D N A BINDING SPECIFICITIES 1 Introduction

In the related br-hlh and br-hlh-lz families of transcription factors, the basic regions of protein dim ers recognise sequences containing CANNTG (the E-box), see Littlewood and Evan, 1995^^^ for review. In view of the large size of these families, it is likely that there m ay be several factors present in a single cell that potentially recognise the same targ et sequence; yet these proteins m u st exercise a high degree of specificity in recognising their ow n targets, given the diversity (such as grow th or differentiation) and fundam ental im portance of their effects. The identification of a transcription factor's binding site is crucial in un d erstan d in g its function, not least as it can indicate potential target genes.

Previous w ork using purified GST-Myc(br-hlh-lz) proteins^^^ w hich bind to DNA as hom odim ers, had identified a 12 base pair DNA binding site containing the sequence CACGTG. In vitro translated Myc br-hlh-lz peptides also recognise a palindrom ic CACGTG motif^'^^. A similar peptide corresponding to the br-hlh-lz region of v-Myc (61 amino acids long) also binds CACGTG^^. Another approach in w hich the basic region of Myc was substituted for the same region of E12^^® generated a fusion p ro tein that also b o u n d CACGTG sequences, and this w as inhibited by CpG méthylation.

Myc dim ers are how ever undetectable in vivoi^^^ and are only form ed at extreme concentrations w ith in vitro translated^'^^ or bacterially produced^'^'^^^ Myc. Studies of Myc trans-activation in both yeasf^^ and m am m alian cells^^^'^^^ (see also chapter 3), as w ell as biochem ical analysis of the Myc proteW ^^, had dem onstrated th at Myc acts as a complex w ith Max. A lthough it was likely that the sites identified by H alazonetis and Kandil^^^ w ere actually b o u n d by a h etero d im er

b etw een the Myc p e p tid e and e n d o g en o u s Max p re se n t in the reticulocyte lysate^"^^, no DNA binding site had been determ ined for the M yc/M ax heterodim er using full-length proteins, and I decided to identify such a site.

There are several possible m echanism s to achieve specificity of DNA binding: differential interactions w ith other DN A-binding proteins on a prom oter m ay restrict the set of sites at w hich a given factor m ay act; specific E-boxes (such as CAC G TG b u t not CAGÇTG) m ay be reco g n ised by d ifferen t factors or co m b in atio n of factors in a

heterodim er^^^d32,338-344. q j. preferences for sequences outside the E-box

m ay be factor specific. Blackwell and Weintraub^"^ dem onstrated that br- h lh -co n tain in g p ro tein s recognise 3 bases on eith er sid e of the hexanucleotide core, and it therefore seem ed im portant to determ ine such extended D N A -binding sequences for the full-length M yc/M ax protein complex to provide the best inform ation as to potential target genes.

The Max dim er had been show n to also recognise the CACGTG sequence^^^'^^^ and to affect trans-activation m ediated th ro u g h such sites^^^'^^^'^^^'^^^ (see also chapter 3). Since Max RNA and protein are very stable com pared w ith Myc and there is an excess of Max over Myc

in it seemed likely that M ax/D N A complexes w ould form in cells. H ow ever, a Max binding site had never been determ ined from scratch, rather Max had only been show n to be able to bind to the Myc b in d in g site. The p o ssib ility rem a in e d th a t M ax w o u ld b in d preferentially to other sequences which m ay be im portant functionally, and I therefore decided to analyse Max binding as well.

To determ ine com plete consensus sequences requires either a binding-site selection from an initially random pool^^^ or comprehensive testing of all possible binding site variants. I decided on the form er

approach to define the M yc/M ax and M ax/M ax DNA binding sites. Further, since Myc is a phospho-protein, I concluded th at in vitro

translation w as superior to purification from bacteria as the protein produced w ould be modified to an extent (for example phosphorylation by casein kinase II (CKII)^^®'^^^, W ork w ith in vitro translated, full-length Myc an d Max in the laboratory (subsequently published^'^^) h a d established a gel retardation and im m uno-precipitation assay for DNA binding. This confirm ed that M yc/M ax can bind to CACGTG and I decided to use this system to analyse the binding specificities of the full- length Myc and Max proteins.

4.2 Selection of binding sites

In an attem pt to investigate the DNA binding preferences of the M yc/M ax and M ax/M ax complexes, I perform ed repeated ro u n d s of selection and am plification from an initially random oligonucleotide pool (R76), as previously described^^^. Accordingly, in vitro translated full-length Myc and Max or Max alone were allowed to bind to a random oligonucleotide pool consisting of 26 random bases flanked by PCR prim ers (see Fig. 4.1). This w as then co-im m uno-precipitated w ith the appropriate antibody (anti-Myc for the M yc/M ax complex and anti-Max for the M ax/M ax complex). Any bound DNA is therefore precipitated w ith the protein complex bound. I then amplified the selected DNA by PCR and perform ed further rounds of selection in the same m anner by using the selected DNA in place of R76. The Max2 form of Max was used since it binds DN A m ore efficiently than M axi, both as a hom odim er and as a heterodim er w ith Myc^'^^