#1PIPE Investor Presentation: Supplementary Materials December 2021 AI ODYSSEY Benevolent ACQUISITION 24#2Presentation Disclaimer For the purposes of this notice, this confidential document (the "Presentation") that follows shall mean and include the slides that follow this notice, the oral presentation of the slides by members of management of Odyssey Acquisition S.A., a public limited liability company (société anonyme) incorporated under the laws of the Grand Duchy of Luxembourg ("Odyssey") or BenevolentAl Limited, a private company registered in England (the "Company") or any person on their behalf, the question-and-answer session that follows that oral presentation, hard copies of this document and any materials distributed at, or in connection with, that presentation. By attending the meeting where theoral presentation is made, or by reading the Presentation, you will be deemed to have: (i) agreed to the following limitations and notifications and made the following undertakings; and (ii) acknowledged that you understand the legal and regulatory sanctions attached to the misuse, disclosure or improper circulation of this Presentation. This Presentation is provided in confidence. By accepting this Presentation, and in consideration of it being made available to recipients, each recipient agrees to keep strictly confidential the information contained in it and any information otherwise made available by Odyssey or the Company, whether orally or in writing. This Presentation has been provided to each recipient solely for their information, and may not be reproduced, copied, published, distributed or circulated to any third party, in whole or in part, without the express prior written consent of Odyssey and the Company. This Presentation is intended solely for investors that are qualified institutional buyers (as defined in Rule 144A under the Securities Act of 1933, as amended (the "Securities Act")), institutional accredited investors (as defined in Rule 501 under the Securities Act) and eligible institutional investors outside the U.S. and has been prepared for the purposes of familiarizing such investors with a potential private placement of securities in connection with the potential business combination between Odyssey and the Company and any related transactions (collectively, the "Proposed Transactions") and for no other purpose. The release, reproduction, publication or distribution of this Presentation, in whole or in part, or the disclosure of its contents, without the prior consent of Odyssey and the Company is unlawfuland prohibited. Persons into whose possession this document comes should inform themselves about, and observe, any such restrictions. By accepting this Presentation, each recipient agrees: (i) that the information included in this Presentation is confidential and may constitute material non-public information, (ii) to maintain the confidentiality of all information that is contained in this Presentation and not already in the public domain, and (iii) to use this Presentation for the sole purpose of evaluating Odyssey, the Company and the Proposed Transactions. This Presentation is not, and should not be construed as, a prospectus for the purposes of Regulation (EU) 2017/1129 (as amended, the "Prospectus Regulation"), and does not constitute or form part of, and should not be construed as an offer or the solicitation of an offer to subscribe for or purchase shares and/or securities of Odyssey or the Company, and nothing contained herein shall form the basis of or be relied on in connection with, or act as any inducement to enter into, any contract or commitment whatsoever, in particular, it must not form the basis of any investment decision. In addition, this Presentation is being furnished on a confidential basis in the European Economic Area to a limited number of "qualified investors" (as defined in the Prospectus Regulation) and, in the United Kingdom, to "qualified investors" (as defined in Regulation (EU) 2017/1129 as it forms part of U.K. domestic law by virtue of the European Union (Withdrawal) Act 2018 (the "EUWA") (the "U.K. Prospectus Regulation")), that are also (i) persons having professional experience in matters relating to investments that fall within the definition of "investment professionals" in Article 19(5) of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2005 (as amended, the "Order"); or (ii)high net worth entities or other persons falling within Article 49(2)(a) to (e) of the Order. In any European Economic Area ("EEA") Member State or in the United Kingdom, this Presentation is not addressed to and is not directed at any retail investor in the EEA or the United Kingdom. For these purposes, the expression "retail investor" means: (A) in an EEA Member State, a person who is one (or more) of: (i) a retail client as defined in point(11) of Article 4(1) of Directive 2014/65/EU (as amended, "MiFID II"); (ii) a customer within the meaning of Directive (EU) 2016/97, where that customer would not qualify as a professional client as defined in point (10) of Article 4(1) of MiFID II; or (iii) not a qualified investor as defined in the Prospectus Regulation; and (B) in the United Kingdom, a person who is one (or more) of: (i) a retail client, as defined point (8) of Article 2 of Regulation (EU) No 2017/565 as it forms part of domestic law by virtue of the EUWA; (ii) a customer within the meaning of the provisions of the Financial Services and Markets Act 2000 (as amended, the "FSMA") and any rules or regulations made under the FSMA to implement Directive (EU) 2016/97 (as amended) where that customer would not qualify as a professional client as defined in point (8) of Article 2(1) of Regulation (EU) No 600/2014 as it forms part of domestic law by virtue of the EUWA; or (iii) not a qualified investor as defined in Article 2 of the U.K. Prospectus Regulation. This Presentation and any oral statements made in connection with this Presentation do not constitute an offer to sell, or the solicitation of an offer to buy, or a recommendation to purchase, any securities in any jurisdiction, or the solicitation of any proxy, consent or approval in any jurisdiction in connection with the Proposed Transactions, nor shall there be any sale, issuance or transfer of any securities in any jurisdiction where, or to any person to whom, such offer, solicitation or sale may be unlawful under the laws of such jurisdiction. This Presentation does not constitute either advice or a recommendation regarding any securities. Any offer to sell securities will be made only pursuant to a definitive subscription agreement and will be made in reliance on an exemption from registration under the Securities Act for offers and sales of securities that do not involve a public offering. Odyssey and the Company reserve the right to withdraw or amend for any reason any offering and to reject any subscription agreement for any reason. The communication of this Presentation is restricted by law; in addition to any prohibitions on distribution otherwise provided for herein, this Presentation is not intended for distribution to, or use by any person in, any jurisdiction where such distribution or use would be contrary to local law or regulation. The contents of this Presentation have not been reviewed by any regulatory authority in any jurisdiction. This Presentation does not purport to be all-inclusive or to contain all the information that a stakeholder may desire to have in evaluating Odyssey, the Company and the Proposed Transactions. This Presentation is qualified entirely by reference to Odyssey's the Company's publicly disclosed information. No representation or warranty, express or implied, is made or given by or on behalf of Odyssey, the Company or any of their shareholders, directors, officers, agents, employees or advisers as to the accuracy, reliability, completeness or fairness of the information, opinions or forward-looking statements contained in this Presentation, or any revision thereof, or of any other written or oral information made or to be made available to any recipient and liability therefore is expressly disclaimed. Accordingly, none of Odyssey, the Company or any of their shareholders, directors, officers, agents, employees or advisers take any responsibility for, or will accept any liability whether direct or indirect, expressor implied, contractual, tortious, statutory or otherwise, in respect of, the accuracy or completeness of the information or for any of the opinions contained herein or for any errors, omissions or misstatements or for any loss, howsoever arising, from the use of this Presentation. In furnishing this Presentation, neither Odyssey nor the Company undertake or agree to any obligation to provide stakeholders with access to any additional information or to update this Presentation or to correct any inaccuracies in, or omissions from, this Presentation that may become apparent. The information and opinions contained in this Presentation are provided as at the date of this Presentation. The contents of this Presentation are not to be construed as legal, financial or tax advice. Each stakeholder should contact his, her or its own legal adviser, independent financial adviser or tax adviser for legal, financial or tax advice. To the extent available, the data contained in this Presentation has come from official or third party sources. Third party industry publications, studies and surveys generally state that the data contained therein have been obtained from sources believed to be reliable, but that there is no guarantee of the accuracy or completeness of such data. While Odyssey and the Company believe that each of these publications, studies and surveys has been prepared by a reputable source, the Company has not independently verified the data contained therein. In addition, certain of the data contained in this Presentation come from the Company's own internal research and estimates based on the knowledge and experience of the Company's management in the market in which the Company operates. While the Company believes that such research and estimates are reasonable and reliable, they, and their underlying methodology and assumptions, have not been verified by any independent source for accuracy or completeness and are subject to change without notice. Accordingly, undue reliance should not be placed on any of the estimates or research contained in this Presentation. Forward-looking statements. Certain information contained in this Presentation, including any information as to the Company's strategy, plans or future financial or operating performance constitutes "forward-looking statements". All statements contained in this Presentation that do not relate to matters of historical fact should be considered forward-looking statements, and these forward-looking statements can be identified by the use of terminology such as, "aims", "anticipates", "assumes", "believes", "budgets", "could", "contemplates", "continues", "estimates", "expects", "intends", "may", "plans", "predicts", "projects", "schedules", "seeks", "shall", "should", "targets", "would", "will" or, in each case, their negative or other variations or comparable terminology. Forward-looking statements appear in a number of places throughout this Presentation and include, but are not limited to, express or implied statements relating to: the Company's business strategy and outlook; the Company's future results of operations; the Company's future financial and market positions; the Company's margins, profitability,cash, borrowings and prospects; expectations as to the Company's future growth; the Company's plans with respect to capital expenditure; general economic trends and other trends in the industry in which the Company operates; the impact of laws and regulations on the Company and its operations; and the competitive environment in which the Company operates. By their nature, forward-looking statements are based upon a number of estimates and assumptions that, whilst considered reasonable by the Company are inherently subject to significant business, economic and competitive uncertainties and contingencies. Known and unknown factors could cause actual results to differ materially from those indicated, expressed or implied in such forward-looking statements. Forward-looking statements are not guarantees of future performance. Any forward-looking statements in this Presentation reflect the Company's current view with respect to future events and are subject to certain risks relating to future events and other risks, uncertainties and assumptions. The forward-looking statements contained in this Presentation reflect knowledge and information available as of the date of preparation of this Presentation. The Company and its directors expressly disclaim any obligations or undertaking to update or revise publicly any forward-looking statements, whether because of new information, future events or otherwise, unless required to do so by applicable law or regulation. Nothing in this Presentation should be construed as a profit forecast. The information contained in this Presentation is not an offer to sell or a solicitation of an offer to purchase interests in any company or a related entity, nor is it intended to provide, and should not be relied on for, investment, tax, legal or financial advice. Certain financial data included in the Presentation may consist of "non-IFRS financial measures", which may not be comparable to similarly-titled measures as presented by other companies, nor should they be considered as an alternative to the historical financial results or other indicators of the Company's cash flow based on IFRS. Even though the non-IFRS financial measures are used by management to assess the Company's financial position, financial results and liquidity and these types of measures are commonly used by investors, they have important limitations as analytical tools, and the recipients should not consider them in isolation or as a substitute for analysis of the Company's financial position or results of operations as reported under IFRS. AstraZeneca's intention to make an equity investment is an indication and not a binding agreement or commitment to purchase and therefore AstraZeneca could determine to purchase more, less or no shares, or we could determine to sell more, less or no shares to AstraZeneca. Neither this offering nor AstraZeneca's equity investment arecontingent upon one another.#3Risk Factors Any investment in Odyssey or the Company involves numerous risks and uncertainties related to the Company's business and the Proposed Transactions that may result for investors in a partial or total loss of their investment. The following is a non-exclusive selection of key risks that, alone or in combination with other events or circumstances, could have a material adverse effect on the Company's business, financial condition, results of operations and prospects as well as the Proposed Transactions. Investors should read, understand and carefully consider the risks and uncertainties described below. This summary is not comprehensive and the below key risks are subject to change. An additional discussion of the risks and uncertainties of the Company and the Proposed Transaction will be included in under the heading "Risk Factors" contained in the circular and prospectus in connection with the proposed business combination. Risks Related to the Company's Business and Industry 12 1. 2. 3. 4. 5. 6. 3456 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 2222 21. 22. 23. We have a history of significant operating losses, and we expect to incur losses over the next several years. Our operating history and business model may make it difficult for you to evaluate the success of our business to date and to assess our future viability, which may depend on us obtaining additional capital, which might not be available on economically acceptable terms, or at all. Our interim and annual results may fluctuate significantly, which could adversely impact the value of our shares. We have no products approved for commercial sale, our revenues to date have been derived from a single source and it may take several years before we generate revenue from product sales, if at all. If we and our present and future collaborators are unable to successfully develop and commercialise drug products, our revenues may be insufficient for us to achieve or maintain profitability. All of our drug candidates are in early-stage preclinical development or in clinical development. If we are unable to advance our drug candidates through clinical development, to obtain regulatory approval and ultimately to commercialise our drug candidates, or if we experience significant additional costs or significant delays in doing so, our business, financial condition, results of operations and prospects will be materially harmed. We are substantially dependent on our technology platform to identify promising drug targets to accelerate drug discovery and development. Our platform technology may fail to discover and design molecules with therapeutic potential or may not result in the discovery and development of commercially viable products for us or our collaborators. If we cannot maintain existing partnerships, including data partnerships, and/or enter into new partnerships or similar business arrangements, our business could be adversely affected. We face substantial competition, which may result in others discovering, developing or commercialising products before or more successfully than we do, requiring us to rapidly adapt our approach to significant technological change and respond to the introduction of new products and technologies to remain competitive. We contract with third parties, including, but not limited to, a number of contract research organisations ("CROS"), site providers, laboratory testing service providers, and universities for assay and experimental work for all of our drug programmes, including where applicable the manufacture of our drug candidates for preclinical development and clinical testing, and expect to continue to do so for commercialisation. This reliance on third parties increases the risk of non-performance or delay to some or all of our drug programmes, or that we will not have sufficient quantities of our drug candidates or products or such quantities at an acceptable cost, which could delay, prevent or impair our development or commercialisation efforts. Because we have multiple programmes and drug candidates in our development pipeline, we may expend our limited resources to pursue a particular drug candidate and fail to capitalise on development opportunities or drug candidates that may be more profitable or for which there is a greater likelihood of success. Clinical development involves a lengthy and expensive process with uncertain outcomes. If our preclinical studies and clinical trials are not sufficient to support regulatory approval of any of our drug candidates, we may incur additional costs or experience delays in completing, or ultimately be unable to complete, the development of such drug candidate. If we are unable to obtain, maintain, enforce and protect patent or other intellectual property right protection for our technology and drug candidates or if the scope of such protection obtained is not sufficiently broad, our competitors could develop and commercialise technology and products similar or identical to ours, and our ability to successfully develop and commercialise our technology and drug candidates, as well as the value of our brand and our business, may be adversely affected. Our internal information technology systems, or those of our third-party vendors (including providers of cloud-based infrastructure), contractors or consultants, may fail or suffer security breaches, loss or leakage of data and other disruptions, which could result in a material disruption of our services, compromise sensitive information related to our business, or prevent us from accessing critical information, potentially exposing us to liability or otherwise adversely affecting our business. If we fail to comply with our obligations under any our existing intellectual property licence agreements and data licensing agreements or under any future such agreements, or otherwise experience disruptions to our business relationships with our current or any future licensors, we could lose intellectual property rights (including access to data) that are important to our business. We make use of the UK's small and medium sized enterprises research and development tax relief regime, through which we have obtained cash tax credits from Her Majesty's Revenue & Customs ("HMRC"). HMRC could seek to challenge the historical cash tax credits paid, or a change of law or our circumstances could restrict our ability to claim additional such cash tax credits. Current and future healthcare and artificial intelligence legislative reform measures may have a material adverse effect on our business and results of operations. Regulatory authorities may implement additional regulations or restrictions on the development and commercialisation of our product candidates. Such changes can be difficult to predict, may require significant systems changes, divert the attention of our personnel, subject us to additional liabilities and may adversely adversely affect our business. Compliance with stringent and evolving global privacy and data security requirements could result in additional costs and liabilities to us or inhibit our ability to collect and process data globally, and the failure to comply with such requirements could subject us to significant fines and penalties, which may have a material adverse effect on our business, financial condition or results of operations. The effects of health epidemics, including the ongoing COVID-19 pandemic, in regions where we, or the third parties on which we rely, have business operations could adversely affect our business, including our preclinical studies and clinical trials, as well as the business or operations of our CROS or other third parties with whom we conduct business. Our current and future clinical trials or those of our current or future collaborators may reveal significant adverse events not seen in our preclinical or nonclinical studies and may result in a safety profile that could inhibit regulatory approval or market acceptance of any of our future drug candidates. Interim, "topline" and preliminary data from our clinical trials that we announce or publish from time to time may change as more patient data becomes available and are subject to audit or verification procedures that could result in material variations in our final data. If we experience delays or difficulties in the enrolment of patients and/or provision of medical data in clinical trials, our receipt of necessary regulatory approvals could be delayed or prevented. Al Benevolent 3#4Risk Factors Risks Related to the Proposed Transactions 123 2. 3. 4. 5. 6. 456 2 CO 7. 8. 9. 10. #2 11. 12. Odyssey and the Company will be subject to business uncertainties and contractual restrictions while the proposed business combination is pending. Odyssey and the Company will incur significant transaction and transition costs in connection with the proposed business combination. Odyssey's sponsor and certain of its directors and officers have interests in the proposed business combination that are different from or are in addition to other shareholders in recommending that shareholders vote in favor of approval of the proposed business combination. Odyssey's sponsor holds a significant number of shares of Odyssey's securities, and their entire investment will be lost if the proposed business combination is not completed. Odyssey's sponsor and its directors or officers or their affiliates may elect to purchase shares from public shareholders, which may influence a vote on the proposed business combination and reduce Odyssey's public float. Odyssey does not have a specified maximum redemption threshold. The absence of such a redemption threshold may make it possible for Odyssey and the Company to complete the proposed business combination with which a substantial majority of Odyssey's shareholders do not agree. Warrants will become exercisable for Odyssey's ordinary shares, which would increase the number of shares eligible for future resale in the public market and result in dilution to Odyssey's shareholders. The ability of Odyssey's ordinary shareholders to exercise redemption rights with respect to a large number of shares could deplete Odyssey's trust account prior to the proposed business combination and thereby diminish the amount of working capital of the combined entity. Goldman Sachs International and J.P. Morgan AG and its or their affiliates (the "Placement Agents") are engaged in a wide range of financial services and businesses (including investment management, financing, securities trading, corporate and investment banking and research) and there may be situations where the Placement Agents and/or its or their clients either now have or may in the future have interests, or take actions, that may conflict with Odyssey's or the Company's interests. For example, the Placement Agents have in the past and may, in the ordinary course of business, engage in trading in financial products or undertake other investments for their own account or on behalf of other clients, including, but not limited to, trading in or holding long, short or derivative positions in securities, loans or other financial products of Odyssey, or other entities connected with the Proposed Transactions. Goldman Sachs International is both acting as a Placement Agent in this proposed private placement of securities and as financial advisor to the Company in connection with the proposed business combination, and a potential conflicts of interest, or a perception thereof, may arise as a result of such relationships. Odyssey has not obtained a third-party valuation or fairness opinion in determining whether or not to proceed with the proposed business combination. As Odyssey may migrate its tax residence to the UK prior to closing the proposed business combination, Odyssey may be subject to both the Luxembourg and UK corporate and tax regimes over the coming accounting periods, which could create a conflict in approach to cross-border and domestic compliance. Odyssey may be adversely affected by amendments to the corporate laws, tax laws or accounting policies of either or both of these jurisdictions, which may also have retrospective effect and be implemented unexpectedly. Future tax audits and other investigations conducted by the competent tax authorities in Luxembourg or the UK in respect of Odyssey's residence could result in the assessment of additional taxes, including corporate income taxes and withholding taxes. Odyssey's entitlement to treaty benefits under the 1967 Luxembourg-UK Double Taxation Convention (as modified by the Multilateral Instrument) (the "Treaty") may be withdrawn or the Treaty may be amended. The materialization of any of these risks could have a material adverse effect on our business, net assets, financial condition, cash flows or results of operations. Al Benevolent#5The Benevolent Platform™ Overview 24#6Agenda Product & Technology Team Knowledge Foundations Target Identification Precision Medicine Chemistry Apps and Infrastructure Recent Publications#7Benevolent's Product & Tech Team . End-to-end Al enabled drug discovery capabilities - from . • Knowledge and Data integration through to application in Precision Medicine, TargetID and Chemistry Team of approx 130 technologists across London and NYC Expertise in Al & Machine Learning, Data Science, Bioinformatics, Genetics, Cheminformatics, Software & Hardware engineering, Product Management, Project Management, Design/UX Experience drawn from Big Tech, Biotech, academic research and healthtech startups Team fully integrated with: › Biologists > Chemists > Informatics and BD Al Benevolent 7#8The Benevolent Platform TM The Benevolent Platform TM is our scientifically-validated computational R&D platform. At its core sits a proprietary knowledge graph, which captures the interconnectivity of scientific literature and relevant, available biomedical data. Our suite of exploratory and predictive Al tools allow scientists to identify novel insights, interrogate data within disease networks, ask biological questions, refine hypotheses and interpret results. Ooo 000 Derived Knowledge Target Identification Precision Medicine Chemistry 1 https://www.labiotech.eu/trends-news/exscientia-drug-discovery-ai/ ΑΙ Benevolent 00 8#9The Benevolent Platform™ Knowledge Foundations#10Proprietary Knowledge Graph, purpose-built for drug discovery The data engine that powers the Benevolent Platform™ COMPREHENSIVE DATA DIVERSITY OF DATA GROWTH OF DATA 400m NLP derived relationships 30m structured relationships 85+ data sources used 1bn relationship edges 22m additional mechanism connections 14X growth over 12 months Literature Scientific Literature Patent Literature Regulatory Documents Pathology Diseases Symptoms Biological Systems Cellular Component Molecular Function Biological Process Mechanism Pathways Benevolent Knowledge Graph Experiments Assay Data (Binding, Omics Comparison, CRISPR Screens) Clinical Trial OMICS Genes Proteins Isoforms Transcripts & Variants Molecules Organic Compounds Preclinical Candidates Approved Drugs Antibodies Other Biologics Pharmacology Pharmacokinetics Uniquely combines public, proprietary & inferred knowledge 60%+ of the most important information used by our models is Al-derived, proprietary knowledge ✓ Therapeutic area and drug modality agnostic ✓ Can be deployed with partners in secure cloud environment Al Benevolent 10#11Domain-specific data processing pipelines support integration of a broad range of data modalities for drug discovery Structured Data e.g. Gene, Pathway, Drug databases etc Unstructured Data e.g. Literature Publications, Clinical Trials etc Genetics and Omics e.g. GWAS, RNAseq, Epigenetics etc Chemistry e.g. Binding, Structural, Mechanism of Action etc BAI Experimental Data e.g. Biochemistry ELNS User Controlled Data Access Patient Level Data e.g. Biobank, Partner cohorts etc Summary Data ΑΙ Benevolent Knowledge Graph 1 Precision Medicine Workflows Model Introspection Al Benevolent 11#12Data Fabric The Data Fabric engineering framework orchestrates the ingestion of data from multiple processing pipelines, providing access to versioned and tracked data releases. Input Data Layer DBs APIS Files Dataset downloaders and importers Managed Data Layer UUID Avro (versioned) UUID Builder Data View Layer Excavator Entity Builder Raw Entity Avro (versioned) Avro (versioned) Synonym Builder Relationship Builder Syn Avro (versioned) Extract-Transform-Load Rel Avro (versioned) A Use Case Data OP Al Benevolent 12#13Al used to extract meaningful information from biomedical literature at scale Automated Data Ingestion Source documents (e.g. Scientific Literature, Patents), extracted and normalised for into format optimised for BenevolentAl Natural Language Processing pipeline. Named Entity Recognition Entities mentioned in unstructured text are identified, normalised and grounded to BenevolentAl entities. Relationship Extraction NLP models extract meaningful relationships between identified entities using contextual information. NAME RELATIONSHIP Amyotrophic Lateral Sclerosis Amyotrophic lateral sclerosis (ALS), also known as motor neurone disease (MND) and Lou Gehrig's... SOD1 Is associated with, Syntactically-linked pairs (35k relations in 4.6k papers) ALS example extract above taken from a readout from the Benevolent Platform™ SOD1 mutation causes ALS phenotype in human MNs SOD1 mutation causes ALS phenotype in human MNs SOD1 ALS INFERRED FROM (CHEMBL) DISEASE CLASSIFICATION Motor Neuron Disease SOD1 Primary lateral sclerosis (PLS) is a rare neuromuscular disease characterized by progressive muscle weakness in the voluntary muscles. PLS belongs to a group of disorders known as motor neuron dise... ALS Al Benevolent 13#14The Benevolent Platform™ Target Identification#15Data Preparation: Optimising the set-up 1. Data Preparation 2. Target Prediction 3. Target Triage 4. Target Validation {!!! Focus and curate a biological question Graph Completion Human Queries 1 2 3 4 MM Reinforcement Learning 'omics Target Safety Drugs? Metadata Score Tgt1 Safe Partial OPC, GCNN 0.71 Tgt2 Unsafe No SQT 0.65 Tgt3 Safe Yes OPC, CROM 0.51 Tgt4 PhaseII Offtgts OPC 0.49 ☐ ☐ ☐ ☐ Model fleet predictions Prediction evaluation Equip experts with data Triage predictions Perform Assays Analysis & retro 1. Disease Data Package 2. Predictions Package 3. Hypothesis Package 4. Validated Hypothesis Package Al Benevolent 15#16Data Preparation Al augmented workflows uncover insights from data 1 Disease mechanism Disease specific datasets 2 ΑΙ Benevolent Knowledge Graph 3 selection Clinical phenotyping Genetic & genomic signature detection & curation Selected analysis workflows Enriched data corpus We turn data into Knowledge by applying our Al augmented analysis workflows Data availability, genetic architecture, and mechanistic understanding varies between disease According to these factors we choose the most appropriate workflows to apply to the disease area Al Benevolent 16#17Data Preparation Biological mechanism selection expands nuance and data coverage, even for rare diseases disease or donder ges dose by subcelu system afec de by car process drupted 0 glucose metabolan diseeseenes dab metus de-755 g type 2 diabetes me Disease/Endotype Mechanism / Pathway Expert exploration Tissue/ Cell Type Recommendation algorithms Response to Insulin Example analysis from Benevolent Platform™: Pulmonary Arterial Hypertension Associated with 32k scientific articles and 33k unique relationships from databases, ontologies, genetic data, etc. Mechanism Expansion Endothelial Cell Proliferation Mechanism X Relevance Novelty Specificity Smooth muscle cell proliferation: 14k publications 6.9k relationships Estrogen metabolism: 3.8k publications 2.4k relationships Vasoconstriction: 107k publications 26k relationships Asking the biological question Biological process curation expands data coverage and reveals novel associations. Relationships from similar biology can be leveraged in predictive models. Algorithms recommend important aspects of biology in the context of a disease. Metrics are computed to evaluate mechanism relevance, novelty, and specificity. Scientists select concepts to form the biological question they wish to ask at the target inference stage. Biologists and data scientists collaborate to guide data buildout predictions Benevolent 17 Al#18Target prediction ° 00 00 00 1. Data Preparation 2. Target Prediction 3. Target Triage 4. Target Validation Focus and curate a biological question Graph Completion Human Queries 1 2 3 4 Target Safety Drugs? Metadata Score MIM Tgt1 Safe Tgt2 Unsafe No Partial OPC, GCNN 0.71 SQT 0.65 Reinforcement Learning 'omics Tgt3 Tgt4 Safe Yes OPC, CROM 0.51 PhaseII Offtgts OPC 0.49 Model fleet predictions Prediction evaluation Equip experts with data Triage predictions Perform Assays Analysis & retro 1. Disease Data Package 2. Predictions Package 3. Hypothesis Package 4. Validated Hypothesis Package Benevolent 18#19Target Prediction Many strands of Al are used to generate target predictions Knowledge Graph 586 Path-based Interpretability Tensor Factorization Human Queries Accuracy Hypotheses Augmentation Learned Weighted Rank Aggregation Multi-prediction Aggregation Specialized models are designed to best produce complementary predictions from the Benevolent Knowledge Graph. Smart aggregation combines predictions by examining which models have been successful in the past, learning what combination of features - and which kinds of evidence - identify a successful target. Al-derived assistance provides support for predictions, for example, from Graph Convolutional Neural Networks. 'Neil et al, 2018 - https://arxiv.org/abs/1812.00279 Data-modality specific Genetic Basis Results in a list of potential therapeutic targets for expert triage Benevolent 19#20Target Prediction Diverse training data yields high-quality Al inferences Literature and Knowledge Graph Training Data [Gene] inhibitor attenuates [Disease] [Gene] regulates [Biological Process] Targeted inhibition of [Gene] limits [Disease] + High-Quality Assay Data Previous Disease Programs CD-8 T Cell Activation Endothelial LDL regulation B Cell CD40 Positive Regulation IL-6 IL-8 IP-10 MCP-1 Algorithms identify sentence forms that suggest strong links by expanding on small curated ground-truth datasets Internal experts define a subset of key relationships that model biomedical knowledge from the scientific literature Our large scientific text corpus results in very high recall across diseases and mechanisms Internal experts curate public and private high-quality transcriptomics datasets Datasets are grounded to diseases, mechanisms, and proteins in the knowledge graph, allowing prediction and evaluation from the knowledge graph representation Prior disease programs provide results for training subsequent models Three kinds of information are routinely captured and available for training: Hit/ no-hit Ranked assay results Triage annotations and reasoning (safety, efficacy, novelty, etc.) 1 By volume of high-quality target associations ΑΙ Benevolent 20#21Target Triage: Prioritising insights and hypotheses ° 00 00 00 1. Data Preparation 2. Target Prediction 3. Target Triage 4. Target Validation Focus and curate a biological question Graph Completion Human Queries 1 2 3 4 000000 Target Safety Drugs? Metadata Score MM Tgt1 Safe Partial OPC, GCNN 0.71 Reinforcement Learning Tgt2 Unsafe No 'omics SQT 0.65 Tgt3 Safe Yes OPC, CROM 0.51 Tgt4 Phase II Offtgts OPC 0.49 Model fleet predictions Prediction evaluation Equip experts with data Triage predictions Perform Assays Analysis & retro 1. Disease Data Package 2. Predictions Package 3. Hypothesis Package 4. Validated Hypothesis Package Al Benevolent 21#22Target Triage The Benevolent Platform TM provides a systematic process for target triage sharing insights and context for each prediction Biological rationale Safety Novelty Progressibility Hypotheses Precision medicine Druggability Target List: Mitochondrion Health focus for PD (Demo) - Mitochondrion Health focus for PD (Demo) 179 targets in total poaded from an unknown source) 26 results 13 Uncen 13 Mach Sort by Sort by QFind target Target Symbol clear all Your criteria for triage Ligendebility must be 3 or 44 Programs Panon's Disease Demo Da CSNK2A1 casein kinase 2 alpha 1 Ligandy&Selety-class (out of 4 Thetic Evidence of Chemical Opportunity-of Bogical Rationale-class 2 of Target Expression-class 1 Uncertain Change Your comment (can be edited when changing target decision) Filter by context VCP valosin containing protein Mechanism Ligand-fety-clef Therapeutic Evidence-Chemical Opportunity-class1 of logical Rationale-class&of Target Expression-f Filter by triage status Match Change Your comment can be edited when changing target decision Triage Status Recommendation/Decision Filter by criteria Has suitable tool compound b Has suitable tool compound in... clear EPAS1 endothelial PAS domain protein 1 Ligandabiy-stof Safety-class 3 out of 4) Therapedic Evidence-cles 3nt of Chemical Opportunity-class 1 out of 4 logical Rationale-less 1 out of Target Expression-of Uncertain Change Your comment (can be edited when changing target decision) Screenshot shows Company triage board. 'Company internal drug progamme statistics Visualise G 51 Target hypotheses undergo expert review with criteria that are tailored to the programme Recommended classifications are made to guide triage decisions using the Knowledge Graph and information presented in the triage tool Decisions are captured in a structured and unstructured manner to both ensure an audit trail and allow the system to learn over time Targets that pass triage are progressed into validation In a recent deployment, 39% of targets were progressed into experimental testing. Of those not progressed; 22% were already known to our program experts, 24% were deemed to have safety concerns and 15% did not have sufficient supporting evidence. Benevolent 22 ΑΙ#23Target Validation: Confirming the hypotheses ° 00 00 00 1. Data Preparation 2. Target Prediction 3. Target Triage 4. Target Validation Focus and curate a biological question Graph Completion Human Queries 1 2 3 4 MM Reinforcement Learning 'omics ☐ ☐ ☐ ☐ Target Safety Drugs? Metadata Score Tgt1 Safe Tgt2 Unsafe No Partial OPC, GCNN 0.71 SQT 0.65 Tgt3 Safe Yes OPC, CROM 0.51 Tgt4 PhaseII Offtgts OPC 0.49 Model fleet predictions Prediction evaluation Equip experts with data Triage predictions Perform Assays Analysis & retro 1. Disease Data Package 2. Predictions Package 3. Hypothesis Package 4. Validated Hypothesis Package Benevolent 23 ΑΙ#24Target Validation Target Validation and Progressability Assessment (TPA) MATCH Targets Tool Selection and TPA 'Primary' Assay Go/ No Go 'Secondary' Assay Deep Dive TPA Validated Hypothesis Review Opportunity to differentiate Can a new therapeutic against this target compete relative to the current and future standard of care? Freedom to Operate Exploitable chemistry/biologics space from an intellectual property perspective Selectivity Are there selectivity challenges with off-targets that are expected to cause tolerability issues? Druggability To what extent is the target amenable to small molecules, siRNA or mAb therapeutics? Assayability Can suitable assays be accessed or developed to support medicinal chemistry? Safety Assessment Are there anticipated safety issues that could be problematic for onward development? Benevolent 24 ΑΙ#25The Benevolent Platform™ Precision Medicine#26Molecularly defined clinical sub-phenotypes At BenevolentAl, we detect subgroups of patients by analysing EHR and other clinical data. By using our genetic tooling, we infer genetic signatures for both entire disease cohorts and more refined subsets of the patient population. Subgroup detection 24 Cluster Cluster 1 12 Cluster 2 Cluster 3 Cluster 4 0 100 200 300 400 500 600 الا Clinical interpretation 20 5 Latent 6: 1052 samples -010 -005 0.00 6 005 H Time (days) Genetic Cohorts UK Biobank • Bespoke cohorts • Partner Cohorts Genetic Summary Stats Ma Including: • UK Biobank • GWAS Catalog 1 pseudonymised or anonymised Electronic Health Records data. 2 QTL = quantitative trait loci genetics data. 3 Whole Exome/Genome sequencing GWAS Pipeline Spark-enabled engineering framework WES/WGS3 Pipelines QTL datasets² Target Prediction Variant-to-Gene Annotation (Coloc, VEP, MR, scoping AI/ML methods currently) Target Triage Chromatin Features Biomarker ID AI Benevolent 26#27Ulcerative Colitis Example: molecular-signature detection linked to outcomes Z Tm (SOW) m k=1...K n=1...N m=1...M xm nd d=1...D Transcriptomes organised into latent structure PATIENT-LEVEL DATA Dataset of IBD patient samples (N = 105 UC/Crohn's patient samples) Generative ML models Inflammatory marker response from suite of models Genes Patients PATIENT SAMPLES CLUSTERED CD56dim NK Pathway marker Y M1 macrophage 11 0.20- R = 0.62 FDR<0.001 10- 0.2- 0.15- 0.10- 0.1- 0.05- R=0.65 FDR<0.001 R=-0.74 FDR<0.001 0.0 0.00 ૐ neutrophil Target X 0.3- 0.2- • 0.1- R=0.63 FDR<0.001 6- R= -0.63 FDR<0.001 0.0 -90 -60 -90 -60 Latent 75 disease state CD Ctr UC Entity Explorer IBD Latent List (modified) Disease program: IBD Demo Data release: 1-27 (latest) How much do shortlisted entities overlap? Number of edges displayed: 1315 Structured data relationships High-confidence literature-derive ☐ ed relationships + Q Search Explore Visualise Shortlists Data-derived mechanism entities are integrated into the Knowledge Graph and connected to enriched biological pathways and relevant biomedical entities (e.g. diseases, tissues, targets) ML models recapture specific subgroups with key inflammatory markers (IL1 and TNFa signalling) and immune cells (M1 macrophages and neutrophils) and uncover mechanistic areas to explore further Al Benevolent 27#28The Benevolent Platform™ Chemistry#29My We combine deep expertise in Drug Discovery with innovative techniques in structure-based design, virtual screening and machine learning Highly experienced Drug Discovery team with a proven track record of taking nascent programme ideas and delivering drugs to the clinic. ✔ Chemoinformatic and Al tools impacting all stages of a drug programme from target selection through to candidate selection, by: Identifying differentiated opportunities through novel binding sites and prioritising previously undrugged targets for exploitation. ✓ Maximising the use of available data to derive new knowledge, at scale, for objective molecular design. ✔ Empowering chemists to design better drugs in fewer cycles - Candidate drugs delivered in as little as 2 years from programme inception, compared to a 3-5 year industry standard². HO OH 1 Based on BEN-8744 timelines - see Portfolio Overview deck November 2021 for more information? https://misciwriters.com/2019/11/15/from-the-lab-to-your-medicine-cabinet-a-timeline-of-drug-development/ ΑΙ Benevolent 29#30Molecular Design - Expertise in structure-based design, virtual screening and machine learning We combine deep expertise in drug discovery with innovative techniques in structure-based design, virtual screening and machine learning. Highly experienced drug discovery team with a proven track record of taking nascent programme ideas and delivering drugs to the clinic. Chemoinformatic and Al tools impacting all stages of a drug programme from target selection through to candidate selection. Empowering chemists to design better drugs in fewer cycles - candidate drugs delivered in as little as 2 years from programme inception compared to 3-5 year industry standard Druggability scoring to prioritise targets Target ID Binding site comparison to identify Hit matter and evaluate selectivity Proprietary pharmacophore | building methodology ML models of activity and ADMET endpoints Hit Identification Hit Expansion Binding site detection to identify differentiating chemistry opportunities Customisable virtual screening pipeline now on >10 billion compound scale Confidential Lead Optimisation Candidate Seeking Protein-Ligand interaction mining to surface protein-centric bioisosteres Programme visualisation Benevolent 30#31Project 'Y' - a Benevolent tech-augmented programme Novel target identified in Target ID No prior literature association with disease of interest One main chemotype reported in the literature Covered extensively by >20 patents Multiple closely-related cores also claimed - challenge to identify novel chemical space Close family members with known safety risks so selectivity important Hit ID & Hit Expansion completed in 7 months Employed both Virtual Screening and focused Fragment Screening approaches Project now in late Lead Optimisation 13 months, 380 compounds Low nM potency ● 200 fold selectivity over all family members Low metabolic clearance in microsomes and hepatocytes (Eh <0.4) Good aqueous solubility (>200uM) Clean in Ames, hERG and Cyp inhibition assays. Internal Company programme - target confidential, no prior literature detected by Benevolent Platform™ Lead Optimisation stats from internal Company experimental data. Al Benevolent 31#32The Benevolent Platform™ Apps, Infrastructure, Research#33Apps & Infrastructure Make the Benevolent Platform TM cheaper, easier to develop, easier to use, and more reliable for running deployments and collaborations'. Apps Engineering & UX Primary focus on development of the platform interfaces used by our drug discoverers Infrastructure Shared Engineering ✓ Site Reliability Engineering Security IT Works across the Delivery Areas Entity Explorer DNA repair Cross-organization delivery Accessible, reusable, and scalable infrastructure for training models and serving predictions Shared and accessible development and deployment platform for all Delivery Areas Common protocols that align product development with drug discovery process Based on internal Company statistics covering time taken to complete insilico steps (data build out, model running, triaging) Target List DNA repair-DNA repair 89 results Your ortena for age Sort by Sotly QFind target Filter by context DNA to G ONA STN1 St of CST comples PARP1 P RPA4 plication protein A4 。 em cell diffentiation GOPROCESSES needed signing#34Some of our recent publications Knowledge Wiatrak et al. (202`1) Zero-Shot Metric Learning Entity Linking submitted to Transactions of ACL. Shah and Fauqueur (2020) Learning Informative Representations of Biomedical Relations with Latent Variable Models, in EMNLP 2020, SustaiNLP Workshop. Wiatrak and Iso-Sipila (2020) Simple Hierarchical Multi-Task End-to-End Entity Linking for Biomedical Text in EMNLP 2020, LOUHI Workshop. Target Identification • Dunbar et al. (2021) Transforming drug target identification in CKD: a multidisciplinary artificial intelligence-based approach submitted to Nature Comms. Paliwal et al. (2020) Preclinical validation of therapeutic targets predicted by tensor factorization on heterogeneous graphs in Nature Scientific Reports. Myszczynska et al. (2020) Applications of machine learning to diagnosis and treatment of neurodegenerative diseases, in Nature Reviews Neurology. Precision Medicine Foster et al. (2021) Contrastive Mixture of Posteriors for Counterfactual Inference, Data Integration and Fairness submitted to NeurIPS 2021. Sim et al. (2021) Directed Graph Embeddings in Pseudo-Riemannian Manifolds in ICML 2021. ⚫. Schneider and Tomlinson (2020) Auxiliary task evaluations to learn meaningful representations from electronic health records in NeurIPS 2020, Workshop on Learning Meaningful Representations of Life. Chemistry Fabian et al. (2020) Molecular representation learning with language models and domain-relevant auxiliary tasks, in NeurIPS 2020, Machine Learning for Molecules Workshop. Simonovsky and Meyers (2020) DeeplyTough: Learning Structural Comparison of Protein Binding Sites, in Journal of Chemical Information and Modeling. Brown (2020) Artificial Intelligence in Drug Discovery, in Royal Society of Chemistry Richardson et al. (2020) Baricitinib as potential treatment for 2019-nCoV acute respiratory disease, in The Lancet Stebbing et al. (2020) COVID-19: combining antiviral and anti-inflammatory treatments, in The Lancet Infectious Diseases Benevolent 34 Al#35BenevolentAl Discovery Portfolio Overview 24#36Agenda Portfolio program examples illustrate the Benevolent PlatformTM in action Inflammatory Bowel Disease - IND-enabling program Amyotrophic Lateral Sclerosis - Candidate seeking program Glioblastoma multiforme - Lead optimisation program о NASH - Hit expansion program О NASH - Hit Identification program for currently undrugged target Atopic Dermatitis - Phase I/II clinical program Benevolent 36 ΑΙ#37Growing number of platform-generated programmes moving into clinical phases Disease Area Atopic Dermatitis (PanTrk inhibitor) Ulcerative Colitis (PDE10 Inhibitor) Amyotrophic Lateral Sclerosis Inflammatory Bowel Disease Glioblastoma Multiforme CNS Diseases Nonalcoholic Steatohepatitis (NASH) Oncology Antiviral Nonalcoholic Steatohepatitis (NASH) Oncology Oncology Chronic Kidney Disease Idiopathic Pulmonary Fibrosis 10+ Early Discovery Programmes (Multiple indications & targets in therapy areas such as Oncology, Immunology, CNS, GI, Metabolic Disorders and Others) Target ID Hit to Lead Lead Opt Preclinical Clinical Commercial AstraZeneca AstraZeneca Phase I start in early 2023 Highlights • All Pipeline assets generated from Benevolent Platform™M Broad therapy area coverage given disease agnostic approach Mix of Best in class, First in class and novel indications Potential for rapid scaling and expansion into new modalities Existing pipeline alone addresses prevalent patient base* of >263m (1) and current market opportunity >$30bn (2) Source: (1) GlobalData, Epidemiology forecasts 2021, Atopic Dermatitis (7MM), IBD (8MM), ALS (8MM), GBM (7MM), NASH (7MM), CKD (7MM), IPF (7MM); 7MM = 7 major markets (US, JP, EU5); 8MM = US, JP, EU5+ Canada; (2) Evaluate Pharma, Current Worldwide Market Size (data pull 22nd Sept 2021) Atopic Dermatitis, IBD, ALS, GBM, NASH, CKD, IPF ΑΙ Benevolent 37#38Inflammatory Bowel Disease (IBD) Ulcerative Colitis (UC) and Crohn's Disease (CD)#39Both the Ulcerative Colitis and Crohn's Disease markets are large, and expected to experience sizable growth USD Millions USD Millions Ulcerative Colitis Market Size, 2019-2029, 7MM(1) 14000 12000 10000 8000 6000 4000 2000 0 6.1% CAGR 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 Crohn's Disease Market Size, 2019-2029, 7MM (2) 14000 12000 10000 8000 6000 4000 5.6% CAGR • Taken together, the 2019 UC and CD market sizes were valued at approx. $14B across the 7 major markets, expected to increase to approx $24B in total by 2029, growing at a ~6% CAGR Growth in the UC and CD markets is driven by: Improved diagnosis and increasing prevalence O Approval of numerous pipeline drugs (both small molecules and biologics) High treatment rates High unmet need for safe & efficacious therapies • Despite a competitive pipeline, there is opportunity to differentiate BEN- 8744 as an oral small molecule with a novel MoA based on safety, efficacy and through pursuing a precision medicine approach 2000 0 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 Source: (1) Global Data, Ulcerative Colitis: Global Drug Forecast and Market Analysis to 2029: (2) GlobalData, Crohn's Disease: Global Drug Forecast and Market Analysis to 2029. 7MM (7 Major Markets): EU5, US, Japan AI Benevolent 39#40BEN-8744: IBD Asset Overview Best-in-class, oral, peripherally restricted potent and selective Phosphodiesterase 10 (PDE10) inhibitor for the treatment of Moderate to Severe Ulcerative Colitis and Crohn's Disease (IBD) Mechanism of Action Formulation Competitive Advantage IP Position Current Status Asset Overview Phosphodiesterase 10 (PDE10) inhibitor, immunomodulatory Small molecule, oral, chronic treatment New immunomodulatory mechanism of action, providing additional option to patients refractory to current treatments Disease modifying treatment targeting the disease mechanisms associated with IBD Targeting induction and maintenance of clinical remission, allowing for reduction in long-term corticosteroid use Peripherally restricted, devoid of on-target central side effects Clean safety profile compared to other oral small molecule competitors (e.g. JAK inhibitors, tofacitinib/Xeljanz - black box warning for infection risk) Precision Medicine approach for patient stratification Second Medical Use & Composition of Matter patent applications filed Commenced preclinical development, with Candidate selection completed August 2021 Benevolent 40 ΑΙ#41PDE10 inhibition: potential to normalise dysregulated mechanisms in IBD Th1 IFNY Dendritic Cell TNF-KB- IL-12 IL-6 Macrophage IL-18 M1 M2 IL-1ẞ IL-23 TNF PGE2 TNFOX IL-2 ILC3 ↑IL-17A IL-22 Th17 ↑ IL-17A IL-17F IL-21 IL-22 IL-23 • Reduced inflammatory cytokine release from intestinal epithelia via ↓ NFkb (1) • Reduced tissue-resident macrophage activation Reduced intestinal inflammation Source: (1) doi:10.1371/journal.pone.0079180: (2) doi:10.1371/journal.pone.0016139; (3) doi:10.3109/00365521.2015.1038849 Na Nutrients PDE10 inhibition ↑ CGMP ↑ CAMP fluid claudins PAMR ZO-1 occludin MLCK MLCK Nutrients TNF • Improved TJ assembly via PKG/PKA-mediated pMLC (2) • Improved fluid/mucus homeostasis. via PKG phosphorylation of intestinal CFTR (3) Improved barrier integrity Benevolent 41 Al#42BEN-8744 is a highly potent and selective PDE10 inhibitor Primary Pharmacology¹ Safety Pharmacology' Human PDE10 IC50 (biochem <InM HERG IC50 13uM & cell-based assays) CaV1.2 IC50 >30uM Mouse PDE10 IC50 <InM Nav1.5 Ic50 >30uM IC50 in IBD biopsy <InM inflammatory cytokine Human iPSC-derived No flags release assay (UC & Crohn's) cardiomyocyte toxicity Selectivity vs other PDE ≥1000 fold family members Selectivity vs broad panel of safety targets (Cerep87) ≥1000 fold Highly potent Good selectivity No safety flags Ames (5 strain, +/- S9) IVMN Hepatic toxicity (hepatocyte cytotoxicity; DILI panel, 14 day 3D liver organoid assay) 7 day rat toxicology study Negative Negative Negative No overt toxicity or clinical observations >100x predicted hAUC Al Benevolent 42 1 Company internal drug programme data#43BEN-8744: Potent activity demonstrated in both UC & CD patient ex-vivo colon biopsies² Individual donor response: UC and CD biopsies Ulcerative colitis donor 1500- 15000- Demonstrated inhibition of proinflammatory cytokine release (IL-6/IL-8) from individual UC and CD patient biopsy samples, comparable to corticosteroids Indicative of a robust anti-inflammatory response with BEN-8744 BEN-8744 has now progressed into preclinical development and a Clinical Trial Application (CTA) is scheduled Q4 2022 ⚫ First-in-Human (SAD/MAD) clinical studies will commence early 2023 ⚫ Subsequently supporting a Ph2a clinical study in Ulcerative Colitis, together with a follow-on clinical study in Crohn's Disease 1SAD/MAD- Single Ascending Dose, Multiple Ascending Dose 2 Company internal drug programme data IL6 pg/ml 0- 1000- 500- DMSO 0.1% Prednisolone 1uM BEN-8744 0.01uM BEN-8744 0.1uM Crohn's disease donor IL6 pg/ml 2500- 2000- 1500- 1000- 500- 0- DMSO 0.1% Prednisolone 1uM BEN-8744 0.01uM BEN-8744 0.1uM IL-6 BEN-8744 1UM IL-6 BEN-8744 1uM IL8 pg/ml IL8 pg/ml IL-8 10000 5000- DMSO 0.1%" Prednisolone 1uM BEN-8744 0.01uM BEN-8744 0.1uM BEN-8744 1uM 100000- IL-8 50000- 0. DMSO 0.1% Prednisolone 1uM BEN-8744 0.01uM ///// BEN-8744 0.1uM BEN-8744 1uM Benevolent 43#44Amyotrophic Lateral Sclerosis (ALS)#45Amyotrophic Lateral Sclerosis (ALS) Affects 0.02% US population over age 40 years(1), ~75 thousand patients in 8MM(2), forecast $1.04bn market by 2029(3) • ALS is a rare and devastating fatal neurodegenerative disease in which the motor neurons degenerate or die, and stop sending messages to the muscles. Fewer than 50% of patients survive 30 months from symptom on-set5 • Efficacy and Safety - Current treatments (riluzole and radicava) are largely ineffective and only extend patient survival by ~6 months. Patients are largely treated with oral riluzole, however radicava is an intense intravenous treatment placing significant burden on patient quality of life • New, safe and effective disease-modifying therapies are urgently needed 10% Familial ALS 90% Sporadic BEN-9160: CNS-Penetrant c-Abl inhibitor for the Treatment of Sporadic and Familial ALS Subtypes, with potential to expand to Parkinson's Disease Deployment of the Benevolent PlatformTM led to the discovery of c-Abl - a target with demonstrated capacity to modulate pathways critical to ALS pathology Our Molecular Design expertise resulted in a potent, brain-penetrant small molecule c-Abl inhibitor BEN-9160 with a pharmacokinetic profile enabling significant target engagement BEN-9160 is expected to provide an efficacious oral treatment for ALS, targeting key disease-relevant mechanisms Disease modifying treatment for the benefit of both the Sporadic and Familial ALS patient populations Delay of disease progression with extension of life significantly better than Standard of Care Clean safety profile (no relevant drug-drug interactions, hepatotoxicity, CV liability, CNS effects on memory or cognition or myelosuppresion) Source (1) GlobalData: Amyotrophic Lateral Sclerosis: Epidemiology Forecast to 2029; (2) and (3) GlobalData,Amyotrophic Lateral Sclerosis (ALS): Opportunity Analysis and Forecasts to 2029; (4) Company internal drug programme data; (5) https://www.sciencedirect.com/science/article/pii/S0140673610611567 https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(10)61156-7/fulltext Benevolent 45#46ALS: Hypothesis Generation with a focus on key mechanisms and Precision Medicine based approaches 1. Targets associated with key mechanisms in neurodegeneration identified using knowledge graph based relational inference models using a “Fleet” of algorithmic models о Key mechanisms: Autophagy, mitochondrial health, proteosome function, and lysosomal function Unstructured data Target Triage TARGET ALS Hypotheses prioritised based on relevance to ALS, neurodegeneration mechanisms, druggability 1 2 3 4 00000 2. Omics/Precision medicine based approach using Target ALS collaboration dataset 612 RNA-Seq samples from 149 individuals identify subgroups of ALS patients (endotypes) with distinct molecular disease mechanisms/targets 830 10 0.00 904 02 0.00 Reported day of symptom onset minus predicted day at which ALSFRS total score should be 48 post symptom onset normal Target Safety Drugs? Metadata Score Tgt1 Safe Partial OPC, GCNN 0.71 120 Structured data Tgt2 Unsafe No Graph Inference SQT 0.65 100 Tgt3 Safe Yes OPC, CROM 0.51 Tgt4 Phase II Offtgts OPC 0.49 Benevolent 46 Al#47Multiple assays used for target validation in ALS, similar approaches being employed for Parkinson's disease 1. Autophagy assay in Hela cells 2. Glutamate induced excitotoxicity in sporadic ALS patient iPSC MNs Glutamate induced excitotoxicity assay (Annexin V ALS Motor Neurons) TR-FRET between LC3-Tb and LC3-D2 or p62- D2 and p62-Tb LC31 is found in the cytoplasm and is conjugated to the lipid phosphatidyl-ethanolamine to form LC311 LC311 binds to the autophagic vacuoles' membrane marker of autophagic vacuoles • SQSTM1/p62 -selective autophagy receptor -sequesters ubiquitinated proteins into AVS by interacting with LC3 - p62 is a substrate for autophagic degradation → its degradation can be used as a marker of autophagic clearance PAGE 2 Astrocytes 3. SITraN ALS iAstrocyte mouse motor neuron co-culture assay Lactate efflux transporters tp75 NGF receptor The University Sheffield Pathophysiologically Relevant in vitro Model of ALS SITraN Human Astrocytes HD9-GFP+ mouse motor neurons % motor i Con155 Con3050 ConAG Pat 78 Pat183 Pat201 Pat009 Pat 12 iAstrocyte line Pat17 Control C9ORF72-ALS Sporadic ALS ALS Patient Derived iAstrocytes are Toxic to Motor Neurons in Co-Culture Impaired axonal transport ALS mechanisms Reduced lactate release Motor neuron Apoptosis Mitochondrial dysfunction NO BAX Glutamate transport Excitotoxicity Ca ROS PGE2 Altered RNA processing Complement activation C1q Autophagy Pro-NGF PGE2 NO Inflammation MCP-1 ER stress Distal axonopathy Microglia Misfolded proteins BEN-3765 Time() + Glbra 10 M 1.M 1000- o Sam Co Annexin V Oy ANOVA The ALS PSC motor neurons were recovered and seeded onto a Matrigel-coated 96-well plate at 15.000/well for 4 days On Day 0, the were treated with vehicle controls or smal molecule at afferent doses for 7 days, and then treated with bet Glutamate for 3 days. The Annexin V red dye (Butun were added to visualize the apoptotic cells using IncuCyte 53 software (red object count 5 images in one wel of each condition were analysed. The average of the Total Red Object Area Well was shown Left panell Dynamic data (Right pane Dyda iXCells BIOTECHNOLOGIES Proteasome impairment 4. Tunicamycin induced motor neuron toxicity Controll Neurite Length pic 6.7 In-house hiPSC- MN assay (% Control) DIC 7.0 -10 -10 · Log [Tunic amycin] (M) Log [Tunic amycin] (M) CYTOPLAN Tunicamycin 1 96h 144 h Perturbation of ER homeostasis Accumulation of RES PERK ATF6 Unfolded Protein Response (UPR) Unfolded Benevolent 47#48Abl target hypothesis has been validated using complex cell-based systems The SITraN assay: a humanised cell model of ALS Working in collaboration with academic experts at SITraN who have developed a patient cell derived assay system ¡Astrocytes (via induced progenitors) obtained from genetically-diverse ALS patients cause motor neuron toxicity in a co-culture assay Can test biological hypotheses by measuring the rescue of motor neuron survival Benevolent proprietary CNS penetrant Abl inhibitors are neuroprotective in this assay Induced Patient Fibroblasts Neuronal Progenitor Cell Induced Astrocytes Motor Neurons uron survival % motor neuron -50 150- 100- 50- Healthy donor ¡Astro/DMSO C9orf72 Patient iAstro/DMSO C9orf72 Patient iAstro/10uM Riluzole C9orf72 Patient iAstro/10uM Gefitinib Non-CNS penetrant TK inhibitor C9orf72 Patient iAstro/1uM BEN-A BAI CNS penetrant Abl inhibitor Benevolent 48#49250 200 150 aSyn area/TH neurons (% of control, TH) 100 50 Control MPP+ 4UM/48h 100- 80- TH number (% of control) 60- 40- 20- Abl inhibitor is neuroprotective in in vitro Parkinson's MMP+ induced TH neuron toxicity and a-synuclein aggregation in mouse primary dopaminergic neuron cultures' 120 disease and ALS models Tunicamycin induced apoptosis of sporadic ALS patient iPSC derived motor neurons² 8.0x 105- Glutamate induced apoptosis of sporadic ALS patient iPSC derived motor neurons² 8.0x 10- Control Control MPP+ 4μM/48h (10 nM) (30 nM) (100 nM)' (300 nM) (1 μM) (3 µM) (10 µM) Ben-B (10 nM) (30 nM) (100 nM) (300 nM) (1 μM) (3 μM) (10 μM) + BDNF (50 ng/mL) + BDNF (50 ng/mL) Total Red Object Area (μm²/Well) 6.0x 105- 4.0x 105- 2.0x 105- 8.0x 105- 6.0x105- 0 T Total Red Object Area (μm²/Well) 4.0x 105- 2.0x105- Ben-B 'Data from Neurosys. 2Data from iXcell (both CROS are engaged by the Company to provide services for this programme) No Stim Ctrl Tunicamycin 10 μM 50 100 150 200 250 Time (h) *** ** Ben-B 1 μM 0.1 M 0.01 μM No Stim Ctrl Tunicamycin -10 μM 1 μM -0.1 μM -0.01 μM Total Red Object Area (μm²/Well) 6.0x 105- 4.0x 105- 2.0x 10- 8.0×105- 6.0x105- Total Red Object Area (μm²/Well) 4.0×105- 2.0x105- 0+ 0 50 100 150 200 250 Time (h) No Stim Ctrl Glutamate 10 μM *** *** **** T 1 μM > 0.1μM 0.01μM Ben-B Benevolent 49 Glutamate 10 μM 1 μM 0.1μM 0.01 μM#50Current Status: Programme in candidate seeking with candidate selection due to complete Q4 2021 Efficacy' Demonstrated with humanised ALS cellular models using motor neurons and motor neuron astrocyte co-cultures Multi-donor screens in these cell types currently ongoing, to provide evidence of effects in sporadic and defined genetic ALS subtypes (including TDP-43 mutants) In vivo efficacy models initiated Safety² In-vitro toxicity and safety assays showing better or equivalent safety to clinical comparators Panel of cellular toxicity screen including ✓ cardiomyocytes, hepatocytes, kidney and HUVEC cells complete In-vivo rat tox study (7 days) complete DMPK Demonstrable target engagement in the mouse CNS PK/PD using target engagement biomarkers in mouse CNS Comprehensive human dose predictions Dose projections commensurate with BID dosing Cyp inhibition/ induction/TDI IP/Chemistry Composition of matter patent applications filed . At the end of candidate seeking (Q4 2021), we will have completed our efficacy, safety/in-vivo toxicity and DMPK studies for BAI-5002 • Planning for preclinical and future clinical studies now ongoing Internal Company drug programme data using SITraN assay. 2 Combined internal Company drug programme data and CRO specific safety assays Benevolent 50 Al#51Glioblastoma Multiforme (GBM)#52Glioblastoma Multiforme (GBM) One of the most lethal and aggressive brain tumours • Extremely poor prognosis and high unmet need Prevalence • Incidence of GBM ranges from 0.59 to 5 per 100,000 (1) • Mean age at presentation 53y, 5 year survival rate - 5% Standard of Care (SoC) Surgery, Radio- & Chemotherapy, Temozolomide (TMZ) • Current therapy rarely curative Glioblastoma Stem Cells - key component • Self renewal Resistant to radio & Chemotherapy Highly infiltrative and heterogeneous MES-Aggressive; Poor survival о PN- Favourable outcome CL- Best response to therapy Reasons why GBM has high unmet need • Tumour intrinsic о 。 Glioblastoma Stem Cells (GSC) ⚫ High level of Tumour heterogeneity О 。 Tumour Micro environment (TME) О ⚫ Rapid evolution of the tumour and its transition into aggressive phenotype KOL most cited reason for therapy failure Lack of effective BBB penetrant molecules Source (1) 1- Grech et al. 2020, DOI: 10.7759/cureus.8195#53GBM: Hypothesis Generation and Validation A therapeutic target which functions as a radiosensitiser identified for Glioblastoma (GBM) using knowledge-graph-based relational inference models Benevolent Knowledge Graph enriched and customised to identify targets modulating viability of Glioblastoma Stem Cells (GSCs) or radiosensitisers Predictions enriched with disease relevance by use of Patient datasets (combination of 'Omics platforms) Target ID Entity selection and data build out around GBM stem cells (GSC) and radiosensitisers. Predictions for GBM using relation inference models on the Benevolent Knowledge Graph Target Triage Hypotheses prioritised based on relevance to GSC modulation, suitable safety profile, "druggability" 'Omics Target expression GBM vs normal brain tissue (Patient dataset; Single Seq), mapped across diverse pathways & mechanisms for GSC Target selection - Novel MoA for GBM - Expression in GBM tumours - Subtype preference - Radiosensitiser Unstructured data 1 2 3 4 Tgt11 ☐ Tgt12 Target Safety Tgt1 Safe Drugs? Metadata Score Partial OPC, GCNN 0.71 Tgt2 Unsafe No SQT 0.65 Structured data Graph Inference Tgt3 Safe Yes OPC, CROM 0.51 Tgt4 Phase II Offtgts OPC 0.49 'Omics Patient Stratification Experimental testing སྱཱ ཎྜ སྠཽ ༔ སྠཽ 206 1.2 0 Gy 3 Gy 1.0 0.8 IC 0.2 0.0 P 0.01 0.1 Concentration (M) Al Benevolent 53#54Target R identified as a therapeutic target for GBM Target R was predicted by the Benevolent Platform TM as a potential therapeutic target to: Modulate viability of glioblastoma stem cells (GSC) Sensitise with radiotherapy (Radiosensitiser) Data from the "stem cell enriched" neurosphere assay indicated that Target R had 'on-target', single agent activity across majority of GBM patient cells but was less sensitive to MES GPCs Data from the 3D clonogenic assay indicated that Target R had both 'on-target', single agent activity and sensitised with ionising radiation Target R % Viability PN-11 PN-20 MES-10 - 60- 40- MES-25 PN-31 CL-12 20- MES-42 MES-40 -4 -3 -2 -1 0 Log[2933] μM -PN-45 Surviving fraction 1.2 0 Gy 3 Gy 1.0 0.8 0.6 0.4 Target R IC50- 0.2 OGY 0.246 (0.201 to 0.291) 0.0 3Gy 0.104 (0.082 to 0.125) p=0.00434 0.01 0.1 10 Concentration (μM) ΑΙ Benevolent 54#55BAI-5028: Target Product Profile Product Properties Mechanism of action Modality Primary indication Patient population Treatment route/duration Target Efficacy Differentiation from other Target R inhibitors in development Other factors for differentiation Profile CNS penetrate, potent and selective inhibitor with Single Agent & Radiosensitising activity Small molecule Glioblastoma Multiforme (GBM) Newly Diagnosed (MGMT methylated and MGMT unmethylated); Recurrent if Radiotherapy (RT) approved Oral, concurrent with RT For recurrent GBM patients: improvement in PFS at 6 months to >40% (based on RANO criteria) compared to 25% typical for SOC. For unmethylated GBM patients at first diagnosis: improvement in overall survival of > 3 months None in development for GBM or other neurological indications CNS penetrant Potential for use in combination with RT and Chemotherapy (CT) in other cancer patients: cancers where RT is established treatment (examples such as lung, breast, head and neck cancers). AI Benevolent 55#56Target R: Chemistry progressed to deliver a potent, selective and highly brain penetrant molecule AIChemOpt Prep & Hit ID / Expansion AIChemOpt Lead Optimisation Feb 2020 9m Nov 2020 12m Nov 2021 Candidate Seeking May 2022 6m Hit ID BEN-9677 Hit-to-Lead BEN-11156 Lead Optimisation Lead compound(s) PIC50 (LLE) Pfizer CNS MPO 6.1 (5.2) 5.4 PIC50 (LLE) Kinases IC50 <1 μM Cerep >50% @ 1μM hERG IC50 (µM) 7.9 (6.7) 6/372 1/87 >30 PIC50 (LLE) Selectivity Efflux ratios ADME >9 (>8.0) Excellent Low / none Good Efflux ratio MDR1/BCRP Крии (rat) 1.7/3.7 0.44 Brain-penetrant Radiosensitiser Yes Yes Novel kinase hinge binder Retained key motif for selectivity From BAI virtual screen Patent ID and data extraction Controlled substance checker Oral bioavailability (%) 51 On-/off-target docking models Reaction enumeration Predictive model suite • о Activity, metabolism, efflux Generative chemistry Advanced profiling underway Upcoming milestone: GBM xenograft TE study Potent backup series with low efflux also in development Our lead series represents a potent, selective and highly brain-penetrant ‘Target R' kinase inhibitor Benevolent 56 Al#57BAI-5028 is approaching Candidate seeking 2019 2020 2021 2022 2023 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Target validation Preparation/Hit ID / Hit expansion Lead Optimisation Advanced Lead selection Candidate profiling Candidate nomination Preclinical Key programme points: Therapeutic target for GBM which functions as a radiosensitiser identified using Benevolent Knowledge Graph-based relational inference models An attractive tech-derived virtual screening hit led to a potent, selective, and highly brain-penetrant series of "Target R" kinase inhibitors "Target R❞ sensitive GBM patient cohort identified using our Precision Medicine workflows for patient stratification Program on track to deliver within company timelines Benevolent 57 ΑΙ#58Non-Alcoholic Steatohepatitis (NASH)#59Non-Alcoholic SteatoHepatitis (NASH) Affects 11% US population (1), 63 million patients in 7MM (2), forecast $27.2bn market by 2029(3) Non-alcoholic fatty liver disease (NAFLD) is a metabolic disorder characterised by accumulation of fatty deposits in the liver; non-alcoholic steatohepatitis (NASH) occurs when NAFLD progresses, and is associated with liver inflammation, fatty deposits, and fibrosis High unmet need - NAFLD and NASH pose a high economic burden, driven by costs to provide chronic care for patients (including liver transplants) in the absence of any disease modifying therapy; currently hard to detect meaning NASH is often diagnosed later in the disease course 30% 27% Fibrosis Staging 0-1 Fibrosis Staging 2 NASH 25% 17% Fibrosis Staging 3 USA, Japan, EU5 (4) Fibrosis Staging 4 (Cirrhosis) NASH Fibrosis Staging (0-4), percentage of diagnosed Prevalent Cases of NASH, ≥18 Years, BAI-5030: (Target A) Best-in-class, potent and selective drug for the treatment of NASH. (Target B) First-in-class programme • Targets A and B were identified by our TargetID platform as entirely novel targets for the treatment of NASH - representing mechanistically distinct approaches • We are currently applying our Hit Identification (Target B) and Hit expansion (Target A) capabilities to support the identification of potent and selective inhibitors • BAI-5030 is expected to provide efficacious, mechanistically differentiated, disease modifying . treatments for NASH, with the potential to reduce fibrosis in both early and late stages of disease BAI-5030 will target fibrosis (and potentially steatosis) in NASH, meeting the unmet need for patients including: О Lack of currently approved therapies High mortality and prevalence of NASH Severe disease progression in absence of disease modifying therapy, including development of cirrhosis and hepatocellular carcinoma Source (1), (2), (3) and (4):GlobalData, Non-Alcoholic Steatohepatitis (NASH): Opportunity Analysis and Forecasts to 2029 Benevolent 59 Al#60Prediction strategy geared for identification of fibrotic regulators driven through oxidative stress Target identification and hypothesis validation strategies were aligned to identify targets that could impact: • Fibrosis (hepatic stellate cell activation assay) ⚫ Steatosis (hepatocyte lipid accumulation assay) Models were trained using datasets focussed on: • NAFLD disease biology • Oxidative stress mechanism 1 Define biological question and assays 2 Transcriptomics data selection 3 Models run Normal liver First hit Insulin resistance ↓ẞ-oxidation ↑ De novo fatty acid synthesis 4 Target Triage NAFLD NASH Fibrosis/Cirrhosis 5 Second hit Hypothesis validation ↑ Oxidative stress Necroinflammation and inflammation Targets A & B ΑΙ Benevolent 60#61Inhibition of Target A has antifibrotic effect in TGF-☐ activated primary human hepatic stellate cells Fibroblast to myofibroblast transition aSMA 150 Run 1 Mean +/- SD Collagen a-SMA/Col 1 mean cellular intensity TGF- HSCs, TGFb, 72h 1750- 1250 1000 1500- 1250- 1000- 750- 500- - 750 - 500 250 250- 0 0 0.01 0.1 1 10 TGF+ [ng/ml] Source: Company internal drug programme O Cells per well % inhibition % inhibition 100- 50- -50- -8 -7 -6 -5 Log Compound [M] 150 Run 2 Mean +/- SD 100 50- Nuceli Count α-SMA → Collagen I Tool compound Target A SPR KD 17nM Cell-based TE assay IC 50 5.4nM WLP >30-fold selectivity Kinome profiling >200-fold selectivity No effect on cell viability a-SMA Collagen 1 Nuclei Count RUN 1 (IC50) 9.5nM 7.OnM 2.3μM D -8 -7 -6 -5 RUN 2 9.6nM 5.3nM -50 Log Compound [M] (IC50) 1.4μM Benevolent 61 Al#62Target A: Scheduled to transition into Lead Op 4Q21 2020 2021 2022 2023 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Target validation Preparation / Hit ID / Hit expansion Lead Optimisation Advanced Lead selection Candidate profiling Candidate nomination Key programme points: ⚫ Best-in-class programme with no clear literature associating Target A as a potential therapeutic option in NASH Compelling anti-fibrotic In vitro target validation data packing to support utility in treatment of NASH • Rapid progress to identify inhibitors of Target A: biophysical screening data and co-crystal structure in-hand to enable identification of novel chemical matter and transition to Lead Optimisation Benevolent 62 ΑΙ#63aSMA readout normalised to shSCR Target B - First in class programme targeting a previously undrugged target for the treatment of NASH • Structure enabled hit identification strategy commenced 3Q21 Al-enabled literature analysis was able to identify data showing target B is reported as being upregulated in NASH, with expression correlating to the degree of fibrosis in the liver • Mechanistically diverse from target A, building depth to our NASH portfolio 150 aSMA Collagen I readout normalised to shSCR 0 SCR shRNA + TGF-B1 hACTA2 shRNA Target shRNA #1 One way ANOVA with Dunnetts 100- 50- **** **** 2012 Target shRNA #2 Target shRNA #3 4 **** 150- 100- Collagen I 50- **** **** **** SCR shRNA + TGF-ẞ1 hTGFBR2 shRNA Target shRNA #1 Target shRNA #2 Target shRNA #3 **** Cell number normalised to shSCR 200- NS 150- 100- 50- I H * Cell count NS SCR shRNA + TGF-B1 hACTA2 shRNA hTGFBR2 shRNA * Target shRNA #1 Target shRNA #2 Target shRNA #3 SI NS TGF-activated immortalized human hepatic stellate cells ⚫ Fibrosis assay established in immortalised hepatic stellate cells • Lentiviral delivery of shRNA constructs to knock down target expression Three separate shRNA constructs were run independently for the target • Full target-story data package complete Al Benevolent 63#64Atopic Dermatitis#65Atopic Dermatitis (AD) • Atopic dermatitis is the most common chronic inflammatory skin disease, characterized by intensely itchy, red, and swollen skin o Affects 10-20% of children and up to 3% of adults (1) o Approximately 60-70% of all cases present with mild-moderate disease severity (2) ○ Prevalence is rising, with market value in 7MM forecast to exceed $14 billion by expected launch of BEN-2293 in 2028 (1) • Skin inflammation and chronic pruritus associated with atopic dermatitis negatively impact quality of life and psychosocial well-being Clear unmet need in mild to moderate patient segment for treatment addressing itch and inflammation, without side effects of steroids BEN-2293: A potent PanTrk antagonist developed to relieve inflammation and provide rapid itch resolution in patients with AD • BEN-2293 is a PanTrk inhibitor targeting TrkA, B and C receptors. The Trk receptors were identified as part of an effort to find mediators of both itch and inflammation in AD. Using our Molecular Design expertise we were able to design a PanTrk inhibitor, equipotent against the 3 receptors BEN-2293 is expected to treat atopic dermatitis by: Inhibiting itch signaling and blocking nerve sensitization (TrkA) in addition to inhibiting Thl and Th2-mediated dermal inflammation (TrkB, TrkC) BEN-2293 will target Mild, Moderate and Severe Atopic Dermatitis patients, addressing unmet need in the treatment of mild to moderate Atopic Dermatitis as a steroid sparing alternative and in more severe patients undergoing treatment with biologics (e.g. dupilumab) that require add-on treatment Source: (1) GlobalData - Atopic Dermatitis: Global Drug Forecast and Market Analysis to 2027; Evaluate Pharma - Eczema/Dermatitis: Worldwide Sales 2026 (2) GlobalData - Atopic Dermatitis: Epidemiology Forecast to 2027 Benevolent 65 ΑΙ HPA-Hypothalamus, Pituitary, Adrenal#66Atopic Dermatitis - BEN-2293, pan-Trk inhibition rationale Healthy skin Non-lesional skin Acute lesional stage Chronic lesional stage Lichenification corneum Skin microbiota Staphylococcus aureus. Barrier dysfunction, innate immune system activation and T2-driven inflammation and/or T 22-driven inflammation Keratinocyte Variable T1 and T17 activation Allergen TrkC • NT3/TrkC potentiates stimulated Th2 T-cell inflammatory responses and synergistically enhances T-cell receptor induced IL-4 production by Th2 cells • Mast cells within AD skin lesions express high levels of NT3 compared to normal controls Stratum Stratum basale TrkB • AD Skin-resident eosinophils express elevated levels of TrkB (together with TrkA and C) and functionally respond to BDNF • BDNF/TrkB inhibit eosinophil • apoptosis and increase chemotactic index Scatum gralosum Stratum spinosum IL-1ẞ IL-33 TARC IL-25 MDC TSLP FCER1 IL-33 TSLP Dermis Blood vessel IL-4 IL-13 OX40L Eosinophil CLA CCR10 H4R CCR4 CRTH2 IL-4 IL-13 IgE IL-31 -Cutaneous sensory neuron ILC2 B cell T cell T2 T22 T1 T17 Trm cell cell cell cell cell cell Langerhans cell O Dermal dendritic cell IDEC TrkA • TrkA levels in skin dramatically increase in response to inflammatory stimuli NGF produced by AD keratinocytes, is a major mediator of cutaneous hyperinnervation • Increased NGF in the skin sensitizes primary afferents contributing to peripheral itch sensitization and chronic pruritus Involved in the inflammatory activation of mast cells and basophils ΑΙ Benevolent 66#67BEN-2293: Excellent skin penetration • Experimental evidence supports high exposure in human skin at >IC90 free, and low exposure in blood with proposed clinical 1% ointment strength. 1% BEN-2293 ointment BID exceeds the exposure needed for PanTrk inhibition in both epidermis/upper dermis and lower dermis even at IC 901 Human in vitro >>IC90 V Healthy skin Non-lesional skin Acute lesional stage Chronic lesional stage Lichenification corneum Skin microbiota Staphylococcus aureus. Barrier dysfunction, innate immune system activation and T 2-driven inflammation and/or T 22-driven inflammation Keratinocyte Variable T1 and T17 activation Stratum Stratum granulosum IL-1ẞ IL-33 TARC IL-25 MDC TSLP Stratum spinosum 000000 Stratum basale Allergen FCER1 IL-33 TSLP Minipig in vitro >> IC90 V Epidermis and upper dermis Minipig in vivo >IC90 V Minipig in vivo Lower Dermis Dermis V Blood Blood vessel Free plasma levels <<400 below IC50 IL-5 IL-13 OX40L IL-4 IL-13 IL-4 IL-13 IgE IL-31 -Cutaneous sensory neuron Eosinophil CLA CCR10 H4R CCR4 CRTH2 ILC2 T2 T22 T1 T 17 Trm B cell T cell Langerhans cell cell cell cell cell cell cell Dermal dendritic cell IDEC CTA-enabling 28d Tox Package: Rat (IV) and Mini-pig (topical) = Safety margins > 20 fold for AUC and > 269 fold for Cmax to dose limited NOAELS 1 (internal Company drug programme data) Benevolent 67 Al#68BEN-2293 is progressing in an adaptive Phase I/II clinical study, with full data expected in mid 2022 Part A Part B 2 3 4 First in Human Dose Escalation 3/4 cohorts completed, data expected late 2021 8 Mild-Moderate AD patients (18-65 years) per cohort, randomised 3:1 BEN-2293: Placebo Safety, Tolerability, PK Adaptive ascending dose cohort design Includes efficacy endpoints MALDI imaging (evaluate human skin PK) Review Late 2021 Part A efficacy readout variability and response. Statistical modelling Finalise Part B design Efficacy Cohort(s) Full data expected by middle of 2022 30-45 Mild-Moderate AD patients (18-65 years) per arm, final design and sample size dependent on Part A outcome Efficacy Outcome measures include itch (NRS) and inflammation (VIGA, EASI) Additional safety, tolerability and PK Biomarker panel (reflects PanTrk mechanism and AD effect) Our intention is to out-licence development and commercialisation of BEN-2293 following completion of this trial, with good interest from key Big Pharma and Dermatology specialists as potential partners Al Benevolent 68#69Given the limitations of current available topical therapies, there is a large unmet need for an efficacious and safe alternative topical therapy for the treatment of patients with Atopic Dermatitis TREATMENT FLOW Atopic Dermatitis Treatment Paradigm Mild Steroid Treatment Moderate Calcineurin inhibitors Eucrisa/PDE4 inhibitors Topical JAKS BEN-2293 (PanTrk) PanTrk positioning Severe Key Insights: • Topical Corticosteroids (with increasing potency) - Poor side effect profile and concern of use by patients Calcineurin inhibitors (pimecrolimus or tacrolimus) - Poor side effect profile with associated black box warning PDE4 inhibitors - Issues with site application irritancy Immunosuppressants (azathioprine, ciclosporin and methotrexate) - Poor side effect profile Topical JAKS - ruxolitinib recently approved in the US, not yet approved in EU, but approval issued with a black box warning Anti IL13/IL4 mAbs (Dupilumab) - high cost treatment only indicated in moderate-severe. Need for better treatment solutions Current Future Source: GlobalData and Company internal drug programmes Immuno- suppressants Dupilumab Other anti-IL13/IL4 mAbs BEN-2293 (PanTrk) - Combined solution addressing itch, inflammation and potential disease modifying effects, together with an improved safety profile and no irritancy on application Potential to displace ineffective 2nd line treatment for chronic use in adults and paediatrics Potential use in a subset of 1st line patients where rapid itch resolution is key Potential for use in the severe patient population as an adjunct treatment option ○ Oral JAKS O Benevolent 69 ΑΙ#70BenevolentAl Drug Programmes Our advanced in house pipeline desery and unique approach in generating drug prog Disease N Targ BenevolentAl Pipeline Market Opportunity derosis Bowel Disease Chemistry & Land Optimisation lioblast forme Oncology Antiviral Nonalcoholic Steatohepatitis NASH#71BenevolentAl pipeline assets target treatment of prevalent diseases with high unmet need BEN-2293 for the treatment of Atopic Dermatitis BEN-2293 is a first-in-class, topical, steroid sparing PanTrk antagonist to address itch and inflammation in mild-moderate atopic dermatitis Illustrative target patient population in 2020 All atopic dermatitis patients¹ Patients with mild-moderate disease¹ Treatable population BEN-8744 for the treatment of Ulcerative Colitis BEN-8744 is an oral, peripherally restricted, potent and selective PDE10 inhibitor for the treatment of moderate-to-severe Ulcerative Colitis Illustrative target patient population in 2020 All ulcerative colitis patients³ 0.62M US | 0.86M EU5 | 0.15M JP 43.4M US | 33.9M EU5 | 5.1M JP 82.6% US | 45.2% EU5 | 55.5% JP 35.8M US | 15.3M EU5 | 2.8M JP Patients with moderate-severe disease³ Treatable population 42.6% US | 39.9% EU5 | 32.0% JP 0.26M US | 0.34M EU5 | 0.05M JP Illustrative approved therapies Dupilumab (Dupixent, AD launch 2017)2 (Subcutaneous injection, mAb, anti IL4/IL13) 2020 Net Revenue $3.2B WW (Atopic Derm Only)² Illustrative pipeline / recently approved therapies Illustrative approved therapies Adalimumab (Humira, UC launch 2012)7 (Subcutaneous injection, mAb, anti TNF) Vedolizumab (Entyvio, UC launch 2014)8 (Subcutaneous injection, mAb, anti a4ẞ7 integrin) 2020 Net Revenue $2.6B WW (UC Only) $2.0B WW (UC Only) Illustrative pipeline / recently approved therapies Ruxolitinib (Opzelura, AD launch 2021)9 (Topical cream, JAK inhibitor; expected cost ~$8,000 patient/year in US5) Peak Sales Forecast $1.1Bn WW (Atopic Derm Only) Ozanimod (Zeposia, UC launch 2021)6 (oral, small molecule, SIP1/S1P5 modulator, $86,000 patient/year in US³) Peak Sales Forecast $3.0B WW (UC Only) 6 Sources: (1) Global Data Atopic Dermatitis: Epidemiology Forecast to 2027, 28 November 2018; (2) Evaluate Pharma Product Report - Dupixent [Accessed 29 Oct 2021]; (3) Global Data Ulcerative Colitis Drug Forecast and Market Analysis to 2029; (4) Endpoints/Andrew Berens at SVB Leerink; (5) Incyte Opzelura approval investor call 22 September 2021; (6) FiercePharma/Salim Syed at Mizuho Securities; (7) Evaluate Pharma Product Report - Humira [Accessed 01 Nov 2021]; (8) Evaluate Pharma Product Report - Entyvio [Accessed 01 Nov 2021]; (9) Evaluate Pharma Product Report - Opzelura [Accessed 29 Oct 2021] Benevolent 71 Al#72BenevolentAl pipeline assets target treatment of prevalent diseases with high unmet need Illustrative Crohn's Disease target market 239k US | 233k EU5 | 17k JP Mod-Sev Crohn's Disease patients 20205 Global Crohn's Disease market value 20292 $11.9B Peak sales forecast Entyvio (2025)10 (Vedolizumab, Subcut. injection, mAb, anti a4ẞ7 integrin, launched 2014, $36,200 patient/year in US) $4.0B WW (Crohn's disease only) Illustrative NASH target market Treatable NASH patients 20204 Global NASH market value 202911 Peak sales forecast Resmetirom (2026)12 (MGL-3196, oral, THRb agonist, expected launch 2022) 13.9M US | 7.8M EU5 | 4.6M JP $27.2B $719M WW (NASH only) Illustrative GBM target market Treatable GBM patients 2020³ 10k US | 12k EU5 | 1.5k JP Global GBM market value 202613 Peak sales forecast Tagrisso (2026)⁹ (osimertinib mesylate, oral, EGFR inhibitor, expected GBM launch 2022,~$185,000 patient/year in US) Illustrative IPF target market Treatable IPF patients 20206 Global IPF market value 202616 Peak sales forecast Ofev (2026)17 (Nintedanib, oral, kinase inhibitor, launched 2014, $85,500 patient/year in US) $1.57B $594M (GBM only) 115k US | 68.9k EU5 | 21.5k JP $3.74B $2.85B WW (IPF only) Illustrative ALS target market Treatable ALS patients 20201 Global ALS market value 202914 Peak sales forecast verdiperstat (2026) 15 (oral myeloperoxidase inhibitor, expected launch 2023) 21k US | 22k EU5 | 11k JP $1.04B $192M WW (ALS only) Illustrative CKD target market Treatable CKD patients 20207 7MM CKD market value 20268 Peak sales forecast Farxiga (2024) 18 (Dapagliflozin, oral, SGLT2 inhibitor, launched 2021, $4,000 patient/year in US) 3.7M US | 3.1M EU5 | 1.8M JP $10.5B $639M WW (CKD only) Sources: (1) Global Data Amyotrophic Lateral Sclerosis: Epidemiology Forecast to 2029, 18 September 2020; (2) GlobalData, Crohn's Disease: Global Drug Forecast and Market Analysis to 2029, calculated for the 7MM (US, EU5, JP); (3) Global Data Glioblastoma Multiforme (GBM): Opportunity Analysis and Forecasts to 2027, 26 October 2018; (4) Global Data Non-Alcoholic Steatohepatitis: Epidemiology Forecast to 2029, 17 June 2020; (5) Global Data Crohns Disease Global Drug Forecast and Market Analysis to 2029, September 2020; (6) Idiopathic Pulmonary Fibrosis: Epidemiology Forecast to 2029, 17 September 2020; (7) Global Data Epidemiology and Market Size Database, Chronic Kidney Disease [Accessed 29 Oct 2021]; (8) Global Data OpportunityAnalyzer: Late-Stage Chronic Kidney Disease - Opportunity Analysis and Forecasts to 2026 22 December 2017 (9) Based on use in NSCLC, Evaluate Pharma Product Report - Tagrisso [Accessed 01 Nov 2021]; (10) Evaluate Pharma Product Report - Entyvio [Accessed 01 Nov 2021]; (11) Global Data, Non-Alcoholic Steatohepatitis (NASH): Opportunity Analysis and Forecasts to 2029; (12) Evaluate Pharma Product Report - Resmetirom [Accessed 01 Nov 2021]; (13) Evaluate Pharma Indication Profile - Glioblastoma Multiforme [Accessed 29 Oct 2021]; (14) Global Data, Amyotrophic Lateral Sclerosis (ALS): Opportunity Analysis and Forecasts to 2029]; (15) EvaluatePharma Product Report - Verdiperstat [Accessed 01 Nov 2021]; (16) Evaluate Pharma Indication Profile - Idiopathic Pulmonary Fibrosis [Accessed 29 Oct 2021]; (17) Evaluate Pharma Product Report - Ofev [Accessed 03 Nov 2021]; (18) based on pricing in Type 2 Diabetes, Evaluate Pharma Product Report - Farxiga [Accessed 01 Nov 2021]; Al Benevolent 72#73BenevolentAl Progress since mid-2020#74Significant progress across all aspects of the business since mid-2020 OOOO Pharma partnerships → Baricitinib; 38% reduction in mortality v SOC. Eli Lilly investment → AstraZeneca; continued delivery, 1st target selected for CKD Pipeline Progress → Phase 1b for Atopic Dermatitis progressing well. Completion by mid-2022 →IND-enabling studies started for novel Ulcerative Colitis asset (PDE10) Platform Enhancements → Mechanism mapping to better represent disease → Improved capacity to ingest human patient level data and genetics at scale Business Model → Ambition to take assets through to commercialisation ourselves including, PDE10 for UC People → Building a world-class Board (Dr. Francois Nader, Dr. John Orloff, Sir Nigel Shadbolt) → Investing in people to support a scaling pipeline (>80 recruited in last year) Al Benevolent 74