 As you have seen, nucleic acid programmable protein arrays or NAPA is very robust technology where many applications can be performed. You have seen Dr. Josh River discussing about the advent of this technology and various type of applications which could be performed on these arrays. However, the nature of microarray experiments are such that you have to perform series of steps before you can make any meaningful signal or the sense out of the data. It is like a western broad when it is starting from blocking your arrays to doing the incubation with the patient samples or primary antibody followed by again washing steps, then secondary antibody incubation and then do the signal detection. So, this whole procedure which is the day long procedure involves series of washing steps drying and again adding the next set of reagents. Imagine that you know you have printed some features on the arrays, some substrate could be glass or nitrocellulose and as the day progresses and you are performing next set of experiments on these chips, then if the reagents are not very tightly bound to the substrate they may slightly wash off or if your binding is not very tight, then probably you will see the loss of the signal. To overcome these technology barriers, there is need to come up with better alternatives and the technology progression has to happen, so that we have more robust microarray platforms available. In this slide, Professor Josh LeBear is today going to introduce you to a newer method which is halotag based NAPA technology which they have very recently developed, it has shown much more promise and very strong signals to do the NAPA arrays with much more efficiently. So, let us welcome Dr. Josh LeBear for today's lecture on Advent of NAPA technologies using halotag methods. Okay, so what I thought we would do today because you are all such advanced scientists in the area of NAPA is fast forward a little bit. So we have talked a lot about the development of the technology, we have talked about the methods for making it, we have talked about some of the applications that we have done with it, what I thought we would talk about today are some of the newer methods that we have been developing in the last year or two. So these are very current, in fact several of them are papers that were published just this last year. So one of the things I mentioned the other day was the halotag and this is what the halotag looks like, this is a chloroalkane, so it is an aliphatic chain with a chloride at the end. You can also do the same thing with bromoalkanes, but chloroalkane is I think one of the preferred substrates. The halotag enzyme is a suicide enzyme, so it binds to the chloride, it forms a covalent attachment to the chloride and then it gets stuck and so now you have this protein that is covalently attached to this chloroalkane. And this R group here can be any functional group that you want and so you can use that to attach the chloroalkane to any surface, you can attach it to beads, you can attach it to a DNA barcode and that means that if you add the halo, if you add this part to your protein just the way you would add the GST protein, now your protein will stick to any of those places covalently, you will capture it in a one directional method that is permanent. This is this could be potentially very useful, so one application is that I think I have emphasized several times that the NAPA technology primarily displays proteins that are functional and folded and for the most part we believe that that is a advantage of the technology, but there may be circumstances in which you want to measure binding to denatured protein, perhaps the epitope you want to find is actually a linear epitope and it is buried inside the protein. So, if you were to try to denature standard NAPA what would happen? So, how does NAPA hold the protein on the array? GST and antigen antibody, so what happens when you denature antigen antibody? They fall apart, right, so if you take standard NAPA and put it under denaturing conditions, the proteins on standard NAPA will all fall off, because they are being held there by a strong, but nonetheless non-covalent interaction. So, imagine now if you could attach the protein to the array in a covalent attachment, now you could treat it with, you could denature it and the proteins would still stay attached, right and so that is what is shown on this slide. So, we made an array and these are with proteins that all have the halo tag on them. So, this is just like the kind of NAPA that you have seen before, but the difference is that in this case instead of using the GST tag, we have a halo tag and it is binding to the ligand and now you can see for example, this is a p53 antibody and the GST tag, so this is the protein array under standard conditions, the protein array under denatured conditions. When I say denatured, I mean we treated it at 55 degrees Celsius with SDS, so that is a pretty harsh treatment and you can see that the GST antibody only binds to folded NAPA, it does not bind to denatured NAPA, but this particular anti-53 antibody could bind both formats, because we knew this antibody here binds to a linear epitope. And so one of the ideas that Ryan had, the student who did this work was is it possible that if you were to take a protein array and display it against serum either in the native format or the denatured format, would you get different immune responses? Could you pick up responses for example, under denatured conditions? And that is what he actually found here. Here you see the denatured protein and then here you see the under standard conditions and here you can see that this antibody here specifically detects under standard conditions and whereas these two antibodies only detect cyclin A and MIC under denatured conditions. So that patient's serum response was different depending on whether they were folded or whether they were not folded proteins. I think you can see that response here that it is binding, this is this response here, binding to the denatured protein whereas you don't see it so strong in this guy here. Any questions on that part? So then no, no, no, no, no, the array was denatured, then it was washed and then treated with physiological. So the proteins were all stretched out on the array surface. You denature the array, then you rinse off the SDS, you rinse and you bring it back to room temperature and now you add it folded antibody because if you denature the antibody it wouldn't work or most antibodies don't work. Okay. All right, so we talked about this a little bit earlier in the course. This issue of some, a couple of people asked the question about protein diffusion and so what, what do I mean by protein diffusion? Well here are three spots on the array kind of in a cartoon and here you see the DNA for each of these genes and one of the concerns would be that if you produce the red protein that the red protein could float over and bind in the spot where the blue protein is supposed to be. Same could be true over here on the green side. So you get mostly red binding to red but maybe red binds here, maybe blue binds over here and so you end up with a circumstance where you have a little bit of mixture at each of these spots and so that would be, that's the concern that a lot of people have. So we actually went to see how much of a problem that actually was. So this is the, this is a configuration of our standard NAPPA. The spacing here between these spots is what we currently print at and what we did in this experiment was we printed the gene here at this spot and in these spots here we printed everything but the gene. So the antibody is still there to capture the GST but there's no gene in that local position. So what that means is that if there's any diffusion from here to here it'll get captured by those sites and you'll see that as signal. And what you see is that in fact there's a little bit of spread around the spot but it doesn't really reach over to these neighboring spots and this is a three dimensional plot of that intensity. So very strong intensity at the main feature, almost no signal at the neighboring area. So in standard NAPPA this is really not a big problem okay but if you start to make NAPPA much smaller, so if you take the 750 micron spacing here and make it 375 spacing so almost cut it in half. So now these spots are really close by. Now you start to see a little bit of signal bleed over. You see that little green signal? So this is the intensity of the spot itself and then these neighboring spots have picked up a little bit of the protein and you can sort of see that in that 3D rendering. So what that tells us is that for the most part under our current conditions we're okay but if we ever wanted to make our arrays much much more dense, so shift from 2,300 proteins let's say to 10,000 proteins, we could run into trouble where there would be neighboring spot intensity. So we've been thinking about ways to get around that and this is the method that we've developed. What we do is we take silicon the same material that you used to make computer chips and we use the same technical approach that they use, what's called photolithography whereby shining light on the surface you create a mask and then you etch it with chemical compounds that etch away the surface and you essentially can wear away the surface of the silicon and you end up what we do is we create these little what we call them nano wells. Nano because they're nanometers in size, in fact in terms of fluid volume they hold picoleters of liquids so they're very very small. And so we etch away those wells, so here this is the process use photolithography to kind of create a mask, you use the acid etching to create these wells and then there's a couple of chemical treatments you have and then we print the Napa mix into the wells. So it's well I describe it like it was easy this is actually quite an involved process to get this to work. It took a lot of different mapping methods on the photolithography side to create wells that had this sort of bowl shape at the bottom because typically photolithography wants to make a straight wall and a flat bottom and it turns out that the signal intensity was not as good in that format. The other thing that's not tricky that's not easy is printing the DNA into the well. So in standard Napa we just have a solid pin printer that just runs along and just makes spots. But here we had to get the liquid right into a much smaller target and so we ended up having to use a piezoelectric printer that has a camera in line with the printhead and by using the camera to align where the spots are we can aim the, it spits the liquid into the tiny wells and it does so quite accurately but it is a little bit tedious. But anyway that's what we do so you end up printing the DNA in the well then once you do that you add cell free lysate across the entire surface of the array and that liquid is intended to get into the wells and then you cover it with a cover slip and you can see that cover slip right there. That also turns out to be non-trivial. Non-trivial because when you have small wells like that there's a tendency for the liquid because it's hydrophilic to not want to go into the wells because air gets captured it basically there's air in the wells and it gets captured. So you have to fiddle a little bit with vacuum pressure and surface pressure to get the air out and get an even distribution of the expression lysate throughout the wells. And actually the solution that the engineer came up with is quite clever I'll show you that in a moment. I just want to mention this is what this is a scanning electron micrograph of these nano wells in silicon and you can see that they have this very nice bowl shape and it turns out that that shape is important. Okay so what what Peter did to seal these wells he developed a system where he had two plastic laminate surfaces that flat plastic surfaces that are like this and sandwiched in the middle he has a form of oil liquid a liquid oil a clear liquid oil and so then what he does and the liquid oil that whole system is connected up to a pressurized system and so the minute we finish putting in the expression lysate he adds pressure to the the oil the oil then takes the plastic and does that it kind of forces it apart and essentially forces the plastic to seal the the nano wells and it does so instantly and at the same time there's they apply a little bit of vacuum to the liquid on the surface of the array that pulls out any excess expression lysate and you end up with a sealed surface where you I don't know if you can quite see that but you end up sealing the silicon well the nano wells with that with a plastic surface so this is what the apparatus looks like it there's it's it's evolved and I need to get a better picture of that this is a little bit of an old slide but the system works pretty well this is the piezoelectric printer by the way that's doing the printing and these are these piezo these special piezoelectric nozzles that are very accurate at delivering fixed volumes to each well one one added benefit for those of you who are Napa aficionados is that with these nano wells we have figured out a way to print the print mix separate from the DNA mix so what one of the things that you may not appreciate when we normal Napa when we print it has a cross linking agent in it and the crossing agent is meant to capture the DNA and the protein BSA to the surface of the slide so it stays put the problem with the cross linking agent is that it's it's there's a time function attached with the minute you activate the cross linking agent it's a free chemistry that starts to act on your sample if you let it go too long everything gets over cross linked and it's no longer functional so the minute you add it to your print mix the clock starts and you have a certain amount of time to print it before everything gets ruined anything that doesn't get printed that day whatever's left in your tube it's gone forever so if you made a lot of DNA to print your rays use a little bit of it to print your rays all the rest of your DNA is lost and remember I mentioned the other day that even though it's not expensive to make DNA whenever you have to make 10,000 of anything it's expensive so now you've essentially wasted all of your 10,000 DNA's one of the advantages of this platform is that we can print the DNA separate from the print mix which means that we don't add the cross linking agent to the DNA when we print it which means that whatever DNA's left over you can freeze it and use it another day and so you don't have to waste everything that you've used so that turns out to be an advantage to us okay so this is what it looks like here we've dispensed genes into nano wells the genes are in this pattern where they're separated by wells that don't have any expression and you can see how accurately it expresses you get very clean expression at each spot and despite the fact that these are very close together you're seeing no intervening spots right this is if you were to just express it without sealing the well so if you just remember I said we sealed the wells of the plastic unpressurized now every now you see how much spreading there is so this is the tendency to spread and this is how well the sealing apparatus prevents the spread essentially blocks out completely and so that's how we got this image here so what you're looking at here is now an app array that has 10,000 features on it all expressed in nano wells this is the DNA print this is the protein print and then we stained it with an antibody to one specific protein that we repeated on the array and you can see how sharp that is single spot single spot no diffusion to any of the neighboring spots and if you plot that in a 3D image you can see it's just exactly where you want the signal to be okay one of the added benefits of this approach that we did not appreciate when we first developed it is that it turns out to be more sensitive than standard NAPPA so we did some comparisons and we looked at this antigen VP1 and we're doing different dilutions of antibody to ask what's the detection limit on the array platform this is standard NAPPA and then this is HD NAPPA we call that high density NAPPA and you can see that pretty much every dilution we're getting much better detection here than we are here in fact it kind of plateaus here at you know maybe 50 or 60 at 1 to 300 dilution at 1 to 300 on this platform it's 450 so the signal intensity is much stronger on these high density NAPPA arrays yeah so this is Ebna up here this is VP1 down here these are two different antigens and you can see that the signal intensity by dilution is much better for the HD than this we even compared the HD NAPPA to the ELISA so you would think that ELISA being a full scale chemical method in a 96 weld tube should be much better in expression but in fact we were able to detect signals here on the HD NAPPA that you could not detect at all on the ELISA and then overall the signal intensity by ELISA compared to signal intensity by HD NAPPA this was nowhere near as strong as that was so it in our lab right now this is probably the most sensitive platform we have for detecting interactions this is just to kind of assure you that we can print these arrays very reproducibly because that's one of the things that you want to be able to do so I don't know if you can see this in this light but this is a single slide array that has four sub arrays on it each of these sub arrays contains 4,000 different spots so we have 4,000 4,000 so a total of 16,000 spots on this slide and we've repeated it in one batch or in a separate batch and then what we've done is we've done an interaction map you know the kind of correlation coefficients that I've been showing you throughout the course every day versus every other day and again as you can see everything here is in the you know close to 1.0 and in the certainly above 95% in terms of its reproducibility so it's every bit as reproducible this platform as the standard NAPPA was and then this is probably more relevant to you all is if you actually that was for protein expression this is now asking if I screen the array with antibodies or serum will the answer I get from array to array from batch to batch be the same and again I don't think can you see the spots there it's a little too dark I think but this is the correlation graph and again you can see that nearly everything is in the high 90% if not 1.0 so the results you get align very nicely and so that led us to this picture here we made the cover of journal of protein research that month and I'll show you this image again what you're looking at here let me see if I can go here hopefully you can appreciate that there's a tiger in that image so this is what you're looking at is an actual protein array we printed different amounts of DNA encoding the p53 protein we then express the p53 protein in the array in these nano wells and then probe the array with anti p53 antibody with a fluorescent tag on it and what you end up seeing is because of the different amounts of DNA you can get an image of the tiger's face so this was the first time we ever did an image using a protein array alright so now I want to move on to a slightly different topic then this is another way about getting a lot of information onto the array so imagine you know we want to test as many proteins as we can when we screen an array in the current NAPPA format on plain glass slides which is what most people can use because all the technology I just showed you is kind of fancy and you have to have special instruments to do it most people would rather work on a plain glass slide the way you have the problem is in our current platform we can only put about 2,300 proteins on that slide the expectation is only true if at each spot you only have one protein but what if you put more than one protein at each spot so maybe what do you all think could that work what would be the issues the idea here one of these ideas that came to me in the shower is that we could print multiple genes at each spot so what would be the issues why would you not want to do that so tell me what you mean by the specificity of the binding say it again that wasn't exactly what I was worried about what other people think say it again so tell me what you mean by quantitation so she is saying how are you going to understand the individual contributions and I think that's a fair concern I've got let's say I put three proteins in the spot right now in the same spot all three proteins will be there if I get a signal I won't know which of those three proteins was the target right I won't be sure yeah will all of them have what the same tag yeah they'll all be captured by the GST tag that's the idea yeah one could imagine a much more sophisticated version of this if you had three sets of proteins with different tags that would be an elaborate method but potentially one that could work for sure right so but nonetheless so that could be an issue so here was my reasoning when the idea occurred to me so whenever I do an experiment on NAPPA and I screen an array of thousands of proteins and I get hits the first thing I do after I get those results is I repeat them right I want to make sure that if my array told me that the antibody bound to protein X that if I really try it again with another protein X it still works right so I believe that all scientists are obliged to repeat their experiments to make sure that they're correct so it occurred to me that if I did the experiment with a multiplexed spot that had multiple proteins I was going to repeat it anyway but at the same time instead of repeating them as mixed I could repeat them as individual proteins and so I would be confirming that they were binding but at the same time I would be identifying which spot was the one that contributed to the signal so I would get sort of two benefits for one in the second round experiment and the net effect would be that I could screen many more proteins on a single slide and then in the end do much less work to get the same information so one of the questions we had now that strategy has limitations to it right one of the assumptions of that strategy is that when you screen the array the first time that the fraction of proteins on the array that will be detected is small right because if the fraction of the proteins on the array is high then the whole time savings thing goes out the window now why is that what do you think so imagine now I have an array that has we'll make a simplified array it has a hundred spots on the array and each of the spots on the array has five proteins in it okay if I screen the array and I get two spots that light up how many possible targets do I have ten possible targets right so my next day when I go to verify I have to do ten different spots and then I'll have done my job right now let's go to the other extreme let's imagine for that hundred spot array that ninety-five spots light up how many potential targets do I have ninety-five times five right and so how many spots am I going to have to do the next day pretty much as if I had started with five arrays each one with one spot each so I'm back to doing the same job I would have done if I had not multiplexed so the multiplex idea works when the hit rate is low and it doesn't work so well when the hit rate is high and so you can actually mathematically evaluate what's the best or most optimal number of spots to mix based on the likelihood of a hit rate and so we actually did that we did the develop the equations and we actually looked at it I'm not going to go through the math we first we looked at the frequency of hit rates for different types of studies that were published in the literature so the first question we asked was on average if you're doing approaching interaction study if you're doing an auto-antibody study of all the targets that people study and when they do their experiments what fraction of proteins light up and so that's what this is this is the percent of identified hits and you can see that this is the five percent mark right here and most of the protein function studies and well I would say all the protein function studies and most of the auto-antibody studies are down in a couple percent range certainly they're less than five percent so that's promising right that means that this strategy could be a big time saver if I can make this strategy work right okay so then the question was what's the optimum number of spots we can do maybe making the assumption that the hit rate is five percent even though I think that's probably a little high for most of them I think it's a fair assumption it's a more conservative estimate if we can satisfy that one we're certainly going to take care of everything more than that and so we did the math and this is the optimum number of genes per spot and you can sort of see that you get more and more savings up until you get to about five spots per gene after which it doesn't really get better and then it gets worse again because of the whole problem of having to do too many duplicates the next day and so the sweet spot here was around five genes per spot right there and this purple line is the this green line here is the five percent see that five percent there that's the five percent line that came out to about five these guys are also pretty good at five when you get up to here when you get up to ten fifteen twenty percent response rates you need maybe to question whether the strategy is a good one for you does that make sense and so the idea here then is you could take on a standard NAPA now that has 2300 spots or 25 let's say for the sake of argument we can do 2500 spots you could print the entire ten thousand open reading frames on one slide doing five proteins per spot and that's what these different colors are meant to indicate five different proteins per spot then you would you would screen that and let's say you get these five hits one two three four five you take each of these five hits that's 25 possibilities you print a second array that has the 25 hits on it and you screen that the next day it does two things for you it tells you which of those five proteins was a hit and it it confirms that it was a hit it tells you for sure yeah that was real and so that's and so that's that would be the strategy does that make sense so you're we call that the deconvolution step it's sort of verification stage and deconvolution stage okay so does it work so what would be an experiment to make sure that it's working so one of the questions that comes up is if I put five spots five protein genes in a spot will they all make protein what if only one makes protein the other four don't now we already know from Napa other other Napa experiments that every almost every gene we print makes protein so we were confident about that piece but you could imagine that somehow mixing them on the spot could be a problem right so how would you test that you could look at you could look at that if you the problem is that for most proteins we don't have the functionality what other ideas we got you could certainly test them one by one in the mix in the mix that's that's how we went about it right so what we did is we said well let's make mixtures of proteins for which we have antibodies so in this case we're testing only proteins that we can come back and test and then we're gonna ask the question if I mix a bunch of proteins together if I test it will I find the protein now is there are there features that we need to consider where one let's if I have a mixture of proteins where one might be made in a greater quantity than another what what kinds of things would I want to think about where could a bias come in it's fine if I have a protein that's 15 kilodaltons and a protein that's 80 kilodaltons would I see a difference what do you think why might I see a difference right so how do how do proteins get made right they get made by adding one amino acid after another using tRNAs on the ribosome right so the amount of amino acids you have to add to get to 15 kilodaltons is a lot shorter than the amino acids you have to add to get to 80 kilodaltons right and so you could imagine that if you have proteins of different sizes in the same spot that the small protein could get churned out a lot faster than the big protein and you might have a bias from that so we tested that too because we want to make sure that the method was going to work okay alright so here here what you have is let me walk you through this experiment on the top where it says M that stands for mixed Napa or multiplex Napa we printed a mixture of five genes and then express them and then we also on the same slide separately printed each of those spots individually okay and then what we did is we probed that array with an antibody that recognized one of the proteins in the five and asked even though it was expressed in the mixture did we detect it and did we detect it as well as we did in the mixture as we did by individual okay and you can see for this IA2 protein we detected it in the mix we detected it much better as a single spot so to some extent this protein did not do as well in this group as it did there but it was still we could still measure it so we wouldn't have missed that in a study here's another one GAD2 we can detect it in the mix we can see it as a single protein here's anti p53 it turns out that the mix was even better than the single protein was here's anti-foss you can see that the mix was about the same as the individual spots and here's I can't read that SFI SFN and again you can see that they're comparable that showed us that the system was basically working now in this top experiment we tried to restrict the study to proteins of similar size so these are all around 100 kilodaltons these are around 60 kilodaltons these are around 50 you can see that they're roughly different sizes so we tried to group the proteins by similar size to avoid that problem I described earlier but then Xiaobo who did this work decided what the heck let's just see what happens if we mix them randomly you know is it a problem and he did that down here so these are this is 100 kilodaltons here's 23 kilodaltons and yet we still detected this one even even though these other even though these guys are much smaller than that one so even though they were smaller they didn't seem to inhibit same as true here this guy is 65 kilodaltons it's with a much bigger protein than some smaller proteins in every case we were able to detect the protein either in the mix or by itself and so that gave us a lot of confidence and these are just some of the data he did much more of it but it gives you the idea that you know if you mix the proteins you can still detect individual proteins in the mix you still have the issue of having to figure out which one is which but that will come later so we decided to try to print a whole array and this is the array we printed we wanted to make sure the array was reproducible so you guys have seen this plot over and over again but we do this on every experiment so I have to show it to you because I want you to get in the get used to the idea that part of the job of doing these sorts of studies is doing the quality control because the experiments only work if you do the quality control alright so this is the array printed with DNA this is the array made with protein and this is doing comparing the DNA from different arrays and then comparing the protein levels from different arrays just to show that they're reliable I'm just going to tell you briefly this is a group of proteins up here printed as a mixed array so that is what we call multiplexed Napa so each of these spots here contains five proteins a piece the same proteins that are here are down here as individual proteins so this was an experiment that we set up so that we could compare how did single individual proteins express compared to the mixed protein expressed right because we're still trying to test the notion that the mixed Napa will the multiplex Napa will give us the same result that we were looking for oh yeah yes yeah we added roughly the same amount of DNA for each one what yeah there's a limit to how much DNA you can print and so I think what we did is we took the normal concentration and cut it by four and then mixed that five times so it was the overall concentration was about 25% higher than normal but it was roughly the same as what we normal print but this time it was made up of five different genes yeah yeah you're yeah so technically if you look at the you know the molarity of the different DNA's that there's technically more moles of the smaller genes it was too complicated to figure that all out and adjust for that and Chavo wasn't willing to do it even though I suggested it so but it seemed to work okay mixing so the idea was you know and I don't remember exactly what our final print concentration is these days I think it's like two or three hundred but they took the standard concentration and cut it by one to one fourth and then mixed that together with the other guys and then and actually what happens is if you just mix one plus one plus one plus one right then each one of those becomes one fifth of the concentration right and then you and the overall concentration is still the same if they're all the same concentration so that's in fact how he did it alright it's a good question though well that's an interesting question in normal biology that would make sense keep in mind that remember here there are no UTR's all of these genes have been cloned into an expression vector they all have identical upstream regions they're yeah it's a T7 polymerase yeah so it's a different circumstance but that's a good point for standard biology yep in today's lecture you have learned that how using a very strong covalent bounding chemistry involving halotags base napa you can now perform high density piece of printing and the assay quality and reproducibility tremendously improved by incorporating these newer methods and that's really a good lesson for all of us to really see that you know a technology can be started but there is a need to improvise it further and bring in the new creative elements so the technology can be much more reproducible and could also serve the much sensitive assays on the same surface in this light the thought of improvising napa for the high density printing as well as much more strong and robust binding was really accomplished by incorporating these new creative methods these concepts will be continued and discussed in the next lecture thank you