The use of random-sequence oligonucleotides for determining consensus sequences

Creating protein biocatalysts as tools for future industrial applications

BACKGROUND

Biocatalysts provide an economical and energy-efficient alternative to traditional chemical manufacturing processes. For processes where biocatalysts currently do not exist or existing protein catalysts function poorly, there is a tremendous need to discover new protein catalysts that function in industrial settings. The protein engineering community has traditionally relied on cell-based techniques in 96-well format to evolve new catalysts or improve existing enzymes.

OBJECTIVE

This review examines recent progress made in many display technologies, providing powerful alternatives for generating novel enzymes with altered specificity or altogether new types of function.

METHODS

Library creation methods and display technologies that are commonly used in conjunction with enzyme evolution are discussed.

CONCLUSION

We conclude with an expert opinion on future trans-disciplinary approaches that combine directed evolution with computational design as novel platforms for rapidly discovering new types of catalytic function.

DNA-binding specificity of the CYS3 transcription factor of Neurospora crassa defined by binding-site selection

The CYS3 transcription factor is a basic region-leucine zipper (bZIP) DNA-binding protein that is essential for the expression of a coordinately regulated group of genes involved in the acquisition and utilization of sulfur in Neurospora crassa. An approach of using binding-site selection from random-sequence oligonucleotides was used to define CYS3-binding specificity. The derived consensus-binding site of ATGGCGCCAT defines a symmetrical sequence (half-site A T G/t G/a C/t) that resembles that of other bZIP proteins such as CREB and C/EBP. By comparison, CYS3 shows a greater range of binding to a central core of varied Pur-Pyr-Pur-Pyr sequences than CREB as determined by gel shift assays. The derived CYS3 consensus binding sequence was further validated by demonstrating in vivo sulfur regulation using a heterologous promoter construct. The CYS3-binding site data will be useful for the genome-wide study of sulfur-regulated genes in N. crassa, which has served as a model fungal sulfur control system.

Expanding the specificity of DNA targeting by harnessing cooperative assembly

The precise control of developmental and regulatory processes in the cell requires accurate recognition of specific DNA sites. For genomes as large as that of humans, single-molecule-DNA binders have difficulties accurately and specifically recognizing the intended targets. Natural transcription factors overcome these difficulties by forming non-covalent complexes on the DNA with other transcription factors. These cooperative complexes overcome the difficulties of single-molecule transcription factors, allowing specific, combinatorial control of a range of transcriptional targets. Artificial transcription factors have been designed to take advantage of this technique of cooperative assembly, facilitating future studies in whole genome targeting. In contrast to a simple model of component independence, cooperative complexes as a whole often display slightly altered DNA specificity from what would be expected from the analysis of their separate components. The true sequence specificity of cooperative complexes, and thus their presumed in vivo targets, have to be experimentally probed. A number of techniques, such as the cognate site identity array, now allow for rapid, high-throughput determination of the specificity of cooperative complexes.

On the generation of information as motive power for molecular evolution

Molecular evolution can be described as a learning process during which previously inanimate matter developed the ability to organize all the reaction pathways that establish a living system. Common to all natural self-organizing procedures is the ability of matter to store, process and evaluate the information achieved by learning. Genetic information which is stored in RNA or DNA is the object of natural evolution. With the recognition of nature's concepts, evolutionary optimization was applied to biopolymers that are not optimally adapted for particular technical or medical purposes. Information can also be stored in molecules with structures and chemical properties that are completely different from nucleic acids. Therefore, optimization processes that mimic the natural evolutionary strategies can also be applied to small organic molecules. Much effort has been made theoretically and practically to find a certain optimized species within the (hyper)astronomical number of possible sequence alternatives. From a series of computer experiments it can be concluded that it is not necessary to search the entire sequence space in order to find a particular structure; this is advantageous because the diversity of mutant libraries that can realistically be achieved in the laboratory never extends to the number of theoretically possible sequences. Molecular mutant libraries that serve as starting populations for in vitro selection have been constructed for nucleic acids, proteins, peptides and small organic molecules.

[Aptamers: selection and scope of applications]

Aptamers are short oligonucleotides selected from large combinatorial pools of sequences for their capacity to bind to many different targets ranging from small molecules (amino acids, antibiotics...) to proteins or nucleic acid structures. Aptamers present the same high specificity and affinity for their targets as antibodies. In addition to efficient binding, aptamers have been shown in many cases to display an inhibitory activity against their targets. Many aptamers are now being developed against biomedical relevant targets, and one aptamer that inhibits the human VEGF165 already received approval for the treatment of age-related macular degeneration. Here we discuss the principles and the practical way of selecting aptamers (SELEX technology) as well as the structural basis for their performance as ligands. A wide scope of applications is described - aptamers have been used as tools for studying nucleic acids/proteins interactions, detecting, purifying or imaging target molecules, regulating gene expression - and includes recent developments of aptamers for therapy and diagnosis.

Some aspects of E. coli promoter evolution observed in a molecular evolution experiment

We devised a molecular evolution procedure to evolve E. coli promoter sequences and applied it to observe an arbitrary, nonfunctional sequence evolving into functional promoters. In the experiments, DNA sequence variations were generated with error-prone PCR and were inserted in the promoter region of the cat (chloramphenicol acetyl transferase) gene on a plasmid. Upon transforming the cells, functional promoters on the plasmid were selected according to the chloramphenicol resistance. Within a few cycles of mutation-selection, promoters emerged, and the sequences converged into a small number of groups. In the process, the extended minus 10 type of promoters emerged quickly, and small deletions were often involved in adjusting the length between the -35 and the -10 elements. Our results also suggest a possible selection for promoter stability against mutation.

Sequence-specific recognition of DNA by hydrophobic, alanine-rich mutants of the basic region/leucine zipper motif investigated by fluorescence anisotropy

We generated minimalist proteins capable of sequence-specific, high-affinity binding of DNA to probe how proteins are used and can be used to recognize DNA. In order to quantify binding affinities and specificities in our protein-DNA system, we used fluorescence anisotropy to measure in situ the thermodynamics of binding of alanine-rich mutants of the GCN4 basic region/leucine zipper (bZIP) domain to DNA duplexes containing target sites AP-1 (5'-TGACTCA-3') or ATF/CREB (5'-TGACGTCA-3'). We simplified the alpha-helical bZIP molecular recognition scaffold by alanine substitution: 4A, 11A, and 18A contain four, eleven, and eighteen alanine mutations in their DNA-binding basic regions, respectively. DNase I footprinting analysis demonstrates that all bZIP mutants retain the sequence-specific DNA-binding function of native GCN4 bZIP. Titration of fluorescein-labeled oligonucleotide duplexes with increasing amounts of protein yielded low nanomolar dissociation constants for all bZIP mutants in complex with the AP-1 and ATF/CREB sites: binding to the nonspecific control duplex was > 1000-fold weaker. Remarkably, the most heavily mutated protein 18A, containing 24 alanines in its 27-residue basic region, still binds AP-1 and ATF/CREB with dissociation constants of 15 and 7.8 nM, respectively. Similarly, wild-type bZIP binds these sites with K(d) values of 9.1 and 14 nM. 11A also displays low nanomolar dissociation constants for AP-1 and ATF/CREB, while 4A binds these sites with approximately 10-fold weaker K(d) values. Thus, both DNA-binding specificity and affinity are maintained in all our bZIP derivatives. This Ala-rich scaffold may be useful in design and synthesis of small alpha-helical proteins with desired DNA-recognition properties capable of serving as therapeutics targeting transcription.

Analysis of promoter recognition in vivo directed by sigma(F) of Bacillus subtilis by using random-sequence oligonucleotides

Formation of spores from vegetative bacteria by Bacillus subtilis is a primitive system of cell differentiation. Critical to spore formation is the action of a series of sporulation-specific RNA polymerase sigma factors. Of these, sigma(F) is the first to become active. Few genes have been identified that are transcribed by RNA polymerase containing sigma(F) (E-sigma(F)), and only two genes of known function are exclusively under the control of E-sigma(F), spoIIR and spoIIQ. In order to investigate the features of promoters that are recognized by E-sigma(F), we studied the effects of randomizing sequences for the -10 and -35 regions of the promoter for spoIIQ. The randomized promoter regions were cloned in front of a promoterless copy of lacZ in a vector designed for insertion by double crossover of single copies of the promoter-lacZ fusions into the amyE region of the B. subtilis chromosome. This system made it possible to test for transcription of lacZ by E-sigma(F) in vivo. The results indicate a weak sigma(F)-specific -10 consensus, GG/tNNANNNT, of which the ANNNT portion is common to all sporulation-associated sigma factors, as well as to sigma(A). There was a rather stronger -35 consensus, GTATA/T, of which GNATA is also recognized by other sporulation-associated sigma factors. The looseness of the sigma(F) promoter requirement contrasts with the strict requirement for sigma(A)-directed promoters of B. subtilis. It suggests that additional, unknown, parameters may help determine the specificity of promoter recognition by E-sigma(F) in vivo.

Short, hydrophobic, alanine-based proteins based on the basic region/leucine zipper protein motif: overcoming inclusion body formation and protein aggregation during overexpression, purification, and renaturation

GCN4 is a yeast transcriptional regulatory protein; its DNA-binding domain is a basic region/leucine zipper (bZIP) structure that comprises a dimer of alpha-helices capable of high-affinity, sequence-specific recognition of the DNA major groove. We are exploiting what nature has evolved by manipulating the bZIP motif as a molecular recognition scaffold; thus we reduced the elegantly minimal bZIP to an even more simplified structure by substitution with alanine residues-hence, a generic, Ala-based, helical scaffold. These Ala-based mutants are unusual proteins for expression as they are short ( approximately 100 amino acids) and hydrophobic (Ala-mutated basic regions, leucine-zipper dimerization domains). Hydrophobicity posed a major problem throughout the expression, isolation, and purification stages; inclusion body formation and protein aggregation were significant hurdles throughout protein production. We describe measures that solved these problems, including use of high concentrations of denaturant in all steps of protein isolation and purification and use of temperature-dependent renaturing techniques to obtain folded, functional protein. Despite these difficulties, we ultimately retrieved 5-10 mg/L of broth of active, correctly folded protein after the complete purification procedure. Homogeneity of the proteins was established by chromatography, electrophoresis, and mass spectrometry. Furthermore, characterization by circular dichroism and DNase footprinting analysis demonstrates that these alanine-based mutants retain the structure and function of the native GCN4 DNA-binding domain. Remarkably, the most heavily mutated protein, containing 24 alanines of 27 total amino acids in the DNA-binding basic region, still binds the AP-1 site, the target of native GCN4.

Sequential determination of ligands binding to discrete components in heterogeneous mixtures by iterative panning and blocking (IPAB)

Biopanning has been used extensively in conjunction with purified components, but there are also examples in which mixtures of targets have been investigated. This study introduces a methodological innovation, termed iterative panning and blocking (IPAB), to extend the range of specific interactions that can be probed in mixtures. Here this procedure is used to probe a mixture of high molecular mass components of human cord blood with phage-peptide display libraries. The initial panning recovered phage that bore the consensus motif Gly-Pro-Arg-Pro, a known fibrinogen-binding motif. These phage bound specifically to purified fibrinogen. A series of peptides containing the Gly-Pro-Arg-Pro motif efficiently blocked the binding of phage having the same motif, presumably by binding to their common target. A second round of panning was performed against the same target mixture in the presence of this blocking peptide. Phage recovered from this second panning exhibited a motif (Ser-His-Tyr) that was subsequently shown to bind specifically to complement component C1q. A second peptide containing this motif specifically blocked the interaction of the phage with C1q. A third round of panning performed in the presence of both the fibrinogen- and the C1q- blocking peptides yielded phage with a new peptide motif (Asn-Pro-Phe) that also bound specifically to C1q, apparently at a new site. The three motifs isolated through this iterative process were distinct in that each was blocked only by its corresponding peptide. This IPAB strategy can be applied to many high diversity selection procedures that target complex mixtures.

Characterization of three related glucose repressors and genes they regulate in Saccharomyces cerevisiae

Mig1 and Mig2 are proteins with similar zinc fingers that are required for glucose repression of SUC2 expression. Mig1, but not Mig2, is required for repression of some other glucose-repressed genes, including the GAL genes. A second homolog of Mig1, Yer028, appears to be a glucose-dependent transcriptional repressor that binds to the Mig1-binding sites in the SUC2 promoter, but is not involved in glucose repression of SUC2 expression. Despite their functional redundancy, we found several significant differences between Mig1 and Mig2: (1) in the absence of glucose, Mig1, but not Mig2, is inactivated by the Snf1 protein kinase; (2) nuclear localization of Mig1, but not Mig2, is regulated by glucose; (3) expression of MIG1, but not MIG2, is repressed by glucose; and (4) Mig1 and Mig2 bind to similar sites but with different relative affinities. By two approaches, we have identified many genes regulated by Mig1 and Mig2, and confirmed a role for Mig1 and Mig2 in repression of several of them. We found no genes repressed by Yer028. Also, we identified no genes repressed by only Mig1 or Mig2. Thus, Mig1 and Mig2 are redundant glucose repressors of many genes.

In vivo nickel insertion into the carbon monoxide dehydrogenase of Rhodospirillum rubrum: molecular and physiological characterization of cooCTJ

The products of cooCTJ are involved in normal in vivo Ni insertion into the carbon monoxide dehydrogenase (CODH) of Rhodospirillum rubrum. Located on a 1.5-kb DNA segment immediately downstream of the CODH structural gene (cooS), two of the genes encode proteins that bear motifs reminiscent of other (urease and hydrogenase) Ni-insertion systems: a nucleoside triphosphate-binding motif near the N terminus of CooC and a run of 15 histidine residues regularly spaced over the last 30 amino acids of the C terminus of CooJ. A Gm(r)omega-linker cassette was developed to create both polar and nonpolar (60 bp) insertions in the cooCTJ region, and these, along with several deletions, were introduced into R. rubrum by homologous recombination. Analysis of the exogenous Ni levels required to sustain CO-dependent growth of the R. rubrum mutants demonstrated different phenotypes: whereas the wild-type strain and a mutant bearing a partial cooJ deletion (of the region encoding the histidine-rich segment) grew at 0.5 microM Ni supplementation, strains bearing Gm(r)omega-linker cassettes in cooT and cooJ required approximately 50-fold-higher Ni levels and all cooC insertion strains, bearing polar or nonpolar insertions, grew optimally at 550 microM Ni.

Mapping sequence specific DNA-protein interactions: a versatile, quantitative method and its application to transcription factor XF1

We have developed a method for the quantitative, exhaustive sequence specificity determination of DNA-binding proteins. The QuESSD method overcomes the limitations inherent in other published in vitro selection methods, not only defining the consensus sequence, but also quantifying the effect on DNA-protein affinity of replacing each base in the recognition domain with every other base. The features distinguishing this method from other in vitro selection approaches are: (1) instead of synthesizing one target oligonucleotide population containing a long randomized domain, we synthesize several oligonucleotide populations, each randomized at two positions. (2) Instead of carrying out several cycles of selection and amplification, we carry out a single cycle. (3) We have developed data collection and analysis procedures that eliminate artifacts and allow generation of quantitative results. The QuESSD method yields accurate measures of: (a) the selectivity of the protein for each base at each position within the recognition domain (normalized relative selectivity), (b) the contributions of individual sites within the recognition domain to the binding affinity (selectivity variance), (c) the relative binding affinity of any given sequence (global selectivity). We confirmed results by (1) tabulating directly the frequency of appearance of individual species in the pool of protein-bound oligonucleotides by cloning and sequencing individual oligonucleotides, and (2) competition EMSA analysis of oligonucleotides designed on the basis of QuESSD data. We have used this method to map the sequence specificity of the nuclear protein XF1 and to distinguish the sequence specificities of XF1 and the AH receptor complex, both of which bind to XRE1, a xenobiotic responsive element (XRE) located upstream of the CYP1A1 gene. Using data obtained by the QuESSD method, we designed oligonucleotides specific for XF1 or for the AH receptor, and prepared CAT reporter gene constructs carrying these oligonucleotides, or wild-type XRE1, upstream of a minimal promoter. Transfection studies using these constructs indicated that XF1 can function as a weak activator of basal transcription, and can, under some circumstances, compete with the AH receptor for binding to XRE1.

Identification of GIYWHHY as a novel peptide substrate for human p60c-src protein tyrosine kinase

We have recently determined that -Ile-Tyr- were the two critical residues as a peptide substrate for p60c-src protein tyrosine kinase (Lou, Q. et al., Lett. Peptide Sci., 1995, 2, 289). Here, we report on the design and synthesis of a secondary 'one-bead, one-compound' combinatorial peptide library based on this dipeptide motif (XIYXXXX, where X = all 19 eukaryotic amino acids except for cysteine). This secondary library was screened for its ability to be phosphorylated by p60c-src PTK using [gamma 32P]ATP as a tracer. Five of the strongest [32P]-labeled peptide-beads were identified and microsequenced: GIYWHHY, KIYDDYE, EIYEENG, EIYEEYE, and YIYEEED. A solid-phase phosphorylation assay was used to evaluate the structure-activity relationship of GIYWHHY. It was determined that Ile2, Tyr3, His5, and His6 were crucial for its activity as a substrate.

Hamming chromatography

Selection of molecules with desired properties from random pools of biopolymers has become a powerful tool in biotechnology. On designing an evolution experiment, a certain knowledge of the concomitant fitness landscape is clearly helpful to set up the optimal experimental conditions. The correlation function is a useful means of characterizing a given landscape, since it can be efficiently measured if one has a method of separating a pool of random sequences according to their Hamming distance from a moderately small number of test sequences. In this paper we describe a special type of hybridization chromatography, where a mixture of oligomers (partially) complementary to a given test sequence is hybridized to the test sequence, covalently bound to a matrix. DNA oligomers are eluted in an 'effective temperature gradient' using conditions that minimize the differences of effects of GC versus AT pairs on the melting temperatures. This method should be a means to quickly separate error classes and thus be the crucial step in characterizing fitness landscapes of biopolymers through an experimental approach. It would also be a useful tool to design sequence pools with a bias towards desired mutant spectra.

Exploring molecular diversity with combinatorial shape libraries

Surface technologies based upon selection of ligands from combinatorial libraries herald a revolution in molecular research and drug discovery. Molecular diversity is generated by random combinations of monomeric building blocks to form polymeric conformers that constitute 'shape libraries'. The media for exploring surfaces of target molecules include synthetic or biological polymers consisting of natural or modified amino acids, nucleotides, carbohydrates and other organic materials. Targets can be any biological surface, including enzymes, antibodies, receptors and other regulatory molecules. The power of combinatorial selection is in finding conceptual leads for designing high-affinity ligands and effector molecules for the analysis and manipulation of biochemical interactions.

Dimerization of leucine zippers analyzed by random selection

The leucine zipper is a coiled coil that mediates specific dimerization of bZIP DNA-binding domains. A hydrophobic spine involving the conserved leucines runs down the coiled-coil and is thought to stabilize the dimer. We used the method of random selection to further define the primary sequence requirements for homodimer formation and heterodimer formation with Fos. When positions on either side of the hydrophobic spine of GCN4 are diversified to include the corresponding residues of Jun, a large percentage of the resulting sequences form homodimers, and a large percentage form heterodimers with Fos. Basic residues were preferred, but not essential, at position e of zippers which heterodimerize with Fos. When random sequences containing 5 heptad repeat of leucines are subject to a selection for homodimer formation, a diverse set of sequences is isolated. Certain residues are preferred at each position in the heptad repeat, although no essential primary sequence determinants could be identified. No pair of residues not involving the conserved leucines could be identified which strongly promotes homodimerization. These results suggest that factors determining leucine zipper dimerization are complex, with numerous interactions contributing to the association.

CASTing for multicomponent DNA-binding complexes

Degenerate oligonucleotides and polymerase chain reaction-based reiterative selection techniques have been used to define the consensus binding sites for an increasing number of transcription factors. The use of crude nuclear extracts rather than purified proteins permits multicomponent complexes to form, and allows the technique to generate information about the combinatorial interactions involved in gene regulation.

Functional roles of amino acid residues involved in forming the alpha-helix-turn-alpha-helix operator DNA binding motif of Tet repressor from Tn10

The Tn10 derived Tet repressor contains an amino acid segment with high homology to the alpha-helix-turn-alpha-helix motif (HTH) of other DNA binding proteins. The five most conserved amino acids in HTH are probably involved in structural formation of the motif. Their functional role was probed by saturation mutagenesis yielding 95 single amino acid replacement mutants of Tet repressor. Their binding efficiencies to tet operator were quantitatively determined in vivo. All functional mutants contain amino acid substitutions consistent with their proposed role in a HTH. In particular, only the two smallest amino acids (serine, glycine) can substitute a conserved alanine in the proposed first alpha-helix without loss of activity. The last position of the first alpha-helix, the second position in the turn, and the fourth position in the second alpha-helix require mostly hydrophobic residues. The proposed C-terminus of the first alpha-helix is supported by a more active asparagine compared to glutamine replacement mutant of the wt leucine residue. The turn is located close to the protein surface as indicated by functional lysine and arginine replacements for valine. A glycine residue at the first position in the turn can be replaced by any amino acid yielding mutants with at least residual tet operator affinity. A structural model of the HTH of Tet repressor is presented.

Parameters affecting transcription termination by Escherichia coli RNA. II. Construction and analysis of hybrid terminators

Rho-independent terminators are characterized by two major functional regions, one upstream from the termination site having a sequence capable of forming an RNA hairpin in the nascent transcript, the second extending, from the base of this hairpin, seven to nine nucleotides along the transcript to the actual sites of termination (3'-tail region). This latter region of the transcript is often rich in uridine residues. Both regions are postulated to play central roles in the termination process. We have constructed a series of hybrid rho-independent, transcription terminators in which sequences upstream and downstream from the RNA hairpin for the Escherichia coli trp attenuator (trpatt+) are interchanged with sequences from trpatt mutant (1419) or from the phage T7 early terminator (T7Te). Similar hybrids have been constructed for T7Te, replacing flanking sequences with trpatt regions. The effects of such changes on transcription termination have been tested in vitro with purified E. coli RNA polymerase to determine the intrinsic termination efficiency (%T) of each hybrid terminator. Both the trpatt+ terminator and T7Te are highly efficient rho-independent terminators in vitro. However, replacement of trpatt+ sequences upstream and downstream from the RNA-terminator hairpin with the comparable T7Te sequences reduces %T dramatically, suggesting that the RNA-terminator hairpin does not function independently from its flanking regions. Regions downstream from the actual termination/release site are shown to be of considerable importance in determining %T for terminators bearing the T7Te or trpatt1419 3'-tail region, but have little effect on terminators with the trpatt+ 3'-tail region. For terminators bearing the T7Te or trpatt1419 3'-tail region that are inefficient, efficient termination is restored by elevated concentrations of KCl in the reaction. The results do not fit well with models for termination in which %T is determined by a two-step process in which the terminator-RNA hairpin, and a seven to 12 base-pair DNA-RNA hybrid structure rich in uridine residues, act independently to cause the polymerase to pause, and to release the transcript, respectively. DNA sequences both upstream and downstream from these regions, as well as DNA sequences downstream from the transcript termination site, can significantly affect the termination process. Conversely, terminators lacking a 3'-tail region rich in uridine residues can be highly efficient, but only when joined with appropriate sequence immediately downstream from the termination site. This suggests that the 3'-tail region acts in some manner other than the formation of an unstable DNA-RNA hybrid that facilitates termination.

In vitro genetics

Understanding the basis of specificity in an intermolecular interaction is a common if difficult task; designing a specific intermolecular interaction is much more challenging. A new technique is described that has applications to both problems, at least with regard to nucleic acids. The power of this method lies in its ability to isolate extremely rare sequences with precisely specified properties from very large pools of random sequences.

Specificity of Escherichia coli endoribonuclease RNase E: in vivo and in vitro analysis of mutants in a bacteriophage T4 mRNA processing site

Endoribonuclease RNase E has an important role in the processing and degradation of bacteriophage T4 and Escherichia coli mRNAs. We have undertaken a mutational analysis of the -71 RNase E processing site of T4 gene 32. A series of mutations were introduced into a synthetic T4 sequence cloned on a plasmid, and their effects on processing were analyzed in vivo. The same mutations were transferred into T4 by homologous recombination. In both the plasmid and the phage contexts the processing of the transcripts was similarly affected by the mutations. Partially purified RNase E has also been used to ascertain the effect of these mutations on RNase E processing in vitro. The hierarchy of the efficiency of processing of the various mutant transcripts was the same in vivo and in vitro. These results and an analysis of all of the known putative RNase E sites suggest a consensus sequence RAUUW (R = A or G; W = A or U) at the cleavage site. Modifications of the stem-loop structure downstream of the -71 site indicate that a secondary structure is required for RNase E processing. Processing by RNase E was apparently inhibited by sequences that sequester the site in secondary structure.

SELEXION. Systematic evolution of ligands by exponential enrichment with integrated optimization by non-linear analysis

Recently, novel technologies for isolation of nucleic acid molecules with specific biological activities have been reported. In each case, the enrichment process involves repeated rounds of selection from complex mixtures of nucleic acid sequences, followed by polymerase chain reaction (PCR) amplification of ligand sequences that function in the desired manner. Particular variations in experimental conditions can dramatically alter the outcome of these processes. In this study, we use mathematical analysis and computer simulation to predict which variations have the greatest impact and to develop strategies and guidelines for enhanced effectiveness. First, we perform reconstruction tests to demonstrate that a mathematical description based on equilibrium binding is sufficient to explain the high levels of enrichment attained in the laboratory after just a few rounds. Then, we show the expected enrichment for an extensive range of conditions; and, finally, we determine the optimum protein and nucleic acid concentrations to use for maximum enrichment, while also ensuring a high likelihood of recovering even the rare molecule that binds well. The strategies and guidelines for enhanced effectiveness are generally applicable to processes for systematic enrichment of DNA, RNA or peptide ligands and have been implemented in an interactive simulation program for integrated non-linear optimization of enrichment using any target of interest.

Cyclic amplification and selection of targets (CASTing) for the myogenin consensus binding site

The consensus binding site for the muscle regulatory factor myogenin was determined from an unbiased set of degenerate oligonucleotides using CASTing (cyclic amplification and selection of targets). Stretches of totally random sequence flanked by polymerase chain reaction priming sequences were mixed with purified myogenin or myotube nuclear extracts, DNA-protein complexes were immunoprecipitated with an antimyogenin antibody, and the DNA was amplified by polymerase chain reaction. Specific binding was obtained after four to six cycles of CASTing. The population of selected binding sites was then cloned, and a consensus was determined from sequencing individual isolates. Starting from a pool with 14 random bases, purified myogenin yielded a consensus binding site of AACAG[T/C]TGTT, while nuclear extracts retrieved the sequence TTGCACCTGTTNNTT from a pool containing 35 random bases. The latter sequence is consistent with that predicted from combining an E12/E47 half-site (N[not T]CAC) with the purified myogenin half-site ([T/C] TGTT). The presence of paired E boxes in many of the sequences isolated following CASTing with nuclear extracts proves that myogenin can bind cooperatively with other E-box-binding factors.

Cyclic amplification and selection of targets (CASTing) for the myogenin consensus binding site

The consensus binding site for the muscle regulatory factor myogenin was determined from an unbiased set of degenerate oligonucleotides using CASTing (cyclic amplification and selection of targets). Stretches of totally random sequence flanked by polymerase chain reaction priming sequences were mixed with purified myogenin or myotube nuclear extracts, DNA-protein complexes were immunoprecipitated with an antimyogenin antibody, and the DNA was amplified by polymerase chain reaction. Specific binding was obtained after four to six cycles of CASTing. The population of selected binding sites was then cloned, and a consensus was determined from sequencing individual isolates. Starting from a pool with 14 random bases, purified myogenin yielded a consensus binding site of AACAG[T/C]TGTT, while nuclear extracts retrieved the sequence TTGCACCTGTTNNTT from a pool containing 35 random bases. The latter sequence is consistent with that predicted from combining an E12/E47 half-site (N[not T]CAC) with the purified myogenin half-site ([T/C] TGTT). The presence of paired E boxes in many of the sequences isolated following CASTing with nuclear extracts proves that myogenin can bind cooperatively with other E-box-binding factors.

DNA-binding specificity of the fushi tarazu homeodomain

The fushi tarazu (ftz) gene of Drosophila melanogaster encodes a homeodomain-containing transcription factor that functions in the formation of body segments. Here we report an analysis of the DNA-binding properties of the ftz homeodomain in vitro. We provide evidence that the homeodomain binds to DNA as a monomer, with an equilibrium dissociation constant of 2.5 x 10(-11) M for binding to a consensus binding site. A single ftz binding site occupies 10 to 12 bp, as judged by the ability of protein bound at one site to interfere with binding to an adjacent site. These experiments also demonstrated a lack of cooperative binding between ftz homeodomains. Analysis of single-nucleotide substitutions over an 11-bp sequence shows that a stretch of 6 bp is critical for binding, with an optimal sequence of 5'CTAATTA3'. These data correlate well with recent structural evidence for base-specific contact at these positions. In addition, we found that sequences flanking the region of direct contact have effects on DNA binding that could be of biological significance.

Defining the sequence recognized with BmFTZ-F1, a sequence specific DNA binding factor in the silkworm, Bombyx mori, as revealed by direct sequencing of bound oligonucleotides and gel mobility shift competition analysis

BmFTZ-F1 is a Bombyx mori homologue of FTZ-F1, a positive regulator of the fushi tarazu gene of Drosophila melanogaster. In order to determine the sequence recognized with this factor, we made three sets of oligonucleotide mixture which contain 4 possible nucleotides at different positions within the previously proposed 12-bp binding consensus sequence. Oligonucleotides which bound to purified BmFTZ-F1 were separated by a gel mobility shift procedure and a binding sequence was determined by direct sequencing through Maxam-Gilbert method. By this analysis, 7 positions showed clear sequence preference and 5 positions showed weak or no sequence preference. The importance of each nucleotide at each position was confirmed by a gel mobility shift competition analysis and results were presented as a quantitative difference in the binding affinity. From these analyses, we conclude that the best binding sequence of BmFTZ-F1 is 5'-PyCAAGGPyCPu-3'. This method may be useful for the determination of a binding sequence of other sequence specific DNA binding factor.

DNA-binding specificity of the fushi tarazu homeodomain

The fushi tarazu (ftz) gene of Drosophila melanogaster encodes a homeodomain-containing transcription factor that functions in the formation of body segments. Here we report an analysis of the DNA-binding properties of the ftz homeodomain in vitro. We provide evidence that the homeodomain binds to DNA as a monomer, with an equilibrium dissociation constant of 2.5 x 10(-11) M for binding to a consensus binding site. A single ftz binding site occupies 10 to 12 bp, as judged by the ability of protein bound at one site to interfere with binding to an adjacent site. These experiments also demonstrated a lack of cooperative binding between ftz homeodomains. Analysis of single-nucleotide substitutions over an 11-bp sequence shows that a stretch of 6 bp is critical for binding, with an optimal sequence of 5'CTAATTA3'. These data correlate well with recent structural evidence for base-specific contact at these positions. In addition, we found that sequences flanking the region of direct contact have effects on DNA binding that could be of biological significance.

Binding site selection analysis of protein-DNA interactions via solid phase sequencing of oligonucleotide mixtures

By combining the concept of degenerate oligonucleotide mutagenesis (1,2,3,4) and the convenience of solid phase chemical DNA sequencing (5), we have developed a rapid procedure for determining the specificity of DNA-binding proteins in vitro. Starting with a degenerate oligonucleotide mixture, the technique assays for alternative nucleotides in fractions that are bound or non-bound to the protein of interest. In contrast to previous approaches using degenerate oligonucleotides, it does not involve cloning but rather employs direct sequencing of the oligonucleotide mixtures after attachment to a solid support. Solid state processing obviates the need for both DNA extractions from polyacrylamide gels and time-consuming ethanol precipitations. Because of its convenience and sensitivity, this binding site selection analysis is well suited to determining rapidly the sequence preference of DNA-binding proteins that are available in small amounts, and complements well established approaches like methylation interference or missing contact assays. The solid phase reaction protocol we propose can also improve these latter approaches.

Differences and similarities in DNA-binding preferences of MyoD and E2A protein complexes revealed by binding site selection

A technique was developed for studying protein-DNA recognition that can be applied to any purified protein, partially purified protein, or cloned gene. From oligonucleotides in which particular positions are of random sequence, that subset to which a given protein binds is amplified by the polymerase chain reaction and sequenced as a pool. These selected and amplified binding site (SAAB) "imprints" provide a characteristic set of preferred sequences for protein binding. With this technique, it was shown that homo- and heterooligomers of the helix-loop-helix proteins MyoD and E2A recognize a common consensus sequence, CA--TG, but otherwise bind to flanking and internal positions with different sequence preferences that suggest half-site recognition. These findings suggest that different combinations of dimeric proteins can have different binding sequence preferences.

Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage T4 DNA polymerase

High-affinity nucleic acid ligands for a protein were isolated by a procedure that depends on alternate cycles of ligand selection from pools of variant sequences and amplification of the bound species. Multiple rounds exponentially enrich the population for the highest affinity species that can be clonally isolated and characterized. In particular one eight-base region of an RNA that interacts with the T4 DNA polymerase was chosen and randomized. Two different sequences were selected by this procedure from the calculated pool of 65,536 species. One is the wild-type sequence found in the bacteriophage mRNA; one is varied from wild type at four positions. The binding constants of these two RNA's to T4 DNA polymerase are equivalent. These protocols with minimal modification can yield high-affinity ligands for any protein that binds nucleic acids as part of its function; high-affinity ligands could conceivably be developed for any target molecule.

A wide variety of DNA sequences can functionally replace a yeast TATA element for transcriptional activation

We created a library of DNA molecules in which the required TATA element of a yeast gal-his3 promoter is replaced by random-sequence oligomers averaging 16 bp in length. Surprisingly, 1% of such random sequences functionally replace the native yeast TATA element. In many cases, sequences completely unrelated to the consensus TATA element (TATAAA) stimulate transcription with equal or increased efficiency. Transcription mediated by these synthetic elements requires GAL4 and is initiated from normal his3 initiation sites, suggesting that it occurs by a mechanism indistinguishable from that involving wild-type TATA elements. Many, but not all, of these elements act as substrates for yeast TFIID in reconstituted transcription reactions in vitro. These observations indicate that yeast TFIID can stimulate transcription from TATA elements whose sequences differ from the consensus, and they suggest the possibility of alternative factors that may provide a related function for transcriptional activation.

An efficient method for generating proteins with altered enzymatic properties: application to beta-lactamase

Random-sequence or highly degenerate oligonucleotides have been useful for defining functionally important sequences both in proteins and in nucleic acids. In this approach, such oligonucleotides are used to replace a segment of DNA required for a desired function, and functional sequences are identified by an appropriate genetic or biochemical selection. Here, a collection of 500,000 [corrected] altered beta-lactamase proteins was generated by cloning a mixed-base oligonucleotide in place of the sequences coding for a 17-amino acid portion of the enzyme's active site. Approximately 2000 enzymes from this collection were able to confer ampicillin resistance on Escherichia coli. Fifty-eight of these were chosen for further study after characterization with various beta-lactam substrates. beta-Lactamases having altered specificity against different antibiotics, resistance to the suicide inhibitors clavulanic acid and sulbactam, and temperature-dependent activities were obtained. The amino acid residues responsible for these altered properties as well as for basic enzyme activity are defined. This approach should prove to be an effective and general tool for creating proteins with novel properties, especially in situations in which a high-resolution structure of the protein is not known.

Defining the sequence specificity of DNA-binding proteins by selecting binding sites from random-sequence oligonucleotides: analysis of yeast GCN4 protein

We describe a new method for accurately defining the sequence recognition properties of DNA-binding proteins by selecting high-affinity binding sites from random-sequence DNA. The yeast transcriptional activator protein GCN4 was coupled to a Sepharose column, and binding sites were isolated by passing short, random-sequence oligonucleotides over the column and eluting them with increasing salt concentrations. Of 43 specifically bound oligonucleotides, 40 contained the symmetric sequence TGA(C/G)TCA, whereas the other 3 contained sequences matching six of these seven bases. The extreme preference for this 7-base-pair sequence suggests that each position directly contacts GCN4. The three nucleotide positions on each side of this core heptanucleotide also showed sequence preferences, indicating their effect on GCN4 binding. Interestingly, deviations in the core and a stronger sequence preference in the flanking region were found on one side of the central C . G base pair. Although GCN4 binds as a dimer, this asymmetry supports a model in which interactions on each side of the binding site are not equivalent. The random selection method should prove generally useful for defining the specificities of other DNA-binding proteins and for identifying putative target sequences from genomic DNA.

Release of the sigma subunit from Escherichia coli RNA polymerase transcription complexes is dependent on the promoter sequence

The sigma subunit of bacterial RNA polymerase is required for the specific initiation of transcription at promoter sites. However, sigma is released from the transcription complex shortly after transcription is initiated, and elongation proceeds in the absence of sigma. In order to study the position of sigma release, we have developed a method to quantify the photoaffinity labeling produced by an aryl azide positioned at the leading (5'-) end of nascent RNA, as a function of the transcript length [Stackhouse, T.M., & Meares, C.F. (1988) Biochemistry 27, 3038-3045]. Here we compare photoaffinity labeling of transcription complexes containing three natural bacteriophage promoters (lambda PR, lambda PL, and T7 A1) and two recombinant constructs, A1/PR (T7 A1 promoter with the lambda PR transcribed region) and PR/A1 (lambda PR promoter with the T7 A1 transcribed region). Significant photoaffinity labeling of the sigma subunit was observed only on the templates containing the lambda PR promoter region, regardless of the sequence of the transcribed region. These results indicate the molecular interactions responsible for the position of sigma release from the transcription complex mainly involve the nucleotide sequence of the promoter region--rather than the transcribed region--of the DNA template. Further studies on transcription complexes containing the A1/PR and the PR/A1 templates were performed, using polyclonal antibodies against the holoenzyme or against the sigma subunit. These experiments corroborate the promoter dependence of sigma release. They also show a correlation between the release of sigma and stable binding of the transcript by the transcription complex.

Defining the sequence specificity of DNA-binding proteins by selecting binding sites from random-sequence oligonucleotides: analysis of yeast GCN4 protein

We describe a new method for accurately defining the sequence recognition properties of DNA-binding proteins by selecting high-affinity binding sites from random-sequence DNA. The yeast transcriptional activator protein GCN4 was coupled to a Sepharose column, and binding sites were isolated by passing short, random-sequence oligonucleotides over the column and eluting them with increasing salt concentrations. Of 43 specifically bound oligonucleotides, 40 contained the symmetric sequence TGA(C/G)TCA, whereas the other 3 contained sequences matching six of these seven bases. The extreme preference for this 7-base-pair sequence suggests that each position directly contacts GCN4. The three nucleotide positions on each side of this core heptanucleotide also showed sequence preferences, indicating their effect on GCN4 binding. Interestingly, deviations in the core and a stronger sequence preference in the flanking region were found on one side of the central C . G base pair. Although GCN4 binds as a dimer, this asymmetry supports a model in which interactions on each side of the binding site are not equivalent. The random selection method should prove generally useful for defining the specificities of other DNA-binding proteins and for identifying putative target sequences from genomic DNA.

Excess information at bacteriophage T7 genomic promoters detected by a random cloning technique

In our previous analysis of the information at binding sites on nucleic acids, we found that most of the sites examined contain the amount of information expected from their frequency in the genome. The sequences at bacteriophage T7 promoters are an exception, because they are far more conserved (35 bits of information content) than should be necessary to distinguish them from the background of the Escherichia coli genome (17 bits). To determine the information actually used by the T7 RNA polymerase, promoters were chemically synthesized with many variations and those that function well in an in vivo assay were sequenced. Our analysis shows that the polymerase uses 18 bits of information, so the sequences at phage genomic promoters have significantly more information than the polymerase needs. The excess may represent the binding site of another protein.

Defining the consensus sequences of E.coli promoter elements by random selection

The consensus sequence of E.coli promoter elements was determined by the method of random selection. A large collection of hybrid molecules was produced in which random-sequence oligonucleotides were cloned in place of a wild-type promoter element, and functional -10 and -35 E.coli promoter elements were obtained by a genetic selection involving the expression of a structural gene. The DNA sequences and relative levels of function for -10 and -35 elements were determined. The consensus sequences determined by this approach are very similar to those determined by comparing DNA sequences of naturally occurring E.coli promoters. However, no strong correlation is observed between similarity to the consensus and relative level of function. The results are considered in terms of E.coli promoter function and of the general applicability of the random selection method.

Saturation mutagenesis of a yeast his3 "TATA element": genetic evidence for a specific TATA-binding protein

The yeast his3 promoter region contains two distinct classes of "TATA elements," constitutive (TC) and regulatory (TR), that are defined by their interactions with upstream promoter elements, selectivity of initiation sites, and chromatin structure. TC is localized between positions -83 and -53, and TR is localized between positions -55 and -35, regions in which there are several TATA-like sequences. In this study, we used saturation mutagenesis to examine the structural requirements of the his3 TR element necessary for transcriptional induction. To avoid the complications of redundant elements, the phenotypic analysis was carried out by using a gal-his3 hybrid promoter whose function depends on a short oligonucleotide containing the prospective his3 TR element. In this context, an oligonucleotide containing the sequence TATAAA is sufficient for TR function. However, 17 out of the 18 possible single-base substitutions and 9 out of 10 double mutations of this sequence abolish TR function. This strict sequence requirement for TR function strongly suggests that the TR element is a target site for a sequence-specific DNA-binding protein. Further, as the region encoding TC and promoters of certain other yeast genes do not contain a sequence that is compatible with TR function, we suggest that yeast cells contain multiple proteins with distinct sequence specificities that carry out a related "TATA function" and that yeast promoters can be divided into classes based on their downstream promoter elements.