Scientist Stories: Kerstin Meyer, Human Cell Atlas

Okay, next we move on to Kirsten Meyer, and Kirsten joined the group as a principal staff scientist about 3 years ago, and she's been an absolutely essential pillar and my, my, my right-hand person all around since then. I think it's about 3 years, isn't it? Um, or is it 4?

4, 4, 4, 4.

Yeah, time flies. Great, and we look forward to your talk.

Great. Well, thank you for inviting me to present here, and I'd like to echo Mike's sentiment that it's just so lovely to give a talk and actually see the people you're talking to and get some feedback. So it's really great to be here, and of course we're here because it's Sarah's 20th anniversary of the founding of her lab, and I just wanted to say a big congratulations, Sarah, both on the amazing scientific achievements that have come out of your lab But in a way, almost more importantly, this morning has been wonderful to see all the people that you've mentored that have gone on to have wonderful careers, both in academia as well as in industry, as we've just heard. So I think that's just really impressive and lovely to see. But as Sarah said, so I'm not going to talk much about my history other than to say that we actually started off in very similar domains, both doing our PhDs at the LMB where I did a PhD on the gene expression of immunoglobulins. I then worked on multiple different things, including breast cancer, but then when I joined Sarah, I did a full loop and came back to immunology. When I joined the lab— how do I move this forward? Yeah, when I joined the lab, we were just in the stages of really expanding the Human Cell Atlas projects, and here's just the mission again that Sarah already went through earlier. But what is very interesting is that right in the very first HCA white paper that was written, there was already a strong emphasis on the fact that in order to understand tissues, you not only need to have the cellular branch, the single-cell component, but you also really need to have a spatial component so that you're able to map your single-cell transcriptomes back into the tissue context. And that is a story I will tell you a little bit more about today after my initial joining in many different projects. Really, I've sort of differentiated out into a lung person, and we've got projects both in adult lung and fetal lung and in pediatric airways. But everything I'm going to talk about today is going to be on the adult lung. And so we set up an experiment where we analyzed multiple different locations in the lung that were then sequenced both by single-cell RNA-seq as well as single-nuclear RNA-seq. And importantly, we also carried out a fair bit of Visian spatial transcriptomics. And this generated a wonderful dataset that's summarized here that comprised data from both cells and nuclei, and also from multiple locations. And this output was really the result of a lot of hard work by Ido Maddussohn, who really spearheaded the analysis and interpretation of the results. But then importantly, we not only had the single-cell data, but also, use the Visium. And so what I'm showing here at the beginning is really a proof of principle that we could take the single-cell data and map it back into a Visium slide that here depicts a small airway. And crucially, the algorithm Cell2Location that Vitaliy Kleshchevnikov developed was really amazing to help us interpret the data.

It's been accepted in Nature Biotech.

Ah, fantastic, congratulations. And so while this is a proof of principle showing that the ciliated cells can sit in the small airways, we then also had from our data set actually chondrocytes that many previous data sets were lacking because we also had nuclear data, because chondrocytes sit in the cartilage and were therefore often hard to obtain. And very satisfyingly, the chondrocytes did indeed map to the locations of the cartilage in the small airways. But despite— so there's actually a huge amount of data in here, and actually we described about 77 different cell types or states, whereas previously there were only 58 known. So we feel this is quite a big step forward, and this work is actually now under review at Cell, and we're keeping our fingers crossed. But so today I just want to pick out a couple little vignettes from this dataset. So here you can see the fibroblast populations that Elo annotated, and actually 7 of these were novel and had not previously been described. And I think this really nicely illustrates the utility of the spatial data. So focusing here on these yellow fibroblasts, and then here the pink fibroblasts, we mapped them back into tissue section via Visium. And here you can see very nicely how these cells here mapped to— here is the mapping of the chondrocytes, and then here we have fibroblasts that map around the chondrocytes, hence we call them perichondrial fibroblasts. Whereas on the other hand, down here, this yellow population really mapped very nicely around the bronchi, so we refer to these as peribronchial fibroblasts. That incidentally had already been described in quite a lot of detail in the mouse. Then moving on, another cell type that we found very exciting were these immune recruiting fibroblasts. And these were interesting because they expressed a number of chemokines that are known to attract lymphocytes. And indeed, when again we did the same thing, mapped that cell type to a visium section, you can see very nicely how this cell type maps into this region of immune infiltrate. And actually, when Ilu then took some data from Rasa's paper in the gut, where she described fibroblastic reticular cells and follicular dendritic cells, so again, immune recruiting cells, they mapped to the same region, further underscoring the importance of this particular fibroblast in orchestrating the immune environment in this tissue. Moving on then to the epithelial cell type populations. Here we described two new populations, the submucosal gland duct and myoepithelial cells. Just for completeness, I've shown some markers here, and then here again indicating that indeed the Visium confirms that these duct cells map to duct tissue and the myoepithelial cells roughly to the same regions. Which allowed us to really have confidence that all the new transcriptomes that we described make up the structure of the submucosal gland together. But the really interesting thing came then in the next step when we looked at the immune cells and annotated them and also had the resolution to distinguish IgG and IgA plasma cells. And again, Same process, mapping it back to the visium, you see very nicely here that the IgA plasma cells specifically map to the regions of the glands. And this is just a single example, but Elo actually took multiple different visium slices and annotated them into different tissue regions. So here, for example, the immune infiltrate, the small airway, the glands, and so forth. And here you can see how the 3 different gland populations mucus, serous, and duct cells completely co-localize with the IgA plasma cells. Whereas in contrast, the IgG plasma cells really map to the immune infiltrate, really speaking to how the zonation of these different immune cells, it must be tightly regulated. And so then, of course, we had the whole transcriptomes, and we had the wonderful tools that Rosa and Mariana had developed. We also used self-chat. And this allowed us now to investigate the underlying signaling circuitry that determines this specificity. And so this work that was done by Amanda Oliver, and one cell, one signaling surface that stood out in particular was the presence of this CCL28 in a number of the different duct cell types that is able to interact with CCR10 on B plasma cells.. And importantly, CCR10, the receptor, is much more highly expressed in IgA than in IgG, really starting to tell us a little bit about how these duct cells are specifically able to recruit IgA rather than IgG. We then went on to look at a number of other signaling pathways. For example, we also— Amanda also noticed that the duct makes APRIL, which can interact with a number of different ligands on both memory as well as plasma cells, so via TAKI and BCMA. And importantly, this interaction is known to stimulate B-cell maturation and also longevity. Similarly, IL-6 was known— was found to be recruited by the duct and is known to upregulate IgA secretion in plasma cells, but also has a strong interaction with, with T cells, naive T cells, potentially to recruit T cell help. So overall then, we're beginning to see that we have a whole immune niche where we have B cells, we have epithelial cells recruiting them, and potentially also T cells. So we wanted to see whether we can actually see this in situ. And here's some beautiful staining that Nathan Ricoz from Mena Klakwer's lab did for us. Here you can just see the nuclear stain. And then the EpCAM beautifully shows us the gland structures here with the vasculature. And then importantly, we see strong overlap of the gland structure with the IgA staining, also quite well with the IgD staining, but very, very little IgG staining. And we do also see T cells in this region, further underscoring that they're likely to be more interesting interactions between these different cell populations.. And so this really brings me to the conclusion of my talk, and the point I'm really trying to hammer home here is that the key aspects to make these conclusions was, one, to have the— first of all, the single-cell data, then the spatial data that allowed us to map the single cells back into the tissue, but then also to have this total unbiased a selection of all different cell types, starting with the endothelial cells that help to recruit immune cells from the vasculature, then the epithelial cells that are able to attract those immune cells into a particular niche, and then further allowing us to really home into the different signaling mechanisms that recruit these cells and then maintain their longevity. And that really brings me to the end of the talk. There's many, many people to thank.. But the people who spearheaded this analysis and the work were really Elo, Amanda, Vitaly, and Anna. But of course, many other people contributed. CGAP, the Celgene wet lab teams, Mena, and of course also all the patient samples that we got via CBTM. And so as my last slide, I just wanted to show our lab. So this was in 2018, not long after I joined. And then we also had a a recent photo which was only taken a few weeks ago. It's got mostly Teichmann members but also a few other people. Sarah unfortunately isn't in here, but therefore I've shown her separately there. But I hope you notice that if you want to stay in the lab longer than 3 years, it's absolutely imperative that you're no taller than 5 foot 3. Anyway, and with that, I think I want to finish. And just thank Sarah for her bold, brilliant, and kind leadership.

Thank you, Kirsten. So that was a great talk. It was a great talk, full stop, and it was also perfect after Mike kind of showing the importance of BCR and BDJ. And of course, both of you worked on that at different times. Well, during your PhDs. So Kirsten did her PhD with Michael Neuberger. On regulation of V segment, and Mike was talking about that, and now you sort of come back to it. Some funny comments about the height on the chat. Questions for Kirsten? I mean, one thing that I would say is that I think it's a beautiful example of interpreting the data And generating data and cluster, processing data in an automatic way is valuable and crucial. And what this shows is also the depth of biological interpretation that you can, that adds knowledge essentially. I think it's on what someone's asking, when is this published?

It's on research school. So, um, we are one of those preprints. We're hoping to put it into BioRxiv next week, but yes, at the moment it's just submitted and we're anxiously waiting for reviews.