Journal of Neuroimmunology 209 (2009) 96–103

Contents lists available at ScienceDirect

Journal of Neuroimmunology

j ourna l homepage: www.e lsev ie r.com/ locate / jneuro im

Lipopolysaccharide (LPS) stimulates adipokine and socs3 gene expression in mousebrain and pituitary gland in vivo, and in N-1 hypothalamic neurons in vitro

Russell Brown a,b,⁎, Syed A. Imran a,c, Michael Wilkinson a,b,c

a Department of Obstetrics and Gynaecology, Faculty of Medicine, Dalhousie University, c/o IWK Health Centre, 5850/5980 University Avenue, PO Box 9700, Halifax,Nova Scotia, Canada B3K 6R8b Department of Physiology and Biophysics, Faculty of Medicine, Dalhousie University, c/o IWK Health Centre, 5850/5980 University Avenue, PO Box 9700, Halifax,Nova Scotia, Canada B3K 6R8c Division of Endocrinology and Metabolism, Faculty of Medicine, Dalhousie University, c/o IWK Health Centre, 5850/5980 University Avenue, PO Box 9700, Halifax,Nova Scotia, Canada B3K 6R8

⁎ Corresponding author. Department of ObstetricsCentre, 5850/5980 University Avenue, PO Box 9700, Ha

E-mail address: Brownre@dal.ca (R. Brown).

0165-5728/$ – see front matter © 2009 Elsevier B.V. Adoi:10.1016/j.jneuroim.2009.02.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 November 2008Received in revised form 28 January 2009Accepted 2 February 2009

Keywords:ResistinSuppressor of cytokine signaling-3Fasting-induced adipose factor (FIAF)Angiopoietin-like 4 (angptl4)Realtime RT-PCRToll-like receptor (TLR4)

Adipokines that modulate metabolic and inflammatory responses, such as resistin (rstn) and fasting-inducedadipose factor (fiaf), are also expressed in mouse brain and pituitary gland. Since lipopolysaccharide (LPS)-induced endotoxinemia provokes an anorectic response via a hypothalamic-dependent mechanism wehypothesized that LPS would also modify hypothalamic adipokine expression. Challenging male CD-1 micewith LPS (5 mg/kg; s.c.) significantly reduced bodyweight (24 h) and realtime RT-PCR revealed time- andtissue-dependent increases in rstn, fiaf and suppressor of cytokine signaling-3 (socs-3) mRNA inhypothalamic, pituitary, cortical and adipose tissues. Gene expression was rapidly increased (3–6 h) in thehypothalamus and pituitary, but returned to normal within 24 h. In contrast, with the exception of rstn in fat,the expression of target genes remained elevated in cortex and visceral fat at 24 h post-injection. In order tomore specifically examine the hypothalamic response to LPS we investigated its effects directly on N-1hypothalamic neurons in vitro. LPS (25 μg/mL; 3 h) had no effect on rstn mRNA, but significantly stimulatedfiaf and socs-3 expression. Although various toll-like receptor 4 (TLR4) antagonists (parthenolide, PD098059,and SB202190) did not prevent the LPS-induced increases in fiaf and socs-3, they did partially attenuate itsstimulatory effects. We conclude that LPS treatment increases the expression of central, and possiblyneuronal, adipokine genes which may influence local tissue repair and function, but could also havedownstream consequences on the hypothalamic control of appetite and energy metabolism following aninflammatory insult.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

Microbial infections, or LPS-induced endotoxinemia, are known toreduce insulin sensitivity and stimulate inflammatory and anorecticresponses via hypothalamic-dependent mechanisms (Huang et al.,1999; Rummel et al., 2008; Ueki et al., 2004). Blockade ofhypothalamic melanocortin receptors (MCR) abrogated LPS-inducedanorexia in rats (Huang et al., 1999), and LPS failed to modify appetitein melanocortin-4 receptor (MC4R) knockout mice (Marks et al.,2001). LPS was also postulated to induce weight loss by increasingleptin secretion (Sachot et al., 2004; Sarraf et al., 1997), though LPSalso induced anorexia in leptin-deficient (ob/ob) and leptin receptor-deficient (db/db) mice (Faggioni et al., 1997). These data suggest thatadditional signals, including other adipokines, might mediate themetabolic and anorectic effects of LPS. For example the expression of

& Gynaecology, IWK Healthlifax, NS, Canada B3K 6R8.

ll rights reserved.

resistin and fasting-induced adipose factor (FIAF; also known asangiopoietin-like 4; ANGPTL4) is also induced by LPS or hypoxia(Belanger et al., 2002; Lu et al., 2002). These adipokines also appear tomodulate glucose and lipid metabolism (Kersten, 2005; Steppan et al.,2001), in addition to influencing appetite, glucose production andbodyweight regulation (Backhed et al., 2007; Muse et al., 2007; Qi etal., 2006; Tovar et al., 2005; Vazquez et al., 2008). In fact, themetabolicresponses induced by the central injection of resistin are strikinglysimilar to those reported following the administration of LPS; i.e.,reduced appetite and increased plasma glucose concentrations (Museet al., 2007; Singhal et al., 2007; Tovar et al., 2005; Vazquez et al.,2008).

Previously we provided evidence, using RT-PCR and immunohis-tochemistry, that rstn and fiaf are transcribed in various regions of thebrain and pituitary gland (Morash et al., 2002; Wiesner et al., 2004;Wilkinson et al., 2007;Wilkinson et al., 2005). In additionwe reportedthat hypothalamic resistin-immunoreactivity (ir) colocalizes withalpha-melanocyte stimulating hormone (αMSH) (Wilkinson et al.,2005), a marker of proopiomelanocortin (POMC) neurons that appear

Table 1Sequences of realtime PCR primers and dual-labeled probes.

Gene Primer sequences Ref.

rstn Sense: 5′ TTCCTTGTCCCTGAACTGCT 3′ (Brown et al., 2005)Antisense: 5′ TGCTGTCCAGTCTATCCTTG 3′Probe: 5′ CACGAATGTCCCACGAGCCACAGG 3′

fiaf Sense: 5′ ATCCTGGGACGAGATGAACTTGCT 3′ (Wiesner et al.,2006)Antisense: 5′ TCGGAGTCTTTGAAGGGTGCATCT 3′

Probe: 5′ ATGGCTGCCTGTGGTAACGCTTGTCA 3′socs-3 Sense: 5′ GCGGGCACCTTTCTTATCC 3′ (Steppan et al., 2005)

Antisense: 5′ TCCCCGACTGGGTCTTGAC 3′Probe: 5′ CTCGGACCAGCGCCACTTCTTCA 3′

cyclophilin Sense: 5′ TGTGCCAGGGTGGTGACTT 3′ (Brown et al., 2005)Antisense: 5′ TCAAATTTCTCTCCGTAGATGGACCT 3′Probe: 5′ ACACGCCATAATGGCACTGGCGG 3′

97R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

tomediate the anorectic responses induced by LPS (Huang et al., 1999;Marks et al., 2003). The immortalized N-1 hypothalamic neuronal cellline also co-expresses rstn and fiaf (Brown et al., 2007; Brown et al.,2008a), in addition to other neuropeptides implicated in centralenergy homeostasis (Belsham et al., 2004). Our use of quantitativerealtime RT-PCR, where the level of adipokine expression is inverselyrelated to its threshold cycle (CT), allowed us to show that the relativeexpression of rstn was 30, 34, 31, and 17 cycles in hypothalamus, N-1neurons, pituitary and adipose tissue respectively, and 32, 28, 31 and24 cycles for fiaf (Wilkinson et al., 2007). Compared to expressionlevels in adipose tissue, the relatively low expression of brainadipokines was not unexpected and matches that of other criticalneuropeptides and growth factors found in the central nervous system(CNS) (Wilkinson et al., 2007). For example nerve growth factor(NGF), which has a CT value of 36 in the CNS, indisputably plays acritical role in brain growth and neuronal plasticity and stronglysuggests that a low level of expression does not equate with lack offunction (Wilkinson et al., 2007). This was further confirmed in N-1neurons where the expression of rstn was specifically silenced usingRNA interference (RNAi) which led to significant changes in theexpression of fiaf and suppressor of cytokine signaling-3 (socs-3) thatwere the opposite of those achieved when cells were treated withrecombinant resistin alone (Brown et al., 2007). The effects on socs-3are particularly intriguing given that it has been implicated in thedevelopment of leptin and insulin resistance, and its increasedexpression has also been associated with the activation of inflamma-tory signaling cascades (Howard and Flier, 2006; Qin et al., 2007).

LPS has been shown to stimulate rstn gene expression in a varietyof mouse, rat and human cell line models (Kaser et al., 2003; Lu et al.,2002) which led us to hypothesize that LPS would induce adipokineexpression in the brain and pituitary, which may, in turn, lead toreductions in bodyweight and altered insulin sensitivity as assessed bysocs-3 expression. The effects of LPS were tested by injecting CD-1mice with LPS (5 mg/kg; s.c.) and gene expression was evaluatedwithin the various tissues at different time points. In addition wedetermined whether LPS could directly modify adipokine expressionin N-1 hypothalamic neurons, but also whether the pharmacologicalblockade of the various downstream components of the toll-likereceptor 4 (TLR4) signaling cascade would prevent the effect of LPS.

2. Materials and methods

2.1. Mice

Malemice (CD-1; postnatal days 30–32)were obtained fromCharlesRiver Laboratories (St. Constant, QC), housed in plastic cages, andmaintained in a 12 h light:12 h dark photoperiod, with free access toPurina Rat Chow and drinking water. Following 1 week of acclimatiza-tion groups of mice were injected with LPS (5 mg/kg; s.c.; from E. coliO26:B6; #L8274; SIGMA, Oakville, ON) or saline, and tissues (hypotha-lamus (HYP), parietal cortex (CTX), whole pituitary gland (PIT) andvisceral (epididymal) adipose tissue (FAT)) were collected from bothgroups at 3 h, 6 h, or 24 h post-injection, snap frozen on dry ice, andstored at −70 °C until RNA isolation. The experimental protocol wasreviewed and approved by the Dalhousie University Committee onLaboratory Animals.

2.2. N-1 cell culture

Immortalized murine N-1 hypothalamic neurons were maintainedin Dulbecco's Modified EagleMedium (DMEM; GIBCO) containing 10%fetal bovine serum (FBS; Invitrogen, Burlington, ON) at 37 °C in ahumid incubator supplied with 5% CO2/95% air (Belsham et al., 2004).Cells were seeded at 100,000 cells/well in Nunc 6-well plates 24 hbefore experimentation. After culturing overnight in serum-freeDMEM, cells were divided into 4 groups and pretreated for 30 min

with either vehicle or 3 different antagonists of the toll-like receptor 4(TLR4) signaling cascade: parthenolide (NFκB inhibitor; 5 μM;SIGMA), PD098059 (MEK 1 inhibitor; 50 μM; SIGMA) or SB202190(p38 MAPK inhibitor;10 μM; SIGMA). The various groups were thenfurther subdivided and treated for 3 h with or without LPS (25 μg/mL;in serum free DMEM) prior to RNA isolation.

2.3. Realtime RT-PCR

Total RNA was isolated from FAT using TRIzol® (Invitrogen), andfrom all remaining tissues and N-1 cells using GenElute MammalianTotal RNA isolation Kit (Sigma-Aldrich, Oakville, ON), and treatedwithDNAse, according to the manufacturer's protocol. Total RNA (1.5 μg)was reverse transcribed (RT) and PCR amplified using the Super-Script™ III Platinum® Two-Step qRT-PCR Kit (Invitrogen) according tothe manufacturer's protocol using previously published primers anddual-labeled probes for rstn, cyclophilin (Brown et al., 2005), socs-3(Steppan et al., 2005) and fiaf (Wiesner et al., 2006) (for sequencessee Table 1). In brief, RNA was diluted to 12 μL and then heatdenatured for 5 min at 70 °C. Samples were returned to ice prior to theaddition of 15 μL of the 2× reverse transcriptionmaster mix and 3 μL ofthe SuperScript™ III RT enzyme master mix (Invitrogen). The RTreaction consisted of a 10 min incubation at 25 °C, 45 min incubationat 42 °C, followed by a 5 min 85 °C termination step, and the resultingcomplementary DNA (cDNA; 30 μL) was stored at −20 °C. For PCRamplifications samples were run in duplicate and only 1 gene wasanalysed per reaction. Individual PCR reactions consisted of a 2×Platinum® quantitative PCR SuperMix-UDG, 7 pmol of the sense andantisense primers, 1 pmol of the appropriate dual-labeled probe, 3 μLof cDNA, to a final volume of 33 μL in sterile water. Reactions wereheated to 95 °C for 10min, followed by 50 amplification cycles of 95 °Cfor 20 s and 60 °C for 60 s using a BioRad iCycler. Standard curves weregenerated using a serial dilution of a reference sample and included ineach realtime run to correct for possible variations in productamplification. Relative copy numbers were obtained from standardcurve values, and were normalized to the values obtained for thehouse keeping gene, cyclophilin. Data are expressed as a percentage ofthe control±SEM. Also note that no significant variations incyclophilin expression were observed between groups when eval-uated using either the threshold cycle (CT), or using the relativeexpression levels, and is consistent with previous reports (Bond et al.,2002).

2.4. Statistics

In vivo LPS experimental data were analysed for each tissue usingthe Student's t-test since values are expressed as a relative percentageof those measured in saline controls for any given time point. Datafrom studies in N-1 cells were subjected to analysis of variance

Fig. 1. The effects of LPS on gene expression in the hypothalamus. [A] Hypothalamicresistin gene expression was significantly increased at 3 h, but had returned to controlvalues at the subsequent time points analysed. [B] In contrast fiaf was significantlyincreased at 3 h and 6 h post-injection, but was fully attenuated by 24 h. [C] Like rstn,socs-3 was strongly induced at 3 h, but had returned to basal levels after 6 h. (⁎pb0.05,⁎⁎⁎pb0.005, ⁎⁎⁎⁎pb0.0001). □ = control.

Fig. 2. The effects of LPS on gene expression in cerebral cortex. [A] Although no changesin cortical rstn were detected after 3 h, its expression was reduced by 6 h, and thenincreased 24 h following the LPS challenge. [B] LPS rapidly upregulated fiaf expression3 h post-injection and these increases were sustained at all further time pointsexamined. [C] Although the increases in socs-3 were greatest after 3 h, its expressionremained significantly elevated for at least 24 h. (⁎pb0.05, ⁎⁎pb0.01, ⁎⁎⁎pb0.005,⁎⁎⁎⁎pb0.0001). □ = control.

98 R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

(ANOVA) followed by the Newman–Keuls multiple comparisons test.Significance was set at pb0.05.

3. Results

3.1. LPS reduces bodyweight in CD-1 mice

As expected, mice challenged with LPS lost a significant amount ofbodyweight 24 h post-injection (27.2±0.4 g vs. 24.1±0.5 g; n=6;pb0.0002), whereas no changes were detected in saline injected mice(27.8±0.6 g vs. 27.3±0.5 g; n=6).

3.2. The effects of LPS on hypothalamic gene expression

Hypothalamic rstn gene expression was acutely upregulated byLPS injection 64% (pb0.005) 3 h post-injection, but had returned tocontrol values by 6 h and 24 h (Fig. 1A). fiaf mRNA was also rapidlyincreased by 94% (pb0.0001) and 113% (pb0.05) at 3 h and 6 h post-injection respectively, but had returned to basal levels by 24 h(Fig. 1B). Like rstn, socs-3 gene expressionwas strongly upregulated inthe hypothalamus 3 h following LPS administration (400%;pb0.0001), but had returned to control levels by 6 h (Fig. 1C).

3.3. LPS modulates gene expression in cerebral cortex

Cortical rstn expression was unchanged 3 h following LPSadministration, was reduced by 35% (pb0.01) after 6 h, but thenincreased by 24 h (+40%; pb0.05) (Fig. 2A). In contrast, fiaf mRNAwas increased by 59% (pb0.005),119% (pb0.05) and 106% (pb0.0001)at 3 h, 6 h and 24 h respectively following the LPS challenge (Fig. 2B).LPS also induced socs-3 expression by 340% (pb0.0001), 73% (pb0.05)and 140% (pb0.0001) at 3 h, 6 h and 24 h post-injection, respectively(Fig. 2C).

3.4. Pituitary adipokines are stimulated by LPS

LPS induced a robust increase in PIT rstn 3 h post-injection(+110%; pb0.005), but its expression had returned to control levelsafter 6 h (Fig. 3A). fiafmRNAwas also increased by 107% (pb0.05) and321% (pb0.005) 3 h and 6 h post-injection, but no differences weredetected 24 h following treatment (Fig. 3B). As seen in Fig. 3C, LPStreatment led to robust increases in the expression of socs-3 after 3 h(440%; pb0.0001), and its expression remained elevated even after6 h (184%; pb0.005), but had returned to basal levels 24 h later.

Fig. 3. The effects of LPS on gene expression in pituitary gland. Like the hypothalamus,rstn gene expressionwas rapidly increased within the pituitary at 3 h, but had returnedto control levels by 6 h and 24 h [A]. Although pituitary fiaf expressionwas increased 3 hfollowing LPS treatment, its expression peaked 6 h post-injection, and had returned tocontrol levels by 24 h [B]. socs-3 expression appeared to be maximally increased after3 h, but remained significantly elevated at 6 h, and returned to basal levels by 24 h [C].(⁎pb0.05, ⁎⁎⁎pb0.005, ⁎⁎⁎⁎pb0.0001). □ = control.

Fig. 4. The effects of LPS on gene expression in adipose tissue. [A] Surprisingly LPS failedto induce any detectable changes in rstn gene expression in adipose tissue at all of thetime points tested. [B] In contrast, fiaf expression was significantly elevated after 3 h,and remained higher 6 h and 24 h following LPS administration. [C] The injection of LPSalso robustly induced socs-3 expression, which peaked after 3 h, and it remainedelevated even 24 h following the endotoxic challenge. (⁎pb0.05, ⁎⁎pb0.01, ⁎⁎⁎pb0.005,⁎⁎⁎⁎pb0.0001). □ = control.

99R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

3.5. LPS modulates gene expression in adipose tissue

LPS failed to modify rstn gene expression in FAT at any of the timepoints analysed (Fig. 4A). In contrast fiaf gene expression wasincreased by 57% (pb0.05), 144% (pb0.01) and 137% (pb0.005) 3 h,6 h and 24 h following LPS injection, relative to the saline injectedcontrols (Fig. 4B). Socs-3 gene expression was also robustly upregu-lated in FAT by 700% (pb0.0001) 3 h post-injection, and remained 4-fold higher (pb0.005) 6 h, and even 24 h, following treatment (Fig.4C).

3.6. LPS increased fiaf and socs-3, but not rstn, in N-1 hypothalamicneurons

As illustrated in Fig. 5A, LPS (25 μg/mL; 3 h) had no significanteffect on rstn gene expression when added directly to cultures of N-1cells, relative to cells treated with vehicle alone, whereas in the samesamples fiaf expression was increased 10-fold (pb0.005). Thisstimulationwas only partially attenuated by pre-treating N-1 neurons

with any of the TLR4 signaling inhibitors. Parthenolide, an inhibitor ofNF-kB activation, partially attenuated (−66%; pb0.005) the LPS-mediated induction of fiaf. Similarly PD098059, a MAPK/MEKinhibitor, reduced LPS-induced fiaf expression (pb0.005; Fig. 5B)compared to LPS alone. SB202190, a p38 MAP kinase inhibitor, had asimilar effect (−35%; pb0.005). Note however that the inhibitionwasincomplete and that none of these inhibitors had an effect on rstnexpression (Fig. 5A). And that the TLR4 signaling antagonists alonehad no significant effect on the basal level of expression for either rstnor fiaf in N-1 neurons.

In contrast to the effects of the TLR4 inhibitors on fiaf expression,the LPS-induced increases of socs-3 expression in N-1 neurons(+37%; pb0.005), were completely prevented by pretreatment withPD098059 and SB202190. A further interesting difference between fiafand socs3 was the significant inhibitory influence of the inhibitorsalone; i.e., both reduced socs-3 mRNA by 40% (pb0.005) relative tountreated N-1 neurons. The effect of parthenolide on socs3 expressionwas also distinct from the effects of PD098059 and SB202190 in that itsignificantly increased socs3 expression in the absence of LPS. Further,

Fig. 5. The effects of LPS on gene expression in N-1 hypothalamic neurons. N-1hypothalamic neurons were pretreated for 30 minwith either vehicle or TLR4 signalingantagonists and then subdivided to receive either vehicle or LPS for a further 3 h.[A] Unlike the hypothalamus, rstn gene expression was unchanged by either LPS or theTLR4 signaling antagonists (parthenolide, PD098059, SB202190) in N-1 cells.[B] However LPS induced a robust increase in fiaf expression in N-1 neurons (10-fold), and this was only partially attenuated by the TLR-4 signaling antagonists. [C]Although LPS modestly increased socs-3 mRNA, treating N-1 cells with the TLR4signaling antagonists alone also had significant effects on socs-3 expression. Data areexpressed as a percentage of the control and were pooled from multiple experiments(pb0.01, ⁎⁎⁎pb0.005).

100 R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

no differences were detected in socs-3 expression between N-1 cellstreated with LPS alone relative to those pre-exposed to parthenolideprior to treatment with LPS.

4. Discussion

These data are consistent with our hypothesis that LPS wouldmodify the central expression of rstn, fiaf and socs-3, genes that areimplicated in the control of central metabolic pathways. A single s.c.injection of LPS induced a significant weight loss in mice in addition totime- and tissue-dependent increases in adipokine gene expression.Although LPS rapidly induced rstn gene expression in the hypotha-lamus and pituitary, cortical rstn mRNA was unchanged after 3 h,decreased 6 h following LPS, but then increased 24 h later. In contrastrstn remained unchanged in adipose tissue at all time points analysed.Although LPS rapidly induced fiaf expression in all tissues itsexpression was fully attenuated 24 h later in the hypothalamus and

pituitary, but remained maximally elevated in cortex and fat. socs-3expression was also induced by LPS in all tissues, with the greatestincreases detected 3 h post-injection. Like rstn, socs-3 mRNA peakedwithin 3–6 h in hypothalamus and pituitary, but remained signifi-cantly elevated in cortical and adipose tissues even 24 h following theendotoxic challenge.

In contrast to the mouse experiments in vivo, LPS failed tosignificantly induce rstn gene expression when added directly to N-1hypothalamic neurons. N-1 hypothalamic neurons are part of an arrayof unique cell line models that express many of the typicalneuropeptides found in the intact hypothalamus, including proopio-melanocortin (POMC) and Neuropeptide Y (NPY), key targetsimplicated in the central control of appetite and bodyweight (Belshamet al., 2004). Moreover these cell lines were shown to respond totreatments in a similar manner to that of the intact hypothalamuswhich strongly suggests their suitability and value for the in vitrostudy of hypothalamic function (Cheng et al., 2007; Cui et al., 2005).However it is also worth noting that within the hypothalamus thereare several neuronal types which appear to differentially respond toexternal treatment (Cai et al., 2007), which might be the case herewith LPS and rstn. This lack of response is therefore quite different tothe effects of LPS on 3T3-L1 adipocytes in vitro, where the same doseof LPS as used in the present study (25 μg/mL; 12 h) increased rstnmRNA levels (Lu et al., 2002). Nonetheless, in spite of the lack of effecton rstn expression, the treatment of N-1 neurons with LPS modestly,but significantly, increased socs-3 mRNA, and markedly induced fiafexpression (10-fold) suggesting they are responsive to an endotoxicchallenge. Our attempts to implicate the TLR4 signaling cascade in themodulation of these LPS-mediated changes in N-1 neurons producedsome complex responses. Treatment of cells with the TLR4 signalingantagonists parthenolide (NFκB), PD098059 (MEK1) and SB202190(p38 MAPK) only partially attenuated the LPS-mediated increases infiaf expression and suggests that other signaling mechanisms arelikely to be involved. This may include activation of signal transducerand activator of transcription 3 (STAT3) or c-Jun-N terminal kinase(JNK) signaling pathways which have also been shown to be activatedin macrophages and microglia following LPS treatment (Qin et al.,2007). However LPS may also simultaneously activate a variety ofpathways in N-1 neurons which would require cells to be treated withmultiple antagonists in order to completely block its effects on fiaf. Incontrast, parthenolide had no effect on LPS-induced socs3 expressionwhereas PD0980509 and SB202190 completely blocked the stimula-tory effects of LPS on socs-3 mRNA, relative to cells treated with thesignaling antagonists alone. Unexpectedly we also observed an LPS-independent effect of these drugs on socs3; i.e., parthenolideincreased socs3 expression, whereas SB202190 and PD098059inhibited its expression.

In agreement with many other reports, we confirmed that micechallenged with LPS underwent a significant weight loss 24 h post-injection. Although the cachectic effects of LPS appear to bedependent on an intact hypothalamic melanocortin system (Huanget al., 1999; Marks et al., 2001), it remains unclear how LPS modulatesthe activity of this neuronal pathway. Several adipokines influencehypothalamic energy homeostasis (Ahima and Lazar, 2008; Ahimaet al., 2006), however their individual roles in inflammatory-inducedanorexia remain uncertain. LPS is known to induce adipose leptin geneexpression, and increase circulating leptin and interleukin-6 levels,which were postulated to mediate the anorectic and inflammatoryresponses to LPS (Rummel et al., 2008; Sarraf et al., 1997). LPS alsoincreased leptin and insulin transport across the blood-brain barrierwhich may contribute to endotoxin-induced anorexia (Banks, 2006).However leptin-deficient (ob/ob), or the leptin receptor-deficient(db/db), mice were not protected from the anorectic effects of LPS(Faggioni et al., 1997), nor were adiponectin-deficient mice (Pini et al.,2006), suggesting that the anorectic effects of LPS are occurring viaalternate mechanism(s). Perhaps redundant pathways have evolved

101R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

to preserve the anorectic and diabetic responses induced by bacterialinfections, and disrupting a single gene would not protect miceagainst LPS-induced cachexia. We are presently unaware of anystudies investigating the effects of LPS on rstn- or fiaf-knockout mice.

The LPS-induced changes in hypothalamic rstn expression areintriguing since resistin-ir colocalizes with the melanocortin system(Wilkinson et al., 2005), a key target implicated in LPS-inducedmetabolic disturbances (Huang et al., 1999; Marks et al., 2001).Additionally the icv injection/infusion of resistin also inhibited foodintake (Tovar et al., 2005; Vazquez et al., 2008), and induced hepaticglucose production in rats (Muse et al., 2007), a finding that isconsistent with the increase in brain rstn expression measuredfollowing the LPS challenge. Such a role for hypothalamic-derivedrstn might be clarified by using our resistin-specific small interferingRNA (siRNA) to block the LPS-mediated increases in hypothalamicrstn (Brown et al., 2007), which will be investigated in future studies.Although LPS failed to modify rstn gene expression in adipose tissue,which is in contrast to the findings of Lu et al. (2002), this mightreflect age- (PD31 vs 9weeks) and species-dependent differences(mouse vs. rat). It is also worth noting that LPS was also found to haveno effect on resistin release from human adipose tissue or fromskeletal muscle (Lappas et al., 2005). Although these preliminaryinvestigations clearly showed an effect of LPS on hypothalamic rstngene expression, an important future goal remains to verify whetherLPS can induce local changes in resistin protein concentration orrelease from the intact hypothalamus in addition to investigating thepotential effects of LPS within distinct hypothalamic nuclei.

Fasting-induced adipose factor (FIAF) is thought to modulate lipidmetabolism through the inhibition of lipoprotein lipase activity, andhas been implicated in the control of body composition and glucosemetabolism (Backhed et al., 2007; Kersten, 2005; Xu et al., 2005). Atpresent the role(s) of FIAF in the central nervous system remainsuncertain. Previously we speculated that FIAF may participate in localinflammatory responses, tissue repair and angiogenesis followingbrain injury, and this was supported by our demonstration that fiafgene expression is robustly increased by cerebral hypoxia/ischemia(H/I) in the cortex and hippocampus of neonatal mice (Wiesner et al.,2006) and in rats subjected to traumatic brain injury (Brown et al.,2008b). However in our hands LPS-treated neonatal mice failed toshow significant increases in cortical fiaf at any of the time pointsanalysed, despite the use of higher doses of LPS (30 mg/kg) (Wiesneret al., 2006). This is probably because neonatal rodents are resistant tothe anorectic effects of LPS and can tolerate much higher doses thatwould be lethal in adult mice (Spencer et al., 2007). Thus in thepresent studies, using pubertal mice, LPS rapidly induced fiafexpression in all tissues analysed. Although these increases weresoon attenuated in the hypothalamus and pituitary, they persisted incortex and adipose tissue for at least 24 h. Although the sustainedincreases in fiaf gene expression could be a secondary consequence ofthe LPS-induced anorexia, since fasting also induces fiaf transcriptionin fat and pituitary gland (Kersten et al., 2000; Wiesner et al., 2004),our in vitro studies in N-1 neurons indicate that LPS is likely having adirect stimulatory effect on hypothalamic fiaf expression. Exploratoryexperiments with various TLR4 signaling inhibitors revealed that LPS-mediated induction of fiaf expression (and socs3; see below) is at leastpartially mediated through NFκB and MAPK signaling pathways,suggesting that N-1 neurons contain an intact TLR4 signaling cascade.The reason for this is unclear, particularly since LPS is unlikely to crossthe blood-brain barrier (Singh and Jiang, 2004). However, assuggested by others, an endogenous ligand for central TLR4 mayexist (Tang et al., 2007). As noted already peripheral treatment withLPS induced inflammatory signals predominantly via IL-6 (Rummelet al., 2008), and we can now add FIAF to this signaling pathway. Thuswhether central TLR4s are also involved remains to be determined.

We also confirmed that socs3, but not rstn, expression is implicatedin the central response to peripheral LPS injection. LPS rapidly induced

socs-3 expression in all tissues, and it even remained significantlyelevated at 24 h in cortex and adipose tissue. Previous studies in miceand rats indicated that most of the central expression occurred inblood vessels, ependymal cells, leptomeninges and in cells of thecircumventricular organs (Lebel et al., 2000). Note however that thislatter study used a dose of LPS (2500 μg/kg; ip) which was 100-foldhigher than that employed in our experiments. In the absence of datafrom in situ hybridization histochemistry it is not possible for us toknow whether LPS increased socs3mRNA in neurons. The cortical andhypothalamic tissue samples obtained in our dissections would haveincluded all of the cell types pinpointed by Lebel et al. (2000).However we also showed that LPS added directly to N-1 hypothalamicneuron cultures modestly increased socs3 mRNA (~35%), suggestingthat these neurons possess TLR4-coupled socs3 expression. Whetherthis is so in the intact hypothalamus is presently unknown, thoughsuch a possibility has been suggested (Zhang et al., 2008). SOCS3 is aknown inhibitor of hypothalamic leptin and insulin signaling such thatincreased hypothalamic socs-3 expression induced leptin and insulinresistance whereas socs-3 haploinsufficiency (+/−) protects miceagainst diet-induced obesity and improves glucose tolerance (Howardand Flier, 2006). Our observation that a single injection of LPS induceda prolonged increase in socs3 mRNA in visceral fat and in brain tissue,coincident with a drop in body weight, would appear to beinconsistent with this theory. However when mice are chronicallyexposed to LPS, via osmotic minipumps, they become obese andhyperinsulinemic (Cani et al., 2007). It is conceivable that the LPS-induced increase in socs3 expression that we have described mayaccount, at least partially, for this result. The concomitant increases inrstn and fiaf expression, following an injection of LPS, could also becomponents of this mechanism. As noted already, FIAF mayparticipate in local inflammatory responses, tissue repair andangiogenesis following brain injury (Brown et al., 2008b; Wiesneret al., 2006), and the icv injection/infusion of resistin inhibited foodintake (Tovar et al., 2005; Vazquez et al., 2008). The latter suggestsitself as a possible hypothalamic counter-measure against LPS-induced obesity.

In agreement with the studies on LPS-induced fiaf expression in N-1 neurons (see above), a TLR4-signaling pathway also appears to beinvolved in the regulation of socs3 expression, though treatment of N-1 neurons with the same inhibitors reveals a more complex relation-ship than that implied for control of fiaf expression; i.e., blockade ofNF-kB with parthenolide had no effect on LPS-induced socs3 mRNA,but when used alone it was more effective than LPS in the induction ofsocs3 expression. In marked contrast, inhibition of p38 and MAPKpathways in the absence of LPS resulted in inhibition of socs3expression. These preliminary data indicate that these pathways areconstitutively active in non-stimulated hypothalamic neurons in vitroand a more intensive study of TLR4-coupled signals is required.

The significance of the gene expression changes detected in thepituitary gland remains obscure, though they appear similar to thoseobserved in the hypothalamus (Fig. 3 vs. Fig. 1). For example rstn geneexpression was increased only 3 h after the injection of LPS in bothtissues, and the increases in fiaf and socs-3 were no longer detected24 h after the injection of LPS, although the expression of both genesremained elevated in the pituitary slightly longer. It remains possiblethat LPS-mediated modifications in hypothalamic function couldinitiate the changes in pituitary gene regulation. However there is alsogood evidence for LPS signaling pathways in both rodent and humanpituitary tissue suggesting that LPS may directly regulate pituitarygene expression. For example the expressions of LPS patternrecognition receptors, tlr4 and CD 14 were detectable in murine andhuman pituitary cell lines (Breuel et al., 2004; Iwasaki et al., 2008;Tichomirowa et al., 2005), in normal and adenoma tissue from humanpituitary (Tichomirowa et al., 2005), and in a porcine pituitary stemcell line (Nagai et al., 2008). Tichomirowa et al. (2005) alsodemonstrated the presence of TLR4 protein in human pituitary tissue,

102 R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

and these receptors responded to LPS (1 μg/mL) stimulation throughan increase in production of IL-6. In addition, proopiomelanocortin(pomc) gene expression was increased in the AtT20 murine cortico-trophic cell line following LPS (1 μg/mL) treatment (Iwasaki et al.,2008). Taken all together, our preliminary data, along with previousreports, strongly suggest that a functional LPS/TLR4 signalingmechan-ism exists in both normal and tumour cells of the pituitary gland.

In conclusion LPS induced a significant reduction in bodyweightand modified brain and pituitary rstn, fiaf and socs-3 gene expressionin a time- and tissue-dependent manner. The increases in hypotha-lamic rstn gene expression raise questions about its possibleinvolvement in central energymetabolism and inflammatory-inducedcachexia. Similarly increased fiaf expression hints at a potential role inlocal tissue injury and inflammation. LPS-induced elevations in socs-3suggest that central cytokine signaling might be impaired, and thiswould be consistent with the known effect of chronic LPS in inducingobesity and hyperinsulinemia (Cani et al., 2007). Our demonstrationthat LPS exerts similar effects in pituitary gland and hypothalamicneurons serves to strengthen our suggestion that LPS would induceadipokine gene expression in multiple tissues, and might haveimplications for the control of bodyweight and energy metabolism.

Acknowledgements

These studies were funded by the NSHRF, the IWK Health Centre,the Atlee Endowment, and UIMRF/Capital Health. RB was therecipient of a NSHRF Graduate Studentship. We are indebted toDiane Wilkinson and Paul Wilkinson for technical assistance. The N-1hypothalamic neuronal cell line was generously provided by Dr.Denise Belsham (University of Toronto).

References

Ahima, R.S., Lazar, M.A., 2008. Adipokines and the peripheral and neural control ofenergy balance. Mol. Endocrinol. 22, 1023–1031.

Ahima, R.S., Qi, Y., Singhal, N.S., 2006. Adipokines that link obesity and diabetes to thehypothalamus. Prog. Brain. Res. 153, 155–174.

Backhed, F., Manchester, J.K., Semenkovich, C.F., Gordon, J.I., 2007. Mechanismsunderlying the resistance to diet-induced obesity in germ-free mice. Proc. Natl.Acad. Sci. U. S. A. 104, 979–984.

Banks, W.A., 2006. Blood-brain barrier and energy balance. Obesity (Silver Spring) 14(Suppl 5), 234S–237S.

Belanger, A.J., Lu, H., Date, T., Liu, L.X., Vincent, K.A., Akita, G.Y., Cheng, S.H., Gregory, R.J.,Jiang, C., 2002. Hypoxia up-regulates expression of peroxisome proliferator-activated receptor gamma angiopoietin-related gene (PGAR) in cardiomyocytes:role of hypoxia inducible factor 1alpha. J. Mol. Cell. Cardiol. 34, 765–774.

Belsham, D.D., Cai, F., Cui, H., Smukler, S.R., Salapatek, A.M., Shkreta, L., 2004. Generationof a phenotypic array of hypothalamic neuronal cell models to study complexneuroendocrine disorders. Endocrinology 145, 393–400.

Bond, B.C., Virley, D.J., Cairns, N.J., Hunter, A.J., Moore, G.B., Moss, S.J., Mudge, A.W.,Walsh, F.S., Jazin, E., Preece, P., 2002. The quantification of gene expression in ananimal model of brain ischaemia using TaqMan real-time RT-PCR. Brain. Res. Mol.Brain. Res. 106, 101–116.

Breuel, K.F., Kougias, P., Rice, P.J., Wei, D., De Ponti, K., Wang, J., Laffan, J.J., Li, C.,Kalbfleisch, J., Williams, D.L., 2004. Anterior pituitary cells express patternrecognition receptors for fungal glucans: implications for neuroendocrine immuneinvolvement in response to fungal infections. Neuroimmunomodulation 11, 1–9.

Brown,R., Imran, S.A., Belsham,D.D., Ur, E.,Wilkinson,M., 2007. Adipokine geneexpressionin a novel hypothalamic neuronal cell line: resistin-dependent regulation of fasting-induced adipose factor and SOCS-3. Neuroendocrinology 85, 232–241.

Brown, R., Imran, S.A., Ur, E., Wilkinson, M., 2008a. Valproic acid and CEBPalpha-mediated regulation of adipokine gene expression in hypothalamic neurons and3T3-L1 adipocytes. Neuroendocrinology 88, 25–34.

Brown, R., Thompson, H.J., Imran, S.A., Ur, E., Wilkinson, M., 2008b. Traumatic braininjury induces adipokine gene expression in rat brain. Neurosci. Lett. 432, 73–78.

Brown, R., Wiesner, G., Ur, E., Wilkinson, M., 2005. Pituitary resistin gene expression isupregulated in vitro and in vivo by dexamethasone but is unaffected byrosiglitazone. Neuroendocrinology 81, 41–48.

Cai, F., Gyulkhandanyan, A.V., Wheeler, M.B., Belsham, D.D., 2007. Glucose regulatesAMP-activated protein kinase activity and gene expression in clonal, hypothalamicneurons expressing proopiomelanocortin: additive effects of leptin or insulin.J. Endocrinol. 192, 605–614.

Cani, P.D., Amar, J., Iglesias, M.A., Poggi, M., Knauf, C., Bastelica, D., Neyrinck, A.M., Fava,F., Tuohy, K.M., Chabo, C., Waget, A., Delmee, E., Cousin, B., Sulpice, T., Chamontin, B.,Ferrieres, J., Tanti, J.F., Gibson, G.R., Casteilla, L., Delzenne, N.M., Alessi, M.C.,

Burcelin, R., 2007. Metabolic endotoxemia initiates obesity and insulin resistance.Diabetes 56, 1761–1772.

Cheng, H., Isoda, F., Belsham, D.D., Mobbs, C.V., 2007. Inhibition of AgRP expression byglucose in a clonal hypothalamic neuronal cell line is mediated by glycolysis, notoxidative phosphorylation. Endocrinology 149, 703–710.

Cui, H., Cai, F., Belsham, D.D., 2005. Anorexigenic hormones leptin, insulin, and alpha-melanocyte-stimulating hormone directly induce neurotensin (NT) gene expres-sion in novel NT-expressing cell models. J. Neurosci. 25, 9497–9506.

Faggioni, R., Fuller, J., Moser, A., Feingold, K.R., Grunfeld, C., 1997. LPS-induced anorexiain leptin-deficient (ob/ob) and leptin receptor-deficient (db/db) mice. Am. J.Physiol. 273, R181–186.

Howard, J.K., Flier, J.S., 2006. Attenuation of leptin and insulin signaling by SOCSproteins. Trends. Endocrinol. Metab. 17, 365–371.

Huang, Q.H., Hruby, V.J., Tatro, J.B., 1999. Role of central melanocortins in endotoxin-induced anorexia. Am. J. Physiol. 276, R864–871.

Iwasaki, Y., Taguchi, T., Nishiyama, M., Asai, M., Yoshida, M., Kambayashi, M., Takao, T.,Hashimoto, K., 2008. Lipopolysaccharide stimulates proopiomelanocortin geneexpression in AtT20 corticotroph cells. Endocr. J. 55, 285–290.

Kaser, S., Kaser, A., Sandhofer, A., Ebenbichler, C.F., Tilg, H., Patsch, J.R., 2003. Resistinmessenger-RNA expression is increased by proinflammatory cytokines in vitro.Biochem. Biophys. Res. Commun. 309, 286–290.

Kersten, S., 2005. Regulation of lipid metabolism via angiopoietin-like proteins.Biochem. Soc. Trans. 33, 1059–1062.

Kersten, S., Mandard, S., Tan, N.S., Escher, P., Metzger, D., Chambon, P., Gonzalez, F.J.,Desvergne, B., Wahli, W., 2000. Characterization of the fasting-induced adiposefactor FIAF, a novel peroxisome proliferator-activated receptor target gene. J. Biol.Chem. 275, 28488–28493.

Lappas, M., Yee, K., Permezel, M., Rice, G.E., 2005. Release and regulation of leptin,resistin and adiponectin from human placenta, fetal membranes, and maternaladipose tissue and skeletal muscle from normal and gestational diabetes mellitus-complicated pregnancies. J. Endocrinol. 186, 457–465.

Lebel, E., Vallieres, L., Rivest, S., 2000. Selective involvement of interleukin-6 in thetranscriptional activation of the suppressor of cytokine signaling-3 in the brainduring systemic immune challenges. Endocrinology 141, 3749–3763.

Lu, S.C., Shieh, W.Y., Chen, C.Y., Hsu, S.C., Chen, H.L., 2002. Lipopolysaccharide increasesresistin gene expression in vivo and in vitro. FEBS Lett. 530, 158–162.

Marks, D.L., Butler, A.A., Turner, R., Brookhart, G., Cone, R.D., 2003. Differential role ofmelanocortin receptor subtypes in cachexia. Endocrinology 144, 1513–1523.

Marks, D.L., Ling, N., Cone, R.D., 2001. Role of the central melanocortin system incachexia. Cancer. Res. 61, 1432–1438.

Morash, B.A., Wilkinson, D., Ur, E., Wilkinson, M., 2002. Resistin expression andregulation in mouse pituitary. FEBS Lett. 526, 26–30.

Muse, E.D., Lam, T.K., Scherer, P.E., Rossetti, L., 2007. Hypothalamic resistin induceshepatic insulin resistance. J. Clin. Invest. 117, 1670–1678.

Nagai, Y., Ogasawara, H., Taketa, Y., Aso, H., Kanaya, T., Miyake, M., Watanabe, K.,Ohwada, S., Muneta, Y., Yamaguchi, T., 2008. Expression of inflammatory-relatedfactors in porcine anterior pituitary-derived cell line. Vet. Immunol. Immunopathol.124, 201–208.

Pini, M., Sennello, J.A., Chan, L., Fantuzzi, G., 2006. Adiponectin deficiency does not affectthe inflammatory response to endotoxin or concanavalin a in mice. Endocrinology147, 5019–5022.

Qi, Y., Nie, Z., Lee, Y.S., Singhal, N.S., Scherer, P.E., Lazar, M.A., Ahima, R.S., 2006. Loss ofresistin improves glucose homeostasis in leptin deficiency. Diabetes 55,3083–3090.

Qin, H., Roberts, K.L., Niyongere, S.A., Cong, Y., Elson, C.O., Benveniste, E.N., 2007.Molecular mechanism of lipopolysaccharide-induced SOCS-3 gene expression inmacrophages and microglia. J. Immunol. 179, 5966–5976.

Rummel, C., Inoue, W., Sachot, C., Poole, S., Hubschle, T., Luheshi, G.N., 2008. Selectivecontribution of interleukin-6 and leptin to brain inflammatory signals induced bysystemic LPS injection in mice. J. Comp. Neurol. 511, 373–395.

Sachot, C., Poole, S., Luheshi, G.N., 2004. Circulating leptin mediates lipopolysaccharide-induced anorexia and fever in rats. J. Physiol. 561, 263–272.

Sarraf, P., Frederich, R.C., Turner, E.M., Ma, G., Jaskowiak, N.T., Rivet III, D.J., Flier, J.S.,Lowell, B.B., Fraker, D.L., Alexander, H.R., 1997. Multiple cytokines and acuteinflammation raise mouse leptin levels: potential role in inflammatory anorexia.J. Exp. Med. 185, 171–175.

Singh, A.K., Jiang, Y., 2004. How does peripheral lipopolysaccharide induce geneexpression in the brain of rats? Toxicology 201, 197–207.

Singhal, N.S., Lazar, M.A., Ahima, R.S., 2007. Central resistin induces hepatic insulinresistance via neuropeptide Y. J. Neurosci. 27, 12924–12932.

Spencer, S.J., Mouihate, A., Galic, M.A., Ellis, S.L., Pittman, Q.J., 2007. Neonatal immunechallenge does not affect body weight regulation in rats. Am. J. Physiol. Regul.Integr. Comp. Physiol. 293, pR581–pR589.

Steppan, C.M., Bailey, S.T., Bhat, S., Brown, E.J., Banerjee, R.R., Wright, C.M., Patel, H.R.,Ahima, R.S., Lazar, M.A., 2001. The hormone resistin links obesity to diabetes. Nature409, 307–312.

Steppan, C.M., Wang, J., Whiteman, E.L., Birnbaum, M.J., Lazar, M.A., 2005. Activation ofSOCS-3 by resistin. Mol. Cell. Biol. 25, 1569–1575.

Tang, S.C., Arumugam, T.V., Xu, X., Cheng, A., Mughal, M.R., Jo, D.G., Lathia, J.D., Siler, D.A.,Chigurupati, S., Ouyang, X., Magnus, T., Camandola, S., Mattson, M.P., 2007. Pivotalrole for neuronal Toll-like receptors in ischemic brain injury and functional deficits.Proc. Natl. Acad. Sci. U. S. A. 104, 13798–13803.

Tichomirowa, M., Theodoropoulou, M., Lohrer, P., Schaaf, L., Losa, M., Uhl, E., Lange, M.,Arzt, E., Stalla, G.K., Renner, U., 2005. Bacterial endotoxin (lipopolysaccharide)stimulates interleukin-6 production and inhibits growth of pituitary tumour cellsexpressing the Toll-like receptor 4. J. Neuroendocrinol. 17, 152–160.

103R. Brown et al. / Journal of Neuroimmunology 209 (2009) 96–103

Tovar, S., Nogueiras, R., Tung, L.Y., Castaneda, T.R., Vazquez, M.J., Morris, A., Williams, L.M.,Dickson, S.L., Dieguez, C., 2005. Central administration of resistin promotes short-term satiety in rats. Eur. J. Endocrinol. 153, R1–5.

Ueki, K., Kondo, T., Kahn, C.R., 2004. Suppressor of cytokine signaling 1 (SOCS-1) andSOCS-3 cause insulin resistance through inhibition of tyrosine phosphorylation ofinsulin receptor substrate proteins by discrete mechanisms. Mol. Cell. Biol. 24,5434–5446.

Vazquez, M.J., Gonzalez, C.R., Varela, L., Lage, R., Tovar, S., Sangiao-Alvarellos, S.,Williams, L.M., Vidal-Puig, A., Nogueiras, R., Lopez, M., Dieguez, C., 2008. Centralresistin regulates hypothalamic and peripheral lipid metabolism in a nutritionaldependent fashion. Endocrinology 149, 4534–4543.

Wiesner, G., Brown, R.E., Robertson, G.S., Imran, S.A., Ur, E., Wilkinson, M., 2006.Increased expression of the adipokine genes resistin and fasting-induced adiposefactor in hypoxic/ischaemic mouse brain. Neuroreport 17, 1195–1198.

Wiesner, G., Morash, B.A., Ur, E., Wilkinson, M., 2004. Food restriction regulates adipose-specific cytokines in pituitary gland but not in hypothalamus. J. Endocrinol. 180,R1–6.

Wilkinson, M., Brown, R., Imran, S.A., Ur, E., 2007. Adipokine gene expression in brainand pituitary gland. Neuroendocrinology 86, 191–209.

Wilkinson, M.,Wilkinson, D., Wiesner, G., Morash, B., Ur, E., 2005. Hypothalamic resistinimmunoreactivity is reduced by obesity in the mouse: co-localization with alpha-melanostimulating hormone. Neuroendocrinology 81, 19–30.

Xu, A., Lam, M.C., Chan, K.W., Wang, Y., Zhang, J., Hoo, R.L., Xu, J.Y., Chen, B., Chow, W.S.,Tso, A.W., Lam, K.S., 2005. Angiopoietin-like protein 4 decreases blood glucose andimproves glucose tolerance but induces hyperlipidemia and hepatic steatosis inmice. Proc. Natl. Acad. Sci. U. S. A. 102, 6086–6091.

Zhang, X., Zhang, G., Zhang, H., Karin, M., Bai, H., Cai, D., 2008. Hypothalamic IKKbeta/NF-kappaB and ER stress link overnutrition to energy imbalance and obesity. Cell.135, 61–73.