Clonal analysis of in vivo hematopoiesis

我们利用分子技术对造血干细胞和祖细胞（HSPCs）在体内的行为有了新的认识。我们继续积极开发和利用具有高多样性的31-35碱基对遗传条形码的慢病毒“条形码”，用于标记单个造血干细胞和祖细胞，以研究非人灵长类动物模型的体内造血。我们的合作者卢荣首先设计了这种非常强大的方法，并将其应用于研究小鼠造血。我们现在已经用条形码的自体CD34+细胞移植了20只猕猴，并且能够在一段时间内（长达5.5年）以定量和高度可重复性的方式跟踪数千个HSPCs个体的造血输出。我们已经取得了许多重要和新颖的发现，包括缺乏证据表明灵长类动物中有共同的淋巴祖细胞产生T细胞和B细胞，直到移植后才有B细胞和T细胞的共同克隆衍生，以及更早的髓细胞和B细胞的共同克隆衍生。我们还首次发现了自然杀伤细胞（NK）主要部分独特的谱系来源和生活史。CD16+CD56-细胞毒性NK细胞在移植后48个月也不与B、T或髓细胞共享条形码。在体外和小鼠模型以前没有能够阐明NK细胞谱系关系。我们与NHLBI的Rick Childs博士小组、Tulane非人灵长类动物研究所的Miti Kaur博士小组和瑞典斯德哥尔摩Karolinska研究所的Yenan Bryceson博士小组合作，继续使用来自这些猕猴的条形码细胞进一步解剖NK细胞的体内个体发育和NK细胞的体外扩增过程，这与过继细胞疗法的发展高度相关。我们已经将克隆独特的条形码定位到具有“适应性”表型的NK细胞，在小鼠研究中与NK记忆有关，在人类研究中与假定的NK对病毒（如巨细胞病毒）的适应性反应有关。在具有适应性表型的NK细胞中，我们发现寡克隆和大量扩增的个体克隆，在其他谱系中没有发现贡献，并且具有起伏模式，表明对特定环境线索（如病毒感染或再激活）的反应。我们的数据提供了克隆NK反应的第一个直接证明，可能解释NK记忆。随着体内NK的消耗，相同的克隆再次出现，而不需要从高多克隆的HSPC中招募。根据人类移植和献血者研究的相关数据，我们假设这些克隆是由对巨细胞病毒的反应产生的，并通过在巨细胞病毒阴性的猕猴中进行条形码移植来验证这一点。我们继续寻找这些细胞的前体，在来自血液、骨髓、淋巴结、肝脏、阴道和肠道淋巴样聚集体的表型纯化样本中追踪优势克隆。我们还应用我们的方法来了解其他种类的先天淋巴样细胞（ILC）的起源，包括肺中的ILC2和肠道中的ILC3。

We have utilized molecular techniques to gain new insights into the behavior of hematopoietic stem and progenitor cells (HSPCs) in vivo. We have continued active development and utilization of lentiviral "barcoding" with high-diversity 31-35 base pair genetic barcodes utilized to label individual hematopoietic stem and progenitor cells to study in vivo hematopoiesis in the non-human primate model. Our collaborator Rong Lu first devised this very powerful approach and applied it to study murine hematopoiesis. We have now transplanted 20 macaques with barcoded autologous CD34+ cells, and have been able to track hematopoietic output from thousands of individual HSPCs over time (up to 5.5 years) and in multiple lineages in a quantitative and highly reproducible manner. We have already made a number of important and novel discoveries, including the lack of evidence for a common lymphoid progenitor producing T and B cells in primates, with no shared clonal derivation of B and T cells until late after transplant, and much earlier shared clonal derivation of myeloid and B cells. We have also for the first time discovered the unique lineage derivation and life history of the major fraction of natural killer (NK) cells. CD16+CD56- cytotoxic NK cells did not share barcodes with B, T or myeloid cells even 48 months post-transplant. In vitro and murine models have not previously been able to shed light on NK cell lineage relationships. We have continued to use barcoded cells from these macaques to further dissect in vivo NK cell ontogeny, and the process of ex vivo expansion of NK cells, highly relevant for adoptive cell therapy development, in collaboration with Dr. Rick Childs' group in NHLBI, Dr Miti Kaur's group at the Tulane Non-human Primate facility, and Dr Yenan Bryceson's group at the Karolinska Institut in Stockholm Sweden. We have mapped the clonally-unique barcodes to NK cells with an "adaptive" phenotype, linked in murine studies to NK memory, and in humans to putative NK adaptive responses to viruses such as CMV. Within NK cells with an adaptive phenotype, we find oligoclonal and massively expanded individual clones, with no contributions found in other lineages, and with waxing and waning patterns suggesting responses to specific environmental cues such as viral infection or reactivation. Our data provides the first direct demonstration of clonal NK responses, possibly explaining NK memory. With profound in vivo NK depletion, the same clones arise again, without recruitment from highly polyclonal HSPC. We hypothesized these clones are generated by a response to CMV, based on correlative data in human transplantation and blood donor studies, and are testing this via barcoded transplantation in CMV negative macaques. We continue to search for the precursor to these cells, tracking dominant clones in phenotypically purified samples from blood, bone marrow, lymph nodes, liver, and vaginal and intestinal lymphoid aggregates. We are also applying our methodology to understand the original of other classes of innate lymphoid cells (ILC), including ILC2 in the lung, and ILC3 in the gut. We validated our model suggesting that adaptive CD56dimCD16+ NK cells are produced independently of multipotent HSPC by taking advantage of two human disease models. In the human disorder PNH, acquired somatic mutations result in loss of GPI-linked proteins on HSPC and their progeny with clonal expansion of GPI- cells over time. We demonstrated that human CD56dimCD16+ NK cells from PNH patients are almost 100% GPI+ despite CD56bright NK cells, B cells and myeloid cells all being GPI-, similar to the pattern in T memory cells, and suggesting that adaptive NK cells share properties with T memory cells and can homotypically self-renew without ongoing production from HSPC (#2). We also demonstrated that human adaptive NK cells are retained patients with GATA2 deficiency syndrome, despite loss of HSPC, CD56bright NK cells, B cells and monocytes (#3). We have extended our analysis of the geographic segregation of individual HSPCs long term in specific marrow sites, confirming our prior findings in mice using LEGO imaging techniques now in the macaque model utilizing barcoding. We can directly demonstrate in situ production of B cells, CD56+ NK cells and myeloid cells in localized marrow niches, and surprisingly, we now have strong evidence for in situ marrow production of T cells (manuscript under review) We have followed the output of thousands of individual HSPC clones in several animals for up to 5.5 years, and demonstrate marked clonal stability of output from long-term repopulating clones producing myeloid, B cell and T cell lineages, along with CD56brightCD16- NK cells. We found no evidence for clonal succession in these young adult animals, directly refuting controversial murine data suggesting contributions from each clone lasting only several weeks. We have shown that myeloid vs lymphoid bias is common, and that this bias remains stable in individual clones over time (#1). We have recently completed a project comparing the clonal behavior of young versus aged HSPC, demonstrating marked differences in clonal patterns, specifically very delayed contributions from multilineage clones with persistence of unilineage contributions, in both myeloid, B and T cell lineages, in contrast to murine models suggesting accumulation of only myeloid-biased clones. In the oldest animal, we observed clonal dropout and clonal expansions, potentially providing a model of human "CHIP" associated with aging (clonal hematopoiesis of indeterminate prognosis). We have continued to analyze clonal patterns following engraftment of ex vivo expanded HSPC. We have derived methodologies allowing investigation of red cell and platelet lineages. Red cell clonal patterns match those of other myeloid lineages very closely. However, we have very interesting data regarding emergence of unique highly biased clones contributing to platelets in the presence of inflammation. We continue to collaboration with Rahul Sajita at NYU to apply single cell RNAseq to barcoded populations in order to further define ontogeny as well as identify rare precursor cell populations. We now have set up the DropSeq single cell methodology at NHLBI in collaboration with Andre Larochelle, and have evidence that we can retrieve both the barcode and the transcriptome from single cells. This approach offers many investigative opportunities, but also challenges in terms of data collection and analytics.