Conserved Overlapping Gene Arrangement, Restricted Expression, and Biochemical Activities of DNA Polymerase ν (POLN)

Background: The biological function of DNA polymerase ν (POLN) is unknown. Results: Vertebrate POLN genes are predominantly expressed in testis, share a first exon with HAUS3, and encode proteins with strand displacement and damage bypass activity. Conclusion: These properties indicate a specific POLN function in testis. Significance: Conserved biochemical activities, expression patterns, and protein interactions suggest a restricted function of POLN in DNA processing.

It is remarkable that the genomes of higher eukaryotes encode so many different DNA polymerases. Each of these enzymes is specialized to operate in some aspect of DNA replication, pathways of DNA repair, diversification of antibody genes, or in translesion DNA synthesis. In human cells, defects or mutations in DNA polymerases increase predisposition to various cancers (1).
The function of POLN is currently uncertain, and several roles have been suggested. It has been reported that siRNA mediated knockdown of POLN-sensitized human cells to DNA cross-linking agents (16,17). However, POLN Ϫ/Ϫ chicken DT40 cells were not sensitive to mitomycin C, cisplatin, or camptothecin (18,19). Instead, it was proposed that POLN functions in homologous recombination reactions in chicken cells, leading to immunoglobulin V gene diversification by gene conversion (19). In mouse tissues, expression of Poln can be detected by northern blotting only in the testis (2). It is uncertain whether POLN is significantly expressed in other tissues or during development and whether the gene is essential for embryogenesis.
Previous studies of recombinant human POLN also hint at diverse functions for the protein by revealing several unique biochemical properties. The human enzyme has efficient strand displacement activity and low fidelity steady-state incorporation of T opposite template G (3,20,21). In vitro, POLN is proficient in the accurate bypass of major groove DNA lesions including a Tg 3 and major groove DNA-peptide and DNA-DNA cross-links (3,22). POLN cannot bypass a number of other DNA modifications, including an abasic (AP) site, a cisplatin-induced intrastrand d(GpG) cross-link, a cyclobutane pyrimidine dimer, a 6-4 photoproduct, or minor groove DNA peptide or DNA-DNA cross-links (3,22). We found that evolutionarily conserved residues in the O-helix of POLN are critical for the low fidelity and bypass activity of human POLN (4). However, when the O-helix of KlenTaq, a high fidelity A-family DNA polymerase, was replaced with the corresponding sequence from POLN, the fidelity of the mutant KlenTaq was higher than that of POLN and similar to that of the parental wild-type KlenTaq (23). The result suggested that the O-helix of POLN is not the only determinant critical for unique properties of human POLN. It is important to examine the fidelity properties of POLN in other species.
The POLN gene is present in the genomes of deuterostomes, including vertebrates. Here, we describe the restricted expression of POLN in the zebrafish Danio rerio. We report the discovery of an unusual overlapping relationship between the POLN and HAUS3 genes in vertebrates. These two genes share the same first exon, but they have very different expression patterns. We also found that ectopically expressed POLN can interact with protein components of the DNA recombination machinery.

Experimental Procedures
Isolation of the Zebrafish DNA Polymerase N (DrPOLN) Gene-Searches of the Zebrafish Model Organism Database revealed a zebrafish chromosome 7 genomic DNA sequence, NW_001879254 (NCBI accession number), which encodes several exons homologous to the human POLN polymerase domain. From this sequence, primers were designed to clone the zebrafish coding sequence by 3Ј-and 5Ј-rapid amplification of cDNA ends (BD Biosciences SMART RACE cDNA amplification kit). Total RNA was prepared from zebrafish testes using TRIzol (Life Technologies, Inc.). The full-length cDNA was cloned into plasmid pCR4-TOPO (Invitrogen), and the DrPOLN cDNA sequence was submitted to NCBI, accession number DQ630550.
Construction of DrPOLN Derivatives-We were unable to express full-length DrPOLN in Escherichia coli, but we succeeded with a construct beginning after the ninth Met, encoding amino acids 276 -1146. This was amplified by PCR from the full-length DrPOLN plasmid using the primers 5Ј-CACCGAA-AACTCTCCAGATGCCAAAAGATG-3Ј (for the 5Ј end) and 5Ј-ATATATGAATTCCTACTTGTCGTCATCGTCTTTGT-AGTCGGCAGAAGTTGCTGTAGCGGTG-3Ј (for the 3Ј end) and cloned into plasmid pENTR/D-TOPO (Invitrogen). After DNA sequencing, the cDNA was transferred into plasmid pDEST17 (Invitrogen) resulting in a protein tagged with six His residues at the N terminus (contributed by the pDEST17 vector), and a FLAG tag at the C terminus. Primers containing DrPOLN point mutations (altered DNA sequences are underlined) were synthesized as follows: 5Ј-CTTTCCTCTCT-GCAGCTTTCTGTCAGGTGGAG-3Ј and 5Ј-CTCCACCTG-ACAGAAAGCTGCAGAGAGGAAAG-3Ј (for D902A); 5Ј-CAGAGAGCAGGCCAAGGCGATCGTCTACTCTGTG-3Ј and 5Ј-CACAGAGTAGACGATCGCCTTGGCCTGCTCT-CTG-3Ј (for R957A). Site-directed mutagenesis was performed by using the QuikChange II site-directed mutagenesis kit (Stratagene). To generate D902A and R957A mutations, the pDEST17 vector carrying DrPOLN (amino acids 276 -1146) was used as a template. Recombinant POLN derivatives were bacterially expressed and purified as reported (3,4). These proteins were concentrated by NANOSEP 30K (PALL) and stored in buffer (50 mM sodium phosphate, pH 7.0, 300 mM NaCl, 10% glycerol, and 0.01% Nonidet P-40). Soluble full-length DrPOLN could not be purified under these expression conditions. Human POLN and RB69 gp43 were purified as reported (3,24) and were used as controls.
Oligonucleotide Substrates-Primer oligonucleotides were purchased from Bio-Synthesis or Sigma Genosys, purified by gel extraction, and 5Ј-labeled using [␥-32 P]dATP with polynucleotide kinase. Oligonucleotides containing a Tg were synthesized as described (3). Substrates for DNA polymerase assays were constructed by annealing 5Ј-32 P-labeled CACTGACTG-TATGATG-3Ј primer to 3Ј-GTGACTGACATACTACXTCT-ACGACTGCTC-5Ј template. The first template base (denoted by X) was T, G, or Tg. To form the nicked substrate, 5Ј-AAG-ATGCTGACGAG was additionally annealed to a template where X ϭ T.
Translesion DNA Synthesis Bypass Efficiency Reactions-The assays were performed as reported (3,4). The 5Ј-32 P-labeled primers (5Ј-CACTGACTGTATGA-3Ј or 5Ј-CACTGACTGT-ATGAT-3Ј) in Fig. 7 are similar but are two or one nucleotide shorter than the primer used for DNA polymerase assays in Fig.  6 (5Ј-CACTGACTGTATGATG-3Ј). The 5Ј-32 P-labeled 14-or 15-mer primer and a 30-mer template (sequences given above) were annealed at a molar ratio of 1:1. Primer extension reactions with DrPOLN and R957A were as described above. 10 l of reaction mixtures were incubated at 37°C for 2, 4, and 6 min and diluted in 10 l of formamide stop buffer. Products were heated at 95°C for 3 min and separated on a denaturing 20% polyacrylamide, 7 M urea gel. Product bands were quantified by PhosphorImager, and the values were used to calculate the probability of termination of processive synthesis and the insertion efficiencies at each template nucleotide. The termination probability at any position (N) is defined as the band intensity at N divided by the total intensity for all bands ՆN. The insertion probability at any position (N) is defined as the intensity at bands ՆN divided by the intensity at bands ՆN Ϫ 1. The extension probability at any position (N) is defined as the band intensity ՆN ϩ 1 divided by the intensity at bands ՆN. The bypass probability at position N is defined as the band density ՆN ϩ 1 divided by the intensity of ՆN 1 . To detect the bypass efficiency, the bypass probability (damaged) is divided by the bypass probability (undamaged) as described (25). The values are averages from two independent experiments at reaction intervals from 2, 4, and 6 min.
Cloning of 5Ј-Untranslated Region (UTR) of Human DNA Polymerase N (HsPOLN)-The 5ЈUTR of HsPOLN was isolated from human testis total RNA (Clontech) using a SMART RACE cDNA amplification kit (BD Biosciences). The 5ЈUTR was cloned into pCR4-TOPO (Invitrogen) and sequenced.
Quantitative PCR (qPCR) Assay-Total RNA was extracted using TRIzol, and RNA integrity was assessed using the Agilent 2100 bioanalyzer (Agilent Technologies, Inc.). Total RNA (1 g) was then used as template to synthesize cDNA with the High Capacity cDNA archive kit (Applied Biosystems). qPCR was then performed on the Applied Biosystems 7900HT fast real time PCR System (Applied Biosystems). Custom assays for zebrafish, DrPOLN, DrPOLQ, DrHAUS3, and Dr␤-ACTIN, and Homo sapiens, HsPOLN, HsPOLQ, and HsHAUS3, were designed using FileBuilder 3.1 software (Applied Biosystems) and ordered from Applied Biosystems. TaqMan primer and probe sets for each gene are shown in Table 1. TaqMan primers and probe set for human GAPDH was purchased from Applied Biosystems. Triplicate qPCRs each containing cDNA representing 40 ng of reverse-transcribed total RNA were then assayed for transcript quantity with Dr␤-ACTIN and HsGAPDH serving as endogenous controls to normalize input RNA levels. The absolute quantity of transcripts for the gene of interest in each sample was determined using the generated standard curves and the Sequence Detection Software version 2.2.2 (Applied Biosystems). Standard curves for each gene were determined using the following plasmids: pDEST17 carrying cDNA coding 276 -1146 amino acids of DrPOLN; full-length open reading frame (ORF) of DrHAUS3 cloned into pCR4; DrPOLQ clone (clone ID 8345083; Open Biosystems); fulllength HsPOLN ORF cloned into pDEST17 (4); HsPOLQ inserted into pFASTBac (11); HsHAUS3 clone (clone ID 3534250; Open Biosystems).
Northern Hybridization-Total RNA was purified from zebrafish tissues using TRIzol. For isolation of poly(A) ϩ RNA from total RNA, an Oligotex Direct mRNA mini kit (Qiagen) was used. 2 g of each poly(A) ϩ RNA was separated on a 1.0% formaldehyde-agarose gel with 1ϫ MOPS buffer. After soaking in 50 mM sodium hydroxide for 25 min and twice in 200 mM sodium acetate, pH 4.0, for 20 min, the gel was transferred to a nylon membrane (BrightStar-Plus; Ambion) with 20ϫ SSPE. The membrane was UV cross-linked, dried at 80°C for 2 h, and stored at Ϫ20°C. After prehybridization with ULTRAhyb (Ambion) containing 1 mg/ml torula yeast total RNA (Sigma), the filter was probed with 32 P-labeled DrPOLN cDNA (DQ630550), DrHAUS3 cDNA (BC124280), or Dr␤-ACTIN cDNA (BC063950) at 42°C for 16 h, followed by washing twice with 2ϫ SSPE ϩ 1% SDS at room temperature for 15 min and twice with 1ϫ SSPE ϩ 0.1% SDS at 50°C for 20 min. Blots were exposed to Kodak X-Omat XAR film.
5Ј-TCTCAACAGATAAATCCAAGGAATACCATTG 5Ј-CTCATAGGCTTTACCAAGTTT 5Ј-AACAATTCTTTTTTCTTATTCTCTCCCTCCA In Situ Hybridization-Zebrafish embryos were fixed in 4% formaldehyde in PBS and processed for whole mount in situ hybridization as described (26). pCR4 plasmid containing 895 bp of POLN cDNA, including 400 bp of its 3Ј-untranslated region, and pCR4 plasmid containing the full-length open reading frame of HAUS3 were used as templates for in vitro transcription to obtain antisense and sense probes. Antisense RNA probes for POLN and HAUS3 were made by digestion of the vectors with PmeI and transcription with T7 RNA polymerase, and sense probes were made by digestion of the same vectors with NotI and transcription with T3 RNA polymerase. Probe RNA was labeled with digoxigenin-dUTP using an RNA labeling kit (Roche Applied Science).
Zebrafish POLN Morpholino Knockdown-Zebrafish embryos were obtained from in-crossing wild-type adults maintained at 28.5°C. A complementary MO (5Ј-TGCAGAGG-TAGCTCTCCATGTTCGT-3Ј) targeting the initiation codon of zebrafish POLN was obtained from GeneTools LLC. Embryos were injected at the one-cell stage with 2.5, 5, or 10 ng of POLN MO. The control group was mock-injected.
Affinity Purification of POLN and FANCL Complexes-HeLa S3 cells stably expressing FLAG-HA epitope-tagged POLN or FANCL were grown to 1.0 ϫ 10 6 cells/ml as 9 liters of suspension cultures (28,29). The supernatant, nuclear extract, and chromatin fractions were prepared from the cells, and the POLN and FANCL complexes were immunoprecipitated from the nuclear extracts by incubating with M2 anti-FLAG-agarose gel for 4 h with rotation in the presence or absence of 50 units/ml Benzonase nuclease (Novagen). After an extensive wash with buffer 0.1B (100 mM KCl, 20 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 10% glycerol, 1 mM PMSF, 0.1% Tween 20, 10 mM ␤-mercaptoethanol), the bound proteins were eluted from M2-agarose by incubation for 60 min with 0.2 mg/ml FLAG peptide (Sigma) in the same buffer. 100 l of FLAG antibodyimmunoprecipitated material was further purified by immunoprecipitation with anti-HA 12CA5 antibody conjugated to protein A-Sepharose (GE Healthcare). The bound proteins were washed with 0.1B and eluted with 17 ml of 0.1 M glycine HCl, pH 2.5. After the elution, the pH was neutralized with 3 l of 1 M Tris-HCl, pH 8.0. To verify all proteins found in each complex by immunoblotting, only one gel was used per complex. Each membrane was cut horizontally into sections to immunoblot specifically for proteins of various sizes. Proteins were identified by LC-MS/MS using either the Proxeon Easy-nLC II or the Dionex Ultimate 3000 RSLCnano LC coupled to the Thermo Velos Pro or the Orbitrap Elite by analysis as reported in detail previously (13). Protein identification established greater than 99.9% protein probability assigned by the Protein Prophet algorithm, with a minimum of two peptides at 95% peptide probability. Peptide and protein false discovery rates were calculated as 0.0% by Scaffold. Abundant proteins commonly found in immunoprecipitation experiments with these epitope tags were eliminated from consideration (30 -33). Protein identifications were checked for agreement with the molecular mass predicted from the relevant gel slice.

Results
Vertebrate DNA Polymerase -To extend studies of the expression and function of POLN, we isolated cDNA for POLN from the zebrafish D. rerio POLN (DrPOLN). Initial database searches found a genomic DNA sequence (NW_001879254) encoding several predicted exons homologous to human POLN (HsPOLN). Although most zebrafish cDNAs can be obtained from RNA of embryos, we were not able to amplify POLN from this source nor were we able to obtain POLN cDNA from pooled RNA extracted from whole adult fish. However, POLN cDNA sequences were readily recovered from RNA prepared from pooling the isolated testis regions of 25 adult fish. The entire DrPOLN cDNA was assembled by reverse transcriptase-PCR using 3Ј-and 5Ј-RACE-PCR techniques. This revealed that the zebrafish POLN genomic DNA includes 26 exons, encoding a protein of 1146 amino acids (sequence deposited as DQ630550). This is larger than human (900 amino acids, AAN52116) or mouse POLN (866 amino acids, AAN39837) (2). Protein alignment shows that DrPOLN has a longer intervening sequence between a short conserved N-terminal end region (designated POLN-N) and the DNA polymerase domain (Fig.  1B). The DNA polymerase domain of DrPOLN is 52.2% identical (71.7% similar) to HsPOLN and 56.7% identical (77.3% similar) to mouse POLN (MmPOLN), and all residues essential for DNA polymerase activity are conserved between predicted fish and mammalian POLN (Fig. 1B).
POLN and HAUS3 Transcripts Overlap, but Have Distinct Expression Patterns-Exons encoding the HAUS3 protein are located just upstream of and closely adjacent to POLN (Fig. 2). HAUS3 encodes subunit 3 of the multisubunit augmin protein complex, critical for regulation of centrosome and spindle integrity. This syntenic relationship appears to be conserved in vertebrates. Surprisingly, our searches of transcript databases revealed that the zebrafish POLN transcript overlaps with the 5ЈUTR of HAUS3, indicating that both share the first exon ( Fig.  2A). To investigate this further, we used RACE-PCR to determine the 5ЈUTR sequence of human POLN. All of the 5ЈUTR sequence clones for HsPOLN identified in this analysis overlapped with the 5ЈUTR of HsHAUS3 (Fig. 3A). Similarly, the 5ЈUTR of the Poln transcript isolated from mouse testis RNA (2) overlaps with the 5ЈUTR sequence of MmHaus3 (Fig. 3B). Thus for zebrafish, human, and mouse POLN, experimental data show that the coding exons for another gene (HAUS3) reside within the first intron (Fig. 2B), a unique arrangement among DNA polymerases.
In human cell lines, the pattern of histone H3K27 acetylation, H3K4 trimethylation, and a CpG island suggests a single promoter upstream of the shared first exon (ENCODE) (Fig. 2C) (34). However, the expression patterns of POLN and HAUS3 are quite different, which can be explained by control at the level of alternative splicing of the primary transcript. In zebrafish, POLN is preferentially expressed in testis (Fig. 4A). In comparison, HAUS3 and POLQ are expressed during embryonic development and in several tissues examined (Fig. 4, A and  B). Consistent with this, HAUS3 and POLQ are also broadly expressed in human tissues (Proteomic DB).
These observations were confirmed by in situ hybridization in zebrafish embryos. HAUS3 transcripts were detected at all embryonic and larval stages analyzed, although POLN was not detectable by this technique (Fig. 4C). We conclude that POLN is expressed weakly or not at all in zebrafish embryos, which agrees with our inability to recover the POLN mRNA from embryos by RT-PCR. We injected a POLN-specific antisense MO into fish embryos. Eighty three, 62, and 22 embryos were injected with 2.5, 5, and 10 ng of POLN MO, respectively. Thirty two control embryos were mock-injected. However, no alteration in development was detected with either POLN-specific or control MO. This suggests that POLN does not have an essential function in early development, consistent with the lack of expression in the embryo.
The expression patterns of POLN and HAUS3 were also distinct in mammals. We compared expression of POLN, POLQ, and HAUS3 in mRNA from human testis and human cell lines (Fig. 5). All three genes were readily detected in testis. In cultured cells, only low levels of partial POLN transcript were detectable by RT-PCR (Fig. 5A) and by real time PCR (qPCR) (Fig. 5B). Using RT-PCR, we designed primers to test whether POLN and HAUS3 are independent mature spliced transcripts. The F1-B2 primer pair amplified a product from human testis of the size expected for POLN but not for a fusion of HAUS3 and POLN transcript (Fig. 5A). Consistently, no evidence for a fused transcript was found in the 5Ј-RACE experiments (Fig. 3A). Full-length POLN transcript was not detected by this primer set in the human 833K or 293T cell lines, although a transcript representing a portion of the mRNA could be detected (primer set F4 ϩ B2). The F1 ϩ B1 primer set yielded two major bands and one minor band. The major bands are consistent with the predicted size of the two documented transcript variants of human HAUS3 (accession numbers NM_001303143 and NM_024511). A third lower molecular weight band arising with this primer set may also be an alternatively spliced product.
We also examined expression of Haus3 and mRNA encoding DNA polymerases in mouse testis and embryonic stem (ES) cells. Transcripts for Haus3, Poln, Rev3L (the catalytic subunit of DNA polymerase ), and PolL (DNA polymerase ) were  readily detected in mouse testis (Fig. 5, C and D). In contrast, mouse ES cells expressed Haus3, Rev3L, and PolL, whereas Poln was not detected (primers 1 ϩ 5 and 4 ϩ 5, Fig. 5C). As found with human cells, no transcript fusing both Haus3 and Poln was detectable (primers 2 ϩ 5, Fig. 5C).

Zebrafish POLN Retains Strand Displacement and Bypass
Activities-The DrPOLN cDNA encoding amino acids 276 -1146 was expressed in E. coli, tagged with six His residues at the N terminus, and a FLAG epitope tag at the C terminus. An active site mutant (D902A) and a mutant with a substitution in an evolutionarily conserved residue of POLN (R957A) were also expressed (Fig. 6A). Asp-902 corresponds to a highly con-served residue in motif 3 of A-family DNA polymerases and is important in coordinating bivalent metal ions to interact with an incoming deoxynucleotide triphosphate (Fig. 1B) (3, 35). Arg-957 corresponds to the Lys-679 of human POLN, important for bypass activity and fidelity (Fig. 1B) (4,5). Proteins sequentially purified on FLAG antibody beads and metal affinity resin migrated near the expected molecular mass of 104 kDa (Fig. 6A). DrPOLN was able to extend DNA on a primed template, although the active site mutation (D902A) abrogated its DNA polymerase activity (Fig. 6B). No exonuclease activity was detected when enzyme and substrate were incubated without dNTPs, consistent with the lack of critical conserved residues  : panels a, c, i, and k are 10 somites; panels b, d, j, and l are 18 somites; panels e, f, m, and n are 24 hpf; panels g, h, o, and p are 48 hpf. for 3Ј-5Ј-exonuclease activity (Fig. 1B) (36), as found with HsPOLN. DrPOLN efficiently bypassed a (5S)-thymine glycol (5S-Tg) in DNA. Quantification ( Fig. 7 and Table 2) showed that the Tg bypass efficiency of DrPOLN was 18.0%, similar to the 15.8% observed for human POLN (4). Like HsPOLN, DrPOLN also showed efficient strand displacement activity on a nicked DNA substrate (Fig. 6B). For comparison, the DNA polymerase RB69 gp43 did not show these activities on the same substrates (Fig. 6B).
A conserved Lys or Arg residue (Lys-679 in HsPOLN and Arg-957 in DrPOLN) was identified in the "O-helix" of motif 4 in the finger subdomain (Fig. 1B) (4). The corresponding residue is one of the most important for controlling fidelity of prokaryotic polymerase I and is a nonpolar Ala or Thr in those enzymes (37,38). The residue was important for low fidelity and bypass activity in HsPOLN. K679A or K679T HsPOLN (polymerase I-like) mutants showed higher fidelity than wildtype HsPOLN but did not bypass a 5S-Tg efficiently. A K679R HsPOLN (DrPOLN-like) mutant bypassed a 5S-Tg as efficiently as wild-type HsPOLN (4). We examined the corresponding residue in zebrafish POLN. The R957A mutation reduced 5S-Tg bypass efficiency by 5-fold, without significantly affecting DNA polymerase activity on undamaged DNA (Fig.  7). Thus, a basic residue at this position is important for translesion synthesis activity of POLN.
Low fidelity favoring incorporation of T for template G is a biochemical property of HsPOLN (3)(4)(5). However, this tendency was not evident in DrPOLN (Fig. 8). Wild-type HsPOLN but not the K679A or K679T HsPOLN mutants efficiently incorporated T for G at the optimal pH of 8.8 (3). When the pH was reduced to 8.0 or 7.2, the DNA polymerase activity and the T misincorporation activity of HsPOLN were both reduced as reported (Fig. 8, B and D) (23). Unlike HsPOLN, wild-type DrPOLN did not efficiently incorporate T opposite template G even at pH 8.8; the Arg residue at 957 did not influence T or G misincorporation (Fig. 8, A and C). In addition, DrPOLN efficiently incorporated G opposite G unlike HsPOLN (Fig. 8, A  and B).
POLN Protein Associations-It has been suggested that POLN interacts with several proteins, including HELQ, Fanconi anemia (FA) core complex proteins, and FANCD2 (16). It is intriguing that this set of proteins has testis-related functions. Helq knock-out male mice have significantly smaller testes (14), targeted disruption of several FA genes caused impaired fertility in mice (39), and infertility is a common feature among male FA patients (40). To test these proposed interactions and to uncover possible molecular pathways relevant to POLN, we searched for proteins with the potential to associate with POLN in vivo. POLN was stably expressed as a FLAG and HA epitope fusion (ePOLN) in HeLa S3 cells. ePOLN was recovered from nuclear extracts by sequential immunoprecipitation with anti-FLAG and anti-HA antibodies (28,29). The immunoprecipitate was separated on a gradient gel (Fig. 9A), and proteins from gel sections were identified by liquid chromatography-mass spectrometry. The data were filtered to eliminate common s. Among the resulting top 150 ranking hits there were nine DNA repair-related proteins (supplemental Table 1). Six of these (BRCA1, BRCA2, BARD1, PALB2, FANCJ, and RBBP8/CtIP) are components of the A, B, and C BRCA1-associated complexes related to homologous recombination ( Fig. 9A and Table  3) (41). None of these proteins were present in a control FLAG-HA purification from HeLa S3 cells transfected with empty control vector. In parallel experiments in the same system using FLAG-HA-tagged HELQ as bait, none of these proteins were detected in the HELQ complex (13). Although FA

TABLE 2 Bypass, insertion, extension probabilities and bypass efficiency
The bypass probability at position N is defined as the band density Ն(N ϩ 1) divided by the intensity of Ն(N 1 ). The insertion probability at any position (N) is defined as the intensity at bands Ն(N) divided by the intensity at bands Ն(N Ϫ 1). The extension probability at any position (N) is defined as the band intensity Ն(N ϩ 1) divided by the intensity at bands Ն(N). To detect the bypass efficiency, the bypass probability (damaged) is divided by the bypass probability (undamaged) (25). core complex proteins FANCD2 and HELQ were previously proposed as POLN-interacting partners (16), no peptides representing any of these proteins were identified as ePOLN-associated proteins (supplemental Table 1), just as POLN was not detected in the HELQ complex in the previous study (13). Immunoblotting confirmed that BRCA1 was present in the immunoprecipitate and that FANCD2 and HELQ were present in the input fraction but undetectable in the POLN immunoprecipitate (Fig. 10A). To confirm POLN-BRCA1 and POLN-FANCJ interactions in another cell line, V5-tagged POLN was coexpressed with FLAG-HA-tagged BRCA1 or FANCJ in 293T cells. After immunoprecipitation from whole-cell extract with anti-FLAG and anti-HA antibodies, V5 antibody was used to identify POLN in the immunoprecipitate. Co-immunoprecipitation was observed between POLN and BRCA1 (Fig. 9B) and POLN and FANCJ (Fig. 9C). We also tested for an association of the FA core complex with POLN. The same system was employed but with FANCL as bait. FANCL was stably expressed as a FLAG and HA epitope fusion (eFANCL) in HeLa S3 cells, with or without exposure to hydroxyurea (HU). After sequential immunoprecipitation with anti-FLAG and anti-HA antibodies, an FA core complex protein, FANCL, was identified in the immunoprecipitate, as expected (Fig. 10B). However, POLN was not present, consis-tent with its normally low or absent expression in human cultured cells.

Discussion
These investigations by biochemical, molecular genetic, and proteomic analysis answer several outstanding questions about vertebrate POLN.
First, the expression of POLN in vertebrate cells is very limited and tissue-specific. In zebrafish, mouse, and humans, POLN is preferentially expressed in testis and very weakly in other tissues or cells in culture. POLN was originally assembled by analyzing transcripts and expressed sequence tags from human cell lines (2). However, only short fragments of POLN are readily isolated by RT-PCR from cDNA of human cell lines (Fig. 5A). RACE-PCR was used to clone the 5ЈUTR sequence of human POLN from testis cDNA (Fig. 3A) without difficulty, but we were unable to do this from human cell lines. In the human cell lines examined so far, POLN is subject to extensive alternative splicing that gives rise to biologically inactive transcripts (2). This is important to consider when evaluating data from microarrays or RNA-sequence experiments, as most POLN mRNA expression will represent biologically inactive transcripts. Second, we report new evidence that the POLN and HAUS3 genes share the same first exon, in an evolutionarily conserved manner. From analysis of mouse database annotations, it was suggested that HAUS3 and POLN do not overlap (42), but our primary analysis of sequences from three vertebrates indicates that POLN and HAUS3 share a single promoter and first exon. Peaks of H3K27 acetylation and H3K4 trimethylation, and a CpG island, which are often found near active regulatory elements, were identified near the first exon of POLN and HAUS3 but not around the other exons. POLN and HAUS3 may initiate transcription from the same promoter, with their expression regulated by tissue/cell type-specific splicing of the transcripts. A recent study underlines that modulation of splicing can indeed influence POLN mRNA expression levels (43). Small molecule compounds were identified that shifted RNA splicing of the SMN2 gene toward production of full-length mRNA.
Intriguingly, POLN was one of only six genes in human cells that increased in expression by a factor of Ͼ2 after treatment with one of the most specific small molecule compounds, SMN-C3 (43). In a large family of genes involved in DNA or RNA metabolism, POLN was the only such regulated gene. POLN is unique among mammalian DNA polymerases in its conserved expression with an overlapping gene. The possible selective advantage for this arrangement is not known, but it is notable that it shares a promoter region with a ubiquitously expressed housekeeping gene (HAUS3). This might provide a mechanism for a small amount of POLN to be produced by alternative splicing in any cell type, when POLN is needed for a specialist function. RNA transcript splicing patterns can vary considerably in different tissues (44), which could allow large amounts of POLN to be produced in specific situations when necessary, as appears to be the case in testis. It remains to be , denoted as wild-type and R957A, respectively, were incubated with 300 fmol of 5Ј-32 P-labeled 16-mer primer annealed to a 30-mer DNA template in the presence of four or one of the indicated dNTPs (100 M) for 10 min. The first template base denoted by X was G or T. Template sequences are indicated above the panel. NE indicates no enzyme. B, as described for A, using HsPOLN and HsPOLN (K679A), denoted as wild-type and K679A, respectively. C, 23 nM DrPOLN and DrPOLN (R957A), denoted as wild-type and R957A, respectively, were incubated with 5Ј-32 P-labeled 16-mer primer annealed to a 30-mer DNA template, in which the first template base was G in the presence of all four dNTPs or dTTP (100 M) for 10 min in indicated pH conditions. D, as described for C, using HsPOLN and HsPOLN (K679A), denoted as wild-type and K679A, respectively. The percentage (%) of the product extension from the primer is shown below each lane in C and D.
seen whether the levels of HAUS3 are also subject to splicing modulation. This could be relevant as HAUS3 is sometimes mutated in breast cancer. In a study following a lobular human breast tumor, a mutation in HAUS3 was one of only five nonsynonymous coding mutations that were prevalent in the primary breast tumor and remained in the metastatic cancer nine years later (45).
Third, it was not known whether POLN is an essential gene for embryonic development. We found that it is not essential in zebrafish. This is because POLN is not appreciably expressed in zebrafish embryos, and because no phenotypic differences were identified after the injection of control and POLN-specific antisense morpholino oligonucleotides.
One possibility is that POLN may have a function in testis, where the gene is preferentially expressed. Morpholino antisense oligonucleotides can be used to analyze phenotypes in early developmental stages (46) but not readily in adult testis. In human cells, POLN is also highly expressed in testis but very weakly expressed in cell lines, even those derived from testicu-lar cancers (1618K, 833K, SuSa, and TERA1). Here, we identified the potential for POLN to interact in a protein complex with homologous recombination-related proteins, including BRCA1 and downstream FA proteins, including FANCJ, BRCA2 (FANCD1), and PALB2 (FANCN). This is consistent with a possible function of POLN in homologous recombination in testis. However, our results do not support the previously proposed interactions of POLN with HELQ, FA core complex proteins, or FANCD2. During the process of meiotic recombination, the evolutionarily conserved strand displacement activity of POLN could be useful to synthesize DNA in D-loop recombination intermediates. Another possible function is chromatin remodeling in the XY body of spermatocytes, because BRCA1 has a role in the establishment of X-pericentric heterochromatin in testis (47).
Finally, we found that the ability to bypass thymine glycol lesions in DNA is a property conserved in the human and zebrafish enzymes and that a basic residue in the O-helix (human residue Lys-679) is crucial for this activity in both spe- FIGURE 9. POLN is associated with BRCA1, FANCJ, and other homologous recombination components. A, POLN-associated proteins were immunopurified from nuclear extracts prepared from HeLa S3 cells expressing FLAG-HA epitope-tagged POLN. The complex was sequentially purified with anti-FLAG and anti-HA antibodies, resolved by SDS-PAGE on a 4 -20% gradient gel, and visualized by silver staining. Approximate migration positions of proteins identified in gel sections are shown. B, V5-tagged POLN and FLAG-HA epitope-tagged BRCA1 were transiently co-expressed in 293T cells. The whole-cell extracts prepared from the transfected cells were sonicated and incubated in the presence of benzonase. FLAG-HA epitope-tagged BRCA1 was immunoprecipitated (IP) with anti-FLAG and anti-HA antibodies. V5-tagged POLN in the immunoprecipitated samples was detected with anti-V5 antibody. C, interaction between V5-tagged POLN and FLAG-HA epitope-tagged FANCJ was examined similarly as described in B.  Q7Z5Q5͉DPOLN_HUMAN  100  1207  349  BRCA1  P38398͉BRCA1_HUMAN  208  23  15  BARD1  Q99728͉BARD1_HUMAN  87  15  8  PALB2  Q86YC2͉PALB2_HUMAN  131  17  11  BRCA2  P51587͉BRCA2_HUMAN  384  7  7  FANCJ  Q9BX63͉FANCJ_HUMAN  141  16  12  CtIP  Q99708͉COM1_HUMAN  102  15  8 cies. Tg is a major DNA lesion generated by reactive oxygen species that blocks the progression of replicative DNA polymerases (48). However, we do not yet know whether the ability of POLN to bypass Tg lesions is physiologically relevant (for example, in a testis-specific role). It is possible that the bypass activity is only an in vitro readout of the unusual active site of POLN, which normally functions in some other challenging role, such as strand displacement. One of the evolutionarily conserved sequence insertions in the DNA polymerase domain of POLN (4), called insert 2, forms a unique cavity in the DNA polymerase domain of POLN (5). The cavity allows POLN to generate and accommodate a looped-out primer strand (5), and it may also help POLN to bypass a Tg lesion by tolerating the distortion generated after incorporation of A opposite Tg (48). Similarly, human POLN is striking in its very high G to A base substitution rate. We found no indication of a marked tendency for zebrafish POLN to incorporate T opposite template G. In steady-state conditions, human POLN preferentially incorporates T opposite G but not in the pre-steadystate (21). The low fidelity of human POLN might be a result of assaying the enzyme at its optimum pH for activity, pH 8.8 (3). Other A-family DNA polymerases also show increased fidelity at lower pH (49,50). It remains to be determined whether nucleotide misincorporation is relevant to an in vivo function of POLN.