RAD51AP2, a novel vertebrate- and meiotic-specific protein, shares a conserved RAD51-interacting C-terminal domain with RAD51AP1/PIR51

Abstract

Many interacting proteins regulate and/or assist the activities of RAD51, a recombinase which plays a critical role in both DNA repair and meiotic recombination. Yeast two-hybrid screening of a human testis cDNA library revealed a new protein, RAD51AP2 (RAD51 Associated Protein 2), that interacts strongly with RAD51. A full-length cDNA clone predicts a novel vertebrate-specific protein of 1159 residues, and the RAD51AP2 transcript was observed only in meiotic tissue (i.e. adult testis and fetal ovary), suggesting a meiotic-specific function for RAD51AP2. In HEK293 cells the interaction of RAD51 with an ectopically-expressed recombinant large fragment of RAD51AP2 requires the C-terminal 57 residues of RAD51AP2. This RAD51-binding region shows 81% homology to the C-terminus of RAD51AP1/PIR51, an otherwise totally unrelated RAD51-binding partner that is ubiquitously expressed. Analyses using truncations and point mutations in both RAD51AP1 and RAD51AP2 demonstrate that these proteins use the same structural motif for RAD51 binding. RAD54 shares some homology with this RAD51-binding motif, but this homologous region plays only an accessory role to the adjacent main RAD51-interacting region, which has been narrowed here to 40 amino acids. A novel protein, RAD51AP2, has been discovered that interacts with RAD51 through a C-terminal motif also present in RAD51AP1.

Figure 1

Structure of the human RAD51AP2 gene, and sequence of the human RAD51AP2 proteins. (A) Comparing the cDNA sequence we have determined with the genomic sequence in GenBank, it was determined that RAD51AP2 consists of one very long exon followed by two short exons, the last of which completely encodes the RAD51-interacting domain. (B) The amino acid sequence of the entire human RAD51AP2 proteins predicted from our RAD51AP2 cDNA sequence (GenBank accession # DQ860102). Residues similar to a Walker A-type nucleotide-binding motif in human RAD51AP2 are in the light gray box. The long underlined region is missing in the mouse homolog, and the short black boxed residues are the two boundaries among the amino acids encoded by the three exons. The C-terminal 33 amino acids (C33) in the red box are sufficient for RAD51-interaction in the Y2H system.

Figure 2

Northern blot analysis of the RAD51AP2 gene in human tissues. Lanes contain ∼2 μg of poly A⁺ RNA from the human tissues indicated. PBL, peripheral blood leukocytes; SI, small intestine. Upper panel, hybridization with RAD51AP2 cDNA probe. Lower panel, hybridization with human β-actin cDNA probe. The positions of RNA size markers, in kilobases, are indicated on the right.

Figure 3

PCR analysis showing meiotic-specific expression of human RAD51AP2 in adult testis and fetal ovary, but ubiquitous expression of RAD51AP1. (A) RAD51AP2 (AP2 in figure) transcript in human adult tissues (note: RAD51AP2 is top band in testis sample only and PCR primers are the bottom bands). LC—leukocytes, SI—small intestine. The size and approximate mobility (based on size standards) of the PCR products are indicated at the right side of each panel. (B) RAD51AP2 transcripts in human fetal tissues. (C and D) RAD51AP1 (AP1 in figure) transcript in human adult tissues. SM—smooth muscle. (E) RAD51AP1 transcripts in human fetal tissues. Note: RAD51AP1 has multiple variants in adult testis and fetal ovary. (B–E) DNA size markers (400 and 800 bp) in right lane. (F) RAD51AP2 transcript is not induced in U20S cells exposed to 2 Gy X-rays. End-point RT–PCR was carried out from 1st strand cDNA samples in a multiplex PCR reaction amplifying both a 838 bp product for β-actin and a 570 bp product for RAD51AP2 (as in A and B). A product for RAD51AP2 was not obtained (as indicated by the absence of a 570 bp fragment). DNA size markers (1200, 800 and 400 bp) in right lane. (G) Same samples as in F: End-point RT–PCR for β-actin and GADD45A (171 bp positive control for the induction of transcript after X-rays). DNA size markers (2000, 1200, and 800 bp in top panel, and 200 and 100 bp in bottom panel) in right lane.

Figure 4

RAD51AP2 and RAD51 proteins interact in human cells, and the interaction is enhanced by DNA damage. (A) Human HEK293 cells were transiently transfected with plasmids encoding the following T7 epitope-tagged proteins: Vmw65 protein of herpesvirus (lane 1); 293amino acid RAD51AP2 polypeptide (lane 2); RAD51AP2 protein deleted for its last 57 residues (lane 3). Immunoprecipitation with anti-T7 antibody followed by protein A-agarose beads was carried out, and fractions of cell lysates before IP, supernatant of IP, and IP pellets were analyzed by Western blotting using anti-T7 (top panel) or anti-RAD51 antibody (lower panel). Secondary antibodies were conjugated to alkaline phosphatase, which was detected using chromogenic substrates. (B) Transfection with plasmid encoding the 293 amino acid RAD51AP2 polypeptide was performed as in panel A, and 0.5 μg/ml MMC, 100 μg/ml MMS or 10 μg/ml cycloheximide (Cyc.) was added to individual cultures 3 h before harvesting cells for IP. IP and Western blotting were done as in panel A, except that chemiluminescent substrates were used to detect peroxidase conjugated to secondary antibodies. To facilitate the relative comparison of RAD51 signals from IP pellets (shown in panel C), the amount of these fractions loaded on the gel was adjusted in an effort to get similar intensities of the T7-tagged RAD51AP2 signal in each lane. The residual differences were normalized to the control (untreated cells) for comparison of the amounts of RAD51 that coprecipitated with RAD51AP2. (C) Quantitation of RAD51-RAD51AP2 coimmunoprecipitation from panel B. The signals from precipitated RAD51 in IP pellet fractions were measured using a densitometer and then normalized for the signal from anti-T7 analysis of the same fractions. The relative increase of RAD51 signal from treated cells as compared to the signal observed from untreated cells (artificially set at 1.0 in left lane) is shown.

Figure 5

Sequence homology between the RAD51AP2 and RAD51AP1 and mutational analysis of their shared RAD51-interacting motifs. (A) Amino acid identities and similarities are indicated by dark blue and gray squares, respectively. Just the C-terminal domains of RAD51AP1 and of RAD51AP2 are shown. Under the sequences are the original residues that were mutated by site-directed mutagenesis. The underlined mutated residues significantly reduce the interaction with RAD51 (see panels B and C). Species abbreviations are as follows: Hs—Homo sapiens; Mm—Mus musculus; Gg—Gallus gallus. (B) Interactions in the Y2H system between RAD51 (i.e. pEG960) and both RAD51AP1 (i.e. pOK31, with all of RAD51AP1) and site-specifically mutated RAD51AP1 (in pOK31). Results are the average from three different colonies, with the standard error of the mean. (C) Interactions in the Y2H system between RAD51 (i.e. pEG960) and both RAD51AP2 (i.e. pDS439; pGBKT7-RAD51AP2-C33) and site-specifically mutated RAD51AP2 (in pDS439). Results are the average from two different colonies, with the standard error of the mean. * Indicate P <0.05 and ** for P ≤0.001 using the student T-test. Please note: the Y190-ura⁻ Y2H strain was used in these studies, and it gives lower β-galactosidase activity that the HF7c Y2H strain used in Table 1 and 2, and in addition, the RAD51AP2-C33 construct is in the pGBKT7 vector that also results in lower β-gal activity than similar pGBT9-derived constructs. The negative control was the original empty vector used for that construct, together with RAD51.

Figure 6

Truncation mapping and mutational analysis of the human RAD54-RAD51 interaction. (A) Amino acid sequence homology between human RAD54 and the RAD51-interacting domains of RAD51AP1 and RAD51AP2. Amino acid identities and similarities are indicated by dark blue and gray squares, respectively. Under the sequences are the original RAD54 residues that were mutated by site-directed mutagenesis. (B) PCR was used to make different truncation constructs of RAD54-N142 (amino acids 1–142) in pGBT9, and these truncations were tested for interaction with RAD51 (pEG960) in the Y2H strain Y190-ura⁻. Results are the average from three different colonies, with the standard error of the mean shown. Only one colony of the negative control (pGBT9 with pEG918) was quantified, but similar negative controls always give about this same result. (C) Mutations were introduced into the N-terminal fragments of RAD54 consisting of 142 or 89 residues (N142 and N89, respectively) and tested for their interaction with RAD51 in the two-hybrid system, using Y2H strain HF7c. The P85A mutation was made in RAD54-N89, while all the rest were made in RAD54-N142. WT, wild-type sequence. For identical constructs, the Y2H strain Y190-ura⁻ gives lower β-galactosidase activity that HF7c.