Westlake CloudValley Therapeutics: Leveraging AI to Decode AAV Delivery and Pioneering In Vivo CAR-T Therapies

Lijia Ma, Founder of Westlake Genetech and PI at Westlake University

Ma Lijia compared human genome research to a child's process of learning from "recognizing characters" to "writing essays." She explained: "If we compare human DNA to a book, then functional genomics is the stage of understanding sentences and paragraphs, while gene editing and therapy are like writing essays—it’s the process of modifying the genome."

In 2018, Ma Lijia returned to China and joined Westlake University, establishing the Functional Genomics Laboratory.

At Westlake University, she continued her postdoctoral work, establishing functional genomics as the cornerstone of her laboratory. However, her vision was not limited to publishing papers. She keenly realized that the "technology development work" focused on in the lab "is a discipline that leans more towards application." In her view, transitioning from scientific research to entrepreneurship is not accidental but a natural extension accompanying "human understanding of the genome."

She hopes to push the technology from the lab into the real world, "to apply it in medicine, to treat diseases and save lives. If we are the ones who understand this technology the best, then it is our responsibility to achieve the transformation from zero to one, or even from one to ten."

It was this "sense of responsibility and passion" that drove her to establish Westlake Genetech in early 2021.

But which specific track should this nascent entity steer toward? Ma Lijia and her team made a strategic focus: gene and cell therapy. Behind this choice lies her冷静 analysis and forward-looking judgment of the global pharmaceutical innovation landscape: about five years ago, it was already recognized that China and the West had essentially started at the same time in the gene and cell therapy赛道.

"This means that in this promising frontier field, China is not facing a huge generational gap with international peers as it does in the traditional pharmaceuticals sector. Instead, there is a tremendous opportunity to achieve genuine innovation and even surpass them, thanks to our accumulated knowledge and talent, as well as the strong resource reserves within the country," analyzed Ma Lijia.

After identifying the grand stage of gene and cell therapy, Westlake Genetech needed to find its first breakthrough. Ma Lijia described the precise decision-making framework used by the team at the time: first, which are the most important issues; second, which critical problems are we good at solving; third, what we are truly passionate about doing.

Based on the three-layer consideration of "real industry pain points, core team capabilities, and intrinsic passion," the answer is self-evident.At that time, the entire gene and cell therapy field had reached an "absolute consensus" on the industry's biggest bottleneck—delivery issues. Therefore, since its inception, Westlake Genetech has precisely targeted the "delivery" challenge, a critical bottleneck, as its strategic focus.

After clarifying the core focus of "delivery," Ma Lijia and her team needed to select a specific vector technology. Among the numerous viral and non-viral vectors, they ultimately chose adeno-associated virus (AAV). This was not an arbitrary decision but one based on a comprehensive evaluation of technical maturity, clinical validation, and room for improvement.

Marija pointed out that AAV is the most widely used in vivo gene therapy vector with the most FDA-approved drugs.This also means that AAV, as a delivery tool, has already had its safety and basic efficacy clinically validated to a certain extent, and is not a completely unknown field.

At the same time, Ma Lijia is also acutely aware of the challenges and skepticism facing AAV at present, but she believes that these are a necessary part of the process for any technology or therapy to mature from development to eventual widespread acceptance."We need to view these challenges with a developmental perspective, seeing them as inevitable growing pains on the path to technical maturity, rather than a final verdict that denies their potential."

To this end, Marija gave a highly convincing example - Zolgensma, a drug launched by Novartis. It is the world's first AAV gene therapy drug for treating Spinal Muscular Atrophy (SMA), delivering a functional copy of the SMN1 gene through a single intravenous injection across the blood-brain barrier into the central nervous system. This drug demonstrated the ability of AAV vectors to address unmet clinical needs but has issues of "high liver toxicity and poor targeting."

"When it successfully delivers one dose of medication to a child's central nervous system, 300 doses end up in the liver." This shocking figure— "300 doses"— vividly reveals the significant challenge natural AAV vectors face in terms of targeting: in order to deliver an effective dose to the affected area, non-target organs (such as the liver) must endure the toxic risks and enormous metabolic burden brought by three hundred doses of the drug.

"This proves that AAV is usable, but there is also enormous room for improvement." Ma Lijia believes this clearly defines the precise problem Westlake Genetech needs to solve: significantly enhancing AAV's extrahepatic targeting efficiency while minimizing its off-target effects.

Ma Lijia also emphasized that their modification of AAV did not start from scratch. "We are standing on the shoulders of giants. When we entered this field, several AAV drugs had already been marketed, and the modification pathways had been preliminarily established. We are not the first team to work on AAV vector evolution, nor are we the only team using AI for evolution."

The core technology of Westlake Genetech lies in the deep integration of AI into AAV vector design. The underlying logic is to "serialize" complex biological components — the Cas9 protein and gRNA of gene editing systems, as well as the capsid proteins that determine AAV targeting, can essentially be represented by nucleic acid sequences. This reduction establishes the foundation for utilizing AI to process and optimize these components.

Having encodable sequences alone is not enough; it is also necessary to obtain massive and precise data corresponding to the functions of these sequences. Ma Lijia pointed out that second-generation sequencing is currently the only technology in the biopharmaceutical field capable of producing high-quality data that is "affordable, precise, and high-throughput." When the two conditions of "components being serializable" and "functions being readable in a high-throughput manner" are met simultaneously, they form the closed-loop logic of AI-assisted research and development: training models to learn the mapping relationship between sequences and functions. This is the unique path of Westlake Genetech—its AI optimization targets are not small molecules but life "programs" themselves, such as gene editors, gRNA, and AAV capsids, achieving fundamental "programming" and "debugging" of life’s software.

The effectiveness of this pathway is evident in the data.TCE001, a vector developed by Ma Lijia and her team in 2023, exhibits a hundredfold increase in affinity for T cells compared to wild-type AAV. In mice and non-human primates, the accumulation of the virus in the liver is reduced to 1/100 to 1/1000.Faced with breakthrough data, the team’s first step is to “repeat and verify,” followed by cryo-electron microscopy structural analysis and functional experiments, transforming “black box” optimization into “mechanism confirmation,” moving from excitement to certainty.

Marija Ma likens AI training to "the process of nurturing talent": starting with a single objective and gradually adding multiple objectives. For instance, the initial evolution of AAV vectors only required crossing the blood-brain barrier, then progressively incorporated objectives like "avoiding liver enrichment" and "maintaining packaging efficiency," which are production-related attributes. This incremental training strategy ensures that the vector is both functionally excellent and industrially viable.

In the public's perception, Westlake Genetech might have gained prominence alongside the rising popularity of the In Vivo CAR-T concept. Over the past year, In Vivo CAR-T has garnered significant attention due to several acquisitions. In March 2025, AstraZeneca acquired EsoBiotech for $1 billion; by the end of June 2025, AbbVie acquired Capstan Therapeutics, Inc. for up to $2.1 billion in cash; in August 2025, Gilead’s Kite announced a $350 million all-cash acquisition of Interius, an In Vivo CAR-T therapy company; and in October 2025, BMS announced it would acquire private biotech company Orbital Therapeutics for $1.5 billion in cash.

However, Mary Ma clarified an important strategic sequence: in vivo CAR-T is not their preset ultimate goal but a natural application outlet for their core platform technology—AAV vector innovation.

"In fact, within our team, the in vivo CAR-T project was a relatively late development," admitted Ma Lijia. Initially, their core focus was on AAV innovation, and targeting T cells was just one direction in their strategic map. It wasn't until they developed TCE001 that they began to consider: what can this vector do?

Ma Lijia recalled that the concept of in vivo CAR-T was conceived about two years ago, triggered by the team's successful development of the AAV vector TCE001, which efficiently targets T cells with significantly reduced liver toxicity. With this "precision missile" in hand, they began exploring its optimal application scenarios. Given that CAR-T therapy had already demonstrated immense value in vitro but faced core challenges such as complex preparation and high costs, a groundbreaking idea emerged: leveraging established scientific and clinical foundations to shift CAR delivery from the complex ex vivo process to a streamlined in vivo procedure, fundamentally transforming the therapeutic paradigm.

However, any disruptive idea often comes with skepticism at its inception. Ma Lijia admitted that when the team proposed the concept of in vivo CAR-T in the second half of 2023, even though they were confident in the vector itself, the idea was still considered "far-fetched" and faced doubts even internally. Fortunately, an open team culture allowed for bold attempts, and the positive results from initial experiments provided the initial support for this approach.

The real "breaking through" test happened on the international academic stage. In May 2024, Li-Jia Ma brought two achievements – AAV-BBB001, which can penetrate the blood-brain barrier, and AAV-TCE001 for in vivo CAR-T – to the annual meeting of the American Society of Gene and Cell Therapy (ASGCT), where she gave oral presentations on each. Her original expectation was that the more groundbreaking in vivo CAR-T work using AAV would cause a huge sensation, while BBB001, though excellent, fell within the scope of exploration by international peers.

However, the reality was quite the opposite: BBB001 garnered significant attention, while the report on in vivo CAR-T was scheduled for the final session of the conference, with few attendees and a lukewarm response, eliciting only polite remarks like "sounds good." This experience gave her a profound realization that true innovation does not always receive applause at its inception.

However, at the annual meeting, Moli Jia and her team also received an unexpected harvest:Professor Guangping Gao, a senior scholar in the AAV field, highly recognized and spiritually encouraged their work after listening to the report, marking the first significant external affirmation. Subsequently, the team compiled two studies into papers and submitted them to top journals in the field (unnamed as they have not been published yet), both of which have now entered the revision stage.

In this regard, Ma Lijia particularly pointed out that these recognitions are extremely hard-earned for her new team in the field of gene and cell therapy. These two events, which occurred successively in mid-2024, have greatly strengthened the team's belief, proving that their chosen AAV in vivo CAR-T technology route is likely correct.

By the time of the 2025 ASGCT conference, the market trend had clearly shifted, with in vivo CAR-T becoming a hot topic. Ma Lijia observed that major mergers and acquisitions within the industry at the time were concentrated on in vivo CAR-T technologies utilizing lentivirus and LNP vectors, with no MNCs (global top pharmaceutical companies) explicitly supporting the AAV route. She analyzed that this precisely reflected the uniqueness and forward-thinking nature of the AAV pathway—few in the industry at the time believed that AAV could be used as an in vivo CAR-T delivery vector. It was for this reason that when she introduced the AAV solution at domestic conferences such as the CSGCT, it also allowed more peers in China to learn for the first time that AAV could be applied to in vivo CAR-T.Mark A. Kay, a professor at Stanford University and guest of the CSGCT conference, gave very high praise to this innovative project after listening to the presentation.

As the results were disclosed, Westlake Genetech began to capture the attention of MNCs.Ma Lijia pointed out that MNCs are increasingly focusing on China's original innovation and have increased their efforts to search for projects in China.A landmark event was that the company unexpectedly received the "Star Start" Innovation Award from Novo Nordisk, a company primarily focused on metabolic diseases. She admitted to being somewhat surprised by the win, as their work does not directly align with Novo Nordisk's main focus. However, this indicates that the breakthrough value of Westlake Genetech’s AAV platform technology transcends disease area boundaries, gaining widespread recognition based on the essence of innovation.

Looking back on the journey from being questioned to being recognized, Ma Lijia summarized that original innovation inevitably goes through a long cycle: in the early stage, the team had doubts about the direction, but as data was generated, the internal team first became confident in the research and development direction; when the trend emerged, the focus of skepticism shifted from "whether the direction is right" to "the choice of technical routes"; however, in the end, everything still has to go through rigorous clinical validation.Ma Lijia also emphasized the significant potential of in vivo CAR-T in reducing costs. Preliminary assessments suggest that it could reduce treatment costs to about 5% (one-twentieth) of current ex vivo autologous CAR-T therapies, which would greatly enhance drug accessibility.

At the same time, Mary Ma predicted that more in vivo CAR-T clinical data would be released within the next year or so based on China's strong CDMO capabilities and clinical development efficiency, which would drive the industry to reach a consensus on dosage and pricing. She even humorously stated that she hoped her technological breakthroughs would boost the development of the entire AAV industry chain, giving CDMO "tanks" a new lease on life.

Marijia Ma summarized the core spirit of Westlake Genetech as "solving real problems." "We haven't followed any trends or worked on me-too projects; all our directions are original and challenging."

She cited the viewpoint of DeepSeek founder Wenfeng Liang, emphasizing that "giving a challenging problem to someone with potential is more important than anything else." In her team, there is an AI director who transitioned from a major internet company, members from developmental biology backgrounds working on the CAR-T project, and also a large number of undergraduates. "We don't care whether you are an industry expert; we value your learning ability, passion, and uniqueness."

Her advice to the younger generation is: "It takes a good blacksmith to make good steel; find a development path that suits you. Secondly, choosing a platform with strong growth potential is far more important than starting salary." She used her experiences at BGI and Westlake University as examples to illustrate that "growing together with a platform can lead to completely different rewards."

"We are not afraid to start from scratch," she concluded. "As long as you are solving truly important problems, the path will become wider and wider."