Mirror-image ligand discovery enabled by single-shot fast-flow synthesis of D-proteins

Callahan, Alex J.; Gandhesiri, Satish; Travaline, Tara L.; Reja, Rahi M.; Lozano Salazar, Lia; Hanna, Stephanie; Lee, Yen-Chun; Li, Kunhua; Tokareva, Olena S.; Swiecicki, Jean-Marie; Loas, Andrei; Verdine, Gregory L.; McGee, John H.; Pentelute, Bradley L.

doi:10.1038/s41467-024-45634-z

Download PDF

Article
Open access
Published: 28 February 2024

Mirror-image ligand discovery enabled by single-shot fast-flow synthesis of D-proteins

Nature Communications volume 15, Article number: 1813 (2024) Cite this article

4276 Accesses
2 Citations
1 Altmetric
Metrics details

Subjects

Abstract

Widespread adoption of mirror-image biological systems presents difficulties in accessing the requisite D-protein substrates. In particular, mirror-image phage display has the potential for high-throughput generation of biologically stable macrocyclic D-peptide binders with potentially unique recognition modes but is hindered by the individualized optimization required for D-protein chemical synthesis. We demonstrate a general mirror-image phage display pipeline that utilizes automated flow peptide synthesis to prepare D-proteins in a single run. With this approach, we prepare and characterize 12 D-proteins – almost one third of all reported D-proteins to date. With access to mirror-image protein targets, we describe the successful discovery of six macrocyclic D-peptide binders: three to the oncoprotein MDM2, and three to the E3 ubiquitin ligase CHIP. Reliable production of mirror-image proteins can unlock the full potential of D-peptide drug discovery and streamline the study of mirror-image biology more broadly.

Pooled multicolour tagging for visualizing subcellular protein dynamics

Article Open access 19 April 2024

De novo design of protein structure and function with RFdiffusion

Article Open access 11 July 2023

Luciferase- and HaloTag-based reporter assays to measure small-molecule-induced degradation pathway in living cells

Article 18 April 2024

Introduction

The biopolymers that carry out the central dogma of all known life are generated from a set of homochiral building blocks: d-deoxyribonucleic acids for DNA, d-ribonucleic acids for RNA, and l-amino acids for proteins. When strung together, these chiral monomers form three-dimensional structures with biological activities that are inherently chiral. Alternatively, unnatural biopolymers made from building blocks of the opposite chirality form mirror-image three-dimensional structures. These compounds retain the activity of their natural counterpart in the mirror-image biochemical space, but are not recognized nor degraded by natural biomolecules^1,2,3,4. As a result, mirror-image biopolymers are useful tools to probe or perturb living systems. One such class of mirror-image compounds, D-peptides, are particularly valuable as therapeutic scaffolds, but remain under-developed.

Metabolic stability is a key bottleneck for the design of peptide-based therapeutics, and mirror-image peptides are uniquely resistant to degradation. The greater chiral and structural complexity of peptides make them valuable scaffolds to target protein surfaces not well-suited to disruption with small molecules⁵. Despite these benefits, unmodified canonical peptides are rapidly degraded in vivo, and chemical modification to extend their biological half-life is critical in the design of peptide-based drugs. Modern techniques for peptide drug discovery integrate stabilizing modifications early into design cycles, but stability is constantly monitored throughout development⁶. Mirror-image d-peptides, on the other hand, are not recognized by natural proteases, and therefore display long half-lives in biological systems on their own^7,8. As such, d-peptides are a privileged scaffold for peptide drug discovery, especially when combined with macrocyclic linkages, also termed staples. Despite these benefits, modern discovery efforts rarely utilize mirror-image peptides, as the high-throughput biological display techniques critical to modern peptide drug discovery efforts are not compatible with d-peptides. Development of mirror-image peptide-based therapeutics is limited to techniques that as of yet have not had sufficient throughput for modern drug discovery efforts.

Mirror-image phage display (MIPD) can generate high affinity d-peptide ligands to challenging protein interfaces, but its scope has remained limited. In MIPD, a library of l-peptides displayed on phage surfaces are screened against a ‘bait’ mirror-image protein⁹. By symmetry, the interaction of a d-peptide with an l-protein is identical to that of the screened l-peptide to a d-protein. As a result, the discovered l-peptide binders provide a blueprint for mirror-image d-peptide binders to the natural l-protein. This strategy has been employed to generate d-peptide ligands to therapeutically relevant protein targets, including mouse double minute 2 homolog (MDM2)¹⁰, epidermal growth factor (EGF)¹¹, and amyloid beta (Aβ)¹². Despite these successes, existing MIPD pipelines are low-throughput and, as a result, are not practical for modern drug discovery; to the best of our knowledge, MIPD has only been applied to nine protein targets^{9,10,11,12,13,14,15,16,17,18,19,20} during the 30 years since its introduction.

A significant limitation to the broad application of MIPD relates to the challenges associated with accessing the required synthetic d-proteins²¹. Modern chemical protein synthesis (CPS) techniques have enabled access to a diverse range of synthetic proteins^22,23, but require individualized optimization for each target. Sequence-dependent synthetic challenges are ubiquitous, and chemical approaches to mitigate these problems are required. For example, aggregation disruptors^24,25,26, solubilizing tags^27,28, and ligation site alterations²⁹ are commonly used in modern CPS efforts. Our group implemented automated fast-flow peptide synthesis^30,31 (AFPS), an efficient technology which may obviate the need for individualized synthesis optimization. To our knowledge, 36 d-protein targets have been reported to date^{2,3,4,11,14,15,19,20,29,32,33,34,35,36,37,38,39,40,41,42,43,44,45}. Despite the success of these strategies, their implementation requires many rounds of trial and error to address the specific complications of each sequence. As a result, each protein target presents a unique synthetic challenge that, although surmountable, requires significant effort. These considerations have rendered MIPD low throughput and, as a result, unable to compete with alternative discovery pipelines. A recently reported screening platform based on phage display generated high-affinity and conformationally constrained α-helical peptide binders⁴⁶.

Here, we use AFPS to prepare a panel of 12 diverse proteins (Fig. 1). We selected intracellular protein targets with underexplored biology (ERG and IRAK2), targets that are historically important to chemical protein synthesis (MDM2 and Barnase), or proteins that are emerging (CHIP, YAP1, NEMO, NEMO_iZIP) or classical (Myc-Max, Max-Max, BCL11a, and FKBP12) therapeutic targets. Of these sequences, eight had not been previously synthesized by chemical methods. We successfully isolate milligram amounts of these 12 proteins in both l- and d-forms (24 synthetic proteins total) (Fig. 1). We discover D-macrocyclic binders to two of the proteins synthesized here, the oncoprotein MDM2 and the C-terminus of Hsc70-interacting protein (CHIP). Analysis of the X-ray co-crystal structures between the discovered d-peptide binders and their protein targets reveals similar side-chain interactions to known binders that utilize different structural scaffolds, including loops and α-helices with opposite handedness. Together, these results represent a major advance in the throughput and general accessibility of d-peptide binder discovery.

**Fig. 1: Rapid generation of d-peptide binders leverages automated flow protein synthesis coupled with mirror-image phage display.**

Results

Synthesis of enantiomeric protein pairs with AFPS

We first defined a panel of 12 structurally diverse single-domain proteins for generation of synthetic l- and d-versions using AFPS (Fig. 2). Two of these proteins, MDM2¹⁰ and barnase⁴⁷, were previously prepared in their all-d forms (see SI Sections 2.3 and 2.5), and therefore can be used to assess the reliability and fidelity of the AFPS approach. We also chose two proteins that had previously only been synthesized in their native l-forms: Myc-Max, and Max-Max⁴⁸ (see SI Sections 2.13–2.19). We chose an additional eight single-domain proteins ranging in length from 70–132 amino acids with no reported chemical synthesis: ERG (PNT domain) (see SI Section 2.4), IRAK2 (death domain) (see SI Section 2.6), CHIP (tetranucleotide repeat domain) (see SI Section 2.7), NEMO (coiled coil domain) (see SI Section 2.8), FKBP12 (see SI Section 2.9), BCL11a (Zn finger domains) (see SI Section 2.10), YAP1 (ww1-ww2 domains) (see SI Section 2.11) and NEMO_iZIP (coiled coil domain with iZip adapters⁴⁹) (see SI Section 2.12). In all cases, we appended a biotin unit at the N-terminus of the protein through a PEG₁₂ linker to facilitate phage display screening.

**Fig. 2: Automated flow synthesis delivers diverse single-domain protein chains in both enantiomeric forms.**

We used AFPS to prepare each polypeptide chain in a stepwise fashion, with total synthesis times ranging from 4–7 h, using three separate AFPS instruments. After we released the proteins from the H-Rink Amide solid support, we isolated them by precipitation with cold diethyl ether (see Methods). We used analytical reverse phase HPLC (RP-HPLC) and liquid chromatography–mass spectrometry (LC-MS) to analyze crude peptide mixtures and preparative mass-directed RP-HPLC to purify them. We typically obtained ~5 µmol of crude polypeptide powder from individual synthesis runs, and purifying ~3 µmol of this material afforded ~0.3 µmol of pure peptide. Gradient shape for the preparative purification was determined by a preliminary run at low protein loading on the same column used for purification (see SI Section 1.4). Typical yields are on the order of 1–10 mg (0.07–0.7 µmol, 0.3%–3% isolated yield based on resin loading). In all instances, a single synthesis experiment afforded sufficient material after folding for phage selections and biochemical validation.

Folding of diverse d-proteins with preparative size exclusion chromatography

Using a generalized folding protocol, each synthetic protein was folded to a homogeneous product as analyzed by size-exclusion chromatography (SEC). We set out to identify appropriate folding conditions for our 12 enantiomeric protein pairs and began by adopting the protocols that had been used to successfully isolate folded protein in an aqueous solution for the few targets for which literature reports exist. Specifically, L/D Barnase³¹ and L/D MDM2⁵² were previously isolated via stepwise dilution from a high concentration of guanidine hydrochloride. Subsequent purification using semi-preparative SEC achieves yields ranging from 4.5 to 10.6 nmol for all four targets. Here, using this technique, we were unable to identify any dilution conditions under which we could isolate folded BCL11a or IRAK2. Instead, we observed significant precipitation for both target pairs, with any remaining solubilized material eluting in the exclusion volume of the column, indicating an apparent molecular weight at least 10-fold over the expected value, consistent with the formation of soluble aggregates.

To address these complications, we adapted a refolding technique utilizing SEC to isolate the remaining protein targets^53,54. Purified proteins were dissolved in a denaturing buffer, typically a 10 mg/mL solution containing 6 M guanidine hydrochloride buffered to pH 7.5 with 50 mM tris(hydroxymethyl)aminomethane hydrochloride (TRIS HCl) and including 50 mM dithiothreitol (DTT). The mixture was then submitted to semi-preparative SEC coupled to an HPLC instrument. The denaturant is removed as the protein progresses through the SEC column, and the resulting folded protein can be separated from off-target aggregates based on their different retention times. The running buffer of the SEC could be tuned to the requirements of each protein and unless otherwise specified was 50 mM HEPES, 150 mM NaCl, 0.5 mM DTT, 5% glycerol (v/v), at pH 7.5. DTT was omitted in cases where the protein contained no free sulfhydryl groups (NEMO, NEMO_iZIP, Myc-Max, and Max-Max). For BCL11a, which contains three Zn-finger motifs, both the denaturing buffer and the running buffer were adjusted to contain ZnCl₂ with additional alterations to prevent precipitation of Zn salts. In each case, the folding protocol afforded material as major elution peaks with retention times consistent with the calculated molecular weights (SI Sections 2.20 to 2.31). This set includes BCL11a, and IRAK2 for which we were unable to isolate folded material by any other technique. Some proteins elute as peaks with apparent molecular weight larger than would be otherwise expected, including CHIP, NEMO, NEMO_iZIP, and BCL11a. In the case of NEMO, and NEMO_iZIP, this behavior has been established and has been attributed to their extended structures^49,55.

Synthetic d- and l-proteins display expected biochemical activity

For each enantiomeric pair generated using AFPS, we used biochemical assays to confirm that the activities of the synthetic proteins are similar to their recombinant versions. For proteins with known binding partners, we measured the affinities of the synthetic proteins to fluorophore-tagged probes of the appropriate chirality using time-resolved fluorescence resonance energy transfer (TR-FRET). Binder substrates were modified with 5(6)-carboxyfluorescein (FAM), and the biotinylated target proteins were complexed with terbium-labeled streptavidin. Binding was measured as an increase in fluorescence quenching between the two fluorophores. For BCL11a binding to DNA (Fig. 3A and Supplementary Table 1), CHIP binding to HSP peptide (Fig. 3B and Supplementary Table 2), MDM2 binding to a p53-derived peptide (Fig. 3I and Supplementary Table 3), Max-Max binding to E-Box DNA (Fig. 3E and Supplementary Table 5), and Myc-Max binding to E-Box DNA (Fig. 3F and Supplementary Table 4), each synthetic chiral protein displayed apparent dissociation constants (K_D) for the l- and d-proteins consistent with those of the recombinant proteins (Fig. 3L). We next used surface plasmon resonance (SPR) binding assays with substrates of the matching chirality to assess binding of NEMO to iKKb peptide (see Supplementary Fig. 2), NEMO_iZip binding to iKKb peptide (see Supplementary Fig. 5), YAP1 binding to dendrin (see Supplementary Fig. 4), and l-FKBP12 binding to rapamycin followed by mTOR (see Supplementary Fig. 3). Again, in all instances, the synthetic proteins displayed binding affinities similar to the recombinantly derived proteins, suggesting that the synthetic proteins derived from AFPS are folding into bioactive tertiary structures.

**Fig. 3: Synthetic d-proteins display similar bioactivity to their synthetic l- and recombinant counterparts.**

Synthetic L/D barnase from AFPS displays ribonucleolytic activity that is selective for the chirality of its nucleotide substrate. Barnase catalytically cleaves a fluorescently quenched reporter RNA substrate⁵⁷ composed of a tetranucleotide flanked with FAM and 6-carboxytetramethylrhodamine (6-TAMRA) labels on either end (FAM-DdA-DrU-DdA-DdA-TAMRA), where each nucleotide is specified as the l- or d-enantiomer and the ribose (r) or deoxyribose (d) sugar. We tracked the barnase-mediated cleavage of the substrate by assessing the FRET between the FAM and TAMRA fluorophores, which is abolished upon cleavage. When the chirality of the substrate and protein were matched, as for l-barnase to the d-substrate FAM-DdA-DrU-DdA-DdA-TAMRA, and d-barnase to the l-substrate FAM-LdA-LrU-LdA-LdA-TAMRA, we observed rapid hydrolysis with extracted kinetic constants that are similar to previously reported values (Fig. 3J, K and Supplementary Table 6)⁴⁷. In the mismatched cases, l-barnase to l-substrate and d-barnase to d-substrate, we observed no catalysis on the timescale of our observation (Fig. 3J, K and Supplementary Fig. 1). These results reconfirm the structural and biological integrity of the folded synthetic proteins, indicating that their enantiomeric purities are retained throughout the synthesis and purification processes.

For protein targets with no known binding partners and no available activity assays (ERG and IRAK2), we recorded circular dichroism (CD) spectra and compared them to recombinantly derived material. Both targets are reported to be mostly alpha-helical^58,59, and we confirmed this feature for the recombinant proteins. The CD spectra for the synthetically derived l-proteins closely matched those of the recombinant versions, indicating that the synthetic material is forming secondary structures in similar proportions (Fig. 3C, G and SI Section 1.12). Both d-ERG and d-IRAK2 display a CD signal with similar absolute intensities as the corresponding synthetic l- and recombinant proteins, but with inverted sign, consistent with the formation of mirror-image secondary structures. Furthermore, we found that both synthetic protein pairs form folded structures with melting temperatures (T_m) close to those of the recombinantly derived proteins, as shown by tracking the intensity of the α-helical signature as a function of temperature (Fig. 3D, H and SI Section 1.12). Together, these data suggest that the synthetically derived l/d-ERG and l/d-IRAK2 form tertiary structures that closely resemble the native proteins.

Generation of macrocyclic d-peptide ligands to MDM2 using mirror-image phage display

Having validated the structure and function of our synthetic d-proteins, we sought to benchmark our MIPD screening platform with a model target, MDM2, an E3 ubiquitin ligase that recognizes the FXXXWXXL motif present on its substrate, p53, a critical mediator of cell cycle arrest, senescence and apoptosis⁶⁰. As a result, p53 degradation driven by upregulation of MDM2 is found to accelerate growth in a variety of cancers⁶¹. To disrupt p53 recognition, and thus cancer progression, a number of high-affinity MDM2 binders have been developed and evaluated in the clinic⁶². In particular, highly-potent biologically stable linear d-peptide ligands to MDM2 have been generated with mirror-image phage display, including ^dPMI-α (K_D = 219 nM)⁶³, ^dPMI-β (K_D = 219 nM)¹⁰, and ^dPMI-γ (K_D = 53 nM)⁶³. We set out to investigate if our MIPD platform could rediscover similar motifs, and perhaps generate new candidates for MDM2 inhibition. To that end, we successfully isolated synthetic mirror image MDM2, and carried it into our MIPD platform⁴⁶.

We identified three binding clusters selective for d-MDM2 with phage display screening (Fig. 4A and SI Section 1.26). In brief, immobilized d-MDM2 from AFPS was screened against an unbiased phage library of 10⁸ members displaying macrocyclic α-helical peptides. We sequenced phage particles that remained bound to the targets after washing using next-generation sequencing. Comparisons of sequencing reads to a spiked-in internal reference allowed for a semi-quantitative read-out of binding affinity as a function of protein bait concentration. The resulting hit sequences that displayed dose-response were clustered into binder families using hierarchical statistical clustering. We identified three such clusters to d-MDM2, and logo plots of their sequence overlaps are shown in Fig. 4A (MDM2.C1-MDM2.C3). Consistent with previously described α-helical peptide binders to MDM2^64,65, all three clusters contain anchoring N-terminal tryptophan residues (dW2 in MDM2.C1, dW1 in MDM2.C2, and dW1 in MDM2.C3), and show conserved hydrophobic residues on the same α-helical face as the tryptophan residue (i,i + 4/5 and i,i + 8/9 relative to dW on each cluster). To validate the specificity of each binding cluster, we synthesized representative binding peptides in the d-form (Fig. 4B, d-H101 from MDM2.C1, d-H102 from MDM2.C2, and d-H103 from MDM2.C3) and measured their affinity to recombinant MDM2 with SPR. We observed binding constants of 5.5 μM, 2.4 μM, and 0.88 μM, respectively (Fig. 4C and Supplementary Figs. 6–8).

**Fig. 4: Mirror-image phage display of d-MDM2 generates macrocyclic d-peptide binders to l-MDM2.**

To further investigate the binding modes of each binding cluster, we solved the X-ray co-crystal structures of d-H101 (Fig. 5A, resolution 1.64 Å), d-H102 (Fig. 5B, resolution 1.30 Å), and d-H103 (Figs. 5C, resolution 1.90 Å) with recombinant MDM2 (see SI Section 3). All three peptides form left-handed α-helices that engage the same hydrophobic groove on MDM2 but make use of different side-chain interactions. d-H101 and d-H102 use similar amino acids to interact with MDM2 at its N-terminus (dW5/dW4, dY6/dH6, and dF9/dY9), but diverge towards the C-terminus where the side-chain conformations are altered (dQ12/dF12 and dE16/dF16) (Fig. 5A, B). Beyond the conserved N-terminal aromatic residues, d-H103 makes use of altered side-chains to interact with MDM2 compared to either d-H101 or d-H102 (Fig. 5C vs Fig. 5A, B).

**Fig. 5: Discovered d-macrocycles bind to a common hydrophobic groove on MDM2.**

Consistent with their sequence similarities, d-H101 utilizes similar side-chain interactions to two previously reported α-helical peptide binders to MDM2. Overlay of the known d-peptide MDM2 binder ^dPMI[1–5, 9–12]⁶⁴ onto the d-H101 MDM2 co-crystal structure revealed that both peptides present into the same groove on MDM2 with the same axial direction (N to C), and make use of similar side-chains (Fig. 5D). Chemically related side-chains project with similar orientations for both peptides, perhaps due to their matched left-handed helices. Matching helicity, however, is not an exclusive requirement to make use of similar side-chain interactions. ATSP-7041 is a high-affinity l-peptide binder to MDM2 (K_D = 0.9 nM)⁶⁵ and its structure overlaid onto the d-H101 MDM2 co-crystal structure is shown in Fig. 5C. Despite ATSP-7041 forming a right-handed alpha helix, both peptides appear to use similar binding modes. Important side-chain interactions of ATSP-7041 closely overlap with equivalent interactions for d-H101 (Fig. 5C, dW5/F3, dY6/Y6, dF9/W7, and dV13/Cba10).

Generation of macrocyclic d-peptide ligands to CHIP using mirror-image phage display

Not limited to known d-proteins, we nominated an original target for MIPD screening, CHIP, and identified three macrocyclic d-peptide binders. CHIP is a member of the RING/U-Box family of E3 ligases that direct the ubiquitination of chaperones⁶⁶ and their bound client proteins^67,68 (Fig. 6A). This interaction is driven by a common (I/M)EEVD motif present on the C-terminus of heat shock protein 70 (Hsp70) and Hsp90 family members that is recognized by the TPR domain of CHIP⁶⁹ (Fig. 6B). Degradation driven by CHIP is active in a diverse range of cellular processes from monitoring protein quality control⁶⁸ to mediating interferon-γ signaling⁷⁰. Masking of the CHIP-TPR domain can in theory modulate these signaling pathways and as a result is actively being pursued^69,71. In addition, CHIP belongs to the same protein family as MDM2 (E3 ubiquitin ligases) and has potential therapeutic relevance by being implicated in neurological disease. For these reasons, we hypothesized that this domain would be a valuable target for d-peptide discovery, and successfully isolated the protein in its mirror image form for MIPD screening. From a single-round phage screen, we identified three clusters that displayed robust binding to CHIP (CHIP.C1–CHIP.C3) (SI Section 1.27). All three clusters feature conserved terminal residues with carboxylate functionalities (dD3 in CHIP.C1, dD1/dE2 in CHIP.C2, and dD12 in CHIP.C3), consistent with the conserved terminal aspartic acid residues found on CHIP substrates. Hydrophobic residues are retained in the i,i + 4/5 positions relative to the anchoring aspartic acid residue: dW6 in CHIP.C1, dW6 in CHIP.C2, and dA8 in CHIP.C3, presumably for interaction with the hydrophobic TPR groove. We validated binding of selected sequences from the three clusters to the recombinant CHIP-TRP domain (IC₅₀ values of 0.47 μM, 0.55 μM, and 1.23 μM, respectively, Fig. 6C) in a competition fluorescence polarization assay against FAM-Hsp70 loaded CHIP (see Supplementary Fig. 15).

**Fig. 6: D-macrocyclic peptides identified from MIPD are high-affinity ligands to CHIP.**

X-ray co-crystal structures of each discovered d-macrocycle with recombinant CHIP revealed a common binding pocket despite variable side-chain interactions. We solved the X-ray co-crystal structures of each peptide with recombinant CHIP (Fig. 6B–D, 1.72 Å, 1.59 Å, and 1.76 Å resolution, respectively) (see SI Section 3). As predicted, all three peptides form left-handed α-helices to engage the same groove, though d-H202 binds in the reverse direction compared to d-201 and d-203 (Fig. 6C). We also solved the co-crystal structure of CHIP in complex with D-H204 at a resolution of 2.21 Å. We identified d-H203 and d-H204 from the same phage cluster. As expected, they share a similar amino acid sequence and binding mode. Despite varying residues, side-chains projecting into the CHIP surface across all binders have significant spatial overlap (Fig. 6D). Two loci for hydrophobic interactions on the CHIP surface in particular show strong side-chain overlap: dW7/dL6/dW7, and dF8/Staple/dM4 (of d-H201, d-H202, and d-H203, respectively). Additionally, the common terminal carboxylate-containing residues engage in similar hydrogen bonding networks with CHIP. These similarities extend to known biological ligands to CHIP, and an overlay of a co-crystal structure of the C-terminus of Hsc70 onto d-H201 is shown in Fig. 6E. Biological ligands to the CHIP TPR domain are known to form distinct kinked backbone conformations⁶⁹, unlike the left-handed alpha-helices of the discovered macrocycles. Despite this feature, the biological ligand utilizes similar side-chain contacts to engage CHIP. Both hydrophobic side-chain sites found in the discovered macrocycles (dF8/V4 and dW7/I1 in d-H201 and Hsc70, respectively) and the terminal carboxylate are also found in the biological CHIP ligand. These similarities indicate that the discovered macrocyclic d-peptides utilize a native binding mode translated into a left-handed alpha-helical interface not characterized previously.

Discussion

Peptide scaffolds are increasingly more often applied to new therapeutic targets and D-peptides in particular have great potential to be broadly applicable. Native, canonical peptides are notoriously unstable in vivo, and this reputation has complicated their application as therapeutics⁷². Increasingly effective and easily accessible chemical stabilization strategies have revitalized peptides as effective therapeutic scaffolds⁶. With judicious optimization, newly reported peptide therapeutics routinely display long biological half-lives. For example, after optimization, the candidate PCSK9 inhibitor MK-0616 displays nearly complete stability against key biological proteases⁷³, and as a result is dosed orally in its Phase 2b clinical trial⁷⁴. However, such achievements still require a complex and meticulous drug development process. Mirror-image D-peptides, on the other hand, should display substantial metabolic stabilities. Routine implementation of screening strategies that afford D-peptide ligands stands to greatly simplify and accelerate peptide drug development.

We demonstrate a robust and scalable platform that uses automated flow synthesis technology to facilitate access to mirror-image proteins for D-peptide discovery using mirror-image phage display. Among the main obstacles to the broad implementation of MIPD are the inconsistent and unpredictable outcomes of the total chemical synthesis of the requisite mirror-image proteins and their fragments. In this work, we demonstrate that AFPS technology can reliably supply mirror-image proteins for MIPD in a single shot. We successfully isolated 12 mirror-image proteins with AFPS, 8 of which have no previously reported syntheses. Without AFPS, these efforts would likely require significant individualized synthetic route optimization and method troubleshooting. We acknowledge that individual optimization may be needed for the flow synthesis of specific protein sequences, identified in our laboratory on a case-by-case basis. This was not the case here, however, as each target protein was synthesized successfully on the initial attempts according to our established protocol.

In our work, these mirror-image proteins were rapidly prepared in a single shot for screening in a standardized format and the ease of synthesis benefitted from their relatively small size (<150 amino acids). Making MIPD routinely applicable to larger protein targets may likely require combining flow synthesis with ligation methods. Our ongoing efforts are focused on leveraging these two powerful synthetic approaches. Continued improvement of d-protein synthesis quality is key to making MIPD practical for modern screening pipelines.

With practical and reliable access to mirror-image proteins, we describe the discovery of six macrocyclic d-peptide ligands to two protein targets. We first validated our discovery platform with screening of MDM2, a negative regulator of p53 known to be overexpressed in a variety of cancers⁶¹. MDM2 is one of a few previously reported targets for mirror-image phage display¹⁰, and we successfully rediscovered sequence motifs that are similar to these earlier reports. Furthermore, we recorded X-ray co-crystal structures of three mirror image binders to MDM2. Collectively, these results confirm the fidelity of our synthesis approach, and in addition provide new therapeutic leads for targeting MDM2. We further nominated CHIP for mirror image phage display screening, and again identified three macrocyclic d-peptide ligands. CHIP is an E3 ubiquitin ligase that directs the degradation of cellular chaperones and their bound client proteins⁶⁶. Inhibition of the CHIP ligand binding site is an emerging approach to modulate interferon-γ signaling⁷⁰. We anticipated that mirror-image peptide ligands could prove a valuable tool to engage this site. The three discovered peptide ligands all bound to recombinant CHIP and were able to displace a native CHIP ligand at low micromolar concentrations. Characterization of each mirror-image ligand by X-ray co-crystal structures with recombinant CHIP revealed a similar binding site to the native CHIP ligands, indicating their potential to disrupt this interaction.

Collectively, the results reported here represent a significant fraction of the successful d-protein syntheses to date. Widespread application of MIPD is hindered by the individualized attention required to prepare the mirror-image d-protein substrates, and we show that APFS addresses these challenges by delivering mirror-image proteins for screening using a standardized, rapid format. We anticipate that MIPD enabled by AFPS is poised to revitalize campaigns to generate mirror-image binders to existing and emerging protein targets.

Methods

General methods

Unless otherwise specified, all reactions with polypeptides, proteins, and protein oxidative addition complexes were set up on the bench top and carried out under ambient conditions. Unless stated otherwise, all small (≤1 mL) volumes were measured using Eppendorf Research® plus, single-channel, variable, mechanical pipettes (referred to as mechanical pipettes). Universal low retention pipet tips (10 µL, 200 µL and 1000 µL sizes) were purchased from VWR International (Philadelphia, PA). Care was taken to use the appropriate mechanical pipette/tip combinations to ensure a dispensing error of ≤ 2% for volumes between 10 µL and 1000 µL and ≤4% for volumes between 1 µL and 10 µL. To avoid loss of protein due to non-specific adsorption, plastic tubes (Eppendorf Protein LoBind® tubes, 0.5, 1.5, and 2.0 mL) were used in all cases after folding. The weight of lyophilized powders of the peptides was measured using analytical scales (XS205DU Analytical Balance, Mettler-Toledo) with an attached SPI Westek Workstation Still Air Ionizer. All fluorenylmethyloxycarbonyl (Fmoc)-protected D-amino acids were lyophilized for at least 12 h before use to remove trace solvents.

Materials for peptide synthesis

All reagents were purchased and used as received unless otherwise noted. Fmoc-protected l-amino acids (Fmoc-Ala-OH•H₂O, Fmoc-Cys(Trt)-OH, Fmoc-Asp(OtBu)-OH, Fmoc-Glu(OtBu)-OH, Fmoc-Phe-OH, Fmoc-Gly-OH, Fmoc-Ile-OH, Fmoc-Lys(Boc)-OH, Fmoc-Leu-OH, Fmoc-Met-OH, Fmoc-Asn(Trt)-OH, Fmoc-Pro-OH, Fmoc-Gln(Trt)-OH, Fmoc-Arg(Pbf)-OH, Fmoc-Ser(tBu)-OH, Fmoc-Thr(tBu), Fmoc-Val-OH, Fmoc-Trp(Boc)-OH, and Fmoc-Tyr(OtBu)-OH) were purchased from the Novabiochem line (Millipore-Sigma). l-Fmoc-His(Boc)-OH was purchased from ChemPep. l-Fmoc-Nle-OH was purchased from Chem-Impex. Fmoc-protected d-amino acids (Fmoc-D-Ala-OH•H₂O, Fmoc-d-Cys(Trt)-OH, Fmoc-d-Asp(OtBu)-OH, Fmoc-d-Glu(OtBu)-OH, Fmoc-d-Phe-OH, Fmoc-d-His(Boc)-OH, Fmoc-d-Ile-OH, Fmoc-d-Lys(Boc)-OH, Fmoc-d-Leu-OH, Fmoc-d-Met-OH, Fmoc-d-Asn(Trt)-OH, Fmoc-d-Pro-OH, Fmoc-d-Gln(Trt)-OH, Fmoc-d-Arg(Pbf)-OH, Fmoc-d-Ser(tBu)-OH, Fmoc-d-Thr(tBu), Fmoc-d-Val-OH, Fmoc-d-Trp(Boc)-OH, and Fmoc-d-Tyr(OtBu)-OH) were purchased from both the Novabiochem line from Millipore-Sigma and Chem-Impex. Fmoc-d-Nle-OH was purchased from Chem-Impex. O-(7-azabenzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium hexafluorophosphate (HATU, ≥97.0%) and (7-azabenzotriazol-1-yloxy)trispyrrolidinophosphonium hexafluorophosphate (PyAOP, ≥97.0%) were purchased from P3 Biosystems. Biosynthesis OmniSolv® grade N,N-dimethylformamide (DMF) was purchased from EMD Millipore, and stored over AldraAmine trapping agents (for 1000–4000 mL, Millipore-Sigma catalog number Z511706). Diisopropylethylamine (DIEA; 99.5%, biotech grade, catalog number 387649) was purchased from Sigma-Aldrich and purified by passage through an activated alumina column (Pure Process Technology solvent purification system). Piperidine (ACS reagent, ≥99.0%), trifluoroacetic acid (TFA, HPLC grade, ≥99.0%), triisopropylsilane (TIPS, ≥98.0%), acetonitrile (HPLC grade), formic acid (FA, ≥95.0%), phenol (ACS reagent, ≥99.0%), diethyl ether (Et₂O, ≥99.7%, containing 1 ppm BHT as inhibitor), and 1,2-ethanedithiol (EDT, GC grade, ≥98.0%) were purchased from Sigma-Aldrich. H-Rink Amide ChemMatrix® (0.49 mmol/g and 0.18 mmol/g loading) resin was purchased from PCAS Biomatrix. Water was deionized using a Milli-Q Reference water purification system (Millipore). Nylon 0.22 μm syringe filters were TISCH brand SPEC17984. 5 mL and 10 mL peptide synthesis reaction vessels were purchased from Torviq (catalog numbers SF-0500, and SF-1000 respectively). Biotin-PEG₁₂-COOH (98%) was purchased from BroadPharm. Syringe tip caps were purchased from VWR (catalog number 97001-202).

Materials for protein folding

Guanidine hydrochloride (Gdn HCl, molecular biology grade, ≥99%), 4-(2-hydroxyethyl)1-piperazineethanesulfonic acid (HEPES, ≥99.5%), tris(hydroxymethyl)aminomethane (TRIS, ≥99.9%), tris(hydroxymethyl)aminomethane hydrochloride (TRIS HCl, ≥99.0%), monopotassium phosphate (KPi, ≥99.0%), dipotassium phosphate (≥99.0%), sodium chloride (NaCl, BioXtra, ≥99.5%), glycerol (≥99.5%), sodium hydroxide (≥98%), and hydrochloric acid (36.5–38%) were purchased from Sigma Aldrich. A 0.5 M solution of tris(2-carboxyethyl)phosphine hydrochloride (TCEP, Bond-Breaker™, catalogue number 77720), was purchased from Thermo Fisher Scientific. 1,4-Dithio-DL-threitol (DTT, ≥99%) was purchased from Chem-Impex.

Automated flow peptide synthesis

All peptides were synthesized on one of three automated fast-flow systems built in the Pentelute laboratory. In all cases, synthesis conditions were identical to the optimized protocol as previously described³¹. Every protein target was synthesized once to yield sufficient amounts for analytical and biophysical characterization, with the exception of l-CHIP, l-IRAK2, l-ERG, l-BCL11a, l-Myc, l-Max, l-Max-nb, l-Max-Max-nb, and l-Myc-Max-nb, which were synthesized twice. Conditions are provided in the Supplementary Methods section.

Manual peptide synthesis

Unless otherwise specified, all manual coupling reactions were carried out in 10 mL disposable peptide synthesis vessels from Torviq attached to vacuum manifolds. To the reaction vessel was added an appropriate amount of resin and DMF was added to completely fill the vessel. The mixture was stirred with a glass rod to remove clumps and fully suspend the resin in DMF. The finely divided resin was incubated in DMF for 30 min to fully swell. To a 500 mL volumetric flask (VWR, part number 10124-384) was weighed 72.24 g (0.19 mol) of HATU and DMF was added to prepare a 0.38 M stock solution. The HATU stock solution was transferred to a 500 mL Corning Pyrex glass media bottle (part number 1395-500) and stored for not more than 1 week. To a 20 mL glass scintillation vial was weighed 5 equivalents (relative to resin loading) of Fmoc-protected amino acid. To the glass scintillation vial was added an appropriate volume of HATU stock to produce a 0.4 M solution of Fmoc-protected amino acid. The amino acid was fully solubilized with sonication and stored for not more than 24 h. In cases where the amino acid was not fully solubilized, DMF was added to adjust amino acid concentration to 0.2 M. After the incubation time DMF was drained from the reaction vessel. To the glass scintillation vial with the solubilized amino acid and HATU was added 15 equivalents of DIEA (relative to resin loading), the screw cap affixed, and the solution mixed rapidly. After a 30 s incubation, the solution containing the now activated amino acid was added to the reaction vessel. The resin was stirred with the same glass rod to produce a homogeneous mixture, and the apparatus was allowed to sit for 15 min. The reaction mixture was drained from the reaction vessel, and the resin rinsed with 10 mL DMF three times. After washing the resin, the Fmoc protecting group was removed from the terminal amine. Separately, a stock solution of 20% piperidine (v/v) in DMF was prepared in a 500 mL Corning Pyrex glass media bottle. To the reaction vessel was added enough of the deprotection mixture to fully cover the resin, and let sit for 15 min. The deprotection mixture was drained from the reaction vessel and the resin rinsed with 10 mL DMF three times.

Cleavage protocols

Following synthesis, peptidyl resins were washed with dichloromethane (5 × 5 mL) and dried under an N₂ stream. The peptidyl resin was transferred to a 10 mL reaction vessel and stored at 4 °C until cleavage. Special care was taken with peptidyl resins for sequences containing methionine to avoid prolonged storage. Cleavage was performed by one of two protocols (see Supplementary Methods). After cleavage, the solid pellet was dried gently with N₂ flow until no visible liquid remained. To the Falcon tube containing the dried peptide precipitate was added a volume equal to 3 times that of the precipitate of a 50:50 mixture of water/acetonitrile containing 0.1% TFA. To aid solubilization of the precipitate, the assembly was re-capped and vortexed. In most cases, the precipitate was fully solubilized. The mixture was left to incubate for 30 min at room temperature. The mixture was then flash-frozen with liquid nitrogen and lyophilized for at least 16 h to provide the freeze-dried crude polypeptide as a white solid.

Analytical reverse phase high pressure liquid chromatography

Analysis of synthetic peptides was primarily carried out by evaluation of the chromatogram produced from the absorbance at 214 nm after separation on a reverse phase column. In each case, an appropriate amount of material was injected to produce peaks on the chromatogram that were within the dynamic range of the UV detector, typically between 100 and 2000 mAU (3–5 µg of the major product) for the major peak. One of three methods was used for each synthetic protein chain (see Supplementary Methods and Supplementary Results).

Reverse phase high-performance liquid chromatography–mass spectrometry (LC–MS)

Solutions of the synthetic peptides were evaluated by LC–MS using one of four methods (see Supplementary Methods and Supplementary Results). Data were processed using Agilent MassHunter Workstation Qualitative Analysis Version B.10.00 or Agilent MassHunter BioConfirm B.10.00. Deconvoluted masses of proteins were obtained using a maximum entropy algorithm. Unless otherwise depicted, the following parameters were used for deconvolution: A range 3000 Da less than the starting mass rounded down to the nearest 10,000 Da was used as a lower limit (not lower than 5,000 Da). For the higher limit, a mass to the nearest 10,000 Da rounding up was used. A limited m/z range from 600 to 3000 was used with a baseline subtraction factor of 3 and a mass step of 1. Unless otherwise mentioned, the mass spectrometry data were extracted from the integrated total ion count of the entire major product peak for all targets.

Preparative reverse phase purification

Crude lyophilized peptide powder was weighed and added to a 50 mL Falcon tube. A denaturing buffer consisting of 6 M Gdn HCl, 50 mM TRIS HCl, 50 mM DTT was freshly prepared, and the pH adjusted with 6 M NaOH and 6 M HCl to a value of 7.5 as measured with a pH electrode. An aliquot of the denaturing buffer was added to the lyophilized peptide to prepare a solution at 10 mg of peptide per mL of buffer. A fresh screw-cap was added to the Falcon tube, and the mixture sequentially vortexed and sonicated until no visible precipitate remained. Depending on the solubility of the peptide sequence, additional denaturing buffer was required. The pH was readjusted to 7.5 with 6 M NaOH and 6 M HCl as measured by pH paper, and the mixture was incubated for 30 min at room temperature. After incubation, the Falcon tube screw cap was removed, and the mixture was transferred to a 20 mL syringe affixed with a 0.22 µM nylon syringe filter. The mixture was forced through the syringe filter into a fresh Falcon tube. The Falcon tube originally containing the unfiltered peptide solution was rinsed with an additional 2 mL of the denaturing buffer that was then filtered using the same filtration apparatus into the Falcon tube containing the filtered peptide solution. Any bubbles generated were removed by centrifugation of the falcon tube containing the dissolved peptide at 3220 × g for 5 min. The clarified peptide mixture was drawn-up with a fresh 20 mL syringe and added to the purification column via multiple injections with a manual injection valve affixed with a 10 mL loading loop. Purification was performed in each case by one of three methods (see Supplementary Methods and Supplementary Results).

Phage display

Phage Library Construction and Screening. Stapled phage-displayed peptide libraries were constructed as previously described⁴⁶. Briefly, Peptide Display Cloning System kit from New England Biolabs is used to construct M13KE-containing stapled phage libraries (New England Biolabs, Ipswich, MA). Library oligonucleotides are chemically synthesized using a mix of trimer phosphoramidites (Glen Research, Sterling, VA) without codons encoding cysteine, lysine, proline, and glycine, annealed, extended, and ligated into a digested M13KE vector. All DNA products are purified using Monarch PCR and DNA cleanup kit (New England Biolabs, Ipswich, MA). The resulted library-containing phage vector is transformed into E. coli strain ER2738 (Lucigen, Middleton, WI) by electroporation and amplified by adding the post-rescue electroporated cells to a 500 mL E. coli culture at early-lag phase (OD600 = 0.01). Phage propagation, purification, and stapling are conducted as described⁴⁶. The quality of the final phage libraries is assessed by Sanger and Next Generation Sequencing. To conduct phage library screening, we followed a previously described procedure⁴⁶. Briefly, peptide-displayed phage libraries are incubated with streptavidin magnetic beads for 1 h at room temperature in a blocking buffer containing 5% w/v nonfat milk. For each screening condition, biotinylated protein is captured with streptavidin-coated magnetic beads at room temperature for 15 min, the supernatant is removed using a plate magnet and the beads are resuspended in 50 µL of the blocking buffer. 150 µL of the depleted phage library is added to each well for 200 µL final volume, plates are sealed, and the screening reactions are incubated at room temperature for 45 min, with rotation to maintain beads in solution. Following binding, beads are washed 5 times with ice-cold washing buffer (1X TBS, 1 mM MgCl₂, 1% (w/v) BSA, 0.1% Tween-20, 0.02% (w/v) sodium azide, 2% (w/v) glycerol), beads containing protein-bound phage are collected and directly processed for NGS.

Phage Next Generation Sequencing, Hit ID and Clustering. Next Generation Sequencing was performed according to a previously described procedure⁴⁶. Briefly, phage particles are denatured from magnetic beads at 95 °C for 15 min with an added spike-in sequence (not a library member that is used to enable cross-well normalization of sequence reads), followed by a two-step low-cycled PCR to introduce Illumina adaptors and 10 bp TruSeq DNA UD Indexes (Illumina, San Diego, CA) according to an Illumina’s 16S Metagenomic Sequencing Library Preparation protocol. The NGS library is sequenced by an Illumina NovaSeq platform using a 2 × 150 bp high-output kit (Illumina, San Diego, CA). Hit ID and clustering was performed as previously described⁴⁶. Briefly, NGS reads are trimmed for quality and filtered for sequences that matched the design of the phage library. Counts for each unique sequence are tallied, and then normalized by the counts of the spike-in sequence added to each sample. A metric called Hit Strength is computed for each sequence as the fold change between the normalized counts in the highest target concentration sample and the normalized counts in the blank samples (averaged across experimental replicates). When 0 counts are observed for a sequence in blank samples, a count of 0.5 is used to prevent dividing by zero.

Determination of protein concentration

The concentration of proteins in aqueous solutions was determined using the solution’s absorbance at 280 nm (A₂₈₀), or the Quick Start™ Bradford Protein Assay (see Supplementary Methods).

Protein folding via size exclusion chromatography

Semi-preparative size exclusion chromatography for the folding of synthetic proteins was carried out on an Agilent InfinityLab LC Series 1260 Infinity II Bio-Inert LC System with a Superdex 75 Increase 10/300 GL size exclusion chromatography (SEC) column. In all cases, the flow rate used was 0.8 mL/min. The sample application was accomplished with the use of an Agilent 1200 series manual injection valve fitted with either a 0.5 mL or 0.25 mL loading loop placed before the entrance to the SEC column. An appropriate running buffer was freshly prepared and filtered through a 0.22 µm PES bottle top filter (Corning, catalog number 431118) and the SEC column equilibrated in the running buffer for at least 2 column volumes before sample application. Unless otherwise specified, the running buffer was 50 mM HEPES, 150 mM NaCl, 0.5 mM DTT, 5% glycerol (v/v) adjusted to pH 7.5 with 6 M NaOH or 6 M HCl as required.

An appropriate amount of lyophilized, purified protein (typically 0.1–1.0 mg) was weighed into a low-protein-binding 2 mL Eppendorf tube. An appropriate denaturing buffer was freshly prepared. Unless otherwise specified, the denaturing buffer consisted of 6 M Gdn HCl, 50 mM TRIS HCl, and 50 mM DTT adjusted to pH 7.5 with 6 M NaOH or 6 M HCl as required. A sufficient amount of denaturing buffer was added to the protein to achieve a final concentration of 10 mg/mL. Buffer was added as required to achieve a completely clear solution with no precipitate, but the total volume of the sample was kept below half the volume of the loading loop. The sample was clarified by centrifugation at 21,000 × g for 15 min. The sample was applied to the SEC column and eluted over 2 column volumes with manual fractionation into 2 mL low-protein-binding Eppendorf tubes. Fractions corresponding to the folded protein were collected, and checked by LC-MS. The resulting mixture was separated into 1 nmol aliquots, flash frozen with liquid nitrogen, and stored at −80 °C until further use.

Protein folding via dilution

An appropriate amount of lyophilized, purified protein (typically 0.1–1.0 mg) was weighed into a low-protein-binding 5 mL Eppendorf tube. Appropriate volumes of a denaturing buffer and a dilution buffer were freshly prepared. Unless otherwise specified, the denaturing buffer consisted of 6 M Gdn HCl, 50 mM TRIS HCl, and 50 mM DTT adjusted to pH 7.5 with 6 M NaOH or 6 M HCl as required, and the dilution buffer consisted of 50 mM HEPES, 150 mM NaCl, 0.5 mM DTT, 5% glycerol (v/v) adjusted to pH 7.5 with 6 M NaOH or 6 M HCl as required. Denaturing buffer was added to the protein to achieve a final concentration of 10 mg/mL. Protein concentration was measured and the stock concentration adjusted with additional denaturing buffer to 150 µM. Dilution buffer was added to the protein in denaturing buffer to afford a final protein concentration of 25 µM. The resulting solution was kept at room temperature for 1 h before further use.

Software

Data collection was performed with the assistance of Agilent MassHunter (Version B.06.01), Agilent ChemStation (Version C.01.10[287]), AVIV CD control software, Tecan SparkControl (Version 3.1), PheraStar Software (V 5.70 R4), Biacore S200 Control Software (V 1.1), Biacore 8K Control Software (V 3.0.12.15655), and CLARIOstar Software (V.5.40 R3). Data analysis was performed with the assistance of Agilent MassHunter Workstation Qualitative Analysis (Version B.10.00), Agilent MassHunter BioConfirm Software (Version B.10.00), GraphPad Prism (Version 9.4.0), Agilent ChemStation (Version C.01.10[287]), Adobe Illustrator (Version 27.4.1), Microsoft Excel (2019, Version 1808), ChemDraw (Version 19.1.0.8), PheraStar MARS (V 3.42 R5), BIAevaluation (V 4.1), and CLARIOstar (V.5.40 R3).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data are available in the main text or the supplementary materials (Supplementary Information and Data Source files). Crystallographic datasets have been deposited in the Protein Data Bank (PDB IDs: 8F14, 8F15, 8F16, 8F17, 8F0Z, 8F12, 8F13, 8F10). The raw sequencing data from mirror-image phage display selections against MDM2 and CHIP was deposited with the NCBI Sequence Read Archive (SRA) and can be retrieved using BioProject accession code PRJNA997450 (ID: 997450) and BioSample accession codes SAMN36679786 and SAMN36679787. Source data are provided with this paper.

References

Milton, R. C., Milton, S. C. & Kent, S. B. Total chemical synthesis of a D-enzyme: the enantiomers of HIV-1 protease show reciprocal chiral substrate specificity. Science 256, 1445–1448 (1992).
Article ADS CAS PubMed Google Scholar
Wang, Z., Xu, W., Liu, L. & Zhu, T. F. A synthetic molecular system capable of mirror-image genetic replication and transcription. Nat. Chem. 8, 698–704 (2016).
Article CAS PubMed Google Scholar
Xu, Y. & Zhu, T. F. Mirror-image T7 transcription of chirally inverted ribosomal and functional RNAs. Science 378, 405–412 (2022).
Article ADS CAS PubMed Google Scholar
Fan, C., Deng, Q. & Zhu, T. F. Bioorthogonal information storage in l-DNA with a high-fidelity mirror-image Pfu DNA polymerase. Nat. Biotechnol. 39, 1548–1555 (2021).
Article CAS PubMed Google Scholar
Wang, L. et al. Therapeutic peptides: current applications and future directions. Signal Transduct. Target. Ther. 7, 48 (2022).
Article CAS PubMed PubMed Central Google Scholar
Muttenthaler, M., King, G. F., Adams, D. J. & Alewood, P. F. Trends in peptide drug discovery. Nat. Rev. Drug Discov. 20, 309–325 (2021).
Article CAS PubMed Google Scholar
Feng, Z. & Xu, B. Inspiration from the mirror: D-amino acid containing peptides in biomedical approaches. Biomol. Concepts 7, 179–187 (2016).
Article CAS PubMed PubMed Central Google Scholar
Kremsmayr, T. et al. On the utility of chemical strategies to improve peptide gut stability. J. Med. Chem. 65, 6191–6206 (2022).
Article CAS PubMed PubMed Central Google Scholar
Schumacher, T. N. M. et al. Identification of d-peptide ligands through mirror-image phage display. Science 271, 1854–1857 (1996).
Article ADS CAS PubMed Google Scholar
Liu, M. et al. A left-handed solution to peptide inhibition of the p53–MDM2 interaction. Angew. Chem. Int. Ed. 49, 3649–3652 (2010).
Article ADS CAS Google Scholar
Díaz-Perlas, C. et al. Protein chemical synthesis combined with mirror-image phage display yields d-peptide EGF ligands that block the EGF–EGFR interaction. ChemBioChem 20, 2079–2084 (2019).
Article PubMed Google Scholar
Wiesehan, K. et al. Selection of D-amino-acid peptides that bind to Alzheimer’s disease amyloid peptide Aβ1–42 by mirror image phage display. ChemBioChem 4, 748–753 (2003).
Article CAS PubMed Google Scholar
Eckert, D. M., Malashkevich, V. N., Hong, L. H., Carr, P. A. & Kim, P. S. Inhibiting HIV-1 entry: discovery of D-peptide inhibitors that target the gp41 coiled-coil pocket. Cell 99, 103–115 (1999).
Article CAS PubMed Google Scholar
Marinec, P. S. et al. A non-immunogenic bivalent d-protein potently inhibits retinal vascularization and tumor growth. ACS Chem. Biol. 16, 548–556 (2021).
Article CAS PubMed Google Scholar
Zhou, X. et al. A novel d-peptide identified by mirror-image phage display blocks TIGIT/PVR for cancer immunotherapy. Angew. Chem. Int. Ed. 59, 15114–15118 (2020).
Article CAS Google Scholar
Chang, H.-N. et al. Blocking of the PD-1/PD-L1 interaction by a D-peptide antagonist for cancer immunotherapy. Angew. Chem. Int. Ed. 54, 11760–11764 (2015).
Article ADS CAS Google Scholar
Mandal, K. et al. Chemical synthesis and X-ray structure of a heterochiral {D-protein antagonist plus vascular endothelial growth factor} protein complex by racemic crystallography. Proc. Natl Acad. Sci. USA 109, 14779–14784 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, M. et al. D-peptide inhibitors of the p53–MDM2 interaction for targeted molecular therapy of malignant neoplasms. Proc. Natl Acad. Sci. USA 107, 14321–14326 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Huang, L. et al. Highly selective targeting of hepatic stellate cells for liver fibrosis treatment using a d-enantiomeric peptide ligand of Fn14 identified by mirror-image mRNA display. Mol. Pharmaceutics 14, 1742–1753 (2017).
Article CAS Google Scholar
Li, Z. et al. Novel strategy utilizing extracellular cysteine-rich domain of membrane receptor for constructing d-peptide mediated targeted drug delivery systems: a case study on Fn14. Bioconjugate Chem. 28, 2167–2179 (2017).
Article CAS Google Scholar
Harrison, K., Mackay, A. S., Kambanis, L., Maxwell, J. W. C. & Payne, R. J. Synthesis and applications of mirror-image proteins. Nat. Rev. Chem. 7, 383–404 (2023).
Article CAS PubMed Google Scholar
El-Faham, A. & Albericio, F. Peptide coupling reagents, more than a letter soup. Chem. Rev. 111, 6557–6602 (2011).
Article CAS PubMed Google Scholar
Dawson, P. E. & Kent, S. B. H. Synthesis of native proteins by chemical ligation. Annu. Rev. Biochem. 69, 923–960 (2000).
Article CAS PubMed Google Scholar
Wöhr, T. et al. Pseudo-prolines as a solubilizing, structure-disrupting protection technique in peptide synthesis. J. Am. Chem. Soc. 118, 9218–9227 (1996).
Article Google Scholar
Wu, H., Wei, T., Ngai, W. L., Zhou, H. & Li, X. Ligation embedding aggregation disruptor strategy enables the chemical synthesis of PD-1 immunoglobulin and extracellular domains. J. Am. Chem. Soc. 144, 14748–14757 (2022).
Article CAS PubMed Google Scholar
Manne, S. R., Rustler, K., Bruckdorfer, T., de la Torre, B. G. & Albericio, F. Incorporation of pseudoproline monomer (Fmoc-Thr[ψMe,Mepro]–OH) facilitates efficient solid-phase synthesis of difficult peptides. Tetrahedron Lett. 115, 154301 (2023).
Article CAS Google Scholar
Tan, Z., Shang, S. & Danishefsky, S. J. Rational development of a strategy for modifying the aggregatibility of proteins. Proc. Natl Acad. Sci. Usa. 108, 4297–4302 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Jacobsen, M. T. et al. A helping hand to overcome solubility challenges in chemical protein synthesis. J. Am. Chem. Soc. 138, 11775–11782 (2016).
Article CAS PubMed PubMed Central Google Scholar
Levinson, A. M. et al. Total chemical synthesis and folding of All-L and All-D variants of oncogenic KRas(G12V). J. Am. Chem. Soc. 139, 7632–7639 (2017).
Article CAS PubMed PubMed Central Google Scholar
Mijalis, A. J. et al. A fully automated flow-based approach for accelerated peptide synthesis. Nat. Chem. Biol. 13, 464 (2017).
Article CAS PubMed Google Scholar
Hartrampf, N. et al. Synthesis of proteins by automated flow chemistry. Science 368, 980 (2020).
Article CAS PubMed Google Scholar
Zuo, C. et al. Chemical synthesis and racemic crystallization of rat C5a-desArg. Chin. Chem. Lett. 31, 693–696 (2020).
Article CAS Google Scholar
Ling, J.-J. et al. Mirror-image 5S ribonucleoprotein complexes. Angew. Chem. Int. Ed. 59, 3724–3731 (2020).
Article CAS Google Scholar
Zuo, C. et al. Chimeric protein probes for C5a receptors through fusion of the anaphylatoxin C5a core region with a small-molecule antagonist. Sci. China Chem. 62, 1371–1378 (2019).
Article CAS Google Scholar
Weidmann, J., Schnölzer, M., Dawson, P. E. & Hoheisel, J. D. Copying life: synthesis of an enzymatically active mirror-image DNA-ligase made of D-amino acids. Cell Chem. Biol. 26, 645–651.e3 (2019).
Article CAS PubMed Google Scholar
Shu, K. et al. Development of mirror-image screening systems for XIAP BIR3 domain inhibitors. Bioconjugate Chem. 30, 1395–1404 (2019).
Article ADS CAS Google Scholar
Wang, M. et al. Mirror-image gene transcription and reverse transcription. Chem 5, 848–857 (2019).
Article CAS Google Scholar
Guo, Q.-Y. et al. Channel activity of mirror-image M2 proton channel of influenza A virus is blocked by achiral or chiral inhibitors. Protein Cell 10, 211–216 (2019).
Article CAS PubMed Google Scholar
De Rosa, L., Di Stasi, R. & D’Andrea, L. D. Total chemical synthesis by native chemical ligation of the all-D immunoglobulin-like domain 2 of Axl. Tetrahedron 75, 894–905 (2019).
Article Google Scholar
Chen, C.-C. et al. Racemic X-ray structure of L-type calcium channel antagonist Calciseptine prepared by total chemical synthesis. Sci. China Chem. 61, 702–707 (2018).
Article ADS CAS Google Scholar
Pech, A. et al. A thermostable d-polymerase for mirror-image PCR. Nucleic Acids Res. 45, 3997–4005 (2017).
Article CAS PubMed PubMed Central Google Scholar
Xu, W. et al. Total chemical synthesis of a thermostable enzyme capable of polymerase chain reaction. Cell Discov. 3, 17008 (2017).
Article ADS PubMed PubMed Central Google Scholar
Noguchi, T. et al. Synthesis of Grb2 SH2 domain proteins for mirror-image screening systems. Bioconjugate Chem. 28, 609–619 (2017).
Article CAS Google Scholar
Shu, K. et al. Synthesis of the Src SH2 domain and its application in bioassays for mirror-image screening. RSC Adv. 7, 38725–38732 (2017).
Article ADS CAS Google Scholar
Guo, C., Zhang, L.-Q. & Jiang, W. Biodegrading plastics with a synthetic non-biodegradable enzyme. Chem 9, 363–376 (2023).
Article CAS Google Scholar
Li, K. et al. De novo mapping of α-helix recognition sites on protein surfaces using unbiased libraries. Proc. Natl Acad. Sci. USA 119, e2210435119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Vinogradov, A. A., Evans, E. D. & Pentelute, B. L. Total synthesis and biochemical characterization of mirror image barnase. Chem. Sci. 6, 2997–3002 (2015).
Article CAS PubMed PubMed Central Google Scholar
Jbara, M. et al. Engineering bioactive dimeric transcription factor analogs via palladium rebound reagents. J. Am. Chem. Soc. 143, 11788–11798 (2021).
Article CAS PubMed Google Scholar
Guo, B., Audu, C. O., Cochran, J. C., Mierke, D. F. & Pellegrini, M. Protein engineering of the N-terminus of NEMO: structure stabilization and rescue of IKKβ binding. Biochemistry 53, 6776–6785 (2014).
Article CAS PubMed Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 50, D439–D444 (2022).
Article CAS PubMed Google Scholar
Zhan, C., Varney, K., Yuan, W., Zhao, L. & Lu, W. Interrogation of MDM2 phosphorylation in p53 activation using native chemical ligation: the functional role of Ser17 phosphorylation in MDM2 reexamined. J. Am. Chem. Soc. 134, 6855–6864 (2012).
Article CAS PubMed PubMed Central Google Scholar
Batas, B. & Chaudhuri, J. B. Protein refolding at high concentration using size-exclusion chromatography. Biotechnol. Bioeng. 50, 16–23 (1996).
Article CAS PubMed Google Scholar
Corbett, R. J. T. & Roche, R. S. Use of high-speed size-exclusion chromatography for the study of protein folding and stability. Biochemistry 23, 1888–1894 (1984).
Article CAS PubMed Google Scholar
Rushe, M. et al. Structure of a NEMO/IKK-associating domain reveals architecture of the interaction site. Structure 16, 798–808 (2008).
Article CAS PubMed Google Scholar
Yang, Y. et al. Structural insights into the recognition of γ-globin gene promoter by BCL11A. Cell Res. 29, 960–963 (2019).
Article PubMed PubMed Central Google Scholar
Kelemen, B. R. et al. Hypersensitive substrate for ribonucleases. Nucleic Acids Res. 27, 3696–3701 (1999).
Article CAS PubMed PubMed Central Google Scholar
Mackereth, C. D. et al. Diversity in structure and function of the Ets family PNT domains. J. Mol. Biol. 342, 1249–1264 (2004).
Article CAS PubMed Google Scholar
Lin, S.-C., Lo, Y.-C. & Wu, H. Helical assembly in the MyD88–IRAK4–IRAK2 complex in TLR/IL-1R signalling. Nature 465, 885–890 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Vassilev, L. T. et al. In vivo activation of the p53 pathway by small-molecule antagonists of MDM2. Science 303, 844–848 (2004).
Article ADS CAS PubMed Google Scholar
Chène, P. Inhibiting the p53–MDM2 interaction: an important target for cancer therapy. Nat. Rev. Cancer 3, 102–109 (2003).
Article PubMed Google Scholar
Khoo, K. H., Verma, C. S. & Lane, D. P. Drugging the p53 pathway: understanding the route to clinical efficacy. Nat. Rev. Drug Discov. 13, 217–236 (2014).
Article CAS PubMed Google Scholar
Liu, M. et al. D-peptide inhibitors of the p53-MDM2 interaction for targeted molecular therapy of malignant neoplasms. Proc. Natl Acad. Sci. USA 107, 14321–14326 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Kannan, S. et al. Macrocyclization of an all-d linear α-helical peptide imparts cellular permeability. Chem. Sci. 11, 5577–5591 (2020).
Article CAS PubMed PubMed Central Google Scholar
Chang, Y. S. et al. Stapled α−helical peptide drug development: a potent dual inhibitor of MDM2 and MDMX for p53-dependent cancer therapy. Proc. Natl Acad. Sci. USA 110, E3445–E3454 (2013).
Article CAS PubMed PubMed Central Google Scholar
Qian, S.-B., McDonough, H., Boellmann, F., Cyr, D. M. & Patterson, C. CHIP-mediated stress recovery by sequential ubiquitination of substrates and Hsp70. Nature 440, 551–555 (2006).
Article ADS CAS PubMed PubMed Central Google Scholar
Jiang, J. et al. CHIP Is a U-box-dependent E3 ubiquitin ligase: identification of Hsc70 AS A target for ubiquitylation *. J. Biol. Chem. 276, 42938–42944 (2001).
Article CAS PubMed Google Scholar
Connell, P. et al. The co-chaperone CHIP regulates protein triage decisions mediated by heat-shock proteins. Nat. Cell Biol. 3, 93–96 (2001).
Article CAS PubMed Google Scholar
Ravalin, M. et al. Specificity for latent C termini links the E3 ubiquitin ligase CHIP to caspases. Nat. Chem. Biol. 15, 786–794 (2019).
Article CAS PubMed PubMed Central Google Scholar
Apriamashvili, G. et al. Ubiquitin ligase STUB1 destabilizes IFNγ-receptor complex to suppress tumor IFNγ signaling. Nat. Commun. 13, 1923 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Ng, S. et al. Discovery and structure-based design of macrocyclic peptides targeting STUB1. J. Med. Chem. 65, 9789–9801 (2022).
Article CAS Google Scholar
Henninot, A., Collins, J. C. & Nuss, J. M. The current state of peptide drug discovery: back to the future? J. Med. Chem. 61, 1382–1414 (2018).
Article CAS PubMed Google Scholar
Tucker, T. J. et al. A series of novel, highly potent, and orally bioavailable next-generation tricyclic peptide PCSK9 inhibitors. J. Med. Chem. 64, 16770–16800 (2021).
Article CAS PubMed Google Scholar
Ballantyne Christie, M. et al. Phase 2b randomized trial of the oral PCSK9 inhibitor MK-0616. J. Am. Coll. Cardiol. 81, 1553–1564 (2023).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

Financial support for this work was provided by FOG Pharmaceuticals (to B.L.P.). We thank Mirella Bucci and Jacob Rodriguez for their valuable comments during manuscript preparation. Please refer to Supplementary Information Section 3 for acknowledgments related to X-ray crystallography data collection.

Author information

Yen-Chun Lee
Present address: Department of Chemistry, National Cheng Kung University, No.1, University Road, Tainan City, 701, Taiwan
Jean-Marie Swiecicki
Present address: Relay Therapeutics, Inc., 399 Binney Street, 2nd Floor, Cambridge, MA, 02139, USA
These authors contributed equally: Alex J. Callahan, Satish Gandhesiri.

Authors and Affiliations

Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA, 02139, USA
Alex J. Callahan, Satish Gandhesiri, Rahi M. Reja, Lia Lozano Salazar, Stephanie Hanna, Yen-Chun Lee, Andrei Loas & Bradley L. Pentelute
FOG Pharmaceuticals Inc., 30 Acorn Park Drive, Cambridge, MA, 02140, USA
Tara L. Travaline, Kunhua Li, Olena S. Tokareva, Jean-Marie Swiecicki, Gregory L. Verdine & John H. McGee
Department of Stem Cell and Regenerative Biology, Harvard University, 7 Divinity Avenue, Cambridge, MA, 02138, USA
Gregory L. Verdine
Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA, 02138, USA
Gregory L. Verdine
Department of Molecular and Cellular Biology, Harvard University, 52 Oxford Street, Cambridge, MA, 02138, USA
Gregory L. Verdine
The Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, 500 Main Street, Cambridge, MA, 02142, USA
Bradley L. Pentelute
Center for Environmental Health Sciences, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA, 02139, USA
Bradley L. Pentelute
Broad Institute of MIT and Harvard, 415 Main Street, Cambridge, MA, 02142, USA
Bradley L. Pentelute

Authors

Alex J. Callahan
View author publications
You can also search for this author in PubMed Google Scholar
Satish Gandhesiri
View author publications
You can also search for this author in PubMed Google Scholar
Tara L. Travaline
View author publications
You can also search for this author in PubMed Google Scholar
Rahi M. Reja
View author publications
You can also search for this author in PubMed Google Scholar
Lia Lozano Salazar
View author publications
You can also search for this author in PubMed Google Scholar
Stephanie Hanna
View author publications
You can also search for this author in PubMed Google Scholar
Yen-Chun Lee
View author publications
You can also search for this author in PubMed Google Scholar
Kunhua Li
View author publications
You can also search for this author in PubMed Google Scholar
Olena S. Tokareva
View author publications
You can also search for this author in PubMed Google Scholar
Jean-Marie Swiecicki
View author publications
You can also search for this author in PubMed Google Scholar
Andrei Loas
View author publications
You can also search for this author in PubMed Google Scholar
Gregory L. Verdine
View author publications
You can also search for this author in PubMed Google Scholar
John H. McGee
View author publications
You can also search for this author in PubMed Google Scholar
Bradley L. Pentelute
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

A.J.C., G.L.V., J.H.M., and B.L.P. conceptualized the research. A.J.C., S.G., R.M.R., L.L.S., S.H., and Y.C.L. carried out protein synthesis and A.J.C. carried out protein folding. T.L.T. and A.J.C. carried out protein biochemical validation. O.S.T., K.L., and J.M.S. carried out phage screening and binder validation. K.L. carried out protein crystallization and structure refinement. A.J.C., S.G., R.M.R., and K.L. generated figures. A.J.C., A.L., J.H.M., and B.L.P. wrote the manuscript with input from all authors.

Corresponding authors

Correspondence to John H. McGee or Bradley L. Pentelute.

Ethics declarations

Competing interests

The authors declare the following competing interest(s): B.L.P. is a co-founder and/or member of the scientific advisory board of several companies focusing on the development of protein and peptide therapeutics. T.L.T., K.L., O.S.T., G.L.V., and J.H.M. are currently employed by FOG Pharmaceuticals, and G.L.V. serves on the board of directors of FOG Pharmaceuticals. FOG Pharmaceuticals has filed provisional patent applications related to the compounds described in this work. J.M.S. was an employee of FOG Pharmaceuticals and is currently employed by Relay Therapeutics, Inc.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Peer Review File

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Callahan, A.J., Gandhesiri, S., Travaline, T.L. et al. Mirror-image ligand discovery enabled by single-shot fast-flow synthesis of D-proteins. Nat Commun 15, 1813 (2024). https://doi.org/10.1038/s41467-024-45634-z

Download citation

Received: 29 June 2023
Accepted: 26 January 2024
Published: 28 February 2024
DOI: https://doi.org/10.1038/s41467-024-45634-z

This article is cited by

Recent advances in chemical protein synthesis: method developments and biological applications
- Suwei Dong
- Ji-Shen Zheng
- Lei Liu
Science China Chemistry (2024)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.