Identification of Gli1 as a progenitor cell marker for meniscus development and injury repair

Meniscal tears are associated with a high risk of osteoarthritis but currently have no disease-modifying therapies. Using Gli1-CreER tdTomato mice, we found that Gli1+ cells contribute to the development of meniscus horns from 2 weeks of age. In adult mice, Gli1+ cells resided at the superficial layer of meniscus and expressed known mesenchymal progenitor markers. In culture, meniscal Gli1+ cells possessed high progenitor activities under the control of Hh signal. Meniscus injury at the anterior horn induced a quick expansion of Gli1+ cells. Normally, the tissue healed slowly, leading to cartilage degeneration. Ablation of Gli1+ cells further hindered this repair process. Strikingly, intra-articular injection of Gli1+ meniscal cells or an Hh activator right after injury accelerated the bridging of the interrupted ends and attenuated signs of osteoarthritis. Taken together, our work identified a novel progenitor population in meniscus and proposes a new treatment for repairing injured meniscus and preventing osteoarthritis.


52
Meniscal tears, with all the morbidity and disability they cause, are among the most 53 common injuries of the knee affecting both the young and aged; and the procedures to 54 address them are among the most commonly performed surgeries in the orthopedics field. 55 Beyond the short-term pain, disability, time from desired activities including work, 56 meniscal injuries are important early events in the initiation and later propagation of 57 osteoarthritis (OA) [1]. From a clinical therapeutic point of view, surgical treatments, 58 including the maximally preserving partial meniscectomy, while improving immediate 59 symptoms, do not delay the natural history progression of OA or may actually accelerate 60 it. As the adult meniscus is predominantly avascular, true biologic healing with surgical 61 repair remains a viable treatment for only a small portion of individuals typically with 62 tears contained within the red vascular zone [2]. For the majorities of injuries, a 63 restorative biologic therapy does not currently exist in practice. 64 Mesenchymal progenitors play a critical role in tissue regeneration. Therefore, 65 identifying and characterizing residential mesenchymal progenitors in meniscus are 66 important for developing novel and effective strategies to treat meniscus injury. Using 67 enzymatic digestion and clonal expansion methods, previous studies have demonstrated 68 that human and rabbit meniscus contain mesenchymal progenitors with multi-69 differentiation abilities [3][4][5][6]. Interestingly, the superficial layer of meniscus was 70 proposed to harbor the progenitors. By collecting cells growing out of mouse meniscus 71 explant, Gamer et al. showed that these cells exhibit stem cell-like characteristic and are 72 located in the superficial zone in vivo [7]. During injury, it has been observed that 73 progenitors on the meniscus surface migrate from vascularized red zone to non-74 5 vascularized white zone for repair [8]. While these cells in culture express several 75 common mesenchymal progenitor markers, such as CD44, Sca1, and CD90, their in vivo 76 properties and regulatory signaling pathways are not known [6,8].

77
Hedgehog (Hh) signaling is essential for embryonic development and tissue 78 homeostasis. It is one of few fundamental pathways that maintain adult stem and 79 progenitor cells in various organs, such as brain, skin, bladder, teeth, and others [9]. 80 Following injury, Hh signaling can trigger stem and other resident cells to participate in 81 repair, and therefore, Hh upregulation is viewed not only as a natural response to injury 82 but also as a way to stimulate tissue repair by activating stem cells. Gli1, an integral 83 effector protein of Hh pathway, was recently recognized as a marker for bone marrow, 84 periosteal, and periarticular mesenchymal progenitors [10][11][12], suggesting that Hh 85 signaling is also functional in skeleton for maintaining tissue-specific stem and 86 progenitors.

87
In this study, we constructed a Hh reporter mouse (Gli1-CreER Tomato, Gli1ER/Td), 88 and found that    Animals. All animal work performed in this report was approved by the Institutional 99 Animal Care and Use Committee (IACUC) at the University of Pennsylvania.  CreER Rosa-tdTomato (Gli1ER/Td) mice were generated by breeding Gli1-CreER mice 101 (Jackson Laboratory, Bar Harbor, ME USA) with Rosa-tdTomato mice (Jackson 102 Laboratory). They were further bred with Rosa-DTA mice (Jackson Laboratory) to 103 produce Gli1-CreER Rosa-tdTomato Rosa-DTA (Gli1ER/Td/DTA). In accordance with the 104 standards for animal housing, mice were group housed at 23-25℃ with a 12 h light/dark 105 cycle and allowed free access to water and standard laboratory pellets. All animal work 106 performed in this report was approved by the Institutional Animal Care and Use 107 Committee (IACUC) at the University of Pennsylvania.

108
To induce Td expression or ablate Gli1-labeled cells, mice (Gli1ER/Td or 109 Gli1ER/Td/DTA) received vehicle or Tamoxifen (Tam) injections at 50 mg/kg at P4 and 110 P5 or 75 mg/kg for 5 days at ages older than 1 week. For EdU labeling of proliferation 111 experiment, mice were injected with daily 1.6 mg/kg EdU (Invitrogen, Carlsbad, USA, 112 A10044) for 4 days before harvesting. For EdU labeling of slow-cycling experiment, 113 mice were injected with daily 5 mg/kg EdU for 4 days at P3-6. 114 Male mice at 3 months of age were subjected to meniscus injury at right knees. To 115 perform the surgery, the joint capsule was opened immediately after anesthesia and the 116 anteriomedial horn of meniscus were cut into two parts using microsurgical scissors. The 117 joint capsule and the subcutaneous layer were then closed with suture followed by skin 118 closure with tissue adhesive. In sham surgery, meniscus will be visualized but not 119 transected. For cell treatment, cells digested from meniscus of Gli1ER/Td mice were 120 7 sorted by FACS to collect Td + and Tdcells. 10,000 cells were injected into the knee joint 121 space of sibling WT mice immediately after meniscus surgery. For activator treatment, 2 122 μl purmorphamine (100 μM) were injected into the knee joint space of WT or Gli1ER/Td 123 mice immediately after surgery. Mice were euthanized at indicated time points for 124 histology analysis.

125
The knee joint pain after meniscus injury was evaluated in mice at 1 month after 126 surgery using von Frey filaments as described previously [13]. applied to the plantar surface of the hind paw until the fibers bowed, and then held for 3 132 seconds. The threshold force required to elicit withdrawal of the paw (median 50% 133 withdrawal) was determined five times on each hind paw with sequential measurements 134 separated by at least 5 min.

135
To induce OA, male mice at 3 months of age were subjected to DMM surgery at right 136 knees and sham surgery at left knees as described previously [14]. Briefly, in DMM 137 surgery, the joint capsule was opened immediately after anesthesia and the medial 138 meniscotibial ligament was cut to destabilize the meniscus without damaging other 139 tissues. In sham surgery, the joint capsule was opened in the same fashion but without 140 any further damage.

141
Human and Mini-pig Meniscus Samples. The meniscus samples were prepared from 142 the de-identified specimens obtained at the total arthroplasty of the knee joints and used 143 8 for histological and immunohistochemical examination. The meniscus degeneration 144 severity was evaluated according to the meniscus surface including lamellar layer, 145 cellularity, collagen organization and safranin O/fast green staining [15]. 6-month-old 146 male Yucatan minipigs were utilized (Sinclair Bioresources) to provide meniscus tissues.

147
Anterior horn meniscus tissue was obtained for following histological analysis.   numbers. To study cell migration, primary meniscus cells were seeded in 12-well plates.

185
When reaching confluency, the cell layer was scratched by a 1000 μL pipette tip and then 186 cultured in FBS free growth medium. Wound closure was monitored by imaging at 0 and 187 48 hr later. To study cell proliferation, primary meniscus cells were seeded at 50,000 188 cells/well in 12-well plates and cell numbers were counted 2, 4, and 6 days later.  Table S1.

216
The expression patterns of Gli1 + cells and their descendants in mouse meniscus. 217 We performed lineage tracing with Gli1ER/Td mice at various ages to identify Gli1 + cells 218 and their descendants at 6 weeks later in meniscus (Fig. S1A). Joints were cut at either 219 sagittal or coronal planes to visualize different parts of meniscus (Fig. S1B). In line with 220 our previous report [11], at 1 week of age, Gli1 + cells were only observed in the 221 periarticular layer of articular cartilage, but not in the meniscus and other joint tissues 222 . Long term tracing also did not detect any Td signal in the meniscus, 223 confirming that neonatal meniscus does not harbor Gli1 + cells (Fig. 1Ad). At 2 weeks of 224 age, most cells in the anterior horn of the meniscus, both medially and laterally, were Td + 225 . Six weeks of tracing confirmed that the entire anterior horn, but not the 226 posterior horn, is labeled by Td (Fig. 1Ah). At 4 weeks of age, Gli1 + cells were 227 concentrated in the superficial layer of the anterior horn; 6 weeks later, most cells in both 228 superficial and central portions of the anterior horn were labeled by Td .

229
Within the posterior horn, very few cells in the center of meniscus were initially labeled 230 but then gave rise to the majority of internal cells 6 weeks later. Quantification along the 231 length of the meniscus over the time indicated that 1-8 weeks of age represents the rapid 232 growing phase for the meniscus (Fig. S2). Taken together, our data suggested that Gli1 + 233 cells represent progenitors for meniscus cells of the horn regions at adolescence stage.

234
Starting from 8 weeks of age, Gli1 + cells were exclusively restricted to the superficial 235 12 layer of the anterior horn throughout tracing . The labeling pattern in the 236 posterior horn was slightly different that Td + cells first appear in the center and then 237 expand to the entire tissue at 6 weeks later . At 12 weeks of age, Gli1 + cells 238 remained restricted to the superficial layer of both anterior and posterior horns throughout 239 tracing ( Fig.1Aq-t). At any given age, Td signal was not detected in the center of the body 240 of either the medial or lateral meniscus regardless of cutting planes (Fig. 1B).

256
The existence of Gli1-labeled cells on the meniscus surface was confirmed by Gli1 257 immunostaining (Fig. S4A). Furthermore, analysis of porcine meniscus revealed a similar   Meniscal tear is a common injury in joints. To mimic this injury, we surgically cut the 301 anteromedial horn of the meniscus into two parts in 3-month-old mice, resulting in 302 disconnected synovial and ligamental ends of the meniscus (Fig. S7A). Gli1ER/Td mice 303 received Tam right before surgery (Fig. S7B). At 1-2 weeks post-surgery, the two ends of 304 15 the meniscus retracted toward the synovium and ligament, respectively (Fig. 4Aa-c). This 305 was accompanied by massive synovial hyperplasia that wrapped around the ends of the 306 meniscus and likely stabilized them. At 4 weeks, the synovium returned to relatively 307 normal thickness and the two cut ends of the meniscus were aligned but not connected 308 (Fig. 4Ad). Over time, the connection between the two ends gradually moved toward re-309 establishment but never reached the normal level even after 3 months post surgery (Fig. 310 4Ae,f). The meniscus repair scores summarized this trend (Fig. 4B), suggesting that 311 meniscus heals slowly in this injury model.

312
Fluorescence imaging was used to analyze the contribution of Gli1 + cells and their 313 descendants during this process. Strikingly, starting from 2 weeks post-injury, Td + cells 314 appeared at the synovial ends and ligamental ends of injured meniscus (Fig. 4Ai). Their 315 number peaked around 4 weeks, and gradually declined thereafter (Fig. 4Aj-l). Total cell 316 density and the percentage of Gli1 + cells at both ends were significantly increased after 317 injury, particularly at the ligamental end (Fig. 4C). EdU incorporation experiment 318 confirmed that many Gli1 + cells and their progenies are proliferative at 2 weeks post 319 surgery (Fig. 4D). In old mice (52 weeks of age), this expansion of Td + cells after injury 320 was remarkably attenuated and the end-to-end reconnection was much less with a lower 321 repair score than young adult mice 4 weeks later (Fig. S8A, B), indicating that aging 322 diminishes the repair ability of meniscus.

323
To further understand the role of Gli1 + cell expansion in meniscus repair, we generated 324 Gli1-CreER Tomato DTA (Gli1ER/Td/DTA) mice for a cell ablation experiment. These 325 mice at 3 months of age received Tam injections followed by meniscus injury (Fig. 4E).

326
One day after the Tam injections, Td + cells in meniscus were drastically decreased by 327 16 54.5%, as shown by both sagittal and coronal views of meniscus horns (Fig. 4E). Three 328 months later, while two meniscus ends loosely reconnected in vehicle-treated mice, those 329 in Tam-treated mice were still well separated, leading to a significant reduction of repair 330 score (Fig. 4F). Fluorescence imaging confirmed no more expansion of Td + cells in Tam

342
Since Gli1 + cells and their descendants were greatly expanded at the early phase of 343 meniscus injury repair, we hypothesized that activation of Hh/Gli1 pathway could 344 stimulate the repair process. We adopted two approaches to test this hypothesis. One was 345 to inject Gli1 + cells freshly isolated from Gli1ER/Td meniscus into injured knees (Fig. 346 5A). Strikingly, a single injection of cells right after injury resulted in a reconnection of 347 the synovial and ligamental ends of injured meniscus at 4 weeks, leading to a repair score 348 of 4.8 (Fig. 5A, B). At the same time, these two meniscus ends were well separated in 349 both vehicle and Gli1meniscus cell-treated groups, with a repair score of only 1.8 and 350 1.9, respectively. Polarizing images clearly showed a disconnection of collagen fibers in 351 mice that received either vehicle or Gli1cells. However, in Gli1 + cell-treated mice, 352 collagen fibers crossed the broken ends of the meniscus, suggesting that the repair does 353 occur at the structural level. Fluorescence imaging revealed that injected Gli1 + cells 354 expand and contribute to the newly formed connection at the injury site (Fig. 5C).

355
In another approach, we injected purmorphamine to the knee joint right after injury 356 (Fig. 5D). Four weeks later, the injured ends of purmorphamine-treated meniscus were 357 reconnected based on gross morphology, safranin O/fast green staining, and imaging of 358 collagen fibers (Fig. 5D), leading to a repair score of 4.9 (Fig. 5E). There were more Td + 359 meniscus cells in purmorphamine-treated joints than vehicle-treated joints at 1 week after 360 injury (Fig. 5F). These data clearly indicated a therapeutic effect of activating Hh 361 signaling.

362
Meniscus repair by enhancing Hh/Gli1 pathway delays OA progression. and reduction in uncalcified cartilage thickness (Fig. 6A, B). Meanwhile, the calcified 368 cartilage layer was not eroded (Fig. 6B), suggesting a moderate OA with a Mankin Score 369 of 6.9 (Fig. 6C). Strikingly, injections of either Gli1 + cells or purmorphamine greatly 370 reduced cartilage degeneration by retaining proteoglycan content, cartilage surface 371 smoothness, and the structure of uncalcified cartilage. These treatments led to a reduction 372 in Mankin Score by 35% and 53%, respectively. Von Frey assay is commonly used in OA 373 18 study as a pain outcome by evaluating mechanical allodynia. Using this assay, we 374 observed that OA knees displayed significantly decreased paw withdraw threshold 375 compared to sham knees. However, this OA-related pain was mostly attenuated in Gli1 + 376 cell-or purmorphamine-treated knees (Fig. 6D).