nuclear factor i (nf i) binds to an nf i-type site but not to the

THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 12, Issue of April 25, pp. 8478-8484,1992 Printed in U.S.A.

Nuclear Factor I (NF I) Binds to an NF I-type Site but Not to the CCAAT Site in the Human a-Globin Gene Promoter*

(Received for publication, August 27, 1991)

Haralabos Zorbas, The0 Rein, Anja Krause, Konstanze Hoffmann, and Ernst-Ludwig WinnackerS From the Institut fur Biochemie, Ludwig-Maximiliam-Universitat Munchen Karlstrape 23, 0-8000 Miinchen 2, Federal Republic of Germany

Methylation interference and missing contact analyses demonstrate that nuclear factor I (NF I) recognizes an NF I-like site (5’-GGG(N)6GCCAG-3’) within the a-globin promoter rather than the adjacent CCAAT box. Consistent with this, mutations within the CCAAT box do not alter significantly the affinity and specificity of the interaction whereas elimination of the 5’- GGG-3’ half-site of the recognition sequence reduces the DNA binding strength of NF I by 2 orders of magnitude down to the range of unspecific interaction. On the other hand, the mutated a-globin promoter sequence that is no longer bound by NF I, although it retains an intact CCAAT box, interacts specifically with a protein component from nuclear extracts of HeLa cells. From these results we conclude that NF I is not the factor that interacts with the CCAAT box and that the second half of the canonical 5’- TGG(N)eGCCAA-3’ NF I binding site cannot be regarded as identical with the CCAAT promoter element, as suggested previously.

Nuclear factor I (NF I)’ is a DNA-binding protein originally described as a cellular factor essential for the efficient repli- cation of adenovirus 2/5 DNA i n vitro and in vivo (1, 2). NF I interacts with the consensus sequence 5’-TGG(N)6GCCAA- 3’, which is found in viral (3-6) and cellular genomes (7-12). Often this sequence resides in regions with promoter or en- hancer function. Because of the resemblance of the second half (GCCAA) of the recognition sequence with the genetically defined CCAAT box of many eukaryotic promoters, a role as CCAAT box transcription factor for NF I has been suggested (10). Since the demonstration that NF I interacts with the CCAAT box region of the human a-globin promoter in vitro (10) and is able to direct transcription from an a-globin promoter in vitro and in vivo (10, 13, 14), NF I is regarded as CCAAT box transcription factor.

In a previous study we determined that NF I from porcine liver binds CCAAT half-sites with an affinity of at least 100 times lower than the consensus NF I site (15). Furthermore, the binding constant for the strongest half-site in the a-globin

* This work was supported by the Deutsche Forschungsgemein- schaft, Forschergruppe Virus/Zell Wechselwirkungen, Fa 124/3-3. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

This paper is dedicated to the memory of Guido Hartmann who died January 30, 1992.

$ To whom correspondence should be sent. Tel.: 49-89-85782467; Fax: 49-89-85782470.

The abbreviations used are: NF I, nuclear factor I; HEPES, 4-(2- hydroxyethy1)-1-piperazineethanesulfonic acid.

promoter (2 X 10‘ M-’) is also far below typical values for other site-specific DNA-binding proteins (see refs. in 15). Therefore, specific interaction of NF I with CCAAT boxes for a concomitant stimulation of transcription seems unlikely. Because NF I interacts with DNA as a dimer (13, 16) it has been suggested for reasons of symmetry that it is not the CCAAT box that is contacted by NF I in the a-globin promoter but a different, NF I-like sequence, located a few nucleotides upstream from the CCAAT box (15). To clarify this issue we performed a study of the NF I/a-globin promoter interaction and present in this report the results of the footprint analyses and gel retentions with mutated oligonu- cleotides of the a-globin promoter.

EXPERIMENTAL PROCEDURES

Protein Extracts-NF I purification from porcine liver is described in (15). HeLa crude nuclear extracts were prepared according to (59) with the exception that glycerol was omitted in the storage buffer. Baculovirus-expressed NF I (amino acids 1-257) was extracted from infected Spodoptera frugiperda (Sf9) cells 4 days after infection, by shaking the cells for 10 min in 50 mM HEPES/NaOH, pH 7.5, 5 mM dithiothreitol, 400 mM NaCl, 1% Nonidet P-40, 1 mM phenylmethyl- sulfonyl fluoride. After centrifugation of the cell debris, NF I was purified to homogeneity by affinity chromatography as described in (15). Escherichia coli total cytosolic extract was prepared by sonifi- cation of the cells in 50 mM Tris-C1, pH 7.5, 100 mM NaCl, 10 mM p- mercaptoethanol. Extracts and purified protein were stored at -80 “C for months without any loss of activity.

Synthetic Oligodeoxynucleotides-Synthetic oligodeoxynucleotides were purified by denaturing polyacrylamide gel electrophoresis, an- nealed in 10 mM Tris-C1, pH 7.5,5 mM NaC1, and repurified as double strands by native polyacrylamide gel electrophoresis. The concentra- tion was determined by UV absorption and comparison with molecular weight standards after ethidium bromide staining in the gel. The oligodeoxynucleotides were radioactively labeled at the 5’ ends with T4 polynucleotide kinase (New England Biolabs) and [y-3ZP]ATP at 3,000 Ci/mmol (Amersham Corp.) for gel retention analyses. The labeling efficiency was 70-loo%, where 100% yield is 6.7 X lo6 cpm/ pmol end at the calibration date.

ynucleotide L1/2, which bears a canonical NF I binding site found in DNA Footprinting Analysis-For footprinting purposes oligodeox-

mouse adenovirus FL (15), and the a-globin wild-type promoter oligodeoxynucleotide were cloned into the vector Bluescript M13- (Stratagene), giving rise to the clones designated pTFL and pTaG, respectively. The DNA strands were radioactively labeled at the 5’ and/or the 3’ ends as described in (22). Partial methylation of the DNA was performed according to the protocol in (19), and hydroxyl radical modification of the DNA was as described (21). Binding reactions were performed in 25 mM HEPES/KOH, pH 7.5, 150 mM NaCI, 0.1 pg/pl poly(dI.dC) for 15-30 min a t room temperature with 100-200 fmol of DNA fragment and enough protein to bind 50-90% of the DNA. After native polyacrylamide gel electrophoresis of the mixes the bands of the bound and free DNA were localized by autoradiography, excised, eluted, and processed as described (22) for analysis on a sequencing gel.

Gel Retention Analysis-Incubations were performed under the

protein and radioactively labeled DNA indicated in the legends of the same conditions as for the footprinting analysis with the amounts of

by guest on February 19, 2018http://w

ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

NF I Is N o t a CCAAT-binding Factor 8479

figures. Native 8% polyacrylamide gel electrophoresis (acrylam- ide:bisacrylamide = 19:l) and autoradiography were as in (22). Ra- dioactive signals were quantified by densitometric scanning of a nonsaturated x-ray film. NF I had the same affinity to the synthetic or the cloned cu-globin oligodeoxynucleotide (not shown).

RESULTS

Eukaryotic NF I from Three Different Sources Interacts with the NF I Recognition Sequence in an Indistinguishable Manner-Comparative binding studies were performed with a purified N F I preparation from porcine liver (15), HeLa crude nuclear extract which contains heterogeneous N F I proteins (lo), a recombinant N F I protein bearing amino acids 1-257 of the porcine NF I that was overproduced and purified to homogeneity from a baculovirus expression system, and a total cytosolic extract from E. coli expressing a full-length porcine NF I cDNA. DNase I footprints (not shown) and methylation interference analysis of the eukaryotic NF I proteins on the canonical NF I recognition sequence are absolutely indistinguishable (examples in Fig. 1) and identical to results of other published work (17). The bacterial product interacts also with the same site on the DNA with some minor quantitative differences (Fig. 1, lanes 7 and 11 ).

From these data we conclude the following. (a) Because porcine N F I, baculovirus-expressed N F I, and HeLa NF I bind DNA in the same manner, the presence or absence of the C terminus, the divergence of the C-terminal amino acid sequence (14), or existing post-translational modifications of the eukaryotic N F I species (18) are without relevance for the DNA interaction. (b) There might be differences in the prokaryotic polypeptide, possibly at the post-translational level, which influence the mode of the NF I/DNA interaction. (c) N F I, as expressed from a defined cDNA in insect cells or in E. coli, and no additional factor, is sufficient for a specific interaction with DNA.’ The binding experiments with the a- globin promoter in the following sections were performed a t the beginning with the porcine liver NF I and subsequently with the baculovirus-expressed N F I protein.

NF I Recognizes a n NF I-like Site (5’-GGG(N)6GCCAG-3’) in the a-Globin Promoter but Not The CCAAT Box-To identify particular DNA residues that are necessary for NF I interaction with the a-globin promoter, we performed footprint analyses of both DNA strands by two complementary methods that provide more detailed information than the DNase I footprints carried out previously (10). First, we checked the contacts to purines by methylation interference

C

D G C

A A A A C C

I ) c T

O G A

C A C

+ c + G

T T

O G A

C A

C A

A

C A

G>A (19); however, this method yields only partial information about the interacting residues. Furthermore, it has been re- ‘ n 1 - 7 ’ ported that under certain circumstances critical guanosine

.- ; ; E - residues can interact with the proteins even if they are meth- .- .- - ; ; E .-

o 2 z L l i o a “ r w o ylated (20). Therefore, we investigated this interaction also by missing contact analysis with hydroxyl radical-modified DNA (21). This method provides information about every nucleotide position. T h e a-globin promoter sequence was cloned in Bluescript M13- and labeled at the 5’ or 3’ end as described in (22). On the (+) strand three methylated Gs were found to interfere clearly with NF I binding (see Fig. 2 A , lanes 6 and 7), which are included in the sequence 5’- G@(N)&CCAG-3’ (indicated by underlining). The pattern of interference with the a-globin promoter is identical to that observed with the canonical 5’-TG3( N)&CCAA-3’ N F I site (11). T h e sequence deviation in the a-globin promoter from tha t of the canonical NF I site is very likely the reason for

c the diminished binding affinity (15). Data from the hydroxyl c radical-modified DNA entirely confirm the participation of

the above sequence in the recognition process. As shown clearly in Fig. 2B, lanes 14 and 15, only residues 5’- GGG. . . .(XX)GCCAG-3’ preclude NF I/DNA interaction when extruded by hydroxyl radicals; this sequence again matches the NF I site. Moreover, there is no indication whatsoever of an interaction with the adjacent CCAAT box.

Footprint analysis with the corresponding (-) DNA strand corroborated these conclusions. By methylation interference we found that it is the sequence 5’-T@(N)&CCCG-3’ on this strand which is recognized by NF I (interfering guanosines are underlined; see Fig. 3A, lane 5 ) , which again resem- bles the NF I consensus and is placed complementary to the 4 5 6 7 8 Y 1 0 1 1 1 2

FIG. 1. Methylation interference analysis of the interaction of NF I from different sources with the canonical NF I recognition sequence (clone pTFL, see “Experimental Proce- dures’’). The amount of NF I from porcine liver (lanes 5 and 9 ) , Hela crude nuclear extract (lanes 6 and I O ) , and bacterial total cytosolic extract (lanes 7 and 1 I ) was adjusted to give in all incubations about 90% complexed DNA. Protein bound (+, lanes 5-7) and free DNA (-, lanes 9-11) were analyzed in a sequencing gel after separation in a native retention gel and G > A cleavage of the purine- methylated DNA as described (39). Labeled, unmodified DNA ( u n ) run in lane 1. 0 are DNA references not incubated with protein; Maxam and Gilbert sequencing reactions are shown in lane 2 (pyrim- idines) and lane 3 (purines). The NF I sequence 5’-TGG(N),GCCAA- 3’ is written in bold letters. Arrows indicate interfering, and squares indicate noninterfering guanosine residues.

sequence determined for the (+j strand. We also see a slight interference of the methylated A immediately upstream of the second half-site, which was not expected because a contact in the minor DNA groove has not been reported in the literature; the significance of this point is unknown at present. However, the two Gs of the sequence 5’-ATTGG-3’, diagnostic for the CCAAT box on this strand, are absolutely irrelevant for the interaction, as they can be methylated without disturbing the N F I binding. From the results with the hydroxyl radical-

’ Protein extracts from insect cells infected with wild-t.ype baculovirus or from t.he isogenic bacterial st.rain without NF I-expressing plasmid do not cont.ain a detectable NF I activity (not shown).


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

N F I Is N o t a CCAAT-binding Factor

FIG. 2. Methylation interference ( A ) and missing contact analysis ( B ) of the interaction of NF I with the + strand of the a-globin promoter (clone pTaG, see “Experimental Procedures”). Modified DNA (purine- methylated, or hydroxyl radical-treated) from complexes with NF I (+, lanes 6 and 14) and nonbound DNA (-, lanes 7 and 15) were analyzed in a sequencing gel after separation in a native retention gel and G > A cleavage of the purine- methylated DNA. The NF I-like sequence is indicated by black arrows in 5’

3‘ direction; the CCAAT box is indicated by dotted bars. Interfering guanosine residues (panel A ) are marked by arrows; regions of sensitivity to hydroxyl radical treatment in the complexed DNA (panel B ) are indicated by brackets. Fur- ther abbreviations and symbols are as in Fig. 1.

G + G +

G I )

un Py Pu 0 + - 0 Pu Py r”l

G+

GI) G*

’ C G .G, G - L

G

C C G G G G A

OH- P y ’ O + - ‘ P y Pu G

1 2 3 4 5 6 7 8 9 1 0 1 1 ; 8

A

* un Pu Py 0 + - ..-.”-~

”

I I I ”

; ? 1 2 3 4 5 6 7 8 9 3’ 10 11 12 13 14 15

A B

FIG. 3. Methylation interference ( A ) and missing contact analysis ( B ) of the interaction of NF I with the - strand of the a-globin promoter (clone pTaG, see “Experimental Pro- cedures”). Modified DNA (purine-methylated, or hydroxyl radical- treated) from complexes with NF I (+, lanes 5 and 14) and nonbound DNA (-, lanes 6 and 15) were analyzed as described in Fig. 2. Abbreviations and symbols are as in Fig. 2.

modified DNA (Fig. 3B, lune 1 4 ) we can draw the clear conclusion that the presence of nucleotides 5’-(G)CTGG-3’, which are part of the first half-site of the NF I sequence, is absolutely required for a productive interaction, as opposed to nucleotides 5’-ATTG(G)-3’, belonging to the CCAAT box, which can be extruded without any effect. Furthermore, we

notice a slight enhancement of the intensity of the first G (underlined in the sequence above) as compared with the control (Fig. 3B, lune 13), which means that NF I binds to this DNA euen better when this position is devoid of this particular nucleoside. Such a positive selection for gapped molecules has been described (23) and is interpreted as a preferred interaction based on the enhanced flexibility of the DNA. Because NF I has been shown to bend its target DNA (22), this observation is consistent with an induced DNA structural alteration of the a-globin promoter. With the sequence 5‘-G=CG-3’, which we consider as the second half of the NF I binding site in this context, there is no striking interference with binding except of the two underlined Cs, for which the signal intensity is slightly reduced in the bound probe (Fig. 3B, lane 14). We do not know the reason for this, especially when there is methylation interference with this strand and hydroxyl radical-induced interference with the (+) strand at this site, as shown above. We can think of two possibilities to explain this phenomenon: either binding of the first half of the recognition sequence suffices for some tolerance of extruded but not methylated residues of the (-) strand which could hinder proper placement of the protein, or the picture is obscured because of overlapping effects of requirements for specific residues and for increased flexibility a t this site. Regardless of the last question, the footprinting data convincingly demonstrate that NF I interacts with an NF I-like site in the a-globin promoter with low affinity (15) but not with the genetically defined CCAAT box (results summarized in Fig. 4).

Mutations of the NF I-like Site but Not of the CCAAT Box Abolish the NF Ila-Globin Promoter Interaction-Although NF I forms stable complexes with the NF I-like sequence as demonstrated by footprints in the previous section, we won- dered if an additional or transient interaction with the CCAAT box does occur in a manner that remains undetected by the footprinting methods. We were also interested in whether a possible NF I interaction with the CCAAT box is masked in the a-globin promoter because of the presence of


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

NF I Is Not a CCAAT-binding Factor 8481

NF I site shining:

FIG. 4. Oligodeoxynucleotides used for quantitative NF I binding studies. The region of the NF I-like sequence is written in bold letters, the CCAAT box is written in italics. Mutated sequence elements are upshifted; the actual mutated residues are underlined. A summary of the methylation interference and missing contact analyses is given on the wild-type oligodeoxynucleotide; asterisks mark interfering methylated guanosines; the nucleoside regions required for the interaction as determined by the hydroxyl radical analysis are emphasized by lines; exclamation marks depict the methylated adenosine, which interferes with binding, and the particular guanosine residue, the extrusion of which enhances binding.

the NF I-like sequence. In this case a possible role of the CCAAT box as an "entry site" for NF I could be considered. For these purposes we analyzed quantitatively by gel retention experiments the interaction of NF I with an a-globin promoter, into which selected mutations were introduced which alter either the CCAAT box or the NF I sequence (cf. Fig 4). Oligonucleotide a is an up mutation of the CCAAT box, which has been shown to enhance binding of a CCAAT-binding protein, C/EBP (24). Oligodeoxynucleotide b is again an up mutation of the CCAAT box from the y-fibrinogen promoter, where CP2, another CCAAT-binding factor, binds with enhanced affinity (25). Oligodeoxynucleotide c is a down double mutant of the CCAAT box, which shuts off interaction with CP1, a CCAAT box factor (25). Oligodeoxynucleotide f rep- resents an inversion of the CCAAT box sequence, which was also expected to be a down mutation of a CCAAT box- recognizing protein; the same sequence (AACC) introduced into the second half of the consensus NF I site abolishes the interaction with NF I (probe IV 1/2, cf. Ref. 15). Oligodeox- ynucleotide g changes TCC to TGG at the appropriate distance from the CCAAT box, which is the first half of the NF I consensus sequence, thus creating a possible second sub- strate sequence for NF I, intermingled with the interacting sequence GGG(N)&CCAG. We expected this oligodeoxynucleotide to be at least as good as the wild-type a-globin sequence. Oligodeoxynucleotide k transforms the first half of the interacting sequence GGG, which has been shown here to be essential for NF 1 binding, to TCC, thus altering severely the NF I-like site and demasking the adjacent CCAAT box. If the intact CCAAT box alone does not suffice for NF I binding (as expected), this oligodeoxynucleotide should abolish or strongly reduce the interaction with NF I. Oligodeoxy-

nucleotide 1 recreates a (shifted) NF I site on the k oligodeoxynucleotide by introducing TGG at the appropriate distance from the CCAAT box. Therefore, we expected oligodeoxynucleotide 1 to interact as the wild-type a-globin sequence, similarly to oligodeoxynucleotide g.

The affinity of NF I to the a-globin wild-type and the above mutant sequences was compared by determining complex formation at increasing protein concentrations in the presence of constant DNA amounts added in excess. The specificity of the interaction was checked by competitions of the NF I/a-globin promoter complex with each of the above oligodeoxynucleotides. Examples of the autoradiographs and the quantitative results from densitometric scanning of representative retention gels are shown in Figs. 5 and 6. From the shape and titration end point of the curves shown in Figs. 5 and 6 the general conclusion can be drawn that all oligodeoxynucleotides except oligodeoxynucleotide k show an affinity to NF I which is the same or very similar to that of the wild- type sequence. There is no difference in the binding affinity among these oligodeoxynucleotides greater than a factor of 2. Neither the up nor the down mutations of the CCAAT box change the interaction with NF I significantly. Small differences that are most obvious, for example with oligodeoxynucleotide b or c, might be attributed to experimental error or to effects of the surrounding residues (26, 27). However, the mutation of the NF I-like site in oligodeoxynucleotide k drastically reduces the affinity of NF I to the a-globin promoter. We have estimated that for the same amount of complex formed between wild-type oligodeoxynucleotide and NF I one needs about 100 times more protein with the k oligodeoxynucleotide. This is in accordance with the competition curve, in which only 100 times the amount of k can compete the wild-type by about 50%. Because the equilibrium binding constant of the a-globin promoter with NF I has been determined to be 2 X 10' M" (E), we tentatively estimate the equilibrium binding constant for k to be 2 X lo6 M-', which is in the range of unspecific interaction (15).

We were surprised to notice that both oligodeoxynucleotides g and 1, which create TGG(N)CGCCAA sequences, are worse substrates for NF I than the wild type. Because it is only the newly created NF I site on oligodeoxynucleotide 1 which could interact with NF I (compare with k), this oligodeoxynucleotide thus demonstrates most clearly the significance of the overall sequence context. From the affinity and competition diagrams one is not able to decide unequivocally which site of oligodeoxynucleotide g is actually bound; we presume it to be the sequence TGG(N)6GCCAA that exclu- sively interacts with NF I on g.

From these binding studies we thus conclude that: (a) NF I is able to bind only the NF I-like sequence within the a- globin promoter; (b) there is no cryptic interaction of NF I with the CCAAT box in the a-globin promoter which significantly influences the thermodynamic equilibrium of binding; (c) the NF I-like site in the a-globin promoter does not mask an interaction of NF I with the CCAAT box.

A Protein Different from NF I Is Able to Interact with the CCAAT Box in the a-Globin Promoter-We considered the remote possibility that the mutations introduced into oligodeoxynucleotide k inactivate this sequence in a general, non- specific way as a target for DNA-binding proteins. In addition, we were interested to see if there are other protein activities than NF I in HeLa extracts which were the original source for NF I (2, lo), capable of interacting with the CCAAT box of the a-globin promoter. In this instance, oligodeoxynucleotide k is the most appropriate target for binding studies because it does not interact with NF I, and hence confusion


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

a482 NF I Is Not a CCAAT-binding Factor

A

-"--I B

- OllY_ll I, oh:n I

FIG. 5. Titration of oligodeoxynucleotides with increasing amounts of NF I. A, examples of selected oligodeoxynucleotides, analyzed by native gel electrophoresis and autoradiography. Generally, 5 fmol of oligodeoxynucleotide (about 5 X lo4 cpm) were used in each incubation reaction. B, quantification of the titration of the oligodeoxynucleotides with the indicated amounts of baculovims-expressed NF I, as determined by densitometric scanning of representative autoradiographs. Percent complex is the amount of the gel-shifted protein-bound DNA.

with overlapping DNA binding activities can be avoided. In a gel retention assay (Fig. 7) we obtain two sharp complex bands with oligodeoxynucleotide k and crude nuclear HeLa extract. The upper band can be specifically competed only with homologous (nonlabeled) competitor k; however, neither oligodeoxynucleotide lacking an intact CCAAT box can compete away the observed binding interaction, irrespective of the absence (IV 1/2, see above and Ref. 26), or the presence (c or f ) of a functional NF I site. Therefore, we suggest that there is indeed a CCAAT box-binding protein in HeLa nuclear extracts interacting with the a-globin promoter; furthermore, i t is obvious that neither half of the NF I-like sequence on the a-globin promoter is mistaken for a CCAAT box by this protein. Thus, this binding activity must be a protein entirely different from NF I. With oligodeoxynucleotides c and f (Fig. 7) a diffuse complex band is obtained, most probably by the interaction of the multiple NF I forms in HeLa cells (10) with these NF I binding targets. These complexes can be competed with oligodeoxynucleotide c or f to the same extent, but not with oligodeoxynucleotides IV 112 or k, which lack a functional NF I site, as expected.

DISCUSSION

NF I has been isolated from HeLa cells as a CCAAT- binding protein by the use of three different target oligonu-

cleotides: the adenovirus 2-ITR, the ha-ras, and the a-globin promoters (lo), of which the first two contain a canonical NF I recognition site. The present study provides evidence that in all three cases NF I-type sites rather than CCAAT boxes are being recognized by NF I.

Because the CCAAT box is a genetically defined cis-acting element (28), the theoretical possibility exists that it is utilized not as the site of stable protein binding but only as a protein recognition sequence (entry site) for a concomitant one-di- mensional diffusion along the DNA, as is known for example for prokaryotic restriction endonucleases (29). Sliding of NF I along the DNA has in fact been suggested (30). However, our results with mutations in the CCAAT box argue against the possibility that the CCAAT sequence is sensed by NF I. On the contrary, DNA binding activities that do interact with high specificity with the CCAAT box of a-globin and which are not NF I have been identified, e.g. CP1 in HeLa cells (25) and a-CP1, a factor from murine erythroleukemia cells (31). In accordance, we also discover a CCAAT binding activity in HeLa cells, presumably CP1, which does not interact with the NF I sequence (this work).

In other genes the CCAAT box has been shown similarly to interact with different proteins but not with NF I (32-37). Nevertheless, in vitro binding of NF I to some CCAAT,


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

N F I Is No t a CCAAT-binding Factor 8483

competilor b

FIG. 6. Competition of the NF I/a-globin promoter interaction by increasing amounts of mutated oligodeoxynucleotides. A, examples of the competition by the indicated nonradioactive oligodeoxynucleotides, analyzed as in Fig. 5A. The noncompeted incubation reaction (corresponding to 5 fmol of labeled a-globin promoter oligodeoxynucleotide and 10 ng of baculovirus-expressed N F I) was in the linear range of complex formation, giving about 20% bound DNA. lo-, 20-, 50-, and 100-fold molar excess of nonradioactive competitor was added to the reaction before NF I binding. B, quantification of the competition of the NF I/a-globin promoter interaction a t the indicated molar excess of the oligodeoxynucleotides, determined as in Fig. 5B. The complex amount formed in the noncompeted reaction was set at 100%.

CCAAT-like sequences, and TGG half-sites, which can be considered as constituents of an inverted CCAAT box, has been reported (10, 15, 32, 34, 38-42). However, the significance of these data can be questioned. First, the affinity of N F I to these sites is consistently lower than to the consensus sequence, as in the case of a-globin; characteristically, in the hydroxymethylglutaryl-CoA reductase promoter, the two (out of six) NF I sites with a perfect (inverted) CCAAT box bind NF I with the lowest affinity (39). Second, in the cases in which purified natural, rather than the cloned NF I protein was used, there can be no certainty as to what protein actually interacts with the target sequence, in accordance with the coactivator concept (43). Third, it is interesting to note that some of these CCAAT sequences, which act as cis-transcriptional elements, recently were shown to interact rather with proteins other than NF I, for example: the CCAAT element of the hsp 70 promoter interacts with CBF (44); the CCAAT box of the tk promoter may interact with YEBP (45); the putative NF I site of the a2(I) collagen promoter binds histone H1 (46); and site E of the mouse albumin promoter interacts with a major distinct liver species designated $-NF I (34). Therefore, in our opinion, no definitive proof for a NF I/

+competitor +cornpetitor + competitor

Is ' 'c 1 - ' ' Q - z k c f - z k c f - z k c f

I 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 14 15

oligo k oligo c oligo f

FIG. 7. Detection of CCAAT binding activities in nuclear HeLa extracts. Gel retention analyses of oligodeoxynucleotides k, c, and f (2.5 fmol, about 2-3 X lo4 cpm/incubation reaction each) were made with 2 pl of crude extracts/incubation reaction. The indicated nonradioactive competitor oligodeoxynucleotides were added a t 50-fold molar excess before the incubation reaction.

CCAAT box interaction was provided in any system as yet. Can an NF I/CCAAT box interaction be achieved by other

means? Alteration in the DNA binding modus can be a consequence of protein/protein interaction (e.g. 47, 48). Het- erodimerization of NF I, as suggested in (49), was in fact reported for an NF I preparation from HeLa cells (25), but it apparently results in an even higher stringency of the interaction of NF I with the NF I sequence, rather than with the CCAAT box of the a-globin promoter (25). Although syner- gism of NF I with other factors was recognized (50-53), physical interaction of NF I has been shown only with adenovirus polymerase (54-56) and polymerase a/primase.3 There is, however, no indication as yet that these associations change the NF I DNA binding requirements. Post-translational modifications of proteins may also alter the DNA binding specificity (57 and refs. therein; 58), but the known modifications of NF I (18) were not reported to cause any change in the DNA recognition.

If for all these reasons, NF I does not bind to the CCAAT transcriptional element, is NF I a transcription factor at all? The potential of NF I to act as a transcriptional factor by binding in vivo to a TGG(N)6GCCAA sequence was recently convincingly demonstrated by Briiggemeier et al. (3). This study was performed with the murine mammary tumor virus promoter and emphasizes the broad versatility of NF I to be able to participate in many aspects of viral DNA metabolism. Although NF I has been also implicated in binding of several cellular DNA sites, as discussed above, to our knowledge the human a-globin promoter is the cellular target, investigated in most detail as yet. However, experiments designed to demonstrate transcriptional enhancement from the a-globin promoter in vitro (10) and in vivo (13, 14) by NF I show a rather marginal stimulation despite the auxiliary NF I sites cloned in front of the a-globin gene (well below 1 order of magnitude; compare on the contrary the transcriptional stimulation from the MMTV promoter by a factor of 20 (3)). Thus, for reasons (a) of weak DNA binding, ( b ) of very weak transcriptional stimulation and (c) of specific interaction of

' I. Dornreiter, unpublished data.


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

8484 NF I Is Not a CCAAT-binding Factor

NF I with an NF I-type site rather than the CCAAT sequence, we conclude that this system is not a suitable paradigm for the cellular action of NF I. The exact function of NF I as a specific DNA-binding protein in the uninfected cell thus remains undecided; however, one can definitely conclude that its transcriptional enhancement activities are exerted as a TGG(N),GCCAA and not as a CCAAT box binding factor.

Acknowledgments-We thank Lars Rogge for splendid ideas and Ellen Fanning, Horst Ibelgaufts, Kari van Zee, Lars Rogge, and Bertram Brenig for critical reading of the manuscript and valuable comments.

REFERENCES 1. Hay, R. T. (1985) EMBO J. 4 , 421-426 2. Nagata, K., Guggenheimer, R. A., Enomoto, T., Lichy, J. H., and

Hurwitz, J. (1982) Proc. Natl. Acad. Sci. U. S. A. 7 9 , 6438- 6442

3. Briiggemeier, U., Rogge, L., Winnacker, E.-L., and Beato, M.

4. Hennighausen, L., and Fleckenstein, B. (1986) EMBO J. 5,1367-

5. Leegwater, P. A. J., van Driel, W., and van der Vliet, P. C. (1985)

6. Nowock, J., Borgmeyer, U., Puschel, A. W., Rupp, R. A. W., and Sippel, A. E. (1985) Nucleic Acids Res. 13 , 2045-2061

7. Borgmeyer, U., Nowock, J., and Sippel, A. E. (1984) Nucleic Acids Res. 12,4295-4311

8. Gronostajski, R. M., Nagata, K., and Hunvitz, J. (1984) Proc. Natl. Acad. Sci. U. S. A. 81,4013-4017

9. Hennighausen, L., Siebenlist, U., Danner, D., Leder, P., Rawlins, D., Rosenfeld, P., and Kelly, T., Jr. (1985) Nature 314 , 289- 292

10. Jones, K. A., Kadonaga, J. T., Rosenfeld, P. J., Kelly, T. J., and Tjian, R. (1987) Cell 48 , 79-89

11. Jose-Estanyol, M., and Danan, J.-L. (1988) J . Biol. Chem. 2 6 3 ,

12. Siebenlist, U., Hennighausen, L., Battey, J., and Leder, P. (1984)

13. Mermod. N.. O’Neill. E. A., Kelly. T. J., and Tiian, R. (1989) Cell

(1990) EMBO J. 9,2233-2239

1371

EMBO J. 4 , 1515-1521

10865-10871

Cell 37 , 381-391

5 8 , 741-753

Nature 334, 218-224

. - .

14. Santoro, C., Mermod, N., Andrews, P. C., and Tjian, R. (1988)

15. Meisterernst, M., Gander, I., Rogge, L., and Winnacker, E.-L.

16. Meisterernst, M., Rogge, L., Foeckler, R., Karaghiosoff, M., and

17. de Vries, E., van Driel, W., van den Heuvel, S. J. L., and van der

18. Jackson, S. P., and Tjian, R. (1988) Cell 5 5 , 125-133 19. Siebenlist, U., Simpson, R. B., and Gilbert, W. (1980) Cell 20,

20. Williams, T., and Tjian, R. (1991) Genes & Den 5,670-682 21. Hayes, J. J., and Tullius, T. D. (1989) Biochemistry 2 8 , 9521-

22. Zorbas, H., Rogge, L., Meisterernst, M., and Winnacker, E.-L.

23. Schreck, R., Zorbas, H., Winnacker, E.-L., and Bauerle, P. A.

24. Graves, B. J., Johnson, P. F., and McKnight, S. L. (1986) Cell

25. Chodosh, L. A., Baldwin, A. S., Carthew, R. W., and Sharp, P. A.

26. Gronostajski, R. M. (1986) Nucleic Acids. Res. 14,9117-9132

(1988) Nucleic Acids Res. 16, 4419-4435

Winnacker, E.-L. (1989) Biochemistry 28,8191-8200

Vliet, P. C. (1987) EMBO J. 6 , 161-168

269-281

9527

(1989) Nucleic Acids Res. 17 , 7735-7748

(1990) Nucleic Acids Res. 18,6497-6502

44,565-576

(1988) Cell 5 3 , 11-24

27. Gronostajski, R. M. (1987) Nucleic Acids. Res. 15,5545-5559 28. Mellon, P., Parker, V., Gluzman, Y., and Maniatis, T. (1981) Cell

29. Yuan, Y. (1981) Annu. Reu. Biochem. 50, 285-315 30. van der Vliet, P. C., Claessens, J., de Vries, E., Leegwater, P. A.

J., Pruijn, G. J. M., van Driel, W., and van Miltenburg, R. T. (1988) in Cancer Cells (Kelly, T., and Stillman, B., eds) vol. 6, pp. 61-70, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

31. Kim, C. G., and Sheffery, M. (1990) J. Biol. Chem. 265 , 13362- 13369

32. Cereghini, S., Raymondjean, M., Carranca, A. G., Herbomel, P., and Yaniv, M. (1987) Cell 50,627-638

33. Dorn, A., Bollekens, J., Staub, A., Benoist, C., and Mathis, D. (1987) CeZl 50, 863-872

34. Lichsteiner, S., Wuarin, J., and Schibler, U. (1987) Cell 5 1 , 963- 973

35. Maity, S. N., Golumbek, P. T., Karsenty, G., and de Crom- brugghe, B. (1988) Science 241,582-585

36. Oikarinen, J., Hatamochi, A., and de Crombrugghe, B. (1987) J. Biol. Chem. 262,11064-11070

37. Raymondjean, M., Cereghini, S., and Yaniv, M. (1988) Proc. Natl. Acad. Sci. U. S. A. 8 5 , 757-761

38. Bradshaw, M. S., Tsai, M.-J., and O’Malley, B. W. (1988) J. Biol. Chem. 263,8485-8490

39. Gil, G., Osborne, T. F., Goldstein, J. L., and Brown, M. S. (1988) J. Biol. Chem. 263 , 19009-19019

40. Jones, K. A,, Luciw, P. A., and Duchange, N. (1988) Genes & Deu. 2,1101-1114

41. Paonessa, G., Gounari, F., Frank, R., and Cortese, R. (1988)

42. Rossi, P., Karsenty, G., Roberts, A. B., Roche, N. S., Sporn, M.

43. Pugh, B. F., and Tjian, R. (1990) Cell 6 1 , 1187-1197 44. Lum, L. S. Y., Sultzman, L. A., Kaufman, R., Linzer, D. I. H.,

and Wu, B. J. (1990) Mol. Cell. BioZ. 1 0 , 6709-6717 45. Zeleznik-Le, N. J., Azizkhan, J. C., and Ting, J. P.-Y. (1991)

Proc. Natl. Acad. Sci. U. 5’. A. 8 8 , 1873-1877 46. Ristiniemi, J., and Oikarinen, J. (1989) J. Biol. Chem. 264,2164-

2174 47. Urban, M. B., and Bauerle, P. A. (1990) Genes & Deu. 4, 1975-

1984 48. Aprhys, C. J. M., Ciufo, D. M., O’Neill, E. A., Kelly, T. J., and

Hayward, G. S. (1989) J. Virol. 6 3 , 2798-2812 49. Gil, G., Smith, J. R., Goldstein, J. L., Slaughter, C. A., Orth, K.,

Brown, M. S., and Osborne, T. F. (1988) Proc. Natl. Acad. Sci.

50. Schule, R., Muller, M., Kaltschmidt, C., and Renkawitz, R. (1988)

51. Strahle, U., Schmid, W., and Schutz, G. (1988) EMBO J. 7,3389-

52. Cleat, P. H., and Hay, R. T. (1989) EMBOJ. 8,1841-1848 53. Stuiver, M. H., and van der Vliet, P. C. (1990) J. Virol. 65,379-

54. Bosher, J., Robinson, E. C., and Hay, R. T. (1990) New Biol. 2,

55. Mul, Y. M., Verrijzer, C. P., and van der Vliet, P. C. (1990) J.

56. Chen, M., Mermod, N., and Horwitz, M. S. (1990) J. Biol. Chem.

57. Hoss, A., Moarefi, I. F., Fanning, E., and Arthur, A. K. (1990) J.

58. Kamata, N., Jotte, R. M., and Holt, J. T. (1991) Mol. Cell. Biol.

59. Dignam, J. D., Lebovitz, R. M., and Roeder, R. G. (1983) Nucleic

27,279-288

EMBO J. 7,3115-3123

B., and de Crombrugghe, B. (1988) Cell 62,405-414

U. S. A. 85,8963-8967

Science 242,1418-1420

3395

386

1083-1090

Virol. 6 4 , 5510-5518

265,18634-18642

Virol. 64, 6291-6296

11,765-772

Acids Res. 11 , 1475-1489


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/

H Zorbas, T Rein, A Krause, K Hoffmann and E L Winnackerhuman alpha-globin gene promoter.

Nuclear factor I (NF I) binds to an NF I-type site but not to the CCAAT site in the

1992, 267:8478-8484.J. Biol. Chem.

http://www.jbc.org/content/267/12/8478Access the most updated version of this article at

Alerts:

When a correction for this article is posted•

When this article is cited•

to choose from all of JBC's e-mail alertsClick here

http://www.jbc.org/content/267/12/8478.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at


ww

.jbc.org/D

ownloaded from

http://www.jbc.org/content/267/12/8478

http://www.jbc.org/cgi/alerts?alertType=citedby&addAlert=cited_by&cited_by_criteria_resid=jbc;267/12/8478&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/267/12/8478

http://www.jbc.org/cgi/alerts?alertType=correction&addAlert=correction&correction_criteria_value=267/12/8478&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/267/12/8478

http://www.jbc.org/cgi/alerts/etoc

http://www.jbc.org/content/267/12/8478.full.html#ref-list-1

http://www.jbc.org/

nuclear factor i (nf i) binds to an nf i-type site but not to the

Documents