Stabilized bioactive peptides and methods of identification, synthesis, and use

Description of equivalent WO 2004011485 A2

STABILIZED BIOACTIVE PEPTIDES AND METHODS OF IDENTIFICATION, SYNTHESIS, AND USE

CROSS-REFERENCE TO RELATED APPLICATIONS

This international application claims priority to U.S. Serial No. 10/210,023, filed July 31, 2002, which is a continuation-in-part of U.S. Serial No. 09/701,947, filed December 5, 2000, which is a National Stage application under 35 U.S.C. §371 of PCT/US99/23731, filed October 12, 1999, which in turn claims the benefit of U. S. Provisional Patent Applications Serial Nos. 60/104,013, filed October 13, 1998, and 60/112,150, filed December 14, 1998, each of which is incorporated herein by reference in its entirety.

TECHNICAL FIELD

This invention relates to stabilized bioactive peptides, and more particularly to bioactive peptides that contain heterologous stabilizing groups attached to one or both of the bioactive peptide 's termini.

BACKGROUND

Bioactive peptides are small peptides that elicit a biological activity. Since the discovery of secretin in 1902, over 500 of these peptides which average 20 amino acids in size have been identified and characterized. They have been isolated from a variety of systems, exhibit a wide range of actions, and have been utilized as therapeutic agents in the field of medicine and as diagnostic tools in both basic and applied research. Tables 1 and 2 list some of the best known bioactive peptides. Table 1: Bioactive peptides utilized in medicine

<img file="WO2004011485A2_D0001.tif" />

Table 2: Bioactive peptides utilized in applied research

<img file="WO2004011485A2_D0002.tif" />

Where the mode of action of these peptides has been determined, it has been found to be due to the interaction of the bioactive peptide with a specific protein target. In most of the cases, the bioactive peptide acts by binding to and inactivating its protein target with extremely high specificities. Binding constants of these peptides for their protein targets typically have been determined to be in the nanomolar (nM, 10^"9 M) range with binding constants as high as 10^"12 M (picomolar range) having been reported. Table 3 shows target proteins inactivated by several different bioactive peptides as well as the binding constants associated with binding thereto. <img file="WO2004011485A2_D0003.tif" />

Recently, there has been an increasing interest in employing synthetically derived bioactive peptides as novel pharmaceutical agents due to the impressive ability of the naturally occurring peptides to bind to and inhibit specific protein targets. Synthetically derived peptides could be useful in the development of new antibacterial, antiviral, and anticancer agents. Examples of synthetically derived antibacterial or antiviral peptide agents would be those capable of binding to and preventing bacterial or viral surface proteins from interacting with their host cell receptors, or preventing the action of specific toxin or protease proteins. Examples of anticancer agents would include synthetically derived peptides that could bind to and prevent the action of specific oncogenic proteins.

To date, novel bioactive peptides have been engineered through the use of two different in vitro approaches. The first approach produces candidate peptides by chemically synthesizing a randomized library of 6-10 amino acid peptides (J. Eichler et al., Med. Res. Rev. 15: 48 1-496 (1995); K. Lam, Anticancer Drug Des. 12:145-167 (1996); M. Lebl et al., Methods Enzymol. 289:336-392 (1997)). In the second approach, candidate peptides are synthesized by cloning a randomized oligonucleotide library into a Ff filamentous phage gene, which allows peptides that are much larger in size to be expressed on the surface of the bacteriophage (H. Lowman, Ann. Rev. Biophys. Biomol. Struct. 26: 401-424 (1997); G. Smith et al., et al. Meth. Enz. 217: 228-257 (1993)). To date, randomized peptide libraries up to 38 amino acids in length have been made, and longer peptides are likely achievable using this system. The peptide libraries that are produced using either of these strategies are then typically mixed with a preselected matri -bound protein target. Peptides that bind are eluted, and their sequences are determined. From this information new peptides are synthesized and their inhibitory properties are determined. This is a tedious process that only screens for one biological activity at a time.

Although these in vitro approaches show promise, the use of synthetically derived peptides has not yet become a mainstay in the pharmaceutical industry. The primary obstacle remaining is that of peptide instability within the biological system of interest as evidenced by the unwanted degradation of potential peptide drugs by proteases and/or peptidases in the host cells. There are three major classes of peptidases which can degrade larger peptides: amino and carboxy exopeptidases which act at either the amino or the carboxy terminal end of the peptide, respectively, and endopeptidases which act on an internal portion of the peptide. Aminopeptidases, carboxypeptidases, and endopeptidases have been identified in both prokaryotic and eukaryotic cells. Many of those that have been extensively characterized were found to function similarly in both cell types. Interestingly, in both prokaryotic and eukaryotic systems, many more aminopeptidases than carboxypeptidases have been identified to date.

Approaches used to address the problem of peptide degradation have included the use of D-amino acids or modified amino acids as opposed to the naturally occurring L-amino acids (e.g., J. Eichler et al., Med Res Rev. 15:48 1496 (1995); L. Sanders, Eur. J. Drug Metabol. Pharmacokinetics 15:95-102 (1990)), the use of cyclized peptides (e.g., R. Egleton, et al., Peptides 18:1431-1439 (1997)), and the development of enhanced delivery systems that prevent degradation of a peptide before it reaches its target in a patient (e. g., L. Wearley, Crit. Rev. Ther. Drug Carrier Syst. 8:33 1-394 (1991); L. Sanders, Eur. J. Drug Metabol. Pharmacokinetics 15: 95- 102 (1990)). Although these approaches for stabilizing peptides and thereby preventing their unwanted degradation in the biosystem of choice (e.g., a patient) are promising, there remains no way to routinely and reliably stabilize peptide drugs and drug candidates. Moreover, many of the existing stabilization and delivery methods cannot be directly utilized in the screening and development of novel useful bioactive peptides. A biological approach that would serve as both a method of stabilizing peptides and a method for identifying novel bioactive peptides would represent a much needed advance in the field of peptide drug development.

Summary of the Invention

The present invention provides an intracellular screening method for identifying novel bioactive peptides. A host cell is transformed with an expression vector comprising a tightly regulable control region operably linked to a nucleic acid sequence encoding a peptide. Typically, the encoded peptide has a stabilizing group positioned at one or both ends of the peptide. The transformed host cell is first grown under conditions that repress expression of the peptide and then, subsequently, expression of the peptide is induced. Phenotypic changes in the host cell upon expression of the peptide are indicative of bioactivity, and are evaluated. If, for example, expression of the peptide is accompanied by complete or partial inhibition of host cell growth, the expressed peptide constitutes a bioactive peptide, in that it functions as an inhibitory peptide.

Intracellular identification of bioactive peptides can be advantageously carried out in a pathogenic microbial host cell. Bioactive peptides having antimicrobial activity are readily identified in a microbial host cell system. Further, the method can be carried out in a host cell that has not been modified to reduce or eliminate the expression of naturally expressed proteases or peptidases. When carried out in a host cell comprising proteases and peptides, the selection process of the invention is biased in favor of bioactive peptides that are protease and peptidase-resistant. The tightly regulable control region of the expression vector used to transform the microbial host cell according to the invention can be derived from the wild-type Escherichia coli lac operon, and the transformed host cell can include an amount of Lac repressor protein effective to repress expression of the peptide during host cell growth under repressed conditions. To insure a sufficient amount of Lac repressor protein, the host cell can be transformed with a second vector that overproduces Lac repressor protein.

Optionally, the expression vector used to transform the host cell can be genetically engineered to encode a stabilized peptide that is resistant to peptidases and proteases. For example, the coding sequence can be designed to encode a stabilizing group at either or both of the peptide's N-terminus or C-terminus. As another example, the coding sequence can be designed to encode a stabilizing motif such as an α-helix motif or an opposite charge ending motif, as described below. The presence of a stabilizing group at a peptide terminus and/or of a stabilizing motif can slow down the rate of intracellular degradation of the peptide.

A plurality of vectors can be used to screen a randomized library of candidate bioactive peptides.

The present invention also provides a polypeptide that includes a bioactive peptide and a stabilizing group coupled to at least one terminus of the bioactive peptide. Preferably, the bioactive peptide is 50 or fewer amino acids in length. The stabilizing group is heterologous to the bioactive peptide and can be, for example, a proline, a proline-containing peptide, a single α-helix or multiple helix bundle, or other polypeptide or small protein such as Rop, human serum albumin, and the like. In one embodiment of the stabilized polypeptide, the stabilizing group(s) lack the capacity to participate in the formation of an intramolecular disulfide bond within the polypeptide. Thus, the stabilizing group is preferably not a thioredoxin polypeptide.

When a polypeptide includes a stabilizing group on each terminus, the stabilizing groups can be the same or different. If different, the stabilizing groups are optionally heterologous to each other, as that term is defined below. The first and second stabilizing groups can, but need not, interact to form a naturally occurring secondary or tertiary structure. Further, the first and second stabilizing groups can, ^" but need not, confine the N-terminus and the C-terminus of the bioactive peptide in close proximity. The invention further includes a nucleic acid encoding the polypeptide of the invention. A vector that contains such a nucleic acid is also included. Preferably the vector contains a tightly regulable expression control sequence operably linked to the nucleic acid sequence encoding the stabilized polypeptide.

The present invention also includes a method for making a stabilized polypeptide that involves coupling a stabilizing group to at least one terminus of a bioactive peptide. Coupling can be achieved chemically or enzymatically, or can occur as the result of translation in a host cell of a vector containing a nucleic acid sequence encoding the stabilized polypeptide. The vector comprises an expression control sequence operably linked to the coding sequence; preferably, the expression control sequence is tightly regulable in said host cell. Optionally the method includes determining stability of said stabilized polypeptide relative to said bioactive peptide. When the method is performed in a host cell, the host cell is first transformed with an exogenous nucleic acid encoding the stabilized polypeptide, then the stabilized polypeptide is expressed and recovered. The host cells can be prokaryotic, such as bacteria, or eukaryotic.

Phage display can be used to identify a bioactive peptide that can be subsequently stabilized according to the invention. When displayed on the surface of a bacteriophage, bioactive peptides are tethered at one end by a bacteriophage protein. The free bioactive peptide (i.e., uncoupled from the bacteriophage protein) may exhibit a lack of stability in vivo. Hence, the invention involves stabilizing these bioactive peptides by coupling them to a stabilizing group at the end that had been tethered during phage display, thereby effectively replacing the bacteriophage protein as a stabilizing feature. Coupling can take place chemically, enzymatically, or by way of recombinant genetic engineering, as described herein. Polypeptides thus stabilized are also included in the invention.

Alternatively, the stabilized polypeptide can be produced as a direct product of phage display. A bacteriophage that contains an exogenous nucleic acid encoding a polypeptide comprising a bioactive peptide (or candidate bioactive peptide), a bacteriophage protein coupled to one terminus of the bioactive peptide, and a stabilizing group coupled to the other terminus of the bioactive peptide is cultured under conditions to cause the bacteriophage to express the stabilized polypeptide and display it on its surface. The stabilizing group can be coupled to either end of the bioactive peptide, and the bacteriophage protein is coupled to the other end. Optionally the stabilized polypeptide is cleaved from the host cell surface to yield a stabilized bioactive peptide comprising the bioactive peptide and the stabilizing group.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, suitable methods and materials are described below. All publications, patent applications, patents, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting.

Other features and advantages of the invention will be apparent from the following detailed description and from the claims.

Brief Description of the Drawings

Figure 1 shows the control region (SEQ ID NO: 1) of the wild-type lac operon from the auxiliary operator O3 through the translational start of the lacZ gene. DNA binding sites include the operators O3 and OI (both underlined), CAP (boxed), the - 35 site (boxed), and the -10 site (boxed), while important RNA and protein sites include the Lacl translation stop site (TGA), the +1 lacZ transcription start site, the Shine Dalgarno (SD) ribosome binding site for lacZ, and the LacZ translation start site (ATG).

Figure 2 is a map of plasmid pLACll. The unique restriction sites and the base pair at which they cut are indicated. Other sites of interest are also shown, including Tet (98-1288), Rop (1931-2122), ori (2551-3138), Amp (3309-25 4 169), and lacVO (4424-4536).

Figure 3 is a map of plasmid pLAC22. The unique restriction sites and the base pair at which they cut are indicated. Other sites of interest are also shown, including Tet (98-1288), Rop (1927-2118), ori (2547-3134), Amp (3305-4165), lacM (4452-5536), and lacPO (5529-5641).

Figure 4 is a map of plasmid pLAC33. The unique restriction sites and the base pair at which they cut are indicated. Other sites of interest are also shown, including Tet (98-1288), ori (1746-2333), Amp (2504-3364), and lacFO (3619-3731). Figure 5 shows the response of the pLACll -lacZ construct (open circles) to varying amounts of isopropyl β-D-thiogalactoside (JPTG). A filled square indicates the β-galactosidase activity that was obtained when MG1655 or CSH27 cells were grown in rich media induced with 1 mM IPTG, while a filled diamond indicates the β- galactosidase activity that was obtained when MG1655 or CSH27 cells were grown in M9 minimal lactose media. Figure 6 shows growth curves depicting the inhibitory effects of a two day inhibitor (pPepl2) versus a one day inhibitor (pPepl). Data points for the control, pLAC 11, for pPep 1, and for pPep 12, are indicated by squares, circles, and triangles, respectively. Figure 7 is a map of the p-Rop(C) fusion vector. The unique restriction sites and the base pair at which they cut are indicated. Other sites of interest are also shown, including Rop (7-198), ori (627-1214), Amp (2245-1385), lacPO (2500-26 12). Figure 8 is a map of the p(N)Rop-fusion vector. The unique restriction sites and the base pair at which they cut are indicated. Other sites of interest are also shown: Rop (7-204), ori (266-853), Amp (1024-1 884), lacPO (2139-2251).

Figure 9 illustrates a peptide (SEQ ID NO:2) having the opposite charge ending motif, wherein the amino and carboxy termini of the peptide are stabilized by the interactions of the opposite charge ending amino acids.

DETAILED DESCRIPTION

The present invention represents a significant advance in the art of peptide drag development by allowing concurrent screening for peptide bioactivity and stability. Randomized recombinant peptides are screened for bioactivity in a tightly regulated inducible expression system that permits essentially complete repression of peptide expression in the host cell. Subsequent induction of peptide expression can then be used to identify peptides that inhibit host cell growth or possess other bioactivities.

Intracellular screening of randomized peptides has many advantages over existing methods. Bioactivity is readily apparent, many diverse bioactivities can be screened for simultaneously, very large numbers of peptides can be screened using easily generated peptide libraries, and the host cell, if desired, can be genetically manipulated to identify an affected protein target. Advantageously, randomized peptides can be screened in a host cell that is identical to or closely resembles the eventual target cell for antimicrobial, anticancer, and other therapeutic applications. An additional and very important feature of this system is that selection is naturally biased in favor of peptides that are stable in an intracellular environment, i.e., that are resistant to proteases and peptidases. Fortuitously, bacterial peptidases are very similar to eukaryotic peptidases. Peptides that are stable in a bacterial host are thus likely to be stable in a eukaryotic cell as well, allowing bacterial cells to be used in initial screens to identify drugs that may eventually prove useful as human or animal therapeutics. The invention is directed to the identification and use of bioactive peptides. A bioactive peptide is a peptide having a biological activity. The term "bioactivity" as used herein includes, but is not limited to, any type of interaction with another biomolecule, such as a protein, glycoprotein, carbohydrate, for example an oligosaccharide or polysaccharide, nucleotide, polynucleotide, fatty acid, hormone, enzyme, cofactor or the like, whether the interactions involve covalent or noncovalent binding. Bioactivity further includes interactions of any type with other cellular components or constituents including salts, ions, metals, nutrients, foreign or exogenous agents present in a cell such as viruses, phage and the like, for example binding, sequestration or transport-related interactions. Bioactivity of a peptide can be detected, for example, by observing phenotypic effects in a host cell in which it is expressed, or by performing an in vitro assay for a particular bioactivity, such as affinity binding to a target molecule, alteration of an enzymatic activity, or the like. Examples of bioactive peptides include antimicrobial peptides and peptide drugs. Antimicrobial peptides are peptides that adversely affect a microbe such as a bacterium, virus, protozoan, or the like. Antimicrobial peptides include, for example, inhibitory peptides that slow the growth of a microbe, microbiocidal peptides that are effective to kill a microbe (e.g., bacteriocidal and virocidal peptide drugs, sterilants, and disinfectants), and peptides effective to interfere with microbial reproduction, host toxicity, or the like. Peptide drugs for therapeutic use in humans or other animals include, for example, antimicrobial peptides that are not prohibitively toxic to the patient, and peptides designed to elicit, speed up, slow down, or prevent various metabolic processes in the host such as insulin, oxytocin, calcitonin, gastrin, somatostatin, anticancer peptides, and the like. The term "peptide" as used herein refers to a plurality of amino acids joined together in a linear chain via peptide bonds. Accordingly, the term "peptide" as used herein includes a dipeptide, tripeptide, oligopeptide and polypeptide. A dipeptide contains two amino acids; a tripeptide contains three amino acids; and the term oligopeptide is typically used to describe peptides having between 2 and about 50 or more amino acids. Peptides larger than about 50 are often referred to as polypeptides or proteins. For purposes of the present invention, a "peptide" is not limited to any particular number of amino acids. Preferably, however, the peptide contains about 2 to about 50 amino acids (e.g., about 5 to about 40 amino acids, about 5 to about 20 amino acids, or about 7 to 15 amino acids). The library used to transform the host cell is formed by cloning a randomized, peptide-encoding oligonucleotide into a nucleic acid construct having a tightly regulable expression control region. An expression control region can be readily evaluated to determine whether it is "tightly regulable," as the term is used herein, by bioassay in a host cell engineered to contain a mutant nonfunctional gene "X."

Transforming the engineered host cell with an expression vector containing a tightly regulable expression control region operably linked to a cloned wild-type gene "X" will preserve the phenotype of the engineered host cell under repressed conditions. Under induced conditions, however, the expression vector containing the tightly regulable expression control region that is operably linked to the cloned wild-type gene "X" will complement the mutant nonfunctional gene X to yield the wild-type phenotype. In other words, a host cell containing a null mutation which is transformed with a tightly regulable expression vector capable of expressing the chromosomally inactivated gene will exhibit the null phenotype under repressed conditions; but when expression is induced, the cell will exhibit a phenotype indistinguishable from the wild-type cell. It should be understood that the expression control region in a tightly regulable expression vector of the present invention can be readily modified to produce higher levels of an encoded biopeptide, if desired (see, e.g., Example 1, below). Such modification may unavoidably introduce some "leakiness" into expression control, resulting in a low level of peptide expression under repressed conditions.

In one embodiment, the expression control region of the inducible expression vector is derived from the wild-type E. coli lac promoter/operator region. The expression vector can contain a regulatory region that includes the auxiliary operator O3, the CAP binding region, the -35 promoter site, the -10 promoter site, the operator OI, the Shine-Dalgarno sequence for lacZ, and a spacer region between the end of the Shine-Dalgarno sequence and the ATG start of the lacZ coding sequence (see Fig. 1). It is to be understood that variations in the wild-type nucleic acid sequence of the lac promoter/operator region can be tolerated in the expression control region of the preferred expression vector and are encompassed by the invention, provided that the expression control region remains tightly regulable as defined herein. For example, the -10 site of the wild-type lac operon (TATGTT) is weak compared to the bacterial consensus -10 site sequence TATAAT, sharing four out of six positions. It is contemplated that other comparably weak promoters are equally effective at the -10 site in the expression control region; a strong promoter is to be avoided in order to insure complete repression in the uninduced state. With respect to the -35 region, the sequence of the wild-type lac operon, TTTACA, is one base removed from the consensus -35 sequence TTGACA. It is contemplated that a tightly regulable lac operon-derived expression control region could be constructed using a weaker -35 sequence (i.e., one having less identity with the consensus -35 sequence) and a wild- type -10 sequence (TATAAT), yielding a weak promoter that needs the assistance of the CAP activator protein. Similarly, it is to be understood that the nucleic acid sequence of the CAP binding region can be altered as long as the CAP protein binds to it with essentially the same affinity. The spacer region between the end of the

Shine-Dalgarno sequence and the ATG start of the lacZ coding sequence is typically between about 5 and about 10 nucleotides in length, preferably about 5 to about 8 nucleotides in length, more preferably about 7-9 nucleotides in length. The most preferred composition and length of the spacer region depends on the composition and length of Shine-Dalgarno sequence with which it is operably linked as well as the translation start codon employed (i.e., AUG, GUQ or UUG), and can be determined accordingly by one of skill in the art. Preferably the nucleotide composition of the spacer region is "AT rich"; that is, it contains more A's and T's than it does G's and Cs. In one embodiment of the method of the invention, the expression vector has the identifying characteristics of pLAC 11 (ATCC No. 207108) and, in a particularly convenient embodiment, is pLAC 11 (ATCC No. 207108).

As used in the present invention, the term "vector" is to be broadly interpreted as including a plasmid, including an episome, a viral vector, a cosmid, or the like. A vector can be circular or linear, single-stranded or double-stranded, and can comprise RNA, DNA, or modifications and combinations thereof. Selection of a vector or plasmid backbone depends upon a variety of characteristics desired in the resulting construct, such as selection marker(s), plasmid copy number, and the like. A nucleic acid sequence is "operably linked" to an expression control sequence in the regulatory region of a vector, such as a promoter, when the expression control sequence controls or regulates the transcription and/or the translation of that nucleic acid sequence. A nucleic acid that is "operably linked" to an expression control sequence includes, for example, an appropriate start signal (e.g., ATG) at the beginning of the nucleic acid sequence to be expressed and a reading frame that permits expression of the nucleic acid sequence under control of the expression control sequence to yield production of the encoded peptide. The regulatory region of the expression vector optionally includes a termination sequence, such as a codon for which there is no corresponding aminoacetyl-tRNA, thus ending peptide synthesis. Typically, when the ribosome reaches a termination sequence or codon during translation of the mRNA, the polypeptide is released and the ribosome-mRNA-tRNA complex dissociates.

An expression vector optionally includes one or more selection or marker sequences, which typically encode an enzyme capable of inactivating a compound in the growth medium. The inclusion of a marker sequence can, for example, render the host cell resistant to an antibiotic, or it can confer a compound-specific metabolic advantage on the host cell. Markers such as green fluorescent protein also can be used to monitor growth or toxicity in host cells in which it is expressed. Cells can be transformed with the expression vector using any convenient method known in the art, including chemical transformation, e.g., whereby cells are made competent by treatment with reagents such as CaCl₂; electroporation and other electrical techniques; microinjection and the like.

The vector may further include a tightly regulable expression control sequence operably linked to the nucleic acid sequence encoding the polypeptide, particularly a stabilized polypeptide, as described herein. In embodiments of the method that use a tightly regulable expression system derived from the lac operon, the host cell is or has been genetically engineered or otherwise altered to contain a source of Lac repressor protein in excess of the amount produced in wild-type E. coli. A host cell that contains an excess source of Lac repressor protein is one that expresses an amount of Lac repressor protein sufficient to repress expression of the peptide under repressed conditions, i.e., in the absence of an inducing agent, such as isopropyl β-D- thiogalactoside (IPTG). Preferably, expression of Lac repressor protein is constitutive. For example, the host cell can be transformed with a second vector comprising a gene encoding Lac repressor protein, preferably lacl, more preferably lacH, to provide an excess source of Lac repressor protein in trans, i.e., extraneous to the tightly regulable expression vector. An episome can also serve as a trans source of Lac repressor. Another option for providing a trans source of Lac repressor protein is the host chromosome itself, which can be genetically engineered to express excess Lac repressor protein. Alternatively, a gene encoding Lac repressor protein can be included on the tightly regulable expression vector that contains the peptide-encoding oligonucleotide so that Lac repressor protein is provided in cis. The gene encoding the Lac repressor protein is preferably under the control of a constitutive promoter. The invention is not limited by the type of host cell used for screening. The host cell can be a prokaryotic or a eukaryotic cell. Preferred mammalian cells include human cells, of any tissue type, and can include cancer cells or cell lines (e.g., HeLa cells) or other immortalized cell lines, hybridomas, pluripotent or omnipotent cells such as stem cells or cord blood cells, etc., without limitation. Preferred yeast host cells include Saccharomyces cerevisiae and Schizosaccharomyces pombe. Preferred bacterial host cells include gram negative bacteria, such as E. coli and various Salmonella spp., and gram positive bacteria, such as bacteria from the genera

Staphylococcus, Streptococcus and Enterococcus. Protozoan cells are also suitable host cells, hi clear contrast to conventional recombinant protein expression systems, it is preferable that the host cell contains proteases and/or peptidases, since the selection will, as a result, be advantageously biased in favor of peptides that are protease- and peptidase-resistant. More preferably, the host cell has not been modified, genetically or otherwise, to reduce or eliminate the expression of any naturally expressed proteases or peptidases. The host cell can be selected with a particular purpose in mind. For example, if it is desired to obtain peptide drugs specific to inhibit Staphylococcus, peptides can be advantageously expressed and screened in Staphylococcus.

There is, accordingly, tremendous potential for the application of this technology in the development of new antibacterial peptides useful to treat various ^■ pathogenic bacteria. Of particular interest are pathogenic Staphylococci, Streptococci, and Enterococci, which are the primary causes of nosocomial infections. Many of these strains are becoming increasingly drug-resistant at an alarming rate. The technology of the present invention can be practiced in a pathogenic host cell to isolate inhibitor peptides that specifically target the pathogenic strain of choice. Inhibitory peptides identified using pathogenic microbial host cells in accordance with the invention may have direct therapeutic utility; based on what is known about peptide import, it is very likely that small peptides are rapidly taken up by

Staphylococci, Streptococci, and Enterococci. Once internalized, the inhibitory peptides identified according to the invention would be expected to inhibit the growth of the bacteria in question. It is therefore contemplated that novel inhibitor peptides so identified can be used in medical treatments and therapies directed against microbial infection. It is further contemplated that these novel inhibitor peptides can be used, in turn, to identify additional novel antibacterial peptides using a synthetic approach. The coding sequence of the inhibitory peptides is determined, and peptides are then chemically synthesized and tested in the host cell for their inhibitory properties.

Novel inhibitor peptides identified in a pathogenic microbial host cell according to the invention can also be used to elucidate potential new drug targets. The protein target that the inhibitor peptide inactivated is identified using reverse genetics by isolating mutants that are no longer inhibited by the peptide. These mutants are then mapped in order to precisely determine the protein target that is inhibited. New antibacterial drugs can then be developed using various known or yet to be discovered pharmaceutical strategies.

Following transformation of the host cell, the transformed host cell is initially grown under conditions that repress expression of the peptide. Expression of the peptide is then induced. For example, when a lac promoter/operator system is used for expression, IPTG is added to the culture medium. A determination is subsequently made as to whether the peptide is inhibitory to host cell growth, wherein inhibition of host cell growth under induced but not repressed conditions is indicative of the expression of a bioactive peptide. Alternatively, a vector encoding a marker such as green fluorescent protein

(GFP) can be used to monitor toxicity of the random peptides in a host. In general, fluorescence can be monitored in cells that are expressing both GFP and a randomized peptide and compared with the fluorescence of control cells, i.e., cells expressing only GFP. If the randomized peptide is toxic to the host, fluorescence would not be observed or would be decreased relative to the control cells. Alternatively, GFP or other markers can be used to monitor the cells for complete or partial inhibition of cell division, or for induction of apoptosis.

For example, to identify a potential anticancer peptide, a cancer cell line such as the HeLa cell line can be used as the host. The cell line can be transfected with one or more vectors such that the cell line expresses both a marker (e.g., GFP) and a peptide from a library of random peptides. It should be noted that the nucleic acid sequences encoding the marker and the random peptides can be on the same or different vectors. Expression of the random peptides can be controlled by a tightly regulable control sequence, although this need not be the case. The transfected cells can be seeded into multi-well plates (e.g., 96-well plates) or into multiple flasks, with each well or plate receiving cells collectively expressing a single random peptide. In one embodiment, the number of cells seeded into each well or flask is chosen to ensure that an adequate number of cells expressing a random peptide and a marker is present in each well or flask. This can depend, for example, on the original transfection efficiency. Under these conditions, the marker will be observed in all wells or in each flask, unless a peptide that is being expressed is toxic to the cells or otherwise exhibits a desirable bioactivity (e.g., causes a complete or partial inhibition of cell division, or induces apoptosis). In wells in which fluorescence is not observed or the level of fluorescence is decreased, the random peptides are candidates for anticancer peptides. Candidate anti-cancer peptides identified by this method can be further screened to determine if the peptide is selectively toxic or otherwise bioactive in cancer cells. For example, the bioactivity can be compared between malignant and non-malignant cells using a 96-well screening format similar to that described above. In a similar fashion, the method of the invention can be used to identify peptides that exert an agonist effect on cell division and growth. For example, stem cells and cord blood cells, which typically do not proliferate well, can be employed as the host cells. Candidate peptides can be assayed for a positive effect on cell division and growth. Agonistic peptides may be useful in wound healing, organ transplantation and cardiovascular applications.

A plurality of vectors (e.g., a library) that encode a population of randomized peptides can be used to identify bioactive peptides (e.g., antimicrobial or anticancer peptides). A library can include at least two different vectors (e.g., at least five, 50, 500, 5000, 50,000, 100,000, 1 x 10⁶, 5 x 10⁶ or more vectors), with each of the vectors encoding different, randomized peptides (e.g., at least five, 50, 500, 5000, 50,000, 100,000, 1 x 10 , 5 x 10 or more different randomized peptides). Bioactive peptides can be identified by screening each of the randomized peptides encoded by the different vectors, for a desired bioactivity (e.g., cell toxicity). Randomized peptides that exhibit the desired bioactivity can be selected as bioactive peptides. During development and testing of the intracellular screening method of the present invention, it was surprisingly discovered that several bioactive peptides identified from the randomized peptide library shared particular structural features. For example, a disproportionately high number of bioactive peptides identified using the intracellular screening method contained one or more proline residues at or near a peptide terminus. A disproportionate number also contained sequences predicted, using structure prediction algorithms well-known in the art, to form secondary structures such as α helices or β sheets; or a hydrophobic membrane spanning domain. Bioactive fusion proteins comprising the randomized peptide sequence fused to the Rop protein, due to a deletion event in the expression vector, were also identified. Accordingly, randomized peptides used in the screening method of the invention can be optionally engineered according to the method of the invention in a biased synthesis to increase their stability by making one or both of the N-terminal or C-terminal ends more resistant to proteases and peptidases. For example, a vector can include a nucleic acid sequence encoding a stabilized polypeptide, wherein the stabilized polypeptide includes a randomized peptide and a stabilizing group positioned at the N- and/or C-terminus of the randomized peptide. The resulting stabilized polypeptide includes the randomized peptide and the stabilizing group coupled to one or both of the randomized peptide 's termini. By "coupled to . . .one or both . . . termini" it is meant that the randomized peptide is covalently linked, at one or both of its termini, to the stabilizing group. The nucleic acid sequence that encodes the randomized peptide in the expression vector or the expression vector itself is preferably modified such that a first stabilizing group is positioned at the N-terminus of the peptide, and a second stabilizing group is positioned at the C-terminus of the peptide.

Notably, the bioactive peptides identified according to the method of the invention are, by reason of the method itself, stable in the intracellular environment of the host cell.- The method of the invention thus preferably identifies bioactive peptides that are resistant to proteases and peptidases. Resistance to proteases and peptidases can be evaluated by measuring peptide degradation when in contact with appropriate cell extracts (e.g., bacterial, yeast, or human cell extracts), employing methods well-known in the art. A bioactive peptide, without stabilization, can be used as a control. For example, degradation of a stabilized, biotinylated peptide can be assessed by electrophoresis through an SDS-polyacrylamide gel and Western blotting using an avidin-horseradish peroxidase conjugate. Alternatively, resistance to proteases and peptidases can be evaluated by measuring peptide degradation when in contact with purified peptidases and/or proteases (e.g., the Lon and Clp proteases from E. coli). A protease- or peptidase-resistant peptide exhibits a longer half-life in the presence of proteases or peptidases compared to a control peptide. In should be noted that the stabilization of peptides (e.g., polypeptides containing about 2 to about 50 amino acids) in accordance with the present invention is an unexpected as peptides, unlike proteins, are relatively unstable in physiological environments. For example, the half-life of most peptides in physiological environments is about 2 minutes, whereas the half -life of most proteins in physiological environments is typically well in excess of 2 minutes and is often measured in hours or days. Proteins possess an inherent stability as a result of complex intramolecular interactions wherein, due to "protein folding" sections of the polypeptide that are distant on the linear chain are close together in space resulting in tertiary and quaternary structure. Peptides, on the other hand, typically possess, at most, one or two secondary structural elements (e.g., α-helix, β-sheet or β-turn). Many peptides possess no apparent secondary structural elements at all.

Stabilizing groups are amino acid sequences that can range in size from a single amino acid to a polypeptide (&gt;50 amino acids). Suitable stabilizing groups do not specifically bind to serum proteins (e.g., albumin) or immunoglobulins, and in many embodiments, are free of disulfide bonds. Thus, the stabilizing groups of the present invention directly stabilize the peptides to which they are attached. Stabilizing groups that do not elicit, or elicit only minimal (i.e., clinically acceptable), immune responses in subject mammals are particularly useful. In one embodiment, the stabilizing group is a stable protein, preferably a small stable protein such as thioredoxin, glutathione sulfotransferase, maltose binding protein, glutathione reductase, or a four-helix bundle protein such as Rop protein, as described below, although no specific size limitation on the protein anchor is intended. Proteins suitable for use as stabilizing groups can be either naturally occurring or non-naturally occurring. They can be isolated from an endogenous source, chemically or enzymatically synthesized, or produced using recombinant DNA technology. Proteins that are particularly well-suited for use as stabilizing groups are those that are relatively short in length and form very stable structures in solution. Proteins having molecular weights of less than about 70 kD (e.g., less than about 65, 60, 50, 40, 25, or 12 kD) are useful as stabilizing groups. For example, human serum albumin has a molecular weight of about 64 kD; E. coli thioredoxin has a molecular weight of about 11.7 kD; E. coli glutathione sulfotransferase has a molecular weight of about 22.9 kD; Rop from the ColΕl replicon has a molecular weight of about 7.2 kD; and maltose binding protein (without its signal sequence) has a molecular weight of about 40.7 kD. The small size of the Rop protein makes it especially useful as a stabilizing group, fusion partner, or peptide "anchor", in that it is less likely than larger proteins to interfere with the accessibility of the linked peptide, thus preserving its bioactivity. Rop's highly ordered anti-parallel four-helix bundle topology (after dimerization), slow unfolding kinetics (see, e.g., Betz et al, Biochemistry 36, 2450- 2458 (1997)), and lack of disulfide bonds also contribute to its usefulness as a peptide anchor according to the invention. Other proteins with similar folding kinetics and/or thermodynamic stability (e.g., Rop has a midpoint temperature of denaturation, T_m, of about 71 °C, Steif et al., Biochemistry 32, 3867-3876 ( 1993)) are also preferred peptide anchors.

Peptides or proteins having highly stable tertiary structures, such as a four- helix bundle topology as exemplified in Rop, are particularly useful. Thus, in another embodiment of the screening method of the invention, the expression vector encodes a stabilizing group comprising an α-helical moiety at the N-terminus, C-terminus, or both, of the randomized peptide. The resulting fusion protein is predicted to be more stable than the randomized peptide itself in the host intracellular environment. Suitable α-helical moieties can range from a single α-helix to two, three, four, or five α-helix bundles. Non-limiting examples of single α-helical moieties that can be used to stabilize a bioactive peptide include the following: a 17 amino acid peptide based on the first α-helix of the α-helix/turn/α-helix peptide of Fezoui et al., Proc. Natl. Acad. Sci. USA 91, 3675-3679 (1994) (Asp-Trp-Leu-Lys-Ala-Arg-Val-Glu-Gln-Glu-Leu- Gln-Ala-Leu-Glu-Ala-Arg, SEQ ID NO: 111); an 18 to 36 amino acid peptide containing only glutamic acid, lysine, and glutamine residues, such as (Glu-Lys-Gln)_y where y is 6 to 12, although no specific arrangement of the three amino acids within the repeating tripeptide is intended; a 20 amino acid peptide containing amino acids 14-33 of Neuropeptide Y (Ala-Glu- Asp-Leu- Ala- Arg-Tyr-Tyr-Ser-Ala-Leu-Arg-His- Tyr-Ile-Asn-Leu-Ile-Thr-Arg, SEQ ID NO:112); a 21 amino acid peptide based on amino acids 88-108 of human mannose binding protein (Ala-Ala-Ser-Glu-Arg-Lys- Ala-Leu-Gln-Thr-Glu-Met-Ala-Arg-He-Lys-Lys-Ala-Leu-Thr-Ala, SEQ ID NO:113); a 24 amino acid peptide based on amino acids 4-27 of helodermin (Ala-Ile-Phe-Thr- Glu-Glu-Tyr-Ser-Lys-Leu-Leu-Ala-Lys-Leu-Ala-Leu-Gln-Lys-Tyr-Leu-Ala-Ser-Ile- Leu, SEQ ID NO: 114); and a 34 amino acid peptide based on amino acids 41-74 of ribosomal protein L9 (Pro-Ala-Asn-Leu-Lys-Ala-Leu-Glu-Ala-Gln-Lys-Gln-Lys-Glu- Gln-Arg-Gln-Ala-Ala-Glu-Glu-Leu-Ala-Asn-Ala-Lys-Lys-Leu-Lys-Glu-Gln-Leu- Glu-Lys, SEQ ID NO: 115).

Non-limiting examples of two-helix bundles include two α-helices connected by a turn region (see, for example, the 38 amino acid α-helix/turn/α-helix peptide of Fezoui et al. (1994), supra); a 42 amino acid peptide based on amino acids 512-553 of the adhesion modulation domain (AMD) of α-catenin; a 64 amino acid peptide based on residues 411-475 of α-catenin; and a 78 amino acid a peptide based on residues 24- 102 of seryl-tRNA synthetase from E. coli. Non-limiting examples of three-helix bundles include a peptide based on residues 410-504 of α-catenin, a (Gly-Pro-Pro-)ιo or (-Pro-Pro-Gly)ιo peptide, and an (Ala-Pro-Pro-) ιo or (-Pro-Pro- Ala)ιo peptide.

Two-helix bundles may dimerize and form a four-helix bundle. As indicated above, Rop, which is 63 amino acids in size, forms a two-helix bundle that automatically dimerizes to become a four-helix bundle. Other useful four-helix bundles include the 35 amino acid and 51 amino acid four-helix bundle peptides of Betz et al. (1997) supra, and the 125 amino acid AMD of the α-catenin protein (residues 509-633 of the α-catenin protein).

Where a small stable protein or an α-helical moiety, such as a four-helix bundle protein, is fused to the N-terminus, the randomized peptide can optionally be further stabilized by, for example, covalently linking one or more prolines, with or without a following undefined amino acids (e.g., -Pro, -Pro-Pro, -Pro-Xaa_n, -Pro-Pro- Xaa_n, etc.) at the C-terminus of the peptide sequence, where n is 1 or 2; likewise, when the α-helical moiety is fused to the C-terminus, the randomized peptide can be further stabilized by, for example, covalently linking one or more prolines, with or without a preceding undefined amino acid (e.g., Pro-, Pro-Pro-, Xaa_n-Pro-, Xaa_n-Pro- Pro-, etc.) at the N-terminus of the peptide sequence, where n is 1 or 2, as discussed in more detail below.

In another embodiment of the screening method of the invention, the stabilizing group can constitute one or more proline (Pro) residues. Preferably, a proline dipeptide (Pro-Pro) is used as a stabilizing group, although additional prolines may be included. The encoded proline(s) are typically naturally occurring amino acids. However, if and to the extent a proline derivative, for example a hydroxyproline or a methyl- or ethyl-proline derivative, can be encoded or otherwise incorporated into the peptide, those proline derivatives are also useful as stabilizing groups.

At the N-terminus of the peptide, the stabilizing group also can be an oligopeptide having the sequence Xaa„-Pro_m-, wherein Xaa is any amino acid (e.g., Ala), m is greater than 0, and n is 1 or 2. For example, m can be about 1 to about 5 (e.g., m can be 2 or 3). An oligopeptide having the sequence Xaa_n-Prθ_m-, wherein m = 2, is particularly useful. Likewise, at the C-terminus of the peptide, the stabilizing group can be an oligopeptide having the sequence -Pro_m-Xaa_n, wherein Xaa is any amino acid (e.g., Ala), m is greater than 0, and n is 1 or 2. For example, m can be about 1 to about 5 (e.g., m can be 2 or 3). An oligopeptide having the sequence Pro_m- Xaa_n, wherein m = 2, is particularly useful.

In one embodiment of the screening method of the invention, the nucleic acid sequence that encodes the randomized peptide in the expression vector is modified to encode both a first stabilizing group linked to the N-terminus of the peptide, the first stabilizing group being selected from the group consisting of small stable protein,

Pro-, Pro-Pro-, Xaa_n-Pro-, and Xaa_n-Pro-Pro-, and a second stabilizing group linked to the C-terminus of the peptide, the second stabilizing group being selected from the group consisting of a small stable protein, -Pro, -Pro-Pro, Pro-Xaa_n and Pro-Pro-Xaa_n. The resulting peptide has enhanced stability in the intracellular environment relative to a peptide lacking the terminal stabilizing groups.

In yet another embodiment of the screening method of the invention, the putative bioactive peptide is stabilized by engineering into the peptide a stabilizing motif such as an α-helix motif or an opposite charge ending motif. Chemical synthesis of an oligonucleotide according to the scheme [(CAG)A(TCAG)] yields an oligonucleotide encoding a peptide consisting of a random mixture of the hydrophilic amino acids His, Gin, Asn, Lys, Asp, and Glu (see Table 14). Except for aspartate, these amino acids are most often associated with α-helical secondary structural motifs; the resulting oligonucleotides are thus biased in favor of oligonucleotides that encode peptides that are likely to form α-helices in solution. Alternatively, the putative bioactive peptide is stabilized by flanking a randomized region with a region of uniform charge (e.g., positive charge) on one end and a region of opposite charge (e.g., negative) on the other end, to form an opposite charge -ending motif. To this end, the nucleic acid sequence that encodes the randomized peptide in the expression vector or the expression vector itself is preferably modified to encode a plurality of sequential uniformly charged amino acids at the N-terminus of the peptide, and a plurality of sequential oppositely charged amino acids at the C-terminus of the peptide. The positive charges are supplied by a plurality of positively charged amino acids consisting of lysine, histidine, arginine or a combination thereof; and the negative charges are supplied by a plurality of negatively charged amino acids consisting of aspartate, glutamate or a combination thereof. It is expected that such a peptide will be stabilized by the ionic interaction of the two oppositely charged ends. Preferably, the putative bioactive peptide contains at least three charged amino acids at each end. More preferably, it contains at least four charged amino acids at each end. In a particularly preferred embodiment, the larger acidic amino acid glutamate is paired with the smaller basic amino acid lysine, and the smaller acidic amino acid aspartate is paired with the larger basic amino acid arginine.

The present invention further provides a bioactive peptide containing one or more structural features or motifs selected to enhance the stability of the bioactive peptide in an intracellular environment. For example, a bioactive peptide of the invention can include any stabilizing group as described above in connection with the screening method of the invention. Thus, stabilized bioactive peptides identified using the screening method of the invention are included in the invention. Likewise, both known bioactive peptides and bioactive peptides subsequently discovered, when linked to one or more stabilizing groups as described herein, are also within the scope of the present invention.

Accordingly, the invention provides a bioactive peptide having a stabilizing group at its N-terminus, its C-terminus, or at both termini. The bioactive peptide of the invention includes a bioactive peptide that has been detectably labeled, derivatized, or modified in any manner desired prior to use, provided it contains one or more terminal stabilizing groups as provided herein. For example, a non-stabilizing moiety (e.g., a label) can be attached to either terminus of the bioactive peptide, which terminus may or may not also include a stabilizing group. The stabilized bioactive peptide of the invention can be synthesized enzymatically, chemically, or produced by recombinant genetic engineering, without limitation, as described in more detail below. In any synthetic peptide having a stabilizing group that includes one or more prolines according to the present invention, the proline is preferably a naturally occurring amino acid; alternatively, however, it can be a synthetic derivative of proline, for example a hydroxyproline or a methyl-or ethyl-proline derivative. Accordingly, where the abbreviation "Pro" is used herein in connection with a stabilizing group that is part of a synthetic peptide, it is meant to include proline derivatives in addition to a naturally occurring proline. In a bioactive peptide stabilized at only one terminus (i.e., at either the N-or the C-terminus), the stabilizing group is preferably an α-helical moiety (e.g., four- helix bundle protein such as Rop protein), or one or more proline residues, with or without an undefined amino acid (Xaa). The resulting polypeptide consists essentially of a bioactive peptide and the stabilizing group coupled to one terminus of the bioactive peptide.

A peptide stabilized at both termini can include a first stabilizing group attached to the N-terminus, and a second stabilizing group attached to the C-terminus, where the first and second stabilizing groups are as defined previously in connection with the method for identifying bioactive peptides. The stabilizing group is covalently attached to the peptide (e.g., via a peptide bond).

In one embodiment of the bioactive peptide of the invention, the first stabilizing group is Xaa_n-Prθm-, with or without a preceding undefined amino acid (e.g., Pro-, Pro-Pro-, Xaa_n-Pro-, Xaa_n-Pro-Pro-, etc.), and the second stabilizing group is -Pro_m-Xaa_n, , with or without a following undefined amino acids (e.g., -Pro, -Pro- Pro, -Pro-Xaa_n, -Pro-Pro-Xaa_n, etc.). In another embodiment, the first (N-terminal) stabilizing group is a small stable protein or an α-helical moiety (e.g., a four-helix bundle protein such as Rop protein); and the second (C-terminal) stabilizing group is - Pro_m-Xaa_n or one or more proline residues (e.g., -Pro-Pro). In yet another embodiment, the second (C-terminal) stabilizing group is a small stable protein or an α-helical moiety (e.g., a four-helix bundle protein such as Rop protein) and the first (N-terminal) stabilizing group is Xaa_n-Pro_m- or one or more proline residues.

The invention further provides a peptide stabilized by flanking the amino acid sequence of a bioactive peptide with an opposite charge ending motif, as described herein. Preferably, the resulting stabilized peptide retains at least a portion of the biological activity of the bioactive protein. The stabilized peptide includes a peptide that has been detectably labeled, derivatized, or modified in any manner desired prior to use.

It should be understood that any bioactive peptide, without limitation, can be stabilized according to the invention by attaching a stabilizing group to either or both of the N- and C-termini. Included in the present invention are various antimicrobial peptides, inhibitory peptides, therapeutic peptide drugs, and the like. Non-limiting examples include adrenocorticotropic hormone, bactericidal/permeability-increasing protein (BPI), brain natriuretic peptide, cercropin, endothelin, pentagastrin, scorpion peptides, teriparatide acetate, and all of the peptides listed in Tables 1 and 2, that have been modified at one or both peptide termini to include a stabilizing group as discussed above. Particularly useful bioactive peptides include insulin, glucagon, calcitonin, somatostatin, gonadotrophin, and secretin.

The invention is exemplified by peptides such as Pro-Pro- Asp-Arg-Val-Tyr- He-His-Pro-Phe-His-Ile-Pro-Pro (SEQ ID NO:3) and Glu-Asp-Glu-Asp-Asp-Arg-Val- Tyr-Ile-His-Pro-Phe-His-Ile-Arg-Lys-Arg-Lys (SEQ ID NO:4), wherein the middle nine amino acids (Asp-Arg-Val-Tyr-ue-His-Pro-Phe-His-Ile-; SEQ ID NO:5) constitute the sequence of angiotensin. In embodiments in which the bioactive peptide is a known peptide (e.g., angiotensin), the stabilizing group or groups that are coupled to one or both of the bioactive peptide's termini are not naturally associated with the peptide. In other words, the stabilizing groups that are coupled to the bioactive peptide are heterologous to the bioactive peptide.

In embodiments in which a first stabilizing group is coupled to the N-terminus of a bioactive peptide and a second stabilizing group is coupled to the C-terminus of the bioactive peptide, the first and second stabilizing groups can be the same or different. In some embodiments where the first and second stabilizing groups are different, they can be said to be "heterologous" to each other, i.e., the stabilizing groups have different amino acid sequences and (1) are from different proteins, or (2) are from the same protein, but the stabilizing groups are not contiguous with each other in a naturally occurring protein, or (3) are produced synthetically and one or both of the stabilizing groups do not correspond to a naturally occurring sequence. For example, a bioactive peptide can be coupled to Rop at the N-terminus and coupled to a two-helix bundle from the α-catenin AMD protein on the C-terminus. Where the first and second stabilizing groups are from the same protein, it is not necessary that these flanking groups interact with each other so as to confine or constrain the N-terminus and C-terminus of the flanked peptide in close proximity to one another. For example, Lavallie and others (Bio/Technology 11:187-193 (1993); McCoy et al., U.S. Pat Nos. 5,270,181; 5,292,646; 5,646,016 and 6,143,524) describe internal peptide fusions at the active site loop of thioredoxin, citing the advantages of strong secondary structure in that region, the absence of tertiary structure, and constraint of the peptide at both ends. In the internal thioredoxin fusion, the inserted peptide is bound at each end by cysteine residues, which may form a disulfide linkage and further limit the conformational freedom of the inserted peptide. However, the present work suggests that neither steric constraint of the peptide ends nor any of the other unique characteristics of thioredoxin polypeptides are necessary. Proteins other than thioredoxin, such as Rop, can be effectively used as first and/or second stabilizing groups for bioactive peptides. Neither the first nor the second stabilizing group needs to possess the capacity to participate in the formation of an intramolecular disulfide bond. That is, the stabilizing group does not need to contain a cysteine or, if it does, the remainder of the polypeptide need not contain a cysteine. The stabilizing groups can be selected such that disulfide formation between the first and second stabilizing groups, if it occurs at all, does not bring the N- terminus and C-terminus of the bioactive peptide into close proximity. For example, flanking stabilizing groups can be selected such that they do not contain, within about three residues of the ends that are connected to the ends of the bioactive peptide, cysteine residues that interact with each other to form an intramolecular disulfide bond.

Moreover, in embodiments in which the first and second stabilizing groups are from the same polypeptide, the stabilizing groups need not, although they may, interact to form a naturally occurring secondary or tertiary structure. Naturally occurring secondary structures include, for example, -helices, β-sheets, β-turns and the like that are present in, for example, the native solution or crystal structure of the protein as determined by X-ray crystallography or nuclear magnetic resonance spectroscopy. Naturally occurring tertiary structures result from the "folding" of sections the polypeptide that may be distant on the linear chain such that they are close together in space. Tertiary structure includes the three-dimensional relationships between and among the secondary structures and unstructured portions of the molecule. Modification of a bioactive peptide to yield a stabilized bioactive peptide according to the invention can be achieved by standard techniques well-known in the arts of genetics and peptide synthesis. For example, where the peptide is synthesized de novo, as in solid state peptide synthesis, one or more prolines or other stabilizing groups can be added at the beginning and the end of the peptide chain during the synthetic reaction. In recombinant synthesis, for example as described in Example HI herein, one or more codons encoding proline, or codons encoding α-helical moieties, for example, can be inserted into the peptide coding sequence at the beginning and/or the end of the sequence, as desired. Preferably, codons encoding N-terminal prolines are inserted after (i.e., 3' to) the initiation site ATG (which encodes methionine). Analogous techniques are used to synthesize bioactive peptides having an opposite charge ending motif. When a known bioactive peptide is modified to yield a stabilized bioactive peptide according to the invention, the unmodified peptide can conveniently be used as a control in a protease- or peptidase-resistance assay as described hereinabove to confirm, if desired, that the modified peptide exhibits increased stability.

A stabilized bioactive peptide according to the invention can include a peptide whose bioactivity is evident from or identified in a "phage display" experiment. In "phage display" peptides are displayed on the surface of phage and assayed for bioactivity. Displayed peptides are tethered at the one terminus, typically the C- terminus, to the bacteriophage surface. Their other terminus, typically the N- ter inus, is free (i.e., non-fused). Structurally, the polypeptide produced in a phage display system is typically a fusion polypeptide that contains a peptide of interest at the N-terminus, followed by a phage protein at the C-terminus. hi some phage display polypeptides, however, the order is reversed and the phage protein is at the N- ter inus of the polypeptide and the peptide of interest is at the C-terminus. The phage protein is selected such that the polypeptide is displayed on the surface of the bacteriophage. Examples include bacteriophage proteins pill and pVIII.

Phage display can be used to screen peptide libraries and identify novel bioactive peptides. Bioactive peptides that are active when displayed typically continue to exhibit bioactivity when fabricated synthetically (i.e., without fusion to the phage protein), but they frequently exhibit instability in vivo. This may be due to the fact that the C-terminus is no longer protected or tethered. Hence, the present invention includes a method for stabilizing a bioactive peptide by linking a stabilizing group to the N-terminus or C-terminus of a peptide, when the bioactive peptide has been identified using phage display.

Additionally or alternatively, the genetic constructs used to produce the fusion protein within the bacteriophage can be engineered encode a stabilizing group at the terminus of encoded fusion polypeptide that would otherwise have been free, such that the fusion polypeptide displayed on the surface of the bacteriophage contains a peptide of interest flanked by a stabilizing group at the terminus and a bacteriophage protein at the other terminus.

The invention thus includes methods for phage display of stabilized bioactive proteins, methods for stabilizing bioactive peptides identified using phage display, and bioactive peptides thus identified and/or stabilized.

The present invention also provides a cleavable polypeptide comprising a stabilized, bioactive peptide either immediately preceded by (i.e., adjacent to the N- terminus of the bioactive peptide) a cleavage site, or immediately followed by (i.e., adjacent to the C-terminus of the bioactive peptide) a cleavage site. Thus, a bioactive peptide as contemplated by the invention can be part of a cleavable polypeptide. The cleavable polypeptide is cleavable, either chemically, as with cyanogen bromide, or enzymatically, to yield the bioactive peptide. The resulting bioactive peptide either includes a first stabilizing group positioned at its N-terminus and/or a second stabilizing group positioned at its C-terminus, both as described hereinabove. The cleavage site immediately precedes the N-terminal stabilizing group or immediately follows the C-terminal stabilizing group. In the case of a bioactive peptide stabilized with an opposite charge ending motif, the cleavage site immediately precedes the first charged region or immediately follows the second charged region. The cleavage site makes it possible to administer a bioactive peptide in a form that could allow intracellular targeting and/or activation.

Alternatively, a bioactive peptide of the invention can be fused to a noncleavable N-terminal or C-terminal targeting sequence wherein the targeting sequence allows targeted delivery of the bioactive peptide, e.g., intracellular targeting or tissue-specific targeting of the bioactive peptide. In one embodiment of this aspect of the invention, a stabilizing group (e.g., one or more proline residues) is positioned at the free (i.e., non-fused) terminus of the bioactive peptide as described hereinabove in connection with the screening method for identifying bioactive peptides. The targeting sequence attached to the other peptide terminus can, but need not, contain a small stable protein such as Rop or one or more proline residues, as long as the targeting function of the targeting sequence is preserved. In another embodiment of this aspect of the invention, the bioactive peptide is stabilized with a charge ending motif as described hereinabove, wherein one charged region is coupled to the free terminus of the bioactive peptide, and the other charged region is disposed between the targeting sequence and the active sequence of the bioactive peptide.

The invention further includes a method for using an antimicrobial peptide that includes covalently linking a stabilizing group, as described above, to the N- terminus, the C-terminus, or to both termini, to yield a stabilized antimicrobial peptide, then contacting a microbe with the stabilized antimicrobial peptide. Alternatively, the stabilized antimicrobial peptide used in this aspect of the invention is made by covalently linking oppositely charged regions, as described above, to each end of the antimicrobial peptide to form an opposite charge ending motif. An antimicrobial peptide is to be broadly understood as including any bioactive peptide that adversely affects a microbe such as a bacterium, virus, protozoan, or the like, as described in more detail above. An example of an antimicrobial peptide is an inhibitory peptide that inhibits the growth of a microbe. When the antimicrobial peptide is covalently linked to a stabilizing group at only one peptide terminus, any of the stabilizing groups described hereinabove can be utilized. When the antimicrobial peptide is covalently linked to a stabilizing group at both peptide termini, the method includes covalently linking a first stabilizing group to the N terminus of the antimicrobial peptide and a second stabilizing group to the C terminus of the antimicrobial peptide, where the first and second stabilizing groups are as defined previously in connection with the method for identifying bioactive peptides. In a preferred embodiment of the method for using an antimicrobial peptide, one or more prolines, more preferably a Pro-Pro dipeptide, is attached to at least one, preferably both, termini of the antimicrobial peptide. Alternatively, or in addition, an Xaa_n-Pro_m- sequence, as described above, can be attached to the N-terminus of a microbial peptide, and/or a -Pro_m-Xaa_n sequence can be attached to the C-terminus, to yield a stabilized antimicrobial peptide.

The antimicrobial peptide thus modified in accordance with the invention has enhanced stability in the intracellular environment relative to an unmodified antimicrobial peptide. As noted earlier, the unmodified peptide can conveniently be used as a control in a protease- or peptidase-resistance assay as described hereinabove to confirm, if desired, that the modified peptide exhibits increased stability. Further, the antimicrobial activity of the antimicrobial peptide is preferably preserved or enhanced in the modified antimicrobial peptide; modifications that reduce or eliminate the antimicrobial activity of the antimicrobial peptide are easily detected and are to be avoided.

The invention further provides a method for inhibiting the growth of a microbe comprising contacting the microbe with a stabilized inhibitory peptide. As described above, the stabilized inhibitory peptide can have a stabilizing group attached at its N- terminus, C-terminus, or both termini.

Also included in the present invention is a method for treating a patient having a condition treatable with a peptide drug, comprising administering to the patient a peptide drug that has been stabilized as described herein. Peptide drugs for use in therapeutic treatments are well known (see, e.g., Table 1). However, they are often easily degraded in biological systems, which affects their efficacy. In one embodiment of the present method, the patient is treated with a stabilized drug comprising the peptide drug of choice and a stabilizing group linked to either the N- terminus, the C-terminus of, or to both termini of the peptide drug. In another embodiment of the present method, the patient is treated with a stabilized drug comprising the peptide drug of choice that has been stabilized by attachment of oppositely charged regions to both termini of the peptide drug. Because the peptide drug is thereby stabilized against proteolytic degradation, greater amounts of the drug should reach the intended target in the patient. In embodiments of the method involving administration of a peptide drug that is covalently linked to a stabilizing group at only one peptide terminus, the stabilizing group is preferably an α-helical moiety, such as a four-helix bundle protein (e.g., Rop), provided that attachment of the α-helical moiety to the peptide terminus preserves a sufficient amount of efficacy for the drug. It is to be nonetheless understood that the group or groups used to stabilize the peptide drug are as defined hereinabove, without limitation. In embodiments involving administration of a peptide drug covalently linked to a stabilizing group at both peptide termini, the peptide drug includes a first stabilizing group linked to the N-terminus of the peptide drug and a second stabilizing group linked to the C-terminus of the peptide drug. Thus, in another preferred embodiment of the treatment method of the invention, the stabilized peptide drug includes one or more prolines, more preferably a proline- proline dipeptide, attached to one or both termini of the peptide drug. For example, the peptide drug can be stabilized by covalent attachment of a Rop protein at one terminus, and by covalent attachment of a proline or proline dipeptide at the other terminus; in another preferred embodiment, the peptide drug can be stabilized by proline dipeptides at each of the N-terminus and C terminus. Alternatively, or in addition, the stabilized peptide drug used in the treatment method can include an Xaa_n-Pro_m- sequence at the N-terminus of the peptide drug, and/or a -Pro_m-Xaa_n sequence at the C-terminus. Optionally, prior to administering the stabilized peptide drug, the treatment method can include covalently linking a stabilizing group to one or both termini of the peptide drug to yield the stabilized peptide drug.

If desired, the unmodified peptide drug can conveniently be used as a control in a protease- or peptidase-resistance assay as described hereinabove to confirm that the stabilized peptide drug exhibits increased stability. Further, the therapeutic efficacy of the peptide drug is preferably preserved or enhanced in the stabilized peptide drug; modifications that reduce or eliminate the therapeutic efficacy of the peptide drug are easily detected and are to be avoided.

The present invention further includes a fusion protein comprising a four-helix bundle protein, preferably Rop protein, and a polypeptide. Preferably the polypeptide is bioactive; more preferably it is a bioactive peptide. The fusion protein of the invention can be used in any convenient expression vector known in the art for expression or overexpression of a peptide or protein of interest. Optionally, a cleavage site is present between the four-helix bundle protein and the polypeptide to allow cleavage, isolation and purification of the polypeptide. In one embodiment of the fusion protein, the four-helix bundle protein is covalently linked at its C-terminus to the N-terminus of the polypeptide; in an alternative embodiment, the four-helix bundle protein is covalently linked at its N-terminus to the C-terminus of the polypeptide. Fusion proteins of the invention, and expression vectors comprising nucleic acid sequences encoding fusion proteins wherein the nucleic acid sequences are operably linked to a regulatory control element such as a promoter, are useful for producing or overproducing any peptide or protein of interest.

EXAMPLES The present invention is illustrated by the following examples. It is to be understood that the particular examples, materials, amounts, and procedures are to be interpreted broadly in accordance with the scope and spirit of the invention as set forth herein. Example I Construction and characterization of a highly regulable expression vector, pLACll, and its multipurpose derivatives, pLAC22 and pLAC33 A number of different expression vectors have been developed over the years to facilitate the production of proteins in E. coli and related bacteria. Most of the routinely employed expression vectors rely on lac control in order to overproduce a gene of choice. While these vectors allow for overexpression of the gene product of interest, they are leaky due to changes that have been introduced into the lac control region and gene expression can never be shut off under repressed conditions, as described in more detail below. Numerous researchers have noticed this problem with the more popular expression vectors pKK223-3 (G. Posfai et al. Gene. 50: 63-67 (1986); N.Scrutton et al., Biochem J.245: 875-880 (1987)), pKK233-2 (P. Beremand et al., Arch Biochem Biophys. 256: 90-100 (1987); K. Ooki et al., Biochemie. 76: 398-403 (1994)), pTrc99A (S. Ghosh, Protein Expr. Purif. 10: 100-106 (1997); J.

Ranie et al., Mol. Biochem. Parasitol. 61: 159-169 (1993)), as well as the PET series (M. Eren et al., J. Biol. Chem. 264: 14874-14879 (1989); G. Godson, Gene 100: 59- 64 (1991)).

The expression vector described in this example, pLACll, was designed to be more regulable and thus more tightly repressible when grown under repressed conditions. This allows better regulation of cloned genes in order to conduct physiological experiments. pLACl 1 can be used to conduct physiologically relevant studies in which the cloned gene is expressed at levels equal to that obtainable from the chromosomal copy of the gene in question. The expression vectors described here were designed utilizing the wild-type lac promoter/operator in order to accomplish this purpose and include all of the lac control region, without modification, that is contained between the start of the O3 auxiliary operator through the end of the OI operator. As with all lac based vectors, the pLACl 1 expression vector described herein can be turned on or off by the presence or absence of the gratuitous inducer IPTG. In experiments in which a bacterial cell contained both a null allele in the chromosome and a second copy of the wild-type allele on pLACl 1 cells grown under repressed conditions exhibited the null phenotype while cells grown under induced conditions exhibited the wild-type phenotype. Thus the pLACll vector truly allows for the gene of interest to be grown under either completely repressed or fully induced conditions. Two multipurpose derivatives of pLACll, pLAC22 and pLAC33 were also constructed to fulfill different experimental needs.

The vectors pLACll, pLAC22 and pLAC33 w ere deposited with the American Type Culture Collection (ATCC), 10801 University Blvd., Manassas, VA, 20110-2209, USA, on February 16,1999, and assigned ATCC deposit numbers ATCC 207108, ATCC 207110 and ATCC 207109, respectively. It is nonetheless to be understood that the written description herein is considered sufficient to enable one skilled in the art to fully practice the present invention. Moreover, the deposited embodiment is intended as a single illustration of one aspect of the invention and is not to be construed as limiting the scope of the claims in any way.

MATERIALS AND METHODS

Media. Minimal M9 media (6 g disodium phosphate, 3 g potassium phosphate, 1 g ammonium chloride, 0.5 g sodium chloride, distilled water to IL; autoclave; add 1 mL m magnesium sulfate (IM) and 0.1 mL calcium chloride (IM); a sugar added to a final concentration of 0.2 %; vitamins and amino acids as required for non-prototrophic strains) and rich LB media (10 g tryptone, 5 g yeast extract, 10 g sodium chloride, distilled water to IL; autoclave) were prepared as described by Miller (J. Miller, "Experiments in molecular genetics" Cold Spring Harbor Laboratory, Cold Spring Harbor, NY (1972). The antibiotics ampicillin, kanamycin, streptomycin, and tetracycline (Sigma Chemical Company, St. Louis, MO) were used in rich media at a final concentration of 100, 40, 200, and 20 ug/ml, respectively. When used in minimal media, tetracycline was added at a final concentration of 10 ϊg/ml. 5-bromo-4-chloro-3-indoyl β-D-galactopyranoside (Xgal) was added to media at a final concentration of 40 ϊg/ml unless otherwise noted. IPTG was added to media at a final concentration of 1 mM.

Chemicals and Reagents. When amplified DNA was used to construct the plasmids that were generated in this study, the PCR reaction was carried out using native Pfu polymerase from Stratagene (Cat. No. 600135). Xgal and IPTG were purchased from Diagnostic Chemicals Limited.

Bacterial Strains and Plasmids. Bacterial strains and plasmids are listed in Table 4. To construct ALS225, ALS224 was mated with ALS216 and streptomycin resistant, blue recombinants were selected on a Rich LB plat that contained streptomycin, Xgal, and IPTG. To constract ALS226, ALS224 was mated with ALS217 and streptomycin resistant, kanomycin resistant recombinants were selected on a Rich LB plate that contained streptomycin and kanamycin. To construct ALSS15, ALSS14 was mated with ALS216 and streptomycin resistant, blue recombinants were selected on a Rich LB plate that contained streptomycin, Xgal, and IPTG. To construct ALS527, ALS524 was mated with ALS224 and streptomycin resistant, tetracycline resistant recombinants were selected on a Rich LB plate that contained streptomycin and tetracycline. To construct ALS535, ALS533 was mated with ALS498 and tetracycline resistant recombinants were selected on a Minimal M9 Glucose plate that contained tetracycline, leucine and thiamine (Bj) (Sigma Chemical Company). To constract ALS533, a PI lysate prepared from E. coli strain K5076 (H. Miller et al., Cell 20: 711-719 (1980)) was used to transduce ALS224 and tetracycline resistant transductants were selected.

Table 4: Bacterial strains and plasmids used in Example I

E. coli Strains

<img file="WO2004011485A2_D0004.tif" /> Plasmids

<img file="WO2004011485A2_D0005.tif" />

Construction of the pLACll, pLAC22, and pLAC33 expression vectors.

To construct pLACl 1, primers #1 and #2 (see Table 5) were used to PCR amplify a 952 base pair (bp fragment from the plasmid pBH20 which contains the wild-type lac operon. Primer #2 introduced two different base pair mutations into the seven base spacer region between the Shine Dalgarno site and the ATG start site of the lacZ which converted it from AACAGCT to AAGATCT thus placing a Bgl II site in between the Shine Dalgarno and the start codon of the lacZ gene. The resulting fragment was gel isolated, digested with Pst I and EcoR I, and then ligated into the 3614 bp fragment from the plasmid pBR322ΔAvaI which had been digested with the same two restriction enzymes. To construct pBR322ΔAval , pBr322 was digested with Aval, filled in using Klenow, and then religated. To constract pLAC22, a 1291 bp Nco I. EcoR I fragment was gel isolated from pLAC21 and ligated to a 4361 bp Nco I. EcoR I fragment which was gel isolated frompBR322/NcoI. To constract pLAC21, primers #2 and #3 (see Table 5) were used to PCR amplify a 1310 bp fragment from the plasmid pMS421 which contains the wild-type lac operon as well as the lacH repressor. The resulting fragment was gel isolated, digested with EcoR I, and then ligated into pBR322 which had also been digested with EcoR I. To construct pBR322/Nco I, primers #4 and #5 (see Table 5) were used to PCR amplify a 788 bp fragment from the plasmid pBR322. The resulting fragment was gel isolated, digested with Pst I and EcoR I, and then ligated into the 3606 bp fragment from the plasmid pBR322 which had been digested with the same two restriction enzymes.

The pBR322Mco I vector also contains added Kpn 1 and Sma I sites in addition to the new Nco I site. To construct pLAC33, a 2778 bp fragment was gel isolated from pLAC12 which had been digested with BsaB I and Bsa I and ligated to a 960 bp fragment from pUC8 which had been digested with Afl III, filled in with Klenow, and then digested with Bsa I. To constract pLAC12, a 1310 bp Pst I, BamH I fragment was gel isolated from pLACll and ligated to a 3232 bp Pst I, BamH I fragment which was gel isolated from pBR322.

Table S. Prime-¹-, employed to PCft ai-ip iffr PiNA l-ragmenfe Hat were g*e In the construct, on or the various plasmids dess-ribrf in liaΛtpis.1

pl*fi»l an pM£--Z j (*ori j-ttι...QEa.A33-,j!-≤a:,i-sa.-S-fi,&amp;τ ^SEO. m sa.≤) a trβv) ATS ≤aa m ΛJΓA M^ aaa; TS..cs„-s^.-fl-aB,.%xp.-.-.ra-^i-2 cac .ssα iB a-^3

3 ⁽gojr, &amp;τr SM 2m vx: -33-* S5,^q?„^«,«„. ^..: e... e.-Mii.₁A <img file="WO2004011485A2_D0006.tif" />

4 .tor. fi .srr-,sse,,A.τι,_&gt;.6≤-ι-₁β-y-,,≤ issa as aosBi s ,».¥* st -is j&amp;-ϊ a≤c das sm sea, 333313&amp; Aas„- ;o.j.a&amp;..j33a

BSΪ.11 - lasS - Hli-d III

&lt;s 4 --at.} rø-e » &amp; Λ£-q lιi ∞..OTJiπ„ ffi..&lt;-δT.ιc-...cϊβ -sEg-

133 IΛ.I13

^■7 &lt;3TBvi TA Jrø MS SEE fi G.,f-3S-,-ΩCq; ,Gqϊ-,-ιaτ.-røϊ_.5as ϊβf .SBC

--» NOiH-i)

Fat 2 - laea - BiπA III a ^{£03,⁾ tAT CAT E5-3 -afi Λ©3 AM. O-G CϊA<img file="WO2004011485A2_D0007.tif" />

---3-£j-_!-C„:K5 -SE8 ID HO i 13

9 (jtev) T-w j-ϊA at (a-e e-ns-- tii-a, ags p-sa.-QCC_-ι-3c--ra-a.jsι:s a'π:

AΛ-t-Aϊt_-πF f≤EO ∑D HD;l.S

BfarøH 1 - lacZ - Bind III ial_5© Hs&amp;e pri-mer #3) iό &lt;forl SA Gftr ≤2&amp; .£££ Ara A&amp;J. ratα cr-i.-3S----- -L^ϊ-^-- a,4ϊ-?-. a-βC_Λ2-C_-rø .SWQ XO SPU5.

Døl E-£ - -t*-?A ~ J-ϊ-ιά ϊtϊ

HI &lt;£Dar]| - TAC T3.T i «EK ft-IS„GCX..AIC..(- C.,S-a,^ ----&amp;-i», K-i ISEQ

ED SO -16,

12 {tt-vj -WEλ. Tffif S SUE <img file="WO2004011485A2_D0008.tif" />

O I EQ J-D WO! XT)

Banes 1 - j-ylJs - EQfeft 1 i3 &lt;&amp;-*^•, τ-5- TOT isaa aaae <img file="WO2004011485A2_D0009.tif" /> ism ID HO.I&amp;J&gt;

14 &lt;j-?ev1 ATT AGT ≤M 3--1-- GC^-GftA-..TCX-.≤-Tfl- J-AS.- a- aτc_i-3. .gJBp

ΪJ -fe.as) In Table 5 the regions of the primers that are homologous to the DNA target template are indicated with a dotted underline, while the relevant restriction sites are indicated with a solid underline. All primers are listed in the 5' — 3' orientation.

Compilation of the DNA sequences for the pLACll, pLAC22, and pLAC33 expression vectors. All of the DNA that is contained in the pLACl 1, pLAC22, and, pLAC33 vectors has been sequenced.

The sequence for the pLACll vector, which is 4547 bp, can be compiled as follows: bp 1-15 are AGATCTTATGAATTC (SEQ ID NO: 20) from primer #2 (Table 5); bp 16-1434 are bp 4-1422 from ρBR322 (GenBank Accession #JOl 749); bp 1435-1442 are TCGGTCGG, caused by filling in the Ava I site in pBR322AAvaI; bp 1443-4375 is bp 1427-4359 from pBR322 (GenBank Accession #JO1749); and bp 4376-4547 are bp 1106-1277 from the wild- type E. coli lac operon (GenBank Accession #J01636). The sequence for the pLAC22 vector which is 5652 bp can be compiled as follows: bp 1-15 are AGATCTTATGAA TTC (SEQ ID NO: 21) from primer #2 (Table 5); bp 16-4370 are bp 4-4358 from ρBR322 (GenBank Accession* J01749); bp 4371-4376 is CCATGG which is the Nco I site from pBR322/Nco I; and bp 4377- 5652 are bp 2-1277 from the wild-type E. coli lac operon (GenBank Accession #J01636), except that bp #4391 of the pLAC22 sequence or bp#16 from the wild-type E. coli lac operon sequence has been changed from a "C" to a "T" to reflect the presence of the lacV mutation (L Brosius et al., Proc. Natl. Acad. Sci. USA. 81: 6929-6933 (1984)).

The sequence for the pLAC33 vector which is 3742 bp can be compiled as follows: bp 1-15 is AGATCTTATGAATTC (SEQ.ID NO: 22) from primer #2 (Table 5); bp 16-1684 are bp 4-1672 from pEk322 (GenBank Accession # 501749); bp 1685- 2638 are bp 786-1739 from pUC8 (GenBank Accession #L09132); bp 2639-3570 are bp 3428-4359 from pBR322 (GenBank Accession #30 J01749); and bp 357 1-3742 are bp 1106-1277 from the wild-type E. coli lac operon (GenBank Accession #J01636). In the maps for these vectors, the ori is identified as per Balbas (P. Balbas et al., Gene. 50: 3-40 (1986)), while the lacV is indicated starting with the O3 auxiliary operatic and ending with the OI operator as per Muller-Hill (B. Muller-Hill, The lac Operon: A Short History of a Genetic Paradigm. Walter de Gruyter, Berlin, Germany (1996)). Construction of the pLACll-, pLAC22-, pLAC33-, pKK223-3-, pKK233- 2-, pTrc99A-, and pET-21(+)- lacZ constructs. To construct pLACl 1-lacZ, pLAC22- lacZ, and pLAC33- lacZ, primers #6 and #7 (see Table 5) were used to PCR amplify a 3115 bp fragment from the plasmid pTer7 which contains the wild- type lacZ gene. The resulting fragment was gel isolated, digested with Bgl II and Hind UI, and then ligated into the pLACl 1, pLAC22 or pLAC33 vectors that had been digested with the same two restriction enzymes. To construct pKK223-3-/αcZ and pKK233-2- lacZ, primers #8 and #9 (see Table 5) were used to PCR amplify a 3137 bp fragment from the plasmid pTer7. The resulting fragment was gel isolated, digested with Pst I and Hind HI, and then ligated into the pKK223-3 or pKK233-2 vectors which had been digested with the same two restriction enzymes. To construct pTrc99A- lacZ and pET-21 (+)-lacZ, primers #9 and #10 (see Table 5) were used to PCR amplify a 3137 bp fragment fro m the plasmid pTer7. The resulting fragment was gel isolated, digested with BamH I and Hind III, and then ligated into the pTrc99A or pET-21(+) vectors which had been digested with the same two restriction enzymes.

Construction of the pLACll-recA and xylE constructs. To construct pLACl 1-recA, primers #11 and #12 (see Table 5) were used to PCR amplify a 1085 bp fragment from the plasmid pGE226 which contains the wild-type recA gene. The resulting fragment was gel isolated, digested with Bgl II and Hind III, and then ligated into the pLACl 1 vector which had been digested with the same two restriction enzymes. To construct pLACl 1-xylE, primers #13 and #14 (see Table 5) were used to PCR amplify a 979 bp fragment from the plasmid pXE60 which contains the wild- type Pseudomonas putida xylE gene isolated from the TOL pWWO plasmid. The resulting fragment was gel isolated, digested with Bgl II and EcoR I, and then ligated into the pLACl 1 vector which had been digested with the same two restriction enzymes.

Assays, β-galactosidase assays were performed as described by Miller (J.

Miller. "Experiments in molecular genetics," Cold Spring Harbor Laboratory, Cold Spring Harbor, NY (1972)), while catechol 2,3-dioxygenase (catO₂ase) assays were performed as described by Zukowski, et. al. (M. Zukowski et al., Proc. Natl. Acad. Sci. U.S.A. 80: 1101-1105 (1983)). RESULTS

Construction and features of pLACll, pLAC22, and pLAC33. Plasmid maps that indicate the unique restriction sites, drug resistances, origin of replication, and other relevant regions that are contained in pLAC 11, pLAC22, and pLAC33 are shown in Figs. 2, 3 and 4, respectively. pLACll was designed to be the most tightly regulable of these vectors. It utilizes the ColEl origin of replication from pBR322 and Lacl repressor is provided in trans from either an episome or another compatible plasmid. pLAC22 is very similar to pLACll, however, it also contains lacl thus a source of Lacl does not have to be provided in trans. pLAC33 is a derivative of pLACll which utilizes the mutated ColEl origin of replication from pUC8 (S. Lin- Chao et al., Mol. Micro. 6: 3385-3393 (1992))and thus pLAC33's copy number is significantly higher than pLACll and is comparable to that of other pUC vectors. Because the cloning regions of these three vectors are identical, cloned genes can be trivially shuffled between and among them depending on the expression demands of the experiment in question.

To clone into pLAC 11, pLAC22, or pLAC33, PCR amplification is performed with primers that are designed to introduce unique restriction sites just upstream and downstream of the gene of interest. Usually a Bgl II site is introduced immediately in front of the ATG start codon and an EcoR I site is introduced immediately following the stop codon. An additional 6 bases is added to both ends of the oligonucleotide in order to ensure that complete digestion of the amplified PCR product will occur. After amplification the double-stranded (ds) DNA is restricted with Bgl II and EcoR I, and cloned into the vector which has also been restricted with the same two enzymes. If the gene of interest contains a Blgll site, then BamH I or Bel I can be used instead since they generate overhangs which are compatible with Bgl II. If the gene of interest contains an EcoR I site, then a site downstream of EcoR I in the vector (such as Hind III) can be substituted.

Comparison of pLACll, pLAC22, and pLAC33, to other expression vectors. In order to demonstrate how regulable the pLAC 11, pLAC22, and pLAC33 expression vectors were, the wild-type lacZ gene was cloned into pLACll, pLAC22, pLAC33, pKK223-3, pKK233-2, pTrc99A, and pET-21(+). Constructs which required an extraneous source of Lacl for their repression were transformed into ALS225, while constructs which contained a source of Lacl on the vector were transformed into ALS224. pET-21(+)constructs were transformed into BL21 because they require T7 RNA polymerase for their expression. Four clones were chosen for each of these seven constructs and β-galactosidase assays were performed under repressed and induced conditions. Rich Amp overnights were diluted 1 to 200 in either Rich Amp Glucose or Rich Amp IPTG media and grown until they reached mid-log (OD₅5₀ =0.5). In the case of PET-21(+) the pLysE and pLysS plasmids, which make T7 lysozyme and thus lower the amount of available T7 polymerase, were also transformed into each of the constructs. Table 6 shows the results of these studies and also lists the induction ratio that was determined for each of the expression vectors. As the data clearly indicate, pLAC 11 is the most regulable of these expression vectors and its induction ratio is close to that which can be achieved with the wild- type lac operon. The vector which yielded the lowest level of expression under repressed conditions was pLAC 11, while the vector which yielded the highest level of expression under induced conditions was pLAC33.

Table 6: β-galactosidase levels obtained in different expression vectors grown under either repressed or induced conditions

<img file="WO2004011485A2_D0010.tif" />

The average values obtained for the four clones that were tested from each vector are listed in the table. Standard deviation is not shown but was less than 5% in each case. Induction ratios are expressed as the ratio of enzymatic activity observed at fully induced conditions versus fully repressed conditions. The plasmid pLysE yielded unexpected results; it was expected to cause lower amounts of lacZ to be expressed from pET-21(+) under repressed conditions and, instead, higher amounts were observed. As a result, both pLysE and pLysS were restriction mapped to make sure that they were correct.

Demonstrating that pLACll constructs can be tightly regulated. pLACll was designed to provide researchers with an expression vector that could be utilized to conduct physiological experiments in which a cloned gene is studied under completely repressed conditions where it is off or partially induced conditions where it is expressed at physiologically relevant levels. Figure 5 demonstrates how a pLAC 11 -lacZ construct can be utilized to mimic chromosomally expressed lacZ that occurs under various physiological conditions by varying the amount of IPTG inducer that is added. ALS226 cells containing pLACll-ZacZ were grown to mid-log in rich media that contained varying amounts of IPTG and then β-galactosidase activity was assayed. Also indicated in the graph are the average β-galactosidase activities obtained for strains with a single chromosomal copy of the wild-type lacZ gene that were grown under different conditions.

To demonstrate just how regulable pLACll is, the recA gene was cloned into the pLACll vector and transformed into cells which contained a null recA allele in the chromosome. As the results in Table 7 clearly show, recombination cannot occur in a host strain which contains a nonfunctional RecA protein and thus PI lysates which provide a TnlOdKan transposon cannot be used to transduce the strain to Kan^R at a high frequency. recA cells which also contain the pLACll-recA construct can be transduced to an^R at a high frequency when grown under induced conditions but cannot be transduced to Kan^R when grown under repressed conditions.

Table 7: The recombination (-) phenotype of a recA null mutant strain can be preserved with a pLACll-recA (wild-type) construct under repressed conditions

<img file="WO2004011485A2_D0011.tif" />

The data presented in Table 7 are the number of Kan transductants that were obtained from the different MCI 061 derivative strains when they were transduced with a PI lysate prepared from strain ALS598 which harbored a TnlOdKan transposon insertion. Overnights were prepared from each of these strains using either rich medium to which glucose was added at a final concentration of 0.2 (repressed conditions) or rich medium to which IPTG was added at a final concentration of 1 mM (induced conditions). The overnights were then diluted 1 to 10 into the same medium which contained CaCl₂ added to a final concentration of 10 mM and aerated for two hours to make them competent for transduction with PI phage. Cells were then spectrophotometrically normalized and aliquots of 5 OD₅₅₀ cell equivalents in a volume of approximately 0.1 ml were transduced with 0.1 ml of concentrated PI lysate as well as 0.1 ml of PI lysates that had been diluted to 10^"1, 10^"2, or 10^"3. 0.2 ml of 0.1 M Sodium Citrate was added to the cell/phage mixtures and 0.2 ml of the final mixtures were plated onto Rich Kanamycin plates and incubated overnight at 37°C. The total number of Kan^R colonies were then counted. ALS225 recA⁺ data points were taken from the transductions which used the 10^"3 diluted phage, while ALS514 recA data points were taken from the transductions which used the concentrated phage. The data points for ALS515 recA^' pCyt-3-recA grown under repressed conditions were taken from the transductions which used the concentrated phage, while the data points for ALS515 recA^' pCyt-3-recA grown under induced conditions were taken from the transductions which used the 10^"3 diluted phage.

Testing various sources of Lacl for trans repression of pLACll. Because pLACl 1 was designed to be used with an extraneous source of Lacl repressor, different episomal or plasmid sources of Lacl which are routinely employed by researchers were tested. Since one of the Lacl sources also contained the lacZ gene, a reporter construct other than pLACl 1-lacZ was required and thus a pLACll- /E construct was engineered. Table 8 shows the results of this study.

All of the Lacl sources that were tested proved to be adequate to repress expression from pLACll, however, some were better than others. The basal level of expression that was observed with F's which provided lacl - or with the plasmid pMS421 which provided lacIQ at approximately six copies per cell was lower than the basal level of expression that was observed with F's which provided lacVλ all three times that the assay was run. Unfortunately, however, the xylE gene could not be induced as high when lacl - on a F' or lacV on a plasmid was used as the source of Lac repressor.

Table 8: Catechol 2,3-dioxygenase levels obtained for a pLACll--ι. /E construct when Lac repressor is provided by various sources

<img file="WO2004011485A2_D0012.tif" />

The wild-type xylE gene was cloned into the pLACll vector and the resulting pLACll-xy/E construct was then transformed into each of the MCI 061 derivative strains listed in the table. Rich overnights were diluted 1 to 200 in either Rich

Glucose or Rich IPTG media and grown until they reached mid-log (OD₅₅₀ = 0.5). Cell extracts were then prepared and catechol 2,3-dioxygenase assays were performed as described by Zukowski, et. al. (Proc. Natl. Acad. Sci. U. S. A. 80:1101-1105 (1983)). The average values obtained in three different experiments are listed in the table. Standard deviation is not shown but was less 15 than 10% in each case.

DISCUSSION Most of the routinely employed expression vectors rely on lac control in order to overproduce a gene of choice. The lac promoter/operator functions as it does due to the interplay of three main components. First, the wild-type lac-10 region (TATGTT) is very weak. c-AMP activated CAP protein is able to bind to the CAP site just upstream of the -35 region which stimulates binding of RNA polymerase to the weak -10 site. Repression of the lac promoter is observed when glucose is the main carbon source because very little c-AMP is present which results in low amounts of available c-AMP activated CAP protein. When poor carbon sources such as lactose or glycerol are used, c-AMP levels rise and large amounts of c-AMP activated CAP protein become available. Thus induction of the lac promoter can occur. Second, Lac repressor binds to the lac operator. Lac repressor can be overcome by allolactose which is a natural byproduct of lactose utilization in the cell, or by the gratuitous inducer, IPTG. Third, the lac operator can form stable loop structures which prevents the initiation of transcription due to the interaction of the Lac repressor with the lac operator (OI) and one of two auxiliary operators, O2 which is located downstream in the coding region of the lacZ gene, or O3 which is located just upstream of the CAP binding site.

While binding of Lac repressor to the lac operator is the major effector of lac regulation, the other two components are not dispensable. However, most of the routinely used lac regulable vectors either contain mutations or deletions which alter the affect of the other two components. The pKK223-3 (J. Brosius et al., Proc. Natl. Acad. Sci. USA. 81:6929-6933 (1984)), pKK233-2 (E. Amann et al., Gene. 40:183- 190 (1985)), pTrc99A (E. Amann et al., Gene. 69:301-315 (1988)), and pET family of vectors (F. Studier, Method Enzymol. 185:60-89 (1990)) contain only the lac operator (OI) and lack both the CAP binding site as well as the O3 auxiliary operator. pKK223-3, pKK233-2, and pTrc99 use a trp-lac hybrid promoter that contains the trp-35 region and the lac\JV5 -10 region which contains a strong TATAAT site instead of the weak TATGTT site. The pET family of vectors use the strong T7 promoter. Given this information, perhaps it is not so surprising researchers have found it is not possible to tightly shut off genes that are cloned into these vectors. The purpose of the studies described in Example I was to design a vector which would allow researchers to better regulate their cloned genes in order to conduct physiological experiments. The expression vectors described herein were designed utilizing the wild-type lac promoter/operator in order to accomplish this purpose and include all of the lac control region, without modification, that is contained between the start of the O3 auxiliary operator through the end of the OI operator. As with all lac based vectors, the pLACll, pLAC22, and pLAC33 expression vectors can be turned on or off by the presence or absence of the gratuitous inducer IPTG. Because the new vector, pLACl 1 , relies on the wild-type lac control region from the auxiliary lac 03 operator through the lac OI operator, it can be more tightly regulated than the other available expression vectors. In direct comparison studies with pKK223-3, pKK233-2, pTrc99A, and pET-21(+), the lowest level of expression under repressed conditions was achievable with the pLACl 1 expression vector. Under fully induced conditions, pLACl 1 expressed lacZ protein that was comparable to the levels achievable with the other expression vectors. Induction ratios of lOOOx have been observed with the wild-type lac operon. Of all the expression vectors that were tested, only pLACll yielded induction ratios which were comparable to what has been observed with the wild-type lac operon. It should be noted that the regulation achievable by pLACl 1 may be even better than the data in Table 6 indicates. Because lacZ was used in this test, the auxiliary lac O operator which resides in the coding region of the lacZ gene was provided to the pKK223-3, pKK233-2, pTrc99A, and pET-21(+) vectors which do not normally contain either the O2 or O3 auxiliary operators. Thus the repressed states that were observed in the study in Table 6 are probably lower than one would normally observe with the pKK223-3, pKK233-2, pTrc99A, and pET-21(+) vectors.

To meet the expression needs required under different experimental circumstances, two additional expression vectors which are derivatives of pLACll were designed. pLAC22 provides lacN on the vector and thus unlike pLACll does not require an extraneous source of Lacl for its repression. pLAC33 contains the mutated ColEl replicon from pUC8 and thus allows proteins to be expressed at much higher levels due to the increase in the copy number of the vector. Of all the expressions that were evaluated in direct comparison studies, the highest level of protein expression under fully induced conditions was achieved using the pLAC33 vector. Because the cloning regions are identical in pLACll, pLAC22, and pLAC33, genes that are cloned into one of these vectors can be trivially subcloned into either of the other two vectors depending on experimental circumstances. For physiological studies, pLACl 1 is the best suited of the three vectors. If, however, the bacterial strain of choice can not be modified to introduce elevated levels of Lac repressor protein which can be achieved by F's or compatible plasmids that provide lacH or lac , the pLAC22 vector can be utilized. If maximal overexpression of a gene product is the goal, then the pLAC33 vector can be utilized.

Numerous experiments call for expression of a cloned gene product at physiological levels; i.e., at expression levels that are equivalent to the expression levels observed for the chromosomal copy of the gene. While this is not easily achievable with any of the commonly utilized expression vectors, these kinds of experiments can be done with the pLACll expression vector. By varying the IPTG concentrations, expression from the pLACll vector can be adjusted to match the expression levels that occur under different physiological conditions for the chromosomal copy of the gene. In fact, strains which contain both a chromosomal null mutation of the gene in question and a pLACll constract of the gene preserve the physiological phenotype of the null mutation under repressed conditions.

Because the use of Lac repressor is an essential component of any expression vector that utilizes the lac operon for its regulation, the ability of different source of Lacl to repress the pLACll vector was also investigated. Researchers have historically utilized either lac¥\ constracts which make 10 fold more Lac repressor than wild-type lacl or lacR¹ constructs which make 100 fold more Lac repressor than wild-type lacl (B. Muller-Hill, Prog. Biophys. Mol. Biol. 30:227-252 (1975)). The greatest level of repression of pLAC 11 constracts could be achieved using F's which provided approximately one copy of the lacΑ¹ gene or a multicopy compatible plasmid which provided approximately six copies of the lacH gene. However, the induction that was achievable using these lacl sources was significantly lower than what could be achieved when F's which provided approximately one copy of the lacH¹ gene were used to repress the pLACll construct. Thus if physiological studies are the goal of an investigation, then F's which provide approximately one copy of the lacH¹ gene or a multicopy compatible plasmid which provides approximately six copies of the lacIQ gene can be used to regulate the pLACll vector. However, if maximal expression is desired, then F's which provide approximately one copy of the lacfl gene can be utilized. Alternatively, if a bacterial strain can tolerate prolonged overexpression of an expressed gene and overexpression of a gene product is the desired goal, then maximal expression under induced conditions is obtained when a bacteria strain lacks any source of Lac repressor.

Example II An in vivo approach for generating novel bioactive peptides that inhibit the growth of E. coli

A randomized oligonucleotide library containing sequences capable of encoding peptides containing up to 20 amino acids was cloned into pLACl 1 (Example I) which allowed the peptides to either be tightly turned off or overproduced in the cytoplasm of E. coli. The randomized library was prepared using a [NNN] codon design instead of either the [NN(G,T)] or [NN(G, ] codon design used by most fusion-phage technology researchers. [NN(G,T)] or [NN(Q ] codons have been widely used instead of [NNN] codons to eliminate two out of the three stop codons, thus increasing the amount of full-length peptides that can be synthesized without a stop codon (J. Scott et al., Science 249:386-390 (1990); J. Delvin et al., Science 249:404-406 (1990); S. Cwirla et al., Proc. Nat'l. Acad. Sci. U. S. A. 87:6378-6382 (1990)). However, the [NN(G,T)] and [NN(G,Q] oligonucleotide codon schemes eliminate half of the otherwise available codons and, as a direct result, biases the distribution of amino acids that are generated. Moreover, the [NN(G,T)] and [NN(G, ] codon schemes drastically affect the preferential codon usage of highly expressed genes and removes a number of the codons which are utilized by the abundant tRNAs that are present in E. coli (H. Grosjean et al., Gene. 18: 199-209 (1982); T. Tkemura, J. Mol. Biol. 151:389-409 (1981)).

Of the 20,000 peptides screened in this Example, 21 inhibitors of cell growth were found which could prevent the growth of E. coli on minimal media. The top twenty inhibitor peptides were evaluated for strength of inhibition, and the putative amino acid sequences of the top 10 "anchorless" inhibitor peptides were examined for commonly shared features or motifs. MATERIALS AND METHODS

Media. Rich LB and minimal M9 media used in this study was prepared as in Example I. Ampicillin was used in rich media at a final concentration of 100 ϊg/ml and in minimal media at a final concentration of 50 ϊg/ml. IPTG was added to media at a final concentration of 1 mM.

Chemicals and Reagents. Extension reactions were carried out using Klenow from New England Biolabs while ligation reactions were performed using T4 DNA Ligase from Life Sciences. IPTG was obtained from Diagnostic Chemicals Limited.

Bacterial Strains and Plasmids. ALS225, which is MC1061/F 'lacN¹Z+Y+A+ (see Example I), was the E. coli bacterial strain used in this Example. The genotype for MC1061 is araD139Δ(araABOIC-leu)7679Δ(lac)X74 galUgalKrpsL hsr- hsm+ (M. Casadaban et al., J. Mol. Biol. 138: 179-207 (1980)). pLACll, a highly regulable expression vector, is described in Example I.

Generation of the Randomized Peptide Library. The 93 base oligonucleotide 5 TAC TAT AGA TCT ATG (NNN)₂o TAA TAA GAA TTC TCG ACA 3' (SEQ ID NO:23), where N denotes an equimolar mixture of the nucleotides A, C, G, or T, was synthesized with the trityl group and subsequently purified with an OPC cartridge using standard procedures. The complementary strand of the 93 base oligonucleotide was generated by an extension/fill-in reaction with Klenow using an equimolar amount of the 18 base Oligonucleotide primer 5' TGT CGA GAA TTC TTA TTA 3' (SEQ ID NO:24). After extension, the resulting ds-DNA was purified using a Promega DNA clean-up kit and restricted with EcoR 1 and Bgl II (Promega, Madison, WI). The digested DNA was again purified using a Promega DNA clean-up kit and ligated to pLACll vector which had been digested with the same two restriction enzymes. The resulting library was transformed into electrocompetent ALS225 E. coli cells under repressed conditions (LB, ampicillin, plus glucose added to 0.2%).

Screening of Transformants to Identify Inhibitor Clones. Transformants were screened to identify any that could not grow on minimal media when the peptides were overproduced. Using this scheme, any transformant bacterial colony that overproduces a peptide that inhibits the production or function of a protein necessary for growth of that transformant on minimal media will be identified. Screening on minimal media, which imposes more stringent growth demands on the cell, will facilitate the isolation of potential inhibitors from the library. It is well known that growth in minimal media puts more demands on a bacterial cell than growth in rich media as evidenced by the drastically reduced growth rate; thus a peptide that adversely affects cell growth is more likely to be detected on minimal media. Screening was carried out using a grid-patching technique. Fifty clones at a time were patched onto both a rich repressing plate (LB Amp glucose) and a minimal inducing plate (M9 glycerol Amp IPTG) using an ordered grid. Patches that do not grow are sought because presumably these represent bacteria that are being inhibited by the expressed bioactive peptide. To verify that all of the inhibitors were legitimate, plasmid DNA was made from each inhibitory clone (QIA Prep Spin Miniprep kit; Qiagen Cat. No. 27104) and transformed into a fresh background (ALS225 cells), then checked to confirm that they were still inhibitory on plates and that their inhibition was dependent on the presence of the inducer, TPTG.

Growth Rate Analysis in Liquid Media. Inhibition strength of the peptides was assessed by subjecting the inhibitory clones to a growth rate analysis in liquid media. To determine the growth rate inhibition, starting cultures of both the peptides to be tested and a control strain which contains pLACl 1 were diluted from a saturated overnight culture to an initial OD₅₅₀ of -.01. All cultures were then induced with 1 mM IPTG and OD₅₅₀ readings were taken until the control culture reached an OD₅₅₀ of ~0.5. The hypothetical data in Table 9 shows that when the control strain reaches an OD₅₅₀ of about 0.64 (at about 15 hours), a strain which contains a peptide that inhibits the growth rate at 50%will only have reached an OD₅₅₀ of only about 0.08. Thus, the growth of a 50%inhibited culture at 15 hours (i.e., the OD₅₅₀ at 15 hours, which is proportional to the number of cells in a given volume of culture) is only about 12.5% (that is, 0.08/0.64 x 100) of that of a control strain after the same amount of time, and the inhibitor peptide would thus have effectively inhibited the growth of the culture (as measured by the OD₅₅₀ at the endpoint) by 87.5% (=100% - 12.5%). Table 9: Hypothetical data from a peptide that inhibits growth rate at 30%, 50%and 70%

<img file="WO2004011485A2_D0013.tif" />

An example is shown in Fig. 6, wherein ALS225 cells containing the pLACl 1 vector (control), and either the one day inhibitor pPepl or the two day inhibitor pPepl2 (see below), were grown in minimal M9 glycerol media with IPTG added to 1 mM. OD₅₅₀ readings were then taken hourly until the cultures had passed log phase. Growth rates were determined by measuring the spectrophotometric change in OD₅₅₀ per unit time within the log phase of growth. The inhibition of the growth rate was then calculated for the inhibitors using pLACll as a control.

Sequencing the Coding Regions of the Inhibitor Peptide Clones. The forward primer 5 ' TCA TTA ATG CAG CTG GCA CG 3 ' (SEQ ID NO:25)and the reverse primer 5 ' TTC ATAC AC GGT GCC TGA CT 3 ' (SEQ ID NO:26) were used to sequence both strands of the top ten "anchorless" inhibitor peptide clones identified by the grid-patching technique. If an error-free consensus sequence could not be deduced from these two sequencing runs, both strands of the inhibitor peptide clones in question were resequenced using the forward primer 5' TAG CTC ACT CAT TAG GCA CC 3' (SEQ ID NO:27)and the reverse primer 5' GAT GAC GAT GAG CGC ATT GT 3' (SEQ ID NO:28). The second set of primers were designed to anneal downstream of the first set of primers in the pLACll vector.

Generating Antisense Derivatives of the Top Five "Anchorless" Inhibitor Clones. Oligonucleotides were synthesized which duplicated the DNA insert contained between the Bgl II and EcoR I restriction sites for the top five "anchorless" inhibitor peptides as shown in Table 12 with one major nucleotide change. The "T" of the ATG start codon was changed to a "C" which resulted in an ACG which can not be used as a start codon. The oligonucleotides were extended using the same 18 base oligonucleotide primer that was used to build the original peptide library. The resulting ds-DNA was then restricted, and cloned into pLACll exactly as described in the preceding section "Generating the randomized peptide library." The antisense oligonucleotides that were used are as follows:

pPepl(antisense): 5'TAC TAT AGA TCT ACG GTC ACT GAA TTT TGT GGC TTG TTG GAC CAA CTG CCT TAG TAA TAG TGG AAG GCT GAA ATT AAT AAG AAT TCT CGA CA 3'(SEQ ID NO: 29);

pPepS(antisense): 5'TAC TAT AGA TCT ACG TGG CGG GAC TCA TGG ATT AAG GGT AGG GAC GTG GGG TTT ATG GGT TAA AAT AGT TTG ATA ATA AGA ATT CTC GAC A 3'(SEQ ID NO: 30)

pPepl2(antisense): 5'TAC TAT AGA TCT ACG AAC GGC CGA ACC AAA CGA ATC CGG GAC CCA CCA GCC GCC TAA ACA GCT ACC AGC TGT GGT AAT AAG AAT TCT CGA CA 3 '(SEQ ID NO: 31)

pPepl3(antisense): 5'TAC TAT AGA TCT ACG GAC CGT GAA GTG ATG TGT GCG GCA AAA CAG GAA TGG AAG GAA CGA ACG CCA TAG GCC GCG TAA TAA GAA TTC TCG ACA 3'(SEQ ID NO: .32)

pPep 19(antisense): 5'TAC TAT AGA TCT ACG AGG GGC GCC AAC TAA GGG GGG GGG AAG GTA TTT GTC CCG TGC ATA ATC TCG GGT GTT GTC TAA TAA GAA TTC TCG ACA 3'(SEQ ID NO: 33)

RESULTS Identifying and Characterizing Inhibitor Peptides from the Library.

Approximately 20,000 potential candidates were screened as described hereinabove, and 21 IPTG-dependent growth inhibitors were isolated. All the inhibitors so identified were able to prevent the growth of the E. coli bacteria at 24 hours, and three of the 21 inhibitors were able to prevent the growth of the E. coli bacteria at 48 hours, using the grid patching technique. These three inhibitors were classified as "two day" inhibitors; the other 18 were classified as "one-day" inhibitors.

Results from the growth rate analysis for candidate peptide inhibitors are shown in Table 10. The % inhibition of the growth rate was calculated by comparing the growth rates of cells that contained induced peptides with the growth rate of cells that contained the induced pLACll vector. Averaged values of three independent determinations are shown.

Table 10: Ability of the Inhibitor Peptides to Inhibit Cell Growth

<img file="WO2004011485A2_D0014.tif" />

Of the 21 peptides that were tested, the one-day inhibitor peptides inhibited the bacterial growth rate at a level of approximately 25%, while the two-day inhibitor peptides inhibited the bacterial growth rate at levels greater than 75%. As can be seen from the hypothetical data in Table 9, a one-day inhibitor which inhibited the growth rate at 25%would have only reached an OD₅₅₀ of 0.226 when the control strain reached an OD₅₅₀ of 0.64. At that point in time, the growth of the culture that is inhibited by a one-day inhibitor (as measured by the end-point OD₅₅₀) only be only 35.3%of that of a control strain at that point; thus the inhibitor peptide would have effectively inhibited the growth of the culture by 64.7%. A two-day inhibitor which inhibited the growth rate at 75% would have only reached an OD₅₅o of 0.028 when the control strain reached an OD₅₅₀ of 0.64. Thus the growth of the culture that is being inhibited by a two-day inhibitor will only be 4.4%of that of the control strain at this point, and the inhibitor peptide would have effectively inhibited the growth of the culture by 95.6%. These calculations are consistent with the observation that two-day inhibitors prevent the growth of bacteria on plates for a full 48 hours while the one- day inhibitors only prevent the growth of bacteria on plates for 24 hours.

All 21 candidates were examined using restriction analysis to determine whether they contained 66 bp inserts as expected. While most of them did, the two- day inhibitors pPep3 and pPepl4 were found to contain a huge deletion. Sequence analysis of these clones revealed that the deletion had caused the carboxy-terminal end of the inhibitor peptides to become fused to the amino-terminal end of the short 63 amino acid Rop protein. The rop gene, which is part of the ColEl replicon, is located downstream from where the oligonucleotide library is inserted into the pLACll vector.

Sequence Analysis of the Top 10 "Anchorless" Inhibitor Peptides. The DNA fragments comprising the sequences encoding the top 10 "anchorless" inhibitor peptides (i.e., excluding the two Rop fusion peptides) were sequenced, and their coding regions are shown in Table 11. Stop codons are represented by stars, and the landmark Bgl II and EcoR I restriction sites for the insert region are underlined. Since the ends of the oligonucleotide from which these inhibitors were constructed contained these restriction sites, the oligonucleotide was not gel isolated when the libraries were prepared in order to maximize the oligonucleotide yields. Because of this, several of the inhibitory clones were found to contain one (n-1) or two (n-2) base deletions in the randomized portion of the oligonucleotide.

Table 11: Sequence analysis of the insert region from the top 10 inhibitory clones and the peptides that they are predicted to encode ppepl - 13 aa

CM; GAA AGA TCT ATG OK- AST GAA T« TG- &lt;-QC MS MC GftC CΛA CTG CC? TAB TAft TA&lt;3 TC≤ «-3 GCT

_{M V T £ s; )}; _{ε l}, B δ 3- P " * * (SK tt NO. _ 3.1 •GAA Kit AAT AAG AW 1£ BEQ *^β HP* 35.

pPepS - 16 as

CRfi eω AGA is ^{AϊG ϊsβ cce eΛt ? ra} ^ ^{A ϊ} *"* ^ ^WB* ^ ^{CTS GGS T7T} *^{ϊfit GGT ΪSA} *** ^Λβϊ κ i E » s s. ι κ c a 5J v e P (, G ^« .SEU ID

NO. 35&gt;

TK n ATA Atu. ATT C (sea ID WO. 3t.

pPe β - 42 aa - last 25 aa could form a ydro hobic memb ane^**<img file="WO2004011485A2_D0015.tif" />

CAB Sftft &amp;» TCT AT5 ^TCA G*^{58 GGR} ■** SK- *03 »3 fi^ffi ?SC SAG TO3 G03 AΪG W MS ∞T TGG ATC

W-5 GTA ATA *aS gre »T- «T (K SS **Λ TCA KM ATA MC TTT WT O0G Gift Bt TA» C^AC AGT _Y V I B l -, M r D S J. S S - -. r -I A V V ϊ « S

TAP. (SK) -&amp; HOS 361 * iSB8 IB bt-.- 39'

pPep? - 6 aa

CAS CAR. ASΛ KI A.B »_Λl IKS TTC W- GSft TAA ^■- AC MA ATC GTC CSC TSS AGA ACT TEA GTT -AA TM

K y L F I G « {SBQ Jt&gt; W-t .«&gt; cu TTC (5BQ IP KSt .1)

ppepa - 21 aa

CAS GAR A©. -CT ATG CT CtA TTT «βS GGS &amp;&lt;kC TEC I5S6 CAS ABΛ GCC 6S-. -AC TTT ACT «K- CTA COS B t l P G e t- C C O K A CS V e T l P

TCA AGO -AA TSA a&amp; -2-U^SSG ^{TO Wt α2J}<img file="WO2004011485A2_D0016.tif" />

pPeplO - ?,Q as - predicted to bo 45% p-s e^βt - sάno acids 6-14

CAS CAR ASA TCT AW ATT GGS «GJ- TCG- TTϋ M9C TΪC GCC. TGS 6CA Λtt^, GTT TBT tof AAG AOT £« CAT _j) G ₁3 S j_i S F A H A l V C l. }. S S K

GJT TSA. {SEQ I t NOi -9) V * &lt;$BQ ID HO. 451 <img file="WO2004011485A2_D0017.tif" /> efts GΛA AS* rer AM ARC GGC CGA ACC AAA ess, ATC csβ &amp;&gt;£ ctt i ecu ace GCC TAA ACA CCT ΛCC røc

•- ". 'S ft - K R Ϊ R r&gt; P P A n » 'S£δ ID MOt .(.J teat ear AAT ωfi ART JCJSEQ IB HO? «!

pPeρl3 -IB aa ~ predicted to be 721 α-helical - amino acids

3-15 m 6ftA ASA re? ATS GAC csr GAA OTG ΛW T** eβs CCA AAU AG GBΛ TOO AAO GAA CO» ACS CCA ϊ-^*-S

-. D R S V M C A A K Q E 1. S E R T Ϊ * J5EQ ID NO; 48. GW SW TAA ΪΛ^*. GAA TTC {SB^*. ID SO' .»)

pPepl? - 12 aa

CAS GAA ftgΛ, TCT ATC TAG CCC AAT fiCA CTG GGA GCA CGC GTG TTA Cδl C*A I-ftA &lt;-∞ SOG TAG CCA TTT

H * H t S ϋ E -l T ^*. -^* -"

1WJ CCA TAA TAA &amp;8A TTC (SBO IP WffJ SO) N 1" * ^♦ &lt;-^*BS ΪB»OE 51)

pPeplS - 5 aa

CftG CM, AW TCT ATS Afiβ GCC GCC AAC TAA S5S (355 S5S AC 9m I» -K CCS TGC ATA ATC TM 66*

B R 5 A S « (S£0 TB HOI 42) GTT GW- TAA -AA GAA jtTC SϊC" IP ϋO: 53-1

Eight out of the top 10 inhibitors were predicted to encode peptides that terminate before the double TAA TAA termination site, which was engineered into the oligonucleotide. Two of the inhibitors, pPepό and pPeplO, which contain deletions within the randomized portion of the oligonucleotide, are terminated beyond the EcoR I site. One of the inhibitors, pPepl7, contains a termination signal just after the ATG start codon. However, just downstream from this is a Shine Dalgarno site and a GTG codon, which should function as the start codon. Interestingly, the start sites of several proteins such as Rop are identical to that proposed for the pPe l7 peptide (G. Cesareni et al., Proc. Natl.35 Acad. Sci. USA.79:6313-6317 (1982)). The average and median length for the 8 peptides whose termination signals occurred before or at the double TAA TAA termination site was 13 amino acids.

The characteristics of the predicted coding regions of the inhibitor peptides proved to be quite interesting. Three out of the top 10 peptides, pPe l, pPepl3, and pPepl7, contained a proline residue as their last (C-terminal) amino acid. Additionally, one of the peptides, pPepl2, contained 2 proline residues near the C- terminus, at the n-2 and n-3 positions. Thus there appears to be a bias for the placement of proline residues at or near the end of several of the inhibitory peptides. Secondary structure analysis predicted that 3 out of the 10 peptides contained a known motif that could potentially form a very stable structure. pPepB, a peptide containing a C-terminal proline, is predicted to be 72% α-helical, pPeplO is predicted to be 45% β-sheet, and pPep6 is predicted to form a hydrophobic membrane spanning domain.

Verifying that the Inhibitory Clones do not Function as Antisense. To rule out the possibility that the bioactivity of the inhibitory clones resulted from their functioning as antisense RNA or DNA (thus hybridizing to host DNA or RNA) rather than by way of the encoded peptides, the insert regions between the Bgl II and EcoR I sites for the top five inhibitors from Table 10 were recloned into the pLACll vector using oligonucleotides which converted the ATG start codon to an ACG codon thus abolishing the start site . In all five cases the new constructs were no longer inhibitory (see Table 12), thus confirming that it is the encoded peptides that causes the inhibition and not the DNA or transcribed mRNA.

Table 12: Antisense test of the top 5 "anchorless" inhibitory peptides

<img file="WO2004011485A2_D0018.tif" />

Growth rates for cells containing the induced inhibitors or antisense constructs were determined and then the % inhibition was calculated by comparing these values to the growth rate of cells that contained the induced pCyt-3 vector.

DISCUSSION

Use of the tightly regulable pLACl 1 expression vector made possible the identification of novel bioactive peptides. The bioactive peptides identified using the system described in this Example inhibit the growth of the host organism (E. coli) on minimal media. Moreover, bioactive peptides thus identified are, by reason of the selection process itself, stable in the host's cellular environment. Peptides that are unstable in the host cell, whether or not bioactive, will be degraded; those that have short half -lives are, as a result, not part of the selectable pool. The selection system thus makes it possible to identify and characterize novel, stable, degradation-resistant bioactive peptides in essentially a single experiment.

The stability of the inhibitory peptides identified in this Example may be related to the presence of certain shared structural features. For example, three out of the top 10 inhibitory "anchorless" (i.e., non-Rop fusion) peptides contained a proline residue as their last amino acid. According to the genetic code, a randomly generated oligonucleotide such as the one used in this Example has only a 6% chance of encoding a proline at a given position, yet the frequency of a C-terminal proline among the top ten inhibitory peptides is a full 30%. This 5-fold bias in favor of a C- terminal proline is quite surprising, because although the presence of proline in a polypeptide chain generally protects biologically active proteins against nonspecific enzymatic degradation, a group of enzymes exists that specifically recognize proline at or near the N- and C-termini of peptide substrates. Indeed, proline-specific peptidases have been discovered that cover practically all situations where a proline residue might occur in a potential substrate (D. F. Cunningham et al., Biochimica et Biophysics Acta 1343:160-1 86 (1997)). For example, although the N-terminal sequences Xaa-Pro-Yaa- and Xaa-Pro-Pro-Yaa (SEQ ID NO: 54) have been identified as being protective against nonspecific N-terminal degradation, the former sequence is cleaved by arninopeptidase P (at the Xaa-Pro bond) and dipeptidyl peptidases IN and π (at the -Pro-Yaa-bond)) (Table 5, G. Vanhoof et al., FASEB J. 9:736-44 (1995); D. F. Cunningham et al., Biochimica et Biophysics Acta 1343:160-1 86 (1997)); and the latter sequence, present in bradykinin, interleukin 6, factor XII and erythropoietin, is possibly cleaved by consecutive action of arninopeptidase P and dipeptidyl peptidase IV (DPPIN), or by prolyl oligopeptidase (post Pro-Pro bond) (Table 5, G. Vanhoof et al., FASEB J. 9:736-44 (1995)). Prolyl oligopeptidase is also known to cleave Pro-Xaa bonds in peptides that contain an Ν-terminal acyl-Yaa-Pro-Xaa sequence (D. F. Cunningham et al., Biochimica et Biophysics Acta 1343:160-1 86 (1997)). Other proline specific peptidases acting on the Ν-terminus of substrates include prolidase, proline iminopeptidase and prolinase. Prolyl carboxypeptidase and carboxypeptidase P, on the other hand, cleave C-terminal residues from peptides with proline being the preferred Pi residue (D. F. Cunningham et al., Biochimica et Biophysics Acta 1343:160-1 86 (1997). Also of interest with respect to the stability of the inhibitory peptides, three of the top ten (30%) contained motifs that were predicted, using standard protein structure prediction algorithms, to form stable secondary structures. One of the peptides (which also has a C-terminal proline) was predicted to be 72% α-helical. Another was predicted to be 45% β-sheet; this peptide may dimerize in order to effect the hydrogen bonding necessary to form the β-sheet. A third was predicted to possess a hydrophobic membrane spanning domain. According to these algorithms (see, e.g., P. Chou et al., Adv. Enzymol. 47:45-148 (1978); J. Gamier et al., J. Mol. Biol. 120:97- 120 (1978); P. Chou, "Prediction of protein structural classes from amino acid composition," In Prediction of Protein Structure and the Principles of Protein

Conformation (Fasman, G. D. ed.). Plenum Press, New York, N. Y. 549-586 (1990); P. Klein et al., Biochim. Biophys. Acta 815:468-476 (1985)), a randomly generated oligonucleotide such as the one used in our studies would have had no better than a 1 in a 1000 chance of generating the motifs that occurred in these peptides. Finally, two of the three two-day inhibitors proved to be fusion peptides in which the carboxyl terminus of the peptides was fused to the amino terminus of the Rop protein. Rop is a small 63 amino acid protein that consists of two antiparallel E- helices connected by a sharp hairpin loop. It is a dispensable part of the ColEl replicon which is used by plasmids such as pBr322, and it can be deleted without causing any ill-effects on the replication, partitioning, or copy numbers of plasmids that contain a ColEl ori (X. Soberon, Gene. 9:287-305 (1980). Rop is known to possess a highly stable structure (W. Eberle et al., Biochem. 29:7402-7407 (1990); S. Betz et al., Biochemistry 36:2450-2458 (1997)),. and thus it could be serving as a stable protein anchor for these two peptides. Table 13 lists naturally occurring bioactive peptides whose structures have been determined. Most of these peptides contain ordered structures, further highlighting the importance of structural stabilization. Research on developing novel synthetic inhibitory peptides for use as potential therapeutic agents over the last few years has shown that peptide stability is a major problem that must be solved if designer synthetic peptides are to become a mainstay in the pharmaceutical industry (J. Bai et al., Crit. Rev. Ther. Drug. 12:339-371 (1995); R. Egleton Peptides. 18:1431- 1439 (1997); L. Wearley, Crit Rev Ther Drug Carrier Syst. 8:331-394 (1991). The system described in this Example represents a major advance in the art of peptide drug development by biasing the selection process in favor of bioactive peptides that exhibit a high degree of stability in an intracellular environment.

Table 13: Structural motifs observed in naturally occurring bioactive peptides

<img file="WO2004011485A2_D0019.tif" />

These peptides belong to multi-member families. Example III Directed synthesis of stable synthetically engineered inhibitor peptides

These experiments were directed toward increasing the number of bioactive peptides produced by the selection method described in Example π. In. the initial experiment, randomized peptides fused to the Rop protein, at either the N- or C- terminus, were evaluated. In the second experiment, nucleic acid sequences encoding peptides containing a randomized internal amino acid sequence flanked by terminal prolines were evaluated. Other experiments included engineering into the peptides an α-helical structural motif, and engineering in a cluster of opposite charges at the N- and C-termini of the peptide.

MATERIALS AND METHODS

Media. Rich LB and minimal M9 media used in this study was prepared as described by Miller (see Example I). Ampicillin was used in rich media at a final concentration of 100 ϊg/ml and in minimal media at a final concentration of 50 ig/ml. IPTG was added to media at a final concentration of 1 mM.

Chemicals and Reagents. Extension reactions were carried out using Klenow from New England Biolabs (Bedford, MA) while ligation reactions were performed using T4 DNA ligase from Life Sciences (Gaithersburg, MD) Alkaline phosphatase (calf intestinal mucosa) from Pharmacia (Piscataway, NJ) was used for dephosphorylation. IPTG was obtained from Diagnostic Chemicals Limited (Oxford, CT).

Bacterial Strains and Plasmids. ALS225, which is MC1061/E'toc/q⁷Z+Y+A+, was the E. coli bacterial strain used in this study (see Example I). The genotype for

MC1061 is araD139 Δ(araAB01C-leu)7679 Δ(lac)X74 galUgalKrpsL hsr-hsm+ as previously described. pLACl 1 (Example I), a highly regulable expression vector, was used to make p-Rop(C) and p(N)Rop-fusion vectors as well as the other randomized peptide libraries which are described below.

Construction of the p-Rop(C) Fusion Vector. The forward primer 5'TAC TAT AGA TCT ATG ACC AAA CAG GAA AAA ACC GCC 3' (SEQ ID NO:55)and the reverse primer 5'TAT ACG TAT TCA GTT GCT CAC ATG TTC TTT CCT GCG 3' (SEQ ID NO:56)were used to PCR amplify a 558 bp DNA fragment using pBR322 as a template. This fragment contained a Bgl II restriction site which was incorporated into the forward primer followed by an ATG start codon and the Rop coding region. The fragment extended beyond the Rop stop codon through the Afl DI restriction site in pBR322. The amplified dsDNA was gel isolated, restricted with Bgl II and Afl m, and then ligated into the pLAC expression vector which had been digested with the same two restriction enzymes. The resulting p-Rop(C) fusion vector is 2623 bp in size (Fig. 7).

Construction of the p(N)Rop-Fusion Vector. The forward primer 5'AAT TCA TAC TAT AGA TCT ATG ACC AAA CAG GAA AAA ACC GC 3' (SEQ ID

NO:57)and the reverse primer 5'TAT ATA ATA CAT GTC AGA ATT CGA GGT TTT CAC CGT CAT CAC 3' (SEQ ID NO:58) were used to PCR amplify a 201 bp DNA fragment using pBR322 as a template. This fragment contained a Bgl II restriction site which was incorporated into the forward primer followed by an ATG start codon and the Rop coding region. The reverse primer placed an EcoR I restriction site just before the Rop TGA stop codon and an Afl III restriction site immediately after the Rop TGA stop codon. The amplified dsDNA was gel isolated, restricted with Bgl II and Afl III, and then ligated into the pLACll expression vector which had been digested with the same two restriction enzymes. The resulting p(N)Rop-fusion vector is 2262 bp in size (Fig. 8).

Generation of Rop Fusion Randomized Peptide Libraries. Peptide libraries were constructed as described in Example II. The synthetic oligonucleotide 5'TAC TAT AGA TCT ATG (NNN)₂₀ CAT AGA TCT GCG TGC TGT GAT 3' (SEQ ID NO:59) was used to construct the randomized peptide libraries for use with the p-Rop(C) fusion vector, substantially as described in Example II. The complementary strand of this oligonucleotide was generated by a fill-in reaction with Klenow using an equimolar amount of the oligonucleotide primer 5 'ATC ACA GCA CGC AGA TCT ATG 3' were used (SEQ ID NO:60). After extension, the resulting dsDNA was digested with Bgl II and ligated into the pLACl 1 expression vector which had been digested with the same restriction enzyme and subsequently dephosphorylated using alkaline phosphatase. Because of the way the oligonucleotide library has been engineered, either orientation of the incoming digested double-stranded DNA fragment results in a fusion product. To construct the randomized peptide libraries for use with the p(N)Rop fusion vector, the randomized oligonucleotide 5'TAC TAT GAA TTC (NNN)₂₀ GAA TTC TGC CAC CAC TAC TAT 3' (SEQ TD NO:61), and the primer 5' ATA GTA GTG GTG GCA GAA TTC 3' (SEQ ID NO:62) were used. After extension, the resulting dsDNA was digested with EcoRI and ligated into the pLACll expression vector which had been digested with the same restriction enzyme and subsequently dephosphorylated using alkaline phosphatase. Because of the way the oligonucleotide library has been engineered, either orientation of the incoming digested double- stranded DNA fragment results in a fusion product.

Generation of a Randomized Peptide Library Containing Terminal Prolines.

Randomized amino acid peptide libraries containing two proline residues at both the amino and the carboxy terminal ends of the peptides were constructed using the synthetic oligonucleotide 5'TAC TAT AGA TCT ATG CCG CCG (NNN)_]6 CCG CCG TAA TAA GAA TTC GTA CAT 3' (SEQ TD NO:63). The complementary strand of the 93 base randomized oligonucleotide was generated by filling in with Klenow using the oligonucleotide primer 5' ATG TAC GAA TTC TTA TTA CGG CGG 3' (SEQ ID NO:64). After extension, the resulting dsDNA was digested with Bgl II and EcoR I and ligated into the pLACll expression vector which had been digested with the same two restriction enzymes. Because the initiating methionine of the peptides coded by this library is followed by a proline residue, the initiating methionine will be removed (F. Sherman et al, Bioessays 3:27-31 (1985)). Thus the peptide libraries encoded by this scheme are 20 amino acids in length.

Generation of a Randomized Hydrophilic α-Helical Peptide Library. Table 14 shows the genetic code highlighted to indicate certain amino acid properties. Table 14: Genetic Code Highlighted to Indicate Amino Acid Properties

TTT phe h_a TCT ser TAT tyr b_a TGT cys

TTC phe ha TCC ser TAC tyr ba TGC cys

TTA leu H_a TCA ser TAA OCH TGA OPA

TTG leu H_a TCG ser TAG AMB TGG trp

CTT leu H_a CCT pro B_a CAT his ha CGT arg

CTC leu H_a CCC pro B_a CAC his ha CGC arg

CTA leu H_a CCA pro B_a CAA gin h_a CGA arg

CTG leu H_a CCG pro B_a CAG gin h_a CGG arg

ATT ile h_a ACT thr AAT asn ba AGT ser

ATC ile ha ACC thr AAC asn b_a AGC ser

ATA ile h_a ACA thr AAA asn h_a AGA arg

ATG met H_a ACG thr AAG asn h_a AGG arg

GTT val ha GCT ala H_a GAT asp h_a GGT giy B_a

GTC val ha GCC ala H_a GAC asp h_a GGC gly B_a

GTA val ha GCA ala H_a GAA asp h_a GGA g y B_a

GTG val ha GCG ala H_a GAG asp h_a GGG giy B_a

Boldface amino acids are hydrophobic while italicized amino acids are hydrophilic. The propensity for various amino acids to form α-helical structures is also indicated in this table using the conventions first described by Chou and Fasman (P. Chou et al., Adv. Enzymol. 47:45-148 (1978)). H_a = strong α-helix former, h_a = α- helix former, B_a = strong α-helix breaker, b_a = α-helix breaker. The assignments given in this table are the consensus agreement from several different sources. Hydrophilic versus hydrophobic assignments for the amino acids were made from data found in Wolfenden et. al. (Biochemistry. 20:849-55 (1981)); Miller et. al. (J Mol. Biol. 196: 641-656 (1987)); and Roseman (J Mol. Biol. 200:5 13-22(1988)). The propensity for amino acids to form α-helical structures were obtained from consensus agreements of the Chou and Fasman (P. Chou et al., Adv. Enzymol. 47:45-148 (1978); P. Chou, "Prediction of protein structural classes from amino acid compositions," in Prediction of protein structure and the principles of protein conformation (G. Fasman, G. D. ed.). Plenum Press, New York, N.Y. 549-586 (1990)); Gamier, Osguthorpe, and Robson (J Mol. Biol, 120:97-120 (1978)); and O'Neill and DeGrado (Science. 250:646-65 1 (1990)) methods for predicting secondary structure. By analyzing the distribution pattern of single nucleotides in the genetic code relative to the properties of the amino acids encoded by each nucleotide triplet, a novel synthetic approach was identified that would yield randomized 18 amino acid hydrophilic peptide libraries with a propensity to form α-helices. According to Table 14, the use of a [(CAG)A(TCAG)] codon mixture yields the hydrophilic amino acids His, Gin, Asn, Lys, Asp, and Glu. These amino acids are most often associated with α-helical motifs except for asparagine, which is classified as a weak α-helical breaker. If this codon mixture was used to build an α-helical peptide, asparagine would be expected to occur in about 17% of the positions, which is acceptable in an α-helical structure according to the secondary structure prediction rules of either Chou and Fasman (P. Chou et al., Adv. Enzymol. 47:45-148 (1978); P. Chou, "Prediction of protein structural classes from amino acid compositions," in Prediction of protein structure and the principles of protein conformation (G. Fasman, G. D. ed.). Plenum Press, New York, N.Y 549-586 (1990)) or Gamier, Osguthorpe, and Robson (J. Gamier et al., J. Mol. Biol. 120:97-120 (1978)). Additionally, several well- characterized proteins have been observed to contain up to three b_a breaker amino acids within a similarly sized α-helical region of the protein (T. Creighton, "Conformational properties of polypeptide chains," in Proteins: structures and molecular properties, W. H. Freeman and Company, N.Y, 182-186 (1993)). Since in most α-helices there are 3.6 amino acids per complete turn, the 18 amino acid length was chosen in order to generate α-helical peptides which contained 5 complete turns. Moreover, the use of hydrophilic amino acids would be expected to yield peptides which are soluble in the cellular cytosol.

Randomized 18 amino acid hydrophilic α-helical peptide libraries were synthesized using the synthetic oligonucleotide 5'TAC TAT AGA TCT ATG (VAN) TAA TAA GAA TTC TGC CAG CAC TAT 3 ' (SEQ ID NO:65). The complementary strand of the 90 base randomized oligonucleotide was generated by filling in with Klenow using the oligonucleotide primer 5 'ATA GTG CTG GCA GAA TTC TTA TTA 3' (SEQ ID NO:66). After extension the resulting dsDNA was digested with Bgl II and EcoR I and ligated into the pLAC 11 expression vector which had been digested with the same two restriction enzymes.

Generating a Randomized Peptide Library Containing the +/- Charge Ending Motif. Randomized peptide libraries stabilized by the interaction of oppositely charge amino acids at the amino and carboxy termini were generated according to the scheme shown in Fig. 9. To maximize the potential interactions of the charged amino acids, the larger acidic amino acid glutamate was paired with the smaller basic amino acid lysine, while the smaller acidic amino acid aspartate was paired with the larger basic amino acid arginine. To construct the randomized peptide libraries, the synthetic oligonucleotide 5'TAC TAT AGA TCT ATG GAA GAC GAA GAC (NNN)_]6 CGT AAA CGT AAA TAA TAA GAA TTC GTA CAT 3' (SEQ ID NO:67)and the oligonucleotide primer 5'ATG TAC GAA TTC TTA TTA TTT ACG TTT ACG 3' (SEQ ID NO: 68) were used. After extension, the resulting dsDNA was digested with Bgl II and EcoR I and ligated into the pLACl 1 expression vector which had been digested with the same two restriction enzymes.

For all libraries of randomized oligonucleotides, N denotes that an equimolar mixture of the four nucleotides A, C, Q and T was used, and V denotes that an equimolar mixture of the three nucleotides A, C and G was used. The resulting libraries were transformed into electrocompetent ALS225 E. coli cells (Example I) under repressed conditions as described in Example II.

Screening of Transformants to Identify Inhibitor Clones. Transformants were initially screened using the grid-patching technique to identify any that could not grow on minimal media as described in Example II when the peptides were overproduced. To verify that all the inhibitors were legitimate, plasmid DNA was made from each inhibitory clone, transformed into a fresh background, then checked to make sure that they were still inhibitory on plates and that their inhibition was dependent on the presence of the inducer, IPTG as in Example II.

Growth Rate Analysis in Liquid Media. Inhibition strength of the peptides was assessed by subjecting the inhibitory clones to a growth rate analysis in liquid media. Minimal or rich cultures containing either the inhibitor to be tested or the relevant vector as a control were diluted to an initial OD₅₅₀ of approximately 0.01 using new media and induced with 1 mM IPTG. OD₅₅₀ readings were then taken hourly until the cultures had passed log phase. Growth rates were determined as the spectrophotometric change in OD₅₅₀ per unit time within the log phase of growth, and inhibition of the growth rate was calculated for the inhibitors using the appropriate vector as a control. RESULTS

Isolation and Characterization of Inhibitor Peptides that are Fused at Their

Carboxy Terminal End to the Amino Terminal End of the Rop Protein.

Approximately 10,000 peptides protected by the Rop protein at their carboxy terminal end were screened using the grid-patching technique described in Example II, and 16 two day inhibitors were isolated. The inhibitory effects were determined as described in the Example π, using pRop(C) as a control. Unlike the anchorless inhibitors identified in Example U that were only inhibitory on minimal media, many of the Rop fusion inhibitors were also inhibitory on rich media as well, which reflects increased potency. As indicated in Table 15, the inhibitors inhibited the bacterial growth rate at levels that averaged 90% in minimal media and at levels that averaged 50% in rich media. The data in Table 15 is the average of duplicate experiments.

Table 15: Inhibitory effects of peptide inhibitors stabilized by fusing the carboxy terminal end of the peptide to the amino terminal end of the Rop protein (Rop(C) fusion peptide inhibitors

<img file="WO2004011485A2_D0020.tif" />

Isolation and Characterization of Inhibitor Peptides that are Fused at Their Amino Terminal End to the Carboxy Terminal End of the Rop Protein.

Approximately 6000 peptides protected at their amino terminal end by Rop protein were screened using the grid-patching technique described in Example II, and 14 two day inhibitors were isolated. As observed for the Rop fusion peptides isolated using the p-Rop(C) vector, most of the inhibitor peptides isolated using the p(N)Rop- vector were inhibitory on rich media as well as minimal media. The inhibitors were verified as described hereinabove and subjected to growth rate analysis using p(N)Rop- as a control in order to determine their potency. As indicated in Table 16, the inhibitors inhibited the bacterial growth rate at levels that averaged 90% in minimal media and at levels that averaged 40% in rich media. The data in Table 16 is the average of duplicate experiments.

Table 16: Inhibitory effects of peptide inhibitors stabilized by fusing the amino terminal end of the peptide to the carboxy terminal end of the Rop protein (Rop(N) fusion peptide inhibitors)

<img file="WO2004011485A2_D0021.tif" />

Isolation and Characterization of Anchorless Inhibitor Peptides Containing Two Prolines at Both Their Amino Terminal and Carboxy Terminal Ends.

Approximately 7500 peptides were screened using the grid-patching technique described in Example II, and 12 two day inhibitors were isolated. As indicated in Table 17, the top ten inhibitors inhibited the bacterial growth rate at levels that averaged 50%in minimal media. The inhibitory effects were determined as described in the text using pLACll as a control. The data in Table 17 is the average of duplicate experiments. Table 17: Inhibitory effects of peptide inhibitors stabilized by two proline residues at both the amino and carboxy terminal ends of the peptide

<img file="WO2004011485A2_D0022.tif" />

Sequence analysis of the coding regions for the top ten inhibitors is shown in Table 19. The landmark Bgl II and EcoR I restriction sites for the insert region are underlined, as are the proline residues.

Since the ends of the oligonucleotide from which these inhibitors were constructed contained Bgl II and EcoRI I restriction sites, the oligonucleotide was not gel isolated when the libraries were prepared in order to maximize the oligonucleotide yields. Because of this, three of the inhibitory clones, pPro2, Ppro5, and pProό were found to contain deletions in the randomized portion of the oligonucleotide.

Table 18: Sequence analysis of the insert region from the proline peptides pProl -21aa

AGA TCT ATG CCG CCG ATT CTA TGG GGC GAA GCG AGA AAG CGC TTG TGG GGT GGG GAT CAT ACA CCG CCG TAA TAA M P P I L G E A R K R G G D H T P P

_{* *}

(SEQ ID NO: 70) GAA TTC (SEQ ID NO: 69)

pPro2 - 27aa

AGA TCT ATG CCG CCG CCG TTG GAT ATT GTG TCG GGT ATT GAG GTA GGG GGG CAT TTG TGG TGC CGC CGT ATT AAG

M P P P L D I V S G I E V G G H L W C R R i K - - -

AAT TCT CAT GTT TGA (SEQ ID NO: 71) N S H V * (SEQ ID NO: 72)

pPro3 - 8aa

AGA TCT ATG CCG CCG GAC AAT CCG GTC CTG TGA TGA AGC GGA GGT CGA CCA AGG GGA TAT CAG CCG CCG TAA TAA

M P P D N P V L * * (SEQ ID NO: 74)

GAA TTC (SEQ ID NO: 73)

pPro4 - 9aa

AGA TCT ATG CCG CCG CTA TTG GAC GGA GAT GAC AAA TAG ATA TAT GCG TGG TTG TTT TTC TGT CCG CCG TAA TAA M P P L D G D D K * (SEQ ID NO: 76)

GAA TTC (SEQ ID NO: 75)

pPro5 - lOaa AGA TCT ATG CCG CCG AGG TGG AAG ATG TTG ATA AGA CAG TGA CAG ATG CGT TCC ATT ACT CCC GCC GTA ATA AGA

M P P R W K M I R Q * (SEQ ID NO: 78) ATT C (SEQ ID NO: 77)

pProό - 7aa

AGA TCT ATG ATG AGA GTA GCG CCG CCG TAA TAA GAA TTC (SEQ ID NO: 79) M M R V A P P * * (SEQ ID NO: 80)

pPro7 - 14aa

AGA TCT ATG CCG CCG TTG CGC GGG GCA TGC GAT GTA TAT GGG GTA AAT TGA ATG TCT TGT GGG CCG CCG TAA TAA P P L R G A C D V Y G V N * (SEQ ID NO: 82) GAA TTC (SEQ ID NO: 81)

pPro8-21aa

AGA TCT ATG CCG CCG GGG AGA GGG GAA GCG GTG GGA GTG ACA TGC TTG AGC GCG AAC GTG TAC CCG CCG TAA TAA

M P P G R G E A V G V T C S A N V Y P P

(SEQ ID NO: 84) GAA TTC (SEQ ID NO: 83)

pPro9-21aa

AGA TCT ATG CCG CCG GGA AGG GTA GTG TTC TTT GTC GCT ATC TTT GTT TCC GCA ATA TGC CTC CCG CCG TAA TAA M P P G R V V F F V A I F V S A I C L P P

(SEQ ID NO: 86)

GAA TTC (SEQ ID NO: 85) TC GCT CAT GAG AGT GTT AAA GGG CTG GGG GAC GTT ACA AAA GCT CCG CCG R F A H E S V K G G D V T K A P

<img file="WO2004011485A2_D0023.tif" />

All the inhibitors were found to contain two proline residues at either their amino or carboxy termini as expected. Four inhibitors contained two proline residues at both their amino and. carboxy termini, five inhibitors contained two proline residues at only their amino termini, and one inhibitor contained two proline residues at only its carboxy terminus.

Isolation and Characterization of Anchorless Hydrophilic Inhibitor Peptides Stabilized by an α-Helical Motif. Approximately 12,000 peptides were screened using the grid-patching technique and 5 two-day inhibitors were isolated. The inhibitors were verified as already described for the Rop-peptide fusion studies and subjected to growth rate analysis using pLACll as a control in order to determine their potency. As indicated in Table 19, the inhibitor peptides inhibited the bacterial growth rate at levels that averaged 50% in minimal media. The averaged values of two independent determinations are shown.

Table 19: Inhibitor effects of the hydrophilic α-helical peptides

<img file="WO2004011485A2_D0024.tif" />

Sequence analysis of the coding regions for the 5 inhibitors is shown in Table 20. The landmark Bgl II and EcoR I restriction sites for the insert region are underlined. Since the ends of the oligonucleotide from which these inhibitors were constructed contained these restriction sites, the oligonucleotide was not gel isolated when the libraries were prepared in order to maximize the oligonucleotide yields. Because of this, two of the inhibitory clones, pHelix2 and pHelix3, were found to contain deletions in the randomized portion of the oligonucleotide. The predicted α- helical content of these peptides is indicated in Table 20 according to the secondary structure prediction rules of Gamier, Osguthorpe, and Robson (J. Gamier et al., J. Mol. Biol.120: 97-120 (1978)) prediction rules.

Table 20: Sequence analysis of the insert region from the fcydrβpMH* - helicai peptides

pHelixl - lβaa, 831 α- elical

AS..CT ATC CΛ* SAC GAA CAA SJβ Θ$ &lt;sfVS CM AM ABB, A&amp;G W«C«Ml-⁽_^*Affl- TAJ. TAA S?Λ ' *- H -. B β B -. -- H W K I- I&gt; » B K E H * * i$t_

IB NO. WJ TΪC »E tB MO. 89 *

pHβlix2 - 22aa, 68% α-helical Sfl-S CAA GCC &amp;33 ΛTC Λfit AAG ΛGS ATG AAS MT R T Wj g ΪCΪ CAT E β -. λ -- S K B H K S -. -. H S B <img file="WO2004011485A2_D0025.tif" />

pHelix3 - 22aa 55% - βlical SS TCT AWJ ΛAΛ CRT CRT TOT GW UCC AΪS A1C ΛSC MS, AT6 #«δWMI AW ASG S TCT CΛΪ

H H R )t i £ A t. l H T H K ϊ R r -. K K -i H <img file="WO2004011485A2_D0026.tif" />

pHelix4 - 18aa, 17% α-helical

AST- let ATC MC QfiC CAC BH CAS PA GfttS GW Krø CW fiat Cftϊ CW AftO G» MΠ ΛAA 1W. TM 6SA

M K -) B -. φ O E D « H D 1J It K D N K * * (ϋeQ <img file="WO2004011485A2_D0027.tif" />

pHelix5 - 18aa_# 50% -helical

%Pb i--» A e t-w. GAS ews &lt;*»» csvs era MT GAT ΛRC M? ew- c-vs -A- AM. αw ana<img file="WO2004011485A2_D0028.tif" /> *w- ϊAft £--&amp;

ID KOS 9Θ&gt;

TTC ISTO iβ BO. -•^".)

According to Gamier, Osguthorpe, and Robson secondary structure prediction, all of the encoded peptides are expected to be largely α-helical except for pHelix4. Interestingly, pHelixl, which had the highest degree of α-helical content, was also the most potent inhibitory peptide that was isolated in this study. Isolation and Characterization of Anchorless Inhibitor Peptides Stabilized by an Opposite Charge Ending Motif. Approximately 20,000 peptides were screened using the grid-patching technique and 6 two day inhibitors were isolated. The inhibitors were verified as already described for the Rop-peptide fusion studies and subjected to growth rate analysis using pLACll as a control in order to determine their potency. As indicated in Table 21, the inhibitor peptides inhibited the bacterial growth rate at levels that averaged 50% in minimal media. The averaged values of two independent determinations are shown.

Table 21: Inhibitory effects of peptide inhibitors that are stabilized by the opposite charge ending motif

<img file="WO2004011485A2_D0029.tif" />

Sequence analysis of the coding regions for the six inhibitors is shown in Table 22. The landmark Bgl II and EcoR I restriction sites for the insert region are underlined. With the exception of p+/-4, which was terminated prematurely, the coding regions for the inhibitors were as expected based on the motif that was used to generate the peptide libraries. Table 22: Sequence analysis of the insert region from the opposite charge ending peptides p+/~l - 25aa

ASA rar ATG earn GRC GBA GAC HVS GGT COS *Λ oes ?c$ OCA GCA CAA cr* ΦSS ΪCΏ ffi CMS TCG ers

H -- tϊ ϊ. -. -. e A S A .. G A -. Iι W S S 'Q 5 V βSt AAA CS? AAA TAA TAA GfA TTC (SEQ XO TO. 99 S ft K R K * * ζgEQ IB -β-ii iDβJ

P+/-2 - 25aa

ASA TC? A7β ft"¹-^*- i^*J«- ^βWl « G3T CTA GGC ASS GCG CGT S5G TΪG SIC AGG CFC ACT TTA T? r F

OGf AAA CST AM- tϊ»A TΛΛ gftft Wg ^SEfi It» HOi Λ01J K K ft K * * jSBQ ID MO. Iβ-i, p+/-3 - 25aa

AGA TCT ATC SB C C 6.A QfcC eS-5<img file="WO2004011485A2_D0030.tif" /> ASS AK. CAS Θ3G ©X CSC W-T CCA (HA GCG CTS βϊA GAT ABA

-- E D B D β E B I P O A R C P V A I- V O R CGT ABA CGT ASA TJ* TW, OJA TtC 'SEft IB (SO. MB) R K ft K » * ISBQ IB HO: 1ft. &gt;

P+/-4 ~ llaa

JMSA *CΓ AW GM. GAC GAA SAC GAC AOG GGO COT ess tee S?AS etf *-« GTT GCG era asr rar GAS ATA

Cβr AAft CSΪ AAA TAA TAA WA HC (SEC- ID MO, IOC. p+/ 5 - 25aa as-, TC &amp;m &lt;3m efte &gt;aftA aac GGG ess GGC &lt;GGG AGG AGS GCC TGP CTT rør τo _J W err en- ess SSA

J. E D I! l^* S G A '-. B -l A C -- C S A lι V &lt;3 E

CGT AAA CGT AW, TAA TAA gftΛ «K 'S-» ID HO. 107. R K ft X ^■» * iS&amp;Q ID MO; 306

p /-6 - 25aa

AGA TCT AT3 GAA SAC GAA GAC AAG CGT CGC GSS ACS ACT GCA APA G«S Cft? CA* C*C OS* Cfifi ?CG AK

-J E O E P K H E E -1 -t A K β l. ll V S If S M CCT AAA C&lt;5T A WA 6AC ΪSS {SEQ ID KO. 109' II ft K * JSEO ID WD! lift.

DISCUSSION

In Example II, where fully randomized peptides were screened for inhibitory effect, only three peptides (one "anchorless" and two unanticipated Rop fusions resulting from deletion) were identified out of 20,000 potential candidates as a potent (i.e., two day) inhibitor of E. coli bacteria. Using a biased synthesis as in this Example, it was possible to significantly increase the frequency of isolating potent growth inhibitors (see Table 23). Table 23: Summary of the frequency at which the different types of inhibitor peptides can be isolated

<img file="WO2004011485A2_D0031.tif" />

Many more aminopeptidases have been identified than carboxypeptidases in both prokaryotic and eukaryotic cells (J. Bai, et al., Pharm. Res. 9: 969-978 (1992); J. Brownlees et al., J. Neurochem. 60:793-803 (1993); C. Miller, In Escherichia coli and Salmonella typhimurium cellular and molecular biology, 2nd edition (Neidhardt, F. C. ed.), ASM Press, Washington, D. C. 1:938-954 (1996)). In the Rop fusion studies, it might have therefore been expected that stabilizing the amino terminal end of the peptide would have been more effective at preventing the action of exopeptidases than stabilizing the carboxy end of the peptides. Surprisingly, it was found that stabilizing either end of the peptide caused about the same effect.

Peptides could also be stabilized by the addition of two proline residues at the amino and/or carboxy termini, the incorporation pf opposite charge ending amino acids at the amino and carboxy termini, or the use of helix-generating hydrophilic amino acids. As shown in Table 23, the frequency at which potent inhibitor peptides could be isolated increased significantly over that of the anchorless peptides characterized in Example II.

These findings can be directly implemented to design more effective peptide drugs that are resistant to degradation by peptidases. In this example, several strategies were shown to stabilize peptides in a bacterial host. Because the aminopeptidases and carboxypeptidases that have been characterized in prokaryotic and eukaryotic systems appear to function quite similarly (C. Miller, In Escherichia coli and Salmonella typhimurium cellular and molecular biology, 2nd edition (Neidhardt, F. C. ed.), ASM Press, Washington, D.C. 1:938-954 (1996); N. Rawlings et al., Biochem J. 290: 205-218 (1993)), the incorporation of on or more of these motifs into new or known peptide drugs should slow or prevent the action of exopeptidases in a eukaryotic host cell as well.

Example IV Confirmation of the stabilizing effects of proline residues using an in vitro system

To extend the in vivo studies described above, an in vitro system for directly assessing peptide stability was developed. In the in vitro system, peptides to be tested were mixed with a cellular extract containing the proteases and peptidases present in a particular cell type. To validate this approach, the stability or half-life of a randomized biotinylated peptide initially was measured using both wild-type bacterial extracts and bacterial extracts that were deficient in known proteases or peptidases.

MATERIAL AND METHODS

Bacterial strains. The bacterial strains used in this study are shown in Table 24. MG1655 clpP::ca was constructed by transducing MG1655 to chloramphenicol resistance using a PI lysate that was prepared from SG22098. Table 24. Bacterial strains

<img file="WO2004011485A2_D0032.tif" />

*Guyer, M.S. et al., Cold Spring Harbor Symp. Quant. Biol. 45: 135-140

(1980).

Media. Bacterial cells were grown in LB media; yeast cells were grown in 1.0% yeast extract, 2.0% peptone, 2.0% glucose; human HeLa cells (ATCC CCL-2) and colon CCD-I8C0 cells (ATCC CRL-1459) were grown in Minimal Essential Medium Eagle (ATCC 30-2003) with Earle's balanced salt solution, 0.1 mM non-essential amino acids, 2.0 mM L-glutamine, 1.0 mM sodium pyruvate, 1.5 g L NaHCO₃, and 10% fetal bovine serum; and human small intestine FHs74 M cells (ATCC CCL-241) were grown in Hybri-Care media (ATCC 46-X) with 1.5 g/L NaHCO₃ and 10% fetal bovine serum.

Preparation of the extracts. For bacteria and yeast, 500 mL of cells were grown to an OD₅₅₀ of 0.5, centrifuged, washed twice with TioEcπ (10.0 mM Tris; pH 8.0, 0.1 mM EDTA; pH 8.0) and resuspended in 2.0 mL of 10.0 mM Tris; pH 8.0. For human cells, 10-5075 cm² T flasks were seeded and allowed to grow to 95% confluency in a 37 C incubator with 5% CO₂ atmosphere. Each flask was then washed with HBSS (0.4 g/L KC1, 0.06 g/L KH₂PO₄, 8 g/L NaCl, 0.35 g L NaHCO₃, 0.048 g/L Na₂HPO₄, 1.0 g/L Glucose) that contained 0.125 mM EDTA; pH 8.0. To liberate the cells, the flasks were treated with 1.5 mL of HBSS that contained 0.25% trypsin and 0.5 mM EDTA; pH 8.0. The trypsin was neutralized by adding 5 mL of media with 10% fetal bovine serum to each flask. The cells were centrifuged, washed with HBSS that contained 0.125 mM EDTA; pH 8.0, washed twice with HBSS lacking glucose and ETDA, and resuspended in 2.0 mL of 10.0 mM Tris; pH 8.0. All cell suspensions were lysed with three passes at 15,000 psi in a French Pressure cell maintained at 4^°C. The lysates were then centrifuged at 15,000 rpm, 4 C, for 10 minutes to pellet debris and unlysed cells and the supernatant was saved as the cell extract. To prepare rat serum, one 300g Sprague Dawley rat was euthanized with CO^and a heart puncture was performed to draw the blood which was immediately transferred to a tube and centrifuged at 4 C, 10,000 rpm, for 10 minutes. The cleared serum was removed with a pipette except for 1 cm of serum at the interface with the blood cell pellet.

Peptide synthesis. The following randomized biotinylated peptides were synthesized by Sigma Genosys (The Woodlands, Texas): Unprotected XXXXXX[KBtn]XXXXXA

P at both ends PXXXX[KBtn]XXXXP

PP at both ends PPXXXX[KBtn]XXXXPP

APP at both ends APPXXXX[KBtn]XXXXPPA

APP amino APPXXXX[K-BtnJXXXXA APP carboxyl AXXXX[K-Btn]XXXXPPA

Acetylated (Ac)AXXXXX[KBtn]XXXXXA

Amidated XXXXXX[KBtn]XXXXXA(NH₂)

Cyclized CXXXXXX[KBtn]XXXXXXC

where A denotes the L-amino acid alanine, P denotes the L-amino acid proline, X denotes an equimolar mixture of the 20 natural L-amino acids except for proline, and KBtn denotes the L-amino acid lysine to which biotin has been attached.

To ensure that the length of the randomized portion of the peptides did not affect the degradation profiles, we also tested the unprotected peptides

XXXX[KBtn]XXXXA and AXXXX[KBtn]XXXXA. Their half-lives were determined to be within 5% of the XXXXXX[KBtn]XXXXXA peptide which was used as the control for these studies.

In vitro degradation assay. All extracts were used at a final concentration of 10 mg/mL, except for the S. typhimurium extracts, which were used at a final concentration of 25 mg/mL. The cell extract (50 μL) was mixed with 50 μL of a peptide at a concentration of 1 mg/mL in 10 mM Tris; pH 8.0 and incubated at 37°C. Aliquots were removed (10 μL) at 30, 60, 90, or 120 minute intervals, placed into 90 μL of SDS-PAGE gradient gel buffer, boiled for 5 minutes, and eiectrophoresed through a 10-20% tricine gradient gel. The gel was blotted onto a nitrocellulose membrane and the resulting Western blot was treated with NeutrAvidin Horseradish Peroxidase Conjugate and SuperSignal West Dura Extended Duration Chemiluminescent Substrate (Pierce, Rockford, IL). The biotinylated peptides were then visualized by exposing the blots to autoradiography film and the resulting bands were quantified using the AlphaEase 5.5 Densitometry Program from Alpha Innotech, San Leandro, CA.

RESULTS The proteases and peptidases have been well characterized in E. coli and S. typhimurium. In E. coli, the two main proteases that have been shown to have a role in peptide degradation are Lon and ClpP, which are encoded respectively by the Ion and clpP genes. In S. typhimurium, numerous peptidases have been identified, and strains have been constructed that delete several of the peptidases. Using extracts prepared from E. coli strains that contained Ion or clpP deletions and a S. typhimurium strain in which nine peptidase genes were deleted, half-lives were determined for the unprotected randomized biotinylated control peptide. As shown in Table 25, deletion of the Lon protease caused the peptide's half-life to increase by 6.5 fold, deletion of the ClpP protease caused the peptide's half-life to increase by 1.8 fold, and deletion of multiple peptidases caused the peptide's half-life to increase by 7.1 fold. These results prove that the in vitro system provides an accurate method by which to assess peptide stability.

Table 25. Peptide degradation in protease and peptidase deficient extracts.

<img file="WO2004011485A2_D0033.tif" />

Because of the decreased potency of S. typhimurium extracts relative to E. coli extracts, the S. typhimurium extracts were used at a concentration of 25 mg/mL.

With the system validated, the stabilizing effects of proline residues were analyzed. Three randomized biotinylated peptides were tested using extracts prepared from bacterial (wild-type E. coli), Baker's yeast (wild-type Saccharomyces cerevisiae), human (HeLa) cells, human intestine and colon cells, and rat serum. One randomized peptide was unprotected, while the other two peptides were stabilized on both the N- and C-termini with a Pro (P) motif, a Pro-Pro motif (PP), or an Ala-Pro- Pro motif (APP). The results are shown in the Table 26. Table 26. The effect of proline-containing stabilizing groups on peptide degradation

<img file="WO2004011485A2_D0034.tif" />

ND: not determined

As the data indicate, the APP motif offered significantly more protection than the PP motif, which provided better protection than the P motif.

Table 27 shows the results of degradation studies on peptides that contain the APP motif at either or both of the amino or carboxyl termini.

Table 27. The effect of APP stabilizing groups on peptide degradation

<img file="WO2004011485A2_D0035.tif" />

The data show that APP at only the amino terminus offers slightly better protection than APP at both termini, and that APP at only the amino terminus offers significantly better protection than APP at only the carboxyl terminus.

Table 28 shows the results of degradation studies on peptides that contain the APP motif at the N- or C-terminus compared to peptides that are acetylated at their amino terminus, amidated at their carboxyl terminus, or cyclized.

Table 28. The effect of APP stabilizing groups on peptide degradation in comparison to acetylation, amidation or cyclization.

oe <img file="WO2004011485A2_D0036.tif" />

The data clearly shows that APP at the amino terminus offers better protection than amidating the carboxyl terminus or acetylating the amino terminus, and is almost as good as cyclization.

Example V Bioactivity of natural galanin, APP-galanin, and APP-galanin-PPA

Radio immunoassays (RIAs) were performed to determine the ability of galanin and its APP derivatives to displace radiolabeled galanin from its receptor. Binding (displacement) constants were then calculated from this data.

Natural galanin Ki = 5.21 x 10^"9

APP-galanin Ki = 6.42 x 10^"9

APP-galanin-PPA Ki = 9.46 x 10^~9

As the data shows the binding constants for the APP derivatives were in the same range as natural galanin and thus these compounds were able to interact with the galanin receptor in a manner similar to natural galanin.

Example Yl In vivo Glucagon, APP-GIucagon and APP-Glucagon-PPA Degradation

A catheter was placed in the right jugular vein of six Male Sprague-Dawley rats for dosing and sampling. Two rats were used for each of the three compounds that were tested. The rats received an intravenous bolus injection of the peptide, and serial blood samples (0.3 ml) were obtained. The glucagon was extracted from plasma by organic protein precipitation and quantified by electrospray LC-MS.

The presence of the APP motif affected both the half-life of glucagon as well as the rate at which it is cleared from the body. The data (Table 29) suggests that a significant portion of the glucagon harboring the APP motif becomes sequestered and thus is much more resistant to degradation. It should be noted that significantly more APP-glucagon-PPA and APP-glucagon is present at 20 and 60 minutes than would be predicted due to its half-life. Table 29.

<img file="WO2004011485A2_D0037.tif" />

ND (not determined)

Sequence Listing Free Text

SEQ ID NO:2 peptide sequence having opposite charge ending motif

SEQ ID NOs:3, 4 stabilized angiotensin

SEQ ID NOs:6-19, 24-28, 55-58, 60, 62, 64, 66, 68 primer

SEQ ID NOs:20-22 primer fragment

SEQ ID NOs:23, 59, 61, 63, 65, 67 randomized oligonucleotide

SEQ ID NOs:29-33 antisense oligonucleotide

SEQ 3D NOs:34, 36, 39, 40, 43, 45, 46, 48, 51, 52, 70, 72, 74, 76, 78, 80, 82, 84, 86, 88, 90, 92, 94, 96, 98, 100, 102, 104, 105, 108, 110 stabilized peptide

SEQ ID NOs:35, 37, 38, 41, 42, 44, 47, 49, 50, 53, 69, 71, 73, 75, 77, 79, 81, 83, 85, 87, 89, 91, 93, 95, 97, 99, 101, 103, 106, 107, 109 nucleic acid encoding stabilized peptide

SEQ ID NO:54

N-terminal protective sequence

SEQ ID NO:lll-115 α-helical moieties The foregoing detailed description and examples have been given for clarity of understanding only. No unnecessary limitations are to be understood therefrom. The invention is not limited to the exact details shown and described, for variations obvious to one skilled in the art will be included within the invention defined by the claim.