Constitutively activated AMPKα1 protects against skeletal aging in mice by promoting bone‐derived IGF‐1 secretion

Abstract Senile osteoporosis is characterized by age‐related bone loss and bone microarchitecture deterioration. However, little is known to date about the mechanism that maintains bone homeostasis during aging. In this study, we identify adenosine monophosphate‐activated protein kinase alpha 1 (AMPKα1) as a critical factor regulating the senescence and lineage commitment of mesenchymal stem cells (MSCs). A phospho‐mutant mouse model shows that constitutive AMPKα1 activation prevents age‐related bone loss and promoted MSC osteogenic commitment with increased bone‐derived insulin‐like growth factor 1 (IGF‐1) secretion. Mechanistically, upregulation of IGF‐1 signalling by AMPKα1 depends on cAMP‐response element binding protein (CREB)‐mediated transcriptional regulation. Furthermore, the essential role of the AMPKα1/IGF‐1/CREB axis in promoting aged MSC osteogenic potential is confirmed using three‐dimensional (3D) culture systems. Taken together, these results can provide mechanistic insight into the protective effect of AMPKα1 against skeletal aging by promoting bone‐derived IGF‐1 secretion.


| INTRODUCTION
Skeletal aging, also called senile osteoporosis (SOP), is characterized by age-related bone loss and deterioration of bone microarchitecture. 1,2 2050, people >60 years old are predicted to number 1.5 billion, >20% of the total population worldwide. 3In this context, SOP and its complications distinctly increase the mortality of aged individuals and impose a huge financial burden on society.In contrast with the hyperactivity of bone resorption in postmenopausal osteoporosis, lack of bone formation is the main cause of SOP. 4 Previous studies have shown that SOP is accompanied by inhibited mesenchymal stem cell (MSC) function, while multiple pathological factors, including intracellular reactive oxygen species (ROS) accumulation, deoxyribonucleic acid (DNA) damage, decreased metabolic activity and diminished osteogenic differentiation, are related to its aetiology. 5To develop novel and effective treatments for SOP, it is essential to understand which regulatory proteins help to maintain bone homeostasis in skeletal aging.
Adenosine monophosphate-activated protein kinase (AMPK) is the major sensor and regulator of cellular-energy metabolism. 6AMPK, which is a serine (Ser)/threonine (Thr) kinase complex, consists of α catalytic subunits (α1, α2), β scaffolding subunits (β1, β2) and γ regulatory subunits (γ1, γ2, γ3). 6,7Phosphorylation of Thr172 (T172) in the α subunits by upstream signals is required for AMPK activation; subsequently, activated AMPK can further phosphorylate a wide range of downstream targets, such as acetyl coenzyme A (acetyl-CoA) carboxylase (ACC), glucose transporter 4 (GLUT4), and cyclic adenosine monophosphate response element binding protein (CREB).Our previous studies found that AMPKα1 is the major isoform expressed in bone tissue and that its activation significantly promotes MSC osteogenic differentiation both in vitro and in vivo. 8Although the importance of AMPKα1 in bone remodelling is known, its role in skeletal aging is still debated.
Insulin-like growth factor 1 (IGF-1) is reported to be positively correlated with bone mass, and it is regarded as an independent predictor for the risk of osteoporosis. 9Notably, autocrine IGF-1 in serum is secreted by the liver, whereas bone-derived IGF-1 deposited in bone matrix and marrow is secreted by skeletal cells, especially MSCs. 10 Bone-derived IGF-1 acts in an autocrine/paracrine manner, showing a stronger effect on bone homeostasis than autocrine IGF-1 does.Previous papers have demonstrated that matrix IGF-1 helps to develop an osteogenic microenvironment and enhances osteogenesis of recruited MSCs. 11Emerging evidence again indicates that the IGF-1 pool in bone matrix might be essential for new bone formation in the aging process. 12However, the regulatory mechanisms governing bone-derived IGF-1 expression during senescence are still not fully understood.
In this study, we found the aging of MSCs to be accompanied by obvious AMPKα1 inhibition.The results of our phospho-mutant transgenic mouse model demonstrated that constitutive AMPKα1 activation protected against age-related bone loss in vivo and delayed MSC aging in vitro with increased bone-derived IGF-1 secretion.Mechanistically, upregulation of IGF-1 signalling by AMPKα1 was found to depend on CREB-mediated transcriptional regulation.We confirmed this essential AMPKα1/IGF-1/CREB axis in both a two-dimensional (2D) cell model and a three-dimensional (3D) culture environment.

| Mice
Prkaa1-T172D fl/fl phospho-mutant knock-in mice were generated by GemPharmatech (Nanjing, China) from embryonic stem cells (ESCs) modified via homologous recombination.To express phospho-mutant AMPKα1 specifically in MSCs, we crossed Prkaa1-T172D fl/fl mice with Prx1-Cre mice to generate conditionally AMPKα1 phospho-mutant mice (AMPKα1 T172D ).All phospho-mutant mice had a C57BL/6 genetic background.All mice were bred in a specific-pathogen-free (SPF) environment.All animal experiments were approved by the Animal Ethics Committee of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China (Nos.HKDL-2016-221 and SH9H-2022-A1-1).Due to oestrogen withdrawal, C57BL/6 female mice undergo a rapid decline in bone mass after maturity.To exclude putative influences of sex steroid hormone changes on the findings, male mice were chosen for this skeletal aging study. 13

| Isolation and expansion of primary mouse MSCs
We isolated primary mouse MSCs following a previous published study. 14ese MSCs were cultured, serially passaged, and expanded using a MesenCult Expansion Kit (#05513; STEMCELL Technologies, Inc., Vancouver, BC, Canada) in vitro.Mouse MSCs in passage 2 were used in the experiments described below unless specifically emphasized otherwise.
For the CFU assay, we plated 1 million nuclear MSCs per well in 12-well plates in MSC expansion medium.After 14 days of cultivation, MSC-derived colonies were stained using a Crystal Violet Staining Kit (#C0121; Beyotime).For the CFU-ALP assay, we plated 0.5 million nuclear MSCs per well in 12-well plates in MSC expansion medium.After 7 days, mouse MSCs were induced for osteogenesis as previously described.At Day 14, we stained the cells for ALP using the ALP Colour Development Kit.For the population-doubling assay, the first passaged mouse MSCs were plated in 10-cm plates (1 Â 10 6 cells).We changed the medium every 2 days and passaged MSCs at weekly intervals.MSCs were counted at each passage using an EVOS FL Imaging System.

| AMPK activity measurement
AMPK activity was quantitatively measured using an AMPK Assay Kit (#GMS50140.1;GemMed, USA).Briefly, per manufacturer's instructions, we collected cells treated with Reagent A and centrifuged them in 15-mL tubes at 500 g for 3 min.The supernatant was then discarded.After adding Reagent B (300 μL), we transferred the mixture to a pre-chilled 1.5-mL microcentrifuge tube and incubated it at 4 C for 30 min.Next, we centrifuged samples at 16,000 g for 5 min to collect the supernatants for protein quantitation.Finally, the supernatants were mixed with kinase activity buffer, and absorbance was measured at 340 nm separately at the first and sixth minutes.

| ELISA assay
Serum concentrations of P1NP and CTX-1 were determined using a Mouse P1NP ELISA Kit (#E-EL-M0233) and a Mouse CTX-1 ELISA Kit (#E-EL-M3023; both Elabscience, Houston, TX, USA) per the instructions provided.In addition, we determined concentrations of IGF-1 in mouse bone marrow and cell supernatant using a Mouse IGF-1 ELISA Kit (#MG100; R&D Systems, Inc., Minneapolis, MN, USA) per manufacturer's instructions.

| Immunofluorescence
Cells were fixed in pre-chilled 4% paraformaldehyde for 15 min, treated with 0.5% Triton for 5 min, blocked with 10% goat serum at 4 C for 60 min, and incubated with primary antibodies at 4 C overnight.
Then, we exported the oxidative-phosphorylation parameters using the XF Mito Stress Test Report.In the extracellular-acidification rate (ECAR) assay, cells were treated with glucose (10 mM), oligomycin (10 μM), and 2-deoxy-d-glucose (2-DG; 50 mM) as protocol.Then, we exported the glycolysis parameters using the XF Glyco Stress Test Report.

| RNA sequencing
MSCs were isolated and expanded from 22-month-old AMPKα1 T172D mice and WT controls.At passage 2, we isolated total RNA from both groups for cDNA library preparation.Sequencing was performed following Illumina's (San Diego, CA, USA) standard protocol.We performed three independent replicates per group.KEGG and GO enrichment analyses and heatmap analysis were performed for all DEGs.

| Cell line construction
We knocked down AMPKα1 using shRNA with lentivirus transfection following the provided instructions.Knockdown efficiency was evaluated using Western blot (WB).Similarly, we overexpressed phosphomutant AMPKα1 (T172D) via lentivirus transfection.Overexpression efficiency was evaluated using WB.
A CREB-deficient cell line was generated using a clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated protein 9 (Cas9) lentivirus based on phospho-mutant C3H10T1/2 cells and primary MSCs.We assessed CREB-KO efficiency via WB.Lentivirus service was provided by Genomeditech (Shanghai, China).All primer sequences used in cell line construction are listed in Table S2.

| Luciferase reporter assay
We constructed an Igf1 full-length promoter reporter and its truncated mutants using the pGL4.10 vector (Promega).A luciferase reporter assay was performed using the Dual-Lumi II Luciferase Reporter Gene Assay Kit (#RG089; Beyotime) following the provided instructions.Briefly, we transfected cells in 96-well plates with indicated vectors and reporter plasmids using Lipofectamine 3000 reagent (Thermo Fisher).Forty-eight hours after transfection, cells were lysed in cell lysis buffer (#RG126; Beyotime), and luminescence signals were detected using a GloMax 20/20 Luminometer (Promega).

| ChIP assay
For our ChIP assay, we used a SimpleChIP Plus Enzymatic Chromatin IP Kit (#9005; CST) per manufacturer's instructions.Briefly, cells were crosslinked using 1% formaldehyde at RT for 20 min.Then, samples were collected, washed with PBS, lysed in lysis buffer, and sonicated to produce $400-bp DNA fragments.Next, we immunoprecipitated these fragments with the indicated antibodies and finally pulled them down using magnetic beads.DNA purification columns from the ChIP Assay Kit were used for DNA purification.We assessed DNA sequence enrichment via RT-qPCR.All primer sequences used in the ChIP assay are listed in Table S3.

| Statistics
All data are presented as mean ± standard deviation (SD), and all results are expressed as bar plots with individual data points.Statistical analysis is performed using Prism Version 9 Software (GraphPad, CA, USA).Significant comparisons between two groups are analysed using unpaired, two-tailed Student's t tests.Significant comparisons of three or more groups are analysed using one-way analysis of variance (ANOVA) with Tukey post-hoc test.For conditions which are affected by two variables, two-way ANOVA with Benjamini, Krieger and Yekutieli post-hoc test is used.p < 0.05 is considered statistically significantly different.

| MSC aging and lineage commitment are accompanied by changes in AMPKα1 activity
Physiological AMPK activation mainly involves specific phosphorylation of the AMPK catalytic α1 subunit at T172 (pAMPKα1-T172), after which the activated AMPKα1 mediates upstream direct phosphorylation of ACC at the Ser79 (S79) residue (pACC-S79; Figure S1).
So far, a range of studies have reported AMPKα1 activity to be positively correlated with the ratios of pAMPKα1-T172/AMPKα1 and pACC-S79/ACC. 16To explore the role of AMPKα1 in MSC aging, a replicative MSC aging model was first established by serial passaging in vitro, and we examined the expression profile of pAMPKα1-T172 and pACC-S79 during this process.MSC purity was determined by flow cytometry at passage 6: there was high expression of CD29 (84.6%),CD44 (99.7%) and CD106 (22.0%), but negative expression of CD45 (2.11%), Ter-119 (0.97%), CD31 (1.73%) and CD11b (2.19%) (Figure S2).Notably, we further found that the ratios of pAMPKα1-T172/AMPKα1 and pACC-S79/ACC in MSCs significantly decreased after three passages, concurrent with an increase of positive senescence-associated β-galactosidase (SA-β-gal) cells (Figure 1A).The quantification of WB results was shown in Figure S3A.Next, we isolated primary MSCs separately from young wild-type (WT) mice (3 months) and old mice (22 months) for immunoblotting to study the changes of AMPKα1 activation between young and old individuals (Figure 1B).Correspondingly, we detected decreased pAMPKα1-T172/AMPKα1 and pACC-S79/ACC ratios in old mice.The quantification of WB results was shown in Figure S3B.
These observations indicated that mouse MSC aging was accompanied with AMPKα1 inhibition.
MSC aging is marked by a switch from osteogenic to adipogenic commitment.Dysregulation of this osteo-adipogenic balance is a hallmark of MSC senescence. 17To explore changes in AMPKα1 activity in the context of adipo-osteogenic balance, we examined the ratios of pAMPKα1-T172/AMPKα1 and pACC-S79/ACC with MSC differentiation.As detected by immunoblotting, expression of both showed a gradual increase during osteogenic induction but an obvious decrease during adipogenic induction (Figure S4).In other words, our results showed that AMPKα1 activity in MSCs was correlated positively with osteogenic commitment and negatively with adipogenic commitment.
Therefore, we speculated that AMPKα1 activation might be involved in MSC aging regulation.

| Generation and characteristics of phosphomutant AMPKα1 T172D mice
Since the average mouse lifespan is >12 months, experiments on MSC senescence require prolonged periods of observation and intervention.However, our previously published studies showed that widely used AMPK agonists such as metformin and 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) failed to continuously activate AMPK for >24 h. 18Therefore, we aimed to develop a transgenic-mouse model with constitutive AMPKα1 activation to study skeletal aging in vivo.To this end, we introduced a phospho-mimetic mutation into AMPKα1 by substituting Thr172 with an aspartic acid (hereafter designated T172D) to produce Prkaa1-T172D fl/fl knock-in mice.Furthermore, to specifically express phospho-mutant AMPKα1 in MSCs, we crossed Prkaa1-T172D fl/ fl mice with Prx1-Cre mice to generate conditionally AMPKα1 phosphomutant (AMPKα1 T172D ) mice (Figure 1C).Sanger sequencing analysis confirmed the successful introduction of phospho-mimetic mutations (Figure 1D), while genotyping confirmed successful insertion of target DNA fragments into the mice's genomic DNA (Figure 1E).
As shown in gross photograph, AMPKα1 T172D mice exhibited no obvious changes in body size compared with their WT littermate controls (Figure 1F).Notably, both immunohistochemical (IHC) staining and immunoblotting revealed obviously higher expression of pACC-S79 in primary MSCs from AMPKα1 T172D mice (Figure 1G,H).The quantification of WB results was shown in Figure S3C.Consistently, more-direct evidence came from the AMPK activity assay, in which AMPK isolated from phospho-mutant MSCs showed higher kinase activity than that isolated from controls (Figure 1I).These results confirmed efficient activation of AMPKα1 in primary MSCs from AMPKα1 T172D mice.Bone marrow adiposity and diminished number of osteocytes are hallmarks of skeletal aging. 14,19Bone histomorphometry based on haematoxylin and eosin (H&E) staining further confirmed our μCT findings.

| Constitutively activated AMPKα1 protects against age-related bone loss in mice
No obvious difference was observed between AMPKα1 T172D mice and WT controls during the adult stage (3 months).However, more trabeculae and diminished adipose tissues were found within the bone marrow cavities of aged AMPKα1 T172D mice (Figure 2E).Both the adipocyte density (N.Ad/T.Ar) and the bone marrow adiposity parameter (Ad.Ar/T.Ar) were markedly decreased by constitutive AMPKα1 activation in aged mice (Figure 2F).Even though constitutive AMPKα1 activation failed to rescue the age-induced reduction of lacunae density (N.Ot.Lc/B.Ar), osteocyte death was prevented to a marked extent, as indicated by the decreased empty lacunae ratio in aged AMPKα1 T172D mice (Figure 2G).
The number of osteoblasts per unit bone surface (N.Ob/BS) and the level of serum procollagen type I intact N-terminal propeptide (P1NP; a bone formation marker) were also increased in aged AMPKα1 T172D mice (Figure 2H,I).No significant difference in osteoclastogenic parameters was noted in aged mice (Figure S5C-E).Collectively, these results indicated that constitutive AMPKα1 activation could enhance bone formation and MSC osteogenic commitment in aging process.

| Constitutively activated AMPKα1 delays primary MSC aging
To explore the role of AMPKα1 activation in modulating MSC aging, we isolated aged primary mouse MSCs (22 months) for a colony forming unit-alkaline phosphatase (CFU-ALP) assay and SA-β-gal staining.Consistent with in vivo data, the aged phospho-mutant MSCs exhibited higher self-renewal and ALP + colony formation capacity with less SA-β-gal expression compared with WT controls (Figure 3A,B).In terms of MSC replication capacity, primary MSCs from aged AMPKα1 T172D mice also displayed accelerated cell replication in vitro (Figure 3C).
Cell senescence is always accompanied by a serial change of biomarkers, such as intracellular ROS, γH2AX (a genomic-DNA damage marker), Ki-67 (a self-renewal capacity marker), and lamina-associated polypeptide 2 beta (LAP2β; a genomic-DNA stability marker).
Although primary MSC cultures from aged AMPKα1 T172D mice had relatively more Ki-67 + and LAP2β + cells than WT controls, we detected fewer γH2AX + cells and reduced intracellular ROS levels in phospho-mutant MSCs (Figure 3D,E).Senescent MSCs attempt to express high levels of aging-associated markers, such as p16 and p21, and a wide range of inflammatory cytokines.This secretome change is known as the senescence-associated secretion phenotype (SASP). 20Cs from aged AMPKα1 T172D mice exhibited significantly lower expression of p16, p21, and SASP biomarkers than WT controls (Figure 3F).Inhibited metabolic activity is another hallmark of cell senescence. 21According to Seahorse XF Analyser stress test reports, constitutively activated AMPKα1 enhanced cellular-energy metabolic activity in aged MSCs, as evidenced by its promotion of not only mitochondrial respiration but also glycolysis (Figure 3G,H; Figure S6).
These results indicated that constitutively activated AMPKα1 delayed MSC aging by promoting self-renewal capacity, mitigating genomic-DNA damage, and reprogramming metabolism.

| Constitutively activated AMPKα1 promotes bone-derived IGF-1 production and activates IGF-1 signalling pathway
To further investigate the molecular mechanism underlying the protective effect of AMPKα1 activation, we performed ribonucleic acid (RNA) sequencing to identify differentially expressed genes (DEGs) between aged AMPKα1 T172D mice and WT controls (Figure 4A).A total of 287 DEGs (fold change [FC] > 2; p < 0.05) were identified; 200 of the DEGs were upregulated, the other 87 downregulated (Figure S7).The top five most significant DEGs in sequencing were listed in Figure 4B (p-values in ascending order).Collagen type XXVIII alpha 1 (Col28a1) and membrane metalloendopeptidase (Mme) are mainly involved in extracellular matrix metabolism 22,23 ; the frizzled-related protein 2 (Sfrp2) is a Wnt signalling inhibitor and long pentraxin 3 (Ptx3) is an innate immune modulator. 24,25By contrast, IGF-1 is well-acknowledged to significantly participate in cell aging and MSC commitment. 12,26Therefore, we speculated that IGF-1 derived from MSCs might be essential to the protective effect of AMPKα1 activation.
To verify this hypothesis, we first assessed bone-derived IGF-1 expression levels in vivo and in vitro.In accordance with RNA sequencing data, aged AMPKα1 T172D mice had higher IGF-1 protein concentrations in bone marrow as detected by enzyme-linked immunosorbent assay (ELISA; Figure 4C).In addition, primary MSCs from aged AMPKα1 T172D mice also showed higher IGF-1 expression both in messenger RNA (mRNA) and protein levels than those from WT controls (Figure 4C; in primary MSCs (Figure 4E).Similarly, LV-T172D cells also showed a higher activation level of IGF-1 signalling (Figure 4F).In contrast, phosphorylation of IGF-1R, Akt and ERK was remarkably inhibited after AMPKα1 knockdown in C3H10T1/2 cells (Figure 4G).The quantification of WB results was shown in Figure S10.In summary, these results demonstrated that constitutively activated AMPKα1 promoted bone-derived IGF-1 secretion and activated IGF-1 signalling pathway both in vivo and in vitro.

| IGF-1 signalling pathway is vital to AMPKα1 modulation of aging and lineage commitment
8][29] Bone-derived IGF-1 has been acknowledged to contribute to the maintenance of skeletal homeostasis and bone mass during skeletal senescence. 12To investigate the role of bone-

| AMPKα1 promotion of bone-derived IGF-1 secretion depends on CREB-mediated transcriptional regulation
CREB is a major phosphorylation target protein of AMPK.Generally, AMPK directly phosphorylates CREB at Ser133 and drives its nuclear translocation for transcriptional regulation. 30Therefore, we speculated that CREB may be essential to bone-derived IGF-1 secretion promoted by AMPKα1 activation.To test this hypothesis, we first assessed changes in CREB phosphorylation under AMPKα1 influence.
As expected, AMPKα1 activation significantly enhanced CREB phosphorylation in both primary MSCs and LV-T172D C3H10T1/2 cells, whereas AMPKα1 knockdown obviously diminished pCREB expression (Figure 6A-C).The quantification of WB results was shown in Figure S13A.Next, we knocked out CREB protein in phospho-mutant cells using a lentivirus-mediated CRISPR-Cas9 system (CREB-KO, with mock as a control) while also overexpressing CREB protein in shAMPKα1 cells using a pcDNA3.1 expression plasmid (CREB-OE, with vehicle as a control).Knockout (KO) and overexpression efficiencies were assessed via immunoblotting (Figure S14).Notably, CREB-KO partially counteracted the increased IGF-1 expression induced by AMPKα1 activation, whereas CREB overexpression rescued the decrease in IGF-1 secretion due to AMPKα1 knockdown (Figure 6D;   In order to verify our findings from the 2D cell model again in the 3D culture environment, bioprinting technology was used to fabricate a 3D MSC culture model (Figure 7A).The characteristics of bone matrix bioink, the feasibility of bioprinting, and the evaluation workflow have been discussed in our previous studies. 15This 3D culture model offers a feasible ex vivo approach to investigating MSC osteogenic potential under various conditions.In the present study, four groups of primary

| DISCUSSION
Bone formation requires energy consumption.Energy metabolism dysregulation is linked with bone loss and deterioration. 31Aging is accompanied by a synchronous decline in energy metabolism activity and bone formation capacity, but the relationship between energy metabolism and skeletal aging is still unclear. 32In this study, we provided first-hand genetic evidence that the energy metabolism enzyme AMPK was involved in bone homeostasis during skeletal aging.Constitutive activation of AMPKα1 in Prx1-Cre + MSCs led to increased bone-derived IGF-1 secretion and protected against age-related bone loss, which we attributed to enhanced CREB-mediated transcriptional regulation.
AMPK is the main cellular energy sensor and regulator.4][35][36] However, activation effect of this mutation site has been confirmed by a wide range of studies. 37,38At the animal level, we found that AMPKα1 In addition to the skeleton, targeting AMPKα1 is also a feasible way to delay the aging processes of other organs.Rangarajan et al.
reported that metformin reversed severe lung fibrosis in an AMPKdependent manner. 41Moreover, Zhang et al. found that AMPK activation delayed ovarian aging and improved female fertility. 42gether, these studies show that AMPKα1 is a potentially important therapeutic target for combating aging.As a next step, a novel agonist with better specificity and pharmacokinetic features should be developed.
Notably, no significant differences in bone volume, bone microarchitecture, bone marrow adiposity, or number of osteogenic cells was observed between AMPKα1 T172D mice and WT controls at 3-month-old age.These results implied that congenital AMPKα1 activation may not directly affect mouse skeletal development before adulthood.Therefore, AMPKα1 activation-mediated protective effects on 22-month-old mouse skeletons could be mainly attributed to delaying skeletal aging.The study by Ruiz et al. reported that AMPK activity peaks early, is sustained during young adulthood, and declines gradually with aging. 43Considering the kinase cascade characteristic of feedback inhibition, we speculate that phospho-mimetic mutation increased the reserve capacity of AMPK cascade to function until the moment that intrinsic AMPK activity is diminished in old age.
Our hypothesis is supported by a previous study that found that AMPKα1 ablation during the embryonic period did not affect mouse endochondral ossification or bone development. 44Thus, the beneficial effect of AMPK activation is strongly age related.
IGF-1 is the most common growth factor deposited in bone throughout a mammal's lifetime.Its secretion declines with time; very low levels are detected in older adults.This phenomenon is called somatopause. 26Intriguingly, IGF-1 plays dual roles in regulating cell senescence.Although overdose of IGF-1 leads to inflammatory reaction and metabolic disorder, an appropriate amount thereof is essential for scavenging intracellular ROS and maintaining cellular mitochondrial biogenesis. 45In the skeletal system, bone-derived IGF-1 contributes to bone homeostasis in an autocrine/paracrine manner.Xian et al. reported that matrix IGF-1 enhanced osteogenic commitment and maintained bone mass. 12This result was once again confirmed in the current study.
In the past few years, many studies have investigated the downstream signalling mechanisms of bone-derived IGF-1, but few have focused on the upstream regulator of bone-derived IGF-1 expression.
CREB is a well-known transcription factor linked to cellular growth and development. 46In several cell types, CREB has been documented as involved in transcriptional regulation of crucial genes such as insulin, growth factor, or cyclin. 47,48Notably, Thomson et al. first identified CREB as a direct downstream target of AMPK. 30Activated AMPK phosphorylates CREB at Ser133 and promotes its nuclear translocation.Interestingly, previous papers have shown that CREB also participates in IGF-1 downstream signalling. 49Therefore, CREB is not only an activator of IGF-1 expression but also a downstream effector of IGF-1 signalling.This forms a positive feedback loop that augments AMPK function.
There are still some limitations and drawbacks in the present study.First, due to the limitation of construction strategy, the transgenic mice express mutated AMPKα1 from the birth.Even though no significant differences in skeletal phenotype were observed between 3-month-old transgenic mice and WT controls, it was still difficult to distinguish indirect contributions of congenital AMPKα1 activation between development and aging.Hopefully, this problem may be overcome through the application of a CreERT-mediated inducible knockout strategy, in which Cre is fused with a functional oestrogen receptor, to be activated by tamoxifen only in the aging process.Second, although data prove our findings in a cell model, 3D culture system, and transgenic mouse model, evidence from human study is still lacking.As a next step, we plan to develop specific, long-term AMPKα1 agonists for clinical translational research.
In conclusion, the findings of this study highlighted AMPKα1 as a critical regulatory protein in protection against skeletal aging.We found that constitutive AMPKα1 activation delaying murine MSC aging and enhanced their osteogenic commitment by promoting CREB-mediated IGF-1 secretion.Further studies are still needed to validate whether AMPKα1 could be a direct therapeutic target for the clinical SOP treatment.

F I G U R E 1
Generation of phospho-mutant AMPKα1 T172D mice for constitutive AMPK activation.(A) Representative SA-β-gal staining images of primary mouse mesenchymal stem cells (MSCs) with serial passaging.Immunoblotting showing pAMPKα1-T172 and pACC-S79 expression during in vitro passaging (P1-P6).(B) Immunoblotting showing pAMPKα1-T172 and pACC-S79 expression in primary MSCs from aged mice (22 months) as compared to those from young mice (3 months).(C) Schematic illustration of the strategy to generate AMPKα1 T172D mice.To constitutively activate AMPK, we introduce a phospho-mimetic mutation into AMPKα1 by substituted Thr172 with aspartic acid (T172D).After Cre recombination, the sequence between loxp sites was deleted, and phospho-mutant AMPKα1 was expressed.(D) Sanger sequencing validated successful phospho-mimetic mutation in primary MSCs.(E) Genotyping of offspring mice.P: plasmid DNA as positive control; B6: B6 genomic DNA as negative control; N: H 2 O as blank control.(F) Representative photographs of AMPKα1 T172D mice and WT littermate controls.(G) IHC staining showing pACC-S79 expression in bone tissues.(H) Immunoblotting showing pACC-S79, ACC, pAMPKα1-T172 and AMPKα1 expression in primary MSCs from AMPKα1 T172D mice and WT controls.(I) Measurement of AMPK activity in primary MSCs from AMPKα1 T172D mice and WT controls (n = 3).Results are presented as bar plots with all data points.**p < 0.01.
Skeletal aging is characterized by age-related bone loss, bone marrow adiposity and osteogenic cell diminishment.To elucidate the role of AMPKα1 activation in regulating bone accrual and age-related bone loss, we analysed bone samples from AMPKα1 T172D mice and WT controls of different ages.Micro-computed tomographic (μCT) analysis of both femoral trabecular bones and L5 vertebrae showed that bone microarchitecture parameters did not differ significantly between AMPKα1 T172D mice and WT controls in the adult stage (3 months) even though mutated AMPKα1 had been expressed from birth.This indicated that constitutive AMPKα1 activation barely affected bone accrual and remodelling during development.In contrast, bone volume per tissue volume (BV/TV), connectivity density (Conn.Dens.), and number of trabeculae (Tb.N) were appreciably increased in aged AMPKα1 T172D mice compared with WT controls (22 months); meanwhile, trabecular spacing (Tb.Sp) was decreased as indicated by μCT analysis (Figure 2A-D).No statistically significant difference in trabecular thickness (Tb.Th), bone mineral density (BMD), or cortical-microarchitecture parameters was noted (Figure S5A,B).

F I G U R E 2
Constitutively activated AMPKα1 protects against age-related bone loss in mice.(A) 3D μCT images of femoral trabecular bones from AMPKα1 T172D mice and WT controls (n = 8).(B) Quantitative μCT morphometric analysis of trabecular bones.(C) 3D μCT images of vertebral bodies (L5) from AMPKα1 T172D mice and WT controls (n = 5-6).(D) Quantitative μCT morphometric analysis of vertebral bodies.(E) Representative H&E staining images of tibia from AMPKα1 T172D mice and WT controls (n = 6-7).Left panel: a panoramic view of the proximal tibia; Right upper panel: a zoom-in view of adipocytes in bone marrow beneath the growth plate; Right middle panel: a zoom-in view of lacunae in tibia cortical bone; Right lower panel: a zoom-in view of osteoblasts lining bony trabeculae.(F) Quantitative morphometric analysis of adipocyte density (adipocyte number per total area, N.Ad/T.Ar) and BM adiposity (adipocyte area per total area, Ad.Ar/T.Ar).BM = bone marrow.(G) Quantitative morphometric analysis of lacunae density (lacunae number with/without osteocyte per bone area, N.Ot.Lc/B.Ar) and empty lacunae ratio.(H) Quantitative morphometric analysis of osteoblast number per bone surface (N.Ob/BS).(I) Serum P1NP concentrations of AMPKα1 T172D mice and WT controls (n = 8).Results are presented as bar plots with all data points.*p < 0.05; **p < 0.01.F I G U R E 3 Constitutively activated AMPKα1 delays primary mesenchymal stem cell (MSC) aging.(A) Representative images of CFU-ALP assay and SA-β-gal staining assay of primary MSCs from AMPKα1 T172D mice and WT controls at 22 months.(B) Quantification of ALP + colonies (n = 3).(C) In vitro population doublings of MSCs from AMPKα1 T172D mice and WT controls at 22 months (n = 3).(D) Fluorescent staining of reactive oxygen species (ROS), γH2AX, Ki-67 and LAP2β in MSCs from AMPKα1 T172D mice and WT controls at 22 months.(E) Quantification of intracellular ROS level based on fluorescent images.Quantitative analysis of Ki6-7 + , γH2AX + and LAP2β + frequencies in primary MSCs based on fluorescent images (n = 3).(F) Expression of senescence biomarkers ( p16, p21, Tnfa, Il-8, Ccl2 and Il-6) as detected by RT-qPCR (n = 3).(G, H) Metabolic analysis (OCR and ECAR) of primary MSCs from AMPKα1 T172D mice and WT controls at 22 months (n = 7 for Mito stress test; n = 9 for Glycolysis stress test).Quantification of basal respiration, adenosine triphosphate (ATP) production, glycolysis, and glycolytic capacity.Results are presented as bar plots with all data points.**p < 0.01.
Figure S8B).IGF-1 is a secreted peptide that functions by binding to its receptor and activating downstream signalling pathways.Therefore, we attempted to clarify whether AMPKα1 could activate IGF-1 downstream signalling pathways.As shown by immunoblotting, AMPKα1 activation stimulated the phosphorylation of IGF-1 receptor (IGF1R), protein kinase B (Akt), and extracellular signal-regulated kinase (ERK)

derived IGF- 1 in
Figure S11).Next, we reconfirmed our findings from MSCs in C3H10T1/2 cell line by treating LV-T172D cells with GSK.Consistent with the primary MSC results, IGF-1 signalling inhibition by GSK downregulated osteogenic-gene expression and attenuated osteogenic differentiation induced by T172D phospho-mutation, with only minimal effect on adipogenic differentiation (Figure5D,E).We also investigated the rescuing effect of IGF-1 activation on lineage commitment in the AMPKα1 knockdown C3H10T1/2 cell line.Exogenous IGF-1 significantly rescued the inhibited osteogenic commitment of shAMPKα1 cells, as evidenced by increased ALP expression and upregulated osteogenic-marker transcripts (Figure5F,G; FigureS12).These results therefore suggested the vital role of IGF-1 in AMPKα1 modulation of aging and lineage commitment.

Figure
Figure S15A).The quantification of WB results was shown in Figure S13B.The above results indicated that regulation of bonederived IGF-1 secretion by AMPKα1 depended on CREB.To clarify the regulatory mechanism of CREB in Igf1 gene transcription, we assessed Igf1 promoter (À2000 to +72 bp) activity using dual-luciferase reporter assays.Remarkably, exogenous CREB stimulated Igf1 promoter-driven luciferase expression in a dose-dependent manner (Figure 6E).Furthermore, consistent with the gene's expression profile, CREB-KO could cancel the increase in Igf1 promoter activity due to AMPKα1 activation, and CREB overexpression could reverse the decrease in Igf1 promoter activity due to AMPKα1 knockdown (Figure 6F,G; Figure S15B).Next, bioinformatic analysis predicted three possible binding sites of CREB on the Igf1 promoter (Figure6H).To further identify the actual CREB binding site, we used the Igf1 full-length promoter reporter and its truncated mutants in dual-luciferase reporter assays with CREB stimulation.The results confirmed that the actual binding site was about À830 bp (Figure6I).Chromatin immunoprecipitation (ChIP) assays also confirmed the binding of CREB to this region on the Igf1 promoter.Notably, constitutively activated AMPKα1 significantly enhanced CREB enrichment on the promoter, but AMPKα1 knockdown partially abrogated this binding (Figure6J; FigureS15C).In summary, these results demonstrated that AMPKα1 activation delayed MSC aging and enhanced osteogenic commitment by promoting bone-derived IGF-1 secretion, and this process depended on CREB-mediated transcriptional regulation (Figure6K).

MSCs were included: ( 1 )
aged WT MSCs; (2) aged AMPKα1 T172D MSCs; (3) aged AMPKα1 T172D MSCs treated with GSK; (4) aged AMPKα1 T172D MSCs with CREB-KO by CRISPR-Cas9.Of note, all groups of 3D culture models showed superior biocompatibility, as evidenced by >80% of cells remaining alive after 21 days' induction (Figure S16).Consistent with results from the 2D culture system, constitutive AMPKα1 activation significantly attenuated ROS accumulation in 3D culture model compared with WT controls, but both GSK treatment and CREB-KO increased ROS production to varying degrees (Figure7B,C).ALP is a biomarker of osteogenic potential.We observed relatively more ALP + MSCs in AMPKα1 T172D 3D culture models than in WT controls, whereas both GSK treatment and CREB-KO diminished ALP expression (Figure7B,D).Furthermore, we directly evaluated mineral deposition within 3D culture systems via ARS staining 21 days after induction.AMPKα1 T172D 3D culture models showed the most enriched calcium deposition, while both GSK treatment and CREB-KO inhibited this deposition process (Figure7B).At the transcriptional level, osteogenic-marker expression (Runx2, Alp, Osx and Col1) in 3D models exhibited a similar trend: either CREB or IGF-1 inactivation could impede the protective effect of constitutively activated AMPKα1 (Figure7E).Collectively, these results once again confirmed the essential role of the AMPKα1/CREB/IGF-1 axis in delaying MSC aging and promoting osteogenic commitment.
AMPKα1 promotion of bone-derived IGF-1 secretion depends on CREB-mediated transcriptional regulation.(A) Protein expression of CREB in primary mesenchymal stem cells (MSCs) from AMPKα1 T172D mice and WT controls.(B, C) Protein expression of CREB in C3H10T1/2 cells with AMPKα1 constitutive activation or AMPKα1 knockdown.(D) Effects of CREB-KO or overexpression on IGF-1 protein expression in MSCs or C3H10T1/2 cells.(E) CREB-enhanced Igf1 promoter transcriptional activity as assessed by dual-luciferase reporter assays in C3H10T1/2 cells (n = 3).(F, G) Igf1 promoter transcriptional activity in indicated cell lines was assessed using dual-luciferase reporter assays (n = 3).1: LV-GFP; 2: LV-T172D + Mock; 3: LV-T172D + CREB-KO; 4: NC; 5: shAMPKα1-3 + Vehicle; 6: shAMPKα1-3 + CREB-OE.(H) Predicted binding sites of CREB on the Igf1 promoter.(I) Dual-luciferase reporter assays were performed on C3H10T1/2 cells with the Igf1 full-length promoter reporter and its truncated mutants (n = 3).(J) ChIP analysis of CREB binding to the Igf1 promoter (À837 to À826 bp) and exon regions (n = 3).(K) Schematic illustration of the proposed molecular mechanism by which constitutively activated AMPKα1 protects against MSC aging by promoting autocrine IGF-1 secretion.CREB-mediated transcriptional regulation is essential for this process.Results are presented as bar plots with all data points.*p < 0.05; **p < 0.01.because of the prolonged time span required for aging research and the lack of long-term AMPK activation strategy in vivo, few studies have focused on the protective effect of AMPK against skeletal aging.Herein, we established an Prx1 + MSC-specific AMPK constitutive-activation mouse model by introducing a phosphomimetic mutation (T172D) into the catalytic subunit AMPKα1.TheF I G U R E 7 AMPKα1/CREB/IGF-1 axis is essential to enhance osteogenic commitment of aged mesenchymal stem cells in 3D culture model.(A) Schematic illustration of study design for 3D bioprinting and for osteogenic induction and evaluation.(B) Representative images of intracellular reactive oxygen species (ROS) fluorescent staining, ALP immunofluorescent staining, and ARS staining of 3D culture models 21 days after printing.(C, D) Quantification of intracellular ROS level and ALP expression (n = 3).(E) Expression of osteogenic genes (Runx2, Alp, Osx and Col1) in 3D culture systems as detected by RT-qPCR 21 days after printing (n = 3).1: WT; 2: AMPKα1 T172D ; 3: AMPKα1 T172D + GSK; 4: AMPKα1 T172D &CREB-KO.Results are presented as bar plots with all data points.**p < 0.01.
T172Dmice exhibited less age-related bone loss, attenuated bone marrow adiposity, and increased numbers of viable osteocyte and periosteal osteoblast at the aged stage.At the cellular level, our data showed that T172D phospho-mutation in aged MSCs enhanced cell self- Yang and Kai Yuan contributed equally to this work.Conceptualization: Yiqi Yang, Yugang Wang, Qiming Fan, Hanjun Li and Tingting Tang; Study conduct: Yiqi Yang, Kai Yuan, Yihao Liu, Qishan Wang, Yixuan Lin, Shengbing Yang, Kai Huang, Tianyou Kan, Yuxin Zhang and Mingming Xu; Data analysis: Yiqi Yang, Kai Yuan, Qishan Wang and Zhifeng Yu; Writing: Yiqi Yang and Kai Yuan; Supervision: Yugang Wang, Hanjun Li and Tingting Tang.All authors reviewed and approved the final manuscript.