0:00
And we will have Sebastian Warth with us from Charles River talking on leveraging humanised mice, developing your therapies.
0:10
Thank you, Sebastian.
0:11
Yes, thank you for the introduction.
0:12
Yeah, so there has been a swap in this slot.
0:15
So if you are looking for the talk of my colleague talking about in vitro immunology, she will be talking just after the break in room 8.
0:24
So that's Elsinore flavour.
0:26
Yeah, so, and I will talk about leverage, how to leverage humanised mice to develop immunotherapy mostly in oncology.
0:33
But yeah, certainly some of these principles can also be taken to autoimmunity or inflammation.
0:41
So that's Charles River's mission.
0:42
And Charles River has acquired many different companies in the past and now has the ability to offer an integrated multidisciplinary drug discovery expertise and has capabilities and many targets, platforms and therapeutic areas.
1:00
So we have depths and breadths in science.
1:03
And what I find very important is that we work in a true collaborative approach with clients.
1:08
So that like first point of contact would be a business development representative that you may even know already.
1:17
And he will work with our subject matter experts to guide you to the right place in Charles River because we are so such a vast company in our services.
1:29
And then you will work with a dedicated project manager, a client manager who helps you to find the right model and set up the model.
1:35
And then as an overarching party, we have our scientific advisory service who helps you guide through your whole programme up to the safety studies and even planning for phase one and two studies.
1:52
Yeah, so as I mentioned, Charles River has invested a lot $4 billion over the last 10 years in acquisitions, co-authored multiple presentation publications.
2:04
And what I find very interesting is that at some point in the process, we were involved in over 80% of the FDA approved novel drugs over the last five years.
2:15
Yeah, this is the Charles River map.
2:17
So using one of these lines, you can get to basically any point in your drug discovery process.
2:25
And I will focus today on the in vivo pharmacology.
2:31
Yeah, in vivo oncology aspects of this.
2:35
So Charles River Oncology is located in North America in the Shrewsbury site just next to the safety side.
2:44
And then our CNS side in Finland that has a lot of imaging capabilities also extended their portfolio to oncology now and then.
2:55
Our original oncology site is in Freiburg in Germany.
2:59
This is where I'm located as well.
3:02
And here we have all our PDX models and we have the 3D centre of excellence.
3:07
So this is what we offer in Charles River Oncology.
3:10
It's at its core are our in vivo models, which would be PDX models, so patient derived xenograft models and CDX models where we have a really great number of them.
3:22
But we also offer syngeneic models.
3:24
We nearly acquired paediatric cancers, which is, yeah, hard to find, but sought for indications.
3:34
And we can offer our models typically in the subcutaneous manner.
3:38
But we also offer other topic implementation using for example, stereotactic implementation into the brain.
3:46
And we can use disseminated models.
3:49
And of course we have imaging capabilities to monitor cells in vivo and also have a number of labelled, many labelled models also.
3:59
But you can also do some sort of standard studies, efficacy PK/PD studies.
4:04
We can also do GvHD studies.
4:07
And now we can also do live PK studies.
4:10
So to support these in vivo programmes, we have also 3D and 2D capabilities that we're expanding all the time.
4:17
And I will circle back in my talk between in vivo and in vitro capabilities that we have.
4:23
Of course, if you want to run these studies, you have to have the analytical capabilities as well.
4:27
So we have a full flow cytometry unit.
4:29
We have Histology on site, which can be very helpful.
4:32
I will show you an interesting example in a minute.
4:36
Yeah, then we can of course leverage all that we can do in Charles River and offer pathological evaluation and PK analysis etcetera.
4:45
So just quickly cell line model, cell line, direct xenograft, but we tumour cell line that you all probably know and then it would be engrafted into an onto an immunodeficient mouse subcutaneously.
4:56
And then the tumour was grow on the mouse.
5:00
And it at some point if you have like of course you implant many mice and then you can recruit into your treatment or control groups and then measure tumour volume.
5:13
If you're working in a subcutaneous setting, measure tumour volume or overall survival and you can see some example data on standard of care treatment.
5:24
Now for a PDX model that's a patient derived xenograft, meaning that primary patient material is implanted onto an immunodeficient mouse and then it's grown until it reaches like termination size and then this tumour model tumour is explanted and cut into pieces and then re implanted into mice.
5:45
So it's propagated only on mice.
5:47
So it has never seen plastic in a sense that is it has never been completely dissociated if it's a solid tumour.
5:57
And yeah, so in that way, the PDX model retains the correct characteristics of the original tumour.
6:05
So in that, for example, here we can see that the tissue composition is here on the right is a breast cancer model.
6:11
So it still has this lumina in there.
6:16
So PDX model models are difficult to work with because they are diverse, have they are heterogeneous.
6:25
But this is something that you want to have because patient tumours are also heterogeneous.
6:30
And this is why it's considered to be the gold standard in Preclinical Research.
6:37
And of course, if you want to, if you're looking for a specific target, you need to have the molecular data available so that you can tailor select the tumour model that you really need to have your therapy.
6:49
And this is also a big advantage, I would say of a cell and derived models because yeah, we just have the whatever we found in the patient and this PDX compendium, like you can access that.
7:01
It's a publicly available data through our Charles River compendium.
7:08
Yeah, now you're here because I'm talking about leverage humanised mice.
7:13
So what is a humanised mice for me?
7:16
I mean some like in some settings the humanised mice would be a genetically engineered mouse model where human genes are knocked in.
7:26
But today I will be talking about humanised mice in a sense that we use an immunodeficient mice that has been transferred with human cells and then carries a CDX or a PDX cancer model.
7:40
So of course the approach with a knocked in genes has some advantages because you work with a full immune system and you work with a mouse tumour.
7:49
But as I said, it's a mouse tumour and most of our agents that we're working for are specifically for human targets.
7:56
And This is why we believe that the these humanised mice are really important because you have a human tumour model and you have human immune cells and still it's a model.
8:08
So it will never have a full immune system such as a mouse model like a pure mouse model could have.
8:14
So I will focus the rest of my talk on this mice.
8:18
So when would you use humanised mice approach?
8:22
It's when you work with human specific agents that mediate activity through immunomodulation of targets that are expressed on human immune cells or on human cancer cells.
8:34
That could be for example human specific antibodies like checkpoint inhibitors.
8:38
We've tested that a lot.
8:40
And then we also work with bispecific antibodies and T cell engages and what you can test assume an immune cell function for example on T cells and K cells, gamma delta cells, that's all.
8:53
Yeah, we have all experience with all the all of these.
8:56
We can select activation, cytokine production, ex vivo functions.
9:00
And now we also have fact sorting capabilities.
9:03
So can also look into gene expression on a single cell level.
9:09
We have kind of our own CAR T cells.
9:14
So if you need them in a reference project, these are also available.
9:18
And of course we can use all these humanised models in the subcutaneous, which is the most simple set up and then also an author topic or disseminated fashion manner.
9:32
So now getting into the models and showing some data, this most simple approach from my point of view would be to use the humanised T cell models where tumours are implanted onto immunodeficient mice, just as I said before.
9:46
And then we would start a T cell culture starting from PBMCs and then would inject the expanded T cells into mice.
9:56
So we have all these expansion capabilities.
9:58
But I made this red box here in the bottom because you can say, yeah, well, it's a T cell culture.
10:06
It's not very interesting, but can be very specific like which donor do you select to expand your cells from?
10:14
Will you be able to recall it in a follow up experiment and how long would you expand them or would you start even with T cells that are expanded or even send us the tell the cells so that we use them just from you.
10:28
So this would be a typical efficacy layout.
10:31
We would use 9 miles and then treat basically with vehicles or no T cells.
10:37
Then include the T cells with vehicle and include the T cells with the compound.
10:42
Why 9 mice?
10:45
This is because we use three donors.
10:47
So 3 animals per group for donor 1, 3 for donor 2, 3 for donor 3.
10:52
So to build like to account for the donor variability.
10:57
And this is an example where a liver cell line was tested, our liver cell line model was tested and you see that with T cells alone, I think that's the blue line, you don't have any response, you don't diminish the tumour growth.
11:16
However, if you use them the, in this case a bispecific T cell binding compound, you'll see a very strong response.
11:24
And this is true for fast growing model on the left and also a slower growing model on the right.
11:33
Then people often time ask us, yeah, can we do something in vitro to derisk that if we want to work with the T cell model.
11:40
And yeah, we have tested that for pacanalotamab in an L363 multiple myeloma model.
11:49
And you can see that there is a very nice correlation and like cytotoxic killing assay with a pacanalotamab concentration and two donors.
11:58
So there's still a little offset between the donors, but yeah, trend is very clear.
12:03
And we can also do that in an LDH cytotoxicity assay, which can also work without having a label on the cells.
12:12
And yeah, this licence correlates.
12:15
So from that we would say it's fine to go in vitro, sorry in vivo, but it's still we have the difficulty with the donor in vivo whether it will engraft and how it will response to the therapy.
12:28
And we also sometimes see that the that we see in response to T cell therapy or yeah, T cell therapy alone.
12:38
So all the things to be discussed and to clarify upfront.
12:44
So the in vivo response is donor dependent.
12:48
Now we're moving one step further.
12:50
Can we can also humanise with modified T cells or CAR T cells would be a very simple model where tumours again explain implanted onto mice, then CAR T cells are thawed and injected.
13:06
And then here you see in the breast cancer model with using HER2 CAR T cells, you see that with two different doses that's 2 times 10 to the six cells or 1 times 10 to the seven cells.
13:20
You see a very strong response.
13:23
Yeah, not so much dose dependency to see here compared to trastuzumab in the upper panel.
13:29
So this is all fine.
13:31
Then we tested another HER2 expressing breast cancer model which is the OE19 cell and you see flow cytometry of HER2 expression in the little upper panel, but there is no, yeah, there's a slight growth delay, but CAR T cells don't work on these.
13:48
How can that be?
13:50
So we leveraged our histology capabilities and here for in the upper panel, you see that in the JIMT-1 model, there's HER2 expressions throughout the tumour.
14:02
But if we look into the OE19 model treated with a vehicle, you see that large parts of the tumour do not express HER2 anymore in vivo.
14:14
So that can also happen that a cell line model that has always been known to express HER2 suddenly in vivo differentiate so that only part of the cells would express HER2 anymore.
14:26
And if you look on the very bottom right panel, this is the CAR T cells.
14:30
So they're perfectly cleared all HER2 positive cells.
14:35
Yeah, but they do.
14:36
They couldn't clear the tumour as a whole, of course, because other cells didn't express that.
14:41
Now didn't that may seem like a failed model in some sense, but you could think whether this could be even a very helpful model if you're looking for NK cells for example that are yeah conferring functions regardless of the target.
14:57
So they are for example, with a bispecific antibody could be activated, but then work on the bystander cells which are antigen negative.
15:10
Then of course it would.
15:12
It may help to evaluate that in vitro a little bit and I will.
15:17
In this experiment, we generated tumour specific T cells by just cultivating PBMCs with the tumour cells and then harvesting them at about 14 days.
15:29
And then at the same time we would spike in CAR T cells, the ones that I just told you and then also use CD8 cells that have not been educated on tumour cells.
15:41
And then we measure in 2D and in 3D the response.
15:45
And what you see here is IncuCyte data, so life-cell imaging where the red curve is the tumour cells, how they would behave without any T cell culture.
15:56
And then CD8 cells would be the blue curve.
15:57
So they'd confer some effect in vitro.
16:00
And then these antigen, sorry, no, it's tumour cell specific T cells, they confer a little bit less of an effect.
16:08
And CAR T cells are very strong in the effect.
16:10
That's the green curve like that was expected.
16:14
And we can also do that on spheroids that we can generate here on the that's again breast cancer cell lines.
16:22
So JIMT-1, SK-BR-3, and MCF-7.
16:25
And like you could think it's of course it's breast cancer.
16:30
It must be HER2 the target.
16:32
But we tested like T cells that have been generated on the MCF-7 cells and added them to the SK-BR-3 cells.
16:44
And in fact, what you see is that they confer less of a killing effect.
16:48
So sorry, I cannot use the pointer, but this means that it's obviously not only one target, but what we would have expected, there are multiple targets on the tumour cells.
16:58
And yeah, so we only get a partial killing effect.
17:06
And it's also true in the 3D setting.
17:10
So now I want to switch a bit to the CD34 humanised mice.
17:16
What is this?
17:16
So it's mice that have been lethally irradiated and then they will be transferred with CD34 positive hematopoietic stem cells.
17:26
And then these cells that's even before tumour implantation, these cells would differentiate them into all this progenitor cell lines, including T cells, including myelin cells, etcetera.
17:41
The development of all these compartments will depend a little bit whether the supporting cytokines are there.
17:46
So in which mouse model this is implanted.
17:49
But in any case we will have T cells and these T cells will be will have been selected in the mouse.
17:55
So they will not be autoreactive to the mouse.
17:59
And then in these mice, the we would implant the tumour and the example that I want to show you is where we combine this with the specific format that we offer, which is the single mouse trial where we only have one mouse per treatment group for one tumour.
18:20
So with the idea that we can look at the population of tumours, because when you have a clinical trial, you only have one tumour, like one patient, one tumour.
18:36
And we tested this with checkpoint inhibitors and then we set out to group the tumours based on what we see here in histology, and we did we performed anti PDL-1 staining.
18:52
And you can see on the right side are hot tumours that in these CD34 mice throat infiltration of immune cells compared to cold tumours which have no infiltration of or below 5% infiltration of immune cells.
19:10
And then we want to test them with checkpoint inhibitors.
19:12
So yeah, this is the response data, but I want to I will come back to this in a second.
19:19
But what I wanted to show you is that here in the bottom panel on the right side in these humanised mice, you see that the in the hot tumours, the PDL-1 expression is much higher than in the cold tumours.
19:34
So this could be that the tumour is counter regulating against the infiltrating immune cells, which is not necessary in the cold tumours.
19:43
But now that if we look at the response, yeah, here I've plotted all the different mice responses.
19:49
That would be the data that you also face in a clinical trial where you have all kinds of responses.
19:57
And then you see in the cold tumours here and the grey dots you see that the that you see a slightly stronger response with PDL-1 inhibitors which was.
20:09
Kind of surprising for me, I thought it would have we would see a stronger response in the hot tumours which where we have also immune cells there that are reactive and that could be disinhibited by PD-1 or therapy.
20:25
Now, this approach is obviously very, yeah, time consuming, difficult.
20:30
We have to solve a lot of PDX models.
20:32
So can we take this in vitro somehow?
20:35
And to achieve this, we started to develop lung cancer tumoroids from PDX tissue.
20:42
It's like we explanted tumours from these mice and resected them to make them, yeah, to have a primary cell suspension.
20:52
And then with two different expansion protocols and testing a lot of different media, we ended up to have tumoroids on.
21:01
You may also call them organoids if you like to.
21:04
I'm not sure there's a specific definition for that.
21:07
So but we wanted to make sure that these tumoroids like that are, that they are really used for it.
21:15
We tested several conditions and most conditions did not end up with such nice results, meaning that here we see like lineage specific markers expressed.
21:26
This was tested by immunofluorescent like multiplex immunofluorescent.
21:36
And yeah, we maintain here the balance in epithelial to mesenchymal states in these tumoroids.
21:43
So we are very confident that they like really represent the PDX model that they are derived from.
21:54
And then we tested that for the in vitro response and we see that, yeah, with a standard of care agents, we see also a mutation specific response.
22:09
So that all made sense.
22:10
But the critical thing is this also correlates with the in vivo response in these PDX models, which is something we not necessarily see that the in vitro response predicts for the in vivo response.
22:26
So what I find very interesting, these tumours comprise hot and cold tumours.
22:32
And when you Co cultured them with NK cells as a first step to go in the direction of IO, we saw that the we had a very strong response of NK cells to this hot tumours.
22:43
Like coming back to the idea that even if the target may not be expressed, NK cells can still be active against the tumour because you have some yeah, immune cell function present there.
22:56
So most importantly, we can buy a biobank these tumoroids and so we can use them for a future study.
23:03
So we have established that for a number of lung models and we want to expand that of course.
23:08
And yeah, the next step would be to use that IO.
23:12
And yeah, we see tumoroids as in vitro avatars for PDX models.
23:16
So last, very last step of bit of data is talking about NK cells.
23:22
We can also adoptively transfer NK cells, of course, very similar setup.
23:27
And here I put in this example because it shows how you can use established players like NK cells and then anti CD20 antibodies like obinutuzumab in this case and combine them to see whether they can leverage the response.
23:45
Until you see that with NK cells alone, that's the blue line, you'll see pretty low response.
23:53
And there's a register model, B cell and former model with obinutuzumab, you'll see quite a good response.
24:01
But if you combine the two, you have a very strong response, like T cells, sorry, NK cells.
24:07
And these NK cells were given weekly.
24:09
So this is also something to consider when we working with NK cell, it's always a little bit difficult.
24:15
So coming to my conclusions, I showed you examples of humanised models that we can do.
24:20
It's not all what we can do, but a good snapshot I would say.
24:26
And then we have PDX models here that I introduced in the beginning.
24:30
We have a quite a number.
24:31
Then we can use humanised mouse models and we have immunodeficient strains and all of these are forming the combinations that we need to sort out.
24:40
So in theory we have all of our cancer models are available for humanised approaches.
24:46
But yeah, and mine, as I see it, most humanised models are a combination, Most humanised models are unique.
24:54
So they're custom models are specified models.
24:59
So this would be a true collaborative approach that we need here to find the right model for you.
25:06
And with that, I would like to end happy to answer a few questions.
