WO1991004745A1

WO1991004745A1 - Compositions for cell adhesion inhibition and methods of use

Info

Publication number: WO1991004745A1
Application number: PCT/US1990/005105
Authority: WO
Inventors: Lee L. Rubin; Chen W. Liaw; Kevin J. Tomaselli
Original assignee: Athena Neurosciences, Inc.
Priority date: 1989-09-27
Filing date: 1990-09-13
Publication date: 1991-04-18
Also published as: CA2067183A1; EP0494175A1; JPH05500504A; EP0494175A4

Abstract

Compositions that disrupt microvascular endothelial and epithelial cell tight junctions, and methods of use, are disclosed. Such compositions comprise agents that inhibit the binding to such cells of cell adhesion molecules. Such inhibitor agents include cell adhesion molecules, fragments of cell adhesion molecules that encompass a cell-binding domain such as HAV, and antibodies directed against cell adhesion molecules and fragments thereof. Also disclosed are drug delivery compositions comprising a therapeutic drug conjugated to an agent that disrupts cell tight junctions.

Description

COMPOSITIONS FOR CELL ADHESION INHIBITION

AND METHODS OF USE

This is a continuation-in-part of United States Serial No. 07/413,332, filed September 27, 1989. Background of the Invention

Field of the Invention

This invention relates to compositions that transiently and reversibly dissociate the blood-brain barrier. More particularly, the invention relates to compositions that dissociate tight junctions between brain capillary endothelial cells that constitute the physiological barrier between the general circulation and the brain.

Detailed Description of Related Art

The entry of drugs from the blood stream to the central nervous system (CNS), i.e., the brain and spinal cord, is restricted by the presence of high resistance tight junctions between brain capillary cells and by the apparently low rate of transport across these endothelial cells (Betz, A.L., et al., Ann. Rev. Physiol., 48:241 (1986); Pardridge, W.M., Ann. Rev. Pharmacol. Toxicol., 28:25 (1988)).

The tight junctions of the blood brain barrier (BBB) prevent diffusion of molecules and ions around the brain capillary endothelial cells. The only substances that can readily pass from the luminal core of the capillary to the abluminal tissues that surround the capillary are those molecules for which selective transport systems exist in the endothelial cells, as well as those compounds that are lipophilic (i.e., hydrophobic). In contrast, drugs, peptides and other molecules that are neither lipophilic nor transported by specific carrier proteins are barred from entry into the brain, or their rates of entry are too low to be useful, thereby imposing a severe limitation upon the physician's ability to treat CNS disorders

pharmacologically.

The carrier-mediated transcellular transport system mentioned above may have limited usefulness for therapeutic modalities under some circumstances.

Transcytotic transport, in general, involves, first, the binding of molecules to specific carrier proteins on the surface of endothelial cells, and, second, the delivery of such molecules across the endothelial cells. Limitations on the usefulness of such a system for treatment of CNS disorders are based on the

following considerations: (1) physiological carrier proteins may not function efficiently, or at all, with non-physiological drugs; (2) even where function occurs, the rate of transport of therapeutic agents will be limited by the rate of transport of the

carrier; (3 ) the overall capacity of cerebral capillary endothelial cells to transport any therapeutic

macromolecules may be simply too low to achieve

therapeutic levels of certain drugs in the brain; and (4) once therapeutic macromolecules enter endothelial cells, depending on their nature, they might be

delivered to any number of organelles, including lysosomes that contain a wide variety of hydrolytic enzymes. For these reasons, creating drug delivery systems that do not rely upon transcytosis will clearly be advantageous.

As tight junctions between brain capillary endothelial cells constitute a major part of the BBB, the possibility of modifying these junctions has been considered. It has been found that tight junctions, including those of the BBB, can be disrupted by

hyperosmotic solutions administered intra-arterially. For example, Polley et al., WO89/04663, published

June 1, 1989, disclose the osmotic disruption of the interendothelial structure of the BBB by the intra- arterial administration of hypertonic solutions of mannitol, arabinose or glycerol as a means of

introducing into the brain genetic material.

Similarly, hyperosmotic solutions of urea have also been used to alter the BBB (Bowman, P.D. et al., Ped. Res., 16:335A (1982)).

Other chemical agents have been reported to disrupt endothelial or epithelial cell tight junctions when administered intravenously, including:

7-fluorouracil (MacDonell, L.A., et al., Cancer. Res., 38:2930 (1978)), degradation by membrane enzymes

(Vincent, P.A., et al., Exp. Mol. Path., 48:403 (1988); Diener, H.M., et al., J. Immunol., 135:537 (1985)), aluminum salts (Zigler, Z.Y., et al., IRCS Med. Sci., 12:1095 (1984)), histamine (Meyrick, B., et al., Exp. Lung Res., 6:11 (1984)), thrombin (Siflinger-Birnboin, A., et al., Microvasc. Res., 36:216 (1988)), phorbol esters (Shiba, K., et al., Exp. Cell Res., 178:233 (1988)), and neutralization of the luminal anionic charge (Hart, M.M., J. Neuropathol. EXP. Neurol.,

46:141 (1987)).

Although the above-listed modalities may disrupt tight junctions and thereby increase permeability of the BBB, problems attendant upon their use make them less than desireable. For example, intra-arterial perfusion with hyperosmotic solutions involves surgery, and this cannot be repeated on a regular basis.

Further, concentrated sugar solutions may not be innocuous, and might be expected to have undesirable side effects. In addition, the aforementioned chemical agents may not be useful for the treatment of chronic neurological disease, their effects on tight junctions are not always reversible, and, as they all are

themselves powerful drugs, there is always the danger that their use will compromise the patient's health generally. For example, 7-fluorouracil is a powerful inhibitor of pyrimidine synthesis, and thus nucleic acid biosynthesis, in animals cells.

Thus, an important need still exists for means which transiently and reversibly disrupt tight

junctions of the BBB in order that administered drugs can reach the brain from the general circulation, and which have no undesirable side effects of their own in the subject.

Attempts have been made to disrupt cell-cell adhesion by modifying the protein(s) responsible for such adhesion, collectively referred to as "cell adhesion molecules" (CAM). One class of CAM is termed "cadherin". "Cadherin" is the term applied to a family of glycoproteins found in most kinds of mammalian tissues and thought to be responsible for Ca²⁺- dependent cell-cell adhesion, (Takeichi, M.,

Development, 102:639 (1988)). Three subclasses of cadherin have been identified, namely, E-cadherin (from epithelial tissues), P-cadherin (from placental tissues), and N-cadherin (from neural tissues)

(Yoshida-Noro, C., et al. Dev. Biol., 101:19 (1984); Nose, A., et al., J. Cell Biol., 103:2649 (1986);

Hatta, K., et al., Nature, 320:447 (1986)).

The different cadherins exhibit distinct tissue distribution patterns (Takeichi, U., (1988) above). E-cadherin, which was found to be distributed

exclusively in epithelial cells of various tissues (Hatta, K., et al., Proc. Nat'l. Acad. Sci. (USA), 82:2789 (1985); Takeichi, 1988, above), appears to be identical to uvomorulin (Hyafil, F., et al., Cell, 21:927 (1986)), chicken liver-cell adhesion molecule (L-CAM, Gallin, W.J., et al., Proc. Nat. Acad. Sci. (USA). 80:1038 (1983)), and cell-CAM 120/80 (Damsky, C.H., et al., Cell, 34:455 (1983)) in terms of

biochemical properties (Cunningham, B.A., et al., Proc. Nat. Acad. Sci. (USA). 81:5787 (1984)) and tissue distributions (Thiery, J.-P., et al., Dev. Biol.,

102:61 (1984)).

N-cadherin, which is expressed in various neural tissues including astrocytes (Hatta, K., et al., Devel. Biol., 120:215 (1987); Matsunega, M., et al., Nature. 334:62 (1988); Tomaselli, K.J., Neuron, 1:33 (1988)), shows 92% amino acid sequence homology between

mammalian and avian homologs, shows from 40 to 50% similarity to epithelial E-cadherin and to placental P-cadherin of the same species, but was immunologically not cross-reactive with other cadherins within the same animal (Miyatani, S., Science, 245:631 (1989)).

Placental P-cadherin has also been cloned, and the deduced amino acid sequence of this glycoprotein was found to exhibit about 58% homology with epithelial E-cadherin (Nose, A., et al., EMBO J., 12:3655 (1987)).

Subsequent to the September 27, 1989 filing of the parent application, Heimark, et al. (Heimark, R.L., et al., J. Cell Biol., 110:1745 (1990) reported on the identification of a Ca²⁺-dependent cell-cell adhesion molecule in aortic endothelial cells.

Although each of the aforelisted cadherins

displays unique immunological and tissue distribution specifications, all have features in common: (1) a requirement for Ca²⁺ for cell adhesion function; (2) protection by Ca²⁺ from proteolytic cleavage; (3) similar numbers of amino acids, i.e., from about 723 to about 822; (4) similar masses, i.e., about 124 kdal. for the glycoprotein; (5) substantial interspecies (50%-60%) overall sequence homology with interspecies homologies increasing to about 56% to 99% in the cytoplasmic region of the protein, suggesting that they constitute a gene family (Nose, A., 1987; Miysysni, D., et al., 1989); and (6) a common mechanism of action, namely, homophilic binding of cadherins on one cell to similar cadherins on the adjoining cell.

CAMs independent of Ca²⁺ are also known, for example, the 125K glycoprotein of Urushihara et al. (Urushihara, H., et al., Cell, 20:363 (1980)); N-CAM (Rutishauser, U., Nature. Lond., 310:549 (1984));

Ng-CAM (Grunet, M. et al., Proc. Nat'l. Acad. Sci.

(USA). 81:7989 (1984)); LI (Rathjien, F.G. et al.,

J. , 3:1 (1984)); G4 (Rathjien, F.G. et al., J. Cell Biol., 104:343 (1987)); and platelet glycoprotein

PECAM-1 (CD 31) (Newman, P.J., Science, 247:1219

(1990)). Ca²⁺-independent CAMs are known to exhibit certain properties of the Ca²⁺-dependent CAMs. Thus, N-CAM and N-cadherin both promote retinal neurite outgrowth on astrocytes (Neugebauer, K.M., et al., J. Cell Biol., 107:1177 (1985)), and on Schwann cells (Bixby, J.L. et al., J. Cell Biol., 107:353 (1988)).

Monoclonal antibodies raised against epithelial E-type cadherins such as uvomorulin are known to disrupt the adhesion of several cell types, including embryo cells, cultured teratocarcinoma cells,

hepatocytes, and MDCK kidney epithelial cells (Ogou, S.-I., et al., J. Cell Biol., 97:944 (1983); Yoshida- Noro, et al., (1984), above; Shirayoshi, Y., et al., Cell Struct. Funct., 11:285 (1986); Gallin, et al., (1983), above; Vestweber, D., et al., EMBO J., 4:3393 (1985); Johnson, M.H., et al., J. Embrol. Exp.

Morphol., 93:239 (1986); Gumbiner, B., et al., J. Cell Biol., 102:457 (1986)). However, prior to the present discoveries

disclosed in the parent applications cadherins had not been found in brain capillary or other endothelial cells (see, Takeichi, et al. (1988), above). Further, the CAMs of microvascular endothelial cells had not yet been identified, nor had such molecules been localized specifically to brain capillary endothelial cells.

Thus, until the present invention no means were known for transiently and reversibly disrupting tight

junctions between microvascular endothelial cells, including those of the BBB, based upon an attack upon the CAM'S of such cells that are responsible for tight junction formation and maintenance.

It has been hypothesized that the cadherins contain a common cell adhesion recognition (CAR) sequence. The CAR sequences of several cell and substratum adhesion molecules are known. Martin, G.R., et al., Ann. Rev. Cell Biol., 3:57 (1987) ; Ruoslahti, E., et al., Science, 238:491 (1987). In general, CAR sequences are composed of at least three amino acid residues. The most rigorously investigated CAR

sequence is RGD which is found in laminin, fribronectin and other basement membrane components that are

responsible for the binding of cells to the substratum.

Blaschuk, et al., in a paper to be published subsequent to the filing of the present application (Blaschuk, O., et al., J. Mol. Biol., in press,

(1990)), disclose the presence of three potential cadherin CAR sequences in the first extracellular domains of liver CAM, E-, P-, and N-cadherin, namely, PPI, GAD and HAV. Blaschuk, et al. (Blaschuk, O., et al., Develop. Biol., 139:227 (1990)), also disclosed recently that synthetic peptides containing the HAV sequence inhibited two biological processes (compaction of 8-cell-stage mouse embryos and rate of neurite outgrowth on astrocytes) that are known to be mediated by cadherins. Effective peptides in these assays were LRAHAVDVNG and AHAVSE; PPI-containing peptides were without effect. However, Blaschuk et al. provide no guidance for determining the regions flanking the HAV tripeptide that are critical for cell-cell adhesion. In the BBB disrupting peptides of the present invention detailed below, we have observed that the mere presence of the HAV sequence in a small cadherin-derived peptide is not the sine gua non for a composition effective to prevent cell-cell adhesion. Indeed, it should be emphasized that neither Blaschuk et al. nor any other publication known to the present inventors suggest that cadherin sequences containing HAV or SHAVS sequences would be effective in opening tight junctions and piercing blood brain barriers formed by E-cadherins in brain microvascular endothelial cells.

SUMMARY OF THE INVENTION

It has now been discovered that molecules

homologous to, and immunologically related to, cadherin cell adhesion molecules are present on brain and non- brain microvascular endothelial cells, such that

junctions between such endothelial cells can be reversibly opened so as to permit passage of

therapeutic drugs by the use of polypeptide and antibody compositions that compete with such cell adhesion molecules for binding to such cells.

It is therefore an object of this invention to provide the identity of microvascular endothelial cell adhesion molecules.

Another object of this invention is to provide DNA sequences of genes, and plasmids containing same, coding for the expression of all or a cell-binding portion of microvascular endothelial cell adhesion molecules.

Yet another object of this invention is to provide means to identify those sequences of cell adhesion molecules responsible for the tight binding of

adjoining endothelial cells.

A further object is to provide therapeutic

compositions comprising polypeptides derived from cell adhesion molecules that reversibly disrupt cell-cell adhesion.

Still another object of this invention is to provide therapeutic compositions comprising polyclonal or monoclonal antibodies or fragments thereof directed against endothelial cell adhesion molecules, or against polypeptides representing cell binding regions thereof, that reversibly disrupt endothelial cell-cell adhesion.

Yet another object of this invention is to provide therapeutic formulations comprising therapeutic drugs conjugated with blood-brain barrier-disrupting

compositions of this invention, that are capable of entering the central nervous system following

disruption of the blood-brain barrier.

These and other objects of this invention will become clear by reference to the following description of the invention and to the appended claims.

DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates the partial cDNA sequence for bovine endothelial cell adhesion molecule homologous to chicken N-cadherin.

Figure 2 illustrates the partial cDNA sequence for bovine endothelial cell adhesion molecule homologous to mouse P-cadherin.

Figure 3 illustrates the cDNA sequence for the MDCK cell adhesion molecule homologous to mouse

E-cadherin.

Figure 4 illustrates the restriction sites in the bovine endothelial cell N- (4-1 to 4-5) and P-cadherin (4-6 to 4-8) cDNA sequences and in the MDCK E-cadherin (4-9 to 4-14) cDNA sequence.

Figure 5 shows the staining of a mouse brain thin section by an antibody raised against a fusion protein derived from amino acids 9-96 of MDCK E-cadherin containing an HAV region.

Figure 6 is a repeat of the experiment of Fig. 5, except that the antibody was raised against the entire E-cadherin molecule.

Figure 7 illustrates the effects of an 18-mer HAV- containing polypeptide on the resistance of tight junction monolayers of MDCK epithelial cells.

Figure 8 illustrates the effects of 11-mer and 18-mer HAV-containing polypeptides on the resistance of tight junction monolayers MDCK epithelial cells.

Figure 9 illustrates the effects of 11-mer and 18- mer HAV-containing polypeptides on the resistance of tight-junction monolayers of brain microvascular endothelial cells. DETAILED DESCRIPTION OF THE INVENTION

It has now been discovered that cell adhesion molecules with characteristics of cadherins are present on the surfaces of brain capillary endothelial cells and of microvascular endothelial cells of non-brain origins. The present invention is based on the

discovery that a polypeptide composition comprising cell binding domains of endothelial cell adhesion molecules may compete against such molecules for binding to such cells, such that by this means the junctions between such cells could be reversibly opened, thereby permitting penetration by therapeutic agents. The present invention also discloses that polyclonal or monoclonal antibodies (or fragments thereof) raised against endothelial cell adhesion molecules or cell-binding domains thereof may also compete for endothelial cell surface binding sites, and, by this means, reversibly disrupt junctions between endothelial cells, thereby permitting entry into the central nervous system of therapeutic agents.

In order to obtain compositions useful for

disrupting tight junctions between microvascular endothelial cells, the cell adhesion molecules

responsible for such junctions were identified.

The endothelial cell cadherins disclosed herein exhibit one or more of several characteristics of E-, P- and N- cadherins, including: characteristics of a transmembrane integral protein, with cytoplasmic, hydrophobic plasma membrane, and extracellular regions; intraspecies DNA sequence homologies of greater than about 50% for the entire molecule; immunological cross- reactivity with antibodies raised against non- endothelial cell cadherins; and containing cell-binding domains. "Immunologically related to" means that these cadherin-like molecules cross-react with antibodies raised against non-endothelial cell cadherins.

E-cadherin-like molecules were localized in brain by immunofluorescence. Cryostat sections of mouse brain were labeled with a rabbit antibody prepared against E-cadherin, and then with fluorescein

isothiocyanate-conjugated goat anti-rabbit

immunoglobulin. There is clear labeling of a capillary in brain sections as shown by immunofluorescence microscopy. Endothelial cells in liver and kidney were not stained by this procedure.

cDNAs coding for the expression of bovine

microvascular endothelial cell (BMEC) cadherins were cloned and sequenced as described below, and the partial sequence of N-cadherin and P-cadherin are disclosed herein in Figures 1 and 2, respectively. In addition, as MDCK dog kidney epithelial cells are known to employ E-cadherin to form high resistance tight junctions, and as the present invention discloses that brain capillary endothelial cell adhesion molecules include E-type cadherin, the DNA of this cadherin was also cloned; its complete DNA sequence is disclosed herein (Fig. 3).

N-, P- and E-cadherin-type clones described herein were deposited in the American Type Culture Collection on September 26, 1989, and were assigned the following accession numbers:

Clone Designation Accession No.

N-cadherin-type clones

pUC19-bNCad 10A 40667 pUC19-bNCad 39A 40669

P-cadherin-type clones

pUC18-bPCad 3B-10 40668 pUC19-bPCad 9B 40670

E-cadherin-type clones

pBluescript MDCKECad 45-30E 40671

The cloning of cadherins was accomplished by taking advantage of the fact that the cadherins

characterized thus far are transmembrane glycoproteins, the cytoplasmic domains of which are highly conserved, that is, are highly homologous.

Two degenerate oligonucleotides flanking the

42-amino acid coding region in the cytoplasmic domain were selected to serve as primers for polymerase chain reaction (PCR) using either BMEC cDNA or MDCK cDNA as templates. The PCR reactions were carried out

essentially according to Saiki, R. K. et al., Science, 239:487 (1988), which is incorporated herein by

reference.

The cloned PCR products from each cell type were sequenced essentially according to the method of

Sanger, F. et al., Proc. Nat'l. Acad. Sci. (USA),

74:5463 (1977), which is incorporated herein by

reference.

It was discovered that BMEC cadherins are of two types - one homologous to chicken N-cadherin (neuronal type, see, e.g., Hatta, K., et al., J. Cell Biol., 106:873 (1988)) and the other homologous to mouse

P-cadherin (placental type, see e.g., Nose, A., et al., (1987) above). It has also been found that there are two species of cadherins in MDCK cells - one homologous to mouse E-cadherin (see, e.g., Nagafuchi, A., et al., Nature, 329:341 (1987)) and the other homologous to mouse P-cadherin (Nose, et al. (1987), above).

The PCR products were then used as probes to isolate the BMEC and MDCK cadherin cDNA clones as follows. A cDNA library was constructed essentially according to Gubler et al . (Gubler, U. et al., Gene, 25:263 (1983), which is incorporated herein by

reference), using poly (A)⁺RNA isolated from either BMEC or MDCK cells. The cDNA was ligated via EcoRI adaptors into gt10 arms (BMEC) or ZAP^R (from

Stratagene, Inc., La Jolla, CA) vector arms (MDCK). cDNA libraries containing 5 x 10⁵ - 1.5 x 10⁶

independent cDNA clones were screened using

radiolabeled PCR products (Benton, W.D. et al.,

Science. 196:180 (1987), which is incorporated herein by reference). Northern blot analysis (Maniatis, T. et al., "Molecular Cloning: A Laboratory Manual", Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., 1982) may be used to determine whether each cDNA species cloned hybridizes to a single mRNA species, as well as the tissue distributions of each cDNA species. cDNA clones for each cadherin were sequenced by the method of Sanger et al. (1977) above.

The partial restriction maps for each cDNA clone based on their sequences are shown in Fig. 4. Some of these restriction sites were confirmed by restriction enzyme digestions, including Hind III, Pst I, Kpn I, Bgl II for N-cadherin; Pvu II, Sac I and Pst I for P-cadherin; Pst I, Pvu II, BamH I, and Sac I for

E-cadherin.

In order to test whether the cloned E-cadherin cDNA contains all the information necessary for cadherin function, full-length E-cadherin cDNA joined to a suitable promoter may be introduced into mouse L-cells that have very little endogenous cadherin activity (Nagafuchi, et al. (1987), supra). To test for expression of E-cadherin in transfectants derived from the introduced cDNA, transfected L-cells may be tested for Ca²⁺-dependent aggregating activity. The extent of this aggregating activity should be closely correlated with the amount of E-cadherin expressed (Takeichi, M. (1988), supra). This same technique may be used for testing cDNAs encoding bovine endothelial N- and P-cadherins, according to the method of Hatta, et al. (Hatta, K., et al. (1988), supra).

In order to identify cell binding domains in, for example, MDCK E-type cadherin, L-cells may be first transfected as above with a cDNA of a size sufficient to cause Ca²⁺-mediated aggregation of transfectants. A series of deletion mutants comprising truncated cDNA species missing different regions of the extracellular domain may be prepared by restriction enzyme digestion and proper end filling or exonuclease digestion to make the deletions in the proper coding frames. These deletion mutants can then be tested for their ability to express in L-cells a protein causing Ca²⁺-dependent aggregation. By correlating a loss of aggregation with deletion of particular fragments, the regions important for cell binding may be determined. A variety of polypeptides corresponding to binding regions of cadherins, as deduced from the nucleotide sequences of deleted cDNA, may be synthesized chemically using an automated peptide synthesizer such as that of Applied Biosystems, Inc., Foster City, CA, or expressed by recombinant DNA methods. Effective polypeptides may be of varying lengths, depending upon the natures of junctions being disrupted and the cell adhesion

molecule present. Nucleotide, and corresponding amino acid,

sequences of cadherins may be analyzed to detect homologous regions. Applying this technique to bovine endothelial cell N- and P-cadherins and to epithelial cell E-cadherin, we have determined that, in the amino acid 80 region of each of these cadherins, there is conserved a triplet HAV (His-Ala-Val) region. We have deduced that this HAV region may be a common cell adhesions recognition (CAR) sequence.

We have chemically synthesized the following polypeptides, each of which containing the HAV

sequence:

6-mer (78-83 ) NH₂-SHAVSS-CONH₂

11-mer (76-86) NH₂-LYSHAVSSNGN-CONH₂

17-mer (74-90) NH₂-YILYSHAVSSNGNAVED-CONH₂

18 mer ( 69-86) NH₂-EQIAKYILYSHAVSSNGN-CONH₂

20-mer(71-90) NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ and have tested each for efficacy in opening brain endothelial cell tight junctions in the BBB model disclosed in copending United States application Serial

No. 07/413,274, and also on kidney epithelial cell tight jucntions..

Polyclonal antibodies raised in rabbits and monoclonal antibodies derived from hybridomas may be generated against each of the chemically-synthesized polypeptides by standard methods. (Harlow, E., et al.,

"Antibodies: A Laboratory Manual", Cold Spring Harbor

Laboratory Press, Cold Spring Harbor, N.Y., 1988;

Goding, J.W., "Monoclonal Antibodies: Principles and Practice", Academic Press, N.Y. 1986). In addition, recombinant antibodies may be prepared. Fragments of antibodies, e.g., Fc, Fab, F(ab)', may be prepared by standard methods.

We have cloned and sequenced fusion proteins derived from amino acids 9-96 of MDCK E-cadherin containing the HAV region. A polyclonal antibody prepared against this fusion protein stained rat

(Fig.55) mouse brain sections as well as did an

antibody raised against the entire E-cadherin (Fig. 6). A polyclonal antibody raised against a fusion protein derived from amino acids 9-37 failed to stain brain sections. These results indicate that the key cell- binding domain of E-cadherin lies in the region of amino acids 37-96.

The ability of CAM-derived polypeptides containing cell-binding domains, and the corresponding polyclonal and monoclonal antibodies, of the invention to disrupt tight junctions may be tested in in vitro and in vivo models of high resistance tight junctions and in animal models. Monolayers of MDCK dog kidney epithelial cells, that are known to contain high resistance tight junctions (Gumbiner, B., J. Cell Biol., 102:457

(1986)), can be used to test for the ability of the polypeptides and corresponding antibodies of the present invention to disrupt such tight junctions.

Polyclonal antibodies prepared as described above may also be used in conjunction with Western blotting (Old, R.W., et al., Principles of Gene Manipulation, 3d ed., Blackwell, Oxford, 1985, p. 10) and a variety of tissue extracts in order to identify cell adhesion glycoproteins in such extracts.

Another embodiment of the present invention is in drug delivery systems. Conjugates between therapeutic drugs and agents that affect cell adhesion molecule function in brain capillary endothelial cells may be used to deliver therapeutic drugs to the CNS. For example, a polypeptide derived from a cell adhesion molecule that contains within its amino acid sequence a cell-binding domain, or antibodies thereto, may be conjugated in biologically-active form to a therapeutic modality. Such conjugates may have the dual effect of opening the BBB and delivering the therapeutic agent to the brain side of the BBB. Delivery of therapeutic drugs to the CNS, either alone or conjugated to agents that disrupt cell-cell adhesion, may be accomplished by administering such drugs to a subject either

simultaneously with or subsequent to the administration of the agents of this invention that disrupt the tight junctions of the BBB. Examples of therapeutic

modalities that may be delivered to the brain by the cell adhesion disruption compositions of this invention include Nerve Growth Factor, anti-Parkinsonian drugs, and brain enzymes known to be missing in

sphingolipidoses, e.g., Tay-Sachs disease. Means of chemically conjugating protein or polypeptide carriers to therapeutic agents such that the biological

integrity of the therapeutic agent is not compromised and such that the therapeutic agent is readily cleaved from the carrier by enzymes present on or within endothelial cells (e.g., amidases, esterases,

disulfide-cleaving enzymes), are well known in the art. It is also apparent that these therapeutic conjugates may be delivered to endothelial cells in encapsulated form (e.g., in liposomes) or as microsuspensions stabilized by pharmacological excipients.

It is known (Jain, R.K., J. Natn'l Cancer Inst., 81:570 (1989)) that many solid tumors develop internal barriers, including high pressure zones and collapsed blood vessels, that make it difficult for blood-borne chemotherapeutic agents to reach the tumor's inner core. The barrier problem is particularly troublesome with therapeutic products drawn from the human immune system, such as monoclonal antibodies conjugated with chemotherapeutic agents, interleukin-2, interferon and activated killer T-lymphocytes, because of their large size. Thus, in another embodiment of this invention, compositions that disrupt the junctions between

endothelial cells, particularly the relatively small peptides that contain one or more cell-binding regions of cell adhesion macromolecules, may be used to enhance drug delivery to tumors with depressed blood flow.

It has been theorized that cancer cells

metastasize by secreting soluble cadherins variously to open tight junctions in cells that block their movement and to prevent their being bound to such cells. We consider it likely that antibodies raised against these cadherins, which are derived from extracellular domains of the cadherins disclosed in this invention, may provide a therapeutic modality that inhibits or

prevents cancer cell metastases.

In another embodiment, the compositions of this invention may also be used to provide penetration for chemotherapeutic agents of other well-known blood- tissue barriers, such as blood-testis barriers and blood-retina barriers. The latter barrier is known to prevent the efficient transport of, for example, administered antibiotics to the retina from the general circulation. The cell adhesion disrupting compositions of this invention may, thus, be used in conjunction with the administration of antibiotics to treat retinal infections.

The following examples are illustrative of several embodiments of this invention, and should not be construed in any way as limiting the invention as recited in the claims.

EXAMPLE 1

EFFECTS OF HAV-CONTAINING POLYPEPTIDES

ON TIGHT JUNCTIONS OF MDCK EPITHELIAL

AND BOVINE ENDOTHELIAL CELLS The BBB model of copending U.S. Serial No.

07/413,332 was used to examine the effects of

polypeptides containing the HAV region on the tight junctions of monolayers of MDCK epithelial cells and bovine capillary endothelial cells as determined by resistance measurements across the monolayers.

The polypeptide was added to the cells either from the apical side (top) or basolateral side (bottom), as shown in the following sketch.

APICAL

EPITHELIAL CELLS ENDOTHELIAL CELLS

Gut Side Blood Side

Blood Side Brain Side

BASOLATERAL Figure 7 illustrates the effects of various concentrations of the aforementioned 18-mer polypeptide on resistance of MDCK epithelial cells. At the lowest concentration tested, 0.5 mg/ml, resistance was

markedly decreased. The polypeptide was more effective when added from the basolateral side, but at high concentrations was quite effective even when added from the apical side. These data indicate that the 18-mer is effective in making tight junctions permeable. The 20-mer was similarly effective, and a 17-mer less effective.

Figure 8 illustrates the effects of the

aforementioned 11-mer and 18-mer on MDCK cell

resistance when added from either the apical or basolateral side of the monolayers. The concentration of polypeptide was about 1 mg/ml. The 11-mer (as well as the 6-mer data not shown) was virtually without effect. With the 18-mer, resistance was almost totally abolished by about 6 hours, indicating disruption of tight junctions. That the effect of the 18-mer is reversible is indicated by the "wash-out" experiment. When the 18-mer was washed out of the MDCK cells at 6 hours, resistance recovered to a substantial extent over the next 21 hours. This recovery was particularly pronounced when the 18-mer had originally been added from the basolateral side of the monolayers. The

20-mer produced results similar to those of the 18-mer, and the 17-mer was effective, but somewhat less so.

Figure 9 illustrates the effect of 1 mg/ml of the 11-mer and 18-mer on high resistance monolayer cultures of brain endothelial cells (see copending United States Serial No. 07/413,332 for method of preparation). As with MDCK cells, the 11-mer (and the 6-mer) failed to reduce resistance values over a 48-hour period of observation. In contrast, the 18-mer (as well as the 20-mer) decreased resistance values markedly when added from either the basolateral or apical side, but the effect of the polypeptide was more rapid and more pronounced when it was added from the basolateral side; the 17-mer was less effective.

The conclusion of these experiments is that a particular set of peptides (but not all peptides) centered around the HAV region of E-cadherin are effective in opening tight junctions of brain

endothelial cell blood-brain barriers, and also of epithelial cells that form such junctions ("gut

barrier"). Both the length and compositon of the amino acid region flanking the HAV triplet thus appear to play a role in the efficacy of such compositions.

While the aforementioned embodiments represent the preferred embodiments of the invention, those skilled in the art may, without undue experimentation, devise other executions of the compositions and methods of use of this invention without departing from the concept and spirit inherent therein.

Claims

What is claimed is:

1. A composition for opening tight

junctions between microvascular endothelial cells of a subject, whereby means are provided for a drug to cross the permeability barrier imposed by such junctions, comprising an agent capable of reacting with at least one type of cell-bound cell adhesion molecule that would otherwise mediate tight junction formation between microvascular endothelial cells, so that cell- cell adhesion is disrupted.

2. A composition of claim 1, wherein said cell adhesion molecule exhibits at least about 50% sequence homology with a cadherin selected from the group consisting of E-cadherin, N-cadherin and

P-cadherin.

3. A composition of claim 1, wherein said cell adhesion molecule is immunologically related to at least one of the group consisting of E-cadherin,

N-cadherin and P-cadherin.

4. A composition of claim 1, wherein the microvascular endothelial cells are brain capillary endothelial cells.

5. A composition of claim 2, wherein said agent comprises an inhibitor of the binding to cells of said cell adhesion molecule.

6. A composition of claim 3, wherein said agent comprises an inhibitor of the binding to cells of said cell adhesion molecule.

7. A composition of claim 5, wherein said inhibitor agent comprises a fragment of said cell adhesion molecule.

8. A composition of claim 7, wherein said cell adhesion molecule fragment includes within its amino acid sequence a cell-binding domain.

9. A composition of claim 8, wherein said cell-binding domain contains an HAV amino acid

sequence.

10. A composition of claim 9, wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

11. A composition of claim 9, wherein said amino acid sequence is

NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

12. A composition of claim 9, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

13. A composition of claim 9, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

14. A composition of claim 5, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against said cell adhesion molecule.

15. A composition of claim 5, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against a fragment of said cell adhesion molecule.

16. A composition of claim 15, wherein said cell adhesion molecule fragment includes within its amino acid sequence a cell-binding domain.

17. A composition of claim 16, wherein said cell-binding domain contains an HAV amino acid

sequence.

18. A composition of claim 17, wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

19. A composition of claim 17, wherein said amino acid sequence is

NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

20. A composition of claim 17, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

21. A composition of claim 17, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

22. A composition of claim 5 or 6 in a pharmaceutically-acceptable vehicle.

23. A method for opening tight junctions between microvascular endothelial cells of a subject, comprising the step of administering to the subject an agent, in an effective amount and in a

pharmaceutically-acceptable vehicle, capable of reacting with at least one type of cell-bound cell adhesion molecule that would otherwise mediate tight junction formation between microvascular endothelial cells, so that cell-cell adhesion is disrupted and whereby means are provided for a drug to cross

permeability barriers imposed by such tight junctions.

24. A method of claim 23, wherein said cell adhesion molecule exhibits at least about 50% homology with a cadherin selected from the group consisting of E-cadherin, N-cadherin and P-cadherin.

25. A method of claim 23, wherein said cell adhesion molecule is immunologically related to at least one of the group consisting of E-cadherin,

N-cadherin and P-cadherin.

26. A method of claim 23, wherein the microvascular endothelial cells are brain capillary endothelial cells.

27. A method of anyone of claims 23-25, inclusive, wherein said agent comprises an inhibitor of the binding to cells of said cell adhesion molecule.

28. A method of claim 27, wherein said inhibitor agent comprises a fragment of said cell adhesion molecule.

29. A method of claim 28, wherein said cell adhesion molecule fragment includes within its amino acid sequence a cell-binding domain.

30. A method of claim 29, wherein said cell- binding domain contains an HAV amino acid sequence.

31. A method of claim 30 wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

32. A method of claim 30, wherein said amino acid sequence is

NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

33. A method of claim 30, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

34. A method of claim 30, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

35. A method of claim 27, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against said cell adhesion molecule.

36. A method of claim 28, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against said fragment of said cell adhesion molecule.

37. A method of claim 36, wherein said cell adhesion fragment includes within its amino acid sequence a cell-binding domain.

38. A method of claim 37 wherein said cell- binding domain contains an HAV amino acid sequence.

39. A method of claim 38, wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

40. A method of claim 38, wherein said amino acid sequence is

NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

41. A method of claim 38, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

42. A method of claim 38, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

43. A drug delivery composition comprising a conjugate between a therapeutic drug and an agent capable of reacting with at least one type of a cell- bound cell adhesion molecule that would otherwise mediate tight junction formation between microvascular endothelial cells, so that cell-cell adhesion is disrupted by said agent, whereby means are provided for said drug to cross permeability barriers imposed by such tight junctions, in a pharmaceutically-acceptable vehicle.

44. A drug delivery composition of claim 43, wherein said cell adhesion molecule exhibits at least about 50% homology with a cadherin selected from the group consisting of E-cadherin, N-cadherin and

P-cadherin.

45. A drug delivery composition of claim 43, wherein said cell adhesion molecule is immunologically related to at least one of the group consisting of E-cadherin, N-cadherin and P-cadherin.

46. A drug delivery composition of claim 43, wherein the microvascular endothelial cells are brain capillary endothelial cells.

47. A drug delivery composition of any one of claims 43-45, inclusive, wherein said agent

comprises an inhibitor of the binding to cells of said cell adhesion molecule.

48. A drug delivery composition of claim 47, wherein said agent comprises a fragment of said cell adhesion molecule.

49. A drug delivery composition of claim 48, wherein said cell adhesion molecule fragment includes within its amino acid sequence a cell-binding domain.

50. A drug delivery composition of claim 49, wherein said cell-binding domain contains an HAV amino acid sequence.

51. A drug delivery composition of claim 50, wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

52. A drug delivery composition of claim 50, wherein said amino acid sequence is

NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

53. A drug delivery composition of claim 50, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

54. A drug delivery composition of claim 50, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

55. A drug delivery composition of claim 43, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against said cell adhesion molecule.

56. A drug delivery composition of claim 43, wherein said inhibitor agent comprises a polyclonal or monoclonal antibody directed against a fragment of said cell adhesion molecule.

57. A drug delivery composition of claim 56, wherein said cell adhesion molecule fragment contains within its amino acid sequence a cell-binding domain.

58. A drug delivery composition of claim 56 , wherein said cell-binding domain encompasses an HAV amino acid sequence

59. A drug delivery composition of claim 58, wherein said amino acid sequence is

NH₂-YILYSHAVSSNGNAVED-CONH₂ .

60. A drug delivery composition of claim 58, wherein said amino acid sequence is NH2-EQIAKYILYSHAVSSNGN-COHN₂ .

61. A drug delivery composition of claim 58, wherein said amino acid sequence is

NH₂-IAKYILYSHAVSSNGNAVED-CONH₂ .

62. A drug delivery composition of claim 58, wherein said amino acid sequence comprises amino acids 9-96 of E-cadherin.

63. A drug delivery composition of claim 43, wherein said conjugate comprises a physiologically- cleavable covalent bond.

64. A drug delivery composition of claim 43, wherein said conjugate is encapsulated within a

physiologically-compatible particle.

65. A drug delivery composition of claim 64, wherein said particle comprises a liposome.