agrdtwuzdflnrfifia 3 try. (1,: 2 r. ..l»; (I 3‘ III! In»... 1‘13]... .I. VI)-.. .le , .356 ‘v r... J a. ? 6... {”1} b .. .3 4 , (Io 3%. 1.319.151}?! 3f.“- ell '53? .nvvwvywv—Tpunwflwuyu ‘ v" , Idol 14:.IIKI. ll‘nx‘vl'iii . 1 ullvdcl‘lll. o. ‘SIIO...‘|!IAIOAA . A1£‘§ It» :11 III .lvllg‘alltllll \ 'lxll.‘ 10VI|I . _ . . v 9., . -3, Iaz. .lYaI.-.... [1.9-193 f! fury n.” ‘ . I‘l»-.l, ; ...!pl ....... :..‘ ......A- .l.:r;f7 H‘l‘ ‘ 9.17.1. . _ .. ‘ p . Aunt-92.6 . L ‘ ‘II‘OL. ,‘rht .5 l‘lildyln"t .. I .‘v. ' "Gu‘n ltllilllllllllllllllHHIIIIIIIIIHIIHI'Jllllllllllllllilli 300885 0178 This is to certify that the dissertation entitled Site-Specific and Random Mutagenesis Studies of the VP16 Transcriptional Activation Domain presented by Jeffrey Lee Regier has been accepted towards fulfillment of the requirements for Ph . D . degree in Genetics Date July 12, 1993 MS U is an Affirmative Action/Equal Opportunity Institution . 0-12771 LIBRARY Michigan State University A] PLACE iN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore data due. ‘ DATE DUE DATE DUE DATE DUE + r \\”\i MSU is An Affirmative ActionlEqual Opportunity Institution 9"”! SITE-SPECIFIC AND RANDOM MUTAGENESIS STUDIES OF THE VP16 TRAN SCRIPTIONAL ACTIVATION DOMAIN By Jeffrey Lee Regier A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Genetics Program 1993 ABSTRACT SITE-SPECIFIC AND RANDOM MUTAGENESIS STUDIES OF THE VP16 TRAN SCRIPTIONAL ACTIVATION DOMAIN By Jeffrey Lee Regier Transcriptional activator proteins are an integral part of the transcriptional regulatory apparatus. However, little is known about the structural features of transcriptional activation domains that are important for their function. We have examined structural features of the activation domain of VP16, a transcriptional activator protein from herpes simplex virus type 1, by oligonucleotide-directed and random mutagenesis methods. Extensive mutagenesis at a phenylalanine residue at position 442 of a VP16 activation domain truncated after amino acid 456 demonstrated the importance of an aromatic amino acid at that position. Based on an alignment of the VP16 sequence surrounding Phe442 and the sequences of other transcriptional activation domains, leucine residues at positions 439 and 444 of VP16 were subjected to mutagenesis. Results from these experiments suggest that bulky hydrophobic residues flanking Phe442 also contribute to the function of the truncated VP16 activation domain. Restoration of aa 457-490 to various truncated Phe442 mutants restored partial activity. Although a pattern of amino acids surrounding Phe473 resembles that surrounding Phe442, mutations of Phe473 (or Phe475) did not dramatically affect activity. We infer that the two regions of VP16 (aa 413-456 and 457-490) possess unique structural features, although neither appears consistent with two prior models (the amphipathic alpha-helix or "acid blob" models) purporting to describe the structure of acidic activation domains such as that of VP16. These results, considered with previous Mm activation and inhibition studies, suggest that the two subdomains of VP16 affect transcription by different mechanisms. To identify mutations that decrease the activity of the distal subdomain (aa 457-490), I employed random mutagenesis combined with a biological selection in yeast. This approach was based on the observation that the GAL4-VP16 fusion protein, when overexpressed in yeast, is toxic to the growth of the host cell; mutations decreasing the activity of the VP16 activation domain relieve this toxicity. Four mutations in the distal subdomain (TA458/DN461/MV478, TA480, N485, and YC465) were identified that decreased the activity of that domain. I praise You because I am fearfully and wonderfully made; Your works are wonderful, I know that full well. Psalm 139:14 iv ACKNOWLEDGMENTS I wish to thank Steve Triezenberg for his scientific excellence, his patience, and for being a mentor in the truest sense of the word. I also wish to thank my wife, Marty Regier, for her understanding and tolerance as a fellow student, and for her encouragement during those all-so-frequent times of frustration. The past and present members of the Triezenberg lab, namely Steve Triezenberg, Doug and Andrea Cress, Rath Pichyangura (S. 6.), Lisa Ortquist, Fan Shen, Lee Alexander, John Stebbins, Jaya Reddy, and Peter ("Chuck") Horn, made graduate school, if not fun, then at least...interesting. I also thank them for their forbearance in putting up with my G.O.W.s. I thank Susan Roehl for her valuable assistance in mutagenesis and helping to spread those yeast plates. I acknowledge the contributions of the members of my guidance committee, Drs. Zach Burton, Michele Pluck, Tom Friedman, and Lee Kroos. I also acknowledge the important contributions of my collaborators, Dr. Shelley Berger (The Wistar Institute) and Dr. Leonard Guarente (MIT); an especially big "thank you" goes to Shelley for helping a rookie learn how to use yeast. West Side / Droletts, the Thai Kitchen, Mancino's, and Lee kept us energized for important Thursday noon discussions. Finally, I thank my parents and parents-in-law for support and encouragement over the years. TABLE OF CONTENTS PAGE LIST OF TABLES ....................................................................................................... viii LIST OF FIGURES ........................................................................................................ ix LIST OF ABBREVIATIONS ....................................................................................... xi CHAPTER I: INTRODUCTION ............................................ , ...................................... 1 Eukaryotic Transcription ................................................................................... 1 Transcriptional Activator Proteins ............................................................... 11 Mechanisms of Transcriptional Activator Function ................................ 12 VP16 ..................................................................................................................... 13 Models for Transcriptional Activation by VP16 ........................................ 14 Overview ............................................................................................................ 18 CHAPTER II: SITE-DIRECTED MUTAGENESIS OF THE VP16 TRANSCRIPTIONAL ACTIVATION DOMAIN ...................................... 22 Introduction ....................................................................................................... 22 Chapter II Methods ........................................................................................... 23 Mutagenesis of Phe442 ..................................................................................... 29 Hydrophobic Residues Flanking Phe442 Also Contribute to Activity ................................................................... 32 Addition of aa 457-490 Restores Partial Activity to Phe442 Mutants ........................................................................................ 35 vi TABLE OF CONTENTS (cont'd) PAGE Proximal and Distal Subdomains of VP16 Have Unique Structural Features ................................................................... 35 Conclusions ........................................................................................................ 40 CHAPTER III: RANDOM MUTAGENESIS OF THE VP16 TRANSCRIPTIONAL ACTIVATION DOMAIN ...................................... 45 Introduction ....................................................................................................... 45 Biological Selection in Yeast ........................................................................... 46 Chemical Mutagenesis ..................................................................................... 48 Chapter 1]] Methods .......................................................................................... 53 Results ................................................................................................................. 58 Conclusions ........................................................................................................ 67 CHAPTER IV: MECHANISM OF ACTION OF THE VP16 ACTIVATION DOMAIN ................................................................................ 70 Introduction ....................................................................................................... 70 Mechanism of Action of the VP16 Activation Domain ........................... 70 Future Studies ................................................................................................... 77 LIST OF REFERENCES ................................................................................................ 80 vii LIST OF TABLES PAGE CHAPTER II Table 1. Mutagenic oligonucleotides ...................................................................... 24 Table 2. Relative activities of truncated (del456) VP16 mutants bearing amino acid substitutions at position 442 ................................ 31 CHAPTER 111 Table 3. Description of plasmids used in the yeast biological selection and fi-galactosidase assay .......................................................................... 49 viii LIST OF FIGURES PAGE CHAPTER I Figure 1. Assembly of basal transcription factors ................................................... 6 Figure 2. Nucleotide and deduced amino acid sequences of the VP16 activation domain (codons 410-490) ............................................ 15 Figure 3. Alignment of hydrophobic residues from various transcriptional activation domains with that of VP16 (revised and extended from Cress and Triezenberg, 1991a) .............. 19 CHAPTER 11 Figure 4. Autoradiogram of primer extension assay reflecting the activities of truncated VP16 proteins altered at Phe442 .............. 30 Figure 5. Determination of mutant protein stability ........................................... 33 Figure 6. Effects of amino acid substitutions at Leu439 or Leu444 of truncated VP16 ....................................................................................... 34 Figure 7. Relative activities of full-length and truncated VP16 mutants bearing substitutions at Phe442 ................................... 36 Figure 8. Schematic representation of the VP16 activation domain (amino acids 413-490) ................................................................. 38 Figure 9. Effects of amino acid substitutions at Phe473 and Phe475 of VP16, tested in the context of the FA442 mutation ....................... 39 CHAPTER 111 Figure 10. Strategy for chemical mutagenesis and subsequent biological selection in yeast .................................................................... 47 ix LIST OF FIGURES (cont'd) PAGE CHAPTER III (cont'd) Figure 11. Summary of mutations resulting from treatment of the gene encoding the VP16 activation domain with nitrous acid ............................................................................................... 62 Figure 12. Mutations in the VP16 activation domain of GALA-VP16 relieve in yivg toxicity .................................................... 64 Figure 13. Relative activities of GAL4-VP16delSm proteins bearing mutations in the distal subdomain of the VP16 activation domain ...................................................................... 66 CHAPTER IV Figure 14. Proposed model of the mechanism of activation by the VP16 activation domain ........................................................................ 76 HSV-l MSV NTP ONPG PEG PMSF RNA pol 11 SDS LIST OF ABBREVIATIONS absorbance at 420 nm absorbance at 595 nm absorbance at 600 nm amino acid amphipathic alpha helix adenosine triphosphate base pair carboxyl terminal domain of the largest subunit of RNA pol II N-[2-hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid] herpes simplex virus type 1 infected cell protein immediate early kilodaltons long terminal repeat murine sarcoma virus nucleotide triphosphate Q-nitrophenyl-B-D-galactoside polyethylene glycol phenylmethanesulfonyl fluoride RNA polymerase II sodium dodecyl sulfate xi LIST OF ABBREVIATIONS (cont'd) SDS-PAGE sodium dodecyl sulfate-polyacrylamide gel electrophoresis TAF TBP associated factor TBP TATA box-binding protein T'FII transcription factor of RNA pol II tk thymidine kinase UAS upstream activating sequence VP16 virion protein number 16 Single letter abbreviations for the amino acids: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. xii CHAPTER I INTRODUCTION Eukaryotic Transcription The ability of a cell to regulate the expression of its genes is important, not only for its own viability but also for its appropriate function within the context of its environment. Genes contain the information for all cellular processes. In eukaryotes, the flow of genetic information from DNA to protein begins with the transcription of a gene into pre-messenger RNA, which after processing is exported from the nucleus to the cytoplasm where it is translated into protein. This multistep process of gene expression contains numerous points of regulation. The state of the DNA template can affect the degree of expression of a particular gene. The methylation of cytosine residues can control gene expression, either positively or negatively, depending on the location of the methylated cytosines relative to the gene. In general, in higher eukaryotes genes which are constitutively expressed are undermethylated. An example of the repressive effects of methylated DNA is the murine or 1(1) collagen gene, in which methylation of a specific region downstream of the transcription startsite represses transcriptional activity (Rhodes and Breindl, 1992). A direct correlation between undermethylation and gene expression is not universal; for instance, there appears to be no correlation between the degree of methylation and the level of expression of the chicken lysozyme gene (wolfl gLaL, 1991). 2 DNA usually exists in complex with histone proteins, forming structures called nucleosomes. Nucleosomes can have a repressive effect on gene transcription by physically excluding the binding of proteins necessary for transcription or blocking the movement of the enzymes involved in transcription. Nucleosome displacement likely involves multiple steps, including nucleosome destabilization followed by transfer of histone proteins onto competitor molecules (Adams and Workman, 1993). Another DNA state which has a role in regulating gene expression is the degree of supercoiling, which can increase the stored free energy of a region of DNA, thereby making it easier to unwind the DNA duplex. The initiation of transcription is also a point of gene regulation. As will be described in greater detail below, a discrete set of proteins must assemble at a promoter for transcription to begin. In principle, factors that promote the binding or activity of one or more of these transcription proteins can enhance the rate of transcription initiation. Besides the assembly of transcription factors at the promoter, the initiation of transcription requires hydrolysis of ATP and the separation of the DNA strands; any factor that facilitates either of those events can also increase transcription initiation. Once transcription has been initiated, the elongation of the nascent RNA transcript affords another opportunity for regulation. A good example of this is found in the human immunodeficiency viruses HIV-1 and HIV-2. Transcription occurs from the viral LTR promoter, but in the absence of the virus-encoded protein Tat little full-length viral RNA is produced. When Tat is present it not only stimulates transcription but, probably in combination with cellular factors, also facilitates efficient elongation complexes (Cullen, 1990). A second example is that of transcription factor 118 (THE), which binds to RNA polymerase II and facilitates its movement past H --—-_,_..... Inn-.31.... 3 intrinsic transcription pausing sites 111111119 (Reinberg and Roeder, 1987b; Rappaport $31., 1988; Reines «11:31., 1989). The diversity of proteins in a eukaryotic cell is determined to a large extent by alternative splicing of precursor messenger RNA, which allows synthesis of multiple proteins from a single gene. An excellent example of this is the a-tropomyosin gene. Tropomyosin proteins are essential components of the contractile apparatus in several different tissues, and each contractile cell type contains a specific tropomyosin that differs from that in other cell types. The diversity of tropomyosin proteins does not arise from multiple genes, but rather from differential splicing of the primary transcript from a single (Jr-tropomyosin gene; at least seven different tropomyosin proteins are known (Breitbart 21:31., 1987). The multiple poly(A) sites (designated L1-L5) of the adenovirus major late transcription unit (MLTU), a complex transcription unit encoding five proteins, exhibit different rates of processing depending upon the stage of infection of the virus. In an early stage of adenovirus infection, the 5' proximal L1 site is used almost exclusively even though transcription extends past the L2 and L3 sites. In later stages of infection, all five poly(A) sites are utilized efficiently (Prescott and Falck-Pedersen, 1992). After processing in the nucleus, the mature mRNA is transported through the nuclear pore complex to the cytoplasm for translation. In the case of HIV-1 mRNAs, however, incompletely spliced forms of certain transcripts are transported to the cytoplasm, despite cellular mechanisms designed to prevent such events. The essential viral proteins Gag and Env are encoded by incompletely spliced HIV-1 mRNAs. The transport of these incompletely spliced transcripts is due to the action of the virus-encoded Rev protein, which binds specifically to a sequence present in the incompletely . 4 spliced viral transcripts and allows their transport into the cytoplasm (Malim £131., 1989). The prevalent model for the initiation of translation in eukaryotes is that the 408 ribosomal subunit binds to the 5' cap structure and migrates 5' to 3' along the mRNA until it encounters the first AUG codon; the complete ribosome initiates translation at that codon if it is in the proper sequence context (Kozak, 1991). However, in some instances alternative translation start site selection occurs. The cAMP-responsive-element modulator encodes both repressors and an actiVator of CAMP-responsive transcription. One of the repressor proteins results from the use of an internal AUG codon within the mRNA that also encodes the activator protein (Delmas 21:31., 1992). Because of the inherent complexity of gene expression, those interested in the regulation of gene expression have many options for study. The research focus of our laboratory, as well as that of many others, is the regulation of gene expression at the level of transcriptional initiation of genes transcribed by RNA polymerase II. Eukaryotic genes that encode proteins are transcribed by the multi- subunit RNA polymerase H enzyme (RNA pol 11), whereas the genes encoding ribosomal RN As and those encoding transfer RN As (and the SS ribosomal RNA) are transcribed by RNA polymerases I and III, respectively (Young, 1991). RNA pol II is comprised of 10 :l: 2 subunits, depending upon the particular eukaryote (Young, 1991). The two largest subunits of RNA pol II are related to the B and [3' subunits of prokaryote RNA polymerase, and are responsible primarily for catalysis of RNA chain elongation (Young, 1991). Additionally, three subunits of 14-28 kDa in RNA pol II are shared with the other two eukaryotic RNA polymerases (Young, 1991). The genes for all eleven subunits of mm RNA pol II have been cloned and their 5 proteins expressed and purified; two of these subunits, numbers four and seven, are not essential for mRNA synthesis 1111119, but together form a subcomplex that appears to influence the efficiency of transcription initiation (Young, 1991). An especially interesting feature of the largest subunit of RNA pol H is a carboxyl terminal domain (CTD) consisting of multiple repeats of the consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser (Young, 1991). This heptapeptide sequence is repeated 52 times in mammals, 44 times in Dmsgphila, and 27 times in yeast (Young, 1991). The CTD can be phosphorylated Mm , and it is the major site of phosphorylation in RNA pol 11 (Young, 1991). RNA pol II binds to the promoter in the unphosphorylated state (denoted form IIa); however, phosphorylation of the CTD (generating form 110) is required for the enzyme to clear the promoter and begin transcription (Cadena and Dahmus, 1987; Payne 2131., 1989; Lu :1; 31., 1991). Potential CTD kinases are discussed later in this chapter. In addition to functioning as a "trigger" mediating the transition of RNA Pol II from DNA binding to elongation (Peterson and Tjian, 1992), the CTD might also function to remove DNA-binding proteins during elongation, to facilitate post-transcriptional RNA processing, or to localize the transcriptional machinery to the nucleus (Corden, 1992). RNA pol 11 cannot accurately initiate transcription by itself, but requires other associated proteins, termed transcription factors (TF5). The steps involved in assembly of transcription factors at a promoter are illustrated in Figure 1. This assembly of RNA pol H and its associated factors into a transcription complex at a promoter is a highly ordered process (Buratowski 3131., 1989). The initial step is the binding of the TATA-binding protein (TBP) to the TATA box sequence, located approximately 30 basepairs upstream from TFIID (A) ={m= +1 Pol lla. TFIIF TFIIE. H. J Figure 1. Assembly of basal transcription factors. The letters for the various transcription factors refer to the order of elution from phosphocellulose chromatography columns. 7 the transcription startsite (Davison £131., 1983; Fire £131., 1984; Reinberg £131., 1987). The gene for the TBP has been cloned from a number of different organisms, including human (Hoffmann £L31., 1990; Kao £131., 1990; Peterson £131., 1990), yeast (Cavallini £131., 1989; Eisenmann £131., 1989; Hahn £131., 1989; Schmidt £131., 1989), 1219531211113 (Hoey £131., 1990), and Arabidgpsis (Gasch £131., 1990). Sequence comparisons between the different TBPs revealed that the carboxyl-terminal half of TBP is highly conserved among species; for example, the carboxyl-terminal 180 amino acids of human and [21125121211113 TBPs are 88% identical (Zawel and Reinberg, 1992). In contrast, the amino-terminal region of TBP is greatly different between species, both in length and in amino acid sequence. In higher eukaryotes, TBP does not bind to the TATA sequence as a monomer but rather as a large multisubunit protein complex denoted TFIID (Reinberg £131., 1987). Endogenous 12195121211113 TBP, for example, exists as a complex with at least six tightly associated proteins, or TAFs (TBP associated factors) (Dynlacht £131., 1991). In addition to the TAFs, other factors can associate with TBP to inhibit transcription by RNA pol II (Meisterernst and Roeder, 1991; Meisterernst £131., 1991;1nostroza £131., 1992). Transcription factor HA has been the subject of much controversy and conflicting data. Depending upon the cell type (or laboratory) from which it was isolated, a chromatographic fraction containing TFIIA activity was or was not required for basal transcription, and the activity could be comprised of one, two, or three polypeptides (Zawel and Reinberg, 1992). Recent reports indicate that Maine TFIIA is composed of two subunits (Ranish £131., 1992) and that TFIIA interacts with TFIID (Buratowski and Zhou, 1992; Lee £1 31., 1992). A report from the Reinberg laboratory (Cortes £131., 1992) may finally 8 have shed light on the role of TFHA. Using yeast TFHD affinity chromatography, they purified TFHA from HeLa cells and showed that it consists of three subunits, two of which correspond to those from TFHA of S, w. TFHA stimulates basal transcription when a native HeLa TFHD is used, but not when transcription is reconstituted using bacterially produced TBP (Cortes £131., 1992). Thus they hypothesize that TFHA counteracts the effects of a negative component which is normally associated with TFHD in vim. Transcription factor HB, like TFIID, is absolutely required for basal transcription (Reinberg and Roeder, 1987a). TFIIB consists of a single polypeptide of approximately 30 kDa (Reinberg and Roeder, 1987a), the gene for which has been cloned from human, yeast, and [2132591211113 (Ha £131., 1991 ; Pinto £131., 1992; Wampler and Kadonaga, 1992; Yamashita £131., 1992). The TFHB protein sequence shows a motif similar to one in prokaryotic sigma factors (Ha £131., 1991). This protein binds to the TFIID-DNA complex (Maldonado £131., 1990), but can also bind to RNA pol H. TFHB is capable of suppressing nonspecific transcription by RNA pol H, probably by a direct interaction between TFHB and RNA pol H (Wampler and Kadonaga, 1992). Although required for transcription, excess TFIIB can inhibit transcription from some promoters 111311113, perhaps by sequestering an unidentified basal transcription factor required by the repressed promoters (W ampler and Kadonaga, 1992). It has also been suggested that a possible function of transcriptional activator proteins (discussed in detail below) may be to induce a conformational change in TFHB that allows it to interact more stably with RNA pol H (Colgan £131., 1993). Transcription factor IIF (also known as RAP30 / 74) consists of two polypeptides of 30 kDa and 74 kDa (Flores £131., 1988). This factor binds tightly 9 to RNA pol H in solution (Reinberg and Roeder, 1987a), and it recruits RNA pol H to the TFIID-B-DNA complex. The small subunit of TFHF is sufficient to recruit RNA pol H to the preinitiation complex (Flores £131., 1991) and it prevents RNA pol H from binding nonspecifically to DNA (Killeen and Greenblatt, 1992). The large subunit of TFHF possesses amino acid sequence similarity to the E4911 transcription factor sigma 70 (McCracken and Greenblatt, 1991); also, this subunit is not required for transcription initiation, but it is required for elongation of mRNA transcripts (Chang et al., 1993). Transcription factor HE consists of two subunits of 57 kDa and 34 kDa in humans (Inostroza £131., 1991). TFIIE enters the preinitiation complex after the binding of RNA pol II and TFHF (Inostroza et al., 1991). The large subunit contains a zinc finger motif and a domain similar to the catalytic loop of protein kinase C (Ohkuma £131., 1991; Peterson £131., 1991). The small subunit contains a putative nucleotide binding site (Peterson £131., 1991). Thus TFHB is a potential CT D kinase. However, homogeneous preparations of TFIIE exhibited no ATPase activity, prompting the notion that other transcription factors are required for TFIIE function (Inostroza £131., 1991). It has also been reported that TFHE increases the processivity of another potential CTD kinase, TFHH (Lu £131., 1992). Transcription factor HH is a multisubunit factor that appears to have two potential roles in transcription. Lu £131. (1992) reported that TFHH possesses a kinase activity which specifically phosphorylates the CTD of RNA pol H in 11119. A second function proposed for TFHH is that of a helicase, which separates the strands of a DNA duplex, an operation required for transcription as well as DNA replication. It has been reported that the 89 kDa subunit of human TFHH possesses an ATP-dependent helicase activity (Schaeffer £131., 1993). This subunit was found to be identical to the product of the BBQ 1 0 gene, a gene encoding a protein involved in DNA excision repair which is mutated in patients suffering from xeroderma pigmentosa and Cockayne's syndrome (Buratowski, 1993; Schaeffer £131., 1993); this result is intriguing, since both DNA repair and transcription require DNA strand separation. Each transcription factor mentioned thus far frequently requires the presence of the others for its function. It is also apparent that not all of the aforementioned factors are required for transcription at every pol H promoter; for example, TFIIB was not required for transcription from the immunoglobulin heavy chain (IgH) gene promoter mum (Parvin £131., 1992). Futhermore, the topological state of the DNA template can affect the transcription factor requirements. The IgH gene promoter can support transcription with only TBP, TFIIB, and RNA pol II if it is in a negatively supercoiled state (Parvin and Sharp, 1993). This requirement for a minimal number of transcription factors is presumably due to the free energy stored in a negatively supercoiled template, energy which can function to partially unwind the DNA and facilitate the transition from a "closed" to an "open" transcription complex (Parvin and Sharp, 1993). Further work is necessary before the generality of this phenomenon is known. The transcription complex described above has been termed the basal transcription complex. This complex forms at the promoters of most genes transcribed by RNA pol H, and facilitates basal transcription. However, many promoters contain upstream sequences which allow the adjacent genes to be expressed at levels higher than basal transcription. Conversely, regulatory sequences known as silencers function to inhibit expression of their cognate genes. Different combinations of the upstream regulatory sequences, both transcriptional activators and silencers, at different promoters allow finely tuned regulation of transcription initiation (Renkawitz, 1990). Sequence 1 1 elements that positively regulate adjacent promoters are termed enhancers (or UASs in yeast) and are bound by a class of proteins known collectively as transcriptional activators (Ptashne, 1988; Mitchell and Tjian, 1989). Activated transcription by a particular activator protein is specific for those promoters containing the binding site (enhancer element) for that activator protein. Transcriptional Activator Proteins Transcriptional activator proteins have been identified in a variety of eukaryotic species, including 1213531211113, yeast, mouse, and human. Many transcriptional activator proteins have two domains: one conferring specific association with promoter sequences, usually a DNA binding domain, and a second domain for regulatory function (Brent and Ptashne, 1985; Ptashne, 1988; Mitchell and Tjian, 1989). The DNA specificity domains of transcriptional activators generally fall into different motifs, defined by primary sequence or three-dimensional structure. Examples of these motifs include zinc fingers, homeodomains, helix-turn-helix, helix-loop-helix, and leucine zippers (Mitchell and Tjian, 1989). Activator proteins have been grouped into several classes, based on the amino acid content of their activation domains. The first activator proteins identified possessed activation domains containing a high percentage of negatively charged residues, hence the name acidic activators (Mitchell and Tjian, 1989). Some examples of acidic activator proteins are GAL4 (Ma and Ptashne, 1987a), GCN4 (Hope and Struhl, 1986), and HAP4 (Forsburg and Guarente, 1989) from yeast, and VP16 from herpes simplex virus type 1 (HSV- 1). The GAL4 protein potently stimulates the transcription of the galactose- inducible genes in yeast (Bram and Kornberg, 1985; Giniger £131., 1985; Johnston, 1987), while the GCN4 protein is involved in the coordinated 1 2 induction of 30-50 genes for amino acid biosynthesis (Hope and Struhl, 1986). Other transcriptional activators identified contain activation domains with high proline or glutamine content, such as CTF and Spl, respectively (Courey and Tjian, 1988; Mermod £131., 1989). It must be emphasized that this classification system is based solely on primary amino acid sequence; a better system would group activator proteins based upon the mechanism of increasing transcriptional initiation. Mechanisms of Transcriptional Activator Function Several models have been proposed to explain how activators work. They may function by directly or indirectly contacting a component of the basal transcription complex, thereby speeding up the ordered assembly of that complex at the promoter (Ptashne, 1988; Buratowski £131., 1989; Lewin, 1990; Ptashne and Gann, 1990; Wang £131., 1992). Potential targets of activator proteins include many of the basal transcription factors, such as TFHD, TFHB, TFHH, and RNA pol H (Stringer £131., 1990; Ingles £131., 1991 ; Lin and Green, 1991; Lin £131., 1991). Activators could also increase the rate of transcriptional initiation by relieving the inhibition caused by histones, thus allowing the transcription complex to form (Croston £131., 1991 ; Laybourn and Kadonaga, 1991 ; Workman £131., 1991 ; Felsenfeld, 1992; Kornberg and Lorch, 1992). With regard to histones, transcriptional activator proteins can increase transcription in two ways: by removing histones from the promoter (antirepression) and/ or by facilitating assembly of a transcription complex (true activation) (Laybourn and Kadonaga, 1992). Activators can produce their antirepressive effects either by competing with histones for binding at a 1 3 promoter or by physically displacing histones from the promoter (Felsenfeld, 1992). A third model posits that activators speed up the transition from a preinitiation or initiation complex into an elongation complex, allowing transcription to proceed (Rougvie and Lis, 1990; Spencer and Groudine, 1990; Kerppola and Kane, 1991). A specific example of this would be facilitating the phosphorylation of the CTD of the largest subunit of RNA pol H. These models need not be mutually exclusive; in fact, evidence exists for the participation of the transcriptional activation domain of VP16 (a prototypical activator) in several of these mechanisms. Because an understanding of how activators function is crucial to our understanding of transcriptional regulation, we have undertaken a detailed and systematic mutagenesis study of VP16. VP16 The expression of the genes of herpes simplex virus type 1 (HSV-l) occurs in a temporally regulated manner (Honess and Roizman, 1974; Honess and Roizman, 1975). Once inside the host cell, the HSV-l virion is transported to the nucleus, at which point the viral DNA is released into the nucleus. During the lytic cycle, the first HSV-l genes to be expressed are the immediate early (IE) genes, followed by the early genes and then the late genes (Honess and Roizman, 1974). The five immediate early genes encode transcriptional regulatory factors, and the expression of these genes is enhanced by the virus-encoded transcriptional activator protein VP16 (also known as a-TIF and Vmw65) (Post £131., 1981; Campbell £131., 1984). VP16 consists of 490 amino acids and has an apparent molecular weight of 65 kDa on SDS-PAGE. VP16 possesses two functional domains pertinent to its role in 1 4 transcription. The amino terminal region of the protein (the specificity domain) interacts with host cell factors that bind to IE gene promoter elements (McKnight £131., 1987; Gerster and Roeder, 1988; O'Hare and Goding, 1988; O'Hare £131., 1988; Preston £131., 1988; Triezenberg £131., 1988a). One of the host factors is Oct-1, a protein which binds to the octamer consensus sequence ATGCAAAT and is found inall tissue types; the octamer consensus sequence frequently overlaps the consensus DNA-binding sequence (TAATGARAT) for VP16 in HSV-l IE promoters (Sturm £131., 1988; Stem £131., 1989). A second host factor, variously called C1, VCAF-l, CFF, and HCF, is also required for VP16 to bind to the IE gene promoters (Kristie £131., 1989; Katan £131., 1990; Xiao and Capone, 1990). The second functional domain of VP16 is involved in transcription activation and resides in the carboxyl terminal 80 amino acids of the protein, and is both necessary and sufficient for activating transcription (Sadowski £131., 1988; Triezenberg £131., 1988b; Cousens £131., 1989; Greaves and O'Hare, 1989; Werstuck and Capone, 1989). The transcriptional activation domain of VP16 contains a high percentage of acidic residues, and has a net negative charge of -21 (Figure 2). Models for Transcriptional Activation by VP16 Much of the research in our lab to date has been focused on elucidating those structural elements critical for the function of the VP16 transcriptional activation domain. The groundwork for my research was laid by Doug Cress (Cress and Triezenberg, 1991 a), and will be briefly described below. Two theories have been proposed to explain key structural elements of acidic activation domains such as that of VP16. The first theory, proposed by crystallographer Paul Sigler (1988), was based on the apparent lack of sequence similarity among acidic activation domains, and suggested that the strength 15 LSTAPPTDVSLGDE CTG TCG ACG GCC CCC CCG ACC GAT GTC AGC CTG GGG GAC GAG 410 420 L H L D G E D V A M A H A D A CTC CAC TTA GAC GGC GAG GAC GTG GCG ATG GCG CAT GCC GAC GCG 430 L D D F D L D M L G D G D S P CTA GAC GAT TTC GAT CTG GAC ATG TTG GGG GAC GGG GAT TCC CCG 440 450 G P G F T P H D S A P Y G A L GGT CCG GGA TIT A00 000 CAC GAC TCC GCC CCC TAC GGC GCT CTG l 460 A456 DMADFEFEQMFTDAL GAT ATG GCC GAC TTC GAG TTI'GAG CAG ATG ITI'ACC GAT GCC CIT 470 480 G I D E Y G G END GGA ATI' GAC GAG TAC GGT GGG TAG 490 Figure 2. Nucleotide and deduced amino acid sequences of the VP16 activation domain (codons 410-490). Bold letters indicate acidic residues; outline letters indicate phenylalanine residues. The truncated VP16 activation domain (del456) is indicated. l 6 of a given acidic activation domain resulted primarily from its net negative charge. No secondary structure was necessary, hence the term "negative noodle." The second theory, advocated by Mark Ptashne (Giniger and Ptashne, 1987), proposed that acidic activation domains form amphipathic alpha helices (AAHs), with acidic residues forming one face of the helix and hydrophobic residues lying along another face of the helix. To support their model, Giniger and Ptashne designed a 15 amino acid peptide which, if in the form of an alpha helix, would be both amphipathic and acidic. A gene fusion was made between sequences encoding the GAL4 DNA-binding domain (aa 1- 147) and those encoding the 15 amino acid peptide. This fusion protein was capable of activating transcription from a test promoter. A peptide having the same residues but in scrambled order such that it would not form an AAH was not able to function as a transcriptional activation domain (Giniger and Ptashne, 1987). However, no evidence was provided that the 15 amino acid peptide actually formed an alpha helix. Doug Cress set out to test these two disparate theories, using site-directed mutagenesis to target specific residues in the VP16 activation domain. A VP16 activation domain truncated after amino acid 456 (del456) was used as the mutagenesis substrate. To test Sigler's "negative noodle" theory, acidic amino acids in the truncated VP16 activation domain were removed and replaced with their neutral counterparts, in increasing number and in various combinations. In general, a decrease in net negative charge was accompanied by a decrease in activator function (Cress and Triezenberg, 1991a). However, some combinations of changes possessing identical charge had significantly different activities, which contradicts the Sigler theory that net charge is the primary determinant of activator strength. This result suggested, perhaps not surprisingly, that some element of structure was an 1 7 important contributor to VP16 activation function. The Ptashne model, which suggested a specific secondary structure for acidic activation domains, was then tested. According to standard secondary structure prediction algorithms (Chou and Fasman, 1978; Garnier £131., 1978), the VP16 activation domain has the potential to form an alpha helix, and such a helix would be amphipathic. Groups of four acidic residues were replaced with neutral counterparts in a circularly permuted manner around the putative helix, generating what were termed "face" mutants. The assumption was that removal of negative charge from the center of the acidic face of an amphipathic alpha helix would be more detrimental to activity than removal of negative charge from the edges of the acidic face. However, no correlation was observed between predicted amphipathy and transcriptional activity of VP16 (Cress and Triezenberg, 1991a). For example, several mutants in which charge was removed from the center of the acidic face had more activity than mutants in which charge was removed from the edges of the acidic face. Since the amphipathy of the VP16 activation domain did not appear to contribute to its activity, additional mutations were made to determine if alpha helical structure was necessary for function. The amino acid proline is generally incompatible with alpha helical structure because the proline side chain is bonded to the nitrogen atom of the polypeptide backbone, which prevents the participation of the nitrogen in hydrogen bonding necessary to stabilize the helix (Creighton, 1984). Insertion of a potentially helix-breaking proline residue at position 425 of the truncated VP16 activation domain had no effect on activity; furthermore, substitution of two prolines simultaneously at positions 432 and 436 within the predicted alpha helix also had no effect on activity (Cress and Triezenberg, 1991a). Conversely, replacing 1 8 a phenylalanine at position 442 with proline abolished activity. Replacing Pro442 with helix-compatible but non-aromatic amino acids (Ala, Ser) did not restore activity to the VP16 activation domain. However, substitution of an aromatic amino acid, tyrosine, at position 442 restored activity to approximately 30% of VP16 del456 activity (Cress and Triezenberg, 1991a). After the discovery of the importance of Phe442, the amino acid sequence of the VP16 activation domain was compared to the sequences of other transcriptional activation domains. The sequences of several transcriptional activation domains were aligned using as a guide six bulky hydrophobic residues of the VP16 activation domain (Cress and Triezenberg, 1991a; Figure 3). In a number of different types of activation domains, bulky hydrophobic residues are observed at positions that can be aligned with Leu439 and Leu444 of VP16, on either side of the critical Phe442 (Figure 3). Overview Cress and Triezenberg (1991a) demonstrated the importance of Phe442 for the activity of the VP16 protein truncated after amino acid 456. Their results also suggested that the important characteristic of phenylalanine at this position was its aromaticity. In the following chapter, I describe work using site-directed mutagenesis to further our understanding of the critical Phe442 and of other regions of the VP16 activation domain important to its ability to activate transcription. The phenylalanine at position 442 of the truncated (del456) VP16 activation domain was thoroughly mutagenized to obtain nearly all other amino acid substitutions at this position. The results of this mutagenesis of position 442 confirm the hypothesis of Cress and Triezenberg (1991a) that the aromatic character at this position is critical; truncated VP16 proteins containing either of the other two aromatic amino 19 VP16type1 HLDGEDTAMAHADA TDDKOTDMLGDGD VPIGIypez nLocEE v DMTPADA L one o L EMLGDVE SPHA) NLONOOV LTGLPGV M PN I o v ovuporo spun) IIRTPTV cpncovs WOTL o L ouLovon ennui-2 LFPDDWY PPSIDPADLDESWD v IFETTE nu LNLDSPLTPELNEI LDTF LNDECL mun) ousripjosuenoa L HGF owseeoous mum NSOALSQ PIASSNV H our: u unsuras com AVVESFF sssrosrp M FEY ENL sousxew mean) TLADNKF sv LPPT L as L M eeoocuu LEU3 NWESDMLWROVDILMNEFA F NPKV on: AIRYPPHLNPODPLKDLVSLACDPA p53 LpspncEj'ooLLLPoov est: FEGPSEAL 817 LACEDNSGLPEESOFOTW L NAVI AH ELGE L GEL GAL Looo Figure 3. Alignment of hydrophobic residues from various transcriptional activation domains with that of VP16 (revised and extended from Cress and Triezenberg, 1991a). VP16 type 1 and 2 refer to VP16 proteins from HSV-l and HSV-2 (Cress and Triezenberg, 1991b), respectively. Sp1 (Courey and Tjian, 1988) and CTF (Mermod £131., 1989) are mammalian glutamine- and proline- rich transcriptional activators, respectively. EBNA-2 and Rta are activators from Epstein-Barr virus (Cohen, 1992; Hardwick £131., 1992). GAL4 (Ma and Ptashne, 1987a), GCN4 (Hope and Struhl, 1986), HAP4 (Forsburg and . Guarente, 1989), and LEU3 (Zhou £131., 1987) are yeast activators. p53 is a human tumor suppressor protein (Harlow M" 1985). B17 is a randomly- cloned E4911 genomic fragment (Ma and Ptashne, 1987b). AH is an artificial sequence designed to form an amphipathic alpha helix (Giniger and Ptashne, 1987). 2 0 acids (T yr and Trp) were approximately 30-35% active relative to wild type del456. Hydrophobic residues supported an activity level 10-15% of wild type del456, while all other amino acids tested at this position (with the exception of asparagine) were less than 10% active. The pattern of bulky hydrophobic residues observed by Cress and Triezenberg suggested that leucines at positions 439 and 444 of VP16 may be important for transcriptional activity (Figure 3). These two leucine residues of VP16 were also targeted for mutagenesis; substitution of small hydrophobic or hydrophilic amino acids for either leucine greatly decreased activity, a result which is supportive of the Cress model. We also show that the addition of amino acids 457-490 to the truncated VP16 activation domain partially restores the activities of truncated VP16 mutants inactivated by changes at Phe442, implying functional redundancy within the full-length VP16 activation domain. Within this added region, a phenylalanine at position 473 is found in a context similar to that surrounding Phe442. Changing this residue to non-aromatic amino acids only modestly affected the activity of this region. Instead, mutations at a neighboring phenylalanine (at position 475) had somewhat greater effects on activity, despite it not agreeing as well as Phe473 with the Cress model. Our work suggests that two regions of the VP16 activation domain have different structural elements and may have different mechanisms for achieving transcriptional activation. Our lack of success in identifying amino acids important to the function of the distal subdomain of the VP16 activation domain led us to try a more efficient method for generating mutations in VP16 and testing their effects on the ability of VP16 to activate transcription. In Chapter HI I describe a method of randomly mutating the VP16 activation domain using DNA- 2 l damaging chemicals, which, combined with an 111311112 selection in yeast, enabled us to identify mutations in residues of the VP16 activation domain which decrease function. This strategy identified the region of VP16 between amino acids 480 and 485 which is important for the function of the distal subdomain. In Chapter IV I discuss our current model of how the two subdomains of the VP16 activation domain might function to activate transcription. This model is based upon our mutagenesis results and the results of collaborations in which the various VP16 mutants described here were used to investigate specific aspects of activated transcription. CHAPTER II SITE-DIRECTED MUTAGENESIS OF THE VP16 TRANSCRIPTIONAL ACTIVATION DOMAIN Introduction Previous work by Cress and Triezenberg (1991 a) revealed the importance of Phe442 to the activity of the truncated VP16 activation domain (del456), and also suggested that bulky hydrophobic residues surrounding Phe442 (Leu439 and Leu444) might also contribute to the activity of VP16 del456. Using site-directed mutagenesis, I extended this work by generating and testing further mutations at Phe442; the results of further Phe442 substitutions suggests that an aromatic or hydrophobic residue is critical at this position. Substitutions of small hydrophobic or hydrophilic amino acids for Leu439 or Leu444 greatly reduced activity of VP16 del456. I also tested the activities of a variety of mutations at Phe442 in the context of the full-length VP16 activation domain. The full-length Phe442 mutants had increased activity compared with their truncated counterparts, suggesting that the added carboxyl-terminal region (aa 457-490, designated the distal subdomain) contributes to the overall activity of the VP16 activation domain. A pattern of amino acids similar to that surrounding Phe442 was observed in the added region. I tested the hypothesis that Phe473 (or possibly Phe475) was analogous to Phe442; however, mutations at these phenylalanines in the distal subdomain did not decrease activity to the same degree as similar mutations at Phe442, suggesting that the proximal and distal subdomains activate 22 2 3 transcription through different mechanisms. These results were published in the PrmeedingmflhLNan'onaLAcademiLQLSciencesllSA (1993, Vol. 90, pp. 883-887). Chapter 11 Methods Mutagenesis and Cloning $311 / BamHI fragments of the VP16 gene corresponding to the truncated (codons 411-456) or full-length (codons 411-490 plus 7 bp of 3' nontranslated sequence) activation domain were cloned into M13mp19 (Norrander £131., 1983). Oligonucleotide-directed mutagenesis was performed as described (Zoller and Smith, 1982; Kunkel, 1985; Cress and Triezenberg, 1991a). I prepared single-stranded uridine-containing template DNA by replicating the template in the £3911 host strain CJ236 [dm-l, nag-1, 1111-1, {£1A-1; pCJ105(Cmr F')] which results in occasional uridines incorporated in place of thyrnidines in all DNA synthesized in the bacterium (Bio-Rad Muta-Gene in 211112 mutagenesis kit instruction manual, catalog #170-3571). The uracil- containing single-stranded DNA was purified by extensive phenol / chloroform and chloroform extraction, Biogel P-60 column chromatography, and ethanol precipitation. The uracil-containing single- stranded DNA was resuspended in 20 ul of TE (10 mM Tris-Cl, pH 7.5, 1 mM EDTA). Second-strand synthesis was performed 111111119 using a phosphorylated mutagenic oligonucleotide as a primer (Table 1). In a reaction volume of 10 ul, 1 pl (approximately 6 pmole) of phosphorylated primer was combined with 1 ul of uridine-containing template DNA in annealing buffer (20 mM Tris-Cl, pH 7.5, 2 mM MgC12, and 50 mM NaCl) and the annealing accomplished by heating the sample to 65°C and cooling gradually to 4°C. 24 I.D. Sequence Mutation sr-17 5'-GTCCAGATCC(a/c)AATCGTCTAA-3' F442 to L, w ST-18 5'-GAAATCGTCA(a/g)(a/c)CGCGTCGGC-3' L439 to F, s, v, A sr-19 5'-CAACATGTC(t/a)(a/g)(c/a)ATCGAAATC-3' L444 to F, s, v, A ST-21A 5'CTCAAACTCA(g/a)(g/c)GTCGGCCAT-3' F473 to P, A, v, L 51222 S'CATCTGCTCMg/a)(g/c)CI'CGAAGTC-3' F475 to P, A, v, L sr-25 5'—GTCCAGATC(c/a)(t/a/c)(t/g/c)ATCGTCTAA-3' F442 to K, N, M, 1, R, Q, H, E, D, v, G, L, s ST-26 5'-CATCTGCTCA(g/a)CCTCGAAGTC-3' F475 to v, A sr-27 5'-CTCAAACTC(c/a)(c/t)AGTCGGCCAT—3' F473 to w, Y, C ST-28 5'-CAACATGTCCACATCGAAATC-B' L444 to v 51229 5'-CATCTGCTC(g/C)(t/c)ACTCGAAGTC-3' F475 to w, Y, C Table 1. Mutagenic oligonucleotides. The sequences of the oligonucleotides used to alter the VP16 gene are listed in this table. Parentheses indicate mixed positions. The specific mutations conferred by each oligonucleotide are also indicated. Fan Shen performed the mutagenesis with ST-25. These oligonucleotides were synthesized by MSU Department of Biochemistry Macromolecular Structure Facility and used without further purification. 2 5 After annealing of the primer, the complementary DNA strand synthesis was performed by adding 1 ul of 10X extension buffer (175 mM Tris-Cl, 37.5 mM MgC12, 215 mM DIT, 7.5 mM ATP, 4 mM each deoxynucleotide triphosphate) and 1 ul (1 unit) each of T4 DNA polymerase and T4 DNA Iigase. The . reactions were incubated on ice for 5 minutes, then at 25°C for 5 minutes, and finally at 37°C for 1.5 hours; the reactions were terminated by addition of 87 pl of TE. f Mutagenic synthesis reactions were then used to transform dut+ ung+ MV1193 cells [Adar-12mm; r1251. thi. endA. spams, 113134, Marl-Les A)306::Tn1_Q(tet’) F'(1mD36, Mt 13:19 MMISH, made competent by treatment with 100 mM CaClz. Transformation of the mutagenized duplex DNA into MV1193 results in selection against the uracil-containing DNA strand due to the action of uracil N-glycosylase. The resulting plaques were purified by picking one with a small pipette tip. Each tip was used to inoculate a 3 ml culture of a 1:60 dilution of an overnight culture of MV1193 in LB media. Mutations were identified by dideoxy sequencing (Sanger £131., 1977) and double-stranded phage DNAs containing the desired mutant VP16 activation domains were harvested. The 5311/ BamI-II fragments of VP16 encoding the altered activation domains were cloned downstream of codons 1—410 using the expression vector pMSVP16 (Triezenberg £131., 1988b). After cloning into the expression vector each mutant VP16 activation domain was sequenced to ensure fidelity. Transient transfection assay Mouse L cells were grown in Dulbecco's modified Eagle medium (DMEM; Gibco) supplemented with 10% fetal calf serum (HyClone Laboratories). Cells (8x105 per 60—min culture plate) were transfected by 2 6 DEAE-dextran (Lopata £131., 1984). In this method, DEAE-dextran at 200 ug/ ml was dissolved in DMEM supplemented with 10 mM HEPES, pH 7.2. This solution was combined with the following plasmids: an expression plasmid pMSVP16 (50 ng), a reporter plasmid (2 ug of pSJT 703) containing the ICP4 IE regulatory sequences fused to the body of the HSV-I 1k gene, and an internal control plasmid (2 ug of pMSV—1k) containing the MSV-LTR promoter fused to the 11; gene body (Graves £131., 1985). The reporter and internal control 1k genes differed in the location of the transcription startsite. Cells were incubated with the DEAE-dextran/ DNA solutions for 5 hours at 37°C. At the end of the incubation period the cells were shocked with a dimethyl sulfoxide solution (10% DMSO, 140 mM NaCl, 5 mM KCl, 0.7 mM NazHPO4, 0.1% dextrose, 20 mM I-IEPES) for 3 minutes. After the DMSO shock, the cells were washed with 2 ml DMEM/HEPES, after which 5 ml of DMEM supplemented with 10% fetal calf serum was added and the plates incubated at 37°C. Total RNA was harvested 48 hours after transfection as described (Triezenberg £131., 1988b). For the RNA harvest, solutions were made RNAase-free whenever possible. Transfected cells were incubated with 2.5 ml of a 200 ug/ml proteinase K solution (10 mM Tris-Cl, pH 7.5, 5 mM EDTA, 1% SDS), the cells were scraped off the plates into 15 ml culture tubes, and the samples incubated at 50°C for 3 hours. The samples were extracted two times with phenol (saturated with DEPC-treated water) followed by chloroform. Nucleic acids were precipitated by addition of ethanol followed by centrifugation. The pellets were resuspended in 400 111 of TE and transferred to 1.5 ml microfuge tubes. The ethanol precipitation was repeated, and the final nucleic acid pellets resuspended in 100 111 TE. The samples were digested for 2 hours at 2 7 37°C with DNAase I (5 ug DNAase I, 50 ug heparin, 20 mM HEPES, pH 7.2, 1 mM CaClz, 1 mM MgC12, 5 mM MnClz). Following the DNAase I digestion, the samples were extracted again with phenol and chloroform and ethanol precipitated. RNA pellets were washed with 70% ethanol and resuspended in 25 ul of TE. Primer extension assay The 11; primer used in the primer extension assay was phosphorylated by combining 45 ng of 1k oligonucleotide, 30 uCi of [7-32P]-ATP, and 15 units of T4 polynucleotide kinase in kinase buffer (70 mM Tris-Cl, pH 7.5, 10 mM MgClz), in a reaction volume of 10 ul. The kinase reaction was incubated at 37°C for 2 hours, and the reaction stopped by addition of 2 u] of 0.5 M EDTA and 50 ul of TE, followed by heating at 65°C for 5 minutes. Unincorporated ATP was removed from the labeled 11g primer solution by binding the oligonucleotide to a DEAE-cellulose column in TEN 100 buffer (10 mM Tris- Cl, pH 7.5, 1 mM EDTA, 100 mM NaCl), washing with 1 ml of TENlOO and then 500 111 of TEN300 (TE plus 300 mM NaCl), and finally eluting the labeled oligonucleotide with 400 111 of TEN600 ('TE plus 600 mM NaCl). The sample was frozen at -20°C until use. Ten ul of each RNA sample was incubated with 3.5 ul of the 3ilP-labeled 1k primer in a buffer composed of 150 mM KCl, 10 mM Tris-Cl, pH 8.3, and 1 mM EDTA. The annealing was performed at 65°C for 1.5 hours, after which the samples were allowed to cool to 25°C. The primer extension reaction was performed by adding 30 ul of primer extension reaction buffer [5 units AMV reverse transcriptase (Life Sciences, Inc.), 20 mM Tris-Cl, pH 8.3, 10 mM MgC12, 6 mM DTT, 150 ug/ ml actinomycin D] to the annealed primer sOlution, and incubation at 37°C for 1 hour. The samples were treated with 20 ug/ ml RNAse A plus 100 ug/ ml salmon sperm DNA for 15 minutes at 37°C, 2 8 and then extracted with phenol / chloroform and ethanol precipitated. Primer extension products were resuspended in 10 111 of formamide loading dye (95% formamide, 20 mM EDTA, and 0.05% bromophenol blue and xylene cyanol). The ICP4-1k and MSV-1k RNAs yielded primer extension products of 81 and 55 bases, respectively; these products were separated by electrophoresis on a 9% acrylamide denaturing (7 M urea) gel, 0.5x TBE buffer, and detected by autoradiography. Quantitation of primer extension products The developed film and the dried gel were aligned and portions of the gel corresponding to bands on the film were cut out. Cerenkov activity of each gel slice was detected by a 5 minute scan on the tritium setting of a Packard 300 liquid scintillation counter. Alternatively, the dried gel was used to expose a phosphor screen, and the image detected by a Molecular Dynamics PhosphorImager. Activity of each band was determined using the ImageQuant program and volume integration with correction for background. Determination of mutant protein stability Mouse L cells (8x105 per 60-mm culture plate) were transfected with 10 ug of wild-type or mutant pMSVP16 as described above. Forty eight hours after transfection, the cells were lysed in 2 ml of SDS buffer (50 mM Tris, 0.2 M NaCl, 20 mM EDTA, 0.5% SDS) and scraped into 15 ml polystyrene conical tubes and sonicated. After sonication, the samples were transferred to 15 ml polypropylene conical tubes, and total protein was precipitated with 8 ml of cold acetone. The protein pellets were resuspended in 200 111 of SDS-PAGE sample buffer (62 mM Tris-Cl, pH 6.8, 2% SDS, 10% glycerol, 5% 2- mercaptoethanol); approximately 40 ul of each sample was electrophoresed on a 4% stacking/10% resolving SDS-PAGE gel. Proteins were electrophoretically 2 9 transferred to nitrocellulose using a Western Mini Transphor TE22 apparatus (Hoefer Scientific) and VP16 protein was detected by C8-series anti-VP16 antisera from rabbit (Triezenberg £131., 1988b). Primary antibody was visualized using a biotinylated goat anti-rabbit secondary antibody and avidin/biotinylated enzyme complex, the substrate for which was 4-chloro-1- naphthol (Vector Labs). RESULTS Mutagenesis of Phe442 Cress and Triezenberg (1991a) suggested that Phe442 is critical for function of the truncated VP16 activation domain (del456). To more thoroughly test this hypothesis, additional amino acid substitutions were made at position 442. Fan Shen in our laboratory was responsible for the generation and testing of eleven Phe442 mutants; I made and tested the FW442 and FL442 mutants, while the remaining mutants have been repoi'ted previously (Cress and Triezenberg, 1991a). Activities of VP16 mutants were determined by transient transfection assays in which a plasmid expressing the VP16 gene was cotransfected with a reporter plasmid bearing the HSV- thymidine kinase (1k) gene under the control of a VP16-responsive promoter and an internal control plasmid consisting of the 1k gene regulated by the MSV LTR promoter. Total RNA was harvested and the amount of HSV-fls RNA was quantitated by primer extension assay and scintillation spectroscopy. Of the nineteen possible changes at position 442, seventeen have been generated and their activities tested. Substitution of the other two aromatic amino acids, tyrosine and tryptophan, for Phe442 decreased but did not abolish function (Figure 4). No other substitution mutants had an activity greater than fifteen percent relative to wild type del456 (Table 2). The no VP16 del456 FY 442 FW 442 FL 442 IE-tk E i | l i l l 1 Control Figure 4. Autoradiogram of primer extension assay reflecting the activities of truncated VP16 proteins altered at Phe442. Positions of the reporter (IE-tk) and internal control primer extension products are indicated. The size and stability of all mutant proteins were confirmed by Western blotting and immunodetection by anti-VP16 antisera (see Figure 5). 31 Amino acid at Relative activity Position 442 (% of wt del456) Tyr" 30 i 10 Trp 36 i 6 Leu 14 :l: 2 Pro" 510 Ala" 14 :l: 3 Ser" 510 Gly $10 Val 11 i 2 Ile 14 i 2 Met 12 i 1 Lys 510 Arg 510 His 510 Asp 510 Glu S10 Asn 11 :1: 2 Gln $10 Table 2. Relative activities of truncated (del456) VP16 mutants bearing amino acid substitutions at position 442. Relative activities are calculated as the ratio of the reporter signal (IE-tk) to the internal control signal, normalized to the activity of wild type del456. Means and standard deviations were calculated from at least four independent transfections. Asterisks indicate activities previously reported (Cress and Triezenberg, 1991a). Fan Shen is responsible for all the data in this figure except FW442, FL442, and those identified with an asterisk. 3 2 substitution of bulky hydrophobic residues (Leu, Ile, and Met) and the smaller hydrophobic residue Ala reduced activity to approximately ten to fifteen percent of del456 activity. All other substitution mutants were no more than ten percent active, with the exception of the asparagine substitution (11%). Interestingly, increasing the net negative charge by substitution of acidic residues at position 442 had no positive effect. To determine the stability of mutant proteins, immunoblots were probed with polyclonal antisera directed against VP16, an example of which is shown in Figure 5. All of the mutant proteins described in this dissertation showed no significant differences in size or stability from their wild type VP16 parent. These results strongly support the suggestion stated previously (Cress and Triezenberg, 1991a) that the aromatic character of amino acid 442 is particularly important, and that hydrophobic residues at this position are less effective but retain some function. Hydrophobic Residues Flanking Phe442 Also Contribute to Activity Alignment of the amino acid sequences of several transcriptional activation domains revealed an intriguing pattern of bulky hydrophobic residues (Cress and Triezenberg, 1991a). In the VP16 activation domain such residues include leucines at both positions 439 and 444 flanking Phe442 (Figure 3). To determine if these leucines contribute to the activity of the truncated VP16 activation domain, single amino acid substitutions were made and tested in transient transfection assays. Substitution at either position of a small hydrophobic (Ala) or a hydrophilic residue (Ser) for leucine diminished activity significantly (Figure 6). Substitution of another bulky hydrophobic residue (Val) or an aromatic hydrophobic residue (Phe) decreased activity only slightly. A bulky hydrophobic amino acid was perhaps 33' Full-length ‘L’.’ (‘0 ('0 00 to d) ‘3 rs I~ 3 rs x <- -...l u.u. FV475 FA475 FP475 Position 473 Position 475 Figure 9. Effects of amino acid substitutions at Phe473 and Phe475 of VP16, tested in the context of the FA442 mutation. Rectangles indicate mean relative activities (normalized to full-length wild type VP16) calculated from at least five independent transfections; error bars represent one standard deviation. FA442 FL represents the activity of the FA442 mutation in full- length VP16 (taken from Figure 7). 4 0 the 473 region differ qualitatively in their effect on VP16 activity. We conclude from these results that the proximal and distal subdomains of VP16 have distinct structural features dependent on different patterns of amino acids. Conclusions Using oligonucleotide-directed mutagenesis and transient transfection assays, we have analyzed the effects of amino acid substitutions on the activities of both truncated and full-length VP16 activation domains. In the truncated (del456) domain, we have now mutated residue Phe442 to near saturation. Aromatic amino acids were the best substitutes for Phe442, decreasing activity to approximately 30% of wild-type del456 activity. Other hydrophobic substitutions reduced activity to between 10% and 15% of wild- type, while all other substitutions reduced activity to no more than 10% of wild-type activity. Limited mutational analysis of leucine codons 439 and 444 revealed that, although these positions are somewhat less sensitive to mutation, hydrophobic residues at these positions also contribute to the activity of the truncated VP16 activation domain. Thus, the pattern of bulky hydrophobic residues previously observed (Cress and Triezenberg, 1991a) among a variety of transcriptional activation domains has accurately predicted the importance of such residues in the truncated VP16 domain. Whether this pattern has predictive value for other activation domains is presently being tested. Replacement of Trp454 with alanine in the activation domain of the EBNA-2 protein from Epstein-Barr virus reduced the activity of this domain by approximately 80% (Cohen, 1992); in the sequence alignment of Cress and Triezenberg (1991a), Trp454 aligns with Phe442. A portion (aa 564-587) of the activation domain of Rta, another transcriptional 4 1 activator protein from Epstein-Barr virus, also fits the Cress and Triezenberg pattern. Simultaneous substitution of glycines for the hydrophobic residues L578, F581, and L582 of Rta (corresponding to L439, F442, and L444 of VP16, respectively) reduced activity of the Rta activation domain by 80 to 90% (Hardwick $31., 1992). Thus it appears that the pattern of hydrophobic and aromatic amino acids observed by Cress and Triezenberg can accurately predict important residues in the activation domains of another herpesvirus transcriptional activator protein. A partial restoration of transcriptional activity of previously inactive Phe 442 mutants was achieved by adding back residues 457-490 of the VP16 activation domain. The activity of the full-length protein reflected the activity of its truncated parent. For example, FA442 del456 was approximately 15% active relative to wildtype del456, and full-length FA442 was approximately 45% active; FY442 del456 was approximately 35% active relative to wildtype del456, and full-length FW442 was approximately 70% active relative to the activity of full-length VP16. An examination of the amino acid sequence of residues 457-490 revealed a pattern similar to that surrounding Phe442, with a phenylalanine at position 473 corresponding to Phe442. However, substitutions at Phe473 were not nearly as deleterious to the activity of the distal subdomain as similar substitutions at Phe442 were to the activity of the proximal subdomain. Mutations at Phe475 had a somewhat greater effect on the activity of the distal subdomain. Therefore, the pattern of bulky hydrophobic residues observed by Cress and Triezenberg, which accurately predicted the importance of two leucines flanking Phe442 to the activity of VP16 del456 (the proximal subdomain), was not sufficient in predicting residues in the distal subdomain important for its function. This suggested that the two subdomains might 4 2 activate transcription through distinct mechanisms. The results described here strengthen our previous arguments against prior hypotheses about structural features of acidic activation domains. On one hand, VP16 seems not to fit the "acid blob or negative noodle" hypothesis (Sigler, 1988), which suggests that the activity of such domains is primarily a function of net negative charge (Gill and Ptashne, 1987; Ma and Ptashne, 1987a). Numerous mutations of VP16 at Phe442 have no effect on net charge, and yet they have dramatic effects on transcriptional activation. Furthermore, increasing the net negative charge by replacing Phe442 with Asp or Glu had detrimental rather than beneficial effects on VP16 activity. Apparently, having an aromatic or hydrophobic residue at position 442 is of greater importance than is the net charge of the proximal subdomain. On the other hand, our results are also inconsistent with a model of VP16 as an amphipathic alpha helix (Giniger and Ptashne, 1987; Zhu $31., 1990). We have previously shown that introducing two potentially helix- breaking proline residues at positions 432 and 436 in the predicted helix of truncated VP16 had no effect on transcriptional activation (Cress and Triezenberg, 1991a), suggesting that the predicted helix (if it does exist) is not necessary for VP16 function. Here, we show the deleterious effect of replacing Phe442 with any of a number of residues predicted to maintain an amphipathic alpha helix, suggesting further that such a structure is not sufficient for activity of the proximal subdomain. These conclusions from our mutational analysis are reinforced by recent spectroscopic studies of the VP16, GCN4, and GAL4 activation domains (Donaldson and Capone, 1992; Van Hoy $31., 1992; Van Hoy $31., 1993) that found little evidence of helical structure. In fact, the GCN4 and GAL4 activation domains appear to form [5- sheets (Van Hoy $31., 1992; Van Hoy $31., 1993). 4 3 Although our mutations in the truncated VP16 activation domain indicate that an aromatic amino acid is strongly preferred at position 442, we do not yet understand the reason for this preference. The two simplest ideas for the role of Phe442 are either that it is necessary for maintaining the structure of the domain or that it is directly involved in interactions with target proteins in the activation mechanism. Likewise, the reason for the preference for hydrophobic amino acids at positions 439 and 444 is also unknown. It is possible that the combination of hydrophobic residues at positions 439 and 444 is important in correctly positioning Phe442 for making contact with its target. These questions can be addressed by probing the structure of VP16, using spectroscopic or crystallographic methods, and by exploring the association of VP16 with putative target proteins. In addition to the analysis of the truncated activation domain, we have begun to examine the role of the extreme carboxyl-terminal or distal subdomain of VP16 (aa 457-490). Adding this subdomain onto defective Phe442 mutants partially restored transcriptional activity (Figure 7). Intriguingly, a pattern of acidic and bulky hydrophobic amino acids surrounding Phe473 in this subdomain strikingly resembles the pattern surrounding Phe442 (Figure 8). Furthermore, insertion of four amino acids at codon 471 reportedly abolished transcriptional activity of VP16 (Werstuck and Capone, 1989). However, the types of amino acid substitutions that significantly affected the activity of the truncated activation domain had quantitatively different effects upon the distal subdomain. Specifically, Phe473 (best aligned with Phe442) was relatively insensitive to mutations, whereas Phe475 (best aligned with Leu444) was somewhat more sensitive. These results imply that the pattern of acidic and hydrophobic amino acids, although useful in predicting important residues surrounding Phe442, does CHAPTERIII RANDOM MUTAGENESIS OF THE VP16 TRANSCRIPTIONAL ACTIVATION DOMAIN Introduction The results of the site-directed mutagenesis work described in Chapter H suggested that the VP16 transcriptional activation domain is composed of two subdomains, designated proximal and distal. However, the kinds of mutations that inactivated the proximal subdomain had only modest effects on the activity of the distal subdomain. This suggested to us that the proximal and distal subdomains activate transcription by different mechanisms. Because we lacked clues for targeting specific residues in the distal subdomain for mutagenesis, we decided to randomly mutate the VP16 activation domain and employ a biological selection in yeast to select for mutations in the VP16 activation domain that decrease its function. Using nitrous acid as the mutagen, several missense mutations in the VP16 activation domain were detected that decreased its activity; four mutants had substitutions in the distal subdomain (TA458/DN461/MV478, TA480, IV485, and YC465). The effects of these mutations on the activity of the distal subdomain were tested by determining the abilities of these mutants to activate transcription of a reporter gene; three of the mutations (TA458/DN461/MV478, IV485, and YC465) significantly reduced the activity of the distal subdomain, while the fourth mutation (TA480) had less of an effect. 45 4 4 not correctly describe key residues in the distal subdomain. It is possible that the sequence pattern observed by Cress and Triezenberg may still be predictive of important amino acids in other activation domains that activate by the same mechanism as the proximal subdomain of VP16. Although secondary structure prediction algorithms suggest that residues 468-478 might fold into an amphipathic alpha helix, our mutational analysis argues against this notion. First, a proline substitution at Phe473 had little effect on the apparent activity of the distal subdomain. Second, the replacement of Phe475 with Ala should be compatible with an amphipathic helix, and yet it had the greatest effect on the activity of this subdomain. However, it is unclear from these results whether the net negative charge of the distal subdomain is a primary determinant of the activity of that domain, since no mutations altering charge were made. We conclude that the proximal and distal subdomains both contribute to the overall activity of VP16, but the structural features necessary for the function of the two subdomains must differ considerably. A difference in structural features between the proximal and distal subdomains implies that they might activate transcription through different mechanisms. Thus the information gained from the mutagenesis of the proximal subdomain is not likely to be applicable to the distal subdomain, and we were left with no real clues as to which amino acids in the distal subdomain are important for its function. We therefore decided to attempt to identify important residues in the distal subdomain by means of a genetic selection. CHAPTER III RANDOM MUTAGENESIS OF THE VP16 TRANSCRIPTIONAL ACTIVATION DOMAIN Introduction The results of the site-directed mutagenesis work described in Chapter H suggested that the VP16 transcriptional activation domain is composed of two subdomains, designated proximal and distal. However, the kinds of mutations that inactivated the proximal subdomain had only modest effects on the activity of the distal subdomain. This suggested to us that the proximal and distal subdomains activate transcription by different mechanisms. Because we lacked clues for targeting specific residues in the distal subdomain for mutagenesis, we decided to randomly mutate the VP16 activation domain and employ a biological selection in yeast to select for mutations in the VP16 activation domain that decrease its function. Using nitrous acid as the mutagen, several missense mutations in the VP16 activation domain were detected that decreased its activity; four mutants had substitutions in the distal subdomain (TA458/DN461/MV478, TA480, IV485, and YC465). The effects of these mutations on the activity of the distal subdomain were tested by determining the abilities of these mutants to activate transcription of a reporter gene; three of the mutations (TA458/DN461/MV478, IV485, and YC465) significantly reduced the activity of the distal subdomain, while the fourth mutation (TA480) had less of an effect. 45 4 6 Biological Selection in Yeast An overarching theme evident from the results of the site-directed mutagenesis of the VP16 activation domain described in Chapter H is that our ability to identify potentially important amino acids using sequence alignments is very limited. Thus, in collaboration with L. Guarente’s laboratory at MIT, we devised a biological selection to identify important amino acid changes from randomly mutated VP16 activation domains. Our system takes advantage of the observation that the GAL4-VP16 fusion protein, when present at high concentration, inhibits the growth of yeast, resulting in very small colonies (Berger $31., 1992). This phenomenon has been referred to by some researchers as "squelching" (Gill and Ptashne, 1988). GAL4-VP16 consists of the DNA-binding domain of the yeast transcriptional activator protein GAL4 (aa 1-147) fused to the VP16 activation domain (aa 411- 490). Interestingly, VP16 activation domains which are partially deficient in activity are also less inhibitory of yeast growth. In fact, the abilities of certain VP16 activation domains to inhibit growth correlates quite well with the activities of these mutants as determined in our transient transfection assays (Berger $31., 1992). The strategy, (illustrated in Figure 10, is to randomly mutate the VP16 activation domain, and then clone the resulting collection of domains downstream of the GAL4 DNA-binding domain. The GAL4-VP16 genes are on a high copy number (2 micron) yeast plasmid, and expressed constitutively from the strong ADHJL promoter. Yeast are transformed with these constructs by the lithium acetate procedure, and then plated out on media which selects for transformants. The growth of yeast bearing GAL4-VP16 proteins which have fully active VP16 activation domains should be strongly inhibited. Conversely, GAL4-VP16 proteins in which the VP16 activation domain is 47 VP16 activation domain . . chemical . > mutagenesis . anneal primer second—strand synthesis GAL4 I VP16 2n 4 excise activation domain; clone into expression vector transiorm into yeast; late onto selective media Figure 10. Strategy for chemical mutagenesis and subsequent biological selection in yeast. 4 8 significantly less active should be less inhibitory of the growth of the host yeast. Thus, we have a positive growth selection for defective VP16 activation domains. The names and salient features of the plasmids used in the yeast experiments are described in Table 3. The GAL4-VP16-expressing plasmid pSB201, containing both promoter and terminator sequences from the yeast A1211]. gene and the GAL4 DNA binding domain (aa 1-147), was modified to easily accept DNA fragments encoding mutated VP16 activation domains. After mutagenesis of VP16 activation domains, the resulting collection of restriction fragments was cloned into the modified expression vector (designated prR92) in a directed, one step ligation reaction. Modification of pSBZOl also allowed easy removal and transfer of mutagenized VP16 activation domains to a new yeast vector; this is an important control to ensure that it is a mutation in the VP16 activation domain, and not in some other part of the expression vector, which gives rise to the large colony phenotype. Once mutations in the VP16 activation domain had been identified, the transcriptional activity of each mutant was quantitated by cloning the gene for the mutant GAL4-VP16 into the low copy expression vector pSB202delSma, and then assaying the abilities of the mutants to activate expression of a reporter gene, B-galactosidase. Chemical Mutagenesis Chemical mutagenesis was employed initially as a method of modifying DNA. The mutagenesis substrate mpJR92, consisting of single-stranded M13 DNA containing the full-length VP16 activation domain, was exposed to two chemicals which modify bases in single-stranded DNA without breaking the 49 Plasmid Name: pDBZOLBng pD820.1 YPIR92 mpJR92 pSBZO2 pSB202delSm3 pLGSD5 Features: 2 micron (high copy) origin of replication; LEUZ selectable marker; A1231 promoter and terminator, in between which is a unique 13ng cloning site. The same as pDB20LBg1H, except that a 5331 site at position 25 has been eliminated. Contains a BamI-II/ 13ng fragment encoding GAL4-VP16 cloned into the fing site of pDB20.1. Used for 1111110 toxicity assay. Contains the sequence of the full-length VP16 activation domain cloned as a 5311/ 1333111 fragment into M13mp19. Also contains a unique BglII site 3' of the BamHI site. Used as the substrate for chemical mutagenesis. An ARS/CEN (low copy) plasmid; 1.13112 selectable marker; ADI-11 promoter driving expression of GAL4-VP16. Used for B- galactosidase assay. The same as pSBZOZ, except that VP16 codons 411-452 have been deleted. 2 micron (high copy) origin of replication; URAfi selectable marker; contains the gene for B-galactosidase driven by the §_Y_C1 promoter and regulated by UASG. Used as the reporter for B-galactosidase assay. Table 3. Description of plasmids used in the yeast biological selection and B-galactosidase assay. 5 0 phosphodiester backbone (Myers $31., 1985; Sambrook $31., 1989; Todo $31., 1990). Nitrous acid deaminates deoxycytidine, deoxyadenosine, and deoxyguanosine, changing these nucleosides to deoxyuridine, deoxyinosine, and deoxyxanthidine, respectively. Formic acid acts by breaking the bonds joining purine bases to deoxyribose, thereby depurinating DNA (Myers $31., 1985). Together these chemicals can potentially damage three out of the four bases at a known frequency at specific concentrations and incubation periods. A potential disadvantage of chemical mutagenesis combined with our biological screen is that only those mutations which have a significant effect on VP16 activity will be identified. Therefore, we will miss mutations that have no effect on activity, which can also be informative. In addition, the cloning strategy subsequent to the chemical mutagenesis to transfer the chemically treated VP16 activation domains into the expression vector prR92 utilizes a convenient restriction site (Sad, at codon 424) that is located within the VP16 activation domain; thus, mutations 5' to the SacI site that might reduce VP16 function would not be detected. We felt that such mutations were unlikely, since previous deletion analyses showed that removing codons 413 to 429 of the VP16 activation domain had no effect on function (Triezenberg $31., 1988b). Two other questions concerning chemical mutagenesis are the efficiency and the distribution of base changes. Based on efficiencies reported in the literature (Myers $31., 1985), we calculated that the chemical mutagenesis described here would be efficient enough so that we could be reasonably confident that the VP16 activation domain target (approximately 250 nucleotides) would suffer frequent damage. Changing incubation times and mutagen concentrations can increase the mutagenesis frequency if that is required. If large colonies were not observed with this mutagenesis protocol, 5 1 we can sequence mutagenized activation domains directly to see if the mutagenesis frequency is what we expect. Since the target DNA is single stranded, and these chemicals are single- strand specific, any potential secondary structure might limit the distribution of base changes. Again, even if certain regions of the VP16 activation domain are damaged less frequently than others, those regions which are damaged and result in the large colony phenotype are of interest. However, if there are regions which consistently escape damage, or if there is a particular region about which we obtain little information, a second mutagenesis method may be employed. One such method utilizes a series of degenerate oligonucleotides which together span the entire VP16 activation domain (I-Iill $31., 1986). This method allows a great deal of control over the frequency and location of mutations. The oligonucleotides are designed so that each has on average one mismatch with the template DNA. A second method would be to use a modified polymerase chain reaction, increasing the concentration of MnClz in the reaction and thereby increasing the misincorporation frequency of lag polymerase (Leung $31., 1989). Once the chemical mutagenesis had been performed and the chemicals removed, a primer was annealed downstream of the VP16 activation domains and a complementary DNA strand synthesized using avian myeloblastoma virus reverse transcriptase. This polymerase was used because it is not inhibited by depurinated template DNA (Myers $31., 1985). When reverse transcriptase encounters a damaged base in the template strand, it incorporates nucleotides essentially at random, resulting in a 75% chance of mutation at each damaged site (Sambrook $31., 1989). After second-strand synthesis, mutated VP16 activation domains were excised from mpJR92 and cloned into the plasmid prR92. The resulting 5 2 collection of GAL4-VP16-expressing plasmids was amplified in E,_c911 and then purified by cesium chloride ultracentrifugation before transformation into yeast. We added this step to the published protocol (Myers, 1989) because initially, when the ligation reactions were used to transform yeast directly, many of the resulting transformants exhibited the large colony phenotype but did contain plasmids of the correct size. We deduced that partial (linear) ligation products transformed directly into yeast underwent recombination by the recipient cell, which kept only the selectable marker. Amplification in E, 9911 avoided this problem, and the majority of large yeast colonies contained plasmids with the activation domain. The amplified vectors were then transformed into the yeast strain BP1, which lacks endogenous GAL4. The majority of the transformants gave tiny colonies, due to toxicity by fully active GAL4-VP16. The few larger colonies were picked and re-streaked onto fresh selective medium. Yeast colonies that grew when re-streaked were grown in selective liquid medium, and plasmid was harvested. The resulting plasmids were electroporated into 1.19911 and then plated out onto medium containing ampicillin. Bacterial transformants were picked and grown, and the sizes of the VP16 activation domains were determined by restriction digestion and agarose gel electrophoresis. Correct sized activation domains were recovered from the gel, again cloned into prR92, and transformed into BPI to see if the large-colony phenotype persisted. This re-cloning and re-transformation was done to ensure that the large-colony phenotype was due to mutation in the VP16 activation domain and not elsewhere in the expression plasmid or in the host cell. If the phenotype persisted, the particular yeast expression vector was purified and the DNA fragment encoding the mutated VP16 activation domain sequenced to determine the location and character of the mutation. 5 3 Chapter III Methods Yeast plasmid construction For 11111319 toxicity assays, wild-type or mutant GAL4-VP16 fusion proteins were expressed from the yeast plasmid prR92. This plasmid was derived from the plasmid pDBZOLBng (Berger $31., 1992), which is a 2 micron, Ampr plasmid with a unique BglII restriction site between the alcohol dehydrogenase (ADH) promoter and terminator; it also contains the LEUZ gene, which encodes B-iSOpropyl malate dehydrogenase, the third enzyme in the leucine biosynthetic pathway. The unique S3£I restriction site of pDBZOLBg1H was destroyed by restriction digestion, filling in, and re- ligation, generating plasmid pDB20.1. The plasmid pJR3.B58 (Berger $31., 1992) served as the source of the GAL4-VP16 gene. This plasmid is derived from pEMBL 19+ (Triezenberg $ 31., 1988a) and has the gene for GAL4 (aa 1-147)-VP16 (aa 413-490), plus 400 bp of HSV-1 DNA from the 1k gene 3' untranslated region, cloned into the $9111 site. To facilitate the cloning of the mutated VP16 activation domains into the yeast expression vector prR92, I introduced a unique 13ng site at the 3' end of the GAL4-VP16 gene by the following strategy. pJR3.BS8 was partially digested with fiamI-H, which cuts at both the 5' and 3' ends of the GAL4-VP16 gene. Linear DNA fragments from the BamI-II partial digest were gel purified and ligated with kinased fing linkers; ligation products were digested with Bng and then recircularized. The recircularized DNAs were transformed into E4911, and the colonies which resulted were grown and plasmid harvested. To screen for plasmids that contained the Bng linker, the plasmids were digested with Bng. Plasmids containing a Bg1H site were then digested with 33111111 and £131 to determine the location of the Bng site. A BamI-H/Bglfl fragment in the orientation of B3mHI-GAL4-VP16 gene-BglII 5 4 was cloned into the Bg1II site of pDB20.1; clones containing the GAL4-VP16 gene in the correct orientation relative to the A1211]. promoter were designated prR92. To generate a single-stranded substrate for chemical mutagenesis, the VP16 activation domain (aa 412-490, plus 7 nucleotides downstream of the stop codon) was cloned as a 5311/ BamI-H fragment into 5311/ BamHI-cut M13mp19, generating mpJR91. A Bng site was introduced adjacent to the BamI-II site by digesting mpJR91 with BamI-II and BglI and cloning in a BamI-H/Bgll fragment from pEMBL 19 / 13ng (S. J.Triezenberg, personal communication), thus generating mpJR92. Chemical mutagenesis of single-stranded DNA Chemical mutagenesis was performed using procedures described by Myers (1989). Forty pl of 1 ug/ul single-stranded mpJR92 was treated either with formic acid (60 ul of 18 M formic acid, 10 minutes at room temperature) or nitrous acid (10 ul of 2.5 M sodium acetate, pH 4.3, and 50 ul of 2 M sodium nitrite, 60 minutes at room temperature). After treatment with the chemicals, to each reaction was added 100 ul of 2.5 M sodium acetate, pH 5.5, 200 ill of water, 3 ul of 10 mg/ ml carrier tRNA, and 1 ml of 100% ethanol. Samples were chilled in a dry ice-ethanol bath, and then spun for 10 minutes in a microcentrifuge. The ethanol wash/ precipitation was repeated twice; the chemically treated DNAs were each resuspended in 80 111 TE. Second-strand synthesis and cloning To each sample of chemically treated DNA was added 10 ul of 10X reverse transcription buffer (0.5 M Tris-Cl, pH 8.2, 50 mM MgC12, 50 mM DIT, 0.5 M KCl, 0.5 mg/ ml BSA) and 100 pmole of the -20 universal sequencing primer for M13mp19 (New England Biolabs, product #1221). Samples were heated for 5 minutes at 85°C, then 15 minutes at 40°C. To each sample was 5 5 added 10 ul of 4dNTP mix (2.5 mM each dATP, dCTP, dGTP, and dIT'P) and 40 units of AMV reverse transcriptase (Life Sciences, Inc.); the samples were incubated for 1 hour at 37°C. The extension reaction was stopped by adding 11 pl of 3 M sodium acetate, pH 6, and extracting once with phenol / chloroform, and the DNA was precipitated with ethanol. The DNA was resuspended in TE and then digested with $391 and Bg1H. The mutated activation domains (250 bp) were purified by agarose gel electrophoresis, and the DNA recovered from the agarose using DNA-affinity resin (Qiagen). Gel-purified mutated activation domains were ligated into 5391/ Bng-cut prR92, and then transformed into competent H8101 E,_£Ql1 cells by electroporation (Bio-Rad Gene Pulser, with Bio-Rad 0.2 cm cuvettes). Electroporated H8101 cultures were grown for 1 hour in 1 ml of SOC broth at 37°C and then diluted to 50 ml with LB plus ampicillin. The cultures were grown overnight at 37°C with shaking; the following day plasmid DNA was harvested by alkaline lysis followed by CsCl gradient ultracentrifugation. Transformation of yeast Yeast strain BPI (MAIa, 11:33:52, 1£u2;3, 3931:1911, GAL4::I-IIS4) was transformed using the lithium acetate procedure (Becker and Guarente, 1991). A 100 ml culture of BPl was grown at 30°C with shaking to A600 = 0.8. The cells were pelleted and then resuspended in a total of 5 ml of TE buffer (10 mM Tris-Cl, pH 7.5, 1 mM NazEDTA, pH 8.0). The cells were pelleted as before and then resuspended in 5 m1 of TE / 0.1 M lithium acetate. The cells were again pelleted and resuspended in 1 ml of TE/ 0.1 M lithium acetate. The cells were incubated at 30°C with shaking for 1 hour. After incubation, 100 ill of 4 mg/ ml salmon sperm DNA (in TE) was added. Each transformation reaction consisted of 100 pl of competent BPI cells and 2 ug of yeast plasmid DNA. Samples were incubated for 30 minutes at 30°C without 5 6 shaking. After the incubation, 700 pl of 35% (w/v) PEG (MW 3350)/ TE/ 0.1 M lithium acetate was added to each sample, and the samples were mixed by vortexing. The samples were incubated for 50 minutes at 30°C without shaking, and then heat-shocked for 5 minutes at 42°C. After the heat-shock, the cells were pelleted with a 4 second spin in a microcentrifuge. Pelleted cells were resuspended in 500 111 TE; this step was repeated once. Finally, cells were pelleted, resuspended in 100 111 TE, and spread onto Leu' synthetic complete medium plates to select for transformants. The plates were incubated for 2-3 days at 30°C until colonies appeared. Recovery of plasmid DNA from yeast Large yeast colonies were used to inoculate 2 ml of Leu" synthetic complete liquid medium and grown overnight at 30°C with shaking. The following day, 1.5 ml of each yeast culture was briefly spun to pellet the cells, which were resuspended in 200 ill of yeast lysis buffer (10 mM Tris-Cl, pH 8, 1 mM EDTA, 100 mM NaCl, 0.1% SDS). Glass beads (0.5 mm diameter, Biospec Products), which had been etched with 1 M HCl, were added to the yeast solution to the level of the meniscus. Two hundred 111 of phenol/ chloroform (1:1 ; TE saturated) was added to each sample; the samples were vortexed for one minute and then spun to separate the aqueous and organic phases. The aqueous layer from each sample was transferred to a fresh 1.5 ml microfuge tube, 15 ill of 3 M sodium acetate and two volumes of 100% ethanol was added, and the DNA was recovered. B-galactosidase assay prR92 plasmids containing mutations in the VP16 activation domain were digested with Smal and the 650 bp Small $11131 fragment was purified by agarose gel electrophoresis and DNA-affinity resin (Qiagen). The recovered fragments were ligated to the 8000 bp $11131 / $11131 fragment of 5 7 pSB202delSmaI, an ARS/CEN (low copy) yeast vector expressing GAL4- VP16delSma from the A2111 promoter. The ligation products were screened for the presence of mutant delSma activation domains by digestion with B_g111, the restriction site for which is present in mutant delSm activation domains and absent in wildtype delSma activation domains. The orientations of mutant delSma activation domains were determined by DNA sequencing. Yeast strain BPI was cotransformed with 2 ug each of pSB202delSmaI containing a mutated VP16 activation domain (amino acids 453-490) and the reporter plasmid pLGSD5 (Guarente et al., 1982). pSBZOZdelSmaI contains the LEUZ gene, and pLGSD5 contains the 1.1%,} gene which encodes orotidine 5'- phosphate decarboxylase, an enzyme required for uracil biosynthesis. The B- galactosidase assays were performed according to Rose $31. (1990). Five ml cultures of cotransformants were grown overnight under selective conditions to A600 = 0.8-1.0. The cells were pelleted and the pellets were each resuspended in 250 pl of breaking buffer (100 mM Tris-Cl, pH 7.5, 1 mM DTT, 20% glycerol). Etched glass beads (0.5 mm) were added to the level of the meniscus; then to each sample was added 12.5 111 of 40 mM PMSF (in 100% isopropanol; Boehringer Mannheim). The samples were vortexed six times each at top speed in 15 second bursts, chilling on ice between bursts. Another 250 pl of breaking buffer was added to each sample, and the liquid extracts removed and transferred to fresh 1.5 ml microfuge tubes. The samples were centrifuged for 15 minutes in a horizontal microcentrifuge to clarify; 100 pl of each sample lysate was added to 900 pl of Z buffer (60 mM Nazi-IP04, 40 mM NaHzPO4, 10 mM KCl, 1 mM MgSO4, 30 mM 2-mercaptoethanol; solution pH 7). The samples were incubated at 30°C for 5 minutes. After incubation, to each tube was added 200 pl of 4 mg/ ml 9-nitrophenyl-B-D-galactoside (ONPG; S 8 Sigma), in 2 buffer. Reactions were incubated at 30°C until samples were moderately yellow (usually less than 5 minutes), and then stopped by addition of 500 ill of 1 M NazCO3; the length of reaction time was recorded. The A420 of the samples was read. The total protein concentration of each sample was determined using the Bradford assay and absorbance at 595 nm. The specific activities were calculated by the following formula: Activity = (A420)(1.7)/ [(0.0045)(protein concentration)(extract volume)(time)]. The factor 1.7 corrects for the reaction volume, the factor 0.0045 is the Optical density of a 1 nmole/ ml solution of 9-nitrophenol, the protein concentration is in mg/ ml, the extract volume is in ml, and the time is in minutes (Rose £1 al., 1990). Results The initial attempt at chemical mutagenesis involved formic acid. The mutagenesis with formic acid was performed as described in the Methods section for this chapter. After cloning of the population of mutated VP16 activation domains into the expression vector prR92 and transformation into yeast strain BPl, large yeast colonies were observed. These colonies were picked and restreaked onto fresh minimal medium plates. Those colonies which still exhibited better growth than colonies transformed with unmutated prR92 were grown up and their plasmids recovered. The VP16 activation domains from these large colony mutants were cloned into fresh prR92 and then transformed into yeast. Fifteen out of 18 activation domains that gave a large colony phenotype initially also gave the same phenotype after recloning into fresh expression vector. Because few mutations other than in the VP16 activation domain were detected, I omitted the recloning step from the subsequent experiments and sequenced the VP16 5 9 activation domains directly from clones exhibiting a large colony phenotype. The following is a summary of the mutations in the VP16 activation domain resulting from treatment with formic acid. Formic acid treatment resulted in one mutation per activation domain sequenced. The most prevalent mutation (observed in 13 large colonies) was in the natural VP16 stop codon, changing it to the codon for leucine (or, less frequently, tyrosine). This mutation resulted in the addition of 23 extra amino acids to the GAL4-VP16 protein before the next in-frame stop codon. The other mutations were either additions or deletions of bases. The addition of an extra cytosine in a run of six cytosines at VP16 codons 458-460 caused a frameth which created a new stop codon 15 codons later; this mutation occurred in seven large colonies. All the other frameshift mutations (a total of seven large colonies) were due to a deletion of single bases in the region of VP16 codons 451-463; in addition to the altered coding sequence, these base deletions resulted in at least 40 extra amino acids at the carboxyl end of the GAL4-VP16 protein. Since formic acid as a mutagen generated many base additions or deletions and few transitions or transversions, I switched to nitrous acid as the mutagen. The mutagenesis using nitrous acid and subsequent biological selection in yeast are described in the Methods section for this chapter. A conscious effort was made to pick yeast colonies that were larger than those transformed with wild type GAL4-VP16 but were not as large as those transformed with the expression vector only. The reasoning was that those colonies which grew as well as the vector only transformants likely had suffered some gross damage to the VP16 activation domain (due to a frameth mutation, for example), while we were interested in more subtle changes. Larger colonies were picked and then restreaked onto selective 6 0 media; those clones which continued to grow better than colonies transformed with wild type GAL4-VP16-expressing plasmids were analyzed further. Approximately 150 larger yeast colonies were picked and restreaked; greater than 80% of the clones continued to grow better than wild type GAL4- VP16 transformants. The sizes of the VP16 activation domains of 53 of these clones that grew when restreaked were determined by restriction digest and electrophoresis. Nineteen clones had activation domains of the correct size and were therefore sequenced. Of the nineteen VP16 activation domains sequenced, six had no apparent mutation in the activation domain; presumably the large colony phenotype was due to host mutations or mutations elsewhere in the expression plasmid. Two clones had mutations that did not alter the amino acid specificity, and two clones had frameshift mutations in the VP16 activation domain (the addition or deletion of a deoxyadenosine at codon 460). The remaining nine clones all had missense mutations, the locations of which are shown in Figure 11 and described below. Five yeast colonies (N93, N101, N109, N 111, and N151) contained identical mutations, a glycine to glutamic acid change at codon 448 and a threonine to alanine change at codon 480. The ambiguity in establishing which mutation caused the large colony phenotype was relieved by the isolation of another large colony mutant (N112) having only the TA480 mutation. Another yeast colony (N105) contained a VP16 activation domain with four amino acid substitutions, aspartic acid to glycine at codon 427, threonine to alanine at codon 458, aspartic acid to asparagine at codon 461, and methionine to valine at codon 478. Mutant N144 had two mutations, a 61 Figure 11. Summary of mutations resulting from treatment of the gene encoding the VP16 activation domain with nitrous acid. The amino acid sequence of the VP16 activation domain is shown. The location and type of mutation, along with the identifier (N for the mutagen nitrous acid, the number referring to the original larger yeast colony picked). 62 N32— viii Neg FTCF [H U mez— 532 :F2 092 2:2 F 82. < 00>w964OwOQJIHwOOAw>QHQa