 So, I guess in this session here we are, instead of in the morning at 30,000 or 40,000 feet with inspiration, here we are down in the trenches in the mud struggling with real, with muddy problems and trying to do better. I also want to just take 15 seconds to, the first time I've been inside the National Library of Medicine and Physical, the actual building and I am very grateful to them because it was the introduction of Medline that, and it was brand new at UCSD in about 1992 and it was because I logged on and typed green fluorescent protein as a keyword. That fact then you had to do everything in text search mode. Without that I would never have started the work that got me to Stockholm. So it's, that's one of the little unsung activities that way back when made an awful lot of possible. Today I'm going to talk about imaging in general both at the whole body level but also at the improvement of surgery and of course as Dr. Seltzer will agree that they work together very closely and these are the people who have been involved. I do want to make a conflict of interest disclosure that we are trying to commercialize these molecules and that's the only way to get them into the clinic but that does mean that there is a potential conflict of interest. So our molecular target has been proteases in cancer at least because most cancer deaths of course occur due to metastases in distant organs rather than in primary tumors and cells have to break through normal tissue either to, and then the basement memory, either to reach the bloodstream or lymphatics and either one of those requires expression and activation of proteases to chew, to help the cancer cells chew their way through the normal tissue also to release angiogenic factors and a whole complicated series of biochemical steps to engender the tumor microenvironment and I don't have time to go through all of those. The cartoon I could even just steal from a commercial website. This activity of proteases is sufficiently known in fact the website even highlighted two of our favorite of the extracellular proteases that are involved here and matrix metalloproteinases2 and MMP9 just two members out of about a 25 member family that mediate many forms of extracellular matrix degradation and EM Sciences makes reagents for the in vitro imaging of MMP2 and 9 but the in vivo imaging in a patient is one of the unsolved challenges that we're working on. So the scientific strategy that we're following at least is based on first a little bit of known biochemistry that we had no part in discovery which is that highly positively charged peptides full of them things like arginines and lysines but particularly arginines those like to stick to the outside of the cell and if you attach a payload of that's going along for the ride covalently attached that will likewise stick to the outside of the cell and people believe that this is due to the electrostatic attraction of the plus charges from the polycadine against the negative surface charges that all cells have some of which come from phospholipids but even more come from proteoglycans and then having absorbed to the surface then some of this stuff gets endocytosed as shown here and then a fraction of what got endocytosed by the most mysterious step in this process manages to escape into the cytosol and nucleus and this is much beloved of people who work in say tissue culture which is where this mechanism has largely been confined if you want to get impermanent payloads into cells and you don't want to have to microinject them individually this is a decent way of doing so this is one of the great hopes for example of delivering si RNA that Phil Sharp had been discussing there are some people who think you don't have to go through the endocytic pathway they're mainly biophysicists who don't work on cells but rather on model systems now the polycadine can be a whole range of things the amino acids are okay there's no stereo specificity and the very simplest type of polycadine is just a row of arginine which I like because it's very simple and the payloads can be all sorts of things up to nanoparticles in size now the problem is is that this sounds great and does work fair fairly well in tissue culture if you get the conditions right but it naturally caused a great deal of excitement all the way turns out back in the early 2000s with the possibility that we could do in vivo drug delivery or macromolecule delivery in a in a live human being this way all sorts of things we would love to be able to get inside cells and reach into cellular targets with reagents that wouldn't normally cross cells but unfortunately this doesn't tends not to work and I was watched on the scientific advisory board of at least one of the companies that I helped start and watched it fail and be unable to translate this into the clinic and one of the problems is that when you first inject the stuff say into a mouse a whole mouse it sure it goes into the cells it goes into the cells of the tail vein into which you did the injection because this is nonspecific and then later some ends up in the liver probably carried by the huge amount of argument that also is in the blood circulation that competes and finally gets excreted at no point does the tumor get a sufficient loading of the of the payload that you want to deliver and if you just increase the dose hoping for to drive a small fraction or a proportion in you end up with a dead mouse so what we decided was that we needed to make a selective and our particular strategy was rather naive and the reason I emphasized electrostatics is that that's all you have to do if the plus charges love to stick to negative charges on the surface of the cell if we build in a whole bunch of negative charges on fact one-to-one with the positive charges built in the polycane that internally pacifies nullifies the plus charges they pair up probably as a hairpin and the polycane is no longer interested because it's found a nonstick backing paper onto which it already attaches itself and so it has no need to go out and buy into the cell but if we come along and cut the linker that was covalently holding these two together then the two parts can drift away from each other the affinity isn't that high and then wherever the protease activity was that cut the linker it's now like we had a local micro injection of the polycane or like we ripped off the backing paper and then this stuff immediately stuck down and then does its usual thing so we've converted a nonspecific indiscriminate loading that tends to go to the wrong place into what we hope is a tumor targeted protease targeted local delivery and indeed I wouldn't be giving you this talk except that it does seem to work and here is an example of such a peptide in which we have nine negative charges in the portion colored red nine positive charges in the portion colored blue that's nine arginines and they are diamino acids and for us lowercase is our convention for diamino acids then in the middle is a cleavable linker PLG LAG this is L amino acids natural amino acids that can be cut by MMP 2 and 9 which like to cut right between the glycine and the leucine and we this is what we inject now the tumor happens to be pre-labeled with green fluorescent protein this is something we can do in a mouse we pre-transfect the cells with GFP before we put it in it's a great pity no human tumor ever came green fluorescent glowing for us they were not prepared that way they grew spontaneously it would be very helpful to us as researchers if more and even as clinicians if more human tumors were already glowing green but sorry that's only something you can do in a mouse and that's the real weakness GFP that got me to Nobel Prize is extremely powerful when you have genetic control and it can be linked to all sorts of molecular biological events in the cell but when you are not allowed to transfect it's not actually a very good chromophore and it has no special advantage and that is the case in human beings so here this is what we can inject as into the into the vein in this case into the mouse and you see that it lights up the tumor in correspondence pretty well with the GFP which is the perfect genetic perfection of this is the gold standard this is where the tumor really is as though you can see it as well by bright field here is a control if we put in the same probe but the only thing is that we flipped over the stereochemistry of the proline dilucine and alanine in the cleavage sequence now the enzyme can't recognize it because it's the wrong stereochemistry it's unnatural amino acids and the same molecule otherwise of the same hydrophobicity and molecular weight when you inject that it it fails to light up the tumor as shown here but there is a big juicy tumor shown by the GFP now we are too thick and opaque to be directly useful in a whole body sense for fluorescence the weakness of fluorescence is that it doesn't penetrate more than maybe a millimeter or two and even into tissue and it even there it gets blurry and if you happen to have skin pigment or a little adipose tissue which all too many of us have maybe more adipose than we would like it's even worse so we need things like MRI but fortunately this is a technology that simply targets the contrast agent so we can switch the contrast agent to being saying a nanoparticle loaded with gadolinium chelates gadolinium chelates being what we use to light up a region of tissue with so-called T1 weighted MRI and it actually does seem to work so here is a mouse that has a tumor implanted in its armpit and before injection this tumor is not particularly different in grayscale here though there's a little bit of fat that and that wasn't perfectly nulled out in the fat suppression sequence but there isn't a big difference in grayscale between the tumor and the rest of the animal the four circles are a little tanning bed made out of four glass capillary filled with differing concentrations of gadolinium to serve as an internal reference now then however then we injected the animal with this nanoparticle which has six copies on average of the each pep of a peptide on its surface the same peptide I've been talking about and a bunch of gadolinium and then we would have to wait a while for the stuff to clear out of the normal tissue and let the normal tissue background go down but what's left is a tumor that is now glowing because it has successfully or the protease in there has successfully cut the green linkers and left us with these a whole bunch of these positively charged spokes sticking out that nailed this then nanoparticle into the tumor where if the normal tissue gets rid of it and once again here we have the control which is the same nanoparticle with D amino acids in the linker and though there is certainly a tumor there it does not accumulate the probe and why is this important well I think as many other speakers will echo or have echoed there is a tremendous value in early cancer detection if we catch cancers late they tend to be relatively poor in survival if we catch them early particularly when they're completely irrespectable we have a much easier time managing it in the imaging term they say in breast cancer you might have had a localized lump here in a PET scan versus here it's metastasized all over the place unfortunately the current clinical attempts are largely focused in this domain up here because this is where the patients are desperate that and will enroll in any clinical trial because they're at their end of the life and the chances that you'll mess something up when they're up I'm sorry when they're essentially fatal cases anyway it is much less danger and of course this is where the money is and what is horrible to the health care system about treating at this stage compared to that stage is precisely what drives drug development the prospect of large profits and relatively easy clinic trials so this is the way we are now and obviously some other speakers have mentioned in the long run the nation has to do better and we have to begin to focus a little bit more on this area even if that isn't quite but it currently is the favorite area a particular area that we are interested in is also detecting metastatic lymph nodes partly because that's a sort of a low hanging fruit in clinical practice we understand that this is one of the more difficult decisions a surgeon has to make on the spot how many lymph nodes do you take out and for us it also has the advantage of a sort of a yes or no answer each lymph node is either positive or negative whereas the boundaries of the primary tumor are sometimes a little bit vague depending on what your criteria are are what I showed you before were primary big fat primary tumors and more recently in unpublished work we've now been able to see metastatic lymph nodes that are even a fraction of a certainly less than a millimeter or so in size in a mouse model of course this is a primary tumor in the ear which is draining and this case has been allowed to go to invade this lymph node we didn't know it at the time we did the MRI first and then subsequently verified it with an independent pathologist who had not been able to see the MRI and we then compare results on the other side here not lit up is the normal lymph node that is not draining here so we think we're having some promise toward finding metastatic lymph nodes by by MRI but the next advantage of this sort of imaging is that the nanoparticles can actually also carry fluorescent dyes as well as the MRI gadolinium reagent and therefore we can begin to do fluorescence guided surgery and I'll show you a little bit of this in what's coming up here is a big fat tumor visible by MRI and then that's an axial scan here we laid the animal on its back opened it up and sure enough there's the big tumor big fat juicy xenograft in this case and you can see the fluorescence my colleague Dr. Nguyen who is the practicing surgeon cut this out but she didn't use white light she only used white light the traditional surgical mode we sewed up the animal again and put it back in the MRI scanner you can see a big hole where the tumor used to be but also unfortunately up in this sort of armpit region some suspicious extra white dots that we thought were a bit not not quite kosher and indeed after we did his sack sack the animal and did the histology they were indeed nests of cancer cells whereas in this case Dr. Nguyen used the fluorescence that she could see to guide her in a very crude original apparatus just sort of a dissecting microscope with filters and we managed to get a much cleaner MRI and this could be confirmed by some actual survival studies where without any surgery in these two different types of one a melanoma and another a breast cancer model both syngenic so they had the full tumor microenvironment if you didn't operate the animals would be dead in two or three weeks because this is quite an aggressive both are aggressive models with the help of white light surgery we get the red survival curves you can occasionally save the one animal out of many but with the fluorescence guided surgery even with crude equipment we were able to increase the survival to a significant extent though not not yet perfection and I'd like to show you a little bit of what this looks like in now somewhat more sophisticated imaging equipment but this is still nothing like what you know people at Nibib are really proud of and all this really heavy multi-modality stuff this is still very very simple minded but I think it makes the point here is a tumor in the ear can you see where the boundaries of the tumor are I think you might find it a little difficult in fact if I hadn't told you that there was a tumor you might not know it this is one of the classic models but in a moment this is the white light viewed if surgeons have been resorting to for thousands of years now in a moment I'm going to switch the illumination and I've stopped it here this is the fluorescence mode we're now looking at the deep red fluorescence of the sci-fi dye attached to the poly-arginine motif and it is retained in the tumor because that's where the proteases are that cut off the poly anion the polyglutamate region and now you can nicely see I hope the boundary of the tumor right it's actually rather hard to do surgery in this mode because you can't see anything else you can't see your instruments you can't see the blood vessels the rest of the ear so we do a very dumb and simple minded thing which we actually the idea I think may have been published first by John Franjione at Brigham which is to a false color this monochrome image in your favorite color that doesn't normally appear in an animal we picked in the following case I think a light blue and then you superimpose it in real time on top of the white light image and therefore you can get to see the both to both so the surgeon chooses with a foot pedal which view to see sometimes you don't want the distraction of the fluorescence so you have the white light sometimes you want just fluorescence alone and then sometimes in this case this is the overlay and every time you press the foot pedal using a foot so your hands are free you get the next view so here we can see the fluorescence into the lymph nodes in the neck of the animal it's only been shaved it's pretty blurry because of course the diffusing effect of the skin and in a moment Dr. Nguyen will start opening up the skin and there she has done so and now we can see pretty clearly that this is the heavily inflamed lymph node there may be a small amount of involvement in this one and these guys are clear no ear was out here and it's not visible in this particular view but right now here in real time the surgeon knows cut me out and without having to resort to sentinel node detection which is only telling you anatomically which node is downstream and we actually in this case from the anatomy we already know so but it doesn't tell you whether there are metastatic cells in them or not okay so even that though it seemed to give us quite a bit of improvement was not good enough so in our latest attempts what we've done is add an extra dye called size 7 on the polyglutamate end and that's all we've done but now we could get fluorescence resonance energy transfer between the sci-5 and sci-7 and that's surprisingly efficient in the case when the probe is not yet cut this is before cleavage but then after we cleave it then of course the two halves drift apart and a familiar story to any of you who've ever done molecular imaging at the microscopy level then you lose the fluorescence resonance energy transfer between the sci-5 and sci-7 it's an independent spectroscopic instantaneous measure of cleavage and the sci-5 goes up eight fold and the sci-7 goes down about five fold it's a 40 fold change in ratio and we love ratios because ratios cancel out all sorts of imaging artifacts including thickness of tissue absolute concentration of dye quality of illumination a sensitivity of the camera and so on and it focuses just on the biochemical activity and so here we can show you that we are even better now looking at lymph nodes but immediately after injection still nothing there isn't time for anything to have developed the enzyme hasn't had time to work but say two hours after injection now even through the skin and it's blurring with this fret system we can not only see the primary tumor as this red which is the pseudo color for the high sci-5 to sci-7 ratio or a lot of cleavage we can even see the lymph nodes through the skin surrounded by lots of other white bright dots and so on that if you were just looking at any one intensity they could fool you very badly but only when you take the ratio of sci-5 to sci-7 and coded in color and do the spectral comparison you see that the primary tumor and the lymph node are privileged even here you could barely even see the primary tumor mainly because it was in the shadow of most of the animal the the fiber optic beam illumination wasn't hitting it quite right we can confirm that the probes are responding really to mmp2 and 9 the previous control i showed you was just changing them to dmino changing the substrate to dmino acid is really not a very incisive control because any enzyme would probably be confused by switching from l to dmino acids but here genetically we have knocked out mmp2 and 9 here is the wild type animal even just 45 minutes after injection there are two tumors lit up but if we have removed genetically knocked down 2 and 9 together in the tumor and in the host then the tumor does not light up and that is telling us that effectively in these mice the dominant biochemical activity has to be something dependent on 2 and 9 we've even now been able to see metastases on top of the liver this used to be the worst possible problem for us because the liver has this wonderful ability to take up peptides and in general cleave them because it's so full of enzymes but now we can actually see the tumors on top of the liver these are breast cancer breast metastases and the gfp here is marking the tumors to serve as our reference are gold standard and if you look at say the sci-fi image it doesn't really look very much like the gfp and there's lots of blotches that i'm pointing to with arrows that are really really bright in the sci-fi image and don't are not really tumor because they don't have gfp in them as reflecting the reference standard but when you after you do the sci-fi to sci-7 ratio and cancel them out they're bright but they're in the wrong ratio and the ones with the correct ratio indeed pretty much match with the gfp not perfection but they shouldn't be absolutely perfect because gfp and sci-fi sci-7 have very different penetration depths and finally with all of that we feel that we've raised our specificity and sensitivity which wasn't perfect before we had a lot of overlap when we were just doing intensity measurements between say cancer and normal there was a statistical difference in the means but there was a lot of individual overlap in the populations hence the sensitivity and specificity were not perfect but now that we've switched to ratioing you can get what seems to be absolute criteria where you can draw separation lines that clearly distinguish the population and at least on very preliminary inadequate admittedly numbers of tests we can get 100 specificity and sensitivity how well this will hold up when we go into people probably not as well but it's a good place to start from and it certainly made us like the chance of doing ratioing and we've tested this type of system with many many of these substrates not all in the most raciometric mode but here i think are about 11 human tumors and six mouse tumors and the point is that basically every tumor type we've tried has been positive and this is one of the virtues of the system these enzymes are so ubiquitous that unlike your typical antibody marker or so on which will work on some fraction of tumors of one anatomical site this seems to be relatively more universal it may be even almost too universal but that's a point of discussion okay so we think that it's worth improving surgery and because nearly all patients with solid tumors start up with surgery so it deserves to be improved and nybib is relatively unique and liking to do this when i talk to my colleagues in cancer biology or go to cancer biology talks 98 or 99 percent of anybody who thinks that you learn something out of something new we learn about the molecular basis of cancer it's assumed that the only way you translate it into the clinic is through making a drug all we hear about small molecules antibodies over and over again nobody thinks that surgery or radiation therapy which are the actual mainstays of clinical cures nobody thinks of them as intellectual enough you know surgeons are just dexterous right literally with their fingers however i would point out that if you can catch the tumor early enough to completely cut it out the result is an immediate cure relatively low cost compared to the lifetime of medication on the wonderful designer drugs that don't kill the cancer and merely give it time to become resistant and one thing a tumor can never become resistant to is being literally chopped out and dropped them from out high i do not care what else it has it how apoptosis resistant is how much like cancer stem cells it is whatever it's gone of course we that's why we really need the early detection to catch the cancer when it's still circumscribed and of course if we fail in all of this and catch it too late maybe one day we will also be able to target chemotherapy by the same technology now one of the big issues that prevents us from cut doing a better job in surgery turns out i gather to be cutting nerves nerves are actually the structure that is most dangerous to cut as a naïve layperson i used to think it was blood vessels you know that if you cut a blood vessel all this blood would come spurting out but the surgeons assure me that nowadays with coterie and all sorts of things they don't move that many patients simply due to loss of blood but if you cut a nerve if it grows back at all which is far from guaranteed it will be very slow and in the meantime you have a loss of sensation if it's a sensory nerve or paralysis or loss of motor function if it's a motor nerve and the current techniques are not very good and they're mostly not very visual i had never realized that this was a problem until a clinician point my colleague Dr. Nguyen pointed this out to me and so we said okay well let's just try to find molecules that like to bind a peripheral nerve in distinction through other tissues so we just did phage display a very standard biochemical technique against that dissected out nerves versus other tissues found peptides and here's an example of one of the sequences in a 12 amino acid library i do not know why ntqtla k8 you know you can read it why that should bind to nerve but it does seem to when you attach fluorophores we can inject the labeled peptide recent the size not made by phage anymore and it lights up the nerve and the contrast lasts for a couple hours it labels both motor and sensory nerve the nerve does not have to be physiologically active and our probes seems to have absolutely no biological effect on the nerve and i'd like to show you what that looks like here again is a mouse the flank in this case and there is once again a tumor in there can you see the boundaries i hope you'll admit that it's fairly tough until we turn on the fluorescence and there's the fluorescence and now you can see the boundaries of the primary tumor it shades off gradually because the tumor is getting thinner and thinner but there's no doubt where it is that's fluorescence only and this is our favorite overlay mode and today we chose to make the tumor bright green not because it has gfp but again green is something that we don't normally have so Dr. Nguyen has taken off the skin and starting to cut away and the sciatic is being revealed here and at this stage anybody can see what the sciatic is and in there i'm going to stop it you remember that this is the tumor mass this is the sciatic and it looks like it's just beginning to branch here and that if you were going to try to preserve the sciatic nerve you would carefully dissect along its margins and separate it from this big great big glop which is the tumor right well that turns out to be a bit wrong here's an even better view at a slightly later stage do you see the beautiful one branch of the sciatic and there's the other but in a moment now i switched to the fluorescence of the nerve highlighting peptide this is today in a different this is a different wavelength and this is just fluorescence at this level but there's the nerve you see but the nerve has been diverted from its normal anatomical course by the growth of the tumor and it now dives in here and if you were carefully dissecting along here from the previous view you would have just transected the main branch of the sciatic nerve and left that animal paralyzed now when we highlight it in overlay mode it becomes pretty obvious and i hope you can see again in white light how invisible it is and how valuable it is to actually see where you're going so this is the value of imaging in real time during surgery we hope an example of it in a place that is actually even more clinically relevant as in prostate surgery the cavernousal nerves are non-myelinated nerves that run right beneath the prostate under a layer of fat and they're very difficult to see in fact their anatomical course and the ability of any surgeon to avoid them was only discovered a few decades ago by someone who's still alive and has won many awards for that discovery i didn't ever think there was any part of gross anatomy that could still be discovered in the 20th century but sure enough there is and this is an amounts that has far less fat than a male human male and you cannot see the nerves but when we turn on the nerve highlighting peptide we can see through the fat down to the level as long as it's not too much we can see through some layer of that and see the course of the nerves and the urologists are fairly excited by the prospect that they will now be able to see what they need to avoid in order to preserve erectile function which may be not life you know essential or life threatening but somehow men seem to be attached to that just to point out that we're not totally limited to cancer for reasons of brevity i have focused on cancer today but all you have to do is change the amino acid sequence it previously i was showing PLG LAG if we change the DPRSFL it turns out now to be a pretty selective substrate for thrombin and we can watch thrombin activity in vivo and here we can light up atherosclerotic plaques which have thrombin activity as it is known in the plaque and so this is a view of the carotid and in a mouse the carotid is somewhat visible to the trained eye it's harder you can't see through the carotid in a human being because it's more opaque but even with the blood coursing through it we can see the fluorescence that marks what turns out and can be proven by subsequent histology to be aggressive thrombotic plaques of mostly filling the lumen and we think this will be helpful in surgery at least to help the surgeon avoid accidentally bumping it which could cause a stroke if you were operating nearby or if you were trying to put in a stent it's always helpful to see exactly where it is in real time not just by previous static CT images or occasional MR images where you get one view every 15 minutes but to see it in real time millisecond by millisecond as you manipulate the tissue and this is ever more important i might say as we move toward robotic surgery because that all that wonderful stuff that Dr. Pugh was showing us about the importance of touch is sort of lost at least in current robotic surgery that's the great downside is that it does currently gives you no haptic feedback so we're ever more dependent on plain old vision so in summary we have a mechanism for delivering cleavage activated contrast agents and if there's enzymatic amplification which was necessary to get the MRI sensitivity the nerve homing peptide is a completely separate technology but happens to be nicely complimentary we think i don't want to pretend that any of these is perfect we're trying to improve both systems and i gave you a little hint of that you know one can be semi-rational and try to make these things better we think in earliest clinical application will be an image guided surgery precisely because we offer a little bit extra to the already trained clinician who can ignore things if we tell for you that this signal is a false positive and will consistently be a false positive you learn to ignore that early detection will be a little harder and drug delivery will be the most challenging though we are finally making some headway at last on that application too which i'm well aware of is by far the one that will make me most popular amongst cancer biologists because again they still think that this is the only thing you want to work on and this has taught us how important is to work in vivo and really with clinicians to have them tell us what's important and not just ourselves as basic scientists trying to imagine what we think will be useful so with that i've already talked too long i can leave this up in case there are questions for what i would have as a long-term dream we all have to have dreams in unrealistic thank you very much time for maybe one question for dr shen yes please that is the job of the biotech company because again i don't think we are qualified in and as a university lab to push it through the fda they have just started doing it they have so far tested it 30 times the in the imaging docents and what it mainly does is it does turn initially the whole mouse turned slightly bluish from the sci-fi visibly through the skin the urine is intensely aqua colored because that's the left the normal stuff being excreted so far no sign of tox but they're they have to go to 100 fold and they have to scale up a little more so it's looking good but you never know obviously we're at a very very early stage and many many things that have looked good at this stage we all know didn't don't make it we know that horrible attrition rate and i we may know be no exception so let's not get our hopes too high uh we have lots of academic publications on this some of what i showed you is not yet published in the company i'm afraid has not published anything but that's nature of companies thank you roger welcome to the uh imaging community