Abstract
Bloom syndrome (BS) is characterized by genomic instability and cancer susceptibility caused by defects in BLM, a DNA helicase of the RecQ-family (J. German and N. A. Ellis, The Genetic Basis of Human Cancer, pp. 301–316, 1998). RecQ helicases and topoisomerase III proteins interact physically and functionally in yeast (S. Gangloff et al., Mol. Cell. Biol., 14: 8391–8398, 1994) and in Escherichia coli can function together to enable passage of double-stranded DNA (F. G. Harmon et al., Mol. Cell, 3: 611–620, 1999). We demonstrate in somatic and meiotic human cells an association between BLM and topoisomerase IIIα. These proteins colocalize in promyelocytic leukemia protein nuclear bodies, and this localization is disrupted in BS cells. Thus, mechanisms by which RecQ helicases and topoisomerase III proteins cooperate to maintain genomic stability in model organisms likely apply to humans.
Introduction
BS 4 is a rare recessive autosomal disorder characterized by genome instability, elevated rates of most types of cancers, immunodeficiency, and hypogonadism (1) . The protein defective in BS, BLM, is a DNA helicase of the RecQ family (2 , 3) , which also includes Escherichia coli RecQ and Saccharomyces cerevisiae Sgs1p (4 , 5) . There are at least four human homologues in addition to BLM, including WRN, which is the protein defective in the human premature aging disorder WS (6) , and RECQL (7) . In all organisms examined, mutations in RecQ proteins cause various forms of genome instability; BS cells are typified by elevated levels of chromosome breaks, rearrangements, and sister chromatid exchange (1) . Genomic instability can account for the elevated rate of cancer in BS, but the precise mechanisms of the genomic instability are unknown. In particular, no interacting proteins have been identified for BLM. The human Topo III proteins are good candidates for RecQ-family partners because in S. cerevisiae the Sgs1p and Top3p proteins interact genetically and physically (4) . Topo III proteins are type IA topoisomerases that transiently cleave one DNA strand and selectively relax negative DNA supercoils. Biochemical experiments using E. coli RecQ and topoisomerase III proteins have demonstrated that together these proteins have a novel activity that enables them to pass double-stranded DNA molecules through one another (5) . It is not known whether human RecQ family proteins interact with topoisomerase III proteins. Here we describe an association between the BLM and Topo IIIα proteins in human cells.
Materials and Methods
Plasmids.
The Topo IIIα COOH-terminal fragment expression plasmid was generated by inserting the 1.2-kb TOP3α PpuMI-EcoRI cDNA fragment (gift of J. Wang, Harvard University) into the NheI and EcoRI sites of pET28a(+) (Novagen). Chicken anti-BLM antibodies were derived against nucleotides 3149–4328 of the human coding sequence cloned into pET28a via PCR.
Cells.
Lymphoblastoid cells AG07877 (WT) and AG04103 (WS, homozygous for R369X), both obtained from the Coriell Repository, and HG1525 (BS, homozygous for S595X, gift of J. German, Cornell Medical School) were grown in RPMI 1640, 4 mm glutamine, 1× Pen-Strep, and 15% heat-inactivated fetal bovine serum. WI38 (WT) and AG2940 (BS, gift of J. German, Cornell Medical School) fibroblasts were grown in DMEM/1 × Pen-Strep/20% fetal bovine serum.
Antibodies.
Topo IIIα antibodies were raised in chickens and rabbits (Covance) using the COOH-terminal 211 amino acids of Topo IIIα fused to a 6XHIS tag that had been expressed in E. coli BL21(DE3) and purified using Ni-NTA resin (Qiagen). Affinity-purified rabbit antibodies directed against the NH2-terminal portion of BLM were generated as described (8) . A BLM fragment encoding amino acids 3149–4328 was used to immunize chickens (Covance). Topo IIIα and chicken BLM antibodies were affinity purified using the immunizing antigen coupled to Aminolink resin (Pierce). RECQL antiserum (1145) was a kind gift of Debbie Stumpo and Perry Blackshear (7) . Anti-PML (PG-M3) was obtained from Santa Cruz Biotechnology. Anti-SCP3 antibody was the kind gift of Christa Heyting (9) .
Immunofluorescence.
Early passage WI-38 or AG2940 (BS) human diploid fibroblasts were grown on multiwell slides. Slides were prepared using preextraction and paraformaldehyde fixation as described (10) . Secondary antibodies were purchased from Vector Laboratories. Chicken antibodies were detected with biotinylated antichicken antibodies and then developed with Cy3-avidin (Amersham or Sigma). Meiotic spreads were prepared by standard techniques (11) from testes removed for prostate cancer or chronic pain, and indirect immunofluorescence was performed as for fibroblasts. To visualize the synaptonemal complex, anti-SCP3 was included in the tertiary incubation, and the slides were washed as above and then incubated in blocking buffer plus 1:100 AMCA-conjugated antirabbit (Vector) for an additional 30 min. Control experiments demonstrated that even when another rabbit antibody had been used during the first incubation, there was no detectable binding of AMCA antirabbit to the initial primary antibody. Stained slides were mounted in Vectashield (Vector Laboratories). Slides were imaged at ×100 on a CCD camera, and the images were pseudocolored and merged in Adobe Photoshop.
Extracts and Immunoprecipitations.
Whole-cell extracts were prepared from log-phase lymphoblastoid cells, essentially as described (12) , except that the final dialysis buffer contained 150 mm KOAc. Extract preparation included centrifugation at 150,000 × g to remove insoluble chromatin, followed by a 0.33 g/ml (NH4)SO4 precipitation step. For IPs, 100 μg of extract protein in a final volume of 25 μl IP buffer [20 mm TrisOAc (pH 7.9), 150 mm KOAc, 1 mm EDTA, 5% glycerol, 0.1 mm DTT, 0.05% NP40, 0.2 mg/ml BSA, 1 mm phenylmethylsulfonyl fluoride, and 5 μg/ml each aprotinin, leupeptin, and pepstatin A] were mixed with 2.5 μl of beads [protein A-Sepharose (Sigma) for rabbit antibodies or anti-IgY agarose (Promega) for chicken antibodies] for 15 min at 4°C, followed by a 16,000 × g spin at 4°C for 5 min to preclear the extract. The supernatant was mixed with 5 μl of fresh beads and antibody (equal concentrations of control or experimental antibody) and mixed at 4°C for 1 h. The supernatant was removed, and beads were washed once with 250 μl of IP buffer and then three times with 250 μl of wash buffer [20 mm TrisOAc (pH 7.9), 150 mm KOAc, 0.1 mm EDTA, 6 mm Mg(OAc)2, 0.1 mm DTT, 0.05% NP40, and 0.2 mg/ml BSA]. Proteins retained by the beads were separated by SDS-PAGE, followed by transfer to nitrocellulose and detection by standard immunoblotting using ECL.
Results
Humans and mice each possess two known TOP3 homologues, called Topo IIIα and Topo IIIβ. Topo IIIα is the most similar to yeast Top3p, sharing 44% sequence identity (13) . Antibodies were raised in chickens and rabbits to the COOH-terminal 211 amino acids of the human Topo IIIα protein. Because these residues are contained within a COOH-terminal extension that is not substantially conserved in Topo IIIβ, this strategy ensured that these antibodies would not cross-react with Topo IIIβ. Affinity-purified antibodies recognized a protein of Mr 110,000 on immunoblots of human cell extracts (Fig. 1A) ⇓ , consistent with the predicted size of Topo IIIα, as well a Mr 70,000 species judged to be a degradation product because it increases in relative abundance with freeze-thaws of the extract (not shown). Immunoprecipitation was used to test for association between BLM and Topo IIIα, using lymphoblastoid cell extracts from which bulk chromatin had been removed (see “Materials and Methods”). BLM coprecipitated with Topo IIIα using the chicken anti-Topo IIIα antibody 1003 (Fig. 1B) ⇓ . A second human RecQ-family helicase, RECQL, also coprecipitated with Topo IIIα (Fig. 1B) ⇓ . Ethidium bromide can disrupt protein-DNA interactions and so has been used to distinguish protein-protein interactions from those mediated by binding of factors to DNA (14) . Coprecipitation of BLM or RECQL with Topo IIIα persisted in the presence of 100 μg/ml ethidium bromide (Fig. 1B ⇓ , top) and also persisted after treatment of the precipitates with DNase I and RNase A at 37°C (Fig. 1B ⇓ , bottom). Thus, the associations did not depend on binding of the proteins to residual nucleic acids in the extracts although: (a) the reduction in the BLM-Topo IIIα association by ethidium bromide might indicate some stabilization of the complex by DNA; and (b) other factors could still be important for the observed associations. Similar results for BLM and RECQL were obtained with extracts of HeLa cells and using the anti-Topo IIIα antibodies 1002 or 1151 (data not shown). Topo IIα, nuclear factor-κB, proliferating cell nuclear antigen, or nucleophosmin did not coprecipitate with Topo IIIα, indicating specificity of the observed interactions (data not shown).
BLM and RECQL helicases coimmunoprecipitate with Topo IIIα. A, specificity of Topo IIIα antibodies. Immunoblots of lymphoblastoid whole-cell extracts were probed with equal concentrations of affinity-purified antibodies (Imm) or preimmune serum (Pre). 1002 and 1003 were raised in chickens, and 1151 was raised in a rabbit. Right, apparent molecular weight of marker proteins (in thousands). The Mr 70,000 band appears to be a degraded form because it increases in abundance with freeze/thaws of the extract. B, Topo IIIα precipitates contain BLM and RECQL. Affinity-purified Topo IIIα antibodies (1003) or preimmune serum was used to immunoprecipitate proteins from lymphoblastoid whole-cell extracts, and an immunoblot of the proteins was probed with antibodies to BLM or RECQL. EtBr, precipitations and washes contained 0.1 mg/ml ethidium bromide. 37°, precipitates were incubated in wash buffer at 37°C for 10 min; 37°/Nuc, incubation also included three units of RQ-1 DNase (Promega) and 1 μg of RNase A. Nuclease treatment was sufficient to digest 1.5 μg of DNA and 10 μg of tRNA added to control reactions, and endogenous nucleic acids were not detectable (not shown). C, BLM precipitates contain Topo IIIα. Affinity-purified rabbit anti-BLM antibodies or preimmune serum (Pre) was used to precipitate protein from whole-cell extracts (from WT cells, or cells lacking BLM or WRN, as indicated) and were probed with antibodies to Topo IIIα. Precipitations and washes contained 0.1 mg/ml ethidium bromide. D, RECQL precipitates contain Topo IIIα. Precipitations and immunoblotting were performed as in C, except that RECQL antiserum or control rabbit antiserum (Con) was used for precipitations.
To confirm the interactions, precipitations were performed in the reverse direction using antibodies to BLM and RECQL in the presence of ethidium bromide. Anti-BLM antibodies (8) coprecipitated Topo IIIα in extracts of WT or WS cells but failed to coprecipitate Topo IIIα from extracts from BS cells, thus demonstrating the specificity of the anti-BLM precipitations (Fig. 1C) ⇓ . Anti-RECQL antibodies (7) also coprecipitated Topo IIIα, and the interaction was independent of the presence of BLM or WRN (Fig. 1D) ⇓ . BLM and RECQL do not coimmunoprecipitate (data not shown), suggesting that Topo IIIα exists in distinct complexes with BLM or RECQL. We have not observed an interaction between Topo IIIα and WRN, despite efficient precipitation of WRN with antibodies directed against either the NH2 or COOH terminus of WRN (data not shown).
To identify cellular loci where Topo IIIα, BLM, and RECQL associate, the localization of these proteins in cultured primary human fibroblasts was examined by indirect immunofluorescence. Topo IIIα and BLM each were found diffusely throughout the nucleoplasm, in bright nuclear foci (Fig. 2A ⇓ , filled arrowheads), and occasionally in the nucleolus (Fig. 2A ⇓ , open arrowhead). No BLM staining was observed in BS fibroblasts (not shown). Costaining with rabbit anti-BLM and chicken anti-Topo IIIα revealed that most of the BLM foci contain Topo IIIα (Fig. 2A) ⇓ . Rabbit anti-RECQL antibodies revealed diffuse nuclear staining, as observed previously (7) , as well as discrete foci similar to those observed with BLM antisera. Costaining for RECQL and Topo IIIα also revealed substantial colocalization of the two proteins in a subset of the Topo IIIα foci (Fig. 2B) ⇓ . Greater than 90% of cells showed colocalization of Topo IIIα with the foci containing BLM or RECQL. Colocalization was apparent in the G1, S, and G2 phases of the cell cycle, as determined using primary fibroblasts synchronized by double thymidine block, but was not observed in cells arrested by serum starvation, indicating that association depends on the cells being in cycle (data not shown). Costaining with chicken anti-BLM and rabbit anti-RECQL revealed that most of the foci contain both helicases (Fig. 2C) ⇓ . The number and appearance of these foci suggested that they might be PML NBs, structures of uncertain function containing over 20 identified proteins (15) , and we confirmed this by costaining with BLM, RECQL, or Topo IIIα antiserum together with PML antiserum (Fig. 2, D and E) ⇓ . Despite their localization at NBs, we do not observe coprecipitation of BLM or RECQL with the PML protein that is in the cell extracts (data not shown), arguing that the protein associations that we do observe are specific and not simply artifacts of nuclear proximity. Topo IIIα was examined in fibroblast lines from two different BS individuals and was found diffusely throughout the nucleus and no longer colocalized with PML, indicating that BLM is required for proper localization of Topo IIIα to NBs (Fig. 2F ⇓ and data not shown).
Localization of Topo IIIα, BLM, RECQL, and PML in cultured primary diploid fibroblasts. Data are from WI38 cells; similar results were obtained for MRC5 fibroblasts. A, colocalization of BLM (red) and Topo IIIα (green), using rabbit anti-BLM and chicken anti-Topo IIIα antibody 1002. Filled arrowheads, staining in PML NBs (see below); open arrowhead, nucleolar staining. B, colocalization of RECQL (red) and Topo IIIα (green), using rabbit anti-RECQL serum. C, partial colocalization of BLM (green) and RECQL (red). BLM was detected using chicken anti-BLM. D, BLM and RECQL partially colocalize with PML antigen. BLM or RECQL (red) were detected with their respective rabbit antisera, and PML (green) was detected using mouse monoclonal antibody PG-M3. E and F, staining of Topo IIIα (red; rabbit antibody 1151) and PML (green; PG-M3) in WT WI-38 fibroblasts (E) and in AG2940 fibroblasts from a BS individual (F).
Several observations argue that BLM and Topo IIIα may play roles in meiosis. In yeast, top3 and sgs1 mutants each display sporulation defects that are suppressed by mutations that abolish meiotic recombination, indicating that Top3p and Sgs1p are likely important for the resolution of meiotic recombination intermediates (16 , 17) . In mice, BLM and Topo IIIα mRNAs are highly expressed in testis (18) , and the BLM protein has been shown recently to localize at synaptonemal complexes (19) . Finally, mates with BS are infertile due to failed spermatogenesis (1) . We examined the possibility that BLM and Topo IIIα might be present together on meiotic chromosomes. Meiotic spreads of human testes were immunostained using chicken anti-Topo IIIα antiserum and a polyclonal rabbit antiserum against SCP3 (9) . This protein marks the axial elements that form between sister chromatids prior to synapsis of homologous chromosomes and later marks the lateral elements of the synaptonemal complex after synapsis. In meiotic spreads in the leptonemal stage of prophase I, prior to the onset of synapsis, Topo IIIα is distributed throughout the nucleus, with no specific staining observed on the axial elements (data not shown). However, during early zygonema, as the autosomes initiate pairing, strong focal Topo IIIα staining is visible on the axial elements (Fig. 3, A and B) ⇓ . This axial Topo IIIα staining continues to be visible throughout zygonema as pairing proceeds (Fig. 3B) ⇓ . The foci are particularly numerous and intense on chromosomes that appear to be in the process of pairing and are found on both asynapsed axial and synapsed lateral elements. Later, in pachynema, a time at which the X and Y chromosomes partially synapse, Topo IIIα staining was found diffusely throughout the nucleus with no axial staining on the X and Y chromosomes (data not shown). BLM has been reported recently to localize to discrete foci along meiotic chromosomes in the mouse (19) , and we asked whether BLM and Topo IIIα colocalize on human meiotic chromosomes as they do in somatic cells. Meiotic spreads were costained with affinity-purified chicken anti-Topo IIIα and rabbit anti-BLM antisera, and substantial coincidence of the BLM and Topo IIIα was observed (Fig. 3C) ⇓ . Meiotic spreads were also stained for an unrelated recombination protein, Rad51, which localizes at meiotic chromosome foci (20) ; no overlap was observed between Rad51 and either BLM or Topo IIIα staining, reinforcing the specificity of the BLM-Topo IIIα colocalization (data not shown). In some cases, meiotic spreads were stained for BLM and Topo IIIα and then stained for axial elements using rabbit anti-SCP3 (Fig. 3D) ⇓ . Colocalization of BLM and Topo IIIα was observed on lateral and some axial elements. Similar results were obtained using the rabbit anti-Topo IIIα antiserum and a chicken antiserum directed against the COOH terminus of the BLM protein (data not shown). Focal RECQL staining was not observed on meiotic spreads (data not shown).
Partial colocalization of Topo IIIα and BLM on meiotic spreads of human testes. A, in early zygonema, Topo IIIα protein is found on the axial or lateral elements, prior to and after synapsis. Topo IIIα (green) was detected using chicken anti-Topo IIIα antibodies (1002), and synaptonemal complexes (red) were detected using rabbit anti-SCP3. B, Topo IIIα continues to be associated with the lateral elements as zygonema proceeds and is clearly concentrated at sites of active synapsis. C and D, partial colocalization of BLM and Topo IIIα during meiosis. Rabbit anti-BLM (red) was used to detect BLM protein.
Discussion
We have shown that Topo IIIα associates with the BLM and RECQL DNA helicases, and that one site of association of these proteins is the PML NB. The subcellular distributions of the proteins are not fully coincident by immunolocalization, indicating that they may have functions apart from one another as well as together. Partial overlap between BLM and Topo IIIα is also observed in meiotic spreads of human testes, suggesting cooperation between these proteins in the germ line as well as in somatic cells. This is the first demonstration of an interacting partner for the BLM protein.
The E. coli proteins RecQ and Topo III together can catalyze the passage of double-stranded DNA, most likely through two sequential single-strand transfers (5) . Our finding that BLM associates with Topo IIIα suggests that these proteins cooperate in a similar fashion in humans. In yeast, Top3p and Sgs1p both function to antagonize recombination (4 , 17) , and BS cells also show elevated recombination (21) . BLM and Topo IIIα thus likely influence the frequency or fidelity of recombination events. One possibility is that they perform this function by disrupting nascent joint molecules; BLM helicase activity would disrupt base pairing between the invading strand and target, whereas Topo IIIα would disentangle the joint molecule by allowing the invading strand to pass through the target. Such a role for a topoisomerase in recombination was proposed by Wang et al. (22) and is mechanistically similar to the catenation reaction catalyzed by E. coli RecQ and Topo III in vitro (5) . Alternatively, the proteins might promote or reverse the process of branch migration of Holliday junctions, either directly or by removing DNA structures that impede branch migration. For example, the BLM helicase could catalyze helix unwinding, and Topo IIIα could relieve negative supercoils which, if the DNA duplexes were not free to rotate, would accumulate behind the migrating Holliday junction. Failure to resolve recombination intermediates could account for the high frequency of chromosomal nondisjunction observed in sgs1 mutants cells and the high incidence of chromosomal breaks and translocations in BS cells. The partial colocalization of BLM and Topo IIIα on meiotic chromosomes of human testes also supports a role for these proteins in meiotic recombination. Our results are consistent with the previous localization of BLM to meiotic chromosomes and with the proposed role for yeast Top3p and Sgs1p in resolving meiotic recombination intermediates (16 , 19) . Failure to resolve recombinational intermediates could account for the meiotic defects in sgs1 and top3 mutants and the hypogonadism of BS patients.
BLM and Topo IIIα colocalize in PML NBs, although association might also occur at other sites not revealed by immunofluorescence. Very recently, others have shown independently that BLM localizes to NBs (23 , 24) . Remarkably, immortalized human cells that lack telomerase and are believed to maintain their telomeres via recombinational mechanisms show telomeric localization of PML bodies containing the recombination proteins Rad51 and Rad52 (25) . Taken together with our findings, these observations implicate NBs in recombination. Loss of Topo IIIα from NBs in BS cells suggests that the failure to localize Topo IIIα to NBs might contribute to the genome instability in BS. Thus, in addition to hypothesized roles in transcription, replication, and apoptosis (15) , NBs might function in recombination.
In summary, here we have shown that BLM associates with the Topo IIIα protein in human cells. Additional studies are necessary to elucidate the precise roles that this complex may play in recombination and in the suppression of cancer.
Note Added in Proof
Similar independent observations of an association between BLM and Topo IIIα are in press: Wu, L., Davies, S. L., North, P. S., Goulaouic, H., Riow, J-F., Turley, H., Gatter, K. C., and Hickson, I. D. The Bloom’s Syndrome gene product interacts with topoisomerase III. J. Biol. Chem. in press, 2000.
Acknowledgments
We thank J. German for BS cells, J. Parvin for advice on extract preparation, D. Stumpo and P. Blackshear for the anti-RECQL antiserum, C. Heyting for the anti-SCP3 antibody, J. Wang for the Topo IIIα cDNA, and the members of the Guarente and Jaenisch laboratories for discussions.
Footnotes
-
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.
-
↵1 This work was supported by National Institute on Aging Grants R01 AG11119 (to L. G.) and K08 AG00775 and a Howard Hughes Medical Institute postdoctoral fellowship (to F. B. J.). D. B. L. is supported by a Medical Scientist Training Program Grant to Harvard Medical School.
-
↵2 These authors contributed equally to this work.
-
↵3 To whom requests for reprints should be addressed, at Department of Biology, 68-280, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139. Phone: (617) 253-6965; Fax: (617) 253-8699; E-mail: leng{at}mit.edu
-
↵4 The abbreviations used are: BS, Bloom syndrome; WS, Werner syndrome; WT, wild type; Topo, topoisomerase; PML, promyelocytic leukemia protein; NB, nuclear body; IP, immunoprecipitation.
- Received November 22, 1999.
- Accepted January 19, 2000.
- ©2000 American Association for Cancer Research.
References
Volume 60, Issue 5
