Urologic Oncology

SLIDES & TRANSCRIPTS
Friday, December 5, 2003

Gene Expression Analysis in Prostate Cancer

Arul Chinnaiyan, M.D., Ph.D.

Slide 1:

Thanks for inviting me. Actually my laboratory has been focused on gene expression profiling of prostate cancer using primarily DNA microarrays to try to develop a molecular signature for prostate cancer and identifying individual biomarkers that we can characterize using DNA microarrays as a platform for discovery, but what I would like to share with you today is some more of our recent work which has actually been focused on defining a humoral response to prostate cancer, what we call sort of cancer epitomics. It is another -omics I am sure that you have added to the list of -omics that you have heard about today, genomics, proteomics, metabolomics and epitomics and I feel that this sort of dovetails nicely with our previous speaker's talk and that we will be focusing on antibody responses against a high-throughput approach to defining antibody responses to tumor antigens.

Slide 2:

Just a little bit about our previous work. We, as a number of other groups have been defining molecular signatures for prostate cancer using high-density DNA microarrays and what I am showing you here is sort of a molecular portrait for prostate cancer. We have been using 20,000 element DNA microarrays to interrogate prostate cancer specimens and what I am showing you here is approximately 1000 genes.

This axis in about 100 of the samples that we have profiled using these high-density DNA microarrays grouped into clinically localized prostate cancer, benign and metastatic which in our heat map convention up regulated being defined by red in color, down regulated by green in color and black meaning equal expression.

Slide 3:

So, the general approach that we have been using has been to use the DNA microarrays as a discovery platform to look at tens of thousands of genes. When we identify potential candidates using the DNA microarrays we then validate them individually on high-density tissue microarrays which contain hundreds of patients that are clinically stratified and most importantly have follow-up information such that we can do clinical outcome studies, say good prognosis versus bad prognosis

Slide 4:

and through our global profiling efforts here are some of the biomarkers that we have identified.

Other groups have identified them as well and I will just list a few of them at now the protein level. For example, hepsin we found as being highly up regulated in the precursor lesion of prostate cancer, prostatic intra-epithelial neoplasia and then there is a down-trend. PIM-1 kinase is up regulated across prostate cancer while EZH-2 which I won't have much time to actually talk about today is a very exciting molecule in that it we feel serves as a marker for aggressive prostate cancer as well as breast cancer but also may have a role in mediating the metastatic process in controlling cell invasion as a protein of the cellular memory machinery.

What is interesting about this molecule,EZH-2 is there is a high expression of EZH-2 in metastatic prostate cancer as well as in more aggressive clinically localized disease.

The marker that I will be focusing on today will be AMACR, alpha-methyl-co-A-racemase which we found as well as a number of other groups as being highly up regulated in clinically localized prostate cancer.

Slide 5:

Our group as well as Angelo DeMarzo and Billy Isaacs at Hopkins as well Jiang's group at University of Massachusetts have all identified up regulation of this molecular, alpha-methyl-co-A-racemase as being up regulated in prostate cancer tissue and in this particular study we characterized it as a tissue biomarker of prostate cancer, again coupling DNA microarray technology with high-density tissue microarrays.

Slide 6:

We identified alpha-methyl-co-A-racemase by doing a meta analysis of the publicly available DNA microarray studies that were again in the public domain and we found that this molecule was consistently up regulated in a variety of data sets whether they were Affymetrics oligonucleotide microarrays or spotted cDNA microarrays. It was nicely up regulated as you can see, benign versus cancer, benign versus cancer and so on across different studies done by different institutions

Slide 7:

but what really got us excited as well as a number of other people about this particular molecule was the sensitivity and specificity of AMACR for cancer epithelia and what I am showing you here is a comparison of PSA and these are antibodies against PSA staining a section of prostate cancer and the brown staining means positive staining, and you can see here that when we stained these sections with antibody against PSA we not only hit the benign glands but we also hit the cancer glands.

When we take a consecutive section and stain it with antibodies against AMACR we nicely hit the cancer gland and spare the benign gland. So, this likely accounts for the non-specificity with the serum-based PSA test because there is high expression of PSA as you know in normal epithelia.

Slide 8:

So, what is AMACR, alpha-methyl-coA-racemase? As the name suggests it is an enzyme. It is a racemase in that it converts R stereo isomer fatty acids to S stereo isomer fatty acids and then pushes it down along the beta oxidation pathway of fatty acid degradation essentially creating energy for the tumor cell.

It has been known for some time that fatty acid synthesis pathways as well as fatty acid synthase itself is up regulated in prostate cancer as well as in a number of other tumors. So, it makes sense that the utilization pathways would also have enzymes that are up regulated in the process.

Slide 9:

Our group as well as others have now been using AMACR in needle biopsies and we have actually introduced it clinically for challenging needle biopsies that are referred to the GU pathologist and here I am basically showing you a prostate needle biopsy stained with a basal cell marker which is nicely staining of course the benign gland while the AMACR marker is staining the cancer gland and it has high sensitivity and specificity in that context.

Slide 10:

So, we have shown that there is up regulation of AMACR transcript as well as protein and then what we did was we went on to describe the enzymatic activity of AMACR in prostate cancer by measuring its activity using a tritiated substrate of AMACR and we took prostate cancer tissue extracts, compared them to benign and found consistently up regulation of AMACR enzymatic activity in prostate cancer relative to benign and that we could immunodeplete that activity using antibodies against AMACR.

Slide 11:

We have next been trying to extend this to try to develop a test that can be used on prostate needle biopsies essentially to both measure AMACR activity as well as AMACR protein in these prostate needle biopsies and here is just a proof of concept study where we have looked at 50 prostate needle biopsies and found that essentially the needle biopsies that contained cancer primarily had increased AMACR activity relative to those that did not, that had basically benign tissue and we could see that both at the protein level as well as when we monitor activity and we are excited about this potentially because it has the opportunity of developing a bedside test where the urologist could essentially assess whether a needle biopsy has cancer or not based on either AMACR enzymatic activity or AMACR protein as in the calorimetric assay.

Slide 12:

So, I guess it is certainly intriguing to characterize AMACR in the context of tissues and utilize it as a potential diagnostic marker in prostate cancer needle biopsies but more importantly we would like to detect this molecule in a more non-invasive way.

We have certainly looked for AMACR in the serum of prostate cancer patients and with some success we have identified it in small cohorts of patients but the response that we have more consistently identified in these patients is actually an antibody or a humoral immune response to AMACR or auto-antibodies against the AMACR protein that appear to be specific to prostate cancer patients

Slide 13:

and the way that we identified this was using protein microarrays essentially microarrays where we generated and we spotted a host of tumor antigen proteins and then exposed these protein microarrays to serums from healthy controls as well as from prostate cancer patients and then we detect the antibodies to these tumor antigens using a biotin-labeled anti-human IgG which we can detect with the Cy5 label streptavidin and basically if somebody has antibodies to that particular protein that spot lights up as a red spot on that protein microarray as a positive humoral immune response or signature, and here is just a representative protein microarray analysis reviewing a humoral immune response to AMACR comparing a control subject versus a prostate cancer patient where we have spotted this AMACR protein recombinant AMACR protein in replicate and these are the circled white areas here and you can see that in the prostate cancer patient relative to the control patient we are seeing consistent humoral immune response

Slide 14:

and then when we zoom into the individual microarray element you can see here that the serum from prostate cancer patients is nicely heading for human immune response against AMACR relative to control patients and this is regardless of their PSA level, whether they are above 4 nanograms per ml or less than 4 nanograms per ml.

This is a non-specific protein and you can see here that antibodies against PSA however are non-specific pretty much similar to PSA itself,

Slide 15:

and here is a summary of our protein microarray analysis where we have looked at a number of tumor antigens and you can see that prostate cancer patients in general make a humoral immune response to AMACR relative to control subjects and this is relative to a host of other tumor antigens that we have looked at. For example, CEA there is no difference and PSA there is no difference between healthy subjects and prostate cancer patients.

Slide 16:

We validated this response by doing conventional immunoblot analysis. We take the recombinant AMACR protein, run it on a gel, create strip westerns and then take serum from the individual patients and essentially look for that band to light up on the immunoblot and you can see here that in control subjects relative to prostate cancer patients we see a nice humoral immune response against AMACR. We show the specificity of this humoral immune response by essentially quenching it by incubating the serum with excess of recombinant AMACR to block that immune reactivity and you can see that we can block it consistently across serum with just adding excess AMACR recombinant protein.

We have then extended our findings to a more high-throughput approach using high throughput immunoblot analysis

Slide 17:

and what this allows us to do is look at hundreds of serum samples and use very small volumes of these serums approximately 1 microliter and the basic idea is that we run the recombinant AMACR protein in a preparative gel. We expose that preparative gel then to nitrocellulose and we transfer that to nitrocellulose and then exposure the nitrocellulose to a mini-blot apparatus which allows us to look at approximately 50 serum in one experiment using again 1 microliter from each serum and then basically compare control subject against prostate cancer and as you can see here then you develop this blot using a secondary anti-human IgG and in a cartoon fashion here we are just showing you prostate cancer patients relative to control subjects and the next slide is actual data

Slide 18:

where on this side we are monitoring the humoral immune response in prostate cancer patients on this side compared to controls, control healthy subjects, and you can see that some of the healthy subjects are lighting up for positive for humoral response, against CMACR but you have to remember that this is a sex-matched and age-matched cohort and theoretically these older patients could have prostate cancer as well or these could be false positives.

Slide 19:

We have titrated this AMACR humoral response shown here. We can titrate this down specifically

Slide 20:

as well as we have developed an ELISA to detect this humoral immune response and this primarily is so that we could potentially bring this to clinical utility and ELISA certainly makes it easier and essentially this just involves coating ELISA plates with AMACR protein and then working out the secondary detection, and here we are basically titrating out the AMACR humoral response in prostate cancer patients relative to control subjects

Slide 21:

and in this slide I am basically just showing you the three independent methods that we show that prostate cancer patients have a humoral response against AMACR, this using the protein microarray technology, control versus prostate cancer.

In this approach we are using the high throughput immunoblot analysis, again control versus prostate cancer and then the ELISA approach which we have more recently developed again comparing control versus prostate cancer and this is the number of patients that we have evaluated using each independent technology

Slide 22:

but the punch line is in this cohort of patients where in the critical diagnostic range of PSA between 4 to 10 nanograms per ml how well does AMACR humoral response perform and this is a traditional ROC curve, receiver operator curve, sensitivity versus one minus specificity. The best test of course would look like this. PSA as has been reported previously does not work very well in this cohort, has an area under the curve of about .5. When we compare that to AMACR humoral response which is shown here it has an area under the curve of about .75 and depending on where you pick on the ROC curve you can achieve a 75 percent sensitivity with almost 80 percent specificity and this is looking at just a humoral response against one particular marker.

Slide 23:

So, we certainly don't believe that we can identify the most specific and sensitive humoral immune response to prostate cancer by just looking at one individual marker. So, we are certainly developing methods where we can basically get an idea of the entire humoral immune signature of a patient essentially the epitomic profile of an individual patient which then we feel we will be able to get a much higher sensitivity and specificity by essentially multiplexing across a humoral response, and so in this effort we have been developing these high-density prostate cancer protein microarrays where we are basically spotting 2000 proteins that we isolate from prostate cancer extract and then in a similar fashion expose to serum from both control subjects as well as prostate cancer patients trying to define the humoral immune signature to prostate cancer.

Slide 24:

The basic approach here is analogous to doing a 2D gel. However, we are doing this in liquid phase. We are basically doing 2D liquid phase fractionations of prostate cancer tissue extract to try to define individual proteins in specific extracts that we develop these prostate cancer protein microarrays from.

So, the basic idea is that in one dimension we separate by chromatofocusing basically separating proteins by PI. In the second dimension we separate the proteins by reverse phase HPLC and this essentially is separating by retention time on the column, and this allows us to get approximately 2000 distinct peaks from our work, and we can actually identify those peaks by mass spectrometry, and we can actually compare profiles, proteomics profiles between prostate cancer and benign but more importantly for this study we can take those fractions, make prostate cancer protein microarrays and expose those prostate cancer protein microarrays to serum from different individuals and try to define a humoral immune signature that is specific for prostate cancer,

Slide 25:

and here is just some of our preliminary work in this area comparing a control subject. Again, these are about 2000 different proteins, prostate cancer protein compared against a serum from a prostate cancer patient.

You can start to see spots lighting up and in this case the humoral response is in the red color and green is a control situation,

Slide 26:

and here is basically a heat map defining a cancer humoral signature here compared to what we see in control subjects and this is again in the early stages, but we feel that we should be able to get a higher sensitivity or a very high sensitivity and specificity when we look at the entire repertoire of the humoral immune response in addition to focusing on single immune responses as we have done with AMACR, and I think I will go ahead and wrap up here.I would like to thank my collaborators at Michigan, Rogal Shah, Ken Pienta, Marty Sanda, John Wei and Deboshus Goshure, as well as David Lubman and Mark Rubin and then critical people in my laboratory that have been doing the experiments that I showed you here.

Thank you.