The EMBO Journal vol. 9 no.6 pp. 1 883 - 1 888, 1 990
The DNA binding domain (POU domain) of transcription factor oct-1 suffices for stimulation of DNA replication
C.Peter Verrijzer, Arnoud J.Kal and Peter C.Van der Viiet Laboratory for Physiological Chemistry, University of Utrecht, Vondellaan 24A, NL-3521 GG Utrecht, The Netherlands Communicated by H.S.Jansz
Oct-i, also referred to as NFIH, OTF-l, OBP100 or NFAl, is a ubiquitous sequence-specific DNA binding protein that activates transcription and adenovirus DNA replication. The protein contains a conserved DNA binding domain (POU domain) present in several transcription factors. We have overproduced oct-i, the related oct-2 and several oct-i deletion mutants in a vaccinia expression system to identify the domains important for activation of DNA replication in vitro. Both oct-i and oct-2 stimulate adenovirus DNA replication in an octamer-dependent manner. From deletion studies it appears that the 160 amino acid long POU domain suffices for stimulation. This domain consists of two subdomains, a POU-specific and a homeo domain. Deletion of the POU-specific domain revealed that the homeo domain has an intrinsic, but weak DNA binding activity and surprisingly, inhibits DNA replication. As the POU domain does not coincide with the transcription activation domain, these results indicate that, although oct-i functions both in DNA replication and transcription, the mechanisms underlying these processes are probably distinct. Key words: adenovirus / DNA replication / octamer sequence/POU domain/transcription factor
Introduction Initiation of DNA replication is the key event in growth and reproduction. Eukaryotic origins of viral DNA replication frequently consist of a core region, which is absolutely indispensable, and an auxiliary region capable of binding cellular transcription factors (De Pamphilis, 1988). In recent years several of these factors have been identified and in some cases their role in initiation of DNA replication has been established. Identification of functional protein domains involved in activation of DNA replication is a first step towards understanding the mechanisms governing regulation of DNA replication. Binding of transcription factors is clearly documented for replication of adenovirus DNA, a system that can be reconstituted in vitro. Replication of human adenovirus DNA occurs by a protein-priming mechanism in which a precursor to the viral terminal protein (pTP) becomes covalently bound to the first nucleotide, a dCMP residue, thus forming a pTP-dCMP initiation complex. The 3'-OH group of dCMP serves as a primer for further polymerization by a displacement mechanism (reviews, Hay and Russell, 1989; Oxford University Press
Challberg and Kelly, 1989; Stillman, 1989). Initiation can be reconstituted in vitro and minimally requires the complex of the viral DNA polymerase and pTP (pTP-pol) and the core origin consisting of nucleotides 1-18. Initiation of DNA replication is considerably enhanced by two cellular proteins, nuclear factor I (Nagata et al., 1982; Leegwater et al., 1985; Hay, 1985; Rosenfeld and Kelly, 1986; Meisterernst et al., 1988) and nuclear factor HI (Pruijn et al., 1986; Rosenfeld et al., 1987). These proteins bind to the auxiliary region of the origin located between nucleotides 25 and 50. NFII was purified from HeLa cells as a 90-95 kd protein which is indistinguishable from the ubiquitous octamer transcription factor OTF-1 (O'Neill et al., 1988; Pruijn et al., 1989). This protein, also called oct-i, OBP100 or NF-A1, binds to the octamer element (Singh et al., 1986; Sturm et al., 1987; Fletcher et al., 1987). This element, 5'-ATGCAAAT-3', is an essential component of promoters and enhancers of cellular genes as diverse as histone H2B, Ul and U2 snRNAs and immunoglobulin light and heavy chains. Besides oct-i, several other octamer recognizing proteins have been detected. Oct-2 proteins (oct-2A, oct-2B) are expressed primarily in lymphoid cells and are presumably involved in cell typespecific immunoglobulin gene expression (Staudt et al., 1986; Landolfi et al., 1986; Lebowitz et al., 1988; Schreiber et al., 1988). Oct-3 to oct-7 have been identified in various mouse tissues at different stages of development (Scholler et al., 1989; Lenardo et al., 1989), while related proteins have been detected in adult brain (He et al., 1989) but these have not been characterized yet. Cloning and sequencing analysis of the oct-I and oct-2 cDNAs has revealed the presence of a DNA binding domain called the POU domain (Herr et al., 1988; Sturm et al., 1988; Clerc et al., 1988; Scheidereit et al., 1988; Muller et al., 1988). This 160 amino acid long domain is also present in the product of the unc-86 gene of Caenorhabditis elegans and Pit-1/GHF-1, a rat pituitary-specific transcription factor. The POU domain can be subdivided into an N-terminal 74 amino acid long POU-specific domain and a C-terminal POU homeo domain. The latter is homologous to the homeo domain found in homeotic genes (review, Scott et al., 1989) and contains a potential tri-a-helical motif thought to be involved in DNA recognition (Garcia-Blanco et al., 1989). Based upon the failure to detect DNA binding of the oct-I POU homeo domain, obtained by in vitro translation, a bipartite DNA binding structure has been proposed (Sturm and Herr, 1988). Previously it was shown by deletion analysis that for nuclear factor I (NFI) the N-terminal region, containing the DNA binding domain and the dimerization domain, but lacking the transcription activation domain, was sufficient for stimulation of adenovirus DNA replication (Mermod et al., 1989;. Gounari et al., 1990). To investigate if such a situation also applies to NFiII, we dissected the oct-I gene 1883
C.P.Verrijzer, A.J.Kal and P.C.van der
Viiet
Fig. 2. The POU domain and the homeo domain differ in binding specificity. Probes containing the X.laevis U2 enhancer (lanes 1,2,3), the Ad2 origin (lanes 4,5,6) or the Ad4 origin (lanes 7,8,9) were incubated with 6 ng purified homeo domain (lane, 2,5,8) or 2 ng purified POU domain (lanes 3,6,9), or in the absence of protein (lanes 1,4,7) followed by gel retardation analysis on a 15% polyacrylamide gel.
Fig. 1. Binding of recombinant vaccinia virus expressed oct-i, oct-I mutants and oct-2 to the octamer. An end-labelled octamer containing probe from the Ad2 origin was incubated with 1 Al cytoplasmic extract obtained from cells infected with control wild-type virus (lane 2) or recombinants expressing: oct-I (lane 3), 212-743 (lane 4), 1-23/269-743 (lane 5), 1-592 (lane 6), 1-440 (lane 7), POU domain (lane 8), oct-2 (lane 9), homeo domain crude extract (lane 10) or purified and concentrated homeo domain (1 jsl fraction 14 from the second fast flow S column, lane 11). Lane 1 shows authentic HeLa NFIII binding activity. Bound and free DNA (F) were separated by electrophoresis (see Materials and methods).
and delimited the domain required for DNA replication. We present evidence that the POU domain suffices for stimulation and that the homeo domain is capable of DNA binding, but inhibits DNA replication. This suggests a role for the POU-specific domain in DNA replication.
Results Overproduction of oct-1 and deletion mutants We employed a eukaryotic expression system which was previously used to overproduce other biologically active components of the adenovirus DNA replication system like pTP, pol and NFI (Stunnenberg et al., 1988; Gounari et al., 1990). The oct-I open reading frame as well as several deletion mutants were cloned downstream of the late 11K vaccinia promoter employing the recombination vector pATA-18. Recombinant vaccinia viruses were produced and used to infect HeLa cells in suspension. At 24 h postinfection nuclei and cytoplasm were separated. Total overproduction was 30- to 60-fold and, dependent on the construct, 20-50% of the recombinant protein was found in the cytoplasm. To prevent contamination with endogenous NFIII, we used cytoplasmic extracts. As shown by gel retardation using an Ad2 origin containing fragment as probe (Figure 1), octamer-recognizing proteins were present abundantly in all recombinant extracts containing the intact oct-I POU domain and were absent from extracts prepared from HeLa cells infected with wild-type vaccinia virus. The oct- 1 homeo domain binds to the adenovirus origin but with a reduced affinity The POU domain can be subdivided into an N-terminal 74 amino acid long POU-specific domain and a C-terminal POU
1884
homeo domain that includes a putative tri-a-helical motif present in many homeo domain containing proteins (Sturm et al., 1988). When we overexpressed the POU homeo-domain, weak binding was observed under conditions in which the POU domain binds strongly (Figure 1, compare lanes 8 and 10). Only when higher concentrations of partially purified POU homeo domain were used was a considerable amount of the probe bound (Figure 1, lane 11). These results suggest that, although the POU homeo domain itself has an intrinsic binding capacity, the POU-specific domain contributes considerably to the binding affinity and possibly the specificity. This could be achieved either by stabilization or by making additional protein -DNA contacts. To investigate this, we purified the POU domain and the homeo domain to homogeneity and compared their binding to the origins of Ad2 and Ad4 and to the Xenopus laevis U2 enhancer by gel retardation. As shown in Figure 2 a clear difference in specificity was observed. No binding to the U2 enhancer was observed for the homeo domain, while the same amount of polypeptide could bind to the Ad4 origin and even better to the Ad2 origin. For the POU domain a reverse binding order was observed: Ad4 > Ad2 or U2 in agreement with published data for the intact NFIII protein (Pruijn et al., 1987). We also compared the DNase I footprint of both polypeptides on the Ad2 origin. Figure 3 shows that the footprint of the POU domain on the bottom strand coincides with published data for NFIII but that the region protected against DNase I by the homeo domain is shorter, in particular at the 5' end of the binding site. The same results were obtained for the top strand (not shown). The difference in specificity and the different footprint borders suggest the existence of DNA contacts by the POUspecific domain. Recently we confirmed this by hydroxyl radical footprinting (C.P.Verrijzer et al., in preparation). The POU domain stimulates DNA replication We assayed the functional properties of oct-i and its deletion mutants, employing an in vitro reconstituted system for Ad DNA replication consisting of the purified viral proteins (pTP-pol, DNA binding protein) and NFI. AdS DNA-TP digested with XhoI was used as a template. Fragments B and C contain the origin. To compare the various recombinant
oct-1 POU domain stimulates DNA replication
.Im 'OL
--
OW
U,
.S
Uumi-,, .I.
Fig. 3. Comparison of the DNase I footprints of the POU domain and the homeo domain on the Ad2 origin. The bottom strand of the origin is shown. Lane 1, G+A sequencing marker (M). Lane 2, without protein. Lane 3, 90 ng homeo domain (H) added. Lane 4, 50 ng POU domain (P) added. The borders of the protected regions, determined by scanning, are given.
POU containing fragments with authentic NFIII, identical amounts of DNA binding units were added. Previous experiments established that the DNA binding activity per /tg protein, corrected for the change in molecular mass, did not differ significantly for fragments containing the intact POU domain (Figure 1 see also Sturm and Herr, 1988). Figure 4a shows a 3-fold stimulation by oct-I indistinguishable from that of HeLa NFHI (lanes 1,3). Upon addition of NFIII/oct-1, replication becomes very efficient as indicated by the appearance of labelled single-stranded B and C fragments, originating from second rounds of replication. When more NFIH/oct-1 was added, up to a 5-fold stimulation was obtained (not shown). We also tested the effects in the absence of NFI (lanes 10-18) which gave qualitatively similar results, although the basal level was reduced 10-fold. No stimulation was observed with the control extract obtained from wild-type vaccinia-infected HeLa cells (lanes 2,11), excluding the possibility that the effect was due to endogenous stimulatory proteins. Stimulation was dependent upon the presence of an intact NFIII binding site since replication of a plasmid containing a point mutation in the octamer, Ad2-pm46 (Pruijn et al., 1986), was not enhanced (Figure 4b). The same stimulation was obtained with four mutants having deletions in either the N-terminal region or the C-terminal region up to the POU domain, or with the POU domain itself. This shows that the POU domain suffices for stimulation of adenovirus DNA replication. No effect was observed with the POU-specific domain which does not bind DNA (not shown). Addition of the purified POU homeo domain inhibited replication to below the basal level without NFIII (Figure 5, lanes 1-3). This inhibition could be overcome by adding increasing amounts of NFIII (Figure 5, lanes 4,5), indicating a competitive binding to the octamer
Fig. 4. Stimulation of Ad DNA replication in vitro by POU containing proteins. (A) In vitro replication of AdS DNA-TP digested with XhoI. B and C are the origin containing fragments, ss-B and ss-C indicate ssDNA, originating from a second round of replication. Equal amounts of binding units (0.6 bu) were added either with (lanes 1-9) or without (lanes 10-18) 3 ng NFI. 1 binding unit (bu) is defined as the amount of protein that binds 1.5 fmol (50%) of an Ad2 probe in a bandshift assay. (B) The stimulation is dependent on an intact octamer. Plasmids containing the Ad2 origin were used as a template in the presence of 3 ng NFI. Ad2-pm46 contains a point mutation in the octamer (Pruijn et al., 1986). The origin containing fragments are indicated (0). Due to the protein-priming mechanism of adenovirus DNA replication, products are covalently attached to pTP, leading to a reduced mobility (arrow). The position of the replication bands of Ad2-pm46 differs slightly from that of Ad2 since different plasmid constructs were used.
sequence. A similar inhibition was observed in the absence of NFI (not shown).
Oct-2 can stimulate DNA replication In addition to the ubiquitous oct-I protein, several other tissue-specific, POU domain containing proteins have been described that are probably involved in developmental regulation. Analysis of the cDNA of one of these, the lymphoid-specific oct-2/OTF-2A, showed the presence of a POU domain that shares 88% similarity with oct-1, while the remainder of the protein shows only weak homology (Muller et al., 1988. Clerc et al., 1988; Scheidereit et al., 1988). The high similarity of the POU domains of oct-2 and oct-I raises the question whether oct-2 can also stimulate adenovirus DNA replication. To investigate this, we overexpressed oct-2 cDNA similarly to oct-I and observed strong
1885
C.P.Verrijzer, A.J.Kal and P.C.van der Viiet
for activation of DNA replication and transcription differ, in spite of the requirement for recognition of the octamer sequence in both processes. Interestingly, a similar result was obtained for NFI for which the conserved DNA binding domain, distinct from the C-terminal transactivating domain, also suffices for DNA replication (Mermod et al., 1989; Gounari et al., 1990). Two hypotheses can be envisaged to explain these results. One is that binding itself is sufficient, leading to a structural change in the origin which enables efficient formation of an initiation complex, for instance by providing access to other viral replication proteins. Alternatively, the POU domain could contain sites for direct protein -protein contacts with other replication factors, leading to formation of an initiation complex or stabilization of such a complex. At present, we cannot distinguish between these possibilities. However, we have not observed any direct interactions between NFII and NFI or between NFHI and viral proteins using cross-linking experiments (Y. M.Mul et al., in preparation).
binding (Figure 1, lane 9) as well as stimulation of DNA replication (Figure 4a, lanes 9,18), similar to NFIII/oct-1. As the POU-specific domains of oct-I and oct-2 are most conserved (one conservative change in 74 amino acids) and as the isolated homeo domain inhibits replication, this strongly suggests a role for the POU-specific domain in DNA replication.
Discussion Our data, summarized in Figure 6 indicate that stimulation of DNA replication requires only a small domain of NFIII/oct-1 to which adenovirus may have adapted during evolution. This domain does not coincide with the transcription activation domains identified in oct-I and oct-2 (Tanaka and Herr, 1990). This suggests that the mechanisms
DNA binding properties of the POU homeo domain The oct-i homeo domain, produced by in vitro translation, does not bind detectably to the octamer (Sturm and Herr, 1988). This may be explained by the low concentration of the polypeptide in such a reaction. When we assayed purified, concentrated homeo domain significant binding to the Ad2 octamer was obtained, even though the apparent binding affinity was considerably lower than that of the intact POU domain. Binding to the Ad4 or U2 octamer was even weaker and would be difficult to detect at low protein concentrations. Not only the binding affinity, but also the binding specificity of the POU domain and the homeo domain differ.
Fig. 5. The homeo domain inhibits DNA replication. Ad5 DNA-TP was used as a template in the presence of 3 ng NFI. The concentrations of NFIII and homeo domain are given in binding units.
DNA binding
Adenovirus replication
C oct-I
] 743
I1 2 12-743
1-23/269-743
I
1-592
1-440
1-23/269-440 (POU)
Il
0
369-440 (homoeo)
1-23/269-368
(POU-specific)
Oct-2
s
sn
++
+
++
+
++
+
++
+
++
+
++
+
+_
0
O
0
++-
+
Fig. 6. Localization of the oct-I domain required for adenovirus DNA replication. Oct-1, oct-I deletion mutants and oct-2 are depicted schematically. Indicated is the POU domain consisting of: the POU-specific (barred) and POU homeo (hatched) domains. Tabulated on the right is the biochemical activity. DNA binding: + + indicates an activity identical to that of oct-i, + indicates the diminished affinity of the homeo domain and 0 indicates no binding. Stimulation of adenovirus DNA replication: + indicates activity indistinguishable from that of HeLa NFIII, 0 indicates no effect and indicates an inhibition of DNA replication.
1886
oct-1 POU domain stimulates DNA replication
While the POU domain, similarly to intact NFHI/oct-1, binds most strongly to the authentic octamer sequence ATGCAAAT present in Ad4 and the U2 enhancer, the homeo domain prefers the Ad2 sequence 5'-ATGATAATGA-3'. Interestingly, this sequence contains TAATG, the consensus sequence for homeotic proteins (Scott et al., 1989; Hoey and Levine, 1988). In this respect it is noteworthy that the footprint borders of the homeo domain are shifted towards the 3' end of the NFIH recognition sequence. Recent hydroxyl radical footprinting has confirmed that the homeo domain lacks contacts at the 5' side of the Ad2 recognition sequence, suggesting that additional contacts are provided by the POU-specific domain. So far, direct binding of the POU-specific domain to DNA could not be
detected. Presently we are investigating whether the conin the core recognition sequence made by the homeo domain are identical to those made by the POU domain. For another POU protein, GHF-I or Pit-1, it has recently been reported that the homeo domain is sufficient for sequencespecific DNA binding (Theill et al., 1989). tacts
Role of the POU-specific domain in DNA replication We were surprised to find that the homeo domain inhibits DNA replication. Several other proteins can bind to the origin without an inhibitory effect, such as origin recognition protein ORP-A (Rosenfeld et al., 1987) or NFIV (de Vries et al., 1989). The inhibition can be overcome by excess NFIIH, showing that binding of the homeo domain is essential for inhibition. This implies that the POU-specific region has at least two functions: (i) contribution to the binding and (ii) stimulation of initiation of Ad DNA replication. These functions might be related, but clearly the effect of the POUspecific domain is not simply due to an increased DNA
binding affinity for two reasons. First, we added an identical amount of DNA binding units to the replication reaction. Second, when DNA binding was the only determinant for stimulation one might expect no effect of the homeo domain, rather than the observed inhibition. Therefore we consider it more likely that non-functional multi-protein DNA complexes are formed in the presence of the homeo domain or that an aberrant origin structure is induced which cannot be adequately recognized by viral replication proteins. So far, no specific role of the POU-specific region in transcriptional control has been demonstrated. Interestingly, this region is even more conserved between oct- I and oct-2 (99%) than the POU homeo domain (88%), and this may explain why oct-2 is also capable of stimulation. Presently we are investigating the effects of other, less conserved POUspecific domains from transcription factors Pit-I and Unc-86 in adenovirus DNA replication.
Materials and methods Construction of recombinant vaccinia viruses The 2.5 kb FnuDII-Hindll fragment frompBSoctl + (Sturm et al., 1988) containing the 743 amino acid long oct-I open reading frame was end repaired and blunt ligated into the filled in EcoRI site of pATA-18 leading pATA-18 is a vaccinia recombinant vector containing to the1 1K late viral promoter followed by the pUC- 18 multiple cloning site, inserted in the thymidine kinase gene (Stunnenberg et al., 1988). For mutant 212-743, the 1.9 kb AhaII -HindIL fragment from pBS octl + was cloned into pATA-32-18 +2 digested with AccI and HindIH. pATA-32-18 +2 is a vector in which translation starts from a synthetic start codon directly
pATA-18/oct-1.
of the multiple cloning site, with +2 indicating the addition of nucleotides to enable expression in the correct reading frame. This leads
upstream two
to a product containing MNWIMCRS followed by amino acids 212 -743 from oct-i. For mutant 1-23/269-743, pBSoct-1+ABH (Sturm and Herr, 1988) was digested with FnuDHI-HindIIIand blunt ligated in the filled in EcoRI site of pATA-18. For mutant 1-592, pATA-18/oct-1 was linearized with NcoI and end repaired with T4 DNA polymerase. The 1.8 kb ClaI-NcoI fragment was cloned in pATA-18-STOP linearized with ClaI and Ball. pATA-18-STOP is a pATA-18 derivative with stop codons in three reading frames directly downstream of the BalI site in the multiple cloning site. This leads to a 592 amino acid long product with an additional C-terminal Pro. Mutant 1-440 was constructed essentially as 1-592, only PflMI was used instead of NcoI, thus encoding amino acids 1-440 to which TIAE was added C terminally. Mutant 1-23/269-440 (POU) was constructed essentially as 1-440 only mutant 1-23/269-743 was used as starting material. The resulting construct encoded amino acids 1-23 and 269-440. For expression of the homeo domain, the 234 bp EcoRI-SphI fragment from mutant 1-440, encoding amino acids 369-440, was cloned in the EcoRI-SphI sites from pATA-32-18, thus donating an initiation codon in front of amino acid 369. For oct-2, the 1.9 kb EcoRI fragment containing the oct-2 cDNA (Muller et al., 1988; Clerc et al., 1988. Scheidereit et al., 1988) was cloned in the EcoRI site of pATA-18. These plasmids were used to prepare recombinant viruses as described (Stunnenberg et al., 1988).
Preparation of extracts HeLa cells (2.5 x 107) were infected with 5 p.f.u./cell and harvested 24 h post-infection. The cells were washed twice with phosphate-buffered saline (PBS) containing 0.5 mM MgCl2 and resuspended in 0.5 ml 20 mM HEPES/KOH (pH 7.5), 5 mM KCl, 0.5 mM MgCl2, 2 mM phenylmethylsulphonyl fluoride (PMSF), 0.1 % Nonidet P40. After swelling the cells were Dounce homogenized and centrifuged at low speed. The resulting cytoplasm was centrifuged for 30 min in an Eppendorf centrifuge at 15 000 r.p.m. and the supernatant (cytoplasmic extract) was mixed with an equal amount of twice concentrated buffer B (25 mM HEPES/KOH, pH 8.0, 1 mM dithiothreitol, 0.1 mM PMSF, 0.02% Nonidet P40, 20% glycerol) containing 250 mM NaCl. Protein purification NFIII was purified from HeLa nuclei as described (Pruijn et al., 1988). NFI was purified by DEAE (200 mM flowthrough), heparin-Sepharose (peak 400 mM) and fast flow S (peak 350 mM) chromatography followed by an affinity column pKB 67-88 (Jones et al., 1987) (peak 520 mM). The purification of pTP-pol and DBP was as described (Mul et al., 1989). The homeo domain was purified as follows: 5 x 108 HeLa cells were infected with the appropriate recombinant virus. Cytoplasmic extracts were adjusted to 30 mM NaCl and applied to a DEAE column equilibrated with buffer B containing 30 mM NaCI. The flowthrough was adjusted to 0.1 M NaCl and applied to a fast flow S column, washed with B/0.1 M NaCl and eluted with a linear gradient of 0.1-0.5 M NaCl. Homeo domain eluted around 0.3 M NaCl. Fractions active in gel retardation were combined, adjusted to 0.1 M and applied to a heparin-Sepharose column. After washing with B/0.1 M Nacl, the column was developed with a linear gradient of 0.1-1 M NaCl in buffer B. Homeo domain eluted around 0.5 M. Active fractions were concentrated by step elution from a second fast flow S column. All purification steps were performed on ice or at 4°C. Details of the purification of the POU domain will be presented elsewhere. Gel retardation and DNase I footprinting The following probes were used: Ad2, a 110 bp EcoRI-XbaI fragment from pHRI, containing the Ad2 origin. Ad4, an 85 bp EcoRl-XbaI fragment from p4A85A (Hay, 1985), containing the Ad4 origin. U2, a 40 bp doublestranded oligonucleotide containing the X. laevis U2 octamer (Mattaj et al., 1985). 1 Al of cytoplasmic extracts were assayed by gel retardation as described (Pruijn et al., 1987). For DNase I footprinting, the bottom strand of the 331 bp NdeI-XbaI fragment of pHRI, labelled at the 5' end of the XbaI site by T4 polynucleotide kinase, was used under conditions described previously (Pruijn et al., 1987).
DNA replication in vitro pTP-pol (50 ng) and 0.6 Ag DBP were added to the incubation mixtures (15 dl). For replication of AdS DNA-TP terminal XhoI fragments 25 ng of template was added per incubation. For replication of origin containing plasmid DNA, pHRI, was digested with EcoRI-AvaH and Xpm46G, containing the Ad2-pm46 origin, with EcoRI before use and 30 ng plasmidA DNA was added. In all cases the amount of extract was adjusted to 1 by addition of extract from wild-type vaccinia-infected cells. Further incubation conditions and analysis of the products were as described (Pruijn et al., 1988).
1887
C.P.Verrijzer, A.J.Kal and P.C.van der Vliet
Acknowledgements Winship Herr for providing us with the pBSoct-l+ as and for making data available before publication. We thank Michael Muller and Walter Schaffner for the oct-2 cDNA clone, Henk G. Stunnenberg for pATA-18 and pATA-32-18, Ron Hay for pHRI and p4A85A and lain Mattaj for X. laevis U2 enhancer DNA. Thanks are also due to Ina Pospiech and Roel Schiphof for tissue culture, to Henk Stunnenberg and Peter H.van Bragt for advice on the vaccinia work and We are
well as
grateful
to
pBSoct-l +2ABH
to Wim van Driel
pol, respectively.
This work was
Foundation for Chemical Research
Organization
Netherlands
providing purified NFIII and pTPin part by the Netherlands
and Yvonne Mul for
supported (SON)
with financial
for Scientific Research
support
from the
(NWO).
References Challberg,M.D. (1988)
and
Rev. Biochem., 58, 671-717. LeBowitz,J.M., Baltimore,D. and Sharp,P.A.
Kelly,T.J. (1989) Annu.
Dev., 2,
1570-1581.
Pamphilis,M.L. (1988) Cell, 52, 635-638. De Vries,E., Van Driel,W., Bergsma,W.G., Arnberg,A.C. and Van der Vliet,P.C. (1989) J. Mol. Biol., 208, 65-78. Fletcher,C., Heintz,N. and Roeder,R.G. (1987) Cell, 51, 773-781. De
Garcia-Blanco,M.A., Clerc,R.G. and
Sharp,P.A. (1989)
Genes
Dev., 3,
739-745.
de Francesco,R., Schmidt,J., Van der Vliet,P.C., Cortese,R. Stunnenberg,H.G. (1990) EMBO J., 9, 559-566. Hay,R.T. (1985) J. Mol. Biol. 186, 129-136. Hay,R.T. and Russell,W.C. (1989) Biochem. J., 238, 3-16. He,X., Treacy,M.N., Simmons,D.M., Ingraham,H.A., Swanson,L.W. and Rosenfeld,M.G. (1989) Nature, 340, 35-42. Herr,W., Sturm,R.A., Clerc,R.G., Curcuran,L.M., Baltimore,D. Sharp,P.A., Ingraham,H.A., Rosenfeld,M.G., Finney,M., Ruvkun,G. and Horvitz,H.R. (1988) Genes Dev., 2, 1513-1516. Hoey,T. and Levine,M. (1988) Nature, 332, 858-861. Jones,K.A., Kadonaga,J.T., Rosenfeld,P.J., Kelly,T.J. and Tjian,R. (1987)
Gounari,F., and
Cell, 48, 79-89.
Landolfi,N.F., Capra,J.D.
and Tucker,P.W. (1986) Nature, 323, 548-551. Kobayashi,T., Staudt,L., Baltimore,D. and Sharp,P.A.
Lebowitz,J.H.,
(1988)
Genes
Dev., 2, 1227-1237.
Leegwater,P.A.J.,
Van
and Van der
Driel,W.
Vliet,P.C. (1985) EMBO
J., 4, 1515-1521.
Staudt,L., Robbins,P.,
Lenardo,M.J.,
Baltimore,D. (1989) Science,
Mattaj,I.W., Lienhard,S.,
Kuang,A., Mulligan,R.C. and
243, 544-546.
Jirichy,J.
and De
Robertis,E.M. (1985) Nature,
163-167.
316,
Meisteremst,M., Gander,I., Rogge,L. and Winnacker,E.-L. (1988) Nucleic Acids Res., 16, 4419-4435. Mermod,N., O'Neill,E.A., Kelly,T.J. and Tjian,R. (1989) Cell, 58, 741-753.
Mul,Y.M.,
(1989)
Van
Miltenburg,R.T.,
Nucleic Acids
Muller,M.M.,
De
Clercq,E.
and Van der
Vliet,P.C.
Res., 17, 8917-8929.
Rupert,S., Schaffner,W.
and
Matthias,P. (1988) Nature, 336,
544-551.
Nagata,K., Guggenheimer,R.A. and Hurwitz,J. Sci.
(1982) Proc. Natl. Acad.
USA, 79, 6438-6442.
O'Neill,E.A., Fletcher,C.,
Burrow,C.R. Heintz,N.,
Roeder,R.G. and Kel-
ly,T.J. (1988) Science, 241, 1210-1213. Pruijn,G.J.M., Van Driel,W. and Van der Vliet,P.C.
(1986) Nature, 322,
656-659.
Pruijn,G.J.M., Van Driel,W., Van Miltenburg,R.T. and Van der Vliet, P.C. (1989) EMBO J., 6, 3771-3778. Pruijn,G.J.M., Van Miltenburg,R.T., Claessens,J.A.J. and Van der Vliet,P.C. (1988) J. Virol., 62, 3092-3102. Pruijn,G.J.M., Van der Vliet,P.C., Dathan,N.A. and Mattaj,I.W. (1989) Nucleic Acids Res., 17, 1845-1863. Rosenfeld,P.J. and Kelly,T.J. (1986) J. Biol. Chem., 261, 1398-1408. Rosenfeld,P.J., O'Neill,E.A., Wides,R.J. and Kelly,T.J. (1987) Mol. Cell. Biol., 7, 875-886.
Scheidereit,C., Currie,R.A.
Scholler,H.R.,
(1989)
Scott,M.P., 989,
1888
Cromlish,J.A., Gerster,T., Kawakami,K., Balmaceda,C.G., and Roeder,R.G. (1988) Nature, 336, 551-557.
Hatzopoulos,A.K., Balling,R., Suzuki,N. and
EMBO
144-158.
Staudt,L.M., Singh,H., Sen,R., Wirth,T., Sharp,P.A. and Baltimore,D. (1986) Nature, 323, 640-643. Stillman,B.M. (1989) Annu. Rev. Cell Biol., 5, 197-225.
Stunnenberg,H.G., Lange,H., Philipson,L., Van Miltenburg,R.T. and Van der Vliet,P.C. (1988) Nucleic Acids Res., 16, 2431-2444. Sturm,R.A. and Herr,W. (1988) Nature, 336, 601-604. Sturm,R., Baumruker,T., Franza,Jr.,B.R. and Herr,W. (1987) Genes Dev.,
1, 1147-1160. Sturm,R.A., Das,G. and Herr,W. (1988) Genes Dev., 2, 1582-1599. Tanaka,M. and Herr,W. (1990) Cell, 60, 375-386. Theill,L.E., Castrillo,J.L., Wu,D. and Karin,M. (1989) Nature, 342, 945-948.
Clerc,R.G., Corcoran,L.M., Genes
Schreiber,E., Matthias,P., Muller,M.M. and Schaffner,W. (1988) EMBO J., 7, 4221-4229. Singh,H., Sen,R., Baltimore,D. and Sharp,P.A. (1986) Nature, 319,
Gruss,P.
J., 8, 2543-2550.
Tamkun,J.W.
and
Hartzell,G.N.
(1989)
Biochim.
Biophys. Acta,
Received
on
February 13, 1990
The DNA binding domain (POU domain) of transcription factor oct-1 suffices for stimulation of DNA replication
C.Peter Verrijzer, Arnoud J.Kal and Peter C.Van der Viiet Laboratory for Physiological Chemistry, University of Utrecht, Vondellaan 24A, NL-3521 GG Utrecht, The Netherlands Communicated by H.S.Jansz
Oct-i, also referred to as NFIH, OTF-l, OBP100 or NFAl, is a ubiquitous sequence-specific DNA binding protein that activates transcription and adenovirus DNA replication. The protein contains a conserved DNA binding domain (POU domain) present in several transcription factors. We have overproduced oct-i, the related oct-2 and several oct-i deletion mutants in a vaccinia expression system to identify the domains important for activation of DNA replication in vitro. Both oct-i and oct-2 stimulate adenovirus DNA replication in an octamer-dependent manner. From deletion studies it appears that the 160 amino acid long POU domain suffices for stimulation. This domain consists of two subdomains, a POU-specific and a homeo domain. Deletion of the POU-specific domain revealed that the homeo domain has an intrinsic, but weak DNA binding activity and surprisingly, inhibits DNA replication. As the POU domain does not coincide with the transcription activation domain, these results indicate that, although oct-i functions both in DNA replication and transcription, the mechanisms underlying these processes are probably distinct. Key words: adenovirus / DNA replication / octamer sequence/POU domain/transcription factor
Introduction Initiation of DNA replication is the key event in growth and reproduction. Eukaryotic origins of viral DNA replication frequently consist of a core region, which is absolutely indispensable, and an auxiliary region capable of binding cellular transcription factors (De Pamphilis, 1988). In recent years several of these factors have been identified and in some cases their role in initiation of DNA replication has been established. Identification of functional protein domains involved in activation of DNA replication is a first step towards understanding the mechanisms governing regulation of DNA replication. Binding of transcription factors is clearly documented for replication of adenovirus DNA, a system that can be reconstituted in vitro. Replication of human adenovirus DNA occurs by a protein-priming mechanism in which a precursor to the viral terminal protein (pTP) becomes covalently bound to the first nucleotide, a dCMP residue, thus forming a pTP-dCMP initiation complex. The 3'-OH group of dCMP serves as a primer for further polymerization by a displacement mechanism (reviews, Hay and Russell, 1989; Oxford University Press
Challberg and Kelly, 1989; Stillman, 1989). Initiation can be reconstituted in vitro and minimally requires the complex of the viral DNA polymerase and pTP (pTP-pol) and the core origin consisting of nucleotides 1-18. Initiation of DNA replication is considerably enhanced by two cellular proteins, nuclear factor I (Nagata et al., 1982; Leegwater et al., 1985; Hay, 1985; Rosenfeld and Kelly, 1986; Meisterernst et al., 1988) and nuclear factor HI (Pruijn et al., 1986; Rosenfeld et al., 1987). These proteins bind to the auxiliary region of the origin located between nucleotides 25 and 50. NFII was purified from HeLa cells as a 90-95 kd protein which is indistinguishable from the ubiquitous octamer transcription factor OTF-1 (O'Neill et al., 1988; Pruijn et al., 1989). This protein, also called oct-i, OBP100 or NF-A1, binds to the octamer element (Singh et al., 1986; Sturm et al., 1987; Fletcher et al., 1987). This element, 5'-ATGCAAAT-3', is an essential component of promoters and enhancers of cellular genes as diverse as histone H2B, Ul and U2 snRNAs and immunoglobulin light and heavy chains. Besides oct-i, several other octamer recognizing proteins have been detected. Oct-2 proteins (oct-2A, oct-2B) are expressed primarily in lymphoid cells and are presumably involved in cell typespecific immunoglobulin gene expression (Staudt et al., 1986; Landolfi et al., 1986; Lebowitz et al., 1988; Schreiber et al., 1988). Oct-3 to oct-7 have been identified in various mouse tissues at different stages of development (Scholler et al., 1989; Lenardo et al., 1989), while related proteins have been detected in adult brain (He et al., 1989) but these have not been characterized yet. Cloning and sequencing analysis of the oct-I and oct-2 cDNAs has revealed the presence of a DNA binding domain called the POU domain (Herr et al., 1988; Sturm et al., 1988; Clerc et al., 1988; Scheidereit et al., 1988; Muller et al., 1988). This 160 amino acid long domain is also present in the product of the unc-86 gene of Caenorhabditis elegans and Pit-1/GHF-1, a rat pituitary-specific transcription factor. The POU domain can be subdivided into an N-terminal 74 amino acid long POU-specific domain and a C-terminal POU homeo domain. The latter is homologous to the homeo domain found in homeotic genes (review, Scott et al., 1989) and contains a potential tri-a-helical motif thought to be involved in DNA recognition (Garcia-Blanco et al., 1989). Based upon the failure to detect DNA binding of the oct-I POU homeo domain, obtained by in vitro translation, a bipartite DNA binding structure has been proposed (Sturm and Herr, 1988). Previously it was shown by deletion analysis that for nuclear factor I (NFI) the N-terminal region, containing the DNA binding domain and the dimerization domain, but lacking the transcription activation domain, was sufficient for stimulation of adenovirus DNA replication (Mermod et al., 1989;. Gounari et al., 1990). To investigate if such a situation also applies to NFiII, we dissected the oct-I gene 1883
C.P.Verrijzer, A.J.Kal and P.C.van der
Viiet
Fig. 2. The POU domain and the homeo domain differ in binding specificity. Probes containing the X.laevis U2 enhancer (lanes 1,2,3), the Ad2 origin (lanes 4,5,6) or the Ad4 origin (lanes 7,8,9) were incubated with 6 ng purified homeo domain (lane, 2,5,8) or 2 ng purified POU domain (lanes 3,6,9), or in the absence of protein (lanes 1,4,7) followed by gel retardation analysis on a 15% polyacrylamide gel.
Fig. 1. Binding of recombinant vaccinia virus expressed oct-i, oct-I mutants and oct-2 to the octamer. An end-labelled octamer containing probe from the Ad2 origin was incubated with 1 Al cytoplasmic extract obtained from cells infected with control wild-type virus (lane 2) or recombinants expressing: oct-I (lane 3), 212-743 (lane 4), 1-23/269-743 (lane 5), 1-592 (lane 6), 1-440 (lane 7), POU domain (lane 8), oct-2 (lane 9), homeo domain crude extract (lane 10) or purified and concentrated homeo domain (1 jsl fraction 14 from the second fast flow S column, lane 11). Lane 1 shows authentic HeLa NFIII binding activity. Bound and free DNA (F) were separated by electrophoresis (see Materials and methods).
and delimited the domain required for DNA replication. We present evidence that the POU domain suffices for stimulation and that the homeo domain is capable of DNA binding, but inhibits DNA replication. This suggests a role for the POU-specific domain in DNA replication.
Results Overproduction of oct-1 and deletion mutants We employed a eukaryotic expression system which was previously used to overproduce other biologically active components of the adenovirus DNA replication system like pTP, pol and NFI (Stunnenberg et al., 1988; Gounari et al., 1990). The oct-I open reading frame as well as several deletion mutants were cloned downstream of the late 11K vaccinia promoter employing the recombination vector pATA-18. Recombinant vaccinia viruses were produced and used to infect HeLa cells in suspension. At 24 h postinfection nuclei and cytoplasm were separated. Total overproduction was 30- to 60-fold and, dependent on the construct, 20-50% of the recombinant protein was found in the cytoplasm. To prevent contamination with endogenous NFIII, we used cytoplasmic extracts. As shown by gel retardation using an Ad2 origin containing fragment as probe (Figure 1), octamer-recognizing proteins were present abundantly in all recombinant extracts containing the intact oct-I POU domain and were absent from extracts prepared from HeLa cells infected with wild-type vaccinia virus. The oct- 1 homeo domain binds to the adenovirus origin but with a reduced affinity The POU domain can be subdivided into an N-terminal 74 amino acid long POU-specific domain and a C-terminal POU
1884
homeo domain that includes a putative tri-a-helical motif present in many homeo domain containing proteins (Sturm et al., 1988). When we overexpressed the POU homeo-domain, weak binding was observed under conditions in which the POU domain binds strongly (Figure 1, compare lanes 8 and 10). Only when higher concentrations of partially purified POU homeo domain were used was a considerable amount of the probe bound (Figure 1, lane 11). These results suggest that, although the POU homeo domain itself has an intrinsic binding capacity, the POU-specific domain contributes considerably to the binding affinity and possibly the specificity. This could be achieved either by stabilization or by making additional protein -DNA contacts. To investigate this, we purified the POU domain and the homeo domain to homogeneity and compared their binding to the origins of Ad2 and Ad4 and to the Xenopus laevis U2 enhancer by gel retardation. As shown in Figure 2 a clear difference in specificity was observed. No binding to the U2 enhancer was observed for the homeo domain, while the same amount of polypeptide could bind to the Ad4 origin and even better to the Ad2 origin. For the POU domain a reverse binding order was observed: Ad4 > Ad2 or U2 in agreement with published data for the intact NFIII protein (Pruijn et al., 1987). We also compared the DNase I footprint of both polypeptides on the Ad2 origin. Figure 3 shows that the footprint of the POU domain on the bottom strand coincides with published data for NFIII but that the region protected against DNase I by the homeo domain is shorter, in particular at the 5' end of the binding site. The same results were obtained for the top strand (not shown). The difference in specificity and the different footprint borders suggest the existence of DNA contacts by the POUspecific domain. Recently we confirmed this by hydroxyl radical footprinting (C.P.Verrijzer et al., in preparation). The POU domain stimulates DNA replication We assayed the functional properties of oct-i and its deletion mutants, employing an in vitro reconstituted system for Ad DNA replication consisting of the purified viral proteins (pTP-pol, DNA binding protein) and NFI. AdS DNA-TP digested with XhoI was used as a template. Fragments B and C contain the origin. To compare the various recombinant
oct-1 POU domain stimulates DNA replication
.Im 'OL
--
OW
U,
.S
Uumi-,, .I.
Fig. 3. Comparison of the DNase I footprints of the POU domain and the homeo domain on the Ad2 origin. The bottom strand of the origin is shown. Lane 1, G+A sequencing marker (M). Lane 2, without protein. Lane 3, 90 ng homeo domain (H) added. Lane 4, 50 ng POU domain (P) added. The borders of the protected regions, determined by scanning, are given.
POU containing fragments with authentic NFIII, identical amounts of DNA binding units were added. Previous experiments established that the DNA binding activity per /tg protein, corrected for the change in molecular mass, did not differ significantly for fragments containing the intact POU domain (Figure 1 see also Sturm and Herr, 1988). Figure 4a shows a 3-fold stimulation by oct-I indistinguishable from that of HeLa NFHI (lanes 1,3). Upon addition of NFIII/oct-1, replication becomes very efficient as indicated by the appearance of labelled single-stranded B and C fragments, originating from second rounds of replication. When more NFIH/oct-1 was added, up to a 5-fold stimulation was obtained (not shown). We also tested the effects in the absence of NFI (lanes 10-18) which gave qualitatively similar results, although the basal level was reduced 10-fold. No stimulation was observed with the control extract obtained from wild-type vaccinia-infected HeLa cells (lanes 2,11), excluding the possibility that the effect was due to endogenous stimulatory proteins. Stimulation was dependent upon the presence of an intact NFIII binding site since replication of a plasmid containing a point mutation in the octamer, Ad2-pm46 (Pruijn et al., 1986), was not enhanced (Figure 4b). The same stimulation was obtained with four mutants having deletions in either the N-terminal region or the C-terminal region up to the POU domain, or with the POU domain itself. This shows that the POU domain suffices for stimulation of adenovirus DNA replication. No effect was observed with the POU-specific domain which does not bind DNA (not shown). Addition of the purified POU homeo domain inhibited replication to below the basal level without NFIII (Figure 5, lanes 1-3). This inhibition could be overcome by adding increasing amounts of NFIII (Figure 5, lanes 4,5), indicating a competitive binding to the octamer
Fig. 4. Stimulation of Ad DNA replication in vitro by POU containing proteins. (A) In vitro replication of AdS DNA-TP digested with XhoI. B and C are the origin containing fragments, ss-B and ss-C indicate ssDNA, originating from a second round of replication. Equal amounts of binding units (0.6 bu) were added either with (lanes 1-9) or without (lanes 10-18) 3 ng NFI. 1 binding unit (bu) is defined as the amount of protein that binds 1.5 fmol (50%) of an Ad2 probe in a bandshift assay. (B) The stimulation is dependent on an intact octamer. Plasmids containing the Ad2 origin were used as a template in the presence of 3 ng NFI. Ad2-pm46 contains a point mutation in the octamer (Pruijn et al., 1986). The origin containing fragments are indicated (0). Due to the protein-priming mechanism of adenovirus DNA replication, products are covalently attached to pTP, leading to a reduced mobility (arrow). The position of the replication bands of Ad2-pm46 differs slightly from that of Ad2 since different plasmid constructs were used.
sequence. A similar inhibition was observed in the absence of NFI (not shown).
Oct-2 can stimulate DNA replication In addition to the ubiquitous oct-I protein, several other tissue-specific, POU domain containing proteins have been described that are probably involved in developmental regulation. Analysis of the cDNA of one of these, the lymphoid-specific oct-2/OTF-2A, showed the presence of a POU domain that shares 88% similarity with oct-1, while the remainder of the protein shows only weak homology (Muller et al., 1988. Clerc et al., 1988; Scheidereit et al., 1988). The high similarity of the POU domains of oct-2 and oct-I raises the question whether oct-2 can also stimulate adenovirus DNA replication. To investigate this, we overexpressed oct-2 cDNA similarly to oct-I and observed strong
1885
C.P.Verrijzer, A.J.Kal and P.C.van der Viiet
for activation of DNA replication and transcription differ, in spite of the requirement for recognition of the octamer sequence in both processes. Interestingly, a similar result was obtained for NFI for which the conserved DNA binding domain, distinct from the C-terminal transactivating domain, also suffices for DNA replication (Mermod et al., 1989; Gounari et al., 1990). Two hypotheses can be envisaged to explain these results. One is that binding itself is sufficient, leading to a structural change in the origin which enables efficient formation of an initiation complex, for instance by providing access to other viral replication proteins. Alternatively, the POU domain could contain sites for direct protein -protein contacts with other replication factors, leading to formation of an initiation complex or stabilization of such a complex. At present, we cannot distinguish between these possibilities. However, we have not observed any direct interactions between NFII and NFI or between NFHI and viral proteins using cross-linking experiments (Y. M.Mul et al., in preparation).
binding (Figure 1, lane 9) as well as stimulation of DNA replication (Figure 4a, lanes 9,18), similar to NFIII/oct-1. As the POU-specific domains of oct-I and oct-2 are most conserved (one conservative change in 74 amino acids) and as the isolated homeo domain inhibits replication, this strongly suggests a role for the POU-specific domain in DNA replication.
Discussion Our data, summarized in Figure 6 indicate that stimulation of DNA replication requires only a small domain of NFIII/oct-1 to which adenovirus may have adapted during evolution. This domain does not coincide with the transcription activation domains identified in oct-I and oct-2 (Tanaka and Herr, 1990). This suggests that the mechanisms
DNA binding properties of the POU homeo domain The oct-i homeo domain, produced by in vitro translation, does not bind detectably to the octamer (Sturm and Herr, 1988). This may be explained by the low concentration of the polypeptide in such a reaction. When we assayed purified, concentrated homeo domain significant binding to the Ad2 octamer was obtained, even though the apparent binding affinity was considerably lower than that of the intact POU domain. Binding to the Ad4 or U2 octamer was even weaker and would be difficult to detect at low protein concentrations. Not only the binding affinity, but also the binding specificity of the POU domain and the homeo domain differ.
Fig. 5. The homeo domain inhibits DNA replication. Ad5 DNA-TP was used as a template in the presence of 3 ng NFI. The concentrations of NFIII and homeo domain are given in binding units.
DNA binding
Adenovirus replication
C oct-I
] 743
I1 2 12-743
1-23/269-743
I
1-592
1-440
1-23/269-440 (POU)
Il
0
369-440 (homoeo)
1-23/269-368
(POU-specific)
Oct-2
s
sn
++
+
++
+
++
+
++
+
++
+
++
+
+_
0
O
0
++-
+
Fig. 6. Localization of the oct-I domain required for adenovirus DNA replication. Oct-1, oct-I deletion mutants and oct-2 are depicted schematically. Indicated is the POU domain consisting of: the POU-specific (barred) and POU homeo (hatched) domains. Tabulated on the right is the biochemical activity. DNA binding: + + indicates an activity identical to that of oct-i, + indicates the diminished affinity of the homeo domain and 0 indicates no binding. Stimulation of adenovirus DNA replication: + indicates activity indistinguishable from that of HeLa NFIII, 0 indicates no effect and indicates an inhibition of DNA replication.
1886
oct-1 POU domain stimulates DNA replication
While the POU domain, similarly to intact NFHI/oct-1, binds most strongly to the authentic octamer sequence ATGCAAAT present in Ad4 and the U2 enhancer, the homeo domain prefers the Ad2 sequence 5'-ATGATAATGA-3'. Interestingly, this sequence contains TAATG, the consensus sequence for homeotic proteins (Scott et al., 1989; Hoey and Levine, 1988). In this respect it is noteworthy that the footprint borders of the homeo domain are shifted towards the 3' end of the NFIH recognition sequence. Recent hydroxyl radical footprinting has confirmed that the homeo domain lacks contacts at the 5' side of the Ad2 recognition sequence, suggesting that additional contacts are provided by the POU-specific domain. So far, direct binding of the POU-specific domain to DNA could not be
detected. Presently we are investigating whether the conin the core recognition sequence made by the homeo domain are identical to those made by the POU domain. For another POU protein, GHF-I or Pit-1, it has recently been reported that the homeo domain is sufficient for sequencespecific DNA binding (Theill et al., 1989). tacts
Role of the POU-specific domain in DNA replication We were surprised to find that the homeo domain inhibits DNA replication. Several other proteins can bind to the origin without an inhibitory effect, such as origin recognition protein ORP-A (Rosenfeld et al., 1987) or NFIV (de Vries et al., 1989). The inhibition can be overcome by excess NFIIH, showing that binding of the homeo domain is essential for inhibition. This implies that the POU-specific region has at least two functions: (i) contribution to the binding and (ii) stimulation of initiation of Ad DNA replication. These functions might be related, but clearly the effect of the POUspecific domain is not simply due to an increased DNA
binding affinity for two reasons. First, we added an identical amount of DNA binding units to the replication reaction. Second, when DNA binding was the only determinant for stimulation one might expect no effect of the homeo domain, rather than the observed inhibition. Therefore we consider it more likely that non-functional multi-protein DNA complexes are formed in the presence of the homeo domain or that an aberrant origin structure is induced which cannot be adequately recognized by viral replication proteins. So far, no specific role of the POU-specific region in transcriptional control has been demonstrated. Interestingly, this region is even more conserved between oct- I and oct-2 (99%) than the POU homeo domain (88%), and this may explain why oct-2 is also capable of stimulation. Presently we are investigating the effects of other, less conserved POUspecific domains from transcription factors Pit-I and Unc-86 in adenovirus DNA replication.
Materials and methods Construction of recombinant vaccinia viruses The 2.5 kb FnuDII-Hindll fragment frompBSoctl + (Sturm et al., 1988) containing the 743 amino acid long oct-I open reading frame was end repaired and blunt ligated into the filled in EcoRI site of pATA-18 leading pATA-18 is a vaccinia recombinant vector containing to the1 1K late viral promoter followed by the pUC- 18 multiple cloning site, inserted in the thymidine kinase gene (Stunnenberg et al., 1988). For mutant 212-743, the 1.9 kb AhaII -HindIL fragment from pBS octl + was cloned into pATA-32-18 +2 digested with AccI and HindIH. pATA-32-18 +2 is a vector in which translation starts from a synthetic start codon directly
pATA-18/oct-1.
of the multiple cloning site, with +2 indicating the addition of nucleotides to enable expression in the correct reading frame. This leads
upstream two
to a product containing MNWIMCRS followed by amino acids 212 -743 from oct-i. For mutant 1-23/269-743, pBSoct-1+ABH (Sturm and Herr, 1988) was digested with FnuDHI-HindIIIand blunt ligated in the filled in EcoRI site of pATA-18. For mutant 1-592, pATA-18/oct-1 was linearized with NcoI and end repaired with T4 DNA polymerase. The 1.8 kb ClaI-NcoI fragment was cloned in pATA-18-STOP linearized with ClaI and Ball. pATA-18-STOP is a pATA-18 derivative with stop codons in three reading frames directly downstream of the BalI site in the multiple cloning site. This leads to a 592 amino acid long product with an additional C-terminal Pro. Mutant 1-440 was constructed essentially as 1-592, only PflMI was used instead of NcoI, thus encoding amino acids 1-440 to which TIAE was added C terminally. Mutant 1-23/269-440 (POU) was constructed essentially as 1-440 only mutant 1-23/269-743 was used as starting material. The resulting construct encoded amino acids 1-23 and 269-440. For expression of the homeo domain, the 234 bp EcoRI-SphI fragment from mutant 1-440, encoding amino acids 369-440, was cloned in the EcoRI-SphI sites from pATA-32-18, thus donating an initiation codon in front of amino acid 369. For oct-2, the 1.9 kb EcoRI fragment containing the oct-2 cDNA (Muller et al., 1988; Clerc et al., 1988. Scheidereit et al., 1988) was cloned in the EcoRI site of pATA-18. These plasmids were used to prepare recombinant viruses as described (Stunnenberg et al., 1988).
Preparation of extracts HeLa cells (2.5 x 107) were infected with 5 p.f.u./cell and harvested 24 h post-infection. The cells were washed twice with phosphate-buffered saline (PBS) containing 0.5 mM MgCl2 and resuspended in 0.5 ml 20 mM HEPES/KOH (pH 7.5), 5 mM KCl, 0.5 mM MgCl2, 2 mM phenylmethylsulphonyl fluoride (PMSF), 0.1 % Nonidet P40. After swelling the cells were Dounce homogenized and centrifuged at low speed. The resulting cytoplasm was centrifuged for 30 min in an Eppendorf centrifuge at 15 000 r.p.m. and the supernatant (cytoplasmic extract) was mixed with an equal amount of twice concentrated buffer B (25 mM HEPES/KOH, pH 8.0, 1 mM dithiothreitol, 0.1 mM PMSF, 0.02% Nonidet P40, 20% glycerol) containing 250 mM NaCl. Protein purification NFIII was purified from HeLa nuclei as described (Pruijn et al., 1988). NFI was purified by DEAE (200 mM flowthrough), heparin-Sepharose (peak 400 mM) and fast flow S (peak 350 mM) chromatography followed by an affinity column pKB 67-88 (Jones et al., 1987) (peak 520 mM). The purification of pTP-pol and DBP was as described (Mul et al., 1989). The homeo domain was purified as follows: 5 x 108 HeLa cells were infected with the appropriate recombinant virus. Cytoplasmic extracts were adjusted to 30 mM NaCl and applied to a DEAE column equilibrated with buffer B containing 30 mM NaCI. The flowthrough was adjusted to 0.1 M NaCl and applied to a fast flow S column, washed with B/0.1 M NaCl and eluted with a linear gradient of 0.1-0.5 M NaCl. Homeo domain eluted around 0.3 M NaCl. Fractions active in gel retardation were combined, adjusted to 0.1 M and applied to a heparin-Sepharose column. After washing with B/0.1 M Nacl, the column was developed with a linear gradient of 0.1-1 M NaCl in buffer B. Homeo domain eluted around 0.5 M. Active fractions were concentrated by step elution from a second fast flow S column. All purification steps were performed on ice or at 4°C. Details of the purification of the POU domain will be presented elsewhere. Gel retardation and DNase I footprinting The following probes were used: Ad2, a 110 bp EcoRI-XbaI fragment from pHRI, containing the Ad2 origin. Ad4, an 85 bp EcoRl-XbaI fragment from p4A85A (Hay, 1985), containing the Ad4 origin. U2, a 40 bp doublestranded oligonucleotide containing the X. laevis U2 octamer (Mattaj et al., 1985). 1 Al of cytoplasmic extracts were assayed by gel retardation as described (Pruijn et al., 1987). For DNase I footprinting, the bottom strand of the 331 bp NdeI-XbaI fragment of pHRI, labelled at the 5' end of the XbaI site by T4 polynucleotide kinase, was used under conditions described previously (Pruijn et al., 1987).
DNA replication in vitro pTP-pol (50 ng) and 0.6 Ag DBP were added to the incubation mixtures (15 dl). For replication of AdS DNA-TP terminal XhoI fragments 25 ng of template was added per incubation. For replication of origin containing plasmid DNA, pHRI, was digested with EcoRI-AvaH and Xpm46G, containing the Ad2-pm46 origin, with EcoRI before use and 30 ng plasmidA DNA was added. In all cases the amount of extract was adjusted to 1 by addition of extract from wild-type vaccinia-infected cells. Further incubation conditions and analysis of the products were as described (Pruijn et al., 1988).
1887
C.P.Verrijzer, A.J.Kal and P.C.van der Vliet
Acknowledgements Winship Herr for providing us with the pBSoct-l+ as and for making data available before publication. We thank Michael Muller and Walter Schaffner for the oct-2 cDNA clone, Henk G. Stunnenberg for pATA-18 and pATA-32-18, Ron Hay for pHRI and p4A85A and lain Mattaj for X. laevis U2 enhancer DNA. Thanks are also due to Ina Pospiech and Roel Schiphof for tissue culture, to Henk Stunnenberg and Peter H.van Bragt for advice on the vaccinia work and We are
well as
grateful
to
pBSoct-l +2ABH
to Wim van Driel
pol, respectively.
This work was
Foundation for Chemical Research
Organization
Netherlands
providing purified NFIII and pTPin part by the Netherlands
and Yvonne Mul for
supported (SON)
with financial
for Scientific Research
support
from the
(NWO).
References Challberg,M.D. (1988)
and
Rev. Biochem., 58, 671-717. LeBowitz,J.M., Baltimore,D. and Sharp,P.A.
Kelly,T.J. (1989) Annu.
Dev., 2,
1570-1581.
Pamphilis,M.L. (1988) Cell, 52, 635-638. De Vries,E., Van Driel,W., Bergsma,W.G., Arnberg,A.C. and Van der Vliet,P.C. (1989) J. Mol. Biol., 208, 65-78. Fletcher,C., Heintz,N. and Roeder,R.G. (1987) Cell, 51, 773-781. De
Garcia-Blanco,M.A., Clerc,R.G. and
Sharp,P.A. (1989)
Genes
Dev., 3,
739-745.
de Francesco,R., Schmidt,J., Van der Vliet,P.C., Cortese,R. Stunnenberg,H.G. (1990) EMBO J., 9, 559-566. Hay,R.T. (1985) J. Mol. Biol. 186, 129-136. Hay,R.T. and Russell,W.C. (1989) Biochem. J., 238, 3-16. He,X., Treacy,M.N., Simmons,D.M., Ingraham,H.A., Swanson,L.W. and Rosenfeld,M.G. (1989) Nature, 340, 35-42. Herr,W., Sturm,R.A., Clerc,R.G., Curcuran,L.M., Baltimore,D. Sharp,P.A., Ingraham,H.A., Rosenfeld,M.G., Finney,M., Ruvkun,G. and Horvitz,H.R. (1988) Genes Dev., 2, 1513-1516. Hoey,T. and Levine,M. (1988) Nature, 332, 858-861. Jones,K.A., Kadonaga,J.T., Rosenfeld,P.J., Kelly,T.J. and Tjian,R. (1987)
Gounari,F., and
Cell, 48, 79-89.
Landolfi,N.F., Capra,J.D.
and Tucker,P.W. (1986) Nature, 323, 548-551. Kobayashi,T., Staudt,L., Baltimore,D. and Sharp,P.A.
Lebowitz,J.H.,
(1988)
Genes
Dev., 2, 1227-1237.
Leegwater,P.A.J.,
Van
and Van der
Driel,W.
Vliet,P.C. (1985) EMBO
J., 4, 1515-1521.
Staudt,L., Robbins,P.,
Lenardo,M.J.,
Baltimore,D. (1989) Science,
Mattaj,I.W., Lienhard,S.,
Kuang,A., Mulligan,R.C. and
243, 544-546.
Jirichy,J.
and De
Robertis,E.M. (1985) Nature,
163-167.
316,
Meisteremst,M., Gander,I., Rogge,L. and Winnacker,E.-L. (1988) Nucleic Acids Res., 16, 4419-4435. Mermod,N., O'Neill,E.A., Kelly,T.J. and Tjian,R. (1989) Cell, 58, 741-753.
Mul,Y.M.,
(1989)
Van
Miltenburg,R.T.,
Nucleic Acids
Muller,M.M.,
De
Clercq,E.
and Van der
Vliet,P.C.
Res., 17, 8917-8929.
Rupert,S., Schaffner,W.
and
Matthias,P. (1988) Nature, 336,
544-551.
Nagata,K., Guggenheimer,R.A. and Hurwitz,J. Sci.
(1982) Proc. Natl. Acad.
USA, 79, 6438-6442.
O'Neill,E.A., Fletcher,C.,
Burrow,C.R. Heintz,N.,
Roeder,R.G. and Kel-
ly,T.J. (1988) Science, 241, 1210-1213. Pruijn,G.J.M., Van Driel,W. and Van der Vliet,P.C.
(1986) Nature, 322,
656-659.
Pruijn,G.J.M., Van Driel,W., Van Miltenburg,R.T. and Van der Vliet, P.C. (1989) EMBO J., 6, 3771-3778. Pruijn,G.J.M., Van Miltenburg,R.T., Claessens,J.A.J. and Van der Vliet,P.C. (1988) J. Virol., 62, 3092-3102. Pruijn,G.J.M., Van der Vliet,P.C., Dathan,N.A. and Mattaj,I.W. (1989) Nucleic Acids Res., 17, 1845-1863. Rosenfeld,P.J. and Kelly,T.J. (1986) J. Biol. Chem., 261, 1398-1408. Rosenfeld,P.J., O'Neill,E.A., Wides,R.J. and Kelly,T.J. (1987) Mol. Cell. Biol., 7, 875-886.
Scheidereit,C., Currie,R.A.
Scholler,H.R.,
(1989)
Scott,M.P., 989,
1888
Cromlish,J.A., Gerster,T., Kawakami,K., Balmaceda,C.G., and Roeder,R.G. (1988) Nature, 336, 551-557.
Hatzopoulos,A.K., Balling,R., Suzuki,N. and
EMBO
144-158.
Staudt,L.M., Singh,H., Sen,R., Wirth,T., Sharp,P.A. and Baltimore,D. (1986) Nature, 323, 640-643. Stillman,B.M. (1989) Annu. Rev. Cell Biol., 5, 197-225.
Stunnenberg,H.G., Lange,H., Philipson,L., Van Miltenburg,R.T. and Van der Vliet,P.C. (1988) Nucleic Acids Res., 16, 2431-2444. Sturm,R.A. and Herr,W. (1988) Nature, 336, 601-604. Sturm,R., Baumruker,T., Franza,Jr.,B.R. and Herr,W. (1987) Genes Dev.,
1, 1147-1160. Sturm,R.A., Das,G. and Herr,W. (1988) Genes Dev., 2, 1582-1599. Tanaka,M. and Herr,W. (1990) Cell, 60, 375-386. Theill,L.E., Castrillo,J.L., Wu,D. and Karin,M. (1989) Nature, 342, 945-948.
Clerc,R.G., Corcoran,L.M., Genes
Schreiber,E., Matthias,P., Muller,M.M. and Schaffner,W. (1988) EMBO J., 7, 4221-4229. Singh,H., Sen,R., Baltimore,D. and Sharp,P.A. (1986) Nature, 319,
Gruss,P.
J., 8, 2543-2550.
Tamkun,J.W.
and
Hartzell,G.N.
(1989)
Biochim.
Biophys. Acta,
Received
on
February 13, 1990
