**3. Methods**

400 Olive Oil – Constituents, Quality, Health Properties and Bioconversions

long-chain FA into mitochondria, mimicked these effects on food intake and HGP induced by icv infusion of OA (Obici et al., 2003). In another study a direct bilateral infusion of OA into the mediobasal hypothalamus decreased hepatic glucose production (Ross et al., 2010). In addition, it seems that the hypothalamus differentially senses FA. For example, icv infusions of OA or docosahexanoic acid, but not palmitic acid, reduce food intake and body weight (Schwinkendorf et al., 2010). However, icv and direct infusions of FA into the brain are not physiological. Thus, they might produce non-specific effects by evoking an inflammatory response by irritating ependymocytes and tanycytes lining the ventricles or

Fig. 1. Fragments of whole cell current clamp recordings of oleic acid (OA) excited (A) and inhibited (B) neuron in arcuate nucleus of rat (Adapted from Wang et al, 2006 et Migrenne et al, 2006).The inhibitory effect of OA on neuronal activity is inhibited by tolbutamide (C),

suggesting involvement of KATP channels in OA effect.

by exciting microglia and astrocytes in the brain parenchyma.

A

B

C

All animal care and experimental procedures were approved by the animal ethics committee of the university Paris-Diderot. Four weeks-old male wistar rats were purchased from Charles Rivers (Lyon, France) and housed at 21°C with normal light/dark cycle and free access to water and food.

#### **3.1 First serie of experiments**

Rats received an intracerebroventricular (icv) infusion of FA during 3 days. Briefly, rats anesthetized with isoflurane were stereotactically implanted with a chronic stainless steel

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 403

chloroform/methanol/HCl (100:100:1, v/v) and 1 M KCl. [-32P]-phosphatidic acid was resolved by TLC with chloroform/acetone/methanol/acetic acid/water (10:4:3:2:1, v/v) and quantified with a Molecular Dynamics Storm PhosphorImager. Known amounts of diacylgycerol and ceramide standards were included with each assay. Ceramide and diacylglycerol levels were expressed as pmol by nmol of phospholipid (PL) levels. Total phospholipids present in cellular lipid extracts used for ceramide analysis were quantified as described previously (Escalante-Alcalde et al., 2003; Le Stunff et al., 2002) with minor modifications. Briefly, a mixture of 10N H2SO4/70% perchloric acid (3:1, v/v) was added to lipid extracts which were incubated for 30 min at 210°C. After cooling, water and 4.2% ammonium molybdate in 4 N HCl/0.045% malachite green (1:3 v/v) was added. Samples

In the first serie of experiment, whatever the FA ie oleate or linolenate), there was no change in food intake during experiment (data not shown). Basal plasma glucose, FA, and insulin concentrations were also similar in all groups. As displayed in figure 2, in response to glucose load in linoleate group time course of glycemia was similar to control but was associated with an increased glucose induced insulin secretion (GIIS), suggesting an insulin resistance state which was compensated by this increased GIIS. In oleate infused group

 Fig. 2. Time course of plasma glucose and insulin concentration in response to glucose injection in oleate, linolenate, and NaCl (controls) icv 24h infused rats. \*\*, p < 0.01 vs oleate

In order to test whether β oxidation was required to relay FA effect, GIIS was measured in presence or not of etomoxir a specifix inhibitor of CPT1 activity, a rate-limiting enzyme of β oxidation (figure 3). Results are expressed as insulinogenic index (ie ratio of areas under the curve of insulin to glucose during GIIS). Effect of linolenate on GIIS was reversed by etomoxir.

were incubated at 37°C for 30 min, and absorbance was measured at 660 nm.

there was no change in plasma glucose or GIIS compared to controls.

**4. Results** 

and controls.

cannula in the right lateral cerebral ventricle. The cannula was connected via a polyethylene catheter to a subcutaneously osmotic minipump filled up with FA (oleate, octanoate or linolenate) or saline. Infusions started 6h after surgery. The rate of infusion was 0.5 µl/h. Blood was daily removed (~ 80 µl) from caudal vessels for measurement of plasma substrate (FA and glucose) and insulin. Food intake was daily measured. At day 3 of infusion glucose-induced insulin secretion (GIIS) was measured in response to a single intraperitoneal injection of glucose (0.5g/kg bw). was made in overnight fasted rats. The glycemia was determined by a glucometer (AccuChek, Rabalot, France) from 2 µl collected from the tip of the tail vein at time 0, 5, 10, 15, 20, 30 and 60 min. In addition 20 µl of blood was sampled at the same time for insulin measurement (RIA, Diasorin, France). In another serie of experiments, etomoxir (CPT1 inhibitor) was concomitantly infused with FA. At the end of experiment brain were removed and five hypothalamus nuclei (arcuate, lateral, ventromedian, paraventricular and dorsomedian) were micropunched in order to measure gene expression (acetylCoA carboxylase, ACC, carnitine palmitoyl transferase, CPT1, FA synthase, FAS, G protein related peptide GPR41). Briefly, total RNA was isolated from the hypothalamus using RNeasy Lipid kit (Qiagen). To remove residual DNA contamination, the RNA samples were treated with DNAse RNAse-free (Qiagen). 4 µg of total RNA from each sample was reverse transcribed with 40 U of M-MLV Reverse Transcriptase (Invitrogen, life technologies) using random hexamer primers.

#### **3.2 Second serie of experiments**

In a second serie of experiments our goal was to test the effect of different triglyceride emulsion on glucose tolerance and both diacylglycerol and ceramides content in hypothalamus. To that end rats received an intracarotid infusion during 24 h of lard oil, mainly composed of saturated FA (SFA), ω3-enriched (Omegaven, Santec, France) or ω6 enriched (Ivelip, Rabalot, France) polyunsaturated FA (PUFA) triglyceride emulsion. The long-term unrestrained infusion technique was used, as previously described (Gilbert et al., 2003). Briefly, 5 days before the beginning of the infusion, rats were anaesthetized with isoflurane for catheterization of right carotid artery, towards the brain. Catheter was then exteriorized at the vertex of the head, and animals were allowed to recover for 5 days. For infusion, catheter is connected to a swiveling infusion device, allowing the animal free access to water and food and infused with a triglyceride emulsion. Food intake was measured after the 24h infusion period. In another set of experiments, oral glucose tolerance test (3g/kg, OGTT) was also performed.

#### **3.3 Extraction and analysis of ceramids and DAG content in the hypothalamus**

Diacylglycerol and ceramide levels in tissues extracts were measured by the diacylglycerol kinase enzymatic method as previously described (Escalante-Alcalde et al., 2003; Le Stunff et al., 2002). Briefly, aliquots of the chloroform phases from cellular lipid extracts were resuspended in 7.5% (w/v) octyl-β-D-glucopyranoside/5 mM cardiolipin in 1 mM DETPAC/10 mM imidazole (pH 6.6). The enzymatic reaction was started by the addition of 20 mM DTT, 0.88 U/ml E. coli diacylglycerol kinase, 5 µCi/10 mM [-32P]ATP and the reaction buffer (100 mM imidazole (pH 6.6), 100 mM NaCl, 25 mM MgCl2, and 2 mM EGTA). After incubation for 1 h at room temperature, lipids were extracted with chloroform/methanol/HCl (100:100:1, v/v) and 1 M KCl. [-32P]-phosphatidic acid was resolved by TLC with chloroform/acetone/methanol/acetic acid/water (10:4:3:2:1, v/v) and quantified with a Molecular Dynamics Storm PhosphorImager. Known amounts of diacylgycerol and ceramide standards were included with each assay. Ceramide and diacylglycerol levels were expressed as pmol by nmol of phospholipid (PL) levels. Total phospholipids present in cellular lipid extracts used for ceramide analysis were quantified as described previously (Escalante-Alcalde et al., 2003; Le Stunff et al., 2002) with minor modifications. Briefly, a mixture of 10N H2SO4/70% perchloric acid (3:1, v/v) was added to lipid extracts which were incubated for 30 min at 210°C. After cooling, water and 4.2% ammonium molybdate in 4 N HCl/0.045% malachite green (1:3 v/v) was added. Samples were incubated at 37°C for 30 min, and absorbance was measured at 660 nm.

### **4. Results**

402 Olive Oil – Constituents, Quality, Health Properties and Bioconversions

cannula in the right lateral cerebral ventricle. The cannula was connected via a polyethylene catheter to a subcutaneously osmotic minipump filled up with FA (oleate, octanoate or linolenate) or saline. Infusions started 6h after surgery. The rate of infusion was 0.5 µl/h. Blood was daily removed (~ 80 µl) from caudal vessels for measurement of plasma substrate (FA and glucose) and insulin. Food intake was daily measured. At day 3 of infusion glucose-induced insulin secretion (GIIS) was measured in response to a single intraperitoneal injection of glucose (0.5g/kg bw). was made in overnight fasted rats. The glycemia was determined by a glucometer (AccuChek, Rabalot, France) from 2 µl collected from the tip of the tail vein at time 0, 5, 10, 15, 20, 30 and 60 min. In addition 20 µl of blood was sampled at the same time for insulin measurement (RIA, Diasorin, France). In another serie of experiments, etomoxir (CPT1 inhibitor) was concomitantly infused with FA. At the end of experiment brain were removed and five hypothalamus nuclei (arcuate, lateral, ventromedian, paraventricular and dorsomedian) were micropunched in order to measure gene expression (acetylCoA carboxylase, ACC, carnitine palmitoyl transferase, CPT1, FA synthase, FAS, G protein related peptide GPR41). Briefly, total RNA was isolated from the hypothalamus using RNeasy Lipid kit (Qiagen). To remove residual DNA contamination, the RNA samples were treated with DNAse RNAse-free (Qiagen). 4 µg of total RNA from each sample was reverse transcribed with 40 U of M-MLV Reverse Transcriptase (Invitrogen, life technologies) using random

In a second serie of experiments our goal was to test the effect of different triglyceride emulsion on glucose tolerance and both diacylglycerol and ceramides content in hypothalamus. To that end rats received an intracarotid infusion during 24 h of lard oil, mainly composed of saturated FA (SFA), ω3-enriched (Omegaven, Santec, France) or ω6 enriched (Ivelip, Rabalot, France) polyunsaturated FA (PUFA) triglyceride emulsion. The long-term unrestrained infusion technique was used, as previously described (Gilbert et al., 2003). Briefly, 5 days before the beginning of the infusion, rats were anaesthetized with isoflurane for catheterization of right carotid artery, towards the brain. Catheter was then exteriorized at the vertex of the head, and animals were allowed to recover for 5 days. For infusion, catheter is connected to a swiveling infusion device, allowing the animal free access to water and food and infused with a triglyceride emulsion. Food intake was measured after the 24h infusion period. In another set of experiments, oral glucose tolerance

**3.3 Extraction and analysis of ceramids and DAG content in the hypothalamus** 

Diacylglycerol and ceramide levels in tissues extracts were measured by the diacylglycerol kinase enzymatic method as previously described (Escalante-Alcalde et al., 2003; Le Stunff et al., 2002). Briefly, aliquots of the chloroform phases from cellular lipid extracts were resuspended in 7.5% (w/v) octyl-β-D-glucopyranoside/5 mM cardiolipin in 1 mM DETPAC/10 mM imidazole (pH 6.6). The enzymatic reaction was started by the addition of 20 mM DTT, 0.88 U/ml E. coli diacylglycerol kinase, 5 µCi/10 mM [-32P]ATP and the reaction buffer (100 mM imidazole (pH 6.6), 100 mM NaCl, 25 mM MgCl2, and 2 mM EGTA). After incubation for 1 h at room temperature, lipids were extracted with

hexamer primers.

**3.2 Second serie of experiments** 

test (3g/kg, OGTT) was also performed.

In the first serie of experiment, whatever the FA ie oleate or linolenate), there was no change in food intake during experiment (data not shown). Basal plasma glucose, FA, and insulin concentrations were also similar in all groups. As displayed in figure 2, in response to glucose load in linoleate group time course of glycemia was similar to control but was associated with an increased glucose induced insulin secretion (GIIS), suggesting an insulin resistance state which was compensated by this increased GIIS. In oleate infused group there was no change in plasma glucose or GIIS compared to controls.

Fig. 2. Time course of plasma glucose and insulin concentration in response to glucose injection in oleate, linolenate, and NaCl (controls) icv 24h infused rats. \*\*, p < 0.01 vs oleate and controls.

In order to test whether β oxidation was required to relay FA effect, GIIS was measured in presence or not of etomoxir a specifix inhibitor of CPT1 activity, a rate-limiting enzyme of β oxidation (figure 3). Results are expressed as insulinogenic index (ie ratio of areas under the curve of insulin to glucose during GIIS). Effect of linolenate on GIIS was reversed by etomoxir.

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 405

Hypothalamic gene expression were also modified in some areas depending on FA (figure 5). For example, FAS expression was inhibited in all nuclei except in VMH of linolenate

Fig. 5. mRNA expressionof target gene in different hypothalamic areas in oleate or linolenate +/- etomoxir. The green line represents gene expression in control rats (infused with NaCl). \*\*, p <0.01 vs controls. §§, p <0.01 vs controls. ARC: arcuate nucleus; VMH:

ventromedian hypothalamus; LH: lateral hypothalamus; DMH: dorsomedian

hypothalamus; PVN: paraventricular nucleus.

infused rats. GPR41 was up-regulated in ARC of linolenate infused rats.

Fig. 3. Insulinogenic index in control rats, oleate and linolenate infused rats with or without etomoxir, a specific inhibitor of CPT1.\*\*\* p<0.01 vs controls

Hypothalamic gene expression have been measured in five areas (figure 4) involved in nervous control of energy balance: arcuate nucleus (ARC), ventromedian (VMH), lateral (LH), dorsomedian hypothalamus (DMH) and paraventricular nucleus (PVN). Studied genes were CPT1 (carnitine palmitoyl transferase 1), FAS (fatty acid synthase), GPR41 (G protein related receptor 41, GPR40 and 43 have been also tested but were not detected in our models), ACCβ (Acetyl carboxylase β) and AMPKα2. Results are displayed in figure 5.

Fig. 4. Map of hypothalamic areas involved in nervous control of energy homeostasis. Arcuate nucleus (ARC), ventromedian (VMH), lateral (LH), dorsomedian hypothalamus (DMH) and paraventricular nucleus (PVN).

Fig. 3. Insulinogenic index in control rats, oleate and linolenate infused rats with or without

Hypothalamic gene expression have been measured in five areas (figure 4) involved in nervous control of energy balance: arcuate nucleus (ARC), ventromedian (VMH), lateral (LH), dorsomedian hypothalamus (DMH) and paraventricular nucleus (PVN). Studied genes were CPT1 (carnitine palmitoyl transferase 1), FAS (fatty acid synthase), GPR41 (G protein related receptor 41, GPR40 and 43 have been also tested but were not detected in our models), ACCβ (Acetyl carboxylase β) and AMPKα2. Results are displayed in figure 5.

Fig. 4. Map of hypothalamic areas involved in nervous control of energy homeostasis. Arcuate nucleus (ARC), ventromedian (VMH), lateral (LH), dorsomedian hypothalamus

(DMH) and paraventricular nucleus (PVN).

etomoxir, a specific inhibitor of CPT1.\*\*\* p<0.01 vs controls

Hypothalamic gene expression were also modified in some areas depending on FA (figure 5). For example, FAS expression was inhibited in all nuclei except in VMH of linolenate infused rats. GPR41 was up-regulated in ARC of linolenate infused rats.

Fig. 5. mRNA expressionof target gene in different hypothalamic areas in oleate or linolenate +/- etomoxir. The green line represents gene expression in control rats (infused with NaCl). \*\*, p <0.01 vs controls. §§, p <0.01 vs controls. ARC: arcuate nucleus; VMH: ventromedian hypothalamus; LH: lateral hypothalamus; DMH: dorsomedian hypothalamus; PVN: paraventricular nucleus.

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 407

\*\*

Fig. 7. Time course of glycemia (left) and area under the curve of glycemia during oral

Interestingly, the effect of lard oil was associated with an accumulation of ceramide in the hypothalamus (figure 8). Therefore, our data suggest that ceramide accumulation in the hypothalamus following icv infusion of saturated fatty acid could contribute to the installation of an insulin resistant state by altering nervous output and consequently

\*\*\* \*\*\*

Fig. 8. Hypothalamic content of diacylglycerol (DAG) and ceramides in rats infused through carotid artery with NaCl (controls), Ivelip, Omegaven or Lard oil during 24h. \*\*\*p<0.001 vs

It is now clearly evidenced that hypothalamic FA sensing is an important regulator of nervous control of energy balance. In the present study we highlighted the differential

glucose tolerance test (right). \*\*p<0.01, \*\*\* p<0.001 vs controls.

nervous control of insulin secretion and action.

controls.

**4. Discussion** 

In the second serie of experiments, we first measured food intake (figure 6). As depicted, there was a decreased in food intake with omegaven and ivelip infusion but not with lard oil.

Fig. 6. Measurement of food intake. \*, p<0.05 vs controls, \*\*p<0.01 vs controls.

Figure 7 depicted time course and area under the curse of glycemia in response to oral glucose tolerance test. Lard oil induced glucose intolerance compared to controls.

In the second serie of experiments, we first measured food intake (figure 6). As depicted, there was a decreased in food intake with omegaven and ivelip infusion but not with lard oil.

\*

Fig. 6. Measurement of food intake. \*, p<0.05 vs controls, \*\*p<0.01 vs controls.

\*\*\* \*\* \*\* \*\*\*

glucose tolerance test. Lard oil induced glucose intolerance compared to controls.

Figure 7 depicted time course and area under the curse of glycemia in response to oral

\*\*

\*\*

\*\*\*

Fig. 7. Time course of glycemia (left) and area under the curve of glycemia during oral glucose tolerance test (right). \*\*p<0.01, \*\*\* p<0.001 vs controls.

Interestingly, the effect of lard oil was associated with an accumulation of ceramide in the hypothalamus (figure 8). Therefore, our data suggest that ceramide accumulation in the hypothalamus following icv infusion of saturated fatty acid could contribute to the installation of an insulin resistant state by altering nervous output and consequently nervous control of insulin secretion and action.

Fig. 8. Hypothalamic content of diacylglycerol (DAG) and ceramides in rats infused through carotid artery with NaCl (controls), Ivelip, Omegaven or Lard oil during 24h. \*\*\*p<0.001 vs controls.

## **4. Discussion**

It is now clearly evidenced that hypothalamic FA sensing is an important regulator of nervous control of energy balance. In the present study we highlighted the differential

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 409

Indeed 24h of lard oil infusion in carotid which had no effect on plasma TG or FA concentrations (data not shown) induced a glucose intolerance suggesting a deregulation of insulin sensitivity and or secretion. This deleterious effect of lard oil in nervous control of glucose homeostasis was associated with an increased in DAG and ceramides content in hypothalamus. An important role for ceramides has emerged from research on the pathogenesis of metabolic diseases associated with obesity, such as diabetes (Holland & Summers, 2008). Indeed, ceramides appear to be particularly deleterious components of the lipid milieu that accrues in obesity, and levels of ceramides are often elevated in skeletal muscle, liver, and/or serum of obese humans and rodents (Adams et al., 2004; Clement et al., 2002). DAG and ceramides are known to activate kinase such as PKC, which phosphorylate insulin receptor substrate and Akt leading to an inhibition of the insulin signaling (Mullen et al., 2009; Newton et al., 2009). A recent study also evidenced that sphingolipids such as ceramide might be key components of the signaling networks that link lipid-induced inflammatory pathways to the antagonism of insulin action that contributes to diabetes (Holland et al., 2011). We also recently demonstrated that the atypical protein kinase C, PKCΘ, is expressed in discrete neuronal populations of the ARC and the dorsal medial hypothalamic nucleus (Benoit et al., 2009). CNS exposure to saturated palmitic acid via direct infusion or by oral gavage increased the localization of PKCΘ to hypothalamic cell membranes in association impaired hypothalamic insulin and leptin signaling (Benoit et al., 2009). This finding was specific for palmitic acid, as the monounsaturated FA, OA, neither increased membrane localization of PKCΘ nor reduced insulin signaling. Finally, ARC-specific knockdown of PKCΘ attenuated diet-induced obesity and improved hypothalamic insulin signaling (Benoit et al., 2009). These results suggest that many of the deleterious effects of high fat diets, specifically those enriched with palmitic acid, are CNS mediated via PKCΘ activation, resulting in reduced insulin activity. Therefore, our data suggest that ceramide accumulation in the hypothalamus following icv infusion of saturated fatty acid could contribute to the installation of an insulin resistant state by altering nervous output and consequently nervous control of insulin secretion and

Further studies are needed to clearly identify molecular mechanism relaying ceramides production. However there is now several experiments highlighting some of these

In FA sensitive neurons, exposure to long chain FA can alter the activity of a wide variety of ion channels including Cl-, GABAA (Tewari et al., 2000), potassium, K+-Ca2+ (Honen et al., 2003) or calcium channels (Oishi et al., 1990). Additionally, FA inhibit the Na+-K+ ATPase pump (Oishi et al., 1990). For example, OA activates ARC POMC neurons by inhibiting ATP-sensitive K+ (KATP) channel activity (Jo et al., 2009) and the effect of OA on HGP is abolished by icv administration of a KATP channel inhibitor (Jo et al., 2009). However, KATP channels are ubiquitously expressed on neurons throughout the brain, not only in FA sensing neurons, making the mechanism and site of such in vivo manipulations difficult to discern (Dunn-Meynell et al., 1998). Using in vivo and in vitro electrophysiological approaches, OA sensitive-neurons have been characterized using whole cell patch clamp

mechanisms in FA sensitive neurons as described below.

**4.1 Molecular mechanisms involved in neuronal FA sensing** 

action.

effects of FA regarding their chain length and degree of saturation. We firstly evidenced here that oleate and linolenate have differential effects in regard to glucose homeostasis and GIIS. Especially linolenate induced increased GIIS compared to both oleate and control group whereas time course of glycemia remained similar. Thus there is a difference between effect of monounsaturated and polyunsaturated fatty acids when infused toward the brain. This suggest activation of different pathways. It must be pointed out that we previously demonstrated that short term infusion of oleate (6h and 24h) induced an increased in insulin secretion induced by glucose compared to control rats (Migrenne et al., 2006; Wang et al., 2006). In the present study infusion was made during 3 days which can explain different effect in short vs long term infusion peridod. Indeed we cannot exclude an adaptation to oleate effect when infusion stay longer. In the same way of ideas, inhibitory effect of central infusion of oleate on food intake was also lost after 3 days of infusion as previously evidenced by obici et al (Obici et al., 2002). In contrast, in the present study effect of linolenate was still present after 3 days of infusion. Linolenate effect may induce an insulinresistance state and increased GIIS could be an adaptation to this insulin resistance state. By acting on FA sensitive neurons, linolenate may affect nervous output from CNS, especially autonomic nervous system. This change in autonomic nervous system balance will in turn modify nervous control of insulin secretion and action. We previously demonstrated that lipid infusion induced changes in both sympathetic and parasympathetic nervous activity in both rodents (Magnan et al., 1999) and humans (Magnan et al., 2001). In both studies decreased sympathetic nervous activity induced an increased insulin secretion in response to glucose and insulin resistance. In addition, in the present study we showed that linolenate effect involved its metabolism since it had no more effect in presence of etomoxir an inhibitor of β oxidation. The involvement of β oxidation to relay FA effect on sensitive neurons have been also described in different models by us (Cruciani-Guglielmacci et al., 2004) and others (Obici et al., 2003). Finally that specific effects of linolenate compared to oleate could be, at least in part, related to differential gene transcription involved in FA metabolism such as CPT1, FAS or ACC in key areas of hypothalamus. More precisely in ARC and LH CPT1 expression was increased in linolenate infused rats and etomoxir induced a decreased in this gene and its return to basal value. However, it is difficult to further analyze these results since in other nuclei, there is no change of CPT1 expression. In addition, in some area others genes are differently expressed such as AMPKα2 or ACCβ, both key enzymes of glucose and FA metabolism. Altogether these data suggest that oleate or linolenate may act on different subpopulations of neurons (or astrocytes) thus highlighting the fact that FAs may have different effect in regard of the area in which they act. It is also interesting that expression of gene such as GPR41 can be also modified by linolenate infusion. Indeed it has been recently evidenced that short-chain fatty acids and ketones directly regulate sympathetic nervous system via GPR41 at the level of sympathetic ganglion (Kimura et al., 2011). Thus changes in hypothalamic GRP41 gene expression may have an impact during starvation, a situation in which ketone bodies production is increased. thereby control body energy expenditure in maintaining metabolic homeostasis.

In the second part of our work we demonstrated a differential role of PUFA vs saturated FA (SFA) regarding induction of insulin resistance and ceramides production in hypothalamus by using triglyceride emulsion infusion, in order to mimic a more "physiological approach".

effects of FA regarding their chain length and degree of saturation. We firstly evidenced here that oleate and linolenate have differential effects in regard to glucose homeostasis and GIIS. Especially linolenate induced increased GIIS compared to both oleate and control group whereas time course of glycemia remained similar. Thus there is a difference between effect of monounsaturated and polyunsaturated fatty acids when infused toward the brain. This suggest activation of different pathways. It must be pointed out that we previously demonstrated that short term infusion of oleate (6h and 24h) induced an increased in insulin secretion induced by glucose compared to control rats (Migrenne et al., 2006; Wang et al., 2006). In the present study infusion was made during 3 days which can explain different effect in short vs long term infusion peridod. Indeed we cannot exclude an adaptation to oleate effect when infusion stay longer. In the same way of ideas, inhibitory effect of central infusion of oleate on food intake was also lost after 3 days of infusion as previously evidenced by obici et al (Obici et al., 2002). In contrast, in the present study effect of linolenate was still present after 3 days of infusion. Linolenate effect may induce an insulinresistance state and increased GIIS could be an adaptation to this insulin resistance state. By acting on FA sensitive neurons, linolenate may affect nervous output from CNS, especially autonomic nervous system. This change in autonomic nervous system balance will in turn modify nervous control of insulin secretion and action. We previously demonstrated that lipid infusion induced changes in both sympathetic and parasympathetic nervous activity in both rodents (Magnan et al., 1999) and humans (Magnan et al., 2001). In both studies decreased sympathetic nervous activity induced an increased insulin secretion in response to glucose and insulin resistance. In addition, in the present study we showed that linolenate effect involved its metabolism since it had no more effect in presence of etomoxir an inhibitor of β oxidation. The involvement of β oxidation to relay FA effect on sensitive neurons have been also described in different models by us (Cruciani-Guglielmacci et al., 2004) and others (Obici et al., 2003). Finally that specific effects of linolenate compared to oleate could be, at least in part, related to differential gene transcription involved in FA metabolism such as CPT1, FAS or ACC in key areas of hypothalamus. More precisely in ARC and LH CPT1 expression was increased in linolenate infused rats and etomoxir induced a decreased in this gene and its return to basal value. However, it is difficult to further analyze these results since in other nuclei, there is no change of CPT1 expression. In addition, in some area others genes are differently expressed such as AMPKα2 or ACCβ, both key enzymes of glucose and FA metabolism. Altogether these data suggest that oleate or linolenate may act on different subpopulations of neurons (or astrocytes) thus highlighting the fact that FAs may have different effect in regard of the area in which they act. It is also interesting that expression of gene such as GPR41 can be also modified by linolenate infusion. Indeed it has been recently evidenced that short-chain fatty acids and ketones directly regulate sympathetic nervous system via GPR41 at the level of sympathetic ganglion (Kimura et al., 2011). Thus changes in hypothalamic GRP41 gene expression may have an impact during starvation, a situation in which ketone bodies production is increased. thereby control body energy expenditure in maintaining metabolic homeostasis. In the second part of our work we demonstrated a differential role of PUFA vs saturated FA (SFA) regarding induction of insulin resistance and ceramides production in hypothalamus by using triglyceride emulsion infusion, in order to mimic a more "physiological approach". Indeed 24h of lard oil infusion in carotid which had no effect on plasma TG or FA concentrations (data not shown) induced a glucose intolerance suggesting a deregulation of insulin sensitivity and or secretion. This deleterious effect of lard oil in nervous control of glucose homeostasis was associated with an increased in DAG and ceramides content in hypothalamus. An important role for ceramides has emerged from research on the pathogenesis of metabolic diseases associated with obesity, such as diabetes (Holland & Summers, 2008). Indeed, ceramides appear to be particularly deleterious components of the lipid milieu that accrues in obesity, and levels of ceramides are often elevated in skeletal muscle, liver, and/or serum of obese humans and rodents (Adams et al., 2004; Clement et al., 2002). DAG and ceramides are known to activate kinase such as PKC, which phosphorylate insulin receptor substrate and Akt leading to an inhibition of the insulin signaling (Mullen et al., 2009; Newton et al., 2009). A recent study also evidenced that sphingolipids such as ceramide might be key components of the signaling networks that link lipid-induced inflammatory pathways to the antagonism of insulin action that contributes to diabetes (Holland et al., 2011). We also recently demonstrated that the atypical protein kinase C, PKCΘ, is expressed in discrete neuronal populations of the ARC and the dorsal medial hypothalamic nucleus (Benoit et al., 2009). CNS exposure to saturated palmitic acid via direct infusion or by oral gavage increased the localization of PKCΘ to hypothalamic cell membranes in association impaired hypothalamic insulin and leptin signaling (Benoit et al., 2009). This finding was specific for palmitic acid, as the monounsaturated FA, OA, neither increased membrane localization of PKCΘ nor reduced insulin signaling. Finally, ARC-specific knockdown of PKCΘ attenuated diet-induced obesity and improved hypothalamic insulin signaling (Benoit et al., 2009). These results suggest that many of the deleterious effects of high fat diets, specifically those enriched with palmitic acid, are CNS mediated via PKCΘ activation, resulting in reduced insulin activity. Therefore, our data suggest that ceramide accumulation in the hypothalamus following icv infusion of saturated fatty acid could contribute to the installation of an insulin resistant state by altering nervous output and consequently nervous control of insulin secretion and action.

Further studies are needed to clearly identify molecular mechanism relaying ceramides production. However there is now several experiments highlighting some of these mechanisms in FA sensitive neurons as described below.

#### **4.1 Molecular mechanisms involved in neuronal FA sensing**

In FA sensitive neurons, exposure to long chain FA can alter the activity of a wide variety of ion channels including Cl-, GABAA (Tewari et al., 2000), potassium, K+-Ca2+ (Honen et al., 2003) or calcium channels (Oishi et al., 1990). Additionally, FA inhibit the Na+-K+ ATPase pump (Oishi et al., 1990). For example, OA activates ARC POMC neurons by inhibiting ATP-sensitive K+ (KATP) channel activity (Jo et al., 2009) and the effect of OA on HGP is abolished by icv administration of a KATP channel inhibitor (Jo et al., 2009). However, KATP channels are ubiquitously expressed on neurons throughout the brain, not only in FA sensing neurons, making the mechanism and site of such in vivo manipulations difficult to discern (Dunn-Meynell et al., 1998). Using in vivo and in vitro electrophysiological approaches, OA sensitive-neurons have been characterized using whole cell patch clamp

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 411

modulation of AMP-activated kinase, a known energy-sensing kinase (Ronnett et al., 2005). Indeed, icv administration of 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR), a 5'-AMP kinase activator, rapidly lowers hypothalamic malonyl-CoA concentration and increases food intake (Tu et al., 2005). These effects correlate closely with the phosphorylation-induced inactivation of ACC, an established target of AMP kinase. Collectively, these data suggest a role for FA metabolism in the perception and regulation of energy balance. However, it must be also pointed out that C75 and AICAR may also have nonspecific or even opposite effects. For example, a major effect of C75 is to activate CPT-1 rather than lead to its inhibition *in vitro* (Aja et al., 2008). Finally the route of administration and the type of FA used are also critical. For example, bolus intracerebroventricular injections of OA, but not palmitic acid, reduce food intake and body weight, possibly mediated through POMC/MC4R signaling (Schwinkendorf et al., 2010). Again, such bolus icv injections could cause non-specific effects related to inflammation of ependymocytes and tanycytes. Also because so much of FA metabolism takes place in astrocytes, such manipulations done in vivo and in slice preparations are likely to alter FA metabolism that takes place in astrocytes which

While intracellular FA metabolism may be responsible for altering neuronal activity in some FA sensitive neurons such as ARC POMC neurons (Jo et al., 2009) it accounts for a relatively small percent of the effects of OA on dissociated VMN neurons (Le Foll et al., 2009). In those neurons, inhibition of CPT1, reactive oxygen species formation, long-chain acyl CoA synthetase and KATP channel activity or activation of uncoupling protein 2 (UCP2) accounts for no more than 20% of the excitatory or approximately 40% of the inhibitory effects of OA (Le Foll et al., 2009). On the other hand, pharmacological inhibition of FAT/CD36, a FA transporter/receptor that can alter cell function independently of intracellular FA metabolism reduced the excitatory and inhibitory effects of OA by up to 45% (Le Foll et al., 2009). Thus, in almost half of VMN FA sensing neurons, CD36 may act primarily as receptor, rather than a transporter, for long chain FA as it does on taste cells on the tongue where it activates store-operated calcium channels to alter membrane potential and release of serotonin (Gaillard et al., 2008). These effects all occur in the presence of nanomolar concentrations of OA, whereas micromolar concentrations are generally required to effect similar changes in neuronal activity in brain slice preparations (Jo et al., 2009; Migrenne et al., 2011; Wang et al., 2006). Thus, in the absence of astrocytes, OA can directly affect VMN neuronal activity through both metabolic and non-metabolic pathways. Alternatively, FA might act as signaling molecules by covalent attachment to proteins (N-terminal acylation) to alter the function of membrane and intracellular signaling molecules. For example, palmitoylation facilitates the targeting and plasma membrane binding of proteins which otherwise would remain in the cytosolic compartment (Resh, 1999). Some membrane proteins (TGF, synaptosomal associated protein of 25KDa (required for exocytosis) and plasma membrane receptors (seven transmembrane receptors such as 2a- and 2 adrenoceptors) are typically palmitoylated on one or several cysteine residues located adjacent to or just within the transmembrane domain (Resh, 1999) Such mechanisms might

could then indirectly alter neuronal FA sensing (Escartin et al., 2007).

**4.3 Non metabolic-dependent neuronal FA sensing** 

also modulate neuronal FA sensing.

records in ARC slices from 14 to 21 day old rats (Wang et al., 2006). Of these 13 % were excited by OA and 30% were inhibited by OA (Oomura et al., 1975). The excitatory effects of OA appeared to be due to closure of chloride channels leading to membrane depolarization and increased action potential frequency (Migrenne et al., 2006). On the other hand, inhibitory effect of OA may involve the KATP channels since this inhibition was reversed by the KATP channel blocker tolbutamide (Migrenne et al., 2006). Using fura-2 Ca2+ imaging in dissociated neurons from the ventromedial hypothalamic nucleus (VMN) neurons, we found that OA excited up to 43% and inhibited up to 29% of all VMN neurons independently of glucose concentrations (Le Foll et al., 2009). However, in these neurons, inhibition of the KATP channel mediated FA sensing in only a small percentage of FA sensing neurons. Importantly, although a relatively large percentage of hypothalamic neurons are FA sensors, a select population that also sense glucose are highly dependent upon ambient glucose concentrations for the resultant effect of FA on the activity of these neurons (Le Foll et al., 2009). Such data suggest that the responses of hypothalamic FA sensitive neurons are dependent upon the metabolic state of the animal and thus might be expected to respond differently during fasting (when FA levels rise and glucose levels fall) vs. the overfed state when glucose levels rise while free FA levels remain relatively unchanged (Le Foll et al., 2009). However, it must be pointed out that FA are naturally complexed to serum albumin in the blood and the concentration of circulating free FA is less than 1% of total FA levels. All the studies investigating FA sensing in the hypothalamus either use non-complexed FA or cyclodextrin-complexed FA *in vitro* or *in vivo*. The concentration of free FA in cyclodextrin-complexed FA preparation is unknown. Whether or not the FA concentration used mimics FA levels in physiological states needs to be determined.

#### **4.2 Metabolic-dependent FA sensing effects**

The effects of FA on activity of some neurons are dependent upon intracellular metabolism of FA. Enzymes involved in FA metabolism such as FA synthase (FAS), CPT1 and acetyl-CoA carboxylase (ACC) are expressed in some hypothalamic neurons as well as in glial cells (reviewed in (Blouet & Schwartz; Le Foll et al., 2009). Malonyl-CoA may be an important sensor of energy levels in the hypothalamus. It is derived from either glucose or FA metabolism via the glycolysis or -oxidation, respectively. The steady-state level of malonyl-CoA is determined by its rate of synthesis catalysed by ACC relative to its rate of turnover catalysed by FAS. The synthesis of malonyl-CoA is the first committed step of FA synthesis and ACC is the major site of regulation in that process. Thus, when the supply of glucose is increased, malonyl CoA levels increase in keeping with a decreased need for FA oxidation. This increase in both malonyl CoA and acyl CoA levels is associated with reduced food intake. Central administration of C75, an inhibitor of FAS, also increases malonyl-CoA concentration in the hypothalamus, suppresses food intake and leads to profound weight loss (Proulx & Seeley, 2005). It has been proposed that centrally, C75 and cerulenin (another inhibitor of FAS) alter the expression profiles of feeding-related neuropeptides, often inhibiting the expression of orexigenic peptides such as neuropeptide Y (Proulx et al., 2008). Whether through centrally mediated or peripheral mechanisms, C75 also increases energy expenditure, which contributes to weight loss (Clegg et al., 2002; Tu et al., 2005). In vitro and in vivo studies demonstrate that at least part of C75's effects are mediated by the

records in ARC slices from 14 to 21 day old rats (Wang et al., 2006). Of these 13 % were excited by OA and 30% were inhibited by OA (Oomura et al., 1975). The excitatory effects of OA appeared to be due to closure of chloride channels leading to membrane depolarization and increased action potential frequency (Migrenne et al., 2006). On the other hand, inhibitory effect of OA may involve the KATP channels since this inhibition was reversed by the KATP channel blocker tolbutamide (Migrenne et al., 2006). Using fura-2 Ca2+ imaging in dissociated neurons from the ventromedial hypothalamic nucleus (VMN) neurons, we found that OA excited up to 43% and inhibited up to 29% of all VMN neurons independently of glucose concentrations (Le Foll et al., 2009). However, in these neurons, inhibition of the KATP channel mediated FA sensing in only a small percentage of FA sensing neurons. Importantly, although a relatively large percentage of hypothalamic neurons are FA sensors, a select population that also sense glucose are highly dependent upon ambient glucose concentrations for the resultant effect of FA on the activity of these neurons (Le Foll et al., 2009). Such data suggest that the responses of hypothalamic FA sensitive neurons are dependent upon the metabolic state of the animal and thus might be expected to respond differently during fasting (when FA levels rise and glucose levels fall) vs. the overfed state when glucose levels rise while free FA levels remain relatively unchanged (Le Foll et al., 2009). However, it must be pointed out that FA are naturally complexed to serum albumin in the blood and the concentration of circulating free FA is less than 1% of total FA levels. All the studies investigating FA sensing in the hypothalamus either use non-complexed FA or cyclodextrin-complexed FA *in vitro* or *in vivo*. The concentration of free FA in cyclodextrin-complexed FA preparation is unknown. Whether or not the FA concentration

used mimics FA levels in physiological states needs to be determined.

The effects of FA on activity of some neurons are dependent upon intracellular metabolism of FA. Enzymes involved in FA metabolism such as FA synthase (FAS), CPT1 and acetyl-CoA carboxylase (ACC) are expressed in some hypothalamic neurons as well as in glial cells (reviewed in (Blouet & Schwartz; Le Foll et al., 2009). Malonyl-CoA may be an important sensor of energy levels in the hypothalamus. It is derived from either glucose or FA metabolism via the glycolysis or -oxidation, respectively. The steady-state level of malonyl-CoA is determined by its rate of synthesis catalysed by ACC relative to its rate of turnover catalysed by FAS. The synthesis of malonyl-CoA is the first committed step of FA synthesis and ACC is the major site of regulation in that process. Thus, when the supply of glucose is increased, malonyl CoA levels increase in keeping with a decreased need for FA oxidation. This increase in both malonyl CoA and acyl CoA levels is associated with reduced food intake. Central administration of C75, an inhibitor of FAS, also increases malonyl-CoA concentration in the hypothalamus, suppresses food intake and leads to profound weight loss (Proulx & Seeley, 2005). It has been proposed that centrally, C75 and cerulenin (another inhibitor of FAS) alter the expression profiles of feeding-related neuropeptides, often inhibiting the expression of orexigenic peptides such as neuropeptide Y (Proulx et al., 2008). Whether through centrally mediated or peripheral mechanisms, C75 also increases energy expenditure, which contributes to weight loss (Clegg et al., 2002; Tu et al., 2005). In vitro and in vivo studies demonstrate that at least part of C75's effects are mediated by the

**4.2 Metabolic-dependent FA sensing effects** 

modulation of AMP-activated kinase, a known energy-sensing kinase (Ronnett et al., 2005). Indeed, icv administration of 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR), a 5'-AMP kinase activator, rapidly lowers hypothalamic malonyl-CoA concentration and increases food intake (Tu et al., 2005). These effects correlate closely with the phosphorylation-induced inactivation of ACC, an established target of AMP kinase. Collectively, these data suggest a role for FA metabolism in the perception and regulation of energy balance. However, it must be also pointed out that C75 and AICAR may also have nonspecific or even opposite effects. For example, a major effect of C75 is to activate CPT-1 rather than lead to its inhibition *in vitro* (Aja et al., 2008). Finally the route of administration and the type of FA used are also critical. For example, bolus intracerebroventricular injections of OA, but not palmitic acid, reduce food intake and body weight, possibly mediated through POMC/MC4R signaling (Schwinkendorf et al., 2010). Again, such bolus icv injections could cause non-specific effects related to inflammation of ependymocytes and tanycytes. Also because so much of FA metabolism takes place in astrocytes, such manipulations done in vivo and in slice preparations are likely to alter FA metabolism that takes place in astrocytes which could then indirectly alter neuronal FA sensing (Escartin et al., 2007).

#### **4.3 Non metabolic-dependent neuronal FA sensing**

While intracellular FA metabolism may be responsible for altering neuronal activity in some FA sensitive neurons such as ARC POMC neurons (Jo et al., 2009) it accounts for a relatively small percent of the effects of OA on dissociated VMN neurons (Le Foll et al., 2009). In those neurons, inhibition of CPT1, reactive oxygen species formation, long-chain acyl CoA synthetase and KATP channel activity or activation of uncoupling protein 2 (UCP2) accounts for no more than 20% of the excitatory or approximately 40% of the inhibitory effects of OA (Le Foll et al., 2009). On the other hand, pharmacological inhibition of FAT/CD36, a FA transporter/receptor that can alter cell function independently of intracellular FA metabolism reduced the excitatory and inhibitory effects of OA by up to 45% (Le Foll et al., 2009). Thus, in almost half of VMN FA sensing neurons, CD36 may act primarily as receptor, rather than a transporter, for long chain FA as it does on taste cells on the tongue where it activates store-operated calcium channels to alter membrane potential and release of serotonin (Gaillard et al., 2008). These effects all occur in the presence of nanomolar concentrations of OA, whereas micromolar concentrations are generally required to effect similar changes in neuronal activity in brain slice preparations (Jo et al., 2009; Migrenne et al., 2011; Wang et al., 2006). Thus, in the absence of astrocytes, OA can directly affect VMN neuronal activity through both metabolic and non-metabolic pathways. Alternatively, FA might act as signaling molecules by covalent attachment to proteins (N-terminal acylation) to alter the function of membrane and intracellular signaling molecules. For example, palmitoylation facilitates the targeting and plasma membrane binding of proteins which otherwise would remain in the cytosolic compartment (Resh, 1999). Some membrane proteins (TGF, synaptosomal associated protein of 25KDa (required for exocytosis) and plasma membrane receptors (seven transmembrane receptors such as 2a- and 2 adrenoceptors) are typically palmitoylated on one or several cysteine residues located adjacent to or just within the transmembrane domain (Resh, 1999) Such mechanisms might also modulate neuronal FA sensing.

Differential Effect of Fatty Acids in Nervous Control of Energy Balance 413

Thus, these neurons are ideally suited to respond differentially under a variety of metabolic conditions such as fasting, feeding, hypo- or hyperglycemia. However, while it is clear that specific neurons can respond to changes in ambient FA levels, many questions remain. We still do not know for certain how FA are transported into the brain, astrocytes or neurons and whether those FA that are transported are derived from circulating free FA or triglycerides. Since most studies suggest that rising FA levels reduce food intake, then we must explain why plasma FA levels are most elevated during fasting when the drive to seek and ingest food should be at its strongest. Another major issue relates to the interaction between astrocytes and neurons with regard to the metabolism and signaling of FA. Also, we still know little about the basic mechanisms utilized by neurons to sense FA, where such FA sensitive neurons reside throughout the brain and what neurotransmitters and peptides

Finally, it has been postulated that diabetes may be a disorder of the brain (Elmquist & Marcus, 2003). If so, dysfunction of these FA sensitive neurons could be, at least in part, one of the early mechanisms underlying impairment of neural control of energy and glucose homeostasis and the development of obesity and type 2 diabetes in predisposed subjects. A better understanding of this central nutrient sensing, including both FA and glucose, could provide clues for the identification of new therapeutic targets for the prevention and

This work was partially supported by an award from European Foundation for Study of

Adams JM, 2nd; Pratipanawatr T; Berria R; Wang E; DeFronzo RA; Sullards MC &

Aid S; Vancassel S; Linard A; Lavialle M & Guesnet P. (2005). Dietary docosahexaenoic acid

Aja S; Landree LE; Kleman AM; Medghalchi SM; Vadlamudi A; McFadden JM; Aplasca A;

Benoit SC; Kemp CJ; Elias CF; Abplanalp W; Herman JP; Migrenne S; Lefevre AL;

Mandarino LJ. (2004). Ceramide content is increased in skeletal muscle from obese insulin-resistant humans. *Diabetes*, Vol. 53, No. 1, pp 25-31, 0012-1797 (Print) 0012-

[22: 6(n-3)] as a phospholipid or a triglyceride enhances the potassium chlorideevoked release of acetylcholine in rat hippocampus. *J Nutr*, Vol. 135, No. 5, pp 1008-

Hyun J; Plummer E; Daniels K; Kemm M; Townsend CA; Thupari JN; Kuhajda FP; Moran TH & Ronnett GV. (2008). Pharmacological stimulation of brain carnitine palmitoyl-transferase-1 decreases food intake and body weight. *Am J Physiol Regul Integr Comp Physiol*, Vol. 294, No. 2, pp R352-361, 0363-6119 (Print) 0363-6119

Cruciani-Guglielmacci C; Magnan C; Yu F; Niswender K; Irani BG; Holland WL & Clegg DJ. (2009). Palmitic acid mediates hypothalamic insulin resistance by

they release when responding to FA.

treatment of both diabetes and obesity.

Diabetes (EFSD)/GSK 2007 (Stéphanie Migrenne).

**6. Acknowledgements** 

1797 (Linking)

1013,

(Linking)

**7. References** 

#### **4.4 Which neurotransmitters or neuropeptides?**

The ultimate consequence of the activation or inactivation of a neuron is the release of neurotransmitters and neuropeptides. Since FA decrease food intake, they might be expected to alter activity neurons specifically involved in the regulation of feeding. In fact, OA activates catabolic POMC neurons directly, apparently via ß-oxidation and inactivation of the KATP channel in hypothalamic slice preparations (Jo et al., 2009). In vivo, Obici et al. (Obici et al., 2003) reported that icv administration of OA markedly inhibits glucose production and food intake, accompanied by a decrease in the hypothalamic expression of the anabolic peptide, neuropeptide Y. This decrease in the expression of such a critical anabolic peptide might contribute to the reduced food intake associated with direct central administration of OA. On the other hand, an n-3 FA enriched diet increases food intake in anorexic tumor-bearing rats, in association with reduced tumor appearance, tumor growth and onset of anorexia (Ramos et al., 2005). In these treated rats, neuropeptide Y immunoreactivity increased 38% in ARC and 50% in paraventricular nucleus, whereas αmelanocyte stimulating hormone (a catabolic peptide cleavage product of POMC) decreased 64% in the ARC and 29% in the paraventricular nucleus (Ramos et al., 2005). Finally, in the hippocampus, docosahexaenoic acid (22:6(n-3) increased the spontaneous release of acetylcholine (Aid et al., 2005).

#### **4.5 Pathological implications of excess FA**

Besides physiological regulation of energy balance by hypothalamic neuronal FA sensing, impaired regulation of such sensing might contribute to the development of metabolic diseases such as obesity and type 2 diabetes in predisposed subjects exposed to a chronic lipid overload (Luquet & Magnan, 2009; Migrenne et al., 2011). Excessive brain lipid levels may indeed alter control of glucose and lipid homeostasis through changes of autonomic nervous system activity. Increasing brain FA levels reduces sympathetic activity and increases GIIS in rats (Clement et al., 2002; Obici et al., 2003) a condition which would exacerbate the development of type 2 diabetes mellitus. Also, a lipid overload due to highfat diet intake alters both hypothalamic monoamine turnover (Levin et al., 1983) and peripheral sympathetic activity in rats (Young & Walgren, 1994). In humans, overweight is often associated with an altered sympathetic tone (Peterson et al., 1988) suggesting a relationship between lipids and autonomic control centers in brain.

### **5. Conclusion**

In conclusion, there is now increasing evidence that specialized neurons within hypothalamus and other areas such as the brainstem or hippocampus can detect changes in plasma FA levels by having FA directly or indirectly alter the of FA sensitive neurons involved in the regulation of energy and glucose homeostasis. Central FA effects on insulin secretion and action are related to their chain length or degree of saturation. Such effects are also mediated through differential changes in gene expression.

The neuronal networks of these FA sensitive neurons that sense and respond to FA are likely very complex given the fact that FA can either inhibit or excite specific neurons. In addition, many of these neurons also utilize glucose as a signaling molecule and there is often an inverse responsiveness of such "metabolic sensing" neurons to FA vs. glucose. Thus, these neurons are ideally suited to respond differentially under a variety of metabolic conditions such as fasting, feeding, hypo- or hyperglycemia. However, while it is clear that specific neurons can respond to changes in ambient FA levels, many questions remain. We still do not know for certain how FA are transported into the brain, astrocytes or neurons and whether those FA that are transported are derived from circulating free FA or triglycerides. Since most studies suggest that rising FA levels reduce food intake, then we must explain why plasma FA levels are most elevated during fasting when the drive to seek and ingest food should be at its strongest. Another major issue relates to the interaction between astrocytes and neurons with regard to the metabolism and signaling of FA. Also, we still know little about the basic mechanisms utilized by neurons to sense FA, where such FA sensitive neurons reside throughout the brain and what neurotransmitters and peptides they release when responding to FA.

Finally, it has been postulated that diabetes may be a disorder of the brain (Elmquist & Marcus, 2003). If so, dysfunction of these FA sensitive neurons could be, at least in part, one of the early mechanisms underlying impairment of neural control of energy and glucose homeostasis and the development of obesity and type 2 diabetes in predisposed subjects. A better understanding of this central nutrient sensing, including both FA and glucose, could provide clues for the identification of new therapeutic targets for the prevention and treatment of both diabetes and obesity.
