WO2001009368A1

WO2001009368A1 - Follistatin antagonists

Info

Publication number: WO2001009368A1
Application number: PCT/US2000/020444
Authority: WO
Inventors: Henry T. Keutmann; Alan Schneyer; Yisrael Sidis; Patrick Sluss
Original assignee: The General Hospital Corporation
Priority date: 1999-07-30
Filing date: 2000-07-27
Publication date: 2001-02-08

Abstract

Follistatin antagonists are disclosed. The antagonists are polypeptides that include a non-activin-binding FS N-Domain or non-activin-binding FLRG N-Domain; and a heparin sulfate-binding moiety; wherein the tryptophan residue (W) at position 4 or 36 of the non-activin-binding N-Domain is substituted by a non-aromatic amino acid such as alamine. An example of a heparin sulfate-binding moiety is Domains I-III of human follistatin.

Description

FOLLISTATIN ANTAGONISTS

Field of the Invention The field of the invention is biochemistry, cell biology, physiology, endocrinology and oncology.

Background of the Invention Follistatin (FS) is a monomeric- glycoprotein whose only known function is to bind and neutralize growth factors, primarily activin. Activins are members of the TGF-jS superfamily of growth and differentiation factors.

Human FS exists naturally in several isoforms. Cloning and characterization of the human FS gene revealed that alternative splicing results in two mRNA transcripts encoding FS polypeptides of 288 (FS288) and 315 amino acids (FS315) , following signal cleavage (Shimasaki et al . , 1988, Proc. Natl . Acad. Sci . USA 85:4218-4222). Proteolytic processing yields a third FS polypeptide, which contains 303 amino acids (FS303) . Due to these differences in polypeptide length and differences in posttranslational glycosylation, the molecular weight of FS ranges from about 32 kD to about 44 kD.

After removal of a 29-amino acid signal sequence (leader) , mature FS contains 5 distinct structural domains, each encoded by a different exon: an N-Domain of 63 amino acids; FS Domains I, II, and III, which are three 73-75- amino acid repeats of the "follistatin module" (Hohenester et al., 1997, EMBO J. 16:3778-3786; Eib et al . , 1996, J. Neurochem. 67:1047-1055; Denzer et al . , 1998, EMBO J. 17:335-343); and, in FS315, a highly acidic C-Domain of 27 amino acids. A notable structural feature of follistatin is the large number of cysteine residues in the polypeptide. FS is expressed in a variety of mammalian tissues, including ovary, testis, prostate, kidney, muscle, uterus, brain, pancreas and pituitary. By virtue of a short, highly basic region in FS Domain I, FS binds heparin sulfated proteoglycans exposed on cell membranes. This results in localized concentration of FS at the cell surface. This cell surface-associated FS retains its activin-binding ability and forms a barrier, which impedes the access of extracellular activin to activin receptors in the cell membrane (Delbaere et al . , 1999, Endocrinology 140:2463- 2470)

Activins are dimeric proteins that function as potent, local regulators of cell growth (Vale et al . , "Reproductive and other roles of inhibins and activins," In: Knobil and Neill (Eds.) The Physiology of Reproduction, Raven Press, New York, pp 1861-78) . Activin action is tissue-dependent. For example, activin stimulates cell growth in ovarian tissues. In contrast, activin inhibits basal and androgen-stimulated proliferation, and induce apoptosis, in human prostatic cancer cells (Wang et al . , 1996, Endocrinology 137:5476-5483) .

FS binds to (and neutralizes) activin in vivo and in vitro. This binding is essentially irreversible under physiological conditions, resulting in biologically inactive FS-Act complexes (Nakamura et al., 1990, Science 247:836- 838; Kogawa et al . , 1991, Endocrinology 128 : 1434-1440; Schneyer et al . , 1994, Endocrinology 135 : 667-674. By binding activin, FS limits the activity of activin in a given tissue, or limits the range of tissues in which activin is active, or both. All known isoforms of FS display similar activin binding characteristics (de Winter et al., 1996, Mol . Cell . Endocrinol . 116:105-114; Inouye et al., 1991, Endocrinology 129:815-822) .

Summary of the Invention

We have discovered that when either of two specific amino acids in human FS is mutated, activin binding by FS is eliminated. Based on this discovery, the invention features an FS antagonist. The FS antagonist includes amino acids 3- 63 of a non-activin-binding FS N-Domain (amino acids 3-63 of SEQ ID NO:l), or amino acids 3-61 of a non-activin-binding FLRG N-Domain (amino acids 3-61 of SEQ ID NO: 2) ; and a heparin sulfate-binding moiety; wherein the tryptophan residue (W) at position 4 or 36 of the non-activin-binding N-Domain (from FS or FLRG) is substituted by A, G, L, I, V, P, S, T, M, C, N, Q, D, E, K, R, Y or H. In preferred embodiments, the non-activin-binding N-Domain is located N-terminally, relative to the heparin sulfate-binding moiety. Preferably, the heparin sulfate-binding moiety contains the "heparin sulfate site" of native FS (amino acids 74-87 of SEQ ID NO:l) . The heparin sulfate-binding moiety can include amino acids 64-136 of SEQ ID NO:l, 64-211 of SEQ ID NO:l, or amino acids 64-288 of SEQ ID NO:l. Preferably, the tryptophan residue (W) at position 4 or 36 of the non-activin-binding N-Domain from FS or FLRG is substituted by A, G, S, T, N, Q, R, D or E. More preferably, it is substituted by A or G, and most preferably, A.

The invention also features a nucleic acid encoding the above-described FS antagonist, and a vector containing that nucleic acid. The vector can include expression control sequences operatively linked to the nucleic acid encoding the FS antagonist. The invention also includes a host cell transformed with the vector.

The invention also includes a method of increasing activin activity in an activin-sensitive tissue in a mammal. The method includes contacting the cells of the activin- sensitive tissue with the FS antagonist. Preferably, contacting the cells of the activin-sensitive tissue with the FS antagonist is accomplished by expressing the FS antagonist in vivo from a vector containing a nucleic acid encoding the FS antagonist polypeptide. This method can be used where the activin-sensitive tissue is prostate tissue.

The preferred methods and materials are described below in examples which are meant to illustrate, not limit, the invention. Those of skill in the art will recognize methods and materials that are similar or equivalent to those described herein, and that can be used in the practice or testing of the present invention.

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

Other features and advantages of the invention will be apparent from the detailed description, and from the claims . Brief Description of the Drawings Fig. 1 is the amino acid sequence of FS288 (SEQ ID NO:l) . The signal (leader) sequence, the N-Domain, Domains I-III, and the C-Domain are indicated. Domains I-III are shown with the 10 cysteine residues in each domain aligned. The heparin sulfate site is indicated in Domain I, and the epitopes recognized by three different monoclonal antibodies are indicated.

Figs. 2A-2C are a histograms summarizing data on activin binding, heparin binding, and 7FS30 binding, respectively, by synthetic peptides corresponding to different segments of human FS.

Fig. 3 is a histogram summarizing data on activin binding by synthetic peptides FS1-26 and FS44-59 each alone, and each in competition with native follistatin FS288, or native follistatin FS315.

Fig. 4A is a line graph showing dose-response curves for activin binding by synthetic peptides FS1-26, FS3-26, FS6-26, FS14-26 and FS10-26. Fig. 4B is a line graph showing done-response curves for activin binding by a series of synthetic peptides corresponding to FS amino acids 44-59, 47-59, 53-59, 49-59, 51-59 and 44-59 scrambled (negative control) .

Fig. 5 is a histogram summarizing data on activin binding by a series of mutant FS288 proteins containing specific, site-directed mutations. Activin binding is expressed as % of control, with native FS288 being the positive control. Non-conditioned medium is the negative control. W4A = tryptophan 4 changed to alanine; R6A = arginine 6 changed to alanine; K9A = lysine 9 changed to alanine; KW(48,49)AA = lysine 48 and tryptophan 49 both changed to alanine; M50E = methionine 50 changed to glutamate . Fig. 6 is a sequence comparison showing the amino acid sequence of the N-Domain of FS (amino acids 1-63 of SEQ ID NO:l) and the N-Domain of FLRG (SEQ ID NO: 2) .

Detailed Description In a preferred embodiment of the invention, the mature FS antagonist is FS288 (amino acids 1-288 of SEQ ID NO:l) in which the critical tryptophan residues at positions 4 and 36 are replaced by alanine residues. Replacement of either tryptophan residue reduces activin binding by more than 90%, and results in a useful antagonist. Replacement of both further reduces activin binding. In this embodiment, the FS antagonist efficiently binds heparin sulfate by virtue of a basic (positively charged) region of the FS Domain I called the "heparin sulfate site" (amino acids 74-87 of SEQ ID NO:l) . The acidic (negatively charged) FS C-Domain (amino acids 289-315 of SEQ ID NO:l) is not included in the FS antagonist polypeptide, because if present, it would undergo intramolecular binding to the heparin sulfate site. This would prevent the FS antagonist from binding to heparin sulfate exposed on cell membranes, thereby impairing the in vivo effectiveness of the FS antagonist.

The above-described FS antagonist can be isolated from a recombinant expression system, or chemically synthesized, and then delivered, as an exogenous polypeptide, to cells in a target tissue. Preferably, however, the FS antagonist is expressed in vivo from a suitable vector introduced into the cells of a target tissue. Following a single gene delivery treatment, in vivo expression provides a continuous supply of the FS antagonist in intimate contact with the target cells (cells of the target tissue) , for a period of several weeks, or indefinitely, depending on the type of vector employed. This is important because optimal effectiveness of the FS antagonist occurs when it has displaced endogenous FS from the heparin sulfate sites on the target cell surface. There is a need for newly- supplied FS antagonist to replace cell surface-associated FS antagonist that is continually lost due to natural biochemical turnover of molecules in and on the target cells. The above-described FS antagonist is expressed initially with a signal sequence (amino acids (-)29-(-)l of SEQ ID NO:l) at its N-terminus. The signal sequence causes the FS antagonist polypeptide to be secreted from the cell. The cell cleaves the signal sequence during the secretion process, thereby yielding the mature form of the FS antagonist. As the mature FS antagonist emerges at the cell surface, it is in proximity to the heparin sulfate groups on the cell surface, and binding of the FS antagonist to the heparin sulfate groups occurs readily (if binding has not already occurred) . When a heparin sulfate group is already associated with an native FS molecule, the native FS can be displaced by the FS antagonist through competitive binding. The extent to which such displacement occurs depends on the relative amounts of native FS and FS antagonist. To promote displacement of cell -associated, native FS by the FS antagonist, the local concentration of FS antagonist is increased, e.g., by use of strong, constitutive promoter to drive in vivo expression of the FS antagonist from an FS antagonist transgene. In the above-described embodiment, the only differences between the mature FS antagonist and native human FS288 is the mutation of tryptophan residues 4 and 36 to alanine residues. Thus, the FS antagonist displays a structure extremely similar to natural, native FS. Because of such similarity, the cellular machinery treats the FS antagonist as if it were native FS, a molecule naturally present in the cell . Consequently, the FS antagonist causes minimal disruption to normal cellular processes, i.e., the FS antagonist's side effects are minimized. In addition, because the cell treats the FS antagonist as if it were native FS, the antagonist undergoes any FS-specific transport or localization that takes place in the target cell or target tissue. This is advantageous, because to the extent that any localization of native FS takes place, the FS antagonist is similarly localized so as to maximize its competition with the native FS.

When a critical tryptophan residue (position 4 or 36 of FS Domain I) is replaced to abolish activin binding, the substituted amino acid is not phenylalanine. Replacement of a critical tryptophan residue by phenylalanine does not substantially diminish activin binding. Substitute amino acids with small, non-hydrophobic chains are preferred, e.g., alanine and glycine.

Those of skill in the art will recognize that when the FS antagonist is expressed initially with a signal sequence, it is not necessary to use the signal sequence of natural, human FS (amino acids (-)29-(-)l of SEQ ID NO:l). Other suitable, eukaryotic signal sequences are known in the art and can be employed in the FS antagonist.

Some embodiments of the FS antagonist incorporate a mutated N-Domain of FLRG instead of a mutated N-Domain of FS. In the N-Domain of FLRG (SEQ ID NO:2), the tryptophan at position 4 or position 36 (or both) is mutated to a non- aromatic amino acid, e.g., alanine, just as in a mutated FS N-Domain. In addition to replacement of the critical tryptophan residues, the FS antagonist can include one or more amino acid substitutions in the non-activin-binding FS N-Domain (SEQ ID NO:l), or the non-activin-binding FLRG N- Domain. Preferably, such substitutions are conservative amino acid substitutions.

If the C-terminal portion of the FS antagonist includes FS Domains I, II or III, those domains can include one or more amino acid substitutions. Preferably such substitutions: (1) do not involve the heparin sulfate site; (2) do not involve cysteine residues; (3) do not create highly acidic regions likely to interfere with the heparin sulfate-binding function of the heparin sulfate site; and (4) are conservative amino substitutions. The complete nucleotide sequence of human FS gene exons 1-6 is found in Shimasaki et al . , 1988, Proc. Natl . Acad. Sci . USA 85:4218-4222; and in GenBank (Accession Nos. M19480 and M19481) . Using this sequence information with commercially available PCR technology, DNA encoding human FS288, including the FS signal sequence can be prepared routinely and reliably, without undue experimentation.

In an exemplary procedure, total RNA is isolated from a sample of a human tissue type that expresses FS, e.g., ovarian tissue. The RNA is then used for synthesis of cDNA. Using primers based on the GenBank FS sequence, the desired coding sequence is amplified by standard PCR techniques. Well-known, PCR-based, site-directed mutagenesis techniques are applied to the FS-encoding DNA to obtain a DNA encoding FS288 in which tryptophan residue 4 and/or tryptophan residue 36 are replaced with a suitable amino acid, e.g., alanine. This yields a complete FS antagonist coding sequence. The FS antagonist coding sequence is then inserted into an expression vector suitable for use in mammalian cells. Suitable expression vectors are commercially available. Those of skill in the art will appreciate that alternative methods of reliably obtaining the DNA of the invention are known and can be used. For recombinant DNA techniques, see generally,

Sambrook et al . , Molecular Cloning - A Laboratory Manual , Cold Spring Harbor Press (1989); Ausubel et al . , (Eds.) Current Protocols in Molecular Biology, John Wiley & Sons, Inc. (1994) . PCR equipment and reagents are commercially available. For PCR techniques, see generally, H.A. Erlich, PCR Technology - Principles and Applications for DNA Amplification, Stockton Press, New York (1989) ; and PCJ? Protocols - A Guide to Methods and Applications, Innis et al . (Eds.), Academic Press, San Diego, CA (1990). The design and construction of suitable expression vectors is within ordinary skill in the art.

Vectors suitable for delivery of DNA encoding an FS antagonist to a target tissue, e.g., the prostate are known in the art. A viral or nonviral vector can be used. At least one nonviral vector has been used successfully for intraprostatic gene therapy. Preliminary studies with this vector have demonstrated the safety and efficacy of intratumoral injection of liposome-DNA complexes. This is described in Naitoh et al . , 1998, "Intraprostatic interleukin-2 (IL-2) gene therapy:

Preliminary results of a phase I clinical trial for the treatment of locally advanced prostate cancer, " AUA Abstract J. Urol . 159:254.

Several different viral vectors have been developed for gene delivery in various gene therapy protocols to treat prostate cancer. See Table 2 in Palaputtu et al . , 1999, "Gene Therapy for Prostate Cancer, " Urology Clinics of North America 26:353-363. A gene therapy protocol based on gene delivery through intraprostatic injection of an adenovirus vector has been tested successfully in an animal model of benign prostate hyperplasia. This is described in Cheon et al., 1999, J^". Urol . 161:305 Supplement, abstract no. 1174. Targeting of gene therapy to the prostate also can be accomplished through the use of a prostate-specific promoter. See Yu et al . , Cancer Res . 1999, 59:1498-1504. It should be noted that in the present invention, the therapeutic gene product is secreted by transformed cells. Consequently, it is not necessary to achieve genetic transformation of all cells in the target tissue to achieve beneficial results. For example, in the treatment of benign prostate hyperplasia or prostate cancer, transformation of a substantial number of prostate cells will result in secretion of the FS anagonist. The secreted FS antagonist then can be utilized by non-transformed prostate cells neighboring the transformed cells.

The anatomic location of the human prostate facilitates intraprostatic delivery of DNA vectors to treat local disease. Preferably, routes are transperineal and transrectal. Typically, in the transperineal approach, a transrectal ultrasound probe is used to guide the placement of needles through the perineum into the prostate . An advantage of the transperineal route is decreased risk of infection, as compared to the transrectal route.

Transrectal delivery typically is performed in a manner similar to an ultrasound-guided biopsy of the prostate.

Experimental Information Recombinant human activin A (lot 15365-36) and recombinant human follistatin-288 (preparation #B4384) were provided by the National Hormone and Pituitary Program of the NIH. High specific-activity radioiodide, tritiated thymidine, and tritiated heparin were purchased from NEN. Sequence grade chemicals were used for chemical modification, peptide synthesis, and protein sequencing. Pre-formed linear gradient SDS gels for electrophoresis of proteins were purchased from Pharmacia. Fuji X-omat film was used for autoradiography.

Peptide synthesis

Fig. 1 shows the peptides synthesized and the position of the follistatin sequences they represent within the native FS protein. Synthetic peptides were designed to: (1) represent the entire FS N-Domain for mapping the activin binding site; (2) represent the heparin sulfate binding site; (3) represent the human-specific antigenic epitope of human FS; (4) represent intra-cysteine sequences from each of the "follistatin module" domains of the FS288 core protein. Peptides were also used to generate site-directed monoclonal antibodies for use as structural probes of the C-termini of FS288 and FS315. Progressive N-terminal truncations were introduced in the two peptides identified as containing potential activin-binding epitopes.

All peptides were prepared in the Peptide Core Facilities of the Reproductive Endocrine Sciences Center at Massachusetts General Hospital, using solid-phase synthesis and Fmoc chemistry. The peptides were purified by reverse- phase HPLC. Mass spectroscopy was used to verify the sequences and to confirm peptide homogeneity.

Monoclonal antibodies

The monoclonal antibody recognizing FS Domain II was generated in mice immunized with recombinant human FS 288 essentially as previously described (MacDonald et al . , Biol . Reprod. 52/Supplement, 77; McConnell et al . , 1998, J. Clin . Endocrinol . Metabol . 83:851-858). Monoclonal antibodies against FS Domain III and the C-Domain were generated against synthetic peptides representing amino acid residues 274-287 and 300-315, respectively. Mice were immunized with protein or synthetic peptide-BSA conjugates emulsified in Freund's Complete Adjuvant (Sigma Chemical Co., St. Louis, MO) . Booster immunizations were composed of protein or peptide conjugated to BSA in Freund's Incomplete Adjuvant emulsifants. Both primary and booster immunizations were administered subcutaneously. After achieving suitable serum titers, spleen cells were harvested and fused with SP2/0 myeloma cells obtained from the ATCC (Bethesda, MD) . Hybridomas were selected by standard dilution cloning and enzyme immunoassay methods. Monoclonal antibodies were purified using protein G affinity chromatography from ascites fluid generated from pristane-primed mice inoculated (ip) with 5 x 10^s hybridoma cells. Harlan Laboratories Inc. (Indianapolis, ID) generated and harvested the ascites fluids. The use of mice in these studies was performed in compliance with NIH guidelines. Experimental procedures were reviewed and approved by an internal review board for the use of animal resources. Animals were housed and cared for in AAALAC accredited facilities operated by the Massachusetts General Hospital or Harlan Laboratories.

Solid-phase binding assays

These assays were used to identify and refine activin, heparin, and antibody binding epitopes. FS peptides or full-length FS were coated onto Immulon-4 ELISA plates (Dynex Technologies, Chantilly, VA) by incubating 100 uL of the appropriate solution at 4° C overnight. FS peptides were initially dissolved in 0.1 M acetic acid to 10 mg/mL and then diluted with 0.1 M sodium bicarbonate to 0.3, 1, 3, 10, 30 or 100 μg/mL. Native FS was coated in a similar fashion using a 0.2 μg/mL solution. Non-specific binding sites on ligand-coated (peptide or full-length FS) plates were blocked by incubation with 200 μL of a blocking solution (PBS containing 0.05% gelatin and 0.05% Tween-20) at room temperature for 2 hrs . Binding of [¹²⁵I] -activin, [³H] -heparin, or monoclonal anti-follistatin IgG was determined by incubating these probes in a total volume of 200 uL in ligand-coated or control wells at room temperature for 4 hrs with gentle shaking. After incubation, unbound probes were decanted and the plates were washed 3 times with PBS-0.05% Tween-20 to reduce non-specific binding.

Bound [¹²⁵I] -activin was measured by direct counting of the wells (inserted into 12 x 75 mm plastic tubes) in a gamma counter. Tritiated heparin bound to peptide or follistatin was measured by submerging washed wells in 10 mL of liquid scintillation fluid in a 20 mL glass scintillation vial and counted in a beta counter (Beckman, Los Angeles, CA) . Bound antibody was measured with an ELISA reader (405 nM absorbance) after sequentially incubating with an anti- mouse IgG-alkaline phosphatase conjugate (Pierce Chemical Co., Chicago, IL; 1/4000 dilution of conjugated antibody) then a PNPP substrate (Pierce Chemical Co.; 2.7 mM solution) .

Iodine-labeled activin was prepared using lactoperoxidase and purified by gel eletrophoresis as previously described (Schneyer et al . , 1994, Enodocrinology 135:667-674). The specific activity of the [¹²⁵I] -activin was approximately 30 μCi/μg. Tritiated heparin was obtained from New England Nuclear (Boston, MA) .

Additionally, a solid phase sandwich-type assay was used to examine the effects of activin on FS Domains I and II. FS288 in the absence or presence (5 and 50 ng/ml) of activin was allowed to bind to monoclonal antibody 7FS30 coated on the bottom of ELISA wells by incubating at room temperature for 12 h. After removal of unbound FS by washing, tritiated heparin was added and incubated at room temperature for an additional 3 hr to allow its binding to follistatin (free or activin-bound) captured by the solid- phase antibody. Upon completion of the incubation, the wells were washed to remove free tritiated heparin, then follistatin-bound tritiated heparin was counted in a scintillation counter. This basic sandwich-type assay format was also employed in a modified form utilizing solid- phase capture antibody covalently coupled to paramagetic particles and a detection antibody conjugated to dimethylacrydinium ester. In this modification, bound reagents were separated from free using magnetic tube racks and the detection antibody was measured in a Ciba-Corning Magic Light Analyzer II (Chiron Diagnostics, Inc.).

Pyridylethylation of follistatin To probe for free sulfhydryl groups within follistatin, the sulfhydryl-reactive reagent vinylpyridine was used for derivatization of native follistatin and a control preparation fully reduced by mercaptoethanol (0.4%, 2 hr, 37 C) . The respective preparations were incubated with vinylpyridine in 6 M guanidine/Tris-HCI buffer (0.25 M, pH 8.5) for 2 hr at 37 C and separated from reagents using a Waters (Milford, MA) Sep-Pak C₁₈ cartridge. The stable pyridylethylcysteine residues formed were quantitated by amino acid analysis (Beckman model 6300 analyzer) after total acid hydrolysis (6N HCI, 110 C, 24 hr) . N-terminal sequence analysis to characterize Cys at position 3 was performed on the Applied Biosystems Model 477A microsequencer.

SDS-PAGE and LIGAND blot analysis were performed as previously described (Wang et al . , 1996, J. Clin . Endocrinol . Metab. 81:1434-1441). X-ray film (XOMAT, Eastman Kodak, Rochester, NY) was exposed to the dried nitrocellulose blots for 3 to 7 days at -80 C.

Activin binding to synthetic mimotopes

Data on binding of recombinant human activin-A to to peptides representing linear sequences from human FS are summarized in Fig. 2A. Activin bound to only two synthetic fragments,

and FS_44-59, of the N-Domain (amino acids 1- 63) . Under similar experimental conditions, no activin binding was observed to synthetic peptides- representing epitopes with the FS Domains I, II, or III. Also shown in Fig. 2 is the mapping of tritiated heparin (Figure 2B) and monoclonal antibody 7FS30 (Figure 2C) using this series of synthetic peptides. Heparin binding mapped to separate epitopes within the N-Domain and FS Domain I, respectively. Heparin binding to the 74-86 fragment confirmed the previously identified heparin binding site. The monoclonal antibody 7FS30 bound only to the 168-178 fragment within the FS Domain II.

Because relatively low affinity interactions can be detected using small synthetic peptides as solid phase ligands, competition between the solid phase peptides and native follistatins was used to define further the specificity of the activin binding observed. As shown in Fig. 3, binding of iodinated activin to the peptides was completely inhibited by native follistatins; both FS288 and FS315 inhibited the binding of activin to the synthetic follistatin epitopes. Thus, the binding of activin to

and FS_44-59 appeared to depend on the same binding regions involved in its binding to native FS.

N-terminal binding sequences

Specific activin binding to two non-adjacent, linear sequences (Figs. 2 and 3) indicated that there may be at least two distinct activin binding epitopes (1-26) and (44- 59) located with the FS N-Domain. To define these activin binding mimotopes more precisely, a series of N-terminal truncated peptides was synthesized. Each series of peptides successively truncated at the N-terminus was directly tested as solid-phase ligands for ability to bind activin. Fig. 4A summarizes the results of mapping the binding of activin to peptide FS_1-26. The equipotent activin binding displayed by peptide

and peptide FS_3-26 showed amino acid residues 1 and 2 of FS not to be required for activin binding.

However, none of the peptides truncated beyond 3-26 were able to bind activin, indicating that amino acids 3 to 5 of follistatin are essential for the binding activity of peptide FS₃.₂₆. Fig. 4B shows data summarizing the mapping of the

FS_44-59 mimotope. Removal of three amino acids from the N terminus did not significantly affect [¹²⁵I] -activin binding. However, the subsequent deletion of amino acids 47-48 resulted in a complete loss of [¹²⁵I] -activin binding to this mimotope. This result indicated those two amino acids to be critical to activin binding to this mimotope, and critical to activin binding to this region in native FS. Because this potential epitope had not been previously identified, an additional control was prepared by synthesizing a full length (44-59) peptide with a scrambled sequence of the identical amino acid composition. This scrambled sequence was inactive (Fig. 4B) , showing the specificity of the native sequence for activin binding.

Activin binding to chemically modified, native FS The binding of activin to synthetic fragments of FS indicated at least two separate regions to be involved in activin binding. Therefore, FS288 was unfolded by reduction of cysteine residues to determine if the linear epitopes,

FS₃.₂₆ and FS_47-59, bind activin in linearized native FS or constitute a conformation-dependent binding site(s) within the N-Domain. Ligand blotting ( [¹²⁵I] -activin) after SDS gel-electrophoresis of reduced or unreduced FS₂₈₈ was used to assess the ability of activin to bind epitopes within the unfolded protein. Activin binding to FS288 depended on the disulfide bonding of the protein (data not shown) . The lower limit of detection by ligand blotting was approximately 1 μg per lane. No activin binding was detected when 20 μg of reduced FS was loaded. The amount of activin binding was near the limit of detection when the gels were deliberately overloaded with 100 μg of follistatin per lane. It was concluded from these considerations that reduction of the disulfide bonds resulted in at least a 95% loss of activin binding.

Pyridylethyl derivatization and N-terminal sequencing of FS

The cysteine residue at position 3 is part of an activin binding mimotope (Figs. 4A and 4B) . Its importance was indicated by loss-of-function mutants created by amino acid insertions between residues 2 and 3 of follistatin. This cysteine could be part of a disulfide bond in the native FS protein, or it could be present in an unbonded, reduced state. If Cys₃ is reduced in the native protein, its free sulfhydryl group might participate in post-binding reactions between activin and follistatin, explaining the essentially irreversible behavior of the follistatin-activin complex. N-terminal composition and sequence analysis was performed after reacting FS288 with

4-vinylpyridine to specifically derivatize free sulfhydrils ._. This was done using methodology described in Jones, 1986, in Methods of Protein Microcharacterization (Shively, ed.) pp. 337-361, The Human Press; and Kuhn et al . , 1986, in Advanced Methods in Protein Microsequence Analysis (Wittmann-Liebold, ed.) pp. 65-76, Springer-Verlag, Berlin.

Compositional analysis revealed the pyridylethyl- cysteine residue eluting at 51 minutes .in the reduced vinylpyridine-treated control preparation to be absent from the derivatized, native FS (data not shown) . To determine precisely the state of the sulfhydryls within the N-Domain (3-26) binding epitope, each preparation was subjected to Edman microsequence analysis through 15 cycles. Phenylthiohydantoins representing pyridylethylcysteine, eluting between Tyr and Pro by on-line HPLC, were observed at positions 3 and 13 of the reduced, pyridylethylated control sample. No pyridylethylcysteine peak was found at either position in the pyridylethylated native preparation (data not shown) . This indicated that cysteine residues in native follistatin, including those at the active N- terminus, were present in the disulfide-bonded form, unavailable for covalent crosslinking with activin ligand. To determine if activin binding influenced the conformation of the non-activin binding domains, probes specific for epitopes located in each of the three 10- cysteine-containing "follistatin module" domains (Domains I- III; Fig. 1), or in the C-Domain (Fig. 1), were utilized before and after activin was bound to FS288 or FS315. Tritiated heparin was used as a probe for the heparan sulfate binding site located in Domain I (Fig. 2B) . Domain II was probed using a monoclonal antibody designated 7FS30, specific to the 168-178 sequence (Fig. 2C) . Activin induced changes in FS Domains I or II were probed simultaneously using a "sandwich" type immunoassay composed of solid-phase antibody bound to plastic plates to capture follistatin and the tritiated heparin to detect the antibody-bound protein. FS response curves (which require binding of both probes) were superimposable . Thus, activin binding did not impair recognition by specific probes to either Domain I or II. In contrast, activin binding resulted in dramatic changes in antigenic epitopes located within Domain III or the C-Domain. These changes were identified using sandwich- type immunoassays again utilizing 7FS30-conjugated paramagnetic particles to capture FS . Domain-specific epitopes were detected and quantitated using DMAE-coupled monoclonal antibodies generated against synthetic peptides representing epitopes in Domain III (amino acid 274 to 287) and in the C Domain (amino acid 300 to 315) . Activin did not bind directly to either of these domain-specific epitopes (Fig. 2C) . However, activin binding to FS resulted in a marked decrease in recognition by the monoclonal antibody directed against the Domain III epitope (data not shown) . Activin binding also effected changes in the C- Domain, found only in FS315. In this case, activin binding enhanced antibody binding to the C-Domain epitope (data not shown) . This observation suggested conformational changes associated with activin binding, resulting in a more favorable C-Domain antigenic/epitope, either through higher affinity recognition or an enriched ensemble of favorable conformers among the population of FS315 molecules.

Critical residues for activin binding To more precisely identify particular amino acids critical for activin binding, a series of site-directed mutations was generated in the N-Domain of FS288. Proteins containing the mutations were recombinantly produced in cultured CHO cells, which secreted the mutant forms of FS288 into the cell culture medium. Conditioned media containing the mutant FS288 proteins were used in solid phase binding assays, essentially as described above, to test for activin binding by the mutant FS288 proteins. In these assays, 100 ng/ml of each expression product, as measured by FS immunoassay, was used. Data from these assays are summarized in Fig. 5. Data are expressed as percent inhibition of activin binding, computed as [1- (%B/B₀) , where B = labeled activin bound in the presence of sample, and B₀ = activin bound in presence of saturating dose of FS. Binding activity was markedly impaired in the Trp-4 mutant, whereas all others displayed binding comparable to that of native ("wild type") FS.

Other embodiments are within the following claims.

Claims

Claims We claim:

1. A follistatin antagonist comprising: amino acids 3-63 of a non-activin-binding follistatin N-Domain (amino acids 3-63 of SEQ ID NO:l) or, or amino acids 3-61 of a non-activin-binding FLRG N-Domain (amino acids 3-61 of SEQ ID NO: 2) ; and a heparin sulfate-binding moiety; wherein the tryptophan residue (W) at position 4 or 36 of the non-activin-binding follistatin N-Domain or non- activin-binding FLRG N-Domain is substituted by an amino acid selected from the group consisting of A, G, L, I, V, P, S, T, M, C, N, Q, D, E, K, R, Y and H.

2. The follistatin antagonist of claim 1, wherein the non-activin-binding follistatin N-Domain or non-activin- binding FLRG N-Domain is located N-terminally relative to the heparin sulfate-binding moiety.

3. The follistatin antagonist of claim 1, wherein the heparin sulfate-binding moiety comprises amino acids 74- 87 of SEQ ID NO:l.

4. The follistatin antagonist of claim 3, wherein the heparin sulfate-binding moiety comprises amino acids 64- 136 of SEQ ID NO:l.

5. The follistatin antagonist of claim 4, wherein the heparin sulfate-binding moiety comprises amino acids 64- 211 of SEQ ID NO:l.

6. The follistatin antagonist of claim 5, wherein the heparin sulfate-binding moiety comprises amino acids 64- 288 of SEQ ID NO:l.

7. The follistatin antagonist of claim 1, wherein the tryptophan residue (W) at position 4 or 36 of the non- activin-binding follistatin N-Domain is substituted by an amino acid selected from the group consisting of A, G, S, T, N, Q, R, D and E.

8. The follistatin antagonist of claim 7, wherein the tryptophan residue (W) at position 4 or 36 of the non- activin-binding follistatin N-Domain is substituted by an amino acid selected from the group consisting of A or G.

9. The follistatin antagonist of claim 8, wherein the tryptophan residue (W) at position 4 or 36 of the non- activin-binding follistatin N-Domain is substituted by A.

10. The follistatin antagonist of claim 1, wherein the tryptophan residue (W) at position 4 and 36 of the non- activin-binding follistatin N-terminal domain or non- activin-binding FLRG N-terminal domain is substituted by an amino acid selected from the group consisting of A, G, L, I, V, P, S, T, M, C, N, Q, D, E, K, R, Y and H.

11. A nucleic acid encoding the follistatin antagonist of claim 1.

12. A vector comprising the nucleic acid of claim 11.

13. The vector of claim 12, further comprising an expression control sequence operatively linked to the nucleic acid encoding the follistatin antagonist of claim 1

14. A host cell transformed with the vector of claim 13.

15. A method of increasing activin activity in an activin-sensitive tissue in a mammal, comprising contacting the cells of the activin-sensitive tissue with the follistatin antagonist of claim 1.

16. The method of claim 15, wherein contacting the cells of the activin-sensitive tissue with- the follistatin antagonist comprises expressing the follistatin antagonist in vivo from a vector comprising a nucleic acid encoding the follistatin antagonist polypeptide.

17. The method of claim 16, wherein the activin- sensitive tissue is prostate tissue.