WO1997015665A1

WO1997015665A1 - Heterotetrameric coiled coil structures

Info

Publication number: WO1997015665A1
Application number: PCT/US1996/016902
Authority: WO
Inventors: Jiri Novotny; Robert Fairman; Thomas B. Lavoie; Hann-Guang Chao; Gary R. Matsueda
Original assignee: Bristol-Myers Squibb Company
Priority date: 1995-10-23
Filing date: 1996-10-21
Publication date: 1997-05-01

Abstract

A composition of matter as shown in the figure wherein: (a) I, II, III and IV are each alpha-helical polypeptides of about 21 to 35 residues; (b) moieties in positions a and d of helices I and IV are each independently amino acid residues; (c) moieties in positions b and c of helices I and III are each independently positively charged amino acid residues; (d) moieties in positions b and c of helices II and IV are each independently negatively charged amino acid residues; (e) moieties in positions e and g of helices I, II, III, and IV are each independently amino acid residues; and (f) moieties in positions f of helices I, II, III, and IV are each independently amino acid residues; and (g) each of I, II, III, and IV is optionally substituted at either or both of its termini with an effector domain or a specific ligand binding domain. Such a composition of matter is useful, among other things, for ligand-targeted drug delivery.

Description

HETEROTETRAMERIC COI LED COIL STRUCTURES

This invention relates to immunoconjugates, drug delivery and immunodiagnosis mediated by complementary, self-associating polypeptide segments. This invention further relates to protein design for development of new proteins containing hetero-activity and/or increased valence.

Coiled coils are protein structural motifs that act as multimerization modules. Intrinsic flexibility of helix-pairing interactions allows this motif to adopt multiple oligomerization states and to dictate heterospecific interactions. A major goal of current research is to understand the rules that govern oligomerization states and heterospecificity in coiled coils.

Nature has taken advantage of electrostatics as one mechanism to drive heteromeric assembly of coiled coils. For example, the oncoproteins fos and jun derive the specificity of their interactions from electrostatic contributions. O'Shea et aj. (1992) Cell 68:699-708. In addition, several recent x-ray crystal structures of coiled-coils have helped us understand at the atomic level the contribution of salt-bridges to coiled coil structure. O'Shea et aj. (1991 ), Science 254:539-44; Harbury et aj. (1993), Science 262:1401-7; Glover & Harrison (1995), Nature 373:257-61. A number of laboratories exploited electrostatics to design specificity into dimeric coiled coil interactions. O'Shea et aj. (1993), Curr, BjoL 3:658-67; Graddis et aj. (1993), Biochemistry 32:12664-71 ; Krylov et aj. (1994), EMBO J. 13:2849-61. Recently, Kim's laboratory (Lumb & Kim (1995), Biochemistry 34:8642-8) found that a specific mutation in the hydrophobic interface in the 'Velcro' peptides (peptides based on the repeating leucine zipper heptads or 7-amino acid segments from GCN4 which form heterodimers through mutations at e and g positions to glutamic acid and lysine) results in the formation of heterotetramers.

Many laboratories have explored multimerization to enhance the avidity of antibodies for their targets. One group incorporated a helix-turn-helix protein dimerization motif into the sequence. Pack & Plϋckthun (1992). Biochemistry 31 : 1579-84. Other groups added free cysteine to achieve dimerization. Better et aj. (1993), Proc. Natl. Acad. Sci. USA 90:457-61 ; Cumber et al. (1992), J. Immunol. 149: 120-126; Carter et aj. (1992),

BIO/TECHNOLOGY 10: 163-7; Kipriyanov et aj. (1994), Mol. Immunol. 31: 1047-58; and Rodrigues et aj. (1993), J. Immunol. 151 : 6954-61. One group shortened the linker between two antibody variable regions, thereby disfavoring self association and biasing the resultant protein towards dimers or 'diabodies.' Hollinger et aj. (1993), Proc. Natl. Acad. Sci. USA 90: 6444-8; and Hollinger & Winter (1993), Curr. Opjn. Biotech. 4: 446-9.

Other groups have produced homotetramers of an antibody variable region. In one approach, corestreptavidin mediates homotetramerization, allowing formation of complexes with biotinylated molecules. Dubel et aj. (1995), J. Immunol. Meth. 178: 201-9. In another approach, a mutant GCN-4 leucine zipper mediates homotetramerization. Pack et aj. ( 1995) , J. Mol. Bioi. 246: 28-34. Previous attempts to produce a tetrameric antibody led to dimeric molecules. Pack and Plϋckthun (1992). Most protein multimerization approaches have a stability problem: To achieve a high stability of interaction, they require cysteine disulfide bridges. Blondel & Bedouelle (1991 ), Protein Eng. 4: 457-61 ; Chang et aj. (1994), Proc. Natj. Acad. Sci. USA 91 : 11408- 12; Pack and Plϋckthun (1992). Alternatively, they require the multimerization domain to have very high intrinsic stability. These approaches work for soluble expressed protein in eukaryotic and prokaryotic expression systems but not for proteins expressed as inclusion bodies. Such proteins denature in the inclusion bodies and so must later properly refold. Additional cysteines decrease the likelihood of proper disulfide formation. High intrinsic stability induces oligomerization, even under relatively denaturing conditions, so properly folded molecules are likely to oiigomerize with improperly folded molecules, complicating purification and reducing yield of fully active molecules.

In the present invention, we designed highly stable, self-associating complementary peptide sequences for selective assembly of two different components with a predictable stoichiometry: 2 + 2. Because there are 7 residues per two turns of α-helix, the inventors and others describe α-helices in units of heptad repeats, denoting each position within a heptad by a-g (Figures 1 (A) and 1(B)).

The present invention concerns a composition of matter as shown in Figure 1(A) wherein:

(a) I, II, III and IV are each alpha-helical polypeptides of about 21 to 35 residues; (b) moieties in positions a and d of helices I and IV are each independently amino acid residues; (c) moieties in positions b and c of helices I and III are each independently positively charged amino acid residues;

(d) moieties in positions b and c of helices II and IV are each independently negatively charged amino acid residues; (e) moieties in positions e and g of helices I, II, III, and IV are each independently amino acid residues; and

(f ) moieties in position f of helices I, II, III, and IV are each independently amino acid residues; and

(g) each of I, II, III, and IV is optionally substituted at either or both of its termini with an effector domain or a specific ligand binding domain.

The present invention also concerns nucleic acids and vectors comprising a nucleic acid sequence coding for heterotetramer-forming polypeptides. The invention further concerns host cells transformed with such vectors. Particularly preferred host cells are BL-21 cells. Most preferred are pLEC-1 and pLKC-1 cells, described hereinafter and deposited as ATCC Ace. Nos.69931 and 69932 (American Type Culture Collection, 12301 Parklawn Drive, Rockville MD 20852-1776).

The present invention also concerns methods for preparing and using compositions of matter as described in Figure 1(A).

Figure 1(A) shows a helical wheel diagram of the composition of matter of the present invention. I, II, III, and IV are polypeptides having an alpha helix secondary structure. The letters a through f represent each such position per two turns of the helix.

Figure 1 (B) shows a helical wheel diagram showing E-K interactions at b and c positions for a preferred embodiment. The peptide sequences (SEQ. ID. NOS.: 1 and 2, respectively) and their heptad positions (H) are listed below, amino terminus to carboxyl terminus.

H: a b e d e f g a b c d e f g a b e d e f g

1: M E E L A D S L E E L A R Q V E E L E S A

2 M K K L A D S L K K A R Q V K K L E S A These polypeptides are based on Lac21 and we refer to them herein as Lac2l t (bLU. IU NO : 1) and Lac21 K (SEQ ID. NO 2). For the biophysical experiments, we acetylated the ammo terminus and amidated the carboxyl terminus

Figure 2 shows circular dichroism spectra of Lac21 E, Lac21 K polypeptides. The concentration of total peptide is 100 μM in 10 mM MOPS, pH 7.5, 25°C.

Figure 3 shows thermal unfolding of 100 μM each a) Lac21 K, b) Lac21 E, and c) Lac21E+Lac21K. (o) 0 M NaCl; (n) 1 M NaCl

Figure 4 shows pH dependence of Lac21 E and Lac21 K, and a 1 :1 mix of the two polypeptides The total polypeptide concentration is 50 μM in 1 mM each of the sodium salts of phosphate, borate, and citrate with measurements at 25 °C.

Figure 5 shows sedimentation equilibrium analysis of (a) Lac21 E, (b) Lac21K and (c) Lac21 E+Lac21 K The total polypeptide concentration is 200 μM in 100 mM NaCl,

10 mM MOPS, pH 7 5 with measurements at 25 °C

Figure 6 shows (A) the DNA sequence (SEQ. ID NO 3) and below it the expressed ammo acid sequence (SEQ. ID NO.- 4) of the synthetic Lac28E gene and (B) the Lac28K peptide (SEQ. ID. NO.: 5)

Figure 7 shows the GPC elution profile of BR96-sFv-Lac28E-PE38 in stabilizing and destabilizing conditions for LacE association

Figure 8 shows an ELISA assay for the effect of Lac28K peptide on the binding of BR96-sFv-PE40 and BR96-sFv-Lac28E-PE38 to LeY-HSA

Figure 9(A) shows the predicted structure of BR96-sFv-Lac28E-PE38 (black) + Lac28K (gray) peptide molecule Figure 9(B) shows a proposed representation of a molecule with 4 antibody heads and 2 PE tails

Figure 10 indicates two different orientations available for combining two proteins

Figure 11 indicates two different ways for bringing four proteins together Figure 12 shows sequences and helical wheel diagrams for a fully heterotetramenc scaffold

In this specification, we may designate helices, helix positions and ammo acid residues as follows We identify each helix by the Roman numeral (I-IV) appearing in Figure 1(A) We identify each position within a helix by the letter (a-f) appearing in Figure 1 (A), with the helix in which it appears written as a superscript Thus, "a¹" refers to position a of helix I, "b¹" refers to position b of helix I, and so forth. We distinguish particular amino acid residues at those positions within the helix by Arabic numeral subscripts, the subscript increasing with distance from the amino terminus. Thus, "a'ι" refers to the amino acid residue in the heptad closest to the amino terminus at position a in helix I, "a^" refers to the amino acid residue in the heptad second closest to the amino terminus at position a, and so forth.

The following definitions apply to the terms and abbreviations used throughout this specification, unless otherwise defined in specific instances.

The term "amino acid residue" refers to both natural amino acid residues (defined below) as well as modified, unusual, and synthetic amino acid residues. Persons of ordinary skill in the art can incorporate such unnatural amino acid residues into polypeptides by synthesis techniques (R. D. Walkup et a . (1995), J. Org. Chem. 60:2630-4) or recombinant strategies (C. J. Noren et aj. (1989), Science 244:182-8). Examples of such amino acids include 2-aminoadipic acid, 3-aminoadipic acid, beta-alanine, 2-aminobutyric acid, 4 aminobutyric acid, 6-aminocaproic acid, 2-aminoheptanoic acid, 2-aminoisobutyric acid, 3-aminoisobutyric acid, 2-aminopimelic acid, 2,4-diaminobutyric acid, desmosine, 2,2'- diaminopimelic acid, 2,3-diaminopropionic acid, N-ethylglycine, N-ethylasparagine, hydroxylysine, allo-hydroxylysine, 3-hydroxyproline, 4-hydroxyproline, isodesmosine, allo- isoleucine, N-methylglycine, N-methylisoleucine, N-methylvaline, norvaline, norleucine and ornithine. Typically, such amino acid residues may be described by the formulas

wherein

R¹ is hydrogen, -R³-R⁴, or

and R² is hydrogen or alkyl, or R¹ and R² together are alkylene; R³ is a single bond, alkyl, alkenyl, alkoxy, or amino(lower alkyl); R⁴ is hydrogen, aryl, cycloalkyl, or cycloalkenyl, wherein the aryl, cycloalkyl or cycloalkenyl group may be substituted with 1 to 5 substituents selected from 1 to 5 halo, 1 to 3 nitro, 1 to 3 cyano, 1 to 3 sulfhydryl, 1 to 3 sulfinyl, 1 to 3 sulfonyl, 1 to 3 sulfoxyl, 1 to 3 hydroxyl, 1 to 3 carboxyl, 1 to 3 haloalkyl, ^ R⁵ ^ R⁵ l-N . j -alkylene-N 1 to 3 « R⁶ , 1 to 3 < R⁶ , and, for cycloalkyl and cycloalkenyl only, 1 to 2 oxo.

R⁵ and R⁶ are each independently hydrogen, alkyl, aryl, cycloalkyl, aralkyl, or cycloalkylalkyl, or R⁵ and R⁶ together are alkylene or alkenylene;

The terms "alkyl," "alk-" and "alkoxy" refer to straight or branched chain hydrocarbon groups having 1 to 12 carbon atoms. The terms "lower alkyl" and "lower alkoxy" refer to groups of 1 to 4 carbon atoms, which are preferred.

The term "aryl" or "ar-" refers to phenyl, naphthyl, and biphenyl.

The term "alkenyl" refers to straight or branched chain hydrocarbon groups of

2 to 10 carbon atoms having at least one double bond, preferably 1 , 2, or 3 double bonds. Groups of two to four carbon atoms are preferred.

The term "alkynyl" refers to straight or branched chain groups of 2 to 10 carbon atoms having at least one triple bond, preferably 1, 2, or 3 triple bonds. Groups of two to four carbon atoms are preferred.

The term "alkylene" refers to a straight chain bridge of 1 to 5 carbon atoms attached by single bonds (e.g., -(CH2)_m- wherein m is 1 to 5), which may be substituted with 1 to 3 lower alkyl groups.

The term "alkenylene" refers to a straight chain bridge of 1 to 5 carbon atoms having one or two double bonds and which is attached by single bonds (e.g., -CH=CH2-

CH=CH-, -CH₂-CH=CH- or -CH₂-CH=CH-CH₂-) which may be substituted with 1 to 3 lower alkyl groups.

The terms "cycloalkyl" and "cycloalkenyl" refers to cyclic hydrocarbon groups of 3 to 8 carbon atoms.

The term "effector domain" refers to a substance having pharmaceutical or diagnostic activity. Exemplary effector domains include antineoplastic agents (e.g., doxorubicin); enzymes or other agents that effect clot lysis (e.g., streptokinase, urokinase); anti-thrombotic agents (e.g., thrombomodulin); substances that initiate complement activation (e.g., kininogen); substances that initiate cell death (e.g., Pseudomonas exotoxin, ricin); proteins or other agents which induce clotting (e.g. tissue factor); bioactive ligands (e.g., epidermal growth factor); enzymes, radionuclides, reporter proteins, or other agents that can be used diagnostically (e.g., Lac-radioactive isotope). The term "ligand binding domain" refers to a polypeptide or other substance that specifically binds to a particular ligand. Exemplary ligand binding domains include antibody variable fragments (e.g., BR96 sFv), CTLA4, FGF, mannose binding proteins, HDL binding proteins, ferritin, and the like. The inventors particularly prefer ligand binding domains specific to antigens found only on tumor cell types, such as BR96. The term "negatively charged amino acid residue" refers to amino acid residues having sidechains that are negatively charged at about pH 6 to 8. Exemplary negatively charged amino acid residues are aspartyl and glutamyl.

The term "neutral amino acid residue" refers to amino acid residues having polar sidechains that have neutral charge at about pH 6 to 8. Exemplary neutral amino acid residues are asparaginyl, cysteinyl, glutaminyl, seryl, threonyl, and tyrosyl.

The term "nonpolar amino acid residue" refers to amino acid residues having nonpolar sidechains. Exemplary nonpolar amino acids are alanyl, isoleucyl, leucyl, methionyl, phenylalanyl, prolyl, tryptophyl, and valyl.

The term "natural amino acid residue" refers to glycyl and the L-form of alanyl, arginyl, asparaginyl, aspartyl, cysteinyl, glutamyl, glutaminyl, histidyl, isoleucyl, leucyl, lysyl, methionyl, phenylalanyl, prolyl, seryl, threonyl, tryptophyl, tyrosyl, and valyl.

The term "positively charged amino acid residue" refers to amino acid residues having sidechains that are positively charged at about pH 6 to 8. Exemplary positively charged amino acid residues are L-forms of arginyl, histidyl, and lysyl. The following definitions apply to abbreviations in this specification, unless otherwise defined in specific instances: BSA bovine serum albumin CD circular dichroism

D TT dithiothreitol EDTA ethylenediaminetetraacetic acid

E G T A 1 ,2-di(2-aminoethoxy)ethane-N, N, W, N> tetraacetic acid ELISA enzyme-linked immunosorbent assay

GPC gel permeation chromatography

HPLC high performance liquid chromatography

HRP horse radish peroxidase

HSA human serum albumin

IPTG isopropyl β-D-thiogalactopyranoside

LeY Lewis Y

MOPS 3-[N-morpholino]propanesulfonic acid (buffer)

PAGE polyacrylamide gel electrophoresis

PBS phosphate buffered saline

PCR polymerase chain reaction

PMSF phenylmethylsulfonyl fluoride (protease inhibitor)

SDS sodium dodecyl sulfate (a detergent)

SE sedimentation equilibrium

For additional abbreviations, see Aldrichimica Acta, Vol. 17, No.

1 (1984).

Persons of ordinary skill in the art can use the heterotetramers of the present invention for a variety of pharmaceutical and diagnostic purposes. The binding domain provides specificity for treatment of particular cells bearing the epitope recognized by the binding domain. Thus, instead of systemic distribution of effector domains, the invention can target them to specific cells. This targeted treatment will decrease the side-effects and allow lower dosing of effector domains. The invention can thus target chemotherapeutic agents to tumor cells, CNS agents to neurons, clot dissolvers to clots. The heterotetramer scaffold provides for multiple binding domains and effector domains. The invention can thus feature multiple binding domains, thus increasing the avidity of the composition. It can also feature multiple effector domains, thus increasing the efficacy per dose of the composition. Alternatively, the invention accommodates different types of binding domains and effector domains. The invention can therefore target different cell types within a tumor, or provide more than one cytotoxic agent to a tumor cell type. For each of the terms mentioned in the summary of the invention, the inventors prefer the moieties listed below. For each of the ammo acid residues, the inventors prefer the L-form. positively charged residues— lysyl (most preferred), arginyl, and histidyl; negatively charged residues— aspartyl, glutamyl (most preferred); nonpolar residues— alanyl, isoleucyl, leucyl, methionyl, phenylalanyl, prolyl, tryptophyl, and valyl (with methionyl, leucyl, and valyl most preferred).

Not all residues at each position within a helix or between helices need be the same. For example, b'ι and b^ are both negatively charged residues, but b'ι may be glutamyl while b>2 is aspartyl.

Peptide synthesis. We decided to test heterotetramer formation not by using e and g positions (as has been done for dimeric coiled coils), rather by considering placement of charged residues at b and c positions (Figures 1(A) and 1(B)) We used as our model system polypeptides based on the tetrameπzation domain of Lac repressor. We had demonstrated that synthetic polypeptides derived from the C-terminal 21 am o acids of Lac repressor form stable four-chain coiled coils. Fairman et aj. (1995), Protein Science 4:1457-69. We studied synthetic polypeptides containing negatively or positively charged residues (preferably glutamic acid or lysine) at both b and c positions in the background of the Lac sequence to determine oligomerization state and heterospecificity. Persons of ordinary skill in the art can prepare the polypeptides I-IV on an automated peptide synthesizer (e.g , Biosearch 9600, Applied Biosystems Model 431 A) following standard protocols.

In such protocols, one begins with the C-terminal residue bound to a resm at its carboxyl terminus and to a protecting group (e.g , t-butoxycarbonyl) at its ammo terminus. One treats the bound residue with

(1) a deprotecting agent such as a mild acid (e.g., trifluoroacetic acid) in an inert solvent (e.g., methylene chloride) preferably in the presence of cation scavengers (e g., anisole);

(2) a tertiary base (e.g., diisopropylethylamine); (3) the next contiguous ammo acid residue in the intended polypeptide sequence with a protecting group at its am o terminus in an inert solvent (e.g., dimethylformamide) in the presence of a coupling reagent (e.g., diisopropylcarbodiimide) in an inert solvent (e.g., methylene chloride), optionally in the presence of an agent that suppresses racemization or dehydration (e.g., hydroxybenzotriazole); and

(4) as an optional step, an amino acid acetylating agent (e.g., acetic anhydride), which caps unreacted amino acids. One repeats this procedure with successive amino acid residues until the polypeptide sequence is complete.

When the protecting group for the starting peptide is base-labile (e.g., fluorenylmethoxycarbonyl), one replaces agents (1) and (2) above with a base (e.g., piperidine, morpholine) in an inert solvent (e.g., dimethylformamide). One may also use multiple peptide synthesis techniques known in the art. See, for example, Tjoeng et aj. (1990), "Multiple Peptide Synthesis Using a Single Support (MPS3)," Int. J. Protein Peptide Res. 35: 141-6.

One may cleave the polypeptide from the resin with an acid, such as hydrofluoric acid, trifluoromethanesulfonic acid, and the like. When the amino protecting group is base-labile, one may deprotect with a mild acid such as trifluoroacetic acid and the like.

Amino acid residues in the polypeptide may have sidechains with reactive functional groups. Examples of such reactive sidechain groups are hydroxyl, carboxyl, amino, mercapto, guanidino, imidazolyl, indolyl, and the like. Such a sidechain may have a protecting group in the foregoing procedure to prevent interference with peptide bond formation. One may conventionally treat the competed polypeptide with deprotecting agents. Those skilled in the art are aware of particular sidechain protecting groups and associated deprotecting agents.

Nucleic acid expression. One may also prepare the polypeptides by known recombinant DNA techniques. See Sambrook J., Fritsch E. F., and Maniatis, T. (1989), Molecular Cloning: A laboratory manual. CSHL Press, Cold Spring Harbor NY.

One may place nucleic acid sequences encoding the polypeptides into expression vectors. The expression vectors preferably comprise one or more regulatory DNA sequences operatively linked to this DNA sequence. As used in this context, the term "operatively linked" means that the regulatory DNA sequences are capable of directing the replication and/or the expression of the DNA sequence. Expression vectors of utility in the present invention are often in the form of "plasmids", circular double stranded DNA loops that, in their vector form, are not bound to the chromosome. However, the invention is intended to include such other forms of expression vectors which serve equivalent functions and which become known in the art subsequently hereto. The expression vectors of the present invention may also be used to stably integrate the DNA sequence into the chromosome of an appropriate host cell (e.g., Chinese Hamster Ovary cells).

Expression vectors useful in the present invention typically contain an origin of replication, a promoter located 5' to (i.e., upstream of) the DNA sequence, followed by the DNA sequence coding for polypeptide, transcription termination sequences, and the remaining vector. The expression vectors may also include other DNA sequences known in the art, for example, stability leader sequences, which provide for stability of the expression product; secretory leader sequences, which provide for secretion of the expression product; sequences which allow expression of the structural gene to be modulated (e.g., by the presence or absence of nutrients or other inducers in the growth medium); marking sequences, which are capable of providing phenotypic selection in transformed host cells; restriction sites, which provide sites for cleavage by restriction endonucleases; and sequences which allow expression in various types of hosts, including prokaryotes, yeasts, fungi, plants and higher eukaryotes. An expression vector as contemplated by the present invention is capable of directing the replication and preferably the expression of the nucleic acids encoding the polypeptides forming the heterotetramer. Suitable origins of replication include, for example, the Col E1 , the SV40 viral and the M13 origins of replication. Suitable promoters include, for example, the cytomegalovirus promoter, the jacZ promoter, the gaj10 promoter and the Autographa californica multiple nuclear polyhedrosis virus (AcMNPV) polyhedral promoter.

Suitable termination sequences include, for example, the bovine growth hormone, SV40, lacZ and AcMNPV polyhedral polyadenylation signals. Examples of selectable markers include neomycin, ampicillin, and hygromycin resistance and the like. All of these materials are known in the art and are commercially available. Suitable commercially available expression vectors into which the DNA sequences of the present invention may be inserted include the baculovirus expression vector pBlueBac, the prokaryotic expression vector pcDNAII and the yeast expression vector pYes2, all of which may be obtained from Invitrogen Corp., San Diego, CA. Other expression vectors include, for instance the pET series of vectors available from Novagen Inc., Madison Wl. The inventors prefer vectors comprising DNA sequences encoding SEQ. ID. NOS.: 1 , 2, 4, or 5. See, for example, the vectors appearing in the examples hereinbelow, which are preferred.

Persons skilled in the art are aware of various methods to introduce expression vectors into host cells. For example, one can transfect host cells by the calcium phosphate precipitation method. Other methods exist, including electroporation, liposomal fusion, nuclear injection, and viral or phage infection. After introducing a vector into a host cell, persons skilled in the art can culture the host cell under conditions permitting expression of large amounts of the desired polypeptide.

Suitable host cells include both prokaryotic and eukaryotic cells. Suitable prokaryotic host cells include, for example, E. opji strains HB101 , DH5a, XL1 Blue, Y1090 and JM101. Suitable eukaryotic host cells include, for example, Spodoptera frugiperda insect cells, COS-7 cells (which are preferred), human skin fibroblasts, and S. cerevisiae cells.

One can identify host cells containing an expression vector having a DNA sequence coding for the desired polypeptide by one or more of five general approaches: (a) DNA-DNA hybridization; (b) the presence or absence of marker gene functions; (c) assessing the level of transcription as measured by the production of mRNA transcripts encoding the desired polypeptide in the host cell; (d) detection of the gene product immunologically; and (e) PCR.

In the first approach, one detects the presence of the desired DNA sequence by DNA-DNA or RNA-DNA hybridization using probes complementary to the DNA sequence. In the second approach, one identifies the recombinant expression vector based upon the presence or absence of such marker gene functions thymidine kinase activity, resistance to antibiotics, and the like. Persons skilled in the art can place a marker gene in the same plasmid as the desired DNA sequence, under regulation of the same or a different promoter. With the same promoter, marker gene expression signals expression of the desired sequence; with a different promoter, it signals incorporation of the vector into the host cell. In the third approach, one assesses production of mRNA transcripts by hybridization assays. For example, one can isolate polyadenylated RNA by Northern blotting or a nuclease protection assay using a probe complementary to the RNA sequence. Alternatively, one can extract the total RNA of the host cell and assay for hybridization to such probes.

In the fourth approach, one can assess expression of the desired nucleic acids immunologically; for example, by immunoblotting with antibody to desired polypeptides (Western blotting).

In the fifth approach, one uses oligonucleotide primers homologous to sequences in the expression system in a PCR to produce a DNA fragment of predicted length. One can determine the DNA sequences of expression vectors, plasmids or DNA molecules of the present invention by various methods known in the art; see, for example, Sanger et aj.(1977), Proc. Natl- Acad. Sc . USA 74: 5463-7; Maxam and Gilbert (1977), Proc, Natl. Acad. Sci. USA 74: 560-4. Those skilled in the art understand that not all expression vectors and DNA regulatory sequences will function equally well to express the DNA sequences of the present invention. Neither will all host cells function equally well with the same expression system. However, one of ordinary skill in the art may make a selection among expression vectors, DNA regulatory sequences, and host cells using the guidance provided herein without undue experimentation and without departing from the scope of the present invention.

One can obtain the desired polypeptides from host cells expressing the DNA sequences by in vitro translation of the mRNA encoded by the DNA sequences. Using known techniques, one may (1) synthesize DNA encoding a polypeptide of SEQ. ID. NOS.: 1, 2, 4, or 5 by PCR, (2) insert the DNA into a suitable expression vector, (3) transform a suitable host cell with the vector, and (4) culture the recombinant host cell to produce the polypeptide.

One can use various protein purification techniques to isolate and purify polypeptides produced in this manner. Such techniques include ion exchange chromatography, gel filtration chromatography, and immunoaffinity chromatography. Linkage to Effector and Specific Ligand Binding Domains. One can make polypeptides linked to the effector and specific ligand binding domains by a number of different procedures.

The effector domain and binding domain may be polypeptides themselves. PE38 and PE40, for example, are polypeptide effector domains. More commonly, the binding domain will be a polypeptide; for example, an F_v fragment from an antibody such as BR96. When the effector or binding domain is itself a polypeptide, one may directly link it by peptide bond to the heterotetramer by simply continuing the foregoing synthetic procedure with the am o acids of the effector or binding domain Alternatively, the aforementioned vector may comprise a chimera of a heterotetramer polypeptide and the effector or binding domain (see "Materials and Methods," below).

We also contemplate using the heterotetramer with nonpeptidic effector and binding domains. One could link such domains to the heterotetramer polypeptides by a number of procedures known in the art. See for example, the linkage procedures described in European Patent Application 554, 708, published 1 1 August 1993; PCT Application WO 91/00295, published 10 January 1991 ; European Patent Application 328,147, published 16 August 1989; and European Patent Application 457, 250, published 21 November 1991. We hereby incorporate these publications by reference.

Figure 6 shows (A) the DNA sequence (SEQ. ID. NO.. 3) and below it the expressed ammo acid sequence (SEQ. ID. NO.: 4) of the synthetic Lac28E gene and (B) the Lac28K peptide (SEQ. ID. NO.: 5) with lower case letters above these sequences representing heptad positions on a helical wheel representation. (A) The gene was assembled from 4 oligonucleotides The first oligonucleotide for the top and the bottom strands are italicized. The sequence will directly ligate into an Sphl site and can be recut by BamHI (GGTACC). Ammo acids starting with capital letters are the Lac28E sequence, those starting with lower case are either a flexible linker to allow the domains to reach each other or result from the cloning site (B) As can be seen, the two internal heptads are identical.

Figure 7 shows the GPC elution profile of BR96-sFv-Lac28E-PE38 in stabilizing and destabilizing conditions for LacE association. Samples were run on a TSK 3000 SW column (7x300 mM, Waters) at 1.4 mL/minute on a Waters Advanced Protein Purification System equilibrated with either 40 mM NaP0₄, 300 mM NaCl (stabilizing) or 100 mM KPO4, 100 mM NaCl, 0.5 mM EDTA, 0.5 mM EGTA, 0.001% sodium azide (destabilizing) as described hereinafter.

Figure 8 shows an ELISA assay for the effect of Lac28K peptide on the binding of BR96-sFv-PE40 and BR96-sFv-Lac28E-PE38 to LeY-HSA. Background binding to BSA has been subtracted. (A) Pre-incubation was with 2 μM immunotoxin and 50 μM Lac28K peptide for 2 hours at 4 °C. Binding of immunotoxin was done at 4 °C for 16 hours at the indicated concentrations of protein. (B) Pre-incubation was with 2 μM immunotoxin and 10 μM Lac28K peptide for 2 hours at 4°C. Binding of immunotoxins was done at 23 °C for 4 hours. Protein concentrations and assay conditions are as described hereinafter. The decrease in binding for BR96-sFv-Lac28E-PE38 + Lac28K at higher protein concentrations is due to an increase in the background. This effect is also seen with the other immunotoxin proteins at sufficiently high concentrations.

Figure 9(A) shows the predicted structure of BR96-sFv-Lac28E-PE38 (black) + Lac28K (gray) peptide molecule. The ^* and @ at the ends of the Lac28K peptide represent the N- and C- terminal blocking groups. These groups could be replaced with either chemically added molecules or biosynthetically added domains as indicated in (B). Figure 9(B) shows a proposed representation of a molecule with 4 antibody heads and 2 PE tails. Such a molecule would be formed by a complex of BR96-sFv-Lac28E-PE38 and BR96-sFv- Lac28K with a stop codon following the Lac28K. Figure 10 indicates two different orientations available for combining two proteins. (A) Both the Lac28E (black) and Lac28K (gray) fusions are expressed with the Lac sequences at the C-terminus of the proteins. (B) For the Lac28E (black) fusion the Lac sequence is at the C-terminus, while for the Lac28K (gray) fusion the lac sequence is at the N-terminus. Figure 11 indicates two different ways for bringing four proteins together. (A)

From N- to C- terminus: Protein 1 is fused with the Lac28E (black) sequence, which is in turn fused with Protein 3, Protein 2 is fused with the Lac28K (gray) sequence, which is in turn fused with protein 4. (B) Protein 1 is fused with the Lac28E (black) sequence, which is in turn fused with protein 4. Protein 3 is fused with the Lac28K (gray) sequence, which is in turn fused to protein 2. Clearly a number of different arrangements are possible.

Figure 12 shows a sequence for a fully heterotetrameric scaffold. Panel A shows four possible sequences (SEQ. ID. NOS.: 6 to 9) designed to limit homoassociation and encourage an association that contains one of each of the sequences. Panel B shows the helical wheel diagram of such a tetramer. As can be seen, the opposite charges align with each other in the heterotetramer but there will be numerous like charge interactions in any other type of tetramer formations (6 and 6, 6 and 7, 6 and 7 and 8, etc.). The wheels are numbered corresponding to the sequence number and the N and C refer to the N or C terminal orientation of the sequence.

Peptide synthesis and purification. Polypeptides were synthesized on an Applied Biosystems Model 431 A automated peptide synthesizer using the Boc/benzyl strategy. Peptidyl-resm was deprotected and cleaved by treatment with hydrofluoric acid containing 5% anisole and 5% thioanisole as scavengers. Polypeptides were purified to homogeneity by reverse phase HPLC. Chao et aj. (1993), J. Org. Chem. 58: 2640-4. Their identity was established by am o acid analysis. Liu & Boykins (1989), Anal. Biochem. 182: 383-7. Their identity was also confirmed by electrospray or fast atom bombardment mass spectrometry analysis. The concentrations of the polypeptides were determined by one or both of the following methods: (1 ) quantitative ammo acid analysis (Liu, & Boykins, 1989); or (2) quantitative ninhydrm analysis. Rosen (1957), Arch. Biochem. Biophys. 67: 10-15.

Circular Dichroism. Data were collected using an Aviv 62DS circular dichroism spectropolaπmeter equipped with a thermoelectric device for temperature control. Spectra were collected using a 0.5 nm step-size, an averaging time of 2 seconds, and a bandwidth of 1.5 nm in 10 mM MOPS, pH 7.5. All other measurements were also made in 10 mM

MOPS, pH 7.5, at 25 °C. Data points in the thermal melts represent a time average of 3 minutes.

Ultracentrifugation. Experiments were performed at 25 °C in a Beckman Model XLA ultracentrifuge using an An 60 Ti rotor. Data for the polypeptides were collected using six-channel Epon, charcoal-filled centerpieces with a 12 mm path length containing 110 μL samples and 125 μL buffer references The peptide loading concentrations were 200 μM in 10 mM MOPS, pH 7.5, 0.1 M NaCl. The density of the solvent was measured gravimetπcally (p=1.0078 gm/mL) The samples were centrifuged at 20,000, 30,000,

40,000 and 50,000 rpm and the protein distribution was monitored at a wavelength of 242 nm. Ten successive radial scans were averaged using a 0.001 cm step-size and equilibrium was assumed if no change in distribution was observed at intervals of 2 hours. The data analysis software ran under Igor (Wavemetrics, Lake Oswego, OR) and incorporated the algorithm of Johnson et aj. (1981), Biophys. J. 36: 575-88. Partial specific volumes (Lac21E: 0.724; Lac21K: 0.776) were calculated from the weight average of the partial specific volumes of the individual amino acids. Cohn & Edsall (1943), Proteins, Amino Acids and Peptides as Ions and Dipolar Ions. New York: Reinhold Publishing Corp.

Vector Construction. Unless otherwise stated, all molecular biology techniques were from Sambrook, Fritsch, and Maniatis (1990). The BR96-sFv-PE38 expression vector was kindly provided by Clay Siegall (BMS, Seattle, WA.) and is a derivative of BR96-sFv- PE40 described in Friedman et aj. (1993), Cancer Research 53: 334-9. The latter construct has the PE38 molecule described in Siegall et aj. (1989), J. Biol- Chem. 264: 14256-61. PE38 replaces PE40, which is described in Chaudhary et aj. (1987) Proc. Natl. Acad. ScL USA 84: 4538-52. and Hwang et aj. (1987) Ceil 48: 129-36. Oligonucleotides were purchased from Genosys, restriction enzymes from NEB, and Sequenase from USB. To construct the BR96-sFv-Lac28E-PE38 expression vector, BR96-sFv-PE38 was digested with Sr hl and treated with Calf Intestine Alkaline Phosphatase (NEB). Oligonucleotides were 5' phosphorylated by T4 Kinase (NEB) and annealed by heating to 95 °C and slow cooling to 23 °C (2 hours). The oligonucleotides were mixed with the vector at molar ratios of 1 :1 to 100:1 and ligated at 16 °C for 16 hours. The ligation mixture was transformed into E. cojj MM294 competent cells and colonies were picked for analysis. Miniprep DNA (Promega Wizard Prep) was analyzed by restriction analysis and sequencing. The final vector containing the BR96-sFv-Lac28E-PE38 coding sequence under control of the T7 promoter was named pBL25LEC.

Protein Expression and Purification. Expression of protein from pBL25LEC was expressed into inclusion bodies essentially as described (Friedman, et a ., 1993). Briefly, E. coli BLR (IDE3) cells (Novagen) were transformed with pBL25LEC and grown on M9 minimal media plates supplemented with Casamino acids (Difco) and 100 μg/mL ampicillin (M9CAGA). Colonies were grown ovemight at 30 °C in M9CAGA. The cells from the overnight culture were washed with fresh M9CAGA and added to 500 mL of M9CAGA in a 2 liter baffle flask (Wheaton) at a 1/20 inoculum. Cells were grown at 30 °C until an

OD600 of 0.7 to 1.0 when IPTG (Fluka) was added to 0.4 mM and the cultures were shifted to 37 °C. After 4 to 6 hours, the culture was pelleted, the cell pellet was washed with PBS containing inhibitor cocktail (0.2 mM PMSF, 0.2 mM Pefabloc SC, 10 μg/mL aprotinin, and 10 μg/mL pepstatin), and the washed pellet was frozen at -20 °C until further processing. All protein steps were conducted at 4 °C unless otherwise noted. Bacterial cells were lysed by sonication (Heat Systems Sonicator XL) 4 x 30 seconds at 90% power in cell lysis buffer (50 mM Tris pH 8.0, 5 mM EDTA with 1% Triton X-100 (Fluka) and 10 μg/mL lysozyme (Worthington) and inhibitor cocktail added. Inclusion bodies were pelleted at 30,000 x g for 25 minutes and resuspended and sonicated 2 x 30 seconds in cell lysis buffer with Triton X-100 and inhibitor cocktail added. Inclusion bodies were pelleted as before and resuspended in cell lysis buffer with 0.5 M deionized (Biorad 501-X8(D)) urea (Fluka, Microselect) by dounce homogenization (Koontes). Inclusion bodies were pelleted as before and washed a final time in cell lysis buffer. The final inclusion body pellet was resolubilized in 20 mM Tris pH 8.5, 50 mM DTT, 9.5 M deionized urea by dounce homogenization, sonication and incubation at 4 °C overnight.

The solubilized pellet was centrifuged at 30,000 x g for 1 hour and the supernatant containing the BR96-sFv-Lac28E-PE38 protein was purified by anion exchange chromatography (Fast Flow-Q, Pharmacia) and eluted with a 0-1000 mM NaCl gradient in 20 mM Tris pH 8.2. Fractions containing the BR96-sFv-Lac28E-PE38 protein as determined by SDS-PAGE (S&S, Novex) minigels and coomassie staining were pooled and dialyzed vs. two changes of 30 mM Na acetate pH 5.3 with 3,500 MW cutoff tubing (Spectrum). Protein was applied to a cation exchange resin (Fractogel EMD -SO3 650(M), E.

Merck: or ToyoPearl SP 650(M), TosoHaas) and eluted with a 0-1000 mM gradient in 30 mM Na acetate pH 5.3. Fractions containing the BR96-sFv-Lac28E-PE38 protein were pooled and protein determined by Coomassie Plus Dye binding assay (Pierce), and UV absorbance. The BR96-sFv-Lac28E-PE38 protein was refolded by dilution to 30 mg/mL in 2 M deionized urea, 50 mM Tris pH 8.2, 1 mM reduced glutathione and 0.1 mM oxidized glutathione (Fluka) which had been purged with N₂ and chilled to 4 °C. The protein was allowed to refold for 2 to 4 days at 4 °C.

The refolded protein was purified by anion exchange chromatography (Fractogel EMD TMAE-650(S)) with elution by a 0-1000 mM gradient of NaCl in 20 mM Tris pH 8.2 or 7.5 which served to remove the urea, concentrate the protein, and separate intermolecularly disulfide bonded protein. Fractions containing monomeric protein were determined by non-reducing SDS-PAGE, then pooled and concentrated by ultrafiltration (Centriprep 30, Amicon). Further purification of the refolded BR96-sFv-Lac28E-PE38 was by Gel Permeation Chromatography (GPC) on a TSK-3000 SW (Waters) using a Waters Advanced Protein Purification System in 100 mM Na phosphate pH 7.0, 100 mM NaCl, 0.5 mM EDTA, 0.5 mM EGTA, 0.0025% NaN3. Fractions containing monomer were analyzed and concentrated as above.

ELISA Assay. Function was determined by an ELISA measuring binding to Lewis Y (LeY) synthetic tetrasaccharide. Briefly, 96 well plates (Costar) were coated ovemight with 100 mL of 500 ng/ of LeY-HSA in PBS at 4°C. Plates were then washed twice with 0.5% BSA/PBS incubated with the same solution for 1 hr to block nonspecific sites. Separately, BR96-sFv-Lac28E-PE38 or BR96-sFv-PE40 protein was pre-incubated for 2 hours at 4 °C with or without Lac28K peptide at a protein concentration of 140 mg/mL (2 mM) in 30 mM Na phosphate pH 7.0, 150 mM NaCl, 0.2 mM EDTA, 0.2 mM EGTA. Samples were serially diluted in 0.1%BSA/PBS and 100? was incubated in the wells coated with Lewis-Y either for 16 hours at 4 °C or 4 hours at 23 °C. The wells were washed with 0.1% BSA, 0.01% Tween 20 (PierceVPBS 4 times and the plate was incubated with a 1/3000 dilution of Goat anti-Pseudomonas Exotoxin (LIST Biologicals) in 0.1% BSA/PBS for 1 hour at 23 °C. The wells were washed as above and then incubated with Rabbit anti-Goat-HRP conjugate (Bio-Rad) at 1/1000 in 0.1% BSA PBS for 1 hour at 23 °C. This incubation was followed by 2 washes with 0.1% BSA/PBS and 2 washes with PBS. Color was developed with TMB Peroxidase EIA Substrate Kit (Bio-Rad) at 23°C for 30-60 minutes. Plates were read in a Molecular Devices plate reader.

The sequences for the polypeptides used here are derived from our work showing that the C-terminal 21 residue sequence from Lac repressor protein could form homomeric four-chain coiled-coils (or four-helix bundles). Fairman et aj. (1995), Protein

Science 4:1457-69. We have replaced the amino acids at both heptad positions b and c (Figures 1(A) and (B)) with either glutamic acid (Lac21E) or lysine (Lac21 K). The goal of these replacements would be to disfavor homotetramer formation through electrostatic repulsionand to simultaneously favor heterotetramer formation through favorable interactions between oppositely charged glutamic acid and lysine residues upon mixing equimolar quantities of Lac21E and Lac21K. This strategy has been experimentally validated Figure 2 shows circular dichroism (CD) spectra for either Lac21 E and Lac21K alone or a 1 :1 mix at a total peptide concentration of 100 μM. Lac21 E and Lac21K both have CD spectra that are highly characteristic of unfolded polypeptides showing a weak minimum at 222 nm and a second, stronger minimum below 200 nm. In contrast, a 1 :1 mix of the two polypeptides results in a spectrum which is consistent with a highly helical structure containing minima of approximately the same intensity at 222 nm and at 208 nm.

A different CD study quantified the stability of the Lac21 E:Lac21K heterotetramer. The relative stabilities of the complexes formed can be obtained from thermal unfolding also as monitored by CD. The Tm of a 1 :1 mix of 50 μM each of Lac21E and Lac21 K is about 75 °C (Figure 3c). This T depends upon peptide concentration (data not shown) as expected for an oligomerizmg system. In contrast, the T 's of Lac21 E and

Lac21 K are well below 0 °C. These data impelled us to conduct a thermodynamic analysis of heteromer formation using thermal unfolding data fitted with the Gibbs-Helmholtz function, using a method similar to that of Thompson et aj. (1993), Biochemistry 32: 5491- 6 This analysis places the stability at about -22 Kcal/mol of tetramer at 25 °C This corresponds to a dissociation constant of 4x10^"17 M³ assuming a monomer-tetramer equilibrium process. While the Tm's for Lac21 E and Lac21K can not be calculated, a reasonable upper limit of -20 °C corresponds to Kd's which would be at least 60,000 times less stable. .

The driving force for the increased stability of the heteromer relative to the homomers is electrostatic in origin as judged by the effects of 1 M NaCl on the stabilities of each of the complexes. A dramatic decrease in the T_m of unfolding for the

Lac21 E/Lac21 K complex (Δ50 °C) is observed in 1 M NaCl (Figure 3c). In contrast, the stability of the Lac21K complex is dramatically increased (Figure 3a). Lac21E is also stabilized by 1 M NaCl, albeit the effect is much smaller. The difference in salt effects on thermal unfolding of Lac21E and Lac21K arises from differences in their salt dependence of stability. Much higher concentrations of NaCl are required to achieve equivalent increases of stability for Lac21E as compared to Lac21K While this phenomenon is not fully understood, it is likely a consequence of the inherent differences in screening charges on glutamic acid vs. lysine residues (i e., preferred rotamer distributions and sidechain length). As a further test of the electrostatic contribution to stability, the polypeptides were analyzed for their pH dependence of stability (Figure 4). At 50 μM peptide, both Lac21E and Lac21 K appear fully unfolded at neutral pH. However, Lac21E is more stable at acid pH, presumably caused by protonation of the glutamic acids at positions b and c. Conversely, Lac21 K shows added stability at alkaline pH as the lysine residues become uncharged, albeit the effect is much smaller. The differences in stabilities at these extreme pH's for Lac21E and Lac21K may be a combination of the intrinsic helix propensities for glutamic acid and lysine and the added flexibility of the lysine sidechain.

The polypeptides were analyzed by sedimentation equilibrium to determine their oligomerization states, either as homomeric species or upon formation of a 1:1 complex. A loading concentration of 200 μM was used and 100 mM NaCl was added to anticipate problems with non-ideality owing to high charge density. Single species analysis is presented in Table 1 and global fits to the 30,000, 40,000, and 50,000 rpm data using single species analysis are presented in Figure 5. The molecular weights of Lac21 E, Lac21K, and Lac21 E/Lac21 K, based on the global fits, are 1 ,700+86%, 3,100+55%, and 9,500±7% (The expected monomer molecular weights for Lac21 E and Lac21K are 2,434 and 2,428 daltons). The experimental molecular weights of the Lac21 E and Lac21 K peptides are within experimental error of the calculated values lending credibility to the determination of the Lac21 E:Lac21 K complex: 9,500+7%, a value agreeing well with the sum of two Lac21 E and two Lac21 K peptides.

These amino acid sequences were incorporated into a fusion protein system to assess the utility of the heterotetramerization scaffold. For the gene construct the motif was lengthened by 1 heptad to 28 amino acids (Figure 6) to increase stability. The Lac28E sequence was incorporated between the single chain variable domain and the Pseuodomas exotoxin portion of the BR96-sFv-PE38 molecule with suitable spacers on either end to allow flexibility of the domains.

After expression, purification and refolding, the oligomerization state of the BR96-sFv-Lac28E-PE38 molecule was assessed by gel permeation HPLC (Figure 7). Under high salt conditions BR96-sFv-Lac28E-PE38 eluted as a single peak at -300,000 daltons while at lower salt concentrations the protein eluted as two peaks at about 300,000 and about 90,000 daltons. This result suggests that, as for the peptides, high salt will shield the repulsive charges and allow homotetramerization while lower salt concentrations shield the charges less well and reduce homotetramerization.

One utility of the heterotetramerization motif is demonstrated in Figure 8. Mixing the BR96-sFv-Lac28E-PE38 with the Lac28K peptide (Figure 6) significantly increases the ability of the molecule to bind to the BR96 antigen Lewis Y. The experiment was done at two temperatures to determine the relative stability of the complexes. At both temperatures the addition of Lac28K to the parent protein BR96-sFv-PE38 has no effect on its ability to bind to Lewis Y, while addition of Lac28K to BR96-sFv-Lac28E-PE38 significantly increases the ability to bind to Lewis Y. The relative difference in binding is about 30 fold for either temperature condition suggesting that the BR96-sFv-Lac28E-PE38 and Lac28K form a stable complex.

There are a number of conceivable constructions which can be made using the heterotetramerization motif as illustrated in Figures 9-11. Figure 12 is a description of a fully heterotetrameric scaffold which has been designed based on the principles described herein. By carefully aligning the charges so that no manner of association is completely stable by virtue of similar charges being forced near each other, except a full heterotetramer it will be possible to bring together 4 independent molecules.

Inspection of a computer-built model of antiparallel four-chain coiled coils suggests that the sidechains of glutamic acid and lysine in the b and c positions come close enough to form salt bridges. Our data support this model. . A large difference in the stability of homotetramers vs. heterotetramers is seen in our tetrameric coiled coils. Both pH dependence and salt screening experiments provide conclusive evidence that heterotetramer formation is driven by favorable electrostatic interactions. Our work stands separate from past work where electrostatics have been invoked. Mixed results are seen in studies of dimeric coiled coilssuch as fos and jun where electrostatics appear to disfavor homomeric interactions (Glover and Harrison (1995), Nature 373: 257-61 ) rather than to favor heteromeric interactions. This mechanism has been suggested for the heterospecificity of the designed 'Velcro' polypeptides which can form heterodimers. O'Shea fit a!. (1993), Curr. Biol. 3: 658-67. as well as heterotetramers. Lumb & Kim (1995) Biochemistry 34:8642-8. Other laboratories have shown that charged residues at the same e and g positions can also act to stabilize heteromeric interactions. At present, it is not clear whether repulsive or attractive electrostatic interactions dominate in the assembly of heterospecific coiled coils. In the case of Lac tetrameric coiled coils, heterospecificity can be driven by residues at b and c positions.

Table 1. Single species analysis of sedimentation equilibrium data rotor speed (rpm) peptide monomer 20,000 30,000 40,000 50,000

Lac21E 2,428 <1 ,000 <1,000 2,918±74% 1 ,642+41%

Lac21 K 2,434 <1,000 803±720% 3,245+46% 2,404±32%

Lac21 E+ 2,431 14,605138% 8,741 ± 12% 10,248± 7% 9,470± 3% Lac21 K

Claims

What We Claimed Is:

1. A composition of matter of the formula shown in Figure 1(A), wherein:

(a) I, II, III and IV are each alpha-helical polypeptides of about 21 to 35 residues;

(b) moieties in positions a and d of helices I and IV are each independently amino acid residues;

(c) moieties in positions b and c of helices I and III are each independently positively charged amino acid residues;

(d) moieties in positions b and c of helices II and IV are each independently negatively charged amino acid residues;

(e) moieties in positions e and g of helices I, II, III, and IV are each independently amino acid residues; and

(f ) moieties in positions f of helices I, II, III, and IV are each independently amino acid residues; and

2. The composition of matter of Claim 1 wherein the positively charged amino acid residues of (c) are selected from L-forms of arginyl, histidyl, and lysyl.

3. The composition of matter of Claim 1 wherein the positively charged amino acid residues of (c) are lysyl.

4. The composition of matter of Claim 1 wherein the negatively charged amino acid residues of (d) are selected from L-forms of aspartyl and glutamyl.

5. The composition of matter of Claim 1 wherein the negatively charged amino acid residues of (d) are glutamyl.

6. The composition of matter of Claim 1 wherein the amino acid residues of (b) are selected from L-forms of alanyl, isoleucyl, leucyl, methionyl, phenylalanyl, tryptophyl, and valyl.

The composition of matter of Claim 1 wherein the ammo acid residues of (b) are selected from methionyl, leucyl, and valyl.

The composition of matter of Claim 1 wherein the am o acid residues of (e) are each independently neutral ammo acid residues or nonpolar am o acid residues

The composition of matter of Claim 1 wherein I and III have the ammo acid sequence of SEQ. ID NO. 1

The composition of matter of Claim 1 wherein II and IV have the ammo acid sequence of SEQ. ID. NO. 2

The composition of matter of Claim 1 wherein I, II, III, and IV each have 21 ammo acid residues.

The composition of matter of Claim 1 wherein the ligand binding domain is an antibody variable region.

The composition of matter of Claim 1 wherein the specific ligand binding domain is BR96 sFv.

The composition of matter of Claim 1 wherein the effector domain is a substance that initiates cell death.

The composition of matter of Claim 1 wherein the effector domain is Pseudomonas exotoxin.

A polypeptide having the am o acid sequence of SEQ. ID NOS 1 , 2, 4, or 5

An isolated nucleic acid encoding a polypeptide of SEQ. ID NOS 1 , 2, 4, or 5

18. An expression vector having a nucleic acid sequence encoding a polypeptide of SEQ. ID. NOS.: 1, 2, 4, or 5.

19. A host cell comprising the expression vector of Claim 18.

20. The host cell of Claim 19 wherein the cell is a BL-21 cell.