Kinetic target-guided synthesis (KTGS) is an emerging protein-templated strategy able to provide potent hits and leads for drug discovery campaigns. By applying the bioorthogonal chemistry-inspired concept, a pair of reagents with complementary reactive moieties are irreversibly linked together under the influence and optimal conformational fitting inside the protein binding pocket of interest. Our team has successfully disclosed the discovery of potent (nanomolar range) and selective insulin-degrading enzyme (IDE) and endoplasmic reticulum aminopeptidase 2 (ERAP2) inhibitors through KTGS by combining a series of biocompatible alkynes and azides for the in-situ “click” generation of 1,2,3-triazoles (2, 3).

In order to highlight the isoform selectivity aspect of KTGS between different members of the M1 zinc-metalloprotease subfamily, we have developed an optimized protocol for identifying potential hits targeting the catalytic site of endoplasmic reticulum aminopeptidase 1 (ERAP1). An in-house available alkyne library (250 entries) of diverse functionalities has been incorporated with 13 in-house synthesized hydroxamate azide warheads to afford the equilibrium-driven synthesis of 1,4- or 1,5-disubstituted hydroxamic acid triazole mixtures, detected by mass-spectrometry. The multi-component format was followed by using clusters of our chemical precursors in order to access a larger number of putative ligands. Multi-parametric assay and analytical optimizations were pursued to enhance our readout data and method sensitivity, including an additional single-component KTGS approach. After attentive cross-examination of our collected data, we successfully extracted 18 putative ligands that were confirmed as hits by displaying a dose-response inhibitory activity on ERAP1 at a low micromolar range and with a few of them reaching an IC50 value below 25 μΜ.

Winner (2nd place) of Best Oral Communication award

In this short presentation, ESR4 describes the process followed to obtain monoclonal antibodies from hybridoma cell lines, together with a brief overview of the next steps in the course of characterizing the immunopeptidome of the A375 human melanoma cell line after genetic and pharmacological inhibition of ERAP1, one of the two main aminopeptidases found to act in the endoplasmic reticulum.

The human immune system utilises small peptides bound to major histocompatibility complexes I (MHC I) - found on the cell surface –to distinguish healthy cells from those that are either infected or cancerous. The MHC-bound peptides are recognised by T- cells in a process called antigen presentation. The set of peptides presented to the immune system, known as the immunopeptidome, is generated intracellularly by proteolytic cleavage by the proteasome and other proteases, followed often by trimming excess N-terminal amino acids by endoplasmic reticulum aminopeptidases (ERAPs). ERAPs have been correlated with tumour escape or immune system hyperstimulation. Recent results have revealed that shifts in the immunopeptidome due to genetic variation or pharmacological modulation of ERAPs can enhance or down-regulate adaptive immune responses, which makes them promising targets for cancer immunotherapy and treatments of autoimmune diseases. In light of these results, the proteomic characterization of the immunopeptidome after a pharmacological or genetic alteration of ERAP activity is of very high importance. However, for the MS-based characterisation of the immunopeptidome, MHC I molecules with their bound peptides need to be purified by immunoaffinity chromatography, which requires monoclonal antibodies specific to MHC I molecules.

M1 aminopeptidases consisting of endoplasmatic reticulum aminopeptidases 1 and 2 (ERAP 1 and ERAP2) and insulin-regulated aminopeptidase (IRAP) are a subfamily of aminopeptidases that play a critical role in the process of antigenic peptides generation and the overall control of adaptative immune system response. The regulation of such enzymes could be a new way to approach autoimmune disease treatment and improve immunotherapy against cancer. Therefore, targeting ERAPs and IRAP has raised significant interest during the last decade because of their potential applications.1

These aminopeptidases are a group of metalloproteases that contain a zinc atom in the active site which is involved in the cleavage of the N-terminal amide bond of peptides. Different methods have been used for the design of inhibitors that could present selectivity using several chelating groups, such as carboxylic acids or phosphorus-containing groups. Phosphinic peptide analogues are pseudopeptides that contain a phosphinic group in lieu of the scissile peptide bond of a natural peptide that chelates with the zinc atom in the active site. This phosphinic group in the carboxylic position, together with the carbon in the place of the amine group mimic the structure of a natural peptide’s transition state during the amide bond cleavage, which results in efficiently potent as well as selective inhibitors.2

Phosphinic peptide inhibitors have already shown good potency and selectivity against other metalloproteases such as Aminopeptidase A or MMP-12.3,4 Regarding ERAPs/IRAP, several potent phosphinic inhibitors have already been reported, however no selective inhibitor has been developed so far for these enzymes.5

In the present work, phosphinic peptide libraries have been designed and synthesized, using combinatorial and solid-phase techniques. These libraries have set the basis for the detailed study of structure-activity relationships for ERAPs/IRAP, aiming to the identification of patterns that will lead to the discovery of selective inhibitors of target enzymes.

Ankylosing spondylitis (AS) is a chronic inflammatory arthritis primarily affecting the spine and sacroiliac joints, that link the pelvis and lower spine, with a strong genetic association, particularly with the HLA-B27 gene. The ERAP1 gene (Endoplasmic Reticulum Aminopeptidase 1), coding for a protein involved in the processing of peptides, is another genetic factor associated with AS. AS-affected patients can manifest inflammatory bowel disease (IBD) with a frequency of 5-10%. Metabolomics studies have already been conducted on an AS preclinical model for biomarkers’ identification. We performed untargeted spatial metabolomic analysis via MALDI-MSI on intestine samples from an AS preclinical model and wild-type control and correlated the spatial distribution of identified biomarkers within histological layers to understand their differential modulation and the link to the disease. Epithelial hyperplasia was observed in samples from the transgenic group. The combination of two data analysis approaches, “targeted” t-test and “untargeted” clustering, allowed for the identification of m/z significantly modulated between the two groups. Among lipids, choline's containing phospholipids were highly modulated between wild-type and transgenic groups. The histological regions in which these biomarkers spatially distribute were mucosa and lumen. Phosphatidylcholines (PCs) are a class of phospholipids with a crucial role in stress and inflammatory responses as well as cellular homeostasis, serving as key components of cell membranes and contributing to signaling pathways. Their downregulation in the mucosal layer of the colon is associated with the reduction of hydrophobic and protective properties of the mucus. Mass Spectrometry Imaging gives more insights into the heterogenous spatial modulation of disease-related biomarkers.

Characterisation of quantitative structure-property relationships (QSPR) of compounds has always been a top topic in drug design and material science. Many machine-learning (ML) methods convert chemical compounds into molecular descriptors or other computer-interpretable representations to develop a reliable prediction model. In this work, we developed models that predict the permeability of compounds on the Caco-2 cell line and on the blood-brain barrier (BBB). The Caco-2 cell line model is widely used to evaluate the in vitro human intestinal permeability of compounds due to its morphological and functional similarity to human enterocytes. BBB permeability is another important property that is used to establish the drug-likeness of a molecule, as it establishes whether the molecule can cross the BBB when desired (for example, for central nervous system (CNS)-active molecules). A self-attention-based message-passing neural network (SAMPN) model, directly learns the most relevant features of each QSPR task and identifies the importance of substructures of compounds in each property. The results of SAMPN are characterised by high evaluation metrics, therefore, the trained models seem promising tools for virtual screening in the early stage of drug discovery. For the BBB prediction, the model achieved an area-under-curve (AUC) of 0.94, and an accuracy (ACC) of 0.89. The Caco-2 permeability model achieved an AUC of 0.87 and an ACC of 0.80. Since SAMPN assigns a degree of importance to substructures, provides interpretable results which can help researchers design novel, advanced molecules.

In this poster, ESR4 describes the effects of allosteric or genetic inhibition of ERAP1 on the Immunopeptidome and Proteome of A375 Melanoma Cells. Endoplasmic reticulum aminopeptidase 1 (ERAP1) is an enzyme that resides in the endoplasmic reticulum (ER). Its role is to fine-tune our immune system's responses. Specifically, ERAP1 trims off certain amino acids from the beginning (N-terminus) of peptides. This trimming process readies these fragments to be loaded onto Major Histocompatibility Complex I molecules (MHC-I). MHC-I, together with the loaded peptides that are collectively known as the immunopeptidome, are then transferred on the cell surface. There they can be recognized by the immune system, which may then respond in case of a potential threat.. While ERAP1 activity is essential for presenting many antigenic peptides, it can also excessively trim others, hindering their MHC-I-mediated presentation. This over-trimming can serve as a mechanism of immune evasion in cancer, as it may lead to the destruction of cancer-specific antigens, thereby reducing cytotoxic T-cell responses. Consequently, inhibiting ERAP1 is an emerging strategy for cancer immunotherapy. Modulating ERAP1 can be simulated in vitro using two distinct methods: genetic knock-out or pharmacological inhibition. Initially, efforts to develop ERAP1 inhibitors were concentrated on the enzyme's active site, but concerns about off-target effects have redirected attention to allosteric sites. In the presented study, Martha treated a melanoma cell line with an allosteric inhibitor of ERAP1 and compared the immunopeptidome with that of wild-type and ERAP1 KO cells. Moreover, she performed a proteomics analysis to explore whether indirect changes in the cells’ proteome can contribute to the alteration of the immunopeptidome.

Kinetic target-guided synthesis (KTGS) is an emerging protein-templated strategy able to provide potent hits and leads for drug discovery campaigns. By applying the bioorthogonal chemistry-inspired concept, a pair of reagents with complementary reactive moieties are permanently linked together. This linkage occurs under the influence and optimal conformational fitting inside the protein binding pocket of interest. Our team has successfully disclosed the discovery of potent (nanomolar range) and selective insulin-degrading enzyme (IDE) and endoplasmic reticulum aminopeptidase 2 (ERAP2) inhibitors through KTGS. This involved combining a series of biocompatible alkynes and azides for the in-situ “click” synthesis of 1,2,3-triazoles.

To highlight the isoform selectivity aspect of KTGS between different members of the M1 zinc-metalloprotease subfamily, we are currently developing a similar optimised protocol to identify potential hit compounds targeting the catalytic site of endoplasmic reticulum aminopeptidase 1 (ERAP1). An in-house available alkyne library (250 entries) of diverse functionalities has been incorporated with 13 in-house synthesized hydroxamate azide warheads. This combination allows for the equilibrium-driven synthesis of 1,4- or 1,5-disubstituted hydroxamic acid triazole mixtures, which are detected using mass-spectrometry. The multi-component format was followed by using clusters of our chemical precursors in order to access a larger number of putative ligands. We pursued multi-parametric assay and analytical optimisations to enhance our readout data and method sensitivity, including an additional single-component KTGS approach.

Ankylosing spondylitis (AS) is a chronic immune-mediated inflammatory disease, whose development is associated with HLA-B27, a gene within the major histocompatibility complex. Endoplasmic reticulum aminopeptidase 1 (ERAP1) is the second most important identified locus in AS and its inhibition can be explored as a therapeutic approach in HLA-B27+ patients. Identifying specific biomarkers for AS will help to assess molecular changes upon ERAP1-inhibitory treatment. In this work, we present an unbiased spatial metabolomic characterisation using MALDI-Mass Spectrometry Imaging (MSI) on an AS preclinical model. We developed a novel data analysis approach, combining targeted and untargeted analysis to identify in-situ biomarkers related to this pathological condition. In the targeted approach, histological regions were defined through hematoxillyne and eosine (H&E) staining and the mass spectra of samples from wild-type (WT) and transgenic groups were compared by statistical analysis. A large number of m/z were modulated between the two groups in colon and cecum samples. In the untargeted approach, a batch integration step was implemented to remove the unrelated analytic bias and improve the biological cluster identification between analytical tissue slides. The analysis showed again a strong differential modulation in cecum and colon samples, in terms of clusters distribution between WT and transgenic samples. In these compartments, statistical analysis, i.e. T-test, was performed on the clusters, to identify the m/z differentially modulated. This novel analysis approach applied to the MSI data allowed for the identification of spatially resolved metabolites, meaningful for the disease, and the investigation of their distribution in specific histological regions.

ERAP1 trimming of antigenic peptides has an important role in anticancer immune responses, by generating optimal peptides for MHC-I loading and presentation to T cells. Previous studies showed that colon carcinomas tend to overexpress ERAP1 and that ERAP1 is able to destroy by over-trimming strong cancer-related antigens, such as GSW11. Moreover, ERAP1 knockdown by RNAi induces a strong anticancer immune reaction against CT26 colon carcinoma cells in vivo and stimulates an immunological memory. We established an ERAP1 knock-out MC38 colon carcinoma cell line and will employ a syngeneic mouse model to analyse differences in tumour progression and anticancer immune responses between wild-type MC38 cells and ERAP1-KO MC38 cells. A CRISPR-Cas9 approach was used to knock out the ERAP1 gene in MC38. Single-cell clones were isolated by limiting dilution and confirmed for ERAP1 knock-out by PCR, Sanger Sequencing and Western Blot. Flow cytometry was used to analyse the difference in MHC-I expression between WT and ERAP1-KO MC38 cells. We obtained one ERAP1-KO MC38 clone where the absence of ERAP1 protein was confirmed by Western Blot and Sanger sequencing. Even if stimulated with Interferon γ ERAP1 knock-out cells didn’t show any expression of ERAP1 protein. Flow cytometry analysis showed an average 20% reduction in surface expression of MHC class I proteins H-2Kb and H-2Db on ERAP1 Knock-out cells compared to MC38 WT cells. Future work will investigate the role of ERAP1 on tumour growth rate and immune cells phenotype of WT vs KO MC38 injected in the sub cute of C57BL/6 mice.

ERAP1, is an enzyme that plays a key role in the antigen processing and presentation pathway as an editor of the peptide repertoire, allowing for the binding of neoantigens to MHC I.  The trimming mechanism of ERAP1 is still not fully understood, if we characterise the effect of ERAP1 allotypes and their inhibition on the peptide repertoire we will gain valuable insights to elucidate this process. It has been seen that the pharmacological modulation of this enzyme has the potential to promote T cell and NK-mediated anti-tumour cytotoxic response (Cifaldi et al. 2011).

By assessing the effect of ERAP1 inhibition on the peptide repertoire presented by tumour cells, and the effect of ERAP1 allotypes and their inhibition on the peptide repertoire we will be able to gain a better understanding of the trimming mechanism of ERAP1 and its implications in cancer.   

We have been able to successfully knockout the ERAP1 gene in CT26 cells and optimised a methodology for knockout detection and validation consisting in Sanger sequencing, Western Blot, MHC I expression and TOPO cloning. We did further analysis to assess the MHC I kinetics and stability of the knockout cell line by investigating the dissociation of pMHC complexes at the cell surface using Brefeldin A decay assay and an acid-stripping test. We now plan to insert the different human allotypes back into the CT26 ERAAP-KO cell line using retroviral transduction, determine the effect of ERAP1 inhibitors such as leucinethiol on the trimming activity of relevant ERAP1 alleles and manipulate ERAP1 function to modulate the presentation of peptide epitopes in the CT26 cell line as well as other cell lines. 

Endoplasmic reticulum aminopeptidases (ERAPs) are multifunctional metalloproteases involved in antigen processing and presentation. By trimming the N-terminus of peptide precursors, ERAPs succeed to generate mature antigenic epitopes ready for loading upon major histocompatibility complex class I (MHC-I) molecules and ultimately eliciting T-cytotoxic cellular responses. The connection of ERAP1/2 genetic variants with putative therapeutic applications in inflammatory and proliferative disorders has often been at the centre of our efforts by developing small-molecule probes that can fine-tune immune dysregulation. Targeted and controlled regulation of ERAP1 constitutes a challenging task because of the high structural similarity and broad substrate specificity inside the M1 enzyme family. As a result, the identification and multiparametric optimization of potent and novel chemical scaffolds for ERAP1 selective modulation is imperative. Here we present different strategies to access new ERAP1 inhibitors. First, carboxylic acids and bioisosteres are an important class of bioactive compounds that can be optimized efficiently to target metalloproteases. Structure-activity relationships (SAR) in combination with computer modelling and silico docking studies could fill the gaps in our existing chemical series. Second, Kinetic target-guided synthesis (KTGS) and fragment-based drug discovery (FBDD) will be used to generate hits and leads for ERAP1 in a systematic and refined manner.