Reparative immunological consequences of stem cell transplantation as a cellular therapy for refractory Crohn's disease

Abstract

Background: Treatment strategies for Crohn's disease (CD) suppress diverse inflammatory pathways but many patients remain refractory to treatment. Autologous haematopoietic stem cell transplantation (SCT) is an emerging therapy for medically refractory CD though the mechanisms through which it circumvents refractory pathophysiology are unknown.

Objective: The objective of this study is to understand how the immune system reconstitutes post-SCT and whether SCT may function as a cellular therapy restoring appropriately responsive immune cell populations from haematopoietic stem cells (HSCs).

Design: Adults with CD with active clinical and endoscopic disease who failed available medical therapies were enrolled in a phase II study of SCT for refractory CD (n=19). Blood and intestinal samples were collected longitudinally and analysed using CyTOF and scRNA-seq. Stem cell autografts were functionally assayed in mouse xenograft models.

Results: scRNA-seq and CyTOF analyses reveal that SCT predominantly affected the intestinal myeloid lineage with loss of inflammatory populations and return of macrophages capable of supporting mucosal healing. Xenograft models using patient HSCs suggested that HSCs support the early reconstitution of the myeloid lineage and reveal an impairment of short and long-term HSC engraftment that may determine SCT outcomes.

Conclusions: This study suggests SCT functions as a myeloid-directed cellular therapy reinforcing the critical role of macrophages in refractory CD pathophysiology and as a target for cellular therapies. Furthermore, we report an unrecognised functional heterogeneity among HSC subpopulations in CD that may be relevant to our understanding of CD treatment and pathophysiology.

Keywords: AUTOIMMUNE DISEASE; CROHN'S DISEASE; MACROPHAGES; STEM CELLS.

Figure 1. SCT is highly effective in refractory CD patients. (A) Table of patient demographics at baseline. (B) Bar graphs of clinical outcomes at 6 months and 12 months post-SCT demonstrate a significant reduction in Harvey Bradshaw Index (HBI) and Crohn’s Disease Activity Index (CDAI) scores (n=14, mean±SEM). (C) Line graph of paired samples for endoscopic outcomes using Simple Endoscopic Score for Crohn’s Disease (SES-CD) at 6 months post-SCT (n=14). (D) Pie graphs of endoscopic response (SES-CD decrease ≥50%) and endoscopic remission (SES-CD≤4 and no subscore >1 in any individual variable) at 6 months post-SCT. All statistical analysis (B and C) was done using paired Wilcoxon signed-rank test with *p<0.05, **p<0.01 ***p<0.001 ****p<0.0001. (E) Bar and pie graph of subjects with all intestinal biopsies at 6 months post-SCT that were histologically normal or inactive with only chronic changes (n=14). (F) Schematic of MASCT-CD clinical trial and sample collection timeline. CD, Crohn’s disease; SCT, stem cell transplantation.

Figure 2. Stem cell transplantation (SCT) has a distinct effect on intestinal immune populations. (A) Unsupervised CyTOF analysis with FlowSOM for blood (top) (baseline n=7, 6 months post-SCT n=9) and intestinal (bottom) (baseline n=13, 6 months post-SCT n=13) samples with each minimal spanning tree diagram displaying 15 metaclusters comprised of 100 clusters. Metaclusters are annotated using canonical cell surface markers or prominent markers if lineage markers are negative (lin- are CD3⁻CD19⁻CD14⁻CD16⁻CD56⁻CD66b⁻). Metaclusters that were significantly changed from baseline to 6 months post-SCT are indicated by red font and red dashed circle and clusters that were significantly changed are indicated by solid red circle (p<0.05, Mann-Whitney test). (B) Bar graph depicts the number of significantly changed clusters from baseline to 6 months post-SCT (p<0.05, Mann-Whitney test). Principal component analysis (PCA) of blood (top) and intestine (bottom) samples comparing baseline with 6 months post-SCT (FlowSOM clusters, PCA analysis 95% CI). (C) Bar graph (mean±SEM) of most significantly changed cell populations in supervised clustering analyses from blood (baseline n=9, 6 months post-SCT n=11) and intestine (baseline n=13, 6 months post-SCT n=13), Mann-Whitney test, ***p<0.001, ****p<0.0001.

Figure 3. Reconstitution of intestinal myeloid populations. (A) CITRUS analysis of CyTOF datasets from blood (baseline n=9, 6 months post-SCT n=11) and (B) intestinal (baseline n=13, 6 months post-stem cell transplantation (post-SCT) n=13) samples. Hierarchical clusters are depicted as a CITRUS tree with each node representing a different cluster coloured coded by immune cell populations identified through expression of canonical protein markers. Cluster networks (>1 cluster) that collectively predict the change in immune cell populations 6 months post-SCT compared with baseline are outlined in red (significance analysis of microarrays, false discovery rate <0.01). Hierarchical relationships between significant clusters are identified with red arrows indicating the direction from parental to child cluster. (C) CyTOF samples were concatenated at each timepoint and CD14⁺ populations were analysed by viSNE. Overlapping tSNE plots demonstrate a change in intestinal CD14⁺ cell phenotypes between baseline and 6 months post-SCT.

Figure 4. SCT restores homeostatic macrophages in refractory CD. (A) scRNA-seq datasets from the intestine were analysed by CellphoneDB (baseline n=4, 6 months post-SCT n=3). Significant clusters pairing expression of ligand-receptor pairs displayed as a heatmap. Clusters with the largest number of interacting pairs at baseline and follow-up are outlined in a dashed black square (ie, hub clusters). The largest increase in interactions from baseline to 6 months post-SCT is between hub clusters and epithelial clusters 3, 4, 14 outlined in black at baseline and red at 6 months post-SCT. (B) UMAP plots from scRNA-seq intestinal dataset with re-clustered myeloid cells annotated (baseline n=4, 6 months post-SCT n=3). Clusters are highlighted based on abundance at each timepoint with clusters common to both timepoints that undergo remodelling indicated by arrows (online supplemental figure 5). Selected significant pathways and marker genes are highlighted for each cluster including DEGs for clusters common to both timepoints. (C) UMAPs demonstrating fold expression of select genes. (D) Schematic of monocyte/macrophage populations and functions that reinforce refractory CD pathophysiology at baseline or intestinal healing post-SCT. CD, Crohn’s disease; SCT, stem cell transplantation.

Figure 5. SCT is a myeloid-directed cellular therapy. (A) Blood scRNA-seq samples from all timepoints (n=20) were clustered with UMAP plots highlighting progenitor clusters specific to stem cell collection (n=4) and engraftment (n=3) that are not found at baseline (n=7). RNA velocity curves using scVelo of stem cell collection demonstrate progenitor clusters related to mature myeloid cells. (B) CD34 expressing clusters were re-clustered and annotated. Clusters are highlighted based on their presence at engraftment or stem cell collection. (C) Bar graph (mean±SEM) of Simpson clonality of TcRβ sequencing of blood (baseline n=10, 6 months post-SCT n=10) and tissue (baseline n=9, 6 months post-SCT n=9) samples at baseline and 6 months post-SCT, statistical analysis with Dunn’s test, *p<0.05. (D) Bar graph (mean±SEM) of max clone frequency for T cell clones present at both timepoints, statistical analysis with Dunn’s test, *p<0.05. The increase in Simpson’s clonality post-SCT in the blood and intestine suggests a decrease in T cell clonal diversity. HSC, haematopoietic stem cell; SCT, stem cell transplantation.

Figure 6. HSC heterogeneity in CD. (A) Flow cytometry analysis of mobilised stem cells from nine patients. Stem cell populations are percent live CD34⁺ cells. (B) Human CD34⁺ stem cells were injected into NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice with mean line graphs of CD45⁺ populations in the bone marrow of the mice from three patients with the highest engraftment (as shown in figure 6C) for granulocytes (CD15⁺), monocytes (CD14⁺), B cells (CD19⁺), T cells (CD3⁺), myeloid cells (CD33⁺) and megakaryocytes (CD41a⁺) (mean, n=2 biological replicates at each timepoint). (C) Analysis of engrafted human CD45⁺ cells in the murine bone marrow at 4, 8 and 16 weeks. Bar graph of % human cells from all cells isolated (mean, n=2 biological replicates at each timepoint). Pt ID key identifies clinical response for each patient. Graph background colour demonstrates the contribution of ST-HSC and LT-HSC to human engraftment at each timepoint. CD, Crohn’s disease; HSC, haematopoietic stem cell; IT-HSC, intermediate HSC; LT-HSC, long-term HSC; MPP, multipotent progenitors; SCT, stem cell transplantation; ST-HSC, short-term HSC.