Mirabegron displays anticancer effects by globally browning adipose tissues

Metabolic reprogramming in malignant cells is a hallmark of cancer that relies on augmented glycolytic metabolism to support their growth, invasion, and metastasis. However, the impact of global adipose metabolism on tumor growth and the drug development by targeting adipose metabolism remain largely unexplored. Here we show that a therapeutic paradigm of drugs is effective for treating various cancer types by browning adipose tissues. Mirabegron, a clinically available drug for overactive bladders, displays potent anticancer effects in various animal cancer models, including untreatable cancers such as pancreatic ductal adenocarcinoma and hepatocellular carcinoma, via the browning of adipose tissues. Genetic deletion of the uncoupling protein 1, a key thermogenic protein in adipose tissues, ablates the anticancer effect. Similarly, the removal of brown adipose tissue, which is responsible for non-shivering thermogenesis, attenuates the anticancer activity of mirabegron. These findings demonstrate that mirabegron represents a paradigm of anticancer drugs with a distinct mechanism for the effective treatment of multiple cancers.

Beyond CL-316,243, norepinephrine is the endogenous neurotransmifter and hormone, and it is the actual physiological acfivator of BAT thermogenesis.Isoproterenol is a non-specific β-AR agonist that matches nearly all of the effects of mirabegron, and then some.Meanwhile, propranolol is a beta-AR antagonist that blocks mirabegron's effects and is useful in establishing its pharmacology.L-748,328 and L-748,337 are β3-AR specific and can help determine which adrenergic receptors are mediafing anfitumor acfivity.
The authors need to repeat several of their experiments with proper posifive and negafive controls to show whether mirabegron alone has anfi-tumor acfivity, or if the effect is a class effect seen in betaadrenergic agonists.Examples of helpful studies: 1. Head-to-head-to-head comparisons of mirabegron, 243,norepinephrine,and isoproterenol,in vitro and in vivo. 2. Dose-response studies of mirabegron's effects.3. Inhibitor studies using mirabegron and propranolol and also L-748,328 or L-748,337 as specific and non-specific β3-AR inhibitors, respecfively.

B. Mechanism
The authors propose that "mirabegron-induced NST results in the downregulafion of mulfiple metabolic pathways in tumors."A diagram is shown in While some of these mechanisms can be tested in future experiments, some have already been done in this study.Figs 2C-E show that blood glucose in the mirabegron-treated mice is lower, but certainly within the normal range.Do the authors believe that a normal fasfing glucose is too low for tumors to grow?
Opfion (e) could be tested by measuring blood lipids.If they are unchanged, then (d) is the only remaining mechanism, the alleviafion of hypoxia; the authors should then explain how mirabegron treatment would lead to this.C. BAT thermogenesis C1.Fig. S6 argues that UCP1 deficiency abrogates mirabegron-induced adipose fissue browning.It is not clear why this would be as UCP1 is downstream and part of a separate pathway inifiated by adrenergic signaling.By what mechanism is this happening?C2.Kazak and Spiegelman have published numerous studies showing that UCP1 is only one of several fufile cycles that leads to thermogenesis in BAT.One notable alternafive is fufile creafine cycling.Meanwhile, UCP1-KO leads to disrupted mitochondrial structure and funcfion beyond uncoupling and thermogenesis.The authors need addifional studies to show that a loss of uncoupling/fufile cycles is what leads to the reported anfi-tumor effects.

Reviewer #2 (Remarks to the Author):
The authors addressed all of my concerns convincingly

Reviewer #3 (Remarks to the Author):
With their revisions, Sun et al have been responsive to the previous round of reviews, significantly increasing the "interpretability" of the manuscript by updafing many of the figures to provide actual levels of growth factors.Despite these improvements having removed the human data from this version of the manuscript it is difficult to say if the current version can support the authors largest claims as wriften.
Figure 1 clearly shows that with Mirabegron treatment that several tumor models show decreased growth rates (not as stated inhibifion, just decreased growth rates).
Figure 2 demonstrates that Mirabegron treatment increases UPC1 expression in adipose fissues, along with a decrease in adipocyte cell size in WAT and BAT.They show that treatment decreases blood glucose and insulin levels while increasing metabolic rate as gauged by O2 consumpfion.
Figure 3 they show that if they remove the BAT that they significantly reduce the impact of Mirabegron treatment upon tumor growth, their markers of choice (Ki67, CA9, CC3), and glucose.Despite requests in the last round of review they did not address the changes in insulin nor c-pepfide in these models.
Figure 4 they argue that Mirabegron treatment inhibits glucose metabolism in the tumor fissue, but they do not control for the impact of insulin/glucose changes in this data set as (again) insulin and glucose levels are associated with the changes they demonstrate in glucose regulafing enzymes in the tumor.This feature is not controlled for in their experimental design.This is a key feature since they go on to show that exogenous glucose reduces the impact of Mirabegron in figure 5. Similarly they show that UCP1 delefion abrogates the impact of Mirabegron in figure 6.
For both figures 5 and 6 the pre-requisite matched controls are not being run in the same experiment.E.g. in figure 5 we only see mice with 15%HG and no experimental control for the effects is shown within the experiment; similarly there is no syngeneic control for the UCP1 experiments in figure 6.This makes it difficult to gauge the impact of their perturbafions as shown/argued.non-specific β-AR agonist that matches nearly all of the effects of mirabegron, and then some.Meanwhile, propranolol is a beta-AR antagonist that blocks mirabegron's effects and is useful in establishing its pharmacology.L-748,328 and L-748,337 are β3-AR specific and can help determine which adrenergic receptors are mediating anti-tumor activity.
The authors need to repeat several of their experiments with proper positive and negative controls to show whether mirabegron alone has anti-tumor activity, or if the effect is a class effect seen in beta-adrenergic agonists.Examples of helpful studies: 1. Head-to-head-to-head comparisons of mirabegron, CL-316,243, norepinephrine, and isoproterenol, in vitro and in vivo.

Response:
We highly appreciate the reviewer for his/her authoritative advice, which is extremely helpful for us to understand the anticancer effect of mirabegron and AR agonists.
On the basis of the reviewer´s suggestion, we have performed new experiments to allow us to head-to-head compare anticancer effects of mirabegron, CL-316,243, and isoproterenol, in vitro and in vivo.In the in vitro experiments, mirabegron, CL-316,243 and isoproterenol had no effect on tumor cell growth (See below Fig. R1).These data show that these adrenergic receptor activators have no direct impact on tumor cell proliferation in vitro.These new results have been incorporated into Fig.S3 of the revised manuscript.In the in vivo experiments, mirabegron, CL-316,243 and isoproterenol significantly inhibited tumor growth, indicating the activation of the sympathetic system by these β3-AR is required (See below Fig. R2).These new results have been incorporated into Fig.S3 of the revised manuscript.Comment: 2. Dose-response studies of mirabegron's effects.
Response: On the basis of the reviewer´s suggestion, we have performed new experiments to study the dose-dependent effect of mirabegron on tumor suppression.In the CRC tumor-bearing mice, the dose of 0.8 mg/kg daily lacked a statistically significant anticancer effect.At the dose of 8 mg/kg daily, mirabegron reached the maximal anticancer effect, and further increasing the dosage to 10 mg/kg daily did not enhance the anticancer effect beyond that of 8 mg/kg daily (Fig. R3).These new results have been incorporated into Fig.S1 of the revised manuscript.but certainly within the normal range.Do the authors believe that a normal fasting glucose is too low for tumors to grow?
Option (e) could be tested by measuring blood lipids.If they are unchanged, then (d) is the only remaining mechanism, the alleviation of hypoxia; the authors should then explain how mirabegron treatment would lead to this.

Response:
We thank the reviewer for this important comment.In general, we believe that all listed mechanisms may play a role in mirabegron-inhibited tumor growth, although further studies are needed to provide supportive evidence.
Using 18 F-FDG, a glucose analogue, we have shown that activated BAT significantly increased glucose uptake and usage, whereas tumor decreased glucose utilization (Fig. 2B).Metabolomics and RNA-seq further confirmed the decreased utilization of glucose in solid tumors.As the reviewer pointed out, the reduced blood glucose levels are within the normal range, and we agree that blood glucose reduction alone may not be sufficient to inhibit tumor growth.Broader mechanisms may be involved.For example, downregulation of GLUT1 as the key limiting step of glucose utilization in tumor cells (Fig. R5).It is highly plausible that downregulation of GLUT1 is the key mechanism for the impairment of the glycolytic pathway in tumors, even though the blood glucose levels are within normal range.Therefore, complex mechanisms are likely to be involved in impaired tumor growth.

Fig.R5.
A. mRNA level of the glucose transporter-related genes in vehicle-and mirabegron-treated mice (n = 6 samples per group).
Regarding the alteration of lipid metabolism, we have performed a new analysis of the metabolic pathways for non-glycolytic energy supplies.Reduction patterns were evident in metabolites such as acyl-CoA and enzymes related to lipid metabolism (Fig. R7).This would support the possible mechanism (e).These new results have been incorporated into Fig.S5 of the revised manuscript.7 Concerning the alleviation of tumor hypoxia by β3-AR activation, there are several possible mechanisms, including 1) delayed tumor growth rate; and 2) reduced accumulation of metabolites owing to impaired glycolysis.This issue has now been discussed in the revised manuscript.
It is not clear why this would be as UCP1 is downstream and part of a separate pathway initiated by adrenergic signaling.By what mechanism is this happening?
Response: This is an excellent question.At the time of this writing, we have no good explanation for this excellent question.However, we speculate that there might be a feedback mechanism between UCP1 activation and the browning phenotype.One of the possibilities is that genetic deletion of UCP1 disrupts mitochondrial structure, which is the key determinant for adipose browning.It is possible that the disrupted mitochondria show reduced expression of other mitochondrial markers, including COX4, which was used in our studies.Through this mechanism, the UCP1-deficient adipose tissues exhibit reduced browning phenotype, even though UCP1 is a downstream component.This important issue warrants further in-depth mechanistic studies.
Comment: C2.Kazak and Spiegelman have published numerous studies showing that UCP1 is only one of several futile cycles that leads to thermogenesis in BAT.One notable alternative is futile creatine cycling.Meanwhile, UCP1-KO leads to disrupted mitochondrial structure and function beyond uncoupling and thermogenesis.The authors need additional studies to show that a loss of uncoupling/futile cycles is what leads to the reported anti-tumor effects.

Response:
We thank the reviewer for this important comment.Indeed, additional studies to show that a loss of uncoupling/futile cycles is what leads to the reported anti-tumor effects are important.However, genetic deletion of the critical components of the futile creatine cycle is not feasible for us to do at this time.The main reason for not being feasible is that we do not have these genetic models in our laboratory.Importing and obtaining ethical permission for using these genetic models will take very long time (estimated more than one year).We do not exclude the possibility that the futile creatine cycle-related thermogenesis participates in tumor suppression.The central focus of this study is the discovery of a new anticancer drugs.Genetic deletion of UCP1 already produced marked neutralization of the mirabegroninduced tumor suppression.While this beneficial effect could be related to structural changes of mitochondria, these are positive and rescue experiments.We also backed up with surgical removal of adipose depots.Given the unavailability of these genetic models and focus of this study on cancer suppression, we discussed this issue in the Discussion section.We thank the reviewer for his or her understanding.mirabegron-or vehicle-treated tumor-bearing mice.Non-fasting blood insulin is not altered by mirabegron treatment.Fasting blood insulin in the mirabegron group was reduced by 10% relative the control group (Fig. R8).We also measured both fasting blood insulin and fasting blood c-peptide.In the vehicle-treated tumor-bearing WT and Ucp1 -/-mice, no difference in blood insulin levels was observed.However, mirabegron significantly reduced the blood insulin levels in WT mice.Interestingly, knockout of Ucp1 completely rescued the insulin levels (Fig. R9), suggesting mirabegronreduced blood insulin is dependent on UCP1.
Blood c-peptide levels in mirabegron-and vehicle-treated WT and Ucp1 -/-mice were altered in the same manner as blood insulin levels (Fig. R9).Thus, alterations of insulin and c-peptide match each other in our models.
In mirabegron-or vehicle-treated sham-operated or BAT-removed mice, we obtained similar results as Ucp1 -/-mice (Fig. R9).However, BAT removal did not completely rescue the mirabegron-reduced blood insulin/c-peptide levels (Fig. R9), suggesting that browning WAT may also play role in reducing blood insulin and c-peptide.
These new experimental results have been included in Figs. 2, 3, and 6 of the revised manuscript.Comment: Figure 4 they argue that Mirabegron treatment inhibits glucose metabolism in the tumor tissue, but they do not control for the impact of insulin/glucose changes in this data set as (again) insulin and glucose levels are associated with the changes they demonstrate in glucose regulating enzymes in the tumor.This feature is not controlled for in their experimental design.This is a key feature since they go on to show that exogenous glucose reduces the impact of Mirabegron in Figure 5. Similarly they show that UCP1 deletion abrogates the impact of Mirabegron in Figure 6.

Response:
We thank the reviewer for raising this issue.Indeed, blood glucose and insulin are critical in the glycolytic metabolism in promoting tumor growth.As shown above, since mirabegron downregulates blood insulin levels, we next investigated whether the antitumor effect is insulin-dependent.Vehicle-or mirabegron-treated tumor-bearing mice were treated three times a day with exogenous insulin at the dose of 2 IU/kg.As expected, the administration of insulin has justified to an similar level of the fasting blood insulin in the vehicle-and mirabegron-treated tumor-bearing mice (Fig. R10).Despite the exogenous administration of insulin to reach high blood levels, high blood insulin levels had little impact on tumor growth in the mirabegron-treated tumor-bearing mice (Fig. R10).To further elaborate the role of insulin in affecting tumor growth, we performed a loss-offunction experiment by impairing insulin production using streptozotocin (STZ) (45 mg/kg) to treat tumor-bearing mice (Fig. R11).At the end of day 5 after STZ injection, the fasting blood insulin levels were reduced by over 60% (Fig. R11).However, low levels of blood insulin did not retard tumor growth rates, albeit a slightly increased tumor growth rate relative to the control group was observed (Fig. R11).These data indicate that mirabegron-mediated tumor suppression is largely independent from the blood insulin effect.
Fig. S6.Potenfial mechanisms they offer include the following: a. Reduced glucose supply in tumors; b.Trimmed glucose transportafion in tumor cells; c.Reduced glucose metabolism in tumor cells; d.Alleviafion of tumor hypoxia; e. Downregulated lipid metabolism in tumor cells

Fig
Fig. R8. A. Non-fast and fast blood insulin level of Vehicle-and Mirabegron-treated PDAC (n = 3 mice per group).

Fig
Fig. R9. A. Fast blood insulin and c-peptide level of Vehicle-and Mirabegron-treated CRC tumorbearing WT and Ucp1 -/-mice (n = 8 mice per group).B. Fast blood insulin and c-peptide level of sham and BAT removal tumor-bearing mice with or without mirabegron treatment (n = 8 mice per group).

Fig
Fig. R10. A. Fast blood insulin level at 1h after insulin injection of Vehicle-and Mirabegron-treated PDAC with or without insulin (n = 3 mice per group).B. Tumor growth and tumor inhibition ratios of Vehicle-and Mirabegron-treated PDAC with or without insulin (n = 3 mice per group).