Hox genes are crucial regulators of periosteal stem cell identity

Abstract

Periosteal stem and progenitor cells (PSPCs) are major contributors to bone maintenance and repair. Deciphering the molecular mechanisms that regulate their function is crucial for the successful generation and application of future therapeutics. Here, we pinpoint Hox transcription factors as necessary and sufficient for periosteal stem cell function. Hox genes are transcriptionally enriched in periosteal stem cells and their overexpression in more committed progenitors drives reprogramming to a naïve, self-renewing stem cell-like state. Crucially, individual Hox family members are expressed in a location-specific manner and their stem cell-promoting activity is only observed when the Hox gene is matched to the anatomical origin of the PSPC, demonstrating a role for the embryonic Hox code in adult stem cells. Finally, we demonstrate that Hoxa10 overexpression partially restores the age-related decline in fracture repair. Together, our data highlight the importance of Hox genes as key regulators of PSPC identity in skeletal homeostasis and repair.

Keywords: Aging; Positional identity; Regeneration; Reprogramming; Skeletal stem cell; Stemness.

Fig. 1.

Hox gene expression is enriched in skeletal stem/progenitor cells. (A) Gene expression pattern for 11 HoxA cluster genes in CD45^–TER119^–CD31^–LEPR⁺ SSPCs or cells of the microenvironment harvested from 12-week-old, freshly isolated tibiae and femurs, as determined by RNA-sequencing (Josephson et al., 2019). HoxA genes are highly enriched in the SSPC population and Hoxa10, with the most normalized reads, is the most highly expressed. n=3. (B) RNA-sequencing gene expression data of the HoxA cluster derived from 12-week-old tibial periosteal cells. (C) Representative flow cytometry plot of periosteal stem and progenitor cells as defined by Debnath et al. (2018) and strategy for isolating periosteal stem cells (PSC), periosteal progenitor 1 (PP1) cells and periosteal progenitor 2 (PP2) cells. (D) Relative expression of Hoxa10 in freshly isolated stem and progenitor populations of tibial periosteum as measured by qRT-PCR. n=3 mice. (E,F) Gene expression of multiple skeletal stem/progenitor cell-associated genes during a 72 h in vitro time course of osteogenic (E) or adipogenic (F) induction of isolated PSCs and PP1 cells, relative to growth media controls. n=7 mice. *P<0.05, **P<0.01 (unpaired, two-tailed Student's t-test). Each point represents data collected from an individual mouse. Data are mean±s.e.m.

Fig. 2.

Loss of Hox genes in stem and progenitor cells triggers a loss of skeletal stem cells and periosteal stemness properties. (A) The relative expression of mouse Hoxa10 in bone marrow samples harvested from the periosteum of young (3-month-old) and aged (21-month-old) mouse tibiae, as measured by qRT-PCR. n=5 (young), n=3 (aged). (B) The relative expression of HOXA10 in bone marrow samples harvested from the fracture sites of young (18-39 years old) and aged (61-86 years old) humans, as measured by qRT-PCR. n=8 (young), n=7 (aged). (C) When tibial periosteal cells were harvested from young (3-month-old) and aged (21-month-old) mice, flow cytometry revealed the frequency of 6C3^–CD90^–CD49f^lowCD51^lowCD200⁺CD105⁻ periosteal stem cells. n=10 (young), n=3 (aged). (D-G) Simultaneous knockdown of Hoxa10, Hoxa11, Hoxd10, Hoxd11 and Hoxc10 (HoxMix) was used to analyze stem cell number (D,E), self-renewal (F) and identity (G) in Hox-deficient tibial PSPCs. (D) PSPCs were pulsed with EdU for 15 h following HoxMix and non-targeting control siRNA knockdown; the number of EdU-positive cells was then measured by flow cytometry. n=5. (E) The absolute number of PSPCs after an equal seeding density and 6 days of transfection with either NT control or HoxMix siRNA. n=3. (F) After 7 days of control and HoxMix siRNA, PSPCs were analyzed for stemness-associated cell-surface marker expression using flow cytometry. n=3 each condition. (G) siControl and siHoxMix tibial PSPCs were also treated with CellTrace and subjected to flow cytometry to categorize cells by generation after 6 days of incubation. The fold change in the percentage of cells treated with siHoxMix in each generation is shown relative to the siControl percentage for each generation. Asterisks indicate statistical significance between siControl and siHoxMix within each generation. n=5 (control), n=7 (HoxMix). (H) Scheme of tibial defects, tamoxifen dosing protocol (2 mg/day) and EdU administration to test the requirement of HoxA in skeletal stem and progenitor cells. (I) Flow cytometry revealed the percentage of 6C3^–CD90^–CD51⁺CD200⁺CD105⁻ skeletal stem cells, 6C3^–CD90^–CD51⁺CD200^–CD105⁻ pre-bone/chondro/stromal progenitors (pre-BCSPs) and EdU⁺ proliferative cells in the non-hematopoietic compartment of Pdgfra^CreERT;HoxA^flox/flox and HoxA^flox/flox control mice 3 days after tibial defect repair. (J) Experimental plan to test the requirement of multiple posterior Hox genes to generate osteoblasts during bone regeneration of Col1a1^CreERT;Rosa^Tomato mice. (K) The frequency of Tomato⁺, osteo-lineage cells in calluses harvested from Col1a1^CreERT;Rosa^Tomato tibiae undergoing fracture repair at 7 days post-injury, as measured by flow cytometry. n=4 mice (control), n=6 mice (HoxMix). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (unpaired, two-tailed Student's t-test). Each point represents data collected from an individual mouse. Data are mean±s.e.m.

Fig. 3.

Hoxa10 overexpression induces a skeletal stem cell-like state. (A) LV-GFP- and LV-Hoxa10/GFP-transduced tibial PSPCs were subjected to flow cytometry after a 7-day incubation to analyze the proportion of infected GFP⁺ cells expressing the skeletal stem cell surface markers PDGFRα/CD51 or PDGFRα/SCA1. n=3 mice. (B) LV-GFP- or LV-Hoxa10/GFP-infected tibial PSPCs were also treated with Cell Trace and subjected to flow cytometry to categorize cells as high- or low-cycling after 6 days of incubation. Gating strategy is presented in Fig. S4F. n=4 for each condition. (C) Experimental plan to test the reprogramming abilities of Hoxa10. Tibial PSCs, PP1 and PP2 cells were separately isolated by FACS. Each cell population was infected with LV-GFP or LV-Hoxa10/GFP and analyzed by flow cytometry after 7 days of incubation. (D-F) Flow cytometric analysis of the distribution of GFP⁺ PSCs, PP1 and PP2 cells within the CD51⁺ stem and progenitor cell compartment 7 days after LV-GFP or LV-Hoxa10/GFP infection of PSCs (D), PP1 (E) and PP2 (F) cells. n=3 (PSC), n=3 (PP1), n=5 (PP2). (G) The frequency of PSCs among total cells 7 days after the infection of PP1 cells with LV-GFP or LV-Hoxa10/GFP. Infected (GFP⁺) and uninfected (GFP^–) are shown separately. (H) The relative fold change in GFP⁺ PSCs within the PSPC compartment after transduction of PP1 cells with LV-GFP or LV-Hoxa10/GFP. n=4 separate experiments. (I) Flow cytometry analysis of the distribution of cells within the PSPC lineage hierarchy after a 7-day treatment of tibial PP1 cells with 10 μg/ml mitomycin C and infection with LV-GFP or LV-Hoxa10/GFP. n=3. (J) Experimental plan to carry out serial transplants of LV-GFP- or LV-Hoxa10/GFP-treated periosteal PP1 cells under the renal capsule to test self-renewal capacity. (K) PP1 cells were first transduced with either LV-GFP or LV-Hoxa10/GFP ,and the prevalence of GFP-labelled PSCs was then assessed by flow cytometry before (Pre) and after each round of transplantation. n=6, Pre-LV-GFP and Pre-LV-Hoxa10/GFP; n=3, 1°-LV-GFP and 1°-LV-Hoxa10/GFP; n=3, 2°-LV-GFP; n=4, 2°-LV-Hoxa10/GFP. (D-F,H,I) Complete results and statistics are provided in Table S1. *P<0.05, **P<0.01 (unpaired, two-tailed Student's t-test). Each point represents data collected from an individual mouse. Data are mean±s.e.m.

Fig. 4.

The regional specificity of Hox function is maintained in the adult skeleton. (A) Diagram of skeletal elements investigated along with the proposed regional restriction Hox expression in adult skeletal tissues (adapted from Rux and Wellik, 2017). (B-E) The expression profile of the Hox cluster containing the highest expressed Hox gene in periosteal cells of the pelvis (B), spine^T5-T8 (C), radius/ulna (D) and anterior rib^1-4 (E) (highest Hox gene expression is highlighted in red). n=4 mice for each skeletal element. Full Hox expression data in Fig. S6. (F-L) The lineage output of stem and progenitors 7 days after infecting PP1 cells derived from the pelvis (F), spine^T5-T8 (G,J), radius/ulna (H,K) and anterior rib^1-4 (I,L) with LV-Hoxa10/GFP (F), LV-Hoxb8/GFP (G,J), LV-Hoxa11/GFP (H,K) or LV-Hoxa5/GFP (I,L), with LV-GFP (Ctrl) and LV-Hoxa10/GFP serving as controls. n=5, pelvis (control and A10); n=4, spine (control and B8); n=3, spine (control and A10); n=4 and n=5, radius/ulna (control and A11, respectively); n=5, radius/ulna (control and A10); n=9, n=8, and n=9, anterior rib (control, A5 and A10, respectively). Full lineage output data and statistics are provided in Table S2. n.s., not significant, *P<0.05 (unpaired, two-tailed Student's t-test). Data are mean±s.e.m.

Fig. 5.

In vivo Hoxa10 overexpression partially rescues age-related fracture-healing deficiency. (A) Experimental plan to test the effect of Hoxa10 transduction during in vivo fracture repair. (B) Representative 3D reconstructions (top) and longitudinal cross-sections (bottom) of young (3-month-old) and middle-aged (13-month-old) tibial fractures transduced with LV-GFP or LV-Hoxa10/GFP and imaged by μCT 14 days post-injury. (C) μCT histomorphometry of the calluses of 14-day post-injury tibia fractures in young and middle-aged mice treated with either LV-GFP or LV-Hoxa10/GFP. BV/TV, bone volume/total tissue volume; Tb. Sp., trabecular spacing; Tb. Nb., trabecular number; Tb. Th., trabecular thickness. *P<0.05, **P<0.01, ***P<0.001 (unpaired, two-tailed Student's t-test). Each point represents data collected from an individual mouse. Data are mean±s.e.m. Scale bars: 1 mm.