Neither glucagon-like peptide 1 receptors nor GDNF family receptor α-like neurons are required for aversive or anorectic response to deoxynivalenol (vomitoxin)

Abstract

Deoxynivalenol (DON), a type B trichothecene mycotoxin contaminating grain, promotes nausea, emesis, and anorexia. With DON exposure, circulating levels of intestinally derived satiation hormones, including glucagon-like peptide 1 (GLP-1), are elevated. To directly test whether GLP-1 signaling mediates the effects of DON, we examined the response of GLP-1 or GLP-1R-deficient mice to DON injection. We found comparable anorectic and conditioned taste avoidance learning responses in GLP-1/GLP-1R-deficient mice compared with control littermates, suggesting that GLP-1 is not necessary for the effects of DON on food intake and visceral illness. We then used our previously published data from translating ribosome affinity purification with RNA sequencing (TRAP-seq) analysis of area postrema neurons that express the receptor for the circulating cytokine growth differentiation factor (GDF15), GDNF family receptor α-like (GFRAL). Interestingly, this analysis showed that a cell surface receptor for DON, a calcium-sensing receptor (CaSR), is heavily enriched in GFRAL neurons. Given that GDF15 potently reduces food intake and can cause visceral illness by signaling through GFRAL neurons, we hypothesized that DON may also signal by activating CaSR on GFRAL neurons. Indeed, circulating GDF15 levels are elevated after DON administration, but both GFRAL knockout and GFRAL neuron-ablated mice exhibited similar anorectic and conditioned taste avoidance responses compared with WT littermates. Thus, GLP-1 signaling and GFRAL signaling and neurons are not required for DON-induced visceral illness or anorexia.

INTRODUCTION

Growth differentiation factor 15 (GDF15) is a peripherally derived cytokine that signals onto its receptor on neurons in the brain, GDNF family receptor α-like (GFRAL). GFRAL has a very narrow distribution in the central nervous system (CNS) that is limited to neurons within the area postrema (AP) and to a lesser extent the nucleus of the solitary tract (NTS) (1–4). GDF15 and analogs of GDF15 potently suppress food intake and cause weight loss via action on GFRAL neurons (2, 5, 6). Despite these profound effects on food intake and weight, there is little evidence that GDF15 or GFRAL regulates body weight under normal circumstances (2–4, 7). Rather, it would appear more likely that this system contributes to anorectic responses to a variety of pathophysiological states such as cancer, cachexia, sepsis, and cardiovascular disease in which circulating levels of GDF15 are elevated (8–10).

Given this small population of GFRAL neurons in the AP/NTS, we hypothesized that these neurons may receive other signals that indicate pathophysiological states. To identify other receptors on these neurons, we used data from translating ribosome affinity purification with RNA-seq (TRAP-seq) (11) to identify mRNA enriched in GFRAL neurons (12). We found that one of the highest enriched mRNAs on GFRAL neurons was the calcium-sensing receptor (CaSR). CaSR is a G protein-coupled-receptor that responds to extracellular calcium. In addition to calcium, CaSR is activated by deoxynivalenol (DON), a type B trichothecene mycotoxin produced primarily from Fusarium fungi. DON is a contaminant of grains and cereals including corn, wheat, oats, and barley and is frequently ingested by animals and humans. The visceral illness effects of DON earned it the colloquial name of “vomitoxin” when it was discovered in the 1970s (13). As summarized in a 1973 study, and later confirmed by dozens of additional studies, oral intake of DON-contaminated grains by swine decreased body weight gain, induced food refusal and emesis (14–16). Short-term exposure can cause gastroenteritis symptoms of acute nausea, vomiting, diarrhea, and pain (17) and long-term exposure can induce a reduction of body weight gain (18–21). These adverse symptoms have since additionally been described in humans, rodents, and mink (17, 22).

The emetic response is a protective mechanism against potentially hazardous ingested compounds. Swine, the most sensitive species to DON (22), and mink exhibit emesis as quickly as 30 min and 15 min after intraperitoneal or oral DON administration, respectively (15, 23). Because rodents lack the vomiting reflex, assessing nausea is typically performed using a conditioned taste avoidance (CTA) assay. A CTA is when a novel tastant is paired with the administration of a toxic compound that subsequently elicits malaise and nausea leading the animal to avoid that tastant when tested at a later time point. Consistent with the documented nausea associated with DON in other species, DON induces a potent CTA in rodents (24, 25).

Like many emetic agents, the administration of DON also activates neurons in the AP (26). This could be the result of direct neuronal activation in the AP since DON appears in the brain within 5 min of administration (27, 28). Given that CaSR is highly expressed in GFRAL neurons, it is likely that these DON-responsive neurons also express GFRAL. Activation of GFRAL neurons (either via GDF15 administration or activation of DREADD receptor) results in a profound CTA and other measures of malaise (6, 29–31). Consequently, we hypothesized that the potent and rapid effects of DON are the result of direct activation of CaSR/GFRAL neurons in the AP. To test this hypothesis, we used a variety of novel mouse models to test the role of GDF15 and GFRAL neurons to mediate effects of DON.

In addition to CaSR and GFRAL, receptors for glucagon-like peptide-1 (GLP-1) are also found in the AP. GLP-1 agonists also cause potent reductions in food intake and induce a CTA that are the result of GLP-1R activation (32). DON increases circulating GLP-1 levels presumably by increasing the secretion of GLP-1 from the intestine (33–35). Thus, it is possible that rather than acting directly on AP neurons, DON activates circuits in the AP via increasing levels of GLP-1 that in turn activates emetic circuits in the AP. To test this hypothesis, we used mice that lacked the ability to produce GLP-1 or to respond to GLP-1 via GLP-1R.

MATERIALS AND METHODS

Animals

All rodent experiments were approved by the University of Michigan Institutional Animal Care and Use Committee at (Animal Use Protocol no. PRO00007908). Animals were single housed (at minimum for 1 wk before experimentation) in temperature-controlled rooms containing a 12 h:12 h light:dark cycle and given ad libitum access to chow (PicoLab 5L0D) or 60% high-fat diet (Research Diets, Inc., D12492).

GLP-1R^KO/GLP-1R^WT, Gcg^KO/Gcg^WT, and GFRAL^KO/GFRAL^WT mice were generated as previously described (36, 37). A loxP-flanked stop cassette was inserted between exons 2 and 3 of the Gcg gene (Gcg^KO mice) and exons 6 and 7 of the GLP-1R gene (GLP-1R^KO) to create null alleles equivalent to whole body knockouts. To generate the GFRAL^eGFP mice, the LSL cassette from the ROSA26^eGFP-L10a allele was excised by Gfral^Cre as previously described (6, 38).

Male and female GFRAL^eGFP-L10a mice were used for cfos immunohistochemical analysis of the area postrema. A separate cohort of male and female Gfral^{cre-eGFP-L10a} were used for TRAP. Gfral^cre-eGFP-L10 mice were bred with mice expressing B6-iDTR, ROSA26iDTR (The Jackson Laboratory) to generate Gfral^eGFP-DTR (Gfral^DTR) mice, which express diphtheria toxin receptor in GFRAL cells. WT littermates were used as control (Gfral^WT).

Pharmacological Agents

Lyophilized deoxynivalenol (Sigma 5 mg D0156) was reconstituted into 1 mg/mL saline aliquots and frozen at −80°C. On the day of the experiment, deoxynivalenol was thawed, diluted with saline to the appropriate concentration (0.25 mg/mL), and filtered through a 0.22-μm PVDF membrane filter (Millipore Sigma, SLGV033RS) for intraperitoneal administration. 0.4 mg/kg of mouse GDF15 (Novo Nordisk) was prepared by diluting with a GDF-15 buffer (5 mM acetate salt, 2.25% glycerol and 70 ppm Tween 20 at pH 4). Mouse GDF15 or buffer was injected subcutaneously in mice daily. All GFRAL^DTR and GFRAL^WT mice were dosed with 200 μg/kg diphtheria toxin dissolved in saline (Sigma, D0564-1MG) before experimentation.

Food Intake Test

Clean cages were provided, and all food was removed to fast mice for 4–5 h (GFRAL^KO/WT; GFRAL^DTR/WT; GLP-1R ^KO/WT; Gcg^KO/WT) with full access to water. Animals were randomized for treatment and administered a single injection of saline or DON (2.5 mg/kg ip) 1 h before dark onset and return of preweighed diet. The remaining food in the cage was measured at 1, 2, 4, and 24 h after food return to determine food intake. The same was repeated for food intake analysis of pharmacological GDF15 except GFRAL^DTR/GFRAL^WT mice were not fasted and measurements were taken at the same time every 24 h.

Conditioned Taste Avoidance Assay

Mice were handled and injected with saline for a minimum of 1 wk before experimentation. On day 0, mice began habituation to restricted drinking water and replacement with two water bottles. On day 3, water bottles were removed 1 h before lights off to induce thirst for 23 h. On day 4 (conditioning day), 2 bottles containing 0.15% saccharine in water solution were given to mice for 2 h before lights out. At lights out, mice received an injection of either saline or DON (2.5 mg/kg ip). Saccharine bottles were then removed, and 2 water bottles were added back to the cages immediately. On day 5, water bottles were removed for 23 h to induce thirst. On day 6 (test day), a preweighed water bottle and saccharine bottle were added to the cages (0 h; baseline) to give the mice a two-bottle choice 1 h before lights out. Bottles were measured for the following 1, 2, 4, and 24 h. Volume of water and saccharine consumed were calculated at each time point and represented as a percentage of total volume intake [saccharine intake/(saccharine + water intake)] × 100.

Body Composition

Fat and lean mass were measured using an EchoMRI (Echo Medical Systems).

Immunohistochemistry

GFRAL^eGFP-L10 mice were injected with 2.5 mg/kg DON ip 2 h before CO₂ asphyxiation. GFRAL^DTR and GFRAL^WT mice were injected with 0.04 mg/kg GDF15 sc 4 h before euthanization with isoflurane. Mice were then perfused with phosphate-buffered saline (PBS) followed by a 10% buffered formalin solution. Brains were harvested and post fixed in a 30% sucrose solution, and 30-μm brain slices were taken using a freezing microtome (Leica). Brain sections were then treated with 1% hydrogen peroxide-0.5% sodium hydroxide, 0.3% glycine, 0.03% sodium dodecyl sulfate, and blocked with 0.1% triton, 3% normal donkey serum in PBS (Fisher Scientific). After an overnight room temperature incubation with anti-FOS antibody (Cell Signaling; AB_2247211; 1:1,000 dilution) or anti-GFP (Aves laboratories; 1:000 dilution), slices were washed and incubated with secondary antibody (1:300 dilution). Microscopy and analysis of images were done (Olympus BX51 microscope).

ELISA

To measure plasma GDF15 levels, a sandwich ELISA (R&D Systems, Inc.; MGD150; Minneapolis, MN;LOQ 7.8 pg/ml; LOD 0.686 pg/mL) was used according to manufacturer’s directions and a 1:1 dilution.

Translating Ribosome Affinity Purification-Seq Analysis

Translating ribosome affinity purification (TRAP-seq) data were retrieved from the NCBI Gene Expression Omnibus (GSE160257). Analysis was conducted in R (version 4.0.3). Enrichment was determined using DESeq2 (version 1.30.1) using a model that accounts for sample pairing (∼ Pair + Fraction).

Statistical Analysis

All statistical analysis was performed in either GraphPad Prism v8.4.2, v9.1.0 or v9.2.0 and is represented as means ± SE. Student’s t test, one-way, two-way, or three-way ANOVA with Tukey’s post hoc tests were utilized to determine significance. Significance was defined as P < 0.05.

RESULTS

Role of GLP-1 and GLP-1R Signaling in the Effects of DON

As GLP-1 increases in circulation with DON injection, we hypothesized that GLP-1 signaling contributes to food refusal and aversive effects. To test the necessity of GLP-1, we used mice lacking either GLP-1R or the preproglucagon gene (Gcg), which encodes GLP-1 and other important peptide products including glucagon. We chose to use a dose (2.5 mg/kg) of DON that has previously been shown to increase GLP-1 into circulation by twofold within 30 min in mice (33). Body weights of GLP-1R^KO and GLP-1R^WT mice were not different (Fig. 1A). To assess the necessity of GLP-1R in inducing the nausea effects of DON, we performed a two-bottle choice CTA assay using 0.15% saccharine solution as the paired tastant. DON induced a strong CTA in both GLP-1R^KO and GLP-1R^WT littermate mice and reduced saccharine intake from 65–70% to merely 10–17% of total volume with no statistical impact of genotype (Fig. 1B). In a food intake test, DON induced potent anorexia as early as 1–2 h after food presentation in both GLP-1R^WT and GLP-1R^KO male (Fig. 1C) and female (Fig. 1D) mice. At 4 h, DON reduced food intake in GLP-1R^KO and GLP-1R^WT mice by 33–35% in males and 29–30% in females. By 24 h, there is no statistical difference between DON and saline-treated groups in males. Female GLP-1R^KO mice continue to show a small but statistically significant decrease in feeding. However, we predict that this response would be eliminated within several hours. Although the data shown is separated by the sex of mice, when male and female data are combined, a similar trend remains, with DON inducing anorexia at 2 and 4 h and no differences in the responses between genotypes.

Figure 1.

Signaling. A: body weights of male wild-type (WT, n = 19), male GLP-1R KO (n = 21), female WT (n = 11), and female GLP-1R KO (n = 21) mice before experimentation two-way ANOVA with Tukey’s multiple comparisons (****P < 0.0001: male vs. female). B: conditioned taste avoidance (CTA) assay in male WT and GLP-1R KO mice conditioned to saline or 2.5 mg/kg DON (n = 8–10). Food intake (FI) test in WT and GLP-1R KO male (n = 8–10; C) and female (n = 9–10; D) mice fasted for 4–5 h and injected with 2.5 mg/kg DON compared with WT saline (**P < 0.01, ***P < 0.001, ****P < 0.0001) and compared with KO saline (#P < 0.05, ##P < 0.01, ###P < 0.001, ####P < 0.0001) by two-way ANOVA with Tukey’s multiple comparison test. E: body weights of WT and Gcg KO male and female mice (n = 19–20) two-tailed Student’s t test; no significant differences. F: CTA assay in male and female WT and Gcg KO mice conditioned to saline or 2.5 mg/kg DON compared with WT saline (**P < 0.01, ***P < 0.001, ****P < 0.0001) and compared with KO saline (#P < 0.05, ##P < 0.01, ###P < 0.001, ####P < 0.0001) (n = 4–14). G: food intake test in WT and Gcg KO male and female mice (n = 4–14) compared with WT saline (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001) and compared with KO saline (#P < 0.05, ##P < 0.01, ###P < 0.001, ####P < 0.0001) by two-way ANOVA with Tukey’s multiple comparison test. All data are represented as means ± SE. DON, deoxynivalenol; Gcg, preproglucagon gene; KO, knockout.

To complement the use of GLP-1R^KO mice, we next tested the necessity of GLP-1 using mice lacking GLP-1 as well as other preproglucagon-derived peptides (Gcg^KO mice). Gcg^KO mice did not exhibit differences in body weight compared with Gcg^WT littermates (Fig. 1E). DON induced a CTA in both Gcg^KO and Gcg^WT littermates, illustrated by a reduction of relative saccharine intake beginning at 1 h and lasting for 25 h (Fig. 1F). There was no significant difference between Gcg^WT and Gcg^KO CTA responses. Furthermore, Gcg^WT and Gcg^KO mice reduce feeding by 35% and 32%, respectively within 1 h of DON exposure (Fig. 1G). Reduction of food intake lasts up to 24 h after DON injection. Although data is represented with male and female mice combined, similar to GLP-1R^KO mice, there is no difference between male and female food intake responses when data are separated by sex. These data demonstrate that neither GLP-1 (made by the gut or by the CNS) nor GLP-1R are necessary for the anorexia and avoidance effects of DON.

Role of GDF15 and GFRAL Signaling in the Effects of DON

We then assessed the effect of DON on the GDF15 and GFRAL signaling system. Wild-type mice injected with 2.5 mg/kg DON exhibited a prominent increase in circulating GDF15 from 53 pg/mL to 4,650 pg/mL within 2 h (Fig. 2A). We next investigated central neuronal activation within the AP using DON-injected GFRAL^eGFP-L10 mice. Two hours after DON injection, we found a high colocalization of eGFP+ GFRAL cells and cfos immunofluorescence within the AP (Fig. 2B). Since DON clearly induces activation of GFRAL neurons, we then asked whether whole body GFRAL signaling was necessary for DON-induced CTA and anorexia. To test this, we used female GFRAL^KO and GFRAL^WT littermate mice. DON-treated GFRAL^KO and GFRAL^WT mice both reduced saccharine intake and increased water intake within 2 h and lasted for 24 h following a two-bottle choice presentation test (Fig. 2, C–E). GFRAL^KO saline-treated mice consumed 66–80% saccharine solution compared with only 10–13% of the total intake in DON-treated mice at 2 (Fig. 2C) and 4 h (Fig. 2E). Although, DON induced an equally potent CTA response in both GFRAL^KO and GFRAL^WT mice. We then did a food intake test, in a separate cohort of male and female GFRAL^KO and GFRAL^WT mice. Within 1 h, GFRAL^KO mice and WT littermates exhibited a strong reduction in food intake with DON compared with GFRAL^KO saline controls (Fig. 2F). DON-induced anorexia was detectable at 2 and 4 h (Fig. 2, F–H), and reduced by 24 h (Fig. 2I) with no difference between the GFRAL^KO and GFRAL^WT DON responses at any timepoint. Taken together, these data support the hypothesis that GFRAL signaling is not necessary for the aversive or anorectic responses to DON.

Figure 2.

GFRAL signaling. A: ELISA of serum GDF15 from 2.5 mg/kg DON intraperitoneally injected wild-type mice (n = 4). ****P < 0.0001 compared with saline by unpaired t test. B: immunohistochemical staining of area postrema (100 μm scale bar) from male and female GFRAL eGFP-L10 mice injected with 2.5 mg/kg DON for cfos (n = 3). Conditioned taste avoidance (CTA) assay of female GFRAL KO and WT mice with intraperitoneal administration at 2 h (C), 4 h (D), and 24 h (E; n = 5–6). ****P < 0.0001 compared with WT saline; ####P < 0.0001 compared with GFRAL KO saline by two-way ANOVA with Tukey’s multiple comparison test. Food intake (FI) test of 5 h-fasted mice conditioned to saline or 2.5 mg/kg DON at 1 h (F), 2 h (G), 4 h (H), and 24 h (I; n = 5–10). ##P < 0.01, ###P < 0.001 compared with GFRAL KO saline by two-way ANOVA with Tukey’s multiple comparison test. I: transcripts enriched in TRAP fraction of the AP in male and female GFRALeGFP-L10 mice. Differential expression was determined by CuffDiff analysis with threshold set to P < 0.05. J: expression is represented as fragments per million (FPM; n = 12). All data are represented as means ± SE. AP, area postrema; DON, deoxynivalenol; DTR, diphtheria toxin receptor; GFRAL, GDNF family receptor α-like; WT, wild type; KO, knockout; TRAP, translating ribosome affinity purification.

Role of GFRAL-Expressing Neurons in the Effects of DON

Although GFRAL is not necessary, other cell surface receptors on AP GFRAL neurons may be responsible for the activation of the neuronal population. Thus, we next isolated the AP and used TRAP-seq data from GFRAL^eGFP-L10 mice (12). There was a high enrichment of CaSR mRNA in GFRAL cells (Fig. 2J). This led us to hypothesize that GFRAL neuronal activation is a primary site of action for the anorectic and aversive effects of DON. To functionally test the necessity of GFRAL neurons (including other cell surface receptor expressed on GFRAL neurons) in mediating DON-induced physiological illness, we used mice with GFRAL neuron ablation. We crossed GFRAL^cre mice onto the cre-inducible diphtheria toxin receptor (DTR) background, permitting the specific ablation of GFRAL neurons by intraperitoneal diphtheria toxin (DT) administration in GFRAL^DTR mice while leaving GFRAL neurons intact in the GFRAL^WT mice. To understand the impact of DT on the phenotypes of mice with and without GFRAL neuron ablation, we measured body weight (Fig. 3A), fat, and lean mass (Fig. 3, B and C) before and after DT injection in a cohort of male and female mice. Weekly body weights, fat, and lean mass did not differ between GFRAL^WT and GFRAL^DTR mice before or after DT administration (Fig. 3, A–C).

Figure 3.

GFRAL neuronal ablation. Body weight 2 wk before and 7 wk after diphtheria toxin injection in GFRAL WT and GFRAL DTR mice (n = 18–20/group; A); two-way ANOVA with Sidak’s multiple comparison test. Time x Genotype P = 0.0093. Body composition of GFRAL WT and GFRAL DTR mice 2 wk before diphtheria toxin administration (n = 18–20/group; B) and 15 wk after diphtheria toxin administration (n = 18–20/group; C); two-way ANOVA with Tukey’s multiple comparison test. Conditioned taste avoidance (CTA) assay in male GFRAL WT and GFRAL DTR mice conditioned to saline or 2.5 mg/kg DON at 4 h (D) and 24 h (E) after the addition of two-bottle choices (n = 9–11/group); two-way ANOVA with Tukey’s multiple comparison test. ***P < 0.001, ****P < 0.0001 compared with GFRAL WT saline; ####P < 0.0001 compared with GFRAL DTR saline. Food intake test after a 4- to 5-h fast in GFRAL WT and GFRAL DTR mice with either saline or 2.5 mg/kg DON injection at 2 h (F), 4 h (G), and 24 h (H) after food return. ***P < 0.001, ****P < 0.0001 compared with GFRAL WT saline and ###P < 0.001, ####P < 0.0001 compared with GFRAL DTR saline; two-way ANOVA with Tukey’s multiple comparison test. All data are represented as means ± SE. D’Agostino-Pearson test for normality was performed on B–H. DON, deoxynivalenol; DTR, diphtheria toxin receptor; GFRAL, GDNF family receptor α-like; WT, wild type.

Seven weeks after DT administration, we conducted a two-bottle choice CTA test in which we paired the flavor of 0.15% saccharine solution with DON. GFRAL^WT and GFRAL^DTR saline-paired mice both drank 67% saccharine solution. GFRAL^WT and GFRAL^DTR mice paired with DON, consumed only 8% and 16% saccharine solution at 4 h, respectively, illustrating that DON induced a potent CTA in both genotypes (Fig. 3D). There was no significant difference between the amount of saccharine solution consumed by GFRAL^WT versus GFRAL^DTR mice at either 4 or 24 h (Fig. 3E), suggesting that GFRAL neurons are not required for the aversive effects of DON. There is greater variability in saccharine intake for GFRAL^DTR mice at 24 h, potentially reflecting differential ablation of GFRAL neurons in each subject. To test whether GFRAL neurons play a role in the strong anorexia induced by DON, we subjected mice to a food intake test after intraperitoneal injection of either saline or 2.5 mg/kg DON. Although DON induced anorexia within 2 h of administration in both GFRAL^WT and GFRAL^DTR mice, we did not observe a difference in food intake between the GFRAL^WT and GFRAL^DTR mice (Fig. 3F) at 4 (Fig. 3G) or 24 h (Fig. 3H). Thus, we conclude that GFRAL neurons are not necessary for DON-induced food refusal.

To functionally confirm the ablation of GFRAL neurons, we next challenged GFRAL^WT and GFRAL^DTR mice with high-fat diet (HFD) feeding and performed a food intake test following 0.4 mg/kg GDF15 administration. Pharmacological GDF15 is a potent anorectic agent, the effects of which are clearer in obese mice (4). Daily administration of GDF15 for 3 days did not change body weight (Fig. 4A) in mice with ablated GFRAL neurons. GFRAL ^WT mice exhibit a decrease in cumulative food intake in response to GDF15 by day 2 (Fig. 4B) and day 3 (Fig. 4C) compared with GFRAL^DTR mice. Next, we perfused and harvested the mouse brains 4 h after 0.4 mg/kg GDF15 administration. GFRAL^DTR mice contained reduced GFP reporter fluorescence in the AP and reduced cFOS in the AP and PBN compared with GFRAL^WT (nonlittermate GFRAL^eGFP-L10) mice (Fig. 4, D–F). In a separate cohort of GFRAL^WT and GFRAL^DTR mice, we next performed a CTA, pairing pharmacological GDF15 to a 0.15% saccharine solution. GDF15 treatment induced a potent CTA in GFRAL^WT but not in GFRAL^DTR mice (Fig. 4G). Taken together, the lack of anorexia and CTA induction in response to GDF15 along with a reduced GFP signal in GFRAL^DTR mice confirmed that GFRAL neurons are depleted by DT, but this did not alter the response to DON.

Figure 4.

Confirmation of GFRALDTR neuron ablation confirmation. Body weights of GFRAL WT and GFRAL DTR mice dosed with subcutaneous vehicle or 0.4 mg/kg GDF15 daily on days 1, 2, and 3 (A) and cumulative food intake at 48 h (B) and 72 h (C; n = 6–7/group); two-way ANOVA with Tukey’s post hoc test. *P < 0.05 compared with GFRAL^WT GDF15 treated. Representative immunostaining (100 μm scale bar) of the AP/NTS and PBN labeling GFRAL neurons (eGFP) and cFOS 4 h after GDF15 treatment in GFRAL DTR and GFRAL WT mice (D). Quantification of immunostained cells labeling GFP (E) and cFOS (F) in the AP and NTS from mice dosed with GDF15 (n = 5–6 mice/group); one-way ANOVA. **P < 0.01; ****P < 0.0001 compared with GFRAL WT. CTA of WT and GFRAL DTR mice conditioned to GDF15 at 4 h after presentation of two-bottle choices (G) two-way ANOVA. ****P < 0.0001 compared with GFRAL WT. All data are represented as means ± SE; AP, area postrema; CTA, conditioned taste avoidance; DTR, diphtheria toxin receptor; GFRAL,GDNF family receptor α-like; NTS, nucleus of the solitary tract; WT, wild type; PBN, parabrachial nucleus.

DISCUSSION

A wide range of hypotheses have been offered to explain the potent effects of DON to induce anorexia and nausea. We tested the necessity of peripheral signaling molecules using GLP-1R^KO, Gcg^KO, GFRAL^KO, and GFRAL^DTR ablation of GFRAL neurons. Enteroendocrine cells are important in detecting luminal content and influencing appetite and satiation through the release of hormones. CaSR is found in enteroendocrine cells and downstream signaling leads to the exocytosis of GLP-1 (34, 39, 40) as quickly as 15 min after DON exposure (23, 41–43). Our results demonstrate that DON is a potent visceral illness-inducing agent that does not require the production of GLP-1 or GLP-1R signaling to cause conditioned taste avoidance or anorexia. GLP-1R^KO and Gcg^KO mice exhibited a similar anorexia and avoidance response as the wild-type littermate controls when peripherally administered DON. These results indicate that GLP-1 and GLP-1R signaling is not required for these effects of DON. It remains possible that these do contribute to effects of DON but do so in the context of other anorectic signals.

Our results are in contrast to Jia et al. (33), in which GLP-1 antagonist Exendin_9-39 pharmacologically increased food intake in mice exposed to oral DON. This is likely due to differences in the genetic versus pharmacological approaches. Interpretation of these experiments is highly dependent on the specifics of the doses used. Simply adding an orexigenic dose of Exendin_9-39 to an anorectic dose of DON does not imply that the effects of DON require the GLP-1R system. Rather, the two agents may simply have additive effects (44). The advantage of the genetic approach taken here is that it asks the direct question about whether a system is required for the effects of DON with the caveat that developmental compensations to the loss of GLP-1R function may provide alternative pathways to respond to DON.

Although we found that GLP-1 is not a necessary component, other hormone signals may be contributing to the response. Many peripheral agents are capable of signaling to the hindbrain to elicit behavioral responses. The AP and NTS have been identified as important regions for DON-induced illness. An early-lesion study showed that the development of a CTA in rats in response to DON is dependent on the AP (24). A population of neurons in the AP/NTS containing GFRAL receptor elicits similar illness behaviors of anorexia and avoidance formation when activated (6, 29). We hypothesized that DON could increase GDF15 levels to activate GFRAL-expressing neurons to produce nausea and food refusal. Indeed, peripheral administration of DON results in increased circulating levels of GDF15. Despite a potent effect to increase levels of GDF15, GFRAL^KO mice exhibit the same anorexia and avoidance response as GFRAL^WT littermates. Thus, we conclude that GFRAL signaling is not required for the aversive and anorectic responses to DON.

After peripheral DON administration, levels of DON in the brain increase rapidly (28). Instead of acting peripherally, DON may activate its receptor, CaSR, directly in the brain. Using TRAP-seq, we found a high enrichment of CaSR on GFRAL neurons. We then hypothesized that although GFRAL signaling is not necessary, these neurons may play a critical role by direct activation of CaSR by DON. To test this, we developed GFRAL^DTR mice that specifically express diphtheria toxin receptor on GFRAL neurons. DT administration ablates GFRAL cre expressing neurons in GFRAL^cre mice (GFRAL^DTR) but not in mice lacking cre (GFRAL^WT). To control for DT, we administered DT to all mice. GFRAL^DTR mice exhibit similar anorexia and CTA formation as GFRAL^WT mice. GFRAL^DTR mice exhibited a greater variability in CTA formation, with some DON-conditioned mice drinking more saccharine than others compared with GFRAL^WT mice. It is possible that there is variable ablation of GFRAL-expressing neurons and that a relatively small number of these cells are all that is required to elicit avoidance.

Perspectives and Significance

It is clear that GFRAL neurons represent just a fraction of neuronal populations within the AP and NTS and only a fraction of the neurons that express CaSR (45–47). Indeed, in our staining performed in DON-treated GFRAL^eGFP-L10 animals, we observe a large number of FOS-expressing cells that do not express GFP. Furthermore, CaSR is expressed in the hypothalamus which could be hypothesized as a target for DON’s anorectic and weight loss effects. A handful of studies have detected alterations in the expression of cFos and mRNA within the arcuate nucleus in POMC and AgRP neurons in response to DON, though with conflicting results (26, 41, 42, 48, 49). Consequently, it remains unclear what the most important target is for the potent actions of DON. Further research using loss or gain-of-function models of CaSR is the most likely approach to yield a specific answer to how DON is sensed and how the physiological responses are produced.

DATA AVAILABILITY

Data will be made available on reasonable request.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants P01 DK117821 (to D. A. Sandoval, M. G. Myers, and R. J. Seeley), P30 DK089503 (to R. J. Seeley), and R01DK119188 (to R. J. Seeley).

DISCLOSURES

R.J.S. receives research support from Novo Nordisk, AstraZeneca, Fractyl, and Pfizer. R.J.S. serves as a paid consultant for Novo Nordisk, Scohia, Fractyl, Eli Lilly, and ShouTi and has equity positions in Rewind and Calibrate. A.R.P., currently employed by Eli Lilly and Company, is acting entirely on her own with this work which was completed before her employment at Lilly; these endeavors are not in any manner affiliated with Eli Lilly. M.G.M. receives research support from Novo Nordisk, AstraZeneca. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

A.R.P. and R.J.S. conceived and designed research; A.R.P. and P.V.S. performed experiments; A.R.P., P.V.S., and A.C.R. analyzed data; A.R.P., H.F.-S., P.V.S., A.C.R., D.A.S., M.G.M., and R.J.S. interpreted results of experiments; A.R.P. and A.C.R. prepared figures; A.R.P. drafted manuscript; H.F.-S., P.V.S., M.G.M., and R.J.S. edited and revised manuscript; A.R.P., H.F.-S., A.C.R., D.A.S., M.G.M., and R.J.S. approved final version of manuscript.