Fibroblast growth factor 19 (FGF19) is an ileal-derived protein hormone that produces potent, anti-diabetic, and anti-obesogenic effects. Although early work suggested that these effects were mediated mainly by FGF19 acting on peripheral targets, several reports have now found that acute intracerebroventricular (ICV) administration of FGF19 may act at multiple distinct sites in the brain to regulate energy balance. Lateral ventricle administration of FGF19 increases metabolic rate (1), while 3^rd ventricular administration was found to decrease food intake, lower insulin resistance, improve glucose tolerance, and reduce blood glucose concentrations in rodent models of diabetes and obesity (2–5). These findings suggest a hypothalamic mechanism since FGF19 was found to suppress AGRP/NPY activity (4) and decrease plasma ACTH (2). However, we demonstrated that 4^th ventricular administration of FGF19 also decreases blood glucose in type 1 diabetic (T1DM) mice, suggesting the hindbrain as an underappreciated target tissue for this system (6).

The brainstem dorsal vagal complex (DVC) is an important homeostatic regulatory center that tightly controls parasympathetic output in response to numerous convergent inputs, both neuronal and humoral. The DVC is principally comprised of the area postrema (AP), the nucleus tractus solitarius (NTS), and the dorsal motor nucleus of the vagus (DMV). The NTS integrates vagal afferent, viscerosensory information with input from several brain areas, including the hypothalamus and AP (7–10). In turn, the NTS regulates neural activity in the DMV through both glutamatergic and GABAergic projections (11, 12). Finally, DMV neurons project cholinergic outputs through the efferent vagus nerve to regulate hepatic glucose production, gastric motility, and pancreatic exocrine secretion, among other visceral regulatory functions (13–18). Additionally, the AP and NTS contain fenestrated capillaries that permit the passage of humoral components that may be excluded by the blood-brain barrier elsewhere (19, 20). DVC neurons respond to many primary metabolic hormones including insulin, leptin, ghrelin, glucagon, GLP-1 (21–25) and importantly, FGF19 (6). The DVC, especially the AP and NTS, contains multiple FGF receptors (FGFR) as well as ß-klotho, an obligate co-receptor (26–28), suggesting that neuronal activity in the DVC may regulate metabolic homeostatic mechanisms in response to endogenous FGF signaling.

In a mouse model of type 1 diabetes (T1DM), FGF19 consistently and robustly increased action potential-dependent excitatory synaptic transmission to the DMV (6). This suggests that FGF19 increases the activity of glutamatergic neurons immediately afferent to the DMV, which remain intact in the slice preparation. A likely source of these glutamatergic inputs is the NTS since the NTS to DMV connection is well documented (12, 29–31) and is open to modulation by various other peptides and neurotransmitters (31–34). Interestingly, receptor expression data suggest that the AP could also be a target for FGF19 since the AP expresses significantly more FGFR than surrounding areas (28) and expresses more ß-klotho and FGFR2 than any other brain area, in addition to high levels of FGFR1 and FGFR3 (26). Moreover, glutamatergic AP neurons project extensively to the NTS and DMV (35). Taken together, these findings suggest that FGF19 may act on intrinsic DVC circuitry to modify synaptic input to the DMV and consequently modulate vagally-mediated glucoregulation.

This study tests the hypothesis that FGF19 increases excitatory neurotransmission in the DVC by altering excitability of NTS and AP neurons. In addition to functioning within local DVC circuits, the NTS and AP communicate bidirectionally with other brain areas thought to be involved in regulating energy metabolism (7, 8, 36). Thus, FGF19-mediated alteration of neuronal activity in these areas is likely to have wide-reaching metabolic effects by modulating vago-vagal circuit dynamics as well as influencing more distal areas that control ingestive behaviors and regulate energy balance through autonomic or neuroendocrine means.

These results indicate that FGF19 increases glutamatergic transmission at multiple points in DVC circuitry in hyperglycemic, but not normoglycemic mice, which implies a mechanistic understanding of FGF19 activity in the hindbrain. We found previously that FGF19 decreased blood glucose in hyperglycemic mice through actions in the dorsal hindbrain, and this effect was abrogated by co-administration of a peripheral muscarinic receptor blocker (6), suggesting a parasympathetic mechanism. Similar to the current findings, we also found that FGF19 altered the excitability of DMV neurons in a complex manner that was heavily influenced by hyperglycemic state. These results suggested that FGF19 functioned in the DVC through a relatively restricted mechanism (i.e., that FGF19 modified the activity of central vago-vagal reflexes, resulting in altered parasympathetic output). The current findings expand upon what was established previously by identifying multiple local, excitatory synaptic circuits within the DVC that are positively modulated by FGF19 to increase glutamate release in the DMV. These results are especially intriguing considering that, while the DMV is typically associated with autonomic regulation, the NTS and AP show extensive interconnectivity with several other brain regions involved in regulating metabolism (8, 35, 36, 44–46). This suggests that FGF19 in the hindbrain may regulate metabolism through multiple mechanisms in addition to autonomic changes, which is consistent with the diverse effects produced by ICV FGF19 reported previously (2–4, 47).

Previous data showed that the effect of FGF19 on EPSC frequency in the DMV was blocked by TTX, suggesting a neuronal source that remained intact within slice preparation. Although the source was unknown, it was hypothesized to be the NTS, since it is a primary source of synaptic input to the DMV and expresses the receptors/co-receptor required to respond to FGF19 (26, 28, 48, 49). However, these same expression data suggest the intriguing hypothesis that the AP may also participate in the response of DMV neurons to FGF19. Indeed, the AP displays significantly higher expression of multiple FGFRs and ß-klotho than the NTS and shows some of the highest expression levels of any brain area (26). Thus, neurons in the NTS and AP were tested regarding the effects of FGF19 on their glutamatergic connections to the DMV.

The glutamate photolysis experiments performed here demonstrated that FGF19 increases the excitatory synaptic input to the DMV arising from both AP and NTS neurons. The involvement of the NTS was predicted, since the excitatory glutamatergic NTS to DMV circuit has been well-characterized (9, 11, 12, 50). However, there is little known about the role of the AP in local vagal circuitry. Interestingly, there is evidence to suggest that the AP may be an unappreciated participant in the canonical vago-vagal reflex circuit. Several early tracing studies (51–54), later confirmed with viral and genetic techniques (35, 55, 56), found that the AP receives vagal sensory input and projects glutamatergic output toward the NTS and DMV. This suggests that the AP may serve as a vagal sensory integration center, in a similar capacity as the NTS. Despite this, electrophysiological studies exploring the role that the AP plays in regulating local DVC circuitry have been few and more work is warranted to understand this connection (57–60).

The current findings confirm that AP neurons can modulate DMV activity via glutamatergic synaptic connections and that FGF19 increases the activity of this input. This does not necessarily imply the existence of a monosynaptic connection, since it is possible that the AP regulates DMV activity through intermediary NTS neurons. However, considering the widespread glutamatergic innervation from the AP throughout the DVC and the extensive dendritic fields of DMV neurons, it is likely that the AP communicates to the DMV both mono- and polysynaptically (35, 61). A priori, in light of the AP’s traditional role as the principal emetic center in the brain, the existence of a monosynaptic connection from AP to DMV seems logical. This would allow for rapid control of gastric motility and stomach muscle contraction in response to noxious stimuli in the blood.

Uncaging glutamate over a neuron causes a transient membrane depolarization that subsides over a period of ~500 ms. Consequently, an increased response to uncaging, as was seen here, is presumed to be the result an increase in excitability in the afferent neuron. Initial uncaging experiments indicated an increase in glutamatergic NTS neuron excitability in T1DM mice. To understand this, we first measured the intrinsic membrane properties of these neurons in both control and T1DM mice. The mouse model used here expresses EGFP in GABAergic neurons, so to increase the likelihood of obtaining responses from glutamatergic neurons, only EGFP-negative neurons were targeted for recording, though this does not imply that all neurons were glutamatergic. FGF19 produced a mixed effect on intrinsic membrane properties in these neurons, which did not appear to be dependent on hyperglycemic state. In both control and T1DM groups, FGF19 failed to alter R_in and RMP in approximately half of NTS neurons. In the neurons that did respond, FGF19 tended to produce relatively subtle inhibitory effects, although a small proportion of neurons were excited by FGF19. Similarly, FGF19 produced a predominately inhibitory effect on evoked action potential firing in response to positive current injections. These responses were mainly similar to our previous results in the DMV, which noted a modest inhibitory effect of FGF19 on intrinsic properties of DMV neurons in both control and T1DM groups, suggesting that the peptide may interact with a common intracellular pathway downstream from the FGFR that is not modified by hyperglycemia (6).

The effects of FGF19 on NTS intrinsic excitability run contrary to the excitation that was predicted by glutamate uncaging. Although it is possible that the small proportion of NTS neurons that were excited by FGF19 was responsible for the increase in evoked EPSCs in T1DM mice, further experiments on synaptic excitability were warranted. FGF19 significantly increased sEPSC frequency in most NTS neurons from T1DM mice, with no overall effect in control mice. This is consistent with the findings from Figure 1 and is likely a key driver of the FGF19-induced increase in NTS to DMV excitatory transmission. Intriguingly, and in contrast to several other metabolic hormones, this effect was abolished with the addition of TTX, suggesting the involvement of intact upstream neurons (21, 62). The small decrease in mean sEPSC amplitude is likely due to an increase in frequency of lower amplitude events and not a postsynaptic mechanism, since this effect was not retained with the addition of TTX. Since FGF19 was found to increase the activity of the AP to DMV glutamatergic connection, the effect of AP input to the NTS was tested. As hypothesized, glutamate release in the NTS that was due to activity of AP neurons was increased by FGF19.

Together, these results suggest that FGF19 increases glutamatergic neurotransmission at multiple points in the DVC circuit of T1DM mice. ICV administration of FGF19 decreases blood glucose concentration in T1DM but not control mice (6, 63, 64). This suggests that the effects on blood glucose are likely to be determined by neurophysiological differences between control and T1DM groups. Although FGF19 produced intrinsic inhibition in some NTS neurons, this effect was broadly similar between both animal groups and was relatively modest. Moreover, the effects on intrinsic excitability of NTS neurons do not appear to be consequential regarding their net excitatory influence on DMV neurons in T1DM mice. Rather, the more substantial effects of FGF19 on excitatory synaptic activity in T1DM mice play a greater role in the peptide’s effect on vagal motor output. The effects on intrinsic properties of NTS neurons observed here are also consistent with our prior results showing that FGF19 produced a small decrease in sEPSC frequency in some DMV neurons from control mice (6).

In addition to neuronal input, the DVC regulates metabolism in response to humoral signals. The AP and NTS lack a fully functional blood-brain barrier due to the presence of local fenestrated capillaries, which could allow the diffusion of humoral components into the DVC that might typically be excluded (19). Additionally, NTS and DMV neurons have wide dendritic fields (61, 65) that can extend to the AP border. As such, it is likely that neurons in all three DVC nuclei can sense and respond to humoral signals. FGF19 crosses the blood-brain barrier, albeit slowly, but it is unknown whether its entry into the brain is enhanced in areas that contain fenestrated capillaries (66). Neurons in the DVC have been shown to respond to many primary metabolic hormones (21–25, 34, 67). Similar to the effects of FGF19 described here, leptin, insulin, CCK-8, and ghrelin modify glutamatergic, but not GABAergic transmission in the DVC (21, 22, 34, 62, 68). Despite this, GABAergic transmission is considered to be a primary determinant of DMV neuron activity during resting conditions, since blockade of glutamatergic transmission in this area fails to produce effects on gastric motility or pancreatic secretion in normoglycemic mice (9, 11). However, an NMDA receptor antagonist applied to the NTS attenuated jejunal nutrient sensing-mediated mechanisms of glucose production (69). Interestingly, chemogenetic silencing of GABAergic NTS neurons fails to alter blood glucose concentration in normoglycemic mice, whereas increasing GABA neuron activity in the dorsal hindbrain increases blood glucose concentration (15). This suggests that glucoregulatory DMV neurons may not show the same GABA-dominant phenotype seen with other functional DMV subgroups.

The changes in cellular excitability in the DVC seen here are consistent with the beneficial metabolic effects associated with ICV administration of FGF19. ICV FGF19 has been shown to decrease hepatic glucose production (2) and hepatic expression of G6Pase, a key enzyme required for gluconeogenesis (3). These effects can be replicated by injection of excitatory neurotransmitters into the DVC (14) or by activating DMV neurons as measured by c-fos staining (70). Additionally, a common hallmark of FGFR activation is induction of phosphorylated extracellular signal-related kinase (pERK). Insulin activation of ERK signaling in the DVC is sufficient to decrease hepatic glucose production, suggesting that FGF19 may share this mechanism (67). Injection of NMDA in the DVC lowers hepatic glucose production and this effect was prevented by a hepatic vagotomy, suggesting that the effect was mediated by the excitation of DMV neurons (14). Additionally, increasing synaptic inhibition to the DMV increases blood glucose concentration (15). Together, these results are consistent with a model in which increased excitatory input to the DMV produces a decrease in blood glucose concentration (71). Accordingly, the effects of FGF19 seen here on DVC circuit dynamics are likely to produce beneficial effects on blood glucose levels in diabetic animals.

The effects of FGF19 in the DVC found here may also modulate metabolism independently of autonomic mechanisms. Central delivery of FGF19 has been shown to decrease food intake, which could be explained by alterations in DVC neuron activity (4, 47). Similar to the effects of insulin on hepatic glucose production, satiation produced by CCK in the DVC requires induction of pERK, suggesting that FGF19 may decrease food intake via a shared mechanism. The NTS and AP also exhibit significant connections to other nuclei associated with regulation of ingestive behavior including the lateral parabrachial nucleus (PBN) (35, 72) and the hypothalamus (73). The experiments performed here focus on local circuitry within the DVC. As such, it is not known whether connections to other nuclei such as the PBN or hypothalamus are altered by FGF19. However, considering the significant expression of FGFR/β-Klotho throughout the NTS and AP, it is likely that FGF19 alters the activity of neurons that project centrally as well as those that participate in local vago-vagal reflexes (26, 28).

The DVC is a key metabolic regulatory area of the brain. Although the DVC is typically associated with autonomic regulation of metabolism, the AP and NTS also serve as a communication hub between the DVC and several other key nuclei. Alterations in DVC neuron activity, such as those seen here, are likely to produce profound effects on multiple aspects of metabolism. Consistent with our previous work, these results suggest that FGF19 alters parasympathetic output in T1DM mice by increasing synaptic excitability of DMV neurons. Interestingly, our findings demonstrate a role for the AP in the direct regulation of vago-vagal reflex mechanisms and that this connection can be modified by a metabolic hormone. Though not confirmed, it is possible that other metabolic hormones work similarly in this area. Overall, the electrophysiological effects seen here are consistent with the beneficial metabolic effects of ICV FGF19. Although these findings implicate a primarily autonomic mechanism, FGF19 was also found to significantly alter synaptic inputs originating from the AP and NTS. This raises the possibility that FGF19 may interact with the DVC to produce an array of beneficial metabolic effects through alteration of vagal parasympathetic output and possibly also through connections with other important metabolic regulatory nuclei.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by University of Kentucky Institutional Animal Care and Use Committee.

JW designed experiments, collected data, analyzed data, prepared figures, wrote, and edited the manuscript. BS designed experiments, edited manuscript. All authors contributed to the article and approved the submitted version.

BS is funded by the National Institutes of Health (grants R01 DK122811 and R01 DK056132).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.