Vagal Blockade of the Brain-Liver Axis Deters Cancer-Associated Cachexia

Summary

Cancer-associated cachexia (CAC) is a multifactorial and currently incurable syndrome responsible for nearly one-third of cancer-related deaths. It contributes to therapy resistance and increases mortality among affected patients. In this study, we show that cancer-induced systemic inflammation alters vagal tone in CAC mouse models. This vagal dysregulation disrupts the brain-liver vagal axis, leading to a reprogramming of hepatic protein metabolism through the depletion of HNF4α, a key transcriptional regulator of liver function. The loss of HNF4α disrupts hepatic metabolism and promotes systemic inflammation resulting in cachectic phenotypes. Interventions targeting the right cervical vagus nerve, surgically, chemically, electrically, or through a non-invasive transcutaneous device, attenuate CAC progression, alleviate its clinical manifestations, and synergize with chemotherapy to improve overall health and survival in mice.

eTOC

Vagal dysfunction mediates the impact of tumor on liver metabolism, leading to cachexia. Blocking the right cervical vagus with various invasive or non-invasive approaches alleviates cachexia, decoupling cachexia progression from tumor load, and synergizes with chemotherapy to improve overall health and survival in mice.

Graphical Absract:

Introduction

Cancer-associated cachexia (CAC) is a complex, multifactorial metabolic syndrome. The primary clinical criteria for CAC include weight loss exceeding 5% in 6 months or more than 2% in individuals with a body mass index (BMI) below 20 kg/m², accompanied by at least three secondary criteria: muscle mass depletion, asthenia (physical weakness), body fat loss, anorexia, or abnormal biochemical markers ¹. The prevalence of CAC manifestations in cancer patients is as high as 85% in some cancers, with pancreatic and lung cancers being among the highest. About 30% of all cancer deaths are directly attributed to CAC ^2–4. CAC plays a crucial role in cancer progression and outcome as it increases chemotherapy toxicity, resistance to therapy, and complications from surgeries, decreases in quality of life, and higher mortality rates ^5–8. While CAC is most prominent in end-stage cancers, for some tumor types CAC is present before the tumor is clinically detectable ^{9; 10}. Unfortunately, there are no effective clinical treatments for CAC; no correlation has been demonstrated between dietary intake and body composition parameters^11–13 and pharmacologic interventions have failed to produce significant improvements in clinical settings ^14–16.

The challenge associated with treating CAC is due to its nature as a complex systemic syndrome involving tumor-derived catabolic factors and pro-inflammatory molecules derived from tumor-immune crosstalk that promote systemic changes, e.g., eating behavior, glycolysis, proteolysis, and lipolysis ^17–20. These metabolic alterations contribute to excess energy consumption beyond that attributable to the tumor, which likely stimulates the breakdown of fat and muscle tissues ^{11; 17; 18; 12; 19; 10}. Notably, the liver primarily regulates key metabolic pathways altered in CAC. In cachectic patients, the liver manifests aberrant function in metabolic pathways such as gluconeogenesis, lipid metabolism, albumin synthesis, and the urea cycle, contributing to the metabolic symptoms present in cachexia ^{21; 19; 10}. CAC results from a complex web of feedback systems that is still not wholly understood, making the treatment of any aspect insufficient. However, early targeting of key mediators of multiple elements may synergistically relieve symptoms ^{11; 18; 6}.

The peripheral parasympathetic nervous system, a bidirectional electrical and chemical signaling pathway that drives appetite, hunger, digestion, metabolism, immune activation, and other essential functions of the gastrointestinal system, may be such a target ^{22–27; 20}. The vagus nerve is the dominant electrical mediator of this system, with direct vagal innervation of the digestive tract and liver ^{28–31; 27; 32}. The vagus plays key homeostasis-preserving roles in peripheral and central metabolic pathways, many of which are disrupted in CAC patients ^{33; 34; 25; 31; 27}. The left and right vagus nerves asymmetrically innervate organs which play roles in essential gut-brain neural circuits, with the right vagus nerve primarily influencing gut-induced reward mechanisms, nutrient preference, and food memory^{35; 29; 24; 36–38; 20; 39}. The vagus nerve’s role in multiple hallmark symptoms and mechanisms of CAC led us to hypothesize that dissecting its function in CAC may shed light on the consequent metabolic manifestations with implications for therapy.

Vagal neuromodulation (VNM) is an FDA-approved therapy that provides promising means to “reset” aberrant signaling along the axis, with recent developments in transcutaneous VNM (tVNM) that overcome the need for invasive surgical implants ^{40–42; 37; 43; 44}. Since the vagus nerve is known to change the parasympathetic “rest or digest” output to many organs, it is under investigation for the treatment of other metabolic disorders in preclinical and clinical trials ^45–48. The involvement of the vagus nerve in bodily metabolic processes associated with CAC led us to probe its potential role in CAC pathogenesis.

Results

The vagus nerve is dysregulated by CCL2 in CAC animals

Flank-injected Lewis Lung Carcinoma (LLC) and orthotopically injected pancreatic Kras G12D, Trp53 R172H, Pdx-1-Cre (KPC) are established models for studying CAC ^{49–53; 36; 54–56; 39; 57}. To test the hypothesis of vagal involvement in CAC during carcinogenesis, we first performed electrical recordings on the common hepatic branch of the subdiaphragmatic vagus nerve in mice with and without LLC or KPC tumors. Mice with either cancer had an altered vagal activity pattern compared to recordings from wildtype mice preceding weight loss (Fig. 1A). Analysis of the recordings suggests that while the noisiness of vagal signaling is significantly increased in cancer mice as measured by quantification of the root mean square, the strength of individual signals is lessened as quantified by peak-to-peak amplitude (Fig. 1B). Together, these measurements indicate that the vagus nerve is sending noisy signals that increase its overall activity.

Figure 1: Cancer-induced neuroinflammation promotes vagal hyperactivity.

A. Left: interoperative photograph showing recording electrodes placed on the common hepatic branch of the subdiaphragmatic vagus nerve. Right: subdiaphragmatic electroneurograms (ENGs) from vagus nerves of WT and cancer-bearing mice 14 days post-injection of PBS or LLC cells, respectively. B. Top: root mean square quantification of the ENG signal (N=15 mice, 5 per group). Bottom: mean peak-to-peak amplitude of spikes in the ENG (N=15 mice, 5 per group). C. Left: Staining of the vagal dorsal motor nucleus (VDMN) for neural activation marker c-Fos (green), vagal lineage marker Phox2b (red), and DAPI (blue). Scale bars 100 μm. Right: Quantification in N=24 animals (8 per group). D: Quantification of CCL2 levels in CSF, brain tissue, and serum. N=24 animals (8 per group) for each plot. E: Quantification of change in neuronal activation as measured by fold change in fluorescence intensity (ΔF/F) of cultured primary neurons with gCaMP before and after the addition of PBS or CCL2. N=12 total wells, 6 wells per group, statistical analysis by Student’s T-test. F. Left: representative staining of VDMN with DAPI (blue), Phox2b (green), and IBA1 (red, neuroinflammation marker). Scale bars 100 μm. Right: quantification of IBA1 positivity area. N=24 (8 animals per group). G. Quantification of CD45+ cells by immunofluorescence (N=24, 8 animals per group) and macrophages by FACS (N=6, 3 animals per group) in mouse hindbrains. H. Bindarit reduces weight loss in LLC and KPC mice (N=40, 8 per group). Statistical analysis by RMANOVA with Tukey post-hoc. I. Survival in KPC and LLC animals is not statistically significantly improved in Bindarit animals by Peto’s Rank Log test. N=32, 8 mice per condition. J. Left: CCL2 measurements and Right: Acetylcholine (Ach) measurements in serum from human pancreatic ductal adenocarcinoma (PDAC) patients. N=29 (9 non-cachectic, 20 cachectic). Statistical analysis by Student’s T-test. K. Correlation between serum CCL2 and weight loss at patient’s final clinical visit before death (left) and ACh and the delay between patient’s cancer diagnosis and patient’s cachexia diagnosis (negative interval indicates cachexia was diagnosed before PDAC). Statistical analysis by Pearson’s. For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by ANOVA with Tukey Post-Hoc unless otherwise stated. See also Figure S1 and Table S1.

The vagal dorsal motor nucleus (VDMN) is the source of most parasympathetic pre-synaptic fibers in the vagus nerve and is responsible for reflex circuits prompting digestive secretion in response to afferent sensory vagal signaling ^{28; 58; 59}. Indeed, c-FOS staining for neuronal activation in the VDMN demonstrated increased activity in mice with LLC or KPC, further supporting the notion that cancer promotes central changes in vagal activity (Fig. 1C).

We and others found that CCL2, also known as monocyte chemoattractant protein-1 (MCP-1), is a macrophage-associated proinflammatory cytokine that recruits innate immune cells to tumors and various organs, such as the liver and brain, in mice with cancers outside these organs ^{53; 60–62}. Supporting the systemic role of CCL2 during carcinogenesis, we find that its levels are elevated in brains, cerebrospinal fluid (CSF), and serum of mice with LLC and KPC cancers (Fig. 1D). CCL2 has been previously shown to sensitize neurons and increase their excitability ⁶³. Adding CCL2 to cultured vagal primary neurons isolated from transgenic mice expressing Thy1-GCamp6 ⁶⁴, increased neuronal activation (Fig. 1E). We also measured CCL2 along with acetylcholine (ACh), the primary neurotransmitter of the vagus nerve, in multiple tissues at 3-, 7-, 14-, and 21-days following cancer induction. We found that CCL2 levels in serum, VDMN, CSF, and liver elevate prior to vagal tone alteration and ACh elevation (Fig. S1A–C).

We recently demonstrated in mice with breast and pancreatic cancers an increasing infiltration of innate immune cells to the liver that drives cancer-associated metabolic dysfunction ⁵³. We now explored whether the systemic immune response to cancer similarly extends to the vagal nuclei as a potential explanation for vagal dystonia. Histologic analysis of brain sections demonstrated inflammatory granulation in the VDMN of cancer mice (Fig. S1D). Using ionized calcium-binding adaptor molecule 1 (IBA1) staining, we further showed that mice with LLC have neuroinflammation (Fig. 1F). Complementary studies with immunofluorescence for CD45+ cells noted a significant immune infiltration in the VDMN (Fig. 1G, S1E). To identify the specific infiltrating immune cell subtypes we performed additional staining and Fluorescence Assisted Cell Sorting (FACS) analyses. We found that brain infiltrating immune cells were mainly CD45+ CD11b+ F4/80+ populations consistent with macrophages, including CD86+ and CD206+ macrophages, while the presence of other myeloid cells and T cells did not differ significantly (Fig. 1G and S1F–G)^{53; 65}.

CCL2 receptor (CCR2) knockout in mice with cancer has previously been demonstrated by us and others to restrict the development of CAC ^{53; 66}. Thus, we first confirmed that CCR2 KO animals had reduced CD45+ cell infiltration into the VDMN and reduced neuroinflammation (Fig. S1H–I). Next, by blocking CCL2 activity with the drug Bindarit ⁶⁶, we found significantly reduced immune cell infiltration and neuroinflammation in the VDMN, though the reduction was less robust than in the CCR2 KO (Fig. S1J–K). In addition, Bindarit-treated animals had lower ACh levels in their serum, and attenuated vagal tone alterations compared to PBS-treated animals (Fig. S1L–M). Bindarit-treated animals showed improvements in cachectic features, including weight loss, muscle atrophy, and anorexic eating behavior, although their improvement in survival was not statistically significant (Fig. 1H–I, S1N–Q). The ability of Bindarit to reduce vagal hyperactivation and improve CAC symptoms in mice suggests that CCL2 likely plays an important role in CAC-associated vagal dysregulation and that increased vagal activity contributes to CAC. Notably, we observed similar elevations in CCL2 and ACh in patients with pancreatic ductal adenocarcinoma (PDAC) who develop cachexia, compared to non-cachectic PDAC patients (Table S1 and Fig. 1J). CCL2 levels were inversely correlated with patient weight loss, and ACh levels were inversely correlated with the length of time lapse between the patient’s cancer diagnosis and cachexia diagnosis (Figure 1K). Moreover, an unbiased 80-plex analysis of human inflammatory factors in PDAC patients with or without CAC identified several inflammatory factors that are differentially elevated during CAC, including other CCL signaling proteins CCL17, CCL24, and CCL25, as well as other cachexia-associated inflammatory factors such as TNFα (Fig. S1R).

Together, these data highlight the importance of neuroinflammatory crosstalk in CAC in both cancer-bearing mouse models and patients and suggest that cancer-induced systemic inflammation alters the activation of the vagus nerve and likely contributes to CAC.

Vagotomy alleviates CAC progression and phenotypes in murine CAC models.

To directly evaluate the potential benefit of disrupting the vagal-liver axis in CAC, we performed a right cervical vagotomy, in both KPC and LLC CAC mouse models alongside wildtype (PBS-injected) controls. Mice received either a sham surgical procedure leaving the vagus nerve intact (WT+Sham vs. KPC+Sham or LLC+Sham), or a vagotomy procedure (WT+Vagotomy vs. KPC+Vagotomy or LLC+Vagotomy) (Fig. 2A). We found that vagotomy in tumor-bearing mice ameliorated welfare changes including hunched posture and poor grooming (Fig. 2B–C, S2A). After a modest 2.5-day delay in reaching their peak weight following the surgery, PBS-injected mice treated with vagotomy did not differ significantly in weight from those that received sham surgeries (Fig. 2C). In addition, we could not detect a measurable sex-related weight difference between female and male tumor-bearing mice in response to vagotomy (Fig. S2B). Notably, while KPC and LLC mice treated with vagotomy developed tumor masses similar in size to KPC− and LLC-Sham mice, they lost significantly less weight (Fig. 2C, S2C). In addition, KPC+Vagotomy and LLC+Vagotomy mice lost less lean and fat tissues than their sham cohorts (Fig. S2D–E).

Figure 2: Right cervical vagotomy alleviates CAC manifestations in mice with lung (LLC) or orthotopic pancreatic (KPC) cancer.

A. Schematic outlining the treatment groups assessed. B. Representative photos of mice from each experimental group. LLC tumors outlined with dashed red lines. C. Weight change in mice with and without vagotomy, with and without cancer. Vagotomy or sham surgeries were performed on day −5, cancer cells or PBS were injected on day 0. N=56 (8 animals per group). Analysis by RMANOVA with Tukey HSD. D. Left: representative photomicrographs of right gastrocnemius muscle sections stained with wheat germ agglutinin (WGA) in green. Scale bars 100 μm. Center: quantification right gastrocnemius muscle fiber cross-sectional area. WT controls are the same for each cancer cohort. Right: Grip strength of WT, KPC, and LLC mice with and without vagotomy. WT controls are the same for each cancer cohort. N=48 (8 animals per group). Statistical analysis by ANOVA with Tukey Post-Hoc. E. Survival curve showing survival duration in cancer-injected mice with vagotomy or sham procedure. Statistical significance by Peto’s Log Rank Test. For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. See also Movie S1, Figure S2.

Since muscle atrophy is an established diagnostic criterion for CAC, we further examined the cross-sectional area of right gastrocnemius muscle fibers, right gastrocnemius muscle weight, and grip strength in the CAC mice. Following right vagotomy, we found a significant attenuation in the reduction of the mean fiber area, loss of gastrocnemius muscle mass, and reduced grip strength in both KPC and LLC mice (Fig. 2D, Fig. S2F). In addition, we found reduced muscle atrophy gene expression in these tissues, further supporting the notion that a main contributor to the restoration of lean body mass by vagotomy is reduced muscle loss (Fig. S2G). The functional benefits of vagotomy on the gastrocnemius muscle of CAC mice were also supported by attenuation of the reduced evoked electromyographic (EMG) response and improved wire hang test performance (Fig. 2D, Movie S1, and Fig. S2H–I). Notably, in all vagotomized mice, no statistically significant alterations were found in heart rate or liver hemodynamics in a liver laceration model (Fig. S2J–K).

Importantly, KPC and LLC mice treated with right vagotomies experienced a statistically significant extension in lifespan compared to their sham-treated equivalents (Fig. 2E). While tumor-bearing sham mice of both models overwhelmingly died due to their illness, vagotomized KPC and LLC mice were euthanized mainly due to reaching humane endpoints stemming from tumor burden. Given that loss of appetite, weight, lean muscle, and fat tissue mass are defining features of CAC clinically, we assessed whether the changes in these variables correlated with survival (Fig. S2L–O). We found that while the preservation of each feature correlated with prolonged survival with no sex-specific differences, muscle mass correlated most strongly (Fig. S2L–P). Compared to sham mice, vagotomized KPC and LLC mice showed reduced levels of CCL2, TNFα, MIP2, IL-1β, IL-2, and CXCL10 at death or humane endpoints, consistent with their ameliorated CAC manifestation (Fig. S2Q).

Furthermore, KPC+Sham and LLC+Sham mice showed behavioral abnormalities including signs of depression and anorexia (Fig. 3A–B, S2R). Vagotomized animals were more active and social compared to sham animals (Fig. 3A). LLC+Sham and KPC+Sham animals also consumed less food than WT shams or vagotomized mice with cancer, regardless of whether food was accessed from a hanging feeder or a trough on the cage floor (Fig. 3B, S2R).

Figure 3: Right cervical vagotomy attenuates behavioral abnormalities and enhances chemotherapy response in lung (LLC) or orthotopic pancreatic (KPC) cancer.

A. Spontaneous motion over 15 minutes in open field assay and socialization in three chamber social assay. WT controls are the same for each cancer cohort. N=48 (8 animals per group). B. Quantification of daily food consumption from cage top feeder. WT controls are the same for each cancer cohort. N=48 (8 animals per group). C. Survival curve showing longitudinal survival of LLC- or KPC-injected mice receiving sham surgeries or vagotomy with or without chemotherapy. Statistical significance by Peto’s Log Rank Test with Bonferroni correction for multiple comparisons (6-fold). D. Weight changes in LLC or KPC bearing-mice which received sham surgeries or vagotomy with or without chemotherapy. N=64 (8 animals per group). Analysis by RMANOVA with Tukey HSD. E. Tumor burden measurements at time of sacrifice in tumor bearing-mice which received sham surgeries or vagotomy with or without cisplatin chemotherapy. N=64, 8 mice per group. F-G. Impacts of hexamethonium, atropine, or PBS treatment in KPC or LLC mice and WT controls on (F) weight loss and (G) survival. N=56, 8 per condition. Statistical significance for survival by Peto’s Rank-Log test with 3x Bonferroni correction. Statistical analysis by ANOVA with Tukey Post-Hoc unless otherwise stated. For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. See also Figure S2–S4.

Finally, the significant improvement in the well-being of vagotomized KPC and LLC mice while still suffering from tumor burden led us to evaluate whether right cervical vagotomy could provide added benefits alongside cisplatin chemotherapy, a commonly used chemotherapy that B6 mice tolerate well ^{67–72; 57}. Encouragingly, although chemotherapy by itself may lead to weight loss ⁷³, vagotomized mice in both models treated with cisplatin chemotherapy lost less weight and had reduced tumor size, resulting in prolonged survival (Fig. 3C–E).

Right cervical vagotomy is most beneficial in alleviating CAC

To confirm the specificity of our results, we performed additional experiments using various controls. First, given that the left and right vagus nerve asymmetrically innervate organs which play roles in essential gut-brain neural circuits ^{35; 29; 24; 36–38; 20; 39}, we compared weight changes in LLC mice receiving either left cervical vagotomy or right cervical vagotomy after 30 days (n=16 per sham group, 8 per vagotomy group) (Fig. S3A–B). While left vagotomy (LLC+LV) produced some attenuation of weight loss relative to WT controls, mice with right vagotomies (LLC+RV) experienced no statistically significant weight loss relative to WT controls (Fig. S3B). We then employed subdiaphragmatic vagotomy of the right posterior vagal trunk to investigate whether downstream branches could be targeted to reduce systemic effects (Fig. S3C). While both subdiaphragmatic and cervical vagotomies attenuated weight loss, mice with subdiaphragmatic vagotomies were less able to maintain their weight than those receiving cervical vagotomies (Fig. S3D). These results suggest that the cervical vagotomy is preferable for restricting weight loss in CAC mouse models as no single subdiaphragmatic branch accounts for the full effect.

To further confirm the specificity of the right vagotomy effect, we microinjected the right vagus with botulinum toxin A, a neurotoxic agent that chemically blocks ACh release ⁷⁴. We found that LLC mice injected with botulinum toxin A achieved a comparable weight loss restriction, muscle atrophy reduction, and survived longer than the sham-treated mice (Fig. S3E–G). Notably, the beneficial effect of vagotomy on weight loss and muscle retention in LLC mice surpassed that observed with treatment with anamorelin, a Growth Hormone Secretagogue Receptor (GHSR) agonist and experimental therapy for cachexia (Fig. S3H–I) ^{75; 76}. As previously reported by others ^{75; 76}, a survival benefit was not seen in anamorelin-treated mice (Fig. S3J).

Thus, we concluded that a right cervical vagotomy is the most effective vagal intervention for restricting CAC development in flank lung and orthotopic pancreatic cancer mouse models, and continued using this approach in all subsequent experiments.

Benefits of right cervical vagotomy are dominated by efferent effects

To better understand the mechanisms by which vagotomy produced benefits in CAC, we perturbed either the muscarinic acetylcholine signaling, which is primarily associated with afferent pathways, or the nicotinic signaling, which is more involved in efferent pathways ^{28; 77; 78; 27; 79}. While atropine is a muscarinic acetylcholine receptor antagonist, hexamethonium is a nicotinic acetylcholine receptor antagonist, and hence, treatment with these drugs ablates afferent and efferent signaling, respectively ^80–82. We found that blocking either muscarinic or nicotinic acetylcholine signaling did not relieve upstream neuroinflammation or immune cell infiltration into the VDMN but had an impact on systemic acetylcholine and CCL2 levels, with hexamethonium generating more significant effects (Fig. S3K–N). Treatment with hexamethonium significantly benefited CAC manifestations, including weight retention, muscle retention, grip strength, behavior, and survival, while atropine’s effects were not statistically significant (Fig. 3F–G, S3O–P). These results indicate that while both afferent and efferent components of vagal signaling likely contribute to CAC, the efferent components dominate CAC symptoms.

We therefore investigated whether these beneficial effect of right vagotomy was related to a secondary impact on tumor or liver innervation by measuring liver innervation and tumor innervation in the LLC subcutaneous and KPC orthotopic tumors (Fig. S3Q–R). We found no significant change in either tumor type following vagotomy in terms of total innervation, intrinsic innervation, autonomic intrinsic neuron fraction, or in liver innervation.

Notably, in LLC and KPC mice, vagotomy reduced ACh levels in both serum and liver, while CCL2 was reduced in serum, brain, and CSF, but not liver (Fig. S4A–C). In these cancer-bearing mice, terminal ACh levels were also elevated in the stomach and pancreas, but not the colon or tumors (Fig. S4D–E). In addition, liver and serum ACh levels from mice that succumbed to disease or were euthanized at humane endpoints negatively correlated with survival (Fig. S4F). Although vagotomized LLC mice had a similar level of macrophage infiltration a week following the injection of LLC cells, at 2 weeks, there was a significant decrease in macrophage infiltration compared to their levels in the VDMN of LLC mice without vagotomy, specifically of CD206+ macrophages associated with tumor progression (Fig. S4G).

Vagotomy alleviates CAC manifestations by preserving liver metabolism

We recently found that CAC-related systemic manifestations in breast and KPC pancreatic cancer mouse models are associated with decreased levels of liver HNF4α, the master regulator of liver metabolism ^{83; 53; 84}. As a result, liver metabolism undergoes significant rewiring, particularly in protein metabolism, evidenced by a reduction in the expression levels of HNF4α targets, such as albumin, and the urea cycle enzymes, mainly ornithine transcarbamylase (OTC), consequently contributing to CAC manifestations ^{53 85}. We hence evaluated whether similar changes occur in the livers of mice with LLC. In mice with LLC, we found an almost complete abolishment of liver HNF4α and a significant decrease in OTC protein levels compared to control mice (Fig. 4A). Longitudinal assessment of HNF4α identified its depletion as early as 7 days following cancer induction notably occurring at the same time changes in ACh are observed in the liver (Fig. S4C, H). Since HNF4α is essential to controlling liver metabolism, its expression in the liver has multiple levels of regulation at different developmental stages and cellular states ⁸⁶. To evaluate whether the parasympathetic system, in general, could also regulate the changes in the liver expression of HNF4α in LLC mice, we supplemented cultured human hepatocytes with ACh. We observed a subsequent decrease in HNF4α RNA and protein levels, suggesting that the parasympathetic system can at least partially regulate HNF4α expression via ACh from the vagus nerve (Fig. 4B).

Figure 4: Right cervical vagotomy preserves liver metabolism during cancer cachexia.

A. Left: A representative western blot demonstrating decreased HNF4α and OTC protein expression in cachectic LLC mice livers, compared to those from control mice injected with PBS (WT). Right: A quantification of blot intensity (N=5 for each group). B. 5μM acetylcholine supplementation in THLE-2 for 48 hours decreases HNF4α at the protein (upper) and RNA (lower) levels in primary human hepatocytes as evaluated by western blot and RT-PCR, respectively. C-D. Wildtype mice injected with shHNF4α compared to shLUCIFERASE. C. Decreased liver expression of HNF4α protein levels (upper panel) and a decrease in RNA expression of HNF4α, OTC, and albumin (lower panels), in wildtype mice injected with shHNF4α compared to shLUCIFERASE-injected mice (shLUCIFERASE: N=4, shHNF4α: N=5). D. Weight measurements and NMR body composition analyses measuring % of initial weight, fat, and lean body mass of mice injected with shHNF4α demonstrate a significant loss of weight, fat, and lean body mass compared to control mice injected with shLUCIFERASE (shLUCIFERASE: N=4, shHNF4α: N=5). E. RT PCR measurements show an increase in LCN2 levels in livers of mice with LLC (left) and mice injected with AAV-shHNF4α (right). F. RT-PCR measurements demonstrate an increase in HNF4α and OTC RNA levels in the livers of mice with LLC treated with vagotomy, compared to LLC mice treated with a sham procedure. G. VetScan analysis of liver lysate demonstrates an increase in albumin levels in LLC mice treated with vagotomy (LLC+sham: N=5, LLC+vagotomy: N=4). H. Mass spectrometry analysis of CAC-associated amino acid relative levels in livers of mice with LLC demonstrates preservation of amino acid levels following vagotomy. Heatmap represents an average of relative values per metabolite per treatment. (For WT sham, n-5, WT+vagotomy n=5, LLC sham n=2, LLC + vagotomy n=4). *All statistical analyses were performed using Student t-test; For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.005, *** represents p<0.0005, **** represents p<0.0005. See also Figure S5.

To confirm that the decrease in HNF4α directly contributes to cachexia manifestations independent of tumor effects, we injected wildtype mice with either Adeno-associated virus AAV8-shHNF4α or AAV8-shLUCIFERASE. We confirmed HNF4α knockdown by demonstrating its depletion at both RNA and protein levels (Fig. 4C). Additionally, we validated the decrease in RNA expression of established HNF4α target genes, OTC, albumin [57], and genes involved in fatty acid β-oxidation ⁸⁷, following its depletion (Fig. 4C, bottom right, S5A). No changes in gross liver histology or steatosis were observed in the livers of shHNF4α-injected mice (Fig. S5B). Notably, wildtype mice injected with shHNF4α had weight loss and reductions in both fat and lean body mass (Fig. 4D, S5C). At the molecular level, shHNF4α increased the expression of muscle atrophy-related genes, supporting a significant and direct role for HNF4α depletion in promoting cachexia manifestations independent of cancer (Fig. S5D).

Importantly, we found no liver malfunction in the plasma biochemical profile of wildtype mice treated with vagotomy (Fig. S5E). Both vagal signaling and HNF4α have been shown to regulate hepatic release of lipocalin-2 (LCN2), which is a cachectic driver that enhances systemic inflammation and crosses the blood-brain barrier to regulate food intake ^88–90. Right-sided vagotomy, which elevates HNF4α levels as shown in Fig. 4A, is expected to reduce circulating LCN2 levels ⁹¹. Indeed, we find that LLC mice, which have HNF4α depletion, have elevated levels of LCN2 in the liver, similar to the WT mice injected with shHNF4α (Fig. 4E). Furthermore, LCN2 is also elevated in the brain of LLC mice (S5F).

Furthermore, we found that vagotomy increased the RNA and protein levels of liver HNF4α and OTC and improved liver function, as demonstrated by an increase in the albumin protein levels (Fig. 4F–G, S5G–H). Notably, vagotomy preserved liver amino acid metabolism as manifested by urea cycle metabolites in the livers and plasma of LLC mice, comparable to LLC sham mice (Fig. 4H, S5G). Additionally, vagotomy restored the levels of branched-chain amino acids (Fig. 4H, S5H), which were shown to correlate with muscle loss due to protein catabolism and contribute to reduced survival ^{92; 93}. Notably, the increase in HNF4α and serum albumin levels correlated positively with survival (Fig. S5I–J). Thus, while disrupted liver metabolism via depletion of HNF4α promotes cachexia manifestations, vagotomy maintains HNF4α-regulated liver protein metabolism, restricts protein breakdown, and improves survival.

Intermittent vagal neuromodulation is sufficient to alleviate CAC phenotypes.

While the permanent and invasive nature of vagotomy makes it non-ideal as a potential clinical intervention, new transcutaneous neuromodulation technologies are an appealing therapeutic approach for cachectic patients. Although vagal stimulation has been previously evaluated to activate the immune system in patients with lung cancer, no survival benefit was noted, and no equivalent study on vagal block or vagal pacing has been conducted ⁹⁴. To assess the impact of electrical vagal block on CAC, we implanted mice with microwire hook electrodes on the right cervical vagus nerve, delivering an electrical pattern designed to evoke Low-Frequency Alternating Current block (LFACb) ^95–99 (Fig. S6A–C, Movie S2).

In recent clinical trials using LFACb for pain relief, 91% of patients reported continued therapeutic effect lasting over 24 hours following cessation of the LFACb therapy ⁹⁷, leading us to interrogate the duration of effect for a single dose of LFACb using our waveform. We performed a series of studies where a single 30-minute dose of LFACb therapy was delivered to KPC or LLC mice. To recapitulate the human CAC scenario, we tested the effectiveness of vagal perturbation following a period of tumor growth without intervention. We injected mice with KPC or LLC cells and assigned them to electrode implant but no current (LLC + Sham, KPC + Sham) or implant and therapeutic LFACb (LLC + LFACb, KPC + LFACb) (Fig. 5A). All LLC mice were followed without intervention for 34 days and all KPC mice for 11 days, the time point at which they showed qualitative indicators of cachectic onset including a 5% difference in weight from WT controls. We then assessed the effect of a single dose of vagal block on vagal tone, liver acetylcholine levels, and HNF4α levels (Fig. 5B, S6D–E) based on measurements taken before the therapy is delivered, immediately after treatment, and 12 hours after therapy, 24 hours after therapy, 36 hours after therapy, and 48 hours after therapy. We found that vagal tone was improved immediately after therapy and gradually returned to no statistically significant improvement versus sham over a two-day period. However, liver acetylcholine levels and, resultingly, HNF4α levels were not significantly impacted until 12 hours following therapy but remained significantly improved over sham in both models for at least 24 and 36 hours respectively. Therefore, ongoing treatment is necessary to maintain the benefits of liver metabolism. However, these results suggest that even a short electrical vagal perturbation without the need for performing a non-reversible vagotomy may provide similar attenuation of cachectic symptoms, likely by resetting the relevant neural circuits. To test this hypothesis, we repeated this experiment and from days 11 and 34 onward KPC and LLC mice, respectively, received therapeutic LFACb for 30 minutes daily at the start of the wake cycle. LFACb-treated mice experienced significantly less weight loss compared to sham-treated mice, and they had dramatically improved body condition (Fig. 5C–D). Compared to sham-treated mice, LFACb-treated mice in both models also experienced statistically significant attenuation of muscle atrophy, as well as functional improvements in grip strength and ability to hang suspended from a wire by the forepaws, which was not statistically different from WT controls (Fig. 5E–G, Fig. S6F). Importantly, LFACb-treated LLC and KPC mice showed a statistically significant improvement in survival (Fig. 5H).

Figure 5: Intermittent vagal blockade for 30 minutes a day is sufficient to rescue CAC.

A. Schematic outlining the experimental time course to test electrical vagal blockade as a potential CAC therapy. B. Effects of a single 30-minute dose of LFACb therapy 0, 12, 24, and 48 hours later on vagal tone root mean square (top), hepatic acetylcholine (middle), and hepatic HNF4α (bottom). C. Weight changes presented as mean percent initial weight plus or minus SEM. Implants performed on day −5, cancer injections performed on day 0. Red arrows indicate daily LFACb therapy start date. N=48, 8 animals per group; statistical analysis by RMANOVA with Tukey HSD. D. Photographs of mice with LLC or KPC tumors and LFACb or sham implants and WT mouse with sham implants. E. Quantification of right gastrocnemius muscle fiber cross-sectional area by ANOVA with Tukey Post-Hoc. WT controls are the same for each cancer cohort. N=40, 8 animals per group. F. Grip strength of WT, KPC, and LLC mice with and without LFACb. WT controls are the same for each cancer cohort. N=40 (8 animals per group). Statistical analysis by ANOVA with Tukey Post-Hoc. G. Hanging wire test outcomes for WT, KPC, and LLC mice with and without LFACb. WT controls are the same for each cancer cohort. N=40 (8 animals per group). Statistical analysis by ANOVA with Tukey Post-Hoc. H. Survival curve for LLC and KPC mice receiving LFACb therapy or with sham implants. Statistical significance by Peto’s Log Rank Test. For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. See also Movie S2 and Figure S6.

Mice receiving LFACb saw similar behavioral benefits to vagotomized mice as well. LFACb mice of both models were more active, more social, and consumed more food than sham mice whether fed from a hanging feed box or a shallow trough on the floor (Fig. 6A–C, Fig. S6G). Similar to vagotomized mice, retention of body weight, muscle mass, and fat mass all correlated with extended survival, with muscle retention correlating most strongly, while no sex-specific differences in survival or weight retention benefit were noted (Fig. S6H–J).

Figure 6: LFACb provides benefits to behavior and chemotherapy response.

A. Spontaneous motion over 15 minutes in open field assay. WT controls are the same for each cancer cohort. N=40 (8 animals per group). B. Socialization in three chamber social assay. WT controls are the same for each cancer cohort. N=40 (8 animals per group). C. Quantification of mean daily food consumption from a cage top feeder. N=24, 8 animals per group. D. Weight changes in LLC bearing-mice receiving sham implants, LFACb only, chemotherapy only, or both therapies. N=32, 8 animals per group; statistical analysis by RMANOVA with Tukey HSD. E. Weight changes in KPC bearing-mice receiving sham implants, LFACb only, chemotherapy only, or both therapies. N=32, 8 animals per group; statistical analysis by RMANOVA with Tukey HSD. F. Quantification of tumor burden at time of sacrifice in LLC1 mice receiving sham implant, LFACb therapy, chemotherapy, or both therapies. N=32, 8 animals per condition. G. Quantification of tumor burden at time of sacrifice in KPC mice receiving sham implant, LFACb therapy, chemotherapy, or both therapies. N=32, 8 animals per condition. H. Survival curve showing longitudinal survival of LLC1-injected mice receiving sham implants or LFACb with or without chemotherapy. Statistical significance by Peto’s Log Rank Test with Bonferroni correction for multiple comparisons (6-fold). I. Survival curve showing longitudinal survival of KPC-injected mice receiving sham implants or LFACb with or without chemotherapy. Statistical significance by Peto’s Log Rank Test with Bonferroni correction for multiple comparisons (6-fold). For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by ANOVA with Tukey Post-Hoc unless otherwise stated. See also Figure S6.

Finally, to evaluate whether vagus blockade therapy could provide additive benefits in combination with chemotherapy, LLC and KPC mice were given no treatment, LFACb only, cisplatin only, or a combination of LFACb and cisplatin injections. Mice receiving any treatment retained weight longer than those without treatment. Notably, LFACb mice retained weight longer than those with chemotherapy alone, and the group which consistently had the most weight was the group receiving LFACb and chemotherapy (Fig. 6D–E). Tumor burden varied between mice that had or had not received chemotherapy, but not between those that had or had not received LFACb, suggesting that any documented benefits resulted purely from the reduction of cachectic symptoms rather than reduced tumor burden (Fig. 6F–G). Both LFACb alone and chemotherapy alone provided survival benefits compared to untreated LLC mice (Fig. 6H). Combined therapy, however, provided significant survival benefits over either therapy alone, with the mean survival benefit over untreated LLC controls nearly double that of single therapy. The survival benefit of the combined treatment was even more significant in KPC-bearing mice (Fig. 6I).

Proof of concept translational therapy using transcutaneous vagal block

For LFACb to be a clinically viable approach, non-invasive methods will be necessary as CAC patients could be too fragile for surgery. To this end, we developed a flexible, adhesive electrode patch to deliver transcutaneous LFACb (tLFACb). The same LFACb waveform was delivered by the headstage previously used in implanted animals, with amplitude empirically determined by downstream vagal recording. The patch was adhered for 30 minutes daily over the right side of the neck near the path of the vagus nerve (Fig. 7A). Delivery of tLFACb was sufficient to recapitulate the benefits of LFACb, including weight retention, improved activity level, normal socialization, and appetite retention (Fig. 7B–E). Gastrocnemius muscle fiber measurements confirmed muscle retention, supported by improved grip test and hanging wire test performance (Fig. 7F–H, S7A). Vagal tone 12-hours after treatment was comparable to those achieved by the implant (Fig. 7I). Similar survival benefits, including survival in conjunction with chemotherapy, were observed following tLFACb (Fig. 7J).

Figure 7: Transcutaneous vagal block for 30 minutes a day rescues CAC.

A. Photo of mouse with tLFACb flexible electrode patch. B. Weight changes presented as mean percent initial weight plus or minus SEM. Cancer injections performed on day 0. Red arrows indicate daily tLFACb therapy start date. N=64, 8 animals per group; statistical analysis by RMANOVA with Tukey HSD. C. Spontaneous motion over 15 minutes in open field assay. N=40, 8 animals per group. D. Socialization in three chamber social assay. N=40, 8 animals per group. E. Quantification of mean daily food consumption from a trough placed on the cage floor. N=40, 8 animals per group. F. Quantification of right gastrocnemius muscle fiber cross-sectional area. N=40, 8 animals per group. G. Grip strength of WT, KPC, and LLC mice with and without tLFACb. N=40, 8 animals per group. H. Number of falls by WT, KPC, and LLC mice with and without tLFACb during hanging wire test. N=40, 8 animals per group. I. Quantification of root mean square of vagal ENG signal 12 hours after administration of 30-minute tLFACb therapy or sham. N=25 mice, 5 per group. J. Survival curve for KPC or LLC mice receiving tLFACb therapy or with sham device with or without chemotherapy; significance by Peto’s Log Rank Test with Bonferroni correction for multiple comparisons (6-fold). For all box plots, dots represent data collected from unique mice. For all tests, * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by ANOVA with Tukey Post-Hoc unless otherwise stated. See also Figure S7.

As a proof-of-concept test of whether tLFACb could be achieved at human scale, we performed tLFACb on swine (Fig. S7B). While changes in heart rate, respiration, and temperature were not observed after 30 minutes tLFACb, vagal signaling was significantly reduced in three pigs during four tLFACb sessions occurring one week apart (Fig. S7C). Simultaneous EKG did not show abnormalities during tLFACb protocols (Fig. S7D). Histology of skin and muscle from under the tLFACb electrode patch and of major organs including heart, lung, kidney, and liver did not reveal damage or gross abnormalities (Fig. S7E).

Discussion

The complex, multi-systemic nature of CAC makes it challenging to address therapeutically. Here, we demonstrate that during carcinogenesis, systemic inflammation, including elevated CCL2 levels, drives vagal dystonia and subsequently levels of vagal neurotransmitter ACh in innervated tissues. In the liver, this results in downregulation of HNF4α. Our findings show that blockade of the right vagus nerve can preserve hepatic HNF4α expression, improve feeding behavior, mitigate CAC hallmarks, extend response to chemotherapy, and improve survival.

Our findings underscore the role of neuroinflammatory crosstalk for systemic cytokines such as CCL2 in CAC. Our previous work showed that injecting KPC cells into CCR2−/− mice, which cannot respond to CCL2, results in less severe CAC phenotypes ⁵³. The importance of CCL2 and neuroinflammation in spurring vagal dysfunction and downstream liver dysfunction highlights that neuro-immune crosstalk is a crucial feature in early disease as a potential driver of cancer-associated behavioral alterations ^{100; 101}. The observation that immune cell infiltration is reduced in CAC mice following right vagotomy suggests that, in addition to disrupting the vagal output to the liver, neuro-immune crosstalk may produce a feedback circuit wherein vagal hyperactivity supports a chronic inflammatory state.

Indeed, both vagal signaling and HNF4α have been implicated in the regulation of hepatic acute phase protein production, including LCN2, which can cross the blood-brain barrier and regulate food intake by activating the MC4R-dependent appetite-suppressing pathway in the hypothalamus ^88–90. Notably, LCN2 has been shown to induce CCL2 expression in neurons and hepatocytes^{102; 103}. LCN2 may be responsible for depression, anhedonia, and anorexic eating in cachexia ^{102; 88; 103; 104; 91; 105}. Hence, hyperactivation of the vagus nerve could lead to elevated LCN2 levels via HNF4α based on our findings.

Current development of symptomatic drug therapies such as anti-GDF15 antibody Ponsegromab highlight the need for cachexia therapies targeting the root cause of CAC ¹⁰⁶. tLFACb may be able to help patients earlier by preventing sweeping alterations that occur before clinical symptoms of cachexia rather than as a reaction to the development of severe symptoms. Unlike biological treatments such as Ponsegromab, which have meaningful liver toxicity ^{107; 108}, tLFACb appears to preserve liver function, making it an ideal addition to chemotherapy regimens with narrow margins for additional toxicity like Folfirinox, the frontline PDAC treatment ¹⁰⁹. The broad range of functional improvements seen with tLFACb could overcome the significant barriers previous drugs have faced in receiving regulatory approval improved quality of life and even survival in addition to weight retention ¹¹⁰. Broadly, our study serves as proof-of-concept for targeting the vagal brain-liver axis to benefit patients with CAC, contributing to the emerging field of neuro-metabolic regulation in health and disease.

Limitations of the Study:

While two well-established models of cancer cachexia were used, the KPC model is orthotopic whereas the LLC model is not. Unlike the KPC model, in which direct vagal innervation of the pancreas is preserved, flank-implanted LLC tumors lack vagal innervation. This anatomical difference limits direct comparability. Furthermore, mice that received vagal treatment frequently reached humane endpoints for reasons unrelated to weight loss. In the LLC model, euthanasia was primarily due to excessive tumor burden, including skin ulceration and impaired ambulation, whereas in the KPC model, mortality was predominantly caused by multi-organ metastases. These considerations highlight the translational relevance and advantage of orthotopic models in cancer cachexia research.

Although our findings identify a mechanism contributing to cachexia and offer a potential therapeutic approach, they likely do not capture the full spectrum of cachectic drivers. While targeting elevated CCL2 significantly attenuates cachectic features, it does not fully reverse the syndrome. This is consistent with prior observations that cachexia involves multiple inflammatory mediators, including IL-6 and TNFα^{49; 111; 104; 112; 16; 113}, as corroborated by our own data. These complexities may underlie the limited efficacy of single-cytokine-targeted therapies in clinical settings. Additionally, CCL2 originates from diverse sources, including immune cells, brain, liver, and tumor tissue. Future studies employing tissue-specific CCL2 deletion models will be instrumental in precisely delineating the contributions of individual compartments.

Finally, the molecular mechanisms linking acetylcholine to HNF4α remain to be elucidated. In particular, the roles of specific muscarinic versus nicotinic receptors and downstream signaling pathways merit investigation in future work.

Resource Availability:

Lead Contact:

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Xiling Shen (xshen3@mdanderson.org)

Materials Availability:

This study did not generate new unique reagents.

Data and Code Availability:

All data reported in this paper and any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. This study did not produce original code.

STAR Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human Samples

Blood was collected from PDAC patients at MD Anderson Cancer Center under the approval of the University of Texas MD Anderson Cancer Center IRB #ID98–155. Anonymized patient data including de-identified patient ID, sex, age, and racial distribution of participants is included in Supplemental Table 1 to provide readers with demographics insights while protecting patient identity. Each patient ID refers to a distinct human patient, and N reports throughout the text and figure legends related to human data refer to individual patients. Interventional studies were not performed on humans, and patients were stratified into groups based on clinical CAC diagnosis status reported in their chart at MD Anderson Cancer Center.

Blood was collected during the patients’ initial presentation at MD Anderson Cancer Center following their PDAC diagnosis using an anticoagulant such as heparin, citrate or EDTA for plasma or without anticoagulant for serum collection. Whole blood was centrifuged at 2500 rpm, and aliquot into 1.5 ml Eppendorf tube then stored at −80 °C until needed. All samples were thawed on ice, vortexed, and centrifuged at 2000×g 3 minutes at 4 °C to remove precipitate before use.

C57BL/6J Mice

C57BL/6J mice were purchased from The Jackson Laboratory (Cat# 000664, RRID:IMSR_JAX:000664) or Envigo. Immunocompetent mice were purchased at four- to six-weeks old and given two weeks to acclimate, with procedures beginning at six- to eight-weeks old. Mice of both sexes were used. Mice of each sex from the same shipment were randomly assigned to treatment groups. Mice were singly housed with a 12-hour day/night cycle in housed indoors in temperature and humidity controlled standard institutional housing with supplemental enrichment in accordance with institutional requirements; single housing was necessary to accurately measure food intake and to protect experimental mice post-operatively. All animal studies were conducted under the approval of the Duke University Institutional Animal Care and Use Committee (Protocol #A088–21-04) or the Lundquist Institute Institutional Animal Care and Use Committee (Protocol #22809–02), or by the Weizmann Institute Animal Care and Use Committee following the U.S. National Institute of Health, European Commission, and Israeli guidelines. For shHNF4α experiment, C57BL/6 male mice age ten- to twelve weeks were purchased from Envigo and randomly assigned to experimental groups. For all mouse studies, N reported in the paper and figure legends refer to unique animals.

CCR2 KO Mice

CCR2 KO mice were a kind gift from Professor Steffen Jung at the Weizmann Institute for Science and have been previously maintained by our laboratory⁵³. Original breeders of CCR2−/− C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, Maine)¹¹⁴. All mice were housed with a 12-hour day/night cycle in housed indoors in temperature and humidity controlled standard institutional housing under pathogen-free conditions due to their immunogenically altered state. Mice were monitored daily for evidence of illness. Tumor diameters were measured using a caliper. All animal studies were conducted under the approval of the Weizmann Institute Animal Care and Use Committee following the U.S. National Institute of Health, European Commission, and Israeli guidelines. For the pancreatic cancer model, 12-week-old CCR2-RFP KO mice were injected with 0.3–0.4 × 10⁶ KPC pancreatic cancer cells (in DMEM 50% Matrigel) in the pancreatic tail; sham controls were injected with PBS.

Swine

Domestic pigs were purchased from the National Swine Resource and Research Center (NSRRC) at the University of Missouri. Immunocompetent male pigs we used beginning at age 8 weeks for studies. All animal studies were conducted under the approval of the Lundquist Institute Institutional Animal Care and Use Committee (Protocol #22809–02) following the U.S. National Institute of Health guidelines. Animals were socially housed indoors in temperature and humidity controlled standard institutional cages with a 12-hour day/night cycle. For all swine studies, N reported in the paper and figure legends refer to unique animals.

KPC4508P Murine Pancreatic Cancer Cell Line

KPC4508P cell lines were a gift from Dr. Janoo Navqi and Dr. Bruce Sullenger, who originally isolated the line from the KPC genetic mouse model of pancreatic ductal adenocarcinoma. Cells were cultured at 37°C with 5% CO2 in Dulbecco’s Modified Eagle Medium (Life Technologies, 11995–065) supplemented with 10% Fetal Bovine Serum (Life Technologies, 16000–044).

LL/2 (LLC1) Lewis Lung Carcinoma Cell Line

LL/2 (LLC1) cell line (CRL-1642) was purchased from the American Type Culture Collection (ATCC, Manassas, VA). Cells were cultured at 37°C with 5% CO2 in Dulbecco’s Modified Eagle Medium (Life Technologies, 11995–065) supplemented with 10% Fetal Bovine Serum (Life Technologies, 16000–044).

Mouse Primary Neuron Cultures

Primary neuronal cultures were prepared by dissecting the vagal nodose ganglion from mice expressing Thy1-GCamp6. The nodose ganglion of the vagus nerve can be identified by isolating the vagus nerve adjacent to the common carotid artery and tracing it up to the point where the vagus enters the skull through the jugular foramen; the nodose ganglion is visible as a bulb-shaped bulge in the nerve just outside the skull. Clean the ganglion of blood and connective tissue in cold media, then wash 3x with cold HBSS with HEPES. Incubate the tissue with 2 mL collagenase for 15 minutes at 37C, then wash 2x with HBSS+HEPES to remove the collagenase. Add 2 mL of trypsin solution and incubate for 30 minutes at 37C. Move the ganglion to 2 mL of culture medium and using a small-tipped Pasteur pipette to gently dissociate the ganglion until clumps are no longer visible. Centrifuge to pellet the cells and remove the supernatant. Add 600 uL culture medium for every 2 ganglia and seed on laminin coated plates (200 uL/well). Add 2 mL of culture media to each well.

Culture the neurons in in 6-well plates coated with 200 μl laminin/well prepared at 10 mg/ml in EBSS. Laminin should be allowed at least 2 hours to incubate at 37C, followed by 2 washes with HBSS prior to use. Cells are cultured in Neurobasal A medium with 10% FBS, 38 mM D-glucose, 100 UI/ml penicillin-100 μg/ml streptomicyn, 2 mM L-glutamine, 50 ng/ml nerve growth factor, and B27 supplement. For all neural cell culture studies, N refers to individual wells of cells.

Primary Human Hepatocytes culture

THLE-2, primary human hepatocytes, were purchased from ATCC (#CRL-2706). Cells were seeded in 6 or 12 wells plates with BEGM BulletKit medium (Lonza; #CC-3171 & CC-4175), treated with 0–5uM of acetylcholine chloride (Sigma-Aldrich; #A2661). Following 48 hours, cells were harvested for RNA and protein extraction, with N referring to unique wells of cells. Cells were checked routinely to exclude mycoplasma infection.

HNF4α Silencing

Cells:

Low passage HEK293T were maintained at 37 °C with 5% CO2 in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum.

Production of rAAV:

To produce rAAV8, a triple co-transfection procedure was used to introduce a scAAV vector plasmid (pAAV-DS-CMV-EGFP-shmHNF4α or pAAV-DS-CMV-EGFP-shRL) together with pXR8, AAV8 helper plasmid carrying AAV rep and cap genes and pXX6–80, Ad helper plasmid, at a 1:1:1 molar ratio (Zincarelli et al. Mol Ther (2008) 16: 1073–1080).

Briefly, HEK293T cells were transfected using poly- ethylenimine (PEI) (linear; molecular weight [MW], 25,000) (Poly- sciences, Inc., Warrington, PA), and medium was replaced at 18 h post- transfection. Cells were harvested at 72 h post-transfection, subjected to 3 rounds of freeze-thawing, and then digested with 100 U/ml Benzonase (EMD Millipore, Billerica, MA) at 37°C for 1 h. Viral vectors were purified by iodixanol (Serumwerk Bernburg AG, Germany) gradient ultracentrifugation (according to Zolotukhin et al, Gene Therapy (1999) 6, 973–985) followed by concentration using Amicon ultra-15 100K (100,000-molecular-weight cutoff, Merck Millipore, Ireland) and washed with phosphate-buffered saline (PBS −/−).

Animal work:

Mice were injected with AAV8-shRenilla-Luciferase (shLUCIFERASE) or AAV8-shHNF4α (shHNF4α) viruses, with the concentration range of 3.5–5E+11 vg per mouse. Body weight and NMR body composition were done bi-weekly to monitor the mice well-being. On day 11–15 following the virus injection, mice were sacrificed, and tissues were harvested for further analysis.

Cancer Cell Injection

For the pancreatic cancer model, 12-week old C57BL/6 and CCR2^rfp/rfp male mice were injected with 3–4×10⁵ cells into the pancreatic tail. KPC4508P cells were counted, suspended in a mix of 50% Matrigel and culture medium, and placed on ice. Mice were anesthetized with 2% vol/vol isoflurane and a small incision was made on the left side 0.5 cm inferior to the ribcage. The spleen was identified, and the pancreatic tail was exposed for orthotopic injection. 3–4×10⁵ cells in 30 μl were injected per mouse. For analgesia, mice were injected subcutaneously with bupivacaine (0.25%, <2 mg/kg).

LL/2 (LLC1) cells were suspended in culture medium, counted, and placed on ice. Mice received subcutaneous flank injections while manually restrained with a bolus of 1×10⁶ cells.

METHOD DETAILS

Human Sample Profiling

Acetylcholine levels were measured from serum by ELISA using Biomatik kit EKN43185 according to the manufacturer’s direction. CCL2 levels were measured from serum by ELISA using PeproTech kit BGK13500 according to manufacturer directions. 80-plex human cytokines were assessed from plasma by Luminex using ProcartaPlex^™ Human Immune Response Panel, 80plex (Thermofisher Scientific EPX800–10080-901) according to the manufacturer’s instructions. Samples were processed in triplicate and investigators responsible for sample processing and assay conduction were blinded to patient status.

Implantable and Transcutaneous Vagal Blockade Devices

The implantable vagal block device consists of four essential elements; a PC running StimWare^® software (Triangle BioSystems International), an RF USB transceiver, the programmable constant current output headstage, and a microwire hook electrode. The RF transceiver facilitates full duplex communication between the operator PC and the headstage, and consists of a USB dongle and antenna that transmits and receives at 2.4 GHz. The headstage is a custom mixed-signal PCB which fundamentally encompasses an RF transceiver, a rechargeable lithium ion battery, a microcontroller, a DAC, a demultiplexer, a dc power supply splitter and a constant current driver with seven channels of biphasic constant current output, only one of which is used in these studies. The microwire hook electrodes were manufactured in-house at Duke University department of Electrical and Computer Engineering by cutting down silicone insulated ultra-fine stainless steel AWG40 hookup wire (Cooner Wire Company, AS 631) to approximately 6 cm. One end of the wire was carefully stripped of the silicone coating for a length of 1 mm under a dissecting microscope. The exposed wire was curled with forceps around the tip of a Lasik Flap Lifter (World Precision Instruments, WP302538) to create a hook with approximate 0.5 mm internal diameter. The exterior surface was lightly coated with Loctite^® 405 (Ellsworth, 135436). Roughly 1 cm of insulation on the other end of the wire was stripped to serve as contact leads.

The transcutaneous block device consists of the same elements, except that the microwire hook electrode is replaced by a flexible electrode patch. Flexible electrode patches were made by graphene printing conductive traces onto silicone elastomer. The edges were lined with double sided medical tape to adhere it to the skin. Copper foils (20 μm thick) were used to fabricate EMG electrode arrays, which were connected to the PCB. To ensure effective adhesion to the silicone elastomer, a polyimide film (PI2545 precursor, HD MicroSystems) was spin-coated onto the copper foil at 4,000 r.p.m. for 60 seconds. The polyimide layer underwent soft baking at 100°C for 5 minutes, followed by hard baking at 300°C for 1 hour in a nitrogen atmosphere. Ultraviolet light was used to activate the polyimide for 3 minutes before laminating the copper foil/polyimide onto polydimethylsiloxane (SYLGARD 184)-coated glass.

The electrodes were fabricated through laser ablation (LaserMarks) with a wavelength of 1,059–1,065 nm, a power of 0.228 mJ, and a speed of 300 mm/s. The patterned electrodes were then transfer-printed onto Ecoflex 00–30 (Smooth-On)-coated glass using water-soluble tape (3M) after 3 minutes of ultraviolet light activation. To enhance bonding, the electrodes and Ecoflex-coated glass were placed in an oven at 80°C for 10 minutes. Finally, the water-soluble tape was washed away.

Sham animals had leads placed but no current delivered via their electrodes. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Chemotherapy

Mice in the therapeutic group received 3mg/kg of cisplatin chemotherapy injected daily interperitoneally beginning on day 8 until death or euthanasia for humane endpoints based on previously reported studies ⁵⁷. Control mice received daily intraperitoneal injections of sterile saline. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Anamorelin Therapy

Anamorelin was delivered orally by dissolution in the sterile free access drinking water at 3 mg/ml. Dosage was determined by usage in previous studies ^{75; 76}. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Bindarit Therapy

Beginning on day 8 post-cancer injection, mice were IV injected with 100 mg/kg Bindarit or PBS every other day. Dosage was determined based on previously published studies ⁶⁶. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Hexamethonium and Atropine Therapy

Mice were injected IP daily with 30 mg/kg atropine, 10 mg/kg hexamethonium, or PBS daily. Dosage was based on previously published studies ^{115; 116}. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Vagotomy

Mice were anesthetized with 2% vol/vol isoflurane. Mice received ophthalmic ointment to prevent damage to the eyes. Mice were injected subcutaneously with bupivacaine (0.25%, <2 mg/kg) and meloxicam (2 mg/kg) prior to surgery to provide analgesia. The surgical site was cleaned using a razor to remove hair, followed by three alternating washes with iodine and 70% ethanol.

Mice receiving cervical vagotomies had a midline incision roughly 5 mm in length made over the throat. The submaxillary glands were gently pushed to either side of the surgical field, with blunt dissection of connective tissue as necessary. Retractors were used to hold the skin and glandular tissue open. Under a dissecting microscope, the sterohyoideus and omohyoideus were also gently parted and retracted. The esophagus was identified as a landmark, and the carotid artery used as an additional landmark; mice received either left vagotomy or right vagotomy as defined from the animal’s perspective, which determined the side of dissection for all further steps. The common carotid artery was gently manipulated with blunt forceps to allow visualization of the vagus nerve. The vagus was isolated from surrounding tissues using delicate blunt dissection with forceps or scraping of connective tissue with the point of a Lasik Flap Lifter. The nerve was then hooked over the tip of a glass rod to maintain access, and cuts were made with a pair of microdissecting scissors to remove a 5 mm segment of nerve. The remaining nerve ends were ensured to be spatially separated to limit the potential for regrowth. The incision was then sutured closed.

Mice receiving subdiaphragmatic vagotomies had a 5–10 mm incision made on the animal’s left side just below the rib cage. The diaphragm, liver, and stomach were visualized as landmarks. Retractors were used to hold open the incision, while forceps were used to gently manipulate the stomach to expose the esophagus. Under a dissecting microscope, the vagus nerve was identified running alongside the esophagus on either side. As the vagus crosses sides between the cervical level and the subdiaphragmatic level, to achieve a subdiaphragmatic right vagotomy the vagal branch extending along the animal’s left side of the esophagus (viewer’s right) is severed. Sharp forceps are used to gently separate the nerve from surrounding tissues, and a 3 mm segment of nerve is severed, maintaining separation between remaining ends. The abdominal muscle and skin are then sutured closed.

Sham animals underwent surgery to expose and visualize the vagus nerve, but no direct manipulation or contact with the nerve was made to prevent unintentional damage.

Mice were monitored closely until fully recovered from anesthesia, and then monitored every 12 hours post-operatively for complications. Mice received meloxicam (2 mg/kg) once daily for 48 hours post-operatively for pain management.

Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group. Vagotomy studies were performed at each of the three publishing institutions to ensure model replicability across locations and investigators.

Implantation of Electrodes for Low Frequency Alternating Current Block

Mice were anesthetized with 2% vol/vol isoflurane. Mice received ophthalmic ointment to prevent damage to the eyes. Mice were injected subcutaneously with bupivicane (0.25%, <2 mg/kg) and meloxicam (2 mg/kg) prior to surgery to provide analgesia. The surgical site was cleaned using a razor to remove hair, followed by three alternating washes with iodine and 70% ethanol.

A midline incision roughly 5 mm in length is made over the throat. The submaxillary glands were gently pushed to either side of the surgical field, with blunt dissection of connective tissue as necessary. Retractors were used to hold the skin and glandular tissue open. Under a dissecting microscope, the esophagus was identified as a landmark, and the carotid artery used as an additional landmark. The right common carotid artery was gently manipulated with blunt forceps to allow visualization of the vagus nerve. The vagus was isolated from surrounding tissues using delicate blunt dissection with forceps or scraping of connective tissue with the point of a Lasik Flap Lifter. The nerve was then hooked over the tip of a glass rod to maintain access while minimizing contact with electrically conductive tools. The nerve is placed into the curve of the microwire hook electrode, and forceps are used to gently pinch the open end of the hook closed. The electrode is then secured in position to minimize tension on the nerve using a combination of sutures and Loctite^® 405 adhesive (Ellsworth, 135436). A ground electrode was then embedded in nearby muscle tissue and similarly secured in place. Contact between the electrode and the vagus was verified by delivering a short activating pulse to evoke a throat muscle twitch. The unsecured ends of both wires were then tunneled subcutaneously to the back of the animal’s neck, leaving sufficient slack length as to not impede motion or cause tension during motion. Wire ends were trimmed to leave approximately 1 cm of exposed lead and secured at the back of the neck using a suture and a dab of Loctite^® 405. Robust contact between the electrode and nerve is verified again to ensure tunneling did not cause electrode displacement, and titration was completed to establish minimal current amplitude required to produce vagal block as established by downstream recording from the subdiaphragmatic vagus nerve. With personalized vagal block program noted, the incision is sutured closed.

Mice were monitored closely until fully recovered from anesthesia, and then monitored every 12 hours post-operatively for complications. Mice received meloxicam (2 mg/kg) once daily for 48 hours post-operatively for pain management. A total of nine days elapse between implant of the device and first therapeutic neuromodulation session; seven days of recovery prior to injection and two days after injection to allow establishment of the tumor prior to initiation of treatment. Sham animals underwent the same implant procedure as block animals, but had no current delivered during treatment sessions. Animals were randomly assigned to intervention group at arrival such that both sexes were evenly represented in each intervention group.

Electrode Placement for Transcutaneous Low Frequency Alternating Current Block

Mice were anesthetized with 2% vol/vol isoflurane the day prior to first treatment. Mice were shaved in an approximately 1 inch region covering the right side of the neck from jaw to clavicle over the approximate path of the vagus nerve to ensure a clear surface for electrode patch placement. The flexible electrode patch was placed over the cleared area with approximately 100–200 ul of Spectra360 conductive gel as an interface. Dose titration was completed to establish minimal current amplitude required to produce vagal block as established by downstream recording from the subdiaphragmatic vagus nerve. On each day thereafter, immediately before the LFACb protocol a depilatory cream was applied to the area then wiped with a clean square of gauze under manual restraint before placement of the patch as described above immediately before beginning the LFACb protocol outlined in the next section.

Pigs were sedated with 5 mg/kg telazol and 2 mg/kg xylazine intramuscular injection, then intubated and anesthetized with 2% vol/vol isoflurane. Necks were shaved in an approximately 3-inch region covering the right side of the neck from jaw to clavicle over the approximate path of the vagus nerve to ensure a clear surface for electrode patch placement. The flexible electrode patch was placed over the cleared area with approximately 200–500 ul of Spectra360 conductive gel as an interface. Dose titration was completed to establish minimal current amplitude required to produce vagal block as established by downstream recording from the subdiaphragmatic vagus nerve.

Vagal Tone Recording in Mice

Mice were anesthetized with 2% vol/vol isoflurane. Mice received ophthalmic ointment to prevent damage to the eyes. Mice were injected subcutaneously with bupivicane (0.25%, <2 mg/kg) and meloxicam (2 mg/kg) prior to surgery to provide analgesia. The surgical site was cleaned using a razor to remove hair, followed by three alternating washes with iodine and 70% ethanol.

Mice undergoing vagal recording had electrodes implanted on the subdiaphragmatic right vagus nerve. To achieve this, a 5–10 mm incision made on the animal’s left side just below the rib cage. The diaphragm, liver, and stomach were visualized as landmarks. Retractors were used to hold open the incision, while forceps were used to gently manipulate the stomach to expose the esophagus. Under a dissecting microscope, the vagus nerve was identified running alongside the esophagus on either side. Sharp forceps are used to gently separate the right subdiaphragmatic nerve from surrounding tissues, and a cardiac pacing electrode (Medtronic, 6494) was shaped into a corkscrew and hooked around the vagus nerve. Another cardiac pacing electrode to serve as the ground electrode was placed in the adjacent abdominal wall. The recording electrodes were connected to a Model 1800 2-Channel microelectrode AC amplifier (A-M Systems) with low cut-off filter set to 0.1 Hz, high cut-off filter set to 20 kHz, and gain set to X100 with output to a National Instruments BNC-2110; recordings were sampled at 50 kHz and saved using LabView SignalExpress (National Instruments). Graphing and post-processing was completed using MATLAB R2022a (MathWorks). Recordings were tied to an animal ID and timepoint, with no indicator of treatment assignment within recording files; individuals were blinded to treatment condition during post-processing, quantification, and graphing.

Low Frequency Alternating Current Block in Mice

To perform awake, non-anesthetized vagal block similar to the expected protocol in humans, mice were gently secured in clear plastic restraint cones (Fisher Scientific, 14–370-110) secured with zip ties. During the week prior to use in the LFACb treatment procedure, mice were habituated to the restraint cones by daily introduction for 30 minutes in absence of any procedure. A small incision was made in the top of the cone to allow passage of the ground and electrode lead wires through the bag. The device headstage was connected to the leads using alligator clips for the duration of the block protocol; 30 minutes of biphasic charge balanced unipolar neuromodulation with frequency 5 Hz, pulse durations 100 ms, and charge amplitude of 50 – 350 μA determined on a per-animal basis as described above. Sham animals were also habituated to the restraint cones and placed in them for 30 minutes daily, with no stimulus delivered. Mice were then weighed, their tumors were measured with calipers to assess tumor burden, and they were returned to their home cage, with approximately 35 minutes elapsed.

Vagal Stimulation, Transcutaneous Block, and Vagal Tone Recording in Swine

Pigs were sedated with 2–8 mg/kg telazol and 1–3 mg/kg xylazine by intramuscular injection, then intubated and anesthetized with 2% vol/vol isoflurane. Pigs received 0.05 mg/kg buprenorphine injected subcutaneously every 8–12 hours for 48 hours after surgery for analgesia. The surgical site was cleaned using a razor to remove hair, followed by three alternating washes with iodine and 70% ethanol.

A 6- to 8-inch incision was opened on the right side of the neck beginning at the jaw line. The esophagus and common carotid artery were visualized as landmarks. Cuff electrodes (Microprobes for Life Science) with 2mm inner diameter, 3 100μm Platinum contacts separated by a distance of 10mm were placed on the vagus nerve at the proximal head of the incision site for stimulation and 5–6 inches distally for recording. Leads were tunneled to the surface of the skin and the incision site sutured closed, with approximate midpoint between electrodes over the vagus nerve marked on the surface of the skin. Stimulation, block, and recording sessions were performed four times per animal, separated by one week.

Stimulation was performed at 14 Hz, 100 mA square wave on the proximal electrode. tLFACb electrodes were placed on the surface of the skin midway between the stimulation and recording electrodes. tLFACb was achieved with 30 minutes of biphasic charge balanced unipolar neuromodulation with frequency 5 Hz, pulse duration of 100 ms, and charge amplitude of 5 – 30 mA determined on a per-animal basis empirically via downstream recording. Recording electrodes were connected to a Model 1800 2-Channel microelectrode AC amplifier (A-M Systems) with low cut-off filter set to 0.1 Hz, high cut-off filter set to 20 kHz, and gain set to X100 with output to a National Instruments BNC-2110; recordings were sampled at 50 kHz and saved using LabView SignalExpress (National Instruments). Graphing and post-processing was completed using MATLAB R2022a (MathWorks). Recordings were tied to an animal ID and timepoint, with no indicator of treatment assignment within recording files; individuals were blinded to treatment condition during post-processing, quantification, and graphing.

Daily Food Consumption Assessment

Mice were provided unrestricted access to food at all times, administered in a cage top basket feeder or a shallow trough on the cage floor as indicated for each experiment. The food trough was weighed daily at the beginning of the wake cycle, coinciding with LFACb delivery in the VNM animal cohort. Daily change was recorded to reflect food consumption, and the food level was topped up as necessary. Cages were regularly checked for unconsumed food, and no hoarding of food pellets in the bedding was noted, nor were large volumes of pellet dust consistent with shredding/gnawing outside mealtimes observed. Daily food weight measurements were taken by facility staff and tied to cage numbers; these staff were blinded to treatment assignment of the animals therein.

Gastrocnemius Electromyogram

Mice were anesthetized with 2% vol/vol isoflurane. Mice received ophthalmic ointment to prevent damage to the eyes. Mice were injected subcutaneously with bupivicane (0.25%, <2 mg/kg) and meloxicam (2 mg/kg) prior to surgery to provide analgesia. The surgical site was cleaned using a razor to remove hair, followed by three alternating washes with iodine and 70% ethanol.

An approximately 1 cm incision was made on the right flank of the animal below the level of the femur. Skin and connective tissue was displaced to visualized the muscles of the leg. The biceps femoris were gently pushed upward with forceps to reveal the sciatic nerve, with effort made to minimize incisions in muscle and surrounding tissue to access the nerve. A microwire hook electrode was placed around the sciatic nerve and connected to a current delivery implant headstage, with ground placed in the adjacent muscle tissue. The stimulation electrodes were connected to the custom PCB headstage. If necessary, the incision was extended to visualize the gastrocnemius muscle. A cardiac pacing electrode (Medtronic, 6494) was implanted directly into the gastrocnemius with sharp forceps and anchored with a drop of Loctite^® 405 (Ellsworth, 135436) between the insulation and the adjacent skin, and a ground electrode placed in the adjacent subcutaneous tissue. The recording electrodes were connected to a Model 1800 2-Channel microelectrode AC amplifier (A-M Systems) with low cut-off filter set to 0.1 Hz, high cut-off filter set to 20 kHz, and gain set to X100 with output to a National Instruments BNC-2110; recordings were sampled at10 kHz and saved using LabView SignalExpress (National Instruments). Recordings were tied to an animal ID and timepoint, with no indicator of treatment assignment within recording files; individuals were blinded to treatment condition during post-processing and quantification.

Anesthetized Heart Rate and Liver Bleed Volume

Heart rate of anesthetized mice was collected via a SomnoSuite^® outfitted with a soft touch SpO₂ paw sensor five minutes after anesthesia induction. Liver bleeding rate was measured by modifying the protocol previously described in ¹¹⁷ by pre-weighing absorptive pads and placing them inside the abdominal cavity and resecting 75% of the left-middle lobe of the liver. The abdominal cavity was closed and the mice were allowed to bleed for 30 minutes, at which time they were euthanized and the absorptive squares were measured to identify the collected blood mass.

Euthanasia and Sample Collection

Mice were anesthetized with 2% vol/vol isoflurane. Due to the nature of the study, some humane endpoint extensions were granted under the protocol to allow assessment of clinically relevant features of cachexia in late-stage cancer. Mice were euthanized at a weight loss of 30% assessed by daily checks, inability to locomote due to tumor burden, ulceration over the tumor site, tumor diameter over 20 mm in any dimension, or other body condition deterioration or behavioral changes suggesting pain outside the acute recovery period for procedures or failure to respond to analgesia during recovery. Mice were weighed and blood was collected via terminal cardiac puncture, followed by secondary euthanasia via decapitation. Serum was isolated by centrifuging blood at 4°C for 10 minutes at 1000×g after being allowed to clot for 30 minutes at room temperature; plasma was isolated by collecting blood in EDTA tubes and storing on ice until centrifugation at 4°C for 10 minutes at 1000×g. Tissues, including tumor, liver, muscle, and vagus nerve, were collected, and either flash frozen in liquid nitrogen for molecular processing or embedded in OCT for sectioning and histology. Additionally, some brain, muscle, and cervical vagus nerve specimens were collected and fixed in 4% PFA solution for 24 hours, then embedded in paraffin blocks for sectioning and histology.

Pigs were euthanized by barbiturate overdose immediately following their terminal block/recording session while still under anesthesia. Tissue samples, including skin, muscle, liver, kidney, heart, and lung, were fixed in 10% formalin for 24 hours then embedded in paraffin for sectioning and histology.

Animals were only excluded if a clear reason for biological outlier status could be identified at euthanasia. Examples of conditions which resulted in exclusion were situs inversus, development of post-operative infection, or removal/damage of implants.

Liver Function Test

Plasma was extracted by collecting whole blood in lithium-heparin coated tubes (VetMarket #5598006) and centrifugation at 1,000×g for 10 minutes at 4°C. Liver function parameters were measured using the VETSCAN machine (Abaxis, VETSCAN VS2) with liver profile cassettes (Abaxis-500–0040). Albumin levels were measured as mentioned above, using 100ul of liver lysates extracted with RIPA buffer. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Gas Chromatography–Mass Spectrometry

Liver samples were vacuum-dried by lyophilizer. Dry tissues (5–20mg) were powdered using 3.2mm beads (Next Advance #ADV-SSB32) on CryoMill (Retscht), resuspended with methanol and ribitol as an internal standard and sonicated for 20 minutes. Extraction of polar metabolites was done by adding of 1 volume of water and 0.5 volume of chloroform. Samples were vortexed and centrifuged at 14,000 RPM for 15 minutes at 4°C, and supernatant was taken for drying over-night. Derivatization was done by the addition of methoxyamine hydrochloride (ThermoFisher # A19188.06) solution (20 mg/mL in pyridine) at 37°C for 90 minutes, followed by incubation with N,O-bis (trimethylsilyl) trifluoroacetamide (Sigma-Aldrich #15222) at 37°C for 30 minutes. Samples (1.2ul) were injected in splitless mode, using an inlet temperature of 270°C, and the gas chromatography oven was held at 100°C for 3 minutes and then ramped to 300°C with a gradient of 3°C/minute followed by 5 minutes after each run at 315°C. The MS system was operated under electron impact ionization at 70 eV, and a mass range of 30 to 500 amu was scanned. Peaks were analyzed in MassHunter software (Agilent Technologies) and normalized to internal standard and dry weight. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Plasma amino acids

Samples were diluted with DDW 1:10 or 1:100 to get signals of proper amino acids in the linear range of concentrations.

LC-MS/MS: Each sample (10 μL) was mixed with 10 μL of internal standard (Stable Isotope Labeled Amino Acid Mixture, Sigma-Aldrich), agitated for 15 min, derivatized using the AccQTag method, and filtered through 0.2 μm PES nanoFilter vials (Thomson). Quantification was done using a standard curve (0.01–10 μM)*. The LC-MS/MS system comprised an Acquity I-Class UPLC and a Xevo TQ-S triple quadrupole mass spectrometer (Waters) with chromatographic separation on a UPLC HSS T3 column (100 × 2.1 mm i.d.) and a 1 μL injection volume, as previously described ¹¹⁸. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

H&E Staining

Paraffin-embedded tissues were sectioned with a microtome to 5 μm. Slides were deparaffinized by 3×5-minute washes in xylenes, 3×5-minute washes in 100% ethanol, 1×5-minute wash in 95% ethanol, and 1×5-minute wash in 70% ethanol. Slides were then washed in distilled water for 5 minutes. Slides were stained for 10 minutes in Weigert’s iron hematoxylin, then rinsed in warm running tap water for 10 minutes. Slides were briefly rinsed with distilled water, then stained with eosin yellowish for 1 minute. Slides were washed with distilled water, then dehydrated and cleared by 30 seconds in 95% ethanol, 30 seconds in 100% ethanol, then 30 seconds in xylenes. Coverslips were mounted with Permount^™ (Fisher Scientific, SP15500). Images were taken on an Echo Revolution microscope. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Gastrocnemius Muscle Staining

OCT-embedded gastrocnemius muscle was sectioned on a cryotome to a thickness of 5 μm and stained with Wheat Germ Agglutinin, Alexa Fluor^™ 488 Conjugate (ThermoFisher Scientific, W11261) according to manufacturer instructions. Briefly, stock solution was prepared by dissolving 5 mg of lyophilized stock in 5 mL of PBS and aliquoted for storage at −20°C. Slides were rehydrated with 3×5-minute washes with HBSS. Stock WGA was diluted 1:200 for a final concentration of 5 μg/mL and applied for 10 minutes at room temperature. Slides were then washed with HBSS 3 times for 5 minutes. Slides were mounted with Fluoromount-G^™ (ThermoFisher Scientific, 00–4958-02) and imaged on a Leica DMIL inverted fluorescence microscope.

Immunofluorescence Staining

Paraffin-embedded tissues were sectioned with a microtome to 5 μm. Slides were deparaffinized by 3×5-minute washes in xylenes, 3×5-minute washes in 100% ethanol, 1×5-minute wash in 95% ethanol, and 1×5-minute wash in 70% ethanol. Slides were then washed in distilled water for 5 minutes. Slides were blocked in 2% BSA for 30 minutes. Antigen retrieval was performed by steaming slides in citrate buffer for 30 minutes. Primary antibodies were used at a 1:500 dilution in blocking buffer (Iba1: sc-32725, Santa Cruz Biotechnology; Phox2b: 83811S, Cell Signaling Technology; CD45: MAB114-SP, R&D Systems; CD80: 12–0801-82, ThermoFisher Scientific; CD11b: 101241, BioLegend; cFos: CBL440, Sigma Aldrich) and incubated overnight at 4C. Slides were washed 3x in PBS, then if antibodies were not pre-conjugated, AlexaFluor secondary antibodies were used to conjugate with spectrally separated fluorophores by incubation at a 1:1000 dilution in blocking buffer at room temperature in a dark chamber for 2.5 hours. Slides were washed 3 times with PBS and coverslips were mounted with Vectashield with DAPI (H-1500–10, Vector Laboratories). Slides were imaged on a Zeiss LSM710 confocal microscope. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

RNA Extraction and Quantitative PCR

RNA was extracted from livers and muscles using a Direct-zol RNA MiniPrep kit (Zymo Research ZR-R2050) according to the kit protocol and stored at −80°. RNA from cultured primary hepatocytes (THLE-2 cells) was extracted by using RNeasy Mini Kit (Qiagen; #74104) and stored in −80°. cDNA was synthesized from 1ug RNA, with the qScript cDNA synthesis kit (Quanta; #95749). For RT-PCR SYBR Green FastMix perfect CT (Quantabio; #95073) was used, with the specific primers (Table S2). Samples were run in triplicate, and identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Western Blotting

Liver tissues were ground with 3.2mm beads (Next Advance; #ADV-SSB32) and lysed in RIPA lysis buffer (Sigma-Aldrich; #R0278–50ML), containing 1% phosphatase inhibitor cocktail (Sigma-Aldrich; #P5726), and 1% protease inhibitor cocktail (Sigma-Aldrich; #P8340). Briefly, samples were incubated on ice for 15 minutes, and centrifuged. Supernatant was collected and protein content was measured by BCA Protein Assay Kit (ThermoFisher; #23225). 30–50ug protein were loaded from each sample into each lane and separated by electrophoresis on a 10% SDS polyacrylamide gel. The proteins were transferred to a nitrocellulose membrane (TAMAR; #10401383). Blocking was done by incubating the membranes with TBST [10 nmol/L Tris-Hcl (pH 8.0), 150 mmol/L NaCl, 0.1% Tween 20], containing 5% skim milk for 1 hour at room temperature. Membranes were incubated with primary antibodies over-night in 4°, followed by an incubation of 1 hour in room temperature with donkey anti-rabbit IgG, or donkey anti-mouse IgG secondary antibodies (Abcam; #ab98799, #ab97085), and ECL (ThermoFisher; #32106). Blots were imaged by Gel Doc XR+ (Bio-Rad) and quantified by ImageLab software (Bio-Rad). The relative intensity of each band was calculated by dividing each specific band intensity by the value of its house-keeping protein.

ELISA Assay

Mouse CCL2 was measured using an ELISA assay (R&D Systems, MJE00B) according to the manufacturer’s instructions. Serum and CSF were used directly for analysis; to assess tissue CCL2 levels organs were lysed in RIPA lysis buffer (Sigma-Aldrich; #R0278–50ML), containing 1% phosphatase inhibitor cocktail (Sigma-Aldrich; #P5726), and 1% protease inhibitor cocktail (Sigma-Aldrich; #P8340). Briefly, samples were incubated on ice for 15 minutes, and centrifuged. 50 μL of tissue lysate was used per well. Acetylcholine was measured using an ELISA assay (BIOMATIK, EKE62518) according to the manufacturer’s instructions. Samples were run in triplicate identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Flow-Cytometry Assisted Cell Sorting

Mouse brains were extracted from euthanized mice and placed in a 15 mL tube with 5 mL of RPMI buffer with 10% FBS. Brains were homogenized by drawing into a 5mL syringe fitted with a blunt-tipped 18g needle 20 times or until well-mixed, using care to minimize bubbles. Homogenates were spun at 800×g for 10 minutes at 4°C. Supernatant and fat were aspirated off, and pellets were resuspended in 3 mL of a solution comprised of 1.25 mL collagenase A, 8.25 mL HBSS, and 0.5 mL of DNAse I by pipetting 10 times with a 1mL pipette, then incubated in a 37°C water bath for 20 minutes. Samples were diluted to 15 mL with RPMI, then spun 800×g for 10 minutes at 4°C. Supernatant was aspirated, and pellets were resuspended in 3 mL of 27% Percoll (Sigma Aldrich P1644) by vortexing aggressively. Samples were spun at 1420×g for 30 minutes at room temperature with the brake off. Myelin and media were aspirated off. Cells were resuspended in 250 uL of RPMI. Cells were stained with a panel of antibodies for FACS sorting including zombie violet fixable viability dye (biolegend 423113), CD4 (invitrogen 416–0042-82), CD8a (biolegend 155006), CD11b (biolegend 101242), F4/80 (biolegend 123108), CD86 (biolegend 105026), and CD206 (biolegend 141706). Cells were then fixed in 4% PFA and sorted on a BioRad ZE5 flow cytometer. Samples were identified by a unique animal ID without indication of treatment group, allowing for blinding during the assessment.

Behavior Assays

Mice underwent a battery of behavior tests including open field assay, three-chamber social assay, grip test, and hanging wire test. Mice were acclimated to the testing room 1 hour prior to testing. Equipment was cleaned with 70% ethanol prior to use and between mice.

For the open field assay, mice were placed in the center of a 3-foot opaque acrylic open topped cube and allowed to explore the area for 15 minutes. Movement was captured with an overhead camera and quantified with MouseMove ¹¹⁹.

For the three-chamber social assay, mice were placed in the center chamber a 3 foot open topped opaque acrylic cube divided into three equal compartments. During the 10-minute habituation phase both outer chambers contain inverted empty mesh cups. During the social phase, a stranger WT B6/J mouse of the same sex was placed under the mesh cup on one side. Preference for novel socialization was measured by comparing the time spent in the stranger chamber versus other chambers. Adapted from previously published protocols ¹²⁰.

For the hanging wire test, a 2mm thick wire was tied between two vertical stands approximately 20 inches apart and 12 inches above a clean cage containing bedding to prevent fall-related injury. Upon grasping the wire, the number of falls the mouse underwent within a 2-minute time period was recorded. Adapted from previously published studies ¹²¹.

For the grip test, a grip strength meter was positioned horizontally and animals were positioned to grasp a mesh grid with all paws before being pulled straight out from the meter by the base of the tail. Peak force was recorded for 5 consecutive trials and the maximum load in grams reported. Adapted from previously published methods ¹²².

QUANTIFICATION AND STATISTICAL ANALYSIS

Cross-Sectional Fiber Area Quantification

Muscle fiber area was assessed using the muscle morphometry plug-in (Anthony Sinadinos using Eclipse IDE) and FIJI¹²³ (FIJI, ImageJ, NIH) with examiners blinded to the experimental group assignment of the source tissue.

Other Quantification

Cell counts or positivity were quantified using threshold levels manually determined by an investigator and used consistently for all images derived from a single staining using FIJI¹²³ (FIJI, ImageJ, NIH), with examiners blinded to the experimental group assignment of the source tissue.

Statistical Analysis

All statistical analyses were completed in JMP Pro 16 (JMP Statistical Discovery) or GraphPad Prism (version 9.4.1). Survival curves were analyzed for significant differences via Peto’s Log-Rank Test. Two-group comparisons were made by t-tests with Bonferroni correction. All other analyses were completed using ANOVA with Tukey HSD for multiple comparisons. For parametric tests, normality was assessed with Shapiro-Wilk test. For all analyses, n represents measurements from unique organisms or wells of cells, and individual measurements are shown as points over the summary statistic box plots. Summary statistic box plots show the mean and inner quartiles with outer quartiles bounded by the whiskers. N and statistical tests for all data are explicitly stated in the relevant figure legends. For all analyses, * indicates p<0.05, ** indicates p<0.01, *** indicates p<0.001, and **** indicates p<0.0001.

Supplementary Material

1

Movie S1. Video of a mouse demonstrating a leg kick following sciatic nerve stimulation, representative of the response measured in the EMG recording setup. Related to Figure 2.

2

Movie S2. Video of contact testing of a hook electrode implanted on the right cervical vagus nerve. When stimulation is elicited, a local muscle twitch is evoked. Related to Figure 4.

figs3

Figure S3: Right vagotomy or chemical block of the right vagus provides unique protective benefits against multiple clinically relevant CAC symptoms without liver or tumor denervation, related to Figure 3. A. Intraoperative photos showing left and right cervical vagus nerves in situ. B. Quantification of weight change at 30 days post-injection, corrected for tumor burden, in mice which received sham surgeries, left vagotomy, or right vagotomy in addition to LLC cell injection, compared to mice which received sham surgeries and PBS injection. N=48 (16 per sham group, 8 per vagotomy group) C. Intraoperative photo showing the right subdiaphragmatic vagus in situ. D. Quantification of weight change at 30 days post-injection, corrected for tumor burden, in mice which received sham surgeries, right subdiaphragmatic vagotomy, or right cervical vagotomy in addition to LLC1 cell injection, compared to mice which received sham surgeries and PBS injection. N=48 (16 per sham group, 8 per vagotomy group). E. Left: quantification of right gastrocnemius muscle fiber cross-sectional area in mice which received vagal botox microinjection compared to sham surgeries in WT or LLC mice. N=24, 8 mice per condition. Right: Representative muscle H&E at 10x magnification. F. Longitudinal weight change of WT control mice with sham surgery compared to LLC mice which received sham surgeries or microinjection of botulinum toxin A (botox) into the right vagus nerve. N=32 (8 animals per group). Statistical significance by RMANOVA with Tukey post-hoc. G. Survival curve for LLC mice which received sham surgery or vagal botox microinjection. Statistical significance by Peto’s Rank Log test. H. Left: Quantification of right gastrocnemius muscle fiber cross-sectional area in LLC-injected mice with sham surgery, vagotomy, or oral anamorelin compared to WT controls which received sham surgeries. N=32, 8 mice per condition. Right: Representative muscle H&E at 10x magnification. I. Longitudinal weight change of WT control mice with sham surgery compared to LLC mice which received sham surgeries, right cervical vagotomy, or oral anamorelin treatment. N=32 (8 animals per group). Analysis by RMANOVA with Tukey HSD. J. Survival curve for LLC mice which received sham surgery, vagotomy, or oral anamorelin compared to WT controls which received sham surgeries. Statistical significance by Peto’s Rank Log test. N=32, 8 mice per condition. K-P. Impacts of hexamethonium, atropine, or PBS treatment in KPC or LLC mice and WT controls on: K. CD45+ cell infiltration into the vagal dorsal motor nucleus (VDMN), L. Neuroinflammation in the VDMN, M. Acetylcholine levels, N. CCL2 levels, O. Muscle atrophy (Left: Representative gastrocnemius H&E at 10x magnification, Center Left: Quantification of gastrocnemius muscle fiber cross sectional area, Center Right: Quantification of gastrocnemius muscle weight, Right: Grip strength), and P. Spontaneous activity and socialization. N=56, 8 per condition. P. Left: representative IF staining of liver for neural fibers with Tuj1. Right: Quantification of liver innervation in KPC and LLC mice with and without vagotomy vs. WT controls (n=25, 5 per group). Q. Left: representative immunofluorescence staining of tumors for nuclear neuronal marker HuC/D and autonomic nerve marker Phox2b. Right: Representative immunofluorescence (IF) staining of tumors for Tuj1, a cytoplasmic neuronal marker. Center: Quantification of tumor innervation in KPC and LLC mice with or without vagotomy (n=20, 5 per group). For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Unless otherwise noted, statistical analysis by ANOVA with Tukey Post-Hoc.

figs5

Figure S5: HNF4α depletion produces cachectic phenotypes; vagotomy does not cause liver dysfunction, related to Figure 4. A. RT-qPCR analysis demonstrates decreased expression of fatty acid oxidation genes in the livers of wildtype mice injected with shHNF4α, compared to mice injected with shLUCIFERASE (for shLUCIFERASE: n=4, for shHNF4α: n=5). B. Representative H&E staining of livers from wildtype mice injected with shLUCIFERASE or shHNF4α show no gross abnormalities. C. Quantification of fat (left) and muscle (middle) masses of shLUCIFERASE and shHNF4α injected mice (for shLUCIFERASE: n=4, for shHNF4α: n=5). Right: correlation of fat change to overall weight change by Pearson’s test. D. RT-qPCR demonstrates decreased expression of muscle atrophy genes in shHNF4α injected mice, compared to control mice injected with shLUCIFERASE(for shLUCIFERASE: n=4, for shHNF4α: n=5). E. VetScan measurement of GGT, BUN, bile acids, bilirubin, and albumin in plasma of WT or vagotomy treated WT mice show no liver dysfunction following vagotomy (for WT: n=3, for WT+vagotomy: n=6). F. LCN2 is elevated in the brains of mice with LLC compared to WT. (N=10, 5 per group). G. Amino acids analysis of plasma during vagal intervention, with and without tumors. Mass spectrometry analysis of amino acid relative levels in the plasma following vagotomy or sham procedure of mice injected with LLC (LLC N=5 or LLC vagotomy N=5 groups) or with PBS (Sham N=5 or vagotomy N=4 groups). The heatmap represents average relative values (uM) per metabolite per treatment. (For healthy: n=5, for vagotomy: n=4, for LLC: n=5, for LLC+vagotomy: n=5). H. Representative western blot showing recovery of HNF4α and OTC in vagotomized LLC mice. I. Correlation between HNF4α and survival by Pearson’s test. J. Correlation between albumin and survival by Person’s test. For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by T-test unless otherwise noted.

figs6

Figure S6: Intermittent Low Frequency Alternating Current Block (LFACb) is sufficient to attenuate cachectic phenotypes, related to Figures 5–6. A. Left: photograph of the custom PCB headstage used to deliver current controlled electrical pulses to the vagus nerve. Scale bar 1 cm. Right: Schematic of the electrical neuromodulation program delivery system. The program, established from the PC software control program, is transmitted wirelessly over radio frequency (RF) to the battery powered headstage which translates the computer-generated signal via a digital to analog converter (DAC) and drives current through the implanted microwire hook electrode. B. Intraoperative photograph of placement of a microwire hook electrode on the vagus nerve. C. Schematic showing the composition of the electrical waveform delivered by the device, with time on the x axis and amplitude on the y axis. The train was delivered for a duration of 30 minutes at the start of the wake cycle, with frequency of pulses at 5 Hz. Each pulse had a duration of 100 ms, with amplitudes of 50–350 μA. D-E. Effects of a single 30-minute dose of LFACb therapy 0, 12, 24, and 48 hours later on vagal peak-to-peak amplitude in D. LLC mice and E. KPC mice. N=120 (n=4 per condition per timepoint). F. Left: Representative gastrocnemius muscle H&E at 10x magnification, Right: quantification of gastrocnemius muscle weight at time of death (N=24, 8 per condition). G. Unrestricted food intake from a shallow trough on the cage floor by WT, KPC, or LLC animals receiving LFACb or sham treatment. N=40 (8 per group). H. Survival benefits from LFACb are not sex-specific. I. Weight retention benefits from LFACb are not sex-specific. J. Left: Correlation between weight change and survival in LFACb-treated animals by Pearson’s test. Middle: Correlation between muscle change and survival in LFACb-treated animals by Pearson’s test. Right: Correlation between fat change and survival in LFACb-treated animals by Pearson’s test. For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by T-test unless otherwise noted.

figs7

Figure S7: Transcutaneous Low Frequency Alternating Current Block (tLFACb) attenuates cachectic phenotypes and safely produces vagal block in swine, related to Figure 7. A. Left: Representative H&E staining of gastrocnemius muscle in tLFACb animals, 10x magnification. Right: Quantification of gastrocnemius muscle mass at time of death (N=24, 8 per condition). B. Left: photographs of flexible electrode patches placed on swine necks over the vagus nerve. Right: Representative recording of the right vagus nerve at baseline, during upstream stimulation alone to mimic cancer-induced vagal hyperactivity, and during tLFACb and upstream stimulation to demonstrate proof of concept vagal blocking. C. No statistically significant changes in heart rate, respiration, or body temperature were observed after 30 minutes of vagal block compared to baseline, though statistically significant reduction in vagal signaling was observed. D. Representative EKG taken at the same time as vagal neuromodulation in panel W. E. Histology of skin from the right side of the neck under the tLFACb patch and skin from the neck of a control pig and neck muscle, heart, lung, kidney, and liver tissue taken from tLFACb treated pigs and control pig showing no gross differences. For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Unless otherwise noted, statistical analysis by ANOVA with Tukey Post-Hoc.

figs4

Figure S4: Vagotomy alleviates elevation of CCL2 and acetylcholine, resulting in higher HNF4α levels, related to Figure 3. A. CCL2 levels in vagotomized KPC and LLC mice serum, CSF, brain, and liver on days 3, 7, and 14 post-cancer induction compared to non-vagotomized controls previously presented in Figure S1. N=5 per treatment per timepoint. B-D. Acetylcholine levels in vagotomized KPC, and LLC mouse serum (B) and liver (C) on days 3, 7, and 14 post-cancer induction and tumors (D) on day 7 and 14 compared to non-vagotomized controls previously presented in Figure S1; tumors were too small to reliably differentiate from healthy tissue on day 3. N=5 per treatment per timepoint. E. Terminal stomach, pancreas, and colon acetylcholine levels in KPC and LLC animals with or without vagotomy and WT controls. N=4 per group. F. Serum and liver acetylcholine levels inversely correlate with survival in both KPC and LLC mice by Pearson’s test. G. Top: FACS of vagal dorsal motor nucleus (VDMN) for immune population markers in vagotomized WT, LLC, and LLC+Vagotomy mice 7- and 14-days after cancer induction. Some panels previously discussed in Figure 1. Bottom: Quantification of total CD45+ F4/80+ cells, CD4+ cells, and CD8+ cells, and relative abundance of CD86+ vs. CD206+ cells in FACS. N=18 (3 per condition per time point). H. Quantification by RT-qPCR of HNF4α levels at 3-, 7-, and 14- days after cancer induction. For all box plots, dots represent data collected from unique mice. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Unless otherwise noted, statistical analysis by ANOVA with Tukey Post-Hoc.

figs2

Figure S2: Right vagotomy of the cervical vagus provides unique protective benefits against weight loss, related to Figures 2–3. A. Photos of hunched mice with poor body condition, typical of CAC. B. Weight retention as a percent of baseline body mass is not differentiated by sex in KPC or LLC mice. N=48, 8 animals per group. WT controls are the same for both models throughout the figure. C. Longitudinal measurements of estimated LLC tumor size based off caliper measurements and of KPC tumor size in euthanized animals. N=8 per group, no significant differences by RMANOVA. D. Terminal fat mass changes in KPC and LLC animals with and without vagotomy. N=48, 8 animals per group. E. Terminal lean mass changes in KPC and LLC animals with and without vagotomy. N=48, 8 animals per group. F. Top: Representative H&E of gastrocnemius muscles, 10X magnification, Center: Gastrocnemius muscle weight at time of death (N=40, 8 per group), Bottom: Histograms of gastrocnemius muscle fiber sizes quantified from WGA staining (composite of N=40, 8 per group). G. RT-qPCR of muscle atrophy genes (n=40, 8 per group). H. Left: photo of mouse undergoing EMG testing. Right: Evoked EMG amplitude (n=40, 8 per group). I: Falls during 2-minute hanging wire assay. N=40, 8 per group. J. No statistically significant changes are seen in anesthetized heart rate of mice one week after vagotomy. (N=30, 5 per group) K. No statistically significant changes are seen in liver blood flow as assessed by bleed rate. (N=30, 5 per group). L. Correlation between average daily food consumption and survival. Statistical analysis by Pearson’s test. N=32, 16 per cancer type/8 per treatment. M. Correlation between terminal weight change and survival. Statistical analysis by Pearson’s test. N=32, 16 per cancer type/8 per treatment. N. Correlation between terminal fat change and survival. Statistical analysis by Pearson’s test. N=32, 16 per cancer type/8 per treatment. O. Correlation between terminal muscle change and survival. Statistical analysis by Pearson’s test. N=32, 16 per cancer type/8 per treatment. P. No sex-based differences are observed in survival benefit. Statistical analysis by Pearson’s test. N=32, 4 per sex per treatment/cancer type. Q. Volcano plot of cytokines elevated in sham KPC mice over vagotomized KPC mice. For all box plots, dots represent data collected from unique mice. R. Unrestricted daily food consumption when fed from a shallow trough on the cage floor. N=40, 8 per group. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Unless otherwise noted, statistical analysis by ANOVA with Tukey Post-Hoc.

figs1

Figure S1: Immune infiltration and neuroinflammation precede cachectic onset, related to Figure 1. A. CCL2 levels in WT, KPC, and LLC mice serum, CSF, brain, and liver on days 3, 7, and 14 post-cancer induction. N=5 per treatment per timepoint. B. Acetylcholine levels in WT, KPC, and LLC mouse serum and liver on days 3, 7, and 14 post-cancer induction and tumors on day 7 and 14; tumors were too small to reliably differentiate from healthy tissue on day 3. N=5 per treatment per timepoint. C. Vagal tone RMS on days 3, 7, and 14 in WT, KPC, and LLC mice. N=5 per treatment per timepoint. D. 4x Photomicrograph of mouse hindbrains. Vagal dorsal motor nucleus (VDMN) in dashed white circles, inflammatory granules indicated with white arrows. E. Representative staining for CD45+ cells in the dorsal motor nucleus of the hindbrain. Nuclear staining (DAPI) is in blue, vagal lineage staining (Phox2b) is in green, and inflammatory cells (CD45) are in red. Scale bars 100 μm. F. Staining of the vagal dorsal motor nucleus for CD45 (red), a marker of leukocytes and microglia, CD80 (green), which is found in microglia and macrophages but not neutrophils, and CD11b (violet), which is found in macrophages but not microglia. Scale bars 100 μm. G. Flow cytometry on WT animals (top) and animals 7 days after LLC injection (bottom). Left to right, gating is for live cells (zombie violet), CD4 and CD8, CD45 and F4/80, and CD206 and CD86. H. CD45+ cell infiltration into the VDMN of WT or CCR2 KO mice injected with either PBS or KPC cells. I. IBA1 staining for neuroinflammation in the VDMN of WT or CCR2 KO mice injected with either PBS or KPC cells. J. CD45+ cell density in the VDMN of LLC and KPC mice is decreased by Bindarit. WT controls are the same for both cancer models throughout these studies. N=40 (8 per group). K. IBA1 expression in the VDMN of LLC and KPC mice is decreased by Bindarit. N=40 (8 per group). L. Serum acetylcholine is reduced by Bindarit in LLC and KPC mice (N=40, 8 per group). M. Vagal tone as evaluated by root mean square and peak-to-peak amplitude is normalized during Bindarit treatment of LLC and KPC mice (N=40, 8 per group). N. Bindarit attenuates anorexic eating in LLC and KPC mice (N=40, 8 per group). O. Gastrocnemius muscle fiber atrophy is attenuated by Bindarit (N=40, 8 mice per group). P. Gastrocnemius muscle weight loss is attenuated by Bindarit in LLC but not KPC animals (N=40, 8 mice per group). Q. Representative H&E of right gastrocnemius muscle at 10x magnification. R. Volcano plot of inflammatory factors differentially expressed in plasma of human pancreatic ductal adenocarcinoma patients with cachexia versus those without. * represents p<0.05, ** represents p<0.01, *** represents p<0.001, **** represents p<0.0001. Statistical analysis by ANOVA with Tukey HSD unless noted otherwise.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse Anti-Iba1	Santa Cruz Biotechnology	Sc-32725
Rabbit Anti-Phox2b	Cell Signaling Technology	83811S
Rat Anti-CD45	R&D Systems	MAB114-SP
Rat Anti-CD80	ThermoFisher Scientific	12-0801-82
Rat Anti-CD11b	BioLegend	101241
Sheep Anti-cFos	Sigma Aldrich	CBL440
Donkey anti-rabbit IgG	Abcam	Ab98799
Donkey anti-mouse IgG	Abcam	Ab97085
Rabbit anti-OTC	Abcam	Ab203859
Mouse anti-β-actin	Sigma-Aldrich	A5441
Rabbit anti-HNF4α	Abcam	ab181604
Rabbit anti-vinculin	Cell signaling	13901S
Bacterial and virus strains
AAV8-shLUCIFERASE	Weizmann institute of science	N/A
AAV8-shHNF4α	Weizmann institute of science	N/A
Chemicals, peptides, and recombinant proteins
DMEM	Life Technologies	11995-065
FBS	Life Technologies	16000-044
BEGM BulletKit	Lonza	CC-3171 & CC-4175
Acetylcholine Chloride	Sigma Aldrich	A2661
Matrigel	Corning	356234
Cisplatin	Sigma Aldrich	232120-50mg
Methoxyamine hydrochloride	ThermoFisher	A19188.06
N,O-bis (trimethylsilyl) trifluoroacetamide	Sigma Aldrich	15222
Wheat Germ Agglutinin, AlexaFluor™ 488 Conjugate	ThermoFisher	W11261
RIPA lysis buffer	Sigma Aldrich	R0278
Protease Inhibitor Cocktail	Sigma Aldrich	P8340
Phosphatase Inhibitor Cocktail	Sigma Aldrich	P5726
ECL	ThermoFisher	32106
Anamorelin	Santa Cruz Biotechnology	SC-480157
CCL2	PeproTech	250-10
Bindarit	TargetMol	T6413
Hematoxylin Stain	Leica Biosystems	3801560
Eosin Y	Sigma-Aldrich	HT110216
Hexamethonium	Sigma-Aldrich	H0879
Atropine	Sigma-Aldrich	PHL84121
Critical commercial assays
VETSCAN Liver Profile Cassette	Abaxis	500-0040
Direct-zol RNA MiniPrep kit	Zymo	ZR-R2050
Rneasy Mini Kit	Qiagen	74104
qScript cDNA Synthesis Kit	Quanta	95749
SYBR Green FastMix perfect CT	Quantabio	95073
BCA Protein Assay Kit	ThermoFisher	23225
Human CCL2 ELISA	PeproTech	BGK13500
ProcartaPlexTM Human Immune Response Panel, 80plex	Thermofisher Scientific	EPX800-10080-901
Mouse CCL2 ELISA	R&D Systems	MJE00B
Acetylcholine ELISA	BIOMATIK	EKE62518
Experimental models: Cell lines
Mouse: KPC4508P pancreatic cancer	Cancer Tools	153474
Mouse: LL/2 (LLC1) lung cancer	ATCC	CRL-1642
Human: THLE-2 hepatocyte	ATCC	CRL-2706
Experimental models: Organisms/strains
Mouse: C57Bl/6J	Jackson Laboratory; Envigo	000664; 00800124-2
Oligonucleotides
See Table S2
Software and algorithms
StimWare	Triangle BioSystems International	N/A
LabView SignalExpress	National Instruments	N/A
MATLAB	MathWorks	N/A
MassHunter	Agilent Technologies	N/A
FIJI	Schindelin, et al.118	https://imagej.net/software/fiji/
Muscle Morphometry Plug-In	Anthony Sinadinos using Eclipse IDE	https://drive.google.com/drive/folders/0B_bBI7SbDQhCR1MxNEVXSlhiekE?resourcekey=0-8wdIKyTc0OKlB7WN67JqIw
ImageLab	Bio-Rad	Version 6.1
JMP Pro 16	JMP Statistical Discovery	N/A
Prism	GraphPad	Version 9.4.1
Other
Stimulation Headstage	Duke University	N/A
AWG40 Hookup Wire	Cooner Wire Company	AS 631
Loctite 405 adhesive	Ellsworth	135436
Restraint Cone	Fisher Scientific	14-370-110
AC Amplifier	A-M Systems	Model 1800
BNC Adapter	National Instruments	BNC-2110
Lithium-Heparin Tubes	VetMarket	5598006
VETSCAN VS2	Abaxis	VETSCAN VS2
3.2mm beads	Next Advance	ADV-SSB32
CryoMill	Retscht	N/A
Nitrocellulose membrane	TAMAR	10401383
Graphene Ink	Sigma-Aldrich	793663
Silicone Adhesive	DIGI-KEY	1067-3M2477P-4-10-ND

Highlights.

• Tumor-induced inflammation drives vagal dysfunction, leading to cachexia phenotypes

• Vagal dysfunction depletes HNF4α and disrupts liver metabolic functions

• Right vagotomy alleviates cachexia, decoupling cachexia severity from tumor load

• Noninvasive vagal block restores feeding, reduces cachexia, and extends survival in mice