A Strategy to Select Macrocyclic Peptides Featuring Asymmetric Molecular Scaffolds as Cyclization Units by Phage Display

Macrocyclic peptides (MPs) have positioned themselves as a privileged class of compounds for the discovery of therapeutics and development of chemical probes. Aided by the development of powerful selection strategies, high-affinity binders against biomolecular targets can readily be elicited from massive, genetically encoded libraries by affinity selection. For example, in phage display, MPs are accessed on the surface of whole bacteriophages via disulfide formation, the use of (symmetric) crosslinkers, or the incorporation of non-canonical amino acids. To facilitate a straightforward cyclization of linear precursors with asymmetric molecular scaffolds, which are often found at the core of naturally occurring MPs, we report an efficient two-step strategy to access MPs via the programmed modification of a unique cysteine residue and an N-terminal amine. We demonstrate that this approach yields MPs featuring asymmetric cyclization units from both synthetic peptides and when linear precursors are appended onto a phage-coat protein. Finally, we showcase that our cyclization strategy is compatible with traditional phage-display protocols and enables the selection of MP binders against a model target protein from naïve libraries. By enabling the incorporation of non-peptidic moieties that (1) can serve as cyclization units, (2) provide interactions for binding, and/or (3) tailor pharmacological properties, our head-to-side-chain cyclization strategy provides access to a currently under-explored chemical space for the development of chemical probes and therapeutics.


Supporting discussion of 1D and 2D NMR experiments
: A comparison of 1 H-NMR spectra obtained for synthetic pep1 before and after modification with 1 shows the appearance of signals consistent with the para-substituted benzene and aldehyde protons of the cyclization unit. Moreover, the two β-protons of cysteine are split in the modified sample, which is consistent with a conformational restriction upon thiol alkylation. Notably, for the 1 H-NMR spectrum obtained following reduction with NaBH3CN individual β-protons of cysteine become even more distinct, which is consistent with cyclization. Similarly, we noted the disappearance of the aldehyde proton and the concomitant shift of one pair of aromatic protons, which are consistent with the change from the benzaldehyde to the corresponding benzylamine moiety. Similarly, we observe two new signals around 4 and 4.3 ppm, which were assigned to the benzylic protons formed upon reductive amination. Figure S2: For the comparison of 1 H total correlated spectra (TOCSY) spectra between the synthetic (A), modified (B) and cyclized peptide (C), we focused on identifying significant shifts of amide protons. As amide protons form a spin-system with protons from their respective side chains, this analysis enabled us to assign individual amide protons to their corresponding amino acids in all three spectra. For example, cysteine alkylation has little effect on the environment of most amide protons, with exception of the small shifts observed for those corresponding to the C-terminal alanine, tryptophan and glycine residues (spectrum A and B).
As such, this result is indicative for an interaction of the newly installed benzene ring from the cyclization unit with these residues. In stark contrast, amide protons in the TOCSY spectrum of the reduced product have undergone drastic changes. Particularly, signals corresponding to serine and glycine residues shift by approximately 0.5 ppm. This difference is consistent with the formation of a cyclic product that will require these flexible amino acids to adopt different S3 conformations. Moreover, the splitting of α-and β-methylene protons of glycine and serine residues, respectively can be observed, which is also consistent with the formation of a MP. Figure S3: As stated in the main text of the manuscript, we performed Nuclear Overhauser Effect Spectroscopy (NOESY) on the crude cyclized product after modification of pep1 with 1 and subsequent reduction with NaBH3CN. Note that the spectrum was recorded at a lower temperature (4 °C vs 25 °C) than the 1 H-NMR and TOCSY spectra discussed before to improve the signal-to-noise ratio of NOEs over background. In addition to the signals observed for the aromatic protons of 1 (Fig. 2C), we also identified NOEs between the benzylic protons and those of cysteine and the N-terminal alanine residue. Particularly, this analysis revealed that the alpha and beta-protons of cysteine exclusively interact with the benzylic protons with a shift of 4.45 ppm, while the corresponding alanine protons give rise to NOEs with the second set of benzylic protons at 3.8 and 4.1 ppm. S4 Figure S1: 1 H-NMR of synthetic pep1, after cysteine alkylation with 1 and following cyclization in presence of NaBH3CN. See Supporting Discussion for an analysis of the most relevant changes. Relevant peaks are highlighted by different colors. Peaks assigned with a * are ascribed to hydrolysis/decomposition products of NaBH3CN.    The structure of the obtained cyclic product from the reaction of pep4 with 1 as well as excerpts from a 2D NOESY spectrum are shown. NOEs between the phenylic protons and amino acid residues are highlighted. Color code: phenylic protons = red, cysteine protons = dark blue, lysine protons = light blue, valine protons = magenta. Critically, NOEs are only observed for the α-and β-protons of lysine but are absent for γ-ε protons. This observation is consistent with the cyclization occurring selectively via the N-terminus instead of the ε-amine. Note that excerpts of the spectrum have been scaled independently to highlight the weak NOEs observed between phenylic protons and the NH-Val as well as the H2Cβ-Lys.     S10: A-B: Representative raw mass spectrum and deconvoluted mass obtained for purified pep-TEV-D1D2 (A) and crude reaction mixtures obtained following modification and cyclization with 1-4 (B). C: Zoom-in of the +26 m/z peak around 1000-1020 Da for pep-TEV-D1D2 (black), following modification (blue) and cyclization (red). The formation of a distinct species following reductive amination demonstrates the selective formation of cyclic peptides on pep-TEV-D1D2. D: Representative raw mass spectrum and deconvoluted mass obtained for the protein fragment following TEV protease treatment. Figure S11: Representative raw mass spectra obtained for purified disulfide-free D1D2 and the crude reaction mixture following modification of D1D2 with 1 and NaBH3CN treatment. As stated in the main text, we were unable to observe any appreciable levels of modification in absence of the appended peptide.

Supporting Figures
Figure S12: A-B: Phage titers obtained for linear, modified and cyclized phages displaying SAV_1, SAV_2 or SAV_3 before (A) and after dilution to approximately 10 11 c.f.u.s (B). C: Phage titers obtained after biopanning bacteriophages featuring linear, modified and cyclized SAV_1, SAV_2 or SAV_3 against streptavidin. Note that no colonies for the linear SAV_1 sample were obtained at the dilutions that were plated, resulting in an upper limit of 30,000 c.f.u.s for this sample. S12 and 4 (SAV3). D: SAV_1-3 cyclized with scaffolds 3 (SAV1 and SAV2) and 4 (SAV3). Total ion counts (TICs) are depicted with masses found for the species inserted. *In panel D the observed mass for SAV1_cyclized is 2 Da higher than calculated due to reduction of the aldehyde on the scaffold that modified the second cysteine residue.      To independently verify the extrapolated Kd values obtained from the HABA competition assay, we performed ITC measurements. The obtained binding affinities are 3-5 times higher than those determined by the competition assay, but do show the same relative difference in binding strength. Note, that data points correspond to values obtained from a single experiment.

Materials & Methods
Chemicals were used without further purification unless otherwise noted. All chemicals used in organic synthesis were purchased from SigmaAldrich or TCI Europe. ChemMatrix® Rink amide resin was purchased from SigmaAldrich. Solid-phase reaction vessels (syringes with filter) were purchased from Torviq. Fmoc-protected amino acids, HCTU, trifluoro acetic acid (TFA) and OxymaPure® were purchased from ChemImpex Inc. Solvents for solid phase peptide synthesis (dichloromethane (DCM) and dimethylformamide (DMF)) were purchased from Biosolve. DIPEA and piperidine was purchased from Iris Biotech GMB. Analytical thinlayer chromatography was carried out on pre-coated silica gel on aluminum sheets (Merck TLC Silica gel 60 / Kieselguhr F254), columns were performed using silica-P flash silica gel from Silicycle (0.040-0.063 mm 230 400 mesh). 1 H-NMR and 13 C-NMR spectra were recorded on a Bruker 400 MHz in Methanol-d4, CDCl3 or DMSO-d6. 1D and 2D NMR spectra of peptides were performed in 90% H2O/ 10% D2O/ trifluoroacetic acid and recorded on a Bruker 600 MHz spectrometer. The strongest solvent peak was suppressed using excitation sculpting. Spectra were recorded either at 25 ℃ or at 4 ℃. HSQC, TOCSY, and NOESY (120 ms mixing time) spectra were performed.
Analytical UPLC-MS analysis was performed on an Acquity UPLC system (Waters) coupled to a quadrupole/time-of-flight (QToF) mass spectrometer (Waters) equipped with a PDA detector. The peptides were separated on an Acquity BEH C8; 150 × 2.1 mm, 1.7 μm (Waters) column operated at 40°C. The eluent system employed was a combination of A (0.1% formic acid in water) and B (0.1% formic acid in acetonitrile) at a flow rate of 0.3 mL/min. The gradient varied linearly from 5 to 60% B (v/v) from 0-10 min., 60 to 95% B from 10-11 min., kept at 95% B from 11-13 min., returning to 5% B in 2 min., re-equilibration to 5% B from 15-20 min.
Isothermal titration calorimetry (ITC) measurements were performed on a MicroCal ® iTC200 MicroCalorimeter. Measurements were performed at 25 °C, with a reference power of 6 μcal/sec and with a stirring speed of 750 rpm.
Escherichia coli strain NEB10-beta (New England Biolabs) was used for cloning and Primers

Solid phase peptide synthesis
Loading of Rink amide resin (coupling of Fmoc-Ala-OH: Rink amide resin loading = 0.4-0.6 mmol/g, 0.6 mmol/g was used as 1 eq.) was swollen in dry DMF for 1 and the resin subsequently washed with DCM (3 x 5 mL) and DMF (3 x 5 mL). Fmoc-Ala-OH (5 eq.) was pre-activated for 2 minutes in DMF (final concentration 0.5 M) following the addition of HCTU (4.7 eq.), OxymaPure® (4.7 eq.) and DIPEA (12 eq.). The preactivated Fmoc-protected amino acid was transferred to the resin and the resulting mixture was agitated (bubbling N2 through the syringe) for 2 hours at room temperature. The resin was subsequently drained and another freshly-prepared batch of pre-activated Fmoc-Ala-OH was added. The resulting mixture was agitated for 3 hours, before the resin was drained and washed with DMF (3 x 5 mL), DCM (3 x 5 mL), and DMF (5 x 5 mL). All remaining, unreacted amine groups were capped by adding a solution of acetic anhydride:pyridine (3:2) to the resin. After agitating the mixture for 30 minutes, the resin was washed with DMF (4 x 5 mL), DCM (2 x 5 mL), DMF (3 x 5 mL), and DCM (5 x 5 mL). The resin was then dried under vacuum and stored at -20 °C until further use.
A small sample (10 mg) of the dried resin was removed to determine the loading efficiency.
For this, the resin was first swollen for 30 minutes in 800 μL of DMF, after which 200 μL of piperidine was added. The mixture was vortexed to ensure good mixing and left in a tabletop shaker (room temperature, 300 rpm) for 15 min. An aliquot of the mixture (100 μL) was diluted to 10 mL with 20% piperidine in DMF and the concentration of the piperidine-fulvene adduct (λ = 301 nm, ε = 7800 M cm −1 ) was determined using a spectrophotometer. The loading of the resin was determined using the following formula with L being the resin loading, A301 the absorbance at 301 nm, and M the weight of the sample.
After an initial washing step (3 x 5 mL DMF) peptides were assembled following a cycle of deprotection and coupling steps.

Deprotection:
The resin was treated with 20% piperidine/DMF (6 mL, 1 x 2 min., 2 x 8 min.) and washed with DMF (3 x 6 mL), DCM (2 x 6 mL), and DMF (3 x 6 mL). An aliquot at the end of every deprotection was taken and the absorption at 301 nm measured on the NanoDrop™ to ensure complete deprotection. Following the deprotection of the N-terminal amino acid, the resin was washed with DMF (2 x 6 mL) and DCM (5 x 6 mL) and subsequently dried under vacuum.
Subsequently, the resin was drained and washed with DMF (5 x 6 mL).

Cleavage and peptide isolation:
A cleavage cocktail containing TFA/TIS/EDT/water (90:4:4:2 v/v/v/v, 10 mL) was added to the dried resin and incubated at room temperature for 2 hours. Subsequently, the resulting mixture was filtered, the resin washed with TFA (2 x 3 mL) and the resulting filtrate concentrated to ~0.5 mL by blowing a constant stream of N2 over the solution. Peptides were then precipitated by the addition of ice-cold diethyl-ether (20 mL) and subsequently pelleted by centrifugation (3,000 rpm). The supernatant was carefully removed by decantation and the precipitate washed twice with ice-cold ether (20 mL) to remove organic impurities. Residual ether was removed by blowing a constant stream of N2 over the sample and the residue subsequently dissolved in 0.1% TFA (aq.) and freeze-dried. Peptides obtained from this procedure were used without any further purification in cyclization experiments. AWWTNDFCA-NH2 (175 mg, 95%) were obtained as white solids following lyophilization. S20 5. Chemical synthesis: S1: Compound S1 was synthesized as described previously. [2] 4-aminobenzyl alcohol (500 mg, 4.06 mmol), TBDMS-Cl (612 mg, 4.06 mmol), and imidazole (553 mg, 8.12 mmol), were dissolved in anhydrous DMF (7 mL). The resulting solution was stirred for 1 h at room temperature. After that, water (20 mL) was added and the product extracted with EtOAc (3 x 50 mL). The combined organic layers were washed with saturated NaHCO3 (aq.), water, and brine, and dried over MgSO4. The solvent was evaporated and the residue purified by silica column chromatography (10-30% EtOAc in heptane) to yield 672 mg (70%) of TBDMS ether S1 as a pale-yellow oil. 1  S2: 2,6-lutidine (0.788 mL, 6.8 mmol) and bromoacetyl bromide (0.329 mL, 3.80 mmol) were added under N2 atmosphere at 4 °C to a solution of TBDMS protected 4-aminobenzyl alcohol (S1, 600 mg, 2.52 mmol) in dry DCM (6 mL). After stirring for one hour, the reaction was quenched by addition of 1 M HCl (aq., 10 mL). DCM was removed under vacuum and MeOH (20 mL) added. The reaction was then stirred for 30 min to allow for the removal of the TBDMS protecting group. Excess methanol was removed and the resulting water phase extracted three times with EtOAc (50 mL). The organic phases were combined, washed with water and brine and subsequently dried over MgSO4. EtOAc was removed in vacuo and the off-white product repeatedly washed with ice-cold DCM (3x 3 mL), to yield 200 mg S2 (68%) as a white solid. 1

2:
The bi-functional scaffold S3 was obtained by adapting a procedure of Toshima et al. [3] Dess- The reaction mixture was stirred for 1 hour at room temperature or until TLC analysis (30% EtOAc in heptane) confirmed full conversion. The reaction was quenched by the slow addition of MeOH (15 mL) and subsequently EtOAc (15 mL) was added. After the solution was stirred for 15 min at room temperature, the solvents were removed in vacuo. Water (25 mL) was added and the product was extracted with EtOAc (60 mL × 3). The extracts were washed with brine (60 mL), dried over anhydrous Na2SO4, filtered, and concentrated in vacuo. The residue was purified by silica column chromatography (45-100% EtOAc in heptane) to yield S3 (444 mg, 2.90 mmol, 48%) as a white solid. 1  S6: (6-Aminonaphthyl)methanol (S6) was obtained following a procedure described by Sergeyev et al. [4] LiAlH4 (1 M, 12.82 mmol) was added dropwise at 0 °C to a solution of (6aminonaphthyl)carboxylic acid (800 mg, 4.27 mmol) in dry THF (20 mL). After the addition was completed, the mixture was stirred for 1 h at room temperature and then heated to reflux for an additional 2 h. The mixture was allowed to reach room temperature and quenched by the dropwise addition of water (2 mL). The inorganic precipitate was filtered and the filtrate dried over MgSO4 and concentrated to give S6 (549 mg, 74%) as a slightly brownish solid. 1

Construction of pET21b(+)-D1D2 and pET21b(+)-AGSSGGC-D1D2:
The gene encoding for the disulfide-free D1D2 domains as well as AGSSGGC-D1D2 were amplified from fdg3p0ss21 [5] using the primers NdeI_D1D2_fw or NdeI_pep_D1D2_fw and XhoI_D1D2_rv, which also installed appropriate restriction enzyme sites for restriction digest (see Section 13) using the following PCR protocol: (1) initial denaturation 98 ˚C for 1 min, (2)    Samples were concentrated to 1 mL using Amicon® Ultra-15 (cut off of 100 kDa) centrifugation tubes. Next, reduced phages were washed three times with 12 ml ice-cold reaction buffer (20 mM HEPES, 5 mM EDTA, pH 8). Phages prone to accumulate on the filter during this procedure were resuspended by pipetting samples up and down repeatedly. After the final wash, samples were concentrated to 1 mL, transferred to a 15 mL tube and the volume adjusted with reaction buffer to 9 mL. At this point, a 'TCEP reduction' aliquot was taken and stored at 4 °C.

S32
Cyclization on phages: Cyclization units 1-4 (1 mL of 0.5-5 mM stock solutions in acetonitrile) were added to phage samples obtained after TCEP reduction and the resulting reaction mixtures were incubated at 30 °C for 1 h. The samples were subsequently concentrated to 1 mL and washed three times with reduction buffer (50 mM MES, pH 6) as described above.
Following the final wash, samples were concentrated to 1 mL, transferred to a 15 mL tube and the volume adjusted with reduction buffer to 9 mL. At this point, a 50 'Cys-modification' aliquot was taken and stored at 4 °C. NaBH3CN (2 x 200 µL from 25 mM stock) was added to the phage samples and the reaction was incubated overnight at 4 °C. The next morning, three more batches of NaBH3CN (3 x 200 µL from 25 mM stock) were added in 2 h time intervals to reach a final concentration of 2.5 mM in the reaction mixture. Samples were subsequently concentrated to 1 mL and washed three times with reduction buffer (50 mM MES, pH 6) as described above. Following the final wash, samples were concentrated to 1 mL, transferred to a 15 mL tube and the volume adjusted with reduction buffer to 10 mL. At this point a 50 µl 'NaBH3CN reduction' aliquot was taken and stored at 4 °C.
Infectivity studies: Stored 50 µL aliquots of the 'supernatant', 'PEG-purifaction', 'TCEP reduction', 'cyc-modification', and 'NaBH3CN reduction' were used to determine the phage infectivity after every handling step. For each sample, seven 10-fold dilutions in 2xYT were prepared. Aliquots (20 µL) of samples corresponding to 10 -5 , 10 -6 , and 10 -7 dilutions were added to 180 µL of E.coli TG1 cells growing in the mid-log phase (OD600 ~0.4). Phages in the samples were allowed to infect cells at 37 °C at 135 rpm. for 90 min. An aliquot (50 µL) of each dilution was then plated onto 2xYT/chloramphenicol agar plates and incubated overnight at 37 °C. The next day, colonies on the plates were counted and the number of infectious phages was calculated by adjusting for the corresponding dilution factors.

Library construction
The phage library displaying peptide variants of A(X)8C, where X is any amino acid, was constructed by adapting a previously established procedure. [6] Library construction consisted of the following steps: (1) PCR amplification of the genetic fragment GGSG-D1D2-SfiI; (2) construction of the library SfiI-A(X)8C-GSGG-D1D2-SfiI; and (3) cloning of the library SfiI-A(X)8C-GGSG-D1D2-SfiI into the recipient vector fd0D1D2 and transformation of the plasmid library into E. coli TG1 cells.

Preparation of GSGG-D1D2-SfiI:
In an initial PCR, the gene encoding for the disulfide-free D1D2 domains was amplified from fdg3p0ss21 [6] using the primers GGSG_fd_fw and fd_SfiI_rv. The PCR, which also appended a GGSG-linker and installed the downstream SfiI restriction site, was performed using the following protocol: (1) initial denaturation 95 ˚C for 2 min, (2) 30 cycles of denaturation at 95 ˚C for 50 s, annealing at 58 ˚C for 50 s and extension at 72 ˚C for 2.5 min; (3) a final extension at 72 ˚C for 10 min. Successful amplification as well as the size of the PCR products was verified by gel electrophoresis (1%). The remaining template was digested with DpnI at 37 ˚C for 16 hours. Following PCR purification, the resulting fragment was used for the construction of the library.

Construction of the fragment library SfiI-A(X) 8 C-GGSG-D1D2-SfiI:
The sequence encoding for the randomized peptide A(X)8C, as well as another SfiI restriction site were installed in a second PCR amplification. To a total of 100 ng of GGSG-D1D2-SfiI PCR product was added nuclease-free water (to obtain 0.5 mL final volume), 500 nM degenerate primer SfiI_AX8C_fw (final conc.), 500 nM fd_SfiI_rv (final conc.), 250 µM dNTP mix (final conc.) and 50 µL 10x Taq buffer. After mixing, the PCR master mix was distributed over ten 0.1 mL tubes on ice and 1 µL Taq polymerase (5 units) was added to each tube. The fragment library was obtained using the following PCR protocol: (1) initial denaturation 95 ˚C for 2 min, (2) 28 cycles of denaturation at 95 ˚C for 50 s, annealing at 60 ˚C for 50 s and extension at 72 ˚C for S34 2.5 min; (3) final extension at 72 ˚C for 7 min. The PCR product was purified by gel electrophoresis (1% agarose) followed by gel extraction (QIAGEN). The fragment library was eluted in Tris-HCl buffer (pH 8.5 to ensure compatibility with the buffer system used for SfiI digestion). 8  To assess the size of the library, a 20 µL aliquot of transformed cells was taken and a series of 10-fold dilutions were plated on 2xYT/chloramphenicol agar plates. The remaining mixture S35 was plated on large 2xYT/chloramphenicol agar plates and incubated overnight at 37 °C. The next day, colonies were counted on the dilution plates and cells were harvested from the large plates with 4 mL 2x YT per plate. To the harvested cells, glycerol was added to 20% and aliquots of the library (0.5 mL) were stored at -80 °C until further use.

Phage selections
Production fd_A(X) 8 C-D1D2 phages: Flasks containing 500 mL 2xYT-medium with 30 μg/mL chloramphenicol were inoculated with E. coli TG1 glycerol stock harboring the A(X)8C library to an OD600 of 0.1. After incubation overnight (30 °C at 200 rpm) phage particles were purified and subsequently all cysteines reduced with TCEP as described in Section 9.
Negative selection: For every 500 mL 2x YT starting culture, 50 µL of streptavidin coated magnetic beads (New England BioLabs) were transferred to a 1.5 mL Eppendorf tube and placed in a magnetic rack (NEB). The supernatant was removed and the beads were washed 3 times with 1 mL of binding buffer (10 mM Tris-Cl pH 7.4, 150 mM NaCl, 10 mM MgCl2, 1 mM CaCl2). After the final wash, the beads were resuspended in 300 µL binding buffer, complemented with 150 µL blocking buffer (binding buffer, 0.3% Tween-20, 3% (w/v) BSA) and incubated on a slowly rocking benchtop shaker for 30 min at room temperature. After that, the beads were washed with 1 mL binding buffer and added to the reduced phages in reduction buffer (2 mL) in 15 mL falcon tubes. The phage-bead suspension was incubated for 30 min at room temperature before being placed back into a magnetic rack. The supernatant was removed and used for the following cyclization on phages (vide infra). In parallel, the remaining magnetic beads were washed twice with 1 mL binding buffer, replacing the Eppendorf tubes after every wash to eliminate phages binding to plastic. Linear peptides binding to streptavidin were next eluted by resuspending the beads in 100 µL elution buffer (50 mM glycine-HCl, pH pelleted phages were re-dissolved in 4.5 mL reduction buffer (50 mM MES, pH 6). NaBH3CN (1 x 100 µL from 50 mM stock) was added to the phage sample and the reaction was incubated overnight at 4 °C. The next morning, three more batches of NaBH3CN (3 x 100 µL from 50 mM stock) were added in 2 h time intervals. The following day, a final batch (100 µL of 50 mM stock) was added to reach a final concentration of 5 mM in the reaction mixture. On the following day, samples were concentrated to 1 mL and washed three times with binding buffer as described before. Following the final wash, samples were concentrated to 2 mL and transferred to a 15 mL falcon tube.
Positive selection: Strep-coated magnetic beads (2x 50 µL) were transferred to a 1.5 mL Eppendorf tube, washed three times with 1 mL binding buffer as described previously (see negative selection) and resuspended in 300 µL binding buffer complemented with 150 µL blocking buffer per sample. The beads were incubated on a slowly rocking benchtop shaker for 30 min at room temperature. In parallel, each phage library in 2 mL binding buffer was blocked by addition of 1 mL blocking buffer and incubated for 30 min at room temperature. During the incubation, flasks containing 2x YT (50 mL per phage library) were inoculated with 500 µL of a 5 mL overnight culture of E. coli TG1 cells (37 °C at 135 rpm). Magnetic beads were added to blocked phage libraries in new 15 mL tubes and incubated on a slowly rocking benchtop S38 shaker for 30 min at room temperature. Next, the tubes were transferred into a magnetic rack and the supernatant removed. The beads were washed eight times with 1 mL washing buffer (binding buffer, 0.1% Tween-20) and twice with 1 mL binding buffer. During the washes, tubes were replaced at least six times to remove plastic-binding peptides as described previously.
After the final wash, the beads were resuspended in 100 µL elution buffer (50 mM glycine-HCl, pH 2.2) for 5 min before placing the tube(s) back into the magnet and transferring the supernatant to new tubes containing 50 µL neutralization buffer (1 M Tris-Cl, pH 8). A 5 µL aliquot of each library was added to 45 µL 2x YT and stored at 4 °C as 'positive selection' sample.
Eluted phage particles (145 µL) were added to 25 mL of the E. coli TG1 culture grown to an OD600 of 0.4 and incubated for 90 min. at 37 °C without shaking. Subsequently, cells were pelleted by centrifugation at 4000 rpm for 8 min at 4 °C. The pellets were resuspended in 600 µL 2x YT medium and plated on three 2x YT/ chloramphenicol plates and incubated at 37 °C overnight. The next day, cells were harvested with 1 mL 2x YT per plate, glycerol was added to 20% and aliquots of 0.5 mL were stored at -80 °C for use in the next round of selection.
Infectivity studies: Stored aliquots of the 'negative selection' and 'positive-selection' samples were used to determine the number of retained phages in the different selections over 3 rounds of phage display (see workflow described in Section 9).

HABA competition binding assay
Production of SAV_1-3 displaying bacteriophages: To validate the selection of cyclic peptide binders over their linear, and modified counterparts, phages displaying the peptide sequences of the two most enriched hits SAV_1, SAV_2 and SAV_3 were panned against streptavidin either in linear, modified or cyclic peptide form. Phage titers pre-and poststreptavidin panning were determined and compared between all the samples.
In brief, 700 μL of densely grown E. coli TG1 overnight cultures harboring the AWWTNDFCAC-fd plasmid (SAV_1), AWWEDRGPPC-fd plasmid (SAV_2) and ARNVGLVGMC-fd plasmid (SAV_3) respectively, were used to inoculate 500 mL 2xYTmedium containing 30 μg/mL chloramphenicol. After incubation overnight (30 °C at 190 rpm) phage particles were purified and all cysteines were reduced with TCEP as described in Section 9. At this point the phage samples were split into 3 fractions of 2.5 mL. One sample was stored at 4 °C as 'linear phage' sample, the other two samples were modified with scaffold 3 (SAV_1 and SAV_2) or scaffold 4 (SAV_3) and purified as described in Section 9. One 'modified phage' sample was stored at 4 °C, while the other was cyclized as described previously, yielding the 'cyclized phage' sample. All samples were subsequently purified by PEG-precipitation and dissolved in 1 mL reduction buffer for consistency (50 mM MES, pH 6). Phage titers of these samples were determined ( Figure S12A) and all samples were diluted to approximately 10 11 colony forming units using reduction buffer. Of all samples, 50 μL 'pre-SAV' aliquots were stored at 4 °C.
Streptavidin panning: Streptavidin-coated magnetic beads (190 μL) were first washed, then resuspended in 540 μL of binding buffer, and 270 μL blocking buffer (binding buffer, 0.3% Tween-20, 3% (w/v) BSA) added as described in Section 11. The resulting mixtures were incubated while slowly rocking for 30 min at room temperature, after which the beads were washed with 1 mL of binding buffer. In parallel, 500 μL blocking buffer was added to 950 μL S40 of the previously prepared phage samples (adjusted for equal phage concentrations). The mixtures were incubated on a slowly rocking benchtop shake plate for 30 min at room temperature. Next, pre-treated magnetic beads were divided over the blocked phage samples (20 μL SAV beads per sample, corresponding to ~1.5 nmol SAV-tetramer), and the resulting mixtures were again incubated while slowly rocking for 30 min at room temperature. The tubes were placed in a magnetic rack, and the supernatant was removed. After this, the beads were washed with washing buffer (7 x 500 μL, binding buffer, 0.1% Tween-20), and the tubes were replaced two times to eliminate phages binding to plastic. After the final washes (2 x 500 μL, binding buffer), the beads were resuspended in 100 μL elution buffer (50 mM glycine-HCl, pH 2.2) for 5 min., before placing the tubes back in the magnetic rack. The supernatant was subsequently transferred to new tubes containing 50 μL neutralization buffer (1 M Tris-Cl, pH 8). The elution mixtures were then used together with all the stored 'pre-SAV' samples to determine the phage titers as described in Section 9 (Figs. S12B-C).

In vitro affinity measurements:
To determine the affinity of the selected peptides for Streptavidin, a competitive binding experiment was performed similarly to previously described literature procedures. [7] In brief, for the competitive binding experiment, increasing amounts of peptide (final concentration 0.01 -500 μM in 12.5% ACN + 0.1% TFA in PBS phosphate buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH = 7.4)) were titrated into a PBS solution containing fixed concentrations of HABA/ Streptavidin monomer (final concentration 1.5 mM/ 40 µM (SAV_1), 33 µM (SAV_2), 32 µM (SAV_3)). The reactions were allowed to reach equilibrium at room temperature (60 min.) and the characteristic absorbance of the HABA/ streptavidin complex at 500 nm was measured.