Graphical abstract

   graphic file with name fx1.jpg
   [71]Open in a new tab

Highlights

     * •
       Obesity increases CD47 expression while exercise downregulates CD47
       expression
     * •
       CD47-blocking antibody ameliorates obesity and enhances exercise
       capacity
     * •
       CD47-HSP90α pathway inhibits AMPK activation in skeletal muscle
     __________________________________________________________________

   Su et al. suggest that obesity upregulates CD47 expression while
   exercise downregulates CD47 expression. Targeting CD47 with a blocking
   antibody confers multiple metabolic benefits through AMPK signaling in
   skeletal muscles. These findings reveal a potential therapeutic
   strategy for obesity-related metabolic disorders.

Introduction

   Obesity results from chronic energy surplus, and prevention or
   treatment of obesity in humans is elusive. AMP-activated protein kinase
   (AMPK), an important regulator of energy homeostasis,[72]^1 is a
   crucial target for treating metabolic diseases such as obesity.[73]^2 A
   large number of AMPK activators have been developed, like metformin and
   canagliflozin, which are clinically approved for the treatment of
   metabolic diseases.[74]^3^,[75]^4 However, most AMPK agonists target
   multiple tissues to activate AMPK. In some cells or tissues, aberrant
   activation of AMPK causes detrimental effects on health.[76]^5 For
   example, prolonged AMPK activation enhances tumor cell viability or
   induces cardiac hypertrophy.[77]^6 Skeletal muscle accounts for
   approximately 40% of total body weight, and AMPK activation in skeletal
   muscle is intrinsically linked to systemic metabolism by increasing
   glucose uptake and fatty acid (FA) oxidation.[78]^7^,[79]^8^,[80]^9
   Consequently, the development of AMPK activators targeting skeletal
   muscle without cardiac hypertrophy is vital for the treatment of
   metabolic diseases.

   CD47 has been widely recognized as a “do-not-eat-me” signal that blocks
   the phagocytic process of macrophages,[81]^10^,[82]^11 and thus
   CD47-blocking antibodies are used in clinical trials for the treatment
   of cancer.[83]^12 A recent study shows that the single-nucleotide
   polymorphism (SNP) in the CD47 locus is associated with variations in
   human body shape and composition.[84]^13 It was further reported that
   systemic genetic deletion of CD47 promotes mitochondrial biogenesis and
   improves physical performance in mice.[85]^14 However, little is known
   about the therapeutic role of CD47-blocking antibody in metabolic
   diseases. Therefore, we evaluated the protective role of CD47-blocking
   antibody as activator of muscle AMPK against high-fat diet
   (HFD)-induced metabolic disorders. Furthermore, previous studies have
   shown that AMPK is a client protein of HSP90α and that reduced levels
   of HSP90α phosphorylation induce AMPK activation.[86]^15 Therefore, we
   hypothesized that the CD47-blocking antibody might promote AMPK
   activation by inhibiting HSP90α phosphorylation.

   Here, we discovered that obesity upregulates CD47 expression in the
   skeletal muscle, whereas exercise downregulates its expression in
   skeletal muscle, indicating CD47 as a potential target for ameliorating
   metabolic diseases. We found that CD47-blocking antibody promotes AMPK
   activation in skeletal muscle and prevents metabolic disorders under
   HFD, conferring metabolic benefits in vivo, including reduced body
   weight gain, improved body composition, enhanced basal metabolism and
   exercise capacity, alleviated glucose tolerance, and improved serum
   lipids and mitochondrial function. Using whole-body or muscle-specific
   CD47-deficient mice, we similarly confirmed that CD47 loss of function
   (LOF) exhibited protective effects against HFD-induced metabolic
   disorders. Our results suggest that CD47-blocking antibody promotes the
   activation of AMPK by reducing HSP90α phosphorylation in skeletal
   muscle. Thus, our results reveal the essential role of CD47 signaling
   in skeletal muscle AMPK activation and explore the potential value of
   CD47-blocking antibody to treat metabolic diseases.

Results

Obesity upregulates CD47 expression in skeletal muscle

   To explore the potential correlation between obesity and CD47
   expression, we analyzed the public data from human skeletal muscle
   (GEO: [87]GSE19420
   and GEO: [88]GSE81965); the expression level of CD47 was positively
   correlated with body mass index (BMI) and homeostasis model assessment
   insulin resistance (HOMA-IR) ([89]Figures 1A and 1B). Meanwhile, we
   constructed a mouse model of obesity induced by HFD. Results showed
   that mRNA and protein expression levels of CD47 were increased in the
   skeletal muscle tissue of HFD-induced obese mice compared to that in
   normal diet (ND) controls ([90]Figures 1C and 1D).

Figure 1.

   [91]Figure 1
   [92]Open in a new tab

   Administration of CD47-blocking antibody confers metabolic protective
   effects in mice under HFD condition

   (A) Correlation of CD47 expression with BMI in individuals with type 2
   diabetes (T2D).

   (B) Correlation between CD47 expression and HOMA-IR in obese
   individuals.

   (C and D) CD47 mRNA and protein expression levels in skeletal muscle of
   control and obese mice (n = 4). Expression levels quantified, GAPDH as
   control.

   (E and F) CD47 mRNA and protein expression in sedentary and
   exercise-trained mice after 8 weeks (n = 4). Expression quantified,
   GAPDH as control.

   (G) Schematic of antibody injection protocol: CD47-blocking antibody
   (CD47 Ab) or control IgG antibody (Cont) every two days under HFD.

   (H) Bodyweight changes in mice injected with CD47 Ab or Cont on HFD (n
   = 6 for Cont and n = 7 for CD47 Ab).

   (I and J) Body composition measured by NMR in mice (n = 6 for Cont and
   n = 8 for CD47 Ab).

   (K) Intraperitoneal glucose tolerance test (IPGTT) and area under the
   curve (AUC) after 5 weeks HFD administration (n = 6).

   (L–N) Oxygen consumption, carbon dioxide emissions, and energy
   expenditure measured using metabolic cage (n = 5 for Cont and n = 6 for
   CD47 Ab).

   (O and P) Spontaneous food intake and physical activity (n = 5 for Cont
   and n = 6 for CD47 Ab).

   (Q and R) Running distance and duration measured on a treadmill (n =
   6).

   Data are presented as means ± SEM and analyzed by two-tailed Student’s
   t test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; ns, not significant).

   Furthermore, we analyzed public transcriptomic data (GEO: [93]GSE165630
   ) and proteomic data[94]^16 (PRIDE: [95]PXD044445) from the human
   skeletal muscle of individuals engaged in regular exercise versus
   sedentary controls. The results indicated that both mRNA and protein
   levels of CD47 were downregulated in the exercise group ([96]Figures
   S1A and S1B), suggesting that regular exercise may regulate CD47
   expression. To further validate this result, we established an exercise
   model of mice. Both transcriptional and translational expression levels
   of CD47 were decreased in the skeletal muscle of the mice with regular
   exercise compared to those in the sedentary controls ([97]Figures 1E
   and 1F). Altogether, these results indicate that low expression of CD47
   is associated with the beneficial metabolic effects.

Administration of CD47-blocking antibody exerts metabolic benefits in mice
against HFD

   To determine whether the functional blockade of CD47 could alleviate
   HFD-induced metabolic disorders, we subcutaneously injected either a
   CD47-blocking antibody or an immunoglobulin G (IgG) control antibody
   into HFD-fed mice ([98]Figure 1G), as described by Kojima et al.[99]^17
   Firstly, the mice injected with CD47-blocking antibody displayed a
   significant decrease in body weight gain on HFD ([100]Figure 1H).
   Moreover, nuclear magnetic resonance (NMR) results showed that
   CD47-blocking antibody significantly ameliorated HFD-induced
   alterations in body composition, as evidenced by a reduction in fat
   mass and an increase in lean mass ([101]Figures 1I, 1J, [102]S1C, and
   S1D). Secondly, mice injected with CD47-blocking antibody were
   protected from HFD-induced glucose intolerance ([103]Figure 1K). To
   further elucidate the possibilities of body composition improvements,
   we examined the energy balance in mice injected with CD47-blocking
   antibody or IgG control. An increase in whole-body basal metabolism
   ([104]Figures 1L–1N) independent of daily food consumption ([105]Figure
   1O) and physical activity ([106]Figure 1P) was observed in mice
   injected with CD47-blocking antibody. Thirdly, to test whether CD47
   function blocking could improve the homeostasis of serum lipids, we
   measured the serum concentrations of free fatty acids (NEFAs),
   triglycerides (TGs), total cholesterol (T-CHO), and low-density
   lipoprotein (LDL-C) in the mice injected with CD47-blocking antibody or
   IgG control. Results showed that injection with CD47-blocking antibody
   improved the serum lipid profiles of mice on HFD ([107]Figures
   S1E–S1H). In parallel with the reduction in serum lipids, we also
   observed a decrease in lipid accumulation in both skeletal muscle and
   white adipose tissue (WAT) ([108]Figures S1I–S1L). Most importantly,
   mice injected with the CD47-blocking antibody exhibited a significant
   increase in both maximum running distance ([109]Figure 1Q) and duration
   ([110]Figure 1R) compared to the IgG control group, indicating an
   improvement in running capacity in the CD47-blocking antibody-treated
   mice. These findings suggest that CD47-blocking antibody treatment in
   mice on HFD confers beneficial metabolic effects, including reduced
   body weight, increased skeletal muscle mass, enhanced basal metabolism,
   improved glucose homeostasis and serum lipid profiles, reduced lipid
   droplet accumulation in skeletal muscle, and enhanced exercise
   capacity.

CD47-blocking antibody improves mitochondrial function by promoting AMPK
activation in skeletal muscle

   To elucidate the metabolic benefits of CD47 blockade, we focused on
   AMPK, a crucial energy sensor that governs metabolic
   homeostasis.[111]^18 Activation of AMPK in metabolic tissues is known
   to mitigate obesity, non-alcoholic fatty liver disease (NAFLD), and
   type 2 diabetes.[112]^7^,[113]^19^,[114]^20^,[115]^21 We hypothesized
   that CD47 blockade modulates AMPK activation, particularly at
   Thr172.[116]^22 Indeed, a CD47-blocking antibody activated AMPK in
   skeletal muscle but not in WAT, brown adipose tissue (BAT), liver,
   hypothalamus, or kidney ([117]Figure 2A). To explore the potential
   mechanisms, AMPKγ3 was initially considered due to its selective
   expression in skeletal muscle,[118]^23 which aligns with our findings
   ([119]Figure S2A). To further investigate the possible role of the
   AMPKγ3 subunit in CD47-blocking antibody-induced AMPK activation, we
   knocked down the γ3 subunit in differentiated myotubes and treated them
   with a CD47-blocking antibody. As shown in [120]Figures S2B and S2C,
   the loss of γ3 expression abolished AMPK activation induced by the
   CD47-blocking antibody, indicating that the γ3 subunit plays an
   important role in CD47-blocking antibody-mediated AMPK activation in
   skeletal muscle. The AMPK activity in the skeletal muscle mainly
   originates from the AMPKγ3 complex and the AMPKγ1
   complex.[121]^7^,[122]^24 To further evaluate the relative
   contributions of γ3 and γ1 subunits to AMPK activation, we measured the
   activities of AMPK complexes containing γ3 or γ1 in the muscle of mice
   treated with a CD47-blocking antibody in vivo, following the protocol
   outlined in [123]Figure S2D.[124]^23 The results showed that the AMPKγ3
   complex was activated to a greater extent than the AMPKγ1 complex
   ([125]Figure S2E).

Figure 2.

   [126]Figure 2
   [127]Open in a new tab

   CD47-blocking antibody promotes skeletal muscle AMPK activation and
   improves mitochondrial function

   (A) AMPK phosphorylation in skeletal muscle, WAT, BAT, liver,
   hypothalamus (Hypo), and kidney (n = 6).

   (B) ACC phosphorylation in skeletal muscle after antibody injection.
   GAPDH as control.

   (C) PGC-1α protein levels in skeletal muscle post-injection. GAPDH as
   control.

   (D) Western blot analysis of AMPK, ACC phosphorylation, and PGC-1α in
   skeletal muscle of mice injected with CD47 Ab (20 mg/kg), Cont (20
   mg/kg), and AMPK inhibitor (Compound C, 10 mg/kg) (n = 3).

   (E and F) AMPK phosphorylation and cytochrome c (cyto c) protein
   expression in myotubes after antibody treatment. GAPDH as control.

   (G) Seahorse mitochondrial stress test in myotubes treated with CD47
   Ab.

   (H–K) AMPK, ACC phosphorylation, and PGC-1α expression in myotubes with
   Cd47 knockdown or overexpression.

   (L) Seahorse mitochondrial stress test in myotubes with Cd47 knockdown.

   Data are presented as means ± SEM and analyzed by two-tailed Student’s
   t test (A and G) and two-way repeated measures (RM) ANOVA (D and L). ∗p
   < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; ns, not significant.

   AMPK activation in skeletal muscle promotes FA oxidation[128]^24 and
   mitochondrial respiration.[129]^25 Accordingly, we examined
   acetyl-coenzyme A carboxylase (ACC), a key FA-oxidation regulator that,
   when phosphorylated by AMPK, increases fat oxidation.[130]^26 In mice
   treated with a CD47-blocking antibody, ACC phosphorylation (pSer79) was
   elevated in skeletal muscle ([131]Figure 2B). Because peroxisome
   proliferator-activated receptor-gamma co-activator-1 alpha (PGC-1α)
   enhances mitochondrial respiration,[132]^27 we next observed that CD47
   blockade also increased PGC-1α expression ([133]Figure 2C). In vivo
   administration of the AMPK inhibitor Compound C suppressed the
   antibody-induced phosphorylation of AMPK and ACC, as well as the
   upregulation of PGC-1α ([134]Figure 2D), confirming an AMPK-dependent
   mechanism. Consistently, CD47 blockade in vitro activated AMPK in
   myotubes and raised cytochrome c (cyto c) levels, improving maximal
   respiration and ATP production ([135]Figures 2E–2G). These findings
   suggest that, under HFD conditions, CD47 blockade augments skeletal
   muscle mitochondrial energetics.

   To further examine the role of CD47 in regulating AMPK activation, we
   knocked down or overexpressed Cd47 in myotubes ([136]Figures S3A and
   S3B). Cd47 deficiency markedly increased AMPK phosphorylation and
   downstream ACC (pSer79), whereas Cd47 overexpression suppressed both
   ([137]Figures 2H and 2I), confirming that CD47 negatively regulates the
   AMPK/ACC axis. Likewise, PGC-1α was upregulated in Cd47 knockdown
   myotubes but reduced upon overexpression ([138]Figures 2J and 2K),
   highlighting the role of CD47 in regulating mitochondrial function.
   Seahorse analysis showed that Cd47 knockdown enhanced maximal
   respiration and ATP production ([139]Figure 2L). Beyond AMPK, the
   mitochondrial energetics regulator SIRT1[140]^28^,[141]^29 was also
   elevated in Cd47-deficient myotubes ([142]Figure S3C) and in skeletal
   muscle of CD47^−/− mice ([143]Figure S3D). Because both AMPK and SIRT1
   converge on PGC-1α to bolster mitochondrial function,[144]^30 these
   results underscore coordinated regulation by CD47 LOF in skeletal
   muscle.

CD47 LOF in skeletal muscle promotes AMPK activation

   Next, we generated CD47 whole-body knockout (CD47^−/−) by the
   CRISPR/Cas9 system ([145]Figures 3A, [146]S4A, and S4B). Firstly, the
   CD47^−/− mice displayed a significant decrease in body weight gain
   ([147]Figures 3B and [148]S4C). Further studies revealed a significant
   improvement in body composition in CD47^−/− mice, as evidenced by a
   decrease in fat mass and an increase in lean mass ([149]Figures S4D and
   S4E). Secondly, an improvement in glucose tolerance was observed in
   CD47^−/− mice ([150]Figure 3C). In addition, CD47 LOF significantly
   improved serum lipid profiles ([151]Figures S4F–S4J). Along with the
   reduction in serum lipids, we observed a decrease in lipid accumulation
   in both skeletal muscle and WAT ([152]Figures S4K and S4L). Finally,
   the maximal running distance ([153]Figure S4M) and duration
   ([154]Figure S4N) were significantly greater in CD47^−/− mice compared
   to wild-type (WT) mice.

Figure 3.

   [155]Figure 3
   [156]Open in a new tab

   CD47 LOF in skeletal muscle promotes AMPK activation

   (A) Schematic of CD47-knockout (CD47^−/−) mouse model.

   (B) Growth curves of WT and CD47^−/− mice (n = 6).

   (C) IPGTT and AUC in WT and CD47^−/− mice (n = 6).

   (D and E) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway
   enrichment analysis of differentially expressed genes in skeletal
   muscle of CD47^−/− and exercise mice.

   (F) AMPK and ACC phosphorylation in skeletal muscle of WT and CD47^−/−
   mice.

   (G) PGC-1α, cyto c, and oxidase (COXIV) expression levels in skeletal
   muscle of WT and CD47^−/− mice.

   (H and I) ATPase and CS activities in skeletal muscle of WT and
   CD47^−/− mice (n = 7).

   (J) Schematic of MCK-CD47 mouse model.

   (K) Growth curves of MCK-CD47 mice on HFD (n = 10 for Con and n = 6 for
   MCK-CD47).

   (L) IPGTT and AUC in MCK-CD47 mice after 3 months of HFD (n = 10 for
   Con and n = 6 for MCK-CD47).

   (M) AMPK phosphorylation and PGC-1α expression in skeletal muscle of
   MCK-CD47 mice.

   (N) CS activity in skeletal muscle of MCK-CD47 mice (n = 11 for Con and
   n = 7 for MCK-CD47).

   (O) Schematic of HSA-CD47 mouse model.

   (P and Q) Growth curves and IPGTT of HSA-CD47 mice on HFD (n = 8 for
   Con and n = 6 for HSA-CD47).

   (R and S) Running distance and duration of HSA-CD47 mice on motor
   treadmill (n = 9 for Con and n = 7 for HSA-CD47).

   (T) CS activity in skeletal muscle of HSA-CD47 mice (n = 9 for Con and
   n = 7 for HSA-CD47).

   (U) AMPK phosphorylation and PGC-1α expression in skeletal muscle of
   HSA-CD47 mice.

   (V) Western blot analysis of CD47 and AMPK phosphorylation in skeletal
   muscle of mice injected with AAV expressing GFP or CD47.

   Data are presented as means ± SEM and analyzed by two-tailed Student’s
   t test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; ns, not significant).

   To further investigate how CD47 LOF exerts beneficial metabolic effects
   in vivo, RNA sequencing (RNA-seq) was performed in the skeletal muscle
   of the CD47^−/− mice, WT mice with regular exercise (Ex), and sedentary
   WT controls (Con). Notably, genes associated with the AMPK pathway were
   highly enriched in the skeletal muscle of both CD47^−/− mice and Ex
   mice compared to the Con mice ([157]Figures 3D and 3E). Consistent with
   findings from Cd47-deficient myotubes, the phosphorylation levels of
   AMPK (pThr172) and ACC (pSer79) were increased in the skeletal muscle
   of CD47^−/− mice compared to littermate WT controls ([158]Figure 3F).
   In addition, PGC-1α ([159]Figure 3G) level was also increased in the
   skeletal muscle of CD47^−/− mice. Meanwhile, mitochondrial enzymes
   including cyto c, oxidase (COXIV) protein level ([160]Figure 3G), and
   adenosine triphosphatase (ATPase) ([161]Figure 3H), as well as citrate
   synthase (CS) ([162]Figure 3I), were increased in skeletal muscle of
   CD47^−/− mice. These results are consistent with those in CD47-blocking
   antibody-injected mice.

   Mitochondria metabolism and function are tightly regulated by protein
   acetylation.[163]^31 Analysis of the mitochondrial acetylome revealed
   that over 60% of mitochondrial proteins contain acetylation sites, with
   most involved in energy metabolism, particularly FA metabolism, the
   tricarboxylic acid (TCA) cycle, the electron transport chain, and
   oxidative phosphorylation (OXPHOS).[164]^32 SIRT3 is one of the
   sirtuins localized in mitochondria that deacetylates key mitochondrial
   proteins essential for metabolism and oxidative
   homeostasis.[165]^33^,[166]^34^,[167]^35 Our findings indicated that
   Cd47 gene deletion was associated with a significant increase in SIRT3
   expression levels ([168]Figure S5A). Given the critical role of SIRT3
   in the regulation of mitochondrial metabolism,[169]^31 we subsequently
   isolated mitochondria from the skeletal muscle of CD47^−/− mice and
   investigated the changes in mitochondrial protein acetylation levels.
   The results showed that CD47 deletion led to a significant reduction in
   protein acetylation levels of mitochondria ([170]Figure S5B). To
   further assess protein acetylation, we performed a proteomic analysis
   that identified 1,530 acetylation sites on 410 skeletal muscle
   proteins. Of these, 53 acetylation sites were significantly altered in
   the skeletal muscle tissues of CD47^−/− mice ([171]Table S1). These
   altered sites include several deacetylated mitochondrial proteins, such
   as Iba57, ldh3a, and Slc25a12, with ldh3a playing a key role in the TCA
   pathway ([172]Figure S5C).

   To investigate the muscle-specific role of CD47 in regulating AMPK
   activation, we generated muscle-specific CD47-knockout (MCK-CD47) mice
   by crossing Cd47^flox/flox mice with MCK-Cre mice ([173]Figure 3J).
   After 20 weeks on an HFD, MCK-CD47 mice exhibited significantly reduced
   weight gain compared with littermate controls ([174]Figures 3K and
   [175]S6A), along with lower random blood glucose ([176]Figure S6B),
   improved glucose tolerance ([177]Figure 3L) and insulin sensitivity
   ([178]Figure S6C), and reduced lipid accumulation in WAT and skeletal
   muscle ([179]Figures S6D and S6E).

   Consistent with results from CD47-blocking antibody-treated mice,
   MCK-CD47 mice displayed enhanced AMPK activation in skeletal muscle
   ([180]Figure 3M). Building on the improved mitochondrial energetics
   observed under CD47 blockade, we found that MCK-CD47 mice showed
   elevated PGC-1α expression ([181]Figure 3M) and CS activity
   ([182]Figure 3N). To extend these findings, we generated human
   α-skeletal actin (HSA)-CD47 mice (Cd47^flox/flox, Hsa-MerCreMer) with
   CD47 selectively deleted in mature muscle cells ([183]Figure 3O).
   Similarly, HFD-fed HSA-CD47 mice exhibited significantly lower body
   weight ([184]Figure 3P), improved glucose tolerance ([185]Figure 3Q),
   and greater running distance and duration ([186]Figures 3R and 3S).
   Their skeletal muscle displayed increased CS activity ([187]Figure 3T),
   AMPK activation, and PGC-1α expression ([188]Figure 3U). Finally, to
   confirm that CD47 indeed regulates AMPK activity, we administered an
   adeno-associated virus (AAV) overexpressing CD47 intramuscularly, which
   significantly suppressed AMPK activation ([189]Figure 3V). Together,
   these data demonstrate that CD47 directly modulates AMPK activation in
   skeletal muscle.

CD47-blocking antibody promotes AMPK activation by reducing phosphorylation
of HSP90α

   To gain insight into the mechanism of AMPK activation by CD47 LOF, we
   constructed a cell line stably expressing FLAG-tagged CD47 in myotubes
   and then performed co-immunoprecipitation (coIP) and mass spectrometry
   (MS) analysis to identify proteins that may interact with CD47
   ([190]Figure 4A). Several proteins related to metabolism were
   identified by MS analyses, including HSP90α, GNB2, GNB4, SLC25A4, UGCG,
   HSPA8, PKM, and UBC ([191]Figures S7A and S7B). The direct interaction
   between CD47 and HSP90α was further verified by coIP assay in 293T
   cells ([192]Figure S7C) and differentiated myotubes ([193]Figure 4B).
   The results indicated that CD47 interacted with HSP90α. Previous
   studies have demonstrated that HSP90α interacts with AMPK[194]^36 and
   that decreased phosphorylation of HSP90α leads to increased activation
   of AMPK.[195]^15 Furthermore, DNA-dependent protein kinase (DNA-PK), a
   serine/threonine kinase, has been implicated in the phosphorylation of
   HSP90α in several studies.[196]^15^,[197]^37^,[198]^38 Here, we
   speculated that CD47-blocking antibody inhibits HSP90α phosphorylation
   via DNA-PK, which in turn promotes AMPK activation. To test this, we
   investigated the effects of Cd47 knockdown, overexpression, and CD47
   functional blockade on DNA-PK activity and phosphorylation of its
   downstream target, HSP90α. DNA-PK activity (p-DNA-PK) was assessed by
   monitoring the autophosphorylation at S2056.[199]^39 The results
   revealed that Cd47 deletion reduced the phosphorylation of both DNA-PK
   and HSP90α ([200]Figure 4C), while Cd47 overexpression had the opposite
   effect, enhancing their phosphorylation ([201]Figure 4D). To determine
   whether p-DNA-PK mediates CD47-induced HSP90α phosphorylation, myotubes
   overexpressing Cd47 were treated with a DNA-PK inhibitor. The results
   showed that DNA-PK inhibition abolished the Cd47 overexpression-induced
   phosphorylation of HSP90α ([202]Figure 4E), suggesting that activated
   DNA-PK is essential for CD47 regulation of HSP90α phosphorylation.
   Consistently, the phosphorylation levels of both DNA-PK and HSP90α were
   also reduced in myotubes treated with a CD47-blocking antibody
   ([203]Figure 4F). Collectively, these findings indicate that the
   CD47-blocking antibody reduces HSP90α phosphorylation via inhibition of
   DNA-PK.

Figure 4.

   [204]Figure 4
   [205]Open in a new tab

   CD47-blocking antibody promotes AMPK activation by reducing
   phosphorylation of HSP90α

   (A) Strategy for analyzing CD47-interacting proteins using liquid
   chromatography-tandem mass spectrometry (LC-MS/MS).

   (B) Endogenous interactions between CD47 and HSP90α detected in
   myotubes by western blotting.

   (C and D) HSP90α and DNA-PK phosphorylation in Cd47 knockdown or
   overexpressing myotubes.

   (E) DNA-PK inhibitor treatment (10 μM) effects on HSP90α and DNA-PK
   phosphorylation in Cd47-overexpressing myotubes.

   (F) HSP90α and DNA-PK phosphorylation in myotubes treated with
   CD47-blocking antibody.

   (G) HSP90α phosphorylation and AMPK activation after treatment with
   tanespimycin (17-AAG).

   (H) AMPK phosphorylation in Cd47-overexpressing myotubes treated with
   17-AAG.

   (I) AMPK phosphorylation after lentiviral treatment with mutated HSP90α
   phosphorylation sites.

   (J) The inhibitory effect of Cd47 overexpression on AMPK activation was
   blocked by mutated HSP90α phosphorylation sites.

   (K) Schematic: CD47-blocking antibody reduces HSP90α phosphorylation,
   promoting AMPK activation.

   Data are presented as means ± SEM and analyzed by two-way RM ANOVA (C,
   right) and Student’s t test (D, right). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p <
   0.001; ns, not significant.

   To verify whether CD47 inhibits AMPK activation by increasing HSP90α
   phosphorylation, an inhibitor of HSP90α, tanespimycin (17-AAG),[206]^40
   was added into Cd47-overexpressing myotubes. Our results showed that
   17-AAG reduced the phosphorylation level of HSP90α and increased the
   activation of AMPK ([207]Figure 4G); meanwhile, the inhibitory effect
   of Cd47 overexpression on AMPK activation was also effectively blocked
   by 17-AAG ([208]Figure 4H). To further demonstrate that CD47 inhibits
   AMPK activation by increasing HSP90α phosphorylation levels,
   lentiviruses with normal and mutant HSP90α (inhibiting phosphorylation)
   sites were utilized to infect Cd47-overexpressing myotubes. The results
   showed that myotube cells infected with the mutant HSP90α could relieve
   the inhibition of AMPK activation by Cd47 overexpression ([209]Figure
   4I). In sum, these results suggest that CD47 promotes HSP90α
   phosphorylation to inhibit AMPK activation ([210]Figure 4J).
   Interestingly, 17-AAG was administered intraperitoneally in mice on
   HFD, and, through long-term monitoring, the results showed that the
   inhibitor effectively alleviated HFD-induced obesity ([211]Figure S7D)
   and glucose homeostasis ([212]Figures S7E and S7F), promoted AMPK
   activation, and elevated PGC-1α levels in skeletal muscle tissue
   ([213]Figure S7G), consistent with the phenotype observed in mice
   injected with CD47-blocking antibody. In conclusion, the CD47-blocking
   antibody promotes AMPK activation by reducing phosphorylation of HSP90α
   ([214]Figure 4K).

   In order to clinically translate CD47 antibodies, it is essential to
   evaluate their side effects in vivo. One of the most significant
   questions is whether their long-term application could lead to cardiac
   hypertrophy caused by persistent AMPK activation. In this study, our
   results revealed no significant difference in heart sizes ([215]Figures
   S8A and S8B); heart weight/body weight ratios ([216]Figure S8C);
   cardiomyocyte cross-sectional area, as assessed by wheat germ
   agglutinin (WGA) staining ([217]Figure S8D); or cardiac fibrosis, as
   assessed by Masson’s trichrome staining ([218]Figure S8E) between
   CD47^−/− and WT mice. Furthermore, no differences were observed in left
   ventricular diastolic and systolic posterior wall thickness (LVPW:d,
   LVPW:s) ([219]Figures S8F and S8G) in CD47-blocking antibody-injected
   mice compared to control IgG antibody-injected mice. In brief, our
   results did not reveal any significant changes in cardiac morphology
   and fibrosis following CD47 deficiency or blockade. Notably,
   CD47-blocking antibody treatment under HFD conditions significantly
   attenuated the reduction in ejection fraction and fractional shortening
   ([220]Figures S8H and S8I), suggesting an improvement in cardiac
   function. Collectively, these findings demonstrate that long-term
   treatment with CD47-blocking antibodies does not induce cardiac
   hypertrophy. Previous studies have shown that both CD47-blocking
   antibodies and Cd47 gene knockout effectively mitigate
   isoproterenol-induced cardiac hypertrophy.[221]^41^,[222]^42 Other
   research suggests that CD47 expression is upregulated during cardiac
   hypertrophy.[223]^43^,[224]^44 In conjunction with our findings, these
   observations strengthen the hypothesis that CD47 deficiency or blockade
   may play a pivotal role in preserving cardiac health under both
   pathological and physiological conditions.

Discussion

   In the current study, we demonstrated a direct causal link between CD47
   LOF and AMPK activation in three CD47-knockout mouse models and further
   presented a rationale for the therapeutic potential of CD47-blocking
   antibodies in metabolic diseases by targeting skeletal muscle AMPK
   activation. CD47 pharmacological intervention can offer a promising
   approach to alleviate HFD-induced metabolic disorders, conferring
   metabolic benefits.

   CD47-blocking antibodies have been used in clinical trials for the
   treatment of cancer.[225]^45 Here, our results revealed that the
   administration of the CD47-blocking antibody exerts metabolic benefits
   against HFD, including reduced body weight gain, improved body
   composition, enhanced basal metabolism and exercise capacity,
   alleviated glucose tolerance, and improved serum lipids and
   mitochondrial function. Notably, most weight-loss drugs tend to reduce
   muscle mass while losing body weight.[226]^46 Here, we found that, in
   addition to its significant therapeutic effect on weight loss,
   CD47-blocking antibody also preserves muscle mass while reducing fat
   mass. This may be attributed to the essential role of CD47 signaling in
   the expansion of muscle stem cells (MuSCs).[227]^47 The discovery has
   important clinical implications. (1) CD47-blocking antibody is an
   important candidate for the development of weight-loss drugs. (2) It
   provides a potential strategy for expanding endogenous MuSCs in vivo
   for therapeutic purposes. In addition, whether CD47-blocking antibody
   treatment has curative effects on skeletal muscle dysfunction due to
   other pathological conditions, such as muscular dystrophy diseases, is
   unclear. Therefore, the applications of CD47-blocking antibodies should
   be expanded in future studies.

   CD47 plays distinct roles in different
   cells.[228]^48^,[229]^49^,[230]^50^,[231]^51 In detail, as a
   transmembrane protein, CD47 is a counter-receptor for signal-regulated
   protein-alpha (SIRPα). The CD47-SIRPα axis is crucial in delivering the
   “do-not-eat-me” signal, protecting cells from phagocytosis by
   macrophages.[232]^52 Thrombospondin-1 (TSP1) is the other receptor for
   CD47, and the TSP1-CD47 axis is necessary for MuSCs proliferation under
   injury.[233]^47 A recent study reveals that TSP1/CD47 paracrine
   signaling inhibits the proliferation of aged MuSCs, and blocking the
   TSP1/CD47 signaling restores the proliferative capacity of
   MuSCs.[234]^53 However, due to the lack of a cytoplasmic domain of
   CD47, these studies mainly examined the role of CD47 in extracellular
   signaling with a limited understanding of intracellular downstream
   signaling.[235]^54 Here, our study elucidates the intracellular pathway
   linking CD47 to AMPK activation and highlights a therapeutic potential
   for CD47-blocking antibodies.

Limitations of the study

   In conclusion, our study highlights that the CD47-blocking antibody
   confers metabolic protective effects against obesity via restoration of
   AMPK activity in skeletal muscles. We show that CD47 inhibits AMPK
   activity through the DNA-PK-HSP90α pathway, which aligns with previous
   reports on the metabolic regulatory function of DNA-PK.[236]^15 Future
   studies should investigate whether this signaling mechanism is also
   operative in other cell types or under different pathological
   conditions, which will further elucidate the therapeutic efficacy of
   CD47-blocking antibody and broaden its potential clinical applications.
   We observe an interaction between CD47 and HSP90α, with CD47 promoting
   HSP90α phosphorylation. These results infer that CD47 may facilitate
   HSP90α phosphorylation by stabilizing the interaction between DNA-PK
   and HSP90α. However, the precise mechanism by which CD47 modulates the
   DNA-PK-HSP90α-AMPK axis—particularly whether this occurs through
   receptor-mediated signaling or direct interaction—requires further
   investigation. Additionally, the potential involvement of the cellular
   AMP/ATP ratio or Ca^2+-CaMKK2 signaling in the downstream effects of
   CD47 and how these signaling pathways interact to regulate AMPK
   activity also warrant future investigation.

Resource availability

Lead contact

   For additional information or material requests, please contact the
   corresponding author, Weida Li (liweida@tongji.edu.cn), who will
   coordinate all related inquiries and requests.

Materials availability

   All materials and reagents generated in this study are available from
   the corresponding author upon execution of an appropriate materials
   transfer agreement.

Data and code availability

     * •
       All data associated with this study are included in the manuscript
       and its [237]supplemental information.
     * •
       The RNA-seq raw data and processed count matrices generated in this
       study have been deposited in the NCBI Sequence Read Archive (SRA)
       under BioProject accession SRA: PRJNA1224594, and the proteomics
       dataset is available via the PRIDE database under accession PRIDE:
       [238]PXD061117. Published transcriptomic data analyzed in this work
       were obtained from GEO: [239]GSE19420
       , GEO: [240]GSE81965, GEO: [241]GSE165630, and proteomic data were
       sourced from PRIDE: [242]PXD044445.
     * •
       This paper does not report original code.
     * •
       Any additional information required to reanalyze the data reported
       in this paper is available from the [243]lead contact upon request.

Acknowledgments