Direct Interaction between Nucleosome Assembly Protein 1 and the Papillomavirus E2 Proteins Involved in Activation of Transcription

Plasmid constructions.

All yeast plasmids were shuttle vectors, and cloning was performed with Escherichia coli strain XL1Blue. The yeast reporter construct expressing the lacZ gene under the control of the synthetic promoter containing four E2 binding sites in front of the TATA box of human PV (HPV) type 18 (HPV18) P₁₀₅ was obtained in several steps. The SalI-BglII promoter fragment was isolated from construct p4E2₂₃T₁₀₅ (24), ligated to an oligonucleotide encoding the junction between the promoter and the lacZ gene and followed by a BspEI restriction site, and cloned into the SalI-BspEI restriction site of pBR322. Subsequently, this promoter fragment was inserted into the XhoI-BspEI restriction sites of pΔUAS (34), resulting in the displacement of the CYC promoter by the 4E2₂₃P₁₀₅ promoter. Finally, the HindIII-BamHI promoter fragment isolated from this construct was inserted into the HindIII-BamHI vector fragment of placZi (Clontech) to give rise to 4E2₂₃P₁₀₅lacZi. 5E2-pHIS_min was constructed by inserting oligonucleotides encoding E2 binding sites into the XbaI binding site of pHIS-i1 (Clontech). The yeast E2 expression vector pGBT9-E2 was obtained by removing the GAL4 DNA binding domain (DBD) from pGBT9 (Clontech) by digestion with BamHI and partial digestion with HindIII. The E2-encoding HindIII-BamHI fragment was isolated from pTZE2mHIII (34) and cloned into pGBT9. The HindIII-BamHI fragment derived from pTZE2Δ192- 282 (42) was inserted in a similar way to obtain pGBT9-E2Δ192- 282.

To express the AD of GAL4 fused to the DBD of E2, the StuI-BamHI fragment of E2 derived from pTZE2mHIII (42) was cloned into vector pGAD424 (Clontech). The full-length hNAP-1 clone was isolated from the keratinocyte cell line HaCaT cDNA library (expressing cDNA-encoded proteins; Clontech) by amplification with specific primers and cloned in frame with the hemagglutinin (HA) tag into vectors pXJ41 (68) and pcDNA3.1 (Invitrogen). Plasmids expressing a glutathione S-transferase (GST)- hNAP-1 fusion protein were obtained by cloning PCR products encoding full-length hNAP-1 into pGEX-5X-2 and pGEX-2T (Pharmacia Biotech) or deletion mutants into pGEX-2T. Some of the hNAP-1 deletion constructs were subsequently transferred into eukaryotic vector pXJ41 to express them fused to the HA tag. To express the E2 proteins of BPV1, HPV8, and HPV18 fused to a FLAG tag, the corresponding open reading frames were amplified by PCR, followed by cloning into vector pCMV2-FLAG (Kodak), as were BPV1 E2 lacking the AD and E2 expressing the AD. Expression vector pC59, expressing BPV1 E2 under the control of the simian virus 40 (SV40) early promoter, was described previously (69). Expression vectors for HPV8 E2 and HPV18 E2 and the HPV8 noncoding region-luciferase reporter construct also were described previously (49), as was the expression vector for p300 (13). The regulatory region of BPV1 from nucleotides 6958 to 475 (long control region [LCR]) was amplified by PCR and cloned into pALuc (12) to obtain BPV1 LCR-Luc. p53 expression vector pCp53wt was described by Nigro et al. (54), and the synthetic p53-responsive reporter construct was described by Funk et al. (17). The pG5-Luc reporter construct and the vector expressing the DBD of GAL4 (pM) were derived from a mammalian two-hybrid system (Clontech). PCR products encoding the AD of TEF-1 or HPV8 E2 were cloned into vector pM.

Yeast two-hybrid system.

Yeast maintenance, transformation, and storage were performed according to the instructions of the manufacturer of the yeast two-hybrid system (Clontech). Yeast strain YM954 (67) (genotype: MATa ura3-52 his3-200 ade2-101 lys2-801 leu2-3,112 trp1-901 gal4-542 gal80-538) was kindly provided by P. Bartel, Stony Brook, N.Y. β-Galactosidase (β-Gal) activity was determined by a filter lift assay and liquid β-Gal assays according to protocols from Clontech.

Cell culture and transfection.

RTS3b, a keratinocyte cell line established from an HPV-negative skin lesion from a renal transplant recipient (57), was cultivated in E medium (43). For transient transfection, 10⁵ RTS3b cells were used to seed one six-well plate on the day before transfection. Transfection was performed with FuGene reagent according to the manufacturer's recommendations (Roche Diagnostics), and the cells were harvested 48 h later. Luciferase activities were determined as described previously (49). To generate a stable cell line expressing the luciferase reporter gene under the control of the BPV1 LCR integrated into the cellular genome, RTS3b cells were transfected with the BPV1 LCR-Luc construct. After 48 h, the cells were placed in medium containing 800 μg of G418/ml. After another 48 h, the medium was replaced with medium containing 400 μg of G418/ml; this medium was changed every second day for 2 to 3 weeks until resistant colonies had grown. The resistant colonies then were pooled. The presence of the intact BPV1 LCR-Luc cassette within the cells was confirmed by PCR (data not shown). At 24 h after transfection of this stable cell line with the E2 or hNAP-1 expression plasmids, the medium was replaced with medium supplemented with 400 μg of trichostatin A (TSA)/ml; incubation was continued for another 24 h before luciferase activities were determined as described above. To detect the effect of hNAP-1 on the activation of p21 expression by p53, RTS3b cells were transiently transfected with the corresponding expression vectors (see Fig. 6). A total of 90 μg of total cell extract was loaded onto a sodium dodecyl sulfate-15% polyacrylamide gel and analyzed for p21 by Western blotting with an antibody directed against p21 (Pharmingen).

Coimmunoprecipitation and pull-down assays.

GST pull-down assays were performed as previously described (49). To detect a direct protein-protein interaction, His-tagged proteins were expressed in bacteria, purified as described previously (25), incubated with immobilized GST- hNAP-1 fusion protein or GST, and further treated as described elsewhere (49). To detect an interaction between various E2 proteins and hNAP-1 in vivo, 293T cells were transfected with an expression vector for FLAG-tagged BPV1 E2 protein (see Fig. 2E) or with vectors for various FLAG-tagged E2 proteins (see Fig. 2D) and for HA-tagged hNAP-1. After 48 h, the cells were washed in ice-cold phosphate-buffered saline, followed by four freeze-thaw cycles in LSDB buffer (50 mM Tris-HCl [pH 7.9], 10% glycerol, 0.5 mM EDTA, 1 mM dithiothreitol, 0.2% NP-40) containing 100 mM KCl. Cell debris was removed by centrifugation for 10 min at 4°C. The supernatants were incubated with 20 μl of anti-FLAG M2 affinity resin (Kodak). The samples were incubated at 4°C for 2 h with gentle mixing, followed by four washes in LSDB buffer with 200 mM KCl or 250 mM KCl and one wash in LSDB buffer with 100 mM KCl. The presence of hNAP-1 was analyzed by Western blotting with a monoclonal antibody directed against the HA epitope (Roche Diagnostics) or with a monoclonal antibody directed against hNAP-1 (kindly provided by Y. Ishimi) to detect an interaction between E2 and endogenous hNAP-1. E2 proteins were detected with anti-FLAG M5 antibody. To detect an interaction between endogenous p53 and endogenous hNAP-1 in vivo, 85 μg of nuclear extracts from normal neonatal human epidermal keratinocytes (purchased from Clonetics) and 85 μg of nuclear extracts from RTS3b cells were incubated with 5 μl of anti-p53-agarose conjugate (DO-1; Santa Cruz) at 4°C for 2 h with gentle mixing, followed by four washes in LSDB buffer with 200 mM KCl and one wash in LSDB buffer with 100 mM KCl. The presence of the endogenous proteins was analyzed by Western blotting with antibodies directed against hNAP-1 and p53 (DO-1).

Density gradient analysis of the ternary complex.

hNAP-1, E2 with a deletion of amino acids 326 to 420 (E2Δ326-420), and a fragment of p300 from amino acids 1195 to 1761 (p300 1195 to 1761) were expressed with a tag of six histidines in bacteria and purified. Totals of 800 ng of p300 1195 to 1761, 600 ng of hNAP-1, and 200 ng of E2Δ326-420 were incubated for 2 h in the presence of 25 mM HEPES-KOH (pH 7.9)-0.1 mM EDTA-12.5 mM KCl-10% glycerol-100 mM KCl in various combinations (see Fig. 5C). Next, the proteins were loaded onto a 2-ml 7.5 to 30% glycerol gradient prepared as described by Tanese et al. (61) and then centrifuged for 8 h at 50,000 rpm with a TL100 tabletop ultracentrifuge (Beckman) and a TLS-55 swinging-bucket rotor. Finally, 36 fractions of 60 μl each were carefully removed beginning from the top of the gradients. The presence of various proteins was analyzed by Western blotting.

Identification of cellular factors binding to PV E2 proteins by the yeast two-hybrid system.

To identify cellular factors which may play a role in E2-mediated activation of transcription, we set up a yeast two-hybrid system. Since we wished to screen with wild-type E2 protein, which is a strong activator of transcription in yeast cells (34), we modified the classical yeast two-hybrid system. Previously, Ham et al. showed that in mammalian cells, E2 could not efficiently activate the transcription of a minimal promoter composed of E2 binding sites and a TATA box only but was an efficient activator of more complex promoters (24). In contrast, most cellular sequence-specific activator proteins, among them GAL4, are able to stimulate such a minimal promoter (66). To test whether E2 would be unable to activate transcription from this minimal promoter in yeast cells, we inserted a fragment expressing such a minimal promoter into vector placZi (Fig. 1). E2 yielded a three- to fourfold activation of the expression of the lacZ gene from reporter construct 4E2₂₃P₁₀₅placZi integrated into the cellular genome. A fusion protein composed of the DBD of E2 and the AD of GAL4 led to a 14-fold activation (Fig. 1B), indicating that the mode of activation mediated by the two ADs is distinct in yeast cells also. Furthermore, E2 only weakly activated the HIS_min promoter located behind five E2 binding sites and controlling the expression of the HIS gene in plasmid 5E2-pHIS_min (Fig. 1A). Therefore, plasmid 5E2-pHIS_min was used to select for yeast growth in the presence of 100 to 300 mM 3-amino-1,2,4-triazole, an inhibitor of the HIS gene product. It was necessary to include this additional selection to suppress yeast cell growth mediated by the weak activation of pHIS_min by E2.

A cDNA library derived from keratinocyte cell line HaCaT and allowing the expression of cDNA-encoded proteins fused to the GAL4 AD was transformed into the yeast strain harboring the two reporter genes in addition to expression vector pGBT9-E2. Out of 10⁷ yeast clones, the two reporter genes were activated significantly in 41 clones. The cDNA-encoded proteins of two independent clones were identical to the 300 C-terminal amino acids of hNAP-1, which has 391 amino acids in total. The interaction between hNAP-1 and E2 could be confirmed with yeast strains which were retransformed with the corresponding plasmids (Fig. 1C). The interaction involved the N-terminal AD of BPV1 E2, since the C-terminal 94 amino acids of BPV1 E2, containing the DBD, did not reveal increased β-Gal activity in the presence of hNAP-1 (data not shown). However, E2Δ195-282, which lacks the internal hinge region, retained similar cooperativity with hNAP-1 (Fig. 1C). The reduced absolute levels of activation of this deletion mutant correlated with reduced protein levels (data not shown).

E2 proteins bind to hNAP-1 in vitro and in vivo.

To confirm an interaction between E2 and h-NAP-1 in vitro, the hNAP-1 fragment isolated by the yeast two-hybrid system and encoding hNAP-1 lacking the N-terminal 91 amino acids was inserted into vector pGEX-5X-2. A GST-hNAP-1 fusion protein retained the full-length BPV1 E2 protein obtained by in vitro translation with a rabbit reticulocyte lysate (Fig. 2A). The E2 proteins of different PV types are rather conserved within the N-terminal AD and the C-terminal DBD, in contrast to the variable hinge region, which was suggested to function as a linker between the AD and the DBD of E2 (20, 42). In vitro-translated E2 proteins of HPV8 and HPV18 bound to GST- hNAP-1 (Fig. 2A), indicating that the interaction with hNAP-1 may be conserved among E2 proteins of different PV types.

In correlation with the data obtained with yeast cells, the AD of BPV1 E2 is required for binding to hNAP-1 in vitro. E2Δ1-161, which lacks the N-terminal 161 amino acids and which is a naturally occurring N-terminally truncated mutant that represses activation by the full-length E2 protein (35), still interacted with hNAP-1; in contrast, E2Δ1-203, which lacks the entire AD, did not bind at all. An E2 protein which lacks the C-terminal DBD, E2Δ386-420, retained binding, as shown in Fig. 2B. The various E2 proteins average 30% amino acid identity within the AD. Many of the conserved residues throughout the AD are important for both replication and transcription. However, two amino acid substitutions clearly separated these two capacities. Changing Ile at position 73 to Ala (I73A) destroyed transcriptional activation while leaving replication function intact, whereas replacing Glu at position 39 with Ala (E39A) had the inverse phenotype (14). Neither amino acid change affected the binding of BPV1 E2 to GST-hNAP-1, as shown in Fig. 2B.

In order to exclude the possibility that the interaction between BPV1 E2 and hNAP-1 is mediated by a factor present within the reticulocyte lysate or in yeast cells, we incubated GST-hNAP-1 with His-tagged, bacterially expressed, purified full-length BPV1 E2 or E2Δ386-420. As shown in Fig. 2C, both E2 and E2Δ386-420 were specifically retained by GST-hNAP-1, indicating that the interaction is direct. In this experiment, the binding of full-length E2 was weaker than that of E2Δ386-420. This result may have been due to the lower concentration of full-length E2. We were not able to purify full-length E2 in the same amounts as E2Δ386-420. Furthermore, a faster-migrating product, which appears to be an N-terminal proteolytic form, since it retains the His tag, showed a stronger interaction with hNAP-1, supporting the notion that the interaction is mediated by the AD (Fig. 2C).

E2 and hNAP-1 also interact within the cell, as shown in Fig. 2D. 293T cells were transiently transfected with expression vectors for FLAG-tagged full-length BPV1 E2, an E2 protein that lacks the N-terminal AD (FLAG-E2Δ1-203), or the AD of BPV1 E2 (FLAG-E2Δ204-420) and an expression vector for HA-tagged full-length hNAP-1. Whereas in the presence of full-length BPV1 E2 or its AD, HA-tagged hNAP-1 could be coprecipitated by the FLAG antibody, E2Δ1-203 did not mediate this effect, confirming that the interaction within the cell is mediated by the AD of BPV1 E2 as well. The E2 proteins of HPV8 and HPV18 also were able to coprecipitate HA-tagged hNAP-1 (Fig. 2D).

Furthermore, E2 binds to endogenous hNAP-1. In all cervical carcinoma cell lines infected with PV, the HPV DNA is integrated into the cellular genome, resulting in the disruption of the open reading frame for E2. Therefore, cell lines expressing E2 at physiological levels during HPV infection do not exist, and to analyze the binding of E2 to endogenous hNAP-1, we overexpressed FLAG-tagged E2 in 293T cells. As shown in Fig. 2E, the FLAG antibody was able to precipitate endogenous hNAP-1 from extracts of cells transfected with the vector for FLAG-tagged BPV1 E2 and not from extracts of cells transfected with the empty expression vector or with a vector expressing a control protein, bacterial alkaline phosphatase fused to the FLAG epitope.

Taken together, the data shown in Fig. 1 and 2 demonstrate that E2 proteins from different PV types interact with hNAP-1. For BPV1 E2, we can demonstrate that this interaction is direct and involves the AD.

E2 and hNAP-1 cooperate in the activation of BPV1 gene expression.

As already mentioned, NAP-1 and NAP-2 act as histone chaperones to which functions in both transcription and DNA replication have been ascribed (4, 26-28, 39, 50, 58). Here, we focused on an involvement of the interaction of E2 with hNAP-1, observed in vitro, in the activation of transcription. In BPV1, E2 strongly activates transcription from several BPV1 promoters by binding to the 12 E2 binding sites located within the regulatory region called the LCR (60). A reporter construct containing the entire LCR of BPV1 in front of the luciferase gene was cotransfected into human cell line RTS3b (57) derived from human keratinocytes, the natural target cells of PV, together with vectors for HA-tagged hNAP-1 and BPV1 E2. As shown in Fig. 3A, E2 could activate 100- to 130-fold, depending on the amount of expression vector transfected. The expression of hNAP-1 on its own stimulated promoter activity 2-fold; however, the coexpression of hNAP-1 and small amounts of E2 increased luciferase activity 500-fold. This cooperativity between hNAP-1 and BPV1 E2 was reduced in the presence of larger amounts of E2. However, at these high E2 concentrations, cooperativity between E2 and hNAP-1 was still observed.

As shown in Fig. 2B, point mutations abolishing the replication or transcription function of E2 did not affect binding to hNAP-1 in vitro. In our experiments, the activation-deficient E2 I73A mutant retained 33% the activation capacity of wild-type E2 (data not shown), a finding which is in good agreement with previous findings (14). However, even this residual activation was stimulated about fivefold after cotransfection with hNAP-1, indicating that the step involving binding to hNAP-1 is not affected by this mutant. The same is true for the replication-deficient E2 E39A mutant (Fig. 3A). As expected, the activation of transcription by E2 is necessary for the effect of hNAP-1, since a mutant lacking the AD, E2Δ1-203, was not stimulated by hNAP-1 coexpression, in contrast to the mutant lacking the internal hinge region, E2Δ195-282 (Fig. 3A). The activity of the SV40 early promoter, which had driven the expression of BPV1 E2 in the previous transient transfection experiments, was not stimulated by coexpression of the amounts of E2 and hNAP-1 used here (data not shown). This result excludes the possibility that the enhancing effect is due to elevation of the expression of E2 by hNAP-1. These results demonstrate that E2 and hNAP-1 cooperate in the activation of transcription, which correlates with a direct interaction of the two proteins.

Because the role of NAP-1 as a histone chaperone is known, it is possible that hNAP-1, in cooperation with E2, also affects the state of chromatin during activation by E2. Until now, we have used an E2-responsive reporter construct, which has been transiently transfected. In order to test the effect of hNAP-1 on activation by E2 on reporter constructs organized into a higher-order chromatin structure, we established a cell line harboring reporter construct BPV1 LCR-Luc integrated into the cellular genome. After transiently transfecting this stable cell line with the expression vector for E2, we were not able to observe any activation, independent of whether the hNAP-1 vector was cotransfected or not (data not shown). Since we suspected that the tight packing of chromatin may inhibit the access of E2 to the DNA, we treated the cells at 24 h after transfection with TSA, an inhibitor of histone deacetylases. This treatment enabled E2 to stimulate luciferase activity weakly; this effect was slightly enhanced by the coexpression of hNAP-1 (Fig. 3B). These results demonstrate that with promoters organized in cellular chromatin, hNAP-1 also stimulates activation by E2, although to a much lesser extent than transiently transfected reporter constructs.

Full-length hNAP-1 is required for cooperation with E2.

In order to gain insights into the mechanism of the cooperativity between hNAP-1 and E2 proteins, we determined the domains within hNAP-1 which are bound by E2 and analyzed their roles in the enhancement of activation by E2. A series of N- and C-terminal deletion mutants of hNAP-1 were expressed as GST fusion proteins. Although the N-terminal region from residues 1 to 162 possessed marginal binding activity (GST-hNAP-1Δ4, GST-hNAP-1Δ7, and GST-hNAP-1Δ10), an internal domain from residues 162 to 290 (GST-hNAP-1Δ6) and the C-terminal region from residues 291 to 392 (GST- NAP-1Δ5) bound with greater efficiency to ³⁵S-labeled BPV1 E2 (Fig. 4A). These results suggest that E2 binds to at least two separable domains within hNAP-1, one internal from amino acids 162 to 290 and one C terminal from residues 291 to 392.

To analyze the contributions of the domains of hNAP-1 that interact with E2 to the stimulation of activation by E2, hNAP-1 deletion mutants (Fig. 4B) were fused at the N terminus to an HA tag and expressed. The results obtained in the cotransfection experiments shown in Fig. 4B demonstrate that the N-terminal 91 residues are indispensable for hNAP-1 to enhance activation by E2. All hNAP-1 mutants lacking the N-terminal 91 amino acids failed to cooperate with E2 in activation. Most of the mutants repressed activation by E2 from 2-fold (hNAP-1Δ8) to 10-fold (hNAP-1Δ1, hNAP-1Δ2, hNAP-1Δ5, and hNAP-1Δ7). This repression is not related to squelching due to unphysiologically high concentrations of these mutants compared to that of full-length hNAP-1 in transfected cells, since all hNAP-1 derivatives were present in similar amounts. The sole exception is mutant hNAP-1Δ8, which was produced at reduced levels, as shown in the Western blot in Fig. 4B. Furthermore, titration of these proteins revealed constant, dose-dependent repression (data not shown). Mutant hNAP-1Δ4, retaining the N-terminal domain but lacking an interaction with E2, had no effect on activation by E2. Mutants hNAP-1Δ3 and hNAP-1Δ9, containing the N-terminal domain and one E2 binding motif, respectively, were able to enhance activation by E2, although to a reduced extent compared to that obtained with wild-type hNAP-1. These results indicate that in addition to the N-terminal 91 amino acids, at least one E2 binding motif is required for hNAP-1 to stimulate activation by E2, although it seems that amino acids 162 to 290 of hNAP-1 are more important.

The various hNAP-1 mutants also modulated the activity of the BPV1 promoters in the absence of E2. However, there was no clear correlation between the effects of each mutant on activation by E2 and on basal promoter activity. For example, hNAP-1Δ5 and hNAP-1Δ9 both slightly reduced basal promoter activity, but hNAP-1Δ5 strongly repressed activation by E2 and hNAP-1Δ9 still weakly stimulated activation by E2. These results suggest that the effects of hNAP-1 may be specific for E2. Moreover, these data imply that hNAP-1 has another function in transcription, since some of the mutants, e.g., hNAP-1Δ7 and hNAP-1Δ10, which do not interact with E2 in vitro, strongly inhibited activation by E2; the latter result may be related to their inhibitory effects on promoter activity per se.

E2, hNAP-1, and p300 can form a ternary complex in vitro.

As already mentioned, NAP-1 and NAP-2 have been shown to interact with p300. Like that of E2, the binding of p300 to hNAP-1 may involve at least two domains, one from amino acids 123 to 230 and a second one C terminal to amino acid 290 (4, 58). Thus, the regions in hNAP-1 required for activation by E2 also include the p300 interaction domains, and we cannot rule out the possibility that the binding of hNAP-1 to p300 may be necessary as well. Since it has been demonstrated that E2 proteins of HPV8, HPV18, and BPV1 directly and functionally interact with p300 (37, 49, 56), as does hNAP-1, we were interested in the effects of hNAP-1 and p300 on E2-mediated activation. Unfortunately, coexpression of p300 with reporter construct BPV1 LCR-Luc, used in the experiments described so far, led to the inhibition of promoter activity (data not shown). However, Müller et al. showed previously that HPV8 E2 and coexpressed p300 cooperated in the activation of HPV8 gene expression (49). Since we demonstrated here that in addition to BPV1 E2, HPV8 E2 also bound to hNAP-1 (Fig. 2A and D) and coexpressed hNAP-1 enhanced activation by HPV8 E2 (data not shown; see Fig. 7), we used the HPV8 system to study the effects of hNAP-1 and p300 on E2-mediated activation. In agreement with previous observations, E2 and p300 cooperated, since they stimulated promoter activity 15.5-fold, compared to 4-fold by each of the proteins alone (Fig. 5A). As shown previously, the overexpression of p300 does not affect the expression of the E2 protein (49). hNAP-1 increased activation by either E2 or p300 by up to 23- or 17-fold, respectively. Thus, hNAP-1 can functionally interact with both proteins, in correlation with the direct binding of hNAP-1 to E2 (Fig. 2) and p300 (4, 58). The overexpression of all three proteins together resulted in 134-fold activation. This activation in the presence of E2, hNAP-1, and p300 was far beyond the sum of the effects of the single components, suggesting that a ternary complex of E2, p300, and hNAP-1 likely contributes very efficiently to the activation of transcription.

To analyze possible ternary complex formation among the three proteins in vitro, we used a fragment of p300 from amino acids 1453 to 1882 fused to GST (GST-p300-4). Müller et al. observed previously that E2 binds to this segment of p300, which colocalizes with the C/H3 domain (49). In addition to the KIX domain (4), NAP-2 and NAP-1 were shown to require amino acids 1572 to 1818 for binding to p300 (58), like E2. To confirm the formation of a ternary complex in vitro, we performed a competition experiment. As shown in Fig. 5B, GST or the GST-p300-4 fusion protein was incubated with increasing amounts of bacterially expressed, purified His-tagged hNAP-1 for 2 h. After several washing steps to remove unbound His-tagged hNAP-1, either ³⁵S-labeled E2 or hNAP-1, each obtained by in vitro translation, was added; after incubation for another 2 h, binding was analyzed. The signals were quantified with a PhosphorImager. While in the absence of unlabeled His-tagged hNAP-1 59% of radiolabeled hNAP-1 was precipitated by GST-p300-4, preincubation of increasing amounts of unlabeled hNAP-1 reduced the binding of labeled hNAP-1 to 19% (Fig. 5B, lanes 6 to 10). This competition is consistent with the notion that hNAP-1 binding sites within p300-4 have been occupied by the nonlabeled protein. In contrast, preincubation of GST-p300-4 with the same increasing amounts of His-tagged hNAP-1 did not affect the binding of radiolabeled E2. A total of 17 to 20% of the input E2 proteins were precipitated by GST-p300-4, regardless of preincubation with increasing amounts of His-tagged hNAP-1 (Fig. 5B, lanes 1 to 5). As expected, purified His-tagged hNAP-1 also was bound by GST-p300-4 in the presence of E2, as revealed by Western blotting with a monoclonal antibody against hNAP-1 (Fig. 5B, lanes 21 to 27). These results suggest that E2 can still interact with p300 when hNAP-1 is bound and that a ternary complex among E2, p300, and hNAP-1 is likely.

To further address the formation of a ternary complex among E2, p300, and hNAP-1 in vitro, we investigated the fractionation of the E2 protein in the presence of hNAP-1 and p300 in a density gradient. After incubation of His-tagged, bacterially expressed, purified E2Δ326-420 with His-tagged hNAP-1 and p300 1195 to 1761, the binding reaction was subjected to glycerol gradient centrifugation as described by Tanese (61). A total of 36 fractions were collected, and the presence of E2Δ326-420, hNAP-1, and p300 1195 to 1761 was monitored by Western blotting. Since all three proteins easily were distinguishable due to their characteristic sizes, we used an antibody directed against the His tag. To ensure that most of E2 was complexed by p300 and hNAP-1, we added E2 in limiting amounts to the binding reaction. An initial analysis revealed that all three proteins sedimented near the middle of the gradient within fractions 10 to 31. Therefore, only these fractions were used for the detailed analysis shown in Fig. 5C. Most of E2 was present in fractions 22 to 24. After incubation with hNAP-1, E2 sedimented more toward the top of the gradient, with the highest concentrations in fractions 19 to 22 (Fig. 5C, panel I, E2 vs. E2+hNAP-1). Thus, the interaction between E2 and hNAP-1 can be demonstrated in this glycerol gradient analysis, as indicated by the observation that hNAP-1 cofractionated (Fig. 5C, panel II, hNAP-1 vs. hNAP-1+E2). In contrast to the addition of hNAP-1, the addition of p300 to E2 had the consequence that E2 sedimented more toward the bottom of the gradient (Fig. 5C, panel I, E2+p300). In addition, E2 cofractionated with p300, which was detectable in higher fractions after incubation with E2, compared to free p300, confirming the interaction between E2 and p300 (Fig. 5C, compare panel I, E2+p300, with panel III, p300+E2). In the presence all three proteins, E2 again was shifted toward the top of the gradient (Fig. 5C, panel I, E2+hNAP-1+p300). However, the E2 pattern differed from that obtained after incubation with hNAP-1 alone, since now a larger portion of E2 was detectable in fractions 13 to 16, accompanied by a reduction in fractions 24 and 25. Significant amounts of p300 and hNAP-1 also were found in fractions 13 to 22 when all three proteins were incubated (Fig. 5C, panels II and III, all three proteins). Thus, the cofractionation of E2, hNAP-1, and p300 and the change in the sedimentation of E2 after simultaneous incubation with both hNAP-1 and p300 correlate with ternary complex formation among E2, hNAP-1, and p300.

Together, transient transfection, GST pull-down, and density gradient sedimentation experiments suggested that multiple protein-protein interactions (E2-p300, E2-hNAP-1, and hNAP-1-p300) may contribute to the formation of a stable ternary complex efficient in the activation of transcription.

A direct interaction between p53 and hNAP-1 also is involved in activation mediated by p53.

Previously, it was suggested that NAP-2, which is closely related to NAP-1, augments transcription by activators such as p53 and E2F, which use p300 as a coactivator, through a direct interaction with p300 (58). In order to test whether, in addition to p300- hNAP-1 binding, a direct interaction between p53 and hNAP-1 (like that for PV E2) may be involved, we incubated His-tagged, bacterially expressed, purified p53 with a GST-hNAP-1 fusion protein. As shown in Fig. 6A, p53 was precipitated by the GST-hNAP-1 fusion protein and not by GST. The interaction could be confirmed with GST-p53 and His-tagged, purified hNAP-1 (data not shown). These results demonstrate that the interaction between hNAP-1 and p53 is direct and not merely mediated by other factors, such as p300, as has been suggested elsewhere (58). When a synthetic reporter construct containing one p53 binding site upstream of the Hsp70 promoter was transiently transfected into the p53-negative cell line RTS3b, hNAP-1 also potentiated activation by p53 (Fig. 6B), confirming that the interaction is functional, as has been shown for NAP-2 (58). On the basis of these results, we suspected that hNAP-1 is also able to stimulate the activation by p53 of an endogenous p53-responsive promoter in the absence of the coexpression of p300. We analyzed the effect of the coexpression of p53 and hNAP-1 on the p21 level in cell line RTS3b. Under the Western blotting conditions shown in Fig. 6C, endogenous p21 was not visible. Cotransfection of the p53 expression vector or of small amounts of the hNAP-1 expression vector did not induce p21 expression to detectable levels. However, the coexpression of p53 and large amounts of hNAP-1 led to the appearance of the p21 protein. In addition, endogenous p53 and hNAP-1 can form a stable complex, as shown in Fig. 6D. Extracts from RTS3b cells, which do not express a detectable level of the p53 protein, and from primary neonatal human epidermal keratinocytes, which express high levels of p53 (Fig. 6D, lower panel), were subjected to immunoprecipitation with a monoclonal antibody against p53. Only in neonatal human epidermal keratinocytes could endogenous hNAP-1 be coprecipitated. The results shown in Fig. 6 imply that p53 and hNAP-1 interact directly with each other.

In order to test whether a functional interaction with hNAP-1 is a general strategy for activators or is restricted to activators interacting with p300, we used the AD of TEF-1. Various investigators have shown that TEF-1 neither binds to p300 in vitro nor cooperates with coexpressed p300 (49, 59). As shown in Fig. 7, the AD of TEF-1 failed to interact with GST-hNAP-1 in vitro. The TEF-1 AD fused to the DBD of GAL4 stimulated transcription from a reporter containing GAL4 binding sites about fourfold; this effect was not enhanced after the coexpression of hNAP-1. However, as a control, activation by a hybrid protein composed of the AD of HPV8 E2 and the DBD of GAL4 was further stimulated by coexpressed hNAP-1.