Cellular senescence by loss of Men1 in osteoblasts is critical for age-related osteoporosis

Recent evidence suggests an association between age-related osteoporosis and cellular senescence in the bone; however, the speci�c bone cells that play a critical role in age-related osteoporosis and the mechanism remain unknown. Results revealed that age-related osteoporosis is characterized by the loss of osteoblast Men1. Osteoblast-specic inducible knockout of Men1 caused structural changes in the mice bones, matching the phenotypes in patients with age-related osteoporosis. Histomorphometrically, Men1-knockout mice femurs decreased osteoblastic activity and increased osteoclastic activity, hallmarks of age-related osteoporosis. Loss of Men1 induces cellular senescence via activation of mTORC1 pathway, rescued by metformin treatment. In bone morphogenetic protein-indued bone model, loss of Men1 leads to accumulation of senescent cells and osteoporotic bone formation, which are ameliorated by metformin. Our results indicate that cellular senescence in osteoblasts plays a critical role in age-related osteoporosis and that osteoblast-specic inducible Men1-knockout mice offer a promising model for developing therapeutics for age-related osteoporosis.


INTRODUCTION
Osteoporosis is a systemic skeletal disease characterized by low bone mass and the deterioration of bone microarchitecture, which increase the occurrence of fragility fractures 1,2 .Among the various factors affecting osteoporosis, aging, menopause, and decreased activity, age-related osteoporosis is becoming important because an increase in aging population, which started in high-income countries is now occurring even in low-and middle-income countries 3,4 .
Accumulating evidence highlights a close relationship between age-related diseases and cellular senescence [5][6][7] , which was discovered by Hay ick et al. 8 and is de ned by an irreversible cell growth arrest regulated through the p16 ink4a /RB and p53/p21 CIP1 pathways 9 .Senescent cells acquire a proin ammatory phenotype, known as the senescence-associated secretory phenotype (SASP), which leads to the disruption of tissue and organ functions and is considered to be a major cause of various chronic diseases 10 .Therefore, treatment strategies for chronic diseases targeting the removal of senescent cells (senolytics) or inhibition of the SASP (senomorphics) have recently come into the research spotlight 11 .
The accumulation of senescent cells with age has also been con rmed to occur in the bones, suggesting cellular senescence as a key regulator of age-related osteoporosis 12 ; thus, it is rational to develop treatment strategies for osteoporosis that target senescent cells 13 .Indeed, inhibition of the negative effects of senescent cells in mice attenuated the rate of bone loss with age in mice 14 .However, the precise mechanism by which bone cells become senescent during the natural aging process remains largely unknown.Understanding this mechanism will enable the development of more precise therapies for age-related osteoporosis.
Men1 gene is originally known to be a tumor suppressor for multiple endocrine neoplasia type 1 (MEN1) syndrome 15 .We previously found Men1 de ciency in T cells induces cellular senescence-associated immunode ciency 16,17 and focused on a possible role of Men1 gene in osteoblast senescence.Our current results indicate loss of Men1 is a critical driver for osteoblast senescence, and osteoblast-speci c Men1 knockout mice could be the rst animal model for age-related osteoporosis.

RESULTS
Men1 RNA levels are signi cantly reduced in the osteoporotic bones of aged mice Recent studies suggest that cellular senescence is a key regulator of age-related osteoporosis 12 .We previously identi ed the tumor suppressor gene Men1 as a regulator of cellular senescence regulator in T cells 16, 17 .These ndings prompted us to examine whether Men1 also played a critical role in osteoporosis during aging.To test this hypothesis, we initially compared femurs of 2-month-old (young mice) and 24-month-old mice (aged mice) using microcomputed tomography (µCT).Evaluation of the femurs of young and aged mice via µCT showed a decrease in the bone volume (BV)/total volume (TV) and thickness of the cancellous and cortical bone and an increase in the bone marrow area in 24-monthold mice compared with those in 2-month-old mice (Fig. 1a).The bone phenotypes in 24-month-old mice resembled those of older humans with osteoporosis.Moreover, the mRNA levels of Men1 were reduced in 24-month-old mice (Fig. 1b), accompanied by increased mRNA levels of cellular senescence genes (p16, p21, p53, Il1α, Il8, MMP3) compared with those of 2-month-old mice (Fig. S1).These results suggest that the loss of Men1 is closely involved in the pathology of age-related osteoporosis.
Men1 deletion in osteoblasts induces osteoporotic changes on cortical bone even in young mice.
To further identify the role of Men1 on age-related osteoporosis, we generated Men1 ox/ ox ; Col1a1cre/ERT2 mice that can delete osteoblast Men1 by tamoxifen (TAM) treatment.Femur bone of 9-week-old Men1 ox/ ox ; Col1a1-Cre/ERT2 mice treated with TAM (four days/week starting at four, six, and eight weeks of age; Fig. 1c) were evaluated by µCT.The µCT images clearly demonstrated thinning of the cortical bone in the distal femur of TAM treated Men1 ox/ ox ; Col1a1-cre/ERT2 (Men1 KO) mice compared to Men1 ox/ ox (Control) mice (Fig. 1d, 1e).Microstructural analyses showed that the cortical bone in the Men1 KO mice was thinner and more porotic, with a wider bone marrow area than that in Control mice (Fig. 1f).These results suggest that Men1 deletion in osteoblasts predominantly affects the osteoporotic phenotype in the cortical bone.Since the cortical bone microstructure has a great in uence on bone strength 18 , we investigated the bone strength of Men1 KO mice by a mechanical testing (Fig. 1g).The ultimate load and displacement at fracture were comparable between Control and MenEN1 KO mice (Fig. 1h); however, the femoral stiffness was decreased in Men1 KO mice (Fig. 1h).Thus, femurs from Men1 KO mice resembled the fragile bones of older humans in terms of both bone structure and mechanical strength.

Men1 deletion induces osteoblast senescence, attenuates osteoblastic activity, and increases osteoclastic activity in vivo
After identifying the structural phenotype, we next investigated the detailed actions of Men1 de ciency histologically.Cortical thinning due to Men1 deletion was observed, similar to the µCT ndings (Fig. 2a).Furthermore, Men1 KO mice exhibited an increase in the number of p16-positive osteoblasts (Fig. 2a), which was consistent with the area where Cre/loxP recombination was con rmed (Fig. S2a).These results suggest that Men1 de ciency in vivo leads to osteoblast senescence.To elucidate the effect of Men1 deletion in osteoblasts on bone formation, bone histomorphometry analysis was performed using the femurs of Control and Men1 KO mice.The spaces between the double-stained lines were narrow (Fig. 2b), which matched the decrease in osteogenic ability (as assessed by the mineral apposition rate [MAR]   and bone formation rate per bone surface ratio [BFR/BS]) in Men1 KO mice (Fig. 2c).Men1 KO mice exhibited decreased bone area and thickness, re ecting their low osteogenic ability (Fig. 2d), along with a decrease in the number of osteoblasts and the area of the osteoblast surface (Fig. 2e).The osteoblastrelated bone formation indices were all generally decreased in Men1 KO mice, matching the general ndings in aged humans.Under normal conditions, the number of osteoclasts generally parallels the number of osteoblasts, regulated by cross-talk mediated through the receptor activator of NF-κB ligand (RANKL) and RANK 19 .Interestingly, histomorphometry analysis of bone resorption parameters showed an increase in the number of osteoclasts and the area of the osteoclast surface in Men1 KO mice (Fig. 2f), although the number of osteoblasts decreased.Thus, Men1 de ciency widened the gap between bone formation and resorption, a characteristic of age-related osteoporosis 1 .
Men1 deletion in osteoblasts promotes replicative stress-induced cellular senescence through presumably mTORC1 The results summarized above demonstrate that Men1 deletion in osteoblasts resulted in induced p16 expression and phenotypes characteristic of age-related osteoporosis, especially in the cortical bone.
However, the mechanistic link between Men1 deletion in osteoblasts and age-related osteoporosis remained unclear.Since Men1 was previously shown to act as a cellular senescence regulator in T cells 16,17 , we next evaluated whether Men1 de ciency directly causes cellular senescence using in vitro adenoviral-mediated Men1 KO in primary osteoblasts from Men1 ox/ ox mice.Osteoblasts were infected with an adenovirus expressing green uorescent protein (Ad-GFP; Control) or Cre (Ad-Cre-GFP; Men1 KO) and cellular senescence was induced by replicative stress, as described in the Materials and Methods section.After multiple passages, the positivity of senescence-associated β-galactosidase (SA-β Gal) staining was signi cantly increased in Men1 KO osteoblasts compared with that in the control osteoblasts (Fig. 3a, b).To further verify the characteristics of each osteoblast type, the mRNA levels of cellular senescence-associated genes and Men1 were investigated.Cre/loxP recombination was con rmed by Men1 reduction (Fig. 3c).The induction of cellular senescence induced by replicative stress was con rmed by higher mRNA levels of p16 (Fig. 3d) in both Control and Men1 KO osteoblasts.Even under the same condition of replicative stress, the mRNA levels of SASP genes (Il1α, IL6, IL8, MMP3) were increased in Men1 KO compared to those in Control osteoblasts (Fig. 3e, 3f, 3g, 3h).Since SASP levels are elevated during the late stages of cellular senescence 9 , these results suggest that the loss of Men1 in osteoblasts accelerates cellular senescence induced by replicative stress.To elucidate the mechanism underlying replicative stress-induced senescence in osteoblasts with Men1 deletion, we investigated the potential involvement of the mTORC1 pathway, one of the major signaling pathways for cellular senescence 20 .Phosphorylation of S6K and 4EBP1, which are downstream effectors of mTORC1, was enhanced by Men1 deletion, and metformin treatment suppressed activation of the mTORC1 pathway caused by Men1 de ciency (Fig. 3i).These data suggest that replicative stress-induced senescence in osteoblasts with Men1 deletion is closely associated with mTORC1 pathway activation, which can be restored by metformin treatment.
Cellular senescence and a decrease in Men1 are closely involved in bone formation during natural aging Femoral analysis of Men1 KO mice represented signs of cellular senescence, and osteoblast senescence presumably via mTORC1 induced by Men1 de ciency was con rmed in vitro.To further clarify the relevance of cellular senescence in vivo due to Men1 de ciency, we established a model of accelerated cellular senescence in bone tissue.Bone morphogenic protein (BMP) is known to activate mTORC1 and regulate cellular senescence [21][22][23] .A previous report showed that the transplantation of a BMP-containing collagen sponge allowed us to observe the bone formation capacity 24 .We hypothesized that the effects of osteoblast senescence would be enhanced by conditions of BMP-induced bone formation.Experiments were conducted using BMP-containing collagen sponges (Fig. 4a).BMP-induced ectopic bones in 24-month-old mice were signi cantly thinner and smaller than those in 2-month-old mice (Fig. 4b, 4c), as observed during human aging 25 .We further investigated cellular senescence in the ectopic bone using histological evaluation and reverse transcription-quantitative polymerase chain reaction (RT-qPCR).Immunostaining revealed that the number of p16-positive cells was signi cantly increased on the surface of the ectopic bone from aged mice (Fig. 4d, 4e).The areas positively stained for β-Gal were also increased in the aged mice (Fig. 4f).The mRNA levels of p16 and SASP factors increased in the ectopic bones of aged mice (Fig. 4g).Additionally, similar to the results from the vertebral bones, the mRNA levels of Men1 decreased in the ectopic bones of aged mice (Fig. 4h).Thus, the BMP-induced ectopic bone model suggests that cellular senescence and a decrease in Men1 mRNA levels are closely associated with low bone formation activity in aging individuals.
Metformin suppresses cellular senescence in ectopic bones from Men1 KO mice Since metformin suppressed the mTORC1 pathway in Men1 KO osteoblasts (Fig. 3i), we sought to determine whether metformin could also reverse the cellular senescence due to the loss of Men1.Using the BMP-induced ectopic bone model (Fig. 5a), we found that Men1 deletion in osteoblasts resulted in an increase in the number of p16-positive cells and the β-Gal-positive areas (Fig. 5b), which were markedly reduced after the administration of metformin (Fig. 5c, 5d).RT-qPCR on BMP-induced ectopic bones con rmed that the mRNA levels of Men1 were decreased in Men1 KO mice (Fig. 5e).Consistent with the histological results, the mRNA levels of the SASP genes p16, IL1α, IL8, and MMP3 were signi cantly increased in the ectopic bones from Men1 KO mice, and this increase was suppressed following metformin administration (Fig. 5f, 5g, 5h, 5i).These results indicate that Men1 de ciency promotes cellular senescence, which could be reversed by metformin treatment.
Metformin partially restores the bone morphometry of BMP-induced ectopic bone in Men1 KO mice Both in vitro and in vivo, Men1 deletion induced cellular senescence in osteoblasts, leading to reduced osteogenesis.Since cellular senescence in Men1 KO mice was restored by metformin treatment, we examined whether metformin could also reverse the reduction in osteogenesis in Men1 KO mice.The µCT images and three-dimensional reconstruction showed bone thinning and a decrease in bone volume, BV/TV in the ectopic bone of the Men1 KO mice (Fig. 6a, 6c).Notably, metformin administration signi cantly increased the volume, but not the thickness, of the ectopic bones from Men1 KO mice (Fig. 6b, 6c).Taken together, our current results suggest that metformin partially restores osteogenesis reduced by the loss of Men1, as a hallmark of osteoblast aging.

DISCUSSION
We found that osteoblast-speci c Men1 KO resulted in signi cant porosity, thinning of the cortical bone, widening of the bone marrow cavity, and decreased mechanical strength.In addition, the bones of Men1de cient mice exhibited a decrease in osteoblast number and an increase in osteoclast number, resulting in a low bone volume and bone thinning.In vitro assays revealed that Men1 deletion activated cellular senescence and mTORC1 in osteoblasts.Metformin suppressed the activation of mTORC1 due to Men1 de ciency and restored the low bone volume due to senescent osteoblasts in Men1-de cient mice.
The present study thus revealed that loss of Men1 is a hallmark of cellular senescence-associated bone aging.As senescent cells accumulate in aged individuals, bone aging is clinically associated with osteoporosis 14,26 .Osteoporosis is caused by aging, menopause, and lack of activity in daily life 1,2 .
Several mouse models have been established for osteoporosis caused by menopause 27 or inactivity in daily life 28 , whereas age-related osteoporosis has generally been investigated using naturally or genetically aged mice 29 .Since aged mice have also undergone menopause and exhibit reduced activity, aged mice do not represent a suitable model for speci cally examining age-related osteoporosis.We identi ed the loss of Men1 as a single factor in the pathogenesis of age-related osteoporosis in terms of cellular senescence.Therefore, this animal model can be used to obtain further insights into bone aging by cellular senescence through the measurement of the degree of age-related osteoporosis and for the discovery of new drugs.
This study characterized osteoblast senescence both in vitro and in vivo.Senescent osteoblasts exhibited an increase in the mRNA levels of SASP factors, including IL1α, IL6, IL8, and MMP3, which were associated with an increase in osteoclast numbers.Consistent with our results, senescent cells around the bone tissue secrete SASP and lead to an increase in osteoclast numbers 14,30,31 .Senescent osteoblasts can result in the fragility of cortical bones in the trunk.During bone formation, osteoblast senescence can lead to a decrease in bone volume and thickness of new bones, and metformin can restore the decreased bone volume.Osteoporosis is characterized by the deterioration of cortical and trabecular microstructures, bone fragility, and decreased osteogenic capacity 1,32 .Therefore, our model of osteoblast senescence-associated bone aging in mice recapitulates some of these characteristics in the elderly human population.A previous report demonstrated that the removal of all senescent cells, including senescent osteoblasts, could alleviate osteoporotic phenotypes such as those pertaining to bone thickness, number of osteoclasts, number of osteoblasts, and mineral apposition rate 14 , which were the same factors found to be exacerbated by osteoblast senescence-associated bone aging in the present study.Although various senolytic drugs are expected to be developed in the future, the pathophysiology and phenotypes revealed in this study can serve as biomarkers for evaluating the e cacy of drugs targeting cellular senescence-associated bone aging.
We focused on Men1 as a regulator of osteoblast senescence.Permanent Men1 de ciency in bone from embryonic period reduces bone mass 33,34 .However, the relationship between Men1 and cellular senescence in adults could not be evaluated because cellular senescence is deeply involved during the embryonic period, thus developmental Men1 effect cannot be eliminated 35 .In this study, by using inducible conditional knockout mice, Men1 de ciency in osteoblasts caused cellular senescence, which was associated with mTORC1 activation and induced bone aging.Consistent with our current results on the role of Men1 in osteoblasts, several reports have shown that Men1 de ciency in other cell types causes age-related phenotypes such as impaired lymphocyte function and the development of dementia 16,17,36 .Loss of Men1 in T cells promotes cellular senescence through mTORC1 activation 17 .MEN1, a tumor suppressor gene, causes MEN1, an autosomal dominant disorder and hereditary tumor syndrome 15 .Patients with MEN1 develop osteoporosis early in life, regardless of tumor development 37 and tend to develop benign tumors, in contrast to patients with other hereditary tumor syndromes, who tend to develop malignant tumors 15 .Indeed, permanent Men1 de ciency in early osteoblasts develop ossifying broma, a benign bone tumor in mice 38 .Senescent cells are identi ed at high frequencies in benign tumors, whereas they are rarely detected in malignant tumors 39,40 .These clinical features of MEN1 support our results on cellular senescence in Men1-de cient mice in vitro and in vivo.In a process known as oncogene-induced senescence (OIS), cells become senescent in response to the activation of oncogenes, which promotes cancer development 41 .The concept of OIS, induced by oncogenes, resembles that of cellular senescence by Men1 gene de ciency.Although the overexpression of oncogenes immediately induces cellular senescence in vitro [42][43][44] and forms malignant tumors in vivo 45,46 , the loss of Men1 appears to gradually accelerate cellular senescence, presumably through mTORC1 activation.Thus, an aging animal model with loss of Men1 is consistent with the slow progression of natural aging.Our results suggest that Men1 is deeply involved in cellular senescenceassociated aging and provide insights into new disease concepts.
This study had several limitations.First, the mechanism by which Men1 de ciency leads to the boneaging phenotype has not yet been clearly elucidated.Men1 is reported to be epigenetically involved in cell cycle progression, apoptosis, and the DNA damage response, which are important factors for cellular senescence 47 .Moreover, the BMP and transforming growth factor-beta (TGF-β) pathways, acting as major contributors to bone metabolism, have been associated with Men1 47 .However, the effect of Men1 on these pathways depends on the cell type and does not necessarily facilitate bone formation 48,49 .

RT-qPCR
Total RNA was isolated using the TRIzol reagent (Thermo Fisher Scienti c, Waltham, MA, USA).Bone preprocessing for in vivo assays was performed as previously described 12 .Brie y, after the bones were minced into small pieces, they were subjected to two sequential 30-min collagenase digestions (Worthington Biochemical, Worthington, OH, USA).The total RNA was extracted from the remaining chips.Abundant osteoblasts in the remaining bone chips were con rmed according to GFP-positive signals in the preprocessed bone chips of Col1a1-GFP mice (Figure S2d).Complementary DNA was synthesized by ReverTra Ace (TOYOBO, Osaka, Japan) and then used as a template in qPCR with FAST SYBR Green Master Mix (Thermo Fisher Scienti c).Target mRNA levels were normalized to reference gene expression.The median threshold cycle was compared with that of the control sample, and the fold difference between the reference and target genes was calculated.The internal controls were Gapdh for in vitro assays (osteoblasts) and Hprt for in vivo assays (osteoblast lineage-enriched bone cells).The primer sequences used in RT-qPCR are listed in Table S1.

Cre/loxP recombination by TAM treatment
The dose and schedule of TAM administration (10 mg/kg/day for 4 days) and age of mice (4 weeks old) were determined according to a previous report, in which the authors con rmed the minimum TAM dosage for reliable Cre/loxP recombination with little effect on the bone formation ratio 52 .The TAM solution (20 mg/mL) was prepared by dissolving TAM powder (Sigma-Aldrich, St. Louis, MO, USA) in ethanol and diluted 20-fold with corn oil (C8267, Sigma-Aldrich).For femur analysis, Men1 ox/ ox (control) and Men1 ox/ ox ; Col1a1-cre/ERT2 (Men1 KO) mice were treated with 10 mg/kg TAM per day for four days at four, six, and eight weeks of age.The mice were euthanized and analyzed at nine weeks of age.Men1 deletion by Cre/loxP recombination was con rmed by PCR of calvaria bone (Fig. S2c).For BMP-induced ectopic bone analysis, control and Men1 KO mice underwent BMP collagen sponge implantation at four weeks of age.Osteoblasts in the BMP-induced ectopic bone were visible 10 days after implantation 24,53 .To delete the osteoblast Men1 gene at the appropriate time, mice were treated with TAM (10 mg/kg/day for four days) at 8 to 11 and 15 to 18 days after the surgery and were euthanized 21 days after the surgery.Men1 deletion by Cre/loxP recombination was con rmed by PCR of the calvarial bone (Fig. S4).Men1 deletion by Cre/loxP recombination was con rmed by PCR of calvaria bone (Fig. S2c).The e ciency of Cre/loxP recombination was con rmed in Col1a1-cre/ERT2; Col1a1-GFP; Ai9 mice (Fig. S2b).

Biomechanical testing
Three-point bending tests were performed using an MZ-500 system (Marutoh, Mie, Japan) to investigate the biomechanical strength of the femur bone.The bones were placed on a pedestal, and force was applied to the center of the diaphysis at a rate of 2 mm/s until failure (ultimate load) (Fig. 1g).The ultimate load (N), displacement at fracture (mm) and stiffness (N/mm) were determined from a loaddisplacement diagram 54 .

Histomorphometric analysis
Double labeling by subcutaneous injection of tetracycline (20 mg/kg) and calcein (10 mg/kg) was performed ve and two days before euthanasia, respectively.The resected samples were xed in 70% ethanol, stained with Villanueva, and embedded in methacrylic acid (Wako Pure Chemical Industries, Kanagawa, Japan).The following histomorphometric parameters were measured: mineral apposition rate (MAR), bone formation rate per bone surface (BFR/BS), bone volume/tissue volume (BV/TV), trabecular thickness (Tb.Th), number of osteoblasts per bone surface (N.Ob/BS), osteoblast surface per bone surface (Ob.S/BS), number of osteoclasts per bone surface (N.Oc/BS), and osteoclast surface per bone surface (Oc.S/BS).

Replicative senescence by long-term culture
One week after adenovirus infection, cells were passaged in a 10-cm dish at 5 × 10 5 cells per 75 cm 2 (passage 2) and then passaged to a 10-cm dish at the same density every other week.Passage-3 and passage-6 cells were used for the experiments as the low replicative stress and high replicative stress group, respectively, since a previous report suggested that cellular senescence starts after passage 6 under normal conditions 56 .SA-β-Gal staining and RT-qPCR were performed using passage-3 and passage-6 cells.Western blotting was performed on passage-3 cells with or without 10 mM metformin treatment for 48 h.
SA β-Gal staining SA β-Gal staining was performed using a Senescence β-galactosidase Staining Kit (#9860, Cell Signaling Technology, Danvers, MA, USA).Green-stained cells were counted in 10 random elds under a microscope (BZ-X800, Keyence, Osaka, Japan) with a 10× magni cation objective lens and calculated as the percentage of positive cells.To avoid non-speci c staining due to cell con uence 57 , the assay was performed using subcon uent cells.

BMP-induced ectopic bone model
A collagen sponge (Colla Cote, Zimmet Dental, CA, USA) was cut into cylindrical shape (diameter of 5 mm and height of 2 mm) for establishing the BMP-induced ectopic bone model.Recombinant human BMP-2 (1.5 µg) was soaked into the cylindrical collagen sponge.The sponge was then freeze-dried (BMP pellets).Under anesthesia, the BMP pellet was implanted underneath the dorsal fascia of mice.
Metformin was administered orally in the drinking water at a concentration of 300 µg/mL starting two weeks after the surgery.The dose of metformin was set based on the effective dose for extending lifespan [58][59][60] .

Histological analysis
The excised samples were xed in 10% neutral-buffered formalin, decalci ed by ethylenediaminetetraacetic acid, embedded in para n wax, and cut at 3 µm thickness.Hematoxylin and eosin staining was performed according to standard protocols.Para n-embedded sections were depara nized and dehydrated for immunostaining.The antigens were activated in 10 mM citrate buffer at 95°C for 10 min.After quenching endogenous peroxidase activity with methanol containing 3% H 2 O 2 for 10 min, the sections were blocked with Blocking One (Nacalai Tesque) for 1 h at room temperature.
Sections were then incubated with a primary antibody overnight at 4°C, followed by incubation with a horseradish peroxidase-labeled secondary antibody for 1 h.Finally, the labeled sections were stained with Histo ne Simple Stain Mouse MAX PO (Nichirei Bioscience, Tokyo, Japan) and counterstained with hematoxylin.Rabbit monoclonal anti-CDKN2A/p16INK4a (ab211542, 1:250; Abcam, Cambridge, UK) was

Figures
Figures

Figure 5 Characteristics
Figure 5