Novel Models for Bladder Cancer Progression and Drug Development - John Taylor III

June 5, 2023

Ashish Kamat welcomes John Taylor, III to discuss the importance of preclinical models in advancing medical innovation. He details two models based on the N-Butyl N(4-hydroxybutyl) nitrosamine (BBN) model, which is a reliable model for bladder-exclusive tumors. Taylor speaks about the benefits of the BBN model in studying muscle invasive bladder cancer, yet also underlines its shortcomings in studying non-muscle invasive disease and metastasis. He explains the connection between mitochondrial DNA and metastatic progression. In an exciting development, he outlines the usage of the mitochondrial nuclear exchange mouse (MINX) and Down Syndrome models in bladder cancer research. Taylor emphasizes the potential of these models in shedding light on the mechanisms behind metastasis and in the discovery of new therapeutic targets. He concludes by recognizing his colleagues and reiterating the relevance of the models in clinical settings.


John Taylor, III, MD, MS, University of Kansas Cancer Center, Kansas City, Kansas

Ashish Kamat, MD, MBBS, Professor, Department of Urology, Division of Surgery, University of Texas MD Anderson Cancer Center, President, International Bladder Cancer Group (IBCG), Houston, Texas

Read the Full Video Transcript

Ashish Kamat: Hello, and welcome to UroToday's Bladder Cancer Center of Excellence. I'm Ashish Kamat, Professor of Urologic Oncology at MD Anderson Cancer Center, and it's always a pleasure to have Dr. John Taylor joining us on this forum.

John's a dear friend of mine, and for the purposes of this presentation, I'm going to formally introduce him. Dr. Taylor is the John W. Weigel, Professor of Urology and Cancer Biology, Deputy Director at the Institute for Advancing Medical Innovation, Co-Leader, Drug Discovery Delivery and Experimental Therapeutics, and the Director of Translational Research, in addition to being a dear friend and an expert in the field at the University of Kansas.

So John, before I use up the whole presentation time just introducing you, let me hand the stage over to you, so you can talk to us about the novel models for bladder cancer progression and drug development.

John Taylor III: Ashish, thank you for having me. It's always a pleasure to be with you in any format, and I appreciate the opportunity to sort of review a little bit of what we presented at the AUA.

So I'm going to be talking to you about some novel models that we are in the works of developing to help with both the evaluation of metastasis in bladder cancer, as well as a potential novel model for drug discovery that has not been identified or used before.

So I thought I'd give you a little bit of a background on preclinical modeling. As we all know, the purpose of preclinical models is to inform the development of both novel diagnostics and novel therapeutics. These typically proceed through animal modeling to both confirm the in vivo relevance to a disease or disease state, as well as to inform decisions on moving to human trials.

And the FDA has recently made a statement that they're trying to move away from the animal modeling, but most of us that work in this field feel pretty comfortable that this will be around for some time.

The two models we're going to talk about are based on the BBN model or N-Butyl N(4-hydroxybutyl) nitrosamine model. This is a very old model that was first described in 1957, and around 1973 to 1976, to be modeled in black six mice as we currently use, was defined, and it hasn't changed all that much since then.

The good thing about this model and why it's considered a validated preclinical model which the FDA recognizes, is that the tumors that develop are exclusive to the bladder. There have been some reports of hepatocellular carcinomas developing in the BBN model, but they're exquisitely rare and far between.

These tumors develop de novo. You don't have to manipulate the animal. There's no viral antigen which drive the tumor genesis. This is as natural as it can get. These tumors arise in an immune intact environment, which is important, as we know, as the development of immuno-therapeutics and immuno-oncology is exploding.

The model is one of progression. It typically proceeds through significant submucosal inflammation through dysplasia carcinoma and site two to large muscle invasive tumors.

It is not a reliable tumor for non-muscle invasive disease, predominantly due to the variability of timing after the anterior rescues group looked at some of these early tumors and found that they didn't necessarily mimic human tumors, but that there was high fidelity in the muscle invasive state.

Also, notably, it's not a reliable tumor from metastasis. These mice will die of obstructive uropathy perhaps due to tumor burden before they develop metastasis, and obviously, animal facilities and care folks don't like that. It's not great for the mice. If you let them go long enough, they will possibly develop a metastasis but not reliably.

The most important thing is that this model has been molecularly credentialed to be phenotypically and genetically similar to human basal subtype of muscle invasive bladder cancer.

Both the anterior rescue group and Josh Meek's group have shown this. This is just a heat map and some data on Josh's paper in 2018, but the model is well characterized and considered valid today.

What we know about metastatic progression is that mitochondrial DNA in metastatic progression is important. Metastasis is related to intrinsic genetic factors, not only in the tumor, but also stromal factors in the tumor microenvironment.

Mitochondrial DNA encodes 13 critical proteins related to oxidative spot correlation, as well as some ribosomal and transfer RNAs. Interestingly, mitochondrial DNA, single nucleotide polymorphisms are shown to represent quantitative trait loci, which are not necessarily mutational drivers in and of themselves, but are known to cooperate with other genes that regulate complex traits such as longterm metastasis.

Murine models have consistently shown that mitochondrial DNA contributes to tumor genesis and metastasis. The most common findings have been decreased apoptosis for increased mitochondrial oxygenation species signaling.

The current tumor models for metastatic progression that involve mitochondrial DNA suffer from some drawbacks. Cybrid models require mutagens and/or drugs which can have problems in and of themselves, and the cogenic models require extensive backcrossing, which can result in nuclear DNA crossover and contamination.

My colleague, Danny Welch, was part of the team that developed the MINX mouse or the mitochondrial nuclear exchange mouse, and they did this through isolation of an embryonic pronuclei with implantation into the nucleated embryo of the opposite strain, and this generates a reconstructed zygote. That allows for testing the mitochondrial DNA haplotypes and how they contribute to pathologies and cancer cells, as well as tumor stoma.

Several models using the MNX mice in breast cancer have shown that the mitochondrial DNA consistently impacted the metastasis efficiency of the tumor cells. They noticed that when there was Balb C or C through H mitochondrial DNA, there was increased efficiency.

Actually, interestingly, they showed that the C 57 mitochondrial DNA component reduced the efficacy of metastasis, and this held true when they controlled the tumor growth rates and/or mets.

Other models using the tail vein metastasis model with syngenetic metastatic lines in breast as well as melanoma showed that the stromal mitochondrial DNA is also impactful, regardless of the nascent tumor DNA. And again, they showed that decreased metastatic efficiency was shown when there was mitochondrial DNA components from the C57 mouse.

In bladder cancer, it's been shown that mitochondrial DNA loss in mutation can increase the aggressiveness of tumors. However, we really don't understand why this is, and we currently don't have any reproducible models to study this.

So we undertook a pilot study using the MNX mouse and the BBN model, and we were able to show that in the C3H wildtype mice, the males had significantly larger tumors and the female's, interestingly lost their sex protection, and we're not quite sure why that is as of yet.

When we had the crossed mouse with the C57 mitochondrial DNA, we saw an overall decrease in the tumor burden and, interestingly, near restoration with sex protection in the female mice.

We didn't see overt differences in the metastatic efficiency in this pilot study, and we think that that's due to several reasons, predominantly due to the small number of mice, as well as the early timeframe that we took them down.

We're currently funded to go back and look at this in a larger model, again, using the C57 and 3H MNX mice and the BBN model. And we're going to do a deeper dive into the analysis of metastasis and we'll hopefully be able to show in the lung, liver and lymph nodes by RNA seq analysis. We'd love to do more, but unfortunately, the funding will only support that.

We're also going to begin to evaluate the stromal contribution of the mitochondrial DNA using a tail vein model with single cell suspension of tumors that are developed in the BBN mice.

So that's a pretty quick overview of the MNX model. We're going to move on to the Down Syndrome model for the novel drug discovery in bladder cancer. And this was really borne out of an interesting conversation between several folks in the Down Syndrome world as well as the NCI and ourselves from KU.

And it was postulated that perhaps in a carcinogen model, we would see lower tumors as we find from epidemiologic studies in these patients. As we all know, Down Syndrome or Trisomy 21 is the most common chromosomal anomaly found in humans. While these patients are shown to have higher rates of both bloodborne and germ cell cancers, or particularly at an early age, they statistically have lower rates of solid tumors as they age.

And several epidemiologic studies have shown, in particular, that there is significant increased risk of dying from urological cancers. When we look specifically at bladder, the relative risk of death in bladder cancer in the that that population is 0.27, which is highly statistically significant, as you can see here.

Some of the early thoughts on this were that perhaps this was due to early mortality. Down Syndrome patients typically only lived into their 20s and 30s through the 1950s, '60s and '70s, but the life expectancy for most patients with Down Syndrome today is around 65. And several age adjusted studies looking at specific incidence rates in this population have challenged this notion.

The second question was that perhaps this is due to their lack of smoking or environmental risk or environmental exposure to potential carcinogens. And again, several large epidemiologic studies have shown that this is likely not caused because there's a reduction in cross spectrum of tumors that are related not only to carcinogen exposure, but also those that were a result from hormonal issues as well.

Putative mechanisms that have been put forth for why patients with Down Syndrome might show less solid tumor growth than the aging population are several, and they include decreased angiogenesis due to increased gene dosage of the Down Syndrome critical region one, which is a known anti-angiogenic factor working through VEGF, that there is increased tumor suppressor in numbers on the Trisomic chromosome 21, such as the ETS2 transcription factor, which induces apoptosis through p53.

It's been noted that Down patients have an aberrant and highly recurrent gene specific changes in DNA methylation patterns, which may suppress a certain gene expression, that's subsequently related. This may be one of the reasons as well.

And while all of these are good, and there are some very good scientific background to it, there is evidence for and against each of these, and I think everyone in the field would agree that there is more involved from what they're using to identify it at this time.

There are several models of Down Syndrome that exist in mice that are readily available. The Ts65Dn mice is available from JAX. This was one of the first models. It's trisomic in the mirroring chromosome 16 and a small portion of 17, which are analogous to human chromosome 21. These mice are phenotypic behavioral as well as expressed learning deficits which are consistent with human Down Syndrome.

A more recent model, which is felt to potentially be a little more representative of what we see in humans is the TcMAC21 mouse, which is a non-mosaic of Down Syndrome. This incorporates a freely segregating human chromosome 21, which codes for 93% of the proteins on that gene, with no mouse protein coding gene, without a human orthologue, and no additional trisomic regions. This mouse is also shown to replicate Down Syndrome phenotypically.

There's some novel models which are coming out which are trisomic only portions of chromosome 21. What these will hopefully allow us to do in future studies is to really get granular and dissect down to which portions of the chromosome are protected in development of solid tumors.
So we undertook another pilot study, again, using the BBN model and Ts65Dn mice. And what we are able to show in this small pilot study was that indeed, the former models of the Down mice have reduced tumor volumes, with the trends in lower stages.

We went for low-hanging fruit to look at whether angiogenesis was the potential or contributing factor to this. We found no specific change in angiogenesis, suggesting that at least in bladder cancer, there's another mechanism with the bladder.

And we did RNA-seq with reactome pathway analysis and found consistent changes across the board might take place. The most chartered role and the most easily identifiable gene that we were able to see was decreased in AURXB, or KB.

AURKB is a protein that's critically involved in mitosis cell cycle, but it's a protein whose activity can be followed with a biomarker like phosphorylation histone H3. We used that as evidence of its activity. When we looked into the literature, we found that this was over-expressed in other cancers, and typically damages to cells when it is over-expressed.

We went to this TCGA database and we looked at the available data on bladder cancer and found that this was over-expressed in the database as well. We felt pretty comfortable that this was a solid plan.

Fortunately, for what we were doing, there was a commercially available inhibitor of AURKB, AZD1152, and we used that in cell culture and xenograft models, and we were able to find in multiple cell lines that just have representative HTB9 cells up here, that when we treated cells with a AZD1152, we had significant increase with decreased phosphate H3 as a marker of activity. We found nanomolar SC50 ranges in proliferative indices suggesting that this is something that could be translatable into a drug.

And when we used this in a xenograft model, a flank model, we found consistently decreased tumor volume in weights. I think one of the important things that I want to emphasize about, at least the Downs model is that we're not specifically saying that that AURKB and Barasertib is the answer here. This was a proof of principle model. We wanted to see, could we take an epidemiological finding, reproduce it in a validated mouse model of bladder cancer, and that could we then use that to explore differences in gene expression to identify potential targets for drug development?

So what I've hopefully done in this brief overview of what we're doing is to describe to you two novel mouse models in bladder cancer that we're working on developing, the MNX or MNX mouse that holds potential for a reproducible model of metastasis. And given that mitochondrial DNA is a recognized regulator of the metastasis efficiency, this will hopefully allow us to be a little bit more in-depth in our dissection of what is causing the metastasis. As you know that from a prognostic standpoint we identified the presence of metastasis is the significant finding.

The Down Syndrome models, I said there's a discovery of targeted pathways for drug development. Again, we're not saying that Barasertib or AURKB is the answer, but in a first pass analysis, we showed proof of principle. We can not only show that Downs mice develop smaller tumors, but we can show that when we can set the reason behind that, we can find target-able entities which might allow us to identify pathways or molecules that haven't been considered before.

I just couldn't really, whether I'm presenting in person or here, I have to recognize the folks that work with me, the brain trust behind what we do; Danny Welch, who developed the MNX mice, the MNX mouse, and works closely with me on that model, and we share the R20 together. And Scott Weir, who worked closely with me on the Down Syndrome model, Ben Woolbright in my lab, who has just transitioned to Independence, and I think will be a force in the field in years to come, and then Binh Vo, who was a remarkable medical student going into urology who did the yeoman's amount of work on the Down Syndrome model.

Ashish Kamat: Thanks so much, John. Thanks for distilling your talk into a concise 10 minutes.

Some of the things that come up when we look at models for bladder cancer are, of course, what's most applicable to the clinic and our patients? And with all the work that you've done in modeling bladder cancer in general, for someone that might be looking to get started in the field, and look at maybe one or two models that he or she can start developing or working on the lab, what would your general advice to that person be?

John Taylor III: I think that there's a consistent way to go about it. Obviously, and I come from the drug development, where we really have a specific and driven pathway of in vitro bench work, and does the compound or molecule that you're exploring have significance in bladder cancer cell lines.
But I think early proof of principle models for drug development and/or other types of analysis, I would start with simply a flank model with a mouse one. Again, there's some limitations and drawbacks to that, specifically a lack of immune system.

If you want to study human cell lines, mouse cell lines are typically difficult to deal with. We all have a distaste for the MB49 cell line. That's an aggressive and horrible line to work with and there aren't many others.

But again, I think a xenograft blank model gives early indication of activity and it's a short model. So remember, animal models are expensive, and you're usually trying to get to an endpoint, and usually that's translational science.

So I think a flank model, a xenograft model, gives you a four or five week answer as to whether it's worth for pursuing further. Once you start getting into the BBN model, which is considered the gold standard, but I think at least in bladder cancer modeling, it's time-consuming and it's expensive. It's four to five months of tumor development. But I think if you can get an answer out of a BBN model, you've got a lot to hang your hat.

Ashish Kamat: Great. And what about the transgenic models? Any experience and advice there?

John Taylor III: Yeah, so we use a lot of this. So you're talking the GEMS?

Ashish Kamat: Uh-huh, yeah.

John Taylor III: The genetic engineered mice model. So those are great models, and I use a lot of those to manipulate gene expression. So you either have a full body knockout, or you can have Cre-Lox, floxed driven knockout, you can have urothelial specific knockout. There are several things... Several mice out there, that are used; there's a lot of work on that at KU. He was kind enough to share some of his mice with us.

Once you get into that, you're getting very granular, and you're also increasing your cost, but they're very valuable. And a lot of grant critiques that I've seen are, "Well, you haven't done the knockout model yet." So I think they're valuable. They add to the data. Certainly, if you're putting together an IND package, it's nice to have GEM model data as well as de novo model data, if you have an inhibitor. So it all plays in, and I think they're all really important.

Ashish Kamat: And then last question, John, if you could share your thoughts on the human xenograft modeling for bladder cancer, because there's a lot of folks that still believe that if you take fresh tissue from patients, or even human cell lines, and implant them into the bladder of mice or develop metastatic disease, that's a more true representation of what happens.

On the other hand, there are folks like yourself and, of course, even I'm partly in that camp that believe, "Well, that doesn't allow you to actually understand the whole progression from early to late stages because you're selecting a particular tumor at a defined stage."

So I know you've done a lot of research and you put a lot of thought into that. At a high level, for our audience, what are your pros and cons for the orthotopic or the transplanted models?

John Taylor III: Yeah, yeah, I think, again, you're talking about cost and you're talking about difficulty. Isolating patients, cell lines and generating primary cultures can be easy, it can be difficult. It depends on what your resources are and who's around you with that experience.

There are humanized mice to try and overcome the xenograft rejection, if you will. If you don't have the humanized mice, you're using nude mice, which now removes the immune related components of the tumor progression and metastasis.

I tend to be a little bit more of a purist, as you were describing. The BBN model is so consistent and so akin to human tumor. And again, I give Dan and Josh a lot of credit for showing that in the work that they've done. It gives you, as you said, a progressive model from nothing, from normal urothelium, to inflammatory changes to CIS, display to CIS early non-muscle invasive tumors and muscle invasive tumors.

The downside of the BBN model is you need to be stuck studying advanced disease. It is just not reliable from non-muscle invasive disease, which is a major part of the research that everybody does.

So I tend to stick with mouse modeling, and we've got great models. I think they offset need for a lot of in complexities of transplanting human cells into a mouse ortho-topically or not. Sometimes I really view, when you do that, you're just really using modified incubator, and you're not looking at the tumor stromal issues because you hold mouse stroma. I think there are just too many issues with transplanting, human cells into mice, that can be overcome just by using this simple validated model.

Ashish Kamat: Right. No, exactly. We could obviously chat on this forever, but in the interest of time, let me ask you; any closing thoughts for the audience that are interested in mouse models in bladder cancer?

John Taylor III: Yeah, I think we've come a long way. I think bladder cancer's come a long way, and you and I have talked about these issues for a long time. Ten years ago, there wasn't really much of anything going, and I think the explosion of research is really exciting, and I think what we've been able to validate, again, if you look at the two papers on the BBN model, they're within the last decade. So I think we've come a long way in our ability to study and model bladder cancer in mice. I think we can still learn a tremendous amount. There's a lot more to do.

Ashish Kamat: Once again, John, thank you for taking the time and spending it with us, and the folks at Urotoday.

John Taylor III: Yeah, I much appreciate it. I always enjoy it.