0:00
Logic modalities including TCRs, genetic medicines, cell therapy, peptides and delivery.
0:08
And we partner across 3 strategic pillars including access to foundational pre competitive drug discovery platform, through access to services to support discovery of therapeutic drug candidates and also through the creations of new companies with our venture Studio 82VS.
0:36
The Alloy ecosystem relies on more than 100 employees across 5 research sites that are located in the US, Europe and Japan, and we work with more than 170 partners, including eighteen top pharma companies.
1:00
The system has been very productive achieving more than 70 active drugs declared by our partners so far and even 10 INDs filed since the first launch of the platform in 2019.
1:23
In terms of in vivo discovery platforms, Alloy offers really an unparalleled diversity of strains.
1:33
Our workhorse is the ATX-GK platform that it's a mouse with a fully humanised heavy chain and Kappa chain.
1:44
It's available in different genetic backgrounds, so BL6, BALB/C and F1 cross and has proved to be very successful across multiple discovery campaigns.
1:57
It can generate a very good coverage of epitopes and sequences, very good cross reactivity and it proved to work with a variety of targets, including difficult proteins like GPCRs.
2:20
To have even more diversity, we also generated a strain called ATX-GL that is a mouse with a fully humanised Lambda light chain.
2:32
More recently, we also launched an hyperimmune strain that shares the same genetic of the GK but expresses additional genes to enhance IgG class switching and improve generation of antigen specific B cells.
2:52
Then through the alloy discovery services, it's also possible to have access to common light chain mice and I will describe those platforms in the next slides.
3:05
But I also wanted to mention that we are actively working on the development of a single domain platforms for the discovery of VHH antibodies.
3:16
So I'm not going to talk about single domain mice today, but if you're interested, please get in touch and we can provide more information.
3:30
So given the audience, I don't think I need to explain why bispecific antibodies are so important and so popular.
3:39
But you are all very aware that they can be employed across multiple modalities and they're very helpful to overcome a variety of challenges in therapeutic context.
3:58
However, despite being very popular and successful, the road to that leads to the development of bispecific antibodies, it's often complex and there are multiple challenges along the way.
4:15
The first one is for sure generating quality compatible antibody components because those are always the starting point of good bispecifics.
4:26
But it's also important to select stable manufacturable bispecific formats.
4:32
And then it's essential to screen for functional and developable leads.
4:42
At Alloy, we tackle those challenges with a three-step approach.
4:46
So the first step is using common light chain discovery solutions to address the heavy chain light chain pairing, followed by an optimised format engineering and then integrated high throughput screening approach to identify functional leads.
5:07
The common light chain mouse it's a well known solution to solve the heavy chain light chain pairing problem.
5:19
You are all very aware that starting from a full diversity light chain transgenic mouse can lead to problems or complications down the line if you're trying to produce bispecific antibodies.
5:40
And common light chain transgenic platforms are an elegant way to overcome this problem and really facilitate the production of bispecific antibodies with high developability profiles and also the validity of a common light chain platforms.
6:06
It's also proved by the number of molecules that are currently in clinical development or that already reached the market.
6:20
So at Alloy, we engineered our own suite of common light chain mice.
6:26
They are all characterised by a full human heavy chain within same diversity of the GK strains.
6:36
And then we selected the light chains based on different properties, including high manufacturability, success of the germ lines in clinical settings, compatibility with a large number of VH germ lines and also structural diversity of the light chains.
6:57
And we came up with three different Kappa CLC mice expressing a single germ line from family one, family 3 and family four and one Lambda CLC mouse from expressing a single germ line from family one.
7:20
And I believe this Lambda CLC is the first Lambda mouse available on the market.
7:29
So we performed an extensive validation of those strains.
7:33
Here is an example of NGS on naive BK3 CLC mice, where you can see that as expected, the light chain usage is restricted to a single germline.
7:51
But on the contrary, the usage of the heavy V genes is not restricted by the common light chain, but it's actually very similar to our controlled GK mice or to human samples.
8:07
Also the heavy J genes usage or the CDR3 length.
8:12
It's comparable to controls and in the range of what you would expect for this type of transgenics.
8:21
We tested those mice also for the ability to respond to antigenic stimulation.
8:29
So in this experiment we immunised our mice with recombinant PD one with a standard 28 day RIMMS protocol including our full diversity GK mice as a control.
8:45
And as you can see, all four CLC strains responded very well and produced robust serum titres.
8:56
In terms of antibody discovery, the Alloy workflow is based on four different steps.
9:07
The first one starts with immunizations of our transgenic platforms.
9:12
They can be either food diversity animals like for example the GK or hyperimmune.
9:19
But for today I'm going to focus mainly on the common light chain strains, but it could be any transgenic platform.
9:27
They can be immunised with a variety of methods, recombinant protein, DNA,RNA or cell lines.
9:35
They all work very well and following immunizations the lymphoid tissues are collected.
9:42
We usually perform B cell enrichment step using different methods like magnetic beads or FACS or even yeast immune libraries really depending on the project and partners needs.
9:57
And this generates 10 of 1000 of reads that are then fed into our proprietary AIML programmes that extracts the best leads that are then expressed and characterised.
10:12
And there is an optional iterative step where the best leads can be analysed again by our AI/ML software to for example, expand clonotypes or improve affinity maturation.
10:39
And this approach proved to work very well.
10:42
So it allows to identify clones that have a very nice diversity in terms of sequence usage, very robust cell binding and good cross reactivity in this case for example with the sign of antigen and also in terms of developability.
11:05
They've been tested across seven different assays and the vast majority of clones have very good or acceptable clinical developability scores.
11:21
Another example of discovery project that was run on common light chain mice basically was against a tumour associated receptor.
11:35
And also in this case, we were able to identify 100 of binding clones with a range of affinity going from picomolar to nanomolar.
11:48
And interestingly, when we're on an epitope binding experiment, we were able to identify several clonotypes targeting each subunit of the receptor.
12:00
And this is really ideal when you want to enable the development of the biparatopics bispecific antibodies.
12:11
The last example that I wanted to share today, it's what we have done with a very challenging target that it's Claudin 18.2.
12:24
This protein has a high homology between mouse and human and also there is an isoform called Claudin 18.1 that is very related to 18.2.
12:39
So Claudin is a transmembrane protein as you can see from the picture on the left.
12:47
And there are only two loops basically exposed in the extracellular domain.
12:54
And if you look at the amino acid sequence on those loops you will see that the homology between claudin 18.2 and 18.1 is very close.
13:04
So it's a really challenging target.
13:08
So we started the campaign with RNA/LNP immunisation using proprietary protocol and that resulted in very robust titres and diverse panel of clonotype as you can see on the right panel.
13:28
And then the output of the immunisation has been used to generate a yeast immune library.
13:37
And then we perform depletion and enrichment steps basically to remove clones that were cross reactive with the 18.1 isotype and enriching for the clone specific for the 18.2 protein.
14:00
And at the end of the process, we were able to identify 61 unique clones specific for Claudin18.2 and 19 of those were in the sub nanomolar affinity range.
14:18
So the next step after identifying your monoclonal antibodies against specific targets, it's obviously expressing diverse bispecific formats with a variety of balances and orientations.
14:35
And this generates hundreds of possible combos.
14:40
And we use functional assay to really rank the clones based on really their functionality.
14:52
Again, this approach proved very successful in our hands because following the reformatting in a bispecific format, we can still detect the original binding properties of the parental monoclonal antibodies, but the same time we can also measure bispecific bridging in both orientation.
15:17
The purity of the bispecific antibodies produced with these approaches is very high as we measured with mass spectrometry in the top right panel and again, we were able to prove that the vast majority of clones have a very good developability score.
15:44
As I mentioned at the beginning, we think that the functional validations of the clones is very important to build confidence in your leads and for this reason our team in Alloy developed a range of in vitro functional assays including different type of reporter assays, cytokine production, T cell engager cytotoxicity but also custom made assays that are can be project dependent.
16:18
And there is an example in the bottom panel when we use the time resolved image-based assay to basically measure antibody internalisation, comparing experimental clones with clinical comparators.
16:36
And it proved to be a very good approach to identify the lead clones to bring forward. So to conclude, Alloy’s bispecific discovery workflow is designed around three major steps.
16:57
The first one relies on our common light chain mice with the support of our display capabilities and machine learning programmes by optimised format engineering and an integrated high throughput screening approach that really allows to provide to our partners, complete that package for lead selections in 6-8 months.
17:35
And I also wanted to mention that our scientists worked on a panel of validated antibodies that can be used as a plug and play for your favourite bispecific or ADC project, really with the goal of expediting your discovery project.
17:58
So if there is any target that is of interest and you want to learn more again, please get in touch.
18:03
And we are happy to provide more information and the final slides.
18:11
I just wanted to say thank you for attending and I'm happy to take questions.
