Nutrient infusion in the dorsal vagal complex controls hepatic lipid and glucose metabolism in rats.

Hypothalamic regulation of lipid and glucose homeostasis is emerging, but whether the dorsal vagal complex (DVC) senses nutrients and regulates hepatic nutrient metabolism remains unclear. Here, we found in rats DVC oleic acid infusion suppressed hepatic secretion of triglyceride-rich very-low-density lipoprotein (VLDL-TG), which was disrupted by inhibiting DVC long-chain fatty acyl-CoA synthetase that in parallel disturbed lipid homeostasis during intravenous lipid infusion. DVC glucose infusion elevated local glucose levels similarly as intravenous glucose infusion and suppressed hepatic glucose production. This was independent of lactate metabolism as inhibiting lactate dehydrogenase failed to disrupt glucose sensing and neither could DVC lactate infusion recapitulate glucose effect. DVC oleic acid and glucose infusion failed to lower VLDL-TG secretion and glucose production in high-fat fed rats, while inhibiting DVC farnesoid X receptor enhanced oleic acid but not glucose sensing. Thus, an impairment of DVC nutrient sensing may lead to the disruption of lipid and glucose homeostasis in metabolic syndrome.

Introduction

Metabolic syndrome is characterized by hyperlipidemia and hyperglycemia due partly to increased hepatic secretion of triglyceride-rich very-low-density lipoprotein (VLDL-TG) and glucose production (DeFronzo et al., 1989; Reaven et al., 1993; Sherling et al., 2017). In recent decades, studies documented the role of the central nervous system (CNS) in lipid and glucose homeostasis (Grayson et al., 2013; Nogueiras et al., 2010; Schwartz et al., 2013; Taher et al., 2017).

The hypothalamus regulates lipid and glucose profiles in response to insulin (Koch et al., 2008; Obici et al., 2002a; Scherer et al., 2016; Yue and Lam, 2012), leptin (Buettner et al., 2008; Liu et al., 1998; Morton and Schwartz, 2011), glucagon-like peptide-1 (GLP-1) (Sandoval et al., 2008; Taher et al., 2014), ghrelin (Stark et al., 2015; Theander-Carrillo et al., 2006), melanocortin (Leckstrom et al., 2011; Nogueiras et al., 2007; Obici et al., 2001), and fibroblast grow factors (Morton et al., 2013; Scarlett et al., 2019) in rodents. In parallel, activation of ATP-sensitive potassium (KATP) channels in the hypothalamus and dorsal vagal complex (DVC) is sufficient and required for insulin to suppress glucose production in rodents during the pancreatic-euglycemic clamps (Filippi et al., 2012; Pocai et al., 2005), while oral KATP channel activator diazoxide intake in humans suppresses glucose production with the CNS implicated as the site of action (Kishore et al., 2011). Further, intranasal insulin delivery in healthy humans increases insulin sensitivity and lowers glucose production independent of changes in circulating insulin levels (Dash et al., 2015; Heni et al., 2014), while genome-wide association study-identified gene DUSP8 mediates hypothalamic insulin sensitivity in both rodents and humans (Schriever et al., 2020). The relative contribution of the hypothalamus vs. DVC in glucose homeostasis remains unknown.

The hypothalamus detects amino acids (Arrieta-Cruz et al., 2013; Su et al., 2012), oleic acids (Obici et al., 2002b; Pocai et al., 2006), and glucose (Lam et al., 2005a) to regulate hepatic glucose production. Hypothalamic KATP channels is required for glucose sensing (Lam et al., 2005a), and such a finding was recently replicated in rodents as well as implicated in humans (Carey et al., 2020). Further, genetic manipulation of glucose-sensitive hypothalamic pro-opiomelanocortin neurons impairs glucose tolerance (Parton et al., 2007), while hypothalamic infusion of glucose blunts counter-regulation during hypoglycemia (Borg et al., 1997; Frizzell et al., 1993). Hypothalamic nutrient sensing regulates lipid homeostasis as well since hypothalamic infusion of oleic acid and glucose suppresses VLDL-TG secretion (Lam et al., 2007; Yue et al., 2015). As hypothalamic oleic acid infusion lowers feeding and neuropeptide Y expression (Obici et al., 2002b), hypothalamic infusion of neuropeptide Y exerts the opposing effect as oleic acid in elevating VLDL-TG secretion (Bruinstroop et al., 2012; Stafford et al., 2008). To date, hypothalamic nutrient sensing mechanism remains elusive.

Glucose transporter-1 in the hypothalamic glial cells and hypothalamic lactate dehydrogenase (LDH)-dependent metabolism as well as glucose/lactate-responsive neurons mediate glucose sensing (Chari et al., 2011; Lam et al., 2005a; Yoon and Diano, 2021), while leptin enhances LDH-dependent hypothalamic glucose sensing mechanism to regulate glucose production as well in high fat (HF) rats (Abraham et al., 2018). On the other hand, hypothalamic esterification of long-chain fatty acids via long-chain acyl-CoA synthase (ACSL) is required for oleic acid sensing (Lam et al., 2005b). NMDA receptor-mediated transmission in the DVC relays both hypothalamic glucose and oleic acid sensing mechanism to lower hepatic glucose production and VLDL-TG secretion, respectively (Lam et al., 2011; Yue et al., 2015). DVC is a highly permeable region in the brainstem which contains a circumventricular organ termed the area postrema as well as the nucleus of the solitary tract and the dorsal motor nucleus of the vagus. DVC relays signals from the gut and midbrain to facilitate energy and nutrient homeostasis (Blouet et al., 2009; Grill and Hayes, 2012; Lam et al., 2011; Waise et al., 2018; Wang et al., 2008; Yue et al., 2015) but also senses insulin (Filippi et al., 2012, 2014, 2017; Patel et al., 2021; Zhang et al., 2020), glucagon (LaPierre et al., 2015), leptin (Hayes et al., 2010; Kanoski et al., 2012), and GLP-1 (Alhadeff et al., 2017; Hayes et al., 2008, 2009) to regulate metabolism. DVC senses amino acid to lower food intake (Blouet and Schwartz, 2012), while DVC glucoprivation activates counter-regulation and increases food intake as well (Andrew et al., 2007; Ritter et al., 2000). Indeed, DVC contains glucose-sensitive neurons (Balfour et al., 2006; Mizuno and Oomura, 1984) and astrocytes that increase calcium influx in response to glucoprivation (McDougal et al., 2013). Given that ACSL (Bauer et al., 2018a; Briski et al., 2017) and LDH (Laughton et al., 2000) are expressed in the DVC, we evaluated whether ACSL or LDH-dependent lactate metabolism is necessary for fatty acid and glucose sensing.

Rats fed with a HF lard oil-enriched diet for 3 days develop insulin resistance, hyperlipidemia, and elevated hepatic secretion of VLDL-TG independent of obesity (Abraham et al., 2018; Chari et al., 2008; Côté et al., 2015; Scherer et al., 2012; Wang et al., 2001; Yue et al., 2012, 2015). Short-term HF feeding increases inflammation and endoplasmic reticulum stress in the hypothalamus (Ono et al., 2008; Thaler et al., 2012) and DVC (Filippi et al., 2017), resulting in insulin resistance. The ability of insulin to suppress glucose production is also blunted in lean and obese women after 3d HF feeding (Cornier et al., 2006) and in healthy men after 5 d of HF feeding (Brøns et al., 2009). While hypothalamic nutrient sensing is disrupted in HF feeding (Abraham et al., 2018; Yue et al., 2015), the impact on DVC nutrient sensing remains unknown. Further, inhibition of DVC farnesoid X receptor (FXR) enhances insulin sensitivity (Zhang et al., 2020) and in the small intestine mediates oleic acids and metformin to regulate glucose homeostasis (Bauer et al., 2018a; Jiang et al., 2015; Sun et al., 2018). Thus, we first tested whether DVC detects nutrients to regulate hepatic nutrient metabolism. Second, we evaluated the role of ACSL- and LDH-mediated DVC nutrient metabolism. Finally, we evaluated DVC nutrient sensing and the role of FXR in HF-fed rats. Our findings have revealed that infusion of nutrients into the DVC impacts systemic nutrient metabolism in rats.

Results

Oleic acid sensing in the DVC lowers VLDL-TG secretion via ACSL

To examine whether the DVC senses oleic acids to regulate lipid homeostasis via a negative feedback pathway in rats in vivo, we have directly infused 1 mM oleic acid into DVC of 8-10 hr fasted, healthy, unrestrained rats at 0.33 ul/hr for 160 min (Figures 1A, 1B, S1A, and S1B). The dosage of oleic acids infusion selected has been documented to lower hepatic VLDL-TG secretion when infused into the hypothalamus of rats (Yue et al., 2015). To evaluate changes in lipid metabolism, lipoprotein lipase was inhibited either by chemical inhibitor tyloxapol or poloxamer to prevent VLDL-TG clearance in the plasma (Figure 1B). Then, plasma triglyceride (TG) accumulation was measured as a function of time to determine the hepatic secretion rate of VLDL-TG. Chylomicrons do not contribute to the rise of plasma TG in rats fasted for at least for 4 hr (Stafford et al., 2008). Herein, we have first found that plasma TG levels and the rate of VLDL-TG secretion of rats given DVC vehicle (2.6% cyclodextrin) infusion with tyloxapol injection were comparable with those of previous studies (Yue et al., 2012, 2015) (Figures 1C and 1D). Importantly, oleic acid administration into the DVC decreased the rise in plasma TG levels in lipoprotein lipase-inhibited rats (Figure 1C) at comparable body weights to vehicle (0.321 ± 0.007 kg vs. vehicle 0.318 ± 0.003 kg; n = 23, 8). This lipid-lowering effect by oleic acid infusion into the DVC corresponded to a reduction in VLDL-TG secretion rate (Figure 1D).

Figure 1

Oleic acid sensing in the DVC lowers VLDL-TG secretion via ACSL

(A) Schematic representation of working hypothesis.

(B) Experimental protocol for VLDL-TG assessments. IDL: intermediate-density lipoprotein; FFA: free fatty acid.

(C and D) (C) Plasma TG levels and (D) VLDL-TG secretion rate of tyloxapol-infused rats with DVC vehicle or oleate infusions (n = 23, 8).

(E and F) (E) Plasma TG level and (F) VLDL-TG secretion rate of tyloxapol- or poloxamer-infused rats with DVC vehicle infusions (n = 23, 14).

(G and H) (G) Plasma TG levels and (H) VLDL-TG secretion rate of poloxamer-infused rats with DVC vehicle or oleate infusions (n = 14, 16).

(I and J) (I) Plasma TG levels and (J) VLDL-TG secretion rate of rats receiving vehicle, ACSL inhibitor triacsin C, triacsin C + oleate, or oleate (n = 23, 5, 6, 8).

∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 vs. all other group(s) as determined by unpaired t test or one-way analysis of variance (ANOVA) with Tukey's test where appropriate. Data are presented as mean ± standard error of the mean.

An alternative lipoprotein lipase inhibitor, poloxamer, that is much easier to dissolve than tyloxapol was also used. In rats with comparable body weight (tyloxapol 0.321 ± 0.007 kg vs. poloxamer 0.329 ± 0.005 kg; n = 23, 14), plasma TG level and VLDL-TG secretion rate of DVC vehicle-infused rats injected with poloxamer vs. tyloxapol had identical plasma TG levels and VLDL-TG accumulation rate (Figures 1E and 1F). Importantly, DVC oleic acid vs. vehicle administration was equally effective in lowering plasma TG levels and the rate of VLDL-TG secretion in poloxamer-injected rats (Figures 1G and 1H) at comparable body weight and food intake (Table 1). Thus, oleic acid infusion into the DVC lowers VLDL-TG secretion via the use of two alternative assessment methods.

Table 1

Food intake and body weight of rats which underwent VLDL (poloxamer-infused) and clamp experiments

	3d Cumulative food intake (kcal)	Body weight (kg)
VLDLRC + DVC vehicle (cyclodextrin) (n = 14)RC + DVC oleic acid (n = 16)RC + DVC vehicle (DMSO) + IV saline (n = 7)RC + DVC vehicle (DMSO) +IV IH (n = 7)RC + DVC triacsin C + IV IH (n = 6)	164.0 ± 5.6183.3 ± 5.0159.0 ± 9.6172.3 ± 6.9157.1 ± 13	0.329 ± 0.0050.336 ± 0.0040.334 ± 0.0090.332 ± 0.0060.327 ± 0.010
HFD + DVC vehicle (cyclodextrin) (n = 13)HFD + DVC oleic acid (n = 10)HFD + DVC GMCA (n = 6)HFD + DVC GMCA + oleic acid (n = 5)	235.8 ± 3.9a232.1 ± 3.2a243.1 ± 6.7a232.4 ± 4.7a	0.340 ± 0.0030.342 ± 0.0050.338 ± 0.0040.359 ± 0.007
Pancreatic clampsEuglycemic clampsRC + DVC saline (n = 9)RC + DVC glucose (n = 24)RC + DVC lactate (n = 14)RC + DVC oxamate + glucose (n = 5)Hyperglycemic ClampRC (n = 7)	185.6 ± 7.8171.0 ± 8.7174.7 ± 15164.3 ± 12150.6 ± 13	0.323 ± 0.0090.324 ± 0.0040.322 ± 0.0060.321 ± 0.0080.331 ± 0.007
HFD + DVC saline (n = 11)HFD + DVC glucose (n = 14)HFD + DVC GMCA (n = 6)HFD + DVC GMCA + glucose (n = 7)	230.1 ± 15a232.5 ± 9.8a252.9 ± 13a238.5 ± 14a	0.334 ± 0.0090.338 ± 0.0060.340 ± 0.0060.339 ± 0.011

p < 0.05 vs. RC groups as determined one-way ANOVA with Tukey's test. Data are presented as mean ± standard error of the mean for 3d cumulative food intake and body weight obtained prior to the VLDL or clamp studies.

Oleic acid sensing in the upper small intestine and hypothalamus requires a long-chain acyl-CoA synthetase to esterify long-chain fatty acids to regulate hepatic glucose production and glucose homeostasis (Bauer et al., 2018a; Lam et al., 2005b; Wang et al., 2008). To begin delineating the mechanism underlying DVC oleic acid sensing in lipid homeostasis, we have infused triacsin C, a long-chain acyl-CoA synthetase inhibitor, to evaluate whether ACSL is required (Figures 1A and 1B). DVC administration of triacsin C alone vs. vehicle had no effect on VLDL-TG profiles in tyloxapol-injected rats at comparable weight (Figures 1I and 1J and Table 1). However, co-infusion of triacsin C with oleic acid into the DVC negated the ability of oleic acid infusion to lower plasma TG levels and VLDL-TG secretion rate (Figures 1I and 1J). Thus, long-chain acyl-CoA synthetase in the DVC is necessary for oleic acid sensing to lower hepatic VLDL-TG secretion in healthy rats.

DVC sensing of circulating lipid is required for VLDL-TG regulation

To begin examining the physiological relevance of DVC oleic acid sensing in whole-body lipid homeostasis, we infused 20% intralipid with heparin intravenously (IV IH) to elevate circulating free fatty acids levels (Chapados et al., 2009; Lam et al., 2003; Zhang et al., 2004) at a dose that would activate hypothalamic lipid sensing (Lam et al., 2005b). Concurrently, we infused triacsin C into the DVC at a dose that will negate DVC oleic acid sensing (Figures 1I and 1J) to evaluate whether VLDL-TG secretion in responses to IV IH would be disrupted (Figures 2A and 2B). We first detected a rise in plasma TG levels after 180 min of IV IH infusion (IV IH + DVC vehicle: −180 min = 0.33 ± 0.02 vs. 0 min = 0.86 ± 0.10 mM, p < 0.001; IV IH + DVC triacsin C: −180 min = 0.31 ± 0.03 vs. 0 min = 0.89 ± 0.06 mM, p < 0.0001). Rats that received intravenous (IV) saline for 180 min did not present any difference in plasma TG levels (−180 min = 0.368 ± 0.04, 0 min = 0.384 ± 0.03 mM). Consistent with previous studies (Chapados et al., 2009; Zhang et al., 2004), IV IH + DVC vehicle vs. IV saline + DVC vehicle infusion increased VLDL-TG secretion 45 min after poloxamer injection (Figures 2C and 2D). Importantly, disruption of lipid sensing mechanism in the DVC by DVC triacsin C infusion in the presence of IV IH infusion resulted in higher plasma TG levels and VLDL-TG secretion as compared to rats receiving IV IH + DVC vehicle infusion (Figures 2C and 2D). These results demonstrate that DVC senses circulating lipids to regulate systemic lipid homeostasis during a physiological increase of plasma lipid levels in rats.

Figure 2

DVC sensing of circulating lipid is required for VLDL-TG regulation

(A) Schematic representation of working hypothesis.

(B) Experimental protocol for intravenous intralipid + DVC infusion followed by VLDL-TG assessments.

(C and D) (C) Plasma TG levels and (D) VLDL-TG secretion rate of poloxamer-infused rats with DVC vehicle + intravenous (IV) saline, DVC vehicle + IV intralipid + heparin (IH), or DVC triacsin C + IV IH infusions (n = 7, 7, 6).

∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ###p < 0.001 as determined by one-way ANOVA with Tukey's test. Data are presented as mean ± standard error of the mean.

Glucose sensing in the DVC suppresses glucose production independent of lactate metabolism

To examine whether DVC detects a rise in glucose to regulate glucose homeostasis via a negative feedback pathway in rats in vivo, we have administered 2 mM glucose vs. saline into the DVC at a dosage that will lower hepatic glucose production when administered to the hypothalamus (Chari et al., 2011; Lam et al., 2005a; Yang et al., 2010) and performed the pancreatic (basal insulin)-euglycemic clamps to evaluate changes in glucose kinetics in steady state (Figure 3A). Rats were comparable in weight and food intake (Table 1) and were randomly selected to receive DVC saline or glucose infusions. DVC glucose vs. saline infusion for 210 min increased the exogenous glucose infusion (Figure 3B) required to maintain euglycemia (Table 2) during the clamps. Analysis using the tracer dilution methodology revealed that the increased demand of glucose infusion was due to a suppression of glucose production (Figures 3C and 3D) and not an increase in glucose uptake (Figure 3E). Thus, glucose infusion into the DVC lowers glucose production in healthy rats.

Figure 3

Glucose sensing in the DVC suppresses glucose production

(A) Schematic representation of working hypothesis and the pancreatic (basal insulin)-euglycemic clamp protocol.

(B–E) (B) Glucose infusion rate, (C) basal and clamp glucose production, (D) percentage suppression of clamp glucose production from basal, and (E) glucose uptake of rats with DVC saline or glucose infusions (n = 9, 24). ∗∗∗∗p < 0.0001 vs. DVC saline as determined by unpaired t test.

(F–I) (F) Glucose infusion rate, (G) basal and clamp glucose production, (H) percentage suppression of clamp glucose production from basal, and (I) glucose uptake of rats with DVC saline, lactate, oxamate + glucose, or glucose infusions (n = 9, 14, 5, 24).

∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 vs. DVC saline and DVC lactate as determined by one-way ANOVA with Tukey's test. Data are presented as mean ± standard error of the mean.

Table 2

Plasma glucose levels of rats during basal and clamp steady states of the pancreatic basal-insulin euglycemic clamps

	Plasma glucose (mM)
	Basal (60-90 min)	Clamp (180-210 min)
RC + saline (n = 9)RC + glucose (n = 24)RC + lactate (n = 14)RC + oxamate + glucose (n = 5)	7.6 ± 0.37.3 ± 0.17.2 ± 0.27.1 ± 0.3	6.4 ± 0.47.1 ± 0.37.6 ± 0.36.5 ± 0.3
HFD + saline (n = 11)HFD + glucose (n = 14)HFD + GMCA (n = 6)HFD + GMCA + glucose (n = 7)	7.8 ± 0.17.9 ± 0.28.2 ± 0.48.1 ± 0.1	7.3 ± 0.36.7 ± 0.27.6 ± 0.37.7 ± 0.3

Data are presented as mean ± standard error of the mean.

To examine whether the astrocyte-neuronal lactate shuttle mediates glucose sensing in the DVC, we have first infused 5 mM lactate into the DVC of rats with comparable body weight and food intake to saline-infused rats (Table 1) to see whether it recapitulates the glucoregulatory effect of glucose. In contrast to the hypothalamus (Lam et al., 2005a; Yang et al., 2010), lactate infusion into the DVC failed to alter the glucose infusion rate and glucose production during the clamps (Figures 3F–3I and Table 2). To alternatively address for a potential glucoregulatory role of lactate metabolism in the DVC, we have co-infused the LDH inhibitor oxamate with glucose into the DVC at a dosage that will negate both lactate and glucose sensing in the hypothalamus (Abraham et al., 2018; Lam et al., 2005a). Co-infusion of oxamate with glucose into the DVC did not negate the ability of DVC glucose infusion to increase the glucose infusion rate (Figure 3F) required to maintain euglycemia (Table 2) during the clamps, which corresponded to a reduction in glucose production (Figures 3G and 3H) and not changes in glucose uptake (Figure 3I). Together, these results demonstrate that lactate is neither sufficient nor necessary for glucose infusion into the DVC to lower glucose production in healthy rats.

To verify that brain glucose infusion targeted the DVC, we independently infused radioactive tracer glucose through the DVC cannula at identical rate and duration as the DVC glucose infusion in clamps and collected DVC, mediobasal hypothalamus (MBH), and cortical tissue at the end. DVC infusion of radioactive glucose was found to selectively increase radioactivity in the DVC but not in the MBH and cortex of the same rats (Figure 4A). Next, to begin examining the physiological relevance of DVC glucose sensing, we measured the glucose levels of DVC and MBH tissues obtained from rats that were clamped under euglycemic conditions receiving DVC saline or glucose and systemic hyperglycemic-pancreatic (basal insulin) clamp conditions. DVC glucose vs. saline infusion elevated glucose levels in the DVC by ∼2 fold (Figure 4B) but not in the MBH (Figure 4C) while maintained under comparable plasma glucose levels (DVC glucose: 6.9 ± 0.33 vs. DVC saline: 6.7 ± 0.15 mM). We next performed systemic hyperglycemic-pancreatic clamps and raised and maintained plasma glucose levels by ∼2.5 fold in rats (to 17.1 ± 0.49 mM, n = 7) by 2 hr that was previously documented to elevate hypothalamic glucose levels and activate hypothalamic glucose sensing to suppress glucose production (Lam et al., 2005a). We here first confirmed that MBH glucose levels were indeed raised (Figure 4C) by circulating hyperglycemia and found that DVC glucose levels (Figure 4B) were similarly elevated in the same rats. More importantly, the degree of elevation of DVC glucose levels incurred by the hyperglycemic clamps was comparable to that of the DVC glucose levels detected in response to DVC glucose infusion (Figure 4B). Together, these results further validate accurate placement and infusion of the brain cannula targeting the DVC and support that direct DVC glucose infusion elevates DVC glucose levels to a comparable extent as a physiological rise in plasma glucose concentration that would lower glucose production in healthy rats.

Figure 4

Infusion of glucose is localized in the DVC and elevates DVC glucose levels at a comparable extent as seen with systemic hyperglycemia

(A) Radioactivity level of DVC, MBH, and cortex tissue after DVC [3-3H]-glucose infusion (n = 9).

(B and C) (B) DVC and (C) MBH tissue glucose levels after DVC saline-euglycemic, DVC glucose-euglycemic, or systemic hyperglycemic clamps (n = 9, 6, 7).

∗p < 0.05, ∗∗∗p < 0.001 as determined by one-way ANOVA with Tukey's test. Data are presented as mean ± s.e.m.

HF feeding disrupts oleic acid sensing in the DVC to suppress VLDL-TG secretion

To address for the pathological relevance of nutrient sensing in the DVC, we have fed surgically recovered rats with 3 days of high-fat diet (HFD) to see whether prior to the onset of obesity and in this short-term metabolically challenged model (Figures 1A, 1B, and 3A) DVC nutrient sensing remains intact. First, we have monitored the food intake and body weight of the HFD rats and confirmed that the rats became hyperphagic (Figure 5A), while no significant changes in body weight (Table 1) were observed. We have further detected a disruption of lipid homeostasis in these HF rats as HF vs. RC rats have elevated basal plasma TG levels (RC: 0.30 ± 0.02 mM vs. HFD 0.51 ± 0.03 mM, p < 0.0001) together with increased plasma TG levels and hepatic secretion of VLDL-TG in poloxamer-injected rats (Figures 5B and 5C). Importantly, oleic acid infusion into the DVC failed to lower plasma TG levels and hepatic secretion rate of VLDL-TG in poloxamer-injected HF rats as compared to the oleate effect that was seen in healthy rats (Figures 5B and 5C).

Figure 5

HF feeding disrupts oleic acid sensing in the DVC

(A) 3d cumulative food intake of DVC vehicle (RC), DVC vehicle (HFD), DVC oleate (HFD), or DVC oleate (RC) rats prior to experimentation (n = 14, 13, 10, 16). ∗∗∗∗p < 0.0001 vs. RC groups.

(B and C) (B) Plasma TG levels and (C) VLDL-TG secretion rate of DVC vehicle (RC), DVC vehicle (HFD), DVC oleate (HFD), or DVC oleate (RC) infusion (n = 14, 13, 10, 16).

#p < 0.05, ###p < 0.001 for DVC vehicle/oleate (HFD) vs. RC groups. ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 for DVC vehicle (RC) vs. DVC oleate (RC) as determined by two-way ANOVA with Tukey's test. Data are presented as mean ± standard error of the mean.

HF feeding disrupts glucose sensing in the DVC to suppress glucose production

Next, in an independent cohort of rats, we have first confirmed again that HF vs. RC rats were hyperphagic (Figure 6A) but with no changes in body weight (Table 1). We have then performed the pancreatic-euglycemic clamps in HF rats that received saline or glucose infusion into the DVC as in healthy rats (Figure 3A). In contrast to healthy rats, glucose administration into the DVC failed to increase glucose infusion rate (Figure 6B) during the clamps in maintaining euglycemia (Table 2). Further, glucose production and glucose uptake remained comparable among groups (Figures 6C–6E). Taken together, these results demonstrate that short-term HF feeding disrupts oleic acid and glucose infusion into the DVC to regulate hepatic VLDL-TG and glucose production.

Figure 6

HF feeding disrupts glucose sensing in the DVC

(A) 3d cumulative food intake of DVC saline (RC), DVC saline (HFD), DVC glucose (HFD), or DVC glucose (RC) rats prior to experimentation (n = 9, 11, 14, 24). ∗∗p < 0.01, ∗∗∗∗p < 0.0001 vs. RC groups.

(B–E) (B) Glucose infusion rate, (C) basal and clamp glucose production, (D) percentage suppression of clamp glucose production from basal, and (E) glucose uptake of DVC saline (RC), DVC saline (HFD), DVC glucose (HFD), or DVC glucose (RC) rats (n = 9, 11, 14, 24).

∗∗∗∗p < 0.0001 vs. all other groups as determined by two-way ANOVA with Tukey's test. Data are presented as mean ± standard error of the mean.

FXR antagonism rescues oleic acid, but not glucose, sensing in the DVC

Next, we have tested whether inhibiting FXR via chemical inhibitor glyco-β-muricholic acid (GMCA) in the DVC could enhance oleic acid and glucose sensing to regulate lipid and glucose homeostasis in HF rats (Figures 1A and 3A). We have administered GMCA in the DVC of HF rats at the same dosage as previously described that would inhibit FXR signaling in the DVC of rats in vivo (Zhang et al., 2020) and first tested the response of DVC oleic acid sensing in VLDL-TG regulation (Figure 7A). In poloxamer-injected HF vs. RC rats that were hyperphagic (Figure 7B) but with comparable body weight (Table 1), GMCA administration in the DVC enhanced the ability of DVC oleic acid infusion to lower plasma TG levels and hepatic VLDL-TG secretion (Figures 7C and 7D) in HF rats to a similar level as seen in healthy rats (Figures 1G and 1H), while GMCA given at the current dosage into the DVC per se did not alter plasma TG levels and VLDL-TG secretion rate (Figures 7C and 7D) as compared to vehicle (Figures 5B and 5C).

Figure 7

FXR antagonism rescues oleic acid, but not glucose, sensing in the DVC

(A) Schematic representation of the experimental protocol.

(B) 3d cumulative food intake of DVC vehicle (RC), DVC GMCA + oleate (HFD), or DVC GMCA (HFD) rats prior to experimentation (n = 14, 5, 6). ∗∗∗∗p < 0.0001 vs. DVC vehicle (RC).

(C and D) (C) Plasma TG levels and (D) VLDL-TG secretion rate of HFD rats with DVC oleate, GMCA + oleate, or GMCA infusions (n = 10, 5, 6). ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 vs. all groups.

(E) 3d cumulative food intake of DVC saline (RC), DVC GMCA + glucose (HFD), or DVC GMCA (HFD) rats prior to experimentation (n = 9, 7, 6). ∗p < 0.05, ∗∗p < 0.01 vs. DVC saline (RC) as determined by one-way ANOVA with Tukey's test.

(F–I) (F) Glucose infusion rate, (G) basal and clamp glucose production, (H) percentage suppression of clamp glucose production from basal, and (I) glucose uptake of HFD animals with DVC glucose, GMCA + glucose, or GMCA infusions (n = 14, 7, 6). Data are presented as mean ± standard error of the mean.

Finally, we have administered GMCA in the DVC of HF rats at the same dosage as the VLDL-TG studies and tested the response of DVC glucose sensing during the clamps (Figure 7A). In HF vs. RC rats that were hyperphagic (Figure 7E) but with comparable body weight (Table 1), GMCA administration into the DVC failed to enhance the ability of DVC glucose infusion to regulate glucose metabolism in HF rats as compared to HF rats infused with DVC glucose (Figures 7F–7I), while DVC GMCA infusion alone did not alter glucose metabolism (Figures 7F–7I) as compared to saline (Figures 6B–6E). Taken together, these findings suggest that direct inhibition of FXR in the DVC of HF rats enhances oleic acid sensing to lower plasma TG and VLDL-TG secretion but fails to enhance DVC glucose sensing to regulate glucose metabolism.

Discussion

The metabolic syndrome is associated with atherogenic dyslipidemia that includes hypertriglyceridemia and increased liver production of VLDL-TG, as well as hyperglycemia resulting from unsuppressed hepatic glucose production (Sherling et al., 2017). Impaired lipid homeostasis, in turn, further exacerbates glucose dysregulation (Fu et al., 2012; Yue and Lam, 2012).

In this study, we have first demonstrated that ACSL is required for oleic acid infusion into the DVC to lower hepatic VLDL-TG secretion in rats. This is consistent with the fatty acid sensing mechanism observed for the upper small intestine (Bauer et al., 2018a; Wang et al., 2008) and the hypothalamus (Lam et al., 2005b). In addition, we have demonstrated that DVC lipid sensing is required for restraining and regulating VLDL-TG secretion in the presence of systemic elevated lipid levels. After short-term HF feeding, animals exhibit elevated basal plasma TG levels and concurrently impaired DVC lipid sensing. Together, these results suggest that failure of DVC lipid sensing to regulate VLDL-TG secretion may contribute to the dysregulation of lipid homeostasis in the short-term HF model and that the metabolic beneficial effects of acute CNS lipid sensing is offset by the long-term effect incurred by HF feeding. Future investigations are warranted to examine the physiological relevance of DVC oleic acid sensing and the associated impact in animal models with obesity and diabetes. In addition, in light of the fact that nutrient sensing in the hypothalamus and leptin action in the brain not only regulate hepatic VLDL-TG secretion but also subsequently affect hepatic TG levels in rodents (Lam et al., 2007; Hackl et al., 2019), further assessment on hepatic TG levels as well as lipogenesis is required in the current experimental settings.

HF not only negates fatty acid sensing in the small intestine (Bauer et al., 2018a; Cheung et al., 2009; Wang et al., 2008), hypothalamus (Yue et al., 2015), and as currently observed in DVC but also suppresses ACSL3 in the small intestine to disrupt fatty acid sensing (Bauer et al., 2018a). Given that ACSL3 is expressed in the DVC (Bauer et al., 2018a; Briski et al., 2017), we postulate that HF disrupts DVC oleic acid sensing via a reduction in ACSL3 expression. This is further supported by the fact that activation of upper small intestinal FXR in HF rats negates the ability of upper small intestinal healthy microbiome transplant to increase small intestinal ACSL3 expression and enhance fatty acid sensing glucoregulatory pathways (Bauer et al., 2018a), while direct inhibition of FXR in the DVC enhances oleic acid sensing to regulate hepatic VLDL-TG secretion in HF rats. In both the current and a previous study where direct infusion of glycine activated NMDA receptors in DVC to suppress hepatic VLDL-TG secretion, the first observance of lowered plasma TG levels was 30 min after lipid clearance is blocked (Yue et al., 2012). In the same study when hepatic vagotomy was performed, DVC glycine failed to alter VLDL-TG secretion. The ability of central secreted frizzled-related protein 5 and hypothalamic oleic acid sensing to suppress VLDL-TG secretion both require DVC NMDA signaling and hepatic vagal innervation (Li et al., 2020; Yue et al., 2015), together suggesting the possibility that DVC lipid sensing may modulate VLDL-TG secretion via NMDA receptors and the hepatic vagus. This warrants future investigations and may also give insights to the neuronal population mediating DVC oleic acid sensing.

On the other hand, DVC glucose sensing mechanism is in contrary to the hypothalamic mechanism (Abraham et al., 2018; Lam et al., 2005a) as lactate was neither sufficient nor necessary to lower glucose production. In addition, direct inhibition of FXR in the DVC failed to enhance glucose sensing in HF rats to lower glucose production. While increased lactate in the DVC blunts blood glucose level recovery during insulin-induced hypoglycemia and reduced lactate level by inhibition of monocarboxylate transporter increases blood glucose level (Gujar et al., 2014; Patil and Briski, 2005), it remains unknown why lactate infusion did not lower glucose production in the current pancreatic clamp setting where plasma glucose level was maintained at basal. The relevance of lactate metabolism lies in its potential to be converted to pyruvate and provide energy for the brain as supported by the astrocyte-neuron lactate shuttle hypothesis (Laughton et al., 2000; Pellerin and Magistretti, 1994), as well as studies demonstrating glucose metabolism-activated neurons also being responsive to lactate (Himmi et al., 2001; Yang et al., 1999). In our current findings, lactate did not mimic the effect of glucose even with substantial expression of LDH-B (converts lactate to pyruvate) detected in the DVC (Laughton et al., 2000), suggesting that the glucose sensing pathway we are examining may be independent of glucose metabolism (i.e. glycolysis). In support, glucose sensing in the small intestine does not require intestinal glucose metabolism to improve glucose tolerance but instead via the activation of sodium glucose cotransporter-1 and subsequent release of GLP-1 (Bauer et al., 2018b; Kuhre et al., 2015; Moriya et al., 2009). Consistently, the DVC contains glucose-sensitive neurons (Balfour et al., 2006), and glucose excites glucose-sensitive neurons via sodium glucose cotransporter-facilitated sodium entry (Gonzàlez et al., 2009; O'Malley et al., 2006). Although no study has directly identified sodium glucose transporter in the DVC, sodium glucose transporter is expressed at various regions of the brain (Poppe et al., 1997; Yu et al., 2013). In fact, a study infusing the sodium glucose cotransporter inhibitor phloridzin into the forth ventricle, which grants immediate exposure to the DVC, has demonstrated an elevation in food intake (Li et al., 2013). This finding may serve as an indicator of functional sodium glucose cotransporters in the region. Moreover, GLP-1-producing neurons are detected in the DVC, and when these neurons are activated in mice, it suppresses glucose production, promotes glucose tolerance, and improves hepatic insulin sensitivity (Shi et al., 2017). In support of a working hypothesis that glucose sensing in the DVC activates GLP-1-expressing neurons to facilitate glucose homeostasis, glucose has been documented to stimulate glutamate release (Boychuk et al., 2015; Roberts et al., 2017; Wan and Browning, 2008) and GLP-1-producing neurons have been characterized as glutamatergic (Zheng et al., 2015). Previous study has also demonstrated that the activation of NMDA receptors in the DVC lowers glucose production (Lam et al., 2010; Yue et al., 2016). Of note, GLP-1-producing neurons in the DVC may also mediate fatty acid sensing in regulating VLDL-TG secretion as fatty acid in the small intestine has been documented to stimulate GLP-1 release (Hirasawa et al., 2005; Iakoubov et al., 2007; Rocca et al., 2001), while activation of NMDA receptor in the DVC is sufficient and necessary to lower hepatic VLDL-TG secretion as well (Li et al., 2020; Yue et al., 2012, 2015). Nonetheless, future studies are required to further elucidate glucose and oleic acid sensing mechanism in the DVC, including the neuron type and neurocircuitry responsible as well as the time-dependent effect of nutrient sensing on VLDL-TG secretion and glucose production regulation, as a limitation of the current study is the incapacity to identify specific neurons within the DVC that are activated by nutrients to mediate the control of systemic metabolism.

It remains unknown why direct inhibition of FXR enhances fatty acid but not glucose sensing in the DVC of HF rats. One immediate postulation is that FXR inhibition may upregulate the expression of genes that mediate fatty acid sensing (i.e., ACSL3) but not glucose sensing (i.e., sodium glucose cotransporter-1) in the DVC. Indeed, HF feeding reduces ACSL3 expression in the upper small intestine, and this reduction is prevented by healthy microbiome transplant in an FXR-dependent manner (Bauer et al., 2018a). On the other hand, no study, to the best of our knowledge, has documented that FXR regulates SGLT expression.

We postulate that excess nutrient availability derived from the hyperphagic response induced by HF feeding may inhibit potential molecules within the CNS that would disrupt nutrient sensing. For instance, 3d of HF feeding reduces glucose transporter of the blood brain barrier and lowers brain glucose uptake (Jais et al., 2016). In an uncontrolled diabetic rat model with 24 hr sustained hyperglycaemia (excess glucose), hypothalamic glial cell glucose transporter-1 expression is suppressed, negating the ability of hypothalamic glucose sensing mechanism to elevate local glucose levels and suppress glucose production (Chari et al., 2011). Further, the ability of glucose to promote its own clearance, termed as glucose effectiveness, is lost in people with type 2 diabetes, while normalization of hyperglycaemia and/or hyperlipidemia restores glucose effectiveness (Hawkins et al., 2002). Nonetheless, it is to be acknowledged that a limitation of the current study is the assessment on DVC nutrient sensing resides only in the 3d HF-fed model. To address the effect of hyperphagia per se on DVC nutrient sensing, a model with high carbohydrate intake could also be tested. Future investigations are called for to investigate the contribution of DVC in regulation of lipid/glucose homeostasis in healthy state and whether restoration of DVC nutrient sensing pathways improves lipid/glucose homeostasis in not only diabetic and obese models.

In summary, we have unveiled fatty acid and glucose infusion into the DVC lower hepatic VLDL-TG secretion and glucose production in healthy but not HF rats and that direct inhibition of FXR in the DVC enhances fatty acid but not glucose sensing in vivo.

Limitations of study

The current study did not address the metabolic impact of nutrient sensing in the DVC of male and female rodents with obesity and/or diabetes and neither identify specific neuron(s) in the DVC that mediate nutrient sensing to regulate hepatic nutrient metabolism. The downstream mechanism of glucose sensing as well as the inability of FXR inhibition to enhance glucose sensing in the DVC remains unknown.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Tony K.T. Lam (tony.lam@uhnresearch.ca).

Materials availability

This study did not generate new unique reagents.

Data and code availability

This study did not use any unpublished custom code, software, or algorithm that is central to supporting the main claims of the paper.

Methods

All methods can be found in the accompanying transparent methods supplemental file.