Diabetes impairs periosteal progenitor regenerative potential

Abstract

Diabetics are at increased risk for fracture, and experience severely impaired skeletal healing characterized by delayed union or nonunion of the bone. The periosteum harbors osteochondral progenitors that can differentiate into chondrocytes and osteoblasts, and this connective tissue layer is required for efficient fracture healing. While bone marrow-derived stromal cells have been studied extensively in the context of diabetic skeletal repair and osteogenesis, the effect of diabetes on the periosteum and its ability to contribute to bone regeneration has not yet been explicitly evaluated. Within this study, we utilized an established murine model of type I diabetes to evaluate periosteal cell differentiation capacity, proliferation, and availability under the effect of a diabetic environment. Periosteal cells from diabetic mice were deficient in osteogenic differentiation ability in vitro, and diabetic mice had reduced periosteal populations of mesenchymal progenitors with a corresponding reduction in proliferation capacity following injury. Additionally, fracture callus mineralization and mature osteoblast activity during periosteum-mediated healing was impaired in diabetic mice compared to controls. We propose that the effect of diabetes on periosteal progenitors and their ability to aid in skeletal repair directly impairs fracture healing.

Graphical Abstract

Introduction

Diabetes mellitus is a chronic metabolic disease characterized by lack of insulin production or low insulin sensitivity¹. The prevalence of type I (T1D) and type II diabetes (T2D) in the United States and worldwide is increasing across multiple age groups; importantly, this disease and the associated comorbidities have a serious health impact, as well as an enormous economic burden on healthcare resources that is expected to keep rising^2,3.

At the molecular level, both T1D and T2D result in oxidative stress due to increased levels of reactive oxygen species (ROS) and harmful advanced glycation end-products (AGEs). AGEs accumulate on proteins via a glycosylation reaction when excess levels of reducing sugars are present⁴, and the receptors for AGEs (RAGEs) are key mediators of the signaling transduction that perpetuates oxidative stress and leaves cells more prone to apoptosis^5,6. ROS and AGE accumulations in the context of high blood glucose amplify this cellular stress that underlies the pathogenesis of diabetes, having detrimental effects on multiple organ systems, most notably the nervous and cardiovascular systems. Diabetes also has major damaging consequences on the skeleton and its associated tissues^1,7.

The proinflammatory state instigated by ROS and hyperglycemia has significant consequences on both osteoblasts and osteoclasts, the cell types responsible for bone remodeling, as well as their precursors^8–11. Under diabetic conditions, there is impaired differentiation of mesenchymal progenitor cells toward the osteoblast lineage with a concomitant increase in adipogenic lineage differentation^9,12–15. A diabetic state promotes the differentiation of myeloid-lineage pre-osteoclasts into mature osteoclasts, increasing bone resorption activity^16,17. Combined, this leads to decoupling of bone-forming and bone-resorbing activities, disturbing homeostatic bone remodeling and prompting an increased risk for osteopenia^4,18,19. This holds direct clinical significance, as osteopenic and osteoporotic bones are more likely to fracture^20–23. Additionally, a recent study has concluded that both T1 and T2 diabetics are at a significantly increased risk for hip and non-vertebral fractures²⁴.

One of the defining characteristics of diabetes is poor wound healing^25–27. Despite the remarkable regenerative capacity of bone, fracture healing in diabetic patients is delayed by an average of 87%²⁸; this is accompanied by a dramatically increased risk of healing complications, including delayed union and nonunion of the bone, further increasing the morbidity of this disease and the overall economic burden²⁹. The fracture healing process requires a complex orchestration of cell types and signaling pathways in order to properly recruit and differentiate progenitors at the site of injury. Osteochondral progenitors from the bone marrow, periosteal layer, muscle, endothelial cells, and various other sources have been reported to participate in this complicated regenerative mechanism; the effort to drive progress towards relevant therapeutics warrants a greater understanding of these cells and the cellular mechanisms that control their proliferation/differentiation^30–33.

The periosteum is the outer connective tissue covering on bones, and consists of an outer fibrous layer composed mainly of fibroblasts, and an inner, cellular cambium layer that harbors periosteal progenitor cells^30,34. Periosteal osteochondral progenitors are able to give rise to both chondrocytes and osteoblasts, and are the major contributors to fracture healing following an injury to the bone^35,36, as removal of the periosteum has been shown to severely impair efficient fracture healing^30,35. Periosteal cells proliferate extensively at the fracture site to help form the callus, stabilizing the fracture by undergoing chondrogenesis and osteogenesis as the bone regenerates³⁷. Recent studies have demonstrated how bone harbors distinct progenitor cell populations, and progenitors within the bone marrow and periosteum are differentiated by specific molecular signatures^31,38,39.

Fracture healing and bone regeneration studies in diabetic animal models have been previously described, where diabetic fracture healing is marked by delayed skeletal healing, higher incidence of delayed union and nonunion, and an irreversible decrease in the amount of bone marrow-derived mesenchymal progenitors that aid in the repair process^9,14,40–42. Additionally, diabetic animal model fractures resulted in reduced structural strength⁴³ and mechanically compromised healed bones even after completed healing⁴⁴. The use of diabetic animals is even becoming a method for evaluating bone healing and skeletal progenitors in a model of impaired osteogenesis⁴⁵. However, the specific effect on the periosteum and its resident osteochondral progenitors in impaired diabetic fracture healing has not been described. We hypothesized that diabetic conditions impair the regenerative potential of periosteal progenitor cells, and that the effect of diabetes on the periosteal layer contributes to poor skeletal healing.

Materials and Methods

Animal Model:

Col2.3GFP reporter mice drive GFP expression by a 2.3kb promoter fragment of type I collagen, specific for committed osteoblasts⁴⁶. We additionally used these mice crossed with the inducible osteoprogenitor cell reporter line, αSMACreER^T2/Ai9³⁶. The αSMACreER^T2/Ai9/Col2.3GFP double-reporter mice are hereafter described as Col2.3/SMA9. Diabetes was induced with high dose I.P. streptozotocin (STZ) injections in citrate buffer (150 mg/kg body weight)^47,48. Weight and fasting blood glucose levels were tested at four weeks post-injection, when mice were 8–12 weeks old. In this experimental model of type I diabetes, each group of STZ-treated mice had significantly higher fasting blood glucose levels with no significant corresponding weight change compared to control groups (Table S1). Both male and female mice were used, and each experiment was appropriately age- and gender-matched.

Periosteal Isolation:

Periosteal cells were isolated from intact femurs and tibiae from 8-week-old mice as previously described with a precise and established technique^49,50. This requires three to four mice per experimental group for appropriate cell numbers, and both males and females were used in equal numbers. Briefly, muscle, ligaments, and connective tissue were carefully dissected away from the bone with a scalpel. Epiphyses were cut from the end of each bone and the marrow was flushed with a 26-gauge needle using basal media comprised of alpha-minimum essential medium (αMEM) (Sigma, St. Louis, MO, USA) supplemented with 10% heat inactivated fetal bovine serum (FBS) (Hyclone, Logan, UT, USA) and 1% penicillin/streptomycin. Periosteum was scraped from flushed, hollow bones into PBS and digested enzymatically for one hour in an orbital shaker (0.05% collagenase P, 0.2% hyaluronidase in PBS). Complete media was added to stop enzymatic digestion and cells were filtered through a 40 μm mesh strainer (Fisher Scientific, Hampton, NH, USA) to form a single cell suspension and remove debris. Cells were grown in 5% oxygen for 4 days before a media change and transfer to 20% oxygen.

Osteogenic and Chondrogenic Assays:

Primary periosteal cells were isolated from long bones and cultured until confluence as described above. Primary cells were passaged and seeded at the same density in 12-well plates (2×10⁵ cells/well). For osteogenic differentiation assays, cells were treated with differentiation media (50 μg/mL ascorbic acid, 4 mM β-glycerophosphate) starting on day 7 of culture. Media was changed every other day for the remainder of the assay. For experiments that tested the effect of an in vitro diabetic environment on osteogenesis, periosteal cells were cultured in either basal media with unaltered glucose concentration (5 mM), in high glucose (25 mM), or in high glucose (25 mM) plus pretreatment one day prior to osteogenesis with advanced glycation end products (AGEs, 2 μg/mL)^15,51; no additional AGEs were added to the media throughout the rest of the osteogenic assay. AGE-BSA (10 mg/mL) was purchased from EMD Millipore (Burlington, MA, USA). When using cells harvested from Col2.3/SMA9 mice, 4-hydroxytamoxifen (1 μM) (Sigma) was added to culture media on days 2 and 5 to induce αSMA-driven Ai9 expression. Cell culture fluorescence was viewed using a Leica microscope. For chondrogenic assays, passaged cells in 12-well plates were treated with chondrogenic differentiation media starting when cells reached 80–90% confluence, as we have previously described^49,52. Chondrogenic differentiation media consisted of high glucose DMEM supplemented with TGFβ3 (10 ng/mL), L-proline (40 μg/mL), dexamethasone (10⁻⁷ M), ITS+1 premix (1x), nonessential amino acids (1%), and ascorbate-2-phosphate (25 μg/mL)^49,52. Media was replaced every other day for the remainder of the assay. We tested the effect of AGEs on periosteal cells in the same manner as in our osteogenic assays (pretreatment with 2 μg/mL). All cell culture images and qRT-PCR analyses of in vitro cell culture are representative of three separate experiments.

RNA isolation and cDNA synthesis for qPCR:

Total RNA was isolated using a TRIzol (Invitrogen, Carlsbad, CA, USA) extraction method and cDNA synthesized using Superscript III cDNA synthesis kit (Invitrogen). qRT-PCR reactions were performed on a Bio-Rad real-time PCR system using SYBR Green (BioRad, Hercules, CA, USA) PCR reagents, and are presented relative to Gapdh (F 5’-AGGTCGGTGTGA-ACGGATTTG-3’, R 5’-TGTAGACCATGTAGTTGAGGT CA-3’). Each graph is representative of at least three separate experiments. Additional primer sequences are listed in Table S2.

Alizarin Red S Assay:

Periosteal cells were isolated from long bones and cultured in osteogenic media as described above^49,50. After 21 days of osteogenic differentiation, wells were stained with Alizarin Red S (ScienCell, Carlsbad, CA, USA). Staining was quantified according to manufacturer’s instructions using colorimetric detection.

Surgical Procedure:

Closed femoral fractures were generated as previously described^49,53. Briefly, mice were anesthetized using 4% isoflurane and a 26-gauge needle (BD) was hand-drilled into the medullary cavity of the left femur under the patella. The needle was removed and immediately replaced with a blunt pin (0.38 mm). After X-ray confirmation of pin placement, femurs were fractured using an Einhorn three-point bending device and evaluated immediately using x-ray. Buprenorphine (0.1 mg/kg) was administered by subcutaneous injection just prior to the procedure, and twice a day for three days following the fracture for pain management. Samples were excluded from the study analysis in cases of comminution, bent pins, and fracture placement that was not mid-diaphyseal. Col2.3/SMA9 mice were injected with tamoxifen (7.5 μg/gram body weight) the day of fracture and one day post-fracture to induce αSMA-driven Ai9 expression. Mice were euthanized via CO₂ and cervical dislocation 3 days post-fracture for flow cytometry analyses, and 14 days post-fracture for histological analyses.

Flow Cytometry:

Diabetic mice and controls underwent unilateral femoral fracture. Periosteal cells from intact or fractured femurs of diabetic and control mice were isolated and digested into a single cell suspension as described above. Following red blood cell lysis in ammonium-chloride-potassium buffer, approximately 2–4×10⁶ cells per mouse were collected in cell staining media containing HEPES and 2% heat inactivated FBS in Hanks balanced salt solution. Cells were incubated with a combination of pacific blue-conjugated anti-CD45, pacific blue-conjugated anti-Ter119, pacific blue-conjugated anti-CD31, A700-conjugated anti-Sca1, PerCP-Cy5.5-conjugated anti-CD105 (BD Biosciences, San Jose, CA, USA), PE/Cy7-conjugated anti-Ly-51, Brilliant Violet 510-conjugated anti-CD90.2, PE-conjugated CD51, and APC-conjugated CD200 (Biolegend, San Diego, CA, USA) at a 1:400 dilution in staining media for 1 hour on ice. Cells were washed in staining media and stained for live-dead with Zombie UV (Biolegend, San Diego, CA, USA) and analyzed using a BD-LSRII flow cytometer (BD Biosciences, San Jose, CA, USA). Following gating for live-dead and background staining using isotype controls, the percentage of CD45⁻CD31⁻Ter119⁻ lineage negative (Lin⁻) cells expressing established mesenchymal markers was examined.

In Vivo Proliferation Assay:

Diabetic mice and controls underwent unilateral femoral fracture. 100 μL 5-ethynyl-2-deoxyuridine (EdU, 10mM) from a Click-iT EdU Alexa Fluor 647 kit (Invitrogen, Carlsbad, CA, USA) in DMSO was injected on the second day post-fracture to label proliferating cells at the fracture site during the early phases of healing. Periosteal cells from intact or fractured femurs of diabetic and control mice were isolated and digested into a single cell suspension as described in methods above. Cells were incubated with pacific blue-conjugated anti-CD45, pacific blue-conjugated anti-Ter119, pacific blue-conjugated anti-CD31, and Alexa Flour 700-conjugated anti-Sca1 at a 1:400 dilution in staining media for 1 hour on ice. Following staining for live-dead, cells were stained using the Click-iT EdU Alexa Fluor 647 kit following manufacturer’s instructions cells were analyzed on a BD-LSRII. The percentage cells expressing Alexa Fluor 647 (EdU+, proliferating cells) was examined. Analysis was performed using DIVA and FlowJo software.

Histology:

Bones were harvested 14 days post-fracture and fixed in neutral buffered formalin for 5 days, soaked in 30% sucrose overnight, and embedded for frozen sectioning. Undecalcified sections were cut using a tape-transfer system at 7 μm on a cryostat. Sections were imaged for fluorescence on a Leica microscope and analyzed using ImageJ. To indirectly quantify osteoblastic activity in periosteal-mediated new bone formation 14 days post-fracture, we analyzed the four areas flanking the fracture site for integrated density and averaged this value for each individual sample (Fig. S1).

Statistical analyses:

Results were analyzed using Student’s t-test using GraphPad Prism v6.0. Data are presented as mean ± SD. p-values < 0.05 were considered statistically significant.

Animal welfare:

This study was approved by the local IACUC and was conducted in accordance with the national legislation on protection of animals and the NIH Guidelines for the Care and Use of Laboratory Animals.

Results

To study the effects of a diabetic state on the periosteum’s contribution to skeletal regeneration, we first evaluated periosteal progenitors under an established system of in vitro diabetic conditions as described in the methods¹⁵. Our methodology was based on the likelihood that diabetes-induced mechanisms, including a hyperglycemic diabetic environment and the accumulation of AGEs, would directly result in decreased differentiation potential of periosteal progenitors; similar results have been seen previously using bone marrow-derived mesenchymal cells⁵⁴. We utilized Col2.3GFP/SMA9 mice, in which GFP expression is driven from a type I collagen promoter and tdTomato reporter expression is activated by an inducible alpha smooth muscle actin (αSMA) CreER (termed SMA9)³⁶. Periosteal cells from Col2.3GFP/SMA9 mice were differentiated toward the osteoblast lineage for 21 days in vitro to determine the effect on mature osteoblast formation (Fig. 1A). Cells exposed to this artificial in vitro diabetic environment prior to differentiation had significantly reduced osteogenic differentiation capabilities after 21 days; this inhibitory effect was observed as a reduction in calcium deposits via alizarin red staining (Fig. 1B), significantly lower Ocn expression indicating fewer mature osteoblasts in culture (Fig. 1C), and reduced formation of Col2.3GFP+ osteoblastic colonies (Fig. 1D). Notably, periosteal cells differentiated under in vitro diabetic conditions exhibited marked retention of Ai9+ (αSMA-derived) cells, further indicating a presence of less mature cells with evident lack of osteogenic differentiation compared to control cutures (Fig. 1D).

Fig. 1.

(A) Schematic of in vitro diabetic environment osteogenic assays. Periosteal cells were isolated from tibias and femurs and cultured to confluence. Cells were replated and exposed to 4-hydroxytamoxifen (1 mM) to induce αSMA-driven Ai9 expression, followed by osteogenic differentiation for 21 days as described. For treatment with both high glucose + AGEs, cells were pretreated for one day prior to osteogenesis with 2 μg/mL AGEs. (B) Cells were stained with alizarin red S to visualize calcium deposits. Insets=10x (C) RNA was isolated from these cells after 21 days of osteogenic differentiation, and analyzed for expression levels of the late osteogenic gene Ocn. N=3 **p<0.005, ***p<0.0005 (D) After 21 days of osteogenic differentiation, cells were analyzed for GFP fluorescence (indicating mature osteoblastic colonies) and αSMA-driven Ai9 (indicative of undifferentiated cells). Scale bar = 250 μm. Images are representative of three separate experiments.

Our findings through the use of in vitro diabetic conditions prompted us to move to an in vivo approach of diabetic induction. We used a well-established murine model of T1D based on stretozotocin (STZ) destruction of pancreatic beta cells^22,47,55. Periosteal cells were isolated directly from these STZ-treated mice and age-matched controls (Fig. 2A, B). Even at confluence prior to induction of osteogenic differentiation, periosteal cells isolated from STZ-treated diabetic mice had a significantly reduced expression level of osterix (Sp7), a transcription factor that is necessary to drive mesenchymal progenitors toward the osteoblast lineage, and Runx2, which is required for osteoblast commitment and differentiation (Fig. 2C). We then evaluated the ability of these cells to differentiate in osteogenic media at multiple time points. Cells isolated from STZ-treated mice exhibited delayed early in vitro osteogenic differentiation after 7 days, as shown by decreased expression of transcription factors important to this phase of osteogenesis, including Sp7 and Runx2, as well as alkaline phosphatase (Alpl), which is associated with bone formation and metabolism (Fig. 2D).

Fig. 2.

(A) Schematic of STZ-induced mouse model. Mice were injected with vehicle or a high dose of STZ, and four weeks, later, fasting blood glucose measurements and weight were recorded to confirm diabetic status. Periosteal cells were isolated from tibias and femurs for in vitro assays. (B) Primary periosteal cells were cultured to confluence in basal media. Cells were isolated at confluence, or seeded at the same density in 12-well plates and induced to differentiate toward the osteoblast lineage for 7 or 21 days. (C) At confluence, periosteal cells were isolated and cultured, then analyzed for relative expression of Sp7 and Runx2 via qRT-PCR. N=3 *p<0.05 (D) After 7 days of osteogenic differentiation, cells were analyzed for expression levels of early osteogenic genes and transcription factors, including Alpl, Sp7, and Runx2. N=3 *p<0.05, **p<0.005 (E) After 21 days of osteogenic differentiation, cells were analyzed for expression levels of late osteogenic genes including Ocn and type I collagen (Col1a1). N=3 *p<0.05 (F) After 21 days of osteogenic differentiation, plates were analyzed for GFP+ mature osteoblastic colonies. Images representative of three separate experiments. Scale bar = 250 μm (D)

Following 21 days of osteogenic induction, decreased expression levels of both Ocn and Col1a1 were detected in cells derived from STZ-treated mice (Fig. 2E). While periosteal cells from control animals were successfully able to differentiate into osteoblastic colonies by this time point, cells isolated from STZ-treated mice exhibited delayed differentiation, as shown via lack of type I collagen-driven GFP fluorescence in culture (Fig. 2F).

As our results suggested that periosteal cells from STZ-treated mice exhibit impaired osteogenic differentiation in vitro, we moved to a fracture model to study their differentiation abilities in vivo. As bones heal, cells from various sources including the periosteum and bone marrow stroma proliferate and differentiate into chondrocytes and osteoblasts³⁰. STZ-treated and control Col2.3GFP/SMA9 mice underwent closed, stabilized, femoral fractures. To delineate how periosteal cells are affected in the earliest fracture healing phase following injury in this diabetic model, we utilized flow cytometry to assess changes in periosteal cell proliferation in STZ-treated mice compared to controls. To capture periosteum in the “activated” state three days after fracture, EdU was injected one day prior to sacrifice to enable the labeling of proliferating cells (Fig. 3A), and we employed a simple gating strategy as defined in Fig. 3B. We observed no difference in proliferation of mesenchymal cells within uninjured femur periosteum between STZ-treated mice and controls; these basal results are expected without an injury, and consistent with similarly performed in vivo proliferation assays in the skeleton^49,56. However, following fracture-induced injury, periosteal cells in diabetic mice exhibited significantly reduced proliferation at the fracture site relative to controls (23.6% reduction in EdU⁺ cells, Fig. 3C). We observed a similarly significant reduction in proliferating cells in STZ-treated mice following fracture when gating on a Sca1⁺ population, further narrowing the results to periosteal progenitors (33.2% reduction in EdU⁺Sca1⁺ cells, Fig. 3D).

Fig. 3.

(A) Schematic of flow cytometry-based approach to assess periosteal proliferation before and after a fracture insult. Mice were injected with vehicle or a high dose of STZ, and four weeks, later, fasting blood glucose measurements and weight were recorded to confirm diabetic status. Mice underwent femoral fracture and were injected with EdU one day prior to sacrifice. Periosteal cells were isolated from tibias and femurs of either intact tibias and femurs, or from fractured femurs three days after injury. Cells were stained for cell surface markers and analyzed by flow cytometry using the gating strategy in (B). Periosteal cells from intact bones and contralateral fractured femurs were analyzed for the total proliferating (EdU⁺) cell populations (C) as well as the Sca1⁺EdU⁺ proliferating population (D). N=4 *p<0.05

We hypothesized that a reduction in periosteal proliferation following skeletal injury may be a result of fewer self-renewing stem cells or osteoprogenitor cells in the periosteal layer available to contribute to fracture healing. The ability of stem and progenitor cells to self-renew and commit to specialized lineages is an integral part of the tissue regeneration process. Therefore, we further examined the effect of diabetes on periosteal progenitors in vivo by quantifying established progenitor populations. Primary periosteal cells were isolated directly from either intact bones or bones three days post-fracture of STZ-treated and control mice, and analyzed by flow cytometry (Fig. 4A, B). Notably, intact femurs from STZ-treated mice had significantly fewer periosteal mesenchymal progenitors compared to controls, defined as a percentage of CD31⁻TER119⁻CD45⁻ (Lin⁻) Sca1⁺CD105⁺ cells even prior to injury (50.6% reduction, Fig. 4C)⁴⁹. We then evaluated the number of periosteal progenitors following a fracture insult; cells from fractured bones were isolated three days post-fracture, to again capture periosteal “activation” as cells robustly proliferate at the site of injury^30,35,49. Fractured femurs from diabetic mice had significantly fewer Lin⁻Sca1⁺CD105⁺ periosteal mesenchymal progenitors three days post-injury compared to controls (48.4% reduction, Fig. 4D), indicating a possible loss of progenitor renewal capacity following a fracture insult. We then explored the presence of a bona fide mouse skeletal stem cell (mSSC) population in the periosteum during early fracture healing, as this is a subset of bone progenitor cells implicated in skeletal regeneration^57,58. We found that STZ-treated mice had a significant reduction in this highly specific population in injured bones compared to controls (34.9% reduction Fig. 4E, F, Fig. S2A) defined as a percentage of CD90-CD200+ cells from the depicted parent population. A similarly significant reduction in mSSCs was seen in bone marrow-derived cells, and this effect of diabetes on bone marrow-derived progenitor and stem cell populations is consistent with findings from other groups^9,13 (46.1% reduction, Fig. S2B). Interestingly, we also observed significantly fewer Lin⁻GFP⁺ cells in periosteum three days after fracture as well as in intact bones (Fig. S2C), indicating fewer Col1a1-expressing osteoblastic cells in the periosteal layer from STZ-treated mice both at baseline and following injury.

Fig. 4.

(A) Schematic of flow cytometry-based approach. Mice were injected with vehicle or a high dose of STZ, and four weeks, later, fasting blood glucose measurements and weight were recorded to confirm diabetic status. Periosteal cells were isolated from tibias and femurs of either intact tibias and femurs, or from fractured femurs three days after injury. Cells were stained for cell surface markers and analyzed by flow cytometry as shown in (B). Periosteal cells from intact femurs (C) in diabetic mice and controls were analyzed for a specific population of mesenchymal progenitors, as defined by Lin⁻Sca1⁺CD105⁺ cells. (D) Periosteal cells isolated from the fracture site in diabetic mice and controls were analyzed for a population of mesenchymal progenitors, as defined by Lin⁻Sca1⁺CD105⁺ cells. (D) Periosteal cells isolated from the fracture site in diabetic mice and controls were analyzed for a population of bona fide mouse skeletal stem cells defined as CD90-CD200+ cells from the parent Lin-CD51+Ly51-CD105- population (E) and the downstream bone, cartilage, and stromal progenitors (BCSPs), defined as CD90-CD105- cells from the parent Lin-CD51+Ly51- population (F)⁵⁷. N=4 *p<0.05

Finally, we sought to understand how a diabetic state modulates the direct contribution of periosteal osteoprogenitors to fracture healing. Using mice that harbor the Col2.3GFP reporter, isolation of injured bones during the healing process allows for a snapshot of the location of GFP+ mature osteoblasts in the fracture callus. To capture a time point where fracture calluses were in a bone-forming phase of fracture healing, limbs were isolated 2 weeks post-fracture when a majority of cartilage had been resorbed, but the bony callus is actively being formed³³. We performed histological analysis of Col2.3GFP expression, with Ai9-reporter fluorescence overlaid to aid in distinguishing the bone and fracture callus morphology (Fig. 5A,B). Importantly, the combination of this fluorescent mouse model and the day 14 post-fracture time point marks a clear distinction between endochondral and intramembranous post-natal bone formation during skeletal regeneration^32,33; therefore, we were able to focus on specific areas of periosteal-mediated intramembranous bone formation during fracture healing⁵⁹. GFP+ cells can be seen throughout the fracture callus, in areas both flanking the callus (derived from periosteal progenitors) and centrally within the callus (derived from bone marrow mesenchymal progenitors); these GFP+ cells indicate differentiated osteoblasts contributing to skeletal repair.

Fig. 5.

(A) Schematic of the fracture methodology using the STZ-induced Col2.3GFP/SMA9 mouse model. Mice were injected with vehicle or a high dose of STZ, and four weeks, later^1,16, fasting blood glucose measurements and weight were recorded to confirm diabetic status. Mice underwent closed, stabilized femoral fracture. 14 days after fracture, limbs were isolated for histological analysis. (B) Microscopy images demonstrating the use of a Col2.3-driven GFP expression to indirectly visualize osteoblastic mineralizing activity within the callus. Insets show an enlarged view of newly formed bone within the fracture callus. CB=cortical bone, SM=skeletal muscle. Inset = 10x (C) GFP fluorescence intensity of the entire callus as defined in Fig. S1A was quantified 14 days post-fracture. (D) Areas of the callus specific to periosteal-mediated bone formation (Fig. S1B) were analyzed for the GFP expression to indirectly visualize osteoblastic mineralizing activity, and this fluorescence intensity specific to periosteal new bone was quantified 14 days post-fracture. N=3–4 individual mice, *p<0.05

Using imaging analyses, we found significantly reduced Col2.3GFP overall fluorescence intensity within the callus as a whole in STZ-treated mice (34.8% reduction in fluorescence intensity, Fig. 5C), indicating fewer collagen-producing osteoblasts aiding in bone repair compared to controls. To understand the effect of this diabetic model specifically on periosteal cell differentiation and osteogenesis, we focused on particular regions of interest where periosteal-mediated bone formation occurs (Fig. S1)^33,36. We found significantly reduced Col2.3GFP fluorescence in areas of periosteum-specific new bone formation flanking the fracture site in diabetic mice compared to controls, indicating less osteoblastic activity arising directly from periosteal progenitors^60,61 (36.4% reduction in fluorescence intensity, Fig. 5D).

Discussion

Poor fracture healing seen clinically in diabetics is a major public health concern with a large economic burden, necessitating a broader understanding of skeletal progenitors in this disease state in order to develop new therapies to combat these complications³. This study evaluated the effect of diabetes specifically on periosteal progenitors and their ability to contribute to skeletal repair under diabetic conditions. Our focus on the periosteum is timely and largely rooted in a recent body of literature demonstrating the importance of this connective tissue layer to bone regeneration^30,31,36,38.

Within this study, periosteal cells differentiated under established in vitro diabetic conditions or isolated directly from diabetic mice showed impaired osteogenic potential. This was demonstrated through osteogenic gene expression analyses, lack of GFP+, type I collagen-producing colonies formed in culture, and a corresponding retention of fibroblastic αSMA-lineage uncommitted cells. Despite this striking impairment of osteogenic differentiation capacity, we observed similar chondrogenic potential in periosteal cells exposed to AGEs compared to control cells in an in vitro differentiation assay (Fig. S3). These results are unsurprising, as diabetes has not been directly linked to any increased chondrogenic potential in murine mesenchymal cells^62,63, and has instead been categorically associated with reduced osteogenic potential^12,18,29. However, the reduction of mesenchymal progenitors we found via flow cytometry analyses could potentially reduce the amount of cartilage formed in the early stages of fracture healing during a progenitor-to-chondrocyte transition⁶⁴. In addition, a reduction in progenitor cells in diabetic animal models has been shown to be a limiting factor in resolving inflammation during healing, which is essential for cartilage formation⁶⁵.

Cells cultured under in vitro diabetic conditions as well as those isolated directly from STZ-induced diabetic mice are subject to modulatory effects by AGEs, which accumulate under hyperglycemic conditions and impair cellular function^66,67. The inhibitory effect of AGEs and a high glucose environment on osteogenic differentiation of mesenchymal lineage cells has been previously explored at the transcriptional level in other cell types. García-Hernández et al. found that in a human osteoblastic cell line, high in vitro glucose concentrations of up to 24 mM alone can regulate expression levels of Ocn, bone sialoprotein, and Runx2⁶⁸. However, a combination of high glucose and AGEs inhibited mineralization in MC3T3-E1 cells by increasing RAGE expression⁵⁴, supporting our results of a synergistic inhibition of osteoblast maturation and function when high glucose is combined with AGEs. Okazaki et al. have demonstrated that AGEs inhibit osteoblast differentiation in stromal ST2 cells by suppressing expression of Sp7, a transcription factor downstream of Runx2 that is a key regulator of progenitor commitment toward the osteoblast lineage^69–71. In our studies on the periosteum, we find similarly reduced levels of Sp7 and Runx2 in confluent periosteal cultures from diabetic mice; these data suggest that a diabetic environment mechanistically impairs osteogenic potential in isolated progenitors even prior to the start of osteogenesis. Interestingly, after 28 days of in vitro osteogenic differentiation, GFP+ colonies had still failed to develop in periosteal cultures from STZ-treated mice (data not shown), indicating that the in vivo consequence of a diabetic state on mesenchymal progenitors has a longstanding effect even after they are isolated and plated in vitro. It is important to note that AGEs independently, without the compounding effects of high glucose, are able to dramatically alter cellular function. Patel et al. show that AGEs alone impair critical cellular processes and lead to matrix disorganization in rat-derived tendon fibroblasts, which may contribute to diabetic tendinopathy^72,73, and others have demonstrated an AGE/RAGE-dependent pathway contributing to mitochondrial dysfunction in osteoblast-lineage cells⁷⁴.

There is prior evidence that the effects of diabetes on differentiation capacity within the mesenchymal lineage can be mediated mechanistically through various signaling pathways important to osteoblast formation and function; Wnt and Indian hedgehog (Ihh) are noted as two critical pathways in mesenchymal progenitor commitment and downstream osteoblast differentiation^75,76. In this regard, Qian et al. have shown that bone marrow-derived stromal cells from diabetic rats have reduced osteogenic gene expression in a β–catenin-dependent manner, implicating dysregulation of the canonical Wnt pathway (an essential pathway in the regulation of cell differentiation toward the osteoblastic lineage) under diabetic conditions⁴³. In the context of the Ihh pathway, Tevlin et al. demonstrate a rescue of mSSC osteogenic potential with Ihh signaling activation using a Leprdb diabetic mouse model, which results in hypoinsulemia and hyperglycemia⁷⁷. While a comprehensive framework of diabetes and its associated effects on signaling pathways remains undefined, it is clear that these conditions have the ability to modulate downstream effects on skeletal progenitor commitment at a mechanistic level. Future studies on the periosteum’s regenerative potential in the context of diabetes may prompt research into different signaling pathways that may be affected as these cells undergo a response to injury. Importantly, our use of the Col2.3GFP/SMA9 diabetic fracture model allowed for indirect visualization of osteoblastic activity within fracture callus to distinguish areas of bone-forming activity⁴⁶. Through this method, we observed that the direct contribution of periosteal progenitors to bone repair and the fracture callus is significantly impaired in diabetic mice, as mature osteoblasts arising specifically from differentiated periosteal progenitors were strikingly reduced. Combined with our other findings that diabetic mice experience a loss of specific self-renewing mesenchymal progenitor populations in the periosteal layer, we show that periosteal progenitor contribution to newly formed bone following fracture is multifactorial. This points to recent work in the field that the periosteum is an attractive target for new therapies to treat poor fracture healing in high-risk populations^31,78,79, including cell-based therapies that rely on very specific subsets of progenitors and stem cells that are proven to have multilineage potential^57,80. Diabetes has been known to directly alter the regulation of stem and progenitor cell populations in various tissues, and this progenitor cell dysfunction aberrantly affects their ability to properly participate in tissue healing and regeneration⁸¹. It has been postulated that these aberrant changes are outcomes of a hyperglycemic microenvironment high in ROS and AGE products, as well as a subsequent increase in inflammation and inflammatory cytokines, of which stem/progenitor cells are especially sensitive⁹. These environmental changes can drastically alter cellular function in terms of cell cycle/quiescence, proliferation, and differentiation; impairment and changes in these parameters can negatively affect tissue homeostasis, but are specifically detrimental to the tissue response to injury and subsequent repair. Further, diabetes-induced inflammation may also contribute to increased mesenchymal stem cell apoptosis and reduced proliferation¹³. In terms of bone formation, diabetes has been shown to directly affect critical transcription factors that are required for progenitor differentation into osteoblasts in an STZ-induced diabetic mouse model, including Runx2 and Dlx5¹⁴. Some have speculated that FOXO1 plays an integral role in the wound-healing aspect of diabetes-induced changes to progenitor potential, although the modulation of healing by diabetic factors is multifactorial^25,82. Others have shown that hyperglycemia-induced Wnt11 upregulation in marrow progenitor cells leads to increased adipogensis and impaired osteogenesis due to elevated Ang2 expression. Elevated Ang2 in diabetes may disrupt Ang-Tie2 signaling, reducing stem/progenitor cells in diabetic bone marrow^12,83.

This study holds some minor limitations in the use of the STZ-induced diabetic mouse model. As previously reported by other groups, we found the STZ-induced model to be inconsistent across experiments and groups of mice in terms of the extent of fasting blood glucose level elevation⁴⁷. However, mice that did not respond to STZ treatment were excluded from experimentation. Overall, our study suggests that impaired diabetic fracture healing can be attributed in part to dysregulation of periosteal progenitor cells. We observed a reduction of periosteal progenitor cell numbers in diabetic mice, and these cells concomitantly experienced impaired proliferation following injury and reduced osteoblast differentiation capacity both in vitro and in vivo. The results additionally indicate that diabetes may cause a loss of mesenchymal progenitors within the periosteum, as well as reduced proliferation capacity of periosteal progenitors post-fracture. Our findings mirror evidence that there must be a tight environmental control of progenitor/stem cell niches to maintain these compartments and their function^81,84, and prompt further research into targeting the regenerative capacity of the periosteum in clinical applications.

Highlights.

• Periosteum-specific effects contribute to impaired diabetic fracture healing

• Diabetes induces a long-term loss of progenitor osteogenic differentiation capacity

• Diabetes causes loss of periosteal proliferation/differentiation post-fracture

• Advanced glycation end products do not alter periosteal chondrogenic potential