Early-Onset Lymphoma and Extensive Embryonic Apoptosis in Two Domain-Specific Fen1 Mice Mutants

The hallmark of most cancers is chromosomal instability, caused by defects in cellular processes (1). Flap endonuclease 1 (FEN1) is associated with many of the processes that help to maintain the integrity of the genome. FEN1 is a multifunctional and structure-specific endonuclease, with roles in DNA repair, replication, and recombination (2). FEN1 recognizes and cleaves single-stranded DNA 5′-flap substrates, which are key intermediates in DNA metabolism. Flap removal by FEN1 is required for the processing of Okazaki fragments (3), long-patch base excision repair (4), and homologous recombination (5). FEN1 interacts with proliferating cell nuclear antigen (PCNA), among other proteins (6). The nuclease activity of FEN1 is greatly improved when bound to PCNA on 5′ flaps and nicked, duplex DNA substrates (5, 7) and the ³³⁷QGRLDDFFK³⁴⁵ motif is required for this binding (8, 9). Replacing the F343 and F344 residues in FEN1 with glycine and alanine residues, respectively, completely eliminates the physical FEN1-PCNA interaction in vitro (10).

The critical role of FEN1 in maintaining genome integrity is shown in studies with S. cerevisiae. In these studies, a deletion of RAD27 (the S. cerevisiae homologue of FEN1) exhibits a complex phenotype, including a marked increase in the rate of spontaneous mutations (11). Based on these studies, a FEN1 deficiency is proposed to contribute to several genetic diseases, including cancer and triplet expansion diseases (11, 12). A targeted deletion of the Fen1 gene in mice causes early embryonic lethality due to proliferation failure (12, 13). Haploinsufficiency of Fen1 in mice, combined with the adenomatous polyposis coli gene (Apc), results in microsatellite instability and can lead to tumor progression (12), thus classifying Fen1 as a novel tumor suppressor gene (14).

To examine the role of FEN1 in DNA repair and tumorigenesis, we have generated several mouse models with mutations in the Fen1 gene. We present here two gene-targeted mice carrying site-specific mutations in the Fen1 gene. The first mutant mouse (Fen1^E160D/E160D) contains a single amino acid mutation in the endonuclease domain of FEN1 (Fig. 1A and C). The second mutant mouse contains a two-amino-acid point mutation (FEN1 F343G/F344A) in the highly conserved PCNA interaction domain of FEN1 (referred to as Fen1^{ΔPCNA/ΔPCNA}). The rationale for the generation of these two models was based on the biochemical and genetic characterization of mutant enzymes (10, 15–17). The Fen1^E160D/E160D mice develop a considerably elevated incidence of B-cell lymphomas beginning at 6 months of age, particularly in the females (Fig. 1B). Fen1^{ΔPCNA/ΔPCNA} embryos show extensive apoptosis in the forebrain and vertebrae area and die around stage E9.5 to E11.5.

Generation of targeted Fen1^E160D/E160D and Fen1^{ΔPCNA/ΔPCNA} mutant mice. See Supplementary Table S1 and Supplementary Figs. S1 to S3.

Morphologic and histologic analyses of mouse embryos. To investigate the stage at which the homozygous Fen1^{ΔPCNA/ΔPCNA} embryos die, mouse embryos derived from Fen1^wt/ΔPCNA × Fen1^wt/ΔPCNA crosses were obtained at several stages of gestation, including E8.5, E9.5, E10.5, E11.5, and E12.5. For E9.5, E10.5, and E11.5, at least four litters were collected per stage. All dissections were done in 1× PBS, and the embryo phenotype was studied by light microscopy.

Apoptosis in Fen1^{ΔPCNA/ΔPCNA} embryos. Deciduas from timed Fen1^wt/ΔPCNA heterozygous mating were isolated at E9.5 to E12.5 days postcoitum. Embryos were frozen in Tissue-Tek optimum cutting temperature compound at −80°C and cryosectioned at −10°C (5–10 μm thick). Total and apoptotic nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) and the In Situ Cell Death Detection Kit, TMR red [terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining; Roche], respectively. Five individual knock-in embryos and three individual wild-type embryos were stained two to six times each, with two to three embryo sections on each slide. Stained sections were analyzed with a Zeiss Axioplan 2 microscope and pictures taken with an AxioCamHR.

High- and low-density growth experiments. For the high-density experiments, the Fen1^E160D/E160D and wild-type mouse embryo fibroblasts (MEF) were plated at 2 × 10⁵ per 6-cm dish. Cells were counted when they reached confluence and replated at 2 × 10⁵ per 6-cm dish. For the low-density experiments, Fen1^E160D/E160D and wild-type MEFs were plated at a density of 0.5 × 10⁵ per 10-cm dish and their growth rate was monitored by daily counting for 10 d.

Serum starvation and thymidine incorporation. Fen1^E160D/E160D and wild-type MEF cells were plated in triplicate at 2 × 10⁵ per 3.5-cm dish. After 48 h, the cells were washed twice with PBS and then incubated in DMEM containing 0.1% FCS for 72 h. The cells were fed with DMEM containing 10% FCS. At each time point, the cells were incubated with 5-μCi [³H]thymidine for 1 h at 37°C, washed with PBS, and harvested. [³H]thymidine incorporation was quantified.

FEN1 endonuclease activity on a synthetic DNA flap substrate. Nuclear extracts used for the flap assays were prepared as described previously (18). The flap assay was done with the oligonucleotides sequences provided in Supplementary Table S1, by the method previously described (19).

Histopathology. Full necropsy was done on mice that were found dead after they spontaneously developed disease and mice that were sacrificed due to signs of disease or discomfort. The following tissues were collected and immersion fixed in 10% neutral-buffered formalin: skeletal muscle, skin, mesenteric lymph node, liver, small and large intestines, pancreas, lungs, spleen, kidney, heart, thymus, adrenal gland, pituitary gland, and brain. Any macroscopic abnormalities were noted. After fixation, tissues were trimmed, embedded in paraffin, sectioned at 5 μm, and stained with H&E and Giemsa stain. To evaluate T-cell (CD3) and B-cell (CD45R) populations within the neoplasms, immunohistochemistry was done on sections of the spleen, thymus, mesenteric lymph node, and any macroscopic abnormalities, as previously described. All tissues were examined by a veterinary pathologist using light microscopy. Lymphoid neoplasms were diagnosed using the Bethesda proposals for classification of lymphoid neoplasms in mice (20).

Somatic hypermutation and in vitro switching. The somatic hypermutation was done on Peyers patches of 4- to 6-mo-old wild-type and Fen1^E160D/E160D mice as described previously (21). The efficiency of in vitro class switching in wild-type and Fen1^E160D/E160D spleen B cells was measured as described (22). See Supplementary data for further details.

Comparative genomic hybridization. See Supplementary data.

Mouse FEN1 structural model. A model of mouse FEN1 interacting with PCNA was generated with standard comparative modeling using SWISS-MODEL (23), based on the crystal structure 1UL1 of the human FEN1-PCNA complex from Sakurai and colleagues (24). A model for the active site of the protein including bound water and the two active-site Mg²⁺ ions was generated from the Methanococcus jannaschii FEN1 crystal structure 1A77 of Hwang and colleagues (25). The protein structure illustrations were prepared with PyMOL.

Targeting strategy for Fen1^E160D/E160D and Fen1^{ΔPCNA/ΔPCNA} mutant mice. To circumvent the embryonic lethality that results from deleting the entire Fen1 gene, we have generated two alternative Fen1 mutant mice (Supplementary Table S1; Supplementary Figs. S1–S3). In the first, we mutated a single active-site residue (E160D) in the nuclease core domain of FEN1; in the second, we mutated two residues of FEN1 that are essential for the FEN1-PCNA interaction (F343G/F344A; Fig. 1A and C). The eight active-site residues of the FEN1 nuclease domain (Fig. 1C) are absolutely conserved from archaea to yeast, nematodes, and vertebrates. These active-site residues coordinate at least two divalent cations in a highly conserved geometry. This geometry makes the nuclease domain of FEN1 ideal for binding DNA and the nucleophilic attack of phosphodiester bonds in the DNA-flap backbone. The E160D mutation has previously been shown to reduce the in vitro endonuclease activity of FEN1 to around 10% to 20%, relative to the wild-type enzyme, whereas the PCNA interaction domain remains intact and functional (15). The two mutated phenylalanines in FEN1 [i.e., the conserved residues F343 and F344 of the classic PCNA-interacting protein motif (PIP-box); ref. 8] are absolutely required for the PCNA interaction (10). These phenylalanines in FEN1 are inserted into a hydrophobic pocket on the surface of PCNA (Fig. 1C). The details of the site-directed mutagenesis and gene targeting are included in full in Supplementary data. The targeted Fen1 loci were expressed at wild-type levels, and the targeting did not affect the expression of other genes approximate to the targeted Fen1 gene (Supplementary Table S2; Supplementary Fig. S4).

The PCNA interaction domain of FEN1 is critical for full neonatal viability. Interbreeding of heterozygous mice carrying the FEN1 F343G/F344A residues failed to produce homozygous Fen1^{ΔPCNA/ΔPCNA} mice. We observed that at E10.5, all Fen1^{ΔPCNA/ΔPCNA} embryos were drastically growth retarded, showing significantly reduced overall size and defective development as compared with their littermates (Fig. 2A). They were characterized by an open neural fold, pale appearance, the lack of the E8.5 mouse embryo changing its body bending pattern of the abdomen and dorsal parts, from a U-shaped inverted blastoderm to a C-shaped body by stage E9.5, and a high degree of apoptosis (Fig. 2). At E11.5 to E12.5, only dead or massively growth-retarded Fen1^{ΔPCNA/ΔPCNA} embryos were evident, with most already in the process of resorption (Supplementary Table S3). Attempts to establish MEF cell lines from E10.5 Fen1^{ΔPCNA/ΔPCNA} failed, despite several attempts. In parallel experiments, cell lines were routinely established from the wild-type and heterozygous littermates. No living Fen1^{ΔPCNA/ΔPCNA} embryos were found at or after stage E12.5, when a total of 193 embryos from heterozygous matings were analyzed (Supplementary Table S3). Mice heterozygous for the F343G/F344A mutation were fertile and phenotypically indistinguishable from wild-type mice.

E160D cells have a proliferation defect under low-density culture conditions. In contrast to the embryonic lethal phenotype observed in the Fen1-null mice, interbreeding of Fen1^wt/E160D led to the recovery of the expected genotypes with Mendelian ratios (data not shown). A DNA 5′-flap endonuclease assay confirmed that the activity of the FEN1 E160D protein is reduced to ∼20% compared with the wild-type enzyme (ref. 17; Fig. 3A), in good agreement with the in vitro biochemical assays. MEF cells from wild-type and Fen1^E160D/E160D littermates were cultured under high- or low-density conditions to determine if the E160D mutation affected the rate of cellular proliferation. At high cell density, the growth rate of Fen1^E160D/E160D cells was comparable to, but lower than, the wild-type control cells. The cellular duplication rate was ∼12 hours for wild-type MEF cells and 17 hours for the Fen1^E160D/E160D MEF cells (data not shown). The proliferation defect of the Fen1^E160D/E160D cells became considerably more apparent under low-density culture conditions (Fig. 3B , left). The extent of apoptosis observed in the wild-type MEFs was similar to the level of apoptosis in the Fen1^E160D/E160D MEFs. These results suggest that the proliferation defect observed in the Fen1^E160D/E160D MEF cells is due to a defect in cell division rather than the induction of apoptosis or premature senescence. These results also support the hypothesis that the endonuclease activity of FEN1 is essential during Okazaki fragment maturation.

Fen1^E160D/E160D MEF cells have a cell cycle defect. To understand the nature of the Fen1^E160D/E160D proliferation defect, we compared the cell cycle progression of wild-type MEFs to that of Fen1^E160D/E160D MEFs. The cells were serum starved for 72 hours before serum addition giving a coordinated reentry into the cell cycle. After the addition of serum, the MEFs were harvested at regular intervals and labeled with [³H]thymidine for 1 hour to monitor DNA synthesis (Fig. 3B , right). The wild-type MEFs began incorporating [³H]thymidine 8 to 12 hours after serum stimulation and showed maximal levels of [³H]thymidine incorporation at 16 to 20 hours. In contrast, the Fen1^E160D/E160D MEFs showed significantly reduced levels and a reduced rate of [³H]thymidine incorporation.

Fen1^E160D/E160D mice develop obesity in the absence of insulin resistance or diabetes. Within the first generation, approximately half of the Fen1^E160D/E160D mice, both males and females, developed marked obesity by 9 months of age, compared with the age-matched wild-type controls (Fen1^E160D/E160D, 48 ± 3 g; wild-type control, 31 ± 1 g; P < 0.001; Fig. 3C). This difference in body weight was even greater at 16 months (Fen1^E160D/E160D, 65 ± 5 g; wild-type control, 39 ± 1 g; P ≤ 0.001). Fen1^E160D/E160D and wild-type control mice were raised under identical conditions and fed standard rodent chow.

A marked increase in the amount of white adipose tissue was seen in the subcutis, retroperitoneum, abdomen, mediastinum, and around the epididymes on necropsy (Fig. 3C , left). All other organs examined, including the interscapular brown adipose tissue, seemed of similar size to those of control mice. Histologic examination of adipose tissues revealed a minimal/mild hypertrophy of the white adipose cells and no changes in the brown adipose tissue. Analysis of the glucose, insulin, and leptin levels revealed the obese Fen1^E160D/E160D mice to be normoglycemic, normoinsulinemic, and normoleptinemic (data not given), indicating that the observed obesity was not associated with an insulin resistance or type 2 diabetes mellitus. Supporting this hypothesis, we noted no histologic changes in the size of the pancreatic islets in the Fen1^E160D/E160D mice compared with age-matched, wild-type controls.

Female Fen1^E160D/E160D mice develop lymphoid hyperplasia and lymphoma with multiorgan involvement. Macroscopic evaluations of the Fen1^E160D/E160D mice revealed that most of the female mice (Fig. 4A) had developed a diffuse enlargement of the spleen, which, when cross-sectioned, showed multifocal to coalescing white areas (Fig. 4; Table 1). The lung (97% of cases), liver (93%), kidneys (90%), mesenteric lymph nodes (62%), and thymus (55%) commonly showed enlargement or focal changes in association with the splenic enlargement (Fig. 4B and C). The lung, liver, and kidneys contained multifocal, white nodular foci randomly distributed through the organs, whereas mesenteric lymph nodes and thymus were often diffusely enlarged. Microscopic evaluation of the spleens showed lymphoid hyperplasia, follicular lymphoma, and diffuse large B-cell lymphoma (DLBCL) of the centroblastic type (Fig. 4D). One mouse was diagnosed with a histiocytic sarcoma.

The lymphoid hyperplasia showed normal splenic architecture, although the follicles were enlarged and more numerous. The marginal zone was often expanded with the periarteriolar sheaths (PALS) appearing to be normal. No abnormalities were detected in the red pulp. The CD3-positive cells (T cells) were concentrated around the PALS and were comparable in size and number with the wild-type controls. The CD45R-positive cells (B cells) were increased in number and were concentrated around the PALS, within the marginal zone, and scattered within the red pulp. Follicular B-cell lymphomas showed the uniform enlargement and coalescence of the follicles of white pulp, and a loss of the distinct marginal zones and PALS was not observed. The white pulp contained a mixed cell population with cytologic features similar to centroblasts and centrocytes. CD45R and CD3 staining showed B cells and T cells, respectively, throughout the altered follicular structure. The red pulp was compressed or reduced. The DLBCL of the centroblastic type consisted of diffuse involvement of the spleen and the effacement of the splenic architecture. Within the white pulp, the majority of the cells were of the centroblastic type, being of medium size and containing little cytoplasm, round vesicular nuclei, and prominent nucleoli. The CD45R and CD3 immunostaining showed B cells and T cells intermingled throughout the spleen.

Metastases were noted within the lungs, liver, and kidneys. These consisted of multifocal infiltrations of CD45R-positive B cells, often admixed with smaller numbers of CD3-positive T cells. The infiltration of CD45R-positive B cells was often located around blood vessels and often compressed, effacing the normal architecture of the organ. When all the lymphoproliferative lesions (lymphoid hyperplasia, DLBCL of the centroblastic type, and follicular lymphoma) were grouped, the Fen1^E160D/E160D mice showed a significant increase (both males and females) in these lesions, compared with the wild-type controls (P < 0.05). Pulmonary proliferative lesions (epithelial hyperplasia, adenoma, and adenomcarcinoma) were noted in female Fen1^E160D/E160D mice although the incidence of these lesions, when grouped together, was not statistically higher than in the wild-type controls.

To investigate the extent of the genomic instability of the tumors in Fen1^E160D/E160D mice, high-resolution array-based comparative genomic hybridization was done using an Agilent 44k oligonucleotide microarray (Supplementary Fig. S5). Fifteen tumors from nine different mice were analyzed against their corresponding controls. Complex genomic rearrangement, involving multiple chromosomes, was observed in three tumors (Supplementary Fig. S5A–D). In addition, moderate rearrangements of chromosome or chromosome arms were seen in three samples, as well as tumors with nearly a normal DNA copy number. In total, >50% of the tumors showed aberrations in DNA copy number. Two minimal recurrent regions of alterations were identified in >20% of the tumors (individual animals; Supplementary Fig. S5E–H). Loss of 8C3 (∼0.9 Mb) was found in 4 of 9 tumors, and gain of 8C4 (∼0.7 Mb) was identified in 5 of 9 (Supplementary Fig. S4). According to Ensembl (NCBI m37, April 2007), these regions contained 7 and 29 genes, respectively (Supplementary Table S4). No aberrations were observed in the APC locus previously found to be linked to tumor formation in Fen1 mutant mice (12).

Normal somatic hypermutation and class-switch recombination of immunoglobulin genes in Fen1^E160D/E160D mice. Defects in class-switch recombination and somatic hypermutation are processes in the B cells, which are often associated with B-cell lymphoma. To examine the effects of reduced FEN1 activity on class-switch recombination, we stimulated B cells purified from spleens of native 4- to 6-month-old littermates of wild-type and Fen1^E160D/E160D mice, using lipopolysaccharide (LPS) or LPS and interleukin-4 (IL-4), which will induce class-switch recombination (Fig. 5). LPS stimulation in vitro in the presence of IL-4 induces B cells to undergo class-switch recombination from IgM to IgG1 (26). Cells were cultured for 4 days and analyzed by flow cytometry for surface expression of the expected isotopes (Fig. 5A). These in vitro data showed that class-switch recombination seemed to be normal in the Fen1^E160D/E160D mice. Additionally, stimulated B cells from Fen1^E160D/E160D and wild-type controls showed no appreciable difference in proliferation, as measured by [³H]thymidine incorporation (Fig. 5B), indicating that reduced FEN1 activity did not result in an obvious defect in B-cell proliferation.

We also analyzed somatic hypermutation in the mutant mice by monitoring mutation accumulation in the intronic V_HJ558/J_H rearrangement-flanking region. This intronic region was PCR amplified from germinal center B cells from Peyer's patches of 5- to 7-month-old mice, and sorted by virtue of their PNA^hi CD45R(B220)⁺ phenotype. No divergence was found from the pattern in control animals (Fig. 5C).

Elucidating the molecular events that occur at the onset of B-cell lymphomas is essential for several reasons. Diffuse large cell lymphoma is an aggressive B-cell malignancy that accounts for 40% of non-Hodgkin's lymphomas and is the most common type of lymphoma in adults (27). Due to its heterogeneous presentation, diffuse large cell lymphoma gives rise to significant individual-level differences in responsiveness to therapy and a 60% mortality, making treatment decisions difficult (28). We show in this study the results of two Fen1 knock-in mice, which show that a decrease in FEN1 activity causes abnormal cell proliferation, genomic instability, and tumorigenesis.

The currently understood role of FEN1 at critical DNA structures generated as base excision repair intermediates possibly explains the tumor phenotype of Fen1^E160D/E160D mice. In mammalian cells, 20,000 to 50,000 DNA-damaging events occur per day; some of these DNA damage events include single-strand breaks, depurination of DNA bases, and oxidation of DNA bases. The base excision repair pathway repairs most of the damages associated with this spontaneous decay of the DNA helix (29).

Several base excision repair–deficient mice have been generated, but the majority of these mice are not associated with a cancer-prone phenotype (30). Two exceptions are the cancer-prone Ogg1/MutY double mutant mice and UNG-deficient mice (22, 31). A UNG deficiency in mice causes a lymphoproliferative disorder with lymphoid hyperplasia and a 22-fold increase in the risk of developing B-cell lymphoma, a phenotype similar to that of a Fen1^E160D/E160D knock-in mouse. Lymphoma development in UNG-deficient mice is explained by reduced class-switch recombination and skewed somatic hypermutation (22). Loss of regulation of somatic hypermutation is observed in >50% of DLBCLs (32). Despite a similar phenotype, these processes do not seem to be affected in Fen1^E160D/E160D mice. Cell lines derived from Fen1^E160D/E160D mutant mice show that the E160D mutation reduces the rate of cellular proliferation in MEF cells. This observation is consistent with the role of FEN1 in removing the RNA primer during DNA replication and several studies conclude that lymphomagenesis involves changes in the expression of genes that are important for proliferation (33).

Fen1^E160D/E160D mice develop a 20-fold increased incidence of early onset hematopoietic malignancy, principally attributable to DLBCL and follicular B-cell lymphoma, and females develop lymphomas at a higher incidence than males. This skewed tumor ratio between the sexes is also seen by others (34). The reason for the gender differences observed in mouse lymphoma is not well understood. One can speculate that sex hormones, such as estrogen, could result in the observed skewed tumor ratio between the sexes. The sex hormones are involved in both normal development and immunomodulating processes of the cell (e.g., estrogen influences immunoglobulin levels in postpubertal female mice and is involved in the adaptive immune system). In a recent study, FEN1 was suggested to modulate the estrogen-responsive gene expression (35).

The finding that female Fen1^E160D/E160D mice have an increased incidence of proliferative pulmonary lesions must be evaluated with caution. It has been documented that the 129 strain mice can have a high incidence (up to 40%) of spontaneous lung tumors (36). Although backcrossing is done to remove extraneous genetic material, the 129 strain may still have an influence on the phenotypic lesion observed (37). Therefore, the different genetic background of two Fen1^E160D/E160D knock-in studies could explain the intriguing difference in the tumor phenotypes (38). Yet, both studies conclude that genomic instability caused by inadequate FEN1 activity leads to a considerably elevated spontaneous tumor incidence. In the study by Zheng and colleagues (38), several mutations affecting the nuclease activity of FEN1 were identified in human cancer specimens. The remarkable different tumor manifestation could imply that in a heterogeneous population, represented in the studies described above by two different inbred mouse strains, mutations or single-nucleotide polymorphisms in other genes might be important for the tumor type. Large-scale evaluations of common genetic variants in the human population support this notion. One such study evaluated the association between single-nucleotide polymorphisms in DNA repair genes and risk of non-Hodgkin's lymphoma in a population-based case-control study. Single-nucleotide polymorphisms in DNA repair genes were associated with altered risk of non-Hodgkin's lymphoma and DLBCL, the major B-cell subtype (39).

The detrimental phenotype found in our Fen1^{ΔPCNA/ΔPCNA} embryos is far more severe than the phenotype observed in another related study (40). In our Fen1^{ΔPCNA/ΔPCNA} embryos, we identified an embryonic lethal phenotype at E9.5 to E10.5; however, Zheng and colleagues reported a phenotype with newborn lethality. Additionally, they reported replication defects, pulmonary hypoplasia, and pancytopenia. Our Fen1^{ΔPCNA/ΔPCNA} embryos showed defective organogenesis, which could explain the significant defects in the growth and the development of the lung and the blood system (40). The noticeable difference between these two FEN1 ΔPCNA models could be explained by different amino acid mutations. In one study, the two phenylalanines, in the conserved PIP box of FEN1 (41), were mutated to alanines (FF to AA), whereas in the other study, the two phenylalanines were replaced with glycine and alanine (FF to GA; 343, 344). Glycine can cause increased peptide backbone flexibility compared with alanine because it is lacking the small aliphatic side chain. Gomes and Burgers (10) previously characterized the FF to GA mutant of the conserved PCNA interacting domain in yeast FEN1. With this mutant of FEN1, they observed a severely decreased interaction with PCNA, which resulted in replication and repair defects in vivo. An FF to AA mutant in the PIP box domain of the CDC9 DNA ligase had a less severe effect, when compared with the FEN1 FF to GA mutation, on the CDC9-PCNA interaction (42). Our Fen1^{ΔPCNA/ΔPCNA} embryos did not properly undergo gastrulation (start at stage E6), neurulation, cardiogenesis, somitogenesis, or body folding (all events start at E7–E7.5), and the defective embryonic development ended in apoptosis and death around midgestation. Embryo stages E6 to E12.5 show rapid cell proliferation in the normal developing embryo. The detrimental phenotype observed in embryos leads us to conclude that the FEN1 interaction with PCNA is crucial for rapidly proliferating cells in the mouse embryo. On the other hand, if FEN1 cannot interact with PCNA, it must be recruited to the replication machinery by other mechanisms to avoid very early embryonic lethality. To date, at least 20 proteins are reported to interact with FEN1. Each of the known interacting proteins can be categorized based on the role in which they assist FEN1 in DNA repair, replication, apoptosis, or cell cycle control (2). During DNA lagging strand replication, in addition to PCNA, FEN1 interacts with DNA polymerase δ, replication protein A, and DNA ligase I, in addition to Werner and Bloom proteins (6). Similar to PCNA, Bloom and Werner proteins can stimulate FEN1 activity in vitro (43). FEN1 also interacts with chromatin to accomplish its role in DNA replication (41). Thus, FEN1 might get access to flap substrates during replication through binding to a core histone tail. Some of these interactions might enable mutated FEN1 to perform limited nuclease activity at the replication fork without interacting with PCNA.

We observed marked obesity by 9 months of age in male and female Fen1^E160D/E160D mice. In the last year, several exciting studies have observed obesity and severe metabolic defects in mice carrying specific defects in base excision repair (44, 45). Moreover, the obesity-associated FTO gene was recently shown to encode a 2-oxoglutarate–dependent nucleic acid demethylase (46). It is now important to determine whether altered DNA repair activities underlie altered food intake or energy expenditure, or both. A systematic review recently identified a link between increased body mass index and increased risk of common and less common malignancies, and for some cancer types association differs between sexes and populations of different ethnic origins (47). Such data also support the notion above about the different tumor types observed in two different mouse strains carrying the Fen1^E160D alleles.

In conclusion, we report here of two different Fen1 mutant mice with a dramatic difference between the nuclease-defective mouse and the strain defective in PCNA binding, showing that the interaction with PCNA is extremely important to ensure FEN1 being accurately recruited to DNA 5′-flap substrates, which are key intermediates in DNA metabolism.

No potential conflicts of interest were disclosed.

Grant support: National Program in Functional Genomics sponsored by the Norwegian Research Council, the Norwegian Cancer Society, and the University of Oslo (EMBiO). This work was supported by the National Microarray platform of the Functional Genomics program of the Research Council of Norway.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Adam Robertson for helpful comments and corrections and the Norwegian Transgenic Center (NTS) and the Center for Comparative Studies at Rikshospitalet Medical Center for excellent service.