Directed Evolution of PD-L1-Targeted Affibodies by mRNA Display.

Therapeutic monoclonal antibodies directed against PD-L1 (e.g., atezolizumab) disrupt PD-L1:PD-1 signaling and reactivate exhausted cytotoxic T-cells in the tumor compartment. Although anti-PD-L1 antibodies are successful as immune checkpoint inhibitor (ICI) therapeutics, there is still a pressing need to develop high-affinity, low-molecular-weight ligands for molecular imaging and diagnostic applications. Affibodies are small polypeptides (∼60 amino acids) that provide a stable molecular scaffold from which to evolve high-affinity ligands. Despite its proven utility in the development of imaging probes, this scaffold has never been optimized for use in mRNA display, a powerful in vitro selection platform incorporating high library diversity, unnatural amino acids, and chemical modification. In this manuscript, we describe the selection of a PD-L1-binding affibody by mRNA display. Following randomization of the 13 amino acids that define the binding interface of the well-described Her2 affibody, the resulting library was selected against recombinant human PD-L1 (hPD-L1). After four rounds, the enriched library was split and selected against either hPD-L1 or the mouse ortholog (mPD-L1). The dual target selection resulted in the identification of a human/mouse cross-reactive PD-L1 affibody (M1) with low nanomolar affinity for both targets. The M1 affibody bound with similar affinity to mPD-L1 and hPD-L1 expressed on the cell surface and inhibited signaling through the PD-L1:PD-1 axis at low micromolar concentrations in a cell-based functional assay. In vivo optical imaging with M1-Cy5 in an immune-competent mouse model of lymphoma revealed significant tumor uptake relative to a Cy5-conjugated Her2 affibody.

Directing evolution of novel ligands by mRNA display.

mRNA display is a powerful biological display platform for the directed evolution of proteins and peptides. mRNA display libraries covalently link the displayed peptide or protein (phenotype) with the encoding genetic information (genotype) through the biochemical activity of the small molecule puromycin. Selection for peptide/protein function is followed by amplification of the linked genetic material and generation of a library enriched in functional sequences. Iterative selection cycles are then performed until the desired level of function is achieved, at which time the identity of candidate peptides can be obtained by sequencing the genetic material. The purpose of this review is to discuss the development of mRNA display technology since its inception in 1997 and to comprehensively review its use in the selection of novel peptides and proteins. We begin with an overview of the biochemical mechanism of mRNA display and its variants with a particular focus on its advantages and disadvantages relative to other biological display technologies. We then discuss the importance of scaffold choice in mRNA display selections and review the results of selection experiments with biological (e.g., fibronectin) and linear peptide library architectures. We then explore recent progress in the development of "drug-like" peptides by mRNA display through the post-translational covalent macrocyclization and incorporation of non-proteogenic functionalities. We conclude with an examination of enabling technologies that increase the speed of selection experiments, enhance the information obtained in post-selection sequence analysis, and facilitate high-throughput characterization of lead compounds. We hope to provide the reader with a comprehensive view of current state and future trajectory of mRNA display and its broad utility as a peptide and protein design tool.

O-GlcNAc modification of small heat shock proteins enhances their anti-amyloid chaperone activity.

A major role for the intracellular post-translational modification O-GlcNAc appears to be the inhibition of protein aggregation. Most of the previous studies in this area focused on O-GlcNAc modification of the amyloid-forming proteins themselves. Here we used synthetic protein chemistry to discover that O-GlcNAc also activates the anti-amyloid activity of certain small heat shock proteins (sHSPs), a potentially more important modification event that can act broadly and substoichiometrically. More specifically, we found that O-GlcNAc increases the ability of sHSPs to block the amyloid formation of both α-synuclein and Aß(1-42). Mechanistically, we show that O-GlcNAc near the sHSP IXI-domain prevents its ability to intramolecularly compete with substrate binding. Finally, we found that, although O-GlcNAc levels are globally reduced in Alzheimer's disease brains, the modification of relevant sHSPs is either maintained or increased, which suggests a mechanism to maintain these potentially protective O-GlcNAc modifications. Our results have important implications for neurodegenerative diseases associated with amyloid formation and potentially other areas of sHSP biology.

Mammalian Expression and In Situ Biotinylation of Extracellular Protein Targets for Directed Evolution.

Directed evolution is a powerful tool for the selection of functional ligands from molecular libraries. Extracellular domains (ECDs) of cell surface receptors are common selection targets for therapeutic and imaging agent development. Unfortunately, these proteins are often post-translationally modified and are therefore unsuitable for expression in bacterial systems. Directional immobilization of these targets is further hampered by the absence of biorthogonal groups for site-specific chemical conjugation. We have developed a nonadherent mammalian expression system for rapid, high-yield expression of biotinylated ECDs. ECDs from EGFR, HER2, and HER3 were site-specifically biotinylated in situ and recovered from the cell culture supernatant with yields of up to 10 mg/L at >90% purity. Biotinylated ECDs also contained a protease cleavage site for rapid and selective release of the ECD after immobilization on avidin/streptavidin resins and library binding. A model mRNA display selection round was carried out against the HER2 ECD with the HER2 affibody expressed as an mRNA-protein fusion. HER2 affibody-mRNA fusions were selectively released by thrombin and quantitative PCR revealed substantial improvements in the enrichment of functional affibody-mRNA fusions relative to direct PCR amplification of the resin-bound target. This methodology allows rapid purification of high-quality targets for directed evolution and selective elution of functional sequences at the conclusion of each selection round.

Enabling Flow-Based Kinetic Off-Rate Selections Using a Microfluidic Enrichment Device.

Modern genomic sequencing efforts are identifying potential diagnostic and therapeutic targets more rapidly than existing methods can generate the peptide- and protein-based ligands required to study them. To address this problem, we have developed a microfluidic enrichment device (MFED) enabling kinetic off-rate selection without the use of exogenous competitor. We tuned the conditions of the device (bed volume, flow rate, immobilized target) such that modest, readily achievable changes in flow rates favor formation or dissociation of target-ligand complexes based on affinity. Simple kinetic equations can be used to describe the behavior of ligand binding in the MFED and the kinetic rate constants observed agree with independent measurements. We demonstrate the utility of the MFED by showing a 4-fold improvement in enrichment compared to standard selection. The MFED described here provides a route to simultaneously bias pools toward high-affinity ligands while reducing the demand for target-protein to less than a nanomole per selection.

mRNA Display Discovery of a Novel Programmed Death Ligand 1 (PD-L1) Binding Peptide (a Peptide Ligand for PD-L1).

Programmed death ligand 1 (PD-L1) is a critical immune checkpoint ligand whose overexpression on tumor cells provides a mechanism of escape from immune surveillance. The interaction between PD-L1 and PD-1 on T cell lymphocytes suppresses both T cell activation and effector function and is engaged by cancers to dampen antitumor immunity. Here, we used mRNA display to engineer an 18-residue linear peptide that binds to human PD-L1. This peptide, which we term SPAM (signal peptide-based affinity maturated ligand), is nonhomologous to known PD-L1 binding peptides and mAbs, with dissociation constants (KD) of 119 and 67 nM for unglycosylated and glycosylated human PD-L1, respectively. The SPAM peptide is highly selective for human PD-L1 and shows no significant binding to either mouse PD-L1 or human PD-L2. Competition binding assays indicate that the SPAM peptide binding site overlaps with the binding site of PD-1 as well as therapeutic anti-PD-L1 antibodies. Taken together, these results suggest that the SPAM peptide specifically binds to human PD-L1 and could potentially serve as a PD-L1 affinity agent and PD-L1/PD-1 pathway modulator.

Human Auditory Detection and Discrimination Measured with the Pupil Dilation Response.

In the standard Hughson-Westlake hearing tests (Carhart and Jerger 1959), patient responses like a button press, raised hand, or verbal response are used to assess detection of brief test signals such as tones of varying pitch and level. Because of its reliance on voluntary responses, Hughson-Westlake audiometry is not suitable for patients who cannot follow instructions reliably, such as pre-lingual infants (Northern and Downs 2002). As an alternative approach, we explored the use of the pupillary dilation response (PDR), a short-latency component of the orienting response evoked by novel stimuli, as an indicator of sound detection. The pupils of 31 adult participants (median age 24 years) were monitored with an infrared video camera during a standard hearing test in which they indicated by button press whether or not they heard narrowband noises centered at 1, 2, 4, and 8 kHz. Tests were conducted in a quiet, carpeted office. Pupil size was summed over the first 1750 ms after stimulus delivery, excluding later dilations linked to expenditure of cognitive effort (Kahneman and Beatty 1966; Kahneman et al. 1969). The PDR yielded thresholds comparable to the standard test at all center frequencies tested, suggesting that the PDR is as sensitive as traditional methods of assessing detection. We also tested the effects of repeating a stimulus on the habituation of the PDR. Results showed that habituation can be minimized by operating at near-threshold stimulus levels. At sound levels well above threshold, the PDR habituated but could be recovered by changing the frequency or sound level, suggesting that the PDR can also be used to test stimulus discrimination. Given these features, the PDR may be useful as an audiometric tool or as a means of assessing auditory discrimination in those who cannot produce a reliable voluntary response.

Identification, characterization and application of a new peptide against anterior gradient homolog 2 (AGR2).

The cancer-associated protein Anterior Gradient 2 (AGR2) has been described, predominantly in adenocarcinomas. Increased levels of extracellular AGR2 (eAGR2) have been correlated with poor prognosis in cancer patients, making it a potential biomarker. Additionally, neutralizing AGR2 antibodies showed preclinical effectiveness in murine cancer models suggesting eAGR2 may be a therapeutic target. We set out to identify a peptide by mRNA display that would serve as a theranostic tool targeting AGR2. This method enables the selection of peptides from a complex (>1011) library and incorporates a protease incubation step that filters the selection for serum stable peptides. We performed six successive rounds of enrichment using a 10-amino acid mRNA display library and identified several AGR2 binding peptides. One of these peptides (H10), demonstrated high affinity binding to AGR2 with a binding constant (KD) of 6.4 nM. We developed an AGR2 ELISA with the H10 peptide as the capture reagent. Our H10-based ELISA detected eAGR2 from cancer cell spent media with a detection limit of (20-50 ng/ml). Furthermore, we investigated the therapeutic utility of H10 and discovered that it inhibited cell viability at IC50 (9-12 µmoles/L) in cancer cell lines. We also determined that 10 µg/ml of H10 was sufficient to inhibit cancer cell migration in breast and prostate cancer cell lines. A control peptide did not show any appreciable activity in these cells. The H10 peptide showed promise as both a novel diagnostic and a potential therapeutic peptide.

Automated, Resin-Based Method to Enhance the Specific Activity of Fluorine-18 Clicked PET Radiotracers.

Radiolabeling of substrates with 2-[18F]fluoroethylazide exploits the rapid kinetics, chemical selectivity, and mild conditions of the copper-catalyzed azide-alkyne cycloaddition reaction. While this methodology has proven to result in near-quantitative labeling of alkyne-tagged precursors, the relatively small size of the fluoroethylazide group makes separation of the 18F-labeled radiotracer and the unreacted precursor challenging, particularly with precursors >500 Da (e.g., peptides). We have developed an inexpensive azide-functionalized resin to rapidly remove unreacted alkyne precursor following the fluoroethylazide labeling reaction and integrated it into a fully automated radiosynthesis platform. We have carried out 2-[18F]fluoroethylazide labeling of four different alkynes ranging from <300 Da to >1700 Da and found that >98% of the unreacted alkyne was removed in less than 20 min at room temperature to afford the final radiotracers at >99% radiochemical purity with specific activities up to >200 GBq/µmol. We have applied this technique to label a novel cyclic peptide previously evolved to bind the Her2 receptor with high affinity, and demonstrated tumor-specific uptake and low nonspecific background by PET/CT. This resin-based methodology is automated, rapid, mild, and general allowing peptide-based fluorine-18 radiotracers to be obtained with clinically relevant specific activities without chromatographic separation and with only a minimal increase in total synthesis time.

RasIns: Genetically Encoded Intrabodies of Activated Ras Proteins.

K- and H-Ras are the most commonly mutated genes in human tumors and are critical for conferring and maintaining the oncogenic phenotype in tumors with poor prognoses. Here, we design genetically encoded antibody-like ligands (intrabodies) that recognize active, GTP-bound K- and H-Ras. These ligands, which use the 10th domain of human fibronectin as their scaffold, are stable inside the cells and when fused with a fluorescent protein label, the constitutively active G12V mutant H-Ras. Primary selection of ligands against Ras with mRNA display resulted in an intrabody (termed RasIn1) that binds with a KD of 2.1µM to H-Ras(G12V) (GTP), excellent state selectivity, and remarkable specificity for K- and H-Ras. RasIn1 recognizes residues in the Switch I region of Ras, similar to Raf-RBD, and competes with Raf-RBD for binding. An affinity maturation selection based on RasIn1 resulted in RasIn2, which binds with a KD of 120nM and also retains excellent state selectivity. Both of these intrabodies colocalize with H-Ras, K-Ras, and G12V mutants inside the cells, providing new potential tools to monitor and modulate Ras-mediated signaling. Finally, RasIn1 and Rasin2 both display selectivity for the G12V mutants as compared with wild-type Ras providing a potential route for mutant selective recognition of Ras.

Directed Evolution of Scanning Unnatural-Protease-Resistant (SUPR) Peptides for in Vivo Applications.

Peptides typically have poor biostabilities, and natural sequences cannot easily be converted into drug-like molecules without extensive medicinal chemistry. We have adapted mRNA display to drive the evolution of highly stable cyclic peptides while preserving target affinity. To do this, we incorporated an unnatural amino acid in an mRNA display library that was subjected to proteolysis prior to selection for function. The resulting "SUPR (scanning unnatural protease resistant) peptide" showed ≈500-fold improvement in serum stability (t1/2 =160 h) and up to 3700-fold improvement in protease resistance versus the parent sequence. We extended this approach by carrying out SUPR peptide selections against Her2-positive cells in culture. The resulting SUPR4 peptide showed low-nanomolar affinity toward Her2, excellent specificity, and selective tumor uptake in vivo. These results argue that this is a general method to design potent and stable peptides for in vivo imaging and therapy.

High-Throughput Measurement of Binding Kinetics by mRNA Display and Next-Generation Sequencing.

There is great demand for high-throughput methods to characterize ligand affinity. By combining mRNA display with next-generation sequencing, we determined the kinetic on- and off-rates for over twenty thousand ligands without the need for synthesis or purification of individual members. Our results are reproducible and as accurate as those obtained with other methods of affinity measurement.

General, Label-Free Method for Determining K(d) and Ligand Concentration Simultaneously.

Some of the most commonly used affinity reagents (e.g., antibodies) are often developed and used in conditions where their input concentrations ([L]0) and affinities (K(d)) are not known. Here, we have developed a general approach to determine both [L]0 and K(d) values simultaneously for affinity reagents (small molecules, proteins, and antibodies). To do this, we perform quantitative equilibrium exclusion immunoassays with two different concentrations of target and fit the data simultaneously to determine K(d) and [L]0. The results give accurate and reproducible measures of both values compared to established methods. By performing detailed error analysis, we demonstrate that our fitting gives unique solutions and indicates where K(d) and [L]0 measures are reliable. Furthermore, we found that a divalent model of antibody binding gives accurate K(d) and [L]0 values in both the forward (antibody immobilized) and the reverse (target immobilized) assays-addressing the long-term problem of obtaining quantitative data from reverse assays.

A Neural Model of Auditory Space Compatible with Human Perception under Simulated Echoic Conditions.

In a typical auditory scene, sounds from different sources and reflective surfaces summate in the ears, causing spatial cues to fluctuate. Prevailing hypotheses of how spatial locations may be encoded and represented across auditory neurons generally disregard these fluctuations and must therefore invoke additional mechanisms for detecting and representing them. Here, we consider a different hypothesis in which spatial perception corresponds to an intermediate or sub-maximal firing probability across spatially selective neurons within each hemisphere. The precedence or Haas effect presents an ideal opportunity for examining this hypothesis, since the temporal superposition of an acoustical reflection with sounds arriving directly from a source can cause otherwise stable cues to fluctuate. Our findings suggest that subjects' experiences may simply reflect the spatial cues that momentarily arise under various acoustical conditions and how these cues are represented. We further suggest that auditory objects may acquire "edges" under conditions when interaural time differences are broadly distributed.

Spike timing precision changes with spike rate adaptation in the owl's auditory space map.

Spike rate adaptation (SRA) is a continuing change of responsiveness to ongoing stimuli, which is ubiquitous across species and levels of sensory systems. Under SRA, auditory responses to constant stimuli change over time, relaxing toward a long-term rate often over multiple timescales. With more variable stimuli, SRA causes the dependence of spike rate on sound pressure level to shift toward the mean level of recent stimulus history. A model based on subtractive adaptation (Benda J, Hennig RM. J Comput Neurosci 24: 113-136, 2008) shows that changes in spike rate and level dependence are mechanistically linked. Space-specific neurons in the barn owl's midbrain, when recorded under ketamine-diazepam anesthesia, showed these classical characteristics of SRA, while at the same time exhibiting changes in spike timing precision. Abrupt level increases of sinusoidally amplitude-modulated (SAM) noise initially led to spiking at higher rates with lower temporal precision. Spike rate and precision relaxed toward their long-term values with a time course similar to SRA, results that were also replicated by the subtractive model. Stimuli whose amplitude modulations (AMs) were not synchronous across carrier frequency evoked spikes in response to stimulus envelopes of a particular shape, characterized by the spectrotemporal receptive field (STRF). Again, abrupt stimulus level changes initially disrupted the temporal precision of spiking, which then relaxed along with SRA. We suggest that shifts in latency associated with stimulus level changes may differ between carrier frequency bands and underlie decreased spike precision. Thus SRA is manifest not simply as a change in spike rate but also as a change in the temporal precision of spiking.

Serum stable natural peptides designed by mRNA display.

Peptides constructed with the 20 natural amino acids are generally considered to have little therapeutic potential because they are unstable in the presence of proteases and peptidases. However, proteolysis cleavage can be idiosyncratic, and it is possible that natural analogues of functional sequences exist that are highly resistant to cleavage. Here, we explored this idea in the context of peptides that bind to the signaling protein Gαi1. To do this, we used a two-step in vitro selection process to simultaneously select for protease resistance while retaining function-first by degrading the starting library with protease (chymotrypsin), followed by positive selection for binding via mRNA display. Starting from a pool of functional sequences, these experiments revealed peptides with 100-400 fold increases in protease resistance compared to the parental library. Surprisingly, selection for chymotrypsin resistance also resulted in similarly improved stability in human serum (~100 fold). Mechanistically, the decreases in cleavage results from both a lower rate of cleavage (kcat) and a weaker interaction with the protease (Km). Overall, our results demonstrate that the hydrolytic stability of functional, natural peptide sequences can be improved by two orders of magnitude simply by optimizing the primary sequence.

Robust, quantitative analysis of proteins using peptide immunoreagents, in vitro translation, and an ultrasensitive acoustic resonant sensor.

A major benefit of proteomic and genomic data is the potential for developing thousands of novel diagnostic and analytical tests of cells, tissues, and clinical samples. Monoclonal antibody technologies, phage display and mRNA display, are methods that could be used to generate affinity ligands against each member of the proteome. Increasingly, the challenge is not ligand generation, rather the analysis and affinity rank-ordering of the many ligands generated by these methods. Here, we developed a quantitative method to analyze protein interactions using in vitro translated ligands. In this assay, in vitro translated ligands generate a signal by simultaneously binding to a target immobilized on a magnetic bead and to a sensor surface in a commercial acoustic sensing device. We then normalize the binding of each ligand with its relative translation efficiency in order to rank-order the different ligands. We demonstrate the method with peptides directed against the cancer marker Bcl-xL. Our method has 4- to 10-fold higher sensitivity, using 100-fold less protein and 5-fold less antibody per sample, as compared directly with ELISA. Additionally, all analysis can be conducted in complex mixtures at physiological ionic strength. Lastly, we demonstrate the ability to use peptides as ultrahigh affinity reagents that function in complex matrices, as would be needed in diagnostic applications.

Recombinant probes for visualizing endogenous synaptic proteins in living neurons.

The ability to visualize endogenous proteins in living neurons provides a powerful means to interrogate neuronal structure and function. Here we generate recombinant antibody-like proteins, termed Fibronectin intrabodies generated with mRNA display (FingRs), that bind endogenous neuronal proteins PSD-95 and Gephyrin with high affinity and that, when fused to GFP, allow excitatory and inhibitory synapses to be visualized in living neurons. Design of the FingR incorporates a transcriptional regulation system that ties FingR expression to the level of the target and reduces background fluorescence. In dissociated neurons and brain slices, FingRs generated against PSD-95 and Gephyrin did not affect the expression patterns of their endogenous target proteins or the number or strength of synapses. Together, our data indicate that PSD-95 and Gephyrin FingRs can report the localization and amount of endogenous synaptic proteins in living neurons and thus may be used to study changes in synaptic strength in vivo.

Antibody-mimetic ligand selected by mRNA display targets DC-SIGN for dendritic cell-directed antigen delivery.

Dendritic cell (DC)-based vaccines have shown promise as an immunotherapeutic modality for cancer and infectious diseases in many preclinical studies and clinical trials. Provenge (sipuleucel-T), a DC-based vaccine based on ex vivo-generated autologous DCs loaded with antigens, has recently received FDA approval for prostate cancer treatment, further validating the potential of DC-based vaccine modalities. However, direct antigen delivery to DCs in vivo via DC-specific surface receptors would enable a more direct and less laborious approach to immunization. In this study, the recombinant extracellular domains (ECD) of human and mouse DC-SIGN (hDC-SIGN and mDC-SIGN) were generated as DC-specific targets for mRNA display. Accordingly, an antibody-mimetic library was constructed by randomizing two exposed binding loops of an expression-enhanced 10th human fibronectin type III domain (e10Fn3). After three rounds of selection against mDC-SIGN, followed by four rounds of selection against hDC-SIGN, we were able to evolve several dual-specific ligands, which could bind to both soluble ECD of human and mouse DC-SIGNs. Using a cell-binding assay, one ligand, eFn-DC6, was found to have high affinity to hDC-SIGN and moderate affinity to mDC-SIGN. When fused with an antigenic peptide, eFn-DC6 could direct the antigen delivery and presentation by human peripheral blood mononuclear cell (PBMC)-derived DCs and stimulate antigen-specific CD8(+) T cells to secrete inflammatory cytokines. Taken together, these results demonstrate the utility of mRNA display to select protein carriers for DC-based vaccination and offer in vitro evidence that the antibody-mimetic ligand eFn-DC6 represents a promising candidate for the development of an in vivo DC-based vaccine in humans.

The role of envelope shape in the localization of multiple sound sources and echoes in the barn owl.

Echoes and sounds of independent origin often obscure sounds of interest, but echoes can go undetected under natural listening conditions, a perception called the precedence effect. How does the auditory system distinguish between echoes and independent sources? To investigate, we presented two broadband noises to barn owls (Tyto alba) while varying the similarity of the sounds' envelopes. The carriers of the noises were identical except for a 2- or 3-ms delay. Their onsets and offsets were also synchronized. In owls, sound localization is guided by neural activity on a topographic map of auditory space. When there are two sources concomitantly emitting sounds with overlapping amplitude spectra, space map neurons discharge when the stimulus in their receptive field is louder than the one outside it and when the averaged amplitudes of both sounds are rising. A model incorporating these features calculated the strengths of the two sources' representations on the map (B. S. Nelson and T. T. Takahashi; Neuron 67: 643-655, 2010). The target localized by the owls could be predicted from the model's output. The model also explained why the echo is not localized at short delays: when envelopes are similar, peaks in the leading sound mask corresponding peaks in the echo, weakening the echo's space map representation. When the envelopes are dissimilar, there are few or no corresponding peaks, and the owl localizes whichever source is predicted by the model to be less masked. Thus the precedence effect in the owl is a by-product of a mechanism for representing multiple sound sources on its map.