Read the Full Video Transcript

Andrea Miyahira: Hi, I am Andrea Miyahira at the Prostate Cancer Foundation. Joining me today is Dr. Jindan Yu, a professor at Emory University. Dr. Yu and her team recently published the paper "Chemokine Receptor CXCR7 Activates Aurora Kinase A and Promotes Neuroendocrine Prostate Cancer Growth." This was published in the Journal of Clinical Investigation. Dr. Yu, thanks for joining us today.


Jindan Yu: Andrea, thank you for the opportunity to discuss our work. It's my pleasure to discuss our work that was recently published in the JCI on CXCR7 in late-stage prostate cancer. So we started to work on CXCR7 when we observed that this gene is among the few that are upregulated in enzalutamide-resistant cell lines compared to their parental cells. So CXCR7 is specifically highly expressed in enzalutamide-resistant cells.

This caught our attention and we looked into CXCR7, which is also called ACKR3. It's an atypical chemokine receptor. Better known in the field is actually CXCR4. So CXCR4 is a G-protein coupled receptor that resides mainly in the membrane, and it interacts with G proteins inside of the cell. This turns on the signal transduction cascade that's active with MAPK/ERK and the downstream events, inducing cell growth.

In contrast, CXCR7 is an atypical chemokine receptor, because it does not interact with G-protein specifically, it actually recruits beta arrestin. So through interacting with beta arrestin, CXCR7 is internalized into the endosomes. So they form as a scaffold of protein to mediate the phosphorylation of a lot of kinases, including MAPK/ERK kinase.

Our initial study, actually published in Cancer Research, was to look at CXCR7 induction of ERK phosphorylation as a mechanism to induce enzalutamide-resistant prostate cancer growth. So because of these different mechanisms between CXCR4 and CXCR7, if you do a staining of the cells, you see CXCR7, a lot of those actually localized within the inside of the cell. And CXCR4 is more of a real membrane protein.

So then we continued this study. We looked at CXCR7 across the whole process of prostate cancer progression in the public database. We found that CXCR7 is very highly upregulated in neuroendocrine prostate cancer, as you can see in here in the main database studying neuroendocrine prostate cancer patient samples.

We also see that even in CRPC samples, CXCR7 expression is already very strongly correlated with neuroendocrine prostate cancer markers, such as ENO-II. So to confirm this pattern of expression, we did an IHC staining of CXCR7 in neuroendocrine prostate cancer PDX models, and the CRPC PDX models. So as you can see, androgen receptor is highly expressed in CRPC, and is known to be down-regulated in NEPC. And in contrast, CXCR7 is very specifically highly expressed in neuroendocrine prostate cancer but not in CRPC.

And in the Cancer Research paper I just mentioned, we actually found that the reason for CXCR7 to be up-regulated in enzalutamide-resistant prostate cancer is because it is an androgen receptor repressed gene. So in CRPC, the patient has undergone androgen receptor pathway inhibitors and that leads to a CXCR7 up-regulation. So here we show that it's further up-regulated in neuroendocrine prostate cancer.

And then when we see some quantification of this, we see CXCR7 is almost not expressed in primary prostate cancer. A little bit high in CRPC, but it's mainly high in more than 50% of neuroendocrine prostate cancer samples. So we try to understand what's the mechanism or function of CXCR7 in prostate cancer. So the way we do this is we knock down CXCR7 and we look at a set of genes that are regulated by CXCR7.

So shown here are CXCR7-induced genes. That means the genes that are down-regulated when we knock down CXCR7. And we did gene ontogenesis to look at the function. We found that there's a very strong correlation between CXCR7-induced genes with cell cycle, cell mitosis, cell cycle, and E2F, which is also a cell cycle regulator. So mainly the idea is that CXCR7 can induce cell cycle progression. And we also did some of the kinase assays.

We found that Aurora kinase A actually is a main target of CXCR7. So shown here is a western blot. When we knocked out CXCR7 in two different models, we see a deactivation of Aurora kinase A. So the activity of Aurora kinase A is indicated by its phosphorylations, it's phosphorokinase A. And looking in the patient samples we see that when CXCR7 is high... So these are four different data sets of prostate cancer patients' data with samples. So each correlates a sample, with a sample sorted by CXCR7 from low to high.

So you can see in CXCR7 high samples there's a stronger or upregulation of the G2M checkpoint or cell cycle signature. A lot of those are downstream of Aurora kinase A. So Aurora kinase A is a major regulator of cell cycle progression, including the G2M checkpoint. So then we tried to understand, how does CXCR7 activate Aurora kinase A? As I mentioned, CXCR7 is mainly a membrane and cytosolic protein, and Aurora kinase A, on the other hand, is known to be mainly in the nucleus, specifically in the centrosome when the cells go through the cell cycle.

So as you can see here, this is a co-staining of a bunch of these proteins. As a control, we look at ARRB2. So these are beta arrestins which are known to interact with CXCR7. So you can see they're co-staining, mainly... So this is pointing to ARRB2 in the centrosome. So it seems like ARRB2 could co-localize with Aurora kinase A to the centrosome, and the CXCR7 is actually more in a broader region covering the centrosome. So we call this paracentrosomal regions, and they co-localize here, and also cover where Aurora kinase A is present.

So it seems that they are interacting because they're physically nearby. And we also see that there's a lot of co-localization of CXCR7 with tubulin proteins. So alpha-tubulin, as you can see, has a very similar pattern. So we wondered whether CXCR7 as a membrane protein might be moving into the paracentrosomal region to interact with Aurora kinase A through the tubulin pathways.

So we looked at the mass spec analysis to look at all the proteins that interact with CXCR7. We pulled out a lot of tubulin proteins, as you can see here. And we also show that using co-IP experiments, CXCR7 can interact with alpha-tubulin. So that supports the hypothesis that the membrane CXCR7, binding to ARRB2, is internalized, and they can walk down the tubulin to get into the paracentrosomal regions to interact with Aurora kinase A.

So this is a proximity ligation assay showing that CXCR7 and Aurora kinase A can co-localize. It's a little hard to see here, but the red dots mean the two proteins are adjacent to each other. And when we use a drug to disrupt the tubulin, we lost the interaction between these two proteins. So that explains the mechanism of how these membrane proteins can interact with a protein that's mainly localized in the nucleus and the centrosome.

So then we tried to understand what's the function of this CXCR7, and how important is the CXCR7-induced Aurora kinase A in regulating its function? So to address that, first we used a CXCR7 overexpression cell line. So 22RV1, a prostate cell line with CXCR7 overexpression. We see a dramatic increase in cell growth, which can be abolished using an Aurora kinase A inhibitor called Alisertib. And you can also see the control. This drug can block Aurora kinase A activation.

So it seems like targeting Aurora kinase A can inhibit CXCR7-induced cell growth. To show this in vivo, we put this in mice, and we also put in control cells as well without CXCR7 expression. So first I want to point out is that CXCR7 overexpression, shown in this dark green line here, dramatically induces tumor growth compared to the control which is GFP here.

And these can be killed or abolished by the Alisertib treatment, which is the red line here. So it blocks the tumor growth from this line to the level of this line, suggesting that targeting Aurora kinase A could be useful to treat CXCR7-induced tumor growth. So to summarize, we showed that CXCR7 is upregulated in treatment-resistant prostate cancer, and is further upregulated in neuroendocrine prostate cancer.

And this CXCR7 forms a protein complex with beta arrestin to activate Aurora kinase A phosphorylation, therefore promoting cell cycle progression and cell growth. So CXCR7 increases tumor growth, and these can be partially blocked by Aurora kinase A inhibitors. So with that, I want to thank the people who worked on this project, especially Galina. Great sign up. She was a graduate student in the lab. And also many thanks to the support from the Prostate Cancer Foundation.

Andrea Miyahira: Thank you, Dr. Yu, for sharing that with us. So does the CXCR7 or Aurora kinase A axis have a function in normal cells?

Jindan Yu: So Andrea, that's a great question. The CXCR7 protein is actually mainly expressed in the stroma. So if you are talking about specific prostate cells, the normal prostate cells don't express CXCR7. But we do see, when the prostate develops prostate cancer, and as the disease progresses, CXCR7 starts to get expressed in the function. But whether or not the stroma or endothelium CXCR7 plays a role in prostate cancer, I think it is very likely they might contribute to it.

Andrea Miyahira: So is CXCR7 a biomarker for Aurora kinase A-driven NEPC? And whether or not would it be an optimal way to identify patients who would best respond to inhibition of this pathway?

Jindan Yu: So that's also a great question. Aurora kinase A actually has a lot of genetic alterations. So CXCR7 is just one mechanism to activate Aurora kinase A. And here particularly it's leading to Aurora kinase A phosphorylation and increased activity. And CXCR7 also has other substrates like I mentioned, ERG and AKT. And on the other hand, Aurora kinase A has many ways to increase its activity in neuroendocrine prostate cancer.

One major mechanism is like amplification of Aurora kinase A, which is seen in a lot of neuroendocrine prostate cancer. I don't remember the exact number, but it could be like 60% of neuroendocrine prostate cancer could have Aurora kinase A amplification. So CXCR7 upregulation is not going to serve as a marker for all the Aurora kinase A-driven neuroendocrine prostate cancer.


Andrea Miyahira: Thank you. So you showed that CXCR7 is negatively regulated by AR. Have you evaluated its levels in AR-positive NEPC?

Jindan Yu: Yeah, that's a great point. So we do look at... We see CXCR7 upregulated, for example, in castration-resistant prostate cancer. A lot of those still have AR expression or pretty high AR expression. But when we look at NEPC specifically, we look at the PDX models, the LuCaP PDX series, for example, the University of Washington had, CXCR7 is not upregulated in the double-positive cells.

So the AR-positive, AE-positive LuCaP models, CXCR7 level is pretty low. It is only upregulated in the AR-negative models. But that said, I think there are very few double-positive cases. So it is hard to make a broad conclusion. But apparently, AR signaling is a major regulator of CXCR7. So CXCR7 is the highest I would say in the AR-negative neuroendocrine prostate cancer.


And there is actually also another study reporting that another pathway can induce CXCR7. So they found in the PTEN knockout mice, they see upregulation of CXCR7, but not as high as what we see in neuroendocrine prostate cancer. But that leads to, as you just asked about, the normal prostate. So the PTEN loss, which is an early event in prostate cancer, can lead to some increase of CXCR7, and then it is further increased with enzalutamide treatment or AR pathway inhibitors, and whether the disease develops into neuroendocrine prostate cancer. But I think the short answer is AR is definitely a major regulator of CXCR7.


Andrea Miyahira: Okay, thank you. And what are the next steps in your studies?

Jindan Yu: That's a very good question. So in terms of next steps, we are still looking at Aurora kinase A. We're expanding our study, not just focusing on targeting Aurora kinase A in the CXCR7-driven case, but also in Aurora kinase A-driven neuroendocrine prostate cancer by amplification, or Aurora kinase A overexpression.

So we're actually thinking about doing a clinical trial of Aurora kinase A inhibitors, selecting CRPC patients with Aurora kinase A amplification or overexpression. But the limitation is there are not too many safe Aurora kinase A inhibitors out there for clinical trials. So we're looking out for opportunities in that kind of drug, because Alisertib had a trial and I think it has quite dramatic toxicity. Then the other way we are looking into is targeting CXCR7 specifically, because targeting CXCR7 will not only target Aurora kinase A, but also AKT, ERK. Those are also major drivers of prostate cancer.

So the challenge there is also the availability of the drug, because the CXCR7 antagonist, it's a membrane receptor. The idea would be to make an antagonist, right? So it has been quite difficult to make a CXCR7 antagonist. So actually, the previously reported CXCR7 antagonists in the literature, there were a couple of them, it turned out later to become CXCR7 agonists. So it binds to CXCR7. And since a lot of things can bind CXCR7 to turn on, it's a function instead of blocking it. So we are looking for ways to really make CXCR7 antagonists and to block CXCR7 in neuroendocrine prostate cancer. So that would be in our future lines of study.


Andrea Miyahira: Okay. Well, thank you again, Dr. Yu, for joining us and sharing this study.

Jindan Yu: Thank you, Andrea, for the opportunity.