Generating thermal stable variants of protein domains through phage display

Novel Approach for Obtaining Variable Domain of New Antigen Receptor with Different Physicochemical Properties from Japanese Topeshark (Hemitriakis japanica)

Diverse candidate antibodies are needed to successfully identify therapeutic and diagnostic applications. The variable domain of IgNAR (VNAR), a shark single-domain antibody, has attracted attention owing to its favorable physicochemical properties. The phage display method used to screen for optimal VNARs loses sequence diversity because of the bias caused by the differential ease of protein expression in Escherichia coli. Here, we investigated a VNAR selection method that combined panning with various selection pressures and next-generation sequencing (NGS) analyses to obtain additional candidates. Drawing inspiration from the physiological conditions of sharks and the physicochemical properties of VNARs, we examined the effects of NaCl and urea concentrations, low temperature, and preheating at the binding step of panning. VNAR phage libraries generated from Japanese topeshark (Hemitriakis japanica) were enriched under these conditions. We then performed NGS analysis and attempted to select clones that were specifically enriched under each panning condition. The identified VNARs exhibited higher reactivity than those obtained by panning without selection pressure. Additionally, they possess physicochemical properties that reflect their respective selection pressures. These results can greatly enhance our understanding of VNAR properties and offer guidance for the screening of high-quality VNAR clones that are present at low frequencies.

Target-template relationships in protein structure prediction and their effect on the accuracy of thermostability calculations

Improving protein thermostability has been a labor- and time-consuming process in industrial applications of protein engineering. Advances in computational approaches have facilitated the development of more efficient strategies to allow the prioritization of stabilizing mutants. Among these is FEP+, a free energy perturbation implementation that uses a thoroughly tested physics-based method to achieve unparalleled accuracy in predicting changes in protein thermostability. To gauge the applicability of FEP+ to situations where crystal structures are unavailable, here we have applied the FEP+ approach to homology models of 12 different proteins covering 316 mutations. By comparing predictions obtained with homology models to those obtained using crystal structures, we have identified that local rather than global sequence conservation between target and template sequence is a determining factor in the accuracy of predictions. By excluding mutation sites with low local sequence identity (<40%) to a template structure, we have obtained predictions with comparable performance to crystal structures (R2 of 0.67 and 0.63 and an RMSE of 1.20 and 1.16 kcal/mol for crystal structure and homology model predictions, respectively) for identifying stabilizing mutations when incorporating residue scanning into a cascade screening strategy. Additionally, we identify and discuss inherent limitations in sequence alignments and homology modeling protocols that translate into the poor FEP+ performance of a few select examples. Overall, our retrospective study provides detailed guidelines for the application of the FEP+ approach using homology models for protein thermostability predictions, which will greatly extend this approach to studies that were previously limited by structure availability.

Deep Learning in Therapeutic Antibody Development

Deep learning applied to antibody development is in its adolescence. Low data volumes and biological platform differences make it challenging to develop supervised models that can predict antibody behavior in actual commercial development steps. But successes in modeling general protein behaviors and early antibody models give indications of what is possible for antibodies in general, particularly since antibodies share a common fold. Meanwhile, new methods of data collection and the development of unsupervised and self-supervised deep learning methods like generative models and masked language models give the promise of rich and deep data sets and deep learning architectures for better supervised model development. Together, these move the industry toward improved developability , lower costs, and broader access of biotherapeutics .

Antibody design using LSTM based deep generative model from phage display library for affinity maturation

Molecular evolution is an important step in the development of therapeutic antibodies. However, the current method of affinity maturation is overly costly and labor-intensive because of the repetitive mutation experiments needed to adequately explore sequence space. Here, we employed a long short term memory network (LSTM)-a widely used deep generative model-based sequence generation and prioritization procedure to efficiently discover antibody sequences with higher affinity. We applied our method to the affinity maturation of antibodies against kynurenine, which is a metabolite related to the niacin synthesis pathway. Kynurenine binding sequences were enriched through phage display panning using a kynurenine-binding oriented human synthetic Fab library. We defined binding antibodies using a sequence repertoire from the NGS data to train the LSTM model. We confirmed that likelihood of generated sequences from a trained LSTM correlated well with binding affinity. The affinity of generated sequences are over 1800-fold higher than that of the parental clone. Moreover, compared to frequency based screening using the same dataset, our machine learning approach generated sequences with greater affinity.

Isolation of highly selective IgNAR variable single-domains against a human therapeutic Fc scaffold and their application as tailor-made bioprocessing reagents

The adaptive immune system of cartilaginous fish (Elasmobranchii), comprising of classical hetero-tetrameric antibodies, is enhanced through the presence of a naturally occurring homodimeric antibody-like immunoglobulin-the new antigen receptor (IgNAR). The binding site of the IgNAR variable single-domain (VNAR) offers advantages of reduced size (<1/10th of classical immunoglobulin) and extended binding topographies, making it an ideal candidate for accessing cryptic epitopes otherwise intractable to conventional antibodies. These attributes, coupled with high physicochemical stability and amenability to phage display, facilitate the selection of VNAR binders to challenging targets. Here, we explored the unique attributes of these single domains for potential application as bioprocessing reagents in the development of the SEED-Fc platform, designed to generate therapeutic bispecific antibodies. A panel of unique VNARs specific to the SEED homodimeric (monospecific) 'by-products' were isolated from a shark semi-synthetic VNAR library via phage display. The lead VNAR candidate exhibited low nanomolar affinity and superior selectivity to SEED homodimer, with functionality being retained upon exposure to extreme physicochemical conditions that mimic their applicability as purification agents. Ultimately, this work exemplifies the robustness of the semi-synthetic VNAR platform, the predisposition of the VNAR paratope to recognise novel epitopes and the potential for routine generation of tailor-made VNAR-based bioprocessing reagents.

Selection and screening strategies in directed evolution to improve protein stability

Protein stability is not only fundamental for experimental, industrial, and therapeutic applications, but is also the baseline for evolving novel protein functions. For decades, stability engineering armed with directed evolution has continued its rapid development and inevitably poses challenges. Generally, in directed evolution, establishing a reliable link between a genotype and any interpretable phenotype is more challenging than diversifying genetic libraries. Consequently, we set forth in a small picture to emphasize the screening or selection techniques in protein stability-directed evolution to secure the link. For a more systematic review, two main branches of these techniques, namely cellular or cell-free display and stability biosensors, are expounded with informative examples.

Arming Filamentous Bacteriophage, a Nature-Made Nanoparticle, for New Vaccine and Immunotherapeutic Strategies

The pharmaceutical use of bacteriophages as safe and inexpensive therapeutic tools is collecting renewed interest. The use of lytic phages to fight antibiotic-resistant bacterial strains is pursued in academic and industrial projects and is the object of several clinical trials. On the other hand, filamentous bacteriophages used for the phage display technology can also have diagnostic and therapeutic applications. Filamentous bacteriophages are nature-made nanoparticles useful for their size, the capability to enter blood vessels, and the capacity of high-density antigen expression. In the last decades, our laboratory focused its efforts in the study of antigen delivery strategies based on the filamentous bacteriophage 'fd', able to trigger all arms of the immune response, with particular emphasis on the ability of the MHC class I restricted antigenic determinants displayed on phages to induce strong and protective cytotoxic responses. We showed that fd bacteriophages, engineered to target mouse dendritic cells (DCs), activate innate and adaptive responses without the need of exogenous adjuvants, and more recently, we described the display of immunologically active lipids. In this review, we will provide an overview of the reported applications of the bacteriophage carriers and describe the advantages of exploiting this technology for delivery strategies.

Phage Display in the Quest for New Selective Recognition Elements for Biosensors

Phages are bacterial viruses that have gained a significant role in biotechnology owing to their widely studied biology and many advantageous characteristics. Perhaps the best-known application of phages is phage display that refers to the expression of foreign peptides or proteins outside the phage virion as a fusion with one of the phage coat proteins. In 2018, one half of the Nobel prize in chemistry was awarded jointly to George P. Smith and Sir Gregory P. Winter "for the phage display of peptides and antibodies." The outstanding technology has evolved and developed considerably since its first description in 1985, and today phage display is commonly used in a wide variety of disciplines, including drug discovery, enzyme optimization, biomolecular interaction studies, as well as biosensor development. A cornerstone of all biosensors, regardless of the sensor platform or transduction scheme used, is a sensitive and selective bioreceptor, or a recognition element, that can provide specific binding to the target analyte. Many environmentally or pharmacologically interesting target analytes might not have naturally appropriate binding partners for biosensor development, but phage display can facilitate the production of novel receptors beyond known biomolecular interactions, or against toxic or nonimmunogenic targets, making the technology a valuable tool in the quest of new recognition elements for biosensor development.

Generating a recombinant phosphothreonine-binding domain for a phosphopeptide of the human transcription factor, c-Myc

Transcription factor c-Myc is an oncoprotein that is regulated at the post-translational level through phosphorylation of two conserved residues, Serine 62 (Ser62) and Threonine 58 (Thr58). A highly specific tool capable of recognizing Myc via pThr58 is needed to monitor activation and localization. Through phage display, we have isolated 10 engineered Forkhead-associated (FHA) domains that selectively bind to a phosphothreonine (pThr)-containing peptide (53-FELLPpTPPLSPS-64) segment of human c-Myc. One domain variant was observed to bind to the Myc-pThr58 peptide with a K_D value of 800 nM and had >1000-fold discrimination between the phosphorylated and non-phosphorylated peptide. The crystal structure of the engineered FHA Myc-pThr-binding domain (Myc-pTBD) was solved in complex with its cognate ligand. The Myc-pTBD was observed to be structurally similar to the yeast Rad9 FHA1 domain, except that its β4-β5 and β10-β11 loops form a hydrophobic pocket to facilitate the interaction between the domain and the peptide ligand. The Myc-pTBD's specificity for its cognate ligand was demonstrated to be on a par with 3 commercial polyclonal antibodies, suggesting that this recombinant reagent is a viable alternative to antibodies for monitoring Myc regulation.

Directed evolution and biophysical characterization of a full-length, soluble, human caveolin-1 variant

Protein engineering by directed evolution can alter proteins' structures, properties, and functions. However, membrane proteins, despite their importance to living organisms, remain relatively unexplored as targets for protein engineering and directed evolution. This gap in capabilities likely results from the tendency of membrane proteins to aggregate and fail to overexpress in bacteria cells. For example, the membrane protein caveolin-1 has been implicated in many cell signaling pathways and diseases, yet the full-length protein is too aggregation-prone for detailed mutagenesis, directed evolution, and biophysical characterization. Using a phage-displayed library of full-length caveolin-1 variants, directed evolution with alternating subtractive and functional selections isolated a full-length, soluble variant, termed cav_sol, for expression in E. coli. Cav_sol folds correctly and binds to its known protein ligands HIV gp41, the catalytic domain of cAMP-dependent protein kinase A, and the polymerase I and transcript release factor. As expected, cav_sol does not bind off-target proteins. Cellular studies show that cav_sol retains the parent protein's ability to localize at the cellular membrane. Unlike truncated versions of caveolin, cav_sol forms large, oligomeric complexes consisting of approximately >50 monomeric units without requiring additional cellular components. Cav_sol's secondary structure is a mixture of α-helices and β-strands. Isothermal titration calorimetry experiments reveal that cav_sol binds to gp41 and PKA with low micromolar binding affinity (K_D). In addition to the insights into caveolin structure and function, the approach applied here could be generalized to other membrane proteins.

A comparison of phosphospecific affinity reagents reveals the utility of recombinant Forkhead-associated domains in recognizing phosphothreonine-containing peptides

Phosphorylation is an important post-translational event that has a wide array of functional consequences. With advances in the ability of various technologies in revealing and mapping new phosphosites in proteins, it is equally important to develop affinity reagents that can monitor such post-translational modifications in eukaryotic cells. While monoclonal and polyclonal antibodies have been shown to be useful in assessing the phosphoproteome, we have expanded our efforts to exploit the Forkhead-associated 1 (FHA1) domain as scaffold for generating recombinant affinity reagents that recognize phosphothreonine-containing peptides. A phage display library of FHA1 variants was screened by affinity selection with 15 phosphothreonine-containing peptides corresponding to various human transcription factors and kinases, including human Myc, calmodulin-dependent protein kinase II (CaMKII), and extracellular-signal regulated kinases 1 and 2 (ERK1/2). The library yielded binding variants against 10 targets (66% success rate); success was largely determined by what residue occurred at the +3 position (C-terminal) to the pThr moiety (i.e., pT+3). The FHA domains binding Myc, CaMKII, and ERK1/2 were characterized and compared against commercially available antibodies. All FHA domains were shown to be phosphorylation-dependent and phosphothreonine-specific in their binding, unlike several commercial monoclonal and polyclonal antibodies. Both the pThr and the residue at the pT+3 position were major factors in defining the specificity of the FHA domains.

Ligand screening using fluorescence thermal shift analysis (FTS)

The fluorescence thermal shift (FTS) method is a biophysical technique that can improve productivity in a structural genomics pipeline and provide a fast and easy platform for identifying ligands in protein function or drug discovery screening. The technique has gained widespread popularity in recent years due to its broad-scale applicability, throughput, and functional relevance. FTS is based on the principle that a protein unfolds at a critical temperature that depends upon its intrinsic stability. A probe that will fluoresce when bound to hydrophobic surfaces is used to monitor protein unfolding as temperature is increased. In this manner, conditions or small molecules that affect the thermal stability of a protein can be identified. Herein, principles, protocols, data analysis, and special considerations of FTS screening as performed for the Center for Structural Genomics of Infectious Diseases (CSGID) pipeline are described in detail. The CSGID FTS screen is designed as a high-throughput 384-well assay to be performed on a robotic platform; however, all protocols can be adapted to a 96-well format that can be assembled manually. Data analysis can be performed using a simple curve fitting of the fluorescent signal using a Boltzmann or double Boltzmann equation. A case study of 100 proteins screened against Emerald Biosystem's ADDit™ library is included as discussion.

Directed evolution of the forkhead-associated domain to generate anti-phosphospecific reagents by phage display

While affinity reagents are valuable tools for monitoring protein phosphorylation and studying signaling events in cells, generating them through immunization of animals with phosphopeptides is expensive, laborious, and time-consuming. An attractive alternative is to use protein evolution techniques and isolate new anti-phosphopeptide binding specificities from a library of variants of a phosphopeptide-binding domain. To explore this strategy, we attempted to display on the surface of bacteriophage M13 the N-terminal Forkhead-associated (FHA1) domain of yeast Rad53p, which is a naturally occurring phosphothreonine (pT)-binding domain, and found it to be nonfunctional due to misfolding in the bacterial periplasm. To overcome this limitation, we constructed a library of FHA1 variants by mutagenic PCR and isolated functional variants after three rounds of affinity selection with its pT peptide ligand. A hydrophobic residue at position 34 in the β1 strand was discovered to be essential for phage display of a functional FHA1 domain. Additionally, by heating the phage library to 50°C prior to affinity selection with its cognate pT peptide, we identified a variant (G2) that was ~8°C more thermally stable than the wild-type domain. Using G2 as a scaffold, we constructed phage-displayed libraries of FHA1 variants and affinity selected for variants that bound selectively to five pT peptides. These reagents are renewable and have high protein yields (~20-25mg/L), when expressed in Escherichia coli. Thus, we have changed the specificity of the FHA1 domain and demonstrated that engineering phosphopeptide-binding domains is an attractive avenue for generating new anti-phosphopeptide binding specificities in vitro by phage display.