 In continuation of Dr. Oomani's previous lectures where he talked to you about basics of reverse phase protein arrays, we are continuing to discuss about this novel technology platform RPPA. In today's lecture where Dr. Oomani is going to mainly focus on RPPA analyzer which is a tool for RPPA data analysis and visualization. So, let me welcome again Dr. Rabesh Oomani to continue his lecture and discussion on reverse phase protein arrays. I will show you what kind of data we have generated. So, here let us say for example, these are the different samples, these are the different mutations in the cell lines. So, you can see that this is the data we could generate. So, in this corner and the drug resistance can be incorporated here, drug resistance can be incorporated. So, this kind of numbers you will get. So, these are the target names. So, your complex matrix will be like this. So, from this if you load this data into commercially available software. So, as per pre-designed analysis part you will get the data output, but you cannot manage to get information what you want from your samples according to your experimental requirements. So, when you have such a high end data set, then you need to worry that how experiment is correct, how much experiment is correct or not. So, one of the sensitive point that Sanjeev touched in the morning very nicely and gently suggested you to follow this normalization methods, particularly in RPPA data normalization is very very difficult, because when you do the protein estimation and when you extract all the steps cannot be implicated for 1000 samples in the same manner, it is very very difficult. If you do 10 samples, you will have 10 different values in terms of protein concentration, in terms of method you followed and all. So, to avoid that there are different normalization methods have been developed to apply for RPPA data. So, loading control and variance stabilization of RPPA data have been challenging. So, different normalization methods were developed for LC that means loading control and variance stabilization. One among them is a supercar normalization, total protein normalization, the variable slope normalization and robust jet score normalization methods. One of the method what we follow in western blot is housekeeping normalization, we all are favorite of doing GAPDH more than the target proteins very often and invariable protein set of normalizations. Normal protein set of normalizations means in all the 1000 samples, we know that one particular protein concentration will not change. So, that will set as a landmark. I think if some of you might have done West sorry 2D gels in the proteomics approach in the earlier in the in the beginning days. So, we used to set landmark spots. So, in that way those proteins expected to not change across sample set will be set as a landmark spots based on those concentrations we can normalize. How a normalized data look different from the unnormalized data, you can clearly see that these bars each bar represents one protein across the controls and treatment conditions then you can get a heat map like this. But after normalization, so there is no difference within the samples and nicely you can get a spread of the signals here. So, if you do not apply any normalization method, you end up identifying the more differentially expressed proteins across the sample set you have analyzed. So, how can we do? As a among the methods previous slide explained at least we used to follow total protein normalization method which is more easy to do and also reliable to do. So, it is a how do we do that? First we print a lot of slides. One slide after every 10 slides will be used for this normalization methods. Every 11th slide, 21st slide, 31st and 41st slides will be used as a normalizing slides. This will definitely give the reproducibility this normalizations based on the total protein per microarray spot how we do. So, we strain the slide with fast green FCF, this is called FCF. So, this will directly bind to the protein and we can screen sorry we can scan the slide at 800 channel in infrared imaging system this green color indicates the total protein. We can quantify this same as like antibody signal quantified and we can see that. So, you can already see that in this plot there are 5 or 6 different samples going in different way depending on the time and these signals can be normalized. After normalization once you plot this normalized raw data you look like this to get a more smooth and data once we normalize this with FCF can get a signals like this. You can clearly see that already how nicely the data can be collected to make it biologically sensible. So with this and also morning he suggested to have what kind of controls to include and how to plan experiment and all. So, it is important to include negative slides. So, these negative slides are as non specific binding of the secondary antibody to the arrays. This array every one of the slide out of 50 will be probed with the only secondary antibody and those signals will be detected from the original signal. And if you see any signal for the secondary antibody alone so those antibodies definitely need to be dropped. There are some antibodies which there are some secondary antibodies labeled for infrared from different companies behave completely different. So, now to summarize here how RPPA is advantages or not advantages over the different methods for detection of proteins. First we look at the western blot. Advantage is separation of proteins according to molecular weight. So, now you know the what molecular weight your target of interest is though Santa Cruz will give 10 different lanes, but you know which molecular weight you should look for. Then Eliza, sorry Eliza, but in the disadvantage is very work intensive, high amount of protein is required and it is very low throughput. Also we can do medium throughput, but which is not sufficient to analyze thousands of samples. And Eliza which is quantitative very sensitive, but it requires high amount of lysate. Then IHC is a cellular localization of protein of interest, it is also semi quantitative and sensitivity often not sufficient to detect phosphorylated proteins. When very very few copies of phosphorylated proteins will be there in the cell and then it is very difficult to look at them in the IHC. And mass spectrometry based technologies I am not saying that it is not good. In fact, it is a Dino discovery platform, but highly multiplexing can be done protein isoforms can be distinguished analysis of thousands of proteins and no protein binding binding reagent is required. It is a just direct injection to the sample, this is the main advantage. But disadvantages if you look for complex sample preparation, poor analytical sensitivity compared to immunosciences and definitely is a low throughput compared to the array platforms. I will not say that it is it is very no throughput. Say compared to array platforms it is little low throughput and forward protein arrays many analytes can be measured in parallel in single sample and it is quantitative. The disadvantage is that too highly specific antibodies are required there also because it is antibody captures the analyte, right. So, capture antibody also should be highly specific, but reverse phase protein arrays, robust quantification, low amount of consumption, high throughput, highly sensitive and detection of phospho proteins. It is not that I am watching that it will meet all the other disadvantages of the methods mentioned here. It also has a limitation that it has highly it requires highly specific antibody and every assay need to be optimized and special devices required for this technique as well. So, this is how the RPPA works and also you might have seen that always I am talking about qualitative. Now mass spectrometry evolved much higher level that even can perform the quantitative absolute quantification even the number of copies of proteins present in per cell and all. So, to some extent here also RPPA is useful to make absolute quantification of proteins. This Q-concat approach that means quantitative concavalent protein quantitative approach that is why it is called as like this. So, let us say for example, you can you wanted to establish Q-concat approach, you select a protein of interest and you selects a p-topes recognized by the antibodies. So, here you have a like and always when you buy antibody that antibody will bind to the certain region of a protein. It will not bind to the entire region of a protein, right. So, you can also develop a method called Q-concat approach by selecting those epitope regions of a proteins and you can express them as a fusion proteins. So, this is definitely useful in all kind of protein array approaches and you can develop a single fragment like this containing different epitopes and those proteins can be printed as a standard here in a so in a concentration dependent manner. And then you can do the RPPA along with unknown samples here, then you can make a graph like this concentration of analytes on array and the signals on y-axis and you can also extrapolate the signals of samples for example, A and B on this plot and you can extrapolate and calculate the concentration exactly. If you know the number of cells you have taken to prepare this sample B, you can also say that this many copies present for this many population of cells. This is a way you can quantify. I will show you one example. In this case at least for NASH project non-alcoholic steatohepatitis project, we try to express all 11 recombinant cytochrome P450 proteins as a fusion proteins and using this vector system we develop the proteins and we purified them and we check the antibody specificity and use them as a standard proteins to quantify the cytochrome P450 protein levels in a certain population of cells. So, if 1 million cells contain how many copies of cytochrome P450 proteins like this 11 different recombinant proteins and also nowadays it is very, very important to do quantitative approach for phospho proteins for example, if you take AKT protein not all AKT copies are phosphorylated when it gets activated only certain population or certain copies of AKT proteins gets phosphorylated how can we do that. So, what we did is we have expressed the purified, expressed the targets of interest in this case JNK, AKT, ARK and P38 expressed purified and in vitro translated for protein modifications sorry, in vitro phosphorylated for protein modifications using the specific kinases and those phosphorylated proteins we spotted in a concentration dependent manner and then later on we did the assays and we can show that phosphorylated proteins on the un-phosphorylated protein detection, the linearity range and above 0.9 and then we looked at the signals detected here different cell lines and different stimulated samples we can nicely detect total values are like this and upregulated, downregulated copies are mentioned here in the similar way for phospho-AKT and AKT here, so I am sorry the scales are different here it is into 10 here is the total number is shown here. So, in this way one we can perform the other one is that absolute quantification can be done in multiplexing manner as well, so on the previous experiment we have spotted separately then we mix all of them spotted as a single spots on a concentration dependent manner and then when we looked is a single spotting is important or the can be multiplexed is possible. So, you can also already see that a little bit signal is down but the behavior is same for all the different proteins, so you can see the power of RPPA is reached a different level that you can also use this method for absolute quantification of a particular protein of interest in a larger population of samples, larger set of samples. So, now just imagine that this is a technique is in your hands, how you imagine to apply into your own area of research is up to you, so to highlight some of them you can say that 2D or 3D what kind of samples can be analyzed, 2D or 3D cell cultures, xenograft, primary tissues, tumor biopsies, tissue sections, LCD materials, laser capture micro detection materials, it is definitely high throughput, it is low sample, comparison signaling pattern from as low as 20 microgram protein which is not even sufficient to have one protein and one housekeeping gene on the western blot approach. The key nodes of signaling validated antibodies to cover total and phosphoproteins so that you can reconstruct the signaling networks when you stimulated the cells with certain factors or the drugs and so on. So, to typical applications you can use this approach for disease mechanisms, signaling pathway profiling, biomarker discovery and validation particularly for early diagnosis, prognosis, prediction treatment and efficacy and drug discovery projects like on and off target activities of new drugs. When a new drug is coming into market, definitely nowadays regulators are asking what are the off target effects of your drug or what is the target of your drug? Even if you do not know off target effects and right target of this molecule, what are the possible signaling mechanisms or signaling nodes targeted by the molecule? This can be definitely achieved by RPPA approach and also right combination of drugs. So, when you are looking at the personalized medicine, so if you using the cell based models and by reconstructing the signaling pathways, you can choose the right combination of drugs. So, always sometimes combination of combinatorial treatment are more effective than the single treatments. So, PKPD studies you can identify the biological active doses of drugs because some certain signaling pathway should be shut down or shut on. So, what is the concentration of drug is required to switch on and switch off? So, our reading or the end point reading readout is a switching on switch off by using the drug molecule. This switch on switch off signaling pathways can be definitely done by RPPA in a high throughput fashion. So, clinical research definitely patient selection for personalized treatment I will touch upon later on. So, using this technique we have done different projects, at least I have done these different projects. So, I will touch only non small cell lung cancer and prostate biomarker a little bit. So, lung cancer I do not want to go more details into the lung cancer and it definitely has two different types non small cell lung cancer and small cell lung cancer. Non small cell lung cancer about 85 percent only 15 to 20 percent a small cell type, but today both cancers are difficult to treat though having lot of drugs it is very difficult to treat and still mortality rate is high and due to the several reasons that the different sub types of non small cell lung cancers are there in the patient. Now, in the clinic even doctors are not able to segregate them before start starting the treatment. So, what kind of drug is good to give to them and all. So, keeping that in mind at least we try to do some good experiment and how we can separate the patients based on their molecular sub types using RPPA approach. I will show only one or two examples why it is required. So, when genomic and transcriptome is not informative we are only left to proteins. So, when doctors do not have any genomic information or molecular diagnostic information by analyzing the proteins they can take a decision what kind of drugs can be given to the patient. So, cancer in every patient is different definitely patients will teach the doctors how what kind of treatment they should receive. In this case patient B you can see that red line is passing differently in two different diagrams, but both of them are lung cancers assume in that way, but the signaling pathways are responsible for non small cell lung cancers are different, but how doctor knows this unless until he analyzes the signaling pathways responsible lung cancer only he will be able to understand how we did it this in this case. So, again predictive and prognostic markers are required whether the disease is getting cured or not cured. So, I will show with some examples. Again as I said in the first couple of slides antibodies selection is very very critical and important to establish this method. Although wide range of genes are mutated in cancer along with lung cancer the apparent driver mutations identified so far are components of 12 core pathways that means all the different mutations which cause lung cancer map into one of the these pathways. That means if you analyze these pathways activation inactivation in lung cancer patients based on the activation status of the pathways you can choose the drug to be given to the patient. This was a motto we decided or we arranged a panel of antibodies which are measuring the activation status of these pathways to proceed forward. Of course it is very very difficult to get patient samples with all possible molecular subtypes, but one of the group in Cologne Roman Thomas group has established 84 lung cancer cell lines from different molecular subtypes of the patients. So, we know what kind of mutation they have to the all the all the genomic data genomic fingerprints are available and those cell lines were established, but what are the different pathways activated inactivated is not known. So, in this case these are the genetic aberrations of which are causing lung cancer today and these 84 cell lines have these 11 different mutations and we perform the protein proteomic profiling again using the 782 antibodies which has from the 12 different pathways in the previous slide ultimately we get a heat map like this and you can see that some of the blue areas concentrated some of the red areas concentrated and on the x-axis you can see the cell line and y-axis are the different proteins we measured and these proteins are divided into almost 5-6 clusters. All the proteins are divided or expressed as different clusters here the cyan color indicates a presence of the mutation and red color is not presence of the particular mutation. When we try to align this some any of these concentrated areas are not aligning with cyan colors. So, wide spread is there you can see that, but here we need to take intelligent decision here what we have done is did a closer look here we can see that hereto amplification and EGFR amplification or mutation signals this activation of EGFR pathways are nicely aligned with this the these 3 places here you do not see any science here right. So, this we looked at again this is some of the AKT pathway proteins are inversely related to here the red means upregulated blue means downregulated yellow means no change indeed. So, this blue color is inversely related inversely proportional to the red color observed here. So, then we looked at the in detail so, major cell lines population is having Keras mutations and also Keras protein can be mutated in 3 different places. So, usually when a patient is declared that Keras mutation is present doctor think that ok X drug can be given, but within the Keras mutated patients if they are different of populations drug works for some set of people some set of people will not work. This can be addressed by RPPA approach how you can see that I will try to convince you that when you look at a closer look here these are the different proteins upregulated blue or downregulated in a G 13 mutation compared to G 12 mutation. There are 3 different mutations 80 to 90 percent of lung cancer patients with G 12 mutations and only 10 percent or less than 10 percent or with G 13 codon mutation only less than 5 percent of population are affected with Q61 mutation. So, we tried to assign this proteomic signatures to individual mutations we observed in the cell lines here. So, with this we for we did some validation studies we propose that so, in the Keras mutations patients usually treated with ras-raff inhibitors. So, we also proposed based on this expression pattern that a set of patients with K13 mutation they should be treated with AKT inhibitors not the ras inhibitors ok. So, if the doctor knows that Keras only then he will give the ras-raff, ras pathway inhibitors definitely set of population with K13 mutation they will not respond here. So, this kind of information need to be miniaturized in the form of arrays and bring into the clinic to help the patient community here. So, in the similar way we also did for another closer look for another set of small cluster here you can see that this is a kind of small cluster with red patches here and here. This is basically the p53 pathway and then we looked at the p53 hotspots and this kind of pathway we reconstructed so, which was not known for p53 patients. That means some of the p53 patients were never receiving the AKT inhibitors earlier they were always treated in a different manner. So, those p53 patients with activation status of AKT should be treated with AKT inhibitors rather than the other inhibitors this information was missing earlier. So, this way we can use reverse phase protein array approach to categorize the patients for treatment approach and another way is at targeted therapy. So, we have a drugs can we predict whether the patient is responsive or non responsive. In that case so, these are the different drugs available in the market we have established the IC50 values for all the drugs against the 84 cell lines. Then we integrated the data using RPPA approach you can see nice cluster here all these cell lines having the aberrations for EGFR amplification and EGFR mutations they are responding to the EGFR targeting drugs. That means already established data we can observe in this whatever the data we generated here is a new. That means this is a kind of case control for us now can we establish some of the new prediction markers using the other data set. So, using this we also established other data sets in this case it is nicely aligning here further we established a set of markers for all the drugs we have screened for their IC50 values against 84 different cancer cell lines. So, this is a way we have established. Now last couple of slides I will sum up now protein analysis of prostate biopsy material using RPPA approach. Then I said in the beginning that any type of sample can be analyzed here. So, the capacity is high throughput fashion the applicability of RPPA towards the biomarker discovery approach. One of the study I did during my post-doc time so we have got a patient sample and we I sat with the pathologist sections do the sectioning and get the protein extracted ultimately end up with a very very small core which can give 50 microgram of sample with a real tumor cell population. With 50 microgram of sample I should identify the differential express proteins I should validate them show the proof that they are really differential express and then write a paper and publish to get my PhD ok. So, that was a challenging but again here we did a traditional 2D ditch nowadays 2D gel is not so encouraging for the people not very attractive method those days it was possible to do and did mass spectrometry finally we did RPPA. So, these are the different proteins which are differentially expressed specifically in cancer patient compared to normal or behind prostate hyperplasia in the surrounding tissues. But again this many list is around 112 to validate 112 proteins I do not have a lysate with me I wanted to analyze and establish the signature from the same set of samples right. So, this network platform we built using systems biology based approach and we enrich the marker analysis using the software tools and using the known information these are the different targets at least I have highlighted to proceed forward. This list is also not small so, almost 10 proteins are there to go proceed forward can see the difference between the tumor and normal very nicely here to proceed forward then I did this reverse phase protein or approach I could see clearly the difference for example I am showing peroxidoxin 3 and 4 can see these each dot represents one patient and it is upregulated in patient samples compared to the normal or control samples and also we can see the signal dynamics depending on the serial dilutions and so on. So, based on this approach we highlighted we could identify that these are the different proteins are really differentially regulated in cancer tissue compared to the normal tissue from 50 gram protein we ran a 2D gel which obviously requires high quantity of protein we perform the validation of 10 proteins and still 2-3 micrograms left over in the tube. So, this is the power of RPPA we could use now where are we standing today and as I said that that time not many people were accepting that it can go into clinic and all. So, now in 2014 RPPA consortium meeting was there people were talking about how can we bring it. So here once you know the level cut-off level let us say for example blood glucose level we know it is 80 to 120 below 80 is like a hypoglycemic then above 120 hypoglycemic. So, in that manner so if you know the levels we could spot some of the known protein known concentrations as a low control and high controls and here we can have them as a calibrators based on the concentration dependent we can construct a plot like this. The moment if you get a patient data in this panel and we can extrapolate on the graph and you can say that what is the quantity if this quantity is below or between the low and high or above high or below low then we can decide that. So, this patient is having X problem or Y problem and so on. So, now people are talking about this kind of mini-agedized RPPA platform to bring into the personalized medicine. So with this I will just sum up it is suitable for a simultaneous study of hundreds to thousands of samples for expression and activation of a protein of interest is correctly identified the cell lines with the EGFR single point mutations did not align with activated pathways but if the closer look for mutations on G12 and G13 results how RAS PA3K signaling rather than RAS RAF pathways the RPPA revealed a crosstalk between P53 activation and AKT which was not known earlier the RPPA data integration with IC50 values have provided additional markers is also useful for validation of proteomic data obtained from limited sample and absolute quantification of selected proteins either it can be total or activated fraction can be achieved per cell in fact. So this kind of experiments can be done using RPPA approach. But all in all now you might have understood that the basic steps necessary steps to follow to printing and then hybridization and all steps are same only difference between forward arrays and reverse arrays is that here we spot a just crude lysate you do not have to follow any separation of samples and so on. So on one experiment on a one slide you can you can understand or you can estimate quantity of a particular protein of interest across the large number of samples. So this is this is what so with this I will just like to acknowledgement some of the people in my lab and most of the wet lab data generated with the help of Ulricho Corf and she unfortunately passed away 2 years back. So she is the one who got me into the reverse phase protein array area in fact. So with this I thank you all of you for patient listening for almost one and half hour probably I am very open to discuss with you. So in this case again and again I wanted to emphasize that I may not be able to show you how the slide looks like and all it is exactly the necessary steps you are following is 100 percent similar unless until if I tell you that this is reverse forward you will not know it that is it. So I hope you will be convinced to do to do the experiment by virtually imagining the steps I have highlighted you. Thank you very much. So this now concludes the session on reverse phase protein arrays. I hope you have learned good workflows and basics of reverse phase protein arrays setting up the experiment leading towards the data analysis. By now you know that there are many applications which can be performed on reverse phase protein arrays platform which Dr. Omani has explained in context of clinically relevant problems. I just like to add that now the technologies are really progressing well and we have you know better printer and the arrays to make these kind of chip platforms we have sophisticated software tools which can do data analysis better. We have much more high throughput capability nevertheless the reverse phase protein arrays have been in used in clinics from long time not as the high throughput technology platform but even to test out abundance of a given protein clinicians have been using it from long time and this also shows the need for having reverse phase protein array based platform for the clinical applications. So by knowing that the basics as well as the possible applications I hope now you are excited to really understand this technology in much more detail which can be really really helpful if your goal was to deliver something to the clinics for translational research. Thank you.