Circulation

ORIGINAL RESEARCH ARTICLE

Cannabinoid Receptor 2-Centric Molecular

Feedback Loop Drives Necroptosis in Diabetic
Heart Injuries
Pan Gao , PhD; Mengying Cao , BS; Xueli Jiang, MD, PhD; Xiaolin Wang , BS; Guoping Zhang, BS; Xinru Tang , BS;
Chunjie Yang, BS; Issei Komuro , MD, PhD; Junbo Ge , MD, PhD; Liliang Li , MD, PhD; Yunzeng Zou , MD, PhD

BACKGROUND: Diabetic heart dysfunction is a common complication of diabetes. Cell death is a core event that leads to
diabetic heart dysfunction. However, the time sequence of cell death pathways and the precise time to intervene of particular
cell death type remain largely unknown in the diabetic heart. This study aims to identify the particular cell death type that is
responsible for diabetic heart dysfunction and to propose a promising therapeutic strategy by intervening in the cell death
pathway.

METHODS: Type 2 diabetes models were established using db/db leptin receptor–deficient mice and high-fat diet/streptozotocin–
induced mice. The type 1 diabetes model was established in streptozotocin-induced mice. Apoptosis and programmed cell
necrosis (necroptosis) were detected in diabetic mouse hearts at different ages. G protein–coupled receptor–targeted drug
library was searched to identify potential receptors regulating the key cell death pathway. Pharmacological and genetic
approaches that modulate the expression of targets were used. Stable cell lines and a homemade phosphorylation antibody
were prepared to conduct mechanistic studies.

RESULTS: Necroptosis was activated after apoptosis at later stages of diabetes and was functionally responsible for
Downloaded from http://ahajournals.org by on December 6, 2022

cardiac dysfunction. Cannabinoid receptor 2 (CB2R) was a key regulator of necroptosis. Mechanically, during normal
glucose levels, CB2R inhibited S6 kinase–mediated phosphorylation of BACH2 at serine 520, thereby leading to
BACH2 translocation to the nucleus, where BACH2 transcriptionally repressed the necroptosis genes Rip1, Rip3, and
Mlkl. Under hyperglycemic conditions, high glucose induced CB2R internalization in a β-arrestin 2–dependent manner;
thereafter, MLKL (mixed lineage kinase domain-like), but not receptor-interacting protein kinase 1 or 3, phosphorylated
CB2R at serine 352 and promoted CB2R degradation by ubiquitin modification. Cardiac re-expression of CB2R rescued
diabetes-induced cardiomyocyte necroptosis and heart dysfunction, whereas cardiac knockout of Bach2 diminished
CB2R-mediated beneficial effects. In human diabetic hearts, both CB2R and BACH2 were negatively associated with
diabetes-induced myocardial injuries.
CONCLUSIONS: CB2R transcriptionally repressed necroptosis through interaction with BACH2; in turn, MLKL formed a
negative feedback to phosphorylate CB2R. Our study provides the integrative view of a novel molecular mechanism loop
for regulation of necroptosis centered by CB2R, which represents a promising alternative strategy for controlling diabetic
heart dysfunction.

Key Words: BACH2 protein, human ◼ diabetic cardiomyopathies ◼ necroptosis ◼ receptors, cannabinoid

Correspondence to: Yunzeng Zou, MD, PhD, Shanghai Institute of Cardiovascular Diseases, Zhongshan Hospital and Institute of Biomedical Sciences, Fudan
University. 138 Yixueyuan Road, Xuhui District, Shanghai 200032, China, Email [email protected]; Liliang Li, MD, PhD, Department of Forensic Medicine,
School of Basic Medical Sciences, Fudan University, 131 Dongan Road, Xuhui District, Shanghai 200032, China, Email [email protected]; or Junbo Ge, MD,
PhD, Shanghai Institute of Cardiovascular Diseases, Zhongshan Hospital and Institute of Biomedical Sciences, Fudan University, 180 Feng Lin Road, Shanghai
200032, China, Email [email protected]
Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/circulationaha.122.059304.
For Sources of Funding and Disclosures, see page XXX
© 2022 American Heart Association, Inc.
Circulation is available at www.ahajournals.org/journal/circ

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 1

 Gao et al CB2R-Centric Necroptotic Loop in DCM

has been estimated that more than one-third of patients

Clinical Perspective with diabetes develop diabetic cardiomyopathy.6,7
ORIGINAL RESEARCH

Multiple mechanisms contribute to cardiomyocyte

What Is New? death, which triggers subsequent inflammatory injury,
ARTICLE

• We have identified necroptosis as the predominant cardiac hypertrophy, and cardiac fibrosis, a process called
cell death type at later stages of diabetic heart. cardiac remodeling. At present, multiple drugs that target
• Cannabinoid receptor (CB2R) recruits transcrip- each of the cardiac remodeling, oxidative stress, inflam-
tion factor BACH2 to repress necroptosis and pro- mation, and metabolic disturbance processes have been
tects against diabetic heart injury, whereas during implicated with therapeutic potential in preclinical stud-
hyperglycemia, MLKL in turn phosphorylates CB2R ies and clinical trials.8 Thus far, these molecule-targeted
to promote ubiquitin-dependent degradation of
approaches, however, have failed to translate to clinical
CB2R, thus forming a CB2R-centric feedback loop
use or to improve residual morbidity and mortality.9,10
of necroptosis.
• Cardiac CB2R or BACH2 expression negatively As is well established, cell death is the core event
correlates with both MLKL expression and the that results from early-stage molecular alterations and
extent of diabetic heart injuries in human cases. precedes cardiac remodeling.11,12 Given that overt cardio-
myocyte death underpins structural changes of the left
What Are the Clinical Implications? ventricular wall that often lead to ventricular dysfunction,
• The CB2R-centric necroptotic loop represents a targeting the cardiomyocyte death pathway at a time
promising target for the clinical treatment of dia- point that kick-starts the manifestation of diabetic heart
betic heart injuries. dysfunction would be of potential translational value.
• CB2R agonists may be potent therapeutic com- Both apoptosis and necroptosis (a more recently
pounds that have translational value in necroptosis- defined inflammatory cell death pathway) are forms of
driven cardiovascular diseases. programmed cell deaths that have received wide attention
in terms of diabetic heart injury. Apoptosis is characterized
by a series of dramatic perturbations to the cellular archi-
tecture that is orchestrated by members of the caspase
Nonstandard Abbreviations and Acronyms family of cysteine proteases (eg, caspase-3).13 Caspase-
CB2R cannabinoid receptor 2 mediated apoptosis shares a morphological definition,
including blebbing, chromatin condensation, nuclear frag-
Downloaded from http://ahajournals.org by on December 6, 2022

FADD Fas associated via death domain

mentation, loss of adhesion and rounding (in adherent
GPCR G protein–coupled receptor
cells), and cell shrinkage.14 Unlike apoptosis, necroptosis
HG high glucose
is typically induced by ligand binding to tumor necrosis
MLKL mixed lineage kinase domain-like
factor family death domain receptors, pattern-recogniz-
Nec-1 necrostatin-1 ing receptors, and virus sensors. The common feature of
p- phosphorylated these receptor systems is the implication of proteins that
RIP receptor-interacting protein kinase contain a receptor-interacting protein kinase (RIP) homol-
S352 serine 352 ogy interaction motif mediating recruitment and activa-
S6K S6 kinase tion of RIP3, which ultimately activates the necroptosis
TRADD TNF receptor superfamily member 1a executioner MLKL (mixed lineage kinase domain-like).15
associated via death domain Activated MLKL then oligomerizes and translocates to
WT wild type cellular membranes to punch holes, triggering release of
a series of cytokines and damage-associated molecular
patterns.16 Necroptotic cells show morphological features

D
iabetes is a chronic metabolic noncommunicable similar to those of other programmed necrosis, namely
disease that has become an epidemic world- cell swelling, organelle dysfunction, and rupture of cyto-
wide.1,2 It could be primary or secondary to other plasm membrane.17–19 However, the executioner MLKL is
factors such as unhealthy eating habits, sedentary life- unique to necroptosis, which differentiates it from apop-
styles,3 and long-term antipsychotic use.4 Diabetes has a tosis and other programmed necroses.18 Moreover, these
strong epidemiological link with cardiovascular diseases cell death pathways are not independent but rather time
and is associated with a higher risk of heart failure inde- sequentially occur by the molecular switcher caspase-8,20
pendently of the traditional risks, including hyperten- indicating that it is possible and important to capture the
sion, coronary heart disease, and valvular heart disease.5 particular cell death type at the start point of heart dys-
Long-standing diabetes results in structural and func- function. The present study tested the hypothesis that
tional changes of cardiomyocytes that are independent targeting the particular cell death type would confer car-
of myocardial ischemic or microvascular atherosclerotic dioprotection in diabetic mouse models and in human
disease, an entity that is called diabetic cardiomyopathy.6 It samples. We also surmised that it would be beneficial to

2 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

identify critical regulators that dominate the particular cell tially, death domain containing adaptor molecules (TRADD
death type. [TNF receptor superfamily member 1a associated via death

ORIGINAL RESEARCH
domain] and FADD [Fas associated via death domain], 2
markers associated with programmed cell deaths) were

ARTICLE
METHODS gradually increased with HG exposure time extension (Fig-
All data, analytical methods, and study materials that support ure 1A). Apoptosis markers caspase-8 and cl-caspase-3
the findings of this study are available from the correspond- were time-dependently increased only in the initial 72
ing authors on reasonable request. Expanded Methods are hours. Necroptotic markers such as phosphorylated RIP1
provided in the Supplemental Material. Raw RNA sequencing (p-RIP1), RIP3 (p-RIP3), and MLKL (p-MLKL) remained
data have been made publicly available in the Beijing Institute
consistently unchanged before 72 hours but accumulated
of Genomics Data Center,21 Chinese Academy of Sciences,
dramatically thereafter. Moreover, the apoptosis and cell
under the accession number CRA007662. Analyzed results
of the RNA sequencing data are provided as Excel File S1. death markers were also dose-dependently promoted by
Quantification data for Western blot and histological staining HG treatments for 48 hours, whereas the phosphorylated
are provided as Excel File S2. Human case-by-case informa- levels of necroptosis proteins remained largely unchanged
tion is provided as Excel File S3. within 48 hours even under high concentrations of glucose
(Figure 1A). Scanning electron microscopy confirmed that
Human Samples the HG-treated myocytes presented with multiple even-
sized apoptotic bodies at 24 hours. The cells were round
The formalin-fixed paraffin-embedded human heart slices from
decedents with diabetes (diabetes group, n=25) or without dia- up with bursting extensions and partially detached from the
betes (control group, n=25) were obtained from the Department culture slide after 72 hours (Figure 1B). Flow cytometry
of Forensic Medicine, School of Basic Medical Sciences, Fudan analysis showed that the percentages of both early (Q3)
University. All patients were anonymized to protect the dece- and late (Q2) apoptotic cells increased in the first 48 hours
dents’ privacy. The use of human heart samples for research but decreased sharply thereafter, whereas the percentage
purposes was approved by the Ethical Reviewing Board at the of necrotic cells (Q1) kept increasing over time (Figure
School of Basic Medical Sciences, Fudan University (approvals S1A). We then extracted total proteins from heart tissues
2019-023 and 2020-009), and signed informed consents were of db/db diabetic mice at different ages and found that the
obtained from the next of kin of each decedent before autopsy. necroptosis markers (p-RIP1, p-RIP3, p-MLKL) began to
increase remarkably at 16 weeks of age. The apoptosis
Mouse Models and Treatments markers caspase-3, Bax, and caspase-8 again increased
Downloaded from http://ahajournals.org by on December 6, 2022

The protocol for the animal study was in accordance with our only before 16 weeks of age (Figure 1C). To confirm the
institutional guidelines and approved by the Institutional Animal time sequences of cell deaths, we modulated caspase-8, the
Care and Use Committee at Zhongshan Hospital, Fudan molecular switch that shifts necroptosis to apoptotic pheno-
University. Type 2 diabetes mouse models were established types,20 after 24- or 72-hour HG treatments. We observed
using both db/db diabetic mice and high-fat diet (HFD) in con- that the necroptosis activated at 72 hours was successfully
junction with low-dose streptozotocin injection,22 and the type 1
switched to apoptosis by caspase-8 overexpression (Figure
diabetes mouse model was constructed by injection.
S1B). In turn, although caspase-8 was knocked down at 24
hours after the HG treatments, the HG-induced apoptosis
Statistical Analysis was switched to necroptosis (Figure S1B).
Data are presented as mean±SEM. For numerical variables, the We further isolated primary cardiomyocytes from both
Student t test was used for comparisons between 2 groups. male and female adult C57BL/6J mice and confirmed
One- or 2-way ANOVA was used to compare means among ≥3 that RIP1, RIP3, and MLKL were consistently increased
groups when appropriate, followed by the least-significance-
in both total and phosphorylated forms by long-term HG
difference post hoc test. For categorical variables, the χ2 or
treatments (Figure 1D for male mice and Figure S1C
Fisher exact test was used for comparisons of proportions
when applicable. Statistical analysis was performed with SPSS for female mice). Of note, the necroptosis markers were
16.0 (SPSS Inc, Chicago, IL) or GraphPad Prism version 8.0 also promoted at their mRNA levels by HG in mouse
(GraphPad, La Jolla, CA) software. Any difference with a value HL-1 myocytes (Figure 1E) and by hyperglycemia in the
of P<0.05 was considered statistically significant. db/db type 2 diabetic mice (Figure 1F). The addition of
insulin (Figure S2A) or the insulin sensitizers metformin
and rosiglitazone (Figure S2B) and genetic knockout of
RESULTS the insulin receptor (Figure S2C) failed to rescue HG-
Necroptosis Was Activated Predominantly in induced detrimental effects (Figure S2D). In support of
this notion, in a streptozotocin-induced type 1 diabetic
Cardiomyocytes at Later Stages of Diabetes In model that is normally considered irrelevant to insulin
Vitro and In Vivo resistance, mice also showed impaired heart function
To depict the time sequence of high glucose (HG)–induced (Figure S2E–S2G) and activated necroptosis pathway in
cell death, we detected cell death types in time series. Ini- both male and female mouse hearts (Figure S2H). These

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 3

 Gao et al CB2R-Centric Necroptotic Loop in DCM
ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 1. Necroptosis was activated in diabetic heart at later stages.

A, Time sequence of necroptosis and apoptosis development in HL-1 cells treated with high glucose (HG; 30 mmol/L) for different times or with
various concentrations of HG for 48 hours. B, Representative scanning electron microscope images of HL-1 cells incubated with HG for 0, 24,
or 72 hours. Scale bar: 10 μm. C, Activation of necroptosis and apoptosis in db/db mouse heart at different ages (weeks). D, Protein levels of
necroptosis markers in low glucose– (LG; 5.6 mmol/L) or HG- (30 mmol/L) treated (72 hours) primary cardiomyocytes isolated from adult male
C57BL/6J mice. E, mRNA levels of Rip1, Rip3, and Mlkl in HL-1 cells with or without HG (30 mmol/L for 72 hours) treatments. F, mRNA levels
of Rip1, Rip3, and Mlkl in the hearts of db/dm and db/db mice at 24 weeks of age. (Continued )

4 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

Figure 1 Continued. G, Immunofluorescence staining of MLKL (mixed lineage kinase domain-like) in db/dm and db/db mouse hearts (24
weeks of age). Insets in the right corner indicate local magnification. Scale bar as indicated. H, HL-1 cells were pretreated with HG (30 mmol/L)

ORIGINAL RESEARCH
for 60 hours and were then observed under time-lapse microscope. Time duration is shown as hours:minutes:seconds.milliseconds. Scale bar:
50 μm. I through N, db/db mice were injected with necrostatin-1 (Nec-1; 1 mg/kg) at 16 weeks of age for 2 continuous months. Mouse heart

ARTICLE
functions were monitored by echocardiography before death, and the heart weight/tibia length (HW/TL) ratio was calculated (I). The hearts were
then examined by quantitative reverse transcription–polymerase chain reaction (J and K), Western blot (L), and histological analyses, including
hematoxylin and eosin (HE) staining, Sirius red staining, Oil red staining, and immunohistochemistry (IHC) staining of MLKL (M; n=6–7 per
group). N, Quantification of MLKL immunohistochemistry staining. FADD indicates Fas associated via death domain; LVEDd, left ventricular
end-diastolic dimension; LVESd, left ventricular end-systolic dimension; p-, phosphorylated; and TRADD, TNF receptor superfamily member 1a
associated via death domain. *P<0.05 as indicated.

data suggested that HG induced necroptosis in both a to screen potential targets for inhibiting HG-induced
sex- and insulin signaling–independent manner. necroptosis on the basis of the notion that ≈40% of
MLKL translocates to cytomembrane to execute cell clinically used drugs target GPCRs.25,26 In view of the
rupture after activation by phosphorylation.23 We also primary transcriptional promotion of necroptosis by HG,
observed membrane-resided MLKL in the db/db but we designed dual-luciferase reporter assays to measure
not the db/dm mouse hearts (Figure 1G). Time-lapse the effects of these GPCR drugs on the transcription of
microscopy, started at 60 hours after the HG exposure, Rip1, Rip3, and Mlkl. It turned out that cannabinoid recep-
further revealed the gradual development of necroptosis tor 2 (CB2R) showed the best dictation of necroptosis
along with extended HG exposure (Figure 1H). All these gene transcriptions (Figure 2A). To verify this finding, we
findings suggested that necroptosis was activated in the added AM1241 and JWH-133, 2 selective CB2R ago-
cardiomyocytes at later stage of diabetes. nists, to the HG-treated primary mouse cardiomyocytes
and found that each agonist obviously dampened HG-
Pharmacological Intervention at a Point induced activation of necroptosis (Figure 2B). To assess
whether the CB2R agonists exerted any off-target ef-
of Necroptosis Development Successfully
fects, Cb2r overexpression and knockdown (shCB2R)
Attenuated Diabetic Heart Dysfunction were performed in primary cardiomyocytes. It was found
Necrostatin-1 (Nec-1) is a specific inhibitor of necroptosis that JWH-133–mediated inhibition of necroptosis was
by allosterically targeting RIP1.24 To assess the therapeu- further enhanced after Cb2r overexpression (Figure 2C)
tic relevance of necroptosis inhibition, we injected Nec- but diminished after Cb2r knockdown in the primary
Downloaded from http://ahajournals.org by on December 6, 2022

1 into type 2 diabetic mice at 16 weeks of age when cardiomyocytes (Figure 2D) and in a cultured mouse
necroptosis started to develop. Echocardiography showed HL-1 cell line (Figure S3A and S3B). CB2R also sup-
that although the db/db mice presented with significantly pressed necroptosis in HL-1 cells at both the protein
impaired heart function compared with their db/dm coun- (Figure 2E) and mRNA (Figure 2F and 2G) levels with
terparts, Nec-1 injection remarkably maintained the left the genetic approaches. We then administrated 2 se-
ventricular end-diastolic diameter, left ventricular end- lective CB2R agonists into db/db mice at 12 weeks of
systolic diameter, ejection fraction, and fractional shorten- age. The mRNA levels of fetal genes such as Anp and
ing (Figure 1I). The ratio of heart weight to tibia length in Bnp and the profibrotic genes Tgfβ1 and Fn were in-
the db/db mice was also significantly decreased by Nec-1 creased in the db/db mice but significantly decreased
treatments (Figure 1I). Moreover, the use of Nec-1 notably when the diabetic mice were treated with any CB2R ag-
ameliorated the mRNA levels of Fn and Il-6 in the db/db onist (Figure 2H). In concordance, both CB2R agonists
mouse heart (Figure 1J and 1K). Nec-1 treatment also ameliorated the hyperglycemia-induced heart pathology,
blocked the necroptosis pathway (Figure 1L), attenuated as evidenced by histological studies (Figure 2I), trans-
heart pathology (Figure 1M), and inhibited hyperglycemia- mission electron microscopy (Figure S3C), and morpho-
induced MLKL protein expression, as well as its mem- metric measurements (Figure S3D), and improved heart
brane translocation, in the db/db mouse heart (Figure 1M function, as evidenced by echocardiography (Figure 2J
and 1N). These data verified the hypothesis that inter- and 2K). As a reflection of myocardial injury attenuation,
vening in diabetes at a stage of predominant cell death RIP1, RIP3, and MLKL were suppressed by CB2R ago-
(necroptosis) development conferred cardioprotection. nists in both total and phosphorylated forms (Figure 2L).
Immunohistochemistry staining showed that MLKL was
Small-Molecule Agent–Based Screening upregulated and translocated to cell membrane in the
Identified Cannabinoid Receptor 2 Agonists diabetic heart but downregulated after pharmacologi-
cal activation of CB2R (Figure 2M and Figure S3E and
as Potent Inhibitors of Cell Necroptosis in the S3F). Scanning electron microscopy further confirmed
Diabetic Heart that HG-induced cell roundup and balloon-like mor-
To figure out how HG activated necroptosis, we used phology were recovered by the CB2R selective agonist
a pool of G protein–coupled receptors (GPCRs) drugs AM1241 (Figure 2N).

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 5

 Gao et al CB2R-Centric Necroptotic Loop in DCM
ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 2. CB2R potently inhibited cell necroptosis in diabetic heart.

A, A pool of G protein coupled receptor (GPCR)–targeted antagonists/agonists (final dose, 1 μmol/L) were added to high glucose (HG; 30
mmol/L)–treated mouse HL-1 cells, and their effects on the luciferase activities of Rip1, Rip3, and Mlkl were assessed. Left lane represents
each type of GPCRs; right lane represents the cumulative fold change (cum-FC), which is calculated as FC (Rip1)×FC (Rip3)×FC (Mlkl).
Cannabinoid receptor 2 (CB2R) is highlighted with red because its agonists downregulated and its antagonists upregulated necroptosis
transcription. B, Protein levels of the necroptosis pathway when primary cardiomyocytes were treated with HG (30 mmol/L) for 72 hours with
or without selective CB2R agonists. The working concentrations for AM1241 and JWH-133 were 1 μmol/L. C and D, Protein levels of the
necroptosis pathway when primary cardiomyocytes were treated with HG (30 mmol/L) and JWH-133 (a CB2R agonist, 1 μmol/L) for 72
hours with or without Cb2r overexpression. E, Protein levels of necroptosis when HL-1 cells were stably overexpressed or depleted of Cb2r. F
and G, mRNA levels of Rip1, Rip3, and Mlkl in HL-1 cells with indicated treatments. H through M, Two CB2R-selective agonists (both 1 mg/
kg) were injected intraperitoneally into db/db mice at 12 weeks of age for 3 consecutive months (n=6–8 per group). H, mRNA levels of Anp,
Bnp, Tgfβ1, and Fn were analyzed by quantitative reverse transcription–polymerase chain reaction. I, Hematoxylin and eosin (HE) staining,
Oil red staining, Masson trichrome staining, and Sirius red staining of mouse heart from each group. Scale bar: 1 mm or 25 μm as indicated.
Insets in the right corner represent local magnification. J and K, Echocardiographic examination of ejection fraction and fractional shortening
of the mice. L, Western blot analysis of the protein levels of necroptosis. M, Representative images of immunohistochemical staining. Black
arrows indicate membrane positivity. Scale bar: 50 μm. Insets in the right corner represent local magnification. N, Representative images of
scanning electron microscopy of cardiac HL-1 cells treated with HG (30 mmol/L) for 72 hours with or without AM 1241 (1 μmol/L). Scale bar
as indicated. FADD indicates Fas associated via death domain; MLKL, mixed lineage kinase domain-like; p-, phosphorylated; and TRADD, TNF
receptor superfamily member 1a associated via death domain. *P<0.05.

6 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

To assess whether this regulation was mouse model Rip3 and that this binding activity was further enhanced
dependent, we established another type 2 diabetes in CB2R-overexpressed cells (Figure 3F). We then

ORIGINAL RESEARCH
model using the high-fat diet/streptozotocin injection mutated each of the binding sites (Figure S5E). Dual
and found that cardiac overexpression of CB2R by AAV luciferase reporter assays showed that mutation of any

ARTICLE
also attenuated necroptosis and cardiac dysfunction binding site led to failure of BACH2 to inhibit Rip1 (Fig-
induced by hyperglycemia (Figure S3G–S3L). These ure 3G) or Rip3 (Figure 3H) luciferase activities. These
data suggested that CB2R potently suppressed cardiac data suggested that CB2R inhibited necroptosis through
cell necroptosis across diabetes models. regulation of the transcription factor BACH2.

CB2R Inhibited Rip1, Rip3, and Mlkl CB2R Enhanced BACH2 Nuclear Translocation
Transcription by Enhancing Transcription Factor by Inhibiting S6K Activity
BACH2 Nuclear Translocation Cytoplasm-nucleus trafficking of BACH2 is phosphory-
Because CB2R primarily regulated necroptosis at mRNA lation dependent. Free-state BACH2 can translocate
levels, we further examined whether CB2R regulated into nucleus, whereas phosphorylation of BACH2 by the
necroptosis during transcription processes. As shown upstream S6 kinase (S6K) fails to translocate.28 Unlike
in Figure S4, HG treatment significantly promoted Rip1, HG, which maintained the cytoplasm location of BACH2,
Rip3, and Mlkl luciferase activities. Genetic overexpres- we found that LJI308, a specific inhibitor of S6K activity,
sion or pharmacological activation of CB2R significantly shifted BACH2 to the nucleus location when cotreated
dampened whereas genetic knockdown of CB2R signifi- with HG (Figure 3I). HG promoted the phosphorylation of
cantly enhanced these HG-induced effects. p70 S6K, whereas overexpression of Cb2r suppressed
To figure out how CB2R inhibited the transcription the phosphorylation of S6K, particularly when LJI308
of necroptosis, we performed RNA-sequencing analysis was cotreated (Figure 3J). As a consequence of S6K
and conducted multiple comparisons, including CB2R activity changes, phosphorylated BACH2 was observed
stably overexpressed versus vector plasmid-transfected to be promoted by HG treatments. Moreover, CB2R
(control) HL-1 cells and CB2R stable-knockdown overexpression retarded the phosphorylation of BACH2
(shCB2R) versus negative control shRNA-transfected while promoting its total protein levels; this was particu-
(shNC) HL-1 cells (Figure 3A). With a threshold of fold larly true when cells were costimulated with LJI308 (Fig-
ure 3K). BACH2 is phosphorylated mostly at the serine
Downloaded from http://ahajournals.org by on December 6, 2022

change ≥2 and adjusted P<0.05, 386 significantly dif-

ferentially expressed genes overlapped by CB2R over- 520 site.28 Therefore, we mutated the serine 520 of
expression versus control and shCB2R versus shNC mouse BACH2 to be alanine (BACH2–serine 520). In
comparisons, including 4 transcriptional factors (Dmrta2, the Bach2-KO cells, transfection of S6K overexpression
Pitx3, Pou4f1, and Bach2; Figure 3B and Excel File S1). plasmid provoked the phosphorylation of BACH2 only in
Because necroptosis was inhibited by CB2R at the tran- BACH2 wild-type (WT) plasmid–retransfected cells but
scriptional level, we focused specifically on these tran- not in BACH2–serine 520 plasmid–transfected cells
scription factors and designed specific siRNAs (Figure (Figure 3L). These data suggested that CB2R regulated
S5A–S5D). We found that knockdown of Bach2, but not the nuclear translocation of BACH2 by disrupting the
the other transcription factors, reversed the CB2R-medi- S6K-dependent phosphorylation process.
ated inhibition of cell necroptosis (Figure 3C). We further
confirmed that BACH2 was indeed positively regulated Pharmacological Inhibition of CB2R Aggravated
by CB2R (Figure 3D). As a transcription factor, BACH2
was observed to be distributed in both cytoplasm and
Diabetic Heart Injury, But This Was Reversed by
nucleus in low-glucose environments but exclusively in Cardiac Overexpression of BACH2
the cytoplasm under HG environments. Overexpression Next, we assessed whether CB2R or BACH2 had any
of CB2R potently maintained the nuclear location of functional interaction with diabetic heart injury. In cul-
BACH2 even in HG environments (Figure 3E). Nucleus- tured cells, it was observed that knockdown of Cb2r
located BACH2 is mostly bound to genes together with generally promoted the levels of necroptosis in BACH2
its heterodimer partner MafK.27 Using the TFSearch and nonoverexpressed cells but not the BACH2 stably over-
PROMO 3.0 online tools, we found 2 independent bind- expressed cells (Figure 4A). We then established heart-
ing sties on the Rip1 promoter (−766/−762, Rip1-1; restricted BACH2-overexpression mice by cardiac in situ
−76/−72, Rip1-2) and Rip3 promoter (−1112/−1106, injection of AAV9-α–MHC–Bach2 into the mice with
Rip3-1; −137/−132, Rip3-2) for BACH2 (Figure S5E); high-fat diet/streptozotocin–induced type 2 diabetes.
however, this method failed to predict binding sites of We observed that although a selective CB2R antagonist,
BACH2 on the Mlkl promoter. Chromatin immunoprecipi- AM630, alone tended to worsen left ventricular func-
tation assays showed that BACH2 and its partner MafK tion, cardiac BACH2 overexpression significantly im-
indeed bound to the specific promoter sites of Rip1 and proved the ejection fraction and fractional shortening

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 7

 Gao et al CB2R-Centric Necroptotic Loop in DCM
ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 3. CB2R inhibited necroptosis transcription through regulation of BACH2.

A, Principal component (PC) analysis (PCA) showing the clustering of samples and volcano plots showing the RNA-sequencing results for
cannabinoid 2 receptor (CB2R) overexpression (OE) vs control (Con) comparison and shCB2R vs shNC comparison (n=3 per group). Genes with
fold-change ≥2 and adjusted P<0.05 were considered significantly differentially expressed genes (DEGs). Red, green, and gray dots represent
significantly upregulated, downregulated, and not significantly altered genes, respectively. B, DEGs were extracted from each comparison in A,
and overlapping DEGs were analyzed. C, Western blot analysis of the necroptosis pathway when cells were knocked down with each of Dmrta2,
Pitx3, Pou4f1, and Bach2. D, Western blot analysis of BACH2 and CB2R in the cells with stable constructs. E, Immunofluorescence staining
of BACH2 when control or CB2R-OE HL-1 cells were incubated with low glucose (LG; 5.6 mmol/L) or high glucose (HG; 30 mmol/L) for 72
hours. F, Chromatin immunoprecipitation assays of BACH2 and its partner, MafK, on the Rip1 and Rip3 promoters. G, Effects of BACH2 on
the HG-induced Rip1 luciferase activities when the 2 binding sites of BACH2 were independently mutated (denoted as Rip1-Δ1 and Rip1-Δ2
mutants). Data were from 3 independent assays. H, Effects of BACH2 on the HG-induced Rip3 luciferase activities when the 2 binding sites of
BACH2 were independently mutated (denoted as Rip3-Δ1 and Rip3-Δ2 mutants). Data were from 3 independent assays. I, Immunofluorescence
staining of BACH2 when cells were treated with HG (30 mmol/L) with or without the S6 kinase (S6K) inhibitor LJI308 (1 μmol/L). J, Western
blot analysis of the phosphorylated (p-) and total protein levels of p70 S6K in HL-1 cells receiving HG (30 mmol/L), LJI308 (1 μmol/L), or
CB2R-OE treatments for 72 hours. K, The BACH2 antibody–precipitated lysates were subjected to Western blot detection of phosphorylation in
the indicated treatments. L, BACH2 knockout (BACH2-KO) cells were transfected with S6K expression plasmid, BACH2–wild-type (WT) plasmid,
or BACH2–serine (Ser) 520A plasmid. Phosphorylation of BACH2 was detected in the BACH2 antibody–precipitated lysates. IB indicates
immunoblot; IgG, immunoglobulin G; IP, immunoprecipitation; and MLKL, mixed lineage kinase domain-like. *P<0.05, as indicated.

8 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 4. Pharmacological inhibition of CB2R aggravated diabetic heart dysfunction, and cardiac overexpression of BACH2
reversed these detrimental effects.
A, Western blot analysis of necroptosis in HL-1 cells stably overexpressing BACH2 (BACH2-OE) or depleting cannabinoid 2 receptor (shCB2R)
with or without high-glucose (HG; 30 mmol/L) treatments for 72 hours. B through K, High-fat diet/streptozotocin–induced diabetic mice were
injected with vehicle (PBS) or AM630 (1 mg/kg) since the development of hyperglycemia; 4 weeks later, the diabetic mice were cardiac in
situ injected with AAV9-BACH2 or AAV9–empty vector. Sixteen weeks after AAV9 injection, all mice were killed (n=5–8 per group). B and C,
Echocardiography of left ventricular ejection fraction and fractional shortening. D, Western blot analysis of related proteins in the heart from
indicated groups. E through G, Quantitative polymerase chain reaction analysis of the mRNA levels of Bach2, Fn, Il-6, Rip1, Rip3, and Mlkl in the
heart from the indicated groups. H, Immunofluorescence staining of BACH2 and MLKL (mixed lineage kinase domain-like). Scale bar=100 μm. I,
Hematoxylin and eosin (HE) staining, Sirius red staining, and immunohistochemistry (IHC) staining of BACH2 and MLKL. Scale bar: 1 mm or 50
μm. J and K, Quantification of BACH2 and MLKL staining intensity from the immunohistochemistry staining in I. Con indicates control; FADD, Fas
associated via death domain; and p-, phosphorylated. *P<0.05, as indicated.

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 9

 Gao et al CB2R-Centric Necroptotic Loop in DCM

even under AM630 treatment (Figure 4B and 4C). West- but again, AM1241 failed to exert these beneficial ef-
ern blot analysis showed that the AM630-enhanced fects in the Bach2-CKO mice (Figure 5H). These data
ORIGINAL RESEARCH

necroptosis and expression of cell death marker (FADD) suggested that the cardioprotective roles of CB2R de-
and profibrotic factor (FN) were dampened by cardiac pended on the presence of BACH2.
ARTICLE

BACH2 overexpression (Figure 4D). Quantitative poly-

merase chain reaction analysis showed that this AAV9
delivery system caused cardiac Bach2 upregulation by MLKL Formed a Negative Feedback to
≈12-fold and that the AM630-mediated inhibition of Phosphorylate and Suppress CB2R in the
Bach2 was stably reversed by the cardiac BACH2 over- Diabetic Heart
expression (Figure 4E). Quantitative polymerase chain These studies gave us a clue that CB2R was an endog-
reaction analysis also confirmed that the mRNA levels of enous cardioprotective factor but was suppressed in the
Fn, Il-6, Rip1, Rip3, and Mlkl were upregulated by AM630 diabetic heart. Indeed, we detected that the protein level
in only the AAV9-GFP–injected diabetes mice but not of CB2R was obviously decreased by HG in in vitro treat-
the cardiac BACH2–overexpressed mice (Figure 4F and ments (Figure 6A and 6B) and in the type 2 diabetes
4G). Immunofluorescence staining showed that cardiac diabetic heart specimens (Figure 6C and 6D). Further-
overexpression of BACH2 inhibited MLKL expression more, in a streptozotocin-induced type 1 diabetes model,
and its membrane location (Figure 4H). Histology ex- the mRNA levels of Cb2r were also significantly down-
amination revealed that AM630 aggravated diabetes- regulated (Figure S6A). HG decreased the CB2R pro-
induced inflammation and fibrosis, whereas cardiac tein level even in the presence of insulin or its sensitizers
BACH2 overexpression dampened AM630-mediated (metformin and rosiglitazone; Figure S6B and S6C). HG
effects (Figure 4I). Indeed, after cardiac BACH2 protein also remarkably decreased the CB2R protein level in the
elevation by ≈3-fold (Figure 4J), the MLKL protein level presence or absence of insulin receptor (Figure S6D).
was significantly decreased by ≈50%, and AM630 could These data suggested that HG decreased CB2R expres-
no longer enhance the MLKL levels (Figure 4K). These sion independently of the insulin signaling.
findings suggested that CB2R depletion promoted dia- As a GPCR, the content and activity of CB2R depend
betic heart injuries and that cardiac overexpression of largely on its cellular location and phosphorylation modi-
BACH2 could reverse these detrimental effects. fications.29 We thus extracted subcellular proteins when
HL-1 cells were treated with HG and found that CB2R at
Downloaded from http://ahajournals.org by on December 6, 2022

Pharmacological Activation of CB2R Attenuated cytomembrane was notably decreased whereas CB2R in
cytoplasm was increased by HG treatments (Figure 6E).
Diabetic Heart Injury in the Presence of BACH2 CB2R, like the vast majority of GPCRs, binds β-arrestin
Expression (more often β-arrestin 2) proteins and internalizes on
In the cultured CB2R-overexpressed cells, we found that stimulation.30,31 We then examined the expression of
necroptosis was inhibited exclusively under the BACH2- β-arrestin 2 and found that it began to increase as early
present conditions (Figure 5A). To validate this, we es- as 12 hours after HG exposure at both the protein and
tablished in vivo mouse models (Figure 5B). Bach2-CKO mRNA levels (Figure S7A and S7B). The elevation of
diabetic mice presented with higher protein levels of β-arrestin 12 hours after HG exposure corresponded to
FADD; showed elevation of RIP1, RIP3, and MLKL in the start of the CB2R decrease (Figure 6C and Figure
both the total and phosphorylated forms (Figure 5C); and S7A). Although CB2R was located mostly in the cyto-
showed significantly higher mRNA levels of proinflam- membrane and nucleus in low-glucose environments,
matory factor Il-6, profibrotic factor Col1α1 (Figure 5D), it translocated to cytoplasm after HG treatments and
Rip1, Rip3, and Mlkl (Figure 5E) compared with Bach2- returned to the cytomembrane when a specific inhibitor
WT diabetic mice. It is important to note that AM1241 of β-arrestin 2 (Barbadin) was cotreated (Figure S7C).
remarkably suppressed cell necroptosis in only the Using 3 specific siRNAs against β-arrestin 2 (Figure
Bach2-WT diabetic mice but not the Bach2-CKO diabetic S7D), we found that knockdown of β-arrestin 2 by either
mice (Figure 5C–5E). The AM1241-decreased expres- siRNA reversed the translocation of CB2R from cellular
sion of Il-6 and Col1α1 was notable only in the Bach2- membrane to cytoplasm on HG treatment (Figure S7E).
WT diabetic mice (Figure 5D). Echocardiography showed These loss-of-function studies suggested that hyper-
that AM1241 improved ejection fraction and fractional glycemia induced cytoplasm translocation of CB2R in a
shortening in the Bach2-WT diabetic mice; however, it β-arrestin 2–dependent manner.
could not trigger beneficial effects in the Bach2-CKO It is interesting to note that the cytoplasmic CB2R
mice (Figure 5F and 5G). Histological studies showed merged well with MLKL on HG treatments (Figure 6F).
that AM1241 improved cardiac hypertrophy and inflam- Using the online Phosphorylation Analysis Tool,32 we
mation (hematoxylin and eosin staining) and fibrosis found that MLKL, but not RIP1 or RIP3, is a poten-
(Sirius red staining) and suppressed MLKL expression tial upstream kinase of human CB2R at serine 352
(immunohistochemistry staining) in the Bach2-WT mice, (S352), a site that is highly conserved across species

10 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 5. Cardiac conditional knockout of BACH2 aggravated diabetic heart dysfunction and dampened CB2R-mediated
cardioprotection.
A, Protein levels of necroptosis markers when cannabinoid 2 receptor (CB2R) and BACH2 were modulated as indicated in HL-1 cells. High-
glucose (HG; 30 mmol/L) treatment for 72 hours. B through H, Type 2 diabetes was induced in Bach2-CKO and Bach2–wild-type (WT) mice
with the high-fat diet (HFD)/streptozotocin (STZ) methods, and mice were intraperitoneally injected with AM1241 (1 mg/kg) for 12 continuous
weeks (n=6–7 per group). B, Schematic diagram showing the study design. C, Protein levels of BACH2, necroptosis, and cell death (FADD [Fas
associated via death domain]) markers in the mouse heart. D, mRNA levels of Bach2, Il-6, and Col1a1 in the mouse heart (n=6 per group). E,
mRNA levels of Rip1, Rip3, and Mlkl in the mouse heart (n=6 per group). F and G, Echocardiography of ejection fraction and fractional shortening
of each mouse. H, Hematoxylin and eosin (HE) staining, Sirius red staining, and immunohistochemistry (IHC) staining of MLKL (mixed lineage
kinase domain-like). Scale bar: 1 mm or 50 μmol/L. *P<0.05, as indicated.

(Figure 6G). We then homemade a phosphorylated increased the level of p-CB2R (S352) while decreas-
antibody targeting the S352 site of CB2R with con- ing the total level of CB2R and that MLKL indeed
siderable specificity (Figure S8). It showed that HG promoted the phosphorylation of CB2R (S352) in

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 11

 Gao et al CB2R-Centric Necroptotic Loop in DCM
ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 6. CB2R internalized by β-arrestin 2 and was phosphorylated at S352 and degraded at cytoplasm by MLKL in
hyperglycemic conditions.
A, Western blot analysis of cannabinoid 2 receptor (CB2R) in HL-1 cells treated with or without high glucose (HG; 30 mmol/L) for 48 hours.
B, CB2R protein levels in primary cardiomyocytes treated with or without HG (30 mmol/L) for 48 hours. C and D, Western blot analysis and
immunohistochemical staining of CB2R in the heart from db/db diabetic (DM) mice at different ages. E, Detection of CB2R in cell membrane and
cytoplasm isolated from HL-1 cells treated with HG (30 mmol/L) for 48 hours. F, Immunofluorescence staining of CB2R in HL-1 cells incubated
with HG (30 mmol/L) for 48 hours. G, Schematic illustration of potential kinases of CB2R at serine 352 (S352)32 and S352 conservation across
species (highlighted in red). H, A homemade phosphorylation antibody against CB2R at S352 was used to detect the phosphorylation of CB2R
when cells were stably overexpressed or knocked with human MLKL (mixed lineage kinase domain-like) in 293T cells. I, Phosphorylation (p-)
of CB2R (S352) was detected when human receptor-interacting protein kinase (RIP) 1, RIP3, or MLKL was individually overexpressed (OE)
with or without an MLKL inhibitor (Necrosulfonamide, 1 μmol/L) cotreatment in 293T cells. J and K, In the CB2R knockout (CB2R-KO) cells,
human MLKL overexpression plasmid was cotransfected with the human CB2R–wild-type (WT) plasmid or human CB2R-S352A plasmid.
Phosphorylation of CB2R (S352) and ubiquitination of CB2R were then detected. L, Dose and time effects of CB2R ubiquitination in response to
HG treatments in HL-1 cells. HE indicates hematoxylin and eosin; and IP, immunoprecipitation.

12 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

HEK-293T cells (Figure 6H). Furthermore, overex- DISCUSSION

pression of RIP1 or RIP3 also enhanced the phos-

ORIGINAL RESEARCH
The present study uncovered that RIP1/RIP3/MLKL
phorylation of CB2R individually but failed to do so
signaling–mediated necroptosis was responsible for
when Necrosulfonamide, a specific inhibitor of human

ARTICLE
diabetic heart dysfunction and identified CB2R as a
MLKL, was cotreated (Figure 6I), indicating that RIP1
critical regulator of the RIP1/RIP3/MLKL transcription
or RIP3 per se had no direct influence on the CB2R
by regulating the nuclear translocation of BACH2. We
phosphorylation. In the in vitro kinase assay using
proposed the CB2R-compromized necroptotic loop as a
CB2R knockout cells, MLKL directly promoted the
driver of diabetic heart injury (Figure 8).
phosphorylation of CB2R when transfected with a WT
Despite multiple preclinical studies and clinical tri-
human CB2R (hCB2R-WT) but failed to do so when
als, these therapeutic drugs have failed to demonstrate
transfected with the mutated human CB2R (hCB2R-
meaningful beneficial effects in patients with acute
S352A; Figure 6J). After phosphorylation by MLKL,
heart failure thus far.9,10 The time of starting the drug, for
CB2R ubiquitination was enhanced, and as a result,
example, has been considered a reason for poor clinical
its protein level decreased accordingly. However, when
outcome.10 A previous study showed that high-fat diet
CB2R S352 was mutated, these effects were no lon-
consumption significantly increased the end-diastolic
ger observed (Figure 6K). Moreover, CB2R ubiquitina-
pressure-volume relationship at 8 weeks and thereaf-
tion was observed to be dose- and time-dependently
ter.7 In a db/db diabetic mouse model, 16-week-old mice
promoted by HG treatments with its protein level
had echocardiographic evidence of well-established
accordingly decreasing (Figure 6L). These findings
cardiac dysfunction.33 In view that clinical intervention
suggested that MLKL in turn phosphorylated CB2R
would possibly be imposed after the diagnosis of heart
at the S352 site and led to CB2R degradation after
dysfunction in patients with diabetes, we started the cell
CB2R internalization.
death–targeted therapy in the 16-week-old db/db mice
to mimic the clinical scenario. We have gathered solid
CB2R and BACH2 Were Negatively Associated data showing that drug therapy initiated at 16 weeks
after hyperglycemia could still confer cardioprotection.
With Myocardial Injuries in Human Diabetic
At this particular time point, we further provided solid
Heart results showing that targeting necroptosis but rather
We analyzed human cardiac samples from deceased pa- apoptosis would be beneficial. Using a time-lapse
Downloaded from http://ahajournals.org by on December 6, 2022

tients with or without a history of diabetes. It showed that investigation, we showed that cardiomyocyte apoptosis
CB2R and BACH2 were notably decreased whereas occurred at an earlier stage (within 12 weeks). RIP1/
the cell death markers (TRADD and FADD) and necrop- RIP3/MLKL–mediated necroptosis occurred at a later
totic proteins were dramatically increased in the diabetic stage (≈16 weeks after diabetes). Furthermore, hyper-
hearts (Figure 7A). Masson trichrome staining showed glycemia activated cardiac necroptosis in both male
significant cardiac fibrosis in the diabetic hearts (Fig- and female mice across diabetes models. Nec-1, the
ure 7B and 7C). Immunohistochemistry staining showed specific inhibitor of RIP124 that was injected 16 weeks
that the protein levels of CB2R and BACH2 were sig- after diabetes, was shown to significantly ameliorate
nificantly decreased whereas that of MLKL was remark- diabetic heart injuries and to improve heart function.
ably increased in the diabetic decedents (Figure 7B and These data validated our hypothesis that drug interven-
7D–7F). These findings indicated that the diabetic con- tion in diabetes at a stage of particular cell death devel-
dition was associated with the expression of these pro- opment confers clinical benefits.
teins, although these associations might be confounded At present, necroptosis-specific inhibitors are under
by patient age because the patients with diabetes were development, and >20 drugs that have already been used
significantly older than the control subjects (Table S1). in the treatment of various diseases have the potential to
The human diabetic hearts showed well-merged signals be repurposing to regulate necroptosis.34 These potential
of RIP1 (purple), RIP3 (red), and MLKL (green) (Fig- drugs, however, have not shown their therapeutic pros-
ure 7G). The protein levels of CB2R, BACH2, and MLKL pects in diabetes and its related complications. Moreover,
did not differ across sexes in the diabetic decedents drugs directly targeting necroptosis might have draw-
(Figure 7H–7K), confirming that HG regulated CB2R backs concerning their clinical applicability. The alloste-
signaling independently of sex. Furthermore, expression ric RIP1-targeting drug Nec-1, for example, has been
of CB2R or BACH2 was significantly and negatively shown to have very limited metabolic stability (T1/2 <5
correlated with both MLKL protein levels and the extent minutes in mouse microsomal assay)35,36 and off-target
of cardiac fibrosis (Figure 7H–7K). These data obtained effects.37,38 Alternatively, to promote the clinical translat-
from human cases support that the CB2R/BACH2 sig- ability, we designed a GPCR-based drug screening based
naling was negatively associated with diabetes-induced on the notion that ≈40% of clinically used drugs are
cardiac necroptosis and myocardial injury in the clinic. GPCR targeted,25,26 and our systemic validation studies

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 13

 Gao et al CB2R-Centric Necroptotic Loop in DCM
ORIGINAL RESEARCH
ARTICLE
Downloaded from http://ahajournals.org by on December 6, 2022

Figure 7. CB2R and BACH2 were negatively associated with myocardial injuries in human diabetic heart.
A, Protein levels of cannabinoid 2 receptor (CB2R), BACH2, necroptosis, and cell death markers in human diabetic (DM) heart tissues (n=5) and
nondiabetic controls (Con; n=7). B, Hematoxylin and eosin (HE) staining, Masson trichrome staining, immunohistochemistry staining of CB2R,
BACH2, and MLKL (mixed lineage kinase domain-like) in diabetic heart slides and nondiabetic controls (n=25 per group). Scale bar: 5 or 10
μm. C, Quantification of cardiac fibrosis from the slides after Masson trichrome staining (n=25 per group). D through F, Quantification of CB2R,
BACH2, and MLKL staining intensity from the slides after immunohistochemistry staining. G, Immunofluorescence staining of receptor-interacting
protein kinase (RIP) 1 (purple), RIP3 (red), MLKL (green), and DAPI (blue) in human heart slides. H, Correlation of CB2R staining intensity with
MLKL protein levels in the 25 diabetic hearts. I, Correlation of BACH2 staining intensity with MLKL protein levels in the 25 diabetic hearts. J,
Correlation of CB2R staining intensity with extent of cardiac fibrosis in the 25 diabetic hearts. K, Correlation of BACH2 staining intensity with
extent of cardiac fibrosis in the 25 diabetic hearts. H through K, Dark red dots indicate male diabetic hearts (n=17); and light red dots, female
diabetic hearts (n=8). FADD indicates Fas associated via death domain; p-, phosphorylated; and TRADD, TNF receptor superfamily member 1a
associated via death domain. *P<0.05, as indicated.

confirmed that CB2R agonists are potent suppressors of is different from our previous publication17 and might be
necroptosis and diabetic heart injuries. CB1R, however, explained by the functional diversity or rivalry of cannabi-
seemed not to regulate necroptosis under HG conditions noid receptors.39–41
because its agonists and antagonists did not yield con- While uncovering the detailed mechanisms under-
sistent results in the GPCR-based drug screening. This lying CB2R-suppressed necroptosis transcription, we

14 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

the kinase activity of MLKL in human diseases. In fact,

other pseudokinases, for example, HER348 and JAK2,49

ORIGINAL RESEARCH
have also been reported to exert catalytic activities under
certain circumstances. It is noteworthy that CB2R in Mus

ARTICLE
musculus does not possess a S352 site, indicating that
there might be other undiscovered sites that are respon-
sible for MLKL-mediated phosphorylation and degrada-
tion of CB2R in mice. Collectively, these positive and
negative feedbacks, which we called a necroptotic loop
that was centric by CB2R, provide insight into our under-
standing of the regulatory network for necroptosis.
There are limitations in the human case studies. First,
the patients with diabetes were significantly older than
the control subjects. Therefore, the purported associa-
tions may be inevitably and inextricably limited by poten-
tial confounding factors related to age. Second, because
of the nature of the autopsy cohorts, some cases do
Figure 8. Schematic illustration of this study. not necessarily have full clinical biochemical data, which
Red lines represent the process under hyperglycemia. Gray lines might cause bias of data presentation. The addition of
represent the process under low glucose conditions. CB2R indicates clinical cases with ascertained diabetic heart injury might
cannabinoid 2 receptor; MLKL, mixed lineage kinase domain-
like; S6K, S6 kinase; ×, signal transduction did not flow under
address such limitations.
hyperglycemia; →, promotion; and −, suppression.

found that CB2R modified the coregulator BACH2.

Conclusions
BACH2 is a canonical transcription factor shaping We demonstrated a novel CB2R-compromized necrop-
the balance between CD4(+) T-cell subsets in innate totic loop as the critical driver of diabetic heart dys-
immune responses.42,43 Its role beyond immune system function. CB2R recruits a critical coregulator, BACH2,
has remained largely unknown. We provided concrete a previously unappreciated transcription factor in car-
data that CB2R inhibited the S6K-dependent phosphor- diomyocytes, to transcriptionally repress the expression
Downloaded from http://ahajournals.org by on December 6, 2022

ylation of BACH2 in cardiomyocyte cytoplasm, leading of necroptosis. Our data suggest that pharmacologi-
to its nucleus translocation, which consequently evoked cal activators of CB2R or BACH2 represent alterna-
transcription suppression of necroptosis. Genetic tive therapeutic agents in diabetes-induced myocardial
knockout of Bach2 in cardiomyocytes dampened CB2R necroptosis in advanced stages.
agonist–mediated cardioprotection in diabetes models,
further verifying the inevitable role of BACH2 in col-
laborating CB2R. ARTICLE INFORMATION
It is interesting that CB2R was downregulated in dia- Received June 8, 2022; accepted October 3, 2022.
betic heart but remained at high levels in nondiabetic
Affiliations
heart tissues. HG activated β-arrestin 2 to induce CB2R
Shanghai Institute of Cardiovascular Diseases, Zhongshan Hospital and Institutes
translocation to cytoplasm, where HG-activated MLKL of Biomedical Sciences, Fudan University, China (F.G., M.C., X.J., X.W., G.Z., C.Y.,
further directly phosphorylated CB2R at S352 and pro- J.G., Y.Z.). Department of Forensic Medicine, School of Basic Medical Sciences,
Fudan University, Shanghai, China (X.T., L.L.). Department of Cardiovascular Medi-
moted ubiquitin-dependent degradation of CB2R in
cine, University of Tokyo Graduate School of Medicine, Japan (I.K.).
human cells (Figure 8). Unlike the G protein regulatory
kinase–dependent phosphorylation that controls CB2R Acknowledgments
internalization,44 the MLKL-dependent phosphorylation Bach2flox/flox mic were a kind gift from Dr Chuanxin Huang at the Shanghai Jiao
Tong University School of Medicine. The authors are grateful to Drs Junyi Lin and
of CB2R represents a novel mechanism accounting for Yu Shao from the Department of Forensic Medicine, School of Basic Medical
CB2R degradation in cytoplasm. Traditionally, MLKL is Sciences, Fudan University, for collection of some human cases. They also thank
classified as a pseudokinase because of the lack of 2 Dr Jun Ren from the Shanghai Institute of Cardiovascular Diseases, Zhongshan
Hospital Fudan University, for his valuable suggestions on the Discussion. P.G.,
conserved catalytic residues His-Arg-Asp (HRD) and J.G., L.L., and Y.Z. designed the project and provided funding. J.G. and I.K. provided
Asp-Phe-Gly (DFG), which are responsible for phosphoryl critical comments to the Discussion. P.G., M.C., X.J., X.W., and X.T. performed the
transfer activity, in the kinase-like domain.45,46 We found in vivo experiments and parts of the in vitro experiments. G.Z. and C.Y. performed
flow cytometry assays. L.L. performed electron microscopy and time-lapse mi-
that human MLKL protein could strikingly exert catalytic croscopy. All authors revised and approved the manuscript.
kinase activity to directly regulate CB2R, a phenomenon
that is rare and beyond our knowledge of MLKL. The Sources of Funding
This work was financially supported by the National Natural Science Founda-
fact that Necrosulfonamide effectively inhibits only the tion of China (82000271, 81730009, and 82070285), Shanghai Rising-Star
activity of human MLKL, not mouse MLKL,47 also implies Program (21QA1401700), Scientific Research Project of Shanghai Health

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 15

 Gao et al CB2R-Centric Necroptotic Loop in DCM

Committee (20204Y0225), and Innovative Research Team of High-Level Local 18. Vanden Berghe T, Linkermann A, Jouan-Lanhouet S, Walczak H,
Universities in Shanghai and a key laboratory program of the Education Commis- Vandenabeele P. Regulated necrosis: the expanding network of non-
ORIGINAL RESEARCH

sion of Shanghai Municipality (ZDSYS14005). apoptotic cell death pathways. Nat Rev Mol Cell Biol. 2014;15:134–146.
doi: 10.1038/nrm3737
Disclosures 19. Weinlich R, Oberst A, Beere HM, Green DR. Necroptosis in development,
ARTICLE

None. inflammation and disease. Nat Rev Mol Cell Biol. 2017;18:127–136. doi:
10.1038/nrm.2016.149
Supplemental Material 20. Fritsch M, Gunther SD, Schwarzer R, Albert MC, Schorn F, Werthenbach
Expanded Methods JP, Schiffmann LM, Stair N, Stocks H, Seeger JM, et al. Caspase-8 is
Tables S1–S12 the molecular switch for apoptosis, necroptosis and pyroptosis. Nature.
2019;575:683–687. doi: 10.1038/s41586-019-1770-6
Figures S1–S8
21. National Genomics Data Center. Accessed xxxx. https://bigd.big.ac.cn/
Excel Files S1–S3
22. Gao P, Li LL, Yang L, Gui DK, Zhang JR, Han JF, Wang JJ, Wang NS, Lu
JX, Chen SZ, et al. Yin Yang 1 protein ameliorates diabetic nephropathy
pathology through transcriptional repression of TGF beta 1. Sci Transl Med.
2019;11:eaaw2050. doi: 10.1126/scitranslmed.aaw2050
REFERENCES 23. Cai Z, Jitkaew S, Zhao J, Chiang HC, Choksi S, Liu J, Ward Y, Wu LG, Liu
1. Unnikrishnan R, Anjana RM, Mohan V. Diabetes mellitus and its compli- ZG. Plasma membrane translocation of trimerized MLKL protein is re-
cations in India. Nat Rev Endocrinol. 2016;12:357–370. doi: 10.1038/ quired for TNF-induced necroptosis. Nat Cell Biol. 2014;16:55–65. doi:
nrendo.2016.53 10.1038/ncb2883
2. Afshin A, Forouzanfar MH, Reitsma MB, Sur P, Estep K, Lee A, Marczak L, 24. Degterev A, Hitomi J, Germscheid M, Ch’en IL, Korkina O, Teng X, Abbott
Mokdad AH, Moradi-Lakeh M, Naghavi M, et al; GBD 2015 Obesity Collab- D, Cuny GD, Yuan C, Wagner G, et al. Identification of RIP1 kinase as a
orators. Health effects of overweight and obesity in 195 countries over 25 specific cellular target of necrostatins. Nat Chem Biol. 2008;4:313–321.
years. N Engl J Med. 2017;377:13–27. doi: 10.1056/NEJMoa1614362 doi: 10.1038/nchembio.83
3. Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 dia- 25. Hauser AS, Attwood MM, Rask-Andersen M, Schioth HB, Gloriam DE.
betes mellitus and its complications. Nat Rev Endocrinol. 2018;14:88–98. Trends in GPCR drug discovery: new agents, targets and indications. Nat
doi: 10.1038/nrendo.2017.151 Rev Drug Discov. 2017;16:829–842. doi: 10.1038/nrd.2017.178
4. De Hert M, Detraux J, van Winkel R, Yu W, Correll CU. Metabolic and car- 26. Yang D, Zhou Q, Labroska V, Qin S, Darbalaei S, Wu Y, Yuliantie E, Xie L,
diovascular adverse effects associated with antipsychotic drugs. Nat Rev Tao H, Cheng J, et al. G protein-coupled receptors: structure- and func-
Endocrinol. 2011;8:114–126. doi: 10.1038/nrendo.2011.156 tion-based drug discovery. Signal Transduct Target Ther. 2021;6:7. doi:
5. Murtaza G, Virk HUH, Khalid M, Lavie CJ, Ventura H, Mukherjee D, Ramu 10.1038/s41392-020-00435-w
V, Bhogal S, Kumar G, Shanmugasundaram M, et al. Diabetic cardiomy- 27. Muto A, Hoshino H, Madisen L, Yanai N, Obinata M, Karasuyama H, Hayashi
opathy: a comprehensive updated review. Prog Cardiovasc Dis. 2019;62: N, Nakauchi H, Yamamoto M, Groudine M, et al. Identification of Bach2
315–326. doi: 10.1016/j.pcad.2019.03.003 as a B-cell-specific partner for small maf proteins that negatively regu-
6. Jia G, Hill MA, Sowers JR. Diabetic cardiomyopathy: an update of mecha- late the immunoglobulin heavy chain gene 3’ enhancer. EMBO J. 1998;
nisms contributing to this clinical entity. Circ Res. 2018;122:624–638. doi: 17:5734–5743. doi: 10.1093/emboj/17.19.5734
10.1161/CIRCRESAHA.117.311586 28. Yoshida C, Yoshida F, Sears DE, Hart SM, Ikebe D, Muto A, Basu S,
7. Tong M, Saito T, Zhai P, Oka SI, Mizushima W, Nakamura M, Ikeda S, Igarashi K, Melo JV. Bcr-Abl signaling through the PI-3/S6 kinase
Downloaded from http://ahajournals.org by on December 6, 2022

Shirakabe A, Sadoshima J. Mitophagy is essential for maintaining cardiac pathway inhibits nuclear translocation of the transcription factor Bach2,
function during high fat diet-induced diabetic cardiomyopathy. Circ Res. which represses the antiapoptotic factor heme oxygenase-1. Blood.
2019;124:1360–1371. doi: 10.1161/CIRCRESAHA.118.314607 2007;109:1211–1219. doi: 10.1182/blood-2005-12-040972
8. Tan Y, Zhang Z, Zheng C, Wintergerst KA, Keller BB, Cai L. Mechanisms 29. Pfleger J, Gresham K, Koch WJ. G protein-coupled receptor kinases as
of diabetic cardiomyopathy and potential therapeutic strategies: pre- therapeutic targets in the heart. Nat Rev Cardiol. 2019;16:612–622. doi:
clinical and clinical evidence. Nat Rev Cardiol. 2020;17:585–607. doi: 10.1038/s41569-019-0220-3
10.1038/s41569-020-0339-2 30. Turu G, Soltesz-Katona E, Toth AD, Juhasz C, Cserzo M, Misak A, Balla
9. Becher PM, Fluschnik N, Blankenberg S, Westermann D. Challenging as- A, Caron MG, Hunyady L. Biased coupling to beta-arrestin of two com-
pects of treatment strategies in heart failure with preserved ejection frac- mon variants of the CB2 cannabinoid receptor. Front Endocrinol (Lausanne).
tion: “why did recent clinical trials fail?” World J Cardiol. 2015;7:544–554. 2021;12:714561. doi: 10.3389/fendo.2021.714561
doi: 10.4330/wjc.v7.i9.544 31. Chen X, Zheng C, Qian J, Sutton SW, Wang Z, Lv J, Liu C, Zhou N. In-
10. Voors AA, van Veldhuisen DJ. Why do drugs for acute heart failure fail? Eur volvement of beta-arrestin-2 and clathrin in agonist-mediated internaliza-
J Heart Fail. 2012;14:955–956. doi: 10.1093/eurjhf/hfs122 tion of the human cannabinoid CB2 receptor. Curr Mol Pharmacol. 2014;
11. Wang YJ, Gao EH, Tao L, Lau WB, Yuan YX, Goldstein BJ, Lopez BL, 7:67–80. doi: 10.2174/1874467207666140714115824
Christopher TA, Tian R, Koch W, et al. AMP-activated protein kinase de- 32. Phosphonet. Accessed xxxx. www.phosphoNet.ca
ficiency enhances myocardial ischemia/reperfusion injury but has minimal 33. Lew JK, Pearson JT, Saw E, Tsuchimochi H, Wei M, Ghosh N, Du CK,
effect on the antioxidant/antinitrative protection of adiponectin. Circulation. Zhan DY, Jin M, Umetani K, et al. Exercise regulates microRNAs to pre-
2009;119:835–844. doi: 10.1161/circulationaha.108.815043 serve coronary and cardiac function in the diabetic heart. Circ Res. 2020;
12. Li L, Gao P, Tang X, Liu Z, Cao M, Luo R, Li X, Wang J, Lin X, Peng 127:1384–1400. doi: 10.1161/CIRCRESAHA.120.317604
C, et al. CB1R-stabilized NLRP3 inflammasome drives antipsychotics 34. Molnar T, Mazlo A, Tslaf V, Szollosi AG, Emri G, Koncz G. Current transla-
cardiotoxicity. Signal Transduct Target Ther. 2022;7:190. doi: 10.1038/ tional potential and underlying molecular mechanisms of necroptosis. Cell
s41392-022-01018-7 Death Dis. 2019;10:860. doi: 10.1038/s41419-019-2094-z
13. Taylor RC, Cullen, S, Martin SJ. Apoptosis: controlled demolition at the cel- 35. Degterev A, Maki JL, Yuan J. Activity and specificity of necrostatin-1, small-
lular level. Nat Rev Mol Cell Biol. 2008;9:231–241. doi: 10.1038/nrm2312 molecule inhibitor of RIP1 kinase. Cell Death Differ. 2013;20:366. doi:
14. Pasparakis M, Vandenabeele P. Necroptosis and its role in inflammation. 10.1038/cdd.2012.133
Nature. 2015;517:311–320. doi: 10.1038/nature14191 36. Teng X, Degterev A, Jagtap P, Xing X, Choi S, Denu R, Yuan J, Cuny GD.
15. Grootjans S, Vanden Berghe T, Vandenabeele P. Initiation and execution Structure-activity relationship study of novel necroptosis inhibitors. Bioorg
mechanisms of necroptosis: an overview. Cell Death Differ. 2017;24:1184– Med Chem Lett. 2005;15:5039–5044. doi: 10.1016/j.bmcl.2005.07.077
1195. doi: 10.1038/cdd.2017.65 37. Takahashi N, Duprez L, Grootjans S, Cauwels A, Nerinckx W, DuHadaway
16. Krysko DV, Garg AD, Kaczmarek A, Krysko O, Agostinis P, Vandenabeele JB, Goossens V, Roelandt R, Van Hauwermeiren F, Libert C, et al. Necro-
P. Immunogenic cell death and DAMPs in cancer therapy. Nat Rev Cancer. statin-1 analogues: critical issues on the specificity, activity and in vivo use
2012;12:860–875. doi: 10.1038/nrc3380 in experimental disease models. Cell Death Dis. 2012;3:e437–e437. doi:
17. Li XQ, Peng Z, Zhou YL, Wang J, Lin XY, Dong XR, Liu XC, Jiang JQ, Jiang 10.1038/cddis.2012.176
Y, Li LL. Quetiapine induces myocardial necroptotic cell death through 38. Amgalan D, Chen Y, Kitsis RN. Death receptor signaling in the heart: cell
bidirectional regulation of cannabinoid receptors. Toxicol Lett. 2019; survival, apoptosis, and necroptosis. Circulation. 2017;136:743–746. doi:
313:77–90. doi: 10.1016/j.toxlet.2019.06.005 10.1161/CIRCULATIONAHA.117.029566

16 January 10, 2023 Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304

 Gao et al CB2R-Centric Necroptotic Loop in DCM

39. Li L, Dong X, Tu C, Li X, Peng Z, Zhou Y, Zhang D, Jiang J, Burke A, Zhao 45. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S. The pro-
Z, et al. Opposite effects of cannabinoid CB1 and CB2 receptors on anti- tein kinase complement of the human genome. Science. 2002;298:

ORIGINAL RESEARCH
psychotic clozapine-induced cardiotoxicity. Br J Pharmacol. 2019;176:890– 1912–1934. doi: 10.1126/science.1075762
905. doi: 10.1111/bph.14591 46. Murphy JM, Czabotar PE, Hildebrand JM, Lucet IS, Zhang JG, Alvarez-
40. Cinar R, Iyer MR, Kunos G. The therapeutic potential of second and third Diaz S, Lewis R, Lalaoui N, Metcalf D, Webb AI, et al. The pseudokinase

ARTICLE
generation CB1R antagonists. Pharmacol Ther. 2020;208:107477. doi: MLKL mediates necroptosis via a molecular switch mechanism. Immunity.
10.1016/j.pharmthera.2020.107477 2013;39:443–453. doi: 10.1016/j.immuni.2013.06.018
41. Tang XR, Liu Z, Li XQ, Wang J, Li LL. Cannabinoid receptors in myocar- 47. Sun L, Wang H, Wang Z, He S, Chen S, Liao D, Wang L, Yan J, Liu W,
dial injury: a brother born to rival. Int J Mol Sci . 2021;22. doi: 10.3390/ Lei X, et al. Mixed lineage kinase domain-like protein mediates necro-
ijms22136886 sis signaling downstream of RIP3 kinase. Cell. 2012;148:213–227. doi:
42. Yang L, Chen S, Zhao Q, Sun Y, Nie H. The critical role of Bach2 in shaping 10.1016/j.cell.2011.11.031
the balance between CD4(+) T cell subsets in immune-mediated diseases. 48. Shi FM, Telesco SE, Liu YT, Radhakrishnan R, Lemmon MA. ErbB3/
Mediators Inflamm. 2019;2019:2609737. doi: 10.1155/2019/2609737 HER3 intracellular domain is competent to bind ATP and catalyze auto-
43. Roychoudhuri R, Hirahara K, Mousavi K, Clever D, Klebanoff CA, Bonelli phosphorylation. Proc Natl Acad Sci U S A. 2010;107:7692–7697. doi:
M, Sciume G, Zare H, Vahedi G, Dema B, et al. BACH2 represses effec- 10.1073/pnas.1002753107
tor programs to stabilize T(reg)-mediated immune homeostasis. Nature. 49. Ungureanu D, Wu JH, Pekkala T, Niranjan Y, Young C, Jensen ON, Xu
2013;498:506–510. doi: 10.1038/nature12199 CF, Neubert TA, Skoda RC, Hubbard SR, et al. The pseudokinase do-
44. Franklin JM, Carrasco GA. G-protein receptor kinase 5 regulates the can- main of JAK2 is a dual-specificity protein kinase that negatively regu-
nabinoid receptor 2-induced up-regulation of serotonin 2A receptors. J Biol lates cytokine signaling. Nat Struct Mol Biol. 2011;18:971–976. doi:
Chem. 2013;288:15712–15724. doi: 10.1074/jbc.M113.454843 10.1038/nsmb.2099
Downloaded from http://ahajournals.org by on December 6, 2022

Circulation. 2023;147:00–00. DOI: 10.1161/CIRCULATIONAHA.122.059304 January 10, 2023 17