Scientist Stories: Ichiro Taniuchi, Phosphorylation of Runx Proteins controls Thymocyte Fate

It's a great pleasure to introduce Ichiro. Ichiro has been a long-standing faculty at the RIKEN Center for Integrative Medical Science in Yokohama, Japan. And he joined the RIKEN over there almost about the same time that Mitch and I joined the La Jolla Immunology Institute for— oh God, this conference has really contaminated me. And it was also during that time that I first met Ichiro, and I was immediately really blown away by Ichiro's many talents as a researcher, his ability to think deep and to also in many innovative ways have always fascinated me, and it also resulted in a long-standing collaboration. And I don't think there's any project in my lab that doesn't involve Ichiro. So Ichiro has done milestone work in transcriptional regulation and post-transcriptional modulation that controls gene expression, especially in T cells, T cell development, also lineage decisions and functional differentiation. The RUNX family of transcription factors has always occupied Ichiro's mind, and also today Ichiro will take you really to the very front lines of his research.

Thank you. Okay, thank you very much. I visited here several times, but this is first time after COVID-19 pandemic. I really happy to come back to this place. So today I mainly talk about is my main the interest how these sinusoids, the development of fate is controlled by the transcriptional levels. So this kind of textbook pictures and then all of you may know this, two lineage conventional T cells like helper T cells, cytotoxic T cells are differentiated from common progenitors. We know that if the TCR is restricted to class I MHCs, they eventually choose the way to become cytotoxic T cells. While these TCR in the precursor is MHC class II specific, they eventually choose the way to become helper T cells. But we still don't know how these two distinct type of MHC mediated positive selection T cell signals induced two different fates. Because this— although this class I, class II MHC are different structures, and then CD4, CD8 co-receptors molecules have different affinity class II or class I, there are molecules that transmit the T cell signal after T cell engagement is common between these 2 types. So Lck showed the higher affinity to CD4 cytoplasmic tails, and some studies proposed cross-mediated T cell signal is maybe stronger than the cross-1 mediated T cell signals. But we still don't know yet how is the real difference in the T cell signaling by cross-2 or cross-1 MHC mediated or engaged types. But to explain how the helper versus CD4 helper versus CD8 cytotoxicity choice is regulated, the Allsingar group proposed one model, so-called kinetic signaling model, in which the CD8 coreceptor is temporarily downregulated after positive selection. Therefore, T-cell signals is stopped in class I restricted T cells, and then shorter T-cell signals induce cytotoxic T cell fate. That is the core concept of the kinetic signaling models. But if we overexpress the CD8 co-receptor from the ROD26 locus, and cross to that transgenic into the MSC class II minus background, we still see CD4 negative conventional T cells. So this is like contradictory to the kinetic signal models, because constitutive CD8 expression is not sufficient to induce CD4 helper fate in all MHC I class selected cells. So I feel still kinetic signaling model is a good model, but cannot explain all aspects of the CD4 CD8 lineage choice. That is my opinion. But my or our approach is to understand how this lineage-specific expression of co-receptor is regulated, like why CD4 is specifically expressed on helper T cells. And when I started or joined this field as a postdoc in Dr. Dan Littmann's laboratory, They take up the bottom-up approach and then already find out this important cis-regulated element in the CD4 locus. So there is the silencer region, intronic silencers in the first introns, that is quite important to shut off CD4 expressions during differentiation towards cytotoxic T cells. So my mission as a postdoc is to isolate the fatty acid nuclear proteins that bind to the silencer and regulate silencer activities. And then I found RUNX transfer factor proteins is essential for activating silencers in the CD4 locus, and RUNX complex is essential for the CD4 silencing during CD8 cytotoxic T cell development. Since then, I have been studying in this field to understand how is transcription regulations to control hepatoblastocytokinesis. In 2000, around 2005, other groups isolated another transcription factor called Th4. This is essential for hepatocyte differentiation, and then Th4 ranks can antagonize each other for to repress each expression. But overexpression of Th17 precursors by transgenic approach induced cytotoxic helper fate in all MHC class I-specific T cells. Therefore, the Th17 populations in cytotoxic T cell class I, MHC class I-restricted T cell differentiation is quite important. Then we asked how this ThPOC is expressed specifically in MHC class II-specific T cells, and then we isolated another silencer in the ThPOC locus upstream of the promoter 1. So this is the genetic study, what happened if silencer is knocked out from CD4 locus. So then CD8 T cells cannot shut off the CD4. They appear as a CD4-positive, CD8-positive, double-positive phenotype in the periphery. Similarly, if Thpo silencer is knocked out from the GFP reporter locus, in the presence of the silencer, the GFP expression from Thpo locus is limited to the CD4-positive T T cells, but removal of the silencers induced Thpoq-GFP expression in CD4+ T cells. So these silencers at each locus is quite important to regulate the lineage-specific expressions. But how is activity of the silencer regulated? That is a key question to link how the T cell signals are eventually translated to regulate the silencer's functional state. In MHC class II signal cells, the silencer state is inactivated or non-functional, therefore Th4 co-stimulatory genes can be turned on and then it starts to be expressed or maintain the CD4 expressions. But MHC class I engagement induced activation and of silencers and they become functional, and then therefore Th4 co-stimulatory gene expression is suppressed in here. As I pointed out, the RUNX proteins is quite important to regulate the silencer activities. And this is a structure of RUNX proteins. They can make heterodimer with common partner C-beta proteins through random domain. And lambda domain is also important for the DNA binding. But there is another evolutionary conserved sequence in the RUNX protein at the C-terminal end of that RUNX protein. It's called BWRPY pentapeptide. But this pentapeptide has been shown to be important to make association with Grou2/TRAe corepressor family proteins. So if we remove this BWAPY motif from both RUNX1 and RUNX proteins, the CD8-positive cells fail to downregulate CD4. It's very identical to CD4 cytosol knockout. Also, the ThPOC gene is not well repressed in CD8 T cells. But this BWAPY sequence is quite important CD4 cell receptor activities. And if we removed the TRE depressor family proteins, 3, 1, 4, 3 proteins together in mice, their CD8 T cells cannot downregulate the CD4. So these are almost very similar phenotypes of the phenotype. So VWP motif and then Grouchy TRE is quite important to activate the CD4 cell activities. So there is another transcription factor having the BH RH type DNA binding domains that belong to— and also this BRH hair-related HES protein families have BWRP double RNA sequence. This is quite similar, just tyrosine and tryptophan is different at this end. And then this is a disordered probability, then having the Y have more disordered structures than the E-tryptophans. So this HES1 protein showed a very high affinity to TRE family proteins through binding to the WD domain. Former studies dissolved structures and double P double pentapeptide pocket into this WD proper structures. And in the same time, this paper showed that using double RPY peptide, the crystallization cannot establish. It's very— show very weak affinities. So there are just one sequence difference between these two transcription factor families, and it has been proposed that having the tyrosine, not tryptophan, would endow the novel regulatory mechanism to RUNX proteins, allowing them to function as both activators and repressors, while HES1 proteins mainly work as a strong repressors. I had a question, is this true? And then test this hypothesis to examine what happened if we change tyrosine residue in RANK protein to tryptophan in RANK1 or RANK3 genes. And then this is the result if try to knock in genome editing onto the RUNX1 endogenous genes, then the tyrosine is changed to tryptophan, but we couldn't get any ripe heterozygous mouse from this approach. And then we changed to analyze the embryos, and there is no clean mice obtained around embryonic 11.5-12.5 days. But they are going to die, right? RUNX1, germline knockout mouse. And then germline knockout of RUNX1 eliminates hematopoietic stem cell generations, but similarly in knockout— knocking mouse, even in the heterozygous, they do not have hematopoietic stem cell generation. So therefore, just one amino acid change at the terminal residues dramatically change the function of RUNX1 proteins. But it was so almost impossible to establish this mutant line because heterozygous mice already die as embryos. And this is the case if we change RUNX3 terminal tyrosine residue to tryptophan in heterozygous or homozygous. In this case, the homozygous can survive and they'll grow, but they show some neurological phenotype and dying around 4 weeks. But in this T cell compartment, in the heterozygous mouse, there is clear severe reduction in the CD8 T cell numbers, and also in the homozygous, even in the mature thymocyte, there is no mature CD8 single positive cells generated. And then if we examine what happened as heterozygous or homozygous in terms of IL-7 receptor expressions, the homozygous mouse show clear reduction of IL-7 receptors, therefore they cannot receive the survival signals, are going to die. Also in the cell cycle analysis showing RUNX3 double WRPW homozygous mouse showed quite G0/G1 arrest, they are not proliferating. And this is the analysis of the histone modifications using the heterozygous RUNX3 WRPW mice, and then we see also we performed the RNA RNA-seq analysis, and then this TNK2 gene is quite repressed in the heterozygous mouse. In this TNK2 locus, we see binding of RUNX even in the wild-type heterozygous mouse, but in the case of heterozygous mouse, repressive histone modification mark is spreading from the RUNX binding site. So this suggests that this double AP double forms may function as strong corepressors, and then RUNX target genes that can be activated with wild type can be suppressed in the presence of double AP double form. That is consistent with the concept maybe double AP double sequence show a higher affinity to TRICOP repressors. So interestingly, this RUNX3 double AP double homozygous mouse also lacked the peripheral lymphoid structure like lymph node, and also Peyer's patch is gone. And consistent with the phenotype in the adult, if we see— examine the gut lymphoid tissue-induced cell populations, RANK3 double haploid double homozygous show the complete lack of lymphoid tissue-induced cell differentiation. This is one of the innate lymphoid cell lineage. But a curious thing is this is the structure of ARHGAMA-C encoding ARHGAMA-T transcription factors, and then we find the RUNX binds to this region within ARHGAMA-C locus, and then these regions contain 4 RUNX binding recognition motif. And if we remove these regions by genome editing from the R-gamma T GFP reporter locus, the enhancer deletions completely eliminated R-gamma C expression in lymphoid tissue inducer cells. That is the reason why this RUNX activity is necessary for lymphoid tissue inducer distribution. Differentiations. By the way, we already published RANK1, not RANK3, is important for lymphoid tissue-induced T cell differentiation. By using the conditional knockout of germline mutants of RANK1, RANK1 is essential for lymphoid tissue-induced T cells, but In this case of RANK3 double P double mutant mouse, RANK3's mutation also shows a phenotype. That indicates RANK3 interferes with RANK1 functions if RANK3 has double P double mutations. So that's another question. RANK3 and RANK1 make heterodimers, and we test these possibilities by using the mouse brains, and the RANK1 is tagged with TY1 and RANK3 protein is tagged with FLAG. And using the RANK1, RANK3 protein tagged with different sequence and show that if FLAG antibody is used for immunoprecipitations, RANK1 is co-immunoprecipitated. So this indicates RANK1, RANK3 can also make complex. And by using the AlphaScreen in the cell-free screen system using the recombinant protein generated and fit and using the different tag, we also confirmed at the protein level directly the RANK1, RANK3 can make heterodimers. So this makes quite a change of the concept because RANK protein is known to make heterodimers with the CBA beta protein. But in this case, we now know that RANK1 and RANK3 also can make interactions, make heterodimers. Okay, so summary of the tyrosine to tryptophan replacement in endogenous RANK proteins. We found that RANK double mutant protein showed dominant effect in a dosage-dependent manner, and RANK double mutant protein might be a strong or constitutive repressor. But what happens if tyrosine is changed to other residues? And then we also generate other mutant lines. Why tyrosine is changed to the phenylalanine or glutamic acid. As a short summary, phenylalanine replacement also mimics the change, tryptophan replacement, but phenotype is slightly milder, showing intermediate. But if we change tyrosine to glutamic acid, it mimics the lacking of the tyrosine. It's rock maybe affinity to the TRE Groucho family proteins. So that means this structure of the amino acid, this YFW is similar. And then Y shows this low affinity to the TRE, but phenylalanine tryptophan may have higher affinity to TRE proteins. I will skip this. It's already shown. But during working with this endogenous RANK3 mutant, RANK3 mutants, because the cells are going to death, it's difficult to do molecular biology or biochemical analysis. And then we then asked what happened if RANK3 double P double protein is expressed in double positive precursors. By using the Rhoda26 snipping system. So we put the RANK3 cDNA in here, but to get the RANK3 expressed, we have to remove the so-called stop locus sequence by CD4-CRISPR. And then this is just the CD4-CRISPR only and the wild-type control. And then I'm showing the CD4, CD8 expression profile in the mature thymocyte populations. And here is a TH work in the RANK3 protein expression by flow. The overexpression of RANK3 wild type didn't change so much the CD4/CD8 expression profile, but here overexpression of RANK3 double P double, we see clearly the deduction of CD4 single positive with increase of CD4 minus CD8 minus corporations. And if we examine the Thpok expression in yeast cells, there is no Thpok expressed. But if we consider what endogenous RANK proteins is expressed in precursors, it's not RANK3, it's RANK1. So we also tested what happened if RANK1 is overexpressed in the same system. And here, result is very similar. If RUNX1RP double protein is expressed, the CD4 single positive population is clearly reduced or almost gone, and then CD4 minor, CD8 populations appears. So that means CD4 single positive thymocyte differentiation is interrupted. That is MHC class II restricted cells. And then before going to the next one, we also tested the effect of methylation to fenylalanine to glutamic acid changes. And again, the fenylalanine replacement showed intermediate phenotype between the tryptophan replacement. Also CD4 single positive population is reduced. And then to examine what happened in MHC class II restricted cell differentiation, we cross to all the ranks from double PD-1 transgenic mouse to OT2 T cell transgenic. So otherwise differentiation of MHC class II restricted T cells. Also are crossed to the beta-2 minus β2�� deficient, this is the MHC class I deficient mouse. And then MHC class II cells, OT2 T cell-expressed cells are converted to the CD8 single positive in this case. And similarly in β2�� background, MHC class II selected cells are also converted CD8 single positive cells. So in the presence of RUNX1 or RUNX3 double RP double forms, redirection of MHC class 3 T cells towards the CD8 happens. And this may be the silencer element is activated during the MHC class 2 cell differentiation. Therefore, across THPOXIDASE knockout, TRE knockout, to RUNX1 double P double mutants, and also found the passery, only the passery, the CD4 single positive cell differentiation is restored. So that means TRE function is involved in the d-direction of the MSC cluster cells and the silencer is involved. Tiro, quick question.

The CD4 cells that you see with the terminal tryptophan silencer, are those OT2 positive? Are these mice not on the Rag background?

So the lower left. These ones? This is the— not Rag background. This is just the RANK1 double PD-1.

OK, so they may be expressing—

Natural T cell. So this is just a—

So they managed to get through and express CD4 despite the repression.

Yeah, but that is variation in the mouse.

Yeah, okay.

So this is the kind of schemes to explain what happens. In normal MHC distal cell differentiation, that the silencer state is nonfunctional, therefore Thfoxcd4 is expressed. But in the presence of RUNX1 double P double forms, the silencer is activated and/or functional, and therefore Thfoxcd4 expression is not well expressed. So in the presence of RUNX1 double P double mutant form, the silencer state linear specificity of the synapse is disrupted. But I wanted to point out, even in the presence of DAPA-P double form, the CD4 genes at preselection stages is not downregulated. Only after the postselection, the CD4 genes is downregulated in the presence of DAPA-P double. Therefore, something happened when cytocytes received the T cell signals. Important biological regulation happened on the terminal tyrosine residue or RANK protein to link T cell signals with cell fate decisions. But these genetic approaches clearly showed the terminal amino acid in Lungs protein must be tyrosine. Why tyrosine? What happened to tyrosine? So you guys may think tyrosine can be phosphorylated, and we search this hypothesis and then try to detect the phosphorylations, and finally we succeeded to optimize the conditions for mass Here we used the JAK-ATC line and expressed TY1-tagged RUNX1 proteins. And first immunoprecipitated RUNX1 protein to enrich the RUNX1 protein for the subject in the mass spec, and then performed the mass spec analysis. And then of course we found nonphosphorylated peptide peak but we also found a phosphorylated peptide peak here. So this means at least in JAKA cells, RUNX1 protein can be phosphorylated at the terminal tyrosine residues. But this is consistent with results deposited in the public database PhosphoSitePlus. Database, and they also use the JAKAT cells and then reported terminal tyrosine residue can be phosphorylated. So this is not just artifact confounded by other laboratories. And having these results and see what is the good condition to detect RUNX tyrosine phosphorylations, we Use OT1, OT2 transgenic mouse crossed with the TY1 tag1 from endogenous protein and separate OT2 CD4 single positive cytocytes, OT1 CD8 single positive cytocytes, and then examine phosphorylated state at terminal tyrosine residues. Here is the peak, and you can see here the OT2 CD4 single positive show less phosphorylated forms. It's very difficult to detect. So there are phosphorylation state of RUNX1 proteins, so the difference between CD4 single positive versus CD8 single positive thymocyte.

Can I ask how many mice you used for this?

I don't remember how many mice.

Was it a lot?

Well, maybe 5 to 10 mice combined, or just to use less than 5 to prepare enough number of CD4 single positive. So now we know this terminal tyrosine can be phosphorylated. The next question is what happens if the terminal tyrosine is phosphorylated? Are there conformational change happens? And we started collaboration with other laboratory in the NML who can do simulations of the protein structures. So here is the Y-type type, DAPY motif. This is the tryptophan, this is the phenylalanine, and this is the phosphorylated forms, and this is glutamic acid. These simulations predict the phosphorylation of tyrosine may close mimic structure to tryptophan or phenylalanine and then do some change of conformations. And then we tested whether binding of RUNX protein to TRE is dependent on phosphorylation by using the adenylylphosphatase treatment in co-IP experiment. So this is just to confirm ARCA phosphatase treatment eliminates any phosphorylated proteins by using the pY100 antibody that detects phosphorylations of any tyrosine residues. And then use anti-TRE3 antibody to immunoprecipitate RUNX Y-type form, but in this case after the algal phosphatase treatment, the association of RUNX with TRE3 is abrogated. But when same experiment performed using WRPF forms, even after the algal phosphatase treatment, still this mutant RUNX proteins can associate with TRE3. So these, and then another transcription factor like TCF1, it also has the association with TRE family with different motif. After the alkylphosphate treatment, its affinity seems to be increased in this transcription factor. So we don't have still direct data showing phosphorylation of the terminal residues can induce the interaction with TRPC3, but given all findings together, I think phosphorylation of RUNX terminal residues may induce conformational change similar to the one by tryptophan or phenylalanine and then induce interaction with TRE proteins. But this is quite a different situation from other tyrosine phosphorylation because people use phenylalanine replacement to eliminate the phosphorylation, and then in most cases phenylalanine replacement abrogates protein-protein interactions, like phosphorylated tyrosines can bind to SH2, but tyrosine to phenylalanine change abrogated such SH2 interaction. But in the case of the long terminal tyrosine residues, phenylalanine phosphorylation may mimic structure to tryptophan or phenylalanine. That will change the concept of what happened at tyrosine on the amino acid residues. So consistent with that, if we perform the TLA-3 binding to CD4 silencers, the CD4 single positive thymocytes showed a higher binding to TLA-3 compared to the CD4. So upon the phosphorylations of tyrosine residues, that was confirmed higher in the CD8 OT1 CD8 CD8-positive thymocyte have more binding of TRAC to the sinusoid regions. So now we know this tyrosine can be phosphorylated, then next question, what kinase is involved in this phosphorylation? So to search for unbiased screening FATS, kinase have showed affinity to RUNX protein, we also set up the alpha screening in cell-free systems. And then we generate the kind of the panels or optimize the custom panels covering around 400, 500 genes encoding any kinase including serine/threonine kinase, tyrosine kinase, but I just show the result of the tyrosine kinase. The top-ranked among like top 75, we noticed that FINN, ES, or RIN belong to non-receptor tyrosine kinase families, come up to the upper cases. And also in the mass spec analysis, we first did immunoprecipitation using anti-TY1, and at the same time we can see fat protein is co-immunoprecipitated with RUNX if we subject these samples for regular mass spec. And we often notice LCK, FINN, ZAP70 can be co-immunoprecipitated with RUNX. So therefore we focus on— also one paper using the HEK293 cells, overexpression of Src kinase and RUNX in these cells, and then Src kinase families induced phosphorylation of RUNX proteins. So this is quite artificial, but Src family proteins can phosphorylate RANK protein as a template. We then asked if Lck is involved in phosphorylation of RANK's terminal tyrosine residues using the variant JAKA cells called JAK1.6 that lack the endogenous Lck activity. So this was cloned by the long history of the study using the JAKA genetic Here is JAK1.6, we never see phosphorylated peptide P. So at least in JAKA, if LCK activity is abrogated, RUNX phosphorylations will not happen. So if LCK is involved in the RUNX phosphorylations, it's also raised another question because LCK is known to be an important molecule to initiate the tissue signaling in the cytoplasm. So RUNX is a transcription factor mainly located in the nucleus. How these two molecules interact? If so, where it happens? And then we utilize inside proximity ligation assay in which these two molecules, using the two different antibodies detecting two different molecules, if two molecules associate or locate in the close proximity, the antibody-mediated amplification signals give us some PLA signal. So this way is very good way to know whether molecules can in the close proximity and where it happens. So to set up these conditions, we use the one setting as a positive control. We have the TbyOne tag at rank 1 at N-terminus, and we have antibody that detects BWIPI sequence specifically. So in this case, by using these two antibodies, we can detect intramolecular, intrarank 1 molecule interaction that can work as a positive control for pairing. And then to address interaction between ranks and LCK, we set up the anti-TY antibody and anti-LCK antibody. If two molecules in the cross-proximistic, we have the PL signals. So this is the result of the PLA assay using RUNX1 and then C-terminal RUNX1 antibodies. As you can see here in DP thymocyte, CD4 single-positive thymocyte, CD8 single-positive thymocyte, the PL signals appear mainly in the nucleus. That's coincident with these RUNX mainly located in the nucleus. So this is the result of RANK1-RCP interactions. I think this PLA system works well, and you can see here the PLA signal showing dead spot mainly located outside of the nucleus. So this is higher resolution of the result, and we clearly see this is maybe nuclear membrane and then signals located in the cytoplasmic or just beneath of the plasma membrane. So RANKs and LCK can interact in the cytoplasmic, and the same analysis is performed onto RANK3 with different flags. And you can see here, then similar PR signals are detected associate it in the cytoplasm. In this case, we can just use the CD8 single positive samosas because RUNX3 is only expressed in CD8 single positive. And we also tested another interactions. For instance, RUNX1 can interact with DAP70 here, and then CBF-beta, it's a partner of RUNX proteins, also can associate with Lck in cytoplasm. Plasmic, but Bcl-irnb, it's another transfer factor, cannot interact with Lck, but RUNX1-Bclb, these two transfer factors can interact in the nucleus. That can be also compounded by co-IPs. So, and then we carefully examined our PRA analysis and ask, is LCK-RANKS association different between two lineage or not? And then this is the summary showing RANKS-YRSK interactions. So RANKS-YRSK expression itself is higher in CD4 single positive, lower in CD8 single positive. However, RANKS-YRSK interaction signal is always dominant over are higher in CD8 single positive. So Lck-RANK3 interactions is always dominant in CD8 single positive than CD4 single positive thymocyte. And then we ask what makes the difference. This is on to the acquired CD8 single positive phenotype, or MHC I-mediated T cell signals induce more LCK-RANKSMAR associations? And to answer that question, we utilized Thpok-GFP knock-in/knockout mouse. In this homozygous mouse, without Thpok expression, there is no CD4+ for single-positive thymocytes. Even the MHC-class II restricted cells show CD8 single-positive phenotype, but these MHC-class II restricted cells can be marked with the GFP expression because they want to be expressed to Th1. So we sought out CD8 single-positive cells with GFP positive or GFP negative, and then performed the PLA. And this is a result of 3 independent experiments, but GFP-negative cells show higher PLA signals than GFP-positive cells. This indicates MHC class I-mediated T cell signals induce more RUNX-LCK Lck association. So this is the current model. If NetTissue is engaged by MHC class I, it induces more Lck-RANK associations, and that eventually induces phosphorylation of RANK at terminal tyrosine residue. Recruits the TRE corepressor family, and then the silencer is activated, and then Thfox4 gene is shut off. In the case of MHC class II mediated T cell signals, RUNX-LCK interaction is weaker, and then phosphorylation of RUNX is not induced, therefore silencer is not activated, and then Thfoxcd4 is expressed. But from our genetic approach, changing the tyrosine residue to phenylalanine or tryptophan, some differentiation of the MHC class I-mediated cells is also disturbed because in that case all ranks forming transporter complex work as a depressor type. But in the natural situations, I think phosphorylated form of RUNX is balanced to non-phosphorylated form, and then only phosphorylated form is recruited appropriately at genomic regions and then exerts their job, while non-phosphorylated RUNX may activate a complex and then activate the caspase-8 or of young gamma genes. So phosphorylation's proportion of the phosphorylation ranks must be balanced, but we don't know how that balancing is regulated. So I think this finding of RSK-RANK1 associations make our understanding be advanced, but still I But sometimes raise important questions. How does MHC class I and MHC II engagement induce different T cell signals? We still don't know how that is, what is the molecular basis. The proportion of RUNX phosphorylated forms should be balanced as I explained you. So then to understand the mechanism For these two questions, I think we need to analyze protein interactions, interactors of Lck-γ7T, CD4, CD8 again in post-selection sinusoids. I mean not in vitro, in post-selection sinusoids in vivo. So we are now trying to establish in vivo bio-ID systems, putting BRA or related molecule-directed endogenous LCK CD4CD8 to have what molecules interact, preselection or postselection in cluster-selective, cluster-selective cells. And now I would say LCK BRA bioid systems works efficiently. Countries, and we are trying to do same things to CD4/CD8 cytoplasmic proteins. But then another question, how is recruitment of phosphorylated RUNX1 to the appropriate genomic region regulated? That's quite mysterious, but it's better to know to rate antibody that decarbonates the transactor with specific post-translational modifications like phosphorylation, acetylation, ubiquitylation, and so on. And then understand, uh, transactor protein with specific modifications are recruited to specific regions or not. Then also it may be ideal to isolate the protein content complex on the specific genomic region, but current technologies still are not sufficient to enable us to do this, but we are trying to do this using so-called on-site BioID system to catch up protein complex on the specific genome, but very difficult and we are still on the way to find the good conditions. But I think these studies provide a new concept. The T-cell signals are converted to the post-translational modification trans factors, and then that specific forms regulates this specific target genes. And then again, then how is other molecules involved in the translocation ranks to to the nucleus and the other factors to control gene expressions. I think one interesting molecule is now the senescence here, the other kyokazu is working because this senescence shuttling is quite important. So I wonder after the phosphorylations of RUNX, how RUNX find a good way to come to this senescence that is maybe controlled shuttle proteins. So that I hope our collaborations will continue to understand these kind of important next questions. So the 3 postdocs are involved in this project, and then Satoshi Kojima, who was the former postdoc in this institute, now going to the Kanosawas is analyzing phenotype of RUNX3 double P mutant mouse, and then Chihiro is analyzing the transgenic RUNX double P double phenotype, and then PRA assays was performed by Kazuki Yokoyamasu. And then other proteomics analysis is performed by Kozakos, and then AlphaScreen provided at the Sawasaki Atakehikomi, and the simulation is performed by this group, and then high-resolution imaging is provided by Yokosuka at Tokyo Medical University. Thank you very much.

Thank you, Chiro, for this very exciting talk.

Sam?

That's very exciting stuff. Have you tried doing chips with your antibody that recognizes the nonphosphorylated C-terminus? So you're talking about doing PTM-specific ChIPs, but that one would tell you everywhere it's sitting where it's not phosphorylated.

But there is antibody can detect RUNX proteins, but we don't have any antibody that detect phosphorylated. So that to do, nonphosphorylated can be organized phosphorylated together.

Couldn't you do a pan-RUNX ChIP and then a non-phosphorylated RUNX ChIP and see where they overlap?

But non-phosphorylated— but antibody can detect both non-phosphorylated and phosphorylated. So there is no antibodies that distinguish phosphorylated.

Well, then how are you doing your PLA assay? What was your PLA assay?

PLA assay is we use the TY1 tag, right? That recognize both phosphorylated, non-phosphorylated. So, right.

No, no, no, your other antibody for your PLA assay for your TY1, and then otherwise it's anti-RCT1, anti-JAP72.

So, can you go back? Do you have that?

Can you go back to the PLA slide?

It brings 4.

So this anti-WrPy antibody, again, you're trying to ChIP with that?

So this is raised against WrPy nonphosphorylated. Right.

ChIP with that. Then you get everywhere in the genome that it's sitting nonphosphorylated.

But the— So chip with this, chip with this one, and you'll see it runs everywhere.

And then chip with this one, and you'll see only where it's not phosphorylated. I know this looks insane, but then you can see everywhere that it's not phosphorylated, right? You subtract the signal between the two.

But we don't know yet this antibody against double PY. Can still recognize phosphorylated or just nonphosphorylated.

Or you can IP, you could use this to IP the peptide and see if you're IPing the phosphopeptide, right? And here you're saying that it doesn't recognize phosphorylation, right?

That's the assumption. I think they're good ideas, but we haven't tested antibodies for CHIP.

Okay, that's something I've always wanted to try, and it sounds like you're set up to—

Yeah, good.

Everybody makes antiphosphor antibodies, but you could also make a naked peptide antibody and then you know it's not phosphor modified at all.

Yeah, we have spending almost 7 years to raise phosphorylated specific antibodies, but that's difficult because this is the terminal sequence. Sure. And we changed company twice, but monoclonal antibodies against phosphorylated form is very difficult.

I still think you could try with these two.

Thank you, thank you very much. Maria?

Thank you very much for this fantastic talk. So you said that RUNX1 and RUNX3 form heterodimers in thymocyte. Do they form heterodimers also in BRIEF?

Difficult, these cells? I just— there's a possibility they can make heterodimers, perhaps without CBF-beta. But if both proteins are expressed in the same cell, there's possibilities to make heterodimers.

Okay, so is it RUNX1 expressed in CD4 itself? And so we know that RUNX3 is expressed in TH1 cells. And what happens with your tryptophan terminal RANK3 mice where you have CD4s maturing? So in the periphery, have you tested if the peripheral CD4 T cells can differentiate into TH1s having this tryptophan at the end?

We haven't tested effect that is differentiation in this system yet. We spend our much effort to understand this differentiation cytosite at this moment. But we have the material if you are interested. You can do it. I will provide.

While you're at this side, I have another question. PLA system, is it possible to pull down or IP the associated DNA with these two proteins that are binding together?

No, no, this is inside staining test system. We cannot pull it down.

You cannot pull the DNA associated. So my other question is about the downstream methods. Are you sure —polycomb spreading, right? Histone modifications. Yeah, polycomb regression.

So how—

you showed one locus, but how widespread it is, and have you looked at other things such as acetylation or chromatin accessibility?

At a genome-wide level, we see increase of histone— depressive histone marks. Other locus too. So this is not just the one locus, but other generations I don't remember this.

So K27me3 is a little bit context-dependent, right? I mean, it is impressive but also not necessarily.

So having another mark for chromatin accessibility and So this trimethylation of 27 is quite induced or enhanced in many locus. This is just one locus, typical ones.

This is just an example of many different.

So to get at your first point, You could do a ChIP for one protein, elute, and then ChIP for a second, and then sequential ChIP, and that will tell you what DNA both proteins are sitting on.

But I think those re-ChIP things don't work very well in general.

I have the same feeling that sequential ChIP is very difficult.

So this was nice that if you can do two, but at the same time anytime through the associated DNA, you can come up with so many different ideas, but through this—

I really liked your conclusion slide when you talked about characterizing the complex, because you had several kinases that came out in your AlphaScreen assay that interacted. Do you think that this is, in these cells, it's a multi-kinase, multi-subunit complex? Do you think you can catch that in your IP with— cross-linking or some kind of strategy.

But first of all, this alpha screen is just use the protein generated in the feed and then just mixed, quite artificial system, but to know whether it can interact in that systems or not. So it's a one-by-one screening. So but actually kinase complex, how this kinase complex involved, then what we know is by using the gel cut, LCK is involved and we don't know yet DAP70 is involved in the phosphorylation, but DAP70 ranks can interact. So I'm not surprised if LCK, DAP70, 70 ranks make kind of the complex, and different kinase complexes are involved in that micro— what's the microenvironment? That's a space.

The nuclear membrane. Do you still think that FINN and YESS are involved, or you've ruled those out? Because as you said, you pulled that out in the in vitro.

Yeah, but these are just expressing on the FINN area, so you can have redundant functions. So maybe the FINN and also in the cross-prosmistase. But limitation of PRA is just we can see interaction between two molecules. Three molecules are very difficult. So we can test it one by one at this assay.

How much RUMS protein is in the cytoplasm in a resting single positive —have you tried to quantify that? And if you give them a TCR signal, does that increase?

We don't have data, but the accurate quantification may be difficult, but we can try to separate cytoplasmic region and the nucleus and see Western blot how many That's what I see with themis, that it's done beautifully.

So that's part of the reason why I asked that question. But in just thinking about it, is your idea that once it's phosphorylated that it might be released and then go to the nucleus? Is that your thinking or speculation?

Well, in other papers, The HEC2 lines? Yeah, HEC2 lines also screened phosphorylation-retained DUNK protein cytoplasmic. But that is kind of opposite to our model. Right. So I don't believe HEC cells' results reflect the results in the degradation of T cells. In the RUNX field, CBF-beta protein is known to be also located in the cytoplasm because they don't have the nuclear localization, but they can make a heterodimer with RUNX and then go to the nucleus. But compared to the CBF-betas, most of RUNX proteins located in the nucleus. That is common, like a consensus in the field. So I don't think Someone carefully examined the cytoplasmic expression of RUNX proteins compared to the nuclear. Yeah. Okay, thank you. Great, and then— So you are saying that your LCP and RUNX1 are interacting while phosphorylation is going on.

So in IP I'm seeing like you have 3 protein complexes. So what is the role of another protein? I mean, LCK is directly interacting to your RUNX1 or like to that third protein?

If you ask, and the fact is—

So like when you are doing IP, I saw like there apart from LCK and RUNX, there were another protein, third protein. Third protein. Yeah, there were, there were like, uh, you, um, in IP from LCK, apart from LCK, there was— I, I forgot the name— another, another protein also you pulled out by IP. So while they are interacting to— what is the role of that third protein that is directly interacting to RUNX, or it is interacting to LCK, then LCK interacting to the, uh, your, uh, Difficult to answer because this co-IP, we cannot judge if the direct information or just molecules involved in the complex.

So, but LCK, FynDAP70 can interact directly or they make a complex and then if BRCA is participate in that complex. That may be the one possibility. Kinase complex recruits RUNX2 temporarily. So it's difficult to say this direct interaction or indirect, but they are in the complex, in the close proximity.

Yeah, really interesting. I really wanted to ask how you envision that as the double-positive thymocyte gets its CCR signal, how can it discriminate between an MHC I and an MHC II signal? Do you think it's the amount of kinase that gets recruited to the synapse is different in both situations and that leads to a a different phosphorylation state of the RUNX protein? And then the other question that relates to that is, so I know you really nicely see a different localization of RUNX in CD8s and CD4s, but do you see different amounts of phosphorylation already at the BP stage in cells that will become CD4 or CD8? So when you look in an OT1 or an OT2 system, there's already something early upon T store engagement that's driving them one way or the other that would be ahead of the kinetic signaling model as has been described?

But the first question, we don't know how they distinguish the class 1, class 2 MHC signals by themselves because some people suppose this cluster signals may be strong, but strong is a word, but what is the actual molecular changes in terms of strong?

Well, I guess what has been described is that CD4 versus CD8 cells CD8 as a co-receptor when it gets recruited has a different amount of LCP associated with it or different dwelling time of LCP there? Do you think that plays a role in what you're seeing?

Well, one of the things I'm now currently thinking is that those experiments show CD4 LCK is high affinities before T-cell stimulation, but after the T-cell stimulation, what happens? It dissociates quickly or still maintains association? I don't think someone carefully compared that. So also the phosphorylation event after the T cell engagement is mainly analyzed in the in vitro system using the T cell lines. So by stimulation of the anti-T cell antibody or anti-CD3 antibodies, and then I don't know much whether the people tested real using the APC-mediated stimulations. Because for the changes between CRISPR-mediated to CRISPR-mediated. So there is no clear data, what is the difference, are there any difference in the TCR signaling mediated by CRISPR and CRISPR-Cas2 MHCs. And then the second point is we are trying to test phospho-DPPB. Thymocyte already have phosphorylated RUNX4 or not. But lacking of the specific antibodies, what we can do is do mass spec analysis. But my assumption is it has already phosphorylated forms because RUNX-mediated repression, so TRE-mediated repression already happened in DP thymocyte by using that. In that sense, there are two possibilities. The crass-mediated signals actively remove the phosphorylations or just do the passive dephosphorylations by like cell divisions mediated or crass1-mediated signals actively work to maintain the phosphorylation or some enhanced phosphorylation there. 2 different possibilities, but currently I don't know which one is the truth. But to understand what molecules regulate transphosphorylation at that site, then we will have answers.

Okay, so again, Ichiro, thank you so much.