CA1340999C - Nuclear factors associated with transcriptional regulation - Google Patents

Abstract

Constitutive and tissue-specific protein factors which bind to transcriptional regulatory elements of Ig genes (promoter and enhancer) are described. The factors were identified and isolated by an improved assay for protein-DNA binding. Genes encoding factors which positively regulate transcription can be isolated and employed to enhance transription of Ig genes.

Description

134pggg NUCLEAF~_FACTORS_ASSOCIATED-WITH
TRAPdSCRIP:TIONAL-REGULATION
Field of the Invention This invention is in the field of molecular biology.
Background-of_the_Invention Traps-acting factors that mediate B cell specific transcription of immunoglobulin (Ig) genes have been postulated based on an analysis of the :LO expression of exogenously introduced Ig gene recom-binants in lymphoid <ind non-lymphoid cells. Two B
cell-specific, cis-acting transcriptional regulatory elements have been identified. One element is located in the intron between the variable and :L5 constant regions of both heavy and kappa light chain genes and acts as a transcriptional enhancer. The second element is found upstream. of both heavy chain 1340ggg and kappa light chain gene promoters. This element directs lymphoid-specific transcription even in the presence of viral enhancers.
Mouse and human light chain promoters contain the octamer sequence ATTTGCAT approximately 70 base pairs upstream from the site of initiation. Heavy chain gene promoters contain the identical sequence in inverted orientation, ATGCAAAT, at the same position. This element. appears to be required for 1C1 the efficient utilization of Ig promoters in B cells.
The high degree of sequence and positional conser-vation of this element as well as its apparent functional requirement suggests its interaction with a sequence-specific transcription factor but no such 1~ factor has been identified.
Disclosure of the Invention This invention pertains to human lymphoid-cell nuclear factors which bind to gene elements asso-ciated with regulation of the transcription of Ig 20 genes and to methods for identification and for isolation of such factors. The factors are involved in the regulation of transcription of Ig genes. The invention also pertains to the nucleic acid encoding the regulatory factors, to methods of cloning factor-25 encoding genes and to methods of altering transcrip-tion of Ig genes in lymphoid cells or lymphoid derived cells, such as hybridoma cells, by trans-fecting or infecting cells with nucleic acid encoding the factors.

1340999 ' Four different factors which bind to tran-sciptional regulatory DNA elements of Ig genes were identified and isolated in nuclear extracts of lymphoid cells. Two of the factors, IgNF-A and E, are constitutive; two IgNF-B and rc-3 (hereinafter NF-~cB) are lymphoid cell specific. Each factor is described below.
IgNF=A (NF-A1) IgNF-A binds to DNA sequences in the upstream lQ regions of both the mu.rine heavy and kappa light chain gene promoters and also to the murine heavy chain gene enhancer. The binding is sequence specific and is probably mediated by a highly con-served sequence motif, ATTTGCAT, present in all three transcriptional elements. A factor with binding specificity similar to IgNF-A is also present in human HeLa cells indicating that IgNF-A may not be tissue specific.
E factors The E factors are expressed in all cell types and bind to the light and heavy chain enhancers.
IgNF=B (NF-A2) IgNF-B exhibits the same sequence-specificity as IgNF-A; it binds to upstream regions of murine heavy and kappa light chain gene promoters and to murine heavy chain gene enhancer. This factor, however, is lymphoid specific; its is restricted to B and T cells.

Kappa-3 (NF-rcB) NF-~cB binds exclusively to the kappa light chain gene enhancer (the sequence TGGGGATTCCCA). It is specific to B-lymphocytes (B-cells) and also appears to be B-cell stage specific.
The factors were identified and isolated by means of a modified DNA binding assay. The assay has general applicability for analysis of protein DNA
interactions in eukaryotic cells. In performing the l0 assay, DNA probes embodying the relevant DNA elements or segments thereof are incubated with cellular nuclear extracts. The incubation is performed under conditions which allows the formation of protein-DNA
complexes. Protein-DNA complexes are resolved from uncomplexed DNA by electrophoresis through polyacryl-amide gels in low ionic strength. buffers. In order to minimize binding of protein in a sequence non-specific fashion a competitor DNA species can be added to the incubation mixture of the extract and DNA probe. In the present work with eukaryotic cells the addition of alternating copolymer duplex poly(dI-dC)-poly(dI-df) as a competitor DNA species provided for an enhancement of sensitivity in the detection of specific protein-DNA complexes and facilitated detection of the regulatory factors described herein.
This invention also pertains to the genes encoding the four factors associated with tran-scriptional regulation.. The invention also pertains 1340ggg -S-to a method of cloning; DNA encoding the transcrip-tional regulatory factors. The method involves screening for expression of the part of the binding protein with binding-site, DNA probes. Identifi-cation and cloning of the genes can also be ac-complished by conventional techniques. For example, the desired factor can. be purified from crude cellu-lar nuclear extracts. A portion. of the protein can then be sequenced and with the sequence information, 1() oligonucleotide probes can be constructed and used to identify the gene coding the factor in a cDNA library.
Genes encoding the regulatory factors can be used to alter cellular transcription. For example, positive acting lymphoid specific factors involved in l~~ Ig gene transcription can be inserted into Ig-producing cells in multiple copies to enhance Ig production. Genes encoding tissue specific factors can be used in conjunction with genes encoding constitutive factors, where such combinations are 2() determined necessary or desireable. Modified genes, created by, for example, mutagenesis techniques, may also be used. Further, the sequence-specific DNA
binding domain of the factors can be used to direct a hybrid or altered protein to the specific binding 2~; site.
DNA sequences complimentary to regions of the factor-encoding genes can be used as DNA probes to determine expression of the factors for diagnostic purposes and to help identify other factor-encoding genes. Antibodies can be raised against the factors which can also be usedl as probes for factor expres-sion. In addition, the cloned genes permit develop-ment of assays to screen for agonists or antagonists of gene expression andl/or of the factors themselves.
Brief-Description_of_t.he_Drawings Figure la is a schematic depiction of the 5' region of the MOPC 41 V gene segment; Figure lb is an autoradiograph of g;el electrophoresis DNA binding assays with the SfaNI-SfaNI x promoter fragment of the MOPC 41 V gene; a.nd Figure lc is an autoradio-graph of gel electrophoresis DNA. binding assay with overlapping rc promoter' fragments.
Figure 2 show autoradiographs of binding competi-1!p tion analysis in nuclear extracts of human (a) EW and (b) HeLa nuclear extracts.
Figure 3 shows th.e results of DNase I foot printing analysis of factor-DNA complexes.
Figure 4a shows the nucleotide sequences of 2() actual and putative binding sites of IgNF-A; Figure 4b is an autoradiograp~h of binding assays with various DNA probes of three Ig transcriptional control elements.
Figure 5a shows t:he DNA sequence of the promoter 2_'i region of MODC41; Figure 5b shows an autoradiograph of RNA transcript generalized in whole cell extracts made from human B lymphoma cell lines RAMOS and EW
and from HeLa cells from the indicated templates.

_7_ Figure 6 shows an autoradiogarph of RNA tran-scripts from templated. containing an upstream dele-tion.
Figure 7 is a rad.ioautograph of the binding of B
cell nuclear extract to the MOPC-41 rc promoter region showing the IgNF-A and. IgNF-B complexes.
Figure 8 shows th.e binding of T cell and non-lymphoid cell nuclear extracts to the MOPC-41 ~c promoter region.
1(l Figure 9a shows a. restriction map of the u-enhancer; Figure 9b shows an autoradiograph binding assay carried out with. ~-enhancer fragments.
Figure l0a shows a restriction map of the ~e300 fragment; Figure lOb shows complexes formed by a~
l~; various subfragments of ~s300; Figure lOc and lOd~show competition binding assays with the subfragment X70.
Figure 11 and llb show a location of binding sites in X50 and x.70 by the methylation interference technique; Figure llc provides a summary of these 2p results.
Figure 12a and 12b show an autoradiograph of binding complexes formed with X50 and X70 in B-cell and non B-cell extracts.
Figure 13a is a restriction map of rc enhancer;
2_'. Figure 13b shows a autoradiograph of binding assays with ~c-enhancer fragments; Figure 13c and 13d show an autoradiograph of competition assays with x-enhancer fragments.
Figure 14 shows location of NK-rcB binding by 30 methylation interference experiments.

-8_ Figure 15a shows binding analysis of NK-xB in various lymphoid and non-lymphoid cells; Figure 15b shows the binding analysis of NK.-xB in cells at various stages of B-cell differentiation.
Figure 16 shows the agtll-f~BNA-1 (aEB) recom-binant and the oriP prabe.
Figure 17 shows t:he sequence of the DNA probe used to screen for an H2TF1 and NF-xB expression.
Figure 18a shows the nucleotide sequence of the lQ oct-2 gene derived from cDNA and the predicted amino acid sequence of encof~ed proteins.
Figure 18b shows the nucleotide sequence of the 3' terminus and predicted the amino acid sequence of the C-terminus derived. from clone pass-3.
l~~ Figure 18c is a schematic representation of the amino acid sequence deduced from oct-2 gene derived cDNA.
Figure 19 is a schemetac representation of expression plasmid pBS-ATG-oct-2.
2I) Figure 20 shows amino acid sequence alignment of the DNA binding domain. of oct-2 factor with homeo-boxes of several other genes.
Best_Mode-of-CarrYiny_.out-the_Invention The transcription.al regulatory factors described herein can be broadly classified as constitutive (nonlymphoid) or tissue (lymphoid) specific. All factors are believed to play a role in transcription of Ig genes. Constitutive factors may have a role in regulating transcription of other genes (as might the lymphoid-specific factors) as they are found in non-lymphoid cells.
The presence of constitutive factors rendered the detection of tissue specific factors more dif ficult. To facilitate detection., a sensitive DNA
binding assay described below, was employed in all studies.
The characteristics of the transcriptional regulatory factors IgNiF-A, E, Ig;NFB and Kappa-3 (or NF-~cB) are summarized in the chart below.
Fact_o_r _Ig_Regulat:ory_Seguence desi nation ___ Promoter Enhance_r: _Lymphoid Nonlymphoid UE____KE; ____ 1-'i IgNF-A + + + - + +
(NF-A1) E factors - - + + + +
IgNF-B + + + - + -(NF-A2) 2C1 Kappa-3 - - - + + -(NF-~cB) As indicated in t:he chart, IgNF-A binds to Ig regulatory DNA elements in the region of mouse heavy and kappa light chain gene promoters and also to 2'~ mouse heavy chain gene enhancer. DNAase I footprint analysis indicates that the binding is mediated by the octamer sequence (ATTTGCAT) which occurs in mouse and human light chain gene promoters approximately 70 1340ggg base pairs upstream from the site of initation and in heavy chain gene promoters at about the same position (in inverted sequence).
Deletion or disruption of the IgNF-A binding site in Ig promoters significantly reduces the level of accurately initiated transcripts in vivo. See, e.g., Bergman, Y. et a:l., Proc. Natl. Acad. Sci. USA, 81: 7041-7045 (1984); Mason, J.O. et al. Cell 41 479-487 (1985). As demonstrated below (See Exemplifi-1() cation), this also occurs in an in vitro transcrip-tion system. IgNF-A appears to be a positive tran-sacting factor.
IgNF-A binding site appears to be a functional component of the B-cell-specific Ig promoter. For example, sequences from this promoter containing the IgNF-A binding site specify accurate transcription in B-cells but not in Hela cells. IgNF-A however, may not be restricted to E.-cells because a factor was detected in Hela cell extracts which generated 2() complexes with similar mobilities and sequence specificity (as tested by competition analysis).
Interestingly, the Ig octamer motif in the IgNF-A
binding site has recently been shown to be present in the upstream region (a.bout 225 bp) of vertebrate U1 and U2 snRNA genes. More importantly, this element dramatically stimulates (20 to 50 fold) transcription of U2 snRNA genes in Xenopus oocytes. Therefore, IgNF-A may be a constitutive activator protein that also functions in the high level expression of U1 and 3() U2 snRNA genes in vertebrate cells.

1340ggg The presence of an IgNF-A binding site in the mouse heavy chain enhsincer suggests the additional involvement of IgNF-A in enhancer function. It is known that deletion of an 80 by region of the en-hancer containing the putative binding site reduces enhance activity approximately tenfold. The oc-cupation of the binding site, in vivo, has been inferred from the fact. that the G residue in the enhancer octamer is protected from dimethyl sulfate modification only in cell of the B lineage. Further-more, IgNF-A also binds in a sequence-specific manner to the SV40 enhancer (J. Weinberger, personal com-munication), which contains the Ig octamer motif, thereby strengthening the notion. that the factor participates in enhancer function.
The E factors are constitutive factors which binds to the Ig light and heavy chain enhancer.
Factor IgNF-B binds to the same regulatory elements as IgNF-A. The binding site for IgNF-B is the octamer motif. In. contrast to IgNF-A, IgNF-B is lymphoid cell specific. It was found in nuclear extracts from pre-B, mature B and myeloma cell lines and in nuclear extracts from some T cell lymphomas.
IgNF-B was undetectable in nuclear extracts of several non-lymphoid cells. The gene encoding IgNF-B
has been cloned (oct-2 clone below) and the its nucleotide sequence has been determined. See Figure 18A.
NF-~cB binds only to the Ig light chain enhancer.
The binding is mediated by the sequence TGGGATTCCCA.

The factor is lymphoid cell specific and also lymphoid stage specific; it is expressed only by mature B-cells. In this case, it is a marker of B
cell maturation (the factor can be used to type B
cell lymphomas).
The transcriptional regulatory factors described above were identified in extracts of cellular nuclear protein by means of an improved gel electrophoresis DNA binding assay with enhanced sensitivity. This improved assay is a modification of an original assay based on the altered mobility of protein-DNA com-plexes during gel electrophoresis. The original assay has been extensively employed in equilibrium and kinetic analyses of purified prokaryotic gene regulatory proteins. See, e.g., Fried, M. and Crothers, D.M., Nucl. Acid Res. 9: 6505-6525 (1981);
Garner, M.M. and Revzin A., Nucl. Acids Res. 9:
3047-3060 (1981). More recently it has been used to identify and isolate a protein that binds to satel-lite DNA from a nuclear extract of eukaryotic cells (monkey cells). See Strauss, R. and Varshavsky, A., Cell 37: 889-901 (1984). In the latter study an excess of heterologous competitor DNA (E. coli) was included with the specific probe fragment to bind the more abundant, sequence non-specific DNA binding proteins in the extract:.
In the improved assay of this invention, the simple alternating copolymer, duplex poly(dI-dC)-poly(dI-dC) was used as the competitor DNA species.
The use of this copolymer as competitor resulted in an enhancement of sensitivity for detection of specific protein-DNA complexes.
The assay is performed essentially as described by Strauss and Varshavky, supra, except for the addition of the poly{dlT-dC)-poly(dI-dC). An extract of nuclear protein is prepared, for example, by the method of Dingnam, J.D. et al., Nucl. Acids Res.
11:1475-1489 (1983). The extract is incubated with a radiolabled DNA probe that is to be tested for 1Q binding to nuclear protein. The incubation is carried out in the presence of the poly(dI-dC)-poly-(dI-dC) competitor in a physiological buffer. DNA
protein complexes are resolved from free DNA probes by electrophoresis through a polyacrylamide gel in a low ionic strength buffer and visualized by auto-radiography.
In a preferred embodiment, protein samples (about 10 ~.g protein) are incubated with approxi-mately 10,000 cpm (about 0.5 rig) of an end-labeled 2C1 32P double-stranded DNA probe fragment in the pres-ence of about 0.8 - 4 qg poly(dI-dC)-poly(dI-dC) (Pharmacia) in a final volume of about 25 ~1.
Incubations are carried out at 30° for 30-60 minutes in 10 mM Tris HC1 (pH 7.5), 50 mM NaCl, 1 mM DTT, 1 25~ mM EDTA. Protein-DNA complexes are resolved on low-ionic strength polyacrylamide gels. Samples are layered onto low ionic-strength 4$ polyacrylamide gels (0.15x16 cm; acryl.amide:bisacrylamide weight ratio of 30:1). Gels are pre-electrophoresed for 30 about 30 min at 11 V/cm in buffer consisting of 6.7 ~ 340999 mM TrisHCl, (pH 7.5), 3.3 mM NaOAc, and 1 mM EDTA.
Buffer is recirculated between compartments. Gels are electrophoresed at the same voltage at room temperature, transferred to Whatman 3MM, dried and autoradiographed.
-The enhanced sensitivity of the assay is il-lustrated by initial experiments leading to iden-tification of the factor IgNF-A. A radiolabeled SfaNI-SfaNI DNA fragment derived from the upstream 1G region of the MOPC 41 ~c light chain gene (Figure la) was incubated with a nuclear extract of a human B
cell line in the absence or in the presence of E.
coli chromosomal DNA or poly(dI-dC)-poly(dI-dC). The resulting complexes were resolved from the free 1~, fragment by electrophoresis through a low ionic strength, non-denaturing polyacrylamide gel and visualized by autoradiography (Figure lb). In the absence of competitor DNA all of the labeled fragment was retained at the top of the gel (lane 1) probably 2G due to the binding of an excess of non sequence-specific proteins. With addition of increasing amounts of either poly(dI-dC)-poly(dI-dC) (lanes 2-6) or E. coli chromosomal DNA (lanes 7-11) as competi-tors, putative protein-DNA complexes which migrated 2G, slower than the free fragment were detected. The relative abundance of the major species of complex (B) as well as that of minor species was signifi-cantly greater in the presence of the alternating copolymer competitor DNA.

The use of a sensitive gel electrophoresis DNA
binding assay in conjunction with the copolymer competitor poly(dI-dC)-poly(dI-d.C) facilitated the identification of the regulatory factors described herein. The simple alternating copolymer probably competes less effectively than h.eterologous DNA
sequences for binding of a sequence-specific factor, thereby significantly increasing; the sensitivity of the assay. The assay has general applicability for elucidation of mammalian gene regulatory proteins.
A further increase in sensitivity in this assay is obtained by the use. of small DNA probes (about 100 by or less) which minimize non-specific binding interactions in a crude extract.
1_'. Employing this assay, binding competition tests can be perfomed to analyze the sequence specificity of protein-DNA interactions. For this purpose, an unlabeled DNA fragment. to be examined for competitive binding to the protein, factor can be added to the incubation mixture of protein extract and labeled DNA
probe (along with the poly(dI-dC)-poly(dI-dC)). The disappearance of protein-DNA probe complex, or its diminishment, indicates that the unlabeled fragments compete for binding of the protein factor. In addition, relative binding affinity of the protein to a probe sequence can be assessed by examining the ability of a competitor to displace the protein at varying concentrations.
In conjunction with the competition assays, DNase I footprint analysis (See Galas, D. and Schmitz A., Nucl. Acids Etes. 5: 3157-3170 (1978) and Exemplification below) and methylation interference experiments (See, e.g., Ephrussi, A. et al., Science 227: 134-140 (1985) can be used to refine analysis of the binding domain of the protein factors.
The functional role of the factors in the regulation of transciption can be assessed in several ways. A preferred technique for lymphoid cell factors entails the use of the in vitro transcription lQ system developed from cells of lymphoid cell lineage.
This system is described in detail in the Exempli-fication section below. The function of a factor can be indirectly assessed in this system by employing templates for transcription which carry deletions of l~~ the binding domain of the factor. As has been noted above, deletion of the upstream sequence located between -44 and -79 by from the cap site of the MOPC41 ~c gene disrupts transcription in this system (This has also been noted in in vivo systems). The 2p deleted region includes the IgNf-A binding site.
This indicates that transcription of the template is dependent upon the factor - binding site and, inferentially, upon the factor itself.
A direct way to assess the function of the factors is to show that transcription can be modu-lated by removal and replacement of the factor in the in vitro transcription system with an appropriate template. For example, the intact MOPC41 ~c promoter gene can be used as a template in the in vitro system 3p described and transcription of this template can be 13~0~99 assessed in the presence and absence of a factor (for instance, NF-~cB, a lymphoid specific factor). The factor can be removed from the lymphoid cell extract by chromatographic fractionation and then replaced.
If the level of transcription is diminished in the absence and restored by replacement of the factor, a direct indication of the factors involvement in tran-scription is provided.
In the alternate approach, antisera or mono-1() clonal antibody can be raised against a purified or enriched preparation of the factor. The antibody can be used to probe for expression of the factor in a library of cDNA of cells known to express the factor.
The genes encoding transcriptional regulatory factors can be isolated by a novel method for cloning genes that encode sequence-specific DNA binding proteins. The method involves screening a library of recombinant expression. vectors for expression of the factor with a DNA probe comprising the recognition 2(I (binding) site for the factor. Expression of the factor is identified by the presence of complex between the DNA probe and the expressed binding protein. The approach has general applicability to the cloning of sequence-specific DNA binding pro-teins.
According to the method, an expression library is created by inserting DNA (e. g., cDNA from a cell which expresses the sequence specific binding pro-tein) into an appropriate expression vector to 3C establish an expression library. A preferred expression vector is the bacteriophage agtll which is capable of expressing foreign DNA inserts within E.
coli. See e.g., Young, R.A. and Davis, R.W, in Genetic_En~ineerin~: Principles_and_Technigues, vol 7 (eds Setlow, J. & Holla.ender, A.) 29-41 (Plenum, New York 1985). Alternatively, plasmid vectors may be used.
The expression library is screened with a binding-site, DN.A probe. The probe comprises the DNA
1Q sequence recognized by the binding protein, for example, an appropriate transcriptional regulatory element such as the actamer or rc-element. In preferred embodiments the probe is less than 150bp in length to reduce nonspecific binding. The probe can be an oligomer of the binding site. Multiple copies of the site provide for multiple protein binding to the probe. The DNA probe is a labeled DNA. The preferred label is 32P.
The binding site probe is incubated with host cell protein under conditions which allow the probe to complex with the any cognate binding protein expressed in the cell. The formation of such com-plexes are determined by detecting label associated with the protein. In a preferred mode, the screening is performed by generating a replica of host cell lysates and by screening the replicated protein with the probe. For example, when the bacteriophage agtll is used, recombinant viruses are plated in arrays onto a lawn of E. coli and replica of the resulting viral plaques is made by transferring plaque protein onto an appropriate adsorbtive surface (e. g. protein replica filters). The adsorbed plaque protein is contacted with the probe under conditions which permit the formation of complexes between adsorbed protein and the probe. The replica is then washed to remove unbound probe and then examined for associated label. The protein can be examined autoradio-graghically for the presence of label.
In other embodiments, a nonspecific competitor DNA can be used along with the recognition site probe to reduce nonspecific binding to the probe. Examples of such DNA include poly (dI-dC)-poly(dI-dC) and denatured calf thymus DNA. In addition, the protein-l~~ probe complexes can be stabilized covalently for detection by, for example, uv irradiation.
This method of screening for sequence specific binding proteins is dependent, inter alia, upon:
i) the functional expression of the binding 2() domain of the desired binding protein in the host cell;
ii) a strong and selective interaction between the binding domain and the DNA probe; and iii) a sufficiently high level of expression of 2.'i the binding protein.
These parameters can be optimimized for different proteins by routine experimentation. Some factors relevant to such optimization are discussed in detail in the exemplification of the cloning of transcrip-tional regulatory factors NF-rcB and NF-A2 given below.
Other modes of cloning the genes may be used.
For example, the factor can be purified chromato-graphically by, for example, ion exchange, gel fil-tration and affinity chromatography or combinations thereof. Once t'he factor is sufficiently purified, it can be partially sequenced from the sequence information, oligodeoxynucleotide probes can be made and used to identify the gene encoding the factor in a cDNA library.
The genes encoding positive transcriptional regulatory factors can provide a means for enhancing lc~ gene expression. Lymphoid-specific factors involved in positive regulation of Ig gene transcription can provide a method for enhancement of immunoglobulin production in lymphoid. cells. Lymphoid cells, such as monoclonal antibody producing hybridomas or 2G myelomas, can be transfected with multiple copies of a gene encoding a regulatory factor to induce greater production of Ig. For this purpose, the gene encoding a regulatory factor can be linked to a strong promoter. (The construct can be designed to 2G incorporate a selectable marker as well). Multiple copies of the construct can be inserted into the cell by transfection procedures such as electroporation.
The cell can be transfected with multiple factors, including constitutive factors, where factors are 30 determined to act in conjunction, possibly syner-gistically. By amplifying factor genes in this manner, overexpr~~ssion of the regulatory factors can be induced in th~~ transfected cell and consequently production of immunoglobulin enhanced in these cells.
The factor-.encoding genes can be modified by, for example, techniques of mutagenesis. Alterna-tively, the genes can be synthesized by methods of nucleic acid synthesis in modified forms. These genes could provide modified regulatory factors which have equivalent ~or improved properties over the 1(1 natural factors. For example, the modified factors could have an improved capability to enhance tran-scription. These modified factors are intended to be encompassed by the present invention.
The gene encoding IgNF-B, for example, has been 1_'i cloned and sequenced and the nucleotide sequence is shown in Figure 18A. For the various utilities discussed below, the modified nucleotide sequence can be obtained either naturally (e. g. polymorphic variants) or by mutagenesis to yield substantially 20 complementary sequences having comparable or improved biological activity. Fragments of the sequence may also be used. This invention encompasses sequences to which the sequence of Figure 18A hybridizes in a specific fashion.
2_'i In addition, the DNA binding domain of the factors, which is responsible far the binding sequence-specificity, can be combined with different "activators" (responsible for the effect on tran-scription) to provide modified or hybrid proteins for 3() transcriptional regulation. For example , with recombinant DNA techniques, DNA sequences encoding ~ ~4a999 .

the binding domain can be linked to DNA sequences encoding the activator to form a gene encoding a hybrid protein. The activator portion can be derived from one of the factors or from other molecules. The DNA binding region of the hybrid protein serves to direct the protein to the cognate DNA sequence. For example, in this way, stronger RNA polymerase ac-tivators can be designed and linked to the appro-priate DNA binding domain to provide for stronger enhancement of transcription.
DNA probes for the genes encoding the regulatory factors can be used to determine expression of the genes or to identify related genes by hybridization techniques. This can have diagnostic value for conditions relating to aberrant expression of the factor. Cells can be typed as positive or negative for expression of a particular factor. The DNA
probes are labeled DNA sequences complementary to at least a portion of nucleic acid encoding a transcrip-tional regulatory factor. The labeled probe is contacted with a sample to be tested (e. g., a cell lysate) and incubated under stringent hybridization conditions which permit the labeled probe to hy-bridize with only DNA or RNA containing the sequence to which the probe is substantially complementary.
The unhybridized probe is then removed and the sample is analyzed for hybridized probe.
The DNA probes can also be used to identify genes encoding related transcriptional regulatory factors. For this purpose, the stringent conditions 134p999 of hybridization are sufficiently relaxed so that related DNA sequences which are not completely homologous to the probe can be detected.
Antibodies can be raised against the transcrip-tional regulatory factars of this invention. The antibodies can be polyclonal or monoclonal and they can be used as diagnostic reagents in assays to determine expression of a factor by particular cells or to quantitate levels of a factor.
1() A gene encoding a. transcriptional regulatory factor can also be used to develop in_vivo or in vitro assays to screen for agonists or antagonists of a factor-encoding gene: or of the. factor encoded by the gene. For examp:le:, genetic constructs can be 1!~ created in which a reporter gene. (e. g., the CAT gene) is made dependent upon the activity of a factor-encoding gene. These constructs introduced into host cells provide a means to screen for agonists or antagonists of the factor-encoding gene. The an-21) tagonists may be used to decrease the activity of the factors and thus may be useful i.n the therapy of diseases associated with the overactivity of a transcriptional regulatory factar. Such agonists or antagonists identified by assays employing the 2!i factor-encoding genes of this invention are within the scope of this invention.
The invention is further illustrated by the following exemplification.

~34Q999 EXEMPLIFIfATION
I. Identification_of-_Nuclear_Factor_IgNF=A
A. METHODS
1. Gel electrophoresis DNA Binding assays with the SfaNI-SfaNI rc promoter' fragment.
The SfaNI fragment was subcloned into the SmaI
site of pS64 (pSPIgV , provided by N. Speck). For binding analysis this fragment was excised from pSPIgV by digesting with Hind LII and Eco RI. These x latter sites flank the: Sma I site in the polylinker of pSP64. After end-Labeling with [a-32P]dATP and the large fragment of E. coli DNA polymerase I, the radiolabeled fragment was isolated by polyacrylamide gel electrophoresis. Binding reactions were per-formed and the reaction mixtures resolved by elec-trophoresis (Fig,ure lb). The 3'~Plabeled fragment (about 0.5 ng, 10,000 cpm) was incubated with a nuclear extract of a Human B lymphoma cell line (EW)(prepared by the method of I)ignam, J.D. et al.
Nucl. Acids Res. 11 1475-1489 (1983)) in the absence (lane 1) or presence of two different non-specific competitor DNAs (lanes 2-11). Binding reactions (25 ~1) contained 10 mM Tris.HCl (pH 7.5), 50 mM NaCl, 1 mM DTT, 1 mM EDfA, 5~ glycerol and 8 ~.g EW nuclear ~;5 extract protein. Reac~.ions 2-6 additionally con-tained 800, 1600, 2400, 3200 an<i 4000 rig, respec-tively, of poly(dI-dC)-poly(dI-dC). Reactions 7-11 contained 300, 600, 900, 1200 and 1500 rig, respec-tively, of Hinf I digested E. coli chromosomal DNA. After a 30 min incubation at room temperature, the resulting complexes were resolved in a low ionic strength 4~ polyacryla.mide gel (acrylamide:bisacrylamide weight ratio of 30:1) containing 6.7 mM Tris.HCl (pH
7.5), 3.3 mM Na-acetate and 1 mM: Na-EDTA. See Strauss, F. and Varsha.vsky, A. Cell 37 889-901 (1984). The gel was preelectrophoresed for 30 min at 11V/cm. Electrophoresis Was carried out at the same voltage gradient for 90 min at room temperature with 1G buffer recirculation. The gel was then dried and auto- radiographed at -70°C with a screen. In Figure lb, F and B indicate positions of free and bound fragments respectively.
Binding assays were performed as detailed above lc, using 2400 ng poly(dI-dC)-poly(dI-dC)) and the following DNA fragments: ~c SfaNI-SfaNI (--0.5 rig, 10,000 cpm, lane 1), ~s PvuII-Kpn.I (-0.5 rig, 10,000 cpm, lane 2), x PvuII-SfaNI (-0.1 fig, 5000 cpm, lane 3) and pSP64 PvuII-EcoRI (-0.2 ng, 5000 cpm, lane 4).
The rc PvuII-SfaNI frag;rnent was derived from the plasmid pSPIgV by digesting with PvuII and EcoRI.
The EcoRI site is in t:he polylinker and therefore this fragment contains 16 by of polylinker sequence.
2. Binding competition analysis in nuclear extracts 25 of human EW and HeLa cells.
EW nuclear extract, Figure 2a. Binding assays were performed as detailed above using radiolabeled K
PvuII-SfaNI fragment (-0.1 rig, 5000 cpm) and 2400 r7g ~34pgg9 poly(dI-dC)-poly(dI-dC). Reacti.ons 2-4 additionally contained 50, 100 and 200r~g, respectively, of the bacterial plasmid pSP64 whereas 5-7 contained 50, 100 and 200 qg, respective:ly, of the. recombinant plasmid pSPIgVx. Assuming that a molecule of pSPIgV~ con-tains a single high affinity site whereas a molecule of pSP64 (3000 bp) contains 6000 non-specific sites, the apparent affinity ratio of the factor for these two types of sites is greater than 6000 x 200/50 =
12.4 x 104 Hela nuclear extract, Figure 2b. Binding assays were performed as detailed above using radio-labeled ~c PvuII-SfaNI fragment (about 0.1 ng, 5000 cpm), 2400 ng poly(dI-dC)-poly(dI-dC) and 6 ~g Hela nuclear extract protein (provided by P. Grabowski).
Reactions 1 and 2 additionally contained 100 qg of pSP64 and pSPIgV , respectively.
3. DNase footprinting; of factor-DNA complexes.
(Figure 3) The B cell nuclear extract was applied to a heparin-sepharose column equilibrated with 10 mM
Hepes pH 7.9, 20~ glycerol, 1 mM DTT, 1 mM EDTA, 5 mM
MgCl2 and 0.1 M KCl. The ~c-promoter binding factor was eluted with a 0.25 M KC1 step. Binding reactions with this fraction (30 ~1) conta.ined 2.5 mM MgCl2, x PvuII-SfaNI (-1 rig, 100,000 cpm) and 4800 rig poly(dI-dC)-poly(dI-dC) in addition to components detailed above. The coding strand of the x promoter probe was 3' end-labeled (Eco RI site) with [a-32P]

~34pgg9 _27_ dATP using the large fragment of E. coli DNA poly-merase. Reaction 1 was digested with DNase I (5 ~g/ml) for 2.5 min at room temperature in the absence of B cell nuclear protein. Reaction 2 was initially incubated with t:he heparin Sepharose fraction of the EW nuclear factor (14 fag protein) for. l5 min at room temperature and then digested with DNase I as above.
Each reaction w<~s stopped with EDTA (5 mM) and the products separated by native polyacrylamide gel electrophoresis as detailed above. After auto-radiography to visualize the various species, DNA was eluted from the free (reaction 1) and bound (B1 and B2, reaction 2) fragment bands by incubating gel slices in 0.5 M ammonium acetate (ph 7.5), 0.1~ SDS
and 1 mM EDTA with shaking at 37°C overnight. The supernatants were extracted sequentially with phenol-chloroformisoamylalcol;~ol (25:24:1 v/v) and chloroform - isoamylalcohol. (24:'1 v/v) and precipitated with 2 volumes of ethanol in the presence of carrier tRNA.
After a repreciF~itation step the products were analyzed by sep~~ration in a 10~ polyacrylamide gel (20:1) in the presence of 8M urea followed by auto-radiography at -70°C with a screen. Lane 1 contains products of free. fragment digestion from reaction 1.
Lanes 2 and 3 contain digestion products eluted from bound bands B1 a.nd B2, respectively, from reaction 1.
Lanes 1', 2' and 3' corresponds to 1, 2 and 3, respectively, with the exception that the former set was digested with DNase 1 for .~ min. A-G chemical * t.radema_rk c cleavage ladders of the K promoter probe were co-electrophoresed to map the binding domain. See Maxam, A. and Gilbert, W. Meth. Enzymol. 65, 499-525 (1980).
4. Binding of a common nuclear factor to three Ig transcriptional control elements.
Nucleotide sequences of actual and putative binding sites, Figure 4a. The VL binding site is defined by the DNase I. protection assay (* indicates 1CI boundaries of the protected region). The VH and JH-CU sequences are putative binding sites in the V17.2.25 promoter and the mouse heavy chain enhancer, respectively. Numbers. in brackets indicate start coordinated of octamer motif. Binding competitions.
1'i Binding assays (10 p.l) were performed as detailed above using 1600 ng poly(dI-dC)-poly(dI-dC) and the heparin Sepharose fraction of the EW nuclear factor (1.5 fag protein). VL probe (about 0.1 rig, 5000 cpm) lanes 1-3, 10, 11. Vii probe (-C1.2 rig, 5000 cpm) lanes 4-6, 12-12. JR-Cp probe (-0.2 rig, 5000 cpm), lanes 7-9, 14, 15. Lanes 2, 5, 8 additionally contained 5 ng of a VIJ promoter oligomer (36 bp, spanning positions -8l_ to -44 ofd the MOPC-41 V gene segment) whereas lane:; 3, 6, 9 contained 50 ng of the same oligomer. Lanes 11, 13, 1-'i additionally con-tained 50 rig of a JH-C;H oligomer (41 bp, spanning positions -1 to 40 of the heavy chain enhancer).
Complementary single-stranded synthetic oligo-nucleotides were kind7_y made by Dr. Ronald Mertz, Genzentrum der I;'niver<.~itat Munchen and Dr. E.

L. Winnacker, In:~titut fur Biochemie der Universitat Munchen. They were annealed prior to use as competi-tion substrates in the binding assay.
B. RESULTS
A radiolabeLed SfaNI-SfaNI DNA fragment derived from the upstream region of the MOPC 41 x light chain gene (Fig. la) was incubated with a nuclear extract of a human B cell line in the absence or presence of two different competitor DNAs. The resulting com-1G plexes were resolved from the free fragment by electrophoresis through a low ionic strength, non-denaturing polyacrylamide gel and visualized by autoradiography (Figure lb). In the absence of competitor DNA all of the labeled fragment was retained at the top of the gel (lane 1) probably due to the binding of an excess of non sequence-specific proteins. With addition of increasing amounts of either poly(dI-dC)-poly(dI-dC) (lanes 2-6) or E. coli chromosomal DNA (lanes 7-11) as competitors, putative 2p protein-DNA complexes which migrated slower than the free fragment were detected. The relative abundance of the major species of complex (B) as well as that of minor species was significantly greater in the presence of the alternating copolymer competitor DNA.
Because substitution of the simple copolymer con-siderably increased the sensitivity of the assay, it was used in all subsequent binding analyses.
To test the sequence-specificity of the major species detected in the binding assay, a set of mutually overlapping Kappa promoter fragments (See Figure la, SfaNI-SfaNI, PvuII-KpnI and PvuII-SfaNI) and a similar length fragment derived from the bacterial plasmid SP64 (EcoRI-PvuII) were indi-vidually assayed (Figu.re lc). Whereas the bacterial DNA fragment showed no appreciable binding (lane 4) all of the rc promoter fragments yielded major dis-crete complexes of similar mobilities (lanes 1-3).
The mobilities of a series of complexes formed with different length fragment probes (75-300 bp) are approximately the same (data not shown). These data therefore suggested the binding of a specific nuclear factor within the region of overlap of the Kappa promoter fragments. This region includes the con-served octameric sequence but not the TATA element.
Note that with the smallest 75 by Kappa promoter fragment (lane 3) no appreciable label was retained at the top of the gel. Thus, as has been noted recently, the use of small probe fragments further enhance the sensitivity of detection of specific protein-DNA complexes.
The sensitivity gained by use of both the copolymer and a small fragment probe permitted the detection of two complexes, B1 and B2 (Figure 2a, lane 1). The major species B1 corresponds to complex B in the earlier figure. The relative affinity of the factors) for rc promoter DNA was estimated by a competition assay (Figure 2a). Whereas a control plasmid (pSP64), when added in the above incubation, failed to compete for binding in the concentration range tested (lanes 2-4), the recombinant plasmid (pSPIgV ) effectively reduced the formation of both species B1 and B2 in t:he same range (lanes 5-7). The latter plasmid was constructed by insertion of the upstream region of the: Kappa promoter into pSP64.
Assuming that the pSPLgV~ plasmid contains a single high affinity binding site these results suggest that the nuclear factor(s;) responsible for B1 and B2 has at least a 104-fold higher affinity for its cognate sequence than for hete:rologous plasmid DNA (see Methods section 2 abov~e).
To determine if t:he factors) responsible for formation of B1 and B2'. was specific to B lymphocytes, a nuclear extract derived from human HeLa cells was l~~ assayed for binding to the ~c promoter probe (Figure 2b). Both species B1 and B2 were generated at similar levels, as than observed with B cell ex-tracts, by the HeLa e~:tract (lane 1). Furthermore, both were specifically competed by the pSPIgV
2C1 plasmid (lane 2). Thus HeLa cells also contain a factors) which binds specifically to the rc promoter upstream region.
DNase I footprint: analysis was used to delineate at a higher resolution the binding domains) of 2~~ factor(s) present in complexes E1 and B2. To facili-tate these studies, the binding factors) from B
cells was partially purified by chromatography of nuclear extract protein on a heparin sepharose column. Most of the Minding activity eluted in a 3C1 0.25 M KC1 step fraction, giving a purification of 1340ggg approximately 5-fold (data not shown; see legend to Fig. 3). For footprint analysis, DNase I was added for a partial digestion after incubation of the partially purified fac.tor(s) with the ~c promoter probe B1 and B2 species were then resolved from free fragment by polyacryla.mide gel electrophoresis.
Bound DNA was eluted from both B1 and B2 bands and examined by denaturing polyacrylamide gel electro-phoresis (Fig. 3, lanes 2, 3 and. 2', 3'). DNase I
1C digests of the ~c promoter probe in the absence of B
cell protein (lanes 1 and 1') and A+C chemical cleavage ladders were coelectrophoresed to map the binding domain. Factors) in the B1 complexes (lanes 2 and 2') appeared to protect a 19 nucleotide region lc, on the coding strand. The 5' and 3' boundaries of the protected region m.ap to positions -72 and -52, respectively, from the site of transcriptional initiation. The region of DNase I protection was centered about the conserved octanucleotide sequence 2G ATTTGCAT suggesting its importance in the recognition of the Ig promoter by the nuclear factor. B2 com-plexes showed a virtually identical DNase I protec-tion pattern as B1 complexes and therefore do not appear to involve additional DNA contacts (lanes 3 2-''~ and 3'). The simplest interpretation of this ob-servation is that the B2 complex is generated by dimerization through protein-protein interactions of the factor responsible for the B1 complex. Al-ternatively, the B2 complex could be formed either by 3C~ the binding of another protein to the factor responsible for the B7. complex or by recognition of the same set of sequences by a distinct DNA binding protein.
Because the octamer sequence motif is present in both light and heavy chain gene promoters as well as in the enhancer elements of both mouse and human heavy chain genes, assoays were performed for factor binding to fragments from a mouse heavy chain promoter (VH) and the mouse heavy chain enhancer.
The VH promoter fragment was derived from the 5' region of the V17.2.25~ gene and included nucleotides between positions -154. and +57 relative to the transcriptional start site. Grosscheal, R. and Baltimore, D. Cell 41 885-897 (1985). In this 1_'i promoter the conservedl octanucleotide spans positions -57 to -50 (Figure 4a). The heavy chain enhancer fragment was derived from the germline JH C region and spanned positions 81 to 251 within a 313 by region implicated in enhancer function. Banerji, T.
2C1 et al. Cell 33 729-740' (1983). The conserved octa-nucleotide is positioned between. coordinates 166 to 173 in the above fragment (Fig. 4a). The B-cell heparin sepharosc fraction (purified on the basis of binding to the Kappa promoter sequence, Fig. 4b, lane 25 1) evidenced binding t.o both the VH promoter fragment (lane 4) and to the enhancer fragment (lane 7). The mobilities of the complexes formed with the three fragments were very similar consistent with the binding of a common factor. Furthermore, binding to 30 these fragments was specifically competed by a 1340ggg synthetic duplex 40-mer that spanned the octanucleo-tide motif of the MOPC 41 ~c light chain gene promoter (lanes 1-9). An oligo~mer of equivalent size con-taining a sequence from a region. of the mouse heavy chain enhancer lacking; the octanucleotide motif (Fig.
4b) failed to compete for binding in the same con-centration range (lanes 10-15). This competition analysis further strengthens the suggestion that a common nuclear factor (IgNF-A) binds to all three Ig 1C1 transcriptional elements. As ha.s been mentioned previously, these three transcriptional elements share an identical sequence motif ATTTGCAT (Fig. 4a).
Thus, the binding of a. common nuclear factor is almost certainly mediated by this motif.
15~ II. Dependency-of-in-.vitro-transcription_of-I~_~enes on an a stream se uence _______P_________g_____ A. METHODS
Figure 5a: Templates. The deletions 5'~5 and 5'~7 have been described before. See Bergman Y. et al 20 PNAS USA 81:7041 (1981). The highly conserved octanucleotide sequence which is found upstream of all sequenced heavy and light chain variable region genes is boxed (labelled "OCTA"). It is located approximately 30 base pairs upstream from the "TATA"
25 box. The plasmids prc and p~rc were constructed by converting the 5'-ends of 5'~5 and 5'~7 into a Hind III site by means of synthetic linkers followed by 1 ~4~99g ' cloning the fragment u.p to the Bgl II site in the J -C major intron into Hind III, Bam HI digested x x pUC-13. pxE and pxE represent plasmids containing x ~c either the kappa enhan,cer of the heavy chain enhaner cloned into the unique Hind III site of pK. The segments used as the enhancers a.re an 800 by Hind III-Mbo II fragment from the J -C intron (Max, E.E.
x x et al. (1981) J. Bio:L. Chem. 256: 5116) and a 700 by Xba I-Eco RI fragment from the JH-CU intron. Gillies, S.D. et al. Cell 33: T17 (1983); Banjerji, J. et al.
Cell 33: 729 (1983). figure 5b; transcription in whole cell extracts made from the human B lymphoma cell lines RAMOS (lane.s 1,2) and EW (lanes 3,4);
transcription in a HeLa whole cell extract (from A
Fire (lanes 5,6)). The expected 2.3 kb run off transcript is indicated.
The cell lines RA.MOS and EW were grown in RPMI
medium containing 10~ inactivate.d fetal calf serum to a density of 5-8x105 cells per ml. Whole cell extracts were generated according to the procedure of Manley et al., Proc.-Nfatl._Acad..-Sci.-USA 77: 3855 (1980), and had a final protein concentration of approximately 18 mg/ml. Run off' transcription reactions were carried. out at 30° for 60' in a reaction volume of 20 ~1. A typical reaction mix contained 9 ul (160 ~.g;) of whole cell extract, 50 uM
each of ATP, CTP and fTP, .5 uM UTP, 10 ~Ci of a-32P
UTP (NEG 007X, 7600 Ci/mM) 5 mM creatine phosphate, 0.3 mg/ml creatine phosphokinase (Sigma), 12 mM Hepes 7.9, 12~ glycerol, 60 mM KC1, 5 mM MG++, 1 mM EDTA, 0.6 mM DTT, linearized template (about 50 rig) and poly (dI-dC)-poly(dI-dc) as a non-specific carrier (about 400 rig). The reactions were terminated by adding 200 ~cl of stop buffer (7M urea, 100 mM LiCl, 0.5$ SDS, 10 mM EDTA, 250 pg/ml tRNA, 10 mM Tris (pH
7.9), followed by two extractions with phenol:
chloroform: isoamyl alcohol (1:1.:0.05), one with chloroform and precipitation with ethanol. The RNA's were treated with glyoxal and analyzed by electro-phoresis through a 1.4.~ agarose gel in 10 mM sodium phosphate (pH 6.8), 1 mM EDTA. See Manley et al.
supra. The gel was then dried for autoradiography with an intensifying screen at -70°C.
Figure 6: Effect: of the upstream deletion 5'~7 on in vitro transcription in B cell extracts utiliz-ing a pre-incubation pulse chase. protocol. Run off transcripts obtained utilizing templates containing either the wild type promoter (Lane 1) or the truncated Kappa promoter (lanes 2,3). Lanes 4-6: In vitro 2G transcription using closed circular templates con-taining the wild type promoter (lane 4) or the truncated x promoter (lanes 5-6). In these reactions 50 ng of a closed circular template containing the adenovirus major late promoter (MLP) was included as 25, an internal control. The transcripts specific to the ~c template or the adenovirus template are indicated as x and MLP, respectively. For a template con-taining the major late promoter the plasmid pFLBH was used. The plasmid contains sequences from 14.7 to 13409gg 17.0 map units of aden.ovirus inserted between the Bam HI and Hind III sites of pBR322 and was the kind gift of A. Fire and M. Samuels.
Either the linearized or the supercoiled tem-plate (50 rig) was incubated in a volume of 20 gel with 9 ~cl (about 150 fig) of EW extract, 6$ (wt/vol) polyethylene glycerol 20,000 and all other components described for Fig. 5b except the nucleotides for a period of 60 minutes at 30°C. Transcription was initiated by the addition of nucleotides and radio-active UTP to the following final concentrations: 60 uM each of ATP, CTP anal GTP and 1 ~M UTP and 10 ~Ci a- P UTP (NEG 007X, 600 Ci/mM). The initiating pulse was maintained for 10' at 30° followed by a 10' chase with a vast excess of non-radioactive nucleo-tides. Final concentrations during the chase were as follows: 330 ~M ATP, C'TP, GTP and 1 mM UTP. The reactions were quenched, worked up and the run off transcripts analyzed a.s described above. Mapping of the initiation site of the transcript was conducted as follows: Transcripts generated from closed circular templates were taken up in 20 ~1 of HE (50 mM Hepes, pH 0.7, 1 mM: EDTA) and. 10 ~1 was used for hybridization selection. A hybridization template complementary to the ~s RNA was constructed by cloning the Pva II-Sau 3A fragment which contains the cap site of the MOPC41 gene (Queen and Baltimore, Cell 33: 741 (1983)) into the M13 phage MP9. Single stranded phage DNA was prepared and purified by 140999 , density gradient centrifugation through cesium chloride. MLP specific transcripts were detected using the M13 clone XH11 provided by A. Fire and M.
Samuels. Hybridizations were done in a final volume of 15 ~1 in the presence of 750 mM NaCl and 100-200 ug of single stranded complementary DNA at 50°C for 2 hrs. The reactions were then diluted with 200 ~1 of cold quench solution (0.2 M NaCl, 10 mM Hepes pH 7.5, 1 mM EDTA) and 2 U of ribonuclease T1 added. Di-gestion of single stranded RNA was allowed to proceed for 30' at 30°C .after which the reactions were extracted once with phenol chloroform isoamyl alcohol (1:1:.05) and precipitated with carrier tRNA. The pellet was wash ed once with cold 70$ ETOH, dried and resuspended in 80~ v/v formamide, 50 mM Tris borate, pH 8.3 and 1 mM EDTA. The RNA was denatured at 95°C
for 3 min and then electrophoresed through a 6$
polyacrylamide 8.3 M urea sequencing gel. The upper of 2 bands (x) derived form the immunoglobulin Promoter represent the correct start for x transcrip-tion. The lower band is seen at variable intensities and probably does not represent a different cap site, as explained below.
B. RESULTS
Whole cell extracts were made from two human Burkitt lymphoma lines, EW and RAMOS, by the pro-cedure of Manley et al. supra. The templates used for in vitro transcription reactions are diagrammed ~~40999 in Fig. 5a. The template representing the wild type gene (pK) was derived from the MOPC41 ~c gene and contained sequences from approximately 100 by up-stream from the transcription initiation site (end point 5'05, Fig. 5a) to the Bgl II site in the major J -C intron Max, E.E., J. Biol. Chem. 256: 5116.
______________ ___ This fragment retains the complete variable region which is rearranged to J 1, but not the rc enhancer which is further downstream of the Bgl II site. See, e.g., Queen and Stafford, Mol.-Cell._Biol. 4: 1042 (1984). This short 5' flank has been shown to be sufficient for accurate initiation and high level of transcription in a transient transfection assay.
Bergman, Y. et al., Proc.-Natl.-Acad._Sci. USA 81:
lc, 7041 (1984). Deletion analysis of the x promoter showed previously that important regulatory sequences are present between nucleotides -79 and -44 because deletion 5'~7 completel.y abolished transcriptional competence of the gene while deletion 5'~5 had no 2C1 effect. See Bergman e.t al. supra. The template representing an inactive promoter mutant (pOrc) was constructed by engineering a Hind III site into the 5' -end of 5'~7 and cloning the segment of the gene up to the Bgl II site into pUCl3 cut with Hind III
and Bam HI.
To examine transcriptional activity in B cell extracts, a linear template truncated at the Sac I
site in the polylinker was used and transcripts ending at this site (run off transcripts) were 3Q examined by electrophoretic separation. A run off ~34pgg9 transcript of 2.3 kb was evident: when RAMOS, EW or HeLa cell extracts were used (Fi.g. 5b, lanes 2, 4 and 6). When a ~c chain enhancer sequence was added to the construct, no effect was evident implying that transcription in these: extracts is enhancer indepen-dent (Fig. 5b, lanes 7., 3 and 5). (In EW and HeLa, the enhancer appears t:o cause a slight increase in the background radioactivity but: not in the 2.3 kb band.) The band at 2.3 kb could be abolished by not adding the template or by transcribing in the presence of 0.5 ~g/ml amanitin. Thus it represents a template-specific, RNA polymera:;e II transcript. The band just below 2.3 kb is not decreased by a-amanitin and presumably reflects end-labeling of endogenous 1_'i 18S rRNA.
To assess whether initiation of transcription occurred at the natural cap site, a second assay was used. See Hansen, U. and P.A. Sharp, EMBOJ, 2: 2293 (1983) For this assay, the uniformly labeled RNA
2() was hybridized to a single stranded DNA probe span-ning the transcription initiation site (generated by cloning the Pvu II-Sau 3A fragment of the rc gene into phage M13). The resulting complex was digested with ribonuclease T1 and the ribonucl_ease-resistant RNA
fragments were analyzed by electrophoresis through a 6~ polyacrylamide gel with 8.3 M urea. Analysis of in vitro synthesized F;NA by thisc method is shown in Fig. 6 (lane4). The upper band (labeled ~c) represents the correct cap site. The band just below it was seen at variable intensities and probably does not represent a different cap site but rather arises from cleavage with ribonuclease T1 at the next G residue from the 3'-end of the protected region. (Examination of the sequence near the Sau 3A1 site shows that the second set of G residues on the RNA lies 19 by upstream from the end of the region of homology with the single stranded DNA probe). Thus the extracts generated from B cells were capable of correctly initiating and transcribing the immunoglobulin promoter in vitro with. approximately the same ef-ficiency as a HeLa cell extract.
To analyze the effect of 5' flanking sequences in vitro, we examined the transcription of the deleted gene, p~K. Be.cause many regulatory effects act upon the rate of initiation of transcription, we chose to use a pre-incubation, pulse-chase protocol which measures the initiation rates. See Fire, A. et al., J. Biol. Chem. 259: 2509 (1984). The template DNA was first pre-incubated with the extract to form a pre-initiation complex. Transcription was then initiated by the addition of nucleotides and radio-labeled UTP. The in:it:iated transcripts were com-pleted during a chase period with unlabelled nucleo-tides and analyzed by electrophoretic separation.
Incorporated radioactivity in this assay is propor-tional to the number o~f correct initiations occurring during the pulse.
In Fig. 6, comparison of lanes 2 and 3 with lane 1 shows that the template p~c, which contains about 100 by upstream of the: initiation site, initiated approximately 10-fold more efficiently than did the deletion mutant, p~c. Again, the presence of the heavy chain enhancer placed at -44 by to the trun-cated promoter did not alter the level of tran-scription. When closed circular templates were used, a similar effect of the. promoter truncation was observed (Fig. 6, lanes 4-6). In these reactions a template containing the major late promoter of adenovirus was included as an internal control; the expected protected RNA. fragment of 180 by is labeled MLP. Comparison of lanes 5 and 6 with lane 4 shows that there was a 10-fold decrease in the efficiency of transcription from the mutant promoter, whereas the transcript of the major late promoter remained l~, constant. The reason for the apparent decrease in the amount of transcription from the supercoiled template containing the heavy chain enhancer has not been further addressed.. It is evident, however, that the dependence of transcription on an upstream 2U sequence between -44 and -79 is observed whether the effect described above was specific to B cell ex-tracts, the same templates were transcribed in the heterologous HeLa whole cell extract. A 4- to 5-fold decrease in transcription was seen with the deleted template when compared with the wild type template (data not shown). Thus, the effect of the deletion is at best, only modestly tissue-specific.
We have reported here the development of transcriptionally competent whole cell and nuclear 3p extracts from two independent human B cell lymphomas.

In such extracts, transcription from the promoter of the MOPC41 x gene was correctly initiated and a promoter deletion significantly reduced the level of initiated RNA. In vivo, the effect of the deletion used here is several hundredfold when analyzed by a transient transfection assay, however the effect we observe in vitro is only about 10- to 15-fold.
Although there are now several examples of upstream sequence requirements for in vitro transcriptions, See, e.g., Groschedl R. and Birsteil M.L., Proc.
Natl. Acad. Sci. USA 79: 297 (1982); Hen, R. et al.
Proc. Natl. Acad. Sci. USA 79: 7132 (1982), the effects have been smaller than the corresponding one in vivo. This is possibly due to the dominance of the TATA box and associated factors in determining the level of transcription in vitro Miyamoto, N.G. et al., Nucl. Acids Res. 1.2: 8779 (1984).
III. Discovery-and-Characterization-of-IgNF=B
The mobility shift gel electrophoresis assay was 2C used to screen nuclear extracts from a variety of cell lines for octamer binding proteins. The band corresponding to IgNF-A was found in all extracts but a second band with distinct mobility from IgNF-A was found only in nuclear extracts from lymphoid cells.
This lymphoid-specific octamer binding protein, termed IgNF-B, was found in nuclear extracts from all pre-B, mature B and myeloma cell lines tested and in nuclear extracts from some T cell lymphomas (see Figure 7 and 8). IgNF'B was not detected in nuclear extracts from the non-lymphoid cell lines, Hela, ~2, Cos and Mel (see Figure 8). IgNF-B was shown to be specific for the same octamer sequence as IgNF-A by competition experiments in which the IgNFB band was selectively competed by unlabelled DNA fragments sharing only the octam.er sequence and not by DNA
fragments lacking the octamer sequence.
The octamer sequence is found at approximately position -70 upstream of the transcription start site of all immunoglobulin (Ig) variable genes which is in the region that has been shown to control the lym-phoid specificity of the Ig promoter. Thus IgNF-B
binds to the upstream octamer sequence in lymphoid cells and activate tra.nscription..
IV. Factors_Binding_t:o_~-Enhanc.er---E_Factor The fully functional ~ enhancer has been localized to a 700 by XbaI EcoRI fragment from the major intron between JH and C This fragment can be further subdivided by cleaving at the PstI site to generate a 400 by Xbal-Pstl fragment (p400) and a 300 by PstI-EcoRI fragment: (p300). It has been shown by transient transfections that 30-50~ of the tissue specific enhancer activity is retained in X300, whereas there is no detectable activity of p400. The gel binding assay was employed to investigate what protein factors may interact with the u-enhancer.
Briefly, end-labelled DNA fragments were incubated with nuclear extracts made from tissue culture cells.
After 20 min at room temperature the mixture was 1340ggg loaded on a low ionic strength polyacrylamide gel and electrophoresis carried out at 120V for 2 hrs. The gels were then dried for autoradiography. When the functional 300 by (~.30~0) enhancer fragment was used in such an assay a DNA.-protein complex was observed in extracts derived from the human B lymphoma cell line EW (Figure 9b, lane 2). To show that this new band represented a specific complex binding reactions were carried out in the presence of varying amounts of non-radioactive competitor fragments (Figure 9b, lanes 3-11). It is easily seen that when ~a300 is added as the competitor fragment, the complex band is completely lost. In contrast, the adjacent u400 fragment (lanes 6,7,8) or a 450 by fragment con-taining the k light chain enhancer (lanes 9,10,11), cause only a minor effect even at the highest con-centrations used. It is interesting to note that there appears to be a slight increase in the amount of specific complex in. the presence of the ~c enhancer fragment (compare lanes 9 and 2). As demonstrated below, both the ~ and the rc enhancers interact with at least one common protein and this is not the factor being detected by binding the u300 fragment.
The increase in the specific complex in the presence of the rc enhancer is probably due to the removal of factors common to both. the enhancers from the reac-tion mix, thus leaving, more of the labelled fragment available to bind to the p specific factor being detected by it.

In order to be able to detect binding sites for less abundant proteins and also to more precisely define the complex detected with ~a300, this fragment was further dissected. Each of the smaller fragments generated were analyzed for their ability to serve as binding sites for nuclear proteins. Figure l0a shows a partial restriction map of the relevant region of the ~ enhancer. x.300 was digested with AluI, HinfI
and DdeI to generate a number of 50-70 by fragments labelled X50, ~c(60)2 (a mixture of AluI-DdeI and HinfI to AluI) and p70 (AluI-AluI). Binding reac-tions were carried out with each of these fragments with nuclear extracts of EW lymphoma cells in the presence of increasing amounts of the non-specific competitor poly (dI-dC)-poly(dI-dC). The results are shown in Figure lOb.
Fragment x.50 forms a major complex band (lanes 2,3,4) that is barely decreased even in the presence of 4 ug of poly(dI-dC)-poly(dI-dC) (lane 4). The mixture of the two 60 by fragments does not give rise to a discrete complex band (lanes 6,7,8). Finally the ~c70 fragment gave 2 faint, but discrete, nucleo-protein complex bands (lanes 10,11,12) of which the lower one is again barely affected by 3 ugm of non-specific carrier poly(dI-C)-(dI-C) (lane 12).
Specificity of the complexes observed were shown by competition experiments using a variety of DNA
fragments, Figure lOc. Thus, the complex generated with ~c50 is specifically competed away in the presence of X300 (of which u50 is a part), or a x promoter fragment, but not by corresponding amounts of X400 or a x enhancer fragment, consistent with the complex being generated by the interaction of the previously described factor IgNF-A with its cognate sequence. (This factor recogns_zes a conserved octanucleotide, ATTTGCAT, found in the promoters of all sequenced immunog~lobulin genes and within this subfragment of the heavy chain enhancer.) The complex observed with X70 was specifically competed away by itself (lane 9) and to some extent with the x enhancer (lanes 5,6 Figure 10d) but not at all by either the Moloney murine leukemia virus enhancer (lanes 3,4) or by ~t400 (lanes 7,8, Figure 10d). Further competition analysis showed that this complex could not be competed away by either (~,60)z (Figure 10e, lanes 7,8), X50 (Figure 10e, lanes 5,6) or X170 (central A~_uI-AluI fragment) (figure 10e, lanes 11,12). The binding sae have observed is therefore specific to thi;~ small fragment and was detected only upon further di;>secting X300 which separated the major observable interaction of IgNF-A with the enhancer sequence to <~nother fragment (~50).
Ephrussi Wit. al. (Science, 227:134-140 (1985)) and Church et al. (I~fa r ,, 313:798-801 (1985) ) have used 2.5 methylation protection experiments to define a set of G
residues within the heavy chain enhancer that are specifically resistant. to methylation by DMS in B cells.
This result lead to the proposa7_ that tissue-specific DNA binding proteins were responsible for this decreased accessibility of the reagent to DNA. The protection was observed in 4 clustere;, the DNA sequences of which were sufficiently homologous to derive a consensus sequence for the binding site of a putcrtive factor. All four pos-tulated binding sites (E1-E4) are found within the 700 by fragment; however ~c300 retains only 2 complete binding domains (E3 and E4) for this factor and the octamer (0) sequence. The Alu-Alu fragment that shows that specific nucleoprotein complex described above contains the complete E3 domain and the factor we detect in vitro presumably is the same as that detected in vivo. Thus, it was unexpected that the HinfI-Dde fragment (u50) containing E4 and 0 should not compete for binding to X70 (Figure 10e, lanes 5,6).
In case th is was due to the fragment predomi-nantly binding I;gNF-A at the octamer site and thus making it unavai:Lable as a competitor for u70, binding reaction: and competitian assay were done for a fraction generated by chromatography of the crude extract over a heparin-sepharose column, that con-tained X70 binding activity and was significantly depleted of IgNF-A. When X50 or ~a170 was endlabeled and incubated with the column fraction, no specific nucleoprotein complex es were seen upon electro-phonetic analysis.. Even in this fraction, X50 and X170 failed to compete successfully for the inter-action between ~a70 and its binding protein (data not shown), thus implying :strongly that the binding site defined as E4 perhaps does not~bir~d the same factor that binds at E3. Similarly, the E1 domain (isolated 1340ggg as a Hinf-PstI fragment) does not compete as ef-fectively as X70 itself for the binding of the factor to ~c70.
To determined the location of the binding sites within individual fragments (p70 and X50), the technique of methylation interference was employed.
End-labelled DNA fragm~.ents were partially methylated on guanines and adenines using dimethyl sulfate (DMS). Methylated DNA. was then used for binding 1() reactions with crude extracts and the complex was resolved from the free. fragment by electrophoresis.
Both complex and free fragment bands were then excised from the gel, and the DNA was recovered by electroelution. Piperidine cleavage of the recovered l~~ fragments was followedl by electrophoresis through 12~
polyacylamide urea sequencing gels. In principle, if any of methyl groups introduced by reaction with DMS
interfere with the binding of a specific protein then that molecule of DNA will be selectively missing in 2p the complex formed andl subsequently the corresponding ladder. The method therefore allows identification of G residues making intimate contacts with the protein. [We have found that the use of DNaseI
footprinting via the g;el binding; assay to be compli-cated in the case of some of these less abundant factors because of the short half lives of the complexes themselves. Thus if a binding incubation is followed by partial DNaseI digestion, it is possible that in the course of time required to load the sample and have th.e complex enter the gel, DNA

134pg~9 -s0_ fragments that were in complex form may exchange with the larger amounts of free fragment in the binding reaction. Thus not lcaading to any distinction in the DNase patterns seen with wither the complex or the free DNA (e. g, the ha7_f life of the nucleoprotein complex in p70 is less; than 1 minute,)].
The result of ,carrying out such an interference experiment using nuclear extracts and on the u50 DNA
fragment shows that the complex observed arises via ZO interaction of the IghIF-A protein at the conserved octameric sequence (Fi.gure lla). The free fragment generates a characteristic G ladder (Figure lla), lane 2,3) and the complex form (lane 1) is speci-fically depleted in DNA molecules carrying a methyl group at the G residue indicated by the asterisk which lies in th~~ middle of the conserved octamer.
Presumably, modi:Eication at this residue seriously impedes the formation of a stable complex between the protein and its cognate sequence. This residue was also shown to be protected against methylation of DMS
in vivo.. Interestingly, however, none of the other G
residues observed to be protected _i_n _v_i_v_o in this region of the ~s enhancers appear to be affected in our in vitro interference experiment. Therefore, if these protections, in vivo are due to the binding of a protein, this factor i;s different from IgNF-A or B
and is not binding to :fragment _i_n _v_i_t_r_o.
On the p70 fragment several G residues were identified as being important in forming intimate contacts with the binding protein (G) (Figure llb).

1340ggg On the coding strand bands the 3 G's are signifi-cantly reduced :in intensity in the complex as com-pared to the free DNA fragment (Figure 11b, compare lanes 1,2), and on the non-coding strand 2, G's are significantly a:=fected (Figure 11b, compare lanes 3 and 4).
The result: of both the in vivo DMS protection experiment and our in vitro methylation interference experiments are summarized in Figure 11c. The open and closed circ7_es above the sequence were the residues identified by Ephrussi ~t al. (Science, 227:134-140 (19Fi5)) to be protected against methylation in vivo whereas the encircled G's are the ones identified by ups 'fin v:itro. The pattern of protection ~5 and interference on the X70 fragment over the consensus sequence is strikingly similar in vivo and in vi ro, which indicated strongly that the protein identified here by means of the gel bind assay is the one that interacts with this sequence in vivo. Analogous to ~t50, however, the second set of protections seen in this region in vivo was not. observed in vitro.
Tissue specificity of the factors detected: In order to determined whether t:he proteins identified are c5 limited to expression only in B cells, a large number of extracts made from B cells and non-B cells were screened (Figure 12). Complexes that co-migrate with the ones generated and characterized (by competition and methylation interference experiments) in the B
cell line EW, were observed on both the fragments X50 and X70 (Figure 12; EA 70) in all the cell lines 1340ggg examined. Although th.e complex generated in each extract has not been further characterized, we interpret this data as indicating that both these factors are non-tissue specific. A second complex (NF-~cB) was observed with the p50 fragment that was restricted to B and T cells only.
A point to note is that although the amount of protein in each lane has been held constant at between 9 and 11 ug, the extent of complex generated was found to vary considerably from extract to extract. Thus showing, that quantitive estimations of the abundance of proteins in different cell lines using this assay if not very meaningful at this stage. (This is presumably due to subtle variations in the state of the cells and the extraction con-ditions for the different cell lines).
In summary, our current analysis of the func-tional 300 by Pstl-EcoRI fragment of the p. enhancer reveals that:
(i) at least 2 different proteins bind within this DNA sequence. On.e protein (IgNF-A) interacts with an octamer sequence (ATTTGCAT) that is highly conserved upstream of all heavy and light chain variable region genes and is also found in the a enhancer. The second protein interacts with a sequence shown by Ephrussi and Church to be protected in a tissue specific manner against methylation by DMS in whole cells;

13409gg (ii) both :Eactc>rs can be detected in nuclear extracts from a variety of cell lines and are there-fore not B-cell specific;
(iii) both E1 and E4 sequences hardly compete for the binding of the factor to X70 (which cor-responds to E3) thus implying that these sequences do not interact wii:h th.e same factor, although the sequence homology amongst the sites would have lead one to expect that they should.
V. Identification of Factors binding to Kappa-light chain enhancer An enhances element has also been identified in the major intron of the x light chain gene. Picard and '5 Schaffner (Natuy, 30'7:80-82 (1984)) showed that the enhancement activity can be localized to a 500 by AluI-AluI fragment and Queen and Stafford (Mol. Cell Biol., 4:1042-1049 (1984)) have further refined the 5' and 3' boundaries so that the enhances may be considered restricted to 275 base pairs wit=hin the larger fragment.
We have dissected thi~~ region into a number of smaller fragments and assayed each of these by means of the gel binding assay for the location of protein binding sites_ A restriction map of the relevant region of the c.5 x enhances is shown .in Figure 13a The black boxes represent sequences :identified by Church a al. (Nature, 313:798-801 (1985)) to be homologous to the putative protein binding domains detected in the ~ enhances in Fragments generated by cutting with Dde and HaeIII (x1, x2, ;~t3, x4 and x5; Figure 13a) were assayed for binding in the presence of increasing amounts of poly(:dI-dC)-(dI-dc) as a non-specific carrier, rc4 and rc5 appeared to be obviously negative (Figure 13b, lames 1-.B) while rc3 and rc2 appeared to be positive (Figure 1:3b, lanes 10-12 and 14-16).
Preliminary results show that the internal fragment does not contain. any ;specific binding sites either.
The nucleoprotein complex bands generated with 0.5-1 rig of radiolabelled probe could be detected even in the presence of 3 ~g of the carrier (lanes 12, 16, Figure 13b).
To show that the bands detected represented a specific interaction between a protein and DNA we carried out competition experiments (Figure 13c and 13d). The competition pattern for rc2 was strikingly similar to what ;had been earlier. observed with the X70 fragment; this relatively large amounts of u400, the Moloney leukemia virus enhancer, the SV40 en-hancer or the K promoter (containing the conserved 2p octa) to rc2 did not compete for binding, however u300 and the k enhancer did (data not shown). Since rc2 contains a putatj_ve E box identified by sequence comparison (as does p7~0) we competed its binding with smaller fragments. from ~.r300 (Figure 13c) . The complex is specifically competed away by the addition of unlabelled ~r70 during the incubation (compare lanes 3 and 4 with lane 2), but not by X60 (lanes 5,6), p170 (lanes 7,8) or the SV40 enhancer (lanes 9,10).
Further, the protein that binds to this sequence co-fractionates with the X70 binding activity through 1340ggg two sequential chromatographic steps (Heparin agarose and DEAE Sepharose). Thus we conclude that the same sequence specific protein binds to both the fragments X70 and ~s2 and therefore there is at least one common protein interacting with both the ~s and the ~c en-hancers.
The ~c3 complex (indicated by the arrowhead, !' Figure 13d) failed to be competed away by X300 (compare lanes 3 and 4 with lane 2), X400 (compare l~~ lanes 5 and 6 with lane 2) as a rc promoter containing fragment (compare lanes 7 and 8 with lane 2).
However, the complex was specifically competed away with both the complete ~c enhancer (lanes 9,10) and the SV40 enhancer (lan.es 11,12). The band below the l~~ major ~c3 complex was seen at variable intensities in different experiments and failed. to compete even with the complete ~c enhancer in this experiment and has not been further investigated at this stage. The observation that the SV40 enhancer specifically 2C competes for binding of this factor is not altogether surprising since this fragment and the SV40 enhancer share an identical stretch of 11 nucleotides.
The binding site of this factor on the ~c3 a, fragment was localized methylation interference 25 experiments. In two different extracts methylation at three of a stretch of 4 residues within this sequence completely abolished binding (Figure 14, compare lane 1 [complex] and 2 [free]; and lanes 3 [complex] and 4 [free]). This stretch of G's form a 30 part of the conserved region (GGGGACTTTCC) between the SV40 enhancer and rc3. Thus the binding site was localized towards one end of the rc3 fragment. The results also served to explain the specific competi-tion observed earlier with the SV40 Enhancer.
Interestingly, deletion mapping of the x enhancer shows that sequences within the ~c3 fragment are extremely important for enhancer function.
The tissue range of this factor was examined by carrying out binding analysis with ~c3 in extracts from a variety of cell lines. Nucleoprotein complex formation ~c3 was detected in a mouse B cell line (Figure 15a, lane 2) but not in 5 other non-B cell lines (Figure 15a, odd. numbered lanes from 5-11).
Even numbered lanes show that the ubiquitous factor detected by p50 is present in all these cell lines and served as a positive control for the experiment.
The factor x3 therefore appears to be restricted to expression to B lymphoid cells.
We then examined extracts made from cells at various stages of B cell differentiation (Figure 15B). Interestingly ~c3 binding protein can be detected in the Abelson murine leukemia virus trans-formed pre-B cell line PD, in two mouse B cell lines (WEH1231 and AJ9, Figure 15B, lanes 12,14), one human B cell line (EW, Figure 15B, lane 16) mouse myeloma line (MPC22, Figure 15B, lane 18) and 2 human mye-lomas (KR12 and 8226, Figure 15B, lanes 20,22).
However it does not appear to be present in a pre-preB cell line (C5, Figure 15B, lane 4) and in mouse pre-B cell lines (HATFL., 38B9, 70Z, Figure 15B, 1340ggg lanes 6,8,10). Thus this factor appears to be not only tissue-specific, i.e., limited to cells of the B
lymphoid lineage, but is also stage-specific within that lineage. In the series of extracts examined, the presence of this factor bears a striking cor-relation with rc expression.
The results with the Kappa enhancer as can be summarized follows: Dissection of the ~c enhancer enabled detection of two distinct binding proteins with this DNA. One of these proteins appears to be ubiquitous and interacts with the p heavy chain enhancer as well. The second protein appears to be highly expressed in a stage-specific manner within the B cell lineage and. can be detected only in those cell lines where the endogenous rc gene is active.
There does not appear to be a binding site for this factor in the heavy chain enhancer although there is one in the SV40 enhancer.
VI---Cloning-of-Transc.riptional-.Regulatory_Factors A. Cloning of putative NF-xB
Experimental-Procedures agtll-EBNA=1-_Recombir~ant A HinfiI-AhaII DfA fragment. of the EBV genome (coordinates 107,946-109,843), that contains the EBNA-1 open reading frame, was subcloned using BamHI
linkers into the BamHI site of pUCl3 (pUCEBNA-1).

The agtll-EBNA-1 recombinant was constructed by inserting the 600 by SamI-BamHI fragment of pUCEBNA-1 (EBV coordinates 109,298-109,893) into the EcoRI site of agtll using an EcoR.I linker (GGAATTCC). A phage recombinant containing, the EBNA-1 insert in the sense orientation was isolated by immunoscreening with EBNA-1 antibodies (see below). In this recombinant, the carboxy-terminal region of EBNA-1 (191 amino acids) is fused in frame to the carboxy-terminus of (3-galactosidase.
agtll_cDNA-Expression_.Library The human B cells (RPMI 4265) cDNA library constructed in the expression vector agtll was purchased from Clontech Laboratories, Inc. The 15. library contains approximately 9x105 independent clones and has an average insert size of 1.2 kb.
E. Coli Strains The standard pair' of agtll host strains, Y1090 and Y1098, were employed. The former was used to 2f screen agtll recombinants and the latter to generate alysogens for the analysis of p-gal fusion proteins.
Plasmids The plasmid pUCoriPl was constructed by sub-cloning the EcoRI-Ncol fragment from the oriP region 25 of the EBV genome into the SmaI site of pUCl3. This fragment contains 20 high affinity binding sites for 934pg99 EBNA-1. pUCoriP2 was derived from pUCoriPl by subcloning of an oriP fragment (EcoRI-BstXI) of the latter into the SmaI site of pUCl3. pUCoriP2 con-tains 11 high affinity binding sites for EBNA-1.
pUCORIa2 was made by insertion of a synthetic binding site for the bacteriophase a0 protein (AAATCCCCTAAAACGAGGGAfAAA) into the SmaI site of pUCl3. The complementary oligonucleotides were a gift of R. MacMacken. pUCMHCI and pUCmhcI were 1C~ constructed by insertion of the following oligo-nucleotides:
GATCCGGCTGGGGATTCCCCAfCT GATCCGGCTGcGGATTCCCaATCT
GCCGACCCCTAAGGGGTAGACTAG GCCGACgCCTAAGGGtTAGACTAG
into the BamHI site of pUCl3. The wild type sequence 1~~ is a binding site for H2TF1 and NF-xB. pUCOCTA is a similarly constructed pUCl8 derivative that contains a synthetic recognition site (ATGCAAAT) for the mammalian octamer binding protein(s). The plasmids p190H2KCAT (-190 to +5) and p138H2KCAT (-138 to +5) 0 contain 5'-deletions o~f the H-2K.b gene promoter fused to the coding sequence. for chloramphenicol acetyl transferase. All plasmids DNAs were purfied by an alkaline lysis protocol followed by two bandings in CsCl-EtBr gradients.

Binding-Site_Probes-Competitor-D,NAs The MHC, mhcl, on and OCTA probes were generated by digesting; the corresponding pUC plasmids with EcoRI and HindIII. The resulting products were end-labeled with [a-32PJdATP using the large fragment of E. coli DNA polymerase I. dCTP, dGTP and dTTP
were included in these reactions so as to fill in the ends of the restriction fragments. The labeled fragments were separated by native polyacrylamide gel 1(~ electrophoresis. The binding site fragments (60-75 bp) were eluted from t:he gel and purified by ELUTIpTM
(Schleicher and Schuell) chromatography. Using high specific activity [a-32PJdATP (5000 Ci/mmol), typical labelings yielded DNA probes with specific activities 1, of 2-4x107 cpm/pmol.
To generate the o~riP probe, pUCoriP2 was di-gested with EcoRI and HindIII, and the oriP fragment (-400 bp) isolated by low melt agarose gel electro-phoresis. This DNA fragment was then digested with 2() HpaII and the products labeled as detailed above.
The smaller of the two HpaII fragments (--90 bp) was isolated for use as th.e oriP probe. The MHCg probe was prepared by digesting p190H2KCAT with XhoI was labeling as before. T'he labeled DNA was then di-gested with HincII and. the 90 by probe fragment purified as before. This probe contains sequence from -190 to -100 of t:he upstream region of the H-2Kb gene.
The ~6MHCg (-190 to +270) and ~llMHCg (-13S to +270) competitor DNAs were prepared by digesting the 1340ggg plasmids, p190H2KCAT and p138H2KCAT, with XhoI and EcoRI. The H2KCAT fragments were isolated by low melt agarose gel electrophoresis.
Results Specific-Detection_of_a_a_Recombinant_Expressing A model system was used to test the notion that a recombinant clone encoding a sequence-specific DNA
binding protein could be specifically detected with a recognition site probe. The Epstein-Barr virus nuclear antigen (EBNA-1) was selected as the model protein. EBNA-1 is required for maintenance of the EBV genome as an autonomously replicating plasmid in human cell lines. It is also a transactivator of viral gene expression. The carboxy-terminal region of EBNA-1 (191 amino acids) has been expressed in E.
coli as a fusion protein and shown,to encode a sequence-specific DNA binding domain. The fusion protein binds to multiple high affinity sites at three different loci in the EBV genome. Two of these loci consitute a cis-acting element' required for maintenance of the plasmid state (oriP). In the agtll-EBNA-1 (aEB) recombinant the carboxy-terminal region of ENBA-1 was fused in frame to the carboxy-terminus of /~-galactosidase (Figure, 16). A lysogen harboring the aEB phage conditionally expressed a ~3-gal-EBNA-1 fusion protein of expected size (ap-proximately M.W. 145"000) that accumulated to a level of about 1$. The DNA binding activity of the fusion 1340ggg protein was assayed with a segment of oriP DNA that contained two high affinity sites for EBNA-1 (Figure 16). Extracts of agtll and aEB-lysogens were in-cubated with labeled oriP DNA and the products resolved by native polyacrylamide gel electro-phoresis. With the aEB extract, a distinct set of protein-DNA complexes was observed. The formation of these complexes was specifically competed by an excess of plasmid DNA containing EBNA-1 binding sites. Thus, the ~-gal-ENBA-1 fusion protein has the expected sequence-specific DNA binding activity.
To establish conditions for detection of EB
plaques with probes of oriP DNA, protein replica filters were generated from platings of the phage.
These filters were screened with a variety of pro-tocols using oriP or control DNAs. Under a defined set of conditions (see Experimental Procedures), aEB
plaques can be specifically detected using radio-labeled oriP DNA. The control probe (ori) contains a high affinity binding site for the bacteriophage a0 protein. The specific array of spots generated by the oriP probe corresponded to plaques on the master plate as well as to spots that reacted with antiserum to ~i-gal on the replica filter. Furthermore, in a similar experiment the oriP probe did not detect control agtll plaques. From a series of such ex-periments the following conclusions were drawn; (i) the specific detection of aEB plaques requires a DNA

probe with at least one binding site for EBNA-1 (a duplex 30-mer with a consensus binding site sequence gave a signal comparable to a probe containing two or more binding sites), (ii) DNA probes longer than 150 by yield higher 'non-specific signals, (iii) the ad-dition of an excess of non-specific competitor DNA
[poly(dI-dC)-pol:y(dI-dC)] to the binding solution reduces the non-specific signal, and (iv) both specific and non-specific interactions of the DNA
probe with proteins on the replica filter are re-versible. In view of this latter point and the fact that non-specific interactions typically have much shorter half-lives than the specific interactions, sequence-specific binding proteins can be detected after a suitable wash time.
Given the ability to specifically detect DEB
plaques with oriP DNA, reconstruction experiments were carried out to test the sensitivity of the screen. In these experiments the aEB phage was mixed with an excess of control agtll recombinants. R Mica filters generated from. such mixed platings wer ~'e screened initially with oriP DNA and subsequently with antibodies to EBNA-1. In an experiment where approximately 5,000 phage were plated, with aEB being present at a frequency of 10 2, a identical number of positives (approximately 50) were detected with both oriP DNA and antibody probes. In fact, the two patterns are superimposable. Furthermore the signal to noise ratio of the DNA binding site probe was 3C~ better than that of the antibody probe. Thus it is 1340ggg possible to screen for the aEB phage with an oriP DNA
probe.
Screening_for-Mammalian-Clones-Encoding Seguence-Specific_DNA-Binding_Proteins A agtll library of cDNAs prepared with mRNA from human B cells was screened using the conditions developed with aEB. The DNA probe used in the screen contained a regulatory element from a mouse MHC class I gene (H-2Kb, Figure 17). This sequence (MHC) was synthesized and cloned into the pUC polylinker. The mammalian transcriptional regulatory factors H2TF1 and NF-xB bind with high affinity to this MHC ele-ment. In a screen of 2.5x105 recombinants, two positive phage, designated ah3 a.nd ah4, were iso-lated. In an autoradiogram of a. filter from the primary screen, a positive spot resulted in the isolation of ah3. Partially purified ah3 and ah4 phage were challenged with other DNA probes to determine if their detection was specific for the MHC
probe. ah3 and ah4 were not detected by the on probe. These phage were also not detected by labeled pUC polylinker DNA or by a related probe (OCTA) containing a recognition site for the immunoglobulin octamer binding protei.n(s). A mutant MHC binding 25, site probe (mhcl Figure 17) was used to more strin-gently test the sequence-specificity of the pre-sumptive fusion proteins. The mhcl probe did not detect either ah3 or a,h4 plaques. These data 1344ggg strongly suggested that the two phage express proteins that bind specifically to the MHC element.
Characterization_of_th.e-DNA_Binding-Proteins_Encoded by-ah3_and_ah4 Direct evidence that the ~-gal fusion proteins encoded by ~h3 and ah4. are responsible for the sequence-specific DNA binding activities was obtained by screening Western blots with DNA and antibody probes. Lysogens of agtll, ah3 and ah4 were isolated 1C1 and induced to generate high levels of their re-spective ~-gal proteins. Western blots of proteins from induced lysogens were prepared and the im-mobilized proteins were briefly denatured with 6M
guanidine and then allowed to renature (see Ex-l~; perimental Procedures above). This treatment in-creased the recovery of active molecules. Two equivalent transfers were initially probed with either the MHC element. or the OCTA control DNA. A
set of four bands specific to the MHC probe and the 2C1 ah3, ah4 tracks was of>served. The two largest species of this set are labeled P1 and P2. The same transfers were then probed with antibodies to ~-gal.
A pair of novel fusion protein bands was observed with each of the two recombinant: lysogens. These bands corresponded to the species P1 and P2 detected with the MHC probe. 7~his shows that ah3 and ah4 encode (3-gal fusion proteins which bind specifically to the MHC element DNA. The two phage may be iden-tical since they encode the same size fusion 1340ggg proteins. P1 (approximate m.w. 160,000) probably represents the full length fusion protein whereas P2 is a presumptive proteolytic cleavage product. Since the ~-gal portion of this fusion polypeptide has a molecular weight of approximately 120,000, the cDNA
encoded portion must have a molecular weight of 40,000.
A gel electrophoresis DNA binding assay was used to confirm the sequence specificity of the ah3 and 1C ah4 fusion proteins as well as to better define their recognition properties. Extracts derived from the agtll, ah3 and ah4 lysogens were assayed, with the MHC probe. A novel DNA binding activity was detected specifically in extracts of the ah3 and J~h4 lysogens.
This activity was IPTG inducible indicating that it was a product of the lacZ fusion gene. A competition assay indicated that the activity represented a sequence-specific DNA binding protein. Two 5' deletion mutants of th.e H-2Kb genomic sequence was 2Ci used as competitor DNA.s. The segment 6MHCg extends to 190 nucleotides upstream of the transcription start site and contains the MHC sequence element.
The segment O1IMHCg, o~n the other hand, only contains 138 nucleotides of sequences upstream of the initia-2~~ tion site and therefore lacks the MHC element. In-creasing amounts of ~6~MHCg specifically competed for the binding of the ah?~ fusion protein to the MHC
element oligonucleotide probe while the control O1IMHCg did not compete. It should be noted that the 30 sequences flanking the: MHC element in the probe used for the initial screening, the cloned oligonucleo-tide, are totally difference from the sequences flanking the same element in the genomic probe, 06MHCg. Therefore, th.e fusion protein appears to exclusively recognize the common MHC element. This was confirmed by a direct DNA binding assay with a genomic sequence probe (MHCg) containing the MHC
element. Both the oligonucleotide (MHC) and genomic (MHCg) probes gave rise to similarly migrating complexes. Furthermore, a double base substitution mutant (mhcl, Figure 17) abolished recognition by the fusion protein. The mmtant sequence contains a transverion in each half of the symmetric MHC ele-ment. These changes destroy the symmetry of the element and abolish binding by either H2TF1 or NF-~cB.
The immunoglobulin ~c chain gene enhancer con-tains a binding site ( EN) for NF-rcB. This site is related in sequence to the MHC element but is recog-nized by H2TF1 with a 10 to 20 fold lower affinity (Figure 17). A mutant.~s enhancer (~cEN) has been characterized both in vivo and in vitro. This mutant sequence has no B cell specific enhancer activity and is not bound by NF-rcB. The mutant contains clustered base substitutions and. an insertion of a base pair in one of the two symmetric half sites (Figure 17). The binding of the ah3 fusion protein to the wild type ~c-element and the mutant version was tested. The ~cEN
probe generated a complex with a mobility similar to those obtained with the MHC probes. No specific complex was formed with the mutant ~c-enhancer DNA.
Experiments in which the MHC and w-enhancer binding sites were tested for competition with binding of the MHC probe showed that the fusion protein bind with 2-5 fold higher affinity to the MHC site (data not shown). The ~cEN site differs, in part, from the MHC
site by the substitution of two adenine residues for guanine residues. As discussed below, these guanine residues are probably contacted by the fusion.
The contacts of the fusion protein with the MHC
element were probed chemically by modification of the DNA with dimethylsulfa.te. After partial methylation at purine residues, th.e modified probe was used in the gel electrophoresis DNA binding assay. Free (F) and bound (B) probe DNA was recovered, subjected to chemical cleavage at m~.ethylated interference experi-ment. On both the coding and non-coding strands strong interference wa.s detected. when any of central guanine residues of each putative half site was modified at the N-7 position in the major groove.
Weaker interference wa.s observed when the external guanine residue in either putative half site was similarly modified. Thus the fusion protein appears to symmetrically contact the MHC element in a manner similar to both H2TF1 and NF-~cB.
Hybridization_Analysis.-with-the-cDNA_Segment-of_the Recombinant-Phage The recombinant phage ah3 and ah4 contain cross-hybridizing and equivalent size (approximately 1Kb) cDNA segments. The inserts also have indis-tinguishable restriction maps and therefore appear to be identical. Southern blot hybridization confirmed that these cDNA segments are homologous to sequences in the human genome. The patterns of hybridization to restriction digests of genomic DNAs of various human cell lines ar.e identical. Furthermore, the fact that restriction digests with Bam HI (no site in cDNA) and Pst I (on site in cDNA) both generate two prominent bands suggests that the cDNAs are derived from a single copy gene. A similarly simple hy-bridization pattern is observed on probing the mouse and rate genomes.
The expression of the human gene was analyzed by 1~~ Northern blot hybridization. A single, large transcript (approximately 10 kb) was observed with polyA(+) RNA from both B (X50-7) and non-B human cells (HeLa). 'this transcript is moderately abundant in both cell types. Since the cDNA library was 2Q constructed by ~~ligo dT.priming, we were probably fortunate to obtain the coding region for the DNA
binding domain within the 1 kb segments of the recombinant pha,ge. However, this only illustrates the power of th~~ screening strategy for the isolation 25 of clones encoding sequence-specific DNA binding domains.
Discussion A novel stoategy is disclosed for the molecular cloning of genes encoding sequence-specific DNA

-,o_ binding proteins. This strategy can be used to isolate genes specifying mammalian, transcription regulatory proteins. An important step in this approach is the detection of bacterial clones syn-thesizing significant levels of a sequence-specific DNA binding protein by screening with a labeled DNA
binding site probe. This approach is similar to that previously developed for the isolation of genes by screening recombinant libraries with antibodies specific for a given protein. In fact, the phage expression vector, agtll, developed previously for immunological screening can be in this approach.
The feasibility of the strategy was established by the specific detection of a phage recombinant, 1~~ aEB, encoding a ~-gal-EBNA-1 fusion polypeptide with oriP DNA. Conditions have also been developed for the selective detection of E. coli colonies expres-sing high levels of EE'~NA-1 or the bacteriophage a0 protein with their respective binding site DNAs. In 2() these cases, a plasmid expression vector was em-ployed. Using the conditions developed with aEB, we have screened phage cDNA libraries with three dif-ference DNA probes. Screening with a probe con-taining the H2TF1 site: in the MHC class I gene H-2Kb led to the isolation of two identical clones that specify a putative trs~nscription regulatory protein (properties discussed below). I.n similar screens with two other DNA probes, positive recombinant phage 1340ggg were also isolated at a frequency of approximately 1/100,000. However, the DNA binding proteins encoded by these phage do not appear to recognize specific sequence elements but rather to bind sequence non-specifically to either single strand or double strand DNA. Although detection of these types of clones represented a troublesome background in this study their isolation suggests that recombinants encoding different types of DNA. binding proteins can be detected by such functional screens of expression libraries. In future screens for recombinants encoding site-specific DNA binding proteins, the detection of these other types of clones might be selectively suppressed. by inclusion of a non-specific 1~; competitor DNA that is structurally more similar to the probe than poly(dI-dC)-poly(dI-dC).
The prospects for' the isolation of other cDNAs encoding sequence-specific binding protein by this strategy can be assessed by examining the three 2G assumptions on which i.t is based; (i) functional expression of the DNA binding domain of the desired protein in E. coli, (i.i) a strong and selective interaction of the binding domain and its recognition site, and (iii) high level expression of the DNA
25 binding domain. A number of eukaryotic sequence-specific DNA binding proteins have been functionally expressed in E. coli. These include the proteins GAL4, GCN4 and MAT 2 of yeast, f.'tz of Drosophila, TFIIIA of Xenopus, E2 of the bovine papilloma virus 30 and EBNA-1 of the Epst:ein Barr Virus. In most cases, 1 ~40ggg the functional DNA binding domain is contained within a short tract of amino acids. Thus it is reasonable to expect the functional expression in E. coli of the sequence- specific DNA, binding domain of most eukaryo-tic regulatory proteins. The equilibrium association constants of site-specific DNA binding proteins range over many orders of magnitude (10~-012 M). The following analysis suggests that: successful screening may be restricted to proteins with relatively high 1C1 binding constants. If' a regulatory protein has an association constant o~f 1010 M, then under the screening conditions (the DNA probe is in excess and at a concentration of (-1010 M) approximately half of the active molecules on the filter will have DNA
l~; bound. Since the filters are subsequently washed for 30 minutes, the fraction of protein-DNA complexes that remain will be determined by their dissociation rate constant. Assuming a diffusion limited associa-tion rate constant of 2G 10~ M 1 S 1, the dissociation rate constant will be 3 S 1. Such a protein-DNA complex will have a half life of approximately 15 minutes. Thus only a quarter of the protein-DNA complexes will survive the 30 minute wash. For a. binding constant of 109 M, 2~ then only about a tenth of the active protein mole-cules will have DNA bound and much of this signal will be lost since the. half-life of these complexes is approximately 1.5 minutes. Isolation of recombi-nants encoding proteins with binding constants of 109 1340ggg or lower may be possible given that the binding of probe to less than l~ of the total fusion protein within a plaque can be detected. The sensitivity of the current methodology for low affinity proteins could be significantly enhanced by covalent stabili-zation of protein-DNA complexes. This might be accomplished by procedures such as UV-irradiation of pre-formed complexes. Since the binding constants of regulatory proteins are dependent on ionic strength, 1C1 temperature and pH, these factors might also be manipulated to enhance detection.
The successful detection of aEB and ah3 recom-binants with DNA binding site probes required high level expression of their fusion proteins. In both 1~ cases, the fusion proteins accumulate, after in-duction, to a level of about 1~ of total cellular protein. This level of recombinant protein expres-sion is typical of agtll as well as other E. coli vectors. The strategy of cloning a gene on the basis 20 of specific detection of its functional recombinant product in E. coli has considered potential. Indeed, while our work was in progress, this approach was used by other to isolate clones encoding a peptide acetyltransferase and a calmodulin-binding protein.
25 Direct screening of clones encoding recombinant protein products has also been used to isolate ras GTP-binding mutants.
The ah3 recombinant expresses a ~-gal fusion protein that recognizes related transcription control 30 elements in the enhanc.ers of the MHC class I and immunoglobulin rc-chain. genes (see Figure 17 for sequences). This protein also binds a similar element in the SV40 en.hancer 72bp repeat. Further-more, there are two putative binding sites in the long terminal repeat (LRT) of the HIV genome (Figure 17). One of these is identical to the site in the SV40 enhancer and therefore should be recognized by the fusion protein. fhe existence of a clone such as ah3 was anticipated since it had previously been shown that a common factor, NF-xB, binds to the three related elements in th.e enhancer, the SV40 72 by repeat an the HIV-LTR. Interestingly, these three binding sites are more closely related to one another than they are to the M:HC site (figure 17). The former set can be viewed as variants of the MHC site which exhibits perfect two-fold symmetry. It should be noted that the pUC polylinker contains the se-quence, CGGGGA, which is a variant of one of the symmetric halves (TGGGGA) of the MHC element. The fusion protein does not bind with detectably affinity to the pUC polylinker. Thus, a high affinity inter-acter appears to require both symmetric halves.
Even though the above control elements represent quite similar sequences, they function in very different regulatory capacities. The MHC element is a component of an enhancer that functions in a variety of cell types that express MHC class I genes.
The ~c-element, on the other hand, is a component of a cell-type specific enhancer that functions only in B
cells. The activity of this enhancer is induced in pre-B cells upon their differentiation into mature B
lymphocytes. Such differentiation, in vitro, is accompanied by transcriptional activation of the chain gene. The x-element appears to dictate the B
cell specificity of the x-enhancer. The different modes of functioning of the MHC and x-elements are correlated with the properties of their corresponding recognition factors, H.2TF1 and NF-xB. H2TF1 activity is detected in a variety of differentiated cell types 1G and this protein appears to stimulate MHC class I
gene transcription approximately 10-fold. On the other hand, NF-xB activity is detected only a mature B cells. In addition, this activity is induced during differentiation, of pre-B cells to mature 1'. lymphocytes. Finally, NF-xB activity is also induced by phorbol ester treatment of non-B cell lines (HeLa, Jurkat). In the case of Jurkat cells, a T4+ human T
cell line, NF-xB appea.rs to stimulate the transcrip-tional activity of the HIV-LTR. It should be noted 20 that induction of NF-xB in non-B cells does not require new protein synthesis. Thus the protein for NF-xB must exist in cells before induction and the activated by a post-translational modification.
The DNA binding properties of the fusion protein encoded by the recombinant ah3 overlap those of H2TF1 or NF-xB. Mutants of the MHC and x-elements that are not recognized by H2TF1 or NF-xB are also not bound by the fusion protein. The recombinant protein binds the MHC element DNA with 2-5 fold higher 3() affinity than the x-element. In this regard, the fusion protein ha.s relative affinities intermediate between those of H2TF1 and NF-xB. H2TF1 binds the MHC element with 10- to 20-fold higher affinity than the x-element while NF~-xB recognizes both elements with roughly equivalent affinity. This intermediate relationship is also observed in the comparison of the methylation interference patterns of the three DNA binding activities. Methylation of any of the central six guanine re:aidues in the MHC site strongly interfers with tt-ie binding of all three activities.
Methylation at either of the two external guanines partially interferes with recognition by the fusion protein. In contrast, H2TF1 binding is strongly suppressed upon methyl,ation of either of these residues while NF-xB binding shows little perturba-tion upon this modification. This analysis of the three DNA bindink; activities is limited by the use of cell extracts anti not purified proteins. Further-more, the properties of a recombinant protein may be different from those of its native counterpart.
Antibodies raised against the ah3 fusion protein will be useful in clarifying its structural relation-ship with H2TF1 ~~nd NF-xB. A definitive relationship will emerge from a comparison of the deduced amino acid sequence of the cDNA and the protein sequences of H2TF1 and NF-xB. It should be noted that in terms of protein expre:>sion, both H2'~F1 and NF-xB are ~ ~4~999 present in a wide variety of mammalian cells.
Furthermore, the DNA binding specificities of these two factors are remarkably similar. These facts as well as the observations that the cDNA in ~h3 hybridizes to a ;single copy gene and to a single mRNA
in both B and non-B cells suggest that all three binding activities .may be products of the same gene.
This hypothesis would imply that H2TF1 and NF-rcB
represent altern;itive modifications of a common protein.
B. Cloning_of_IgNFB
Methods DNA_Seguencing DNA sequencj~ng was performed on double stranded plasmid DNA temp7~ates according to the Sanger dideoxy-nucleotide proto<:ol as modified by United States Biochemical for use with bacteriophage T7 DNA
polymerase (Seque~nase~'"). The entire sequence was confirmed by sequencing the opposite strand and in the GC-rich regions by sequencing according to Maxam and Gilbert (Methods-Enzymol_, 65: 449-560, (1980).
Plasmids-Constructions cDNA's were subcl~oned from agtll to pGEM4 (Promega), and tluese plasmids were used for DNA
sequence analysis and in_vitro'transcription.
Plasmid pBS-ATG was kindly provided by H. Singh and K. LeClair and was constructed by ligating a 27 by long oligonucleotide containing an ATG codon sur-rounded by the appropriate boxes for efficient initiation, TGCACACCAfGGCCATCGATATCGATC, into the Pstl site of pBS-/+Bluescript~'plasmid (Stratagene).
The expression vector pBS-ATG-oct-2 depicted in Figure 19 was designed for transcription and trans-lation in_vitro .and was constructed by cleaving pBS-ATG with Sma:L and ligating the blunt-ended EcoRI
1.2 kb cDNA fragment from plasmid 3-1 (position 655 to 1710 in Figure 18A).
In_Vitro-TranscrilptionLTranslation In_viCro transcription and translation reactions were performed a:. recommended by the manufacturer (Promega).
i DNA-Binding-Assay=
The EcoRI/Hi.ndIII 50 by fragment containing the wild type octanuc:leotide sequence ATGCAAAT in the BamHI site of pUC;lf3 polylinker was 32P-labeled (50,000 cpm/ng) a.nd 1 ng DNA probe was incubated with 1.~1 of the reacted/un:reacted reticulocyte lysate.
The binding reactions were incubated at room tempera-ture for 30 win and contained 10 mt-t Tris HC1 pH7.5, 50 mM NaCl, 1 mPt DTT, :l mM ED'i'A p118, 5$ glycerol, 25 ~g/ml sonicated denatured calf thymus DNA in 2.5 pg/ml sonicated native calf thymus DNA as nonspecific competitors. The complLexes were resolved by electro-phoresis in 4$ polyacrylamide gel (acrylamide:
>..;.

bisacrylamide weight ratio of 29:1), containing as buffer 25 mM Tris HC1 pH8.5, 190 mM glycine, 1 mM
EDTA buffer as previously described (Singh _e_t__a_1_., Nature 319: 154-158 (7.986)).
Purification_of_NF=A2 NF-A2 was purified to >90% homogeneity from nuclear extracts de,ri«ed from the human Burkett's lymphoma cell line, BJ~AB. Purification was ac-complished by sequential fractionation on Zetachrom"'' QAE discs (Cuno Inc.), heparin sepharose (Pharmacia), ssDNA cellulose (Pharmacia), and on a DNA affinity column which contained) an immobilized double stranded (ds) segment containing the octanucleotide sequence.
In_vitro translated, 35S-methionine-labeled, oct-2 protein was purified by chromatography on dsDNA
cellulose followed by affinity chromatography on the octanucleotide DIVA affinity column.
Tryptic-Digestio~as_of_NF_A2-and_oct=2_Protein Tryptic digests were performed at room tempera-tune in a buffer consisting of 20 mM Hepes,KOH, pH

7.9, 20%.glycero:l, 0.5 M KC1, 0.'? mNI EDTA, 0.5 mM
DTT. Aliquots o:E purified NF-A2 (.-250 ng) or of affinity purified oct-2 protein (90,000 cpm) were incubated with v~3rying amounts of trypsin (affinity purified trypsin was a gift of Dan Doering). After 60 minutes, reac~_ions were terminated by the addition of 2.5 volumes o1~ SDS-PAGE sample buffer and were boiled for 5 mi.n:Ltes. The tryptic digests were resolved Otl 10% polyacrylamide~ gels. Tryptic frag-meats of NF-A2 wE~re visualized by silver-staining.

Tryptic fragments of 'l5S-methionine labeled oct-2 protein were visualiz<>.d by autoradiography after treatment of the gel with En'ihance'~°' (Dupont) .
Results To clone the gene encoding the lymphoid-specific octamer binding pro.te:Ln, IgNF-B (NF-A2), a randomly primed, non-size selected cDNA library in agtll was constructed using cytoplasmic poly (A)-containing mRNA from a human B cell lymphoma cell line, BJAB.
We had previously observed that this cell line contained a particularly large amount of NF-A2 when 28 lymphoid cell linea were surveyed. By randomly priming the cDNA. synthesis we expected to obtain recombinant phag,e encoding the octamer motif binding domain even if that domain was encoded by the 5' end of a long mRNA. The randomly primed cDNA library in agtll was generated by standard methods (Gubler, U.
and Hoffman, B.J., Genes 25:263-269 (1983)). Random hexamers (Pharmacia) were used to prime the first strand cDNA synthesis. The unamplified library contained 500,0f0 recombinants. This library was screened by the method described above using a radiolabelled D~fA probe consisting of four copies, in direct orientation, of a 26bp oligonucleotide derived z5 from the Vk41 promoter. The probe was constructed by cloning four copies of the oligonucleotide in direct orientation into the BamHl site of the pUC polylinker and radiolabelli.ng the 112 by Smal-Xbal fragment.
The library was screened with the tetramer probe (at 1X106 cpm/ml) as described above for the cloning of NF-rcB with the following modification. Previous screens using po:Ly(dI-dC)-poly(dI-dC) as the non-specific competitor DNA yielded recombinant phage encoding single atranded DNA binding proteins. The signal from these phage but not phage encoding sequence-specifi~~ DNA binding proteins could be efficiently competed with denatured calf thymas DNA
(5 ~cg/ml) and therefore this nonspecific competitor was substituted for poly(dI-dC)-poly(dI-dC) in all subsequent screens.
From a primary screen of 450,000 phage plaques, three plaques w ere isolated which bound this tetramer probe. Two of these phage, phage 3 and phage 5, were found to give plaques that bound specifically to the tetramer probe in that they did not bind DNA probes which lacked the octamer motif. These two phage bound probes containing one copy of the rc promoter octamer motif with a much lower affinity than they bound the tetramer probe. Even when four-fold more monomer probe was used then tetramer probe, the tetramer probe still gave a greater signal suggesting that the better 'binding of the tetramer probe was not merely a result of increasing the molar concentration of binding sites in the screen. Certainly in the case of phage 5, which showed dramatically better binding to the tetramer probe, it seems most likely that the tetramer probe was able to bind simultaneous-ly to multiple phage fusion proteins on the filter.

~ ~4ns9s -g2_ This multipoint ~3ttachment would be expected to dramatically decrease the dissociation rate and thus, increase the avidity of the interaction. Genes encoding DNA binding proteins with relatively low binding affinities could be cloned by screening agtll expression libraries with such multimer probes.
The specifi~~ity of the DNA binding proteins encoded by the recombinant phage was investigated by preparing extracts of induced phage lysogens.
Lysogen extracts from both phages bound to the tetramer probe in a mobility shift assay whereas lysogen extracts from non-recombinant agtll showed no binding to this probe. Only the phage 3 extract bound strongly to the ~c promoter probe. Because the 15, inserts of phage 3 and phage 5 (1.2 kb and 0.45 kb in size, respectively) were found to cross-hybridize by Southern blotting analysis, phage 3 was chosen for further analysis.
Phage 3 encoded an octamer binding protein as demonstrated by a competition mobility shift assay in which the lysogen extract was bound to the rc promoter probe in the presence of competing unlabelled DNA
fragments containing either the wild type or mutant octamer motifs. Phage lysogen extracts were prepared as described above for NF-~cB cloning. The extracts were assayed in a mobility shift assay as described above using the octamer-containing PvuII-EcoRl fragment from pSPIgVk as the radiolabelled probe.
Binding reactions were carried out in the absence or 30 presence of 24r~g of cold competitor DNA containing no 1340ggg octamer motif, the wild type octamer motif or mutant octamer motifs as described.
The wild type octamer motif competed efficiently for binding but the octamer motifs containing point mutations either did not compete or competed less well than the wild type motif. In fact, the two mutants which showed slight competition for the binding of the lysogen protein, TCATTTCCAT and ATATTGCAT, were the only mutants which somewhat competed the binding of NF-A1 and NF-A2 in a WEHI 231 nuclear extract.
The phage-encoded. octamer binding protein was further compared to Nf-A1 and NF-A2 using a methyla-tion interference footprinting assay. Methylation interference was performed as described using the non-coding strand of the octamer-containing PvuII-EcoRl fragment of pSPlgVk as radiolabelled probes.
The probes were partially methylated and used in preparative mobility shift DNA binding assays. DNA
present in the bound bands (NF-A1 and NF-A2 bands from a nuclear extract: from the BJAB cell line (or phage 3 lysogen extract bound band or free bands) was isolated, cleaved at t:he modified purine residues and subjected to denaturing polyacrylamide gel electro-phoresis. The footprint obtained using the lysogen extract was centered over the octamer motif and was very similar to the footprints of NF-Al and NF-A2 from a BJAB nuclear extract and from a WEHI 231 nuclear extract (see a.bove). Minor differences were seen between the footprints of the lysogen and nuclear extract proteins which could reflect changes in affinity and/or specificty of DNA binding as a result of fusion of the insert-encoded octamer binding protein with ~-galactosidase. Alternatively, the phage insert could encode an octamer binding protein distinct from NF-A1 and NF-A2.
The phage-encoded ~-galactosidase fusion protein was directly shown to be the octamer binding protein in the phage lys~ogen extracts. Phage lysogen ex-tracts were sub jected to SDS polyacrylamide gel electrophoresis and transferred to nitrocellulose filters. After a denaturation/renaturation procedure (Celenza, J.L. and Carlson, M. Science 233: 1175-1180 (1986)), the filters were probed with either the radiolabelled octamer-containing tetramer probe (OCTA) or a non-specific DNA probe (pUC). The OCTA
probe specifically bound to the ~-galactosidase fusion proteins of phage 3 and phage 5 to a much greater extent than the pUC probe thus formally showing that the octamer binding activity was encoded by the phage inserts. The apparent molecular weights of the largest fusion proteins of phage 3 and phage 5 lysogens are consistent with the entire phage inserts contributing coding sequences to the fusion proteins.
Prototeolysis was presumed to account for the heter-ogeneity in apparent molecular weight of the fusion proteins.
The insert of phage 3, which defines what we term the oct-2 gene, was used in a Southern blot analysis to probe human and mouse genomic DNA

_85_ digested with several restriction enzymes. Restric-tion enzyme digested ~;enomic DNA was electrophoresed through a 1% agarose ~;el and transferred to Zetabind'T' (CUNO Laboratory, Inc.) by standard techniques (Maniatis, 'C., Frisch, E.F, and Sambrook, J. Molecu-lar Cloning. A Laboratory Manual. Cold Spring Harbor Laboratory Press-, N,Y. (1982)). The phage 3 insert was radiolabelled by randomly primed synthesis using hexanucleotides (Pharmacia). Following standard prehybridization high-stringency hybridization (Maniatis, supra.) with the oct-2 probe the filters were washed with 0.2X SSC, 0.1% SDS or 2X SSC, 0.1%
SDS.
One or two bands were observed in each restric-tion enzyme digest which is consistent with oct-2 being a single genetic: locus. No rearrangements or amplifications of the gene were observed in a survey of 8 lymphoid and non-lymphoid cells lines including BJAB. The strength of-_ the signal on the mouse Southern blot at high stringency suggested that the gene is highly conserved between human and mouse.
The oct-2 cDNA sf:gment (1.2 kb) of phage 3 was used to identify additional overlapping recombinants in the same library. One of these phage (pass-3) contained a 1.8 kb DNA insert. Sequence analysis of the cDNA segment in the original agtll phage (3-1) revealed a long open reading frame (ORF) which was ended with multiple nonsense codons at its 3' ter-minus. Sequence anal~,rsis of the pass-3 segment yielded an identical sequence through the open ~ 340ggg reading frame but an abrupt transition to a novel sequence occurred at the C-terminus (Figure 18B; see below). The N terminus of the open reading frame in both of these cD~lA segments was not represented in the cDNA inserts. Additional recombinants from the agtll library were identified by screening with a probe from the N-terminal portion of the pass-3 segment. This resulted in the isolation of a 0.75 kb cDNA segment (pass-5.5) whose sequence extended the N-terminal portion of the previously identified open reading frame. In this cDNA segment, a nonsense codon is found 36 by upstream of the first AUG in the open frame. The sequence context of this AUG con-forms well to that e~:pected for an initiation codon (Kozak, Cell 44: 283-292 (1986)). Two other AUG
codons occur at positions 6 and 13 in the reading frame. Each of these also has an excellent context for initiation. The N terminus of the protein has been arbitrarily assigned to the 5'-most AUG codon.
The cDNA sequence extends 66 bases 5' from this position but the total length of the 5' untranslated region has not been determined.
The sequences of pass-5.5, pass-3 and 3-1 were combined to form an open reading frame encoding a protein of 466 amino acids in length as shown in Figure 18. Figure 18 shows the amino-acid sequence of oct-2 protein depicted in plain capital letters.
cDNA-clone pass-5.5 spans from position 1 (5' end) to position 750 (3' end). cDNA clone pass-3 5' end and 3' end are respectively at position 92 in Figure 18a _87_ and 1847 in Figure 18b. cDNA clone 3-1 starts at position 650 and ends at position 1710. The nucleo-tide sequence shown in. panel A was reconstructed by merging the DNA sequences from clone pass-5.5 from position 1 to 100, from clone pass-3 from 100 to 660 and from clone 3-1 from position 660-1710. Extensive nucleotide sequence overlaps were available to allow unequivocal merges. Sequence of protein encoded by the long overlapping open reading frame (LORF, 277aa) 1C1 is shown in italic letters. Wavy arrows delimit the glutamine (Q)-rich, glutamic and aspartic (E/D)-rich and glycine (G)-rich regions, respectively. Solid arrows delimit the helix-turn-helix motif. Boxed leucine (L) residues are spaced exactly by seven lc~ residues. Vertical arrow indicates the position where the nucleotide sequence diverges with that shown in panel B. Stars indicate stop codons.
Figure 18b shows the nucleotide sequence of the 3' terminus and predicted amino acid sequence of the 20 C-terminus derived from clone pass-3. The code is the same as in A and the vertical arrow denotes the divergence point.
Figure 18c is a schematic representation of the amino acid sequence deduced from oct-2 gene derived 25 cDNA. The code is as in panel A. The DNA binding domain is depicted as DNA and the region containing the four regularly spaced L residues is boxed-in.
LORF stands for long open reading frame, N stands for N-terminus and C for COOH-terminus.

_88_ Data presented below suggests that this ORF
encodes one form of Nf-A2 (oct-2). The amino acid sequence of oct-2 has several interesting features (Figure 18C). It contains three glutamine (Q) rich blocks (ranging from 50$ of Q content) in the N-terminal part of the polypeptide, beginning at nucleotide positions 376, 448, and 502, and a comparably acidic region [aspartic (E) or glutamic (D) amino acids] between positions 648 and 678.
Clusters of Q residues as well as E or D amino acids have been described previously in many transcription factors. Such acidic regions in other factors have been shown to be important in activation of transcrip-tion (Gill and Ptashne, Cell 51: 121-126 1987; Hope et-al., Nature 333: 635-640 (1988)).
The region of oct-2 responsible for sequence-specific DNA binding, depicted "DNA", is discussed below. Downstream of this position is a series of four leucine residues separated by exactly seven 2C amino acids (position 1227 to 1293 in Figure 18A). A
similar configuration of leucine residues in the transcription factor C/EBP has been suggested to form an amphipathic a-helical structure where the leucine residues are arranged along one side of the helix.
Two such helices are thought to interact by a "leucine zippper" mechanism generating a dimeric protein (Landsch~.~ltz et al., Science 240: 1759-1764 (1988); Landschultz et-al-, Genes & Development 2:
786-800 (1988)).

Consistent 'with this suggestion, no helix disrupting proline residue is present in oct-2 in the 22 amino acid tract defined by the four leucines.
However, unlike the first example of a "leucine zipper", protein C/EBP, the potential a-helical region in oct-2 does not possess a high density of paired charged residues which could stabilize the structure. Also, unlike the C/EBP protein, which binds DNA specifically as a homodimer probably by pairing through the "leucine zipper", the oct-2 protein appears to specifically bind DNA as a mono-mer. It is interesting to speculate that the "leucine zipper" region of oct-2 might be important for interaction 'with other proteins as there is no obvious reason to restrict the binding of such a structure to self-recognition.
Searches for sequence similarities in the GenBank library revealed that a region of the oct-2 protein from position 952 to 1135 was distantly related to a family of proteins containing homeo-boxes. The 60-residue homeobox domain is highly conserved among 16 examples in different Drosophila genes (Gehring, Science 236: 1245-1252 (1987)). This level of conservation extends to homeobox sequences found in vertebrates and worms. Among this total family, nine of the 60 residues are invariant. The oct-2 protein only contains six of these nine resi-dues and four of these six sites are clustered in the subregion of the homeobox thought to be related to the helix-turn-h~alix structure (see Fig. 20). As shown in Figure 20 a 60 amino acid region of oct-2 contains 30o identit=y with the prototype homeobox sequence in the Antennapedia (Antp) protein.
Figure 20 shows the amino acid sequence align-s ment of the DNA binding domain of oct-2 factor with homeoboxes from An . (Schneuwly e~ ~, EMBO J. 5-733-739 (1986), cut (Blochlinger, Nature 333: 629-635 (1988), ~n (Poo.le _e~ ~, Cell 40: 37-43 (1985), proteins (boxed-in amino-acid sequences) and with homeobox-related amino acid sequences from the S.
cerevisae proteins Matal (Miller, EMBO J. #: 1061-1 065 (1 984) , MaJ~~&2 (Astell gt al. , Cell 27: 1 5-23 (1981), pho2 (Sengstag and Hinnen, Nucl Acids Res 15: 233-246 (19f37; Burglin, Cell 53: 339-340 (1988) and ~ eleyans protein mec-3 (Way and Chalfie, Cell 54: 5-16 (1988).. The nine invariant residues in canonical homeobox sequences Ante, cu , and en are listed below the boxed-in amino acid sequences and shown in bold print if present in the amino acid sequences. The stars indicate the hydrophobic amino acids that are critical for the protein to maintain the helix-turn-helix structure (Pabo and Sauer, Annu.
Rev. Biochem. 5~i:293-:321 (1984) ) . Solid arrows delimit the helix-turn-helix domain.
:?5 That the homeobo;~ specifies a sequence-specific DNA binding domain is most strongly argued by its homology with the DNA binding doma:i_n of the yeast mating regulatory proi~ein, MATa (Astell et al., Cell 27: 15:23 (1981 ) ; Scot=t and Weiner, Proc. Natl. Acad Sci. USA ~1: 4115-4119 (1984)), which also has homology through this subregion of the homeobox but does not conserve the other invarient of the homeo-box. The homolo;~ous regions in these proteins can be folded into a he:iix-turn-helix structure similar to that first identified in the structural analysis of phage a repressor (for a review, see Pabo and Sauer, Ann.-Rev.-Biochem. 53: 293-321 (1984)). A prediction of the most prob~ible secondary structure of oct-2 also revealed a helix-turn-helix structure between the residues of :isoleucine (position 1041) and cysteine (position 1090). Thus, by analogy, we propose that this region of oct-2 specifies the sequence-specific binding of the protein.
As mentioned above, sequences at the 3' end of the pass-3 recombinant abruptly diverged from that of recombinant 3-1 .at the position (1463) of its termina-tion codon (see vertical arrow in Figure 18B). The substituted sequences in the second recombinant, pass-3, extended the reading frame of the oct-2 related protein by an additional 16 amino acids. To rule out a possihle artifactual sequence generated by the insertion of a fragment during construction of the cDNA library, total polyA(+) RNA from the BJAB
cell line was analyzed by Northern blot with a DNA
fragment from th~~ novel 3' terminal portion of the pass-3 cDNA. This specific probe hybridized only to the two fastest migrating mRNAs of the total family of six mRNAs whi~~h were detected by hybridization with the total 3-1 cDNA. A similar specific probe was excised from the 3' terminus of the 3-1 cDNA. In contrast, this probe only hybridized to the two slowest migrating mRNAs in the total family of six.
This suggests that the two cDNA segments correspond to different populations of oct-2 mRNAs.
The proteins encoded by the two cDNAs should only differ at their C terminus by 16 amino acids or approximately 1.5 kD. In_vitro transcription/trans-lation of subfragments of the 3-1 and pass-3 recom-binants was used to confirm this prediction. Frag-ments representing the 3' portions of 3-1 and pass-3 were subcloned into the expression plasmid pBS-ATG.
The resulting pl.asmid DNAs were transcribed with bacteriophage T7 RNA polymerise and were subsequently translated in a reticulocyte system. The resulting polypeptides migrated with the mobilities of the anticipated mole,~ular weights 34kD and 32.4 kD. The polypeptide from the pass-3 cDNA was 1.6 kD larger than that from tine 3-1 cDNA. Both polypeptides specifically bound a probe containing the octa-nucleotide sequence, producing a readily detectable DNA-protein complex in the gel mobility assay. This suggests that th~~ oct-2 gene is expressed as a family of polypeptides in B-cells.
The potential significance of these additional 16 amino acids i,s unclear. These two cDNAs almost certainly differ by alternative splicing patterns of RNA transcribed :from the oct-2 gene. Furthermore, it is likely that the oct-2 gene encodes a more diverse set of mRNAs than those partially defined by these two cDNAs. Six different length mRNAs are produced at significant levels in mature B cells. The rela-tive amounts of these mRNAs vary between pre-B, B and plasma cell lines (Staudt et_al., Science 241:
577-580 (1988)). This population could reflect variations in sites of initiation of transcription and of polyadenylation as well as further differences in splicing patterns.
The expression of the oct-2 gene was assessed by Northern blot analysis of mRNA from 13 lymphoid and non-lymphoid cell lines and was found to be predominant-ly restricted to lymphoid cells. Poly(A)- containing mRNA (3pg, or 20~g) or total mRNA (30~g) was analyzed from the following cell lines. 1. NIH 3T3: mouse fibroblast; 2. 38B9: mouse pre-B cell line; 3.
WEHI 231: mouse mature B cell line; 4. A431: human epidermal cell line; 5. U1242: human glioma cell line; 6. RB27: human retinoblastoma cell line; 7.
Jurkat: human T cell line; 8. Namalwa: human mature B cell line; 9. BJAB: human mature B cell line (poly(A)-containing mRNA); 10. BJAB (total mRNA); 11. Hut78: human T
cell line; 12. HeLa: human cervical carcinoma cell line; 13. EL4: mouse T cell line. mRNA was electro-phoresed through a formaldehyde-containing 1.3~
agarose gel and transfered to a nitrocellulose filter by standard techniques (Maniatis, supra.). Following prehybridization, the filter was hybridized at high stringency with radiolabelled oct-2 probe (above).
The filter was washed in 0.2XSSC, 0.1~ SDS at 68°C.

~34pgg9 and autoradiogra~phed with an intensifying screen at -70°C for 24 hrs. The filter was stripped by washing in 50$ formamide, lOmM Tris (pH 7.4), 1mM EDTA at 68°C for 1 hr. and rehybridized with a radiolabelled rat alpha tubulivn cDNA probe (Lemischka, I.R., Farmer, S., Racaniello, V.R. and Sharp, P.A., J. Mol.
Biol. 151: 101-120 (1981)) to control for the amount of mRNA loaded.
All five B lymphoma cell lines, including pre-B
and mature B cell lines, and one of three T lymphoma cell lines expressed a family of 6 transcripts. Of the five non-lymyphoid cell lines tested, only a glioma cell line, U1242( ), showed detectable ex-pression of this gene. Even at low stringency we were unable to detect a transcript present in all cell lines which might correspond to NF-A1. The various transcripts, estimated to be 7.2 kb, 5.8 kb, 5.4 kb, 3.7 kb, 3.1 kb and 1.2 kb long, were ex-pressed in somewhat varying amounts relative to each other in the positive cell lines. Whether these transcripts represent alternative mRNA splicing or highly specific mRNA degradation remains to be determined. In 'this regard, it is interesting that highly purified ~~reparations of NF-A2 consist of three or more major polypeptides with distinct molecular weight: which could be the products of the family of transcripts that we have observed.
Previously, we and others (See above and Gerster, T. et a:L. EMBO J. 6: 1323-1330 (1987);
Landolfi et al., Nature 323: 548-51 (1986)) showed that the octamer binding protein NF-A2 varied considerably in expression among lymphoid cell lines.
We therefore investigated the relationship between levels of expression of the oct-2 gene and levels of NF-A2 as judged by mobility shift analysis. BJAB, the cell line which expressed the largest amount of transcript showed the largest amount of NF-A2.
Nuclear extracts from the pre-B cell lines, 38B9 and 70Z, showed very little NF-A2 and, correspondingly, expressed very little transcript (more poly(A)-containing mRNA from these two cell lines was loaded to see a readily detectable signal). Of the three T
lymphoma cell lines tested, Jurkat, HUT78 and EL4, EL4 was the only line that showed large amounts of NF-A2. Although NF-A2 was previously believed to be expressed only in lymphoid cells we found that nuclear extracts from the glioma cell line that expressed the oct-2 gene contained an octamer binding protein which comigrated with NF-A2 in the mobility shift assay. Nuclear extracts from two glioma cell lines which were negative for oct-2 expression did not contain NF-A2. We have at present no explanation for this apparant non-lymphoid expression of NF-A2 and the cloned octamer binding protein gene. Pre-viously, we had shown that NF-A2 but not NF-A1 was inducible in pre-B cells by treatment of the cells with bacterial lipopolysaccharide (LPS) and that this induction required new protein synthesis. Therefore, we prepared poly(A)-containing mRNA from the pre-B
cell line 70Z/3 before and after LPS treatment and observed that LPS increased the expression of the 1340ggg oct-2 gene. Thus, in every instance, the expression of the oct-2 gene correlated with the presence of NF-A2 and is thus a good candidate for the gene which encodes NF-A2.
Further evidence that the oct-2 gene encodes NF-A2 was discovered when the NF-A2 factor was purified from nu~~lear extracts of BJAB cells by conventional chromatography followed by multiple passages over an affinity column containing im-mobilized oligomers of the octanucleotide sequence.
The purified NF-A2 consisted of three bands as resolved by gel electrophoresis, a major band and two minor bands with deduced molecular weights of 61 kD
and 58 kD, and 6:3 kD, respectively. A cDNA (pass-3) for the oct-2 gene was inserted into the polylinker of the pGEM (Promega) expression vector. Translation of RNA transcribed from the SP6 promoter-pass-3 cDNA
construct yielded a major polypeptide of 61 kD which comigrated with the prominent polypeptide from the purified sample ~of NF-A2.
The mobility of a DNA-protein complex in the gel assay is primarily determined by the molecular weight of the protein. Complexes were generated with the affinity purified NF-A2 and the products of transla-tion in vitro of RNA from the oct-2 cDNA. These complexes co-migrated during electrophoresis in a native gel, again suggesting that the oct-2 cDNA
encodes the major form of the NF-A2 factor.
The affinity purified NF-A2 protein and the polypeptide translated in~vitro from the oct-2 cDNA

-97_ were also compared by partial tryptic digestion.
Samples from dif'feren't digestion times of NF-A2 were resolved by dena.turin;e~ gel electrophoresis and detected by staining with silver. The mobility of these partial fragments was compared with those observed after a~ parallel analysis of 35S-methionine labelled polypeF~tide from transcription/translation of the pass-3 cDNA in._vitro. The two samples generated a similar set of digestion fragments, again suggesting that NF-A2 is encoded by the oct-2 cDNA.
Protein sequence comparisons suggested that the DNA binding dome~in of oct-2 was specified by a domain (positions 952 t:o 1135) that was distantly related to both the helix-turn-helix structure of bacterial repressors and t:he homeobox-proteins. To directly test this analol;y a fragment of the cDNA encompassing this region (65'.i to 1710) was inserted into the expression vector pBS-ATG so that RNA could be transcribed from the truncated templates by bacterio-phage T7 RNA po:Lymerase as indicated in Figure 19.
The polypeptides translated in_vitro from these RNAs were tested for specific DNA binding by addition of the total trans:Latian mix to the DNA-protein gel assay. Polypeptides produced from RNAs terminating at positions 17:L0 (Kpnl), 1443 (Stul), and.1134 (Pstl) specific,311y bound the octanucleotide con-taining probe, while the polypeptide translated from RNA terminating at the 945 (Gagl) site did not specifically bind. The region'containing the helix-~ 340999 turn-helix portion ofd oct-2 is deleted in the latter protein. Since the truncated polypeptide encoded by RNA from the latter template was efficiently trans-lated in the reticulc>cyte reaction, this suggests that the specific binding of the oct-2 protein requires the helix-turn-helix structure.
Two distinct but similarly migrating protein-DNA
complexes were detectEed in the sample generated by translation of RNA from the Stul cleaved template.
Faint slower migratin.c~ complex comigrated with the complex generated with templates cleaved by Kpnl.
The presence of the two complexes in the Stul-sample is due to a partial digestion of the plasmid DNA.
The slower migrating complex is probably produced by protein terminated at the stop codon TAA located at position 1465. The faster migrating complex probably results from molecu7_es terminated at the Stul site.
This interpretation was supported by the resolution of two 35S-labeled pol.ypeptides during gel electro-phoresis of the Stul. sample and confirms the position of the termination codon of oct-2.
Many sequence-specific binding proteins have an oligomeric structure. For example, bacterial repres-sor proteins typically bind sites with two-fold Z5 rotational symmetry by forming a similarly symmetric dimer (Ptashne PEI., A Genetic Switch,Cell Press and Blackwell Scieni=ific :Publications, Cambridge, Massachusetts ('1986)). It should be noted that the binding site sequence of the oct-2 protein is not symmetric but o_Ligome.ric proteins could bind to non-symmetric sites.. Other examples of oligomerization ~of sequence-specific binding proteins are the GCN4 protein of yeast (Hope and Struhl, EMBO
J. 6: 2781-2784, (1987)) and the C/EBP protein of mammals. In the latter case, an a-helical region with an amphipat~hic character reflected in the spacing of four leucine residues by exactly seven residues is thought to be responsible for dimer formation (Landschultz et al., Science 240: 1759-1764 (1988); Landschultz et al., Genes & Development 2:
786-800 (1988)). A convenient assay for detection of dimerization of sequence-specific binding proteins is to co-translate :LtNAs encoding two different size forms of the protein and test whether protein-DNA
complexes with novel mobilities are generated (Hope and Struhl, EMBO_-J. 6: 2781-2784 1987). If only monomers bind to the probe, the sample containing the co-translated polypeptides will generate only the complexes detected when either RNA is assayed singular-ly. This was the case with combinations of different length RNAs transcribed from the oct-2 cDNA segment.
Specifically, cotranslation of RNAs from templates cleaved at Stul (1443) and Pstl (1134) did not generate novel bands in the gel mobility assay.
Thus, on the basis of this negative evidence, we suggest that a single molecule of the oct-2 protein is present in the resolved DNA-protein complexes and that it does not require dimerization for binding to DNA.
Anti-sera raised in rabbits against a bacterial fusion protein containing oct-2 encoded sequence ~34pgg9 (prepared employing the vector pRIT2T (Pharmacia)) recognized the native oct-2 protein in metabolically labeled (35S- methionine) human B cells.
The molecular cloning of a lymphoid-restricted octamer binding protein gene demonstrates that higher eukaryotes have adopted a strategy of genetic diversi-fication of transcriptional regulatory proteins which bind a common regulatory motif. The ubiquitous and lymphoid-specific octamer binding proteins have indistinguishable DNA binding sites yet appear to have distinct functional properties (Staudt, L.M. et al, Nature 323: 640-643 (1986)).
Structure-function analysis of cloned yeast transcrip-tion factors (Petkovich, M. et al., Nature-330:444-450 (1987); Giguere, V. et: al., Nature 330: 625-629 (1987)) and steroid receptor related transcription regulatory activity of a transcription factor often reside in discrete protein domains that can be experimentally interchanged. The present findings suggest that similar dliversification of function among proteins which bind the octamer motif has occurred during evolution. The octamer motif has been shown to be necessary and sufficient for lymphoid-specific promoter activity (Fletcher, C. et al., Cell 51: 773-781 (1987)) and NF-A2 has been shown to function as a transcription factor using octamer containing templates in vitro (Scheidereit, C. et al., Cell 51: 7ft3-793 (1987)). A further understanding of the 7_ymphoid-specific activity of immunoglobulin promoters may now come from an understandin<~ of t:he mechanisms underlying the lymphoid-specific expression of the oct-2 gene.
Deposit The clone ~,h3 was deposited at the American Type Culture Collection on February 12, 1988, and was assigned the ATCC accession number 67629 and the clone oct-2, also .referred to as ~,h3-1, was deposited at the American Type Culture Collection on February 12, 1988, and was assigned the ATCC accession number 67630.
Ectuivalents Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein. Such equivalents are intended to be encompassed by the following claims.

Claims (32)

1. An isolated transcriptional regulatory factor Oct-2.

2. The isolated transcriptional regulatory factor of claim 1, wherein Oct-2 comprises an amino acid sequence encoded by the nucleotide sequence set forth in figure 18.

3. The isolated transcriptional regulatory factor of claim 1, wherein Oct-2 is encoded by the nucleic acid having ATCC Deposit No. 67630.

4. The isolated transcriptional regulatory factor of claim l, which is encoded by a nucleic acid which hybridizes to the nucleic acid having ATCC Deposit No. 67630 or to a nucleic acid having the nucleotide sequence set forth in figure 18 or complement thereof.

5. The isolated transcriptional regulatory factor of any one of claims 1 to 4, which is a human factor.

6. The isolated transcriptional regulatory factor of any one of claims 1 to 5, which is a recombinantly produced protein.

7. The isolated transcriptional regulatory factor of claim 6, which is a mutated protein.

8. The isolated transcriptional regulatory factor of any one of claims 1 to 7, which is purified from a cell.

9. An isolated nucleic acid encoding a transcriptional regulatory factor or polypeptide of any one of claims 1 to 8.

10. The isolated nucleic acid of claim 9, which hybridizes to the nucleic acid having ATCC Deposit No. 67630 or a nucleic acid having the nucleotide sequence set forth in figure 18 or complement thereof.

11. The isolated nucleic acid of claim 10, which comprises a nucleotide sequence set forth in Figure 18.

12. The isolated nucleic acid of any one of claims 9 to 11, wherein the polypeptide comprises a domain of the transcriptional regulatory factor.

13. The isolated nucleic acid of claim 12, wherein the domain is a DNA binding domain.

14. The isolated nucleic acid of claim 13, comprising a nucleotide sequence from about nucleotide 952 to about nucleotide 1135 set forth in figure 18.

15. A vector comprising an isolated nucleic acid from any one of claims 9 to 14.

16. A cell comprising an isolated nucleic acid or vector from any one of claims 9 to 15.

17. A method for identifying an agonist or an antagonist of an IgNF-A, IgNF-E, Oct-2 or NF-kB transcriptional regulatory factor, comprising (i) contacting an IgNF-A, IgNF-E, Oct-2 or NF-kB
transcriptional regulatory factor and a nucleic acid comprising an enhancer or promoter sequence of a kappa light chain gene and a test compound, under conditions in which, but for the presence of the test compound, the transcriptional regulatory factor forms a protein-DNA
complex with the nucleic acid; and (ii) determining the level of protein-DNA complex formed in the presence and in the absence of the test compound, such that a different level of the protein-DNA complex in the presence, relative to the absence, of the test compound, indicates that the test compound is an agonist or an antagonist of the transcriptional regulatory factor.

18. The method of claim 17, wherein the enhancer or promoter sequence comprises the nucleotide sequence GGGACTCCC.

19. The method of claim 18, wherein the transcriptional regulatory factor is NF-kB.

20. The method of claim 17, wherein the enhancer or promoter sequence comprises the nucleotide sequence ATTTGCAT or inverted sequence thereof.

21. The method of claim 20, wherein the transcriptional regulatory factor is IgNF-A.

22. The method of claim 20, wherein the transcriptional regulatory factor is Oct-2.

23. A method for identifying an agonist or an antagonist of an IgNF-A, IgNF-E, Oct-2 or NF-kB transcriptional regulatory factor, comprising (i) contacting a cell or cell extract comprising a reporter gene operably linked to a nucleic acid comprising a promoter or enhancer whose activity is dependent on the transcriptional regulatory factor, with a test compound; and (ii) determining the level of expression of the reporter gene in the presence and in the absence of the test compound, such that a different level of expression of the reporter gene in the presence relative to the absence of the test compound indicates that the test compound is an agonist or an antogonist of the transcriptional regulatory factor.

24. The method of claim 23, wherein the enhancer or promoter sequence comprises the nucleotide sequence GGGACTCCC.

25. The method of claim 23, wherein the enhancer or promoter sequence comprises the nucleotide sequence ATTTGCAT or inverted sequence thereof.

26. A method for modulating the transcription of a gene comprising an enhancer or promoter sequence of a kappa light chain gene, comprising modulating the level of an IgNF-A, IgNF-E, Oct-2 or NF-kB transcriptional regulatory factor in the cell.

27. The method of claim 26, wherein the enhancer or promoter sequence comprises the nucleotide sequence GGGACTCCC or ATTTGCAT or inverted sequence thereof.

28. The method of claim 27, wherein modulating the level of a transcriptional regulatory factor comprises introducing into the cell a nucleic acid encoding the transcriptional regulatory factor operably linked to a strong promoter.

29. The method of claim 27, wherein modulating the level of a transcriptional regulatory factor comprises introducing into the cell a nucleic acid inhibiting the expression of the transcriptional regulatory factor.

30. A method for modulating transcription of a gene comprising an enhancer or promoter sequence of a kappa light chain gene in a cell, comprising contacting the cell with an agonist or antagonist identified by a method of any one of claims 17 to 25.

31. The method of claim 30, wherein the enhancer or promoter sequence comprises the nucleotide sequence GGGACTCCC or ATTTGCAT or inverted sequence thereof.

32. An antibody to an isolated transcriptional regulatory factor or polypeptide of any one of claims 1 to 8.