WO2000073342A1

WO2000073342A1 - Structural models for cytoplasmic domains of transmembrane receptors

Info

Publication number: WO2000073342A1
Application number: PCT/US2000/015153
Authority: WO
Inventors: Mark H. Ginsberg; Martin Pfaff; Shouchun Liu
Original assignee: The Scripps Research Institute
Priority date: 1999-06-01
Filing date: 2000-06-01
Publication date: 2000-12-07
Also published as: EP1180122A1; EP1180122A4; JP2003504308A; CA2374133A1; AU5177300A; AU765990B2

Abstract

Polypeptides containing a series of heptad-repeats that mimic a transmembrane domain and a selected cytoplasmic domain attached to the series of heptad repeats are provided which can be used in construction of structural models for evaluating the structure and activity of occupied and clustered transmembrane proteins and identifying therapeutic compounds.

Description

STRUCTURAL MODELS FOR CYTOPLASMIC DOMAINS OF TRANSMEMBRANE RECEPTORS

Introduction This invention was made in the course of research sponsored by the National Institutes of Health. The U.S. Government may have certain rights in this invention.

Background of the Invention In eukaryotic cells, many proteins extend through the cell membrane and therefore contain a cytoplasmic domain, a transmembrane domain and an extracellular domain. Many of these proteins are involved in signal transduction, cell adhesion and cell-cell interactions. Among the proteins that fall into this category are the integrins . Integrins are involved in a number of pathological and physiological processes, including thrombosis, inflammation, and cancer. Other physiological and pathological conditions involving changes in cell adhesiveness are also mediated through integrins.

Many transmembrane proteins are oligomeric, being noncovalent associations of two or more different types of polypeptide subunits. In particular, integrins are heterodimers of two different protein subunits, designated α and β. The subunits vary in size between 120 and 180 kDa and are each noncovalently associated with a β subunit. The extracellular domain of the integrin molecule forms a ligand binding site; both the α and β subunits are involved in forming the ligand binding site. A number of different ligands for integrins are known, including collagens, laminin, fibronectin, vitronectin, complement components, thrombospondin, and integral membrane proteins of the immunoglobulin superfamily such as ICAM-1, ICAM-2, and VCAM-1. The integrins recognize various short peptide sequences in their ligands. Examples of these are Arg-Gly- Asp (RGD) , Lys-Gln-Ala-Gly-Asp-Val (KQAGDV; SEQ ID NO: 1), Asp-Gly-Glu-Ala (DGEA; SEQ ID NO: 2), and Glu- Ile-Leu-Asp- Val (EI DV; SEQ ID NO : 3) . Variations in integrin function are often caused by changes in the ligand binding affinity of the extracellular domain of the integrins (J.S. Bennett & G. Vilaire J. Clin . Invest . 64:1393-1401 (1979); Altieri et al. J^". Cell Biol . 107:1893-1900 (1988); Faull et al . J. Cell Biol . 121:155-162 (1993); Lollo et al . J. Biol . Chem . 268:21693-21700 (1993)).

Integrin _IIbβ₃ (platelet GPIIb-IIIa) , a heterodimer of two type I transmembrane protein subunits, manifests highly regulated changes in ligand binding affinity. Affinity state-specific antibodies, e.g., PAC1 (Shattil et al . J. Biol . Chem . 260:1107-1114 (1985)), are useful for analysis of recombinant _IIbβ₃ in heterologous cells (O'Toole et al . Cell Regulation 1:883-893 (1990)). Platelet agonists increase the affinity of α_IIbβ₃ (activation) probably by causing changes in the conformation of the extracellular domain (O'Toole et al . Cell Regula tion 1:883-893 (1990); Sims et al . J. Biol . Chem . 266:7345-7352 (1991)). Cytoplasmic signaling pathways involving heterotrimeric GTP binding proteins, phospholipid metabolism, and serine- threonine kinases initiate these conformational changes in the extracellular domain; these changes may also involve calcium fluxes, tyrosine kinases, and low molecular weight GTP binding proteins (Sims et al . J. Biol . Chem . 266:7345- 7352 (1991); Shattil et al . J. Biol . Chem . 267: 18424-18431 (1992); S.J. Shattil & J.S. Brugge Curr. Opin . Cell Biol . 3:869-879 (1991); Ginsberg et al . Cold Spring Harbor Symposium of Quanti ta tive Biology: The Cell Surface 57:221-231 (1992); Ginsberg et al . Curr. Opin . Cell Biol . 4:766-771 (1992); Nemoto et al . J. Biol . Chem . 267:20916- 20920 (1992)) . How cytoplasmic signals result in changes in the conformation and ligand binding affinity of the extracellular domain ("inside-out signal transduction" ) of the integrin remains unknown. Studies with chimeras containing the cytoplasmic domains of various α and β subunits joined to the transmembrane and extracellular domain of _IIbβ₃ indicate that integrin cytoplasmic domains transduce cell type-specific signals that modulate ligand binding affinity. These signals require active cellular processes in both α and β cytoplasmic tails of the integrin, suggesting that they reflect physiologically relevant signals. In addition, deletion of a highly conserved motif, Gly-Phe-Phe-Lys-Arg (GFFKR; SEQ ID NO: 4), at the amino-terminus of the α subunit cytoplasmic domain, also resulted in high affinity binding of ligands to integrin _ZIβ₃. In contrast to the chimeras, high affinity ligand binding to GFFKR deletion mutants was independent of cellular metabolism, cell type, and the bulk of the β subunit cytoplasmic domain. Thus, integrin cytoplasmic tails are targets for the modulation of integrin affinity. However, technical difficulties have greatly limited the application of high resolution techniques for determination of the structures of these proteins . In fact, molecular structures are available for only two intact transmembrane proteins, a bacterial photoreaction center (Deisenhofer et al . Nature 318:618-624 (1985)), and a porin (Weiss et al . FEBS Lett . 267:268-272 (1990)). Structures of receptor extracellular domains have been determined using soluble truncated extracellular domains as models (DeVos et al . Science 255:306-312 (1992); Milburn et al. Science 254:1342-1347 (1991)). These structures have contributed to the understanding of the basis of ligand recognition, but have provided less insight into the mechanism of signal transduction. Many membrane proteins that transduce signals are members of the Type I transmembrane protein family, the defining feature of which is a single membrane spanning region. These include the T cell receptor (A. Weiss Cell 73:209-212 (1993)); growth factor receptors (L. Patthy Cell 61:13-14 (1990)), and cytokine receptors (Miyajima et al . TIBS 17:378-382 (1992)) . In general, the cytoplasmic domain of these proteins is critical for signaling. Thus, to understand signal transduction through such receptors, it is essential to understand the structure and function of the cytoplasmic domain. This is especially difficult for multisubunit Type I proteins .

A strategy for the chemical synthesis of structural models of the cytoplasmic domain of multisubunit transmembrane receptors has been previously proposed (Muir et al . Biochemistry 33:7701 (1994)). The cytoplasmic domains of integrin _IIbβ₃ were covalently linked via a helical coiled-coil made up of a series of identical heptad repeats. Coiled-coil tertiary structure was utilized to mimic the presumed helical membrane spanning domain and as a topological constraint, fixing the two integrin tails in a parallel orientation with the appropriate vertical stagger (Muir et al . Biochemistry 33:7701 (1994)). However, this synthetic approach poses limitations upon the polypeptide length and has a relatively modest yield. Accordingly, there is a need for improved methods of producing structural models of the cytoplasmic domain of multisubunit transmembrane receptors . These models are useful in evaluating agents which control and modulate the activity of integrins and other transmembrane proteins, detecting their activity, and modulating their activity to detect and control physiological conditions. Summary of the Invention

In the present invention, a method is provided for preparation of proteins for use in structural models or mimics of the cytoplasmic face of multimeric transmembrane proteins such as integrins . Proteins of the present invention may be prepared recombinantly or synthetically. However, by using recombinant proteins, limitations of polypeptide length and modest yield encountered in the initial synthetic approaches of the prior art are avoided. Accordingly, it is preferred that at least a portion of the structural model of the present invention be prepared recombinantly. In the model of the present invention, the heterodimeric nature of the β cytoplasmic domain is mimicked by use of covalent heterodimers of these domains. Helical coiled-coil architecture provides the desired parallel topology and vertical stagger of the tails. The model is useful in studying protein interactions with transmembrane proteins such as integrin and screening agents for integrin inhibitory activity and in obtaining structures of integrin cytoplasmic domains. For example, using a model comprising an α4 cytoplasmic tail, it has now been found that paxillin and paxillin related molecules such as leupaxin and Hie- 5 have high affinity interactions with 4 integrin. Accordingly, agents which inhibit the interaction of paxillin and paxillin related molecules with α4 integrin are believed to be useful in inhibiting biological responses associated with α4 integrins. Thus, these agents may be useful in inhibiting normal α4 integrin activity such as that occurring in wound healing which can lead to scarring. These agents can also be used in inhibiting pathological responses of α4 integrin such as in atherosclerosis and immune responses associated with conditions including, but not limited to inflammatory bowel disease, arthritis, multiple sclerosis and asthma. Brief Description of the Drawings

Figure 1 exemplifies amino acid sequences of recombinant model proteins of integrin cytoplasmic domains. Figure 1A shows the N-terminal (SEQ ID NO : 5) and heptad- repeat (SEQ ID NO: 6) structures common to all constructs. In the example shown, these are connected to the Gl-βlA cytoplasmic domain (SEQ ID NO: 7) . Arrows indicate the positions of hydrophobic residues corresponding to positions a and d of the heptad repeats. Positions of the additional Gly insertions in the

G2-, G3- and G4 -constructs are also indicated. Figure IB shows the integrin-specific sequences of the constructs used in experiments described herein including B1A (SEQ ID NO: 8), B1A (U788A) (SEQ ID NO: 9), BIB (SEQ ID NO: 10), BIC (SEQ ID NO: 11), BID (SEQ ID NO: 12) and B7 (SEQ ID NO: 13) . All integrin peptides correspond to the reported human integrin sequences . Detailed Description of the Invention

The present invention relates to the production of mimics of the cytoplasmic face of occupied and clustered transmembrane proteins such as integrins consisting of polypeptides comprising a series of α-helical heptad repeats, preferably 2 to 20, more preferably 3 to 6 , most preferably 4, that mimic a transmembrane domain connected to a cytoplasmic domain of a selected multisubunit transmembrane receptors such as integrins. By "mimic" it is meant that the series of heptad repeats, imitates or replaces the structural features of the transmembrane domain. In one embodiment, an immobilizing epitope such as a His-Tag sequence or glutathione-S-transferase, is linked to the N-terminus for immobilization of the polypeptide in affinity chromatography . In this embodiment, it is preferred that the immobilizing epitope be linked to the polypeptide via a Cys-Gly linker. For convenience, a prokaryotic or chemical cleavage site such as a thrombin cleavage site can also be incorporated into the polypeptide at this linkage site.

For the purposes of the present invention, by "α- helical heptad-repeat " it is meant a sequence consisting of substantially helical amphiphilic amino acids having hydrophobic residues at selected positions in the repeat, preferably positions a and d as depicted in Figure 1. In such an embodiment, each repeat is seven a ino acids with hydrophobic residues at the first and fourth positions. For example, in a preferred embodiment, the heptad repeat comprises the amino acid sequence G-X₁-L-X₂-X₃- -X₄-G, (SEQ ID NO: 14) wherein X_l is a lysine, arginine or ornithine, X₂ and X₄ are glutamic acid or aspartic acid, and X₃ is alanine, serine or threonine. The heptad repeats of the polypeptide are preferably identical. However, in some embodiments, each heptad repeat may differ in amino acid sequence .

In a preferred embodiment, the cytoplasmic tail of a transmembrane receptor such as an integrin is linked to the heptad repeat via a glycine residue at the C-terminus of the heptad repeat. In this embodiment the polypeptide is predicted to form parallel coiled-coil dimers under physiological conditions. However, trimers and tetramers can also be designed based upon current methods for coiled coil protein design. These coiled-coil structures are likely to better mimic the proximity of transmembrane helices in the natural system and also ensure that a defined topology is maintained between the and β cytoplasmic tails. In other words, the coiled-coil of the α-helical heptad repeat can act as a structural template onto which the cytoplasmic domain of the integrin or other transmembrane protein is attached. This ensures that the two cytoplasmic tails are staggered with respect to one another in a manner that approximates the intact protein. A cystine bridge ensures a parallel orientation and a correct stagger of the coiled-coil sequences within this dimer configuration. Examples of cytoplasmic tails of integrins which can be used include, but are not limited to which, integrin β subunits such as βlA (SEQ ID NO: 8), βlA(Y788A) (SEQ ID NO: 9), βlB (SEQ ID NO: 10), βlC (SEQ ID NO: 11), βlB (SEQ ID NO: 12), β7 (SEQ ID NO: 13), and β3 and integrin subunits such as αllb, α4 , α3A, α5 or α6A.

It is preferred that at least a portion of the polypeptides used in the mimics of the present invention be prepared recombinantly. Recombinant preparation of polypeptides overcomes limitations of polypeptide length and modest yield encountered in the initial synthetic approaches of the prior art . Methods for recombinant preparation of at least a portion of a polypeptide are well known in the art. Polypeptides of the mimics or portions thereof may also be prepared synthetically. Methods for synthetic preparation of polypeptides are well known in the art. Further, methods for combining portions of synthetically and recombinantly prepared peptides into a single polypeptide are known. In the present invention, if both polypeptides of the mimic are prepared synthetically, at least one heptad repeat in the series of heptad repeats forming the coiled-coil sequences must differ in amino acid sequence from the other heptad repeats in the series.

Polypeptides of the model of the present invention are preferably >90% homogenous as determined by reverse phase C18 high pressure liquid chromatography and have a monomer mass that varies by less than 0.1% from that of the desired monomer sequence as determined by electrospray mass spectrometry . In this embodiment, formation of covalent dimers in aqueous solution can be observed by mass spectrometry and by SDS-PAGE, thus confirming the parallel orientation of the helices. In this embodiment, the beginning of the integrin cytoplasmic domain sequence provides the hydrophobic residues of a fifth heptad repeat (Figure 1) . Consequently, direct linkage of the coiled-coil sequence of the α-helical heptad repeat could induce helical structure in the tail. To address this possibility, embodiments of - 9 - the protein model containing additional glycines between the α-helical heptad repeats and the cytoplasmic domain sequence were synthesized (Figure 1) . Comparison was made of the CD-spectra of βl integrin constructs containing either only one glycine (Gl-βlA) or three additional glycines (G4-βlA) between the heptad repeats and the cytoplasmic domain. Insertion of glycines sharply reduced the minima at 208 and 222 nm. Consequently, predicted α- helical content in the protein model was reduced from 65% to 36%. The four heptad repeats constitute 27% of the mass of the construct; therefore, 36% helical content is consistent with the helical structure being limited to these repeats. Thus, the Gly insertion appears to eliminate α-helical structure induced in the cytoplasmic domain coiled-coil sequence.

To study possible influences of the structural changes induced by the Gly insertions, protein models were produced having the βlA cytoplasmic domain with one, two and three additional Gly residues inserted after the heptad-repeat motif (G2-, G3-, G4-βlA) and compared with the Gl-βlA construct. As an additional control, a variant of the G4- βlA peptide was produced with a Tyr to Ala substitution in the membrane-proximal NPXY-motif (G4-βlA-Y788A) (Figure 1) . This mutation interferes with focal adhesion targeting and activation of integrins. The purified proteins were bound via their N=terminal His-Tag to a Ni^2+~resin and used in affinity chromatography experiments with lysates of NHS- biotin- labeled human platelets. Marked changes in the pattern of protein binding were observed as a consequence of the Gly insertions. Polypeptides migrating at 45, 56, 58, 140 and 240 kDa bound only to the mimics with Gly insertions. The Y788A mutation in the G4-βlA construct (YA) suppressed the interaction with the 240 kDa, but not with the other components. Using monoclonal antibodies, the 240 kDa and 45 kDa proteins were identified as filamin and actin, respectively. The enriched 56, 58 and 140 kDa polypeptides have not been identified but have failed to react with antibodies specific for pp60^Ξrc, paxillin, ppl25^fak, α-actinin, vinculin and pp72^Ξyk in Western blotting experiments. Talin bound to the Gl- and G4-βlA construct but not to the Y788A-G4βlA construct. Thus, the structural changes in the model induced by the insertion of glycines into the coiled-coil motif and the integrin cytoplasmic domain sequence alter interactions of these proteins with cellular components. Alterations of the βlA tail that block cytoskeletal interactions, such as the Y788 mutation and βlB- and βlC-splice variants also abrogate binding to talin and filamin. Consequently, the observed in vi tro interactions are likely to be biologically relevant. Models of the present invention were also constructed with Gl- and G4- polypeptides of the muscle-specific splice variant βlD and the β7 integrin subunits (Figure 1) to study binding interactions of various integrin binding proteins. When used with NHS-biotinylated platelet lysates, the βlD constructs bound more talin and β7 constructs bound more filamin, compared to βlA. In addition, these differences in binding were consistently observed when lysates of a human T-cell leukemia cell line (Jurkat) , a human fibrosarcoma cell line (HT 1080) , and a differentiated myotubes derived from a mouse myoblast cell line (C2C12) , were used for affinity-chromatography . Moreover, stronger binding of the βlD constructs to talin and of the β7 constructs to filamin was independently observed, both with the Gl- as well as the G4 -variants of the model proteins, indicating that the structural changes induced by Gly insertions do not strongly influence these differential interactions.

Purified preparations of these proteins were then used to demonstrate that the observed interactions with talin and filamin in the cell extracts are direct. The relative amounts of purified filamin and talin bound to the model proteins were similar to those observed with cell lysates. Specifically, βlD constructs bound more talin and β7 constructs bound more filamin than βlA protein models. In addition, binding of both cytoskeletal proteins to the G4- Y788A-βlA construct and to the G4-βlB and G4-βlC variants was functionally reduced compared to G4-βlA. Moreover, G4- constructs of βlA, βlD and β7 integrin cytoplasmic domains bound more purified filamin than the corresponding Gl- constructs. However, the Gl-β7 model protein still bound more filamin than G4-βlA or G4-βlD. A densitometric evaluation of the Coomassie blue-stained gels indicated that the βlD construct bound about nine times more talin, and the β7 construct bound 8.4 times more filamin than the βlA model protein. In these experiments, there was a >10 fold molar excess of model proteins relative to the quantity of talin and filamin. Thus, the affinity of βlA for filamin is at least eight fold less than that of β7, and its affinity for talin is at least nine fold less than that of βlD.

Cytoplasmic domain mimics of the α4 integrin have also been prepared in accordance with the present invention. The α4 integrin subunit is indispensible for embryogenesis , hematopoiesis and the immune response (Stewart et al . Curr. Opin. Cell Biol. 7, 690-696 (1995); Shimizu et al . Adv. Immunol. 72, 325-380 (1999)). Because of their central role in the immune response α4 integrins are strongly implicated as potential therapeutic targets for inflammatory bowel disease, arthritis, multiple sclerosis and asthma. It has been suggested that α4 may regulate cell migration, cytoskeletal organization and gene expression differently from other integrin α subunits (Hemler et al . Cold Spring Harbor Symposia on Quantitative Biology: The Cell Surface 57, 213-220 (1992)). These biological properties are dependent on the α4 cytoplasmic domain (Stewart et al . Curr. Opin. Cell Biol. 7, 690-696 (1995); Hemler et al . Cold Spring Harbor Symposia on Quantitative Biology: The Cell Surface 57, 213-220 (1992) ; Newton et al . J. Leukocyte Biol. 61, 422-426 (1997)). Structural mimics of the present invention comprising the α4 cytoplasmic tails were prepared and used to identify molecules involved in α4 integrin-specific signaling.

To identify biochemical bases for the signaling properties of the α4 integrins, the binding of cellular proteins to structural mimics of dimerized α4 integrin cytoplasmic domains was analyzed. These structural mimics were formed by fusing the cytoplasmic tail of the α4 or βlA subunit to an N-terminal sequence containing 4 heptad repeat sequences which form the coiled-coil di ers so that the cytoplasmic domains are parallel dimerized and held in a fixed vertical stagger.

Lysates of Jurkat T lymphoblasts were then incubated with immobilized α4 cytoplasmic domain mimics. Bound proteins were detected by immunoblotting for previously identified integrin cytoplasmic domain binding proteins. Within the bound fraction, it was found that paxillin was enriched greater than 57 fold as compared to the cell lysate. In contrast, while the βlA cytoplasmic domain bound paxillin, there was no enrichment relative to the cell lysate. The interactions with both the α4 and βlA tails were specific in that binding was not seen to resin bearing no protein nor to the αllb cytoplasmic domain. Heterodimers of the α4βlA tails were also produced. These heterodimers bound similar quantities of paxillin to the α4 tail alone. The α4 tail also bound small amounts of the actin-binding proteins filamin and tails. However, these proteins were not enriched relative to the cell lysate. Further, the α4 tail did not bind to vinculin or α-actinin. There are seven conserved N-terminal residues in the α integrin subunit cytoplasmic tails (Hemler et al . Cold Spring Harbor Symposia on Quantitative Biology: The Cell Surface 57, 213-220 (1992); Williams et al . Trends Cell Biol. 4, 109-112 (1994); and Sastry S.K. and Horwitz, A.F. Curr. Opin. Cell Biol. 5, 819-831 (1993)). Accordingly, the specificity of the interaction of paxillin with the α4 cytoplasmic tails was determined by examining paxillin binding to a series of α cytoplasmic domains. Paxillin failed to bind to the αllb, α3A, α5 or α6A tails. Thus, conservation of seven residues of the N-terminal by integrin α cytoplasmic domains does not appear to be sufficient to mediate paxillin binding.

In addition to paxillin, experiments were also performed to determine whether the paxillin related proteins Hic-5 and leupaxin also bind to the α4 cytoplasmic tails. In some experiments, a minor 55 K band, the size of Hic-5, was observed bound to the α4 column. Using platelet extracts as a source of Hic-5, a 7.4 fold enrichment compared to cell lysate was observed. In similar experiments in Jurkat cells, a 1.9 fold enrichment compared to starting lysate was observed for leupaxin.

Experiments were also performed to confirm that paxillin is also associated with intact α4 integrins and intact α4βl integrins. In these experiments, Jurkat cell lysate was immunoprecipitated with monoclonal antibodies reactive with α4 , βl or α5 integrin subunits or monoclonal antibodies reactive to paxillin or an irrelevant IgG, respectively. Paxillin was present in the α4 and βl immunoprecipitates, but not in the immunoprecipitates formed with α5 antibody or irrelevant IgG. Immunoprecipitates of the surface biotin-labeled cells confirmed the immunoprecipitation of α4βl by the α4 antibody, α5βl by the α5 antibody and a mixture of these two plus a band with mobility of αl by the anti βl antibody. In cell lysate immunoprecipitated with monoclonal antibody to paxillin or an irrelevant IgG, α4βl integrin, but not α5βl integrin co-precipitated with paxillin. α4βl did not co-precipitate with the irrelevant IgG.

Paxillin' s tight association with the α4 tails and its ready isolation with α4 integrins is indicative of a significant fraction of α4βl being associated with paxillin in the cells. To determine this fraction, surface biotin labeled Jurkat cell lysate was sequentially immunoprecipitated with anti -paxillin antibody or irrelevant IgG. Western blotting with anti -paxillin antibody confirmed depletion of virtually all paxillin. Paxillin depletion resulted in almost complete loss of α4 in the lysate. In contrast, there was little depletion of α5. Immunoprecipitation with an irrelevant IgG did not result in significant loss of either α4βl or α5βl. Accordingly, a majority of or all of the α4 appears to physically associate with paxillin.

A chimera consisting of the αllb extracellular and transmembrane domain and the α4 cytoplasmic domain was then constructed to determine whether the α4 cytoplasmic tail alone is sufficient to connect paxillin to an integrin. To provide appropriate β tail partners, the extracellular and transmembrane domains of β3 were joined to the βlA or β7 cytoplasmic domain. The αIIbα4β3βlA and αllbα4β3β7 chimeric integrins were expressed in CHO cells. A chimera in which the α6A cytoplasmic domain was joined to αllb and expressed as αIIbα6Aβ3βlA chimera in CHO cells was used (Hughes et al . Cell 88, 521-530 (1997)). When lysates from these cells were immunoprecipitated with antibodies against the extracellular domain of αllbβ3, similar quantities of recombinant integrin were precipitated from each cell line. Only cells containing the α4 tail-bearing chimeric integrin manifested substantial paxillin co- immunoprecipitation. Thus, the α4 cytoplasmic domain must mediate the association of intact integrins with paxillin. The functional effect of the α4 tails was then examined by assaying cells adhesion and spreading on the αllbβ3 ligand, fibrinogen. The α4 tail did not alter αllbβ3 -dependent cell adhesion. However, the α4 tail opposed αllbβ3 -dependent cell spreading. These two cell lines adhered and spread equally well on a ligand for endogenous α5βl, fibronectin, confirming that the effect was specific to the recombinant integrin. In assaying the paxillin-binding site within the α4 tail, an amino acid residue was identified, Y991A, that disrupted binding of paxillin. This mutation was introduced into α4 chimera and αllbα4 (Y991A) β3βlA was expressed in CHO cells. This mutation restored αllbβ3 -dependent cell spreading, but did not alter either αllbβ3 -dependent cell adhesions or cell spreading on fibronectin. Thus, interaction of α4 tail with paxillin results in diminished cell spreading.

To confirm that paxillin is required for α4- inhibition of cell spreading, the α4 subunit was expressed in primary fibroblasts derived from wild-type or paxillin-deficient mice and cell spreading on VCAM-1 , an α4 integrin- specific ligand, was assayed. Primary mouse embryonic fibroblasts from two paxillin-null embryos spread where those from littermate wild-type embryos failed to spread. To determine whether other cytosolic proteins may be mediating the observed binding of paxillin to the α4 complex in whole cell extracts, a recombinant human paxillin-GST fusion protein was prepared. Purified recombinant paxillin-GST fusion protein quantitatively bound to the α4 cytoplasmic domain. In contrast, paxillin binding was not detectable on the αllb tail. Further, there was no binding of GST to the α4 tail. Since both binding partners are recombinant bacterial proteins, a requirement for tyrosine phosphorylation in the direct interaction of paxillin with the α4 tail can be excluded. - 16 -

Paxillin binding to the α4 tails was saturable and of high affinity.

These experiments with the structural mimics of the present invention demonstrate that paxillin binds directly and tightly to the α4 cytoplasmic tail . Paxillin is therefore believed to play an important role in the signaling properties of α4 integrins. In particular, it is believed that direct binding of paxillin to α4 tail opposes α4 -dependent cell spreading. Thus, blockade of the binding of α4 to paxillin should inhibit α4 -mediated cell migration. Since a major function of α4 is the migration and trafficking of leukocytes, inhibitors of the binding of paxillin to α4 are expected to be useful in blocking immune responses. α4 integrin activation has also been associated with atherosclerosis. Accordingly, agents which inhibit activation will also be useful in inhibiting atherosclerosis. Further activation of α4 integrin occurs during wound healing. More specifically, α4 integrin activation signals monocytes to aggregate at the wound site. However, this aggregation can lead to scarring.

Accordingly, inhibition of α4 integrin activation is also useful in inhibiting scarring during wound healing.

This structural model was used to identify a 15 mer peptide, SILQEENRRDSWSYI (SEQ ID NO: 15) derived from the α4 cytoplasmic domain as an inhibitor of the binding of paxillin and the α4 tail. The IC50 of inhibition of the interaction of paxillin and the α4 tail by this peptide was 150 μM. Similar experiments with additional 15 mer peptides, KAGFFKRQYKSILQE (SEQ ID NO: 16) and RRDSWSYINSKSNDD (SEQ ID NO:17), showed no inhibition.

Further substitution of various single amino acids within SEQ ID NO: 15 with alanine also abolished inhibitory activity. Thus, inhibition by the 15 mer peptide SILQEENRRDSWSYI (SEQ ID NO: 15) is structurally specific. The core active sequence of this peptide has been determined to comprise the 9 amino acid sequence ENRRDSWSY (SEQ ID NO: 18) . Knowledge of this core sequence and its structure are useful in the rational design of therapeutic agents which inhibit α4 integrin biological responses. As demonstrated by these experiments, the structural models of the present invention provide a novel experimental tool for the analysis of various proteins associations with integrin tails in vi tro and the structural aspect of the cytoplasmic face of integrins . The structural models of the present invention thus have a number of applications based upon their ability to maintain the cytoplasmic tails of the construct in a configuration that is equivalent or similar to the configuration predominating in vivo while maintaining solubility and stability in an aqueous system, namely in staggered, parallel, and proximal topology. As demonstrated herein, these models can be used to detect intracellular molecules capable of binding to integrins and modulating signals by inside-out signaling. Alternatively, these molecules can be used in vivo to disrupt or modulate inside-out signaling by binding to the cells in a manner such that the cytoplasmic domains of these recombinant models compete for intracellular molecules with the natural integrins. Because these structural models do not contain the extracellular ligand-binding sites of integrins, they would then disrupt inside-out signaling. This would be particularly useful in conditions in which overactivity of integrins is involved, such as inflammation, thrombosis, and malignancy. This would provide a new method of treating such conditions or their sequelae; because these molecules mimic the orientation of the natural integrins within the membrane, they would not disrupt membrane structure and would therefore be better tolerated and avoid side effects. Additionally, structural models of the present invention can be used to detect molecules capable of binding to the intracellular or cytoplasmic domain of integrins and other transmembrane molecules in vivo, such as by affinity chromatography . Accordingly, these models are useful in identifying various therapeutic compounds for selected cytoplasmic domains. By "therapeutic compounds" it is meant to include, but is not limited to, molecules which are found to bind to a selected cytoplasmic domain of the model, molecules which bind to proteins that bind to the cytoplasmic domain of the model, and the models themselves. For example, in one embodiment, a structural model or mimic comprising an α4 integrin cytoplasmic tail can be used in a high throughput screening assay to identify agents which inhibit binding of paxillin to the α4 cytoplasmic tails. In this assay, the structural model comprising an α4 integrin cytoplasmic tail is exposed to paxillin or a paxillin related molecule in the presence or absence of a test agent. Binding of paxillin or the paxillin related molecule to the structural model in the presence and absence of the test agent is then determined. A test agent which decreases binding of paxillin or the paxillin related molecule to the structural model as compared to binding of paxillin or paxillin related molecules to the structural model in the absence of the test agent can inhibit biological responses relating to α4 integrins. For example, these agents may be useful in inhibiting normal wound healing response of α4 integrin which can lead to scarring. These agents can also be used in the inhibition of pathological responses of α4 integrin such as those involved in atherosclerosis and immune responses in conditions such as inflammatory bowel disease, arthritis, multiple sclerosis, and asthma. Compositions comprising such agents and a known pharmaceutically acceptable vehicle are believed to be useful therapeutically to inhibit biological responses of α4 integrins. The following examples are provided for illustrative purposes only and are not intended to limit the invention. EXAMPLES

Example 1 Antibodies and cDNAs Antibodies for the analysis of proteins bound to cytoplasmic domain model proteins on Western blots included: goat serum against filamin (Sigma Chemical Co., St. Louis, MO), rabbit serum against α-actinin (Sigma Chemical Co.), mAbs against talin (clone 8d4) (Sigma Chemical Co.), vinculin (clone hVIN-1) (Sigma Chemical Co.), pacillin (clone Z035) (Zy ed Laboratories Inc., S. San Francisco, CA) , filamin (MAB1680) (Chemicon International Inc. Temecula, CA) , α-actinin (MB75.2) (Sigma Chemical Co.), actin (clone C4 ) (Boehringer-Mannheim Corp., Indianapolis, IN) , mAb against pp60^src (clone 327) , polyclonal rabbit serum against pp^125FAK (BC3) and rabbit anti-pp72^syk, b against human βl integrin (B-D15, BioSource, International), mAb against human α4 integrin (HP2/1, ImmunoTech) , mAb human against human α5 integrin (PharMingen), mAb against HA-tag (12C5, ATCC), mAb against paxillin (clone 349, Transduction Laboratories), and mAb against GST (B-14, Santa Cruz). Polyclonal antibody against FAK (C-20, Santa Cruz was also used. Biotin labeled anti -paxillin antibody was prepared by labeling commercial anti-paxillin (clone 349) with NHS-Biotin (Pierce) according to the manufacturer's instructions. Rabbit polyclonal anti- leupaxin was raised against the N- terminal 14 amino acids of human leupaxin (Lipsky et al . J. Biol. Chem. 273 11709-11713 (1998)). Human cDNA used in these experiments included: βlC cDNA; βl cDNA with the point mutation, Y788A1; a cDNA for the cytoplasmic domain of human integrin βlD obtained by RT-PCT of heart muscle total RNA; cDNA of human integrin β7; and a cDNA coding for the human βlB subunit cytoplasmic domain synthesized in PCR reactions using a human βlA vector with a partially overlapping reverse-oligonucleotide containing the human βlB sequence.

Example 2 Recombinant cytoplasmic domain models

Oligonucleotides were synthesized and used in PCR reactions to create a cDNA for the α-helical heptad repeat protein sequence KLEALEGRLDALEGKLEALEGKLDALEG (SEQ ID NO: 6) Gl- ( [heptad] ₄ ) . Variants containing 1 to 3 additional Gly residues (G2-4- ( [heptad] ₄) ) at the C-terminus were synthesized by modification of the antisense oligonucleotide . These cDNAs were ligated into a Ndel- Hindlll restricted modified pET15b vector (Novagen, Madison, WI) . Integrin cytoplasmic domains were joined to the helix as a Hindlll-BamHI fragments. The final constructs coded for the N-terminal sequence GSSHHHHHHSSGLVPRGSHMCG (SEQ ID NO: 5) [heptad] ₄ linked to the cytoplasmic domains of integrins. Different cytoplasmic domain cDNAs were cloned via PCR from appropriate cDNAs using forward oligonucleotides introducing a 5 ' -Hindlll site and reverse oligonucleotide creating a 3 ' -BamHI site directly after the Stop-codon.

PCR products were first ligated into the pCR™ vector using the TA cloning^® kit (Invitrogen Corp., San Diego, CA) . After sequencing, Hindlll/BamHI inserts were ligated into a modified pET15b vector. Recombinant expression in BL21 (DE3)pLysS cells (Novagen) and purification of the recombinant products were performed according to the pET System Manual (Novagen) with an additional final purification step on a reverse phase C18 HPLC column (Vydac, Hesperia, CA) . Products were analyzed by electrospray mass spectrometry on an API -III quadruple spectrometer (Sciex, Toronto, Ontario, Canada) . Example 3 Ultraviolet circular dichroism spectroscopy

Far UV CD spectra were recorded on an AVIV 60DS spectropolarimeter with peptides dissolved in 50 mM boric acid pH 7.0. Data were corrected for the spectrum obtained with buffer only and related to protein concentrations determined from identical samples by quantitative amino acid analysis. From these values, the percentage of helical secondary structure was calculated in accordance with procedures described by Muir et al . Biochemistry 33:7701 (1994) . Example 4 Cells and cell lysates

Human platelets were obtained by centrifugation of freshly drawn blood samples at 1000 rpm for 20 minutes and sedimentation of the resulting platelet-rich plasma at 2600 rpm for 15 minutes. They were washed twice with 0.12 M NaCl, 0.0129 M trisodium citrate, 0.03 M glucose, pH 6.5, and once in Hepes-Saline (3.8 mM Hepes, 137 mM NaCl, 2.7 mM KC1, 5.6 mM D-Glucose, 3.3 mM Na₂HP0₄ , pH 7.3-7.4). Human Jurkat and HT1080 cells and mouse C2C12 cells were obtained from the American Type Culture Collection (Rockville, MD) and cultured in RPMI1680 (Jurkat) or DMEM with 10% fetal calf serum. For differentiation to myotubes, C2C12 myoblasts were kept confluent in DMEM with 5% horse serum for 6 days. Cultured cells were washed twice in phosphate- buffered saline (PBS) and biotinylated with 1 mM NHS-biotin (Pierce) in PBS during 30 minutes at room temperature. Platelets were biotinylated in Hepes-Saline. After two additional washes with TBS, cells were lysed on ice with buffer A (1 mM Na₃V0₄, 50 mM NaF, 40 mM NaPyrophosphate, 10 mM Pipes, 50 mM NaCl, 150 mM sucrose, pH 6.8) containing 1% TRITON X-100, 0.5% sodium deoxycholate, 1 mM EDTA and protease inhibitors (1/lOOth volume of aprotinin (Sigma A- 6379) , 5 μg/ml leupeptin, 1 mM PMSF) . To platelet lysates 0.1 mM of the calpain inhibitor E-64 (Boehringer Mannheim) were added in addition. Lysates were sonicated 5 times on ice for 10 seconds at a setting of 3 using an Astrason Ultrasonic Processor (Heart Systems, Farmingdale, NY). After 30 minutes, lysates were clarified by centrifugation at 12,000 g for 30 minutes. Example 5 Affinity chromatography experiments with integrin cytoplasmic domain mimics

Purified recombinant cytoplasmic domain proteins (500 μg) were dissolved in a mixture of 5 ml 20 mM Pipes, 50 mM NaCl, pH 6.8 and 1 ml 0.1 M sodium acetate, pH 3.5 and bound overnight to 80 μl of Ni^2t saturated His-bind resin (Novagen) . In control experiments, it was found that this leads to approximate saturation of the resin with peptide. Resins were washed twice with 20 mM Pipes, 50 mM NaCl, pH 6.8, and stored at 4°C with 0.1% sodium azide as suspensions with one volume of this buffer. Fifty microliters of such a suspension were added to 4.5 ml of cell lysates which had been diluted tenfold with buffer A containing 0.05% TRITON X-100, 3 mM MgCl₂ and protease-inhibitors . After incubation overnight at 4°C, resins were washed five times with this buffer and finally heated in 50 μl of reducing sample buffer for SDS PAGE. Samples were separated on 4-20% SDS polyacrylamide gels (NOVEX) and either stained with Coomassie or transferred to Immobilon P membranes (Amersham Corp., Arlington Hts, IL) . Membranes were blocked with TBS, 5% nonfat -mild powder and stained with streptavidin- peroxidase (VECTASTAIN) or specific antibodies. Bound peroxidase was detected with an enhanced chemiluminescence kit (Amersham) . Example 6 Binding to purified talin and filamin

Human uterus filamin (ABP-280) was prepared as a 1.5 mg/ml solution in 0.6 M KC1, 0.5 mM ATP, 0.5 mM DTT, 10 mM imidazole, pH 7.5. For binding assays performed as described in Example 5, this solution was diluted 1/12 with buffer A, 0.05% TRITON X-100, 3 mM MgCl₂, 2 mg/ml BSA, protease-inhibitors (see Example 5) , omitting the 50 mM NaCl (see Example 5) , and resins with bound model proteins were added. Washing was performed in this buffer without BSA and with additional 50 mM Kcl. Talin was purified from human platelets in accordance with well known procedures with an additional purification step using chromatography on phosphocellulose and stored at 1 mg/ml in 10 mM NaCl, 50% glycerol . This solution was diluted to either 87 or 17 μg/ml talin with buffer A, 0.05% TRITON X-100, 3 mM MgCl₂, 2 mg/ml BSA and protease inhibitors (see Example 5, including 0.1 mM E-64) and processed as indicated in the binding assays with cell lysates. For densitometric analysis, scans of Coomassie- stained gels were processed using the program NIH- Image

(NIH, Bethesda, MD) . Equal loading of gels was controlled in Coomassie-stained gels of the recombinant cytoplasmic domain polypeptides coeluted with the ligand from the resins . Example 7 Chimera formation

The αllbα4 and αIIbα4*Y991A) chimeras were formed by connecting human αllb extracellular and transmembrane domains to human α4 or α4 (Y991A) cytoplasmic domain. β3βlA or β3β7 chimeras were formed by connecting human β3 extracellular and transmembrane domains to human βlA or β7 cytoplasmic domains. CHO cells stably expressing αIIbα4β3βlA, αllbα4 (Y991A) β3βlA, or cell lines expressing these chimeras were transfected and isolated as described by Hughes et al . Cell 88, 521-530 (1997). Primary mouse embryonic fibroblasts from paxillin-null and littermate matched wild-type embryos were isolated by standard methods such as those described by Thomas et al . Nature 376 (6537), 267-71 (1995) ) . Example 8 Immunoprecipitation and Western blot analysis Jurkat T cells or CHO cells were cell surface-labeled with sulfo-NHS-Biotin (Pierce) in accordance with the manufacturer's instructions. Cell lysate was prepared and immunoprecipitation was performed as described by Chen et al . Blood 84, 1857-1865 (1994). Precipitated cell surface biotin-labeled polypeptides were separated under non- reducing conditions and detected with streptavidin- peroxidase followed by ECL (Amersham) . Immunoprecipitation for detection of co-precipitated paxillin was performed as above except cells were not surface-labeled with biotin; and immunoprecipitated proteins were separated under reducing conditions and paxillin co-precipitation was detected with biotin-labeled anti-paxillin . For co- precipitation of α4βl with paxillin, surface biotin-labeled Jurkat cell lysate was precipitated with antibodies reactive to paxillin, α4 , α5 or irrelevant IgG.

Immunoprecipitates were separated on 6% SDS-PAGE under non- reducing conditions and surface polypeptides were detected with streptavidin-peroxidase and ECL. For paxillin- depletion assay, aliquots of cell surface biotinylated Jurkat T cell lysate were subjected to varying rounds of immunoprecipitation using anti-paxillin antibody or irrelevant IgG. The degree of paxillin-depletion in the cell lysate was assessed by Western blot analysis. Cell lysates with or without paxillin-depletion, as well as with the irrelevant IgG precipitation, were then immunoprecipitated with either anti-α4 or α5 antibody. Immunoprecipitates of surface proteins were separated on 6% SDS-PAGE under non-reducing conditions and polypeptides were detected with streptavidin-peroxidase and ECL. Example 9 Cell Adhesion and Spreading Assays

Assays of cell adhesion and spreading on fibrinogen or fibronectin for different CHO cell lines were performed in accordance with procedures described by Ylanne et al . J. Cell Biol. 122, 223-233 (1993). For cell spreading assay of mouse fibroblasts, paxillin knock-out as well as wild- type cells were transfected with human α4 integrin subunit using retroviral infection. Forty-eight hours after transfection, equal expression of α4 integrin in wild-type and knock-out cells was observed by FACS using anti-α4 antibody. Cells resuspended in DMEM plus 1 mg/ml of BSA were plated on coverslips coated with 10 μg/ml of either VCAM-1 (Biogen Inc. Cambridge MA) or fibronectin (Sigma Chemical Co.) and incubated at 37°C for 1 hour. Unattached cells were washed away with PBS. Attached cells were fixed with 3.7% paraformaldehyde and examined by phase microscopy. Photo images were taken with a Nikon Diaphot microscope equipped with a Sensys cooled CCD video camera. Example 10 Production and Binding of Recombinant Paxillin Recombinant human paxillin was expressed and isolated in accordance with procedures described by Salgia et al . J. Biol. Chem. 270, 5039-5047 (1995). Aliquots of recombinant GST-paxillin or GST alone were mixed with 300 μl of buffer A plus 20 μg/ml of aprotinin, 5 μg/ml of leupeptin, 1 mM PMSF, 0.1% Triton X-100, 3 mM MgCl2 , and 1 mg/ml of BSA, added to model protein- loaded resins, and incubated at room temperature with rotation for 2 hours. Both bound and unbound proteins were collected and detected with antibodies specific for HA-tag or GST. For determination of EC50 of paxillin binding to α4 tail, different amounts of recombinant paxillin were added to α4 or αllb tail- loaded resins and bound paxillin was assayed as described above .

Claims

What is Claimed is:

1. A polypeptide comprising:

(a) a series of heptad-repeats that mimic a transmembrane domain; and (b) a selected cytoplasmic domain attached to the heptad repeats, wherein at least a portion of the polypeptide is prepared recombinantly .

2. The polypeptide of claim 1 wherein the selected cytoplasmic domain is an integrin cytoplasmic domain.

3. The polypeptide of claim 2 wherein the integrin cytoplasmic domain is a β or α integrin subunit.

4. The polypeptide of claim 1 further comprising one or more glycine residues inserted between the heptad repeats and the selected cytoplasmic domain.

5. The polypeptide of claim 1 further comprising an immobilizing epitope linked to the series of heptad repeats of the polypeptide via a Cys-Gly linker.

6. The polypeptide of claim 5 wherein a chemical or prokaryotic cleavage site is inserted between the immobilizing epitope and the Cys-Gly linker.

7. A polypeptide comprising:

(a) a series of heptad-repeats that mimic a transmembrane domain; and (b) a selected cytoplasmic domain attached to the heptad repeats, wherein at least one heptad repeat in the series has a different amino acid sequence to other heptad repeats in the series .

8. A structural model of a selected cytoplasmic domain comprising a polypeptide of claim 1 for evaluating structure and activity of a selected occupied and clustered transmembrane protein having the selected cytoplasmic domain .

9. A structural model of a selected cytoplasmic domain comprising a polypeptide of claim 4 for evaluating structure and activity of a selected occupied and clustered transmembrane protein having the selected cytoplasmic domain.

10. A structural model of a selected cytoplasmic domain comprising a polypeptide of claim 1 for use in identification of therapeutic compounds.

11. The structural model of claim 10 wherein the polypeptide comprise an α4 cytoplasmic domain.

12. A method of identifying agents as inhibitors of α4 integrin biological responses comprising contacting the structural model of claim 11 with paxillin or a paxillin related molecule in the presence and absence of a test agent; and determining binding of paxillin or the paxillin related molecule to the structural model in the presence and absence of the test compound wherein a decrease in binding of the paxillin or paxillin related molecule to the structural model in the presence of the test agent as compared to binding in the absence of the test agent is indicative of the test agent being an inhibitor of α4 integrin biological responses.

13. A composition for inhibiting an α4 integrin biological response comprising an agent which inhibits binding of paxillin or paxillin related molecules to α4 integrin and a pharmaceutically acceptable vehicle.