Congenital immunodeficiency with a regulatory defect in MHC class II gene expression lacks a specific HLA-DR promoter binding protein, RF-X

Cell, Vol. 53, 897-906, June 17, 1988, Copyright © 1988 by Cell Press

Congenital Immunodeficiency with a Regulatory Defect in MHC Class II Gene Expression Lacks a Specific HLA-DR Promoter Binding Protein, RF-X W. Reith,* S. Satola,* C. Herrero Sanchez,* I. Amaldi,* B. Lisowska-Grospierre,t C. Griscelli,t M. R. Hadam,~ and B. Mach* * Department of Microbiology University of Geneva Medical School Geneva, Switzerland tlNSERM U 132 H6pital des Enfants Malades Paris, France University of Hannover Medical School Hannover, Federal Republic of Germany

Summary The expression of MHC class II genes is tightly regulated. One form of congenital severe combined immunodeficiency (SClD) is characterized by a regulatory defect that precludes expression of HLA class II genes. B lymphocyte cell lines from such SCID patients provide a tool for identifying putative regulatory proteins that bind to class II gene promoters. We have identified three proteins binding to specific segments of the HLA-DRA promoter, two of which interact to form the predominant DNA-protein complex observed. One of these proteins, defined as an X box binding protein (RF-X), is specifically missing in cells from class II deficient SClD patients. We propose that the molecular defect in this congenital HLA class II regulatory deficiency is a lack of RF-X and that this factor plays an important role in the normal regulation of MHC class II gene expression.

Introduction The expression of class II antigens on antigen presenting cells mediates the activation of T lymphocytes, and therefore, the regulation of MHC class II gene expression plays a key role in the control of the immune response (Benacerraf, 1978, 1981). in addition, the aberrant expression of class II antigens is thought to be involved in the pathogenesis of autoimmune diseases (Bottazzo et al., 1986; Massa et al., 1987). Major histocompatibility complex (MHC) class II antigens are transmembrane glycoproteins consisting of and [3 chains (Korman et al., 1985). In man, class II antigens are encoded by a family of homologous genes clustered on chromosome 6 in the HLA-D region of the MHC (Trowsdale et al., 1985; Mach et al., 1986). The ~ and 13chain genes of the HLA-DP, -DQ, and -DR subregions encode the different types of class II products. The expression of class II genes is tightly regulated in terms of tissue distribution and developmental control. Their expression is largely limited to certain cells of the immune system, primarily B lymphocytes, activated T cells, macrophages, and dendritic cells (Hammerling, 1976; Flavell et al., 1986). With the exception of certain

specialized cell types, such as Kuppfer cells in the hver (Nadler et al., 1980), and Langerhans cells in the skin (Stingl et al., 1978), most tissues are class II negative. Class II gene expression can be induced in a number of class II negative cells by stimulation with interferon-7 (IFNy) (Basham and Merigan, 1983; Collins et al., 1984). In most instances, the entire family of 0~and 13chain genes, including the two functional DR[3 chain genes, is subjected to a coordinate regulation (Collins et al., 1984; Paulnock-King et al., 1985; Berdoz et al., 1987). The study of nuclear proteins that bind specifically to enhancer and promoter regions contributes to the understanding of gene regulation. In several systems, such proteins have been identified and purified (Briggs et al., 1986; Kadonaga and Tjian, 1986; Rosenfeld and Kelly, 1986; Jones et al., 1987; Bohmann et al., 1987a). A simple comparison of the DNAoprotein interactions occuring in cells that are negative or positive for the expression of a given gene is not necessarily informative as to the actual regulatory role of the DNA binding proteins. We have therefore made use of B lymphocyte cell lines characterized by a recessive mutation affecting a class II regulatory gene (de Preval et al., 1985). Such mutant cell lines have allowed us to establish a direct correlation between the activity of class II genes and the presence of a specific DNA binding protein. Class II deficient severe combined immunodeficiency (SCID) is a congenital defect characterized by an absence of class II gene expression in all tissues (LisowskaGrospierre et al., 1984). Moreover, class II genes cannot be induced by INF-7 in cells from SCID patients (de Pr~val et al., 1985), although other INF-7-inducible genes, such as class I genes and the invariant (In) chain gene (Strubin et al., 1984), respond normally. It has been shown that the SCID mutation affects a trans-acting class II regulatory gene that lies outside the MHC (de Preval et al., 1985). Several B lymphocyte cell lines have been established from such SCID patients (Niethammer et al., 1986; Lisowska-Grospierre and Griscelli, unpublished data), and these lines have maintained their class II negative phenotype after years in culture. In this study, we have used gel retardation assays, DNAase I footprinting, and methylation interference experiments to ~dentify and compare proteins binding to the HLA-DRA (previously called HLA-DR~) promoter and present in normal and in class II deficient SCID B lymphocyte cell lines. Our results show that a factor identified as an HLA class II box binding protein (RF-X) is present in normal B cells but absent or defective in cells from patients with a congenital regulatory defect in HLA class II gene expression.

Results Specific Binding of Nuclear Proteins from B Lymphocytes to the Promoter of the HLA-DRA Gene Several conserved DNA sequence motifs are found in the promoter region of the DRA gene (Figure 1). A TATA box,

Cell 898

-108 - 9 5 -74 -63 -52 -45 -28 - 2 4 CCTAGCAACAGATG CTGATTGGCCAA ATTTGCAT TATTA

A.

B,

1 ~

~

k

~

'

~

~

x

~

¥

BstNI

,

~ / ' ~ F

o

IT HmfI

BstNI -132

Y olil]o

I

i

3

4

5

6

7

.~B2--~ ' ~ "-B3-~_ .... """ B 4 "

-,--B5---~

Hlnf][ -32 - 85

-132

I SacI

SacI +31

BstN! -132

2

.... 1

x1 ohQo

I

- 34

i -80

Ftgure 1 Map of the DRA Promoter Regton Exon 1 ts shown by an open box, and the transcnpttonal start stte is mdlcated by an arrow The sohd black boxes represent potential regulatory elements; the conserved class II boxes (X and Y), the octamer mottf (O), and the TATA box (T). The sequence of these elements and thetr posibon m nucleotldes relattve to the cap site are indtcated above the map. The BstNI-Sacl and BstNI-Hmfl fragments used for gel retardation, footpnntmg, and methylation mterference experiments, and the X1 and Y ohgos used for competition expenments are mdlcated below the map.

thought to be important for accurate init=ation of transcnption, ts found at positions - 2 4 to - 2 8 upstream of the cap site. An octamer motif (ATTTGCAT) is located at positions - 4 5 to -52. This octamer sequence is present in a number of other promoters including those of the immunoglobulin heavy and light chain genes, SV40, human histone 2B, and human U2 small nuclear RNA genes (Parslow et al., 1984; Sive and Roeder, 1986, Staudt et al., 1986; Rosales et al., 1987; Bohmann et al., 1987b; Landolfi et al., 1987). It is however not found in other class II gene promoters. Upstream of the octarner motif there are two sequences called the X ( - 9 5 to -108) and Y ( - 6 3 to -74) boxes. The X and Y boxes are sequences strongly conserved among all murine and human class II genes (Mathts et al., 1983; Saito et al., 1983; Kelly and Trowsdale, 1985; Okada et al., 1985; Auffray and Strominger, 1986). They appear to be cis-acting elements involved in the regulation of class II gene expression (Boss and Strommger, 1986; Dorn et al., 1987a; Sherman et al., 1987), but their precise function is as yet unclear. The bmding of nuclear factors from normal B cells to the DRA promoter was studied by means of gel retardatton assays. Briefly, end-labeled DNA fragments were incubated with nuclear extracts prepared from the B lymphocyte cell line Mann under conditions that favor sequence-specific protein-DNA interactions. Protein-DNA complexes were then separated from unbound DNA by nondenaturing polyacrylamtde gel electrophoresis. With the BstNI-Sacl fragment of the DRA promoter (Ftgure 1), a complex binding pattern is observed (Figure 2A). Ftve distinct protemDNA complexes, B1 to B5, are observed reproducibly. The pattern observed with the BstNI-Hinfl fragment is essentially the same as the one obtained when the BstNI-Sacl fragment (Figure 1) is used. The same five complexes are

Ftgure 2. Detectton of Nuclear B Cell Factors That Brad to the DRA Promoter The end-labeled BstNI-Sacl fragment of the DRA promoter was mcubated wtth a nuclear extract from Mann B cells. Stable protem-DNA complexes were then separated from unbound DNA by gel electrophoresls (A) The gel retardatton profile obtained shows the presence of five protem-DNA complexes (B1 to B5). The strong bottom band (F) represents unbound DNA. (B) Specifictty of complexes B1 to B4 as shown by compettt[on expenments. Binding was done tn the absence of competitor DNA (lane 1), m the presence of a 20-, 100-, and 400-fold molar excess of unlabeled BstNI-Sacl fragment (lanes 2-4) or of an unlabeled fragment of pBR322 (lanes 5-7)

obtained with extracts from two other EBV transformed B lymphocyte lines, QBL and HHK (Figure 5B), and from the B lymphoblastoid cell line Rajt (data not shown). B1 to B4 are competed out when the binding reaction ts done in the presence of an excess of unlabeled DRA promoter fragment, while the presence of an equivalent excess of a fragment derived from pBR322 has no effect (Figure 2B). These four bands are therefore sequencespecific protein-DNA complexes. B5 is probably due to an aspecific protein-DNA interaction as it is not competed out efficiently by an excess of unlabeled BstNI-Sacl fragment. Moreover, the intensity of B5 varies s~gnificantly from one experiment to the next with the same or different extracts (Figure 5B). Band B1 is the most abundant retarded complex. Complex B3 frequently migrates close to complex B4 such that it is not always resolved as a distinct band (see Figures 2B, 3A, and 5B). Binding of certain proteins can introduce bending of the DNA such that migration of the complexes varies according to the position wtthin the DNA fragment at which the proteins are bound. This may explain fluctuations observed in the relative positions of B1, B3, and B4 when fragments of different sizes are used for binding (BstNI-Sacl versus BstNI-Hinfl). I d e n t i f i c a t i o n of Proteins B i n d i n g to t h e Y Box and Octamer Sequences

It has recently been shown that the Y box of the murine IE ~ chain gene and the octamer motif of the human DRA gene bind nuclear factors present in B ceils (Dorn et al., 1987a, 1987b; Sherman et al., 1987). We therefore carried

MHC Class II Gene Expression 899

A.

2

1 BI-~,~ B2--.~ .

3

1 234

4

B4...-~

N

Ra

pSP64

B1 B2--.~ B 3 - - " " ~-~ °~ B4,,~"

,

M

5 6

Actin DR~

-.,~.., ,~-

F=gure 4 Comparison of the Level of Transcription of the DRA Gene =n Normal and SCID B Cells

F---~

C.

D.

1 B1 B2"~ ~ B3~,. B4 - I

....... .==.m. ,mmw~

Transcription of the DRA and ~-Actm genes was monitored by nuclear run-off experiments. Nascent RNA chains were elongated =n vitro =n =solated nucle= from a normal B cell hne~ Mann (M), and in two different SCID cell lines, RamJa (Ra) and Nacera (N). Labeled nuclear RNA was hybridized to a dot blotted DRA cDNA clone, pSP64 DNA and actln cDNA were used as controls.

2

3

4

1

2

3

B3--~

F,gure 3. Identification of Proteins Binding to the DRA Promoter by Competition Experiments Gel retardat,on assays were performed as described in Figure 2. The positions of protein-DNA complexes (B1 to B4) and free DNA (F) are mdlcated Binding react,ons were done in the absence or presence of an X-fold molar excess of competitor DNA as indicated. (A) Competition with octamer motif: no competitor (lanes 1 and 3), 70fold of an SV40 promoter fragment (lane 2); 200-fold of a doublestranded oligo containing three copies of the octamer motif (lane 4). (B) Competition with Y oligo: no competitor DNA (lane 1); 2-fold (lane 2); 10-fold (lane 3); 40-fold (lane 4); 200-fold (lane 5); or 1000-fold (lane 6) of a double-stranded Y ohgo (see Figure 1). (C) Competition with X1 oligo: no competitor (lane 1); 700-fold of Y oligo (lane 2), 700-fold of Y oligo plus 200-fold of either X1 ohgo (lane 3) or a fragment of pBR322 (lane 4). (D) Competnt,on with DRB1 promoter: 700-fold of Y ohgo (lane 1); 700fold of Y ohgo plus 200-fold of either homologous DRA promoter fragment (lane 2), or a DRB1 promoter fragment (lane 3)

out competition experiments to determine which of the complexes identified above were due to the binding of nuclear factors to each of these two conserved sequences. When the BstNI-Sacl fragment of the DRA promoter is incubated with B cell nuclear extracts in the presence of an excess of a double-stranded oligonucleotide containing three copies of the octamer motif, complex B2 ,s competed out specifically (Figure 3A, lane 4). The same result is obtained when an SV40 promoter fragment containing the octamer motif is used as competitor DNA (Figure 3A, lane 2). We can thus conclude that complex B2 is due to a protein that binds to the octamer sequence of the DRA promoter.

A 52 bp double-stranded oligonucleotide covering both the Y box and the octamer sequence of the DRA promoter (Figure 1) was used to identify proteins binding to the Y box. When binding reactions are performed with either the BstNI-Sacl or the BstNI-Hinfl fragment in the presence of increasing amounts of this Y oligo, complexes B1, B2, and B4 are specifically competed out (Figure 3B). Since the Y oligo contains the octamer motif, the loss of B2 is consistent with the competition experiments performed with the octamer oligo and the SV40 promoter (Figure 3A). The loss of bands B1 and B4 indicates that their formation involves the binding of nuclear factors to the Y box. The only specific band remaining when the Y oligo is used as competitor is B3, which must therefore correspond to a protein binding 5' of the Y oligo sequence, between positions - 8 6 and -132. This localization was confirmed by a competition experiment in which the specific competitor DNA was a double-stranded oligonucleotide centered on the box (oligo X1; Figure 1). Oligo X1 specifically competes out the formation of B3, an observation that is particularly clear when the gel retardation pattern is simplified by the presence of an excess of oligo Y (Figure 3C). Class II Deficient SCID Cells Are Blocked at the Level of Transcription Three SCID B cell lines, Robert (from M. H.), Ramia and Nacera (from B. L.-G. and C. G.), have remained consistently negative for HLA class II antigens and mRNA for 2 years (Robert) and for 1 year (Ramia and Nacera) in this laboratory. Analysis of these cells by run-off transcription experiments shows clearly that the defect in class II expression is transcriptional (Figure 4). We therefore explored if the absence of transcription of class II genes in SCID B cells could be due to a deficiency m one of the nuclear factors binding to the HLA-DRA promoter described above. Binding of Nuclear Proteins from SCID B Cells to the DRA Promoter Figure 5A shows the result of a gel retardation assay done with the BstNI-Sacl fragment of the DRA promoter and nuclear extracts from the normal B cell line Mann (Figure

Cell 900

g.

A. B

C, H

S

B I ~ ' ~

BI~"",'.

B4,~,v ~*--~ '~*' :

B3---D,- B 4 ,,,~F

Q

M

N

Ra

Ro

H

Q

M

N

Ra

Ro

: . .,7;

-~

B3~

~"~

."

,k.~, L,.~.

"

Figure 5. Comparisonof Nuclear Factors Binding to the DRA Promoterm Extractsfrom Normal and SCID B Cells Gel retardationassayswith the mdtcatedB cell extractsweredone as descnbedin Ftgure 2. Stableprotem-DNAcomplexes(B1 to B4) are md~cated where detectable. (A) Comparisonof the gel retardationprofiles obtainedwtth the normal B cell hne Mann (lane B) and the SCID B cell line Nacera (lane S). (B) Comparisonof the profiles obtainedwith the three normal B cell lines HHK (H), QBL (Q), Mann (M), and the three SCID B cell hnes Nacera (N), Ramla (Ra), and Robert (Ro). (C) Comparisonof the same cell hnes as m (B) when a 700-fold molar excessof Y ohgo is added durmg the bmdmg reactton.

5A, lane B) or the SCID B cell line Nacera (Figure 5A, lane S). Of the four specific protein-DNA complexes, B1 to B4, obtained with the normal B cell, only B2 and B4 are observed at equivalent levels with the SCID B cell. Complexes B1 and B3 are absent, or nearly absent, in the SCID cell profile. The lack of the prominent band B1 is particularly striking. This difference in the protein binding pattern is not restricted to the SCID cell line Nacera, as is shown by the comparison of three normal B cell lines (HHK, QBL, and Mann) and three SCID cell lines (Nacera, Ramia, and Robert) in Figures 5B and 5C. When the formation of B1, B2, and B4 with normal extracts is competed out by the addition of an excess of Y oligo to the binding reactions, the complexity is greatly reduced and complex B3 can be identified very clearly as the only remaining specific band (Figures 3B and 5C). Under these conditions, it can be demonstrated that complex B3 is totally missing in all three SCID cell lines (Figure 5C). SCID extracts are thus deficient in a DRA promoter binding factor that is responsible for the formation of B3 as well as the larger complex BI.

SOlD B Cells Are Defective in a Protein That Binds to the Conserved Class II X Box To identify the promoter sequences that are involved in the formation of these two complexes and to further characterize the SCID defect that leads to the absence of complexes B1 and B3, we performed DNAase I protection or footprint experiments (Galas and Schmitz, 1978). The BstNI-Hinfl fragment of the DRA promoter (see Figure 1) was either 3' end-labeled (coding strand) or 5' end-labeled (noncoding strand) at the BstNI extremity. The labeled fragments were incubated with a nuclear extract from normal B cells (Mann) under conditions that favor the

formation of either B1 (i.e., in the absence of Y oligo) or B3 (i.e., in the presence of Y oligo). Free and complexed DNA in the incubation mix was then treated lightly with DNAase I, resolved by gel electrophoresis, eluted from the appropriate bands in the gel (B1 or B3), and analyzed by sequencing gel electrophoresis to reveal the profiles of DNAase I cuts (Figure 6A). Comparison between the profiles of free and bound DNA shows that in both B1 and B3, a region containing the conserved class II X box is protected against digestion by DNAase I. Protection is stronger and more extensive on the coding strand, and tt is clearer in B1 than in B3. Within the protected region, three enhanced DNAase I cutting sites are observed in the noncoding strand in complex B1. Interestingly, protection is not confined to the X box but extends upstream, in some cases up to position -122. In complex B1, there is an additional protected region covering the conserved class II Y box (Figure 6A), whereas protection of the Y box is not detected in complex B3. This is consistent with the observation that in gel retardation assays, B1 is competed out by the Y oligo while B3 is not (Figure 3B). Thus, while B3 is due to the binding of a factor to the X box alone, both the X box and Y box are bound to proteins in the B1 complex.

Analysis of DNA-Protein Contact Points by Methylation Interference Methylation interference assays were used to confirm the footprinting data and to identify contact points on the DNA more precisely. The BstNI-Hinfl fragment was labeled as above, partially methylated with dimethyl sulfate, and incubated with a normal B cell extract. After gel purification of free and complexed fragments, DNA was cleaved at methylated G residues and analyzed by sequencing gel

MHC Class II Gene Expression 901

A.

B, NC 'G

C

NC

" B

YI

C

r - ~ l ,

G'

G~

+ 8B BB ÷ AF13FF13 FA

+

"- ~mm IY YI

-,

X

. . . . . .

IY

-

×

C, t

B1 B3

1

B1

X

V

Y

uV

V

TGCAAGAACCCTTCCBCCTAGCAACAGATGJCGTCATCTCAAAATATTTTTBCTGATTGGCCAA 1 AGAG ACGTTCTTGGGAAGG G G A T C G T T G T C T A C B G C A G T A G A G T T T T A T ~ A J k A A ~ G A C T ~ A C C G G T T ~ TCTC u

I

B3

F~gure 6 DNAase I Footprmt and Methylation Interference Analysts of the Protem-DNA Complexes B1 and B3 The BstNI-Hmfl DRA promoter fragment was 5' end-labeled on the coding strand or 3' end-labeled on the noncodmg strand. Footprint (A) and methylation interference (B) analysts was then performed as descnbed m Expertmental Procedures. F, B1, and B3 represent cleavage products of unbound DNA (F) and complexed DNA (bands B1 and B3) G+A represents Maxam and Gnlbert G and A sequencing reactions (Maxam and Gtlbert, 1980). The positions of the X and Y boxes are indicated Enhanced DNAase I cutting sttes are mdtcated by dots. Arrow heads mdtcate G residues at which methylatmn spectfically interferes with protem bmdmg. The arrow indicates a G restdue at whnch methylatton enhances binding. The footprint and methylatlon interference data are summanzed schematically rn (C). The regions of protectton against cleavage by DNAase I in complexes B1 and B3 are indicated by brackets above (coding strand) and below (noncoding strand) the sequence. Arrow heads mdmate G residues at whtch methylatlon interferes with bmding, and the arrow indmates a G residue at which methylatnon enhances bmdmg. The A resndues at whmh methylatton interferes wtth bmding (data not shown) are indicated by open arrows.

I

I

B1

B1

electrophoresis (Figure 6B). Comparison of the cleavage profiles obtained with free and bound DNA reveals that the methylation interference profile within and just upstream of the conserved class II X box is identical in B1 and B3. Methylation of six G residues, two in the coding strand and four in the noncoding strand, interfere with binding. Weaker interference by methylation of G residues further upstream of the X box has also been observed (see Figure 6B). Methylation of one G residue in the noncoding strand appears to enhance binding. Methylation of one G residue in the conserved class II Y box interferes with binding in complex B1 but not in complex B3 (Ftgure 6B). This is consistent with the DNAase I protection experiments showing that B1 contains a protein binding to the Y box while B3 does not (Figure 6A). The results of DNAase I protection and methylation interference experiments are summarized in Figure 6C. We have also included data obtained by a methylation interference experiment in which DNA cleavage was enhanced at A residues (data not shown; Maxam and Gilbert, 1980). From these results and from the data obtained by gel retardation assays, two conclusions can be drawn. First, the protein-DNA complex B3 is due to the binding of a factor, RF-X, to the conserved class II box. Second, complex B1 is due to the joint binding of two factors, RF-X and RF-Y, to the class II X and Y boxes. The same RF-X protein is probably involved in binding to the box in both B1 and B3, since the contact points within and close to the X box are the same in B1 and B3. Binding of RF-X is not limited to

the DRA promotor since it can be competed out by other HLA class II promoters. Figure 3D shows that an excess of a DRB1 (previously called DRI31) promoter fragment can abolish the formation of B3 as efficiently as the homologous DRA promoter fragment. Neither B1 nor B3 are detected in gel retardation assays done with nuclear extracts from SCID B cells. This implies that SCID cells are deficient in the HLA class II X box binding protein (RF-X), which contributes to the formation of both B1 and B3. The absence of RF-X binding may therefore explain why none of the class II genes are transcribed in SCID cells. RF-X may thus be a regulatory factor that is essential for the constitutive and coordinate transcription of class II genes.

Discussion Class II gene promoters show a remarkable degree of sequence homology. In particular, two short DNA segments, called the class II X and Y boxes, are found in the promoters of all human and murine class II genes (Mathis et al., 1983; Saito et al., 1983; Kelly and Trowdale, 1985; Okada et al.. 1985; Auffray and Strominger, 1986). Their conservation and location suggests that the X and Y boxes are cis-acting control elements involved in the coordinate regulation of class II gene expression. Moreover, the functional importance of the X and Y sequences in both constitutive and inducible expression has been demonstrated by transfection experiments (Boss and

Cell 902

B1

-132

-108

B3

B4

B2

L

1

l

-95

-74

® -63

ON 1

-52 -45 HmfI

1

I SacI

BstNI

Bst NI

I

I Hinf I Yoligo

I -85 Xl

I -132 I -124

t SacI +31

oligo

X 2 olJgo

] -34

I -80 I -70

Ftgure 7. Model for Potenttal RegulatoryProteins That Brad to the DRA Promoterin B Cells In nuclearextractsfrom normal B cell lines, we have tdenttftedthree DNA bindmgproteins:an octamerbmdmgprotein(OBP),a Y boxbmdmg protein(RF-Y),and an X box bindingprotein(RF-X) In gel retardation expenments,the protein-DNAcomplexesB2, B3, and B4 are due to OBP,RF-X,and RF-Y,respecttvely.B1 is a complexmwhichboth RFX and RF-Yare bound. Fragmentsof the DRA promoterwith which complexesB1 to B4 can form are indicated.The Y ohgowtll compete with thesefragmentsfor bmdmgof RF-Yand OBPsuchthat complexes B1, B2, and B4 are competedout, leavingonly complexB30ligo X1 competesfor bindingof RF-Xwhtle ohgoX2 does not The open box mdicatesa promotersequencethat, in additionto the X box, appears essentralfor the bindingof RF-X. In SCID B cells, RF-Xts mtssmgor macttvesuch that neithercomplexBt nor complexB3 can be formed.

Strommger, 1986; Sherman et al., 1987) and in transgenic mice (Dorn et al., 1987a). We have therefore studied the interaction of putative regulatory factors with these conserved class II promoter sequences. In order to identify factors that bind to class II promoters and that are essential for the constitutive expression of class II gene transcription in B lymphocytes, we have compared the binding of nuclear proteins to the DRA promoter in normal B cell lines and in class II negative SCID B cell lines, characterized by a defect in a regulatory gene controlling class II gene transcription. This comparison has allowed us to identify an X box binding protein (RF-X) that is missing or mutated in SCID B cells.

DRA Promoter Binding Proteins Present in B Lymphocytes In gel retardation assays performed with extracts from normal B cells, four specific protein-DNA complexes (B1 to B4) are formed with fragments of the DRA promoter containing the first 132 bp upstream of the cap site. Results derived from competition experiments, DNAase I footprinting, and methylation interference assays indicate that these four complexes are due to three different proteins, an X box binding protein (RF-X), a Y box binding protein (RF-Y), and an octamer binding protein (OBP). The slowest migrating complex, B1, involves the joint binding of both RF-X and RF-Y, while complexes B2, B3, and B4 are due to OBP, RF-X, and RF-Y, respectively (Figure 7).

From the disappearance of band B2 in competition experiments involving the Ig octamer motif, it can be concluded that B2 is due to the binding of a protein (OBP) to the octamer sequence (ATTTGCAT) found at positions - 5 2 to - 4 5 in the DRA promoter. Binding of a B cell nuclear protein to the DRA octamer sequence has also been observed by others (Sherman et al., 1987), and it has been proposed that it might be involved in B cell-specific expression of the DRA gene. However, no similar sequence is found at equivalent positions in other human or murine class II genes. Complexes B1, B2, and B4 are not formed in the presence of an excess of a double-stranded oligonucleotide covering both the Y box and the octamer sequence of the DRA promoter (Figures 1 and 7). Since only B2 is competed out by the octamer, the formation of B1 and B4 involves a protein (RF-Y) that binds to or near the Y box. This has been confirmed by footprinting and methylation interference experiments for both complex B1 and B4. Methylation of three A residues and one G residue interferes with the binding of RF-Y. These four residues fall within and just upstream of a reversed CCAAT box sequence included in the Y box. Interestingly, methylation of the same four residues in the Y box of the murine class II E~ chain gene inhibits the binding of a CCAAT box binding protein, NF-Y, identified in extracts from murine B lymphoma cells (Dorn et al., 1987a, 1987b). Consequently, the Y box binding protein identified here (RF-Y) is a CCAAT box binding protein and may represent the human equivalent of the murine NF-Y factor. Upon competition with the Y oligo, the complexity of the binding pattern ts greatly simplified and only complex B3 remains, whether the BstNI-Sacl or the BstNI-Hinfl fragment is used for the binding reaction. This maps the sequence responsible for the formation of B3 to a region situated upstream of the Y box, between positions - 8 6 and -132. Furthermore, B3 can be competed out by oligonucleotide Xl, which covers this region of the promoter. B3 is also competed out by the promoter of another class II gene, coregulated with DRA. The results of footprinting experiments on complex B3 show protection against DNAase I at and upstream of the box. Moreover, methylation of six G residues and two A residues within and just 5' of the box inhibits the formation of B3. It can be concluded that B3 is due to a protein, RF-X, that binds to the class II X box. The binding of RF-X can be competed out efficiently by an oligonucleotide covering nucleotides -132 to - 8 0 of the DRA promoter (oligo X1; Figure 7). Interestingly, an oligonucleotide of similar size covering nucleotides -124 to -71 (oligo X2; Figure 7) is very inefficient at competing out complex B3 (data not shown). This indicates that the nucleotides -132 to -124, although located about 20 nucleotides upstream of the X box, are essential for stable binding of RF-X. This is consistant with DNAase I footprinting and methylation interference data that show that contact of RF-X with the promoter extends upstream of the X box, including two G residues at positions -126 and -127. These two observations suggest that the binding of RF-X involves, in addition to the box itself, the -132 to -124

MHC Class II Gene Expression 903

promoter region. It is of interest that this sequence falls within a region exhibiting striking homology between different class II promoters (Miwa et al., 1987).

Complex B1 May Involve an Interaction between RF-X and RF-Y From competition, footprinting, and methylahon experiments, we have concluded that complex B1 ts due to the joint binding of RF-Y and RF-X. A detailed analysis of the footprintmg data suggests that RF-X and RF-Y might interact with one another ~n the formation of complex B I DNAase I protection of the box is stronger and more extenswe tn B1 (DNA complex with both RF-X and RF-Y) than in B3 (RF-X alone), particularly on the noncoding strand. Moreover, three hypersensitwe DNAase I sites are detected on the noncoding strand in B1 but not in B3. We therefore propose that the interaction of RF-X with RF-Y in complex B1 results in a change in the conformation of RFX and perhaps m an enhanced affinity of RF-X for the DNA. Another observation relevant to the possible interaction between RF-X and RF-Y is that complex B1 is the most abundant complex observed, and that it increases relattve to B3 and B4 with longer binding reactions. This suggests the posstbility of a cooperative interaction between factors RF-X and RF-Y. A more detaded kinettc analysis and a study of the relattve stability of B1, B3, and B4 at different salt concentrations should help clarify this question. SOlD Patients with Congenital Immunodeficiency Are Deficient in a Regulatory Protein, RF-X Gel retardation assays have shown that with B cell extracts from three different SCID patients, complexes B1 and B3 are not formed. Cells from the class II regulatory mutants are thus deficient in RF-X, the protein that contributes to the formation of both B1 and B3. The absence of RF-X btnding in SCID B cells is correlated with a congenital regulatory defect in HLA class II gene transcription in these patients. We therefore propose that RF-X is a regulatory factor that is indispensible for constitutive class II gene transcription in B lymphoeytes. This ts, to our knowledge, the first example of a congenital disease involving a defined molecular defect in a regulatory DNA-binding protein. One should also envisage the implications of a possible structural and functional polymorphism of factor RF-X, with certain alleles expressing abnormal levels of HLA class II antigens. This could be relevant to the pathogenesis of certain autoimmune d=seases associated wtth a genetically controlled aberrant expression of class II antfgens (Bottazzo et al., 1986; Massa et al., 1987). At least three explanations can account for the absence of RF-X binding in SCID cells. First, RF-X could be completely absent, possibly as the result of a gene deletion. The fact that in certain SCID patients one observes a faint trace of HLA-DR mRNA (de Preval et al., 1988) suggests that this explanation is unlikely. Second, RF-X could be structurally altered, as the result of mutations in the DNAbinding domain of the regulatory protein, such that it precludes normal binding to the X box Finally, the genetic defect might concern another factor, essential to convert RF-X from an inactive nonbindmg form, to an active form

capable of specific binding to the class II X box. We have recently shown that activation of both class II and In genes by INF-7 in class II negative cells is mediated by the de novo synthesis of an Inducer protein at an early stage in the induction process (Reith et al., in press). Stimulation of class II negative SCID fibroblasts with IFN-7 does not lead to class II gene activation (de Preval et al., 1985). RF-X therefore seems to be essential for induced as well as constitutive expresswon of class II genes. MHC class II genes and the Invariant chain gene are normally coregulated. Surprisingly, although the In promoter contains a sequence homologous to the class II X box (O'Sullivan et al., 1986), both constitutive and induced expression of the In gene are unaffected in SCID cells (de Pr~val et al., 1985). RF-X does not therefore seem to be essential for In gene expression. The dissociation between induced class II and In gene expression m IFN-y-treated SCID cells indicates that this Inducer protein must be distinct from RF-X. The Inducer protein may act as an upstream s~gnal by activating transcription factors required for expression, including RF-X tn the case of class II genes. Alternatively, it could cooperate with these factors in promoting transcription. Parallel experiments have shown that two strong DNAase I hypersensitive sites, present at posptions - 5 0 and -180 of the DRA promotor, are absent in SCID cells (G6nczy et al., unpublished data), suggesting that these two sites reflect, at the level of chromatin structure, the presence of RF-X. Interestingly, these two sites are also present in normal fibroblasts, whether class II negative or activated by INF-~, to express class II genes. Thus, in normal individuals, RF-X might be present but inactive on class II promoters in class II negative cells, presumably being activated following INF-7 induction. In vitro-generated B cell variants, characterized by an absence of class II mRNA, have been derived from established B cell lines of man (Gladstone and Pious, 1978; Accolla, 1983; Long et al., 1984; Levine et al., 1985) and mouse (Polla et al., 1986). Fusion experiments have shown that these variants also revolve trans-acting regulatory defects (Gladstone and Pious, 1980; Accolla et al., 1985; Polla et al., 1986). We have recently shown that, in one of these mutants (Accolla, 1983), the regulatory defect is clearly distinct from the one described here in the case of SCIDs, since this in vitro-generated regulatory mutant exhibits normal DNA-binding protein and DNAase I sensitivity patterns in the HLA class II promoter (Satola et al., unpublished data). If the regulatory defect of the in vitro mutant concerns the same RF-X protein, it must affect a site distinct from the DNA-bindmg domain of RF-X. In the case of SCIDs, the genetic defect might well be pleiomorphic and therefore affect other genes that might also be under the control of RF-X. In this regard, the analysis of a cDNA subtraction library has allowed us to isolate interesting clones corresponding to mRNA not expressed in SCID cells and yet dtstmct from HLA class I1. Finally, it should be pointed out that the different SCID patients studied here originate from different regions and ethnic groups and nevertheless exhibit the same defect in the binding of a specific protein to class II promoters.

Cell 904

Experimental Procedures Plasmids, Fragments, and Oligonucleotides DRA promoter fragments used for binding and competition experiments were derived from two subclones, pDRa2 and pDRa3, of the DRA promoter, pDR~2 was constructed by inserting the 1181 bp BgllI-Sacl promoter fragment of the DRA gene (Das et al., 1983; Schamboeck et al., 1983) into the BamHI-Sacl sites of pUC12, pDR~3 was constructed by inserting the 171 bp BstNI-Sacl promoter fragment of the pDRa2 subclone into the SmaI-Sacl sites of pUC13 after flushing the BstNI extremity with Klenow polymeraee The 400 bp HmdllI-Pstl fragment containing the SV40 promoter used for competition experiments was derived from p33-143 (Strubin et al., 1986). The DRB promoter fragment used for competit0on was a 330 bp ScaI-Sacl fragment from a DRB1 gene (Rollim et al., 1985). The region of pBR322 used for competition was the Sall(651)-Haelll(830) fragment. The DRA cDNA clone used for the nuclear run-off experiment was the 447 bp PstI-Hinfl fragment of a DR~ cDNA clone (Wake et al., 1982) subcloned mto the PetI-Smal sites of pSP64. As control DNAs, actin cDNA (Gunning et al., 1983) and pSP64 were used. To prepare double-stranded oligonucleot=des, complementary strands were synthesized on a Gene Assembler (Pharmacia) using the phosphoramldlte method as suggested by the suppher; the two strands were combined, bo~led for 10 mm m 10 mM Tris (pH 7,5), 1 mM EDTA, 5 mM MgCI2, and allowed to anneal by cooling to room temperature over a period of 3-4 hr. An 80 bp double-stranded oligonucleot~de containing three copies of the octamer motif (ATTTGCAT) was a generous gift from the laboratory of W Schaffner (Institute for Molecular Biology II, University of Zurich, Switzerland).

Cell Lines and Nuclear Extracts The EBV transformed human B lymphocyte cell lines Mann (kindly provided by J. Bodmer), HHK, and QBL (kindly provided by M. Giphart) and the EBV transformed SCID B lymphocyte cell lines from SCID patients Robert (contributed by M. H.), Nacera and Ramia (contributed by B. L-G. and C, G.) were grown in RPM11640 (Seromed, Basel, Switzerland or Gibco, Bethesda, MD) in the presence of 2 mM glutamine, 10% fetal calf serum, 10 U/ml penicillin, and 10 pg/ml streptomycin. Nuclear extracts were prepared according to Dignam et al. (1983), except that 1 M NaCI was used to extract nuclear proteins. The protein concentration was in the range of 1-2 mg/ml, and extracts were stored at -70°C.

Run-Off Transcription Assays Isolabon of nuclei and in vitro elongatton reactions were carried out according to Schibler et al. (1983) except that cells were lysed for 10 min at 0°C in buffer A containing 0.5% Nonidet-P40 and that nuclei were only pelleted once through a 10 ml cushion of 30% sucrose in buffer A for 10 min at 3000 rpm at 4°C in a Sorvall HB4 rotor. Prehybndization and hybridization of in vitro elongated RNA with plasmid DNA spotted onto nitrocellulose filters were done at 42°C in 50% formam~de, 50 mM NaPe4 (pH 7.0), 0.75 M NaCI, 0.5% SDS, 2 mM EDTA, 10x Denhardt (lx Denhardt = 0.02% bovine serum albumin, 0.02% Ficoll, 0.02% polyvmylpyrrolidone), and 500 pg/ml of sonicated denatured salmon sperm DNA. Filters were washed down to a fmal stringent at 65°C in 0.1x SSC, 50 mM NaPe4, (pH 7.0), 0.5% SDS, and then treated for 10 min at 37°C with 2 p.g/ml of RNAase A m 2x SSC, 50 mM Tris~HCI (pH 70), and 0.1% SDS.

Gel Retardation Assays For gel retardation assays (Fried and Crothers, 1981; Strauss and Varehavsky, 1984; Carthew et al., 1985), binding was performed in 20 ~1 of 12% glycerol, 12 mM HEPES (pH 7.9), 60 mM KCI, 5 mM MgCI2, 0.12 mM EDTA, 0.3 mM PMSF, 0.3 mM DTT using 10,000-20,000 cpm of 32p-labeled DNA (0.1-0.5 ng), 1-2 I~g of poly(dl-dC).poly(dl-dC), and 2-4 Ilg of nuclear extract. In some experiments, reactions were optimized by the addttion of 500 ng sonicated denatured E. coh DNA, and the MgCI2 concentration was lowered to 3 mM to improve RF-X bindmg (experiments presented in Figures 3C, 3D, and 5C). 32p-labeled fragments were added to the reaction mixtures after 5 mm of prebmdmg at 0°C of the nonspecific competitor and the nuclear extract. Binding was allowed to proceed for 30 rain at 20°C. Samples

were then analyzed by electrophoresls for 3 hr at 4°C and at 200 V (10 V/cm) on 4% polyacrylamlde (acrylamide:bisacrylamide, 30"1) or 3.5% polyacrylamide (Ftgure 3B), 0.25x TBE gels (lx TBE = 89 mM Tris, 89 mM boric acid, 2 mM EDTA). Electrophoresis buffer was recirculated. Prior to loading the samples, gels were pre-electrophoresed for 1 hr under the same conditions. After electrophoresis, gels were fixed in 10% TCA, 10% acetic acid, 30% methanol, dried, and exposed.

DNAase I Footprinting Binding reactions were scaled up by a factor of eight and set up as described above. For footprinting on band 3, the binding was camed out in the presence of 800 ng of unlabeled Y oligo. After 30 man of incubation at 20°C, DNAase I (2 p.I) was added to a final concentration of 1 p.g/ml. The samples were incubated for 1 rain at 20°C and the digestion was stopped by the addition of EDTA to a final concentration of 10 mM. DNAase I was stored in 2 mM HCI at a concentration of 5 mg/ml and activated by dilution in 10 mM HEPES (pH 7.6), 25 mM CaCI2, and 100 ng/ml of BSA The samples were immediately loaded onto 4% polyacrylamtde gels and electrophoresed as described above. After electrophoresie, the wet gels were autorad=ographed to Iocahze the positions of the retarded and free fragments. The appropriate bands were cut out of the gels, and DNA was eluted overnight at 37°C as described by Landolfi et al. (1987). After precipitation, the DNA was resuspended in 90% formamide, 10 mM EDTA, heated to 90°C for 3 rain, and analyzed on 8% polyacrylamide-7 M urea sequencing gels (Maxam and Gilbert, 1980). After electrophoresis, the gels were fixed in 20% ethanol, 10% acetic acid, dried, and exposed.

Methylation Interference Assays Methylation interference assays were performed essentially as described by Staudt et al, (1986), except that end-labeled DNA fragments were partially methylated for only 2 rain. The binding reactions with methylated DNA, gel electrophoresis, elution of the DNA, and sequencing gel analysis following plpendine cleavage (Maxam and Gilbert, 1930) was done as for DNAase I footprmting experiments

Acknowledgments We would hke to thank C Berte and M Zufferey for skillful help with the cell cultures and U. Schibler, M Strubin, and A. Israel for helpful advise Th~swork was supported by the Swiss National Fund for Scientific Research. The costs of publication of this article were defrayed m part by the payment of page charges This article must therefore be hereby marked "advertisement" in accordance wtth 18 U S.C. Section 1734 solely to indicate thts fact. Received December 7, 1987; revised March 9, 1988.

References Accolla, R. S. (1983). Human 8 cell variants immunoselected against a single la antigen subset have lost expression of several la antigen subsets J Exp. Med. 157, 1053-1058. Accolla, R S., Carra, G , and Guardiola, J (1985) Reactivation by a trans-acting factor of human major histocompatibility complex la gene expression in interepecies hybrids between an la-negatlve human B-cell variant and an la-positive mouse B-cell lymphoma Prec. Natl. Acad. Sci. USA 82, 5145-5149. Auffray, C., and Strominger, J. (1986). Molecular genetics of the human major histocompatibllity complex Adv. Human Genet. 15, 197-247. Basham, T Y., and Merigan, £. C. (1983) Recombinant interferon-~' increases HLA-DR synthesis and expression. J. Immunol. 130, 14921495. Benacerraf, B. (1978). A hypothes~s to relate the specificity of T lymphocytes and the activity of I region-specific Ir genes in macrophages and B lymphocytee. J. Immunol, 120, 1809-1812. Benacerraf, B. (1981). Role of MHC gene products in immune regulation. Science 212, 1229-1238 Berdoz, J, Gorskl, J., Terrnijtelen, A.-M., Dayer, J. M , IriS, C., Schendel, D., and Mach B. (1987) Constitutive and induced expression ofthe

MHC Class II Gene Expression 905

individual HLA-DR 13and c{ chain loci in different cell types. J. Immunol. 139, 1339-1341.

R (1987) A cellular DNA-binding protein that activates eukaryottc transcription and DNA replication Cell 48, 79-89

Bohmann, D., Bos, T., Admon, A., Nishimura, T, Vogt, R, and Titan, R (1987a) Human proto-oncogene c-jun encodes a dna binding protern with structural and functional properties of transcrtption factor AP1 Science 238, 1386-1392

Kadonaga, J. T., and Tjian, R. (1986) Affinity purification of sequencespecific DNA binding proteins. Proc Natl. Acad. Scl USA 83, 58895893.

Bohmann, D, Keller, W., Dale, T., SchSler, H. R, Tebb, G , and Mattaj, I. W. (1987b). A transcriptional factor which binds to the enhancers of SV40, immunoglobulin heavy chain and U2 snRNA genes. Nature 325, 268-272. Boss, J. M., and Strominger, J. L. (1986). Regulation of a transfected human class II major hlstocompatibility complex gene in human fibroblasts. Proc. Natl Acad Sc~ USA 83, 9139-9143. Bottazzo, G. F, Todd, I, Mirakian, R., Belfiore, A , and PujoI-Borrell, R (1986) Organ-specific autotmmumty a 1986 overwew Immunol. Rev. 94, 137-169. Briggs, M R, Kadonaga, J. T., Bell, S. P, and Titan, R. (1986) Purification and biochemical characterization of the promoter-specific transcription factor, Spl. Science 234, 47-52. Carthew, R. W, Chodosh, L. A , and Sharp, P. A. (1985) An RNA polymerase II transcription factor binds to an upstream element in the adenovirus major late promoter Cell 43, 439-448 Collins, T., Korman, A. J., Wake, C. T., Boss, J. M., Kappes, D. J., Flers, W, Ault, K A., Gimbrone, M. A., Jr., Strommger, J. L., and Pober, J. S. (1984). Immune interferon activates multiple class II major histocompatibility complex genes and the associated invariant chain gene in human endothelial cells and dermal fibroblasts Proc. Natl Acad. Sci. USA 81, 4917-4921. Das, H K., Laurence, S. K., and Weissan, S M (1983). Structure and nucleotide sequence of the heavy chain gene of HLA-DR. Proc Natl. Acad Sci USA 80, 3543-3547. de Pr~val, C., Hadam, M , and Mach, B. (1988). Regulation of genes for HLA class II antigens m cell lines from patients w~th severe combined immunodeficiency. New Engl. J Med, in press de Preval, C, Lisowska-Grospierre, B, Loche, M, Gnscelh, C, and Mach, B. (1985). A trans-acting class II regulatory gene unhnked to the MHC controls expression of HLA class II genes. Nature 318, 291-293. D~gnam, J D., Lebovitz, R. M, and Roeder, R. G (1983) Accurate transcription initiation by RNA polymerase II in a soluble extract from ~solated mammalian nuclei Nucl Acids Res. 11, 1475-1489

Kelly, A., and Trowsdale, J (1985). Complete nucleotide sequence of a functional HLA-DPI3 gene and the region between the DP~I and DPal genes: comparison of the 5' ends of HLA class II genes Nucl Acids Res. 13, 1607-1619. Korman, A J, Boss, J M, Spies, T, Sorrentino, R., Okada, K, and Strominger, J. L. (1985). Genetic complextty and expression of human class II hlstocompatibility antigens Immunol Rev 85, 45-86 Landolfi, N F, Capra, J D, and Tucker, P W. (1987) Protemnucleotide contacts in the immunoglobulin heavy-chain promoter region. Proc Natl Acad. Scl USA 84, 3851-3855 Levme, F, Erhch, H. A., Mach, B., and Pious, D (1985). Transcriptional regulation of HLA class II and invariant chain genes. J. Immunol. 134, 637-640 Lisowska-Grospierre, B, Charron, D. J., de Preval, C., Durandy, A , Turmel, P., Gnscelh, C., and Mach, B. (1984) Defect of expression of MHC genes responsible for an abnormal HLA class I phenotype and the class II negative phenotype of lymphocytes from patients with combined Jmmunodeficiency. In Histocompatlbility Testing, (Berlin: Sprmger-Verlag), pp 650-655. Long, E. O., Mach, B., and Accolla, R. S. (1984). la-negative B-cell variants reveal a coordinate regulation in the transcription of the HLA class II gene family. Immunogenetics 19, 349-353. Mach, B, Gorski, J., Rollini, P., Berte, C., Amaldl, I., Berdoz, J, and Ucla, C. (1986). Polymorphism and regulation of HLA class II genes of the major histocompat~bihty complex Cold Spring Harbor Symp. Quant. Biol. 51, 67-74. Massa, R T., ter Meulen, V., and Fontana, A. (1987) Hyperinducibil~ty of la anttgens on astrocytes correlates with strain-specific susceptibility to experimental autoimmune encephalomyehtis Proc. Natl Acad Sci. USA 84, 4219-4223. Mathis, D.J., Benoist, C. O., Wdhams, V. E, II, Kanter, M R., and McDewtt, H. O (1983) The murine E(z immune response gene. Cell 32, 745-754. Maxam, A., and Gilbert, W. (1980). Sequencing end-labeled DNA with base-specific chemical cleavage. Meth. Enzymol. 65, 499-560.

Dorn, A., Durand, B., Marring, C, Le Meur, M., Benoist, C, and Mathis, D. (1987a) The conserved MHC class II boxes--X and Y--are transcriptional control elements and specifically brad nuclear proteins. Proc. Natl. Acad. Sci. USA 84, 6249-6253.

Miwa, K., Doyle, C., and Strommger, J L. (1987). Sequence-specific interactions of nuclear factors with conserved sequences of human class II major histocompatibthty complex genes Proc Natl Acad ScJ USA 84, 4939-4943.

Dorn, A , Bollekens, J, Staub, A., Benoist, C, and Mathls, D. (1987b). A multiplicity of CCAAT box-binding proteins. Cell 50, 863-872.

Nadler, P. I., Khngenstein, R. J., Richman, L K, and Ahmann, G B. (1980). The murine Kuppfer cell J. Immunol. 125, 2521-2525.

Flavell, R. A , Allen, H., Burkly, L C., Sherman, D. H., Waneck, G L, and Widera, G. (1986). Molecular biology of the H-2 histocompat~bihty complex. Science 233, 437-443.

Niethammer et al (1986) Congenital agammaglobulinemia assoctated with malabsorbtion no expression of MHC-class II antigens due to a regulatory gene defect? In Recent Advances in Primary Immunodefic~encies. Sereno Symposia 28, F Aiuti, F. Rosen, and M. D. Cooper, eds. (New York: Raven Press), pp 185-193

Fried, M., and Crothers, D. M. (1981). Equilibria and kinetics of lac repressor-operator interactions by polyacrylamide gel electrophoresls. Nucl. Acids Res 9, 6505-6525. Galas, D. J, and Schmitz, A (1978) DNAase footprinting, a simple method for the detection of protem-DNA binding specihclty. Nucl Acids Res 9, 3047-3060. Gladstone, P., and Pious, D. (1978) Stable vanants affecting B cell alIoantigens in human lymphord cells. Nature 27/, 459-461. Gladstone, R, and Pious, D. (1980) Identification of a trans-acting function regulating HLA-DR expression in a DR-negative B cell variant Som. Cell Genet. 6, 285-298. Gunning, P., Ponte, P., Okayama, H., Engel, J., Blau, H., and Kody, L. (1983). Isolation and characterization of full-length cDNA clones for human c~-, ~, and 7-actm mRNAs: skeletal but not cytoplasmic actins have an amino-terminal cysteme that is subsequently removed Mol. Cell Biol. 3, 787-795. Hammerhng, G. J. (1976). T~ssue distribution of la antigens and their expression on lymphocyte subpopulations Transplant. Rev. 30, 64-82. Jones, K A., Kadonaga, J. T., Rosenfeld, R J., Kelly, T J., and Tjlan,

Okada, K., Boss, J. M., Prentice, H, Spies, T, Mengler, R., Auffrey, C, Lillies, J., Grossberger, D, and Strominger, J L. (1985) Gene orgamzation of DC and DX subregions of the human major histocompatibihty complex. Proc. Natl. Acad. ScL USA 82, 3410-3414 O'Sullivan, D. M., Larhammar, D., Wilson, M.C., and Peterson, P. (t986) Structure of the human la-associated invanant (~,)-cham gene: identtficatton of 5' sequences shared with major histocompatiblhty complex class II genes. Proc. Natl. Acad. Scl. USA 83, 4484-4488 Parslow, T. G , Blair, D L., Murphy, W J, and Granner, D K. (1984). Structure of the 5' ends of immunoglobulin genes: a novel conserved sequence. Proc. Natl. Acad Sci USA 81, 2650-2654 Paulnock-King, D., Sizer, K., Freund, Y., Jones, P., and Parnes, J. (1985). Coordinate induction of la ~, ~, and I, mRNA in a macrophage cell line. J. Immunol 135, 632-636. Polla, B S., Pol]ak, A., Geier, S. G , Nathenson, S. G , Ohara, J., Paul, W E, and Glimcher, L H. (1986) Three distinct signals can induce class II gene expression in a murine pre-B-cell line. Proc. Natl. Acad. Sc~. USA 83, 4878-4882

Cell 906

Retth, W, Satola, S., Amaldi, I., Herrero, C, Berte, C, Ulevwtch, R., and Mach, B (1988), Regulation of HLA-class II genes: identificatton of a regulatory promoter-binding protein m~ssmg m class II defictent congemtal immunodeficiency. In Immunobiology of HLA, B Dupont, ed (New York Sprmger-Verlag), in press Rolhni, P., Mach, B., and Gorsk~, J (1985). Linkage of three HLA-DR I}-cham genes evLdence for a recent duphcatton event Proc. Natl. Acad. Sci. USA 82, 7197-7201 Rosales, R, Wgneron, M, Maccht, M, Dawdson, I, Xlao, J H., and Chambon, P (1987). In wtro binding of cell-specific and ubiquitous proteins to the octamer motif of the SV40 enhancer and related motifs present in other promoters and enhancers. EMBO J 6, 3015-3025. Rosenfeld, R J, and Kelly, T. J. (1986) Purification of nuclear factor I by DNA recogmt~on site affimty chromatography. J. B~ol Chem 261, 1398-10015. Satto, H, Maki, R A., Clayton, L K., and Tonegawa, S (1953) Complete primary structures of the E!3 chain and gene of the mouse major histocompattblllty complex Proc. Natl Acad Sct. USA 80, 5520-5524. Schamboeck, A., Korman, A. J, Kamb, A, and Strommger, J L (1983) Orgamzatton of the transcnpttonal umt of a human class II hstocompatlbihty anttgen HLA-DR heavy chain Nucl Acids Res. 11, 8663-8673 Schibler, U, Hagenbuchle, O., Wellauer, R K., and Pittet, A C (1983) Two promoters of different strengths control the transcription of the mouse alpha-amylase gene amy 1a in the parotd gland and the hver Cell 33, 501- 508. Sherman, R A., Basta, P V, and Ting, J. P.-Y (1987) Upstream DNA sequences required for tissue-specific expression of the HLA-DRa gone. Proc Natl Acad. Sc~ USA 84, 4254-4258. Sive, H, and Roeder, R. G. (1986) Interactton of a common factor with conserved promoter and enhancer sequences in hlstone H2B, immunoglobufin, and U2 small nuclear RNA (snRNA) genes. Prec. Natl. Acad Sci. USA 83, 6382-6386 Staudt, L M, Singh, H., Sen, R., Wtrth, T, Sharp, R A , and Balttmore, D (1986) A lymphoid-specfftc protein binding to the octamer motif of tmmunoglobulin genes Nature 323, 640-643. Stlngl, G S, Katz, I., Clement, L, Green, I., and Shevach, E M (1978) Immunologtc functions of la-beanng epidermal langerhans cells J Immunol 121, 2005-2013. Strauss, R, and Varshavsky, A. (1984) A protein binds to a satelhte DNA repeat at three specific sites that would be brought into mutual proxtmity by DNA folding tn the nucleosome Cell 37, 889-901. Strubm, M., Mach, B., and Long, E O. (1984). The complete sequence of the m RNA for the HLA-DR-assoctated invanant chain reveals a polypeptide with an unusual transmembrane polarity EMBO J 3, 869-872 Strubm, M., Long, E. O, and Mach, B. (1986). Two forms of the la anttgen-associated mvanant chain results from alternatwe initiattons at two In-phase AUGs. Cell 47, 619-625 Trowsdale, J., Young, J. A. T., Kelly, A. P, Austin, P. J., Carson, S, Meunier, H, So, A , Erhch, H A, SpJelman, R. S, Bodmer, J., and Bodmer, W. E (1985). Structure, sequence, and polymorph~sm in the HLA-D region. Immunol. Rev. 85, 5-43. Wake, C. T., Long, E. O., Strubm, M., Gross, N., Accolla, R. S., Carrel, S., and Mach, B. (1982) Isolation of cDNA clones encoding HLA-DR alpha chains Proc. Natl. Acad SCL USA 79, 6979-6983 Note Added in Proof

With more concentrated extract, we have recently been able to obtain binding of RF-X to the X2 ohgo suggesting that DNA upstream of nucleotJde 124, although enhancing binding, is not absolutely necessary