https://orcid.org/0000-0002-8042-932X
Abstract
Regulator of G protein signaling 5 (RGS5) is a GTPase activator for heterotrimeric G-protein α-subunits, shown to be a marker of pericytes. Bone marrow stromal cell population (BMSCs) is heterogeneous. Populations of mesenchymal progenitors, cells supportive of hematopoiesis, and stromal cells regulating bone remodeling have been recently identified. Periosteal and bone marrow mesenchymal stem cells (MSCs) are participating in fracture healing, but it is difficult to distinguish the source of cells within the callus. Considering that perivascular cells exert osteoprogenitor potential, we generated an RGS5 transgenic mouse model (Rgs5-CreER) which when crossed with Ai9 reporter animals (Rgs5/Tomato), is suitable for lineage tracing during growth and post-injury. Flow cytometry analysis and histology confirmed the presence of Rgs5/Tomato+ cells within CD31+ endothelial, CD45+ hematopoietic, and CD31-CD45- mesenchymal/perivascular cells. A tamoxifen chase showed expansion of Rgs5/Tomato+ cells expressing osterix within the trabeculae positioned between mineralized matrix and vasculature. Long-term chase showed proportion of Rgs5/Tomato+ cells contributes to mature osteoblasts expressing osteocalcin. Following femoral fracture, Rgs5/Tomato+ cells are observed around newly formed bone within the BM cavity and expressed osterix and osteocalcin, while contribution within periosteum was low and limited to fibroblastic callus with very few positive chondrocytes. In addition, BM injury model confirmed that RGS5-Cre labels population of BMSCs expands during injury and participates in osteogenesis. Under homeostatic conditions, lineage-traced RGS5 cells within the trabecular area demonstrate osteoprogenitor capacity that in an injury model contributes to new bone formation primarily within the BM niche.
Cells expressing RGS5 have recently been shown to be a marker of pericytes and its role in bone has not been studied. Therefore, we developed an animal model suitable for lineage tracing of these cells and their progeny and studied their role during post-natal growth and after injury. This population of RGS5/Tomato cells has an osteogenic phenotype and contributes to the formation of bone under homeostasis and bone repair with stronger osteogenic potential within bone marrow compartment.
Introduction
Skeletal regeneration and repair are of clinical relevance due to an aging population with associated conditions such as osteoporosis, diabetes, and non-union fracture healing, the last accounting for up to 10% of all fractures.1-3 A better understanding of skeletal progenitors involved in maintaining bone homeostasis and regeneration during injury/disease is needed to make progress in skeletal regenerative therapies. The bone marrow niche harbors a heterogenous pool of progenitors that give rise to cells of the hematopoietic, endothelial, and mesenchymal/stromal cell lineages.4-6 Progenitor cells of the mesenchymal lineage have been described as multipotent cells with the ability to produce osteoblasts, chondrocytes, and adipocytes7,8 and are important for supporting hematopoiesis.9 Perivascular cells in various organs and within the bone marrow microenvironment have been shown to differentiate into cells of the mesenchymal lineage.10,11 Reporter mice and lineage tracing experiments have been utilized to study the role and development of both skeletal progenitors and pericytes during homeostasis and after injury.8,12-16
Previous work in our lab has identified αSMA as a robust marker of osteoprogenitors which can become bone marrow trabecular osteoblasts and osteocytes. αSMA also identifies periosteal progenitor cells that form new bone after fracture.8,14,17 Using single-cell analysis on sorted αSMA+ periosteal cells, we recently stratified an αSMA labeled progenitor population into 4 different clusters, one of which represented a perivascular segment that expressed RGS5 along with PDGFRβ and CXCL12.14
Regulator of G-protein signaling 5 (RGS5) is a member of the subfamily of small RGS proteins which also include RGS4 and RGS1618 and serve to inactivate signal transduction initiated by G protein-coupled receptors via GTPase activation.19 All RGS proteins have an 120 amino acid conserved RGS domain which directly interacts with Gαi and Gαq.20 RGS5 as a members of the small RGS protein family have short peptide sequences flanking the RGS domain. These short peptide sequences are responsible for binding signaling factors.21 RGS5 is strongly expressed in all major arteries and in all major organs except for liver and lungs with very high expression seen in heart, kidney blood vessels, and glomerular mesangial cells.22 More recently, RGS5 has been shown to be expressed in PDGFRβ pericytes22 and to be a measure of pericyte coverage of blood vessels,23 which coincides with active vessel remodeling during neovascularization.24 Rgs5-GFP knock-in mice in which the RGS domain was replaced by GFP coding sequence show no developmental or vascular defects and appear to have proper pericyte coverage.25 Mice with global deletion of RGS5 had similar growth to controls; however, experienced higher blood pressure with effects on blood-brain barrier and stroke severity.26
Very little is known about RGS5 as a marker of stem/progenitor cells and the contribution of RGS5-expressing cells to osteogenesis. The perivascular expression of RGS5 suggested that it might mark cells that play a role in homeostatic bone regeneration and injury healing. To examine the role of RGS5 expressing cell population in these dynamic processes, we generated an inducible Rgs5-CreER transgenic animal model. These mice enabled us to label and lineage trace RGS5/Tomato+ cells and determine their osteogenic potential in long bones during normal growth and skeletal trauma and to define their progenitor potential.
Materials and Methods
Animals
All experiments were performed in an AAALAC accredited facility and approved by the Uconn Health Institutional Animal Care and Use Committee. Experiments were done in both male and female mice. Animals were housed in ventilated cages at 22 °C. Water and irradiated rodent chow (Teklad 2918, Invigo, Indianapolis, IN) were provided ad libitum. Cre-positive littermates treated with corn oil served as non-tamoxifen controls in lineage tracing and fracture studies.
For in vivo adipogenic studies, mice were fed a high-fat diet containing rosiglitazone pellets (0.02 mg per 1 g of diet)27 (Teklad) for 4 weeks. To facilitate adaptation to high-fat diet food, we provided a mixed diet for a first week (rosiglitazone diet with regular rodent chow, 80/20%).
Generation of Bacterial Artificial Chromosome (BAC) Transgenic Rgs5-CreER Mice
DNA manipulation followed the procedures detailed in Ref28 All primers are listed in Supplementary Table S1. Briefly, murine BAC clone RP24-102N16 was obtained from the Children’s Hospital Oakland Research Institute (CHORI) and checked for correctness by PCR. This 187 kbp clone was chosen because it contains no flanking genes and stops short of including the Rgs4 gene, which is only ~48 kbp downstream of Rgs5. A 500 bp homology arm ending 2 bp upstream of the Rgs5 translational start site was amplified from this clone by high-fidelity PCR and inserted into the multiple cloning region of the previously described construct pLD53.SC2-CreER. Bacterial recombination was then used to insert pLD53.SC2-CreER into the original BAC clone and recombinants clones were screened by colony PCR and verified by diagnostic restriction enzyme digestions (Supplementary Fig. S1). Selected fractions of modified BAC fragments were chosen for next-day pronuclear injection at the Uconn Health gene targeting and transgenic facility. PCR genotyping for Cre recombinase was used to identify transgenic animals with a total of 5 founder lines being identified. All 5 founders were then crossed with Ai9 tdTomato reporter mice (Jackson lab stock #007909) and genotyped using primers for Cre and Ai9. Animals were injected intraperitoneally 2 times (unless otherwise indicated) with 75 mg/kg tamoxifen (Sigma) to induce Cre recombination and then analyzed at the indicated time points after last injection. One Cre line was chosen for extended breeding and experimentation.
Bone Injury Models
A closed transverse diaphyseal fracture on the femur was created using a drop-weight blunt guillotine device. Prior to the fracture, a 25G needle was inserted into the intramedullary canal to stabilize the fracture.29 X-rays (Faxitron LX60, Faxitron X-ray LLC, Lincolnshire, IL, USA and Parameter 2D, Kubtec, CT, USA) were used to confirm pin placement, fracture creation, and to monitor fracture healing. Exclusion criteria were destabilization of the pin, absence of a fracture callus by day 10, multi-fracture, and loss of sensation/movement to the limb. Mice were administered 0.08 mg/kg Buprenorphine HCl (Reckitt Benckiser Pharmaceuticals, England) every 12 h for 48 h following time of fracture. Modified procedure from bone fractures procedure was used for bone marrow injury model. Pin was inserted into intramedullary space to generate a direct injury to bone marrow.30
Histology
For histology on frozen sections, soft tissue and long bones were fixed in 4% paraformaldehyde (PFA) at 4 °C, placed in 30% sucrose overnight, and embedded in OCT. Using a tape transfer method (Cryofilm 2C, Section lab), frozen sections were cut with a Leica Cryostat at 7 or 20 µm and glued to slides using Norland Optical Adhesive 61 (Norland Optical). After UV crosslinking, sections were hydrated or immunostained and then mounted in 50% glycerol following 4ʹ,6-diamidino-2-phenylindole (DAPI) staining of nuclei. All sections were imaged by an Axioscan Z.1 (Zeiss) and representative images are shown in figures from experiments with at least 3 mice per group. For histology on paraffin sections, soft tissues, and long bones were fixed in PFA as described and then bones were decalcified in 14% EDTA for 2 weeks, after which samples were dehydrated with ethanol and xylene prior to embedding in paraffin. Serial sections were cut at 5 µm and then deparaffinized and rehydrated prior to staining.
Immunohistochemistry
For CD31, osterix, and perilipin stainings, frozen femoral sections were hydrated and then incubated with 0.3% triton for 20 min. Blocking was done for 20 min with 1x Power block (BioGenEx) for CD31 staining, and 1 h for both osterix (10% goat serum) and perilipin (10% donkey serum) stainings. For osteocalcin staining, frozen femoral sections were incubated first in Epitope Retrieval Solution (IHC World) at 60 °C for 6 h, washed in 0.1% Tween 20, and then blocked with 10% goat serum in 0.1% Tween 20 for 1 h. Osteocalcin antibody was incubated overnight at 4°C at a dilution of 1:200, washed in 0.1% Tween 20, and then 50 mM Tris–Cl-150 mM NaCl prior to incubation with secondary antibody. All sections were counterstained with DAPI for imaging of nuclei. For RGS5 protein colocalization staining with tamoxifen-induced Rgs5/Tomato cells using an antibody against RFP, deparaffinized sections underwent antigen retrieval in Tris–EDTA buffer pH 9, followed by 0.3% triton and blocking with 10% donkey serum. Primary antibodies to RGS5 and RFP, both diluted at 1:200, were incubated for 1.5 h at room temperature followed by appropriate secondaries for 1 h both at room temperature. After imaging, sections were counterstained with Gill 1 hematoxylin. Specifics on primary and secondary antibodies can be found in Supplementary Table S2.
Quantification of Images
Day 60 osteocytes were counted manually in bone located within a 5 mm area under the growth plate using Zen Blue software (Zeiss). All Osterix+Tomato+ quantification was done in Image J Fiji.14 All bone marrow quantification was done in an area defined by 5 mm below the growth plate. Total fracture callus was quantified.
Flow Cytometry
For lineage tracing and marrow injury model experiments, bones devoid of periosteum were crushed and enzymatically digested with a mixture of 0.5% Collagenase P (Roche) and 2 mg/mL hyaluronidase (Sigma) for 1 h at 37 °C with shaking. For fracture experiments, callus was removed from the bone or periosteum from intact contralateral bone and were digested as described above. After digestion, cells were washed and incubated in red blood cell lysing buffer (Sigma) for 5 min and then washed and filtered through Nitex. Antibody staining was done for 45 min at 4 °C, then cells were washed in staining media and analyzed on a BD-LSRII flow cytometer (BD Biosciences) after the addition of DAPI (Molecular probes). All flow cytometry analysis was done in FlowJo v10 software (FlowJo LLC). Antibody information can be found in Supplementary Table S3.
Statistical Analysis
Statistical analysis of flow cytometric experiments with at least 3 mice per group was done using GraphPad PRISM 6 and shown as mean ± SEM. Student unpaired t-test or one-way ANOVA with Tukey post-test were performed and significance was shown for P values <.05*, <.01**, and <.001***.
Results
Generation and Validation of Rgs5-CreER Mice
To define RGS5 expressing cell populations in intact bone and during bone regeneration, we developed a new animal model that uses large region of DNA flanking the endogenous gene to drive an inducible Cre transgene (Rgs5-CreER). Recombination allowed insertion of the pLD53.SC2-CreER construct into an Rgs5 bacterial artificial chromosome (BAC) in a way that substituted expression of CreER for Rgs5 (Supplementary Fig. S1). For lineage tracing Rgs5-CreER mouse was crossed with the Ai9 reporter animal (Rgs5/Tomato). Rgs5/Tomato+ cells were evaluated by histology for tdTomato expression in soft tissues including lung, liver, kidney, and heart (Supplementary Fig. S2) 2 days following Tamoxifen (Tx) treatment. Reporter expression was absent in lung tissue and only present within blood vessels of liver (Supplementary Fig. S2A, S2B). More robust expression was seen in kidney blood vessels, glomerular mesangial cells as well as cardiac muscle (Supplementary Fig. S2C, S2D).
To validate Rgs5-CreER/Ai9 transgenic mice, endogenous RGS5 expression was evaluated by histology and co-expression of tdTomato determined in heart and kidney—tissues with reported abundant levels of RGS5. Co-localization of endogenous RGS5 protein and Rgs5/Tomato+ reporter (anti-RFP) was seen in smooth muscle cells within renal artery and cardiac muscle and vessels within heart tissues (Fig. 1A, 1B). Rgs5/Tomato+ cells were observed within the bone marrow compartment, but not within cortical bone (Fig. 1C, 1D). Staining for both endogenous RGS5 and RFP (Rgs5/Tomato+) was seen in blood vessels and lower expression on perivascular cells and hematopoietic cells within bone marrow (Fig. 1C inset 1-2). Importantly, no RFP or RGS5 staining was seen in cells within bone (Fig. 1D). A few cells lining the endosteal surface were expressing RGS5 but are not labeled with Rgs5/Tomato. This could represent less efficient recombination of the CreER transgene in some cell types or Rgs5 expression that occurred after tamoxifen exposure (not shown).
Validation of Rgs5/Tomato reporter expression. (A-D) Paraffin histology of Rgs5/Tomato+ mice 2 days following Tx injection stained with anti-RFP (red), anti-RGS5 (white) and counterstained with hematoxylin. (A) Renal artery, (B) Heart, (C) Bone marrow; 1. Blue arrows indicate co-staining of vascular endothelial cells and yellow arrows indicate co-staining of perivascular cells; 2. Yellow arrow indicates co-staining of stromal cell and blue arrow indicates co-staining of hematopoietic cell. (D) Within cortical bone (CB) there is no staining of RFP and RGS5; DAPI staining shown in blue. For color figure refer to online version.
Osteogenic Potential of Rgs5/Tomato+ Cells
Three-week-old Rgs5-CreER/Ai9 mice were treated with Tx. Two days after Tx treatment Rgs5/Tomato+ cells are present primarily within the trabecular regions (Fig. 2A) with occasional Rgs5/Tomato+ chondrocytes within growth plate (Supplementary Fig. S3A, S3B). Eight-week-old animals treated with Tx had similar distribution of Rgs5/Tomato+ as younger ones (Supplementary Fig. S3A, S3C). By day 20 post Tx, lineage-traced cells within the trabecular region are located between endosteal/trabecular surfaces and CD31+ vasculature (Fig. 2B, 2C and Supplementary Fig. S3D, S3E) and co-expressing osterix (Osx+) (Fig. 2D and Supplementary Fig. S3F). Six percent of total Osterix+ cells were Rgs5/Tomato+ cells at day 20 post-Tx treatment. Rgs5/Tomato+ cells 60 days post-Tx treatment, are found on cortical endosteal surfaces expressing osteocalcin (Fig. 2E, 2F) and osteocytes embedded within cortical bone (Fig. 2G). We quantified 5.25% Rgs5/Tomato+ osteocytes in trabecular bone and 6.47% in cortical bone at day 60 post Tx (Fig. 2H). At that time point, expansion of Tomato+ BMSC is also observed within the bone marrow compartment as well as robust lineage tracing to skeletal muscle surrounding bone (Fig. 2E). Importantly, Rgs5/Tomato+ animals treated with corn oil had no leaky Tomato expression in long bones (Supplementary Fig. S4).
Rgs5/Tomato+ cells lineage trace to osteoblasts and osteocytes. Three-week-old male and female mice (n = 4 per group) were analyzed 2, 20, and 60 days post tamoxifen (Tx). (A) Representative histology of day 2 after Tx. 1-3 show magnified areas of cell location, yellow line indicates growth plate (GP) and white arrows indicate Tomato+ cells on periosteam (P) next to cortical bone (CB). Blue indicates DAPI stained nuclei in periosteum. (B-D) Histology of Rgs5 lineage tracing cells at day 20 stained for CD31 in cyan (C) and osterix in green (D), white arrows in (C) indicate perivascular Tomato+ cells and in (D) co-expressing osterix (yellow). (E-G) Histology of day 60 lineage tracing. (E) 1. shows magnified area with Tomato+ cells embedded in cortical bone (white arrows) and Tomato+ cells on endosteal surface (yellow arrows). Lineaged traced skeletal muscle (blue arrow) (F) Osteocalcin staining (green), arrows indicate Tomato+ cells positive for osteocalcin on endosteal surface. (G) Tomato+ cells imaged in cortical bone, arrows indicate Tomato+ cells with osteocyte processes. (H) Quantification of Tomato+osteocalcin+ osteocytes day 60 post Tx. (I) Flow cytometric analysis of bone digests at day 2, 20, and 60 post Tx injection for total Tomato expression. (J) Bar graphs of flow cytometric analysis within the Tomato+ population identifies 4 subsets of cell types based on CD45 and CD31 expression. Percentages of each population is shown for day 2, 20, and 60 post Tx. (K) Bone marrow histology at days 2 and 60 post-Tx treatment. Rgs5/Tomato—red, DAPI stained nuclei—blue. CB—cortical bone. For color figure refer to online version. One-way ANOVA with Tukey post hoc test analysis of each population. P values indicated by **<.01,***<.001.
Next, we analyzed bone digests by flow cytometry 2, 20, and 60 days post Tx (Fig. 2I, 2J; Supplementary Fig. S5). The frequency of Rgs5/Tomato+ cells is highest at day 2 post Tx and then decreased (Fig. 2I). To understand this decrease in lineaged traced cells, we evaluated the distribution of cluster of hematopoietic or endothelial populations using CD45 and CD31 antibodies within total Rgs5/Tomato+ cells. By day 2 post Tx, 78% of these cells are hematopoietic (CD45+CD31−) which decreases to 59% by day 20 and to 30% by day 60 post Tx (Fig. 2J).
Of Rgs5/Tomato+ cells, 2% are lineage-traced vascular endothelial cells (CD45-CD31+) at day 2 post Tx, and then the proportion increases to 7% and 12% by days 20 and 60 post Tx respectively. Within the Rgs5/Tomato+ cells, there are 16% of lineage-traced mesenchymal subset (CD45-CD31−), which increases over time to 29% at day 20 and 56% at day 60 post Tx (Fig. 2J). This dynamic change in cell types during lineage tracing within the bone marrow microenvironment is evident by histology as seen on day 2 post Tx, when the majority of Rgs5/Tomato+ cells are small, round hematopoietic cells with some mesenchymal-like cells, but by day 60 post Tx, the population is almost all stromal/mesenchymal (Fig. 2K).
Lineaging Tracing of Rgs5/Tomato+ Hematopoietic Cells
Neutrophils have a short lifespan of about 14 days and express RGS5.31,32 We hypothesized that the decrease seen in lineaged traced hematopoietic cells was due to neutrophil death or egress from the bone marrow (Fig. 2J). As shown in Supplementary Fig. S6A, S6B, bone marrow isolates gated on CD45+ cells at day 2 were 7.88% Tomato+ which decreased to 0.18% by day 20 and 0.11% by day 60 of lineage tracing. We confirmed that 99.7% of CD45+Tomato+ cells at day 2 post-Tx treatment are CD11b+/Gr-1+ neutrophils (Supplementary Fig. S6C). Small Rgs5/Tomato+ cells (white arrows in Supplementary Fig. S6D) co-localize with CD45, while more stromal-like cells with bright Tomato expression did not coexpress CD45 (yellow arrow).
Rgs5/Tomato+ Lineage Express Markers of Mesenchymal/Stromal Cells
We used flow cytometry to assess osteoprogenitors phenotype of Rgs5/Tomato+ within bone digests (Fig. 3). Endothelial and hematopoietic lineages (Fig. 3A) were excluded by using CD31, CD45, and Ter119, respectively and then cells gated on Tomato+ cells. The frequency of Rgs5/Tomato+ cells within the digests is 3.1% at day 2 post-Tx, rises to a peak of 7.4% by day 20 and then declines to 1.1% by day 60 post Tx (Fig. 3B, 3C), a time when most lineage traced cells have embedded into matrix as osteocytes (Fig. 2E). Rgs5/Tomato+ cells at day 2 were 10.5% Sca1+, which increased to 24.6% by day 20 and to 69.6% by day 60 of lineage tracing, indicative of expansion of these cells (Fig. 3D, 3E and Supplementary Fig. S7).33 This coincides with expansion of Tomato+ cells in trabecular region (Fig. 2B). Appearance of Tomato+/Osx+ cells and expression of CD51—markers of more committed cells (Fig. 2B–2D). Importantly, by day 60 there is histological evidence that some Rgs5/Tomato+ cells are embedded in bone matrix as osteocytes, resulting in a decrease in Sca1-/CD51+ cells as these cells are not present in the bone digest (Figs. 2E–G, 3E). To evaluate further the identity or Rgs5/Tomato+ cells at day 2 post Tx, digests from 5-month-old mice were analyzed by flow cytometry for the mesenchymal markers PDGFRβ and LeptinR (Fig. 3F–3H). At that time point, 29% of Lin-Rgs5/Tomato+ cells were PDGFRβ+ and 28% of lin-Rgs5/Tomato+ cells are LeptinR+.
Mesenchymal/stromal Rgs5/Tomato+ lineage traced cells in bone. Bone digests from 3-week-old male and female mice (n = 3 per group) were analyzed 2, 20, and 60 days post Tx by flow cytometry. (A) Dot plots presenting gating strategy to isolate single, live cells that are lineage negative (CD31 and CD45, Ter119) and Tomato positive. (B) Dot plots (n = 3) of lineage traced animals 2, 20, and 60 days after Tx treatment which is quantified in (C) as frequency of Tomato+ cells within lineage-gate. (D) Dot plots gated on Lineage-Tomato+ cells for Sca1 and CD51 expression and (E) One-way ANOVA with Tukey post hoc test analysis of each population. P values indicated by *<.05, **<.01. (F–H) Lineage-Rgs5/Tomato+ concatenated flow cytometric dot plots (n = 3) of bone digests 2 days post Tx for PDGFRβ (G) and LeptinR (H). Unstained control is shown for both PDGFRβ and LeptinR.
Rgs5/Tomato+ Cells Do Not Contribute to Adipogenesis
We tested if Rgs5/Tomato+ cells could differentiate into adipocytes. To induce bone marrow adipogenesis, Tx injected mice were fed with a high-fat diet containing rosiglitazone for 1 month (Fig. 4A). Histological frozen sections immunostained for the adipocyte marker perilipin showed no co-localization with Rgs5/Tomato+ cells (Fig. 4B).
Rgs5/Tomato+ do not contribute to adipogenesis. (A) Experimental design of adipogenic induction in Tx treated Rgs5/Tomato animals. (B) Representative histology of bone marrow and trabecular region of femoral bone stained for perilipin (yellow) to identify adipocytes (white arrows). Red—lineage traced Rgs5/Tomato+ cells do not co-localize with perilipin staining in metaphyseal or epiphyseal regions. For color figure refer to online version. Four animals were analyzed and 3 sections per femur.
Rgs5/Tomato+ Cells are Primarily Contributing to Bone Marrow Regenerative Response
The primary goal was to evaluate contribution of Rgs5 lineage traced cells toward repair after injury. Two different models were utilized that allow to distinguish contribution from periosteal versus bone marrow derived cell compartments. Femoral fracture injury model was used to assess complex bone repair with contribution of periosteum and bone marrow to repair versus a bone marrow pin insertion injury that excludes contribution of periosteum.
Fractured mice were injected with Tx and then analyzed by histology at 10 days post fracture (dpf) (Fig. 5A). We confirmed this injury model had no leaky Tomato expression in fractured Rgs5-CreER/Ai9+ animals with no Tx treatment (Fig. 5B). Interestingly, by 10 dpf, Rgs5/Tomato+ cells were localized in the area of woven bone around the pin (Fig. 5C1) and fibroblastic cells close to fracture line (Fig. 5C2). Rgs5/Tomato+ were positive for osterix (Fig. 5D) and osteocalcin (Fig. 5E) confirming their differentiation ability into osteoblasts. Fewer Rgs5/Tomato+ cells were observed within the periosteal callus and they do not appear to contribute to osteogenic Osx+ cell population (Fig. 5C3). The total number of Osx+ cells that were Tomato+ in entire callus was quantified to be 4.1% and 18.3% in bone marrow (Fig. 5F).
Contribution of Rgs5/Tomato+ cells to fracture healing. Histology sections and flow cytometry of 8-week-old male and female mice (n = 4) after fracture. Dotted line indicates fracture callus, solid line fracture site. (A) Experimental design. (B) Ten dpf histology for osterix (green) and DAPI (blue) in corn oil-treated control. (C) Representative histology of Rgs5/Tomato+ lineage traced cells at 10 dpf fracture stained with osterix (green); 1. Higher magnification of woven bone within marrow space demonstrating Tomato+osterix+ cells (white arrows). 2. Rgs5/Tomato+ fibroblasts in callus. 3. Indicates new bone in fracture callus with no lineage traced Rgs5/Tomato+ cells. (D-E) High power magnification of Tomato+ cells within woven bone in marrow space co-stained with osterix (D) and osteocalcin (E). White arrows indicate co-localization with lineage traced Rgs5/Tomato+ cells. (F) Quantification of Tomato+Osterix+ cells in callus and bone marrow. (G-I) Flow cytometric analysis at 7 dpf. (G) Dot dot plots of CD31 and Tomato cells gated from live lineage negative cells of digests from control periosteum, and Tx treated intact and fracture callus periosteum. (H-I) Proportion of CD51 and Sca1 in Lineage-CD31-Tomato+ fractions from intact periosteum and callus digests. For color figure refer to online version. *P < .05.
Flow cytometry analysis shows increase of Rgs5/Tomato+ cells within the periosteal callus tissue. Within the lineage-gate (CD45-Ter119−), the percentage of CD31+Tomato+ vasculature was notably decreased in callus, from 1.0% to 0.1%, showing that RGS5 lineage traced cells have low endothelial progenitor potential in response to bone injury (Fig. 5G; Supplementary Fig. S8A). Lineage-CD31-Tomato+ population increased from 1.7% to 5.0% during fracture healing. Skeletal progenitor cell markers CD51 and Sca1 were analyzed (Fig. 5H–5I; Supplementary Fig. S8B, S8E). Almost all Lin-/Rgs5/Tomato+ cells were CD51+ with no change in CD51 marker expression seen between intact periosteum and callus. However, within Rgs5/Tomato+, a population of Sca1+ cells decreased from 74.6% to 39.4% during fracture healing indicating lower proportion of Rgs5/Tomato+ progenitors within callus. This low progenitor activity is also indicated by low contribution of Rgs5/Tomato+Osx+ population in callus at 10 dpf (Fig. 5C3).
Rgs5/Tomato+ contribution to bone marrow repair was evaluated using a bone marrow injury model generated by pin insertion. Injured bone and intact contralateral bone were analyzed 7 days post-injury by histology, and enzymatically digested samples of bone were analyzed by flow cytometry (Fig. 6). Histological evaluation of injured tibias showed that some lineage traced Rgs5/Tomato+ cells express Osx+ in the newly formed bone (Fig. 6C), and also Osteocalcin+ (Fig. 6E). At 7 days post-injury, 4.8% of Osx+ cells are Tomato+ compared to 2.0% in intact contralateral bones (Fig. 6D). Bone digests analyzed by flow cytometry showed that a proportion of lineage-CD31-Tomato+ cells increased after injury, from 0.36% to 2.44% (Fig. 6F). Similar to periosteum, the majority of lineage-CD31-Tomato+ cells were CD51+ with no difference observed after injury (Fig. 6G). However, following bone marrow injury the frequency of Sca1+ progenitor cells increases from 16.6% to 31.2% (Fig. 6H) and contributes to osteogenesis as observed by co-expression of Tomato and osterix expression.
Contribution of Rgs5/Tomato+ cells following bone marrow injury. (A) Experimental design. Bone marrow injury was induced in 6-week-old mice (n = 5). To induce TdToamto expression in RGS5-Cre cells, Tx was injected on days 0, 2, and 4 days after the injury. Animals were sacrificed 7 days post-injury. (B) Uninjured contralateral tibia as control. (C) Cryosection stained for osterix (green) in Rgs5/Tomato bone after marrow injury. White boxes indicate magnified areas. (D) Quantification of Tomato+Osterix+ cells. (E) Osteocalcin staining, arrows indicate colocalization of osteocalcin and Tomato. (F) Dot plots of CD31 and Tomato cells gated from Lineage- cells of digests from intact and injured bones. (G-H) Frequencies of CD51 and Sca1 in Lineage-CD31-Tomato+ fractions from intact and injured digests. For color figure refer to online version.
Discussion
We have developed an animal model suitable for lineage tracing of RGS5-expressing cells. Crossing it with Ai9 reporter mouse (Tomato) revealed that Rgs5/Tomato+ cells are heterogenous detected within bone including the periosteum, cortical channels, bone marrow, and trabecular surfaces (Fig. 7A). Lineage tracing experiments (day 20) revealed that Rgs5/Tomato+ cells trace to the trabecular and endosteal surfaces as osteoblasts expressing osterix. By day 60, a portion of labeled cells express osteocalcin and have become osteocytes embedded in matrix. Rgs5/Tomato+ cells were also detected in bone marrow as stromal cells.
Lineage traced Rgs5/Tomato+ cells within intact and injured bone. (A) Heterogeneous population of Rgs5/Tomato+ cells is present within the bone 2 days post-Tx treatment. Majority of RGS5/Tomato+ mesenchymal/stromal cells resides within metaphyseal region, and within bone marrow they mainly represent neutrophil population. Rgs5/Tomato+ vascular endothelial cells are also observed within bone marrow 2 days post Tx. (B) After fracture injury, Rgs5/Tomato+ lineage trace to Osterix+ cells within woven bone in marrow, while in periosteal callus they do not contribute to osteoblasts. (C) RGS5/Tomato+ cells after marrow pin injury lineage trace to Osterix+ cells within areas of newly forming bone.
With available osteoprogenitor labeling transgenic models like αSMACreER it is not possible to distinguish contribution of progenitor populations deriving from bone marrow versus periosteum in fracture healing. Our results indicate that a population of Rgs5/Tomato+ cells contributes to injury repair via expansion and osteogenesis from cells within bone marrow (Fig. 7B–7C). In contrast to αSMA population, RGS5 labeled cells did not show major contribution to osteogenesis within fracture callus at that healing stage.
Therefore, the bone formation seen in an bone marrow injury model, being mostly a marrow-driven response from Rgs5/Tomato+ cells, may be merely due to the difference in these subsets with a more Sca1+ osteoprogenitors present in the bone marrow compared to periosteum. We did not observe difference in expression of CD51 within Rgs5/Tomato+ cell population. Rgs5/Tomato+ marker expression varies depending on the microenvironment from within it develops; with cells from within the metaphyseal growth plate area being more committed osteolineage cells (CD51+Sca1−) and those expressing Sca1+ representing earlier skeletal progenitor.34,35 During lineage tracing, Sca1+ population increased within Rgs5/Tomato+ cells within the bone marrow population, while commitment of cells into mature osteobalsts/osteocytes results in a decrease of Sca1-CD51+ cells.
We have shown that Rgs5-Cre also labeled populations of endothelial, and hematopoietic cells within bone marrow (Fig. 7A). The proportion of endothelial cells (CD31+) shows some increase during lineage tracing, while Rgs5/Tomato hematopoietic proportion (CD45+) decreases. The main component of cells targeted within hematopoietic system are granulocytes and the pulse-chase experiments show a decrease in hematopoietic Tomato+ cells as a result of their turnover.
Although we did not evaluate the involvement of RGS5-expressing cells in osteogenesis during development, it is possible this cell type is more pronounced during embryonic skeletogenesis. In fact, Rgs5 mRNA expression peaks at around day E14.5 in the mouse embryo,22 a time just prior to when fetal bone is beginning to be vascularized (E15.5), followed by similar activity in primary ossification centers at E16.5.36 Another limitation of this study is that we did not evaluate Rgs5-CreER cells after depletion or by using genetic knockouts—to determine their functional role. Although we did lineage tracing in both growing and adult mice, we did not analyze differences in the potential of these cells within these 2 groups. Rgs5/Tomato+ cells were labeled by injecting tamoxifen twice in 2 days (for injury, 3 times in 2 days) and it is plausible that longer treatment would label more cells tracing into osteogenic population within bone marrow and bone itself.
The primary focus of our work was to evaluate a RGS5 population of cells for their contribution to bone repair process. We have utilized 2 systems of injury, one that involves bone marrow and periosteum (femoral fracture) and other that results in response restricted to bone marrow (bone marrow pin insertion). In contrast to previously developed models such as SMACreER, PrxCreER, PDGFRαCreER, and GliCreER that target populations of progenitors in both periosteum and bone marrow, Rgs5-Cre contribution to osteoblast lineage is primarily restricted to bone marrow response with minimal osteoblast contribution of Rgs5/Tomato+ cells within periosteum.8,15,37-39 Furthermore, Rgs5/Tomato cells have low level of contribution to cartilage.
In conclusion, Rgs5 in the context of bone labels a heterogeneous population of cells including some of the hematopoietic and endothelial cell populations and mesenchymal/stromal cells with osteoprogenitor potential. Rgs5 cells lineage trace to osteoblasts and osteocytes under homeostatic conditions and contributes to woven bone within the marrow after fracture and marrow injury. Osteoprogenitors are a heterogenous population of cells and Rgs5 expressing cells are likely part of progenitor reservoirs within bone marrow that await certain microenvironmental cues to expand and differentiate.
Acknowledgments
We would like to thank Erin Wilkie and Charlie Leary for technical assistance in weaning and genotyping the transgenic mice.
Funding
This work has been supported by National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases grants AR055607 and AR070813 to IK.
Conflict of Interest
All of the authors declared no potential conflicts of interest.
Author Contributions
S.H.R., S.N.: conception and design, collection/and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript. I.V.M: collection/and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript. M.S.K.: collection/and assembly of data, data analysis and interpretation, final approval of manuscript. Y.C.: data analysis and interpretation, final approval of manuscript. I.K.: conception and design, financial support, collection/and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript.
Data Availability
The data that support the findings in this manuscript areavailable from the corresponding author.
