This story as been corrected to show that Ensoma's vector is based on CRISPR-Cas12a, not CRISPR-Cas9, and that it integrates into the nuclear DNA of the target cell.
NEW YORK – In cell and gene therapies, the genetic machinery is key to producing the proteins missing or dysregulated in a disease, but whether the therapeutic cassette reaches target cells and how well patients respond are often dictated by the vehicle carrying this precious cargo.
Recognizing that the delivery system can often make or break a cell and gene therapy product when it comes to its safety, efficacy, and tolerability in patients, biotechnology firms are investing in a new generation of viral and nonviral vectors. When considering gene delivery vectors specifically for precision oncology applications, drugmakers must take into account whether the gene is being delivered in vivo or ex vivo and if the engineered gene will be used to directly target and kill cancer cells or to propagate and expand with cells as they divide.
Vectors typically fall into the categories of viral and nonviral. Viral vectors commonly used in cell and gene therapy include adenovirus, adeno-associated virus (AAV), lentivirus, and retrovirus, while some examples of nonviral vectors used in the field include lipid vesicles, virus-like particles (VLPs), and exosomes.
Next-generation adenoviral vectors
Adenovirus, the first virus adapted for use within gene therapies in the early 1990s, offers high transduction efficiency and scalability for production. However, adenoviral vectors can also present challenges, including immunogenicity and cellular toxicity. Because preexisting immunity against adenovirus is common in human populations, gene therapies using adenoviral vectors are typically developed as "one-shot" treatments that do not require repeat dosing.
Adenovirus comprises an icosahedral protein capsid of about 100 nanometers surrounding a double-stranded DNA genome of 26 kilobases to 45 kilobases, and early adenoviral vectors could accommodate a payload of about 4.5 kilobases of DNA. However, many newer adenoviral vectors typically have had all viral sequences deleted, except the minimum required to function as a vector. These high-capacity adenoviral vectors can carry up to 37 kilobases of DNA payload, enabling them to transport the larger, multi-gene cassettes needed for some cancer therapies.
Boston-based Ensoma, for example, develops off-the-shelf, in vivo cell therapies that use a helper-dependent adenovirus to deliver a CRISPR-Cas12a-based gene-editing toolkit. Also known as a virus-like particle, the helper-dependent adenovirus has had all of its viral genes removed and requires a helper virus to propagate. These "gutless" viruses can hold a DNA payload of up to 35 kilobases and are able to transduce a variety of cell types.
Ensoma CSO Robert Peters said that although it is a blank slate on the inside, the company's helper-dependent adenovirus "looks like" an adenovirus on the outside and has the same ability to deliver DNA to the endosome, where cellular proteins can traffic the message to the nucleus and insert it into the nuclear DNA of hematopoietic stem cells (HSCs). These cells then generate in vivo engineered T cells, NK cells, and macrophages and marshal a multipronged attack on cancer cells.
Ensoma's most advanced product, EN-374, is an in vivo therapy for chronic granulomatosis, an X-linked genetic disease that affects the immune system. EN-374 is designed to deliver a corrected copy of the CYBB gene directly to the patient's HSCs, restoring the body's ability to fight infection. While Ensoma is preparing to file an investigational new drug application for that program in the first half of 2025, it is also advancing in vivo cell therapies for sickle cell disease, solid tumors, and other hematology indications.
Ensoma's virus-like delivery system plays a critical role in getting all of these therapies to their intended destination. "It's that combination of effects leveraging the biology of an adenovirus that we're using for [multi-lineage] approach for oncology," Peters said.
The firm's initial target in its cancer pipeline is HER2, which Peters noted is a "proven target in a range of cancers." HER2-targeted therapies, which were initially approved for treating breast and gastric cancer, have now become an option for other advanced HER2-expressing solid tumors. Ensoma aims to program the patient's HSCs to produce cancer-fighting macrophages in vivo. Engineered macrophages, or CAR M cells, have the potential to infiltrate an immunosuppressive tumor microenvironment and turn an immunologically cold tumor hot, Roberts said, enabling treatment of a wider range of tumors.
Other companies have attempted to develop CAR macrophage or similar CAR monocyte therapies against cancer targets, including HER2. For example, Carisma Therapeutics began a Phase I trial of its ex vivo gene-modified autologous CAR-monocyte therapy in May 2024, only to wind down its HER2 program in December, citing a tough competitive landscape. Before it shuttered the program, Carisma was able to demonstrate that its anti-HER2 CAR monocyte-cell therapy was safe, well tolerated, and feasible to manufacture.
Ensoma is tackling what Peters sees as the main challenge in CAR M-cell therapy — a short half-life — by programming hematopoietic stem cells to produce an endless supply of the engineered macrophages, rather than infusing a one-time dose of mature engineered macrophages. And while in comparison lentiviral vectors are the workhorse of ex vivo cell therapies, Peters said that they lack the precision to effectively target HSCs. Ensoma is trying to get around this problem with its vector, which binds CD46. This surface receptor that the firm identified enables in vivo transduction of HSCs, which then continuously produce engineered immune cells of different lineages, not just macrophages or monocytes.
Yinghua Wang, VP of oncology at Ensoma, noted that the ability to produce multiple types of cells is a major advantage over most other CAR therapies. "[Other cell therapies] cannot fully leverage the full immune system, meaning both innate and adaptive immune cells. We can really address and fully leverage the multiple lineages of cells," Wang said. "We will have unlimited production and supply of these effector cells [in the body]." In contrast, he noted, ex vivo cells become exhausted.
Although the helper-dependent adenovirus is less immunogenic than its wild ancestor, Peters said there is some immune response. Thus, Ensoma's therapies are designed for single dosing, and Peters envisions that patients will receive the treatment on an outpatient basis.
Another family of viruses, the adeno-associated viruses, are also proving useful in gene therapy. AAVs are viruses that rely on adenovirus to complete their life cycle. AAVs do not cause any human disease and do not have genes needed for replication and expression on their own. While AAV vectors are less immunogenic than other viral vectors, making them an appealing option for gene therapy, preexisting and therapy-induced immune responses remain an obstacle to their use.
Dyno Therapeutics is addressing this issue by using a deep learning model to optimize AAV capsids to improve targeting, increase payload size, and reduce immune reactions. Dyno specializes in optimizing AAV capsids and partners with biotech and pharma companies who develop and commercialize products made with those capsids. "The challenge is to safely and effectively deliver therapeutic genes into a patient's body right where they're needed and nowhere else," said Dyno CEO Eric Kelsic. "We want to do that for all patients and at low cost."
Kelsic said that researchers and drugmakers have been on a quest to engineer better AAV capsules for more than 25 years and have made little progress. That's why Dyno turned to artificial intelligence to accelerate its capsid engineering programs and tackle challenges like organ targeting. AAVs, when given systemically, mostly end up in the liver, Kelsic said. However, there remains an unmet need for gene therapies directed to other tissues and organs. "We're engineering the capsid to increase the efficiency for those target organs like the brain or the muscle and de-target other organs like the liver," Kelsic said. In so doing, Dyno is also trying to increase the efficiency of AAVs by "hundreds of times" compared to natural vectors for targeting the central nervous system, according to Kelsic, while reducing targeting of the liver more than 10-fold.
The more precise targeting and higher efficiency also may solve another challenge the field has encountered when using AAVs, which is that high doses are required to achieve a therapeutic effect, which can increase toxicity risks. "Increasing the efficiency of engineered vectors is a very excellent path for still having a therapeutic effect, but doing that at a lower dose, which is potentially safer," Kelsic said.
In January, Roche announced the successful completion of a 2020 partnership with Dyno to develop an AAV capsid for a neurological indication. That program is now advancing in Roche's pipeline toward clinical testing, according to Kelsic. Following the success of that project, Dyno has inked a new partnership with Roche focusing on developing AAV capsids for central nervous system delivery, again focusing on improved efficiency while also optimizing for properties like on-target versus off-target effects. Kelsic highlighted as an example that its Dyno bCap1 capsid crosses the blood-brain barrier about 100 times more efficiently than a natural AAV.
Retroviral vector advantages
The retroviral family of vectors offers a different set of advantages for cell and gene therapy. Retroviruses are spherical viruses with a single-stranded RNA genome. Two of the most popular retroviruses used in cell and gene therapy include the simple murine leukemia virus and the more complex lentivirus, derived from human immunodeficiency virus.
Adaptimmune uses lentiviral vectors in its T-cell receptor (TCR) T-cell therapies. The firm engineers the sequences that encode the T-cell receptors within a T cell to create a cell therapy that harnesses the patient's own immune system to target certain cancer antigens. The process begins with the collection of T cells from the patient at the clinic or hospital. The cells are transported to a central manufacturing facility where they are washed and cryopreserved. The cells then undergo a selection process for CD4 helper T cells and CD8 cytotoxic T cells followed by lentiviral transduction to deliver genes encoding affinity-enhanced TCRs. Adaptimmune then returns the engineered cells to the infusion center for the patient to receive.
Adaptimmune's lead product is Tecelra (afamitresgene autoleucel), a TCR T-cell therapy directed at MAGE-A4, which is expressed in multiple solid tumor types. The US Food and Drug Administration (FDA) approved Tecelra in August 2024 for metastatic synovial sarcoma patients with certain human leukocyte antigen (HLA) types and also approved Thermo Fisher Scientific's HLA typing kit as a companion diagnostic for identifying treatment-eligible patients.
Adaptimmune has plans to submit a rolling biologics license application with the FDA for another TCR T-cell therapy, letetresgene autoleucel, this year for patients with one of several specific HLA types. The company also has several preclinical-stage programs spanning a range of solid tumors and hematological malignancies.
According to Phil Bassett, Adaptimmune's VP of process development, the company chose lentiviral vectors for its cell therapies because of their ability to integrate the gene of interest into the target cell and enable long-term gene expression. Although the exact nature of the vector is proprietary, Bassett said that it is a third-generation, self-inactivating lentiviral vector that has been modified with a viral envelope studded with vesicular stomatitis virus glycoprotein (VSVG) to enhance gene transfer into the cell.
Bassett said the self-inactivation feature "significantly reduces" the risk of recombination of lentiviral vector sequences with other sequences that could lead to generation of a replicating virus, and that the vesicular stomatitis viral protein "pseudotype" confers stability on the vector to survive processing conditions during manufacturing and storage. "Additionally, a VSVG pseudotyped lentiviral vector has broader potential for transduction of multiple different target cell types, such as CD4 and CD8 T cells and CD34 stem cells," Bassett said.
Bassett noted that AAVs, the other main vectors currently being used in cell and gene therapy, tend to be more suited for in vivo gene delivery than ex vivo. "AAV does not integrate the gene of interest into the target cell's chromosomal DNA," he said. "Instead, the gene is expressed transiently. Expression is, therefore, lost over time with AAV as the target cells divide and copies of the gene are diluted in the daughter cells."
Denovo Biopharma, a San Diego-based company that develops precision medicines for central nervous system and oncology indications, uses a retroviral replicating vector for its investigational gene therapy products. Replication-competent vectors first attracted attention as a potential cancer therapy in 1904, when a physician published a case report describing how a leukemia patient infected with influenza went into remission. Replication-competent viruses have subsequently been investigated as oncolytic viral therapies. During the '80s and '90s, researchers studied a range of replication-competent viruses looking for a potential vector for cancer therapy, and retroviruses emerged as a virus well-suited for use in cancer as a non-lytic gene transfer vehicle. In the past, their utility for gene therapy has been limited by low levels of gene transfer and poor efficacy. However, recent innovations in RRV design have improved those features, making them more attractive for gene therapy applications.
The RRV Denovo uses for its investigational glioblastoma therapy, DB107 (vocimagene amiretrorepvec), is based on a murine leukemia virus. Denovo CEO Wen Luo said that Denovo's next-generation RRV technology has advantages compared to other common vectors for gene delivery, such as AAV, including an ability to deliver a high continuous local concentration of the gene product to the tumor or the tumor microenvironment. The vector also features enhanced viral replication and transgene expression, and it can carry and express multiple transgenes.
Denovo acquired the RRV technology from Tocagen in 2020, along with Tocagen's gene therapy Toca 511, which it renamed DB107. The RRV infects and integrates only in dividing cells, preferentially cells with defective immunity, including malignant cells. Rather than immediately killing the target cell like an oncolytic virus, DB107 selectively infects and spreads among residual tumor cells. DB107, which is administered during tumor resection surgery, introduces the gene for cytosine deaminase into the tumor cells, which then begin producing the enzyme.
Separately, patients receive the pro-drug 5-fluorocytosine, which in the tumor cells is converted to the chemotherapy drug 5-fluorouracil by cytosine deaminase. This results in selective targeting of cancer cells with potent chemotherapy without systemic toxic effects. In a Phase III trial conducted by Tocagen, the therapy did not improve patients' overall survival compared to standard-of-care therapy. However, an analysis carried out by Denovo showed that a subset of patients with a germline genetic biomarker, dubbed Denovo Genomic Marker 7 (DGM7), lived longer on the combination regimen versus those in the comparator arm.
DGM7 is located in an intron of the SHROOM3 gene. Luo said the SHROOM3 alteration may impact viral entry into the cells, making some patients more sensitive to the virus. "Patients that carry the biomarker can be infected by [the RRV] more efficiently, which leads to better overall survival than in a [biomarker-negative] patient," Luo said.
According to Luo, in animal models, the drug worked as designed, selectively killing tumor cells while sparing normal cells. The FDA gave Denovo the go-ahead to begin a Phase II trial of the drug in 2023. It is also being evaluated in two investigator-initiated studies at the University of Miami in Florida and the University of California, San Francisco.
Denovo is also exploring the utility of another Tocagen product based on the same RRV backbone. RRV-scFv-PDL1 encodes an antibody targeting PD-L1 that competes with PD-1, essentially forcing the tumor to secrete its own checkpoint inhibitor. The firm has preclinical data showing "compelling efficacy," according to Luo, and it plans to pursue investigational new drug-enabling activities. Luo further noted that there are opportunities for developing RRV-based therapies targeting CTLA4 and cancer-associated antigens in various solid tumors.
Although many companies, as a practical matter, work with a single vector platform, Sana Biotechnology takes an approach that pairs the optimal vector system with each indication and class of therapy. Sana's pipeline includes ex vivo, in vivo, and stem cell-derived therapies, and the company's portfolio of delivery systems includes electroporation, lentiviral vectors, and virus-like particles.
"You can do anything you want to a cell in a Petri dish," Sana CEO Steve Harr said. "Our goal is to be able to deliver any payload — mRNA, RNA, DNA, protein — right to any cell in a specific and repeatable way."
Harr said that when starting a drug discovery program, it's important to begin by defining the cell type targeted for gene delivery and what type of genetic material is being delivered, followed by a consideration of whether the material should be delivered temporarily or permanently. Additional factors to weigh include whether the cells will be dividing and growing, whether the genetic payload should be integrated into the cell's chromosomal structure and passed on to progeny, and whether manufacturing will happen inside or outside the body.
In Harr's view, gamma-retroviral vectors, such as the murine leukemia virus, are the easiest to manufacture, while lentiviral vectors have a strong track record in autologous CAR T-cell therapy and AAVs are good for delivering DNA not meant to integrate into the cellular genome. Each of those vectors also has a downside, Harr noted. For example, although lentiviral vectors have a strong safety record, experts have been exploring what may be causing new cancers in patients receiving BCMA- and CD19-directed CAR T-cell therapies like Bristol Myers Squibb's Abecma (idecabtagene vicleucel) and Janssen's Carvykti (ciltacabtagene autoleucel) that happen to incorporate them. And for AAVs, he said they are both harder to manufacture and harder to get into cells.
Sana is developing an in vivo CD19 CAR T-cell therapy, dubbed SG299, using a lentiviral vector that has features from a paramyxovirus to facilitate entry into the cell. The company has demonstrated that this system, dubbed the paramyxovirus fusogen system, allowed the construct to avoid off-target delivery, which is common with lentiviral vector-based in vivo cell therapies, and successfully deliver its payload to CD8 T cells.
Harr said the company will develop SG299 both for cancer and autoimmune disorders and will pursue a blood cancer as the initial indication. Later, the firm plans to expand studies to include patients with B-cell mediated diseases such as lupus, multiple sclerosis, and myasthenia gravis. "Our goal is an [investigational new drug application] next year," Harr said.
For its pipeline of allogeneic CAR T-cell therapies, Harr said the company uses one of two methods, depending on whether the intention is to knock out or knock in a gene. For therapies in which a gene is knocked out, Sana uses electroporation to deliver CRISPR-Cas12B gene-editing machinery, and for knock-in constructs, the firm uses a VSVG pseudotyped lentiviral vector. Harr noted that lentiviral vectors work well ex vivo, but in the body, they aim for the liver, making them less useful for in vivo applications.
The synthetic vector
Although nature has provided powerful tools for delivering the genetic payloads of cell and gene therapies, viruses and virus-like particles have their limitations. These vectors require significant engineering and may still trigger immune reactions in their hosts, and even then, every viral vector comes with restrictions, such as whether it can deliver the genetic payload to dividing cells or not.
One researcher looking beyond those boundaries is Joshua Leonard, a professor of chemical biological engineering at Northwestern University. Leonard said his group is dedicated to developing "adjacent technology" and anticipating the needs in the biotech field. One of Leonard's interests has been developing extracellular vesicles (EVs) as an alternate gene delivery vector for cancer immunotherapy, which he believes may be a much more flexible option that researchers could program with functions bespoke for their applications.
Leonard and his collaborators developed a suite of methods for producing EVs from human cells for targeted delivery of genes and validated them by delivering a CRISPR gene-editing toolkit into primary human CD4 T cells. When Leonard's team introduced the EVs into a T-cell culture, they bound efficiently and successfully delivered their cargo, inactivating a gene encoding a receptor used by HIV to infect cells.
Through a startup called Syenex, Leonard aims to develop delivery vehicles based on this EV technology to enable the discovery and development of new cell and gene therapies. Although Leonard is looking to the future of the field with his technology, he doesn't envision that the field will totally abandon viral vectors. In fact, he's seen a lot of interest from prospective customers looking for help delivering lentiviral or retroviral cargo.
"It's not necessarily an either/or proposition when you get it down to individual components," Leonard said, noting that lentiviral and retroviral vectors have an established manufacturing and safety record. Instead of replacing viral vectors completely, he speculated that EVs could offer complementary benefits. For example, EVs might have a therapeutic role in certain oncology indications that could benefit from a transient approach, he suggested, or where researchers may prefer to trade a bit of the efficiency they'd get with viral vectors for a safer, less immunogenic EV.
"You might want to lean towards something that's more like an EV in the sense that it may have a little bit less risk of immunogenicity, just because it uses more human components, even if it's maybe less efficient," Leonard said.
He also sees EVs as a departure from the time-consuming and burdensome ex vivo cell therapy manufacturing process. "For many patients, the barrier to getting treated is related to the complexity of manufacturing these therapeutics, not the way that the cell ultimately works."
In contrast, in vivo approaches facilitated by next-generation delivery vectors could make cell therapies available to patients who don't have access to ex vivo CAR T-cell therapy or need to begin treatment sooner. "I'm trying to think about how many of these approaches we can make safe and effective enough that you can do in vivo delivery of these genetic cargos," he said. "Ultimately, it's the way we're going to have to move to get to the full spectrum of patients who could benefit."