Combination of PD-1 Checkpoint Blockade and Botulinum Toxin Type A1 Improves Antitumor Responses in Mouse Tumor Models of Melanoma and Colon Carcinoma

ABSTRACT Background Tumor innervation has been shown to be utilized by some solid cancers to support tumor initiation, growth, progression, and metastasis, as well as confer resistance to immune checkpoint blockade through suppression of antitumor immunologic responses. Since botulinum neurotoxin type A1 (BoNT/A1) blocks neuronal cholinergic signaling, its potential use as an anticancer drug in combination with anti-PD-1 therapy was investigated in four different syngeneic mouse tumor models. Methods Mice implanted with breast (4T1), lung (LLC1), colon (MC38), and melanoma (B16-F10) tumors were administered a single intratumoral injection of 15 U/kg BoNT/A1, repeated intraperitoneal injections of 5 mg/kg anti-PD-1 (RMP1-14), or both. Results Compared to the single-agent treatments, anti-PD-1 and BoNT/A1 combination treatment elicited significant reduction in tumor growth among B16-F10 and MC38 tumor-bearing mice. The combination treatment also lowered serum exosome levels in these mice compared to the placebo control group. In the B16-F10 syngeneic mouse tumor model, anti-PD-1 + BoNT/A1 combination treatment lowered the proportion of MDSCs, negated the increased proportion of Treg cells, and elicited a higher number of tumor-infiltrating CD4+ and CD8+ T lymphocytes into the tumor microenvironment compared to anti-PD-1 treatment alone. Conclusion Our findings demonstrate the synergistic antitumor effects of BoNT/A1 and PD-1 checkpoint blockade in mouse tumor models of melanoma and colon carcinoma. These findings provide some evidence on the potential application of BoNT/A1 as an anticancer drug in combination with immune checkpoint blockade and should be further explored.


Introduction
Immune checkpoint blockade (ICB) is a novel cancer treatment approach that directs the patient's own immune system to recognize and eliminate cancer cells (Pico de Coaña et al. 2015). Immune checkpoints function to maintain immune homeostasis by acting as inhibitory regulators of T cell activation during strong immunologic responses to infections (Riva and Chokshi 2018;Zhang and Zheng 2020). Unfortunately, some cancers exploit the immune checkpoint pathway to evade immune surveillance (Qin et al. 2019;Vinay et al. 2015;Wang et al. 2016). Immune checkpoint inhibitors (ICIs) disrupt the interaction between coinhibitory signaling molecules, thereby allowing the activation and effector functions of T cells to proceed unimpeded (Wei et al. 2018).
Since 2011, the US Food and Drug Administration has approved a total of seven ICIs as monotherapy or in combination with other anticancer treatment modalities for 19 types of cancer. These include monoclonal antibodies α-CTLA-4 (ipilimumab), α-PD-1 (nivolumab, pembrolizumab, and cemiplimab), and α-PD-L1 (atezolizumab, avelumab, and durvalumab) (Lipson and Drake 2011;Twomey and Zhang 2021). Cancer treatments using ICIs have resulted in longer progression-free survival benefits in some cancer patients, but a large number have been found to be refractive or relapsed due to acquired resistance (Seidel et al. 2018). Thus, it is imperative to elucidate their underlying mechanisms and develop strategies to overcome ICB resistance (Barrueto et al. 2020;Fares et al. 2019;Kalbasi and Ribas 2020).
Tumor innervation plays a key role in the development of solid cancers. Recent studies have identified molecular mechanisms that enable reactivation of developmental neurogenic pathways in tumors and the roles of peripheral nerves in the initiation, development, progression, and metastases of different types of cancers (Zahalka and Frenette 2020). Autonomic and sensory nervous systems engage with the tumor microenvironment (TME) to promote cancer cell proliferation and migration, as well as recruit immunosuppressive cells to the TME which trigger signaling pathways related to anti-inflammatory reflexes (Fujii et al. 2017;Jiang et al. 2020;Kuol et al. 2018;Magnon et al. 2013;Mohammadpour et al. 2021). Blocking the communication between the peripheral nervous system and TME by transection, genetic manipulation, and chemical denervation has been shown to inhibit tumor growth and progression of certain cancers, suggesting that tumorassociated nerves are potential targets for anticancer therapy (He et al. 2016;Hunt et al. 2020;Zhao et al. 2014).
Botulinum neurotoxin type A (BoNT/A) is one of the several highly potent neurotoxic protein serotypes produced by neurotoxigenic strains of the bacteria Clostridium botulinum. BoNT/A induces flaccid paralysis by blocking the release of acetylcholine (Ach) from presynaptic cholinergic nerve endings at neuromuscular junctions (Poulain and Popoff 2019). BoNT/A was first used clinically in 1980 as a nonsurgical treatment to correct strabismus in pediatric patients, then later received approval for various indications involving muscle spasticity and hyperkinetic movement disorders, neurological disorders, as well as temporary correction of facial wrinkles, symmetry, and contours (Pirazzini et al. 2017;Scott 1980;Wollmer et al. 2022). At present, the active ingredient of all clinically approved BoNT/A formulations is BoNT/A1 and is the most characterized BoNT/A subtype to date (Whitemarsh et al. 2013).
BoNT/A1 has been used in cancer therapy to manage pain associated with tumor growth, postsurgical or radiation treatment, but has also been reported to exhibit antitumor properties (Mittal and Jabbari 2020). In a previous study, we demonstrated the antitumor effect of BoNT/A1 on the growth of B16-F10 tumor grafts in syngeneic mice (Kang et al. 2022). Here, we investigated the antitumor properties of BoNT/A1 in combination with PD-1 checkpoint blockade using four different syngeneic mouse tumor models. The findings revealed synergistic antitumor effects between a single intratumoral injection of BoNT/A1 and anti-PD-1 therapy in two of the four tumor models based on delayed tumor growth and reduced serum exosome levels observed in tumor-bearing mice, and how this combination treatment affects cellular immune response towards tumors.

Cells cultures
Mouse cancer cell lines were purchased from various sources. Melanoma B16-F10 was obtained from the Korean Cell Line Bank (KCLB), colon adenocarcinoma MC38 from Kerafast Inc., mammary carcinoma 4T1, and Lewis lung carcinoma LLC1 from the American Type Culture Collection (ATCC). B16-F10, MC38, and LLC1 were cultured in DMEM (Gibco), while 4T1 was cultured in RPMI1640 (WelGene). All cell culture media contained 10% heat-inactivated fetal bovine serum and 1% penicillin-streptomycin (Gibco). MC38 culture medium was also supplemented with 25 mM HEPES. The cells were maintained in a 5% CO 2 atmosphere at 37°C.

Tumor cell implantation and animal enrollment
All experimental procedures involving laboratory animals received approval from the Institutional Animal Care and Use Committee (IACUC) of Medytox, Inc. (Approval No. A-2020-004). Male and female C57BL/6 and female Balb/c mice, 6 weeks of age, were purchased from Orient Bio, Inc. and housed at an animal facility of Medytox, Inc. under specific pathogen-free conditions. After a one-week quarantine period prior to the start of the experiments, the animals were placed under general anesthesia with an intraperitoneal (IP) injection of 100 mg/kg ketamine hydrochloride (Yuhan Corp.) and 10 mg/kg xylazine (Rompun®, Bayer) preparatory to tumor cell implantation at Day 0. B16-F10, LLC1, and MC38 tumor cells, freshly harvested from cultures and resuspended in cold PBS (Gibco) at appropriate density, were injected subcutaneously into the right flanks of sex-matched C57BL/6 mice at a volume of 0.1 mL, while 4T1 tumor cells were injected directly into the right abdominal fat pad of female Balb/c mice at a volume of 0.05 mL. Specifically, male C57BL/6 mice were injected with 5 × 10 5 B16-F10 or LLC1 cells, female C57BL/6 mice with 2.5 × 10 5 MC38 cells, and female Balb/c mice with 5 × 10 5 4T1 cells. Mice that developed tumors at the implantation sites were selected and assigned to the placebo, BoNT/A1, α-PD-1, and α-PD-1 + BoNT/A1 groups (n = 10 per group) by stratified randomization based on tumor volume, calculated from the tumor length (long diameter) and width (short diameter) using the formula ½ × length × width 2 (Faustino-Rocha et al. 2013). Briefly, 4T1 tumorbearing mice were grouped at day 5, each having an average tumor volume of 50 mm 3 ; MC38 tumor-bearing mice were grouped at day 7, each having an average tumor volume of 40 mm 3 ; B16-F10 and LLC1 tumor-bearing mice were grouped at day 8, each having an average tumor volume of 30 mm 3 and 15 mm 3 , respectively.

Administration of articles, assessment of tumor growth, and body weight changes
After group assignment, tumor-bearing mice in the BoNT/A1 group were administered a single IT injection of 15 U/kg BoNT/A1 (1 mL/kg dose volume), while the α-PD-1 group received a total of four IP injections of 5 mg/kg α-PD-1 antibodies (5 mL/kg dose volume), administered at an interval of twice a week. The α-PD-1 + BoNT/A1 group was administered both α-PD-1 and BoNT/A1 following the same dosing regimen described for each article. The placebo group received sterile saline injection as sham control following the same dosing schedule as the two articles. Tumor volume was measured twice a week using a digital Vernier caliper (Misumi), and body weight was measured once a week. Tumor growth inhibition (TGI), defined as the percent reduction in tumor size relative to the placebo group observed at the point in time, was calculated using the formula (MTV con − MTV tx ) ÷ MTV con × 100%, where MTV con and MTV tx represent mean tumor volumes of the placebo and test groups, respectively. Body weight changes were calculated using the formula BW X ÷ BW i × 100%, where BW i and BW X represent body weights at the time of initial treatment and X days after treatment, respectively. Animals with tumor volumes exceeding 2000 mm 3 , body weight loss of >20%, or exhibited signs of moribundity were immediately euthanized in accordance with IACUC guidelines.

Serum exosome analysis
At the end of tumor growth evaluation, 0.1 mL whole blood was drawn from the retroorbital sinus of each tumor-bearing mouse and collected in individual serum separation tubes (SST™, BD Biosciences). Whole blood was also drawn and collected from agematched naive C57BL/6 and Balb/c mice and served as baseline controls. Serum was separated from the other blood components by centrifugation at 1000 × g for 15 min at 4°C. Exosomes were isolated from serum samples using ExoQuick™ exosome precipitation solution (System Biosciences) and quantified using the ExoELISA-ULTRA Complete Kit, CD63 detection (System Biosciences) following manufacturer's instructions. Absorbance at 450 nm was measured with the BioTek Synergy HTX Multimode Reader (Agilent).

Statistical analysis
Experimental data are presented as mean ± standard deviation or mean ± standard error. Graphical representation and analysis of statistical data were generated using Prism 7.05 (GraphPad). Groupwise comparisons for tumor growth and body weight changes used Dunnett's or Tukey's test; serum exosome levels, TIL quantities and percentages by Dunn's, Tukey's, or Dunnett's test; and Kaplan-Meier survival curves by MantelCox Logrank test.
Placebo and α-PD-1 groups showed a 7.6% and 9.8% average increase in mouse body weight 10 days after the initial treatment, respectively. In contrast, BoNT/A1 and α-PD-1 + BoNT/A1 groups showed a 1.9% and 3.5% average decrease in body weight, respectively ( Figure 1f). The lower mean body weights observed compared to the placebo (p < .001 and p < .0001, respectively) and α-PD-1 groups (p < .05 and p < .01, respectively) were expected as lower body weight gains or body weight loss has previously been reported as a transient secondary pharmacological effect of BoNT/A1 without any serious implications when exposed only at therapeutic doses (Seo et al. 2019;Torii et al. 2015).

Antitumor effects of PD-1 checkpoint blockade on MC38 tumors is enhanced by BoNT/A1
Twenty days after the subcutaneous implantation of MC38 tumors, the placebo, BoNT/A1 and α-PD-1 groups each showed eight mice (80%) with tumors approximating or exceeding Tumor growth and changes in body weight among MC38 tumor-bearing mice administered with a single IT injection of 15 U/kg BoNT/A1, four repeated IP injections of α-PD-1 (5 mg/kg/dose), or their combination. (a-d) Tumor growth curves of individual mouse, with arrows indicating time of administration for BoNT/A1 (red) and α-PD-1 (blue). Data points and error bars represent mean ± standard error for tumor volume (e) and mean ± standard deviation for body weight (f). Data analyzed by Dunnett's test or Tukey's test (*p < .05, **p < .01). 1000 mm 3 (Figure 2a-c), whereas the α-PD-1 + BoNT/A1 group only showed two mice (20%) with tumor volumes >1000 mm 3 . Moreover, one mouse in the α-PD-1 + BoNT/A1 group recorded a tumor volume of 103 mm 3 , which was only twice its initial size at the start of treatment (Figure 2d). The placebo group reached a mean tumor volume of 1569 ± 768 mm 3 , while the BoNT/A1 and α-PD-1 groups displayed similar tumor growth, reaching mean tumor volumes of 1530 ± 722 mm 3 (2.5% TGI), and 1412 ± 587 mm 3 (10.0% TGI), respectively. In contrast, the mean tumor volume of the α-PD-1 + BoNT/A1 group (806 ± 310 mm 3 , 48.7% TGI) was significantly lower than the other groups (p < .05). These observations suggest that the combination of α-PD-1 and BoNT/A1 elicited synergistic antitumor effects on MC38 tumor growth (Figure 2e). Both groups administered with BoNT/A1 recorded lower mean body weights than the placebo and α-PD-1 groups at the end of the evaluation period (Figure 2f), though only the difference between the α-PD-1 and α-PD-1 + BoNT/A1 groups showed statistical significance (p < .01).

Discussion
Tumor innervation has been associated with tumor cell survival, proliferation, and migration. Elevated expression of proteins involved in tumor survival such as NFкB, c-Myc, GSK-2, PIM-2, SKP, SRF, PTEN, androgen receptor, and estrogen receptor α have been found in densely innervated prostate tumors, and correlated with increased cancer cell proliferation, metastasis, poor patient prognosis, and decreased survival (Magnon et al. 2013;Olar et al. 2014). Innervation by both sympathetic and parasympathetic nerves has been reported to activate adrenergic, muscarinic, and nicotinic acetylcholine receptors on the surface of cancer cells, enhancing tumor budding and tumor cell migration along neurite extensions leading to perineural invasion, a pathological feature that serves as an alternative route for tumor metastases (Amit et al. 2016;Gil et al. 2010;Guo et al. 2013;Zhang et al. 2017). Densely innervated tumors also showed increased recruitment of bone marrow-derived cells and M2 macrophages into the tumor stroma, as well as activation of regulatory T cells and myeloid-derived suppressor cells that are involved in arresting immune cell maturation and help maintain local immunosuppression. These, along with other tumor intrinsic and extrinsic factors, may contribute to ICB resistance (Barrueto et al. 2020;Cervantes-Villagrana et al. 2020;Fares et al. 2019;Vinay et al. 2015).
Several groups have suggested the use of pharmacologic agents that induce denervation for anticancer therapy. Adrenergic antagonists that interfere with β-adrenergic receptor signaling (beta-blockers) have been reported to inhibit cancer progression and improve survival, including a recent Phase I clinical trial showing the synergistic antitumor effect of pembrolizumab and propranolol on metastatic melanoma (Al-Wadei et al. 2009;De Giorgi et al. 2011;Gandhi et al. 2021;Powe et al. 2010). Meanwhile, cholinergic antagonists that interfere with nicotinic and muscarinic receptor signaling pathways have been found to inhibit proliferation of tumor cell growth dependent on parasympathetic innervation (Calaf et al. 2022;Russo et al. 2014;Xie and Raufman 2016;Zhang et al. 2010). Botulinum neurotoxin type A, a bacterial exotoxin that blocks the Ach release from cholinergic nerve terminals, has also been reported to elicit antitumor effects and potentiate other anticancer modalities (Ansiaux and Gallez 2007;Ansiaux et al. 2006;Coarfa et al. 2018;Zhao et al. 2014).
The present study found evidence that combining BoNT/A1 intratumoral injection with anti-PD-1 therapy improved antitumor responses in certain cancers. While MC38 subcutaneous tumors have been reported to be responsive to repeated IP doses of 12.5 mg/kg α-PD-1 (Grasselly et al. 2018;Ngiow et al. 2015), it has been demonstrated here that a suboptimal dosage (5 mg/kg/IP dose) was not sufficient to inhibit MC38 tumor growth. However, when this dosage of α-PD-1 was preceded by a single IT injection of 15 U/kg BoNT/A1, a significant delay in MC38 tumor growth was observed. Moreover, this combination treatment improved the survival rates of MC38 tumor-bearing mice compared to the α-PD-1 treatment alone (Supplementary Figure S2). Meanwhile, the growth of B16-F10 subcutaneous tumors, which have been reported to be refractive for repeated IP injections of 12.5 mg/kg α-PD-1 (Tomita et al. 2018;Triplett et al. 2018), was strongly inhibited by repeated IP injections of 5 mg/kg α-PD-1 after receiving a single IT injection of 15 U/kg BoNT/A1. Based on the known mode of action of BoNT/A1 and these findings, it can be inferred that cholinergic innervation plays a role in MC38 and B16-F10 tumor resistance to anti-PD-1 therapy. Previous studies would suggest cellular apoptosis induced by BoNT/A intratumoral injection as a possible mechanism of action for the potentiation of PD-1 checkpoint blockade. However, some of those studies have also reported that other tumor cells originating from similar types of cancer were not affected by exposure to BoNT/A (Mittal and Jabbari 2020). Moreover, the mouse tumor models investigated in the present study did not show any significant changes in tumor growth after intratumoral injection of BoNT/A1 alone, suggesting that cellular apoptosis in tumors through BoNT/A1-mediated denervation or direct uptake of the neurotoxin is not likely the major contributing factor in precluding resistance to PD-1 checkpoint blockade. Further studies are needed to elucidate the physiological events related to BoNT/A1mediated tumor denervation and PD-1 checkpoint blockade that contribute to tumor growth inhibition observed in the B16-F10 and MC38, but not in the LLC1 and 4T1 syngeneic mouse tumor models.
In addition to delayed tumor growth, the α-PD-1 + BoNT/A1 combination therapy also showed synergistic effects in the reduction of serum exosome levels among mice implanted with B16-F10 and MC38 tumors. Lower serum exosome levels observed among B16-F10 and MC38 tumor-bearing mice treated with the α-PD-1 + BoNT/A1 combination therapy could be directly correlated with tumor growth inhibition. Interestingly, an in vitro assay with B16-F10 cells exposed to 10 U/mL BoNT/A1 for 24 h had shown a statistically significant reduction in exosomes release compared to the vehicle control, while not having any significant influence on cell viability (Supplementary Figure S3). This would suggest that the reduction in serum exosome levels attributed to tumor growth inhibition by the α-PD-1 + BoNT/A1 combination treatment may be complemented by a direct antagonistic effect of BoNT/A1 on tumor-derived extracellular vesicle release through a yet-to-bedefined mechanism. In recent years, tumor metastases and disease progression of various cancers have been attributed to tumor-derived exosomes (Whiteside 2016). These extracellular vesicles reportedly carry (1) axonal guidance molecules that induce neurite outgrowth such as EphrinB1; (2) microRNAs that promote tumor growth by participating in blood vessel permeability, angiogenesis, and polarization of macrophages; or (3) longnoncoding RNAs associated with anti-apoptotic activity, resistance to cisplatin, protection against oxidative stress, and increases in cancer cell proliferation, migration, and invasiveness (Lucido et al. 2019;Madeo et al. 2018;Smolarz and Widlak 2021). Therefore, reducing the levels tumor-derived extracellular vesicles in circulation may have a crucial impact on preventing cancer progression and improving success rates of anticancer therapy. The size and morphology of the serum exosomes investigated in the present study was not verified, and biomarkers such as MET (melanoma), Hsp70 (colon CA), and miR10b (breast CA) were not used to determine what percentage of the exosomes are tumor-associated. Nevertheless, the significantly higher levels of serum exosomes found in tumor-bearing mice compared to agematched naïve mice strongly suggest the presence of tumor-derived extracellular vesicles in the circulatory system of tumor-bearing mice. While the present study has established growth inhibition of MC38 and B16-F10 tumors and reduced exosome levels as antitumor effects of the α-PD-1 + BoNT/A1 combination therapy, further studies are needed to identify the mechanisms that BoNT/A1 uses to suppress the release of tumor-derived exosomes and define their actual impact on tumor metastases and cancer progression.
LLC1 and 4T1, two other mouse tumors investigated in this study, have also been reported to be refractive to PD-1 checkpoint blockade (Ajona et al. 2020;Grasselly et al. 2018;Triplett et al. 2018). Combination treatment with the same dosage of α-PD-1 and BoNT/A1 effective against MC38 and B16-F10 tumors had shown no apparent changes in tumor growth and serum exosome levels among LLC1 and 4T1 tumor-bearing mice. A higher dose of BoNT/A1 for the combination therapy was initially considered since the previous study using B16-F10 mouse tumor model had shown a delay in tumor growth following a single IT injection of 50 U/kg BoNT/A1 (Kang et al. 2022). However, 4T1 tumor-bearing mice administered with BoNT/A1 of varying dosage had shown no significant tumor growth inhibition of up to 50 U/kg BoNT/A1. More importantly, severe adverse events such as body weight loss were evident at 50 U/kg BoNT/A1 and significant mortality had occurred at doses above 50 U/kg BoNT/A1 (Supplementary Figure S4). Based on these evidence, cholinergic innervation of LLC1 and 4T1 tumors is not likely involved in conferring resistance to PD-1 checkpoint blockade.
A recent study on pancreatic ductal adenocarcinoma reported that elevated levels of Ach impaired the recruitment of CD8 + T cells via HDAC1-mediated suppression of CCL5, inhibited IFNγ production by CD8 + T cells, and favored Th2 over Th1 differentiation, consequently suppressing intratumoral T cell responses (Yang et al. 2020). Furthermore, nerve-derived neurotransmitters have been reported to induce tumor infiltration of immunosuppressive macrophages that compromise the effectiveness of ICB (Gysler and Drapkin 2021). As shown in our B16-F10 syngeneic tumor model, combining PD-1 checkpoint blockade with BoNT/A1 intratumoral injection lowered the proportion of immunosuppressive cells such as regulatory T cells and myeloid-derived suppressor cells, while also promoting the infiltration of a higher proportion of CD8 + T cells to the tumor microenvironment compared to α-PD-1 treatment alone. These likely contributed to the higher number of CD4 + and CD8 + T cells found at the later time point and inhibitory effects on B16-F10 tumor growth. On the basis of these observations, the synergistic antitumor effects of α-PD-1 and BoNT/A1 combination therapy may involve the suppression of immunosuppressive cell recruitment through inhibition of Ach release from tumor-associated cholinergic nerves by BoNT/A1, thereby allowing favorable conditions for launching antitumor T cell responses elicited by PD-1 checkpoint blockade. Additional studies are needed to define the exact mechanism and ascertain what other types of tumors are susceptible to the combination therapy of PD-1 checkpoint blockade and BoNT/A1. Furthermore, a detailed characterization of CD4 + and CD8 + T cell states within the tumor microenvironment, as well as in tumor-draining lymph nodes, may provide more insights on the effects of BoNT/A1 intratumoral injection on antitumor immunity, and how this may impact the efficacy of PD-1 and other immune checkpoint blockades.
In conclusion, there is vast information related to tumor innervation and its role in cancer pathology. Nonetheless, more in-depth studies are necessary to establish a better understanding of how different types of tumor innervation affect the physiology of different solid cancers, and also define their influence on tumor responses to immune checkpoint