Enhancement of Binding Affinity of Folate to Its Receptor by Peptide Conjugation

(1) Background: The folate receptor (FR) is a target for cancer treatment and detection. Expression of the FR is restricted in normal cells but overexpressed in many types of tumors. Folate was conjugated with peptides for enhancing binding affinity to the FR. (2) Materials and Methods: For conjugation, folate was coupled with propargyl or dibenzocyclooctyne, and 4-azidophenylalanine was introduced in peptides for “click” reactions. We measured binding kinetics including the rate constants of association (ka) and dissociation (kd) of folate-peptide conjugates with purified FR by biolayer interferometry. After optimization of the conditions for the click reaction, we successfully conjugated folate with designed peptides. (3) Results: The binding affinity, indicated by the equilibrium dissociation constant (KD), of folate toward the FR was enhanced by peptide conjugation. The enhanced FR binding affinity by peptide conjugation is a result of an increase in the number of interaction sites. (4) Conclusion: Such peptide-ligand conjugates will be important in the design of ligands with higher affinity. These high affinity ligands can be useful for targeted drug delivery system.


Introduction
Traditional cancer therapy involves removal of tumor cells by surgery, radiation and non-selective types of chemotherapy [1,2]. Surgery and radiation are often effective with tumors that are primary or localized and have not metastasized to multiple sites throughout the body [3]. Chemotherapy is effective in the treatment of metastatic cancers because typical chemotherapeutic agents focus on rapidly growing tissues, which is a property common to cancer cells. Nonetheless, chemotherapy also often has a high incidence of unwanted and damaging side effects in normal tissues because these tissues are also undergoing growth [4,5]. Therefore, monoclonal antibodies against cellular targets that are unique to cancer cells have been developed [4,6], and antibody-drug conjugates (ADCs) have also been developed [6]. Targeted treatments exert their anticancer effects through multiple mechanisms, including proliferation inhibition [6], apoptosis induction [7], metastasis suppression [8], immune function regulation [9] and multidrug resistance reversal [5,10]. A few ADCs have been used successfully in clinical trials [5,10,11]. However, there are several points to consider when using an antibody as A few ADCs have been used successfully in clinical trials [5,10,11]. However, there are several points to consider when using an antibody as a drug-transporter that targets tumors. Limitations owing to poor therapeutic efficacy of ADCs include: (i) manufacturing procedures that create heterogeneous mixtures of ADCs with a number of drug molecules conjugated inconsistently; (ii) the synthesis costs are extremely high with difficulties in quality control; and (iii) the larger size of ADCs hampers penetration of ADCs into tumor tissue [12]. Small molecules or peptides are potential therapeutic molecules that overcome these problems [2]. In contrast to antibodies, these agents provide advantages such as reduced immunogenicity, quick clearance, increased diffusion and tissue penetration, chemical stability and ease of synthesis [2,6].
Due to the remarkable expression of the folate receptor (FR) on the surface of tumor cells, the FR can be exploited as a cancer diagnostic and therapeutic target [13]. Folate is an intrinsic ligand of the FR, consisting of a pterin ring, a central p-amino benzoic acid and an L-glutamic acid tail [4, 14,15], and has been conjugated with anti-cancer drugs [4,16] and drug carriers [17][18][19][20][21][22][23] for targeted delivery of drugs to tumor cells. For example, a peptide that binds to the α isoform of the FR, which is a subtype of FRs, was selected by phage display; however, the affinity of this peptide was low when compared with that of folate [24].
In this report, we conjugate folate with peptides to enhance binding affinity toward the FR. Previously, Li and Roberts [25] prepared a penicillin-peptide conjugate that has at least 100-fold higher activity than penicillin. Wang et al. introduced aminophenylalanine coupled with purvalanol into peptides to enhance the inhibitory activity of purvalanol against kinases [26,27]. Peptide conjugation should increase the affinity between the target protein and ligand by increasing the number of interaction sites, as shown in Figure 1. For conjugation, we added propargyl or dibenzocyclooctyne (DBCO) to folate and 4-azidophenyalanine (AzPhe) in the peptide for the "click" reaction, as shown in Figure 2, because it is possible to introduce the azidophenyl groups into proteins by bio-orthogonal approaches [28,29]. The binding assay of the synthesized folate-peptide conjugates with FR was performed by biolayer interferometry (BLI), and the association rate constant (ka) and dissociation rate constant (kd) were determined. The study demonstrated that the conjugation of folate with peptides enhanced the affinity of folate toward the FR. For conjugation, we added propargyl or dibenzocyclooctyne (DBCO) to folate and 4-azidophenyalanine (AzPhe) in the peptide for the "click" reaction, as shown in Figure 2, because it is possible to introduce the azidophenyl groups into proteins by bio-orthogonal approaches [28,29]. The binding assay of the synthesized folate-peptide conjugates with FR was performed by biolayer interferometry (BLI), and the association rate constant (k a ) and dissociation rate constant (k d ) were determined. The study demonstrated that the conjugation of folate with peptides enhanced the affinity of folate toward the FR.

Folate-Phe Conjugation by Click Reactions
Two types of folate analogues were prepared by addition of the propargyl group ( Figure 2A) and DBCO ( Figure 2B), and both were adjacent to the γ-carboxyl group of folate. The additions enabled confirmation of the click reaction between folate analogues and AzPhe-Fmoc. Folate-propargyl was used for the Cu(I)-catalyzed alkyne-azide cycloaddition (CuAAC) click reaction with AzPhe-Fmoc. To promote the CuAAC reaction, Cu(I) stabilizing ligands such as Tris (2-benzimidazoylmethyl) amine (BimH3) and microwaves were also employed at 50 °C. However, absorbance from the triazole ring on the target compound was not detected under the conditions shown in Table 1.  The strain-promoted alkyne-azide cycloaddition (SPAAC) click reaction between DBCO (green dotted circle) and the azide group (cyan dotted circle) to conjugate folate via dibenzocyclooctyne triazole (red dotted circle).

Folate-Phe Conjugation by Click Reactions
Two types of folate analogues were prepared by addition of the propargyl group ( Figure 2A) and DBCO ( Figure 2B), and both were adjacent to the γ-carboxyl group of folate. The additions enabled confirmation of the click reaction between folate analogues and AzPhe-Fmoc. Folate-propargyl was used for the Cu(I)-catalyzed alkyne-azide cycloaddition (CuAAC) click reaction with AzPhe-Fmoc. To promote the CuAAC reaction, Cu(I) stabilizing ligands such as Tris (2-benzimidazoylmethyl) amine (BimH 3 ) and microwaves were also employed at 50 • C. However, absorbance from the triazole ring on the target compound was not detected under the conditions shown in Table 1.
Currently, some groups have reported success of the CuAAC reaction between folate-propargyl and polymers containing an azido group [30][31][32][33][34]. However, their folate-propargyl conjugates were a mixture of propargyl groups bound to the C α and C γ of the glutamic acid part of folate. The present conjugate is the first example of a folate-propargyl with the propargyl group specifically linked to the C γ of folate. The results in Table 1 indicate that the C γ -binding propargyl group shows low reactivity in the CuAAC reaction. The other possibility is that coordination by the -N and -NH groups of the folate-propargyl with Cu(I) interferes with alkyne-Cu(I) complexation.
In contrast, the strain-promoted azide-alkyne cycloaddition (SPAAC) "click" reaction between folate-DBCO and AzPhe-Fmoc was successful ( Table 1). The yield increased up to 88% by using twice the molar ratio of folate-DBCO against AzPhe-Fmoc, and the reaction temperature did not affect yields noticeably. Golas et al. [35] studied the substituent effect on azide reactivity in CuAAC using various azide compounds with propargyl alcohol. The electronic properties and steric congestion near end groups are major determinants for the reactivity of azide compounds. Azide with electron withdrawing groups, such as ethyl azido-acetate, methyl 2-azidopropionate and azidoacetonitrile, react faster than similar compounds with a neighboring aromatic ring (benzyl azide and 1-phenylethyl azide). In addition, primary azides such as benzyl azide and ethyl azido-acetate react faster than their secondary analogues, 1-phenylethyl azide and methyl 2-azidopropionate, respectively. In this case, AzPhe is less reactive because the electron-withdrawing is affected by the aromatic ring. Nonetheless, AzPhe can be more reactive through SPAAC because DBCO enhances the reactivity by its resonant structure [36]. Two types of folate analogues were prepared by addition of the propargyl group ( Figure 2A) and DBCO ( Figure 2B), and both were adjacent to the γ-carboxyl group of folate. The additions enabled confirmation of the click reaction between folate analogues and AzPhe-Fmoc. Folate-propargyl was used for the Cu(I)-catalyzed alkyne-azide cycloaddition (CuAAC) click reaction with AzPhe-Fmoc. To promote the CuAAC reaction, Cu(I) stabilizing ligands such as Tris (2-benzimidazoylmethyl) amine (BimH3) and microwaves were also employed at 50 °C. However, absorbance from the triazole ring on the target compound was not detected under the conditions shown in Table 1.
The number corresponds the compound number in Figure 6. b High performance liquid chromatography (HPLC) yields; c Microwave conditions; N.D. Not detected.

Preparation of Folate-Peptide Conjugates by the SPAAC Click Reaction
Since folate-DBCO was demonstrated to conjugate efficiently to AzPhe by the SPAAC reaction, the preparation of folate-peptide conjugates was performed by this reaction (Figure 3). Three peptide sequences, GF[AzPhe]IQ, SE[AzPhe]KA and DSE[AzPhe]KAY, were synthesized. The folate-peptide conjugates were designed by the program ICM-Pro (Molsoft L.L.C., San Diego, CA, USA). After successful conjugation of folate with AzPhe by SPAAC, we considered the folate-conjugated AzPhe as one unit and increased the length of the peptide by adding amino acids at N-terminal and C-terminal of the AzPhe. This length was increased by trial and error procedure. The peptides were synthesized by a conventional solid phase synthesis method. For BLI measurements, in which a biotin group binds to streptavidin bound to coated sensor chips, the N-terminus of the peptides was modified with biotin-(PEG 24 )-NHS. The coupling was performed before release from the solid phase synthesis resin ( Figure 3A) [37,38]. The polyethylene glycol (PEG) linker functions as a spacer between the immobilized and interaction sites and as a solubilizer of the folate-peptide conjugates in aqueous solutions.
The same coupling reaction conditions were used for peptide conjugation. After the click reaction and purification, each folate-peptide conjugate was identified by matrix assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS). From the mass spectra, folate was confirmed to bind successfully to the side chain of AzPhe in the peptides.  Table 2 and Figure 4 show the results of the BLI measurements to evaluate the affinities of the folate-peptide conjugates toward folate receptor alpha (FRα).   Table 2 and Figure 4 show the results of the BLI measurements to evaluate the affinities of the folate-peptide conjugates toward folate receptor alpha (FRα).

BLI Measurement
Commercially available folate-PEG 8 -biotin was used as a control for BLI analysis. The equilibrium dissociation constant (K D ) between FRα and folate was 1.14 nM. Wibowo et al. [39] and Chen et al. [14] used a radiolabeled ligand assay and isothermal calorimetry for measurement of the K D of folate with FRα and yielded values of~10 pM and~190 pM, respectively. Combined with our results, the differences in K D values indicate that the method used to measure the K D has a strong influence on the outcome.  Commercially available folate-PEG8-biotin was used as a control for BLI analysis. The equilibrium dissociation constant (KD) between FRα and folate was 1.14 nM. Wibowo et al. [39] and Chen et al. [14] used a radiolabeled ligand assay and isothermal calorimetry for measurement of the KD of folate with FRα and yielded values of ~10 pM and ~190 pM, respectively. Combined with our results, the differences in KD values indicate that the method used to measure the KD has a strong influence on the outcome.
An advantage of BLI is evaluation of the ka and kd. The binding mode of folate to FRα shows a non-equilibrium binding mode, in which the kd (7.69 × 10 −3 s −1 ) was ~10 3 times slower than that of the association rate (6.74 × 10 6 M −1 s −1 ). This difference between the ka and kd corresponds well with the scenario previously proposed for folate binding to FRα [39]. In crystallographic work that compared the apo-and folate binding forms of FRs, large conformational changes around the folate binding pocket upon folate binding were observed, i.e., from the relaxed (open) to tight (closed) forms. In the closed form, the inhibitory loop, basic loop and switching helix around the binding pocket cooperatively undergo conformational changes to bind the folate tightly. The bound folate in the FRs then dissociates from the receptors after endocytosis of the FRs into cells, which is triggered by the acidic environment of the cells. Such a non-equilibrium-binding mode promotes efficient uptake of An advantage of BLI is evaluation of the k a and k d . The binding mode of folate to FRα shows a non-equilibrium binding mode, in which the k d (7.69 × 10 −3 s −1 ) was~10 3 times slower than that of the association rate (6.74 × 10 6 M −1 s −1 ). This difference between the k a and k d corresponds well with the scenario previously proposed for folate binding to FRα [39]. In crystallographic work that compared the apo-and folate binding forms of FRs, large conformational changes around the folate binding pocket upon folate binding were observed, i.e., from the relaxed (open) to tight (closed) forms. In the closed form, the inhibitory loop, basic loop and switching helix around the binding pocket cooperatively undergo conformational changes to bind the folate tightly. The bound folate in the FRs then dissociates from the receptors after endocytosis of the FRs into cells, which is triggered by the acidic environment of the cells. Such a non-equilibrium-binding mode promotes efficient uptake of folate into cells. Thus, our BLI data provide the first indication that the proposed trafficking mechanism of folate is valid by revealing the asymmetric binding kinetics of FRs.

Interaction of Folate-Peptide Conjugates with FRα
By conjugation with peptides, the affinity increased to sub-nanomolar (~10 −10 M) K D values ( Table 2). The peptide-conjugates showed slower k a values that ranged from 8.91 × 10 4 to 1.10 × 10 6 (M −1 s −1 ). Results presented in Figure 4B-D show significantly slow dissociation even after incubation in buffer. As a result, the k d slows from 7.53 × 10 −5 to 2.65 × 10 −4 s −1 , which increases the K D values. These observations suggest that peptide modification further stabilizes the complex formed between the peptide-conjugates and FRs, most probably by increasing the number of interaction sites between them.
In the peptide-conjugates, SEZKA and DSEZKAY share the common SEZKA sequence. Addition of aspartic acid (D) at the N-terminus and tyrosine (Y) at the C-terminus leads to a 12-fold faster association constant and 3-fold faster dissociation constant for the DSEZKAY peptide-conjugate, resulting in a 4-fold lower K D . This increase in affinity occurs by lengthening SEZKA to DSEZKAY. This result indicates that we can alter the affinity of peptide-conjugate compounds by increasing the length of the peptides at both the N-and C-termini. This may provide a way to manipulate binding properties of peptide-conjugated compounds by increasing the length of the peptide part of the conjugates, which may increase the number of interaction contacts with the target protein. Figure 5 shows the results of the docking simulation, which demonstrates the interaction mode of DSEZKAY with FRα. As expected in Figure 1, the structure of the complex shows an increase in the number of interactions to FRα from the peptide portion around the folate-binding pocket. Previous reports have demonstrated greater than 100-fold increases in binding affinity by peptide conjugates [25][26][27], whereas the present result was lower than these previous increases in affinity. However, the present investigation also revealed that peptide conjugation is a useful tool to enhance the binding affinity to the target molecule. Future efforts will focus on using the folate-peptide conjugate to target anti-cancer drug delivery. folate into cells. Thus, our BLI data provide the first indication that the proposed trafficking mechanism of folate is valid by revealing the asymmetric binding kinetics of FRs.

Interaction of Folate-Peptide Conjugates with FRα
By conjugation with peptides, the affinity increased to sub-nanomolar (~10 −10 M) KD values ( Table 2). The peptide-conjugates showed slower ka values that ranged from 8.91 × 10 4 to 1.10 × 10 6 (M −1 s −1 ). Results presented in Figures 4B-D show significantly slow dissociation even after incubation in buffer. As a result, the kd slows from 7.53 × 10 −5 to 2.65 × 10 −4 s −1 , which increases the KD values. These observations suggest that peptide modification further stabilizes the complex formed between the peptide-conjugates and FRs, most probably by increasing the number of interaction sites between them.
In the peptide-conjugates, SEZKA and DSEZKAY share the common SEZKA sequence. Addition of aspartic acid (D) at the N-terminus and tyrosine (Y) at the C-terminus leads to a 12-fold faster association constant and 3-fold faster dissociation constant for the DSEZKAY peptide-conjugate, resulting in a 4-fold lower KD. This increase in affinity occurs by lengthening SEZKA to DSEZKAY. This result indicates that we can alter the affinity of peptide-conjugate compounds by increasing the length of the peptides at both the N-and C-termini. This may provide a way to manipulate binding properties of peptide-conjugated compounds by increasing the length of the peptide part of the conjugates, which may increase the number of interaction contacts with the target protein. Figure 5 shows the results of the docking simulation, which demonstrates the interaction mode of DSEZKAY with FRα. As expected in Figure 1, the structure of the complex shows an increase in the number of interactions to FRα from the peptide portion around the folate-binding pocket. Previous reports have demonstrated greater than 100-fold increases in binding affinity by peptide conjugates [25][26][27], whereas the present result was lower than these previous increases in affinity. However, the present investigation also revealed that peptide conjugation is a useful tool to enhance the binding affinity to the target molecule. Future efforts will focus on using the folate-peptide conjugate to target anti-cancer drug delivery.

Materials
Fmoc-Phe(4-N 3 )-OH (AzPhe-Fmoc) was purchased from Watanabe Chemical Industries, Ltd. (Hiroshima, Japan) to incorporate non-natural amino acids during the solid phase peptide synthesis procedure. BimH 3 was purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Biotin-PEG 24  NHS was purchased from Thermo Fisher Scientific (Waltham, MA, USA) for biotin-PEG 24 modification at the N-terminus of the folate-peptide conjugates. For the activity assay, streptavidin (SA) biosensors were purchased from ForteBio (Fremont, CA, USA). Folate-PEG 8 -biotin was purchased from Nanocs (New York, NY, USA). Reagents used for reversed-phase high performance liquid chromatography (RP-HPLC) were of HPLC grade. All other chemicals used were of biochemical research grade. MALDI-TOF MS (Microflex, Bruker Daltonics, Billerica, MA, USA.) was employed for molecular weight measurement.

Synthesis of Folate-Propargyl and Folate-DBCO
Synthetic schemes of folate derivatives are presented in Figure 6. Each compound was synthesized and confirmed as follows.

Materials
Fmoc-Phe(4-N3)-OH (AzPhe-Fmoc) was purchased from Watanabe Chemical Industries, Ltd. (Hiroshima, Japan) to incorporate non-natural amino acids during the solid phase peptide synthesis procedure. BimH3 was purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Biotin-PEG24-NHS was purchased from Thermo Fisher Scientific (Waltham, MA, USA) for biotin-PEG24 modification at the N-terminus of the folate-peptide conjugates. For the activity assay, streptavidin (SA) biosensors were purchased from ForteBio (Fremont, CA, USA). Folate-PEG8-biotin was purchased from Nanocs (New York, NY, USA). Reagents used for reversed-phase high performance liquid chromatography (RP-HPLC) were of HPLC grade. All other chemicals used were of biochemical research grade. MALDI-TOF MS (Microflex, Bruker Daltonics, Billerica, MA, USA.) was employed for molecular weight measurement.

Synthesis of Folate-Propargyl and Folate-DBCO
Synthetic schemes of folate derivatives are presented in Figure 6. Each compound was synthesized and confirmed as follows.

Compound 2
To a solution of folate 1 (10 g, 0.022 mol) and 100 mL anhydrous tetrahydrofuran (THF) in a three-neck flask, 24 mL, 0.176 mol trifluoroacetic anhydride [(CF 3 CO) 2 O] was slowly added at 0 • C for 30 min. The dark brown homogeneous mixture was stirred at room temperature (RT). After 10 h, the reaction mixture was filtered through a pad of celite to remove the small amount of solid residue. The filtrate was concentrated under reduced pressure and the viscous liquid was dissolved with a minimum amount of THF (5 mL), which was slowly transformed into a flask of well-stirred diethyl ether (Et 2 O). The yellow precipitate formed in Et 2 O was collected by filtration and washed with Et 2 O (25 mL ×2) to yield the crude compound 2.

Compound 3
The crude compound 2 (6 g) was dissolved in THF (50 mL) followed by the addition of ice (~10 g) with stirring for 5 h. The mixture was slowly transferred into stirred Et 2 O (200 mL). The yellowish precipitate was collected by filtration, washed with Et 2 O (200 mL ×3) and dried for 24 h under vacuum. To the suspension of yellowish precipitate, conc. HCl (60 mL) was added and refluxed at 60 • C overnight and then 100 • C for 2.5 h. The reaction mixture was poured into water (100 mL). The precipitate formed in the solution was collected by filtration and washed with Et 2 O to afford compound 3 (75%). 1

Compound 7
Compound 6 (620 mg, 1.8 mmol) was dissolved in CH 2 Cl 2 (5.4 mL) and cooled to 0 • C. To the solution, 12.6 mL trifluoroacetic acid (TFA) was added while stirring. After stirring at RT for 4 h, the solvent was evaporated under reduced vacuum. MeOH was added to dissolve the crude powder and then solidification was performed by the addition of Et 2 O. The solvent was evaporated and the precipitate dried to give compound 7 (340 mg, 100%). 1

Compound 11
A mixture of compound 10 (3 g, 9.9 mmol) and CDI (1.60 g, 9.9 mmol) in CH 2 Cl 2 (30 mL) was stirred at RT for 1 h, followed by the addition of 1.46 mL of 9.9 mmol tetramethylsilane ethanol (TMS EtOH), and this sample was stirred for a further 18 h. H 2 O (150 mL) was added to the reaction mixture and the resulting mixture was partitioned. The organic layer was dried with anhydrous Na 2 SO 4 and the solvent evaporated under reduced pressure. The residue was purified on a silica gel column with 25% (v/v) ethyl acetate in hexane to give a colorless oil 11 (3.46 g, 87%). 1

Compound 12
A mixture of compound 11 (2 g, 4.9 mmol) and TFA:CH 2 Cl 2 (1:2, v/v) (15 mL) was stirred at 0 • C for 30 min. The reaction mixture was then allowed to acquire at RT for 4.5 h. while stirring. The solvent of the reaction mixture was evaporated and the material purified on a silica gel column with 20-35% (v/v) MeOH in CHCl 3 to give compound 12 (0.842 g, 69%, as a colorless semisolid). 1

Compound 14
To a solution of compound 13 (1 g, 1.5 mmol) and 5 mL N,N-dimethylformamide (DMF), NHS (202 mg, 1.7 mmol) and EDC (279 mg, 1.5 mmol) were added. The resulting mixture was stirred at RT for 18 h. The reaction mixture was poured into water (300 mL) and the yellow precipitate was collected by filtration to afford compound 14 (1.03 g, 90%). 1

Compound 16 (Folate-DBCO)
To a solution of compound 15 (100 mg, 0.1 mmol) in DMSO (1 mL), TBAF [1.14 mL of 1 M in anhydrous THF, 10 equivalent (eq.)] was added and then stirred at RT. After 10 h stirring, AcOH (1.25 mL) was added and the mixture was poured into a mixture of CHCl 3 and ethyl acetate (4:1, 25mL). The yellowish precipitate formed in the solution was collected by filtration and then recrystallized in a mixture EtOH:MeOH to give the yellow solid compound 16 (folate-DBCO). 1 H NMR data is displayed in Figure S2A 13 C NMR data is displayed in Figure S2B. 13

Click Reaction of Folate-Propargyl or Folate-DBCO with AzPhe-Fmoc
Reaction schemes for the click reactions of folate-propargyl and folate-DBCO with AzPhe-Fmoc are shown in Figure 2. A 1 mM stock of folate-propargyl and a 10 mM stock of BimH 3 were prepared in DMSO for CuAAC. A 10 mM stock of AzPhe, sodium ascorbate, 2 mM stock of copper (II) sulfate (CuSO 4 ) and copper (I) chloride (CuCl) were prepared in H 2 O.
A 10 mM folate-DBCO stock was prepared in DMF and a 10 mM stock of AzPhe-Fmoc was dissolved in H 2 O for SPAAC. Several trials were performed for both compounds listed in Table 1. RP-HPLC using an Inertsil ODS-3 column (Nacalai tesque Inc., Kyoto, Japan) at 25 • C for 55 min was performed with H 2 O containing 0.1 % (v/v) TFA (solvent A) and acetonitrile containing 0.1 % (v/v) TFA (solvent B) as a solvent system with a gradient from 0-0.10 min at 90% A, 5-40 min at 90-50% A, 40-45 min at 50-0% A and 45-47 min at 0-90% A, and flow rate of 1 mL/min. In some cases, a gradient from 0-10 min at 90% A, 5-40 min at 90-30% A, 40-43 min at 30-0% A and 43-45 min at 0-90% A was used, and a flow rate of 1 mL/min.

Synthesis and Purification of Peptides with N-terminal Biotin-PEG 24
The folate-peptide conjugates were synthesized by conventional Fmoc based solid-phase synthesis methods using a high purity single peptide synthesizer MultiPep CF and micro-column (INTAVIS Co. Ltd., Cologne, Germany). During synthesis, coupling and deprotection steps were carried out in the peptide synthesizer. All peptides were synthesized at the 10 µmol scale. Peptide synthesis is as follows: Preloaded 0.21 mmol/g of fmoc-Gln(Trt)-NovaSyn TGA (Novabiochem, Darmstadt, Germany), 0.19 mmol/g of fmoc-Ala-NovaSyn TGA (Novabiochem) or 0.24 mmol/g of fmoc-Tyr (tBu)-NovaSyn (Novabiochem) was used for synthesis of GF[AzPhe]IQ, SE[AzPhe]KA and DSE[AzPhe]KAY, respectively. Fmoc deprotection was performed by using 20% (v/v) piperidine in N-methyl-2-pyrrolidone (NMP) or 1% (v/v) formic acid + 20% (v/v) piperidine in NMP, depending on the amino acid content of the peptide. For the coupling step, the corresponding amino acid (5 times mol with respect to resin) was incubated with the resin for 30 min in the presence of NMP (8 µL), 0.5 M (2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (150 µL) and 4 M N-methylmorpholine (45 µL). AzPhe (65 ng) was used during each synthesis. After confirming the mass of the synthesized peptides by MALDI-TOF MS, the beads were incubated overnight on a shaker with a mixture of biotin-PEG 24 -NHS (34 mg, 1.5 mol eq.), hydroxybenzotriazole (8.3 mg, 0.6 mol eq.) with the addition of NMP (300 µL). The reaction scheme for N-terminal peptide modification with biotin-PEG 24 modification is shown in Figure 3A. After confirming the mass of the product with MALDI-TOF MS, the peptides were cleaved from the resin using a cleavage cocktail [95.0% (v/v) TFA, 2.5% (v/v) triisopropylsilane and 2.5% (v/v) H 2 O]. Depending on the amino acid content of each peptide, resins were incubated with the cleavage cocktail for 2-4 h in a light protected container with intermittent shaking. The cleavage mixture was then filtered to remove the beads and peptides were precipitated using cold Et 2 O. The resulting precipitate was centrifuged and washed three times with Et 2 O. Et 2 O was removed by overnight vacuum lyophilization and peptides were obtained in powder form. The products were further purified by RP-HPLC using the Inertsil ODS-3 column at 25 • C for GFZIQ and SEZKA: 50 min with a gradient of 1-51% (v/v) acetonitrile in water containing 0.1% (v/v) TFA. An Inert Sustain C18 column (Nacalai tesque Inc.) at 50 • C was used to further purify the DSEZKAY peptide. The gradient was 25-55% (v/v) acetonitrile in water containing 0.1 % (v/v) TFA for 30 min. Peptides were purified as shown in Figure S3 and analyzed by MALDI-TOF MS ( Figure S4A-C) and the results are summarized in Table S1. The purified peptides were lyophilized and stored until required.

Purification and Refolding of FRα
All steps performed for purification and refolding of FRα were carried out according to our previous report [41]. However, some different reagents were used. For cell body washing, we used 4 M urea instead of Triton X-114. The inclusion bodies were solubilized and purified with 8 M urea instead of 6 M guanidine HCl. Purification and refolding data are shown in Figure S5, and Tables S2 and S3.

BLI Measurements
The binding affinity of refolded FRα toward folate and folate conjugated peptide aptamers was measured by biolayer interferometry at 25 • C using a BLItz system (ForteBio) with kinetics buffer [10 mM PBS, pH 7.4, 0.5% (w/v) BSA and 0.01% (v/v) Tween 20]. The measurement procedure has been reported previously [41]. Streptavidin-coated biosensors (SA sensors were hydrolyzed for 2 h in 250 µL kinetics buffer and then soaked with 250 µL folate-PEG 8 -biotin (2.5 µM), or a variety of concentrations of 250 µL folate-peptide-PEG 24 -biotin conjugates at a stirring speed of 2200 rpm. Two baselines were measured for each sensor in kinetics buffer for 30 and 300 s prior to the immobilization and association step, respectively. Folate-PEG 8 -biotin or folate-peptide-PEG 24 -biotin conjugates immobilized to SA biosensors were dipped into FRα solutions for the association step. Dissociation was monitored in 250 µL immune assay kinetics buffer. To eliminate errors from non-specific binding of the analyte (FRα) on the SA biosensor chips, reference data with the same concentrations of analyte were also measured.
The obtained binding data were analyzed using a 1:1 local analysis mode applied with association and dissociation step corrections by the BLItz Pro1.2 software (ForteBio). The reference measurements were subtracted during data analysis to determine k a , k d and K D .