The prostate-specific tumor suppressor homeodomain protein NKX3.1 is inactivated by a variety of mechanisms in the earliest phases of prostate carcinogenesis and in premalignant regions of the prostate gland. The mechanisms by which NKX3.1 exercises tumor suppression have not been well elucidated. Here, we show that NKX3.1 affects DNA damage response and cell survival after DNA damage. NKX3.1 expression in PC-3 prostate cancer cells enhances colony formation after DNA damage but has minimal effect on apoptosis. NKX3.1 also diminishes and regulates total cellular accumulation of γH2AX. Endogenous NKX3.1 in LNCaP cells localizes to sites of DNA damage where it affects the recruitment of phosphorylated ATM and the phosphorylation of H2AX. Knockdown of NKX3.1 in LNCaP cells attenuates the acute responses of both ATM and H2AX phosphorylation to DNA damage and their subnuclear localization to DNA damage sites. NKX3.1 expression enhances activation of ATM as assayed by autophosphorylation at serine 1981 and activation of ATR as assayed by phosphorylation of CHK1. An inherited mutation of NKX3.1 that predisposes to early prostate cancer and attenuates in vitro DNA binding was devoid of the ability to activate ATM and to colocalize with γH2AX at foci of DNA damage. These data show a novel mechanism by which a homeoprotein can affect DNA damage repair and act as a tumor suppressor. Cancer Res; 70(8); 3089–97. ©2010 AACR.
NKX3.1 is a haploinsufficient prostate cancer suppressor protein that has reduced expression in the majority of human primary prostate cancers (1). NKX3.1 downregulation also predisposes to the formation of invasive prostate cancer and prostate carcinogenesis, because decreased expression is seen in premalignant lesions and prostatic intraepithelial neoplasia (1, 2). In mouse models, loss of Nkx3.1 causes prostatic dysplasia and intraepithelial neoplasia but does not by itself cause invasive cancer. During prostate cancer progression, there is progressive loss of NKX3.1 expression, suggesting that there is an ongoing selective pressure against expression of this suppressor protein. Indeed, in up to 80% of metastatic prostate cancer foci, there is complete loss of NKX3.1 expression (3). NKX3.1 downregulation not only results from genetic loss, DNA methylation, or both, but also NKX3.1 turnover is accelerated by ubiquitination and proteasomal degradation triggered by cellular exposure to inflammatory cytokines, such as interleukin-1β and tumor necrosis factor-α (1, 4). In this way, regions of inflammatory atrophy may cause both downregulation of NKX3.1 and oxidative damage, which accompanies inflammation (2).
In trying to understand the mechanism by which NKX3.1 loss predisposes to prostate carcinogenesis, it was observed that NKX3.1 positively affects the expression of antioxidant enzymes and downregulates the expression of prooxidants, thereby acting to protect the cell and the genome from oxidative stress and DNA damage (5). NKX3.1 may also play a broader role in the cellular response to DNA damage. When we reported that NKX3.1 activated topoisomerase I, we observed that both proteins in concert changed their nuclear location in response to DNA-damaging agents. We have since pursued this observation by asking whether NKX3.1 affected cell survival after DNA damage and whether NKX3.1 was directly involved with the DNA damage response. In this report, we show that NKX3.1 expression has a protective effect against DNA-damaging agents. Moreover, NKX3.1 expression affects the activation of ATM and ATR, two members of the phosphoinositide 3-kinase (PI3K) family that activate the cell's response to DNA damage by initiating a signal transduction cascade.
Materials and Methods
Cell culture, transfection, and vectors
LNCaP, 293T, and PC-3 cells were cultured in IMEM (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen). Cells were transfected with either Lipofectamine 2000 (Invitrogen) or Fugene HD (Roche) according to manufacturer's instructions. Luciferase siRNA control and shNKX3.1 lentivirus vectors, which were custom-designed by Open Biosystems, were stably transfected with Lipofectamine 2000 into LNCaP cell to generate LNCaP(siLuc), LNCaP(si471), and LNCaP(si3098) cell lines. The stable cell lines were selected with 5 μg/mL puromycin and maintained in media with 2 μg/mL puromycin.
Colony formation assay
For colony formation assay, PC-3 cells were transiently transfected with either empty pcDNA3 or pcDNA3-NKX3.1 and the cotransfected selection marker green fluorescent protein (GFP) with Lipofectamine 2000. Cells were sorted by fluorescence-activated cell sorting (FACS) based on expression of GFP, placed into six-well plates at 2,000 cells per well, and incubated in IMEM supplemented with 10% FBS for 24 h. Cells were subjected to treatment with UV, γ-irradiation, or mitomycin C for 1 h. Cells were incubated for 7 to 10 d to score for colony formation, fixed with 4% formaldehyde, and stained with crystal violet. All experiments were repeated. Apoptosis was assayed in PC-3(pcDNA3.1) and PC-3(NKX3.1) cells (6) after exposure to 10 mJ/cm2 UV. Cells were fixed at indicated times by 4% formaldehyde, and apoptosis was detected by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay (7).
Cell extracts were prepared by lysing cells with radioimmunoprecipitation assay buffer containing protease inhibitors (Complete mini, Roche) for detecting phosphorylation of proteins with phosphatase inhibitors (PhosStop, Roche). Proteins were resolved by SDS-PAGE and transferred to a nitrocellulose membrane. The primary antibodies used were polyclonal NKX3.1 (3), ATR (ABR), monoclonal anti-γH2AX (Upstate), anti-pATM-S1981 (Rockland), anti-pChk1-S345 (Cell Signaling), ATM (GeneTex), Chk1 (Santa Cruz), anti-β-actin (Sigma). The corresponding horseradish peroxidase–labeled secondary antibody was detected with SuperSignal West Pico Chemiluminescent substrate (Thermo Scientific) or SuperSignal West Dura Extended Duration Substrate (Thermo Scientific).
Cells were fixed with 4% formaldehyde (Polysciences) or Streck Tissue Fixative in PBS for 15 min and permeabilized with either 0.5% Triton X-100 in PBS for 15 min at room temperature or ice-cold methanol for 10 min at −20°C. Cells were blocked with 3% bovine serum albumin (BSA) in PBS and incubated with primary antibody in 1% BSA in PBS followed by incubation with FITC-conjugated (Vector) or Alexafluor secondary antibody (Invitrogen). Cells were visualized with a Zeiss LSM 510 META scanning confocal microscope. Antibodies for immunofluorescence are polyclonal antibody against NKX3.1 (3), anti-ATR (C-19; Santa Cruz), monoclonal γH2AX (Upstate), anti-RAD51 (GeneTex), anti-poly(ADP-ribose) polymerase 1 (Biomol), anti-ATM (GeneTex), anti-pATM-S1981 (Rockland), and anti-53BP1 (Cell signaling).
Cells were placed on polylysine-coated (Sigma) 35-mm glass bottomed microwell dishes (MatTek). Hoechst dye #33258 (Sigma) was added to the medium to a final concentration of 10 μg/mL. After 30 min of incubation at 37°C, the medium was aspirated from the cells, which were then washed and covered with phenol red–free IMEM supplemented with 10% FBS. A PALM MicroBeam IV Laser Capture Microdissection System (Zeiss) combined with laser scanning confocal microscopy was used for generation of local double-strand breaks by laser microirradiation and image acquisition. The energy output of the microdissection laser was 36%. After microirradiation, the recruitment of DNA repair proteins at defined DSB regions was identified by immunofluorescence as described above.
We first determined whether NKX3.1 affected cell survival after DNA damage by assaying colony formation after cells were exposed to DNA-damaging agents. PC-3 prostate cancer cells were transiently transfected with either an empty expression vector or a plasmid expressing NKX3.1. Cells were cotransfected with a GFP expression vector used as a selection marker. After transfection, cells were subjected to DNA damage by exposure to UV, γ-irradiation, or mitomycin C. Cells were then sorted by FACS to select the transfected cells, and transfected and untransfected cells were then plated for colony formation. Transfection with NKX3.1 altered survival curves after DNA damage, as shown by increased colony formation after exposure to UV, γ-irradiation, and mitomycin C (Fig. 1A). The Western blot in Fig. 1A shows the expression of NKX3.1 in the transfected cells. The differences in colony formation were not due to an effect of NKX3.1 on apoptosis, as NKX3.1 expression did not affect induction of apoptosis in PC-3 cells (Fig. 1B).
One of the earliest indications of cellular response to DNA damage and an indicator of the degree of DNA damage, in particular double-stranded DNA breaks, is the phosphorylation of histone 2AX to form γ-histone 2AX (γH2AX; ref. 8). We examined the presence of γH2AX in PC-3 cells and PC-3(NKX3.1) cells during 6 hours after exposure to UV or mitomycin C. Expression of NKX3.1 attenuated the appearance of γH2AX over a 6-hour time span after exposure either to UV irradiation or mitomycin C (Fig. 2A). This result was consistent with the interpretation that more DNA damage accumulated in cells over 6 hours in the absence of NKX3.1 than in its presence. In a second experiment LNCaP cells that express endogenous NKX3.1 were subjected to NKX3.1 knockdown and in each of two independently derived LNCaP cell lines, LNCaP(si471) and LNCaP(si3098). Knockdown of NKX3.1 expression is shown in Fig. 2B (bottom). Knockdown of NKX3.1 resulted in increased accumulation of γH2AX over 6 hours (Fig. 2B). Thus, NKX3.1 expression correlated with decreased overall accumulation of γH2AX.
The DNA damage response and phosphorylation of γH2AX by ATM and ATR occurs within minutes of DNA strand breakage. To examine the effect of NKX3.1 on the shorter-term response of cells to DNA damage, LNCaP cells and NKX3.1 knockdown cells were examined at 10 and 30min after exposure of BrdUrd-treated cells to UV irradiation. BrdUrd sensitizes DNA to UV and predisposes to UV-induced double-stranded DNA breakage. LNCaP(siLuc) cells showed expression of NKX3.1 by immunohistochemical staining, which was substantially decreased in LNCaP(si471) cells (Fig. 3A, compare panels b). Within 10 minutes after exposure to UV LNCaP(siLuc), cells showed formation of γH2AX foci as shown at both lower power and higher power (Fig. 3A, compare d and g for both derivative cells). Figure 3A (g–i) shows in a higher power view of a single nucleus that staining of γH2AX and NKX3.1 aggregated in granular regions. The insets for LNCaP(siLuc) show magnified views of the regions marked with the arrow heads (Fig. 3A, left, g–l). In contrast, there was minimal formation of γH2AX foci in LNCaP(si471) cells at 10 minutes after UV exposure (Fig. 3A, right, d and g). γH2AX focus formation in LNCaP(si471) cells was seen to have increased at 30 minutes after UV exposure (Fig. 3A, right, compare j and g). This was in contrast to γH2AX focus formation in LNCaP(siLuc) cells that had plateaued by 30 minutes (Fig. 3A, 30-minute panels j–l and B, graph on the left). Expression of NKX3.1 in LNCaP(siLuc) cells was also associated with a more rapid response to DNA damage as measured by the fraction of γH2AX-positive cells (Fig. 3B, right, graph). In contrast, the fraction of γH2AX-positive LNCaP(si471) cells increased more slowly and was still rising at 30 minutes after exposure to UV irradiation (Fig. 3B, right, graph).
H2AX is a target for phosphorylation by the PI3K family members ATM and ATR that are activated in response to a variety of stimuli, including DNA breaks (9, 10). Thus, the effect of NKX3.1 on the phosphorylation of H2AX suggested that NKX3.1 affected the activity of ATM, ATR, or both. The effect of NKX3.1 on the formation of ATM and ATR nuclear complexes after DNA damage was studied in LNCaP cells expressing either siLuc or si471. Ten minutes after exposure to mitomycin C, ATM, ATR, and γH2AX nuclear foci were seen in the presence of NKX3.1 as shown by confocal immunofluorescent microscopy of the single nucleus shown in Fig. 4A. As in Fig. 3A, effective knockdown of NKX3.1 expression was seen in the LNCaP(si471) cells (Fig. 4A, b, e, and h). Each panel in Fig. 4A shows a single nucleus in which NKX3.1, γH2AX, ATM, and ATR aggregate after the cells were exposed to irradiation. At higher magnification (figure insets), colocalization of NKX3.1 with γH2AX, ATM, and ATR is seen in nuclear granules marked with arrows in the lower power images. However, in the presence of NKX3.1 knockdown, very few γH2AX, ATM foci, or ATR foci were seen at the 10-minute time point (Fig. 4A, right, a–i). Quantitation of the nuclear focus formation in cells at 10 minutes is seen in Fig. 4B. NKX3.1 knockdown resulted in a decreased number of detectable foci per nucleus (Fig. 4B, left) and a decrease in the fraction of cells with visible signals (Fig. 4B, right).
Recognition of and response to DNA damage occurs within seconds to a few minutes of DNA damage. To determine whether NKX3.1 affected the immediate response to DNA damage, we used laser microirradiation of LNCaP(si471) cells transfected with different MYC-tagged NKX3.1 constructs. We expressed MYC-tagged NKX3.1 and two constructs with inactivating mutations in the homeodomain. One mutant expression construct, MYC-tagged NKX3.1(T164A), coded a protein found in a family with hereditary early incidence of prostate cancer (11). The missense mutation alters the α-helical cap amino acid of the third helix of the homeodomain and diminishes DNA binding to 5% of the wild-type protein. A second NKX3.1 construct with a missense mutation in the homeodomain (N174Q) was also studied. Asparagine 51 of the Drosophila NK-2 homeodomain is juxtaposed to the major groove of DNA when the protein binds to its cognate hexanucleotide (12). A missense mutation changing asparagine 174 at the 51st amino acid of the homeodomain to glutamine (N174Q) reduces DNA binding by a factor of >10−3.1
MYC-tagged NKX3.1 was expressed in LNCaP(si471) cells by transient transfection, and cells, pretreated with Hoechst 33258 dye, were subject to laser microirradiation to induce double-stranded DNA breaks in response to UV. Expression of MYC-NKX3.1 was seen with MYC antibody, and MYC-NKX3.1 localized to the site of UV irradiation. This was not seen in untransfected cells (Fig. 5A, e and f). MYC-NKX3.1 enhanced the formation of γH2AX at the site of UV within 2 to 3 minutes of double-stranded DNA damage (Fig. 5A, compare a and b). Moreover, MYC-NKX3.1 and γH2AX were seen to colocalize at the site of UV (Fig. 5A, compare i and j). Both MYC-NKX3.1(T164A) and MYC-NKX3.1(N174Q) were expressed in LNCaP(si471) cells but did not accumulate the sites of laser microirradiation (Fig. 5A, g and h, respectively). γH2AX was seen to accumulate in the laser microirradiation sites of both transfected and untransfected cells, but there was no difference in intensity of γH2AX staining between cells expressing the NKX3.1 mutant constructs and those that did not (Fig. 5A, c and d). Also, the mutant NKX3.1 proteins did not localize to the regions of DNA damage (Fig. 5A, g and h) and did not colocalize with γH2AX (Fig. 5A, k and l).
Before phosphorylation of γH2AX, the PI3K-like protein ATM is rapidly phosphorylated on serine 1981 in response to DNA damage (13). To determine whether NKX3.1 affected phosphorylation of ATM, we examined the appearance of pATM-S1981 in the presence of different NKX3.1 constructs within 1 to 3 minutes of DNA damage induced by laser microirradiation. In the top row of images in Fig. 5B, pATM-S1981 is seen to accumulate at the sites of damage in all three cell lines. The fluorescent signal was strongest in cells expressing wild-type MYC-NKX3.1 (Fig. 5B, a, b, and c). As seen in Fig. 5A, wild-type MYC-NKX3.1, but not mutant protein, localized to the site of UV irradiation (Fig. 5B, d, e, and f). In the merged images at the bottom, colocalization of MYC-NKX3.1 and pATM-S1981 is seen, but not the mutant proteins (Fig. 5B, g, h, and i). This result is consistent with the notion that NKX3.1 expression facilitates activation of ATM and downstream phosphorylation of H2AX. Thus NKX3.1 affects the earliest stages of the cellular DNA damage response.
To show further the effect of NKX3.1 expression on DNA damage response by assaying cellular levels of activated ATM and ATR, we did Western blotting of cell extracts after exposure to mitomycin C (Fig. 6A). NKX3.1 knockdown is seen in the Western blot with NKX3.1 antibody. Knockdown of NKX3.1 markedly attenuated phosphorylation of ATM. In addition, LNCaP(si471) cells had decreased activation of ATR as shown by diminished phosphorylation of the ATR target CHK1 at serine 345 (14). Importantly, NKX3.1 did not affect protein levels of the DNA response kinases but rather seemed to affect kinase activation and downstream signaling. Further demonstration that the effect of NKX3.1 on ATM and ATR activation was not due to off-target effects of the siRNAs was seen in PC-3 cells transfected with NKX3.1 and subjected to UV irradiation after treatment with Hoechst dye. NKX3.1 expression accelerated activation of ATM in PC-3 cells (Fig. 6B).
A substantial weight of evidence now implicates NKX3.1 asa key tumor suppressor and the lead candidate for the gatekeeper of prostate cancer (15). This haploinsufficient protein has been shown in mouse models to be a critical determinant of prostate epithelial differentiation and growth control (16–19). In premalignant lesions NKX3.1 protein levels are subjected to reductions due to the actions of inflammatory cytokines (2, 4). In human prostate cancer, NKX3.1 is subject to reductions in levels of cellular protein by gene methylation and loss of heterozygosity (1). Numerous studies have shown that loss of the 8p21 locus, including NKX3.1 is a very common event in human prostate cancer accounting for NKX3.1 loss in a majority of cases (20, 21). In human cancer specimens, levels of NKX3.1 protein vary widely around a median level of 0.67 of expression in the adjacent normal prostate epithelium. Interestingly, Nkx3.1+/− mice have a level of Nkx3.1 protein in prostate epithelium that is also 0.67 of wild-type levels, suggesting that the residual contralateral allele compensates somewhat for loss of one gene copy (1). Nevertheless, it is apparent that a variety of factors influence NKX3.1 protein levels in human prostate epithelium.
Our data show that expression of NKX3.1 affects early events in the response to DNA damage and results in increased cell survival and clonogenicity. Conversely, decreased levels of NKX3.1 attenuate the acute response to DNA damage and extend the time during which the cell activates phosphorylation of H2AX, presumably due to an extended time for recognition and remediation of DNA strand breaks. This is an important example of a homeodomain protein directly influencing detection of DNA damage and the DNA repair process. The effect of NKX3.1 on DNA repair may be by direct interaction with DNA repair proteins or by an indirect effect. Experiments to address this point are well under way in our laboratory. Knockdown of NKX3.1 affected the DNA damage response. This effect was not due to off-target effects of siRNA for at least two reasons. In Fig. 5, MYC-NKX3.1 was expressed in LNCaP(si471) knockdown cells and reversed the effect of NKX3.1 knockdown on the DNA damage response. We also studied ATM activation in derivative PC-3 cells to show that NKX3.1 expression affected the timing of ATM phosphorylation after DNA damage.
NKX3.1 has a manifold role in both protecting and repairing DNA, suggesting that DNA protection and integrity may be essential elements of its role in the cell. We previously showed that NKX3.1 binds to topoisomerase I and enhances its DNA-binding and supercoil resolving activity in a stoichiometric fashion (22). By interacting with topoisomerase I, NKX3.1 may also affect DNA integrity because topoisomerase I has been implicated in DNA repair. Topoisomerase I colocalizes with P53 at sites of DNA damage in response to UV irradiation (23). Moreover, knockdown of topoisomerase I predisposes cells to chromosomal instability, suggesting that topoisomerase I has an ongoing function in maintenance of DNA integrity (24). Our findings represent the second example of a homeodomain protein interacting with DNA repair enzymes. The HOXB7 homeodomain protein that functions in body axis patterning is able, when overexpressed, to transform mammary epithelial cells and enhance nonhomologous end-joining after DNA damage. HOXB7 was shown to interact with components of the DNA-dependent protein kinase holoenzyme via helix 3 of its homeodomain (25).
The role of NKX3.1 in maintaining DNA stability is further suggested by studies in gene-targeted mice, wherein Nkx3.1 expression affects the transcriptional profile of a range of prooxidant and antioxidant enzymes and dysplastic prostate epithelium of Nkx3.1 deletion mice has increased levels of 8-oxoguanine (5). The effect of Nkx3.1 on transcription of genes that affect oxidative stress is likely to occur via indirect mechanisms. Although, as is expected of a homeodomain-containing protein, NKX3.1 binds to DNA (26). However, NKX3.1 cannot initiate formation of a transcription complex and by itself does not activate transcription but rather suppresses transcription when targeted to a promoter region by placement of cognate DNA-binding sequences (26). Transcriptional activation by NKX3.1 is seen when NKX3.1 serves as a coactivator for transcription factors like serum response factor. NKX3.1 binds both to DNA sequences adjacent to the serum response element (27) and to SRF directly (28). Protein-protein interaction between NKX3.1 and SRF is mediated by the homeodomain and influenced by NH2 and COOH terminal flanking regions of the protein (28).
Many activities of NKX3.1 are mediated by interactions of other proteins with the homeodomain. The homeodomain is the locus for interaction with the major groove of DNA and responsible for DNA binding (11, 12, 28). However, the NKX3.1 homeodomain is also the site of interaction with proteins, such as topoisomerase I. In fact, topoisomerase I was able to titrate NKX3.1 off its cognate DNA-binding sequence (22). Our results suggest that the homeodomain is also important for the effect of NKX3.1 on early steps in DNA break recognition. A construct that had lost 95% or more of wild-type DNA-binding affinity had essentially no detectable effect on DNA break recognition.
Our finding that NKX3.1 plays a role in maintaining DNA integrity of prostate epithelial cells is consistent with the notion that decreased levels of NKX3.1 in premalignant prostate tissue may predispose to accumulation of mutations and transformation to prostate cancer. If regions of inflammatory atrophy are present throughout the prostate gland then it follows there may be multiple independent oncogenic events leading to independent foci of transformed epithelial cells. This prediction is supported by analyses of prostatectomy specimens that showed multiple foci of cancer with distinct genetic signatures in a single gland (29). Thus, we envision that stochastic inflammatory foci in the prostate result in reductions of NKX3.1 half-life in adjacent prostate epithelial cells, diminishing the response to DNA damage. Inflammation also increases the local production of reactive oxygen species, thus leading to the formation of 8-oxoguanine and subsequent mutations. Decreased levels of NKX3.1 also decrease the production of antioxidant enzymes and increase the synthesis of prooxidant enzymes, further exacerbating the chance of mutagenesis and facilitating steps to carcinogenesis. Our report that NKX3.1 affects DNA integrity in prostate epithelial cells thus adds further evidence to the critical role this protein plays in the pathogenesis of prostate cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
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.
↵1Our unpublished observation.
- Received August 24, 2009.
- Revision received December 22, 2009.
- Accepted January 21, 2010.
- ©2010 American Association for Cancer Research.