WO1993015110A1

WO1993015110A1 - Amino acid sequences which pair specifically

Info

Publication number: WO1993015110A1
Application number: PCT/US1993/000884
Authority: WO
Inventors: Erin K. O'shea; Peter S. Kim
Original assignee: Whitehead Institute For Biomedical Research
Priority date: 1992-01-31
Filing date: 1993-02-01
Publication date: 1993-08-05

Abstract

Pairs of synthetic peptides designed in such a manner that they pair specifically with one another to form a heterodimer and then, once paired, preferentially fold as a helical heterodimer. The peptide members of the pair may be of any length, provided they are sufficiently long that they are stable in the heterodimeric form and are able to fold into a helical configuration. The synthetic peptide pairs of the present invention can be used as affinity reagents to isolate associated molecules. In addition, because the heterodimer is preferentially formed and very stable under physiological conditions, the synthetic peptides of the present invention are suitable for use in vivo. The synthetic peptide pairs can be used, for example, for in vivo applications in which two molecules or components necessary for a given event must be brought together for the event to occur. The synthetic peptide pairs of the present invention make it possible to bring the components together with great specificity and affinity. Alternatively, the synthetic peptide pairs can be designed to prevent binding or interaction of molecules (e.g., DNA and its DNA binding protein) necessary for an event to occur. The synthetic peptide pairs can also be used as a biodegradable molecular velcro, such as in grafting and for artificial sutures.

Description

AMINO ACID SEQUENCES WHICH PAIR SPECIFICALLY

Description

Background Specificity in protein-protein interactions is not well understood, although considerable effort has been expended in trying to explain what determines such inter¬ actions. It would be very useful to know what character¬ istics of an amino acid sequence contribute to protein- protein interactions.

Summary of the Invention

The present invention relates to pairs of synthetic peptides designed in such a manner that they pair specifi¬ cally with one another to form a heterodimer and then, once paired, preferentially fold as a helical heterodimer. The present invention further relates to a method of making pairs of synthetic peptides which bind prefer¬ entially to one another and to methods of producing such peptide pairs. The peptide members of the pair may be of any length, provided they are sufficiently long that they are stable in the heterodimeric form and are able to fold into a helical configuration. The two members of a pair will^" typically be of the same length, although this is not required. In general, the individual peptides will be at least 6 to 8 amino acid residues in length, generally at least 12-14 amino acid residues in length and will prefer¬ ably be at least 16-20 amino acid residues in length. In one embodiment, the individual peptides will be at least 20-23 amino acids in length. In another embodiment, the individual peptides will be 24-30 amino acids in length, particularly to 30 amino acids in length. There is no upper limit on individual peptide length. More than one peptide "repeat" or unit may be combined in a peptide pair of the present invention, if desired. That is, more than one peptide which pairs specifically with another peptide to form a heterodimer and, once paired, forms a coiled- coil helical heterodimer can be present in a synthetic peptide of the present invention; the second member of the peptide pair also includes multiple peptide repeats or units, whose amino acid sequences are designed to pair specifically and avidly with the first peptide repeats and preferentially form a coiled-coil helical heterodimer. In particular, it relates to pairs of synthetic peptides (whose members are designated ACID-pl and BASE-pl herein and A-l and B-l in U.S. Application, Serial No. 07/829,140, to which the subject application claims prior¬ ity) in which the amino acid sequence has been designed in such a manner that the two members of the pair "stick" to each other with great specificity and avidity. In con¬ trast, neither peptide sticks efficiently to itself. The amino acid residues which comprise the members of the peptide pairs can be naturally-occurring amino acid resi¬ dues, non-naturally occurring amino acid residues or modified amino acid residues. As described herein, under physiological conditions of temperature and pH, homodimers of the synthetic peptides are very unstable, relative to heterodimers of the synthetic peptides and, in an equil¬ ibrium mixture of the two peptides, the heterodimer is favored over the homodimers by at least 1-million-fold. In one embodiment of the present invention, two peptides (designated A-l and B-l in U.S. Application,

Serial No. 07/829,140) designed to bind one another spec¬ ifically and avidly and to preferentially form a helical heterodimer have been synthesized and characterized, using known methods. As described herein, under physiological conditions, individual peptides have been shown, using circular dichrois (CD) spectroscopy, to be predominantly unfolded in isolation and when combined, to associate preferentially to produce stable, parallel, coiled-coil (helical) heterodimers. Further, the degree of preference for the heterodimers has been estimated by studying the stability of the heterodimers and homodimers, using CD methods; results showed that the heterodimer has much greater stability than either of the homodimers. The observed difference in stability between the homodimers and the heterodimer suggests that the heterodimer is preferred over the homodimers by at least 1-million-fold. In addition, the oligomerization state, helical content and helix orientation can be assessed, using, respective¬ ly, sedimentation equilibrium studies, CD and disulfide bonding of the peptides in the desired parallel orient¬ ation, followed by measurement of the concentration depen¬ dence of stability.

In a specific embodiment of the present invention, two peptides, each containing two and 14 amino acid re- peats designated, respectively, ACID-pl and BASE-pl, (designated A-l and B-l in a related U.S. Application, Serial No. 07/829,140) were designed to bind to one another specifically and avidly and to preferentially form a helical heterodimer. The two peptides have been syn- thesized and characterized using known methods. Using the same methods as described above and further described herein, under physiological conditions, the individual peptides (ACID-pl and BASE-pl) were shown to be pre¬ dominantly unfolded and an equimolar mixture of ACID-pl and BASE-pl has been shown to preferentially form hetero¬ dimers which are stable, parallel in orientation and highly helical. The heterodimer has been shown to have greater stability than either of the homodimers. The amino acid composition and order of the constituent pep- tides were designed to introduce destabilizing electro- static interactions in the homodimers that would be re¬ lieved in the heterodimers. The two peptides differ only at two positions (designated e and g, as defined below) by a single amino acid. The ACID-pl peptide contains, at these two positions, an acidic amino acid, such as glu- tamic acid; the BASE-pl peptide contains, at these two positions, a basic amino acid, such as lysine. Measure¬ ments of the specificity of dimer formation demonstrate that these peptides have at least 10⁵-fold preference for the heterodimeric state. Studies of the pH and ionic strength dependence of stability confirm that electro¬ static destabilization of the homodimers provides the primary driving force for the specificity of heterodimer formation. Finally, the oligomerization state, helical content and helix orientation were assessed using respec¬ tively, sedimentation equilibrium studies, CD and di- sulfide bonding of the peptides in the desired parallel orientation, followed by measurement of the concentration dependence stability. The synthetic peptide pairs of the present invention can be used as affinity reagents to isolate associated molecules. For example, a synthetic peptide pair, such as ACID-pl and BASE-pl, can be used in place of biotin-avidin (e.g., in a biotin/streptavidin affinity method), epitope tagging or immunoaffinity purification methodology. In addition, because the heterodimer is preferentially formed and very stable under physiological conditions, the syn¬ thetic peptides of the present invention are suitable for use in vivo. In this embodiment, for example, one member of the synthetic peptide pair (e.g., ACID-pl) can be produced as part of a chimeric or hybrid peptide product (chimeric peptide 1) , which includes the synthetic peptide (one member of the peptide pair) and an additional com¬ ponent which is not a synthetic peptide of the present invention. The second component can be a peptide, poly- peptide, glyco- or other protein, a detectable label or a small organic molecule, which is to be joined or brought into contact with another molecule. The other member of the synthetic peptide pair can be, for example, included in a second chimeric or hybrid product (chimeric peptide 2) which includes the synthetic peptide (e.g., BASE-pl) and the molecule with which the additional component of the first chimeric product (chimeric peptide 1) is to bind or otherwise interact with. Such chimeric or fusion peptides are also the subject of this invention. They can be produced using known techniques, such as recombinant production methods or chemical (synthetic) methods. They can be produced as one product (e.g., a chimeric peptide which includes both components as produced) or the com- ponents can be produced individually and then joined, using known methods. The synthetic peptide pairs can be used, for example, for in vivo applications in which two molecules or components necessary for a given event must be brought together for the event to occur. The synthetic peptide pairs of the present invention make it possible to bring the components together with great specificity and affinity. Alternatively, the synthetic peptide pairs can be designed to prevent binding or interaction of molecules (e.g., DNA and its DNA binding protein) necessary for an event to occur. The synthetic peptide pairs can also be used for biodegradable procedures involving grafting and artificial sutures.

Brief Description of the Drawings

Figure 1 is a coiled-coil helical wheel represen- tation of amino acid residues present in peptide A and peptide B at the locations indicated.

Figure 2 is a coiled-coil wheel representation of the sequence of the ACID-pl/ACID-pl peptide homodimer (Panel A) , the BASE-pl/BASE-pl peptide homodimer (Panel B) and the ACID-pl/BASE-pl peptide heterodimer of the present invention (Panel C) . In this representation, the sequence of the peptide ACID-pl and peptide BASE-pl is arrayed on a coiled-coil diagram in which the helices of the dimer are viewed from the N-terminus (with the helix axis projecting into the page) . The sequence of the coiled-coil peptide is divided into positions of the heptad repeat, labeled a- g; amino acid residues at positions a and d make up the 4- 3 hydrophobic repeat characteristic of coiled coils. Amino acids are represented by their respective one-letter cod .

Figure 3 presents circular dichroism spectra of the ACID-pl peptide, the BASE-pl peptide and the ACID-pl/BASE- pl heterodimer at 37°C in phosphate buffered saline (PBS) , pH 7.0 (physiological conditions).

Figure 4 presents a circular dichroism melting curve of the ACID-pl/BASE-pl heterodimer at a wavelength of 222nm.

Detailed Description of the Invention The present invention is based on Applicant's dis¬ covery of characteristics of pairs of peptides necessary and sufficient for such peptides to bind or pair specific¬ ally and avidly to one another, to form heterodimers, and for the resulting heterodimers to preferentially fold as a helical heterodimer. As a result of this discovery,

Applicant has designed and produced peptide pair members which, when mixed, associate preferentially to form a stable, parallel coiled-coil heterodimer, rather than their respective homodimeric forms. The peptides of the present invention may be of any length, provided that they are sufficiently long to be stable when paired with the other member of the peptide pair (i.e., in the hetero¬ dimeric form) and to be able to allow the heterodimer to fold into a helical configuration. Generally, the pep- tides will be at least 6 to 8 amino acid residues, gener¬ ally at least 12 to 14 amino acid residues in length and preferably, at least 16 to 20 amino acid residues in length or between 20 or 30 amino acid residues long. In one embodiment, each member of the peptide pair has two 14 amino acid residue "repeats" (two 14 amino acid units) resulting in each pair member being 28 amino acids long. In another embodiment, each pair member is 30 amino acids long. In another embodiment, each pair member is 100 amino acids long. There is no upper limit on the length of peptide members and the appropriate length will be determined by such considerations as the context in which a peptide pair is to be used (e.g., n vitro or in vivo, isolated method, therapeutic or diagnostic techniques) . More than one peptide repeats or units of the present invention can be combined (i.e., present in a single multi-unit peptide) if desired. For example, more than one repeat peptide designed to pair specifically and avidly to a second peptide, as described herein (e.g, designated peptide ACID-pl herein and A-l in U.S. Appli¬ cation Serial No. 07/829,140, to which the subject appli¬ cation claims priority) , can be combined in a first multi- unit peptide and more than one repeat of the second pep¬ tide, (e.g., designated peptide BASE-pl herein and B-l in U.S. Application Serial No. 07/829,140) also described herein, can be combined in a second multi-unit peptide and the wo multi-unit peptides used as the two members of the peptide pair. The individual peptide units in the multi- unit peptide can be the same or different. In one embodi- ment described herein, the individual peptide units differ within a peptide (e.g., there are two or more peptide units which are not identical in sequence) but are the same between peptides (e.g., the two members of the pep¬ tide pair each contain the same peptide unit(s) as the other member) . Amino acids in the peptides can be natu- rally-occurring amino acids, non-naturally-occurring amino acids or modified amino acids. The members of a peptide pair stick avidly to the other member of the pair, but do not bind efficiently to a like member of the pair. The following is a description of the rationale underlying the design of the peptide pairs; design synthesis and charac¬ teristics of peptide pairs, particularly of peptides ACID- pl and BASE-pl, of the present invention; and uses there¬ for. Figures 1 and 2 are coiled-coil helical wheel repre¬ sentations of 30 amino acid residues of the ACID-pl/ACID- pl peptide homodimer (Figure 2, Panel A) , the BASE-PI/ BASE-PI peptide homodimer (Figure 2, Panel B) and the ACID-pl/BASE-pl peptide heterodimer (Figure 2, Panel C) . The sequence of the coiled-coil peptide represented is divided into positions of the heptad repeat, which are labeled a through g. Selection of the appropriate type of amino acid residue for each of the positions relied on studies of naturally-occurring leucine zipper peptides, particularly the Fos/Jun leucine zipper heterodimer and the GCN4 leucine zipper homodimer, as described below. The peptide pairs described herein were designed based on principles learned from Applicant's study of coiled-coil amino acid sequences, including the hetero- dimeric Fos/Jun leucine zipper. The peptide pairs are designed in such a manner that, in isolation, the indi¬ vidual peptides are unfolded and do not join or stick to a like peptide efficiently (i.e., two of the same peptide pair members do not stick efficiently together to form homodimers) . In contrast, when mixed together, the pep¬ tide pair members join preferentially to form a stable heterodimeric coiled-coil. The primary driving force for the specificity of binding to form a heterodimer is elec¬ trostatic destabilization of the homodimers. As shown by results presented herein, the designed peptides fold as parallel, helical dimers which have a great (at least about 10⁵-fold) preference for the heterodimeric state.

Previous work had demonstrated that an approximately 35 amino acid leucine zipper region from Foε and from Jun is necessary for preferential heterodimer formation.

Landschulz, W.H. ≤t_s. S- _t, Science. 240:1759-1764 (1988); Ransone, L.J. and I.M. Verma, Ann. Rev. Cell Biol. 6:539- 557 (1990) . Studies of the Fos and Jun leucine zipper heterodimer, as described in Example 1, showed that eight residues at positions e and g of the coiled-coil peptide (heptad repeat) are the primary determinants of specific¬ ity of pairing of the two peptides (Fos and Jun) to form a heterodimer. As described in Example 1, 8 residues from Fos and from Jun, in the background of the GCN4 leucine zipper, have been shown to be sufficient to mediate pre¬ ferential heterodimer formation. Amino acid residues at all other positions of the heptad can be substituted without significant loss of specificity of Fos/Jun pair¬ ing. Residues present at positions e and g in the Fos peptide are predominantly acidic. There are many basic residues at positions e and g in the Jun peptide. Studies of the pH dependence of stability suggest that, at neutral pH, Fos is destabilized by these acidic residues and Jun is slightly destabilized by basic residues. Under the same conditions of pH, the Fos/Jun heterodimer has greater stability, due to relief of destabilizing interhelical electrostatic interactions and potential ion pairs.

Examination of the x-ray structure of the GCN4 leu¬ cine zipper shows that amino acid residues at positions e and g^» (and g and e') are close in space. The residues present at these positions are of opposite charge in the GCN4 homodimer and are seen to participate in interhelical ion pairs. O'Shea, E.K. et al. , Science. 254:539-544 (1991) . Relying on the above-described information about the Fos/Jun leucine zipper and the GCN4 leucine zipper, pairs of peptides which pair specifically and avidly to one another and subsequently preferentially form a helical heterodimer have been designed and these peptides have been synthesized and characterized.

In the peptide pairs of the present invention, charged amino acid residues are present at positions e and g (and e' and g^»; Figure 1). In one peptide (in Figure 1, designated peptide ACID-pl) , a negatively charged amino acid, such as glutamate (glutamic acid) , is present and in the second peptide (in Figure 1, designated peptide BASE- pl) , a positively charged amino acid, such as lysine, is present. As shown in Figure 1, a negatively charged amino acid residue is present at e (amino acid residues 4, 11, 18, 25) and g (amino acid residues 6, 13, 20, 27) in peptide ACID-pl and a positively charged amino acid is present at e^» (amino acid residues 4, 11, 21, 25) and g* (amino acid residues 6, 13, 20, 27) in peptide BASE-pl. A hydrophobic amino acid, such as leucine, is present at positions a and d (peptide ACID-pl) and a¹ and d' (peptide BASE-pl) . Leucine is a preferred choice because it is the most common amino acid at these positions in naturally- occurring coiled coils. However, not all amino acid residues at these positions must be hydrophobic; there may be other (non-hydrophobic) amino acids present at these positions, provided that those present at these positions in aggregate are predominantly hydrophobic. In deter¬ mining appropriate residues for these positions, reference can be made to work by Hodges and co-workers insert (Hodges, R.S. et al.. J. Biol. Che .. 256:1214-1224 (1981); Hodges, R.S. et a , Peptide Res.. 1:19-30 (1988); Lau, S.Y. et aL., J. Biol. Chem.. 259:13253-13261 (1984)) and Conway and Parry fflnt. J. Biol. Macromol.. 12:328-334 (1990) and Example 1. In one embodiment of the present invention, an asparagine is present at the second a posi¬ tion (in peptide ACID-pl, position a and in peptide BASE- pl, position a¹, both of which represent amino acid 14) in order to favor the parallel orientation and discourage higher order oligomerization. Asparagine is present at the corresponding position in the GCN4 leucine zipper, which is a peptide which folds as a two-stranded parallel coiled coil. The amino acid residues present at positions b, c and f, as represented in Figure 1 can be very varied; almost any combination of amino acid residues can be used, provided that there is an appropriate distribution of hydrophilic amino acid residues at these positions. Selection of amino acid residues appropriate for inclusion at positions b, c and f can be made with reference to Applicants' work (e.g., Example 1) and work by Conway and Parry (see above) .

In one embodiment of the present invention, as repre¬ sented in Figure 2, small, uncharged amino acid residues, particularly those which are good helix formers, are present at positions b and c and b' and c'. Such small, uncharged amino acid residues as alanine and glutamine are used at these locations in order to prevent residues at b and c and b¹ and c* from interacting with residues at positions e and g and e' and g' and, as a result, co - peting with the desired interhelical interactions. In addition, polar residues included in the peptide pair members increase solubility of the peptide. In this embodiment, charged residues, such as glutamate and ly- sine, are included at position f (and f, see Figure 1) because many charged residues are found at this location in naturally-occurring coiled coils. These residues also serve to increase solubility and discourage higher-order oligomerization. A single tryptophan at position f and f• has been included in peptides ACID-pl and BASE-PI as a means of facilitating concentration determination by absorbance. However, they will generally be a charged residue, such as glutamate or lysine.

Thus, the two members (peptide ACID-pl and peptide BASE-pl) of the synthetic peptide pairs of the present invention have been a general formula which can be des¬ cribed with reference to the coiled-coil helical wheel representation shown in Figure 1. The following amino acid residues are present:

In a peptide A 1. at positions e and g. negatively charged amino acids residues, such as glutamate (glutamic acid) ;

2. at positions a and d. a hydrophobic amino acid res¬ idue, such as leucine; and

3. at positions b. c and f. almost any amino acid res- idueε, provided that there is an appropriate distri¬ bution of hydrophilic amino acids, such as a nega¬ tively charged amino acid (e.g., alanine, glutamine) at b and c and such as a positively charged amino residue (e.g., glutamate, lysine) at f.

In peptide B

1. at positions e' and g' . positively charged amino acid residues, such as lysine;

2. at positions a¹ and d'_f a hydrophobic amino acid residue, such as leucine; and 3. at positions b^f. c¹ and f. almost any amino acid residues, provided that there is an appropriate distribution of hydrophilic amino acids, such as a negatively charged amino acid residue (e.g., alanine, glutamine) at b' and c* and such as a positively charged amino acid residue (e.g., glutamate, lysine) at f' . In general, proline will not be used in these peptides. The amino acid residues present at each of these positions in one embodiment of the two pair members are shown in Figure 2 (by their one-letter code) and in Table 1 (peptide ACID-pl, peptide BASE-pl, respectively, by their three-letter codes) .

TABLE 1 AMINO ACID SEQUENCE OF PEPTIDE ACID-Pl AND BASE-Pl

Peptide ACID-pl Peptide BASE-pl

1

2

3

4

5

6

7

8

9

21

22

23

24

25

26

27

28

29

30

*One letter amino acid symbol.

Figure 2, the amino acid residues present at posi¬ tions a through g of peptide ACID-1 in the heterodimer (Panel C) correspond to amino acid residues 1 through 30 for peptide ACID-pl in Table l and as shown in Table 2. Amino Acid residue

Leu Asn Leu Leu

Glu Lys Trp Lyε

Glu Glu Glu Glu

In Figure 2, the amino acid residueε present at positions a' through g' of peptide BASE-pl in the hetero¬ dimer (Panel C) correspond to amino acid residues 1 through 30 for peptide BASE-pl in Table 1 and as shown in Table 2. Amino Acid residue

Leu Asn Leu Leu

Ala Gin Ala Gin Ala

Gin Ala Gin Ala Gin

Leu

Leu Leu

Leu

Lys

Lys Lys

Glu Lys Leu Glu Lys Lys Lyε Lyε

The amino acid residues can additionally include Cys-Gly- Gly, which is generally added at the N-terminal of the peptide and not part of the 30 amino acid residue peptide of the present invention but, rather, is included for assay purposeε only. For example, Cyε-Gly-Gly can be added to the peptide (before amino acid reεidue 1) to assiεt in the aεεeεsment of helix orientation (parallel vs. antiparallel) of the heterodimer. To do this, a disulfide-bonded peptide (disulfide-bonded in parallel orientation) is assessed by measuring its stability as a function of peptide concentration; alternatively, molecu¬ lar weight can be measured as a function of peptide con¬ centration by sedimentation. If the helices are parallel, the disulfide-bonded peptide is expected to have stability independent of peptide concentration and molecular weight equal to the molecular weight of the dimer independent of peptide concentration.

Other short peptide pairs which pair specifically and avidly and subsequently preferentially form helical heter¬ odimers can be produced, using the teachings of the sub- ject application as to the appropriate amino acid residues and known synthetic methods. Once a peptide pair has been produced, its ability to pair specifically and avidly and to preferentially form helical heterodimers can be as- εeεεed, as described herein for peptide ACID-pl and pep- tide BASE-pl, using known methods.

The members of the peptide pairs of the present invention can be produced using known methods, such aε chemical synthesiε or recombinant/genetic engineering technology. For example, they can be synthesized as described in Example 2 or in much the same manner as described in Example 1 for the synthesis of the Fos and Jun leucine zipper peptides. Alternatively, the peptides can be produced in an appropriate host cell by expresεing DNA or RNA encoding the peptide sequence. As used herein, the term synthetic refers to peptides of the present invention made by any method (e.g., chemically or by recombinant or genetic engineering methods) . Peptide pairε of the preεent invention have many uses, in both in vitro and in vivo contexts. For example, peptide pairs can be used as affinity reagents, in much the same way as or as a replacement for other binding pairs. For example, they can be used in place of biotin-streptavidin, epitope tagging methods or immunoseparation methods. In this use, a member of a peptide pair can be attached or linked to another molecule (e.g., another peptide, polypeptide, glyco- or other protein, or a detectable label or small organic molecule) or to a solid support (e.g., a column, particle, filter, plastic plate) by known methods, such as a component of a fusion protein or through a linker (e.g., the Cys/Gly/Gly referred to above) for attachment to a solid surface. A molecule to which the second member of the peptide pair is attached (e.g., by chemical or recom- binant methods) can be separated or isolated by contacting a mixture containing the molecule-peptide pair with the solid surface bearing the second member of the peptide pair, under conditions appropriate for sticking or pairing of the peptide pair members. The fraction of the mixture which is not the molecule to be separated or isolated will not become affixed to the solid support and can be removed simply by separating the solid support from the remainder of the mixture. The bound molecule (bound as a result of pairing of the peptide pair members) can be released from the solid support by, for example, changing the pH and/or temperature of the bound fraction. Such a method of separating or isolating a molecule in this manner can be used, for example, for purification of a molecule to be used for other purpoεeε (e.g. , where presence of a par¬ ticular substance is indicative of the preεence or absence of a diseaεe or condition) . Peptide pairs of the present method can alεo be uεe for in vivo purpoεeε, such aε to block, induce or enhance an event in cellε (e.g., to interfere with binding of two components in a cell where binding is necesεary, thus inducing or enhancing the event) . In this embodiment, peptides of the present invention can be produced in cells in which they are to act by, for example, expression from a vector, such aε a retroviral vector(ε) containing DNA or RNA encoding a chimeric peptide or peptideε. The chimeric peptide in¬ cludes the amino acid residues of the peptide of the present invention and, if desired, a peptide or a poly- peptide which is not a synthetic peptide of the present invention, such aε a peptide which is to act in the cell. Alternatively, peptides of the present invention can be used for radioimaging or to treat diseases, such as malig- nancies. In the case of malignancies, one member of a peptide pair iε expressed in the malignant cell (e.g., from an appropriate vector) . The second member of the peptide pair can be labeled, thereby capable of detection or can be joined with an agent which is capable of detec- tion (e.g., a radioactive molecule or substance, such as ricin, toxic to cells) which binds specifically to the cell expresεing the firεt member of the peptide pair. Such an approach can also be used, for example, in treat¬ ing hyperthyroidism. One member of the peptide pair can be expreεεed (e.g., from a retroviral or other vector) in the thyroid and the second member of the peptide pair joined to radioactive iodine can be administered to an individual in need of treatment. Pairing of the two peptide pair members results in delivery of the radio- active iodine to the thyroid; continuous delivery occurε until the peptide pair is degraded by the body or other¬ wise^' separated.

Members of peptide pairs can be joined to other peptides or proteins (e.g., peptides or proteins to be delivered) by chemical means or included with/incorporated into another peptide or protein which is made by recombi- nant DNA methods. In either case, a peptide pair member may be joined or present at an end of the peptide or protein or at any internal site at which the protein can tolerate insertion of the peptide pair member (i.e., any site at which the peptide pair member can be present and not interfere with the desired function of the other peptide or protein) .

EXAMPLE 1 Assesε ent of the Mechanism of Specificit^y in the Foε-Jun Heteroprotein

Peptide Synthesis and Purification

Peptides were εynthesized using t-BOC chemistry on an Applied Biosystems model 430A peptide syntheεizer with εtandard reaction cycleε modified to include acetic an- hydride capping (for a review εee Kent, S.B.H., Annu. Rev. Bioche . 57:956-989 (1988). Peptide Jun N corresponds to residueε 286-317 of the c-Jun protein (Bohmann, D. et al.. Science 238:1386-1392 (1987); Maki, Y. ≤£ Al_s., proc. Natl. Acad. Sci. USA 84:2848-2852 (1987)), and peptide Foε N corresponds to residueε 162-193 of the c-Foε protein (Van Beveren, C. et aJ , Cell 32:1241-1255 (1983); Van Straaten, F. et al.. Proc. Natl. Acad. Sci. USA 80:3183- 3187 (1983)). Ser-295 of c-Jun and Ser-177 of c-Foε have been replaced with tyroεine to facilitate concentration determination by UV abεorbance measurements. Peptide GCN4 N conεiεts of residues 250-281 of the GCN4 protein (Hinne- buεch, A.G., Proc. Natl. Acad. Sci. USA 81:6442-6446 (1984)). All peptides have the sequence Cys-Gly-Gly appended to the N-terminus, are acetylated at the N-termi- nus, and are amidated at the C-terminus. Peptideε were cleaved by either low/high HF cleavage (Immunodynamicε, Inc., San Diego, CA) or by trifluoromethanesulfonic acid cleavage (Kent, S.B.H., Annu. Rev. Biochem. 57:956-989 (1988) and were desalted on a Sephadex G-10 column (Pharm- acia) in 5% acetic acid. Final purification was by re- verεe-phaεe high performance liquid chromatography (HPLC) (Waterε, Inc. and Applied Bioεystems) using a Vydac pre¬ parative C18 column (2.2 x 25 cm) at 25°C or 50^βC. A linear acetonitrile-H₂0 gradient with segments of 0.1% to 0.2% buffer B increase per minute was used with a flow rate of 20 ml/min. Buffer A consisted of water with 0.1% trifluoroacetic acid, and buffer B consisted of 90% aceto- nitrile, 10% water with 0.1% trifluoracetic acid. The identity of each peptide was confirmed by fast atom bom¬ bardment mass spectrometry (M-Scan, Inc., West Chester, PA or Masε Search, Inc., Modesto, CA) and was found to be within 1 dalton of the expected mass.

Circular Dichroism Studies

Circular dichroism (CD) studies were performed using a 1 cm or 1 mm cuvette (Helma or Uvonic) on an Aviv CD spectrophotometer (model 60DS or model 62DS) equipped with a thermoelectric controller. The buffer used for all CD experiments except pH titrations was 50 mM NaCl, 20 mM NaP0₄ (pH 7.0). All peptide concentrations were deter¬ mined by tyrosine abεorbance (Edelhoch, H. Biochemiεtry 6:1948-1954 (1967)) at 275.5 nm in 5.4-6 GuHCl (Schwarz/- Mann Biotech Ultra-Pure grade) uεing an Aviv UV/VIS εpec- trophotometer (model 18DS or 14DS) . The molar ellipticity at 222 nm, 0^βC, was measured for the Fos and Jun N, N_in, ^N _out' ^N _a._d' ^N _e.a disulfide-bonded homo- and heterodimers and for the GCN4 N homodimer. All valueε were found to be within the range -27,000 to -33,000 deg cm² dmol"¹, indi- eating that the peptideε are > 80% helical. Thermal melting curveε were determined by monitoring the CD signal at 222 nm as a function of temperature. The pH dependence of stability was meaεured in 50 mM NaCl, 10 mM NaP0₄ at variouε pH values. Because the diεulfide-bonded GCN4 N dimer is very stable, 2 M GuHCl (Schwarz/Mann Biotech Ultra-Pure grade) was included in the buffer used for monitoring the pH dependence of stability of this peptide. The same overall shape of the pH dependence curve iε obtained in the abεence of GuHCl. Reverεibility was checked for all thermal melts. In general, melting curves obtained above pH 8 are not reverεible, presumably because of chemical modification and/or degradation (observed by HPLC) . Disulfide-bonded dimers that have irreversible melting curves (<75% of folded signal is recovered upon cooling) at pH 7.0 are marked (asterisk). These peptides have undergone chemical modification and/or degradation during the thermal melt, as judged by subsequent HPLC analysis.

Determination of the τ_m

The T_m was determined by curve fitting the thermal denaturation curve to the following equation using a nonlinear least squares-fitting program (Kaleidagraph,

Synergy Software) : θ = Θ_F (O K) + m_FT + [θ_y (O K) + n ,T = G_f (O K) - m_FT]

-ΔH/RT + ΔS/R (-ΔH/RT + ΔS/R) [e /l + e ]

where T is temperature in K; θ is the CD signal at 222 nm; Θ_F (O K) is the value for the CD signal of the folded peptide extrapolated linearly to 0 K; m_F is the slope of the temperature dependence of the CD signal for the folded peptide; m_y is the slope of the temperature dependence of the CD signal for the unfolded peptide; θ,_j (0 K) iε the value for the CD εignal of the unfolded peptide extra- polated linearly to 0 K; ΔH is the enthalpy of unfolding at the midpoint of the thermal denaturation curve; and ΔS is the entropy of unfolding at the midpoint of the thermal denaturation curve. The T_Λ is the temperature at which the fraction unfolded is equal to the fraction folded (ΔG = 0), and it is equal to ΔH/ΔS. The aεεumptionε that were made in using the equation above are as follows: thermal melting curves are two state, described by an equilibrium between unfolded and folded peptide; and the enthalpy and entropy of unfolding are independent of temperature. Becauεe some of the Fos peptide homodimerε and Foε N_in-GNC4 N_in peptide heterodimerε are not completely folded at O'C, the slope of the folded baseline (m_F) and the value for the CD signal of the folded baseline extra- polated to O K (Θ_F (O K) ) for each of these dimers were ^• determined from a melting curve at pH 2.0 (conditions where the Fos-containing peptides are more stable) .

For each peptide the T,,, was also determined by taking the first derivative of the CD signal (θ) with respect to temperature^'1 (temperature in K) and finding the minimum of this function (Cantor, C.R. and P.R. Schimmel, Bio¬ physical Chemistry. W.H. Freeman, New York (1980)). All reported values of T_m are those determined from curve fitting. The error in the measurement of T (estimated from the width of the dθ/d(l/T) plot and from repeated curve fits starting from independent points) is ±2*C except in caseε in which 20'OT,, > 80*C, where the error is ±5°C. The determinations of T_m by dθ/d(l/T) and by curve fitting agree to within the estimated errors. Additionally, the T_m for each disulfide-bonded dimer was measured as a function of peptide concentration (over at least a 2.5-fold range of peptide concentration in the low micromolar range, as estimated by the ratio of the CD signal at low and high concentration) to determine if the dimers were associating to higher order oligomers (O'Shea, E.K. et al.. Science 243:538-542 (1989); O'Shea, E.K., et al. , Science 245:646-648 (1989)). Peptide dimerε that were found to have a T_m that iε dependent upon peptide concentration (~3^*C change in T_m over a 3- to 4-fold concentration range) , and the T_m reported iε the higher of the two measurements.

Redox Experiments

The disulfide-bonded heterodimer was incubated in redox buffer consisting of 100-500 μM reduced glutathione (Sigma) , 100-5— uM oxidized glutathione (Sigma) , 50 mM NaCl, 10 mM NaP0₄ (pH 7.4) at -23*C in an anaerobic cham¬ ber (Coy Laboratory Products, Inc.). Reactions were equilibrated at a total peptide concentration of -10-50 μM for 6-16 hr and quenched under anaerobic conditions with concentrated formic acid to a final concentration of 5% by volume (pH < 2) . The reaction products were analyzed by microbore HPLC (Waters, Inc.) using a linear-acetonitrile- H₂0 gradient with segments of 0.1% to 0.25% increase in buffer B per minute. Analytical Vydac C-18 column (0.46 x 25 cm) was used at a column temperature of 25^*C, 40^βC or 50^βC with a flow rate of 0.2 ml/ in. Relative concentra¬ tions of the disulfide-bonded hetero- and homodimerε were determined by integration of the corresponding peaks (absorbance at 229 nm was monitored) . Each redox reaction was determined to be at equilibrium by repeating the reaction using an equimolar mixture of reduced peptideε aε the εtarting material. The valueε for ΔGs_p„c. obtained from theεe two different εtarting points agreed to within 0.1 kcal/mol.

RESULTS

pH Dependence of Stability in the Fos and Jun Leucine Zipper Peptides

The relative stabilities of the Fos and Jun peptide heterodimer and homodimers (Table 1) indicate that the Fos peptide homodimer is subεtantially leεs stable than the Fos Jun heterodimer and the Jun homodimer (O'Shea, E.K. et al.. Science. 245:646-648 (1989)); the new peptideε are εhorter, lacking both an N-terminal residue and 7 amino acids following the last leucine of the leucine repeat (see Experimental Procedures) . As a first step toward identifying sources of stabilization and destabilization in the new Fos and Jun leucine zipper dimers, the pH dependence of the disulfide-bonded peptide pairs was studied.

The thermal stability of the Fos and Jun homodimerε iε pH dependent in a dramatic way; the T_B (melting temper- ature determined from the midpoint of the thermal unfold¬ ing transition, see Experimental Procedures) of the Fos homodimer increaseε -40*C from neutral to acidic pH, and the stability of the Jun homodimer increases -20'C from neutral to basic pH. The stability of the heterodimer changes approximately as expected from an average of the pH dependence of the homodimers.

These large effects of pH on stability can be ex¬ plained in part by examining the sequences of the Fos and Jun leucine zippers. A particularly uεeful repreεentation of the sequence is the coiled-coil diagram in which the helices of the dimers are viewed from the N-terminus with the helix axis projecting into the page. The sequence of the coiled-coil proteins can be divided up into positions of the heptad repeat, labeled a-g (Hodges, R.S. gt al.. Cold Spring Harbor Symo. Quant. Biol.. 37:299-310 (1972); McLachlan, A.D. and M. Stewart (J. Mol. Biol. 98:293-304 (1975)). Residues at positions a and d comprise the 4-3 hydrophobic repeat characteristic of coiled coils, and residueε at positions e and g are predominantly charged amino acids that can be involved in intra- or interhelical electrostatic interactions (Hodges, R.S. e£ al. _r Cold Spring Harbor Symp. Quant. Biol._f 37:299-310 (1972); McLachlan, A.D. and M. Stewart fJ. Mol. Biol. 98:293-304 (1975)) . The Fos leucine zipper is very acidic, with a high concentration of acidic amino acids at positionε e, g and b. Becauεe each chain has a large net negative charge at neutral pH (each chain has a net charge of -5) , one ex¬ pects that dimer formation would be disfavored due to general electrostatic repulsion between monomers. Ad- ditionally, the alignment of four negatively charged side chains at poεition g along one face of the helix iε ex¬ pected to be a source of intrahelical destabilization. The large increase in stability of the Fos homodimer upon titration to low pH can, thus, be explained to result from the relief of destabilizing intra- and interhelical elec¬ trostatic interactions between acidic residues close to the hydrophobic interface of the dimer.

The Jun leucine zipper has a slight net positive charge at neutral pH (dimer has a net charge of +2 at pH 7) ; in addition, the concentration of charge in the Jun leucine zipper is more spread out than in Fos. These properties are consistent with the lesε dramatic increase in stability of the Jun homodimer at high pH. Qualita- tively, the pH dependence of stability for the Fos-Jun heterodimer changes as expected from an average of the pH dependenceε for the homodimers. This result suggests that the Fos-Jun leucine zipper lackε dominant stabilizing electrostatic interactions that are unique to the hetero- dimer; in such a caεe, a bellshaped pH dependence curve would be expected. In contrast, the stability of the heterodimer increases at acidic pH values, suggesting that intrahelical repulsion (expected from the Fos sequence) is strong. Electrostatic effects provide a posεible explanation for preferential heterodimer formation in the Fos-Jun system. The data suggest that the peptide homodimers are destabilized at neutral pH by residues of like charge, the fos homodimer by acidic reεidueε, and the Jun homodimer, to a leεεer extent, by baεic residues. The interhelical component of thiε electroεtatic deεtabilization iε re¬ lieved in the heterodimer becauεe the Foε and Jun monomerε are of opposite charge. The Inside Residues Are Responsible for Specificity and pH-Dependent Stability

To probe the specificity of the Fos-Jun leucine zipper interaction further, structurally based hybrid peptides were made by replacing portions of the Fos and Jun sequences with sequence from GCN4. A peptide cor¬ responding to the leucine zipper region from GCN4 forms very stable homodimers (O'Shea, E.K. et al.. Science 245:646-648 (1989)). As Fos and Jun are likely to fold aε coiled coilε, the boundary between the Foε or Jun sequence and the GCN4 sequence was set by dividing the helical wheel diagram into two groups of residueε: the "inεide" group, conεisting of the predominantly charged residueε at poεitionε e and g, and the "outside" group, consisting of residueε from poεitionε b, c and f.

Two sets of hybrid leucine zipper peptides were constructed. One set of peptides has native sequence (N) from Fos or Jun at the inside positionε and outside se¬ quence from GCN4; these peptides are referred to as N_in. The other set of peptides contains GCN4 sequence inside and Fos and Jun sequence outside; these peptides are referred to aε N_(χJt.

The preference for heterodimer formation was quantit- ated from a redox experiment in which an equimolar mixture of the cysteine-containing Fos and Jun peptides is equili¬ brated in a redox buffer that facilitates disulfide bond formation. K_redox is determined from the ration of di¬ sulfide-bonded heterodimer to homodimers. The free energy of specificity for heterodimer formation (ΔG ) is equal to -RTlnK_red0X + RTln2. These experimentε indicate that there is -2.3 kcal/mol (-100:1) preference for heterodimer in the native peptides.

The N_in peptides also form heterodimers preferential¬ ly, but with reduced specificity (ΔG_spec iε -1.2 kcal/mol). The decreaεe in specificity of the N_jn peptides appears to ariεe from a decrease in stability of the N_in heterodimer; the N_;i„n heterodimer is lesε stable than the native hetero- dimer but the stabilities of the N_in homodimerε are the same as the corresponding native homodimers (Table 1) . In contrast, the ^ peptides show essentially no specificity (ΔG_βpec is -0.1 kcal/mol). As expected if there is no preference for heterodimer over homodimers, the N_out het¬ erodimer has a stability that iε intermediate between that of the two N_out homodimers (Table 1) . Therefore, the inside residueε of the Foε and Jun leucine zipper are neceεsary and sufficient to mediate preferential hetero¬ dimer formation.

The N_jn peptides show pH-dependent stability that closely resembles that of the native Foε and Jun peptideε. In contraεt, the pH dependence for each of the ^ dimerε doeε not resemble that of the corresponding native dimer, but resembles that of the GCN4 leucine zipper peptide. Therefore, residues at the inside positions (positions a, d, e and g) are alεo largely reεponsible for the pH-depen- dent stability observed with the Fos and Jun peptide dimers.

Seguence Requirements for Specificity

To investigate more thoroughly the sequence require¬ ments for specificity, all posεible combinationε of the previously described seven peptides were made (there are 28 possible combinations, and 21 of these are hetero¬ dimers) , and the stability of each disulfide-bonded dimer was measured by thermal denaturation. There are large differences in stability; the T_ms range from 17^βC-38^βC. The free energy of εpecificity can be approximated by the difference in T_m between a given heterodimer and the average of the T_ms for the corresponding homodimers (ΔT = T_m(heterodimer AB) - 1/2 [T_m(homodimer AA) + T_m(homodimer BB) ]) . For the Fos-Jun peptideε, ΔT appearε to be a quantitative measure of specificity, as it is linearly related to the free energy of specificity, ΔG_βpec, with a *. proportionality constant of 7.4'C/kcal.

Using ΔT_m as a measure of specificity, the hetero- dimeric peptide pairs can be grouped into three classes: specificity, antispecificity and additive. The specifici¬ ty classes includes peptides pairs with positive, nonad- ditive differenσeε in T_m ( T_m ≥+8'C); the antiεpecificity class contains peptide pairs with negative, nonadditive differences in T_m HI (^λΔT„ID ≤-8'C)'; and the additive class consiεtε of peptide pairε in which the stability of the heterodimer is intermediate between that of the homodimers (+8°C > ΔT_m -8^βC) .

All peptide dimers combining Jun sequence inside with Fos sequence inside fall into the specificity claεs.

Although the mechanism of antispecificity is not apparent at this time, all members of the antispecificity class have Fos sequence inside combined with GCN4 sequence inside. The other peptide combinationε fall into the additive class, with one exception (the Jun N_in._{Jun Nout} heterodimer falls into the specificity class, for reasons that are not readily apparent) . The most striking result is that specific heteromdimer formation is observed with all peptide pairs containing Fos sequence inside combined with Jun εequence inεide, regardless of the sequence at the outside position, reinforcing the previous conclusion that the inside residues (positions a, d, e and g) are the major determinant of peptide pairing.

The inside residues consist of the predominantly hydrophobic positions (a and d) and predominantly charged positions (e and g) . GCN4-based hybrid peptides contain¬ ing native Fos and Jun εequence at the hydrophobic posi¬ tions (N_ad) or charged positionε (N_{e g}) were made to evalu¬ ate the contribution of these groups of residueε to εpeci- ficity. The N_{e g} peptides form heterodimerε with specific- ity (ΔT_m and AG^,.) and stability (T at least as great as that of the native sequences (Table 1) . In contrast, the - ^N _a._d peptides are slightly anti-specific.

Thus, 8 residues at positions e and g of Fos and Jun are sufficient to mediate preferential heterodimer for¬ mation. Although the residueε from Fos and Jun that comprise the hydrophobic interface between the helices (poεitionε a and d) are undoubtably important for stabil¬ ity (Smeal, T. et al.. Genes Dev. 3:2981-2100 (1989); Ransone, L.J. and I.M. Verma Annu. Rev. Cell Biol. j6:539- 557 (1990)) , the reεidueε do not appear to be important for εpecificity. We conclude that van der Waalε packing differences do not have a dominant role in the discrimi¬ nation between the Fos-Jun heterodimer and the correε- ponding homodimerε. Rather, the mechaniεm of εpecific heterodimer formation appearε to be predominantly electro- εtatic in nature.

Thiε concluεion iε supported further by the finding that the N_{e g} hybrids εhow pH-dependent εtability similar to that observed with the native peptides. In particular, the Fos N_e-8 homodimer exhibits very εtrong pH-dependent εtability (T_B = 41*C at pH 7 and >90"C at pH 4). In contrast, the pH dependence of the N_ad hybrid dimers does not resemble that of the native peptides. Thus, reεidues at positions e and g in the Fos and Jun sequences also account in large part for the dramatic pH-dependent sta¬ bilities observed with the native peptides.

Discussion

The requirements for εpecificity in the Fos-Jun εyεtem appear to be simple: 8 residues from Foε and from Jun, in a background of the GCN4 leucine zipper, are εufficient to mediate preferential heterodimer formation. pH dependence studies suggest a mechanism for εpecificity in which destabilization of the Fos homodimer by acidic reεidueε (at positions e and g) shifts the dimerization equilibrium toward the Fos-Jun heterodimer. Therefore, preferential heterodimer formation by the Fos and Jun leucine zipper peptides iε largely a thermodynamic con- sequence of Foε homodimer instability (O'Shea, E.K. ≤t al.. Science 245:646-648 (1989)). Destabilization of a homodimer iε also used to provide specificity in the case of the tropomyosin αβ heterodimer (Lehrer, S.S. et al.. Science 246:926928 (1989); Lehrer, S.S. and W.F. Stafford, III, Biochemistry 30:5682-5688 (1991)).

The coupling of the ionization state of residueε at positions e and g to the stability of the Fos and Jun leucine zippers can be rationalized by using the crystal εtructure of a peptide corresponding to the GCN4 leucine zipper (O'Shea, E.K. et al.. Science 254:539-544 (1991)). In thiε crystal structure, the methylene groups of the predominantly charged residues at positions e and g pack against the predominantly hydrophobic.residues at posi¬ tions a and d. Thus, the hydrophobic interface is actual- ly formed by side chains from 4 residues of the heptad repeat. Additionally, terminal charged groups of reεidueε at positions e and g of the preceding heptad are close to each other. It iε likely that the close proximity of negatively charged residueε at poεitions e and g of oppos- ing Fos monomerε would diεrupt the complementary packing εeen at the dimer interface of the coiled coil, accounting for the inεtability of the Foε homodimer at neutral pH.

Example 2 Svntheεis of Peptide ACID-pl and Peptide BASE-

Ei Peptides were εyntheεized uεing small-εcale FMOC HBTU reaction cycleε and acetic anhydride capping on an Applied Biosystems Model 431A peptide εyntheεizer (for a review, εee S.B.H. Kent, Annu. Rev. Biochem, 57, 957 (1988)). Peptideε were cleaved from the reεin with trifluoroactetic acid (TFA) using standard cleavage methodε and were de- εalted on a Sephadex G-10 column in 5% acetic acid. Final purification was by reverse-phase high performance liquid chromatography (HPLC) with a Vydac preparative C18 column (2.2 x 25 cm) at 25*C. A linear acetonitrile-H₂0 gradient with segmentε of 0.1% to 0.2% buffer B increase per minute was used with a flow rate of 10 or 20 mL/ in. Buffer A consisted of water with 0.1% TFA and buffer B consiεted of 90% acetonitrile, 10% water with 0.1% TFA. Peptideε were >90-95% pure, aε judged by analytical HPLC. The identity of each peptide waε confirmed by mass εpectrometry on a Finnegan Laεermat and was found to be within 2 daltons of the expected maεε.

Example 3 Characterization of Homodimerε fACID-pl/ACID-pl and BASE-pl/BASE-Pl) and Heterodimerε fACID-Pl/BASE-Pl CD εtudieε were performed with a 1 mm, 0.5 cm or 1 mm cuvette on an Aviv Model 60DS or 62DS CD spectrophotometer equipped with a thermoelecric controller. All CD studies were done in the presence of phoεphate-buffered εaline (PBS: 150mM NaCl, 10 mM Na phoεphate, pH 7.0). All pep¬ tide concentrations were determined by absorbance at 280 nm in 6 M GuHCl (26) . Thermal melting curveε were deter¬ mined by monitoring the CD εignal at 222 nm aε a function of temperature. Melting temperatureε were eεtimated by taking the firεt derivative of the CD εignal with reεpect to temperature^"1 (temperature in K) and finding the mini¬ mum of thiε function (35) . Reverεibility waε checked for all thermal meltε and, in general, melting curveε per- formed at pH <8 are reverεible (>90% recovery of εtarting CD εignal) . The ACID-pl/BASE-pl Heterodimer is Helical and Stable

Circular dichroism (CD) spectra of ACID-pl and BASE- pl at 37*C in phosphate-buffered saline (PBS), pH7 indi¬ cate that the ACID-pl and BASE-pl peptideε are predomi- nantly unfolded) . In contrast, the characteristic the helical minima at 222 and 208 nm in the spectrum of an equi olar mixture of ACID-pl and BASE-pl indicate that the mixture iε highly helical. At O'C, a 100 uM mixture of ACID-pl and BASE-pl iε - 100% helical({θ}^ = -34,000). The helical εtructure formed by the ACID-pl and BASE-pl mixture iε εtable, aε it undergoes a cooperative unfolding transition when denatured with urea. In contrast, the isolated peptides εhow no evidence for cooperative unfold¬ ing. Sedimentation equilibrium studies of ACID-pl and BASE-pl at 20"C in PBS, pH 7 indicate that this mixture is heterodimeric. Thuε, the ACID-pl and BASE-pl peptides asεociate preferentially and fold aε a stable, helical heterodimer.

The amide proton exchange behavior of the ACID-pl and BASE-pl heterodimer waε characterized by nuclear magnetic reεonance (NMR) . One-dimensional NMR spectra were re¬ corded on a Bruker AMX 500 MHz spectrometer at 20°C with a εweep width of 6024.1 Hz and a recycle time of 1.2 εec. 8192 pointε were collected and water waε suppresεed by continuous irradiation. Trimethylsilylpropionic acid was used aε a εtandard (A. DeMarco, J. Magn. Reson. 26, 527 (1977)). Data were processed with the FTNMR software package (proviced by Dr. Dennis Hare, Hare Research, Inc.) and spectra were apodized with a gausεian window. Since the global εtability of the ACID-pl and BASE-pl hetero¬ dimer under the exchange conditionε iε -6.7 kcal/mol, the expected maximum degree of protection from exchange iε - 10⁵. At a time when protonε that are protected by a factor of -10³ are expected to be half exchanged, many proton resonances have full intensity. Additionally, at a later time point correεponding to a protection of 10⁴, twelve amide proton reεonanceε are present in the spec¬ trum. Therefore, the ACID-pl and BASE-pl heterodimer has slowly-exchanging amide protons with protection factors near those expected from the global stability of the heterodimer, suggesting that the molecule has a well- packed 3° interface.

Characterization of the Diεulfide-Bonded Heterodimer and Homodimers Our design predicts that the ACID and BASE peptides will fold aε parallel, helical dimerε, but at reasonable peptide concentrations the iεolated ACID-pl and BASE-pl are too unstable to εtudy. One way to εtabilize dimeric coiled-coil peptideε iε to join the peptideε with a flex- ible disulfide bond. Therefore, versionε of the ACID and BASE peptideε (ACID-pln and BASE-pln) containing an N- terminal cyεteine followed by two glycineε were εynthe- εized (glycines were added to allow disulfide bond for¬ mation without the distortion of the coiled-coil εtruc- ture.) CD εtudieε demonεtrate that the diεulfide-bonded heterodimer and homodimerε are > 80% helical at 0°C and that each dimer εhowε a cooperative thermal uinfolding tranεition indicating that the ACID and BASE peptides can be stabilized sufficiently with a disulfide bond to permit folding as stable, helical dimers.

^"The concentration dependence of stability was εtudied for each of the diεulfide-bonded peptide dimers to deter¬ mine if the orientation of the helices is parallel and if the dimers aεεociate to a higher-order oligomers. When the ACID-plN and BASE-plN heterodimer and homodimer are joined in the parallel oreintation, the stabi;lity of each dimer is independent of peptide concentration over a range of concentration from -2.5 uM to 170 uM. In contraεt, a heterodimer disulfide-bonded in the antiparallel prientat- ion haε a CD εignal that is dependent on peptide con¬ centration, indicating that the antiparallel dimers are aεεociating to higher-order oligomers. The antiparallel heterodimer consiεtε of the peptide ACID-plN (N-terminal Cyε-Gly-Gly) diεulfide bonded to a BASE peptide that haε a C-terminal sequence of Gly-Gly-Cys. In PBS containing 2.25 M GuHCl at 0°C, a 0.9 μM sample of the antiparallel dimer has a [θ]₂₂₂ value of -11,080 and a 24.5 μM sample has a [θ]₂₂₂ value of -25,300. In contraεt, the helicity of the parallel, diεulfide-bonded ACID and BASE hetero¬ dimer and homodimers iε independent of peptide concen¬ tration over a sample concentration range from -2.5 μM to 170 μM. Collectively, these experiments indicate that the ACID and BASE peptides are parallel and dimeric.

Mechanism of Specificity

Our design strategy sought to drive preferential heterodimer formation by destabilizing the homodimerε rather than by εtabilizing the heterodimer. If eletro- εtatic destabilization of the homodimers is occurring, the homodimers will be unstable in conditions of neutral pH and low ionic strength and will become more stable when the charges are titrated at extremes of pH or when the ionic strength iε increaεed. In contraεt, if stabili¬ zation of the heterodimer by ion pairs is present, the heterodimer will be most stable at neutral pH and will become less stable as the charge on residues involved in ionic interactions is titrated at pH extremes. Thus, a bell-shaped pH dependence of εtability curve would be obεerved. Additionally, if electroεtatic stabilization is important, the heterodimer will be moεt stable at low ionic strength and will become less stable as residues involved in ionic pairs are shielded by ions.

The increased εtability of the disulfide-bonded species allowed us to investigate the pH and ionic εtrength dependence of εtability of the homodiemrε. The εtability of both disulfide-bonded homodimers is strongly pH and ionic strength dependent. The ACID-plN homodimer is > 80% more stable at acidic pH than at pH 7, demon- εtrating clearly that it iε destabilized by acidic resi¬ dues at neutral pH. Similarly, the BASE-plN homodimer is deεtabilized at neutral pH by positively charged basic residues, as it becomes -20°C more stable as the pH is changed from neutral to basic pH. Additionally, the di- εulfide-bonded homodimers are stabilized by increasing ionic strength. These results indicate that the homo¬ dimers are deεtabilized by unfavorable electroεtatic interactionε.

Intereεtingly, at neutral pH the stability of the BASE-plN homodimer is much greater than that of the ACID- plN homodimer. It is likely that part of the reason for the greater εtability at neutral pH of the BASE-plN homo¬ dimer as compared to the ACID-plN homodimer iε the differ¬ ence in the length of the sidechains of reεidueε at posi- tionε e and g. Whereaε ACID-plN has glutamate (containing two methylene groups) at position e and g, BASE-plN has lysine (four methylene groups) . The longer lysine side- chain allows for more flexibility and solvation of the terminal charged group. The idea that the length of sidechains at positions e and g is important in determining homodimer stability is supported by studies of BASE peptides containing the non-natural amino acids ornithine (three methylene groupε) or diaminobutyric acid (two methylene groups) . BASE homodimers containing these non-natural amino acids are predominantly unfolded and the stability of the correεponding ACID/BASE heterodimer decreaεes with the decreasing sidechain length. Two analogueε of the BASE-plN peptide were studied; one con¬ taining diamobutyric acid at positions e and g (DAB-plN) , and the other containing ornithine at these positions (ORN-plN) . Although the disulfide-bonded BASE-plN homo¬ dimer (containing lysine at poεitionε e and g) has a T_B of 66*C, neither the disulfide-bonded DAB-plN homodimer nor the ORN-plN homodimer shows evidence for structure at O'C or evidence for a cooperative thermal unfolding trans¬ ition. Additionally, the disulfide-bonded heterodimers formed between ACID-plN and DAB-plN or ORN-plN have T_m' ε of -63"C ad -80'C, respectively (the ACID-plN/BASE-pOlN heterodimer has a T_B >100*C) . In contrast to the stability of the homodimers, the stability of the heterodimer is relatively independent of ionic εtrength and pH. Becauεe the diεulfide-bonded ACID- plN/BASE-plN heterodimer cannot be thermally denatured in the abεence of chemical denaturant, the pH and ionic strength dependence of the ACID-pl + BASE-pl mixture was studied in the absence of a disulfide bond. From pH 4 to pH 9 the εtability of the ACID-pl + BASE-pl heterodimer iε relatively independent of pH. It iε difficult to assess the stability of the heterodimer at pH extremes (3.5 > pH > 10.5) because the ACID-pl + BASE-pl thermal denaturation curve contains more than one transition, suggeεting tl-at homodimerε are populated εignificantly. It is puzzling that the heterodimer does not exhibit pH and ionic strength dependent stability, given that the heterodimer has charged sidechains positioned to be able to form salt bridges ( of the same type seen in the GCN4 leucine zipper structure) . One poεεible explanation for the lack of pH and ionic εtrength dependent stability iε that salt bridg- eε form in the heteordimer but they are not significantly εtabilizing. There iε precedent for thiε explanation, aε the GCN4 leucine zipper peptides show little pH and ionic strength dependent stability even though salt bridges are seen in the crystal structure. Alternatively, the stabi¬ lizing effect of salt bridges could be coincidentally offset by destabilizing intrahelical repulsion. It is plauεible that such compensating effectε exiεt, aε the Fos/Jun heterodimer is more stable at low pH than neutral pH. The greater εtability at low pH suggeεtε that there is substantial intrahelical repulsion at neutral pH. In any case, these studies of pH and ionic εtength dependence suggeεt that deεtabilization of the homodimerε makeε a major contribution to preferential heterodimer formation.

Ouantitation of Specificity

Because the preference for heterodimer in the ACID and BASE peptides is so large, the ratio of heterodimer to homodimer cannot be measured readily from an equilibrium mixture of the two peptideε. However, the degree of εpecificity can be eεtimated becauεe Ks._p„ec (the equilibrium conεtant deεcribing the ratio of heterodimer to homodimer) iε linked thermodynamically to the diεεociation conεtantε for each of the dimerε. Therefore, the dissociation constants for each dimer were determined so that the degree of specificity, AG_βpec (*-RTlnK_βpec) , could be esti¬ mated. The following quantitation of specificity should be considered an estimate becauεe the linkage relationεhip relating the diεsociation constantε relieε upon the as- εumption that the monomer-dimer equilibria are two-εtate. The diεεociation conεtant for the ACID-plN and BASE- plN heterodimer waε determined by monitoring the CD εignal aε a function of urea concentration. Theεe reεulting denaturation curve waε fit to a two-state model for mono¬ mer-dimer equilibrium to obtain a disεociation conεtant for the heterodimer of 3 x 10^"8 M. A two-εtate model for unfolding waε aεεumed and the urea denaturation curve for the heterodimer waε fit with a non-linear least squares fitting program (KaleidaGraph, Synergy Software) to the following equation that describeε the GuHCl dependence of the CD εignal for a monomer-dimer equilibrium: θ = θd + (Θ_B - θd) [-_e ⁽'^AGβ*^{,nCGuHC1 ) RT}/4C_τ + ( (e⁽-^{AGβ+ra[GuHC1] ) RT)2}/16C_τ ²+e-^{( '}

AG"+m[GuHC1] )/ T /2 C ) ^{1 2}1 where θ = CD signal at 222 nm; θ_d = guanidine dependence of the CD signal of the dimer; Θ_B = guanidine dependence of the CD of the monomer; ΔG^# = free energy of folding under standard state conditionε of 20^βC and 1 M concen¬ tration of peptide chainε; ■= εlope of the guanidine dependence of ΔG; R = gaε conεtant; T ■= temperature in K; C_τ = total concentration of peptide chainε. From thiε fit, a diεεociation conεtant of 3 x 10^"8 waε obtained. An estimate of the disεociation conεtant for each of the homodimerε was obtained from measurementε of the helical CD εignal as a function of peptide concentration. The disεociation conεtant for the BASE-pl homodimer iε -1 x 10^"3 M and that for the ACID-pl homodimer iε > 5 x 10^'3 M. Theεe diεsociation constantε place a lower limit on the degree of preference for heterodimer, ΔG.s-p,ec., of -6.5 kcal/mol (> 10⁵-fold preference for heterodimer) .

Specificity can alεo be eεtimated by meaεuring the difference between the melting temperature (T_m) of the heterodimer and the average of the T_ms (melting tempera¬ ture) for the homodimerε (ΔT_m) . ΔT_m haε been eaεured for other diεulfide-bonded leucine zipper peptideε and haε been εhown to be related to ΔG by a proportionality conεtant of 7.4°C/kcal mol^"1. Δτ_m for the diεulfide-bonded ACID-plN and BASE-plN peptideε iε > 56°C. If the εame proportional relationεhip between ΔT,,. and ΔG exists with the ACID and BASE peptides, this lower limit for ΔT implies that ΔG is at least -7.5 kcal/mol (>10⁵-⁶-fold preference for the heterodimer) .

Thuε, the heterodimer is preferred over the ACID-pl and BASE-pl homodimers by at least -10⁵-fold. This degree of specificity is much greater than that observed for the Fos and Jun peptides where the Fos/Jun heterodimer is preferred by only -10²-fold. Eouivalentε

Thoεe εkilled in the art will recognize, or be able to ascertain using not more than routine experimentation, many equivalents to the specific embodiments of the inven¬ tion described herein. Such equivalents are intended to be encompasεed by the following claims.

Claims

CI-AIKS

1. A εynthetic peptide conεiεting eεεentially of 30 amino acid residues, wherein the amino acid residues are as follows: a) amino acid residueε 4, 11, 18, 25, 6, 13, 20 and 27 are negatively charged amino acid residues; b) amino acid residueε 7, 14, 21, 28, 3, 10, 17 and 24 are hydrophobic amino acid reεidueε; c) amino acid reεidueε 1, 2, 8, 15, 22, 29, 9, 16, 23 and 30 are small, uncharged amino acid resi¬ dues; and d) amino acid reεidueε 5, 12, 19 and 26 are charged amino acid residues.

2. The synthetic peptide of Claim 1 wherein: a) the amino acid reεidueε of (a) are glutamate reεidueε; b) the amino acid reεidueε of (b) are leucine reεi¬ dues or aspartate residueε; c) the amino acid reεidues of (c) are alanine reεi- dues or glutamine residueε; and d) the amino acid reεidueε of (d) are glutamate residues, lysine residueε or tryptophan reεi¬ dues.

3. A synthetic peptide consisting esεentially of 30 amino acid reεidues, wherein the amino acid residueε are as follows: a) amino acid residueε 4, 11, 18, 25, 6, 13, 20 and 27 are poεitively charged amino acid reεidues; b) amino acid residueε 7, 14, 21, 28, 3, 10, 17 and 24 are hydrophobic amino acid reεidueε; c) amino acid reεidueε 1, 2, 8, 15, 22, 29, 9, 16, 23 and 30 are εmall, uncharged amino acid reεi- dues; and d) amino acid reεidues 5, 12, 19 and 26 are charged amino acid residueε.

4. The εynthetic peptide of Claim 3 wherein: a) the amino acid reεidueε of (a) are lysine resi- dues; b) the amino acid residueε of (b) are leucine reεi¬ dues or aspartate residues; c) the amino acid residueε of (c) are alanine reεi¬ dues or glutamine residues; and d) the amino acid residues of (d) are glutamate residues, lysine residues or tryptophan resi¬ dueε.

5. A helical heterodimer consiεting essentially of two synthetic peptideε, deεignated peptide ACID-pl and peptide BASE-pl, wherein: a) peptide A conεiεtε eεsentially of 30 amino acid residues, wherein the amino acid residues are as follows:

(1) amino acid residues 4, 11, 18, 25,

6, 13,

20 and 27 are negatively charged amino acid residues; (2) amino acid residues 7, 14, 21, 28, 3, 10,

17 and 24 are hydrophobic amino acid resi¬ dueε;

(3) amino acid residues 1, 2, 8, 15, 22, 29, 9, 16, 23 and 30 are small, uncharged amino acid residueε; and

(4) amino acid residues 5, 12, 19 and 26 are charged amino acid residues b) peptide B consiεtε eεεentially of 30 amino acid reεidues, wherein the amino acid residues are as follows:

(1) amino acid reεidueε 4, 11, 18, 25, 6, 13, 20 and 27 are positively charged amino acid residueε;

(2) amino acid reεidueε 7, 14, 21, 28, 3, 10, 17 and 24 are hydrophobic amino acid resi¬ dues;

(3) amino acid residues 1, 2, 8, 15, 22, 29, 9, 16, 23 and 30 are small, uncharged amino acid residueε; and (4) amino acid reεidueε 5, 12, 19 and 26 are charged amino acid reεidueε. 6. The helical heterodimer of Claim 5 wherein: a) in peptide ACID-pl:

(1) the amino acid reεidueε of (a) are gluta¬ mate reεidueε; (2) the amino acid reεidueε of (b) are leucine residues or aspartate reεidueε;

(3) the amino acid reεidueε of (c) are alanine reεidueε or glutamine reεidueε; and

(4) the amino acid reεidueε of (d) are gluta- mate reεidueε, lyεine reεidueε or tryp- tophan reεidues. b) in peptide BASE-pl:

(1) the amino acid reεidueε of (a) are lyεine reεidueε; (2) the amino acid reεidueε of (b) are leucine residues or aspartate residueε;

(4) the amino acid reεidueε of (d) are gluta- mate reεidueε, lyεine reεidueε of tryp- tophan reεidues.

A coiled-coil heterodimer comprising at least one pair of εynthetic peptideε, repreεented by the fol¬ lowing helical wheel:

wherein the firεt member of the peptide pair iε deεignated peptide ACID-pl and the εecond member of the peptide pair iε deεignated peptide BASE-pl and: a) the amino acid residues at positions e and g of peptide A are negatively charged amino acid residues; b) the amino acid residueε at poεitionε e' and g' of peptide B are poεitively charged amino acid reεidueε; c) the amino acid reεidueε at poεitionε a and d of peptide A and poεitions a' and d' of peptide B are hydrophobic amino acid residueε; d) the amino acid reεidueε at poεitionε b and c of peptide A and positions b' and c' of peptide B are uncharged amino acid residues; and e) the amino acid residueε at poεition f of peptide A and poεition f' of peptide B are charged amino acid reεidueε.

8. The coiled-coil helical heterodimer of Claim 7 where¬ in: a) the amino acid residueε of (a) are glutamate reεidueε; b) the amino acid residues of (b) are lysine resi¬ dueε; c) the amino acid reεidueε of (c) are leucine reεi¬ dueε; d) the amino acid reεidueε of (d) are alanine reεi- dues or glutamine reεidueε; and e) the amino acid reεidueε of (e) are glutamate reεidues or lyεine reεidueε.

9. A multi-unit peptide compriεing at leaεt two εynthe¬ tic peptides of Claim 1.

10. A multi-unit peptide comprising at least two εynthe¬ tic peptides of Claim 3.

11. A coiled-coil heterodimer of Claim 7, further com¬ prising more than one pair of synthetic peptideε ACID-pl and BASE-pl.

12. A chimeric peptide compriεing a synthetic peptide of Claim 1 and a εecond peptide which is not a εynthetic peptide of Claim 1.

13. A chimeric peptide of Claim 12, wherein the εecond component iε εelected from the group conεiεting of: peptideε, polypeptideε, glycopeptideε, small organic molecules and detectable labels.

14. A chimeric peptide of Claim 13, which iε detectably labeled.

15. A multi-unit peptide comprising at least two chimeric synthetic peptides, each chimeric synthetic peptide comprising at least one synthetic peptide of Claim 1 and a second component which is not a synthetic peptide of Claim 1.

16. A multi-unit peptide of Claim 15, wherein the second component is selected from the group consiεting of: peptideε, polypeptideε, glycopeptides, small organic molecules and detectable labels.

17. A multi-unit peptide of Claim 16 which is detectably labeled.