Hematopoietic Cells from Gadd45a- and Gadd45b-Deficient Mice Are

Hematopoietic cells from Gadd45a- and Gadd45b-deﬁcient mice are sensitized to genotoxic-stress-induced apoptosis

Mamta Gupta1, Shiv K Gupta1, Arthur G Balliet1, Mary Christine Hollander2, Albert J Fornace Jr2, Barbara Hoffman1 and Dan A Liebermann*,1

1Fels Institute of Cancer Research and Molecular Biology, Temple University School of Medicine, Room 331, 3307 North Broad Street, Philadelphia, PA 19140, USA; 2Center for Cancer Research, National Cancer Institute, NIH, Bethesda, MD 20892, USA

Gadd45a, gadd45b and gadd45g (Gadd45/MyD118/CR6) encoded by gadd45a and gadd45b are remarkably are genes that are rapidly induced by genotoxic stress. similar, sharing 55–57% overall identity at the amino- However, the exact function of Gadd45 proteins in the acid level (Zhan et al., 1994). Data accumulated so far response of mammalian cells to genotoxic stress is suggest that gadd45a/gadd45b/gadd45g serve similar, unclear. Here, advantage was taken of gadd45a- and but not identical, functions along different apoptotic gadd45b-deficient mice to determine the role gadd45a and and growth-inhibitory pathways. For example, gadd45b, gadd45b play in the response of bone marrow (BM) cells but not gadd45a, is activated upon TGFb-induced to genotoxic stress. BM cells from gadd45a- and gadd45b- apoptosis (Selvakumaran et al., 1994a, b). On the other deficient mice were observed to be more sensitive to hand, gadd45a, but not gadd45b and gadd45g, was ultraviolet radiation chemotherapy (UVC), VP-16 and identified as a target for p53 function (Kastan et al., daunorubicin (DNR)-induced apoptosis compared to wild- 1992; Selvakumaran et al., 1994a, b; Guillouf et al., type (wt) cells. The increased apoptosis in gadd45a- and 1995). gadd45b-deficient cells was evident also by enhanced In recent years, several lines ofresearch have activation of caspase-3 and poly-ADP-ribose polymerase implicated gadd45 genes in cell cycle arrest (Beadling cleavage and decreased expression of c-inhibitor of et al., 1993; Zhang et al., 1999), DNA repair (Smith apoptotic protein-1, Bcl-2, Bcl-xL compared to wt cells. et al., 1994; Vairapandi et al., 1996), innate immunity Reintroduction of gadd45 into gadd45-deficient BM cells (Yang et al., 2001), maintenance ofgenomic stability restored the wt apoptotic phenotype. Both gadd45a- and (Hollander et al., 1999) and apoptosis (Takekawa and gadd45b-deficient BM cells also displayed defective G2/M Saito, 1998). It is believed that the functions of Gadd45 arrest following exposure to UVC and VP-16, but not to proteins are mediated via interactions with other cellular DNR, indicating the existence of different G2/M proteins implicated in cell cycle regulation and the checkpoints that are either dependent or independent of response ofcells to extrinsic stress. For example, gadd45. Taken together, these findings identify gadd45a Gadd45 proteins have been shown to interact with and gadd45b as antiapoptotic genes that increase the cdc2 (cdk1), thereby disrupting interaction between survival of hematopoietic cells following exposure to UV Cdc2 and cyclin B1 to induce G2/M cell cycle arrest radiation and certain anticancer drugs. (Vairapandi et al., 2000; Azam et al., 2001). All three Oncogene (2005) 24, 7170–7179. doi:10.1038/sj.onc.1208847; Gadd45 proteins were found also to interact with PCNA published online 19 September 2005 (Smith et al., 1994; Vairapandi et al., 1996; Azam et al., 2001), a nuclear protein that plays a central role in DNA Keywords: gadd45-deficient mice; apoptosis; hemato- repair. poietic cells; genotoxic stress; DNA damage The exact function of Gadd45 proteins in apoptosis remains unclear. Several lines ofevidence suggest that Gadd45 proteins can play a proapoptotic role. For example, studies performed in this laboratory have Introduction shown that blocking gadd45b (MyD118) expression via antisense transgenes in M1 myeloblastic leukemia cells The Gadd45/MyD118/CR6 (also termed as gadd45a/ impaired TGFb-induced cell death, thereby implicating gadd45b/gadd45g) family of genes has been shown to be Gadd45b as a positive modulator ofTGF b-induced rapidly induced by a variety ofgenotoxic stresses as well apoptosis (Selvakumaran et al., 1994a, b). Moreover, as by terminal differentiation and apoptotic cytokines in IPTG-inducible expression of gadd45b, gadd45a and almost all mammalian cells (Fornace et al., 1989; gadd45g was observed to enhance apoptosis induced by Papathanasiou et al., 1991; Zhan et al., 1994). Proteins genotoxic stress in both M1 and H1299 cells (Zhang et al., 2001). Also, Takekawa and Saito (1998) have *Correspondence: DA Liebermann; E-mail: [email protected] reported that gadd45a, gadd45b and gadd45g each Received 7 March 2005; revised 12 May 2005; accepted 12 May 2005; interact with and activate MTK1/MEKK4, where published online 19 September 2005 activation ofMTK1 is thought to activate its gadd45a and gadd45b in hematopoietic stress M Gupta et al 7171 downstream targets P38 and JNK and induce apoptosis mice has been detailed in Materials and methods. As (Takekawa and Saito, 1998). shown in Figure 1a, the entire coding region ofthe Generation of gadd45a and gadd45b null mice has gadd45b gene was replaced with the neo cassette. allowed us to address the role gadd45a and gadd45b Southern blot analysis of BamHI digested genomic play in the response ofhematopoietic cells to genotoxic DNA from transfected embryonic stem (ES) cell clones, stresses, such as ultraviolet radiation chemotherapy probed with a BglII–BamHI flanking probe (Figure 1a), (UVC) and two well-known anticancer drugs, namely has shown that the BamHI site introduced with the neo VP-16 and daunorubicin (DNR). Interestingly, in this cassette, converted the 5.8 kb wt BamH1 band ofthe study, we show that gadd45a and gadd45b function to gadd45b gene into a 3.3 kb band for the mutated allelle increase survival ofhematopoietic cells following (Figure 1b). To facilitate genotyping, polymerase chain exposure to the indicated genotoxic stress agents. Data reaction (PCR) was used to distinguish mutant and wt obtained also provide evidence for the existence of gadd45b alleles (Figure 1c) in littermates used in different G2/M checkpoints that are either dependent or experiments (Materials and methods). independent ofgadd45a and gadd45b function. Genotoxic stress induces gadd45a and gadd45b expression in wt bone marrow (BM) cells Results To determine how gadd45a or gadd45b deficiency may Generation of gadd45b null mice influence the response ofhematopoietic cells to genotoxic stress agents, first it was in our interest to Gadd45aÀ/À mice were generated as indicated in determine the expression profile of gadd45a and Hollander et al. (1999). Generation ofgadd45b À/À gadd45b genes in BM-derived hematopoietic cells

Figure 1 Targeted disruption ofthe Gadd45b gene. ( a) The murine Gadd45b gene is shown with the four exons represented by numbered black bars, and the BglII–BamHI fragment used as a flanking probe to confirm that homologous recombination is labeled at the three prime end ofthe gene. The gene targeting vector was pG118KO constructed by replacing a 2290 bp region ofthe gadd45b gene sequence in the plasmid pG118Bgl, containing 301 bp ofthe proximal promoter, the 5 0 untranslated region ofthe mRNA, the entire protein coding region and 425 bp ofthe 3 0 untranslated region ofthe mRNA, with the 1.9 kb neo cassette fromthe plasmid pGT- N29 containing a BamHI site. The transcription ofthe neo cassette, as indicated by arrow, is antisense with respect to the Gadd45b gene. (b) Southern blot analysis of BamHI digested genomic DNA from transfected ES cell clones, probed with the 461 bp BglII– BamHI flanking probe. The BamHI site introduced into the gene-targeting vector with the neo cassette, converted the 5.8 kb wt BamH1 band into a 3.3 kb band for genes that had undergone homologous recombination. The sizes of BamHI fragments are indicated alongside bands, and genotype ofeach clone is designated above each lane. ( c) PCR genotyping ofmouse-tail DNA was performed with three primers (Materials and methods). A wt specific upstream primer was located in the fourth exon, five prime to the SacII site, in the region ofthe Gadd45b gene that was replaced by the neo cassette. The neo, or null, specific primer was located in the PGK promoter ofthe neo cassette. The common downstream primer was located in the fourthexon, three prime to the SacII site that formed the three prime junction between the neo cassette and the fourthexon ofthe Gadd45b gene. The size in bp ofbands in the molecular weight marker lane (MW), and bands for wt and null alleles, are indicated alongside the bands, and the genotypes of the individual mice are designated above each lane

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7172 BM underwent apoptosis compared to 11% ofwt BM cells. Finally, it can be seen that the percentage of apoptotic cells after DNR treatment also was significantly higher in the gadd45aÀ/À and gadd45bÀ/À BM cells compared to wt cells (2.5- and twofold, respec- tively). In conclusion, both gadd45a- and gadd45b- deficient BM cells consistently showed a higher sensitivity to genotoxic-stress-induced apoptosis compared to wt BM cells. Figure 2 Gadd45a and gadd45b transcripts are induced under To confirm the higher apoptotic index in gadd45aÀ/À various stress conditions. BM cells from wt and both the gadd45 and gadd45bÀ/À BM cells compared to wt cells, the KO mice were treated with VP-16 (100 mM), DNR (1 mM) and UV TUNEL assay was employed, using terminal deoxy- (25 J/m2). After 4 h of treatment, total RNA (15 mg/lane) was run nucleotidyl-transferase to label the ends of apoptotic on formaldehyde gels and hybridized with 32P-labeled Gadd45a and Gadd45b cDNA probe as indicated DNA breakdown fragments with fluorescein- conju- gated dUTP to allow detection ofapoptotic cells by FACS analysis. BM cells from both gadd45À/À mice and control gadd45 þ / þ littermates were treated with obtained from either wt or gadd45a- and gadd45b- VP-16, DNR and UV at the same concentration as used deficient mice following treatment with stress agents. To in Annexin assay above, cells were cultured for 48 h, and this end, wt and gadd45aÀ/À or gadd45bÀ/À BM cells, the percentage ofTUNEL-positive apoptotic cells was consisting mainly ofcells ofthe myeloid lineage determined by FACS analysis. As shown in Figure 3c, (Materials and methods), were treated with UVC there was an increase in the percentage ofTUNEL- radiation, VP-16 and DNR, and expression ofgadd45a positive cells following various stresses in both gadd45- and gadd45b was assessed using RNA blot analysis. As deficient BM cells compared to wt cells. Following shown in Figure 2, gadd45a and gadd45b mRNA treatment with UV and VP-16, the percentage of expression were increased 8–10-fold above untreated TUNEL-positive BM cells was about twofold higher levels after4 h oftreatment ofBM cells with the in both gadd45aÀ/À and gadd45bÀ/Àcells compared to indicated stress agents. As expected, no gadd45a wt cells, whereas treatment with DNR resulted in a 3–4- transcripts were detected in gadd45a null cells, and no fold increase in the percentage of TUNEL-positive cells gadd45b transcripts were seen in gadd45bÀ/À BM cells. in gadd45a- and gadd45b-deficient BM cells compared Taken together, these data indicate that in hematopoie- to wt cells. Taken together, these results are consistent tic cells gadd45a and gadd45b genes are induced with the observations made using AnnexinV to analyse following exposure to either UV or chemotherapeutic apoptosis, indicating that deficiency in gadd45a or drugs, suggesting that they may play a role in the gadd45b results in increased apoptosis ofBM cells cellular response ofhematopoietic cells to stress. following exposure to the genotoxic stress agents UVC radiation, VP-16 and DNR. Gadd45a- and gadd45b-deficient BM-derived myeloid cells have increased sensitivity to genotoxic-stress-induced Introduction of wt gadd45 into gadd45aÀ/À BM cells apoptosis restored the wt phenotype A number ofprevious reports have suggested a role for To confirm that the higher apoptosis in gadd45a null gadd45a and gadd45b in mediating the cellular response BM cells is due to gadd45a deficiency and not due to to DNA damage, but it remained controversial whether other genetic alterations, BM cells from gadd45a null these genes have a pro- or antiapoptotic role. Thus, we mice were transiently transduced with retroviral vector aimed to determine how loss ofgadd45a or gadd45b MIGW–green fluorescent protein (GFP)–gadd45a. BM function affects genotoxic-stress-induced apoptosis in cultures transduced with MIGW–GFP–gadd45a or hematopoietic cells. BM cells from gadd45aÀ/À, empty control vector MIGW–GFP were exposed to gadd45bÀ/À and wt littermate mice were treated with UVC (15 J/m2), and cells were cultured for 24 h. Flow UV, VP-16 and DNR, and the effect of loss of gadd45a cytometry was employed to identify transduced cells or gadd45b on cell survival was determined via staining which are positive for GFP (emission spectra read in with Annexin V, which is an early indicator of FL-1), ofwhich the fraction ofapoptotic cells was apoptosis. Figure 3a depicts a representative dual-color identified via staining with Annexin V conjugated to flow-cytometric analysis, whereas Figure 3b summarizes PE (emission spectra read in FL-2). As shown in the results ofat least three independent experiments. As Figure 4, about 36% ofthe GFP-positive gadd45a À/À shown in Figure 3a, higher apoptosis was observed BM cells transduced with control MIGW–GFP vector following UV exposure in gadd45 null BM cells were apoptotic, as indicated by staining for Annexin– compared to wt cells; on the average, 24% of PE, whereas only 12% ofthe same cells transduced gadd45aÀ/À BM cells and 18.9% ofgadd45b À/À cells with MIGW–GFP–gadd45a appeared to undergo apop- were apoptotic, compared to 9.6% ofwt cells. Follow- tosis. This result indicated that transduction of gadd45a ing treatment with VP-16 (25 mM), about 26.5% of into gadd45aÀ/À BM cells protects hematopoietic gadd45aÀ/À BM cells and 20.6% ofgadd45b À/À null cells from UV-induced apoptosis. Similar results were

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7173 4

Con4 UV VP-16 DNR a 4 4 10 10 10 10 3 3 3 3 10 10 10 10 2 2 2 2 10 10 10 10 Wild- 1 1 1 1

FL2-Height FL2-Height FL2-Height FL2-Height type 10 10 10 2.5% 9.6% 11% 10 12.5% 0 0 0 0 10 10 0 1 2 3 4 0 1 2 3 4 10 0 1 2 3 4 10 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 FL1-Height FL1-Height FL1-Height FL1-Height 4 4 4 4 10 10 10 10 3 3 3 3 10 10 10 10 gadd45a-/- 2 2 2 2 10 10 10 10 FL2-Height FL2-Height 1 1 1 1 FL2-Height FL2-Height 10 10 10 10 30.5% 3.5% 24.0 26.5% 0 0 0 PI 0 10 10 10 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 10 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 FL1-Height FL1-Height FL1-Height FL1-Height 4 4 4 4 10 10 10 10 3 3 3 3 10 10 10 10 2 2 2 2 10 10 10 10 gadd45b-/- FL2-Height FL2-Height 1 FL2-Height 1 1 1 FL2-Height 10 10 10 10 3.0% 18.9% 20.6% 25.5% 0 0 0 0 10 10 10 0 1 2 3 4 0 1 2 3 4 10 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 FL1-Height FL1-Height FL1-Height FL1-Height

Annexin V

70 b wt c 35 wt gadd45a-/- 60 gadd45a-/-gadd45a-/- 30 gadd45b-/- 50 gadd45b-/-gadd45b-/- 25 40 20 30 15

10 20

5 10 % Tunel positive cells

% Annexin V positive cells 0 0 c UV VP-16 DNR con UV VP-16 DNR Figure 3 Gadd45a- and gadd45b-deficient BM-derived myeloid cells are sensitized to genotoxic-stress-induced apoptosis. (a) Flow- cytometric analysis showing the percentage ofAnnexin-positive cells. Results shown in the upper panel are representative ofat least three independent experiments whose results are summarized in the bar graph shown in the lower panel (b). Wt, gadd45aÀ/À and gadd45bÀ/À BM cells, were treated with VP-16 (25 mM), DNR (0.125 mM) and UV (15 J/m2). After 24 h of incubation, cells were assessed for apoptosis by flow cytometry after labeling with Annexin V–FITC and PI as described in Materials and methods. (c) Percentage ofTUNEL-positive cells followingstress stimuli. Wt, gadd45a À/À and gadd45bÀ/À BM cells were treated with VP-16, DNR and UV at the same concentration as used in Annexin assay above. DNA fragmentation was analysed by TUNEL assay. Data represent the mean7s.d. of three different experiments obtained upon analysis ofgadd45b À/À BM cells activation ofcaspase-3 in BM cells was manifestedby transduced with MIGW–GFP–gadd45b retrovirus (data detection ofthe p19 and p17 cleavage products ofpro- not shown). caspase-3. After 1 h of exposure to UV, gadd45a or gadd45b null BM cells showed the appearance ofthe p19 and p17 cleavage product, while UV exposure ofwt Increased apoptosis in gadd45a- and gadd45b-deficient cells showed significantly less cleavage ofthe full-length cells is evident by enhanced caspase-3 activation and poly- caspase-3. ADP-ribose polymerase (PARP) cleavage To further confirm that UV enhanced caspase-3 One ofthe major biochemical events that commit a cell activation in gadd45 null cells, we analysed the specific to apoptosis is activation ofcaspases, in which caspase-3 cleavage ofthe caspase-3 substrate PARP. As shown in represents one ofthe key executors in apoptosis. Thus, it Figure 5a, there was enhanced appearance ofthe 89- was ofinterest to see whether the higher apoptosis in kDa PARP cleavage product and loss ofthe full-length gadd45a null and gadd45b null BM cells entails PARP protein after UV exposure in gadd45a and enhanced caspase-3 activation. As shown in Figure 5a, gadd45b null cells compared to wt cells.

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7174 bcl-xL) also play an important role in protection from apoptosis and inhibition ofcaspase activation. Thus, it was ofinterest to determine the expression ofthese antiapoptotic proteins following UV exposure in the different gadd45 null BM cells compared to their control wt cells (Figure 5b). It can be seen that in wt BM cells cIAP-1, Bcl-2 and Bcl-xL were upregulated following UV exposure, whereas in both gadd45a- and gadd45b- deﬁcient BM cells cIAP-1and bcl-xL expression was not induced. Following UV exposure, also the level ofbcl-2 was much higher in wt cells compared to that in both gadd45a and gadd45b null cells.

Figure 4 Transduction ofgadd45a into gadd45a-deficient BM cells restores the wt apoptotic phenotype. Effect of UV on Gadd45aÀ/À and gadd45bÀ/À BM cells exhibit gadd45aÀ/À BM cells transduced with MIGW-GFP-gadd45a decreased in vitro clonogenicity following treatment and control MIGW-GFP vector. Transduced BM cells were with genotoxic stress washed and exposed to UV (15 J/m2) and were cultured for 24 h. Apoptosis was evaluated by flow cytometry using PE-conjugated To determine how the absence of gadd45a or gadd45b Annexin V (Materials and methods). Bar diagram shows the genes may affect in vitro colony formation following GFP þ Annexin–PE/total GFP cells. Data are means7s.d. ofthree independent experiments genotoxic stress, BM cells from gadd45aÀ/À, gadd45bÀ/À and wt littermate mice were treated with various stresses prior to assessment for colony formation in the presence of IL-3, conditions under which the majority ofthe colonies (>95%) are myeloid colonies (Krishnaraju et al., 2001), including blast colonies (B4–6%), granulocytic colonies (B40%) and macrophage colonies (50%). Upon UVC exposure, gadd45aÀ/ÀBM cells yielded sixfold fewer colonies compared to similarly treated wt BM cells. BM cells from gadd45bÀ/À mice showed a 3.7-fold reduction in colony formation (Figure 6). When treated with chemotherapeutic drugs (VP-16, DNR), colony formation was decreased fourfold in wt cells compared to untreated cells, and was further decreased (8–16-fold) in gadd45a- and gadd45b-deficient BM cells. These data complement the previous observations, revealing that gadd45a and gadd45b also play a pronounced role in protection ofIL-3-responsive myeloid colony-forming progenitor cells from genotoxic-stress-induced cell death.

Figure 5 (a) Caspase-3 activation and PARP cleavage in gadd45a and gadd45b null BM-derived myeloid cells. BM cells were exposed to UV (15 J/m2) and cells were cultured for 1 h and lysed. From each lysate, 50 mg ofprotein was analysed by immunoblotting for the proform and the p19 and p17 cleavage products of caspase-3, as well as for PARP and its 89 kDa cleavage product. (b) Gadd45a and gadd45b deﬁciency impairs induction ofantiapoptotic proteins: BM cells from wt and both gadd45aÀ/À and gadd45bÀ/À mice were exposed to UV(15 J/m2) and cell were cultured for 1 h. Cells were lysed and 50 mg ofprotein was analysed by immunoblotting for expression of cIAP-1, Bcl-2 and Bcl-xL proteins. b-Actin was used as an internal loading control for immunoblots shown in (a) and (b)

Figure 6 Myeloid progenitors from gadd45a- and gadd45b- c-Inhibitor of apoptotic protein (IAP)-1, Bcl-2 and deﬁcient mice exhibit decreased clonogenicity following stress. BM cells from wt and both gadd45 null mice were treated with Bcl-xL induction is impaired in gadd45a and gadd45b VP-16 (2.5 mM), DNR (0.02 mM) and UV (15 J/m2). After 24 h, deﬁcient cells 50 000 cells were seeded in Metho-Cult (supplemented with 10% Wehi-3B conditioned medium (source ofIL-3)). The bar diagram Apoptosis can also be modulated by downregulation of displays the mean percentage ofCFU colonies 7s.d. oftriplicate IAP activity. Members ofthe Bcl-2 family(bcl-2 and experiments

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7175 Gadd45aÀ/À and gadd45bÀ/À BM cells do not undergo and gadd45b are required for the DNA damage-induced G2/M arrest following genotoxic stress induced by UV or G2/M checkpoint in hematopoietic cells, we compared VP-16 the cell cycle distribution ofthe control gadd45 þ / þ BM cell population with that of gadd45aÀ/À and Earlier studies have implicated the gadd45 family of gadd45bÀ/À cells following treatment with the various genes in cell cycle regulation. To test whether gadd45a stress stimuli. Figure 7a depicts a representative ﬂow

a Con UV VP-16 DNR

G2/M 1000 G2/M G2/M G2/M 800 800 280 2.8% 10.5% 12.5% 15.0% 800 210 600 600 600 140 400 400 WT Number Number 400 Number Number

70 200 200 200

0 0 0 0 0 40 80 120 160 200 0 50 100 150 200 250 0 50 100 150 200 250 0 100 150 200 250 Channels (FL2-A) Channels (FL2-A) Channels (FL2-A) Channels (FL2-A)

G2/M G2/M 800 G2/M 400 280 G2/M 240 4.2% 16.0% 4.0% 3.8%

600 300 210 180 Gadd45a-/- 400 200 140 120 Number Number Number Number

200 100 70 60

0 0 0 0 0 40 80 120 160 200 0 40 80 120 160 0 40 80 120 160 200 0 50 100 150 200 250 Channels (FL2-A) Channels (FL2-A) Channels (FL2-A) Channels (FL2-A) 350

320 280 800 G2/M G2/M G2/M 600 G2/M 3.5% 3.9% 3.6% 14.6% 500 240 600 210 400 Gadd45b-/- Number 400 160 140 300 Number Number Number 200 200 80 70 100

00 0 0 0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200 0 50 100 150 200 250 Channels (FL2-A) Channels (FL2-A) Channels (FL2-A) Channels (FL2-A)

Relative DNA content b

Figure 7 Gadd45a- and gadd45b-deficient BM cells are defective in G2/M arrest induced by UV and VP-16. (a) Representative histograms showing cell cycle distribution following exposure to stresses as indicated. Following 8 h treatment of wt and gadd45- deficient cells with UV (25 J/m2), VP-16 (100 mM) and DNR (1 mM), cells were fixed by 70% ethanol at À201C overnight. Subsequently, DNA was stained with PI and examined by flow cytometry. Results shown in (a) are representative ofthree independent experiments whose results are summarized in the bar graph shown in (b); s.d. was up to 10% (i.e. 3073%)

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7176 cytometry analysis, whereas Figure 7b summarizes the (Maeda et al., 2002). Additional evidence for a results ofat least three independent experiments, proapoptotic function of gadd45a and gadd45g has including the analysis shown in Figure 7a. It can be been documented recently using either overexpression of seen that cell cycle profiles ofboth gadd45 À/À BM cells these proteins in prostate and breast cancer cell lines or and their control wt BM cells were similar in the absence siRNA-mediated knockdown that was observed to ofstress, indicating that loss ofeither gadd45a or block IkB-mediated apoptosis in these cell lines (Zerbini gadd45b did not interfere with normal cell cycle et al., 2004). On the other hand, our results are progression. However, following treatment with UV or consistent with an earlier report where it was shown VP-16, BM cells from gadd45aÀ/À and gadd45bÀ/À that p53 null MEFS that lack gadd45a are sensitized to mice failed to arrest at G2/M. It can be seen that only UV- and cisplatin-induced cell death (Smith et al., 2000). 2.8% ofuntreated wt BM cells were in the G2/M phase, Also, a recent study on B cells showed that gadd45b whereas following UV treatment 10.5% of wt BM cells mediates the protective effects of CD40 co-stimulation accumulated in the G2/M phase, after VP-16 treatment against Fas-induced apoptosis (Zazzeroni et al., 2003). 12.5% ofthe wt cells were in G2/M, and following DNR Thus, it is possible that the stress stimulus encountered, treatment 15.0% ofwt cells accumulated in the G2/M the cell type and interaction with other proteins that phase ofcell cycle. In contrast, both gadd45a À/À and modulate gadd45 function ultimately determine whether gadd45bÀ/À BM cells failed to arrest in G2/M after the outcome will be DNA repair and cell survival, or exposure to UVC and VP-16. Interestingly, after DNR apoptotic cell death. This notion is supported by our treatment, both gadd45aÀ/À and gadd45bÀ/À BM cells observation that the gadd45bÀ/À BM cells, though were observed to accumulate in G2/M similar to wt sensitized to genotoxic-stress-induced apoptosis, are cells. These data indicate that BM cells deficient in largely resistant to TGF-b-induced apoptosis compared gadd45a or gadd45b were defective in the G2/M to wt BM cells (data not shown). checkpoint following exposure to UVC or VP-16, yet Generation ofgadd45a and gadd45b null mice retained a normal G2/M checkpoint in response to allowed us to address the role ofgadd45 in apoptosis. DNR. We showed that induction ofan important antiapoptotic gene, cIAP-1, along with bcl-2 and bcl-xL are reduced in both gadd45 null mice. cIAP-1 inhibits the Discussion active forms of caspase-3, which are responsible for PARP cleavage. This might explain the increased In this study, we addressed the role gadd45a and sensitivity ofgadd45 null BM cells compared to wt cells gadd45b play in the response ofhematopoietic cells to to genotoxic-stress-induced apoptosis. Ongoing studies genotoxic stress, taking advantage ofBM cells obtained are targeted at identifying the signaling pathways which from gadd45a- and gadd45b-deficient mice. Our data gadd45a and gadd45b utilize to protect hematopoietic show for the first time that hematopoietic cells from cells from UV-, VP-16- or DNR-inflicted genotoxic gadd45a- and gadd45b-deficient mice are sensitized to stress. genotoxic-stress-induced apoptosis. We also demon- Several earlier studies have demonstrated that gad- strated that deficiency in either gadd45a or gadd45b d45a and gadd45b mediate a G2/M cell cycle checkpoint impairs the capacity ofBM-derived myeloid progenitors in both human and murine cells (Wang et al., 1999; to form colonies in vitro following exposure to genotoxic Zhan et al., 1999). In the present study, cell cycle stress. Furthermore, by reintroducing gadd45a or analysis ofthe gadd45a and gadd45b BM null cells in gadd45b into the gadd45 null BM cells, it was clearly response to various DNA-damaging agents indicated demonstrated that gadd45a or gadd45b deficiency, and that these cells are deficient in both UV- and VP-16- not any additional genetic alterations, is responsible for induced G2/M arrest. These data are consistent with an the increased apoptosis ofthe gadd45 null cells in earlier report showing that antisense gadd45a expressing response to genotoxic stress. Taken together, these data human colon carcinoma cells are defective in the UV- clearly indicate that Gadd45a and Gadd45b proteins induced G2/M checkpoint (Vairapandi et al., 2002). In protect hematopoietic cells from genotoxic stress in- contrast, when either gadd45a- or gadd45b-deficient BM duced by UVC, VP-16 or DNR, and thus can be cells were treated with DNR, they still retained a assigned an antiapoptotic function in this context. functional G2/M checkpoint. Thus, the DNR-mediated Our findings are surprising in light ofprevious data G2/M checkpoint pathway differs from the UV- and that have identified gadd45 genes as proapoptotic. It was VP-16-mediated pathways. Wang et al. (1999) reported shown that ectopic expression ofgadd45a or gadd45b that gadd45a-deficient cells still retain a functional G2/ sensitizes H1299 human lung carcinoma cells to M checkpoint after IR, yet they are deficient in UV- and apoptosis induced by genotoxic stress (Zhang et al., MMS-induced G2/M (Zhan et al., 1999). These 2001). Recently, by taking advantage on the gadd45bÀ/ observations are consistent with the existence ofmulti- À mice, it has been documented that gadd45bÀ/À ple G2/M checkpoints that are stimulus specific, and are hepatocytes are resistant to TGF-b-mediated apoptosis either gadd45 dependent or gadd45 independent. Our (Yoo et al., 2003), which was consistent with previous data suggest that both gadd45a and gadd45b are observations (Selvakumaran et al., 1994a, b). Also, using required in hematopoietic cells for a normal G2/M the gadd45aÀ/À mice, it has been reported that gadd45a checkpoint in response to certain types ofDNA damage. promotes UV-induced apoptosis in skin keratinocytes Further studies are needed to decipher the underlying

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7177 mechanisms. Nevertheless, it is plausible that the EcoRI digestion and gel purification, the fragment was cloned increased apoptotic response triggered by UV or VP- into the EcoRI site ofpG118neo to generate the gene-targeting 16 in gadd45a- and gadd45b-deficient BM cells also vector pG118KO, where 301 bp ofthe gadd45b proximal promoter, the 50 untranslated region ofthe mRNA, the entire reflects the failure of these cells to arrest in G2/M to 0 allow repair ofdamaged DNA. protein coding region and 425 bp ofthe 3 untranslated region ofthe gadd45b mRNA were replaced by the 1.9 kb neo cassette Overall, our data identify both gadd45a and gadd45b from the plasmid pGT-N29, and where transcription of the as prosurvival gene(s), that protect hematopoietic cells neo cassette is antisense with respect to the gadd45b gene. The against certain DNA damage agents, and suggest that targeting plasmid (pG118KO) was linearized using the PmeI cells disabled for either gene may display higher site in the pNEB193 polylinker and electroporated into mouse susceptibility to killing by UV and certain anticancer ES cells (129VJ-1183) and selected with G418. ES cell drugs. Together, these findings demonstrate that loss of transfectants were assayed for homologous recombination of either gadd45a or gadd45b function may have a the Gadd45b gene by Southern blot analysis. BamHI digested profound impact on the apoptotic efficacy of certain genomic DNA was Southern blotted and probed with a 461 bp 0 anticancer agents. This knowledge should contribute to BglII-BamHI fragment that was located immediately 3 to the a greater understanding ofthe genetic events involved in region ofgenomic DNA corresponding to the targeting vector. The wt allele displayed a 5.8 kb fragment, while the recombi- the pathogenesis of different leukemias and response of nant, null allele, displayed a 3.3 kb BamHI fragment. normal and malignant hematopoietic cells to chemo- Gadd45b-targeted ES cell clones were injected into C57/Bl6 and radiation therapy. These observations set the stage (B6) blastocytes that were used to generate Gadd45b þ /À B6/ to evaluate, in clinically relevant settings, the impact 129VJ germline chimeric mice as described in Hollander et al. that the status ofgadd45a and gadd45b might have on (1999). Germline gadd45b þ /À chimeric males were back- the efficacy of DNR or VP-16 in killing leukemic cells. crossed with C57/BL6 females and resulting Gadd45 þ /À B6/ 129VJ offsprings were used in experiments.

Materials and methods Mice used and genotyping Mice null for gadd45a have been established in the laboratory Chemicals and antibodies ofDr Albert Fornace, NIH (Hollander et al., 1999). VP-16 and DNR were purchased from SIGMA (St Louis, Gadd45aÀ/À and gadd45bÀ/À mice included in this study MO, USA). VP-16 was reconstituted in dimethyl sulfoxide were housed at the animal facility at Temple University and (DMSO) and DNR was diluted in phosphate-buffered saline used at the age of6–8 weeks. Both gadd45a À/À and (PBS) to make a 5 mM stock solution. Propidium iodide (PI) gadd45bÀ/À mice were maintained in a mixed (BL6/129VJ) and RNase were also purchased from Sigma Chemicals. genetic background and propagated using heterozygous Annexin V–fluorescein isothiocyanate (FITC) and TUNEL breeding pairs. Mating between gadd45b heterozygous mice staining kit were obtained from BD Pharmingen (San Diego, gave the expected frequency of gadd45b null (gadd45bÀ/À) CA, USA). Growth factors rIL-3 and Flt-ligand were mice and their wt (gadd45b þ / þ ) littermates that were used as purchased from R&D system (MN, USA), other recombinant age-matched controls for all indicated experiments. PCR cytokines (rIL-6 and rSCF) were generous gifts from Amgen genotyping ofthe Gadd45b gene, using mouse-tail DNA, Inc. (Thousand Oaks, CA, USA). Retronectin was purchased was performed with three primers. A wt specific upstream from Takara, Fisher Scientific (Park lane, PA, USA) and primer (50 GCTGTGGAGCCAGGAGCAGCA 30) was lo- reconstituted in sterile water at 25 mg/ml. Antibodies against cated in the fourth exon, 50 prime to the SacII site, in the bcl-2, bcl-xL, caspase-3 and PARP were obtained from Cell region ofthe Gadd45b gene that was replaced by the neo Signaling Technologies (Beverly, MA, USA). Antibody cassette. The neo, or null, specific primer (50 AAGCG against b-actin was obtained from Santa Cruz Biotechnology CATGCTCCAGACTGCCTT 30) was located in the PGK Inc. (Santa Cruz, CA, USA). Methocult and ammonium promoter ofthe neo cassette. The common downstream primer chloride were purchased from Stem Cell Technologies (50 GCTGTGGAGCCAGGAGCAGCA 30) was located in the (Vancouver, Canada). TRIZOL reagent was purchased from fourth exon, 30 prime to the SacII site that formed the 30 prime Gibco BRL, Invitrogen. junction between the neo cassette and the remainder ofthe fourth exon of the gadd45b gene. The wt gadd45b PCR product was 310 bp long and the null PCR product was 190 bp Targeted deletion of the murine gadd45b gene and generation of in length. The gadd45a genotype was determined using primers gadd45b / mice À À derived from the genotyping primers used by Hollander et al. The targeting vector pG118KO was derived from the plasmid The wt specific upstream primer (50 CACCTCTGC pG118Bgl (Balliet et al., 2001). This plasmid encoded a 9513 TACCTCTGCACAAC 30) was modified from the sequence base pair (bp) mouse genomic, BglII DNA fragment contain- ofprimer CH215. The common downstream primer (5 0 ing the gadd45b gene cloned into the BamHI site ofpNEB193 CCAGAAGACCTAGACAGCACGGTT 30) was derived (New England Biolabs). The targeting vector was constructed from primer CH216 (Hollander et al., 1999). The neo specific in two steps. First, the EcoRI-blunt end HindIII fragment upstream primer was the same used for the PCR genotyping of containing the neomycin resistance gene with the PGK gadd45b mice and was derived from the primer CH21714. The promoter from the plasmid pGT-N29 (New England Biolabs) wt gadd45a PCR product was 310 bp long and the null PCR was cloned into the EcoRI-blunt-ended SacII site of product was 190 bp in length. pG118Bgl, creating the plasmid pG118neo. The first 4430 bp ofpG118Bgl were then amplified using PCR with Vent BM cell preparation polymerase (New England Biolabs), and primers that encoded EcoRI restriction sites. The amplified region was sequenced to For each experiment indicated, gadd45aÀ/À, gadd45bÀ/À confirm accurate copying ofthe original sequence. After and their wt littermate mice (6–8-weeks old) were killed and

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7178 both the femurs were removed. BM cells were flushed kind gift from Dr Luk Van Parjis. Infectious virus particles from each femur and cultured in Iscove modified Dulbecco’s were produced by transient transfection of amphotropic medium (IMDM, Invitrogen). For all indicated experiments, phoenixTm retroviral packaging cells (Orbigen, San Diego, BM cells were cultured in IMDM (with 10% heat-inactivated CA, USA) with the retroviral plasmids, MIGW and MIGW- FBS and supplemented with growth factors (rIL-3 (10 ng/ml), gadd45a, and supernatants were collected 48 h after transfec- rSCF (25 ng/ml), rIL-6 (10 ng/ml))) and cultured for 48 h at tion. BM cells were cultured in IMDM medium supplemented 371C humidified atmosphere with 10% CO2 prior to treatment with high growth factors Flt-3 ligand (100 ng/ml), rSCF with stress agents. Cytospin preparations ofthe BM cells, (50 ng/ml), rIL-6 (10 ng/ml) and rIL-3 (25 ng/ml) at 371C following culturing as indicated, have shown that >90% of for 48 h. Subsequently, cells were re-suspended in virus the cell population consisted ofmyeloid cells at the blast and supernatant, and plated on retronectin-coated plates. Cells early intermediate stages of differentiation for all the were harvested after 48 h, washed and exposed to UVC (15 J/ genotypes, and no significant difference in cell composition m2), and cells were cultured for 24 h. Flow cytometry was observed among different genotypes. was employed to identify transduced cells which are positive for GFP (fluorescence emission spectra read in FL-1), of RNA extraction and Northern blot analysis which the fraction of apoptotic cells was identified using Annexin V conjugated to PE (fluorescence emission spectra Following culturing for 48 h as indicated above, BM cells read in FL-2). were treated with 100 mM ofVP-16, 1 mM ofDNR or 25 J/m 2 UVC. After 4 h of incubation period, total RNA was extracted using TRIZOL reagent according to the manufacturer’s Western blot analysis instructions. Total RNA (15 mg per lane) was electrophoresed BM cells cultured for 48 h, as indicated above, were exposed to on a 1% agarose gel containing formaldehyde (0.7%), and the UV (15 J/m2), cultured for 24 h, washed with cold PBS, loading ofequal amounts was confirmed by comparing subsequently lysed with 1 Â lysis buffer (Cell Signaling; intensity ofGAPDH (internal control) band. For Northern supplemented with 1 mM PMSF) and lysates were quantified. Blot analysis, gels were denatured and then transferred onto a In all, 50 mg oftotal protein was separated on 10% SDS– nitrocellulose membrane (Stratagene, La Jolla, CA, USA). The PAGE and transferred onto poly-vinylidene difluoride 1 membranes were baked at 80 C for 2 h and subsequently (PVDF) membrane (Millipore, MA, USA). Blots were probed hybridized with different probes overnight at 421C. Blots were 1 with primary antibodies against caspase-3, PARP, Bcl-2, Bcl- washed at 60 C with SSC buffer, and exposed to X-ray films xl, cIAP-1 and b-actin used as internal loading control. After for 24 h. To re-hybridize with other probes, the RNA blots three washes, the blots were incubated with 1 : 4000 dilution of were stripped with boiling stripping solution (0.1 Â SSC and peroxidase-conjugated secondary antibody, developed with 0.1% SDS) 5 min each for 4–5 times and exposed to X-ray ECL (Pierce Biotechnology, Rockford, IL, USA) detection films for 24 h. Plasmids and DNA probes were prepared as reagent and autoradiography was carried out. previously described (Selvakumaran et al., 1994a, b). Blots were quantified by the aid ofa Fuji BAS 2000 Phosphorimage analysis. Cell cycle analysis BM cells following culturing for 48 h, as indicated above, were Assay for apoptosis treated with VP-16, DNR or UV at the same conditions as To induce apoptosis, BM cells following culturing for 48 h, as used in apoptosis assay above. After 8 h of incubation, cells indicated above, were seeded at 1.0 Â 106 cells/30 mm plate were washed with PBS and fixed in ice-cold 70% ethanol 2 overnight at 201C. Cells were then treated with RNAase with VP-16 (25 mM) or DNR (0.125 mm) or UVC (15 J/m ) À and cells were cultured for 24 h (Annexin V–FITC) and (50 mg/ml) and subsequently stained with PI (100 mg/ml) for 1 48 h (TUNEL). Cells were then harvested and washed once 30 min at 37 C. Cell cycle was analysed by flow cytometry with PBS (pH 7.2). Two alternative methods were used to using Modfit software. assess the apoptosis before (time 0 h) and after (time 24 or 48 h) culture. Colony-formation assay BM cells following culturing for 48 h, as indicated above, were Annexin staining Apoptosis was measured by staining treated with 2.5 mM VP-16 or 0.025 mM DNR or 15 J/m2 UV. with Annexin V–FITC and PI, and analysed by flow cyto- After 24 h, 50 000 cells were seeded in Metho-Cult (supple- metry (FACScan, Becton Dickinson, Cell Quest software). mented with 10% Wehi-3B conditioned medium (source ofIL- Annexin V-positive cells were defined as percentage of 3)) in 35 mm petri-plates. Petri-plates were incubated in 371C apoptotic cells. Apoptosis, as measured by Annexin V, was humidified chamber with 5% CO2. Colony-forming unit- first detectable at 6 h, and optimal at 24 h; doses ofVP-16, granulocytic and macrophage (CFU-GM) colonies were DNR and UVC used in experiments represent optimal doses scored on the seventh day under an inverted microscope and for apoptosis at 24 h. cytospin smears were made to assess the morphology ofcells in individual colonies. The majority (>95%) ofthe colonies TUNEL assay To confirm the apoptosis, the TUNEL assay appeared to be myeloid colonies, including blast colonies was performed using a TUNEL assay kit. Briefly, fixed cells (4–6%), granulocyte colonies (B40%) and macrophage were labeled with TdT and FITC dUTP, and quantification of colonies (B50%). apoptotic cells was performed by flow cytometry. Acknowledgements Retroviral transduction of BM cells We thank Ms Diana Vesely (graduate student) for construc- The MIGW-gadd45a retroviral construct was generated in tion ofMIGW-gadd45a retroviral construct. This work was this laboratory. The original MIGW vector (that has the GFP supported by NIH grants 5 RO1 CA89718-02 (DAL) and RO1 gene downstream ofthe internal ribosomal entry site) was a HL 70530-01 (DAL).

Oncogene gadd45a and gadd45b in hematopoietic stress M Gupta et al 7179 References

Azam N, Vairapandi M, Zhang W, Hoffman B and Smith ML, Ford JM, Hollander MC, Bortnick RA, Amund- Liebermann DA. (2001). J. Biol. Chem., 276, 2766–2774. son SA, Seo YR, Deng CX, Hanawalt PC and Fornace Jr Balliet AG, Hollander MC, Fornace Jr AJ, Hoffman B and AJ. (2000). Mol. Cell. Biol., 20, 3705–3774. Liebermann DA. (2001). DNA Cell Biol., 20, 239–247. Takekawa M and Saito H. (1998). Cell, 95, 521–530. Beadling C, Johnson KW and Smith KA. (1993). Proc. Natl. Vairapandi M, Azam N, Balliet AG, Hoffman B and Acad. Sci. USA, 90, 2719–2723. Liebermann DA. (2000). J. Biol. Chem., 275, 16810–16819. Fornace Jr AJ, Nebert DW, Hollander MC, Luethy JD, Vairapandi M, Balliet AG, Fornace Jr AJ, Hoffman B and Papathanasiou M, Fargnoli J and Holbrook NJ. (1989). Liebermann DA. (1996). Oncogene, 12, 2579–2594. Mol. Cell. Biol., 9, 4196–4203. Vairapandi M, Balliet AG, Hoffman B and Liebermann DA. GuilloufC, Grana X, Selvakumaran M, De Luca A, Giordano (2002). J. Cell. Physiol., 192, 327–338. A, Hoffman B and Liebermann DA. (1995). Blood, 85, Wang XW, Zhan Q, Coursen JD, Khan MA, Kontny HU, Yu 2691–2698. L, Hollander MC, O’Connor PM, Fornace Jr AJ and Harris Hollander MC, Sheikh MS, Bulavin DV, Lundgren K, Augeri- CC. (1999). Proc. Natl. Acad. Sci. USA, 96, 3706–3711. Henmueller L, Shehee R, Molinaro TA, Kim KE, Tolosa E, Yang J, Zhu H, Murphy TL, Ouyang W and Murphy KM. Ashwell JD, Rosenberg MP, Zhan Q, Fernandez-Salguero (2001). Nat. Immunol., 2, 157–164. PM, Morgan WF, Deng CX and Fornace Jr AJ. (1999). Nat. Yoo J, Ghiassi M, Jirmanova L, Balliet AG, Hoffman B, Genet., 23, 176. Fornace Jr AJ, Liebermann DA, Bottinger EP and Roberts Kastan MB, Zhan Q, el-Deiry WS, Carrier F, Jacks T, Walsh AB. (2003). J. Biol. Chem., 278, 43001–43007. WV, Plunkett BS, Vogelstein B and Fornace Jr AJ. (1992). Zazzeroni F, Papa S, Algeciras-Schimnich A, Alvarez K, Melis Cell, 274, 29592–29599. T, Bubici C, Majewski N, Hay N, De Smaele E, Peter ME Krishnaraju K, Hoffman B and Liebermann DA. (2001). and Franzoso G. (2003). Blood, 102, 3270–3279. Blood, 97, 1298–1305. Zerbini LF, Wang Y, Czibere A, Correa RG, Cho JY, Ijiri K, Maeda T, Hanna AN, Sim AB, Chua PP, Chong MT and Wei W, Joseph M, Gu X, Grall F, Goldring MB, Zhou JR Tron VA. (2002). J. Invest. Dermatol., 119, 22–26. and Libermann TA. (2004). Proc. Natl. Acad. Sci. USA, 101, Papathanasiou MA, Kerr NC, Robbins JH, McBride OW, 13618–13623. Alamo Jr I, Barrett SF, Hickson ID and Fornace Jr AJ. Zhan Q, Antinore MJ, Wang XW, Carrier F, Smith ML, (1991). Mol. Cell. Biol., 11, 1009–1016. Harris CC and Fornace Jr AJ. (1999). Oncogene, 18, Selvakumaran M, Lin HK, Miyashita T, Wang HG, 2892–2900. Krajewski S, Reed JC, Hoffman B and Liebermann D. Zhan Q, Lord KA, Alamo Jr I, Hollander MC, Carrier F, Ron (1994a). Oncogene, 9, 1791–1798. D, Kohn KW, Hoffman B, Liebermann DA and Fornace Jr Selvakumaran M, Lin HK, Sjin RT, Reed JC, Liebermann DA AJ. (1994). Mol. Cell. Biol., 14, 2361–2371. and Hoffman B. (1994b). Mol. Cell. Biol., 14, Zhang W, Bae I, Krishnaraju K, Azam N, Fan W, Smith K, 2352–2360. Hoffman B and Liebermann DA. (1999). Oncogene, 18, Smith ML, Chen IT, Zhan Q, Bae I, Chen CY, Gilmer TM, 4899–4907. Kastan MB, O’Connor PM and Fornace Jr AJ. (1994). Zhang W, Hoffman B and Liebermann DA. (2001). Int. J. Science, 266, 1376–1380. Oncol., 18, 749–757.

Oncogene