Genomic instability in mice is greater in Fanconi anemia caused by deficiency of Fancd2 than Fancg

Abstract

Fanconi anemia (FA) results from mutations in the FANC genes and is characterized by bone marrow failure, birth defects and a high incidence of cancer. FANCG is a part of the FA core complex that is responsible for monoubiquitination of FANCD2 and FANCI. The precise role of the FA pathway is not well understood, although it may be involved in homologous recombination (HR), non-homologous end joining (NHEJ) and translesion synthesis (TLS). Fancd2^−/− mice have a more severe phenotype than Fancg^−/− and other FA core complex deficient mice, although both Fancg and Fancd2 belong to the same FA pathway. We hypothesized that Fancd2 deficiency results in a more severe phenotype because Fancd2 also has a FA pathway-independent function in the maintenance of genomic integrity. To test this hypothesis we determined the level of DNA damage and genomic instability in Fancd2^−/−, Fancg^−/− and wildtype controls. Fancd2^−/− mice displayed a higher magnitude of chromosomal breakage and micronucleus formation than wildtype or Fancg^−/− mice. Also, DNA strand breaks were increased in Fancd2^−/− but not in Fancg^−/− mice. In addition, Fancd2^−/− mice displayed an elevated frequency of DNA deletions resulting from HR at the endogenous p^un locus. In contrast, in Fancg^−/− mice, the frequency of DNA deletions was decreased. Thus, Fancd2 but not Fancg deficiency results in elevated chromosomal/DNA breakage and permanent genome rearrangements. This provides evidence that Fancd2 plays an additional role in the maintenance of genomic stability than Fancg, which might explain the higher predisposition to cancer seen in the Fancd2^−/− mice.

INTRODUCTION

Fanconi anemia (FA) is a rare, recessive (autosomal or X-linked) disorder characterized by progressive bone marrow failure, birth defects, and a markedly increased risk of cancer (1, 2). FA patients are particularly susceptible to developing acute myelogenous leukemia, but also develop solid tumors, often squamous cell carcinomas of the head and neck, and gynecological and gastrointestinal systems (3). The most prominent cellular phenotype of FA is hypersensitivity to DNA cross-linking agents, such as cisplatin, mitomycin C and diepoxybutane (4, 5). FA cells treated with DNA cross-linking agents display increased chromosomal breakage, radial chromosomes and other cytogenetic abnormalities that are commonly used as diagnostic tests for FA (6). Modestly elevated sensitivity to ionizing radiation (IR) and oxygen has also been observed in cultured FA cells (7, 8).

FA is a genetically heterogeneous disease. Currently 13 FA genes (A, B, C, D1, D2, E, F, G, I, J, L, M, and N) have been identified (9–11). Upon DNA damage multiple FANC proteins (FANCA/B/C/E/F/G/L/M and FAAP24/100) form a nuclear core complex that promotes monoubiquitination of the effector protein FANCD2 (12–15). Monoubiquitinated FANCD2 appears during the S-phase of the cell cycle and translocates into the nuclear foci formed at sites of DNA damage (16, 17). A recently identified FANCI forms an interdependent complex with FANCD2 and together localize to the nuclear foci following monoubiquitination by the FA core complex (11). The precise functions of FANC proteins are not well understood, although they appear to promote the repair of blocked or broken replication forks by several DNA repair pathways, which include HR, NHEJ and TLS. Co-localization of FANCD2 with HR proteins RAD51, BRCA1 and BRCA2, functional interaction of FANCD2 with BRCA2, and an important finding that the FANCD1 gene is identical to BRCA2 suggested the involvement of the FA pathway in HR (16–19). However, because BRCA2-dependent RAD51 nuclear focus formation is not affected in FANCD2 deficient cells and FANCD2 nuclear focus formation is normal in BRCA2 deficient cells, FANCD2 appears to be dispensable for RAD51-dependent HR (20). The role of FANC proteins in HR is difficult to understand, in part due to mixed study results showing that HR in FA cells is decreased (21–24), increased (25, 26) or not affected (20). Similarly, the role of the FA pathway in NHEJ is not established given the conflicting results that the FA pathway is either required for the efficient DNA end joining (27, 28) or not (21). There is substantial evidence that FANC proteins regulate TLS (23, 29). A recent report also suggests that the FA proteins facilitate DNA repair by promoting several pathways, such as HR, NHEJ and TLS (30).

Several mouse models for FA have been generated. These include the FA core complex subunit deficient mice, such as Fanca^−/−, Fancc^−/− and Fancg^−/− mice, as well as downstream effector deficient, Fancd2^−/− mice (31–35). Interestingly, Fancd2^−/− mice have a more severe phenotype than the FA core complex deficient mice (35). For example, Fancd2^−/− mice display perinatal lethality, micropthalmia and more severe hypogonadism than FA core complex deficient mice. Most notably, Fancd2^−/− mice but not Fanca^−/−, Fancc^−/− or Fancg^−/− mice are prone to developing epithelial (breast, ovarian and liver) cancers. To understand these differences and further investigate the role of Fanc genes in the maintenance of genomic stability we measured and compared DNA damage endpoints in the core complex deficient, Fancg^−/−, mice and the effector subunit deficient, Fancd2^−/−, mice. Namely, we determined the magnitude of chromosomal damage, DNA strand breaks and HR-mediated deletions in the knock-outs and their respective wildtype controls. In most reported studies, HR was measured in the synthetic reporter genes following artificial DNA double strand break (DSB) induction. In contrast to that, we studied spontaneous HR at an endogenous murine locus (36). This assay measures HR between two 70 kb tandem repeats at the p^un locus resulting in a deletion of one of the repeats (Fig. 1). Such deletions can occur by several HR subpathways, such as intrachromosomal crossing-over, single-strand annealing, unequal sister chromatid exchange and sister chromatid conversion. The deletion is visualized as a black pigmented cell or a clone thereof on the unpigmented retinal pigment epithelium (RPE) of the eye. We found that the frequency of HR-mediated deletions was decreased in Fancg^−/− mice and, in contrast, increased in Fancd2^−/− mice, as compared to wildtype littermates. In addition, Fancd2^−/− mice but not Fancg^−/− mice displayed significantly increased DNA damage and chromosomal instability. This suggests that Fancd2 has a more important role in the maintenance of genomic stability than Fancg.

Figure 1.

Homologous recombination leading to a DNA deletion in the murine p^un locus. A 70 kb deletion spanning exons 6–18 reconstitutes the wildtype p gene, which results in black pigment accumulation in the affected cells of RPE of the developing mouse. Schematic representation of the p^un locus and dissected RPE is shown.

MATERIALS AND METHODS

Mice

Fancd2 deficient mice were generated in Alan D’Andrea’s lab, Dana-Farber Cancer Institute, Harvard Medical School, Boston, USA and were in a mixed genetic background of 129/Sv and C57BL/6J (37)(K. Parmar and A. D’Andrea, unpublished data). Fancg deficient mice, were also generated in the same genetic background in Alan D’Andrea’s lab (129/Sv and C57BL/6J) (34). C57BL/6Jp^un/p^un mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). The C57BL/6Jp^un/p^un background is essentially identical to C57BL/6J with an exception of the recessive 70 kb internal duplication in the pink-eyed dilution (p) gene, termed the pink-eyed unstable (p^un) allele, resulting in a light gray coat color and pink eyes. The Fancd2 and Fancg mutations were crossed into the C57BL/6Jp^un/p^un genetic background 3 times. The resulting p^un/p^un Fancd2 ^+/− mice and p^un/p^un Fancg^+/− mice had a genetic background containing 93.75% of C57BL/6J and were morphologically similar to the parental C57BL/6Jp^un/p^un strain. Fancd2 and Fancg deficient mice and wildtype controls were obtained from heterozygous crosses. We also attempted to generate double mutant, Fancd2^−/−Fancg ^−/− mice (by intercrossing double heterozygous Fancd2 ^+/− Fancg^+/−mice) but no live double mutant offspring were obtained (p = 0.01 as compared with the expected Mendelian segregation, total number of offspring = 99) due to apparent embryonic lethality resulting from combined inactivation of Fancd2 and Fancg.

Mice were bred in an institutional specific pathogen-free (SPF) animal facility under standard conditions with a 12 h light/dark cycle and were given a standard diet (Harlan Teklad No 8656) and water ad libitum. Pregnancy was timed by checking for vaginal plugs, with noon of the day of discovery counted as 0.5 days post coitum (dpc). Similarly, the time of birth of a litter was timed with the noon of discovery counted as 0.5 days post partum (dpp).

Chromosome breakage analysis

Primary fibrobast cultures were established from individual mouse ear clips. The ear clips were rinsed in sterile HBSS supplemented with antibiotics (1% Penn-Strep) media and digested in 2–3 ml of collagenase (500 U/mL) solution at 37°C for up to 120 minutes. The digested cells were washed and plated on in-situ coverslips in RPMI 1640 media containing 10% serum. After 7–10 days of primary growth, standard cytogenetics methods were used to harvest the cells and obtain metaphases. The metaphases cells were analyzed with Giemsa-trypsin (GTG) banding and all available metaphase cells from four dishes were analyzed. The number of these cells varied from each animal but usually ranged from 20–50 in number. Chromosome and chromatid breaks and gaps were scored, recorded and imaged on a Zeiss Axiophot microscope. The total number of cells and breaks/cell analyses were done in a blinded fashion.

In vivo micronucleus assay

The in vivo micronucleus assay was performed in normochromatic erythrocytes from peripheral blood (38). Blood was collected from the mouse facial artery and 3 μL was spread onto microscopic slides for staining with Giemsa (Sigma-Aldrich, St. Louis, MO). Two slides per mouse were prepared. In total, 6000 cells per mouse (3000 cells per slide) were analyzed in a blinded fashion.

The comet assay

The comet assay was performed essentially as described elsewhere (39). Blood drawn from the mouse facial artery was diluted 1:50 with phosphate buffered saline (PBS). Sixty μl of blood suspension was mixed with 200 μl of 1% low melting point (LMP) agarose in PBS, ph 7.4, at 37°C and 25 μl aliquots of this mixture were loaded in two replicate molds imprinted on Gelbond film (FMC Bioproducts, Rockport, ME) precoated with 1% normal melting point agarose in PBS. The agarose was allowed to set for 5 min at 4°C and the slides were incubated in lysis solution (2.5 M NaCl, 10 mM Tris, 100 mM Na₂EDTA, pH 10.0, with 1% v/v Triton X-100 and 10% DMSO added fresh) at 4°C for 1 h to remove cellular proteins. The Gelbond film was removed and placed in a horizontal electrophoresis chamber (BioRad, Hercules, CA) filled with fresh, chilled electrophoresis buffer (300 mM NaOH and 1 mM EDTA, ph 13) and left at 4°C for 20 min to allow for DNA unwinding. Electrophoresis then was performed in the same buffer at 20V and 300 mA at 4°C for 40 min. After electrophoresis the Gelbond film was washed three times at 4°C for 5 min each with neutralizing buffer (400 mM Tris-HCl, ph 7.5) before staining with SYBR Gold (1/10,000 dilution of stock solution from Molecular Probes, 495 nm excitation, 537 nm emission). The Gelbond film was rinsed with distilled water and visualized using a fluorescent microscope (Olympus Ax70, Tokyo, Japan; max absorption 495 nm; max emission: 520 nm) through a 10x objective lens. Comet images were captured with a camera (Model 11.2 Color Mosaic, Diagnostic Instruments Inc., USA) and analyzed with the CASP image analysis system (40)(http://casp.sourceforge.net). Several features for each comet were determined including percentage of DNA in the comet tail, tail length, comet length, and tail moment. The tail moment (TM), which is defined as the product of percentage of DNA in the comet tail and tail length, was used to quantify the extent of DNA damage. TM is considered one of the best indices of comet formation by computerized analysis (41). About one hundred randomly captured comets (~ 50 comets on each two replicate molds) per mouse were graded according to degree of damage into 5 classes and given a value 0 – 4 (0, undamaged, 4, maximally damaged). The overall score or damage index for 100 cells can range from 0 (all undamaged) to 400 (all maximally damaged). Because of inter-experimental variability in the background level, the damage index was normalized to a control. The same sex wildtype littermate served as a control for each mutant mouse. Statistical analysis was performed on the actual damage index values prior to data normalization.

HR-mediated deletion assay

We determined the frequency of HR events between two 70 kb DNA repeats in the pink-eyed unstable (p^un) locus spanning exons 6–18 of the pink-eyed dilution (p) gene resulting in a deletion of one of the repeats (Fig. 1) (36). The p^un allele is reverted to the functional p gene by such a deletion event, allowing black pigment accumulation in the cells of the RPE of the eye of developing p^un/p^un mice. Such HR-mediated deletions or p^un reversions can be visualized microscopically after birth, as described in the following section.

Dissection of the retinal pigment epithelium

Offspring were sacrificed at 20 days of age and their eyes were dissected. Whole mount RPE slides were prepared for microscopic analysis of eye-spots. Eyes were processed to expose the RPE layer as previously described (36). The eye was removed from its orbit and immersed in fixative (4 % paraformaldehyde in 0.1 M phosphate buffer, pH 7.4) for one hour and then in PBS until dissection. An incision was made at the upper corneo-scleral border to allow removal of the cornea and lens. To flatten the eye-cup, six to eight incisions were made from the corneo-scleral margin towards the centrally positioned optic nerve, and the dissected eyecup was placed on a glass slide with the retina facing up. The retina was then gently removed and the residual specimen consisting of sclera, choroid and RPE, with the RPE facing up, was mounted in 90% glycerol and visualized under microscope using 10 x objective.

Scoring of HR-mediated deletions visualized as eye-spots

A pigmented cell or a group of adjacent pigmented cells separated from each other by no more than 5 unpigmented cells was considered as one eye-spot that resulted from one HR-mediated deletion (p^un reversion) event (42). The number of eye-spots in each RPE was counted. Positions of eye-spots in the RPE were mapped.

The RPE was scanned with a DC120 digital camera (Eastman Kodak Company) mounted on a DMLB microscope (Leica Microsystems, Inc.) using a 2.5 x N-plan objective. The images were assembled and examined in Adobe Photoshop 5.0 on a Macintosh Power Computer. All data were stored and processed with Microsoft Excel.

Distance analysis of eye-spots from the optic nerve

Eye-spots were identified under the microscope and compared with their scanned digital images. Distances were measured with the Adobe Photoshop 5.0 Measurement Tool. Two distances were measured for each eye spot: the eye-spot distance is the distance from the center of the optic nerve head to the most proximal edge of the eye-spot, and the RPE distance of an eye-spot is the distance from the optic nerve through the eye-spot to the outer edge of the RPE. Dividing the eye-spot distance by the radius gives the proportional distance of each eye spot from the center of the RPE, or its position. The position of each eye-spot was determined in this manner to compensate for minor differences in the size of the eyes.

Statistical analysis

Comparison between events in different groups was done by a Student t-test. P < 0.05 was considered significant.

RESULTS

Fancd2^−/− mice displayed higher frequency of chromosomal breaks than Fancg^−/− mice

Metaphase spreads of fibroblasts showed that Fancg^−/− mice had somewhat elevated frequency of chromosomal breaks than wildtype controls but it did not reach statistical significance (Table 1). In contrast, Fancd2^−/− mice had significantly elevated frequencies of chromosomal breaks as compared to their wildtype controls (p < 0.0001). When Fanc mutants were compared with one another, Fancd2^−/− mice had 4.4-fold more chromosomal breaks than Fancg^−/− mice (p < 0.001) (Table 1). Thus, Fancd2 deficiency resulted in a more pronounced chromosomal instability than Fancg deficiency.

Table 1.

Chromosomal breakage analysis in Fancg^−/− and Fancd2^−/− mice.

Genotype	Chromosomal breaks per cell	number of mice	P, mutant vs. wildtype
Fancg−/−	0.26 ± 0.25	4	NS
Fancg+/+	0	3
Fancd2−/−	1.15 ± 0.12	4	< 0.0001
Fancd2+/+	0	3

Mean ± SD is shown; p < 0.001, Fancg^−/− versus Fancd2^−/− mice; NS, not significant

Micronucleus formation was increased in Fancd2^−/−mice but not Fancg^−/− mice

Micronucleus formation was determined in normochromatic peripheral blood erythrocytes, as a measure of chromosomal damage. Fancg^−/− mice had 23% more micronuclei than Fancg^+/+ mice, but was not significantly increased (p > 0.05) (Fig. 2). The difference was also not statistically significant when both wildtype controls (Fancg^+/+ and Fancd2^+/+)were combined into one control group. In contrast, the micronucleus frequency was significantly increased (by 40%) in Fancd2^−/− mice compared with Fancd2^+/+ mice (p < 0.05) and combined controls (p < 0.01) (Fig. 2). Although, Fancg deficient mice displayed a somewhat increased frequency of micronuclei, Fancd2 deficiency led to a more severe chromosomal damage than Fancg deficiency, similar to chromosomal breakage data reported here.

Figure 2.

The frequency of micronucleus formation in Fancg deficient mice, Fancd2 deficient mice and respective wildtype controls. Mean ± SEM (error bars) is shown, n = 7, Fancg^+/+, n = 8, Fancg^−/−, n = 7, Fancd2^+/+, n = 8, Fancd2^−/− mice, *, p < 0.05, compared to respective wildtype controls. In total 6000 RBC/mouse were analyzed.

DNA strand breaks were increased in Fancd2^−/− but not in Fancg^−/−mice

DNA strand breaks were determined by the alkaline comet assay in whole blood collected from Fancg^−/− and Fancd2^−/− mice and respective wildtype littermate controls. Fancg^−/− micehad a similar frequency of DNA strand breaks as Fancg^+/+ mice(Fig. 3). Fancd2^−/− micehad 1.8-fold more breaks than their respective wildtype controls and Fancg^−/− mice (Fig. 3). This observation is consistent with the data from Table 1 and Fig. 2 showing that Fancd2 deficiency has a more severe phenotype in terms of DNA damage than Fancg deficiency.

Figure 3.

The frequency of DNA strand breaks in Fancg deficient mice, Fancd2 deficient mice and respective wildtype controls. Mean ± SEM (error bars) is shown, n = 5 mice per genotype, *, p < 0.05, compared to the respective wildtype control. Damage index was calculated using TM and was normalized to controls.

HR-mediated deletions were increased in Fancd2^−/− mice and decreased in Fancg^−/− mice

The frequency of HR events resulting in 70 kb DNA deletions was determined in Fancg^−/− mice, Fancd2^−/− mice and wildtype littermate controls as the number of eye-spots per RPE (Fig. 4). The frequency of eye-spots was reduced by 36% in Fancg^−/− mice (p < 0.001) and, in contrast, increased by 30% in Fancd2^−/− mice (p < 0.01) compared to respective wildtype littermates. When Fanc mutants were compared with each other, Fancd2^−/− mice had 48% more eye-spots than Fancg^−/− mice (p < 0.001).

Figure 4.

The frequency of eye-spots in the RPE of Fancg deficient mice, Fancd2 deficient mice and respective wildtype controls. Mean ± SEM (error bars) is shown: n= 37, Fancg^+/+, n = 30, Fancg^−/−, n = 20, Fancd2^+/+, n = 20, Fancd2^−/− RPEs. ***, p< 0.001, **, < 0.01 compared to respective control. A dashed line indicates a significant difference (***, p< 0.001) between Fancg^−/− and Fancd2^−/− mice.

Because the number of eye-spots did not differ significantly between Fancg^+/+ and Fancd2^+/+ mice, they were combined into one control group. When compared to combined wildtype controls, Fancg^−/− mice still had significantly fewer eye-spots (p < 0.001) and Fancd2^−/− significantly more (p < 0.01). The average number of eye-spots/RPE was 6.35 ± 0.34 (n = 57), 4.27 ± 0.28 (n = 30) and 8.25 ± 0.6 (n = 20) in combined wildtype controls, Fancg^−/− mice and Fancd2^−/− mice, respectively.

These findings demonstrated that a deficiency in Fancg versus Fancd2 has different consequences in spontaneous HR involving direct repeats: Fancg deficiency reduces HR resulting in the frequency of deletions below the normal range, while Fancd2 deficiency increases HR resulting in elevated frequency of permanent genome rearrangements (i.e. DNA deletions).

Positional distribution of HR-mediated deletions in Fancg^−/− and Fancd2^−/− deficient mice

The mouse RPE develops during a second half of gestation in a radial pattern, away from the optic nerve head to the edge of the RPE (Fig. 1) (36). We mapped relative positions of eye-spots within the RPE to determine the approximate time during embryo development at which an HR event occurred (36). For that we divided the RPE into 10 concentric regions. The concentric region closest to the optic nerve head, ‘0.0–0.1’, corresponds to the beginning of the RPE development, while the most remote region, ‘0.9–1.0’, indicates the end of RPE development (Fig. 1). We determined the position of eye-spots in the RPE of Fancg^−/− mice, Fancd2^−/− mice and wildtype littermate controls. Fancg^−/− mice had fewer eye-spots than Fancg^+/+ mice in most (9 of 10) regions of the RPE (Fig. 5A). In contrast, Fancd2^−/− mice had more eye-spots than Fancd2^+/+ mice in most regions of the RPE (Fig. 5B). The same pattern was observed when mutant mice were compared with a combined wildtype control. When mutants were compared with each other, Fancd2^−/− mice had ~2-fold (2.3 ± 0.4) more eye-spots in a given RPE region than Fancg^−/− mice in most (8 of 10) regions. Thus, in Fancg^−/− mice, HR was reduced and, in Fancd2^−/− mice, HR was increased continuously during RPE development.

Figure 5.

Positional distribution of eye-spots in the RPE of Fancg deficient mice (A) and Fancd2 deficient mice (B) and their respective controls. Mutant mice are indicated in black bars, wildtype controls are shown in open bars.

DISCUSSION

Non-epistatic interaction of Fancg and Fancd2

We found that mice deficient in the FA core complex subunit, Fancg, and the downstream effector, Fancd2^−/−, have distinct phenotypes in terms of spontaneous chromosomal instability and HR. Chromosomal and DNA damage was significantly elevated in Fancd2^−/− but not Fancg^−/−mice. Also, the outcome of HR at the endogenous p^un locus was different: HR-mediated deletions were increased in Fancd2^−/− mice and reduced in Fancg^−/− mice. In this study, we also attempted to produce double mutant mice but encountered embryonic lethality (data not shown). These findings were unexpected given that both Fancg and Fancd2 are the members of the FA pathway and a deficiency in one or another gene or both genes of the same pathway presumably leads to similar outcomes. The observed differences reported here suggest that Fancg and Fancd2 are non-epistatic genes. That is the Fancd2 protein plays another role in the maintenance of genomic stability than just being a downstream effector of the FA pathway. In support of this notion, it has been demonstrated in tissue culture experiments that FANCD2 participates in the intra S-phase checkpoint in an FA core complex-independent manner in response to IR (43). FANCD2 is phosphorylated by ATM to halt DNA replication. FANCD2 deficient cells display radioresistant DNA synthesis, a phenotype of ATM deficient cells, while other FA complementation groups do not (43). Thus, pronounced chromosomal damage and hyperrecombination in Fancd2^−/− mice observed in this study may result from an intra S-phase checkpoint defect. In support of this, elevated HR at the p^un locus has also been observed in mice characterized by impaired cell cycle checkpoint control, such as Atm and p53 deficient mice (44).

Fancd2 may also have other functions that are independent of the FA core complex. It has been demonstrated that unmodified Fancd2 binds directly and specifically to DNA double strand ends and Holliday junctions in vitro (45). Fancg^−/− mice contain non-ubiquitinated Fancd2 protein (34), which would allow for the latter Fancd2-DNA interaction. However it’s not clear if unmodified Fancd2 binds directly to DNA double strand breaks and/or enhances its repair in vivo. On the other hand, Fancd2 can be modified by other cellular proteins. In particular, Fancd2 is subject to phosphorylation by CHK1 kinase (46). This phosphorylation occurs independently of the FA core complex and is important for interaction with BRCA2 (46). It’s tempting to speculate that Fancd2 plays a role in the maintenance of genomic stability through interactions with DNA or other DNA repair proteins, however, further studies are needed to characterize the FA-pathway independent Fancd2 functions.

Taken together, the observed phenotypic differences in Fancd2^−/− and Fancg^−/− mice and embryonic lethality of double mutant mice suggest a non-epistatic interaction between Fancg and Fancd2.

Comparison of HR assays in our and other studies

The mild defect in HR in Fancg^−/− mice in our study is in agreement with studies that showed reduced HR repair of I-SceI generated DNA DSBs in chromosomally integrated substrates in FA patient-derived FANCG cell lines (21) and Fancg deficient chicken DT40 cells (22). In contrast, increased HR in Fancd2^−/− mice observed in this study conflicts with the studies reporting that HR repair in FANCD2 deficient cells is either decreased (21, 24) or unaffected (20). However, these findings are difficult to compare given that not only different experimental systems but also different HR assays were used. The published studies were carried out in experimental systems ranging from chicken DT40 cells (24) to FA patient derived cells (20, 21), while our study was done in knock-out mice. Also, previous studies measured the repair of DSBs generated by I-SceI enzyme in the synthetic reporter genes, while, in the current study we determined frequencies of spontaneous HR events at the endogenous murine p^un locus. The previously reported HR assays measure repair of a very specific subset of DNA damage, namely repair of a DSB following enzymatic digestion, while the HR assay used in this study can detect a wider variety of DNA strand breaks including those caused by oxidative stress and replication fork arrest.

The high frequency of DNA breaks found in Fancd2^−/− mice can promote recombination resulting in deletions. This is a likely scenario in Fancd2^−/− mice. The Fancd2 protein seems to be dispensable to complete HR processes under these circumstances (high chromosomal/DNA instability). In Fancg^−/− mice, the lower frequency of DNA strand breaks was associated with lower frequency of deletions. In fact, the deletion frequency fell below the normal range possibly because of a mild defect in the HR machinery. Thus, the role of Fanconi genes, Fancd2 and Fancg, in the HR outcome may depend on the frequency of DNA strand breaks.

Hyperrecombination in Fancd2 deficient mice may play a role in carcinogenesis

Mouse models for FA revealed that Fancd2^−/− mice have a more severe phenotype than the FA core complex deficient mice, such as Fanca^−/−, Fancc^−/− and Fancg^−/− mice (35). Most notably, Fancd2 deficient mice but not Fancg deficient have an increased incidence of epithelial cancers, such as breast, ovarian and liver cancers. In this study, Fancd2^−/− mice displayed elevated chromosomal instability and an increased frequency of HR-mediated DNA deletions. In contrast, Fancg^−/− mice did not show significantly increased genomic instability. Therefore the elevated frequency of chromosomal damage and aberrant HR resulting in DNA deletions may play a role in carcinogenesis in Fancd2 deficient mice. Chromosomal instability and abnormally increased HR has been associated with carcinogenesis in experimental systems and human cancer-prone syndromes (47, 48). For example, cancer-prone Atm deficient mice and p53 deficient mice display a hyperrecombination phenotype at the same genomic locus and to similar extent as Fancd2 deficient mice in this study (44). Thus, the observed aberrant HR may account for cancer development in Fancd2^−/− mice.