Characterization of MyD118, Gadd45, and proliferating cell nuclear antigen (PCNA) interacting domains. PCNA impedes MyD118 AND Gadd45-mediated negative growth control.

MyD118 and Gadd45 are related genes encoding for proteins that play important roles in negative growth control, including growth suppression and apoptosis. MyD118 and Gadd45 are related proteins that previously were shown to interact with proliferating cell nuclear antigen (PCNA), implicated in DNA replication, DNA repair, and cell cycle progression. To establish the role of MyD118 and Gadd45 interactions with PCNA, in this work we sought to identify the interacting domains and analyze the significance of this interaction in negative growth control. Using complementary in vivo and in vitro interaction assays the N-terminal (1-46) and middle (100-127) regions of PCNA were identified as harboring MyD118- and Gadd45 interacting domains, whereas PCNA interacting domains within MyD118 and Gadd45 were localized to the C termini of these proteins (amino acids 114-156 and 137-165, respectively). These findings provide first evidence that similar domains within MyD118 and Gadd45 mediate interactions with PCNA. Importantly, ectopic expression of MyD118 or Gadd45 N-terminal peptides, lacking the PCNA interacting domain, was found to suppress colony formation or induce apoptosis more efficiently than the full-length proteins. These findings suggest that interaction of MyD118 or Gadd45 with PCNA, in essence, serves to impede negative growth control.

The MyD118 gene was originally discovered as a novel myeloid differentiation primary response (MyD) 1 gene, activated in Ml myeloblastic leukemia cells by interleukin-6 upon induction of terminal differentiation associated with growth arrest and apoptosis (1). The MyD118-encoded protein was found to be remarkably similar to the protein encoded by Gadd45, a growth arrest-and DNA damage-induced gene, regulated in part by the p53 tumor suppressor (2,3).
In recent years, several lines of evidence have established the role MyD118 and Gadd45 play in negative growth control, including inhibition of cellular growth and apoptotic cell death. MyD118 was also found to be a primary response gene to the growth inhibitory and apoptotic cytokine TGF-␤, which induces M1 cells for growth arrest and apoptosis uncoupled from differentiation (4). Blocking MyD118 expression in M1 cells by MyD118 antisense RNA was found to compromise TGF-␤-mediated apoptosis. Ectopic expression of bc1-2 in M1 cells, which blocked TGF-␤-induced apoptosis, resulted in reduced levels of MyD118 transcripts. In contrast, deregulated expression of either c-myc or c-myb in M1 cells, which accelerated TGF-␤induced apoptosis, markedly elevated the level of MyD118 transcripts. Taken together, these findings are consistent with MyD118 being a positive modulator of TGF-␤-induced apoptosis (4). Other studies have shown that ectopic expression of MyD118 or Gadd45 in a variety of human tumor cell lines or in NIH3T3 fibroblasts negatively regulated cellular growth resulting in suppression of colony formation (5)(6)(7). In addition, MyD118, Gadd45, and the cyclin-dependent kinase inhibitor p21 were observed to synergize in colony suppression (5,7).
It was shown that MyD118 and Gadd45 encode for nuclear proteins that interact with PCNA, as well as with the cyclindependent kinase inhibitor p21 (6 -8). More recently (upon completion of this study), it became clear that MyD118 and Gadd45 also interact with several other cellular proteins, including the core histones, the stress-inducible MTK1/MEKK4 kinase, and Cdk1 (cdc2-p36) (9 -11). How this multitude of interactions modulates the functions of MyD118 and Gadd45 in negative growth control has not been established.
PCNA, a normal component of multiple quaternary complexes, which include the cyclins, CDKs, and the cyclin-dependent kinase inhibitor p21 WAF1/CIP1 , plays a central role in cell cycle regulation (12)(13)(14). Ample evidence has established the important role of PCNA in various multiprotein complexes that move along the DNA and operate as a communication point and signal-processing center for a variety of cellular processes, including DNA replication, nucleotide excision repair, post replication repair, mismatch repair, base excision repair, and apoptosis via at least one pathway (15)(16)(17).
Given the central role of PCNA in cell proliferation and that it was found to interact with both MyD118 and Gadd45, in this work we sought to dissect the domains that mediate these interactions and analyze their relevance for negative growth control. Having identified the interacting domains, we show that the interaction of MyD118/Gadd45 with PCNA, in essence, impedes the functions of these proteins as negative regulators of cellular growth.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-H1299 and 293T cells were obtained from ATCC. All the cells were cultured at 37°C in Dulbecco's modified Eagle's medium (Cellgro) with 10% fetal bovine serum in a humidified atmosphere of 10% CO 2 (293T were cultured at 5% CO 2 ). All the cells were cultured to give 60 -80% confluency at the time of transfection. One day before transfection H1299 cells were plated at 1.2 million cells/100-mm dish, and 293T cells were plated at 3 million cells/100-mm dish).
Yeast Two-hybrid Analysis-Manipulation of Escherichia coli and DNA was performed according to standard methods (18). Yeast twohybrid analysis was performed with the CLONTECH Matchmaker Two-Hybrid System-2, essentially as described in the CLONTECH protocol. Full-length murine MyD118 cDNA in pBluescript (1), human Gadd45 cDNA in pET-14B (7,19), and human PCNA in p3038-T7hPCNA (20) were used to generate all the deletion constructs of MyD118/Gadd45/PCNA. Deletion mutants were constructed by taking advantage of unique restriction enzyme sites within the coding sequence of these cDNAs. Full-length cDNAs and truncated cDNAs encoding for deletion peptides were cloned in-frame into pAct2 (GAL4 activation domain "trap") as well as into the modified pAS2.1 (GAL4 DNA binding domain "bait") vector. All of the constructs were sequenced to verify in-frame cloning. Transformation into yeast strain Y187 was performed according to CLONTECH's protocol. Yeast strain Y187 was used for testing interaction by activation of the ␤-galactosidase gene (blue selection). Yeast strains CG1945 or Y190 were used to test interaction by activation of the histidine gene. Interaction assays were conducted only with pAct2/pAS2.1 constructs that tested negative for self-activation. Interactions either positive (ϩ) or negative (Ϫ) were ascertained by both ␤-galactosidase expression (blue selection) and histidine expression (histidine selection). Quantitation of relative binding affinities of protein and deletion peptides was determined in yeast strain Y187. Three positive double transfectants were selected on minimal medium plates for each transfection and grown overnight in liquid selection minimal medium. The expression of ␤-galactosidase was quantified, according to the CLONTECH protocol, with o-nitrophenyl ␤-D-galactopyranoside as the substrate, and the intensity of blue color developed was measured colorimetrically at 420 nm.
In Vitro Interaction Assays Using Coupled Transcription/Translation-For in vitro association assays, full-length PCNA, MyD118, Gadd45, and Bax cDNAs, and C-terminal peptide coding sequences were cloned into pET14b (7). N-terminal peptides encoding cDNAs were generated by truncating and linearizing full-length MyD118/Gadd45 cDNAs in pET14b with unique restriction enzymes at the required sites and were then used in the TNT reaction. PCNA, aa 87-261 (EcoRV-NdeI fragment), was cloned in-frame into the pAS2.1 vector. Subsequently, PCNA, aa 87-127 (NdeI-EcoRI fragment from pAS2.1), and PCNA, aa 87-149 (NdeI-BglII fragment from pAS2.1), were excised and cloned in-frame into pET23b (Novagen), which was used in the TNT reaction. Constructs were purified by CsCl gradient centrifugation and suspended in DNase-and RNase-free H 2 O. The T7-based coupled transcription and translation system, with rabbit reticulocyte lysate (TNT, Promega) and Easy Tag [ 35 S]methionine-L (NEN Life Science Products), was used to produce radiolabeled proteins and peptides. In vitro interaction experiments were done as described (7). Briefly, equal amounts of 35 S-labeled PCNA, Bax, and either MyD118 or Gadd45 proteins/peptides were mixed together (10 5 cpm each) in 200 l of interaction buffer A (20 mM Tris (pH 8.3), 150 mM NaCl, 1% Nonidet P-40, 0.1% Tween 20, and 1 mg/ml bovine serum albumin). 35 S-Labeled Bax was added to all reactions as an internal negative control for interaction. Each interaction reaction was carried out, in triplicate, for 30 min on ice. Subsequently each reaction mixture was nonspecifically immunoprecipitated with rabbit IgG or mouse IgG for 1 h. Following pulling down of nonspecific immune complexes with protein-A/G-agarose beads (Oncogene Sciences), supernatants were subjected to coimmunoprecipitation with specific primary antibodies: PCNA (PC10, Santa Cruz), Bax (N 20, Santa Cruz), Gadd45 (H165, Santa Cruz), or MyD118 (7). Immune complexes were pulled down with protein A/Gagarose beads. Beads were washed (4ϫ) with interaction buffer A, suspended in Laemmeli protein loading buffer, heated at 75°C for 10 min, and loaded on 15% SDS-polyacrylamide gel electrophoresis gel for analysis. Following electrophoresis the gels were fixed with 25% isopropyl alcohol in 10% acetic acid, subjected to fluorography with Amplify (Amersham Pharmacia Biotech), and vacuum-dried before exposing to x-ray film.
In Vivo Interaction Assays Using Transient Transfection into 293T or H1299 Cells-For the in vivo interaction studies, the HA tag epitope sequence was inserted into the mammalian expression vector pcDNA3 (Invitrogen). The pAct2 vector of the YTH system (CLONTECH) has a BglII fragment encoding for the HA tag epitope and multiple cloning site. This fragment was excised and cloned into the BamHI site of pcDNA3 to create pcDNA3/HA. Full-length MyD118/Gadd45 cDNAs and cDNAs encoding for N-terminal MyD118/Gadd45 (MyD118/aa 1-40 and Gadd45/aa 1-94) with the HA tag were excised with BglII from the pAct2 vector of the YTH system and cloned into the BamHI site of pcDNA3. cDNAs encoding for C-terminal MyD118/Gadd45 (MyD118/aa 114 -156 and Gadd45/aa 95-165) were excised with the appropriate restriction enzymes from the multiple cloning site of pAS2.1 and cloned in-frame to the HA tag of pcDNA3/HA. 10 g of the pcDNA3 constructs encoding for HA-tagged MyD118/Gadd45, fulllength proteins, or deletion peptides were transiently transfected into 293T or H1299 cells, using LipofectAMINE (Life Technologies, Inc.). After 24 h, transfected cells were trypsinized, washed twice with PBS, and lysed in 3 ml of interaction buffer containing protease inhibitors. Triplicate 1-ml aliquots of cell lysates were subjected to co-IP, as described above for the in vitro association assay, using HA antibodies (Y-11, Santa Cruz), PCNA antibodies, or nonspecific rabbit or mouse IgG. Immunecomplexes were separated on 15% SDS-polyacrylamide gel electrophoresis, and protein bands were transferred to nitrocellulose membrane (Hybond-EC, Amersham Pharmacia Biotech), Western probed with HA antibodies, and visualized by ECL (Amersham Pharmacia Biotech). Membranes were re-probed with PCNA antibodies after stripping the blot according to manufacturer's protocol.
ELISA Binding Assay to Determine Dissociation Constants (K d ) for the Binding of MyD118/Gadd45 to PCNA-A novel ELISA based immunoassay (21) was used to measure the binding affinity of MyD118/ Gadd45 proteins to PCNA. Expression cloning of MyD118, Gadd45, and PCNA into pET14 was described (7,19). These vectors were used to express the proteins in BL21(DE3) E. coli. The expressed proteins were purified from the bacterial extract by passing through a His bind affinity column, as suggested by the manufacturer (Novagen). After removal of the His tag by thrombin cleavage, proteins were again passed through a His bind affinity column to trap the excised histidine tag. ELISA assays were performed with purified proteins in Corning 96-well polystyrene disposable sterile ELISA plates. PCNA protein (50 or 100 ng), in 100 l of interaction buffer (20 mM Tris (pH 8.3), 150 mM NaCl, 0.2% Tween 20), was absorbed to the wells of an ELISA plate by overnight incubation at 4°C. Subsequently, unbound PCNA was removed, and wells were washed three times with interaction buffer. Remaining nonspecific binding capacity of the wells was eliminated by a 1-h incubation with blocking buffer (interaction buffer containing 5% bovine serum albumin). Then the blocking buffer was removed, and the wells were washed with interaction buffer. 2-Fold serially diluted recombinant MyD118 or Gadd45 protein, in 100 l of interaction buffer plus 0.1% bovine serum albumin, was then applied to the wells, and proteins were allowed to interact with plate-bound PCNA for 1 h at room temperature. Following the interaction period, unbound MyD118/ Gadd45 protein was removed, and the wells were washed with interaction buffer. MyD118/Gadd45 antibodies were then applied to the wells for 1 h and unbound antibodies were removed. Next, anti-rabbit IgG coupled with horseradish peroxidase (HRP) (Amersham Pharmacia Biotech), was applied to the wells for 1 h, and unbound HRP-conjugated anti-rabbit IgG was removed. Retained HRP-conjugated anti-rabbit IgG bound to the MyD118/Gadd45 antibody-MyD118/Gadd45 protein-PCNA complex was detected using the chromogenic substrate 3,3Ј,5,5Јtetramethyl benzidine (1-step slow 3,3Ј,5,5Ј-tetramethyl benzidine ELISA, Pierce). Color development was stopped after 15 min by adding 1 M H 2 SO 4 , and the absorbance was measured at 450 nM using a Bio-Rad model 450 microplate reader.
Colony Suppression Analysis-A short term transfection assay was used to assess the ability of full-length MyD118/Gadd45 proteins and N-or C-terminal peptides to suppress colony formation in H1299 cells (5,7). Briefly, H1299 cells were co-transfected by the LipofectAMINE method with 0.25 g of pMSCVpac together with 5 g of empty pcDNA3 construct (negative control) or pcDNA3 encoding for HA-tagged fulllength MyD118, Gadd45 proteins or for N/C-terminal peptides. pM-SCVpac was used for puromycin selection and as an internal normalization control for transfection efficiency (22). Co-transfections were performed in 60-mM tissue culture dishes. One day following transfection, the cells were washed with PBS and trypsinized. The cells were replated in duplicates at 1:15, 1:30, and 1:150 dilutions with puromycin containing medium (3 g/ml) and incubated for 14 days to allow colonies to develop. Then the medium was removed; the colonies were washed once with PBS and fixed with 75% methanol in 25% acetic acid for 5 min, and the plates were dried. Colonies were stained with Lillie's crystal violet (2 g of crystal violet, 0.8 g of ammonium oxalate in 100 ml of 80% ethanol) for 5 min and subsequently washed with de-ionized water to remove excess stain. Stained colonies containing more than 10 cells were scored and counted. The percentage of colony formation was normalized to colonies formed following transfection with empty pcDNA3. pcDNA3 encoding p53 or p21 were co-transfected with pMSCVpac as positive controls. pcDNA3 encoding for antisense p53 was used as an additional negative control.
Apoptosis Analysis-H1299 cells growing on coverslips in 35-mm tissue culture dishes were co-transfected with 2 g of MyD118 or Gadd45 constructs (in pcDNA3) and 0.2 g of ␤-galactosidase expression vector (pSV-␤-galactosidase, Promega) by the LipofectAMINE method. ␤-Galactosidase expression and apoptosis were analyzed 72 h following transfection. The cells were washed once with PBS and fixed for 10 min with 0.05% gluteraldehyde in PBS. Following three washes with PBS to remove the fixative, X-gal solution (20 mM potassium ferricyanide, 20 mM potassium ferrocyanide, 1 mM magnesium sulfate in PBS; X-gal is added to a final concentration of 1 mg/ml just before use, from a 20 mg/ml stock solution in N,N-dimethyl formamide) was spread over the dishes, and the dishes were incubated overnight at room temperature. The next day, the X-gal solution was removed, and cell nuclei were stained for three minutes with 0.1 g/ml Hoechst No.33342 (Sigma catalog number B 2261) and then washed three times with PBS. The coverslips were mounted in Vectashield mounting medium H-1000 (Vector Labs Inc.) and analyzed under a Leitz fluorescent microscope. The percentage of apoptotic cells was determined by dividing the number of ␤-galactosidase-expressing blue cells that exhibit apoptotic nuclear morphology (condensed/fragmented nucleus) by the total number of blue cells. At least 200 cells from five randomly chosen fields were analyzed for each experiment.

RESULTS
Identification of MyD118, Gadd45, and PCNA Interacting Domains Using the Yeast Two-hybrid System-As a first approach to identify PCNA domains that interact with either MyD118 or Gadd45, the yeast two-hybrid system was employed ("Experimental Procedures"). Deletion constructs of PCNA cDNA, encoding for N-terminal, middle, or C-terminal peptides of PCNA, were cloned into the Gal4 DNA binding domain of the pAS2.1 yeast two-hybrid system vector. Each of these constructs were co-transfected into yeast with pACT2 yeast twohybrid system constructs encoding for either MyD118 or Gadd45, fused to the Gal4 activation domain. Interactions, either positive (ϩ) or negative (Ϫ), were ascertained by both ␤-galactosidase expression (blue color) and histidine expression (histidine selection) ("Experimental Procedures"). Results are shown in Fig. 1. The relative binding affinities of different PCNA domains with MyD118/Gadd45 were quantified using the yeast twohybrid system liquid culture assay (see "Experimental Procedures"). In this assay, the level of ␤-galactosidase expression in yeast, under control of the interacting partner proteins encoded by pAct2 and PAS2.1, was quantified by measuring the intensity of blue color development using o-nitrophenyl ␤-D-galactopyranoside as substrate. As shown in Fig. 1, the interaction between full-length PCNA and full-length MyD118/Gadd45 was weaker than the interaction when using PCNA-(1-46), PCNA-(87-127), or PCNA-(224 -261) peptides. This observation is consistent with the notion that full-length PCNA contains, in addition to MyD118/Gadd45 interacting domains, domains which hinder interaction.
Having mapped domains within PCNA that interact with MyD118 and Gadd45, next it was in our interest to map domains within MyD118/Gadd45 that mediate interaction with PCNA. The YTH vector pAS2.1 was used to construct yeast expression vectors that encode for N-and C-terminal deletion mutants of MyD118 and Gadd45. These deletion peptides were tested for interaction with PCNA cloned in the YTH vector pAct2 (Fig. 2 and legend).
The relative binding affinities of full-length and C-terminal peptides of MyD118/Gadd45 with PCNA were quantified, using ␤-galactosidase expression in yeast, and are shown in Fig. 2B. Notably, the binding affinity to PCNA of the MyD118-(114 -156) C-terminal peptide was higher than the Gadd45-(95-165) C-terminal peptide. Taken together these findings indicate that similar domains mediate interaction between either MyD118 or Gadd45 to PCNA.
Analysis of MyD118/Gadd45 and PCNA Interactions in Vitro-To determine if the interacting domains identified by the YTH approach are capable of mediating direct association between these proteins in vitro, binding assays were performed with 35 S-labeled PCNA, MyD118, and Gadd45 full-length proteins and deletion peptides that were generated by coupled transcription/translation. The murine Bax protein (pK I ϭ 4.69), a bcl-2 partner protein with a mass and charge similar to MyD118 (pK I ϭ 3.92) and Gadd45 (pK I ϭ 4.08) (24), was included as an internal control to monitor for the specificity of the binding assay. Equal amounts of 35 S-labeled proteins were mixed, and following incubation, the protein mixtures were immunoprecipitated with antibodies specific to PCNA, MyD118, Gadd45, or Bax. Because the PCNA monoclonal antibody used (PC10) was specific for amino acids 111-120 of PCNA, it could be used only to test for interaction of the middle domain of PCNA with MyD118/Gadd45. Thus, MyD118-and Gadd45-specific antibodies were used to test for interactions with other PCNA domains.
As shown in Fig. 3A, MyD118 and Gadd45 were contained within PCNA immune complexes, and PCNA was contained within MyD118 and Gadd45 immune complexes. The specificity of these interactions was inferred by the observation that Bax was not contained within PCNA, MyD118, or Gadd45 immune complexes, consistent with previous observations (7). As shown in Fig. 3, B-D, immune complexes of peptides corresponding to PCNA N-terminal (aa 1-46) and middle (aa 87-127), but not the C-terminal (aa 224 -261) domains, also contained MyD118 or Gadd45. This indicated that the middle and N-terminal regions of PCNA contain domains that were capable of association with both MyD118 and Gadd45 in vitro.
Coupled transcription/translation also was used to test the C-terminal peptides of MyD118 and Gadd45 (aa 114 -156 and 95-165, respectively) for their ability to associate with PCNA in vitro. As shown in Fig. 4, A and B 2. Identification of domains within MyD118/Gadd45 that interact with PCNA using the yeast two-hybrid system. Schematic diagram of fulllength MyD118 and MyD118 deletion peptides tested for interaction with PCNA in the yeast two-hybrid system ("Experimental Procedures"). Positive (ϩ) and negative (Ϫ) interactions were ascertained by both ␤-galactosidase expression (blue selection) and histidine expression (histidine selection), as described under "Experimental Procedures." Interaction assays were conducted with MyD118/ Gadd45 in pAS2.1 and PCNA in pAct2, except for constructs marked with an asterisk, where MyD118 was in pAct2 and was tested with PCNA in pAS2.1 (because of self-activation of reciprocal constructs). In these cases interactions were ascertained by histidine selection with both reciprocal combinations of YTH vectors, which in this assay tested negative for self-activation. Numbers in parentheses indicate relative binding affinities and standard deviations determined by using the yeast two-hybrid liquid culture assays, as indicated under "Experimental Procedures." immune complexes contained PCNA. Again, the specificity of these associations was indicated by the observation that PCNA immune complexes did not contain peptides corresponding to N-terminal MyD118/aa 1-113 and N-terminal Gadd45/aa 1-94 regions, respectively, whereas N-terminal MyD118 or N-terminal Gadd45 immune complexes did not pull down PCNA (Fig. 4,  A and B). In conclusion, in vitro binding assays have shown that domains contained within the N-terminal (aa 1-46) and middle regions (aa 87-127) of PCNA are capable of associating in vitro with domains contained within C-terminal regions of MyD118 (aa 114 -156) and Gadd45 (aa 95-165).
To determine the overall binding affinity of MyD118 and Gadd45 to PCNA, an ELISA-based immunoassay (21) was used, where PCNA was coated onto the wells of ELISA plates and allowed to interact with 2-fold diluted MyD118/Gadd45 protein. After extensive washing, the amount of MyD118/ Gadd45 protein bound to PCNA was assessed using either MyD118-or Gadd45-specific antibodies, which, in turn, were reacted with HRP-coupled secondary antibody. Bound HRP was detected and quantitated using the chromogenic substrate TMB (where color development was directly proportional to MyD118 or Gadd45 bound to PCNA (see "Experimental Procedures")) and was used to generate the binding curves (Fig. 5, A1 and A2). K d values for MyD118 and Gadd45 binding to PCNA were derived by Scatchard plot analysis of the binding curves (Fig. 5, B1 and B2).
The two slopes of the Scatchard plots indicated that binding of either Gadd45 or MyD118 to PCNA has biphasic kinetics. This predicts two binding sites on PCNA for each MyD118 and Gadd45, with different affinities. The calculated K d values for MyD118 binding to PCNA were 1.6 ϫ 10 Ϫ9 M (moderate affinity) and 10 ϫ 10 Ϫ9 M (low affinity). For Gadd45 binding to PCNA the K d values were 1.2 ϫ 10 Ϫ9 M (moderate affinity) and 0.25 ϫ 10 Ϫ9 M (high affinity).
Dissection of MyD118/Gadd45 and PCNA Interactions in Mammalian Cells-The work described so far has identified domains within MyD118 and Gadd45, which mediate interaction between these proteins and PCNA in yeast cells and in vitro. To test the biological significance of MyD118/Gadd45 interaction with PCNA in negative growth control, it was important to establish that these domains similarly mediate interaction in mammalian cells. Toward this end, pcDNA 3 expression vectors encoding for full-length MyD118, Gadd45 (or Bax, as negative control), and deletion peptides, all with an HA tag, were transiently transfected into 293T cells, and 24 h later cell extracts were tested for interaction of the encoded proteins or peptides with endogenous PCNA by co-immunoprecipitation with either PCNA or HA antibody.
The results of the in vivo interaction assays are shown in Fig.  6. It can be seen that PCNA immune complexes, obtained from 293T cell extracts transfected with HA-MyD118/pcDNA3, contained HA-MyD118. Reciprocally, HA tag immune complexes obtained from the same cells contained PCNA (Fig. 6). This is consistent with our previous observations indicating that MyD118 and PCNA interact in M1 and NIH3T3 cells following induction of MyD118 by TGF-␤ or methyl methanesulfonate (7). Transfection of 293T with pcDNA3 encoding for either C-terminal or N-terminal HA-MyD118/HA-Gadd45 peptides has demonstrated that PCNA immune complexes contained C-terminal but not N-terminal MyD118/Gadd45 peptides. Likewise, HA immunocomplexes brought down PCNA from cells transfected with pcDNA3 encoding for C-terminal but not N-terminal MyD118/Gadd45. The specificity of the interactions in 293T cells was further demonstrated by the observation that PCNA immune complexes obtained from 293T cells transiently transfected with HA-Bax/pcDNA3 did not contain Bax, and HA immune complexes did not contain PCNA. Taken together these results indicate that MyD118 and Gadd45 interact with PCNA in mammalian cells via their C-terminal regions.
Interaction with PCNA Impedes MyD118/Gadd45-mediated Colony Suppression of Human Tumor Cells-Previously, using a short term transfection assay (5-7), we have shown that MyD118 and Gadd45 suppress colony formation of NIH3T3 and H1299 cells. Having identified the PCNA interacting domains within MyD118 and Gadd45, clearly it was of interest to explore the significance of MyD118 and Gadd45 interactions with PCNA for negative growth control using the colony suppression assay.
To accomplish this, a puromycin selection plasmid (pMSCV pac) together with pcDNA3 encoding for HA-MyD118 or HA-Gadd45 deletion peptides or empty pcDNA3 control plasmid at a 1:20 ratio was transfected into H1299 cells. Following 2 weeks of selection in puromycin-containing medium surviving colonies were fixed, stained, and scored. In all the experiments empty vector was used as negative control, and p53 and its target gene p21 were used as positive controls. The results of the short term colony suppression assays are shown in Fig. 7. Representative tissue culture plates that were fixed, stained, and scored are shown in Fig. 7A. Results of all experiments are summarized in Fig. 7C, where colony formation using empty FIG. 5. Determination of dissociation constants (K d ) for MyD118 and Gadd45 interaction with PCNA. A1 and A2, binding curves of MyD118 and Gadd45 to PCNA. Binding curves were obtained using the ELISA binding assay as described under "Experimental Procedures." Each data point on the curve represents the least mean square values obtained from three independent experiments done in duplicate. The MyD118/ Gadd45 binding curves were obtained using 50 and 100 ng of PCNA coated onto the wells of the ELISA plates. B1 and B2, Scatchard plot analysis and determination of K d values for MyD118 and Gadd45 binding to PCNA. Binding curves were used for Scatchard analysis, where the slope was used to calculate the inverse of the dissociation constant (K d ) values. The data points of the Scatchard curves grouped on two slopes, predicting at least two binding sites on PCNA for MyD118 or Gadd45, with relatively high and low affinities. MyD118/Gadd45 bound to PCNA were computed using the equation A ϭ ␣[JF] ϩ ␤ applied to the binding curves where A is the measured ELISA signal; ␣ is the effective molar absorptivity that was calculated by nonlinear regression analysis (Cricket Graph III) from the binding curves; and [JF] is the molar quantity of the PCNA-MyD118⅐Gadd45 complex, measured from the ELISA binding curve, and from the total amount of MyD118/Gadd45 applied to each well; ␤ is a term that takes into account loss of signal during washing, so as not to affect the shape of the binding curve and not to alter the calculated K d value. Thus, it was not necessary to know a priori the amount of plate-bound PCNA to estimate the K d of the binding affinity of it's partners. As indicated by the Scatchard plots, the slope of MyD118/Gadd45 interaction with PCNA did not change in both 50 and 100 ng of PCNA coating experiments. Only a shift in the positions of the curve was noticeable, reflecting reduced number of binding sites with 50 ng of PCNA coating and demonstrating the validity of this approach. pcDNA3 vector is the standard for no colony suppression and is, therefore, set at 100%. Data used for calculating colony suppression has been derived only from experiments where the initial expression of the various proteins and peptides did not vary significantly, as shown in Fig. 7B. Association of fulllength MyD118 and Gadd45 proteins and deletion peptides with endogenous PCNA in H1299 cells, analyzed 24 h following transfection by immunoprecipitation and Western blotting, was similar to what is shown in Fig. 6 (data not shown).
As seen in Fig. 7, p53 suppressed colony formation by about 85%, whereas p21, MyD118, and Gadd45 inhibited colony formation by 40 -60%. Interestingly, the C-terminal MyD118 and Gadd45 peptides, which harbor the PCNA interacting domain, suppressed colony formation significantly less efficiently than the full-length proteins (42 and 20%, respectively), whereas the non-PCNA interacting N-terminal MyD118/Gadd45 peptides suppressed colony formation much more efficiently than the full-length proteins (90 and 80%, respectively). Taken together, these findings suggest that interaction of MyD118 and Gadd45 with PCNA serves to impede the function of these proteins in negative growth control.
Interaction with PCNA Impedes MyD118/Gadd45-mediated Apoptotic Cell Death-To further examine the role of MyD118/ Gadd45 interaction with PCNA in H1299 colony suppression and determine whether it was because of apoptotic cell death, H1299 cells growing on coverslips were co-transfected with a limiting amount (0.2 g) of ␤-galactosidase expression plasmid together with excess (2 g) pcDNA3 encoding for full-length or MyD118/Gadd45 deletion peptides. Following 72 h the transfected cells were analyzed for ␤-galactosidase expression (blue color), and nuclei of the transfected cells were stained with Hoechst to detect apoptotic morphology characterized by nuclear shrinkage and chromatin condensation (Fig. 8). The percentage of apoptotic cells was determined as the number of ␤-galactosidase-expressing blue cells that exhibited apoptotic nuclear morphology divided by the total number of ␤-galactosidase-expressing blue cells.
As shown in Fig. 8, ectopic expression of either MyD118 or Gadd 45 induced apoptosis in H1299 cells. Furthermore, as observed for H1299 colony suppression, C-terminal MyD118 or Gadd45 peptides, which harbor the PCNA interacting domain, induced apoptosis appreciably less efficiently than the fulllength proteins (34 and Ͻ5% compared with 65 and 40%, respectively). In contrast, induction of apoptosis by N-terminal MyD118 or Gadd45 peptides, which do not interact with PCNA, was significantly more efficient than what was observed with the full-length proteins (80 and 61% compared with 65 and 40%, respectively).
Taken together, these findings suggest that apoptotic cell death is a major factor that contributes to the apparent ability of MyD118 and Gadd45 to suppress colony formation of tumor cells. Interestingly, these findings also extend the notion that interaction of MyD118 and Gadd45 with PCNA impacts negatively on the ability of these proteins to induce cellular death. DISCUSSION PCNA, a normal component of multiple quaternary complexes, which includes the cyclins, CDKs, and the cyclin-dependent kinase inhibitor p21 WAF1/CIP1 (25-27), plays a pivotal role in cell cycle regulation (13-14, 25), DNA replication (28), and repair of damaged DNA (29). Given the central role of PCNA in cell proliferation and that it was found to interact with both MyD118 and Gadd45, we sought to dissect the domains that mediate these interactions and to analyze the significance of these interactions for negative growth control. To unequivocally identify amino acid regions that mediate interactions between either MyD118 or Gadd45 and PCNA, several complementary in vivo and in vitro methods were used, including the yeast two-hybrid system, in vitro transcription/translation, and transient expression in 293T cells.
The YTH approach has led to the identification of three MyD118/Gadd45 interacting domains within PCNA that have been localized to the 1-46 aa N-terminal, 87-127 aa middle, and 224 -261 aa C-terminal regions of PCNA. Evidence also was obtained for regions within PCNA, i.e. aa 47-86 and 150 -223, that compromised the binding of these PCNA domains to MyD118/GAdd45 (Fig. 1). It is possible that such regions may play a physiological role in modulating PCNA interactions with MyD118 and Gadd45 in response to or following exposure of cells to genotoxic stimuli. Only the N-terminal and the middle FIG. 6. Analysis of MyD118 and Gadd45 interactions with PCNA in 293T cells. Transfection of pcDNA3 expression vectors encoding for HA-tagged MyD118/deletion peptides (A) or HAtagged Gadd45/deletion peptides (B) into 293T cells was performed as described under "Experimental Procedures." Cells were collected 24 h following transfections, whole cell extracts were prepared, subjected to co-IP with the indicated antibodies, and Western blot analysis was done with the indicated antibody probes. Empty HA-pCDNA3 or HA-Bax vector were transfected as negative controls.
domains of PCNA were found to associate with MyD118 and Gadd45 in vitro. Thus, the apparent interaction of MyD118 and Gadd45 with C-terminal PCNA in yeast may reflect indirect interaction. The C terminus of PCNA contains the PCNA⅐PCNA interacting domain (30). It is possible that following co-transfection of YTH vectors into yeast, the C-terminal region of PCNA encoded by one YTH vector interacts with endogenous PCNA, which is associated with MyD118 or Gadd45 encoded by the other YTH vector. Another possibility is that, unlike in vitro conditions, under in vivo conditions the C-terminal region of PCNA may assume a tertiary structure and/or may be post-translationally modified to facilitate interaction with MyD118/Gadd45. Whatever may be the case, these findings provide first evidence that similar domains within PCNA mediate interactions with both MyD118 and Gadd45.
Our results regarding PCNA domains that interact with Gadd45 are consistent with the findings of Hall et al. (23). Initial information regarding the strength of interactions between PCNA and MyD118/Gadd45 was obtained using the yeast two-hybrid approach in a semiquantitative assay. Based on this assay, evidence was obtained that the middle domain of PCNA (aa 100 -127) has similar affinities for MyD118 and FIG. 7. Colony suppression assay for MyD118 and Gadd45 deletion peptides. Short term transient transfection assays of the indicated expression vectors into H1299 cells were performed as described under "Experimental Procedures." Empty pcDNA3 was used as negative control, whereas pcDNA3 encoding for p53 and p21 were used as positive controls. One set of representative results is shown in A. Expression of full-length MyD118/Gadd45 and deletion peptides in 293T cells, determined by immunoprecipitation and Western probing, 24 h after transfection, is shown in B. Results of three independent experiments along with standard deviations are shown in C. The percentage of colony formation is based on results using empty vector, which is designated 100%.
Gadd45 (1.2 units), whereas the N-terminal (1-46) domain displayed a lower binding affinity for MyD118 than for Gadd45 (1.67 versus 2.66 units, respectively). An ELISA-based immunoassay (21) was used to measure the binding affinity (K d ) of MyD118 and Gadd45 to PCNA. The biphasic kinetics of binding that were obtained predict two binding sites on PCNA for MyD118 and Gadd45 that vary in their affinities (low and moderate). Thus, the stoichiometry of binding may vary from one up to six molecules of MyD118 and/or Gadd45 to trimeric PCNA (31,32).
Recent work has established a central role for trimeric PCNA as a moving platform on DNA that is a communication point and signal processing center for a variety of cellular processes, including DNA replication, nucleotide excision repair, post replication mismatch repair, base excision repair, and at least one apoptotic pathway (15)(16)(17). Central to these roles of PCNA is the interaction with the RNA polymerase accessory factor (TFIIH) complex, (33)(34)(35) as well as its interactions with DNA polymerase ␦, Fen1, DNA ligase I, and p21, via the interdomain connector loop (aa 100 -150) (36 -39). It is clear, therefore, that these later proteins compete for accessibility to bind to the same domains of PCNA, which have been identified to interact with MyD118, Gadd45, and p21 (23, 37, 40 -42). This picture is further complicated by the ability of MyD118 and Gadd45 to interact with each other as well as with p21 2 (7). The biological significance of this intricate myriad of interactions for the role MyD118 and Gadd45 play in negative growth control and for the functions of PCNA in DNA replication and repair remains to be elucidated.
Mapping MyD118 and Gadd45 regions that interact with PCNA has shown that both proteins interact with PCNA through their C termini (aa 114 -156 and 137-165, respectively). Importantly, using MyD118/Gadd45 deletion peptides in colony suppression and apoptosis assays has provided evidence that interaction with PCNA, in essence, impedes the function of these proteins in negative growth control. MyD118 and Gadd45 C-terminal peptides, which harbor the PCNA interacting domain, were observed to suppress colony formation and induce apoptosis less efficiently than the full-length proteins, whereas the non-PCNA interacting N-terminal peptides suppressed colony formation and induced apoptosis much more efficiently than the full-length proteins. Understanding the mechanism by which MyD118/Gadd45 exert their function and how PCNA may modulate it, should explain how the C-terminal peptides of these proteins still retain residual negative growth control function. Also, given the evidence that MyD118 and Gadd45 can stimulate DNA repair in vitro (6 -7), clearly also of interest will be to determine the role MyD118/Gadd45 interaction with PCNA plays in PCNA-mediated DNA repair.
While this work was in progress, a third member in the MyD118 and Gadd45 family, termed CR6, has been identified (43). Evidence was obtained also that CR6, like MyD118 and GAdd45, interacts with PCNA and p21, 3 as well as with the stress-inducible MTK1/MEKK4 kinase (9). Localization of the interacting domains should determine to what extent CR6 interactions with PCNA are similar to what was observed for MyD118 and Gadd45 and whether PCNA may also impede CR6-mediated negative growth control.
Recently, upon completion of this study, it became evident that MyD118, Gadd45, and CR6, in addition to interacting with PCNA and p21, also interact with several other cellular proteins. It was reported that all three proteins (termed Gadd45␣, Gadd45␤, and Gadd45␥) interact with and activate the stressresponsive MTK1/MEKK4 kinase that mediates activation of the p38/JNK kinase pathway and apoptosis in response to environmental stress stimuli (9). Evidence was obtained that Gadd45 also associates with CDK1 (cdc2-p34) and inhibits the kinase activity of the CDK1⅐cyclin B1 complex, implicated in Gadd45 induction of a G 2 /M cell cycle check point in response to certain genotoxic stresses (10,44). In addition, it was documented that Gadd45 can recognize damaged chromatin and modify chromatin accessibility by direct interaction with the core histones (11).
Using a battery of genetically engineered domain-deficient mutant MyD118 and Gadd45 proteins, work is underway to determine what role interactions of MyD118/Gadd45/CR6 with any one or a combination of these other proteins may play in negative growth control and how the interaction with PCNA may modulate these functions. FIG. 8. Analysis of apoptosis induced by MyD118 or Gadd45 deletion peptides. H1299 cells growing on coverslips were co-transfected with 2 g of MyD118 or Gadd45 constructs (in pcDNA3) and 0.2 g of ␤-galactosidase expression vector using the LipofectAMINE method. ␤-Galactosidase expression and apoptosis were analyzed 72 h following transfection, as described under "Experimental Procedures." The percentage of apoptotic cells was determined by dividing the number of ␤-galactosidase-expressing blue cells that exhibit apoptotic nuclear morphology (condensed/fragmented nucleus) by the total number of blue cells. At least 200 cells from five randomly chosen fields were analyzed. Note: in H1299 cells nuclear apoptotic morphology was observed to precede membrane blebbing and cellular shrinkage.