Parallel transcriptional programs driving early-starting inheritable lineage bias of myeloid, lymphoid and dendritic cells are initiated by distinct dosage of PU.1 and IRF8 in HSCs. Homo sapiens

In the traditional model of hematopoiesis, blood lineages diversify from HSCs through a sequential process of hierarchical bifurcation. Each of these bifurcation nodes comprises multipotent progenitors that are assumed to be homogeneous and equipotent in terms of their potency. The isolation of dendritic cell (DC) progenitors with either myeloid or lymphoid potency has led to the notion of “convergent development”, and an ongoing debate on the origin of DCs and their relationship with myeloid and lymphoid lineages. We have used a clonal assay combining with statistical modeling to quantify the yield of granulocytes (G), monocytes (M), lymphocytes (L) and three subsets of DCs from single human CD34+ progenitor cells of nine different phenotypes, and traced the potency of their individual progeny. We show that multipotent progenitors are not in fact equipotent, but instead exhibit a bias toward one specific lineage; this lineage bias is established at the HSC stage and transmitted to most progeny, with bifurcation to other lineages only occurring infrequently. Critically, these lineage biases can be correlated to lineage-specific transcriptional programs that are reinforced during division. We performed computational analysis to correlate transcriptomes and lineage composition of individual phenotype-defined progenitor cells. This enabled us to identify 32 common transcription factor (TF) candidates that interact physically and form genetic regulatory networks, with lineage-specific developmental programs initiated and reinforced based on the distinct dosage combination of these common TFs. Our analysis reveals that early expression of high levels of IRF8 and intermediate levels of PU.1 protein correlates with a bias towards CD141+ DC and plasmacytoid DC lineages in vivo, and that this pattern is inheritable from progenitors to mature cells. We have therefore arrived at a coherent model of DC development driven by parallel and inheritable programs defined by distinct dosages of TFs including PU.1 and IRF8, a developmental model that may apply to other lineages as well. Overall design: mRNA profiles of human cord blood CD34+ hematopoietic progenitors from healthy donors

在传统的造血模型中，血液谱系经由层级分支的连续过程，从造血干细胞（Hematopoietic Stem Cells, HSCs）分化而来。这些分支节点中的每一个均由多能祖细胞构成，这类细胞曾被认为在分化潜能上具有均质性与等潜能性。分离获得兼具髓系或淋系分化潜能的树突状细胞（Dendritic Cell, DC）祖细胞后，学界提出了“趋同发育”的概念，同时也引发了关于DC起源及其与髓系、淋系谱系关系的持续争论。

本研究结合克隆检测法与统计建模，对9种不同表型的人CD34+祖细胞单克隆产生的粒细胞（G）、单核细胞（M）、淋巴细胞（L）以及3种DC亚群的产出量进行定量分析，并追踪了其子代细胞的分化潜能。研究结果表明，多能祖细胞实际上并不具备等潜能性，反而倾向于偏向某一特定谱系；这种谱系偏向性在造血干细胞阶段即已确立，并传递至大部分子代细胞，仅极少数子代细胞会发生向其他谱系的分支分化。

尤为关键的是，这些谱系偏向性可与谱系特异性转录程序相关联，此类程序在细胞分裂过程中得到强化。本研究通过计算分析，将经表型定义的单个祖细胞的转录组与谱系构成进行关联分析。该分析使得我们得以鉴定出32种常见转录因子（Transcription Factor, TF）候选分子，它们可发生物理相互作用并构成基因调控网络；基于这些常见TF的不同剂量组合，谱系特异性发育程序得以启动并强化。

分析结果显示，体内早期高表达IRF8且中等表达PU.1蛋白的模式，与偏向CD141+ DC及浆细胞样DC谱系的特性相关，且该模式可从祖细胞传递至成熟细胞。综上，我们提出了一个连贯的DC发育模型：该模型由包括PU.1与IRF8在内的转录因子不同剂量所定义的平行且可遗传的程序驱动，这一发育模型或许也可推广至其他谱系。

整体实验设计：来自健康供者的人脐带血CD34+造血祖细胞的mRNA转录组谱。