In Vivo Tracking of Human Hematopoiesis Reveals Patterns of Clonal Dynamics during Early and Steady-State Reconstitution Phases

Abstract

Hematopoietic stem/progenitor cells (HSPCs) are capable of supporting the lifelong production of blood cells exerting a wide spectrum of functions. Lentiviral vector HSPC gene therapy generates a human hematopoietic system stably marked at the clonal level by vector integration sites (ISs). Using IS analysis, we longitudinally tracked >89,000 clones from 15 distinct bone marrow and peripheral blood lineages purified up to 4 years after transplant in four Wiskott-Aldrich syndrome patients treated with HSPC gene therapy. We measured at the clonal level repopulating waves, populations' sizes and dynamics, activity of distinct HSPC subtypes, contribution of various progenitor classes during the early and late post-transplant phases, and hierarchical relationships among lineages. We discovered that in-vitro-manipulated HSPCs retain the ability to return to latency after transplant and can be physiologically reactivated, sustaining a stable hematopoietic output. This study constitutes in vivo comprehensive tracking in humans of hematopoietic clonal dynamics during the early and late post-transplant phases.

Graphical abstract

Figure 1

Tracking of Engineered Population Dynamics and Clonal Abundance over Time (A) Diversity index calculated on number of ISs and relative sequence counts of pooled PB lineages purified from four patients over time. Black asterisk on the y axis indicates the average diversity index of pre-infusion samples. (B) Diversity index of BM CD34⁺ cells (in red) and individual samples over time (in gray) from a representative patient (patient 2). Red asterisk shows the diversity index of BM CD34⁺ cells before infusion. (C) Estimates of population clonal abundance of PB mature lineages from four WAS patients. Time intervals are reported as months after GT on the x axis of each graph. (D) Overall clonal size estimates of BM CD34⁺ and PB mature populations at the last follow-up when both BM and PB were collected (36 months after GT for patients 1, 2, and 3 and 24 months for patient 4). FU, follow-up. See also Figures S1 and S2.

Figure 2

IS Sharing and Clonal Output of BM CD34⁺ Progenitors over Time (A) Bar charts showing the percentage of IS from BM CD34⁺ cells shared with only myeloid (green), only lymphoid (blue), and both myeloid and lymphoid lineages (orange) from the same time point (months after GT). The fractions of ISs detected only in BM CD34⁺ cells at each given time are shown in gray. Data have been filtered for potential cross-contaminations as described in Supplemental Experimental Procedures. (B) Circos plots showing the levels of CD34⁺ output measured as multilineage sharing of ISs from BM CD34⁺ cells over time (ordered in columns per months after GT). The rainbow area of the circle is composed of ribbons showing relative IS sharing toward different BM and PB lineages listed on the left. The red area comprises the total number of IS from CD34⁺ cells. The larger the rainbow area compared to the red area, the higher the CD34⁺ output. The more colored the rainbow area, the more diverse and multilineage the CD34⁺ output. (C) Heatmap using blue-to-green color intensity to show the frequency of IS shared by each lineage (in rows) with BM CD34⁺ cells (in columns) over time as a measurement of clonal input received with BM CD34⁺ progenitors. Unsupervised clustering was performed as described in Supplemental Experimental Procedures. See also Table S1 and Supplemental Experimental Procedures.

Figure 3

Analysis of Purified HSC and MPP Output over Time (A) Output of HSC and MPP progenitors over time isolated from patient 2 (Pt2) and Pt3 at 36 months after GT. Bar graphs on the left show the fraction of HSC or MPP ISs (in blue) detected in multiple lineages and time points. Heatmaps on the right show the detection of each shared IS (in rows) over time. Each column represents an individual lineage and time point. Blue color intensity is proportional to the level of multilineage and over-time detection of each shared IS. The number of clones (15 HSCs and 8 MPPs) mentioned in the text is the sum of the clones identified as sharing ISs with at least one other lineage in the two patients over time (HSC clones: 6 (heatmap above) + 9 (heatmap below) = 15; MPP clones: 4 (heatmap above) + 4 (heatmap below) = 8). (B) Horizontal column graphs showing the real-time input of HSC or MPP shared ISs as relative abundance (percentage of sequence counts) within whole BM, whole PB, BM MNCs, or PBMCs isolated from the same patient at the same time point. (C) Line plots showing the sharing scores (on y axis) of HSC or MPP ISs toward individual myeloid and lymphoid lineages over time (months after GT on the x axis). Sharing scores were calculated according to the rules reported in Supplemental Experimental Procedures.

Figure 4

Hierarchical Relationships among Hematopoietic Lineages (A) Dot plots representing, from left to right, the frequency of ISs shared with at least one other cell type, number of unique ISs, diversity index, and vector copy number of different cell compartments isolated from four WAS patients at the last time point when both BM and PB were collected (Mann-Whitney U test: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001). (B) Schematic representation of two alternative models of hematopoietic hierarchies implying distinct myeloid-lymphoid (left top graph) or myeloid-based (right top graph) branching of hematopoietic differentiation. A Bayesian network approach was used to visualize, estimate, and compare complex dependence structures among lineages of the first three WAS patients at last time point after GT (Supplemental Experimental Procedures) starting from IS data. The corresponding two different model constraints used for the test, not allowing or allowing probability dependencies between ISs retrieved in myeloid precursors and lymphoid mature cells, are schematically shown respectively on the left and right panels. A measure of the strength of informativeness of each model is provided using the Bayesian information criterion (BIC) score, which allows the statistical comparison of different hierarchical structures. The resulting BIC scores are shown above each panel. The resulting best-fitting model is framed with a red square. Models of hematopoietic hierarchies are adapted from Kawamoto et al. (2010b). Differentiation potential are labeled as follows: E, erythroid; M, myeloid; B, lymphoid B; and T, lymphoid T. Common progenitors are labeled as follows: CMP, common myeloid progenitor; CLP, common lymphoid progenitor; CMEP, common myeloerythroid progenitor; CMLP, common myelolymphoid progenitor; MTP, myeloid-lymphoid T progenitor; and MBP, myeloid-lymphoid B progenitor. (C) Dynamic model of lineage similarities designed on m-dimensional continuous-time stochastic Markov process, observed at fixed time points (schematic representation and calculation formulas of different rates are shown on the left). Principal-component analysis (PCA) performed on estimated death, duplication, and differentiation rates during the 12, 24 and 36 months after GT calculated on IS sharing and sequence reads belonging to each IS. (The first two principal components are shown on the PCA plot on the right.) Clustering of BM and PB lineages is shown as rounded areas colored in blue and orange, respectively. Similarities among lineages are shown by the relative distance or proximity of the boxes tagged with their relative cellular surface markers (Supplemental Experimental Procedures). See also Figures S3 and S4.