 So, our next speaker is Wes Pike from Madison, and he's going to be talking about the role of enhancers in control of gene expression. And Wes, for people who don't know him, is one of the pioneers in understanding the regulation of gene... I'm sorry, bone metabolism. Thanks, John. I really would like to thank the members of the organizing committee for inviting me here. I don't actually know who that was, but I think it was John. I got an email at some late night email asking whether I was available, which of course I said yes, because last year I attended the meeting and I'm thrilled to be here again. And I'm also happy to be in a session on common diseases because I will touch a little bit on that at the end of my talk. So I would like to first acknowledge some of the people that have been involved in the work that I'm going to do, that I'm going to be talking about. Mark Meyer in particular who's here, Songman Lee, Nancy Minkowski, and Melda Onal have been integrally involved in some of the work that we're going to talk about today. Now I don't really need to show this slide because you're all aware of enhancers and the question is, what I would like to do is tell you a little bit about our enhancer work and specifically that involved in bone cell-specific enhancers. But of course that raises the question of why one would want to study these interesting regulatory regions of genes. And of course I don't have to go through all this except to say that enhancer governors certainly govern cellular phenotyping through selective control of gene expression. And this I think is actually a rather daunting task because while we're talking about cell selectivity and so forth today, there's at least 700, there's at least a couple of hundred cell types in most organisms. And then there's also the transitions that occur within these cell types as they undergo differentiation. And there's also the environmental impact that's actually been shown with regard to epigenetic landscapes with respect to the same type of cells, for example, macrophages that Chris Glass has shown and others have shown that could impact what we see there. So there's some major reasons why we should understand these elements. And of course the single nucleotide variants and the SNPs here in human disease are extremely important. But as was just mentioned, the context in which these SNPs or variants exist are primary determinants of how that SNP will actually behave regardless of whether it's located in a binding site for a particular transcription factor or not. And then I also happen to think that understanding enhancer properties are important for therapeutic development because I do think that there's a unique specificity there with respect to the regulation of genes that may have therapeutic value. So everyone needs an entry point in terms of trying to understand how enhancers work and that can certainly be evolution, it can be differentiation, it can be disease as we're talking about at this meeting extensively. But also if you're going to study enhancers you may be interested in systemic regulation of these regulatory elements by such hormonal systems as the vitamin D system. And that's of course what we have used to not only understand the vitamin D system itself but also to get into the regulation of enhancers. Now the focal point of the vitamin D system of course is the vitamin D receptor itself which is a nuclear receptor and following its cloning a number of years ago there was much work that was accomplished to determine basically three principles. The first one was to identify the motifs in which the vitamin D receptor interacted. And here the second was to realize that the vitamin D receptor acted as a heterodimer complex with another nuclear receptor called RxR and of course that complex RxR participates as a heterodimer partner for other nuclear receptors as well. So there's a incredible complexity involved in that. And then finally we actually really discovered perhaps to our disappointment and this was a number of years ago that really the function of the receptor was simply to provide a binding a redirect a binding site for other co-regulatory complexes that include the epigenetic histone acetylation complexes, nuclear soma remodeling complexes and so forth. And this is just three of many that have been identified. Now the problem with all those studies early on was the fact that it was very receptor centric and in fact there was only a few genes that we could look at. So a few years ago we began to with the availability of new approaches we began to look at use chip-chip and then chip-seq analysis to actually search for the vitamin D receptor cystrom and to understand where it was binding on in various cell types including bone which was a particularly interesting one and so this is a study that we did in bone cells and one of the important aspects of this we can isolate bone cells, grow them in culture and we can actually also differentiate them. So we can actually look not only at the vitamin D receptor cystrom that is the binding sites in what we're going to call POBs or pre-ostioblasts or they can be oftentimes they can be MSCs, mesenchymal stem cells and we can actually look at the transition then when we differentiate them into mature mineralizing osteoblasts and so we took these cells and basically treated them with either a vehicle or 125 dihydroxy vitamin D which of course is the ligand for the vitamin D receptor and then just did a chip-seq analysis of these cells and the first thing you'll see here is that in the primary osteoblasts, the precursor osteoblasts there's a large dependency upon 125 for binding so we get about 1,000 binding sites in the absence of the hormone and about 7,000 in the presence so very dependent for binding on the ligand itself which is not actually true for a lot of the other nuclear receptors that utilize RXR as their partner. Now when we looked at the RXR cystrom it was actually much larger than that for the VDR but the important point here was when we looked at the co-localization of the two, about 60% of the vitamin D receptor sites co-localized with RXR suggesting of course that we're supporting the idea that in fact this was truly a primary function of the receptor was to utilize RXR as a partner and when we looked at the response elements we looked at a de novo examination of these binding sites. The most common element that turned up was an AGT-TCA duplex separated by three based pairs and this actually was pretty satisfying because this was the first, this was consistent with exactly the sequence that we identified in the first gene and I won't tell you how many decades ago that was but in any event these were then features that were consistent and validated those earlier studies but as has been talked about today already and was a surprise to us we learned that in fact that most of the binding sites were not located to promote proximal but were in fact entronic and intergenic and could be 10 to 100 kilobases away from the promoter itself and this has major implications has already been discussed in actually now identifying the sites of action of transcription factors because this is the general feature and the target genes they regulate. This has actually created a big problem for everyone in the room who's studying gene regulation. Now the final thing is when we look at the osteoblast mature mineralizing osteoblast what you can see here is an incredible down regulation of the cisterome for the vitamin D receptor going from around 7,000 binding sites down to about a little bit less than a thousand and this was actually rather surprising to some extent. So we were curious about whether there was a consequence to this and so we actually did some micro rays and some RNA sequencing analysis of the precursors in the mature osteoblasts and you can see that there's a striking decrease in the number of genes that are targeted from about a thousand down to maybe around 400 or so. So there was incredible down regulation distributed both in genes that were up regulated as well as down regulated. Now this actually wasn't entirely a surprise because we know that 125D3 is actually trophic for the vitamin D receptor and therefore in the absence of 125D3 the receptor actually decreases its expression over time as these cells actually mature. So this wasn't an incredible surprise but what was the surprise actually was that we found a collection of genes actually a fairly large collection of those 400 genes which showed differential target gene responsiveness to 125D3 that was entirely due to differentiation and here's some examples here of the comparison of expression of response to 125D3 and then the data tracks, the chip seek tracks for those individual genes across those sites. So the first one would be called 2A1 which you can see is down regulated in precursor cells but is relatively resistant to 125 action when the cells are actually mature and I think it's not surprising if you look at the track down here there's a very strong vitamin D receptor binding peak right here and in fact that's almost completely obliterated when the cells are mature. So this pretty much accounts for at least some of the segmented genes that are downright that are lost, the responsiveness is lost, however if you look at ENPP1 and 3 these are mineralizing regulators involved in the osteoblastin ability to mineralize their matrix you can see that in fact their responsiveness to vitamin D over time is actually increased during the differentiation process. And if you look at this locus these are genes that are actually separated or closely spaced and in fact they're regulated by the same enhancer but the surprise is that even though these are increased in responsiveness to the vitamin D hormone the amount of vitamin D receptor that's bound here is decreased by at least tenfold or more and which was a surprise. Finally if you look at IGFBP5 you can see it's not responsive in earlier cells and then becomes responsive to vitamin D as the cells are mature and again in the same way we've lost a lot of vitamin D receptor binding, this is the scale over here so we've lost a lot of vitamin D receptor binding and the real question here then is how do we sensitize these cells to the effects of 125 D3 even though the receptor is now dramatically down regulated and its binding activity is not near as extensive and there's two answers to this and they're not simple but one is epigenetic and the other is transcription factor activity separate from that of the vitamin D receptor itself. So I don't have time to go through the epigenetic changes that are involved in differentiation. These two descriptors here are really things that are one of the great contributions of the ENCODE project and it's something that we have utilized extensively and I won't go through these but there are certainly signature histone modifications that are dynamic and that are changed rather dramatically in response differentiation and they contribute to the responsibility of secondary regulators that certainly includes the vitamin D receptor and other signal regulated nuclear proteins. We also noticed that vitamin D actually can provoke changes in histone modification particularly acetylation and this goes back to the one of the co-regulatory complexes that I talked about. So the second part of it however is that there are transcription factors that are involved that are localized to many of these regulatory regions that bind the vitamin D receptor. We noticed in fact that if one looked at, we had some hints that two master regulators of the osteoblast rungs two and CEVP beta might be bound to those sites that the vitamin D receptor and its partner RXR were bound to and in fact it turned out when we did a chip seek analysis you can see about 40% of the vitamin D receptor binding sites actually have both rungs two and CEVP beta and the binding sites actually correlate quite well and in fact what was most interesting about this was when we really looked at this there was a definite organization of these binding sites where the vitamin D receptor was in the center and rungs two and CEVP beta were on either side of the binding activity for these. We call this a consolidated enhancer and we gave it the name osteoblast enhancer complex. I'm sure that it probably works in other cell types besides the osteoblast but certainly there was an organization here at the same, at a specific enhancer there was other proteins that were involved in potentiating or controlling the regulatory capability of the vitamin D receptor itself and we think that that's not unexpected and certainly interesting. So key features of these enhancers thus far, distal binding site locations, you're all pretty familiar with this, modular features where there are multiple transcription factors that can be bound, epigenetic enhancer signatures can be seen in these sites and in fact the most interesting is transcription factor systems are highly dynamic not only in the differentiation which I've showed you but clearly and well I'm sure there'll be others that will talking about this in differentiation, maturation and disease activation and they have a major consequence on gene expression. So I want to touch on now just two more genes that highlight some of the additional features of enhancers that we discovered. The first one is MMP13 which is regulated by vitamin D and during differentiation it produces collagenase 3 which degrades exocellular collagen as skeletal sites and bone but it also has very profound effects in cancer and in fibrosis and a variety of other areas, it's regulated by many things besides vitamin D and the interesting thing about this is that all previous work on this gene has focused exclusively on the promoter proximal region of MMP13 and this is true of so many genes now that when we begin to study a gene, we try to ignore all the previous data that's been generated by standard Luciferase assays and transient transfection analysis because it's almost always wrong. So you can see and we contribute a little bit of that although most of our stuff turned out to be not unreasonable but I like to say that it's either 5 percent right or completely wrong but in any event you can see here this is another gene in that collection where there's really very little regulation in the precursor cells but when you mature the cells there's an increased sensitivity to 125 D3. So here's the chip seek analysis that looks at this locus, here's the MMP13 in this direction and here's the promoter region and what we're looking at is both precursor cells and osteoblastic cells and we're looking at a variety of different measures and I think the first thing you can see is that in addition to the promoter itself there are three different regions here at 10 kilobase, 20 kilobase and 30 kilobase that suggest that they might be enhancers and interestingly this one binds the vitamin D receptor, the 10K binds the vitamin D receptor, the 20K tends to bind CEP beta and the more distal one at 30K seems to bind largely rungs to which is a master regulator and so we were curious about this if you look at these sites they all align with certainly with H3K4 monomethylation but you can also see some examples here where there's a difference between the precursor cells and the mature osteoblasts suggesting that there may be some differential activity and that correlates with the expression as well particularly down here with H3K36 tri-methylation. You can also see some valleys here that John Stam has highlighted and others have highlighted where the binding activities actually are lodged within the valleys that are associated with this thing. So to actually explore this in more detail we've used CRISPR analysis to CRISPR deletion gene editing methods and in this particular case we took these cells the parental cells and we made a variety of deletions using CRISPR and these are the deletions that we made. We did delete a region around proximal to the promoter but not including the promoter we deleted the 10 kilobase region as well as the 30 kilobase region. We also knocked out the VDR and we also knocked out rungs too. So we had a collection of daughter cell lines and in addition to that the beauty of CRISPR is that you can make sequential deletions so that you can examine the combinations and we did that as well. And then we looked at these daughter cells for the activity of MMP13 both basely and in response to 125. Now these are 125 D3. So these are the data in these individual daughter cell lines. This is the parental cell line here basal and 125 inducible. This is just a clonal cell line as another control. And I'm not going to take you through all this because it's relatively complicated here. This is the general activity we just expanded the axis in this cell on this side. But what I want to show you now is that in fact when you delete the promoter proximal region right here you do lose some basal activity of MMP13 expression but you retain to a large extent the 125 D3 responsive induction. If you remove the 10K enhancer you'll see here that you also get a reduction in the baseline expression of MMP13 transcripts. But interestingly you lose the induction as might be expected. You lose the induction by 125 D3 which you can see here is going down. But in fact now you get actually a suppression and I'm going to talk about this in the next slide. Just a comment on what our hypothesis is. Finally if you look at the 30K deletion you can see you lose all basal activity and almost all the 125 D3 responsiveness. And if you look at the transcription factor knockouts you get the similar correlations. So these were pretty interesting to us and these data along with some other data that I don't have time to show led to this kind of not unexpected chromatin interaction model where we've centered the 30K base enhancer and we have aligned it with the 20K base, the 10K base in the promoter proximal region. Largely because of 30K enhancer impacts which binds rungs this master regulator but not a pioneering factor influences the activities of these other enhancer regions and they then in turn determine the output of the gene. And one of the additional experiments that we did was simply to see if we deleted the 30K base enhancer what happened to vitamin D receptor binding and rungs binding to some of these other sites and CBP beta binding activity. And it was almost completely obliterated when we removed this distal 30K base enhancer. So we think that this is very central and we also would suggest that this could be very complicated. For example if you found a SNP in this region in this 30K enhancer that would affect the binding activity of rungs and therefore impact the ability of this gene to be regulated by vitamin D but you might search long and hard to find a vitamin D responsive region in this 30K element because it's not there, it's over here. And so I think this has an impact on these ideas. Just to address the repression the 125D3 actually down regulates rungs expression and the dependency of this gene on rungs is extremely important and so if you down regulate the rungs by a secondary activity or a separate activity by 125D3 we think that leads to something that looks like suppression. Now I'm going to turn to this last gene ranked like in the ranked ligand gene and I'll go quickly through this. Ranked ligand is an osteoclastogenic factor that we know is involved in bone remodeling and certainly involved in osteoporosis but it's also involved in immune function, it's also involved in smooth muscle activity and we think it's involved in athosclerosis as well. And so we were interested in that, this is the gene and you can see immediately we've looked at a variety of enhancer regions upstream of this particular gene, you can see there's at least 10 different enhancers, these are the binding activities of various factors that we've looked at, I won't go through this in detail but the key elements here are that these individual enhancers, here's the gene itself actually is TNF SF11, you can see that these enhancers actually are mediating largely mesenchymal or osteoblast lineage type cells and these actually are regulating the expression of ranked ligand in the, in T cells and B cells. So we've knocked, we went in and knocked out these five different enhancer regions and we've looked at the phenotype of expression of ranked ligand in these cells and I won't go through all the details of this because it's rather extensive but the reality is that what we were able to learn from these is that they have individual temporal, hormonal and tissue specific activities that are very important. So just in the last three slides and I'll go through this very quickly, we've done a lot of other work but we're actually interested in the impact of ranked ligand in the atherosclerotic plaques and what we did then was cross the enhancer deleted mouse, the D5 enhancer deleted mouse into the apoenol mouse, treated these animals then with a high fat diet and looked at some of the features of this animal, phenotypic features at 12 and 18 weeks and the first thing you'll see is that ranked ligand is up-regulated in the atherosclerotic plaque but when you cross it with the D5 you can see a complete down-regulation. Bone is affected as a control in these animals and we get an induction of osteopatrosis due to a down-regulation of ranked ligand. So when we looked at the atherosclerotic plaques by micro-CT and by histology, this is sort of the method, you can see the calcification that shows up in the aortic plaque, this is the region that we're looking at. One can actually quantitate the plaque size as well as the degree of calcification and this is a summary of that experiment, these are the visual aspects of it, this is the histology but at 12 weeks what we see was that in the apoenol, the single mutation, most of the animals developed calcification, however in the apoed5 double null where we've reduced the expression of ranked ligand, only one of the animals out of eight actually developed calcification although this was lost by 18 weeks. So the conclusion here is that ranked ligand plays a significant role in the atherosclerotic plaque calcification perhaps by promoting bone formation which if you notice the bone activity ranked ligand is a bone resorber on bone but in fact it may promote bone formation. This is the last slide, I won't go through this in any detail because I'm out of time but this is just a list of all the different aspects that we've learned from some of these studies and we would anticipate that these would be the kinds of things that one would have to do after we learned bioinformatically how to link enhancers to the genes that they regulate. Thank you very much. Nice work, so I'm wondering about the deletion of the 10KV enhancer at the MMP9 locus when you do the CRISPR deletion of that region your results tend to suggest that you completely lose the regulation of the MMP9 so I was wondering if you've actually assessed whether the 3D structure was completely disrupted by the deletion of that element and or if you could specifically just delete the VDR element or the RXR component to actually just specifically target the binding of those factors to minimize the new potential impact that there would be in the CRISPR deletion that would influence a 3D structure of that regulatory unit. Yeah, so we wanted to go in and in essence a gross way to look at it but that was also the reason why we knocked out the vitamin D receptor itself so I didn't talk about that, the data were there but I didn't talk about that but basically when you do that then you do get a very specific response that's dependent upon the receptor for MMP13. The problem with that experiment or not the problem but the consequence of that is that the vitamin D receptor regulates other genes and runks actually if you knock it out actually has a profound effect on the differentiation process itself but in essence you get the same kind of data when you do the combination. We would probably go back in and make the mutation and you're right that would isolate the vitamin D effects specifically. Well so yeah, so those kinds of studies that are being done and have been done, the ligand binding site is actually buried within the receptor itself and so once the ligand gets in this is sort of standard nuclear receptor sort of study is basically it's buried there and it rearranges the receptor and it actually opens up a binding site in LXL motif interaction that will facilitate interaction of co-regulators and there are a number of them, they can't bind all of them at the same time clearly so we don't really know much about the dynamics of which or where, when and how so you're probably right that there may be some sequential, first of all there's probably selectivity at individual enhancers that may convey unique activities out of that particular complex as it's bound to DNA. It's very possible but I don't believe that there are studies that really have now looked at that yet.