EP4677595A2 - De novo designed high affinity protein binders to convex protein target sites on tgfbrii, ctla-4 and pd-l1 - Google Patents

Abstract

Polypeptides having the amino acid of one of SEQ ID NO: 1-9 are provided that bind to convex protein target sites on transfomimg growth factor beta receptor type 2 (TGFbRII), cytotoxic T-lymphocyte associated protein 4 (CTLA-4), or programmed death-ligand 1 (PD-L1), and methods for their use in treating cancer or tissue fibrosis.

Description

De Novo Designed High Affinity Protein Binders to Convex Protein Target Sites on TGFbRIl, CTLA-4 and PD-L1

Federal Funding Statement

This invention was made with government support under Grant No. HR0011-21-2- 0012, awarded by the Defense Advanced Research Projects Agency and Grant No. R01AG063845-01, awarded by the National Institute on Aging. The government has certain rights in the invention.

Sequence Listing Statement

A computer readable form of the Sequence Listing is filed with this application by electronic submission and is incorporated into tins application by reference in its entirety. The Sequence Listing is contained in tire file created on March 5, 2.024 having the file name “23- 0158-WO.xml” and is 342,816 bytes in size.

Background

Naturally occurring high affinity' protein-protein interfaces generally exhibit considerable shape complementarity, which enables concerted interatomic interactions and solvation free energy reduction needed to overcome the entropic cost of macromolecular association. Design of proteins that bind to convex protein target sites are difficult due to the requirement for overall shape matching. Methods for designing proteins which bind to convex target sites could considerably expand the power and scope of de novo binder design.

Summary

In one aspect, the disclosure provides polypeptides comprising the amino acid sequence selected from the group consisting of SEQ ID NO: 1-9. In one embodiment, the polypeptides comprise the amino acid sequence selected from the group consisting of SEQ ID NO: 1-3, wherein the polypeptide binds to CTLA-4. In another embodiment, the polypeptides comprise the amino acid sequence selected from the group consisting of SEQ ID NO:4-6, wherein the polypeptide binds to PD-Ll . In a further embodiment, the polypeptides comprise the amino acid sequence selected from the group consisting of SEQ ID NO:7-9, wherein the polypeptide binds to TGFbRII. In various embodiments, the polypeptides comprise an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence selected from SEQ ID NO: 10-16, not including any insertions. In another embodiment, the disclosure provides fusion proteins comprising the polypeptide of any embodiment herein and one or more functional domains at the N-terminus and/or at the C-terminus of the polypeptide.

The disclosure also provides nucleic acids encoding the polypeptide or fusion protein of any embodiment or combination of embodiments herein, expression vectors comprising tlie nucleic acid operatively linked to a promoter, host cells comprising the polypeptide, fusion protein, nucleic acid, or expression vector of any embodiment, and pharmaceutical compositions, comprising the polypeptide, the nucleic acid, the expression vector, and/or the host cell of embodiment; and a pharmaceu tically acceptable carrier.

In another aspect, the disclosure provides methods for treating cancer or tissue fibrosis, comprising administering to a subject in need thereof an amount of the polypeptide, fission protein, nucleic acid, expression vector, host cell, and/or pharmaceutical composition effective to treat the cancer or tissue fibrosis.

Description of the Figures

Figure 1. Design of 5HCS scaffolds to target convex interfaces. (A) Distribution of protein-protein interface curvatures from the PDB and designed protein binders. Previously designed protein binders (for these, the designed binders are partner 1 and the targets, partner 2), Previously designed protein binders have been limited to binding to flat or concave interfaces (receptor convexity <=0). Examples of native protein complexes, v: PDB ID, 5XXB; vi: TGFpIIl/TGFpRll complex, PDB ID. 1KTZ, vii: CD86/CTLA-4 complex, PDB ID, 1185, viii, PD-1 /PD-Ll, PDB ID, 3bik. The TGFpRII and CTLA-4 functional interfaces showed high convexity, which we used as case studies to design concave binders. The 5HCS scaffolds described in the examples can target convex binding sites. The distribution of convexity of the 5HCS scaffolds (upper part of panel a) shows that the 5HCS scaffolds are diverse enough to cover most of the naturally existing convex interfaces, (B) Design models of complexes highlighted in panel a. i,ii,ii are PDCsFR, 1GF1R, H3 in complex with corresponding de novo minibinders; iv, 5HCS binder in complex with TGFpRII; v, PDB ID: 5XXB; vi, TGFpiII/TGFpRII complex, PDB ID: 1KTZ. (C) Design workflow. Column 1 : 5FICS concave scaffolds with a wide range of curvatures were designed with three helices forming the concave surfaces (Cbeta labeled as spheres ) and two helices butressing at the back side. Column 2: Docking of 5HCS scaffolds to target binding sites. Column 3: Following docking, the interface sequencing is optimized for high affinity binding.

Figure 2. Concave 5HCS binder to TGFpRII. (A) Design model of 5HCS_TGFBR2_1 (cartoon) binding to TGFpRII (PDB ID: 1KTZ). 5HCS_TGFBR2_I is shown by Shannon entropy from the site saturation mutagenesis results at each position from low entropy, conserved, to high entropy, not conserved. (B) Circular dichroism spectra from 25 °C to 95 °C for 5HCS TGFBR2 1. (C) Biolayer interferometry characterization of 5HCS_TGFBR2_1. Biotinylated TGFpRII were loaded to Streptavidin (SA) tips and incubated with 2.7 nM, 0.9 nM and 0.3 nM of 5HCS TGFBR2 1 to measure the binding affinity'. The binding responses are shown in solid lines and fited curves shown in dotted lines. (D) Heat map of the log enrichments for the 5HCS_TGFBR2_1 SSM library’ selected with 1.6 nM TGFpRII at representative positions. Tire annotated amino acid in each column indicates the residue from the parent sequence.

Figure 3. Designed 5HCS CTLA-4 binder. (A) Model of 5HCS CTLA4J (cartoon) binding to CTLA-4 (PDB ID: 1185 ) shown by Shannon entropy’ from site saturation mutagenesis results. (B) Circular dichroism spectra from 25 °C to 95 °C tor 5HCS__CTLA4 __1. (C) Biolayer interferometry characterization of 5HCS__ CTLA4__1. Biotinylated CTLA-4 was loaded to Streptavidin (SA) tips and these were incubated with 2.7 nM, 0.9 nM and 0.3 nM of 5HCS CTLA4 1 to measure the binding affinity. (D) Log enrichments for the 5HCS CTLA4 1 SSM library selected with 10 nM CTLA-4 at representative positions. The annotated ammo acid in each column indicates the residue from the parent sequence.

Figure 4. Designed 5HCS binder to PD-Ll . (A) Model of 5HCS PDL1 1 (cartoon) binding to PD-Ll (PDB ID: 3BIK), with 5HCS PDL1 1 shown by Shannon entropy from site saturation mutagenesis results. (B) Circular dichroism spectra from 25 °C to 95 °C for 5HCS PDL1 1. (C) Biolayer interferometry characterization of 5HCS PDL1 1. Biotinylated PD-Ll was loaded to Streptavidin (SA) tips and these were incubated with 8 nM, 2.7 nM and 0.9 nM of 5HCS PDL1 1 to measure the binding affinity . (D) Heat map representing the log enrichments for the 5HCS_PDL1_1 SSM library selected with 6 nM PD- Ll at representative positions. The annotated amino acid in each column indicates the residue from the parent sequence. Figure 5. Computational Characterization of 5HCS scaffolds library. (A) Distribution of curvatures of the 5HCS library. (B) Distribution of pLDDT predicted by AlphafoldZ. (C) Distribution of plDDT predicted by DeepAccNet.

Figure 6. TGF0RII binding protein binding site and SSM analysis. (A) Complex structure of the TGF0RH and the TGF0-3. (B) 5HCS_TGFBR2J binds to the TGF0-3 binding site on the TGF0RII. (C) Heat map representing the log enrichments for the 5HCS_TGFBR2_1 SSM libraiy selected with 1.6 nM TGF0RII.

Figure 7. Examples of Immunoglobulin domain head-to-head interactions. PDB ID: 1181, liqd, 2jjs, 3r08, 3s35, 4g6m, 4hcr, 5dlq.

Figure 8. CTLA-4 binding protein binding site and SSM analysis. (A) Complex structure of the CTLA-4 and the CD86. (B) 5HCS_CTLA4_1 binds to the CD86 binding site on the CTLA-4. (C) Heat map representing the log enrichments for the 5HCS CTLA4 1 SSM library selected with 10 nM CTLA-4.

Figure 9. PD-L1 binding protein binding site and SSM analysis. (A) Complex structure of the PD-L1 and the PD-1. (B) 5HCS PDL1 1 binds to the PD-1 binding site on the PD-L1. (C) Heat map representing the log enrichments for the 5HCS PDL1 1 SSM library' selected with 6 nM PD-L1.

Detailed Description

All references cited are herein incorporated by reference m their entirety. Within this application, unless otherwise stated, the techniques utilized may be found m any' of several well-known references such as: Molecular Cloning: A Laboratory Manual (Sambrook, et al.,

1989, Cold Spring Harbor Laboratory'’ Press), Gene Expression Technology (Methods in Enzymology, Vol. 185, edited by D. Goeddel, 1991. Academic Press, San Diego, CA), “Guide to Protein Purification” in Methods in Enzymology (MLP. Deutshcer, ed., (1990) Academic Press, Inc.); PCR Protocols: A Guide to Methods and Applications (Innis, et al.

1990. Academic Press, San Diego, CA), Culture of Animal Cells: A Manual of Basic Technique, 2^nd Ed. (R.I. Freshney. 1987. Liss, Inc. New York, NY), Gene Transfer and Expression Protocols, pp. 109-128, ed. E.J. Murray, The Humana Press Inc., Clifton, N.J.), Dang, B. et ai. and the Ambion 1998 Catalog (Arnbion, Austin, TX).

As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise.

As used herein, the amino acid residues are abbreviated as follows: alanine (Ala; A), asparagine (Asn; N), aspartic acid (Asp; D), arginine (Arg; R), cysteine (Cys; C), glutamic acid (Glu; E), glutamine (Gin; Q), glycine (Gly; G), histidine (His; H), isoleucine (He; I), leucine (Leu; L), lysine (Lys; K), methionine (Met; M), phenylalanine (Phe; F), proline (Pro; P), serine (Ser; S), threonine (Thr; T), tryptophan (Tip; W), tyrosine (Tyr; Y), and valine (Vai; V).

Any N-terminal methionine residue in any polypeptide of the disclosure may be present or may be deleted. In all embodiments of the polypeptides disclosed herein, 1, 2, 3, 4, or 5 residues may be deleted from the N-terminus and/or the C-terroinus of the polypeptide while retaining activity.

All embodiments of any aspect of the disclosure can be used in combination, unless the context clearly dictates otherwise.

Unless the context clearly requires otherwise, throughout the description and the claims, the words ‘comprise’, ‘comprising’, and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in the sense of “including, but not limited to”. Words using the singular or plural number also include the plural and singular number, respectively. Additionally, the words “herein,” “above,” and “below” and words of similar import, when used in this application, shall refer to this application as a whole and not to any particular portions of the application.

In a first aspect, the disclosure provides polypeptide comprising the amino acid sequence selected from the group consisting of SEQ ID NO: 1-9. The polypeptides of the disclosure are high affinity binders to convex protein target sites on transforming growth factor beta receptor type 2 (TGFbRII), cytotoxic T-lymphocyte associated protein 4 (CTLA- 4), or programmed death-ligand 1 (PD-L1). As described in the examples, the inventors designed de novo binding proteins for these targets and conducted extensive site saturation mutagenesis studies to identify permissible substitutions, provided in Tables 1-3. Hie ammo acid sequence of SEQ ID NO: 1-9 are shown in Tables 1 -3.

Table 1 provides sequences and permissible substitutions for CTLA4 mini-binders (SEQ ID NO: 1-3), Table 2 provides sequences and permissible substitutions for PD-L1 minibinders (SEQ ID NO:4~6), and Table 3 provides sequences and permissible substitutions for TGFBR2 minibinders (SEQ ID NO:7-9). In each case, each row in the table represents one residue in the polypeptide and the different columns show permissible substitutions from the “native” amino acid residue at that position (i.e., “native” meaning a residue present in a specific design shown in Table 4 below), based on site saturated mutagenesis studies as described in the examples. The first column in Tables 1-3 provides the residue number of the reference polypeptide, the second column lists the amino acid sequence of an exemplary design listed in Table 4, the third column indicates whether the residue is present in a loop of the polypeptide, and the fourth column indicates whether the residue is present at an interface between the polypeptide and its target (i.e., TGFbRII, CTLA-4, or PD-L1).

Columns 5-7 provide all residues that can be present at defined positions (the starting residue from column 2, and permissible substitutions), as determined by site saturation mutagenesis studies. For example, in SEQ ID NO: 1, residue 1 can be P, Q, T, S, M, E, A, L, N, W, D, V, H, F, Y, C, K, I, or G. In SEQ ID NO:2, residue 1 can be N, W, M, D, V, E, A, T, L, P, Y, S, H, K, I, G, or Q. In SEQ ID NO:3, residue 1 is N. Those of skill in the art will readily understand the sequences of each SEQ ID NO: 1 -9 as presented in the Tables.

Table 1. CTLA4 mini-binders

Table 2. PD-L1 minibinders

Table 3. TGFBR2 minibinders

Table 4 provides exemplary binder designs, which are described in detail in the examples. Table 4. Specific Designs

In one embodiment, the polypeptide comprises the amino acid sequence selected from the group consisting of SEQ ID NO: 1-3, wherein the polypeptide binds to CTLA-4. In these embodiments the polypeptides may be used, for example, in cancer immunotherapy. Exemplary tumor types that can be treated with the CTLA-4 binders of the disclosure include, but are not limited to, melanoma, lung cancer (such as non-small cell lung cancer), bladder cancer, head and neck cancer, renal cell carcinoma, ovarian cancer, and colorectal cancer. In one embodiment, the polypeptide comprises the amino acid sequence of SEQ ID NO:3. In another embodiment, the polypeptide comprises an amino acid sequence at least 50%, 55%. 60%, 65%, 70%. 75%, 80%, 85%. 90%, 91%, 92%. 93%, 94%, 95%. 96%, 97%, 98%, 99%, or 100% identical to the ammo acid sequence selected from SEQ ID NO: 10-13. Exemplary substitutions are as disclosed in Table 1 above. The amino acid sequences of SEQ ID NO: 10-13 are provided in Table 4, and are discussed in detail in the examples. In another embodiment, the polypeptide is identical relative to tire reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, I I, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues. Interface residues are shown in column 4 of Table 1, and are at positions 8, 11-16, 18-20, SO- 51, 54-55, 58, 89-90, 92-94, 96-98, and 100.

In another embodiment, the polypeptide comprises the amino acid sequence selected from the group consisting of SEQ ID NO:4-6, wherein the polypeptide binds to PD-L1. In these embodiments the polypeptides may be used, for example, in cancer immunotherapy. Exemplary tumor types that can be treated with the PD-L1 binders of the disclosure include, but are not limited to, non-small cell lung cancer, melanoma, bladder cancer, head and neck squamous cell carcinoma, Hodgkin lymphoma, renal cell carcinoma, gastric or gastroesophageal junction adenocarcinoma, cervical cancer, and breast cancer (including but not limited to triple-negative breast cancer). In one embodiment, the polypeptide comprises the amino acid sequence of SEQ ID NO:6. In another embodiment, the polypeptide comprises an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the ammo acid sequence selected from the group consisting of SEQ ID NO: 15 or 16, Exemplary substitutions are as disclosed in Table 2 above. The amino acid sequence of SEQ ID NO: 15- 16 are provided in Table 4, and are discussed in detail in the examples. In another embodiment, the polypeptide is identical relative to the reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues.

Interface residues are shown in column 4 of Table 2, and are at positions 2, 6, 9-10, 13, 45, 48-49. 52-53, 56, 59, 86-88, 91-92, 95-96. 99, and 102.

In another embodiment, the polypeptide comprises the amino acid sequence selected from the group consisting of SEQ ID NO:7-9, wherein the polypeptide binds to TGFbRII. In these embodiments the polypeptides may be used, for example, in in treating cancer and tissue fibrosis, by interfering with the TGF-p pathway. Exemplary tumor types that can be treated with the TGFbRII binders of the disclosure include, but are not limited to, pancreatic cancer, breast cancer, colorectal cancer, lung cancer, prostate cancer, liver cancer, gastric cancer, ovarian cancer, melanoma, and glioblastoma. Exemplary uses for treating tissue fibrosis include, but are not limited to, idiopathic pulmonary’ fibrosi s (IPF), cirrhosis of the liver, renal fibrosis, cardiac fibrosis, scleroderma, keloids, and hypertrophic scarring. In one embodiment, the polypeptide comprises the amino acid sequence of SEQ ID NO:9. In another embodiment, the polypeptide comprises an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 14. Exemplary’ substitutions are as disclosed in Table 3 above. The amino acid sequence of SEQ ID NO: 14 is provided in Table 4, and is discussed in detail in the examples. In another embodiment, the polypeptide is identical relative to the reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues. Interface residues are shown in column 4 of Table 3, and are at positions 3, 6, 10, 42-43, 45-46, 48-50, 52-53, 56-57, 85-86, 88-89, 92-93, 95-96, and 100-101.

In another embodiment, the disclosure provides polypeptides comprising an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence selected from SEQ ID NO: 10-16, not including any insertions.

Tables 1-3 show in column 3 the position of loop regions in the polypeptides of the disclosure. The loop regions can accommodate amino acid insertions. Tirus, in another embodiment the polypeptides comprise an insertion in one or more loop regions of the polypeptide. In certain embodiments, ammo acids or amino acid domains (such as a functional domain) may be inserted in the loop region. The polypeptides of the disclosure may include any such insertion, and in these embodiments the polypeptide would still comprise the reference amino acid sequence, with an interruption at tire site of insertion.

In another embodiment, substitutions relative to the reference sequence are conservative amino acid substitutions. Such conservative amino acid substitutions involve replacing a residue by a residue having similar physiochemical characteristics, e.g., substituting one aliphatic residue for another (such as He, Vai, Leu, or Ala for one another), or substitution of one polar residue for another (such as between Lys and Arg; Glu and Asp; or Gin and Asn). Other such conservative substitutions, e.g., substitutions of entire regions having similar hydrophobicity characteristics, are known. Amino acids can be grouped according to similarities in the properties of their side chains (in A. L. Lehninger, in Biochemistry, second ed., pp. 73-75, Worth Publishers, New York (1975)): (1) non-polar: Ala (A), Vai (V), Leu (L), He (I), Pro (P), Phe (F), Trp (W), Met (M); (2) uncharged polar: Gly (G), Ser (S), Thr (T), Cys (C), Tyr (Y), Asn (N), Gin (Q); (3) acidic: Asp (D), Glu (E); (4) basic: Lys (K), Arg (R), His (H). Alternatively, naturally occurring residues can be divided into groups based on common side-chain properties: (1) hydrophobic: Norleucine, Met, Ala, Vai, Leu, He; (2) neutral hydrophilic: Cys, Ser, Thr, Asn, Gin; (3) acidic: Asp, Glu; (4) basic: His, Lys, Arg; (5) residues that influence chain orientation: Gly, Pro; (6) aromatic: Trp, Tyr, Phe,

In all of these embodiments, the position of residues in the polypeptides of the disclosure are “relative to” the position of residues in the reference sequence; this does not necessarily mean that the residue number in the polypeptide of the disclosure will be identical to the residue number in the reference sequence. Those of skill in the art will understand that the polypeptides of the disclosure may be fused to other functional domains (such as the fusion proteins or polypeptides with insertions described below), including N-terminal domains, or comprise insertions, such that the residue numbers in the polypeptides relative to the reference polypeptide sequence may differ.

In another embodiment, the disclosure provides fusion proteins, comprising: (a) the polypeptide of any embodiment or combination of embodiments herein; and

(b) one or more functional domains at the N-terminus and/or at the C-terminus of the polypeptide.

In these embodiments, any functional domain may be inserted (as an insertion at a loop region, and/or at one or both termini of tire fusion protein). In various non-limiting embodiments, the functional domain may comprise, for example, a targeting domain, a detectable domain, a. scaffold domain, an oligomerization domain, a secretion signal, an Fc domain, or a further therapeutic peptide domain. In one embodiment, the functional domain comprises an oligomerization domain, As described in the examples, fusing the binders to an oligomerization domain, optionally via a linker sequence (including but not limited to a flexible linker such as GS-rich linker), can improve avidity of binding to the target. One example of such a fusion is 5HCS__CTLA4__l_c6 (SEQ ID NO: 12). In which the oligomerization domain comprises the amino acid sequence of SEQ ID NO: 17. Those of skill in the art will understand that any oligomerization domain may be used, as suitable for an intended purpose.

DEEEAR.ELEEPAR.EAAKPAIEAAKRTGDPRVRELAEELVKLAIWAAVEWLDPSSSDVNEALKLIVEAIEAAVPA

LEAAERTGDPEVRELARELVRLAVEAAEEVQRNPSSSDVNEALKLIVIAIEAAVPA,LEAAERTGDPEVRELAREL

VRLAVEJGAEEVQRNPSSEEVNEALRKI I KLILFAVEWLELAEE I GDPTWREPLARRZAVREAVELAEEVQRDPSGWL

GH ( SEQ ID NO : 17 )

In one embodiment of any embodiment disclosed herein, the polypeptide or fusion protein binds its target with nanomolar or picomolar affinit y, as determined using biolayer interferometry and a. protocol as defined in the examples and figure legends.

In another aspect the disclosure provides nucleic acids encoding the polypeptide or fusion protein of any embodiment or combination of embodiments of the disclosure . The nucleic acid sequence may comprise single stranded or double stranded RNA or DNA in genomic or cDNA form, or DNA -RNA hybrids, each of which may include chemically or biochemically modified, non-natural, or derivatized nucleotide bases. Such nucleic acid sequences may comprise additional sequences useful for promoting expression and/or purificati on of the encoded peptide or chimeric molecular construct, including but not limited to poly A sequences, modified Kozak sequences, and sequences encoding epitope tags, export signals, and secretory signals, nuclear localization signals, and plasma membrane localization signals. It will be apparent to those of skill in the art, based on the teachings herein, what nucleic acid sequences will encode the polypeptide or fusion protein of the disclosure.

In a further aspect, the disclosure provides expression vectors comprising the nucleic acid of any aspect of the disclosure operatively linked to a suitable control sequence, such as a promoter. “Expression vector” includes vectors that operatively link a nucleic acid coding region or gene to any control sequences capable of effecting expression of the gene product. “Control sequences” operably linked to the nucleic acid sequences of the disclosure are nucleic acid sequences capable of effecting the expression of the nucleic acid molecules. The control sequences need not be contiguous with the nucleic acid sequences, so long as they function to direct the expression thereof. Thus, for example, intervening untranslated yet transcribed sequences can be present between a promoter sequence and the nucleic acid sequences and the promoter sequence can still be considered “operably linked” to the coding sequence. Other such control sequences include, but are not limited to, polyadenylation signals, termination signals, and ribosome binding sites. Such expression vectors can be of any type, including but not limited plasmid and viral-based expression vectors. The control sequence used to drive expression of the disclosed nucleic acid sequences in a mammalian system may be constitutive (driven by any of a variety of promoters, including but not limited to, CMV, SV40, RSV, actin, EF) or inducible (driven by any of a number of inducible promoters including, but not limited to, tetracycline, ecdysone, steroid-responsive). The expression vector must be replicable in the host organisms either as an episome or by integration into host chromosomal DNA. In various embodiments, the expression vector may comprise a plasmid, viral-based vector, or any other suitable expression vector.

In another aspect, the disclosure provides host cells that comprise the polypeptide, fusion protein nucleic acid or expression vector (i.e.: episomal or chromosomally integrated) disclosed herein, wherein the host cells can be either prokaryotic or eukaryotic. The cells can be transiently or stably engineered to incorporate the expression vector of the disclosure, using techniques including but not limited to bacterial transformations, calcium phosphate coprecipitation, electroporation, or liposome mediated-, DEAE dextran mediated-, polycationic mediated-, or viral mediated transfection.

In another aspect, the disclosure provides pharmaceutical compositions, comprising:

(a) the polypeptide, tire nucleic acid, the expression vector, and/or the host cell of any preceding claim; and

(b) a pharmaceutically acceptable carrier. In certain embodiments, the carrier in a pharmaceutical composition can be either aqueous or non-aqueous in nature. For example, in certain embodiments, a suitable vehicle or carrier can be water for injection, physiological saline solution or artificial cerebrospinal fluid, possibly supplemented with other materials common in compositions for parenteral administration. In certain embodiments, the saline comprises isotonic phosphate-buffered saline. In certain embodiments, neutral buffered saline or saline mixed with serum albumin are further exemplary vehicles. In certain embodiments, pharmaceutical compositions comprise Tris buffer of about pH 7.0-8.5, or acetate buffer of about pH 4.0-5.5, which can further include sorbitol or a suitable substitute therefore. In certain embodiments, a composition comprising an immunomodulatory fusion protein is prepared for storage by mixing the selected composition having the desired degree of puri ty with optional formulation agents (Remington's Pharmaceutical Sciences, supra) in the form of a lyophilized cake or an aqueous solution. Further, in certain embodiments, a composition comprising an immunomodulatory fusion protein is formulated as a lyophilizate using appropriate excipients such as sucrose.

The compositions may be used, for example, in the methods disclosed herein. The compositions may further comprise (a) a lyoprotectant; (b) a surfactant; (c) a bulking agent; (d) atonicity adjusting agent; (e) a stabilizer; (f) a preservative and/or (g) a buffer. In some embodiments, the buffer in the pharmaceutical composition is a Tris buffer, a histidine buffer, a phosphate buffer, a citrate buffer or an acetate buffer. The composition may also include a lyoprotectant, e.g. sucrose, sorbitol or trehalose. In certain embodiments, the composition includes a preservative e.g. benzalkonium chloride, benzethonium, chlorohexidine, phenol, m-cresol, benzyl alcohol, methylparaben, propylparaben, chlorobutanol, o-cresol, p-cresol, chlorocresol, phenylmercuric nitrate, thinierosal, benzoic acid, and various mixtures thereof. In other embodiments, the composition includes a bulking agent, like glycine. In yet other embodiments, the composition includes a surfactant e.g., polysorbate-20, polysorbate-40, polysorbate- 60, polysorbate-65, polysorbate-80 polysorbate-85, poloxamer-188, sorbitan monolaurate, sorbitan monopalmitate, sorbitan monostearate, sorbitan monooleate, sorbitan trilaurate, sorbitan tristearate, sorbitan trioleaste, or a combination thereof. The composition may also include a tonicity adjusting agent, e.g., a compound that renders the formulation substantially isotonic or isoosmotic with human blood. Exemplary tonicity adjusting agents include sucrose, sorbitol, glycine, methionine, mannitol, dextrose, inositol, sodium chloride, arginine and arginine hydrochloride. In other embodiments, the composition additionally includes a stabilizer, e.g., a molecule which substantially prevents or reduces chemical and/or physical instability of the nanostructure, in lyophilized or liquid form. Exemplar}' stabilizers include sucrose, sorbitol, glycine, inositol, sodium chloride, methionine, arginine, and arginine hydrochloride.

The polypeptide, fusion protein, nucleic acid, expression vector, and/or host cell maybe the sole active agent in the composition, or the composition may further comprise one or more other agents suitable for an intended use. In various non-limiting embodiments, such other agents may include angiogenesis inhibitors (including but not limited to) axitinib, bevacizumab, cabozantinib, everolimus, lenalidomide, lenvatinib mesylate, pazopanib, ramucirumab, regorafenib, sorafenib, sunitinib, thalidomide, vandetanib, and ziv-aflibercept), immune checkpoint inhibitors (including, but not limited to, pembrolizumab, nivolumab, and cemiplimab as anti-PD-1 antibodies, ipilimumab as an anti-CTLA-4 antibody, and atezolizumab, avelumab, and durvalumab as anti-PD-Ll antibodies), and other cancer growth inhibitors including but not limited to tyrosine kinase inhibitors (including but not limited to alectinib, brigatimb, ceritinib, crizotinib, entrectinib, lorlatinib, ALK, I, bosutinib, dasatinib, imatinib, nilotinib, ponatinib, afatinib, dacomitinib, erlotinib, gefitinib, lapatinib, neratinib, osimertinib, vandetanib, gilteritinib, rnidostaurin, erdafitinib, mxolitinib, larotrectinib, axitinib, carbozantinib, lenvatinib, pazopanib, regorafenib, sorafenib, sunitinib, dabrafenib, encorafenib, vemurafenib, acalabrutinib, ibrutinib, binimetinib, cobimetinib, trametinib, abemaciclib, palbociclib, or ribociclib), proteasome inhibitors (including but not limited to ortezomib, carfizomib, ixazomib, delanzomib, oprozomib, and manzomib), mTOR inhibitors (including but not limited to everolimus, sirolimus, temsirolimus, everolimus, sirolimus, sirolimus protein-bound, and everolimus), PI3K inhibitors (including but not limited to copanlisib, alpelisib, idelalisib, duvelisib and umbralisib), histone deacetylase inhibitors (including but not limited to vorinostat, romidepsin, panobmostat, and belinostat), and Hedgehog pathway blockers (including but not limited to vismodegib, sonidegib, and glasdegib).

In another aspect, the disclosure provides methods for treating cancer, comprising administering to a subject in need thereof an amount of the polypeptide, fusion protein, nucleic acid, expression vector, host cell, and/or pharmaceutical composition of any preceding claim effective to treat the cancer. Exemplary tumor types that can be treated with the CTLA-4 binders of the disclosure include, but are not limited to, melanoma, lung cancer (such as non-small cell lung cancer), bladder cancer, head and neck cancer, renal cell carcinoma, ovarian cancer, and colorectal cancer. Tirus, in one embodiment the Exemplary' tumor types that can be treated with the PD-L1 binders of the disclosure include, but are not limited to, non-small cell lung cancer, melanoma, bladder cancer, head and neck squamous cell carcinoma, Hodgkin lymphoma, renal cell carcinoma, gastric or gastroesophageal junction adenocarcinoma, cervical cancer, and breast cancer (including but not limited to triple-negative breast cancer).

In these embodiments tire polypeptides may be used, for example, in in treating cancer and tissue fibrosis, by interfering with the TGF~p pathway. Exemplaiy tumor types that can be treated with the TGFbRII binders of the disclosure include, but are not limited to, pancreatic cancer, breast cancer, colorectal cancer, lung cancer, prostate cancer, liver cancer, gastric cancer, ovarian cancer, melanoma, and glioblastoma.

Exemplary’ uses for treating tissue fibrosis include, but are not limited to, idiopathic pulmonary fibrosis (IPF), cirrhosis of the liver, renal fibrosis, cardiac fibrosis, scleroderma, keloids, and hypertrophic scarring.

As used herein, "treat" or "treating" means accomplishing one or more of the following: (a) reducing the severity of the disorder; (b) limiting or preventing development of symptoms characteristic of the disorder(s) being treated; (c) inhibiting worsening of symptoms characteristic of the disorder! s) being treated; (d) limiting or preventing recurrence of the disorder! s) in patients that have previously had the disorder(s); and (e) limiting or preventing recurrence of symptoms in patients that were previously symptomatic for the disorder(s).

When the method comprises treating cancer, "treat" or "treating" means accomplishing one or more of the following: (a) reducing the size or volume of tumors and/or metastases in the subject; (b) limiting any increase in the size or volume of tumors and/or metastases in the subject; (c) increasing survival; (d) reducing the severity of symptoms associated with cancer; (e) limiting or preventing development of symptoms associated with cancer; and (f) inhibiting worsening of symptoms associated with cancer.

Hie subject may be any subject that has a relevant disorder. In one embodiment, the subject is a mammal, including but not limited to humans, dogs, cats, horses, cattle, etc.

As used herein, an "‘effective” amount refers to an amount of the polypeptide, fusion protein, nucleic acid, expression vector, and/or host cell that is effective for treating the disorder. The polypeptides, fusion proteins nucleic acids, expression vectors, and/or host cells are typically formulated as a pharmaceutical composition, such as those disclosed above, and can be administered via any suitable route, including but not limited to orally, by inhalation spray, ocularly, intravenously, subcutaneously, intraperitoneally, and intravesicularly in dosage unit formulations containing conventional pharmaceutically acceptable carriers, adjuvants, and vehicles.

Any suitable dosage range may be used as determined by attending medical personnel. Dosage regimens can be adjusted to provide the optimum desired response. A suitable dosage range for the polypeptides or fusion proteins may, for instance, be 0. 1 ug/kg- 100 mg/kg body weight; alternatively, it may be 0.5 ug/kg to 50 mg/kg; 1 ug/kg to 25 mg/kg, or 5 ug/kg to 10 mg/kg body weight. In some embodiments, the recommended dose could be lower than 0.1 mcg/kg, especially if administered locally (such as by mtra-tumoral injection). In other embodiments, the recommended dose could be based on weight/m² (i.e. body surface area), and/or it could be administered at a fixed dose (e.g., .05-100 mg). The polypeptides, fusion proteins, nucleic acids, expression vectors, and/or host cells can be delivered in a single bolus, or may be administered more than once (e.g., 2, 3, 4, 5, or more times) as determined by an attending physician.

The polypeptides, fusion proteins, nucleic acids, expression vectors, and/or host cells made be administered as the sole therapeutic agent, or may be administered together with (i.e.: combined or separately) one or more other therapeutic agents, including but not limited to tumor resection, chemotherapy, radiation therapy, and immunotherapy (such as checkpoint inhibitors). In various non-limiting embodiments, other agents that may be administered in the methods of the disclosure include angiogenesis inhibitors (including but not limited to) axitinib, bevacizumab, cabozantinib, everolimus, lenalidomide, lenvatmib mesylate, pazopanib, ramucirumab, regorafenib, sorafenib, sunitinib, thalidomide, vandetanib, and ziv-aflibercept), immune checkpoint inhibitors (including, but not limited to, pembrolizumab, nivolumab, and cerniplimab as anti-PD-1 antibodies, ipilimumab as an anti- CTLA-4 antibody, and atezolizumab, avelumab, and durvalumab as anti-PD-Ll antibodies), and other cancer growth inhibitors including but not limited to tyrosine kinase inhibitors (including but not limited to alectinib, brigatinib, ceritinib, crizotinib, entrectinib, lorlatinib, ALK, 1, bosutinib, dasatinib, imatinib, nilotinib, ponatinib, afatinib, dacomitinib, erlotinib, gefitmib, lapatinib, neratmib, osimertinib, vandetanib, gilteritinib, midostaurin, erdafitinib, ruxolitinib, larotrectinib, axitinib, carbozantinib, lenvatinib, pazopanib, regorafenib, sorafenib, sunitinib, dabrafenib, encorafenib, vemurafenib, acalabrutinib, ibrutinib, binimetinib, cobimetinib, trametmib, abemaciclib, palbocichb, or ribociclib), proteasome inhibitors (including but not limited to ortezormb, carfizomib, ixazomib, delanzomib, oprozomib, and marizomib), mTOR inhibitors (including but not limited to everolimus, sirolimus, ternsirolimus, everolimus, sirolimus, sirolimus protein-bound, and everolimus). PI3K inhibitors (including but not limited to copanlisib, alpelisib, idelal i sib, duvelisib and ombralisib), histone deacetylase inhibitors (including but not limited to vorinostat, romidepsin, panobinostat, and belinostat), and Hedgehog pathway blockers (including but not limited to vismodegib, sonidegib, and glasdegib).

Examples

Here, we describe a general approach to generate computationally designed proteins which bind to convex target sites that employ geometrically matching concave scaffolds. We used this approach to design proteins binding to TGFpRII, CTLA-4 and PD-L1 which following experimental optimization have low nanomolar to picomolar affinities and potent biological activity. Co-crystal structures of the TGFpRII and CTLA-4 binders in complex with the receptors are in close agreement with the design models. Our approach provides a general route to generating very high affinity binders to convex protein target sites.

We reasoned that to enable systematic design of binders to convex protein target sites, it would be necessary' to generate scaffold sets with overall concave shapes. Three additional properties would further facilitate binder design and characterization. First, varying curvature: protein surfaces vary considerably in shape, and to enable close complementary matching of a wide range of targets, a set of proteins with varying curvature and surface topography would be ideal. Second, high stability: the higher the stability of the base scaffold, tire more room for customizing the binding interface for high affinity binding, and the more robust the resulting binders. Third, minimal size: to make design cost effective for gene synthesis and generation of oligonucleotide libraries for initial screening, and for applications such as tumor penetration in oncology, the overall length of the binder scaffolds should be minimal (80-120 aa). With these selection criteria, we set out to construct a set of scaffolds.

Results

Computational design of 5HCS scaffolds

We reasoned that scaffolds resembling designed helical repeat proteins (DHRs) but with fewer helices could satisfy the four target properties (concave interaction surface, tunable curvature, high stability, and minimal size). We focused on 5 helix bundle scaffolds with three helices forming the concave interface and two helices providing structural support (Fig. 1 c). Scaffolds were generated using a library' of ideal helical and loop fragments combined first to create helix-tum-helix-tum modules. The length of each helix was constrained to 18 to 22 amino acids (5 to 6 helical turns) balancing stability and overall length constraints. These modules were repeated 3 times to generate 3 unit repeat proteins, and either the N- or C-terminal helix was truncated to generate five helix proteins with fewer than 120 amino acids. We evaluated the curvatures of the surfaces formed by the three interfacial helices and filtered out the backbones with convex surfaces. To more extensively diversify the backbones and break the repeat symmetry, we remodeled the backbones by randomly replacing short scaffold structural elements with alternative local structures (fragments) from the PDB followed by Cartesian-space minimization and full-atom optimization. Following Roseta^1M sequence design, designs with sequences predicted to fold to the designed structure with AlphaFokl2^IM and to have high accuracy using DeepAccNet™ were selected (Fig. 5). The selected 7476 scaffolds, which we refer to as 5HCS (5 helix concave scaffolds) throughout the remainder of the text, have a wide range of curvatures (Fig. 5).

To guide selection of representative convex targets for 5HCS scaffolds, we systematically analyzed the convexities of protein-protein interfaces from the PDB. Pairs of interacting proteins were extracted from PDB entries with multiple chains and were grouped into 2411 clusters. We calculated the convexities of representatives of each cluster by fitting the interfacial heavy atoms to spherical surfaces using the random sample consensus (RANSAC) algorithm. Because of the overall shape matching constraint, the convexity of the two binding partners for each complex are negatively correlated: when one partner is convex, the other is almost always concave (Fig. la). The convexities of the 5HCS scaffolds covers the range of convexities we analyzed from PDB (Fig. la).

Design and structural validation of TGFpRII binders

We selected as a representative convex target site that of the transforming growth factor-p3 (TGF-p3) on the TGF~p receptor type-2 (TGFpRII) (PDB ID: 1KTZ). Binders to modulate TGF~P pathways have considerable interest as therapeutics in oncology, tissue fibrosis, and other areas’ ^!. We used the RIF based docking protocol of Cao et. Al. to dock both the 5HCS scaffolds described above and a globular miniprotein scaffold library’ used in the previous studies^2,12 to the TGF-P binding site (Fig. 9a). Following design and filtering for binders with the concave surface of the 5HCS interacting with the target, and AIphafold2^1M based confirmation of structure and binding mode, we encoded 67 designs using oligonucleotide arrays and cloned into a yeast, surface-expression vector to enable high throughput assessment of binding affinity. After two rounds of fluorescent activated cell sorting for binding to biotinylated TGFpRII, sequencing revealed 2 5HCS hits but no mini- protein hits despite the nearly 100-fold greater representation of the latter in the library (see Methods). The sequences of the two hits are identical to the two designed sequences. We further optimized the most enriched design 5HCS_TGFBR2_0, by resampling the sequences of interfacial residues in the bound state using ProteinMPNN^TM and filtering the complex models using Alphafold2^TM. We combined the mutations predicted to improve binding affinities and encoded a combinatorial library with these mutations included using degenerate codons (Fig.6a). Finally, we sorted the library for the optimized binders using yeast display selection . Four of the optimized binders obtained after several rounds of yeast display selection were produced in E. coli. The highest affinity binder, 5HCS_TGFBR2_1, was found using biolayer interferometry to have an affinity less than 1nM for TGFȕRII (Fig.2c, Fig.8a). The sequence identity between 5HCS_TGFBR2_0 and 5HCS_TGFBR2_1 is 88.12% (Fig.7a). The circular dichroism spectra indicates a helical structure with peaks at 208 nm and 222 nm, consistent with the design model (Fig.2a,b), and was only slightly changed by heating to 95 °C, indicating high stability (Fig.2b). We determined co-crystal structures of 5HCS_TGFBR2_1 with TGFȕRII. The high resolution (1.24 Å) X-ray crystal structure is very close to the computational design model (Fig.2f, g; root mean square deviation (rmsd) over CĮ atoms of 0.55 Å over the full complex), showing 5HCS_TGFBR2_1 binds to the TGF-ȕ3 binding site on TGFȕRII utilizing the concave surface as designed. To further investigate the sequence dependence of folding and binding, we generated site saturation mutagenesis (SSM) libraries in which each residue was substituted with all other nineteen amino acids one at a time, and sorted the library using fluorescence-activated cell sorting (FACS) with fluorescent TGFȕRII. Deep sequencing results were closely consistent with the design model and crystal structure. Both the core residues and interfacial residues were highly conserved, while surface residues not at the interface were quite variable (Fig.2 a,e). Helices H1, H3 and H5 which form the concave binding surface interact with TGFȕRII, and the most highly conserved non-core residues are in these helices. In H1, N10 hydrogen bonds with TGFȕRII D142 (Fig 2g, top panel); in H3, S46 and S49 hydrogen bond to the backbone atoms of strand S72 - S75 (Fig 2g, middle panel); and in H5, N93 hydrogen bonds to the backbone atoms of I76 (Fig 2g, lower panel). A hydrophobic patch composed of F48, L50 and I76 on TGFȕRII critical for TGF-ȕ3 binding packs tightly on a hydrophobic groove formed by L6 from H1, M50, V52, K53 from H3 and V96, K99, V100 from H5. All the key interactions described above are recapitulated in the crystal structure with high side- chain orientation consistency (Fig.2f,g). Design of such extended grooves and pockets is nearly impossible using small globular miniproteins; the high affinity binding and crystal structure of 5HCS_TGFBR2_1 demonstrates that 5HCS scaffolds can indeed be used to target convex binding sites. We assessed the biological activities of 5HCS_TGFBR2_1 in cell culture signaling assays. HEK293 cells with luciferase reporter for the TGFȕ SMAD2/3 signaling pathway were stimulated using 10 pM TGF-ȕ3 and varying concentrations of 5HCS_TGFBR2_1. Dose-dependent inhibition of the TGFȕ SMAD2/3 signaling was observed with an IC₅₀ of 30.6 nM (Fig.2d). Design and structural validation of CTLA-4 binders An important class of convex targets are the portions of the extracellular domains of transmembrane receptors which interact with their biological partners. These frequently consist of immunoglobulin fold domains, which our large-scale shape analyses indicate are generally quite convex. Immunoglobulin domain recognition plays important roles in immune receptor functions; in particular the cancer immunotherapy target Cytotoxic--T- -lymphocyte--antigen--4 (CTLA-4) have extracellular Ig fold domains that are the targets of therapeutic antibodies. Because of the therapeutic importance of the target, and receptor extracellular Ig domains more generally, we next sought to evaluate the generality of our approach by designing 5HCS based binders to CTLA-4. CTLA-4 plays an important role in peripheral tolerance and the prevention of autoimmune disease by inhibition of T cell activation. Antibody CTLA-4 targeting checkpoint inhibitors have been used for melanoma and non-small cell lung cancer (NSCLC) therapy. We targeted the region surrounding the beta-turn (132-140) of CTLA-4 which is buried in the interface between CTLA-4 and CD86 (PDB ID: 1I85 ) using the methods described above. FACS of yeast libraries displaying 4600 the concave designs identified six CTLA-4 binders. The sequences of the six hits match their designs with 100% sequence identity. Deep sequencing of a site saturation mutagenesis library of the most enriched binder, 5HCS_CTLA4_0, showed that the designed core and interfacial residues of the binder were highly conserved, suggesting the design folds and binds target as in the computational model (Fig.3a,e,). As the Alphafold2^TM predicted models were not consistent with the designed complex model, we combined the most enriched substitutions from the SSM heatmap, instead of using the ProteinMPNN^TM resampling followed by Alphafold2^TM filtering method mentioned above. We synthesized the combinatorial library with these mutations included using degenerate codons (Fig.6b). After additional rounds of yeast display selection of the combinatorial library, we expressed four of the best binders in E. coli. The highest affinity optimized binder, 5HCS_CTLA4_1 has a sequence similarity of 82.86% compared to 5HCS_CTLA4_0 (Fig.7b).5HCS_CTLA4_1 had an off rate too slow and a binding affinity for CTLA-4 too tight (<100 pM) to be measurable by biolayer interferometry (Fig.3c, Fig. S5b). We determined co-crystal structures of 5HCS_CTLA4_1 with CTLA-4 and unbound crystal structures of 5HCS_CTLA4_2. The unbound crystal structure of 5HCS_CTLA4_2 aligns with the structure of binder in 5HCS_CTLA4_1 bound structure well with a rmsd. of 0.416 Å. The design model of 5HCS_CTLA4_1 in complex with CTLA-4 also closely agrees with the crystal structure, with a very low rmsd of 0.34 Å (Fig 3f,g). 5HCS_CTLA4_1 binds to the CD86 binding site on CTLA-4 using a concave binding surface formed by H1, H3 and H5 covering both the CTLA-4 beta-turn (L98 to Y104) and hydrophobic pocket which interacts with CD86. H1 interacts with the hydrophobic beta-turn (L128 to Y136) through hydrophobic interactions between Y18 and M135 and aromatic interactions between H19 and Y136 (Fig.3g, top panel). Substitution of this residue with H or Y improves binding affinity (Fig 3E). S54 and I55 on H3 interact with Y139 on CTLA-4 (Fig 3g, middle panel), and N89 on H5 hydrogen bonds with Q90 on CTLA-4. (Fig.3g, lower panel). All of these interactions are closely recapitulated in the crystal structure (Fig 3g). The circular dichroism spectra indicates a helical structure with peaks at 208 nm and 222 nm consistent with the design model and was unchanged by heating to 95 °C, indicating high thermal stability (Fig.3b). We tested the biological activity of 5HCS_CTLA4_1 in cell culture using an immune checkpoint functional assay in which stably expressing CTLA-4 Jurkat cells with a luciferase reporter for TCR/CD28 activation were incubated with activating Raji cells expressing the CTLA-4 ligands CD80 and CD86. Inhibition of the inhibitory CTLA-4-CD86 interaction results in TCR pathway activation, and hence can be directly read out using this assay. We co-cultured the cells with a range of concentrations of the CTLA–4 binder, and observed dose-dependent activation of CTLA-4 effector cells with an EC50 of 53.3 nM (Fig.3d). Surprisingly, this is higher than the EC50 (15.0nM) of the anti-CTLA-4 antibody ipilimumab (MDX-010, Yervoy), despite the at least two order weaker binding affinity for CTLA-4 (18.2 nM). Steric or avidity effects may contribute to the potency of the antibody, which can interact with two receptors through the two Fabs. To explore the effect of avidity, we flexibly fused 5HCS_CTLA4_1 to previously designed domains which oligomerize into different symmetric architectures. We found that a highly expressed and monodisperse hexameric version, 5HCS_CTLA4_1_c6 had an EC50 of 16.1 nM, comparable to the antibody (Fig.3d). Design and structural validation of PD-L1 binders Programmed death-ligand 1 (PD-L1), is upregulated on many tumors, and interacts with PD-1 on T-cells to downregulate T-cell activation. Therapeutic antibodies against PDL1 have shown considerable promise for checkpoint inhibition in cancer immunotherapy. Considering the therapeutic importance of the target, and to test the generalizability of our approach towards flatter protein surfaces, we designed binders using the methods described above to target the binding site of PD-1 on PD-L1 (PDB ID:3BIK ) and block the interaction between the two proteins (Fig.1a). Two PD-L1 binders were obtained from a set of 96 designs. We optimized the stronger binder, 5HCS_PDL1_0, by resampling the residues at the designed interface using ProteinMPNN^TM followed by Alphafold2^TM filtering. We used yeast display to sort a library with degenerate codons encoding mutations (Fig.6) predicted to improve binding, expressed ten of the most enriched binders in E. coli, and measured their binding affinities by biolayer interferometry. The highest affinity binder, 5HCS_PDL1_1, which has 93.2% similarity with the sequence of 5HCS_PDL1_0 (Fig. S4c), is expressed at high levels, very stable (Fig 4b) and has an affinity of 646 pM (Fig.4c, Fig.8c). To examine the sequence determinants of folding and binding of 5HCS_PDL1_1 and to provide a structural footprint of the binding site, we generated a SSM library and sorted the library using FACS with fluorescent PD-L1. The conservation of both the core residues and the interfacial residues (Fig.4a,e,) suggests the binders fold and bind to the models as designed. As with 5HCS_TGFBR2_1 and 5HCS_CTLA4_1, the interfacial helices H1, H3 and H5 of 5HCS_PDL1_1 have an overall concave shape (Fig.4a). The key interactions between H1 and PD-L1 include aromatic packing of Y9 and Y123 on PD-L1 and electrostatic interactions between D10 and E13 with K124 and R125 on PD-L1 (Fig.4h). H3 binds to the hydrophobic pocket formed by Y56, M115, A121 and Y123 (Fig.4h). Residues Y9, E13, K56 and Q99 spanning the three helices satisfy the hydrogen bonding requirements of both the side chains and backbone of the PD-L1 edge beta strand (A121-R125) buried at the interface. We solved the high resolution crystal structure of 5HCS_PDL1_1. The refined structure has excellent geometry and reveals the expected helical assembly with five antiparallel helices (Fig.4g). The crystal structure of 5HCS_PDL1_1 superimposes on the computational design model with a rmsd of 0.75 A over 105 aligned Ca atoms (Fig. 4g; the substitutions which increase affinity relative to 5HCS PDL1 0 do not alter the backbone structure). Not surprisingly, given the near identity' between the computational designs (Fig. 7c) and the crystal structures, the shape and electrostatic potential of the designed target binding interfaces are nearly identical between the crystal structure and the computational design model.

We solved the high resolution crystal structure of 5HCS PDL1 1 . The refined structure has excellent geometry' and reveals the expected helical assembly with five antiparallel helices (Fig. 4g). The crystal structure of 5HCS PDL1 1 superimposes on the computational design model with a rmsd of 0.75 A over 105 aligned Ca atoms (Fig. 4g; the substitutions which increase affinity’ relative to 5HCS PDL1 0 do not alter the backbone structure). Not surprisingly, given the near identity between the computational designs (Fig. 7c) and the crystal structures, the shape and electrostatic potential of the designed target binding interfaces are nearly' identical between the crystal structure and the computational design model.

Discussion

Our method for computationally designing small but concave proteins to bind to convex protein target sites expands the space of the protein surfaces that can be targeted by de novo protein design. Despite their relatively small size (120 residues), the 5HCS scaffolds span a wide range of concave shapes and have high stability' (Fig, 2 b, Table 5). The designed surfaces are more extensive and more concave than those obtained using previous miniprotein binder approaches: the curvature reaches -0.067 while minibinders range from -0.012 to 0.073 (more negative indicates greater curvature) and the largest distance between target bound hot-spot residues is 33 A while minibinders range from 15 A to 20 A. Because of this, the 5HCS Hl, H3 and H5 interface helices interact with hydrophobic pockets and patches on tire target surface in ways not possible with 50-65 residue miniprotein scaffolds in which the secondary’ structure elements at the interface are necessarily all very close together. As illustrated by the binders designed to TGFpRII and PD-L1 , the dense and extended networks of hydrogen bonding residues that the 5FICS designs are able to satisfy the hydrogen bonding requirements of exposed target beta-strand backbone polar atoms, which enables binding modes which span both sides of the beta sheet; this is almost impossible to achieve with smaller miniproteins. Tire 5HCS binders can interact with beta-stands either parallelly using helix H3 with Hl and H5 flanking the sheet (5HCS TGBR2 1) or perpendicularly with sides chains from all HI, H3 and H5 (5HCS PDL1 1).

Table 5. Physicochemical properties and interface profiles of the optimized de novo 5HCS binders

The high affinity and potent signaling pathway modulation possible with the TGFpRIL PD-L1, and CTLA-4 convex binders described here demonstrate the considerable potential of our approach for targeting critical cell surface receptors. The current designs provide new routes for manipulating signaling and checkpoint blockade to be explored in future in vivo studies, and more generally our approach considerably expands the scope of de novo binder design.

Methods and Protocols:

5HCS Scaffolds Library Design

Backbone generation. The backbones were designed by taking a library' of loops and helices drawn from previous successfill mini-proteins and assembling them into helix-tum- helix-tum modules of 30-50 amino acids. The modules were then repeated 3 times to give a repeat protein. Ail possibilities of N- and C- terminal truncation were assessed and the most concave compact structure under 120 amino acids was chosen . The backbones were diversified using the Rosetta^1M HybrizeMover using the backbones themselves as templates.

Sequence design and filtering. The generated backbones were designed using standard Rosetta™ LayerDesign protocol. The heavy atoms from the residues at the concave surfaces were selected by secondary structure and Rosetta^IM layerselector. RANSAC was used to fit spherical surfaces from the coordinates of the interfacial atoms with a threshold of 1 A and max iteration 100k. 'The algorithm was implemented by python. By definition, convexity of the surface is the reciprocal of the radius. The designed scaffolds were later filtered by the AlphaFold2™ wi th a mean plDDT cutoff of 80 and AccNet wi th a mean plDDT cutoff of 0.8. There are finally 7476 scaffolds meeting all the criteria. Protein Surface Convexity Calculation

Protein complex structure extraction. Pairs of interacting chains were extracted from high quality crystals from PDB. The pairs of protein complex structures were filtered by interfacial profiles, including the length of each partner's and delta solvent accessible surface area (dSASA). Then we clustered them 40% sequence identity on both chains, and selected representatives favoring higher resolution and shorter proteins.

Convexity Calculation. We calculated atomic SASA for both partners from proteinprotein complex pairs in apo and holo structures. Heavy atoms with a difference of 0.5 A² are defined as interfacial residues. RANSAC was used to fit spherical surfaces from the coordinates of the interfacial atoms with a threshold of I A and max iteration 100k. The algorithm was implemented by python. By definition, convexity of the surface is the reciprocal of the radius.

Concave and Convex Definition. To define whether the surface is concave or convex, the geometry centers of heavy atoms of the proteins and interfacial atoms were firstly calculated. The inner product of interfacial atoms centers to protein centers and interracial atoms to fitted centers was calculated. Those surfaces with minus results are defined concave, vice versa.

Interface Design and Filtering

Docking and Interface design. For TGF pR 11 binder design, the 5HCS or mini protein libraries were docked to the target-binding site using the previously reported method³. Docked poses of the 5HCS library were filtered by binding orientation. Only designs with interfacial residues as the concave surfaces were kept. Interface sequence design was performed using ProteinMPNN^IM with target sequences fixed as native sequences as previously reported. The designs were later filtered by ddG (less than -40), contact molecular surface (larger than 400) and pAE (less than 10) from AlphaFold2™ initial guess. Finally, 67 and 4310 designs from 5HCS and mini protein libraries passed the filters and were tested experimentally, respectively .

For CTLA-4 binder design, the 5HCS libraries were docked to the target-binding site using the previously reported method³. Docked poses of the 5HCS library were filtered by binding orientation. Only designs with interfacial residues as the concave surfaces were kept. Interface sequence design was performed using previously reported protocol. Tire designs were later filtered by ddG (less than -40), contact molecular surface (larger than 400). Finally, 4600 designs from 5HCS passed the filters and were tested experimentally.

For PD-L1 binder design, the 5HCS libraries were docked to the target binding site using the previously reported method³. Docked poses of the 5HCS library were filtered by binding orientation. Only designs with interfacial residues as the concave surfaces were kept. Interface sequence design was performed using ProteinMPNM™ with target sequences fixed as native sequences as previously reported. The designs were later filtered by ddG (less than - 40), contact molecular surface (larger than 400) and pAE (less than 10) from AlphaFold2^1M initial guess. Finally, 96 designs from 5HCS libraries passed the filters and were tested experimentally.

Combinatorial Library Design. The hits screened from the initial designs were further optimized by the virtual optimization protocol. Interfacial residues were re-sampled massively (5000 replicates) using ProteinMPNN™ with a higher temperature of 0.4. As the binding pattern stays mostly the same, the re-sampled designs were later assessed by delta ddG predicted by AlphaFold2^1M initial guess. Designs with lower ddG than the initial hits were aligned by primary sequences. At each residue position, the more times of one type of mutation showed up the more likely the mutation will improve the binding affinity. We then ordered Ultramer™ oligonucleotides (Integrated DNA Technologies) containing the degenerate codons for the mutations predicted to be beneficial. The constructed libraries were transformed into Saccharomyces cerevisiae EBY 100. The transformation efficiencies were around 10⁷.

Yeast Surface Display

Saccharomyces cerevisiae EBY 100 strain cultures were grown in C-Trp-Ura medium supplemented with 2% (w/v) glucose. For induction of expression, yeast cells were centrifuged at 4,000g for 1 min and resuspended in SGCAA medium supplemented with 0.2% (w/v) glucose at the cell density of 1 x 10⁷ cells per ml and induced at 30 °C for 16 -24 h. Cells were washed with PBSF (PBS with 1% (w/v) BSA) and labeled with biotinylated targets using two labeling methods: with-avidity and without-avidity labeling. For the with- avidity method, the cells were incubated with biotinylated target, together with anti-c-Myc fluorescein isothiocyanate (FITC, Miltenyi Biotec) and streptavidin-phycoeiythrin (SAFE, ThermoFisher). The concentration of SAPE in the with-avidity method was used at one- quarter of the concentration of the biotinylated targets. For the without-avidity method, the cells were first incubated with biotinylated targets, washed and secondarily labeled with SAPE and FITC. AH the original libraries of de novo designs were sorted using the with- avidity method for the first few rounds of screening to exclude weak binder candidates, followed by several without-avidity sorts with different concentrations of targets. For SSM libraries, two rounds of without-avidity sorts were applied and in the third round of screening, die libraries were titrated with a series of decreasing concentrations of targets to enrich mutants with beneficial mutations. The combinatorial libraries were enriched at medium concentration of target for two rounds by collecting the top 10% of the binding population. In the third round of sorting, the enriched library was titrated to with a series of decreasing concentrations of targets. The several binding populations with lowest concentration of target were collected.

Protein Expression and Purification

Synthetic genes were optimized for E. colt expression and purchased from IDT (Integrated DN A Technologies) as plasmids in pET29b vector with a TEV -cleavable hexahistidine affinity tag. Plasmids were transformed into BL21* (DE3) E. coli competent cells (Invitrogen). Single colonies from agar plate with 100 ing/L kanamycin were inoculated in 50 mL of Studier autoinduction media 45, and the expression continued at 37 °C for over 24 hours. The cells were harvested by centrifugation at 4000 g for 10 min, and resuspended in a 35 mL lysis buffer of 300 mM NaCI, 25 mM Tris pH 8.0 and 1 mM PMSF. After lysis by sonication and centrifugation at 14000 g for 45 min, the supernatant was purified by Ni^?i immobilized metal affinity chromatography (IMAC) with Ni-NTA Superflow^1M resins (Qiagen). Resins with bound cell lysate were washed with 10 mL (bed volume 1 mL) of washing buffer (300 mM NaCI, 25 mM Tris pH 8.0, 60 mM imidazole) and eluted with 5 mL of elution buffer (300 mM NaCI, 25 mM Tris pH 8.0, 300 mM imidazole). Both soluble fractions and full cell culture were checked by SDS-PAGE. Soluble designs were further purified by size exclusion chromatography (SEC). Concentrated samples were run in 150 mM NaCI, 25 mM Tris pH 8.0 on a Superdex™ 75 Increase 10/300 gel filtration column (Cytiva). SEC-purified designs were concentrated by 10K concentrators (Ami con) and quantified by UV absorbance at 280 nm.

Biolayer interferometry

Binding assays were performed on an OctetRED96^1M BLI system (ForteBio) using streptavidin-coated biosensors. Biosensors were equilibrated for at least 10 min in Octet^1M buffer (10 mM Hepes pH 7.4, 150 mM NaCI, 3 mM EDTA, 0.05% Surfactant P20) supplemented with 1 mg/mL bovine serum albumin (SigmaAldrich). For each experiment, the biotinylated target protein was immobilized onto the biosensors by dipping the biosensors into a solution with 50 nM target protein for 200 to 500 s, followed by dipping in fresh Octet^TM buffer to establish a baseline for 200 s. Titrations were executed at 25 °C while rotating at 1,000 rpm. Association of designs to targets on the biosensor was allowed by dipping biosensors in solutions containing designed proteins diluted in Octet^TM buffer for 800 to 3600 s. After reaching equilibrium, the biosensors were dipped into fresh buffer solution in order to monitor the dissociation kinetics for 800 to 3600 s. For binding titrations, kinetic data were collected and processed using a 1:1 binding model using the data analysis software 9.1 of the manufacturer. Global kinetic fitting using three concentration data was performed for KD calculations. Circular dichroism Far-ultraviolet circular dichroism measurements were carried out with a JASCO-1500 instrument equipped with a temperature-controlled multi-cell holder. Wavelength scans were measured from 260 to 190 nm at 25 and 95 °C and again at 25 °C after fast refolding (about 5 min). Temperature melts monitored the dichroism signal at 222 nm in steps of 2 °C min–1 with 30 s of equilibration time. Wavelength scans and temperature melts were performed using 0.3 mg ml^–1 protein in PBS buffer (20 mM NaPO₄, 150 mM NaCl, pH 7.4) with a 1 mm path-length cuvette. Cell assays TGF-ȕ luciferase reporter assay. The TGF-ȕ inhibition assays utilizing HEK-293 cells stably transfected with the CAGA₁₂ TGF-ȕ reporter²³ were performed as previously described²⁴. Cells were maintained in DMEM containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Cells were plated at 3x10⁴ cells per well in a treated 96-well plate. After 24 hours, the media was removed and replaced with fresh DMEM containing 0.1% bovine serum albumin (BSA) and a two-fold concentration series of 5HCS_TGFȕR2_1. After 30 minutes, cells were stimulated with 10 pM TGF-ȕ3. Twenty-four hours after stimulation, the cells were lysed and luciferase activity was measured using luciferin. The measurements for each condition were made in triplicate. IC50 values were calculated using the four parameters logistic regression by python scripts. CTLA-4 blockade cell assay. The CTLA-4 Blockade Bioassay (Promega) was used as described in the product literature to compare bioacitivity of our novel high affinity CTLA-4 binders with Ipilimumab. Briefly, 25 uL of CTLA-4 effector cells prediluted into complete RPMI media supplemented with 10% FBS were added to wells of a 96-well flat- bottomed white luminescence plate (Costar). In a separate 96-well assay plate, antibodies and binding reagents to be tested were serially diluted into RPMI media at three times the intended final concentration. Activity of the CTLA-4 binders was compared to a control hIgG antibody (Biosciences) and the FDA-approved anti-CTLA-4 mAb ipilimumab. From this 25uL of each diluted reagent was transferred to the wells containing CTLA-4 effector cells and subsequently 25uL of the aAPC/Raji Cells were also added. The resulting reactions were incubated for 16 hours at 37C in a humidified CO 2 incubator. After incubation, 75uL of prepared Bio-Glo^TM reagent (Promega) was added to each well, incubated for 5min at room temperature with gentle shaking at 300 rpm and luminescence measured on an Envision plate reader (Perkin Elmer). The raw luminescence data was normalized using the following formula: (RLU signal–background)/(RLU no antibody–background), where the background and no antibody control values were each calculated from an average of three wells with no cells or cells but no antibody respectively. EC50 values were calculated using the four parameters logistic regression by python scripts. PD-L1 blockade cell assay. The assays were performed according to manufacturer’s instructions (Promega). Briefly, PD-L1 aAPC/CHO-K1 cells were thawed in a 37 ^ water bath until just thawed and transferred to pre-warmed media (90% Ham’s F12 / 10% FBS). Cells were mixed and immediately seeded to the inner 60 wells of a 96 well flat bottom white cell culture plates at 100 ul volume.100 ul of media was also added to the outside wells to prevent evaporation. Cells were incubated for 16 hours in a 37 ^, 5% CO^ incubator. At the end of the incubation period, 95 ul of media was removed from each of the wells. Immediately after 40 ul of appropriate antibody or binder dilutions were added to individual wells. PD-1 effector cells were thawed in similar fashion as for PD-L1 aAPC/CHO-K1 cells and transferred to pre-warmed assay buffer (99% RPMI 1640 / 1% FBS).40 ul of PD-1 effector cells were added to the inner 60 wells of the assay plate.80 ul of assay buffer was added to outside wells to prevent evaporation. The assay plate was incubated for 6 hours in a 37 ^, 5% CO^ incubator. At the end of incubation plates were removed from the incubator and equilibrated to ambient temperature (22~25 ^).80 ul of Bio-Glo^TM reagent was added to each well and incubated for 10 mins. Luminescence was measured using the BioTek^TM Synergy Neo2^TM multi-mode reader. EC50 values were calculated using the four parameters logistic regression by python scripts. Specificity Determination Cell surface receptor knockouts A431 cells had PD-L1 knocked out via CRISPR RNP transfection. RNP complexes were formed by incubating 4 ul of 80 uM guide RNA (IDT guides: Hs.Cas9.CD274.1.AA, Hs.Cas9.CD274.1.AB) with 4 ul of 80 uM tracrRNA (IDT cat.1072533) at 37°C for 30 minutes. To generate complete RNPs, 4 ul of 40 uM guide complex was incubated with 4 ul of 40 uM cas9-NLS (Berkeley MacroLab) at 37°C for 30 minutes. For electroporation, 2x10⁵ cells of each cell type in 20 ul of electroporation buffer (Lonza, cell line SF for A431) were mixed with 1 ul of electroporation enhancer (IDT cat.1075916) and 2 ul of assembled RNPs prior to loading 20 ul into an electroporation cuvette strip (Lonza cat. V4XC-2032). Cells were electroporated with appropriate settings (A431:EQ-100). Cells were immediately rescued with warm complete media and transferred to a 24 well plate to grow after resting for 5 minutes at 37°C with 5% CO2. Cells were tested for knockout efficiency by TIDE analysis. Genomic DNA was extracted with Lucigen Quickextract^TM (Lucigen cat. QE0905T) and amplified with NEBNext^TM high-fidelity polymerase (NEB cat. M0541S). Cellular surface staining A431 cells were stained with 5HCS_PDL1_1 or antibody to compare specificity of de novo binders to commercial antibodies. For staining, 5x10⁵ cells were washed twice with 200 ul cell staining buffer (Biolegend cat.420201) in a 96 well u-bottom plate. Cells were then resuspended in 50 ul of staining mixture (cell staining buffer and fluorophore-conjugated binder or antibody (Biolegend cat.329713) and incubated on ice in the dark for 30 minutes. Cells were washed three times with 200 ul staining buffer and then analyzed on a ThermoFisher Attune. Structure Determination Expression and Purification. The coding sequence for residues 46-155 of human TȕRII (UniProt P37173) was inserted into plasmid pET32a (EMD-Millipore) between the NdeI and HindIII sites without inclusion of any expression tags, transformed into chemically- competent E. coli BL21(DE3) (EMD-Millipore), expressed at 37 °C in the form of insoluble inclusion bodies, and refolded and purified to homogeneity as previously described²⁴. The 5HCS_TGFBR2_1 used for crystallization was prepared as described above, followed by digestion for 12 h at 25°C with TEV protease (1:25 mass ratio) in 25 mM Tris, 100 mM Tris, pH 8.0, 1 mM DTT, 1 mM EDTA. Identity of the isolated protein products was verified by measuring their intact masses, which were found to be within 0.5 Da of the calculated masses (Thermo UltiMate™ UHPLC coupled to Broker Compact QqTOF ESI quadrupole TOF mass spectrometer). The TbRII:5HCS_TGFBR2_l complex was isolated by size exclusion chromatography using a HiLoad Superdex™ 75 26/60 column (GE Healthcare, Piscataway, NJ) in 25mM HEPES pH 7.5, 100 mM NaCl at a 1: 1.1 ratio, with 5HCS TGFBR2 1 being in slight excess. The complex peak fractions were pooled and concentrated to 33 mg/mL for crystal screening.

For large-scale purifications of the CTLA-4 and PD-L1 binders for crystallization, 2-liter bacterial cultures were grown in Super Broth™ (Teknova) media supplemented with antibiotics and antifoam 204 (Sigma) at 37 °C in LEX 48 airlift bioreactors (Epiphyte3, Canada) to an A600 of 3. The temperature was then reduced to 22 °C, isopropyl-P-D- thiogalactoside (1PTG) was added to 0.5 mM, and the cultures were incubated overnight. Cells were harvested by centrifugation at 14,000 x g and suspended in buffer containing 20 mM HEPES (pH 7.5), 500 mM NaCl, 20 mM imidazole, 0.1% IGEPAL, 20% sucrose, 1 mM P-mercaptoethanol (BME). Cells were disrupted by sonication and debris was removed by centrifugation at 45,000 x g. The supernatants were applied to a chromatography column packed with 10 ml His60 SuperFlow™ resin (Clontech Laboratories) that had been equilibrated with buffer A (50 mM HEPES pH 7.5, 30 mM imidazole, 500 mM NaCl, and 1 mM BME). The columns were washed with buffer A and the Hise-binder proteins were eluted with buffer B (20 mM HEPES, pH 7.5, 350 mM NaCl, 400 mM imidazole, and 1 mM BME), The His* tags were removed by overnight digestion at 4 °C with the TEV protease at a 1500: 1 ratio of binder: TEV. Tire tag-free binders were then separated from Hise-tags by Superdex™ 200 gel filtration equilibrated with a buffer containing 20 mM HEPES pH 7.5, 350 mM NaCl . The CTLA-4 and PD-L1 binders migrated through gel filtration as discrete peaks with estimated molecular weights of 14 kDa and 12 kDa, respectively, indicating that they are monomers in solution. The purity of the binders was judged by SDS-PAGE and Coomassie blue staining. The peak fractions from the gel filtration step were pooled and concentrated to 20-25 mg/ml in a buffer containing 20 mM HEPES (pH 7.5) and 150 mM NaCl. 5HCS_CTLA4_l:CTLA-4 complex were purified using size exclusion chromatography (Superdex™ S200) equilibrated with a buffer containing 20 mM HEPES pH 7.5, 150 mM NaCl. The peak fractions were pooled and concentrated to 7.5 mg/ml The preparations were flash frozen in liquid nitrogen and stored at -80 °C for long-term storage.

Protein crystallization and crystal harvesting. Crystals of the TpRII:5HCS__TGFBR2_l complex were formed using hanging drop vapor diffusion in 24-well plates with 300 pL of well solution and siliconized glass cover slips. Crystals formed in 1- 2 days at 25 °C with drops prepared by mixing 0,4 pL 25 mg/mL protein complex and 0.4 pL of 20% (w/v) PEG-MME 5K, 0.4 M (NH02 SO-i, 0.1 M Tris pH 7.4, and 16 - 32 % glycerol. The crystals were mounted in nylon loops without additional cryoprotectants and with excess well solution wicked off.

Screening of 5HCS_CTLA4_2 and 5HCS_PDL1_1 for crystal formation was performed using 0.8 uL (protein : reservoir solution=l: 1) crystallization drops at a concentration of 15 mg/rnl with a Crystal Gryphon (Art Robbins Instruments) robot, using MCSG (Microlytic), Index HT, Crystal Screen HT, and Peg Ion HT sparse matrix crystallization suites (Hampton Research). Initial crystals obtained from the sparse matrix screening were further optimized with several rounds of grid screening using a Forrnulator™ robot (Art Robbins Instruments).

Data collection and processing, structure refinement and analysis.

The diffraction data for the TpRII:5HCS_TGFBR2_l complex was collected at the Southeast Regional Collaborative Access Team (SER-CAT) 22-ID beamline at the Advanced Photon Source, Argonne National Laboratory. The data was integrated with XDS²⁵ and the space group (P2i2i2i with dimensions a,b,c = 47.98 A, 57. 17 A, 78.80 A and a,0,y = 90°, 90°, 90°) was confirmed via pointless²⁶. Hie data, was reduced with aimless²⁷, ctruncate²⁸ and the uniquify script in the CCP4 software suite³³. Phasing was performed with Phaser³⁴, initially with the 1.1 A TpRII X-ray structure (PDB 1M9Z), followed by the predicted 5HCS_TGFBR2_1 structure. Several cycles of refinement using RefinacS³’"⁴² and model building using COOT⁴³ were performed to determine the final structure.

Data from the crystals of CTLA-4 binder were collected on a Dectris Pilatus 6M detector, with a wavelength of 0.98 A, on the ID-31 (LRL-CAT) beamline at the Argonne National Laboratory. Single crystal data were integrated and scaled using iMosflm™ and Aimless™, respectively. Diffraction was consistent with the orthorhombic space group P2i2i2.i and extended to 1 .85 A resolution with one molecule (chain A) in the asymmetric unit. Data for the PD-L1 binder crystals were collected on a Dectris EIGER™ X 9M detector, with a wavelength of 0.92 A, on the 17-ID-l (AMX) beamline at the Brookhaven National Laboratory. Data for the CTLA-4-binder complex crystals were collected on a Dectris EIGER X 9M detector, with a wavelength of 0.98 A, on the 17-ID-2 (FMX) beamline at the Brookhaven National Laboratory. Tire datasets were indexed, integrated, and scaled using fastDP, XDS²³ and aimless⁴⁴, respectively. The PD-L1 crystals belong to tetragonal space group and diffracted to 1.88 A. Hie CTLA-4-binder complex crystal belongs to C2 space group and diffracted to 2.72 A. Initial phases of 5HCS CTL.A4 2, 5HCS PDL1 1 and 5HCS_CTLA4_2:CTLA-4 complex were determined by molecular replacement (MR) with Phaser³⁴, using coordinates of the computationally designed respective binders and binder complex; the initial MR coordinate was manually inspected and corrected using Coot⁴³. The model was refined with Phenix-Refine⁴³. Analyses of the structures were performed in Coot an d evaluated using Mol Probity’⁴”; B-factors were calculated using Baverage program in CCP4 suite⁴⁷. The crystallographic model exhibited excellent geometry with no residues in disallowed regions of the Ramachandran plot⁴⁸. All figures depicting structure were generated with PyMol, unless stated otherwise.

Reference

1. Brady, G. P. & Sharp, K. A. Entropy in protein folding and in protein — protein interactions. Curr. Opm. Struct. Biol. 1, 215—221 (1997).

2. Chevalier, A. el al. Massively parallel de novo protein design for targeted therapeutics. Nature 550, 74-79 (2017).

3. Cao, L. et al. Design of protein-binding proteins from tire target structure alone. Nature 605, 551-560 (2022).

4. Brunette, T. J. et al. Exploring the repeat protein universe through computational protein design. Nature 528, 580-584 (2015).

5. Hicks, D. R. et al. De novo design of protein homodimers containing tunable symmetric protein pockets. Proc. Natl. Acad. Sei. 119, e21 13400119 (2022).

6. Park, H., Ovchinnikov, S., Kim, D. E., DiMaio, F. & Baker, D. Protein homology model refinement by large-scale energy optimization. Proc. Natl. Acad. Set. 115, 3054-3059 (2018).

7. Alford, R. F. et al. Tire Rosetta All-Atom Energy Function for Macromolecular Modeling and Design. J Cheni. Theory Comput. 13, 3031-3048 (2.017).

8. Jumper, J. etal. Highly accurate protein structure prediction with AlphaFold, Nature 596, 583-589 (2021 ).

9. Hiranuma, N. et al. Improved protein structure refinement guided by deep learning based accuracy estimation. Nat. Commun. 12, 1340 (2021).

10. Fischler, M. A. & Bolles, R. C. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM 24, 381— 395 (1981). 11. Derynck, R., Turley, S. J. & Akhurst, R. J. TGFȕ biology in cancer progression and immunotherapy. Nat. Rev. Clin. Oncol.18, 9–34 (2021). 12. Rocklin, G. J. et al. Global analysis of protein folding using massively parallel design, synthesis, and testing. Science 357, 168–175 (2017). 13. Hart, P. J. et al. Crystal structure of the human TbetaR2 ectodomain--TGF-beta3 complex. Nat. Struct. Biol.9, 203–208 (2002). 14. Bennett, N. R. et al. Improving de novo protein binder design with deep learning. Nat. Commun.14, 2625 (2023). 15. Dauparas, J. et al. Robust deep learning–based protein sequence design using ProteinMPNN. Science 378, 49–56 (2022). 16. Cannon, J. P. et al. Recognition of additional roles for immunoglobulin domains in immune function. Semin. Immunol.22, 17–24 (2010). 17. Sharma, P. et al. The Next Decade of Immune Checkpoint Therapy. Cancer Discov.11, 838–857 (2021). 18. Wojtukiewicz, M. Z. et al. Inhibitors of immune checkpoints—PD-1, PD-L1, CTLA-4— new opportunities for cancer patients and a new challenge for internists and general practitioners. Cancer Metastasis Rev.40, 949–982 (2021). 19. He, M. et al. Remarkably similar CTLA-4 binding properties of therapeutic ipilimumab and tremelimumab antibodies. Oncotarget 8, 67129–67139 (2017). 20. Edman, N. I. et al. Modulation of FGF pathway signaling and vascular differentiation using designed oligomeric assemblies.2023.03.14.532666 Preprint at doi.org/10.1101/2023.03.14.532666 (2023). 21. Jost, C. & Plückthun, A. Engineered proteins with desired specificity: DARPins, other alternative scaffolds and bispecific IgGs. Curr. Opin. Struct. Biol.27, 102–112 (2014). 22. Fleishman, S. J. et al. RosettaScripts: A Scripting Language Interface to the Rosetta Macromolecular Modeling Suite. PLOS ONE 6, e20161 (2011). 23. Tries, R. S. et al. Gdf-8 Propeptide Binds to GDF-8 and Antagonizes Biological Activity by Inhibiting GDF-8 Receptor Binding. Growth Factors 18, 251–259 (2001). 24. Kim, S. K. et al. An engineered transforming growth factor ȕ (TGF-ȕ) monomer that functions as a dominant negative to block TGF-ȕ signaling. J. Biol. Chem.292, 7173– 7188 (2017). 25. Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr.66, 125–132 (2010). 26. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D Biol. Crystallogr. 62, 72–82 (2006). 27. Evans, P. R. & Murshudov, G. N. How good are my data and what is the resolution? Acta Crystallogr. D Biol. Crystallogr.69, 1204–1214 (2013). 28. Zwart, P. H. Anomalous signal indicators in protein crystallography. Acta Crystallogr. D Biol. Crystallogr.61, 1437–1448 (2005). 29. Yeates, T. O. Simple statistics for intensity data from twinned specimens. Acta Crystallogr. A 44 ( Pt 2), 142–144 (1988). 30. Padilla, J. E. & Yeates, T. O. A statistic for local intensity differences: robustness to anisotropy and pseudo-centering and utility for detecting twinning. Acta Crystallogr. D Biol. Crystallogr.59, 1124–1130 (2003). 31. French, S. & Wilson, K. On the treatment of negative intensity observations. Acta Crystallogr. A 34, 517–525 (1978). 32. Dauter, Z. Estimation of anomalous signal in diffraction data. Acta Crystallogr. D Biol. Crystallogr.62, 867–876 (2006). 33. Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr.67, 235–242 (2011). 34. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr.40, 658–674 (2007). 35. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr.53, 240–255 (1997). 36. Pannu, N. S., Murshudov, G. N., Dodson, E. J. & Read, R. J. Incorporation of prior phase information strengthens maximum-likelihood structure refinement. Acta Crystallogr. D Biol. Crystallogr.54, 1285–1294 (1998). 37. Murshudov, G. N., Vagin, A. A., Lebedev, A., Wilson, K. S. & Dodson, E. J. Efficient anisotropic refinement of macromolecular structures using FFT. Acta Crystallogr. D Biol. Crystallogr.55, 247–255 (1999). 38. Winn, M. D., Isupov, M. N. & Murshudov, G. N. Use of TLS parameters to model anisotropic displacements in macromolecular refinement. Acta Crystallogr. D Biol. Crystallogr.57, 122–133 (2001). 39. Steiner, R. A., Lebedev, A. A. & Murshudov, G. N. Fisher’s information in maximum- likelihood macromolecular crystallographic refinement. Acta Crystallogr. D Biol. Crystallogr.59, 2114–2124 (2003). 40. Winn, M. D., Murshudov, G. N. & Papiz, M. Z. Macromolecular TLS Refinement in REFMAC at Moderate Resolutions. in Methods in Enzymology vol.374300–321 (Academic Press, 2003).

41. Skubak, P., Murshudov, G. N. & Pannu, N. S. Direct incorporation of experimental phase information in model refinement. Acta Crystallogr. D Biol. Crystallogr. 60, 2196—2201 (2004).

42. Vagin, A. A. el al. REFMAC5 dictionary: organization of prior chemical knowledge and guidelines for its use. Acta Crystallogr. D Biol. Crystallogr. 60, 2184-2.195 (2004).

43. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486-501 (2010).

44. Battye, T. G. G., Kontogiannis, L., Johnson, ()., Powell, H. R. & Leslie, A. G. W. iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM. Acta Crystallogr. D Biol. Crystallogr. 67, 271-281 (2011).

45. Liebschner, D. et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr. Sect. Struct. Biol. 75, 861- 877 (2019).

46. Chen, V. B. et al. MoIProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 66, 12-21 (2010).

47. Project, C. C. & Number 4. The CCP4 suite: programs for protein crystallography . Acta Crystallogr. D Biol. Crystallogr. 50, 760-763 (1994),

48. Ramachandran, G. N., Ramakrishnan, C. & Sasisekharan, V. Stereochemistry- of polypeptide chain configurations. J. Mol. Biol. 7, 95-99 (1963).

The description of embodiments of the disclosure is not intended to be exhaustive or to limit the disclosure to the precise form disclosed. While the specific embodiments of, and examples for, the disclosure are described herein for illustrative purposes, various equivalent modifications are possible within the scope of the disclosure, as those skilled in the relevant art will recognize.

Claims

We claim

1. A polypeptide comprising the amino acid sequence selected from the group consisting of SEQ ID NO: 1-9.

2. The polypeptide of claim 1, comprising the amino acid sequence selected from the group consisting of SEQ ID NO: 1-3, wherein the polypeptide binds to CTLA-4.

3. The polypeptide of claim 1 or 2, comprising the amino acid sequence of SEQ ID NO:.'.

4. The polypeptide of any one of claims 1-3, comprising an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 10-13.

5. The polypeptide of any one of claims 2-4, wherein the polypeptide is identical relative to the reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues.

6. The polypeptide of claim 1, comprising the amino acid sequence selected from the group consisting of SEQ ID NO: 4-6, wherein the polypeptide binds to PD-L1.

7. The polypeptide of claim 1 or 6, comprising the amino acid sequence of SEQ ID NO:6.

8. The polypeptide of any one of claims I and 6-7, comprising an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91 %, 92%, 93%, 94%, 95%,

96%, 97%, 98%, 99%, or 100% identical to tire ammo acid sequence of SEQ ID NO: 15 or

16.

9. The polypeptide of any one of claims 6-8, wherein the polypeptide is identical relative to the reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues.

10. The polypeptide of claim 1, comprising the amino acid sequence selected from the group consisting of SEQ ID NO:7-9, wherein the polypeptide binds to TGFbRII.

11. The polypeptide of claim 1 or 10, comprising the amino acid sequence of SEQ ID NO:9.

12. The polypeptide of any one of claims 1 and 10-11, comprising an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the ammo acid sequence of SEQ ID NO: 14.

13. The polypeptide of any one of claims 10-12, wherein the polypeptide is identical relative to the reference polypeptide at 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, or all interface residues.

14. A polypeptide comprising an amino acid sequence at least 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence selected from SEQ ID NO: 10-16, not including any insertions.

15. The polypeptide of claim 14, wherein substitutions relative to the reference sequence are conservative amino acid substitutions.

16. The polypeptide of any one of claims 1-15, comprising an insertion in one or more loop regions of the polypeptide relative to the reference sequence.

17. A fusion protein, comprising:

(a) the polypeptide of any one of claims 1-16; and

(b) one or more functional domains at the N-tenninus and/or at the C-terminus of the polypeptide.

18. The fusion protein of claim 17, wherein the one or more functional domain comprises an oligomerization domain.

19 The polypeptide or fusion protein of any one of claims 1-18, wherein the polypeptide binds its target with nanomolar or picomolar affinity.

20. A nucleic acid encoding the polypeptide or fusion protein of any one of claims 1-19.

21. An expression vector comprising the nucleic acid of claim 20 operatively linked to a promoter.

22. A host cell comprising the polypeptide, fusion protein, nucleic acid, or expression vector of any one of claims 1-21.

23. A pharmaceutical composition, comprising:

(a) the polypeptide, the nucleic acid, the expression vector, and/or the host cell of any one of claims 1 -22; and

(b) a pharmaceutically acceptable earner.

24. The pharmaceutical composition of claim 23, further comprising one or more of an angiogenesis inhibitor, an immune checkpoint inhibitor, a tyrosine kinase inhibitor, a proteasome inhibitor, an mTOR inhibitor, a PI3K inhibitor, a histone deacetylase inhibitor. and a Hedgehog pathway blocker.

25. A method for treating cancer or tissue fibrosis, comprising administering to a subject in need thereof an amount of the polypeptide, fusion protein, nucleic acid, expression vector, host cell, and/or pharmaceutical composition effective to treat the cancer or tissue fibrosis.

26. The method of claim 25, wherein the method comprises administering to the subject an amount of the polypeptide of any one of claims 2-5, or a fusion protein or pharmaceutical composition thereof, wherein the subject has melanoma, lung cancer, bladder cancer, head and neck cancer, renal cell carcinoma, ovarian cancer, or colorectal cancer.

27. The method of claim 25, wherein the method composes administering to the subject an amount of the polypeptide of any one of claims 6-9, or a fusion protein or pharmaceutical composition thereof, wherein the subject has non-small cell lung cancer, melanoma, bladder cancer, head and neck squamous cell carcinoma, Hodgkin lymphoma, renal cell carcinoma, gastric or gastroesophageal junction adenocarcinoma, cervical cancer, or breast cancer.

28. Tire method of claim 25, wherein the method comprises administering to the subject an amount of the polypeptide of any one of claims 10-13, or a fusion protein or pharmaceutical composition thereof, wherein the subject has pancreatic cancer, breast cancer, colorectal cancer, lung cancer, prostate cancer, liver cancer, gastric cancer, ovarian cancer, melanoma, or glioblastoma.

29. The method of any one of claims 25-28, further comprising administering to the subject an amount effective to treat cancer of one or more of an angiogenesis inhibitor, an immune checkpoint inhibitor, a tyrosine kinase inhibitor, a proteasome inhibitor, an mTOR inhibitor, a P13K inhibitor, a histone deacetylase inhibitor, and a Hedgehog pathway blocker.

30. The method of claim 25, wherein the method comprises administering to the subject an amount of the polypeptide of any one of claims 10-13, or a fusion protein or pharmaceutical composition thereof, wherein the subject has tissue fibrosis.

31. The method of claim 30, wherein the tissue fibrosis is selected from the group consisting of idiopathic pulmonary fibrosis (IPF), liver cirrhosis, renal fibrosis, cardiac fibrosis, scleroderma, keloids, or hypertrophic scarring.

32. The method of tiny one of claims 25-31, wherein the subject is a human subject.