Discovery Technologies for Cell and Cell-Based Therapies

Dr Ryan Cawood
Dr Robert Leydon
20 September 2022
Cell Therapy
Cell-Based IO
Viral/Non-Viral Vectors
Ryan Cawood and Robert Leydon from OXGENE, a WuXi Advanced Therapies company, discuss current challenges and opportunities for discovery technologies for cell and cell-based therapies. Click here to download the full interview as a PDF
First of all, could you please introduce yourselves and tell us a little about your roles?

I’m Ryan Cawood (RC) and I’m the Chief Scientific Officer of both OXGENE and WuXi Advanced Therapies. My role is primarily to think about where we want to take both the discovery and
manufacture of viral vectors in the future, identifying the limitations in the industry and then trying to structure the teams and give them the objectives that allow us to solve those problems. My background is in virology and I have spent the last 15 years working on viral vectors.

I’m Robert Leydon. (RL) I run a group that specializes in CAR-T generation and antibody services, which includes a variety of commercial offerings such as antibody developability assessment and CAR-T construction. Taken together, these services encompass the steps a CAR-T developer would need to take to facilitate the discovery, generation, and development of a CAR-T against their target. These processes range from hybridoma sequencing or antibody discovery, through to humanization, affinity maturation, and finally CAR-T construction. My background is largely in molecular biology, and recombinant expression systems, whilst immunotherapy is my current focus.

What are some of the key challenges associated with cell therapy discovery stages?

RC – The discovery of cell therapies is challenging. Firstly, how do you identify the particular antigen that you want to develop a CAR-T against, particularly for things like solid tumors which have so far proven relatively refractory to a lot of the existing CARs that we’ve already developed?

Trying to identify the target that you’re going after is one of the major challenges that comes before the CAR-T design; this is the piece that Robert is doing a lot of his work on.

We have developed CAR-Ts against a number of different conditions, and they don’t always work, so trying to understand why some CAR-Ts work and others don’t, is another one of the major challenges.

Once you have identified the target, the challenge then becomes the design of the CAR itself.

How do you ensure that the CAR is going to express well within the T cells once you’ve added it? How do you ensure that those CAR-Ts are going to go after the right cells and not cause toxicities?

Then after the discovery stage, once you have actually got your CAR-T, you have to manufacture it, which brings an entirely separate series of challenges.

RL – I think a lot of the challenges with antibody scFv (single chain fragment variable) discovery have to do with library design, including the complexity of those libraries, which to a large degree, determines the quality of the resultant hits.

There are additional challenges in designing the CAR. For instance, if we’re taking our scFv from a murine source, there may be issues with immunogenicity that need to be mitigated, for example by antibody humanization.

When constructing the CAR construct, there are challenges relating to the complexity of the interface between the CAR-T cell and the target antigen.

Effectively, we are trying to mimic the natural interaction between T cells and antigen presenting cells, which is far more complex than perhaps we see with CAR-T antigen binding.

Enjoying this interview? Download the full eBook, here >>

What can be done to mitigate the challenges associated with CAR-T discovery?

RL – We have produced a variety of innovations at OXGENE, particularly involving the discovery stages. For example, our SLIM affinity maturation platform allows us to express the scFv library on the surfaces of mammalian cells which also overexpress the target antigen. We can then use an excess of decoy cells to ‘mop up’ any secreted, unbound scFv, before employing a Förster resonance energy transfer (FRET) assay to detect ‘true binders’, so that we only obtain a signal when the antigen and an scFv are in very close proximity, i.e. bound.

However, for clients using murine antibodies as a source of scFvs, humanization will be a requirement. We’ve developed a sophisticated software application which enables selection of optimal human templates from a germline antibody database. Antibody humanization is an effective strategy for reducing the otherwise very high risk of immunogenicity that comes from introducing a murine antibody fragment into a human patient via the CAR-T construct.

Our humanization software allows us to graft the CDRs (complementarity determining regions) from a mouse antibody onto the optimal human template, while simultaneously predicting T-cell epitopes to assess potential immunogenicity. Greater numbers of murine residues are expected to lead to greater numbers of epitopes. Our scheme provides a responsive output when we’re designing humanized variants. We can see the potential T-cell epitopes melt away as we progress from more mouse-like sequences humanlike sequence. Problematic murine residues can be easily spotted during the design process and unintentionally introduced epitopes avoided. This is an integrated approach to designing humanized variants whilst simultaneously assessing the immunogenicity of those variants relative to the murine parent, providing instantaneous, visual indications of the consequences of specific design changes.

Enjoying this interview? Download the full eBook, here >>

How can optimization help to improve safety and efficacy for CAR-T?

RC – In terms of optimizing CARTs, there are a lot of different parts that you can consider in terms of optimization. There’s the raw DNA sequence itself, considering how efficiently and when that particular gene is expressed in T cells.

At OXGENE, we’ve been developing a promoter discovery workflow that we’ve so far used to identify DNA promoters that drive gene expression very specifically in T cells. This adds an advantage from a safety perspective because rather than having a CAR-T vector that expresses in any cells it gets into, the CAR-T is only expressed in the particular cells that you want it to be expressed within.

There’s also the gene itself. There are billions, if not, trillions of ways to encode a CAR-T molecule. There are many ways to get it wrong, and there are many ways to get it right, and so optimizing that DNA to ensure that you have good expression of the CAR-T in your T cells is certainly another area where you can improve expression.

For most CAR-T products, lentiviral vectors are the vehicles of choice for delivering the CAR-T cassette into the T cells.

Over the last few years, we’ve been developing improved plasmid systems for the manufacture of lentiviral vectors. So once we’ve got the promoter driving the CAR-T, we can put that into our optimized plasmid system which ensures that we produce the high levels of lentiviral vector we’re going to need when we add it to the CAR-T cells. Next we can consider how efficiently the viral vector is expressed and produced within the T cells, as well as how efficiently we can deliver that vector to the T cells.

RL – Something I would add is that there are multiple generations of the CAR-T technology, with the very earliest generations composed more or less of an scFv and some elements of the TCR (T-cell receptor). More recent generations – and we’re up to four generations now – involve the use of intracellular signaling domains from a variety of costimulatory and inhibitory receptors which are expressed on the surface of T-cells, and which also respond to signals from the antigen presenting cell itself. More recent generations of CAR-T therapy involve the inclusion of some of these intracellular domains as part of the CAR, such that when the scFv finds its antigen, you get signaling from all of these different domains.

There’s an opportunity for optimization here in choosing the most appropriate domains and in how you combine inhibitory and stimulatory domains simultaneously within the molecule to achieve the most optimal outcome. I think deconvolution of the relationship between T-cells and antigen presenting cells is going to provide the solutions to a lot of the current challenges, particularly in regard to cytokine release syndrome, a common side effect of CAR-T cell therapy, the effect of which can range from mild to extremely severe.

Enjoying this interview? Download the full eBook, here >>

What would be the impact of optimizing gene expression?

RC – One of the things that we do differently at OXGENE in terms of optimizing genes is that we work in high throughput. Most algorithms work by simply making a single new version of a particular gene and saying “there is the optimized version”. What we do differently is that we will make millions of variants of that gene and then rank them all based on a number of different parameters, including GC content, a codon bias, homology, direct repeats, splice sites, maybe restriction sites as well, and then we filter those million variants that we’ve made in silico, selecting the top few that we believe will have the best parameters. That’s a little bit different to the way most optimization algorithms work and certainly from the results that we’ve seen, it does seem to improve expression.

RL – At OXGENE, we have a proprietary codon optimization algorithm, which Ryan designed, for improved expression of transgenes in a recombinant expression system of choice. Additionally, selecting the optimal human germline antibody templates is essential for humanization. Antibody expression levels are largely determined by the composition of the amino acid sequences of
the variable regions of antibodies, and this can change them from being terrible expressers to very good expressers, so you need to consider templates carefully, both for similarity to the murine
parent sequence, and for their likelihood of being abundantly expressed.

There are also all kinds of manufacturing liabilities that can be assessed at the sequence level. This isn’t the same as experimentally determining expression levels or stability, but you can make predictions about how easy the scFv might be to manufacture based upon the identification of certain motifs within their amino acid sequences. Potentially, amino acid substitutions can be made to improve the manufacturability of antibodies and scFvs.

If you do a combination of all these things, then you stand to have the best possible chance of obtaining good expression of your scFv on the cell surface, which aids discovery and CAR-T generation from identified hits.

There are sometimes quirks that you will discover in the very early research phases that are critically important for later manufacturing.

Enjoying this interview? Download the full eBook, here >>

How could an end-to-end discovery to development solution accelerate delivery to patients?

RC – A lot of the issues that companies have in developing their products is that there’s a lot of tech transfer, so you have to go from one company to another. That company then has to pick up where the other company left off, and all the details that are associated with that don’t always necessarily get transferred over.

I think it’s really good that we, as a business, can go all the way from the DNA: the design, the initial concept of what the customer wants to work with, all the way through optimization and humanization, to actually demonstrating that the platform works. Then we can continue with lentiviral production and test how well these vectors work in house. We can take customers all the way along that journey, and when they leave OXGENE, tech transfer to our colleagues at WuXi Advanced Therapies is still an internal process – so all the quirks and details of the process are accounted for – and they can do the GMP manufacturing and testing of the viral vector and the cell therapy product as well.

It’s rare that you would have one company that you could work with from the start all the way to the very end. This saves a lot of headaches and reduces timelines because while I’m working
on my piece, I’m already talking to the next group about the next piece of work, and the next piece of work, and so on and so forth. So everything is lined up, planned in and scheduled in advance to
ensure a smooth transition along each step of the project.

RL – The only other thing I’d add here are the benefits in terms of traceability and data documentation, particularly if the aim is to go into GMP manufacture. There are also benefits to having ownership of the complete process within a single organization, particularly when it means that innovations can be applied across the company and to different stages of your project workflow. It means that we can apply lessons learned on company-wide basis.

Enjoying this interview? Download the full eBook, here >>

What are the next challenges that need to be overcome?

RL – I think the main challenge that we are faced with now regarding CAR-T cell therapy is that while effectiveness has been demonstrated quite convincingly for hematological cancers, there is still a lot to be demonstrated in terms of how suitable this form of treatment is for solid tumors, and there are a few different reasons why.

One reason has to do with the different ways that cells within tumors will express antigens, and if there’s heterogeneity among the tumor cells as to how well, if at all, they express the target antigen, it can make the tumor impenetrable to early generations of CAR-T cell therapy.

The latest generation of CAR-T cells – 4th generation CARTs – involve the expression of transgenes for the secretion of signals aiming to recruit the innate immune system to the site of CAR-T binding and exert their own effects, which may help to overcome this difficulty with solid tumors.

RC – I think the challenges can be really divided into two areas, and one is the area that Robert was alluding to, considering how to treat intractable diseases using those standard early generation CARs that aren’t working in the way that they are for the hematological conditions.

The other area, which is the area I work on very often, is the manufacturing side. How do you make them more cost effective? How do you solve the challenge of taking patient cells from the hospital, to the manufacturing and testing sites, and then get them back in a very quick turnaround time? Or, how do you transition from an autologous product to an allogeneic product, which might solve some of these difficulties, but would present alternative challenges instead.

Enjoying this interview? Download the full eBook, here >>

What does the future hold for these cell therapies?

RL – There are a few new innovations that we might expect to see. There are now bispecific CARs which can take different formats. This is where you aim to target two different antigens expressed on the surface of the same tumor cells, which gives the CAR-T a level of redundancy such that binding to either target leads to cytotoxicity and tumor death. This may help make CAR-T cell therapies more successful for tumors with heterogeneous antigen expression on the cell surface.

Similarly, tandem CAR constructs also have two scFvs, but instead of being expressed separately, they’re linked together so they can have proximal effects, again potentially making the CAR-T cell therapy more powerful against tumors with heterogeneous antigen expression.

Inhibitory CARs could target immune checkpoints, and there are opportunities, I think, to combine many of these different innovations. That’s the sort of thing I would expect to be seeing
in the next few years.

RC – I think Robert made an important point there; now we’ve seen success with individual CAR-Ts, we need to look at what other weapons we have in the arsenal that we  can combine with CAR-Ts to provide therapeutic benefit for those diseases that are currently proving really challenging. For example, combining CARs with CRISPR technology; perhaps there are genes you can knock out in the T cells when you introduce your CAR that can improve efficacy? Or can we express more than one CAR? Can we do some further optimization to improve the interaction of the CAR-T cell with the tumor’s immunosuppressive environment so that the CAR-T cells can be more active? I think we’re just starting to scratch the surface.

Obviously, this isn’t a quick fix. Clinical trials take a long time. But over the next 10 years or so, I expect to see a lot of clinical data showing what works and what doesn’t. Accompanying that
will be a major transition in the way in which we manufacture CAR-T cells and deliver quality products to patients, which will both reduce timelines and improve efficacy.

Enjoyed this interview? Download the full eBook, here >> 

This article has been produced in partnership with WuXi Advanced Therapies