Deficient DNA End Joining Activity in Extracts from Fanconi Anemia Fibroblasts*

Fanconi anemia (FA) is a genetic disorder associated with genomic instability and cancer predisposition. Cultured cells from FA patients display a high level of spontaneous chromosome breaks and an increased frequency of intragenic deletions, suggesting that FA cells may have deficiencies in properly processing DNA double strand breaks. In this study, an in vitro plasmid DNA end joining assay was used to characterize the end joining capabilities of nuclear extracts from diploid FA fibroblasts from complementation groups A, C, and D. The Fanconi anemia extracts had 3–9-fold less DNA end joining activity and rejoined substrates with significantly less fidelity than normal extracts. Wild-type end joining activity could be reconstituted by mixing FA-D extracts with FA-A or FA-C extracts, while mixing FA-A and FA-C extracts had no effect on end joining activity. Protein expression levels of the DNA-dependent protein kinase (DNA-PK)/Ku-dependent nonhomologous DNA end-joining proteins Xrcc4, DNA ligase IV, Ku70, and Ku86 in FA and normal extracts were indistinguishable, as were DNA-dependent protein kinase and DNA end binding activities. The end joining activity as measured by the assay was not sensitive to the DNA-PK inhibitor wortmannin or dependent on the nonhomologous DNA end-joining factor Xrcc4. However, when DNA/protein ratios were lowered, the end joining activity became wortmannin-sensitive and no difference in end joining activity was observed between normal and FA extracts. Taken together, these results suggest that the FA fibroblast extracts have a deficiency in a DNA end joining process that is distinct from the DNA-PK/Ku-dependent nonhomologous DNA end joining pathway.

Fanconi anemia (FA) 1 is an autosomal recessive disease characterized by developmental abnormalities, progressive bone marrow failure, chromosomal instability, and predisposition to cancer (1,2). Somatic cell fusion studies have demonstrated the existence of at least eight complementation groups (FA-A through FA-H) (3). Presently, five of the FA genes have been cloned, FANCA, FANCC, FANCE, FANCF, and FANCG (4 -8). The biochemical functions of these proteins are unknown; thus, the underlying defect of this disease has not been established.
In vitro analysis of cultured cells obtained from FA patients reveals an elevated level of spontaneous chromosome breaks. The frequency of these chromosomal lesions is amplified following exposure to DNA cross-linking agents (1). FA cells also experience spontaneous and psoralen-induced DNA deletions at a higher frequency than normal cells. These DNA deletions have been detected both within the endogenous hypoxanthineguanine phosphoribosyltransferase gene and within a target gene present on an autonomously replicating plasmid (9,10). These cellular phenotypes suggested that FA cells may have deficiencies in processing DNA double strand breaks.
Recent reports have supported the hypothesis that FA cells have deficiencies in rejoining double strand breaks (DSBs) (11,12). In these studies, linearized plasmid DNA was transfected into immortalized FA lymphoblasts and recovered after 48 h. Analysis of recircularized products revealed that the overall efficiency of plasmid end joining was normal in FA lymphoblasts from complementation groups B, C, and D, but error-free processing of blunt-ended substrates was significantly compromised in these cells.
To gain further insight into the process of DNA end joining in FA cells, we used an in vitro assay to examine the ability of nuclear protein extracts prepared from diploid FA fibroblasts to rejoin linear plasmid DNA substrates. Nuclear extracts from diploid fibroblasts from patients from complementation groups A, C, and D had 3-9-fold less end joining activity and rejoined linear substrates imprecisely at a higher frequency than extracts from normal donors. This end joining deficiency was not due to the presence of an inhibitor in the FA extracts or to deficiencies in proteins or activities known to be involved in the well characterized DNA-PK/Ku nonhomologous DNA end joining pathway (13). Wild-type end joining activity could be reconstituted by mixing FA-D extracts with FA-A or FA-C extracts but not by mixing FA-A with FA-C extracts. The end joining activity that was deficient in the FA extracts was not sensitive to the DNA-PK inhibitor wortmannin or dependent on Xrcc4. When a lower substrate DNA/protein ratio was used in the end joining assay, the end joining activity was wortmannin-sensitive, and indistinguishable end joining levels were observed between normal and FA extracts.

EXPERIMENTAL PROCEDURES
Cell Culture and Conditions-The diploid FA fibroblast cell strains PD.134.F (FA-C), PD.220.F (FA-A), and PD.20.F (FA-D) as well as the normal diploid cell strains PD.715.F, PD.13.F, PD.793.F, PD.792.F, and PD.751.F were kindly provided by Dr. Markus Grompe (Oregon Health Sciences University). These cells were maintained in minimum essential ␣-medium supplemented with 2 mM glutamine and 15% fetal bovine serum. Normal diploid strains CRL-2115, CRL-2068, and CRL-2072 were purchased from the American Type Culture Collection (Manassas, VA) and were maintained in Eagle's minimum essential medium supplemented with 2 mM glutamine, 1 mM sodium pyruvate, and 10% fetal bovine serum. All cells were maintained at 37°C in a humidified, 5% CO 2 environment.
Nuclear Protein Extracts-Nuclear extracts were prepared as previously described (14). Briefly, cells harvested from confluent 100-mm tissue culture dishes were washed three times with ice-cold phosphatebuffered saline and resuspended in 2 ml of hypotonic buffer A (10 mM KCl, 10 mM Tris (pH 7.4), 10 mM MgCl 2 , and 10 mM dithiothreitol) and kept on ice for 15 min. Phenylmethylsulfonyl fluoride was added to 1 mM, and cells were disrupted using a Dounce homogenizer (20 strokes with a tight pestle). The released nuclei were pelleted and resuspended in 2 ml of buffer A containing 350 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, 1.0 g/l aprotinin, and 0.7 g/ml pepstatin and incubated for 1 h on ice. The nuclei were centrifuged at 70,000 rpm in a Beckman TL-100.3 rotor at 4°C for 30 min, and the clear supernatant was adjusted to 10% glycerol and 10 mM ␤-mercaptoethanol. The resulting extracts were dialyzed against a buffer containing 25 mM Tris (pH 7.5), 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol. Protein concentrations were determined by the Bradford method (15).
End Joining Reactions-DNA end joining reactions were carried out essentially as described previously (16). Circular pCR 2.1 plasmid DNA (Invitrogen, Carlsbad, CA) was linearized by restriction digestion with KpnI to create linear substrates with 3Ј-cohesive ends, EcoRI to produce 5Ј-cohesive ends, or EcoRV to generate blunt-ended substrates (all endonucleases were from New England Biolabs, Beverly, MA). After restriction digest, substrates were ethanol-precipitated and resuspended in TE buffer, pH 8.0. 1 g of linearized DNA was incubated with 5 g of nuclear protein extract in 70 mM Tris (pH 7.5), 10 mM MgCl 2 , 10 mM dithiothreitol, and 1 mM ATP in a total volume of 50 l. The reaction was carried out at 14°C for 12 h unless otherwise noted. The reaction mixture was then treated with proteinase K at 37°C for 30 min and electrophoretically separated on a 0.8% agarose gel in Tris borate-EDTA buffer at 0.55 V/cm for 12-15 h. After staining in ethidium bromide, gels were scanned on a Bio-Rad scanner using the Molecular Analyst program and quantified using IP Lab Gel (Signal Analytics Corp., Vienna, VA). A band that migrated with form II of the uncut substrate DNA was labeled CC and called closed circular; a band that migrated at the predicted size of a linear dimer was labeled D; and bands that migrated larger than the linear dimer were labeled as higher molecular weight products (HM). To quantitate the percentage of rejoining, total product formation was calculated (CC ϩ D ϩ HM) and divided by the sum of total substrate DNA in the reaction (L ϩ CC ϩ D ϩ HM).
Antibodies-The anti-Ku70 and anti-Ku86 antibodies were purchased from Serotech, (Raleigh, NC). The anti-Xrcc4 and anti-DNA ligase IV antibodies were kindly provided by Drs. Susan Critchlow and Stephen Jackson (Welcome/CRC Institute, Cambridge, UK). All primary antibodies were raised in rabbits against recombinant human proteins.
Western Blot Analysis-Protein samples (10 g) were resolved on SDS-polyacrylamide gels (17) and transferred to nitrocellulose membranes (Bio-Rad). After a 1-h incubation in 5% bovine serum albumin in Tris-buffered saline, the membrane was probed with antibody (1:2000 dilution for anti-Ku70 and anti-Ku86 antibodies; 1:1000 dilution for anti-Xrcc4 and anti-DNA ligase IV antibodies). The membrane was then washed three times in 0.1% bovine serum albumin in Tris-buffered saline and incubated with diluted (1:5000) alkaline phosphatase-conjugated goat-anti-rabbit IgG (Sigma) for 1 h at room temperature. This was followed by three additional washes in 0.1% bovine serum albumin. Incubation with Sigma Fast (Sigma), which contains the alkaline phosphatase substrate 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium, was then carried out to detect the presence of the proteins.
Electrophoretic Mobility Shift Assay-A modification of the assay used by Rathmell and Chu (18) was used to detect DNA end binding activity in the nuclear extracts. A 394-base pair fragment from an EcoRI digest of the plasmid pREP4 was radioactively end-labeled with [␣-32 P]dATP by using the Klenow fragment of Escherichia coli DNA polymerase. 0.2 ng of probe was incubated with 0.3 g of nuclear extract in 12 mM HEPES, 5 mM MgCl 2 , 4 mM Tris (pH 7.9), 100 mM KCl, 0.6 nM EDTA, 0.6 mM dithiothreitol, and 6% glycerol in the presence of 200 ng of unlabeled supercoiled DNA. To specifically compete for the DNA end binding activity, 200 ng of unlabeled linear plasmid DNA was added in place of circular DNA to the reaction mix. Reactions were carried out at 14°C for 30 min in a final volume of 10 l. Following incubation, 5ϫ loading dye (0.25% bromphenol blue, 0.25% xylene cyanol, 30% glycerol) was added to each reaction and subjected to electrophoresis on a 5% polyacrylamide gel in TBE (90 mM Tris borate, 2 mM EDTA) running buffer. The gel was run at 20 V/cm for 2 h at 4°C. Detection of radioactivity was achieved using a PhosphorImager (Molecular Dynamics, Inc., Foster City, CA).
DNA-PK Assay-The SignaTECT DNA-dependent Protein Kinase Assay System (Promega, Madison, WI) was used to detect DNA-PK activity following instructions provided by the manufacturer.
Wortmannin Inhibition-A 0.1 mM stock of wortmannin (Sigma) was prepared in 10% Me 2 SO. Wortmannin was incubated with 5 g of nuclear extracts for 30 min on ice. Plasmid DNA end joining experiments as previously described were then performed at 14°C.
Immunodepletion-Extracts (50 g) were incubated with antisera for 1 h at 4°C on a rotary wheel. The extract/antibody mixture was added to 25 l of protein A-Sepharose beads (Sigma) and incubated with rotation for 3 h at 4°C. The beads were removed by repeated centrifugation at 12,000 ϫ g, and the supernatant was used for Western blot analysis and end joining experiments.
Bacterial Transformation Assay-End joining assays were carried out as described above. After the 14°C incubation, the reaction mixture was incubated at 37°C with calf intestinal alkaline phosphatase (Roche Molecular Biochemicals) for 1 h. Samples were then extracted with phenol/chloroform, ethanol-precipitated, and resuspended in 10 l of TE buffer, pH 8.0. Electrocompetent E. coli (strain DH10B) was then electroporated with 1 l of recovered plasmid DNA using a Life Technologies, Inc. Gene Pulser at a field strength of 2.44 kV/cm and plated on ampicillin-containing LB plates containing isopropyl-1-thio-␤-D-galactopyranoside and 5-bromo-4-chloro-3-indoyl-␤-D-galactoside. Bacterial transformants containing imprecisely end-joined plasmids were detected as white colonies, whereas transformants harboring precisely rejoined plasmids formed blue colonies. Plasmid DNA was isolated by the alkaline lysis method, characterized by restriction mapping, and sequenced using a PerkinElmer Life Sciences automated DNA sequencer (Microchemical Facility, University of Minnesota).

RESULTS
To test the end joining activity of FA fibroblasts, an in vitro DNA end joining assay that has been previously described was employed (16). Nuclear protein extracts were prepared from diploid fibroblasts from normal donors and diploid fibroblasts from FA patients of complementation groups A (FA-A), C (FA-C), and D (FA-D). 5 g of extract was incubated for 12 h at 14°C with 1 g of plasmid pCR2.1 DNA that had been linearized by restriction digestion to have either blunt or cohesive ends. End joining activity was detected by the presence of closed circular, linear substrate, and high molecular weight products when the reaction mixture was analyzed by agarose gel electrophoresis. In end joining experiments performed with blunt-ended substrate, closed circular was the predominant product formed. When cohesive-ended substrates were used, closed circular, linear substrate, and high molecular weight products were detected on the ethidium bromide-stained gels. Fig. 1a shows the results of an end joining experiment performed with blunt-ended substrate. Scanning laser densitometry was used to quantitate the bands, and the percentage of linear substrate that had been rejoined was determined. This analysis revealed that the normal extract rejoined 34% of the linearized substrate as compared with 3, 7, and 7%, respectively, by the FA-A, FA-C, and FA-D extracts.
To confirm that the FA-A, FA-C, and FA-D extracts were deficient in DNA end joining, two or three independent extracts were prepared from each cell line and tested multiple times in end joining experiments with blunt and cohesive-ended substrates. Also, nuclear extracts were prepared from seven normal fibroblast strains in addition to the normal fibroblast strain used in Fig. 1a (PD.715.F) and tested for end joining activity. Fig. 1b depicts the average end joining activity (percentage rejoined Ϯ S.E.) in the normal, FA-A, FA-C, and FA-D extracts. The mean end joining activity of blunt-ended substrate in extracts from the normal strains ranged from 19 to 45% with a cumulative mean of 31 Ϯ 4%. Rejoining of cohesiveended substrate by the normal extracts ranged from 16 to 37% (cumulative mean, 26 Ϯ 3%). The normal strain, PD.715.F, used in Fig. 1a and subsequent figures, rejoined 41 Ϯ 3% of blunt-ended substrate and 35 Ϯ 4% of cohesive-ended substrate. In comparison, the extracts from the FA-A cell strain rejoined 4 Ϯ 1% of both substrates, the extracts from the FA-C cell strain rejoined 6 Ϯ 2% of blunt-ended substrate and 12 Ϯ 2% of cohesive-ended substrate, and the extracts from the FA-D cells rejoined 15 Ϯ 3% and 10 Ϯ 2% of the blunt-and cohesiveended substrates, respectively. Comparison of the mean end joining activities in the eight normal fibroblast strains with the mean end joining activities in the FA extracts revealed that the FA-A, FA-C, and FA-D extracts were significantly less able to rejoin linearized plasmid substrates with blunt and cohesive ends than the normal extracts.
In the above experiments, the cohesive-ended substrate had a 5Ј-overhang. To determine whether the FA extracts were also deficient at rejoining substrates with 3Ј-cohesive ends, end joining experiments were performed with a substrate that was linearized by restriction digest with KpnI. Nuclear extracts from the eight normal fibroblast strains rejoined a mean of 48 Ϯ 4% of the substrate, with a range from 35 to 64%. The FA-A, FA-C, and FA-D extracts rejoined 3, 2, and 12%, respectively, of the 3Ј-cohesive-ended substrate (data not shown). We concluded that FA-A, FA-C, and FA-D extracts were deficient in rejoining plasmid DNA substrates with blunt-, 5Ј-cohesive, and 3Ј-cohesive ends.
To determine whether the observed end joining deficiency in the FA extracts was due to reduced kinetics of activity, a time course experiment was performed with normal and FA-C extracts. The plasmid end joining assay was performed using a blunt-ended substrate, and aliquots were removed for analysis at 4, 8, and 12 h. As shown in Fig. 2a, the normal cell extract rejoined the blunt-ended substrate in a linear fashion between 4 h (25% rejoined) and 8 h (44% rejoined) before reaching saturation at 12 h (52% rejoined). On the other hand, no activity was detected in samples incubated with the FA fibroblast extract for up to 8 h. Following a 12-h incubation period, 4% rejoining was detected. Similarly, we tested the end joining activity of the normal and the FA-C fibroblast extracts as a function of protein concentration. Plasmid DNA end joining with blunt-ended substrate was performed using 2.5, 5, 7.5, 10, and 12.5 g of protein.
As seen in Fig. 2b, there was a noticeable difference in end joining activity between the two extracts at all protein concentrations tested. Similar observations were made when cohesive-ended substrates or FA-A and FA-D extracts were used (data not shown).
Deficient end joining capabilities of the FA extracts can be explained by two alternate hypotheses; either the extracts derived from the FA fibroblasts contain an inhibitor of the end joining reaction, or, conversely, a factor or factors essential for the end joining reaction may be absent from the FA extracts. To distinguish between these two possibilities, an end joining experiment with blunt-ended substrate was performed using a mixture of equal amounts of nuclear protein extract from normal and FA-C cells. To keep the total protein present in the reaction at 5 g, 2.5 g of the normal extract was mixed with 2.5 g of the FA extract. As seen in Fig. 3, 5 g of normal extract alone yielded 42% product formation. When a mixture of normal and FA extracts was tested, 34% product formation was detected. This level of end joining is consistent with the percentage of end joining obtained when 2.5 g of normal extract is used in end joining reactions (see Fig. 2b). The same results were obtained when FA-A or FA-D extracts were mixed with normal extracts or when cohesive-ended substrates were used (data not shown). The wild-type level of plasmid end joining activity present in the mixed sample is inconsistent with the notion that an inhibitor of end joining is present in the FA nuclear extract. This finding indicates that nuclear extracts from FA fibroblasts lack a factor or factors essential for the end joining reaction.
The reduced DNA end joining activity in extracts prepared from FA-A, FA-C, and FA-D fibroblast strains raised the possibility that wild-type end joining activity could be reconsti-

FIG. 2. Time course (A) and protein concentration dependence (B) of blunt end plasmid rejoining in nuclear protein extracts.
End joining activity is depicted as percentage rejoined and was determined as described under "Experimental Procedures." Squares represent data from an extract prepared from the normal cell strain PD.715.F; circles represent data from an extract from FA-C cell strain PD.134.F. *, Ͻ1% of plasmid was rejoined.
tuted by mixing combinations of the FA extracts. We therefore mixed 2.5 g of extract from FA-A, FA-C, and FA-D fibroblasts in combination with one another and performed DNA end joining experiments. As seen in Fig. 4, 5 g of normal extract alone rejoined 33% of the linearized DNA substrate. Combining 2.5 g of FA-A or FA-C extracts with 2.5 g of FA-D extract resulted in 29 and 32%, respectively, of DNA end joining activity, whereas a mixture of the FA-A and FA-C extracts had no effect on end joining levels (7% end joining activity).
It has been previously established that immortalized FA lymphoblasts rejoin blunt-ended DNA substrates in an errorprone manner in vivo relative to control lymphoblasts (12,13). We therefore examined the nature of DNA end joining in nuclear extracts from FA-C fibroblasts and nuclear extracts from normal donors. A bacterial transformation assay was employed to analyze the fidelity of end joining in the extracts (see "Experimental Procedures"). Using blunt-ended substrate, a significantly higher number of white colonies were recovered from rejoining by the FA-C extract as compared with the normal extract, 27% (217 of 588) versus 18% (171 of 780) (p Ͻ 0.005, 2 ϭ 36.67) (Fig. 5). Similarly, a significantly higher number of white colonies were recovered from the FA-C extract when cohesive-ended substrates were used. 4% (33 of 899) of the colonies resulting from the rejoining by the FA-C extract were white compared with 0.5% (15 of 2917) from the normal fibroblast extract. Statistical analysis again revealed that the difference was significant (p Ͻ 0.005, 2 ϭ 52.8).
Restriction digest analysis of imprecisely rejoined plasmids from the FA-C (101 plasmids) and normal (69 plasmids) end joining experiments revealed that all imprecise rejoining events resulted in deletions (data not shown). Sequence data of the region flanking the rejoining site obtained from 26 plasmids that resulted from imprecise rejoining (13 from FA and 13 from normal extracts) showed no difference in the range of the sizes of the deletions (2-47 base pairs). It also showed that all deletions, irrespective of whether obtained from the FA-C or normal extract, were flanked by direct repeat sequences of between 1 and 7 base pairs in length (data not shown). From this analysis, we concluded that there was no difference in the nature of the imprecise end joining events between the normal and FA-C extracts.
Currently, two independent pathways are known to repair DSBs in cells: homologous recombination and nonhomologous DNA end joining (NHEJ). In contrast to genetic screens of x-ray-sensitive yeast cells, which identified genes involved in homologous recombination (HR) (19, 20), mutant screens of x-ray-hypersensitive mammalian cells have led to the identification of factors involved in NHEJ (21). This led to the sugges-tion that DSBs are preferentially repaired by NHEJ in mammalian cells (13).
A NHEJ pathway, generally referred to as the DNA-PK-or Ku-dependent pathway, has been well characterized in recent years. It has been demonstrated to be minimally dependent upon five proteins: Xrcc4, DNA ligase IV, Ku70, Ku86, and the catalytic subunit of DNA-dependent protein kinase (DNA-PK cs ) (22). To test if any of these proteins were absent from FA-C extracts, Western blot analysis was performed using antisera specific for the Ku70, Ku86, DNA ligase IV, and Xrcc4 proteins. The levels of these four proteins present in nuclear extracts from FA-C fibroblasts were essentially identical to those seen in the normal fibroblast extract (Fig. 6a), suggesting that low expression of any of these four proteins was not the cause of the deficient end joining in the FA extracts.
The presence of wild-type levels of these proteins does not prove that they are functional. However, the amount of DNA end binding activity, which is dependent upon the functional Ku70/Ku86 heterodimer as determined by the ability of Ku70 and Ku86 antibodies to supershift the end binding band (23), was identical in control and FA-C extracts (Fig. 6b). Similarly, DNA-PK activities in FA-C extracts were indistinguishable from those present in extracts from control fibroblasts (not shown). Western blot analysis, DNA end binding, and DNA-PK assays performed with FA-A and FA-D extracts revealed no differences from normal fibroblasts in these extracts as well (data not shown). Previous studies performed with FA lymphoblasts also found no deficiencies in these factors or activities (11,12).
We next wished to determine whether the assay used in this study was measuring NHEJ activity. To do this, we determined if the end joining activity was dependent on DNA-PK or Xrcc4, two components of the DNA-PK/Ku-dependent NHEJ pathway (22). Wortmannin is a potent inhibitor of phosphatidylinositol 3-kinases (24) and has been demonstrated to inhibit DNA-PK activity (25). 5.0 g of nuclear extract prepared from normal diploid fibroblasts was incubated for 30 min on ice with 5.0 M wortmannin. This concentration of wortmannin abolished DNA-PK activity in the extract as measured by an in vitro DNA-PK activity kit (Fig. 7a). DNA end joining reactions with 5Ј-cohesive-ended substrates were then performed at 14°C for 4 h. Wortmannin had no inhibitory effects on the end joining activity of these extracts (Fig. 7b). Wortmannin also had no inhibitory effects on the end joining activity detected in FA extracts (Fig. 7c).
To determine whether Xrcc4 was required for the end joining activity, antiserum raised against the human Xrcc4 protein was used to immunodeplete Xrcc4 from a nuclear extract prepared from fibroblasts from a normal donor. Western blot analysis confirmed that the extract was depleted of this protein (Fig. 8A). DNA end joining experiments were performed at 14°C for 4 h with the Xrcc4-immunodepleted extract. As shown in Fig. 8B, the Xrcc4-immunodepleted extract had wild-type end joining activity. We concluded that the end joining assay as used in this study was measuring an activity that was not dependent on the NHEJ factors DNA-PK or Xrcc4.
Baumann and West described an in vitro end joining assay that was dependent on the NHEJ factors Ku70, Ku86, Xrcc4, DNA ligase IV, and DNA-PK cs (22). Under the conditions of this assay, 10 -100 ng of substrate DNA was incubated with 20 -60 g of cellular extract (0.001 g of DNA/g of protein as compared with 0.2 g of DNA/g of protein in the current study). To more closely resemble the conditions described by Baumann and West, we performed end joining experiments for 2 h at 14°C using 50 ng of linearized DNA and 5 g of nuclear extracts (0.05 g of DNA/g of protein) from normal and FA fibroblasts. As shown in Fig. 9, similar to what was reported by Baumann and West, the end joining activity was sensitive to wortmannin under these conditions. The end joining activity observed in the normal and FA extract were identical under these conditions. The substrate DNA concentration was the only variable changed under the wortmannin-sensitive condi- A, 10 g of nuclear protein extract immunodepleted with preimmune antisera (Ϫ) or Xrcc4 antisera (ϩ) was resolved on a 10% SDS-polyacrylamide gel, transferred to nitrocellulose membrane, and subjected to Western blot analysis against a human Xrcc4 antiserum. B, DNA end joining experiments with 5Ј-cohesive-ended substrate were performed with 5 g of normal extract that was immunodepleted with preimmune antisera (Ϫ), or Xrcc4 antisera (ϩ). End joining was performed for 4 h at 14°C. L represents the mobility of the linear substrate DNA, and CC indicates the mobility of closed circular product.
tions, suggesting that the DNA/protein ratio is crucial in achieving a DNA-PK-dependent end joining activity.

DISCUSSION
The data presented here demonstrate that extracts from diploid Fanconi anemia fibroblasts are deficient in end joining blunt and cohesive-ended linear plasmid DNA substrates. The end joining deficiency appears to be a common feature of FA, since extracts from fibroblasts from complementation groups A, C, and D share the defect.
Several lines of in vitro evidence indicate that eukaryotic cells rely upon more than one DNA end joining pathway. In a study performed by Johnson and Fairman (26), calf thymus extracts were fractionated into four biochemically distinct fractions. Despite the presence of Ku70 in only one of the fractions, end joining activity was detected in all fractions. In a second study, extracts that were prepared from the DNA-PK mutant tumor cell line MO59 had wild-type end joining activity, suggesting that a DNA-PK-independent end joining pathway was functioning in these cells (27).
In vivo studies have also supported the hypothesis that multiple DNA end joining pathways exist. Genetic inactivation studies carried out in the yeast Saccharomyces cerevisiae convincingly demonstrate that this organism repairs DNA double strand breaks through precise (Ku-dependent) and imprecise (Ku-independent) pathways (28,29). The same appears to be true in mammalian cells. Both wild-type and xrs6 hamster cells, which lack a functional Ku86 protein, were able to rejoin transfected linear plasmids. However, rejoining occurred in a more error prone manner in the xrs6 cells, and the deletions formed were on average 3-4-fold larger than in control cells (30). Taken together, these in vitro and in vivo studies strongly suggest that more than one NHEJ pathways exist.
The data presented in this study suggest that the FA extracts are deficient in a DNA end joining pathway that is distinct from the well characterized NHEJ pathway. First, FA extracts have wild type levels of the NHEJ factors Xrcc4, DNA ligase IV, Ku70, and Ku86 as well as normal levels of DNA-PK and DNA end binding activities. Second, the end joining activity in which the FA extracts are deficient is not inhibited by the DNA-PK inhibitor wortmannin. Third, extracts immunodepleted of Xrcc4 have wild-type end joining activity as measured by the assay. Finally, when end joining reactions are carried out similar to those previously reported for measuring NHEJ (22), there is no difference in end joining between normal and FA extracts, indicating that the well characterized NHEJ pathway is fully functional in the FA extracts.
The possibility cannot be ruled out that Ku is involved in this alternate end joining process. However, given that FA extracts have wild-type DNA end binding activity, it seems unlikely that the defect in FA cells is the result of defective Ku proteins. In addition, cells with Ku defects are sensitive to ionizing radiation (13, 21), a phenotype not associated with FA cells (31). Regardless of the involvement of Ku in this pathway, the data indicates that the end joining activity that is deficient in the FA extracts is independent of DNA-PK and Xrcc4. Thus, this end joining pathway appears to be distinct from that classically referred to as NHEJ.
It is tempting to speculate that the DNA end joining deficiency observed in FA extracts is representative of a cellular defect in rejoining double strand breaks. This alternate end joining pathway may represent an additional end joining pathway in cells. This could explain why end joining activity is still detected in Ku-deficient cells (28 -30). This could also explain why Escarceller et al. (11,12) found no deficiency in the overall end joining efficiency of linearized plasmid substrates but did see error-prone rejoining of blunt-ended substrates in FA lymphoblasts. In vivo, a defect in this additional end joining pathway would be masked by other DNA end joining pathways.
An end joining deficiency in FA cells could account for many of the cellular phenotypes associated with this disorder such as high levels of spontaneous and DNA cross-link-induced chromosomal breaks and high frequencies of spontaneous and psoralen-induced deletions. An end joining deficiency could also explain the predisposition to cancer associated with these patients, since unrepaired or misrepaired DNA lesions could ultimately lead to loss of function of genes essential for proper cellular maintenance and growth.
In addition, a DNA end joining defect in FA cells could potentially explain a previous result observed by our laboratory. HR activity was found to be elevated in Fanconi anemia fibroblasts and in nuclear extracts prepared from FA cells as compared with HR activity in fibroblasts from normal donors (32). We have also observed that Rad51, the mammalian homologue of the bacterial recombination protein RecA, is substantially elevated in extracts prepared from FA fibroblasts. 2 Recently, it was demonstrated that the human Rad52 protein, a protein involved in mammalian HR, binds double strand breaks (33). We speculate that HR and Rad51 may be elevated in FA cells in response to an increased number of unrepaired DSBs that result from the described end joining deficiency.
Finally, the data indicate that mixing FA-A or FA-C extracts with FA-D extracts is able to reconstitute wild-type end joining activity levels, while mixing FA-A and FA-C extracts has no effect on end joining levels. While FA-D patients have the same clinical symptoms as FA patients from the other complementation groups, there are reports of FA-D cells having unique biochemical characteristics (34 -36). In particular, a multiprotein complex of four cloned FA gene products (FANCA, FANCC, FANCG, and FANCF) is detected only in wild-type and FA-D cells, indicating that all of the FA proteins, with the exception of FANCD, are required for the proper formation of this complex (36). One could imagine that a preassembled "FA complex" is required for wild-type end joining activity in the extracts. This could only be provided by extracts from FA-D cells. Thus, when FA-A and FA-C extracts are mixed, neither would provide the "FA complex," and wild-type end joining activity would not be reconstituted. FIG. 9. Wild-type end joining activity in FA extracts in a wortmannin-sensitive end joining assay. DNA end joining experiments with 50 ng of 5Ј-cohesive-ended substrate were performed for 2 h at 14°C with 5 g of nuclear protein extract from normal cell line PD.715.F (N) or with 5 g of nuclear protein extract from the FA-C cell line PD.134.F (FA) that had been preincubated with 10% Me 2 SO (Ϫ) or 5.0 M wortmannin (ϩ) for 30 min on ice. The products were then resolved on a 0.8% agarose gel. Shown is an inverted image of the ethidium bromide-stained gel. L represents the mobility of the linear substrate DNA, while CC and D indicate the mobility of closed circular and linear dimer products, respectively.