The temporal placement of MCM8/9 in the HR pathway will perhaps provide insight into whether their role is prior to or after D-loop formation with Rad51. Currently, there are conflicting results showing that MCM8/9 acts prior to Rad51 (during resection) [10, 12] as well as downstream of Rad51 in DT40 cells, where knockout of MCM8 or 9 has no effect on DNA damaged induced Rad51 foci formation [11, 66]. A role for MCM8/9 in the regulation of Rad51 activity during HR remains to be definitively addressed. However, we hypothesize that in mammalian cells Rad51 regulation at sites of DNA damage prior to synapsis is dependent on an interaction with the MCM9 CTD. However, Rad51 has multiple functions during DNA replication and repair independent of MCM8/9. This would suggest a context-dependent interaction likely once a stalled replication fork progresses into collapse and subsequent HR repair.

It would be remiss to focus solely on helicases involved in mitotic HR over those that act more specifically in meiosis. For example, HFM1 acts as a meiosis specific DNA helicase (Mer3 homolog in yeast and of the DEAD family) and is a member of the ZMN group of proteins required for formation of the synaptonemal complex and promotion of CO in meiosis. HFM1^−/− mice are infertile, stages of meiosis are delayed, and recombination COs are reduced. The initial meiotic recombination steps (including DSB formation) appear unaffected, but synapsis is incomplete, suggesting a role for HFM1 in DNA unwinding downstream of D-loop formation and aiding in dHJ establishment [96]. These defects are highly similar to those observed with MCM8^−/− and MCM9^−/− mice which are also infertile, meiosis is blocked early in prophase 1, and females having arrested primary follicles and developing ovarian cancers [16, 97]. HELQ-1 is similar (albeit acting just after) in its role for late stage D-loop dissolution in meiosis [98], although it also associates with Rad51 paralogs during ATR-mediated ICL repair [99]. Even though Drosophila lacks an MCM9 homolog, REC is responsible for 95% of meiotic COs and its absence results in incomplete recombination transitioning towards SDSA and NCO products [15]. Further evidence in plants suggests that MCM8 is involved primarily with Rad51 in meiosis promoting NCO when the primarily CO pathway directed by Dmc1 fails [100].

Interestingly, many recombination (RecQ1, WRN, BLM, and FancJ) helicases [101–103] also interact with mismatch repair (MMR) proteins similar to MCM9 [104]. In this situation, DNA helicases will either direct MMR to homeoduplex pairing, remove homeologous duplexes recognized by MMR proteins in mitotic repair for NCO, or promote recombination and CO during meiosis [105, 106]. In mitotic DNA repair, homeologous duplex rejection would be a primary activity for DNA helicases. This is well established in E. coli where the UvrD helicase interacts with MMR proteins and dismantles RecA filaments to regulate homeologous recombination [107]. Both yeast helicases, Sgs1 and Srs2, interact with MMR proteins to suppress CO [108, 109], but the human helicase responsible to homeologous duplex rejection is not currently known. One candidate is BLM which also interacts with MMR proteins [110], but knockout of BLM in mammalian cells has no effect on the degree of homeologous recombination [111]. It is tempting to speculate that MCM8/9 is directed to MMR bound homeologous duplexes to either reject mismatched synapses for mitotic recombination or promote downstream dHJ formation for CO in meiosis.

4.7. HJ Branch Migration and Resolution.

dHJ formation requires the downstream unwinding of the dsDNA target for efficient extension of the D-loop and DNA synthesis (Figure 4). This activity is suppressed in mitotic HR but is promoted in meiotic recombination [112]. Dissolution of dHJ occurs though action of a higher order ‘dissolvasome’ consisting of the BTRR complex to release NCO [113, 114]. Resolution of the dHJ results in both NCO and CO and is processed primarily by two structure specific nuclease complexes: SLX-MUS-EME1 and GEN1 [115, 116]. Both resolvases are constrained by multiple cellular mechanisms until mitosis giving the BTRR complex priority in S-phase for NCO. However, in meiosis, there is sequential activation of SLX-MUS-EME1 followed by GEN1 to enhance CO products [84].

Prior to dissolution or resolution, DNA helicases can migrate synapses and HJs to further DNA synthesis or generate specific structures. RecQL1 can form a tetramer that recognizes and migrates HJs in a coordinated unwinding and annealing mechanism [117]. RecQL5 can also branch migrate HJs in a similar mechanism, although its main cellular role may be elsewhere aiding fork progression [118, 119]. Both BLM and WRN are important in later stages of HR by mediating the topological disentanglement of D-loops or HJ intermediates and suppressing CO products [113]. WRN can also migrate HJ on its own [120], however there is more support for WRN in regulating pathway choice towards alt-NHEJ earlier in this pathway [121]. BLM has enhanced unwinding processivity for branch migration and HJ dissolution within the BTRR complex to yield NCO [122]. Finally, ultra-fine bridges visualized during late stage mitosis are likely unresolved fork stalling or HR products (likely catenanes or hemi-catenanes) that are resolved by BLM/TopIIIα or GEN1 [123].

5. Conclusions and unanswered questions

With all of these DNA helicases (Table 1) acting similarly and in overlapping pathways, there is a question on whether there is redundancy of function or some yet unknown substrate specificity for each. Currently, there are more open questions than there are answers, which predicts an exciting time of research not just for MCM8/9 but also for distinguishing the specificities of the many replication/recombination helicases. There are clear links between germ line mutations in several DNA helicases that give rise to a variety of diseases including cancer, progeria, anemia, reproductive deficiencies, infertility, and an untold number of somatic mutations. The cellular and mechanistic effects of most of these helicase mutations are unknown, except in a few cases [124]. For MCM8/9, mutations can only be classified into categories such as damaging, variant of unknown significance (VUS), or benign [21] based on homology of the mutation, what is known about the structure/function relationship of the MCM family of proteins, and patient outcomes.

Of the numerous unanswered questions regarding the role of MCM8/9 in HR; the biggest of which are characterization of the enzymatic activity, interactions with other HR members, and specificity for particular HR intermediates. The potential for MCM8/9 to act during telomere maintenance as shown for many other related DNA helicases including BLM, WRN, RecQL4, and RTEL [125, 126], remains to be addressed. We are naive about how MCM8/9 is localized or recruited to replication forks or DSBs. It is unknown whether MCM8/9 aids in fork progression past difficult DNA secondary structures or sequences including G-quadruplexes or common fragile sites. Finally, no posttranslational modifications are known to regulate MCM8/9 function as are common with many other human DNA repair proteins.

There are multiple research priorities to pursue that should provide more insight into these scientific gaps of knowledge for MCM8/9 function compared with related DNA helicases. 1) Research should continue to characterize the mitotic DNA damage response and incorporate patient mutations for better functional characterization. 2) Efforts should be made to obtain purified protein complexes for structure/function studies to define substrate specificities. 3) Experiments should move past noting global meiotic deficiencies and instead should focus more specifically on recombination outcomes. 4) Compensation of helicase function should be examined by comparing patient cell lines to cells that are more abruptly knocked down/out. 5) DNA sequencing of mutations correlating with infertility and cancer should continue to be cataloged and functionally characterized. 6) Links between infertility (meiotic recombination) and cancer (DNA damage response) should be explored. These many open research questions will undoubtedly attract significant research attention in the coming years that will provide a much clearer picture for the role of MCM8/9 within the context of many other analogous DNA helicases.