chemically defined projections linking the mediobasal hypothalamus and the lateral hypothalamic area

Chemically Defined Projections Linkingthe Mediobasal Hypothalamus

and the Lateral Hypothalamic Area

CAROL F. ELIAS,1 CLIFFORD B. SAPER,1 ELEFTHERIA MARATOS-FLIER,2

NICHOLAS A. TRITOS,2 CHARLOTTE LEE,3 JOSEPH KELLY,3 JEFFREY B. TATRO,4

GLORIA E. HOFFMAN,5 MICHAEL M. OLLMANN,6 GREGORY S. BARSH,6

TAKESHI SAKURAI,7 MASASHI YANAGISAWA,7 AND JOEL K. ELMQUIST1,3*1Department of Neurology, Beth Israel Deaconess Medical Center and Program

in Neuroscience, Harvard Medical School, Boston, Massachusetts 022152The Joslin Diabetes Center, Harvard Medical School, Boston, Massachusetts 02215

3Department of Medicine and Division of Endocrinology, Beth Israel DeaconessMedical Center, Harvard Medical School, Boston, Massachusetts 02215

4Division of Endocrinology, Metabolism, and Molecular Medicine, Tupper Research Institute,Tufts University School of Medicine, Boston, Massachusetts 02111

5Department of Anatomy and Neurobiology, University of Maryland School of Medicine,Baltimore, Maryland 21201

6Howard Hughes Medical Institute and Departments of Pediatrics and Genetics,Stanford University School of Medicine, Stanford, California 94395

7Howard Hughes Medical Institute and Department of Molecular Genetics,University of Texas Southwestern Medical Center at Dallas, Dallas, Texas

ABSTRACTRecent studies have identified several neuropeptide systems in the hypothalamus that

are critical in the regulation of body weight. The lateral hypothalamic area (LHA) has longbeen considered essential in regulating food intake and body weight. Two neuropeptides,melanin-concentrating hormone (MCH) and the orexins (ORX), are localized in the LHA andprovide diffuse innervation of the neuraxis, including monosynaptic projections to the cerebralcortex and autonomic preganglionic neurons. Therefore, MCH and ORX neurons may regulateboth cognitive and autonomic aspects of food intake and body weight regulation. The arcuatenucleus also is critical in the regulation of body weight, because it contains neurons thatexpress leptin receptors, neuropeptide Y (NPY), a-melanin-stimulating hormone (a-MSH),and agouti-related peptide (AgRP). In this study, we examined the relationships of thesepeptidergic systems by using dual-label immunohistochemistry or in situ hybridization in rat,mouse, and human brains. In the normal rat, mouse, and human brain, ORX and MCHneurons make up segregated populations. In addition, we found that AgRP- and NPY-immunoreactive neurons are present in the medial division of the human arcuate nucleus,whereas a-MSH-immunoreactive neurons are found in the lateral arcuate nucleus. Inhumans, AgRP projections were widespread in the hypothalamus, but they were especiallydense in the paraventricular nucleus and the perifornical area. Moreover, in both rat andhuman, MCH and ORX neurons receive innervation from NPY-, AgRP-, and a-MSH-immunoreactive fibers. Projections from populations of leptin-responsive neurons in the

Grant sponsor: U.S. Public Health Service; Grant numbers: NS33987,MH56537, MH44694, DK48506, DK53301, and DKR3728082; Grant spon-sor: The American Heart Association; Grant number: 9413110; Grantsponsor: Eli Lilly and Company.

*Correspondence to: Joel K. Elmquist, D.V.M., Ph.D., Division of Endocri-nology, Beth Israel Deaconess Medical Center, 325 Research North, 99Brookline Avenue, Boston, MA 02215. E-mail: [email protected]

Received 3 June 1998; Revised 2 September 1998; Accepted 3 September1998

THE JOURNAL OF COMPARATIVE NEUROLOGY 402:442–459 (1998)

r 1998 WILEY-LISS, INC.

mediobasal hypothalamus to MCH and ORX cells in the LHA may link peripheral metaboliccues with the cortical mantle and may play a critical role in the regulation of feeding behaviorand body weight. J. Comp. Neurol. 402:442–459, 1998. r 1998 Wiley-Liss, Inc.

Indexing terms: obesity; feeding; arcuate hypothalamic nucleus; melanin-concentrating hormone;

orexins

The seminal work of Hetherington and Ranson (1940)and Anand and Brobeck (1951) introduced the model oflateral hypothalamic ‘‘feeding’’ centers and ventromedialhypothalamic ‘‘satiety’’ centers in the mammalian hypo-thalamus. These early studies, based on hypothalamiclesions and electrical stimulation, suggested that loss offunction in the mediobasal hypothalamus resulted inmorbid obesity, decreased spontaneous activity, and neuro-endocrine derangements. An incidental finding of thestudies of Hetherington and Ranson was that lesions of thelateral hypothalamic area (LHA) could result in decreasedfood intake. This observation was extended by Anand andBrobeck, who found that circumscribed lesions in the LHAat tuberal levels of the hypothalamus resulted in a markeddecrease in food intake that could lead to death bystarvation. This dual-center hypothesis has survived inthe literature for decades, although it has been challengedfrequently (see, Gold, 1973; Stricker and Verbalis, 1990).

The reasons for this ambiguity stem from many factors,not the least of which is the inherent difficulty in theinterpretation of the specificity of lesions. Several distinctlesion models that damage mediobasal hypothalamic cellgroups, including the arcuate nucleus and the ventrome-dial nucleus, all produce obesity. For example, chemicallesions of the arcuate nucleus with systemic monosodiumglutamate (MSG) result in obesity and hyperphagia (Ol-ney, 1969). Lesions of the mediobasal hypothalamus involv-ing the ventromedial nucleus by using gold thioglucosealso induce obesity (Debons et al., 1977). Lesions centeredon the ventromedial nucleus induce morbid obesity, markedhyperphagia, and central hypogonadism (Hetheringtonand Ranson, 1940). Conversely, later studies showed thathyperphagia and obesity could be produced by lesions that

were located slightly more dorsally, involving the paraven-tricular nucleus of the hypothalamus (Gold, 1973). Lesionsof the dorsomedial nucleus alter long-term growth andbody composition (Bernardis and Bellinger, 1987). There-fore, lesions of multiple mediobasal cell groups may causeoverlapping alterations in body weight regulation, foodintake, and endocrine status.

Similarly, the effects of lesions of the LHA have beencontroversial (Bernardis and Bellinger, 1996). It has beensuggested that aphagia following LHA lesions is due todamage to fibers of passage, not to destruction of keypopulations of neurons residing in the LHA (Stricker,1984). Thus, the neuroanatomical and neurochemical ba-sis for lesion-induced changes on body weight homeostasisobservations have remained unclear (Bernardis and Bell-inger, 1996).

Recent molecular, physiologic, and anatomic studieshave begun to provide insight on the role of the LHA andthe mediobasal hypothalamus in regulating food intakeand body weight. The LHA contains neurons that inner-vate the entire neuraxis, including monosynaptic projec-tions to several regions of the cerebral cortex (Saper, 1985;Saper et al., 1986). Included in this projection are neuronsthat contain melanin-concentrating hormone (MCH), apeptide that is found in neurons only in the perifornicalarea of the hypothalamus, the LHA, and the zona incerta(Bittencourt et al., 1992). Two recent studies have identi-fied another family of LHA peptides with a similar ana-tomic distribution of cell bodies. The orexins (ORX; Saku-rai et al., 1998), or hypocretins (de Lecea et al., 1998), arefound exclusively in cell bodies of the LHA and theperifornical area of the rat and mouse brain, although asystematic report of ORX projections has not appeared. Itis noteworthy that intracerebroventricular (icv) injectionsof both MCH and ORX increase food intake, and mRNAlevels of both are elevated during fasting (Qu et al., 1996;Sakurai et al., 1998). These observations suggest that theMCH- and ORX-containing LHA neurons play an impor-tant role in integrating and influencing the complex physi-ology underlying feeding behavior.

The cloning of the leptin gene (Zhang et al., 1994) andthe subsequent demonstration that leptin administrationto leptin-deficient (ob/ob) mice corrects morbid obesity andneuroendocrine and autonomic abnormalities seen in thesemice (Campfield et al., 1995; Halaas et al., 1995; Pelley-mounter et al., 1995) established leptin as an essentialcomponent of body weight homeostasis. One proposed siteof leptin action is the arcuate nucleus of the hypothala-mus. Specifically, neuropeptide Y (NPY) neurons withinthe arcuate nucleus contain leptin receptor mRNA (Merceret al., 1996a). Furthermore, leptin regulates the levels ofNPY mRNA (Stephens et al., 1995; Ahima et al., 1996;Schwartz et al., 1996)

Central melanocortin peptides synthesized in the arcu-ate nucleus are also important regulators of body weightand food intake (Fan et al., 1997; Huszar et al., 1997;

Abbreviations

3v third ventriclea-MSH a-melanin stimulating hormoneAgRP agouti-related peptideARC arcuate nucleus of the hypothalamusCP cerebral peduncleDMH dorsomedial nucleus of the hypothalamusfx fornixic internal capsulelf lenticular fasciculusLHA lateral hypothalamic areaLHAp lateral hypothalamic area, posterior levelLHAt lateral hypothalamic area, tuberal levelMB mammillary bodyMCH melanin-concentrating hormoneME median eminencemt mammillothalamic tractNPY neuropeptide YORX orexinsot optic tractPHA posterior hypothalamic areaPOMC proopiomelanocortinPVH paraventricular nucleus of the hypothalamusVMH ventromedial nucleus of the hypothalamusZI zona incerta

CONNECTIONS OF HYPOTHALAMIC FEEDING CENTERS 443

Ollmann et al., 1997). The proopiomelanocortin (POMC)gene product, a-melanin stimulating hormone (a-MSH)inhibits feeding (even in fasted animals) by acting oncentral melanocortin receptors (Tsujii and Bray, 1989;Ludwig et al., 1998). POMC neurons within the arcuatenucleus contain leptin receptor mRNA (Cheung et al.,1997), and leptin regulates the levels of POMC mRNA inthe arcuate nucleus (Schwartz et al., 1997; Thornton et al.,1997; Mizuno et al., 1998). In addition, melanocortinreceptor blockade by ectopic expression of the agoutiprotein (including within the brain), as seen in the lethalyellow (Ay) mouse, produces adult-onset obesity with lep-tin resistance (Fan et al., 1997). In addition, targeteddeletion of the melanocortin 4 receptor (MC4-R) results inan obesity syndrome that is indistinguishable from thatobserved in the Ay mouse (Huszar et al., 1997). Theendogenous ligand for the MC4-R has not been identifieddefinitively, but it likely includes a-MSH. a-MSH cellbodies have a very limited distribution in the brain and arefound in the arcuate nucleus of the hypothalamus and thenucleus of the solitary tract (Jacobowitz and O’Donohue,1978; Watson et al., 1978; Watson and Akil, 1979; Joseph etal., 1983; Bronstein et al., 1992). Recently, a novel endog-enous melanocortin receptor antagonist, agouti-relatedpeptide (AgRP), has been identified and is expressed in thearcuate nucleus of the hypothalamus (Ollmann et al.,1997; Shutter et al., 1997). Transgenic overexpression ofAgRP results in obesity (Ollmann et al., 1997). Takentogether, these findings suggest that the arcuate nucleusof the hypothalamus is essential in the regulation of bodyweight and food intake. Furthermore, many of the biologi-cal effects of leptin likely are due to engaging pathwaysthat originate in the arcuate nucleus. However, the effer-ent pathways that are distant to the arcuate underlyingthese effects are not well understood.

In this study, we investigated the link between thearcuate nucleus of the hypothalamus and the LHA byassessing the innervation of MCH-immunoreactive (MCH-IR) and ORX-IR neurons by NPY-, a-MSH-, and AgRP-IRterminals in the rat and human brain. We also investi-gated the degree of colocalization of MCH and ORX in theLHA of the rat, mouse, and human brain. Our resultsdemonstrate that MCH and ORX neurons comprise dis-tinct populations that are innervated by AgRP-, NPY-, anda-MSH-IR fibers in the rat and human hypothalamus.These connections may provide a functional link from themediobasal hypothalamus to neurons in the LHA thatinnervate autonomic preganglionic neurons and the corti-cal mantle. This projection may be fundamental in theregulation of feeding behavior in the mammalian brain.

MATERIALS AND METHODS

Tissue preparation

Adult male, pathogen-free, Sprague Dawley rats (250–350 g; Taconic, Germantown, NY) and C57BL/6 rats(Jackson Laboratories, Bar Harbor, ME) were housed in alight- and temperature-controlled environment (12 hourson/ 12 hours off) with food and water available ad libitum.The animals and procedures used were in accordance withthe guidelines and approval of the Harvard Medical Schooland Beth Israel Deaconess Institutional Animal Care andUse Committees. Animals were deeply anesthetized withintraperitoneal (i.p.) chloral hydrate (7%; 350 mg/kg) andwere perfused transcardially with diethyl pyrocarbonate

(DEPC)-treated 0.9% saline followed by 10% neutral buff-ered formalin. The brains were removed, stored in thesame fixative for 4 hours, submerged in 20% sucrose, andcut at 30 µm into five (rats) or four (mice) equal series on afreezing microtome. The sections used for immunohisto-chemistry were stored at 4°C in phosphate buffered saline(PBS) containing 0.02% sodium azide until immunohisto-chemical staining was initiated. Sections used for in situhybridization were stored at 220°C in an antifreezesolution (Simmons et al., 1989) until sections were pro-cessed for in situ hybridization histochemistry.

Tissue blocks containing the preoptic area and hypothala-mus from four human brains were obtained at autopsy andwere immersion fixed in 10% neutral buffered formalin.Following fixation, brains were blocked in the coronalplane and immersed in 20% sucrose in PBS azide. Similarto previous studies from our laboratory, brains were sec-tioned on a freezing microtome at 50 µm (1:24 series) andwere stored in PBS azide at 4°C until immunohistochemi-cal processing (De Lacalle et al., 1993, 1994; Holt et al.,1997; Lim et al., 1997a,b). All subjects died from nonneuro-logic causes and were judged to be normal neurologicallyat the time of death (see Table 1).

Dual-label in situ hybridizationhistochemistry

The protocol for dual-label in situ hybridization histo-chemistry was a modification of that reported previously(Marks et al., 1992, 1993). Antisense MCH and ORXriboprobes were generated from cDNA templates thathave been described previously (Qu et al., 1996; Sakurai etal., 1998). For generation of sense and antisense 35S-labeled cRNA ORX probes, both the sense and antisenseplasmids were linearized by digestion with BamHI andwere subjected to in vitro transcription with T7 polymer-ase according to the manufacturer’s protocols (Promega,Madison, WI). The digoxigenin-labeled MCH probe wasmade by using a plasmid linearized by digestion with XhoIor HindIII for antisense and sense, respectively, and wassubjected to in vitro transcription with SP6 or T7 polymer-ase and 400 µm digoxigenin-labeled-UTP (BoehringerMannheim, Indianapolis, IN); 100 µM unlabeled UTP; and500 µM GTP, ATP, and CTP. The mixture was thendigested with DNAase, and the labeled probe was precipi-tated from unincorporated nucleotides with 4.0 M LiCl and100% ethanol. Rat hypothalamic tissue sections weremounted onto SuperFrost slides (Fisher, Pittsburgh, PA),air dried, and stored in desiccated boxes at 220°C. Prior tohybridization, the slides were immersed in 10% neutralbuffered formalin, incubated in 0.001% proteinase K (Boeh-ringer-Mannheim) for 30 minutes then in 0.25% aceticanhydride for 10 minutes, and dehydrated in ascendingconcentrations of ethanol. Dual labeling was accomplishedby using a 35S-labeled cRNA for ORX and a digoxigenin-labeled cRNA probe for MCH. The ORX probe was dilutedto 106 cpm/ml, and the MCH probe was used at a dilution of

TABLE 1. Summary of Human Cases

Caseno. Age (years)/ sex Cause of death

Postmorteminterval (hours)

H164 50/Female Myeloma 20H184 74/Female Bronchopneumonia 14H186 64/Male Lymphoma 3.5H189 21/Male Endocarditis 14.5

444 C.F. ELIAS ET AL.

1–2 ng/µl in a hybridization solution of 50% formamide; 10mM Tris-HCL (Gibco-BRL, Bethesda, MD), pH 8.0; 0.01%sheared salmon sperm DNA; 0.01% yeast tRNA and 0.05%total yeast RNA (Sigma, St. Louis, MO); 10 mM dithiothre-itol; 10% dextran sulfate; 0.3 M NaCl; 1 mM EDTA, pH 8;and 1 3 Denhardt’s solution (Sigma). The hybridizationcocktail (containing the ORX and MCH probes) and a glasscoverslip were applied to each slide, and sections werethen incubated for 12–16 hours at 56°C. The followingmorning, the coverslips were removed, and the slideswashed four times with 2 3 sodium chloride/sodiumcitrate buffer (SSC). Sections were then incubated in0.002% RNAase A (Boehringer-Mannheim) with 0.5 MNaCl, 10 mM Tris-HCl, pH 8, and 1 mM EDTA for 30minutes at 37°C. Sections were rinsed in decreasingconcentrations of SSC containing 0.25% dithiothreitol(DTT) 3 2 at 50°C for 1 hour, 3 0.2 at 55°C for 1 hour,and 3 0.2 for 1 hour at 60°C.

Sections were next placed in 2 3 SSC including 0.05%Triton X-100 and 2% normal sheep serum for 2 hours.Then, they were incubated in sheep antidigoxigenin pri-mary antisera conjugated to alkaline phosphatase (1:1,000; Boehringer-Mannheim) overnight at room tempera-ture. The sections were washed and then incubated in anitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (Boehringer-Mannheim) chromagensolution (Marks et al., 1992) for 2 hours.

Sections were air dried and then dipped in 3% parlodion(Fisher Scientific; Fair Lawn, NJ) dissolved in isoamylacetate. Slides were air dried overnight and placed in x-rayfilm cassettes with BMR-2 film (Eastman-Kodak, Roches-ter, NY) overnight. Slides were next dipped in NTB2photographic emulsion (Eastman-Kodak), dried, and storedwith desiccant in foil-wrapped slide boxes at 4°C for 5days. Slides were developed with D-19 developer (Eastman-Kodak), then quickly immersed in graded ethanols, clearedin xylene, and coverslipped with Permaslip. Control proce-dures to confirm the specificity of our in situ hybridizationprocedures included hybridization with sense probes andtissue pretreatment with RNAase A (200 µg/ml).

Immunohistochemistry

The procedures for single- and double-label immunohis-tochemistry were performed as reported previously fromour laboratory (Elmquist and Saper, 1996; Sherin et al.,1996; Elmquist et al., 1997; Holt et al., 1997). Rat andhuman hypothalamic sections were pretreated with hydro-gen peroxide (0.3% for rat tissue; 3% for human tissue) andthen incubated in MCH, ORX, or AgRP rabbit primaryantisera or with NPY (Chemicon, City, CA) or a-MSHsheep primary antisera (all at 1:10,000) overnight at roomtemperature. After washing in PBS, sections were incu-bated in biotinylated donkey anti-rabbit immunoglobulin(Ig) G (AgRP, MCH, and ORX) or in biotinylated donkeyanti-sheep IgG (NPY and a-MSH; Jackson Laboratories;1:1,000 for rat tissue and 1:500 for human tissue) for 1hour at room temperature. Primary and secondary anti-sera were diluted in 3% normal donkey serum (JacksonLaboratories) and 0.25% Triton X-100 in PBS containing0.02% sodium azide (PDT) for rat tissue or in 5% powderedmilk in PBS azide/Triton X-100 for human tissue. Sectionswere then incubated with avidin-biotin complex (ABC;Vector Elite Kit; Vector Laboratories, Burlingame, CA;1:500 in PBS for rat tissue and 1:250 for human tissue) for1 hour, rinsed, and incubated in 0.04% diaminobenzidine

tetrahydrochloride (DAB; Sigma) and 0.01% hydrogenperoxide dissolved in PBS. The reaction was terminatedafter 6–10 minutes with two successive rinses in PBS. Thetissue sections were mounted onto subbed slides, air dried,dehydrated in alcohol, cleared in xylene, and then cover-slipped with Permaslip. Prior to coverslipping, the DABreaction product was intensified by using a silver-goldintensification procedure that has been described previ-ously (Breder et al., 1992, 1995; De Lacalle et al., 1993).

For double-label immunohistochemistry, the tissue sec-tions were incubated in NPY, a-MSH, or AgRP antiseraand were processed as described above, except that 0.01%nickel sulfate and 0.01% cobalt chloride (Fisher Scientific)were included in the DAB solution, resulting in a blackreaction product. Sections were then reexposed to 0.3%hydrogen peroxide, rinsed in PBS, incubated in PDT for 2hours, and then incubated in ORX or MCH primaryantiserum overnight at room temperature. The sectionswere processed as described above by using DAB as thechromogen, resulting in a brown reaction product. Thetissue sections were mounted onto subbed slides, dehy-drated in alcohol, cleared in xylene, coverslipped withPermaslip, and analyzed with a Zeiss Axioplan lightmicroscope (Thornwood, NY).

Dual-label in situ hybridizationhistochemistry/immunohistochemistry

The protocol used for combined in situ hybridization andimmunohistochemistry was a modification of that de-scribed previously (Hoffman et al., 1995). Tissue sectionswere processed first for in situ hybridization by using thedigoxigenin-labeled MCH riboprobe and free-floating sec-tions. Tissue was pretreated as described above, exceptthat proteinase K was replaced by a 1% sodium borohy-dride treatment, and sections were not dehydrated inethanols and chloroform but, instead, were rinsed in 2 3SSC. The free-floating sections were hybridized at 37°Covernight and then rinsed in decreasing concentrations ofSSC at 42°C. Sections were next incubated in the antidi-goxigenin primary antisera overnight at room tempera-ture and incubated in alkaline-phosphatase color solutionfor 1 hour as described previously (Marks et al., 1992).Following the chromogen reaction, the sections were rinsedseveral times in PBS and then were processed for ORXimmunohistochemistry as described above by using DABas the chromogen. These procedures resulted in a bluereaction product in neurons that contained MCH mRNAand a brown reaction product in ORX-IR cells.

Immunohistochemical controls

For single-label controls, rat and human hypothalamicsections were incubated in MCH, ORX, AgRP, NPY, orMSH antisera that had been preadsorbed with theirrespective antigens (50 µg/ml of diluted antisera). Allspecific staining was abolished following preadsorptionwith the respective peptide. In addition, negative controlswere generated by omission of each primary antiserum.For double-label controls using the AgRP antisera coupledwith the MCH or ORX antisera (because these threeantisera were made in rabbits), sections were stainedsequentially but with the second (MCH or ORX) antiseraomitted or replaced with normal rabbit serum. This wasnot an issue for the other double-label experiments, be-cause both the NPY and a-MSH antisera were raised in


sheep. These procedures resulted in specific staining onlyfor the first (AgRP) antiserum.

Immunostaining of a-MSH-containing neurons was per-formed with an ovine anti-a-MSH antiserum raised againstan immunogen consisting of a-MSH conjugated to bovinethyroglobulin (J. Tatro and S. Reichlin, unpublished obser-vations). Prior to immunocytochemical studies, antiserumspecificity was determined with radioimmunoassay (RIA)by using iodine 125 (125I)-a-MSH as tracer and assayconditions described earlier (Tatro and Reichlin, 1987).The antiserum is highly specific for a-MSH (with noapparent cross reactivity with MCH) and is dependent onthe amidated C-terminal region for recognition, becauseeither its modification (in a-MSH-free acid), absence (inACTH[1–10]) or its extension (in ACTH) reduced crossreaction in the RIA by at least three to four orders ofmagnitude. Cross reactivities of the antiserum with re-lated peptides, expressed as molar potency of a-MSH withrespect to that of each respective peptide in the RIA[inhibitory concentration (IC) 50 ratio for inhibition of125I-a-MSH binding], were desacetyl-a-MSH, 82–100%;a-MSH-free acid, 0.018%; b-MSH (monkey), , 0.0018%;a-MSH(11–13), g1-MSH, and g2-MSH (each undetectableat 1 µM); ACTH(1–10), , 0.018%; ACTH(1–24), 0.02%;ACTH(1–39) (human), 0.022%; and b-LPH (human),0.022%; b-endorphin (human), and g-endorphin (each un-detectable at 1 µM). In preliminary experiments, theantiserum showed intense, uniform staining of rat pitu-itary intermediate lobe cells that was blocked completelyby preadsorption with excess a-MSH.

Production of photomicrographs

Photomicrographs were produced by capturing imageswith a digital camera (Kodak DCS; Eastman-Kodak)mounted directly on the microscope and a Apple MacintoshPower PC computer (Apple Computers, Cuptertino, CA).Image-editing software (Adobe Photoshop; Adobe Systems,Mountain View, CA) was used to combine photomicro-graphs into plates. Only the sharpness, contrast, andbrightness were adjusted. All figures were printed on adye-sublimation printer (Kodak 8600; Eastman-Kodak).Cytoarchitectonic details were added by using a cameralucida.

RESULTS

MCH and ORX are distinct systems

Rat and mouse brain. By using standard double-labelin situ hybridization methods (Marks et al., 1992), wefound very little colocalization of MCH and ORX mRNA.The MCH mRNA was detected with a digoxigenin-labeledriboprobe, and the ORX mRNA was detected by using a35S-labeled riboprobe and standard autoradiography (Fig.1). In general, there were more MCH neurons than ORXneurons in the rat brain, especially in the most lateralregions, near the internal capsule (Figs. 1, 2). Moremedially, in the perifornical area, the MCH and ORXneurons had a more similar and coextensive distribution.However, careful analysis of the material revealed thatvery few cell bodies containing blue reaction product(MCH mRNA) also contained overlying silver grains (ORXmRNA; Fig. 1). Double-labeled cells could be observed, butthey made up less than 1% of either the MCH populationor the ORX population. Specifically, tissue sections fromthree different rats were examined carefully (14 sections

from each rat). We found a total of ten double-labeled cells(five in the perifornical area, four close to the internalcapsule, and one in the lateral zona incerta) in these threecases. In the three rats examined, only one tissue sectioncontained two double-labeled cells, and most sections didnot contain any double-labeled cells.

We also stained rat (n 5 3) and mouse (n 5 2) hypotha-lamic sections by using a free-floating in situ hybridizationtechnique with the digoxigenin-labeled MCH riboprobefollowed by standard immunohistochemistry for ORX.These experiments resulted in a blue reaction product inneurons containing MCH mRNA and brown ORX-immuno-reactive (IR) cells (Figs. 3A,B, 4). Although both labelswere cytoplasmic, this preparation afforded excellent colorcontrast, allowing assessment of the degree of colocaliza-tion of both labels. In agreement with the dual in situmethod, we found very little colocalization of MCH withORX (Figs. 3A,B, 4; less than 1% of the MCH and ORXpopulations), although rare, double-labeled cells could beobserved. Therefore, both procedures demonstrated that,in the normal rat brain, ORX and MCH neurons aresegregated populations with very little overlap.

Fig. 1. A,B: Dual-labeling in situ hybridization histochemistryreveals that mRNA for melanin-concentrating hormone (MCH) andorexin (ORX) mRNA are in distinct sets of neurons. The neuronscontaining MCH mRNA were hybridized with a riboprobe labeled withdigoxigenin (MCH-Dig; horizontal arrows). The neurons containingclusters of silver grains were hybridized with an ORX 35S-labeledriboprobe (ORX-35S; vertical arrows). B is a higher magnification of Ain the perifornical area of the rat lateral hypothalamic area. fx, fornix.Scale bars 5 200 µm in A, 100 µm in B.


Human brain. Similar to the rat brain, many moreMCH neurons than ORX neurons were observed (Fig. 5).No differences in the distribution and localization of MCH-and ORX-IR cells were observed between male and femalecases. A group of MCH-IR cells was located in the rostralzona incerta and extended from the paraventricular nucleusof the hypothalamus caudally to the posterior hypothala-mus (Figs. 5, 6). Another group of cells abutted the internal

capsule ventrally and dorsally (Fig. 6A–C). A prominentcluster of MCH-IR neurons ran along the entire rostrocau-dal extension of the fornix, surrounding it mainly in adorsomedial and lateral position (Figs. 5, 6). Between thefornix and the third ventricle, smaller numbers of cellswere observed dorsal to the ventromedial nucleus ofhypothalamus. In addition, MCH cells were observedextending back into the posterior hypothalamic area, justabove the mammillary body and close to the third ven-tricle.

The MCH cells located in the perifornical area, dorsal tothe ventromedial nucleus of the hypothalamus (VMH), inthe posterior hypothalamus, and in the core of the LHAwere medium sized and multipolar in shape (Fig. 6D,E),whereas the cells bordering the internal capsule werefusiform. The tissue sections used allowed assessment ofthe MCH-IR innervation of some brain areas. Similar tothe rat brain, we observed MCH-IR fibers in the cingulateand insular cortex (Fig. 6G), the amygdala, and thehippocampus. Dense concentrations of fibers were alsoobserved in the anterior thalamic nucleus, the preopticarea of the hypothalamus, and the mammillary body (Fig.6F).

The highest concentrations of ORX-IR cells were ob-served at the tuberal level of the hypothalamus dorsal tothe fornix in a region that comprises the lateral hypotha-lamic area and the zona incerta (Figs. 5, 7). These neuronswere fusiform and multipolar in shape. Fewer ORX-IRcells than MCH-IR neurons were observed, and the distri-bution of the two chemical types seemed to be complemen-tary, with minimal spatial overlap. At the level of theanterior hypothalamus, many of the ORX-IR cells werelocated ventral to the fornix, moving dorsally at tuberaland mammillary levels (Fig. 5A,B). At the most posteriorlevels, the cells were located in the posterior hypothala-mus, in the zona incerta, along the lenticular fasciculus,and close to the medial edge of the cerebral peduncle (Fig.5C,D). Immunoreactive fibers were found throughout thehypothalamus, including the preoptic area, the anteriorhypothalamic area, the paraventricular nucleus, the ven-tromedial nucleus, the arcuate nucleus, the median emi-nence, and the periventricular area (Fig. 7E). Similar toMCH, the brain sections used allowed the observation ofimmunoreactive fibers in the insular cortex (Fig. 7F) andthe cingulate cortex. The cortical innervation by ORX-IRfibers was less dense than that observed with MCH. Inaddition, we found fibers in the basal nucleus of Meynert(Fig. 7G), the bed nucleus of the stria terminalis, theanterior amygdaloid area, and the dorsomedial and mid-line thalamic nuclei.

MCH- and ORX-IR cell bodies receive NPY,a-MSH, and AgRP innervation

Rat brain. Analysis of the single-label immunohisto-chemical material demonstrated a dense NPY-, a-MSH-,and AgRP-IR innervation of sites that contained MCH-and ORX-IR neurons in the rat hypothalamus (Figs. 8, 9).For the three neuropeptides, the immunoreactive fibers inthe LHA were distributed in a similar pattern, with thedensest innervation in the perifornical area (Figs. 8, 9).Double-label immunohistochemistry in the rat brain re-vealed that MCH and ORX neurons receive a denseinnervation from NPY-, AgRP-, and a-MSH-IR fibers (Fig.3C–H). The NPY, AgRP, and a-MSH terminals had charac-teristic bouton morphology and decorated MCH- and

Fig. 2. A,B: Line drawings of two rostral to caudal levels of the rathypothalamus illustrate the distinct localization patterns of neuronscontaining MCH mRNA and ORX mRNA. The sections were processedfor dual-label in situ hybridization histochemistry. The neuronscontaining MCH mRNA were hybridized with a digoxigenin-labeledriboprobe (MCH-Dig; crosses). The neurons containing ORX mRNAwere hybridized with an ORX 35S-labeled riboprobe (ORX-35S; circles).Very few double-labeled neurons were observed. For abbreviations, seelist. Scale bar 5 1 mm.


Fig. 3. A,B: Immunohistochemistry coupled with in situ hybridiza-tion histochemistry reveals that neurons containing mRNA for MCHand ORX-immunoreactive neuron (ORX-IR) do not colocalize. Theneurons containing MCH mRNA were hybridized with a riboprobelabeled with digoxigenin (blue neurons). The ORX-IR neurons containbrown reaction product. C–N: Series of color photomicrographs illus-trate the innervation of MCH- and ORX-IR neurons in the perifornical

region of the lateral hypothalamic area by agouti-related peptide(AgRP)-, neuropeptide Y (NPY)-, and a-melanin-stimulating hormone(a-MSH)-IR terminals in the rat brain (C–H) and in the human brain(I–N). Black immunoreactive axons can be observed closely associatedwith MCH- and ORX-IR cell bodies and dendrites. Scale bars 5 500µm in A, 100 µm in B, 20 µm in C (also applies to D–N).

ORX-IR dendrites. Apparent somatic appositions werealso observed. This innervation was particularly dense inthe perifornical region of the hypothalamus, because asmost MCH- and ORX-IR cells in this region receivedapparent appositions of NPY-, AgRP-, and a-MSH-IRterminals. The innervation was less intense in morelateral regions near the internal capsule, but many MCHand ORX cells received apparent contacts. It should benoted that NPY-, AgRP-, and a-MSH-IR terminals alsowere seen to outline many cells and dendrites in the

perifornical region that did not contain ORX or MCHimmunoreactivity.

Human brain. Single-label immunohistochemistry re-vealed that AgRP-, NPY-, and a-MSH-IR neurons werepresent in the arcuate (infundibular) nucleus of the hu-man hypothalamus. No differences in the distribution andlocalization of cells were observed between male andfemale subjects. Similar to the rodent, the human AgRP-IRneurons were found in the medial regions of the arcuatenucleus of the hypothalamus (Fig. 10; Ollmann et al.,1997; Shutter et al., 1997). These cells were small andround in shape and were surrounded by a dense net ofimmunoreactive fibers (Fig. 10E). Extremely dense projec-tions could be traced leaving the arcuate nucleus at itsdorsal edge and running along the wall of the thirdventricle to innervate densely the dorsal perifornical area,the lateral hypothalamus, and the paraventricular hypo-thalamic nucleus. AgRP-IR fibers were also observed scat-tered widely within the hypothalamus, including the pre-optic area, the periventricular area, the ventromedialnucleus, and the posterior hypothalamus (Fig. 10B,D,F).Outside the hypothalamus, we observed immunoreactivefibers in the bed nucleus of stria terminalis, anterioramygdaloid area, central nucleus of amygdala, stria termi-nalis, and subthalamic and midline thalamic nuclei. Withrespect to NPY, our findings confirm those of previousstudies using immunohistochemistry, in situ hybridiza-tion, or RIA (Adrian et al., 1983; Pelletier et al., 1984;Brene et al., 1989; Ciofi et al., 1990; Walter et al., 1990). Wefound NPY-IR cells in the medial part of the arcuatenucleus surrounded by a dense plexus of immunoreactivevaricosities. The NPY-IR fibers were scattered throughoutthe hypothalamus, from the preoptic area to the mammil-lary levels. An especially dense innervation was observedin the periventricular area, in the paraventricular nucleus,and around the fornix. Similarly, our findings of a-MSH-IRneurons in human arcuate nucleus confirms previousimmunohistochemical and RIA studies (Desy and Pel-letier, 1978; Parker and Porter, 1979; Gramsch et al.,1980). The cell bodies were dispersed in the arcuatenucleus but were concentrated most in the lateral regionsof the arcuate nucleus (Fig. 11). We found immunoreactivefibers in the preoptic area, in the paraventricular nucleus,in the periventricular area, and in the perifornical area(Fig. 11).

We found that NPY-, AgRP-, and a-MSH-IR fibers inner-vated the human LHA and the perifornical area. Someaxons outlined cell bodies and dendrites. Similar to the ratbrain, innervation of MCH-, and ORX-IR neurons by NPY-,a-MSH-, and AgRP-IR terminals were prominent in theperifornical area of the hypothalamus (Fig. 3I–N). Theinnervation was less intense in more lateral regions of theLHA, but most MCH- and ORX-IR cells received innerva-tion from all three neuropeptides. However, as in the rat,dense NPY-, AgRP-, and a-MSH-IR terminals were alsoobserved on cells that did not contain ORX or MCHimmunoreactivity.

DISCUSSION

In this study, we found that ORX and MCH neuronscomprise distinct populations in the LHA in the rat,mouse, and human brain. Moreover, we found that, in both

Fig. 4. A,B: Line drawings of two rostral-to-caudal levels of the rathypothalamus illustrate the lack of colocalization of MCH mRNA andORX immunoreactivity. The sections were processed for in situhybridization histochemistry followed by immunohistochemistry forORX. The neurons containing MCH mRNA were hybridized with adigoxigenin-labeled riboprobe (MCH-Dig; crosses). The neurons con-taining ORX immunoreactivity are visualized as a brown cytoplasmicreaction product (ORX-IR; circles). Very few double-labeled neuronswere observed. For abbreviations, see list. Scale bar 5 1 mm.


Fig. 5. A–D: Line drawings of four rostral-to-caudal levels of thehuman hypothalamus illustrate the distributions of MCH- and ORX-IRneurons. The drawings were made from adjacent sections from caseH186 that were stained singly for MCH or ORX. Note that theMCH-IR neurons (crosses) and the ORX-IR neurons (circles) are found

in a complementary pattern in the lateral hypothalamic area, theperifornical area of the hypothalamus, and the posterior hypothala-mus. For abbreviations, see list. Scale bars 5 3 mm in A,B, 2 mmin C,D.

Fig. 6. Series of photomicrographs demonstrate that MCH-IRneurons are present in the human hypothalamus. A–C: Three rostral-to-caudal, low-power photomicrographs demonstrate that MCH-IRcells are concentrated in the perifornical and lateral hypothalamicareas. D,E: MCH-IR neurons with prominent dendrites are observedin the perifornical area. E is a higher magnification of D (arrows

indicates orientation). F,G: MCH-IR terminals are present in themammillary body and in the insular cortex. For abbreviations, see list.A–C are at same the magnification, and F and G are at the samemagnification. For abbreviations, see list. Scale bars 5 1.5 mm in C(also applies to A,B), 200 µm in D, 100 µm in E (also applies to F,G).

the rat and the human hypothalamus, MCH and ORXneurons receive terminal appositions from NPY-, AgRP-,and a-MSH-IR fibers. The innervation of LHA neurons bypeptidergic fibers corresponding to leptin-responsive celltypes that reside in the arcuate nucleus may be critical inlinking peripheral metabolic cues to autonomic regulatorysites and the cerebral cortical mantle, providing a neuro-anatomic basis for regulation of feeding behavior.

Technical considerations

Our dual in situ hybridization results indicate that thereis very little overlap of ORX and MCH neurons in thebrains of normally fed rats and mice. However, fastingincreases both MCH and ORX mRNA (Qu et al., 1996;Sakurai et al., 1998). Hence, food restriction may alter thisrelation between MCH and ORX neurons. In addition, it is

Fig. 7. Series of photomicrographs demonstrate that ORX-IRneurons are present in the human hypothalamus. A,B: Two rostral-to-caudal, low-power photomicrographs demonstrate that ORX-IR cellsare found in the perifornical and lateral hypothalamic areas.C,D: ORX-IR neurons are observed in the perifornical area. C and Dare higher magnifications of A and B, respectively (boxes in A and B

indicate orientation). E–G: ORX-IR terminals are present in theperiventricular region of the hypothalamus, in the insular cortex, andin the basal nucleus of Meynert. For abbreviations, see list. Scalebars 5 1 mm in A (also applies to B), 100 µm in C (also applies to D–F),50 µm in G.


possible that our immunohistochemical and in situ hybrid-ization methods are not sensitive enough to account for allneurons containing MCH or ORX.

The double-label immunohistochemical results indicatethat AgRP neurons innervate MCH- and ORX-IR neurons.In these experiments, we used two rabbit antisera for thiscolocalization. However, replacing either the MCH or theORX antisera with normal rabbit serum abolished theMCH and ORX cell body labeling, indicating that it wasnot due to cross reactivity of the secondary antibodies. Inaddition, the patterns of immunoreactive label were di-rectly comparable to single-label material, further suggest-ing antibody specificity.

Our present results suggest that, in both the rat and thehuman hypothalamus, ORX and MCH neurons receiveinnervation from neurons containing AgRP, a-MSH, andNPY. Although this hypothesis is attractive, interpreta-tions must be tempered based on the limitations of thetechniques employed. Without the use of electron micros-copy, it is not possible to demonstrate synaptic contacts ofimmunohistochemically stained neurons. Therefore, theresults from this study will identify only potential syn-apses, and immunocytochemistry coupled with electronmicroscopy will be needed to verify synaptic contacts.Nonetheless, our technique is a high-resolution, lightmicroscopic method that provides information regarding

the topographic distribution of immunohistochemicallyidentified fibers and the detailed morphologic relations ofthose fibers and varicosities with MCH- and ORX-IRneurons.

MCH and ORX neurons comprisedistinct populations

Despite the similar distributions of MCH and ORX cellbodies, we found very little colocalization of MCH mRNAand ORX mRNA or of MCH mRNA and ORX-IR neurons.The lack of colocalization may suggest that ORX and MCHmay affect feeding behavior through distinct neuronalpathways. The MCH neuronal system was identified firstas cells that innervate the cerebral cortex and spinal cordby using retrograde tracing and antisera raised againsta-MSH (Kohler et al., 1984; Saper et al., 1986). Nearly 95%of the cells in the LHA that innervated the cerebral cortexalso stained with the a-MSH antisera. However, it wasapparent that these neurons did not stain with otherantisera for other products of the POMC gene, suggestingthat the antisera cross reacted with an unidentified anti-gen, and the identity of the immunoreactive molecule wasunknown. Subsequently, Bittencourt and colleagues dem-onstrated that these neurons contained MCH by using insitu hybridization and immunohistochemistry (Nahon et

Fig. 8. Series of photomicrographs demonstrate the distribution of MCH-IR (A) and ORX-IR (B)neurons and AgRP-IR (C), a-MSH-IR (D), and NPY-IR (E) fibers in the perifornical area of thehypothalamus in the rat brain. Note the overlapping distribution of MCH- and ORX-IR cells and AgRP-,a-MSH, and NPY-IR fibers. For abbreviations, see list. Scale bar 5 100 µm.


al., 1989; Bittencourt et al., 1992) and that MCH neuronsinnervate the entire neuraxis diffusely, including theentire cortical mantle. Our findings confirm and extendthe findings of other studies that described the localizationof MCH cells in the human hypothalamus using anantiserum raised against salmon MCH or mRNA mapping(Pelletier et al., 1987; Sekiya et al., 1988; Bresson et al.,1989; Mouri et al., 1993; Takahashi et al., 1995; Viale etal., 1997). These findings all suggested that MCH neuronsare in a unique position to regulate physiological processes

involving the cerebral cortex, but functions of central MCHsystems remained obscure. More recently, Qu and col-leagues reported that MCH mRNA was elevated in thebrains of hyperphagic ob/ob mice. In addition, MCH mRNAincreased with fasting in both ob/ob and normal mice (Quet al., 1996). Moreover, MCH administered into the lateralventricle robustly increased food intake (Qu et al., 1996;Rossi et al., 1997).

Two recent studies identified another family of lateralhypothalamic peptides with an MCH-like anatomic distri-bution of cell bodies within the brain. The orexins (Sakuraiet al., 1998), also called hypocretin (de Lecea et al., 1998),are found exclusively in cell bodies of the LHA, and,similar to MCH, fibers are found throughout the brain.Moreover, food restriction increased ORX mRNA, andintracerebroventricular injections of ORX increased feed-ing behavior (Sakurai et al., 1998). Although a systematicanalysis of the sites receiving ORX-IR innervation has notappeared, our observations indicate that ORX neuronsalso innervate the cerebral cortex as well as centralautonomic and limbic sites. Thus, the mammalian MCHand ORX systems seem to be positioned ideally to affectfeeding behavior, and MCH and ORX may provide a linkbetween the hypothalamus and the cerebral cortex andkey autonomic regulatory nuclei (Saper et al., 1986; Bitten-court et al., 1992; Elias and Bittencourt, 1997; de Lecea etal., 1998; Sakurai et al., 1998).

Central NPY and melanocortin systems:Role in body weight regulation

Several recent pieces of evidence point to a critical rolefor brain NPY and melanocortin systems in the regulationof food intake and body weight and in responding tocirculating leptin. Evidence indicates that NPY neurons inthe arcuate nucleus regulate food intake and are targets ofleptin. Expression of NPY mRNA in the arcuate nucleus isincreased in response to fasting in normal rats (whenleptin levels rapidly fall) and is markedly increased inob/ob and db/db mice (for review, see Spiegelman and Flier,1996). Repletion of leptin in ob/ob mice and fasted ratssuppresses this elevated NPY expression in the arcuatenucleus (Stephens et al., 1995;Ahima et al., 1996; Schwartzet al., 1996). Targeted deletion of the NPY gene partiallyameliorates the ob/ob phenotype, providing evidence for arole of NPY neurons in regulating food intake in responseto leptin (Erickson et al., 1996). The colocalization of leptinreceptor mRNA and NPY mRNA in neurons in the arcuatenucleus also supports this hypothesis (Mercer et al.,1996a). Presently, it is not clear through which receptorsubtype NPY exerts its effects on feeding. However, onelikely candidate is the Y5 receptor (Gerald et al., 1996). Itis noteworthy that the perifornical region of the LHAcontains high levels of Y5 mRNA (Gerald et al., 1996) andis the most sensitive site at which NPY administrationincreases feeding behavior (Stanley et al., 1993). It shouldbe noted that, although the source of the NPY innervationof MCH and ORX neurons likely includes cells in thearcuate nucleus, NPY neurons are found throughout thebrain. In fact, some of the hypothalamic NPY innervationmay be from cells in the ventral lateral medulla (Saw-chenko et al., 1985) that are known to innervate thehypothalamus. Nonetheless, these findings suggest thatthe projections from the arcuate NPY neurons to the LHAmay play an integral role in regulating body weight.

Fig. 9. A series of photomicrographs demonstrating the AgRP-IR(A), a-MSH-IR (B), and NPY-IR (C) innervation of the lateral hypotha-lamic area of the rat. Note the dense innervation of the perifornicalarea. For abbreviations, see list. Scale bar 5 400 µm.


Fig. 10. Series of photomicrographs demonstrating that AgRP-IRneurons are present in the human hypothalamus. A,B: Two rostral-to-caudal, low-power photomicrographs demonstrate that AgRP-IR neu-rons localize to the arcuate nucleus of the hypothalamus (ARC). In B,immunoreactive fibers are also observed streaming dorsally out of thearcuate. C,E: AgRP-IR neurons are observed in the arcuate nucleus. Cis a higher magnification of B, and E is a higher magnification of C (box

in C indicates orientation). D,F: AgRP-IR terminals are observed inthe perifornical region of the human hypothalamus. A and B are atsame magnification, C and D are at same magnification, and E and Fare at same magnification. For abbreviations, see list. Scale bars 5 2mm in B (also applies to A), 200 µm in D (also applies to C), 100 µm in F(also applies to E).

MC4-R also has been identified as critical in the regula-tion of body weight. The yellow (Ay) mouse has ectopicoverexpression (including the brain) of the MC4-R antago-nist, agouti protein (Fan et al., 1997). Melanocortin recep-tor blockade by the agouti protein in the brains of thesemice produces an adult-onset obesity with leptin resis-tance. In addition, targeted deletion of the MC4-R resultsin an obesity syndrome that is indistinguishable from thatin the Ay mouse (Huszar et al., 1997). Finally, transgenicoverexpression of AgRP, which presumably antagonizesthe MC4-R, results in obesity (Ollmann et al., 1997).Potential ligands for the MC4-Rs in the brain includea-MSH (and related melanocortins) and AgRP, both ofwhich are found in neurons in the arcuate nucleus of thehypothalamus (Fan et al., 1997; Ollmann et al., 1997). Inaddition, POMC and leptin receptor mRNA colocalizewithin the arcuate nucleus (Cheung et al., 1997). More-over, leptin regulates the levels of POMC mRNA, becauseob/ob mice and food restricted rats have lowered levels ofarcuate POMC mRNA compared with controls. Repletionof leptin elevates arcuate POMC mRNA levels to those ofcontrols (Schwartz et al., 1997; Thornton et al., 1997;Mizuno et al., 1998). Similarly, AgRP mRNA is elevated inleptin-deficient ob/ob mice and in leptin-resistant db/dbmice (Shutter et al., 1997). It is interesting that the acuteeffects of leptin on feeding can be attenuated by blocking

melanocortin receptors (Seeley et al., 1997). Therefore,some of the biologic effects of leptin likely are due toantagonistic actions of a-MSH and AgRP acting on sub-populations of neurons containing melanocortin receptors.MC4-R mRNA and MSH binding sites are located in theparaventricular hypothalamic nucleus, the dorsomedialhypothalamic nucleus, and the LHA (Tatro, 1990; Mount-joy et al., 1994).

It is clear that NPY, a-MSH, and AgRP neurons play animportant role in regulating body weight, but the specificpopulations of neurons through which they exert theireffects are not yet well understood. Our results suggestthat two important populations may be ORX- and MCH-IRneurons in the LHA, a region that contains MC4 and Y5receptor mRNA as well (Gerald et al., 1996; Fan et al.,1997; Huszar et al., 1997). At present, it is not knownwhether ORX- and MCH-containing neurons also containMC4 or Y5 receptors, but this certainly deserves carefulanalysis.

Linking the mediobasal hypothalamusand the lateral hypothalamic area:

The dual-center hypothesis revisited

Since the middle of this century, it has been suspectedthat distinct populations of hypothalamic neurons exist

Fig. 11. A–D: Series of photomicrographs demonstrating thata-MSH-IR neurons are present in the human hypothalamus. A:a-MSH-IR neurons localize to the arcuate nucleus of hypothalamus.Note the clusters of cells in the lateral part of the nucleus. B:

a-MSH-IR fibers are dense in the perifornical area of the hypothala-mus. C is a higher magnification of A (box in A indicates orientation),and D is a higher magnification of B. For abbreviations, see list. Scalebars 5 200 µm in A (also applies to B), 100 µm in C; 50 µm in D.


that have differential effects on food intake and bodyweight (Hetherington and Ranson, 1940; Anand and Bro-beck, 1951). The ‘‘VMH syndrome’’ could be produced bylesions of cell groups of the mediobasal hypothalamus thatcentered on the ventromedial nucleus. This syndrome ischaracterized by morbid obesity, marked hyperphagia, andcentral hypogonadism. The neurochemical and neuroana-tomical basis for this constellation of effects has remainedunclear. However, the phenotype of leptin-deficient ob/obmice could be characterized as a constellation of thephenotypes seen in various models that damage the medio-basal ‘‘satiety centers.’’ It is noteworthy that the anatomicdistribution of leptin receptors in the mediobasal hypo-thalamus mirrors the physiologically effective lesion sites.Specifically, dense collections of leptin receptor mRNA arefound in the arcuate, ventromedial, and dorsomedial hypo-thalamic nuclei (Mercer et al., 1996b; Schwartz et al.,1996; Fei et al., 1997; Elmquist et al., 1998a,b). Moreover,intravenous administration of leptin to fed rats activatesseveral nuclear groups in the rat brain, including theventromedial and dorsomedial hypothalamic nuclei andthe lateral divisions of the arcuate nucleus (Elmquist etal., 1997, 1998a,b). These findings all suggest that theintegrity of the mediobasal hypothalamus is essential fornormal body weight regulation and that ablation of leptinreceptors and leptin signaling in combinations of any ofthese sites may explain the ‘‘VMH syndrome.’’

Despite the increasing understanding of the role of themediobasal hypothalamus in regulating food intake andbody weight, the role of the LHA is still debated (Bernardisand Bellinger, 1996). Undoubtedly, projections to auto-nomic and endocrine regulatory sites of medial cell groups,such as the paraventricular and dorsomedial nuclei, areimportant in these processes. However, additional path-ways that innervate higher cortical structures also mustplay a role in the complex regulation of feeding and bodyweight. The LHA has long been considered to be a regionthat regulates food intake, because cell-specific lesions ofthis region can result in decreased food intake and bodyweight (Stricker and Verbalis, 1990; Bernardis and Bell-inger, 1996). Attractive candidates for regulating foodintake are the ORX and MCH cells, because both increasefeeding after icv injections and increase expression of theirmRNA during fasting. Thus, it is plausible that destruc-tion of these neuronal populations may explain the effectson food intake following lesions of the LHA ‘‘feedingcenters.’’

Our present findings suggest that populations of AgRP,a-MSH, and NPY neurons in the arcuate nucleus mayprovide a missing link between mediobasal hypothalamicsatiety and LHA phagic centers. Leptin receptors arepresent in the LHA but at much lower levels than medio-basal cell groups (Fei et al., 1997; Elmquist et al., 1998a,b).Leptin receptor mRNA colocalizes in neurons that alsocontain NPY and a-MSH (Mercer et al., 1996b; Cheung etal., 1997). Taken together, these findings suggest thatdistinct populations of leptin receptor-containing neuronsin the arcuate nucleus that also express NPY, AgRP, orPOMC are in a position to regulate the activity of MCHand ORX cells, which, in turn, directly innervate criticalregions of the amygdala, the thalamus, autonomic pregan-glionic neurons, and the cerebral cortex, thus linking themediobasal satiety and lateral hypothalamic feeding cen-ters. These projections may underlie some of the extremely

complex responses associated with hunger, food intake,and satiety.

ACKNOWLEDGMENTS

The authors thank Quan Ha for expert technical assis-tance and Dr. Seymour Reichlin for advice on the genera-tion of anti-MSH antiserum. We also thank Dr. RobertSteiner for assistance and advice in the optimizing of thedual-labeling in situ hybridization procedures. J.K.E. re-ceived research support from Eli Lilly and Company, andC.F.E. received a postdoctoral fellowship from FAPESP(96/7884–4), Sao Paulo, Brazil. G.S.B. and M.Y. are Inves-tigators and T.S. is an Associate of the Howard HughesMedical Institute.

LITERATURE CITED

Adrian TE, Allen JM, Bloom SR, Ghatei MA, Rossor MN, Roberts GW, CrowTJ, Tatemoto K, Polak JM. 1983. Neuropeptide Y distribution in humanbrain. Nature 306:584–586.

Ahima RS, Prabakaran D, Mantzoros C, Qu D, Lowell B, Maratos-Flier E,Flier JS. 1996. Role of leptin in the neuroendocrine response to fasting.Nature 382:250–252.

Anand BK, Brobeck JR. 1951. Localization of a ‘‘feeding center’’ in thehypothalamus of the rat. Proc Soc Exp Biol Med 77:323–324.

Bernardis LL, Bellinger LL. 1987. The dorsomedial hypothalamic nucleusrevisited: 1986 update. Brain Res 434:321–381.

Bernardis LL, Bellinger LL. 1996. The lateral hypothalamic area revisited:Ingestive behavior. Neurosci Biobehav Rev 20:189–287.

Bittencourt JC, Presse F, Arias C, Peto C, Vaughan J, Nahon JL, Vale W,Sawchenko PE. 1992. The melanin-concentrating hormone system ofthe rat brain: an immuno- and hybridization histochemical characteriza-tion. J Comp Neurol 319:218–245.

Breder CD, Smith WL, Raz A, Masferrer J, Seibert K, Needleman P, SaperCB. 1992. Distribution and characterization of cyclooxygenase immuno-reactivity in the ovine brain. J Comp Neurol 322:409–438.

Breder CD, Dewitt D, Kraig RP. 1995. Characterization of induciblecyclooxygenase in rat brain. J Comp Neurol 355:296–315.

Brene S, Lindefors N, Kopp J, Sedvall G, Persson H. 1989. Regionaldistribution of neuropeptide Y mRNA in postmortem human brain.Brain Res Mol Brain Res 6:241–249.

Bresson JL, Clavequin MC, Fellmann D, Bugnon C. 1989. Human hypotha-lamic neuronal system revealed with a salmon melanin- concentratinghormone (MCH) antiserum. Neurosci Lett 102:39–43.

Bronstein DM, Schafer MK, Watson SJ, Akil H. 1992. Evidence thatbeta-endorphin is synthesized in cells in the nucleus tractus solitarius:Detection of POMC mRNA. Brain Res 587:269–275.

Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P. 1995. Recombinantmouse OB protein: Evidence for a peripheral signal linking adiposityand central neural networks. Science 269:546–549.

Cheung CC, Clifton DK, Steiner RA. 1997. Proopiomelanocortin neuronsare direct targets for leptin in the hypothalamus. Endocrinology138:4489–4492.

Ciofi P, Tramu G, Bloch B. 1990. Comparative immunohistochemical studyof the distribution of neuropeptide Y, growth hormone-releasing factorand the carboxyterminus of precursor protein GHRF in the humanhypothalamic infundibular area. Neuroendocrinology 51:429–436.

De Lacalle S, Hersh LB, Saper CB. 1993. Cholinergic innervation of thehuman cerebellum. J Comp Neurol 328:364–376.

De Lacalle S, Lim C, Sobreviela T, Mufson EJ, Hersh LB, Saper CB. 1994.Cholinergic innervation in the human hippocampal formation includ-ing the entorhinal cortex. J Comp Neurol 345:321–344.

de Lecea L, Kilduff TS, Peyron C, Gao X, Foye PE, Danielson PE, FukuharaC, Battenberg ELF, Gautvik VT, Bartlett FS, et al. 1998. The hypocre-tins: Hypothalamus-specific peptides with neuroexcitatory activity.Proc Natl Acad Sci USA 95:322–327.

Debons AF, Krimsky I, Maayan ML, Fani K, Jemenez FA. 1977. Goldthioglucose obesity syndrome. Fed Proc 36:143–147.

Desy L, Pelletier G. 1978. Immunohistochemical localization of alpha-melanocyte stimulating hormone (alpha-MSH) in the human hypothala-mus. Brain Res 154:377–381.


Elias CF, Bittencourt JC. 1997. Study of the origins of melanin-concentrating hormone and neuropeptide EI immunoreactive projec-tions to the periaqueductal gray matter. Brain Res 755:255–271.

Elmquist JK, Saper CB. 1996. Activation of neurons projecting to theparaventricular hypothalamic nucleus by intravenous lipopolysaccha-ride. J Comp Neurol 374:315–331.

Elmquist JK, Ahima RS, Maratos-Flier E, Flier JS, Saper CB. 1997. Leptinactivates neurons in ventrobasal hypothalamus and brainstem. Endo-crinology 138:839–842.

Elmquist JK, Ahima RS, Elias CF, Flier JS, Saper CB. 1998a. Leptinactivates distinct projections from the dorsomedial and ventromedialhypothalamic nuclei. Proc Natl Acad Sci USA 95:741–746.

Elmquist JK, Bjorbaek C, Ahima RS, Flier JS, Saper CB. 1998b. Distribu-tions of leptin receptor mRNA isoforms in the rat brain. J Comp Neurol395:535–547.

Erickson JC, Hollopeter G, Palmiter RD. 1996. Attenuation of the obesitysyndrome of ob/ob mice by the loss of neuropeptide Y. Science 274:1704–1707.

Fan W, Boston BA, Kesterson RA, Hruby VJ, Cone RD. 1997. Role ofmelanocortinergic neurons in feeding and the agouti obesity syndrome.Nature 385:165–168.

Fei H, Okano HJ, Li C, Lee GH, Zhao C, Darnell R, Friedman JM. 1997.Anatomic localization of alternatively spliced leptin receptors (Ob-R) inmouse brain and other tissues. Proc Natl Acad Sci USA 94:7001–7005.

Gerald C, Walker MW, Criscione L, Gustafson EL, Batzl-Hartmann C,Smith KE, Vaysse P, Durkin MM, Laz TM, Linemeyer DL, SchaffhauserAO, Whitebread S, Hofbauer KG, Taber RI, Branchek TA, WeinshankRL. 1996. A receptor subtype involved in neuropeptide-Y-induced foodintake. Nature 382:168–171.

Gold RM. 1973. Hypothalamic obesity: The myth of the ventromedialnucleus. Science 182:488–490.

Gramsch C, Kleber G, Hollt V, Pasi A, Mehraein P, Herz A. 1980.Pro-opiocortin fragments in human and rat brain: beta-endorphin andalpha-MSH are the predominant peptides. Brain Res 192:109–119.

Halaas JL, Gajiwala KS, Maffei M, Cohen SL, Chait BT, Rabinowitz D,Lallone RL, Burley SK, Friedman JM. 1995. Weight-reducing effects ofthe plasma protein encoded by the obese gene. Science 269:543–546.

Hetherington AW, Ranson SW. 1940. Hypothalamic lesions and adiposity inthe rat. Anat Rec 78:149–172.

Hoffman GE, Berghorn KA, Knapp LT, Le WW, Sherman TG. 1995.Physiological stimulation of vasopressin and oxytocin neurons: perspec-tives from Fos activation. In: Saito T, Kurokawa K, Yoshida S, editors.Neurohypophysis: recent progress of vasopressin and oxytocin re-search. Amsterdam: Elsevier, p 151–164.

Holt DJ, Graybiel AM, Saper CB. 1997. Neurochemical architecture of thehuman striatum. J Comp Neurol 384:1–25.

Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q, Berke-meier LR, Gu W, Kesterson RA, Boston BA, Cone RD, Smith FJ,Campfield LA, Burn P, Lee F. 1997. Targeted disruption of the melanocor-tin-4 receptor results in obesity in mice. Cell 88:131–141.

Jacobowitz DM, O’Donohue TL. 1978. Alpha-Melanocyte stimulating hor-mone: Immunohistochemical identification and mapping in neurons ofrat brain. Proc Natl Acad Sci USA 75:6300–6304.

Joseph SA, Pilcher WH, Bennett-Clarke C. 1983. Immunocytochemicallocalization of ACTH perikarya in nucleus tractus solitarius: Evidencefor a second opiocortin neuronal system. Neurosci Lett 38:221–225.

Kohler C, Haglund L, Swanson LW. 1984. A diffuse alpha MSH-immunoreactive projection to the hippocampus and spinal cord fromindividual neurons in the lateral hypothalamic area and zona incerta. JComp Neurol 223:501–514.

Lim C, Blume HW, Madsen JR, Saper CB. 1997a. Connections of thehippocampal formation in humans: I. The mossy fiber pathway. J CompNeurol 385:325–351.

Lim C, Mufson EJ, Kordower JH, Blume HW, Madsen JR, Saper CB. 1997b.Connections of the hippocampal formation in humans: II. The endfolialfiber pathway. J Comp Neurol 385:352–371.

Ludwig DS, Mountjoy KG, Tatro JB, Gillette JA, Frederich RC, Flier JS,Maratos-Flier E. 1998. Melanin-concentrating hormone: a functionalmelanocortin antagonist in the hypothalamus. Am J Physiol 274:E627–E633.

Marks DL, Wiemann JN, Burton KA, Lent KL, Clifton DK, Steiner RA.1992. Simultaneous visualization of two cellular mRNA species inindividual neurons by use of a new double in situ hybridization method.Mol Cell. Neurosci 3:395–405.

Marks DL, Smith MS, Clifton DK, Steiner RA. 1993. Regulation ofgonadotropin-releasing hormone (GnRH) and galanin gene expression

in GnRH neurons during lactation in the rat. Endocrinology 133:1450–1458.

Mercer JG, Hoggard N, Williams LM, Lawrence CB, Hannah LT, MorganPJ, Trayhurn P. 1996a. Coexpression of leptin receptor and preproneu-ropeptide Y mRNA in arcuate nucleus of mouse hypothalamus. JNeuroendocrinol 8:733–735.

Mercer JG, Hoggard N, Williams LM, Lawrence CB, Hannah LT, TrayhurnP.1996b. Localization of leptin receptor mRNA and the long form splicevariant (Ob-Rb) in mouse hypothalamus and adjacent brain regions byin situ hybridization. FEBS Lett 387:113–116.

Mizuno TM, Kleopoulos SP, Bergen HT, Roberts JL, Priest CA, Mobbs CV.1998. Hypothalamic pro-opiomelanocortin mRNA is reduced by fastingin ob/ob and db/db mice, but is stimulated by leptin. Diabetes 47:294–297.

Mountjoy KG, Mortrud MT, Low MJ, Simerly RB, Cone RD. 1994. Localiza-tion of the melanocortin-4 receptor (MC4-R) in neuroendocrine andautonomic control circuits in the brain. Mol Endocrinol 8:1298–1308.

Mouri T, Takahashi K, Kawauchi H, Sone M, Totsune K, Murakami O, ItoiK, Ohneda M, Sasano H, Sasano N. 1993. Melanin-concentratinghormone in the human brain. Peptides 14:643–646.

Nahon JL, Presse F, Bittencourt JC, Sawchenko PE, Vale W. 1989. The ratmelanin-concentrating hormone messenger ribonucleic acid encodesmultiple putative neuropeptides coexpressed in the dorsolateral hypo-thalamus. Endocrinology 125:2056–2065.

Ollmann MM, Wilson BD, Yang YK, Kerns JA, Chen Y, Gantz I, Barsh GS.1997. Antagonism of central melanocortin receptors in vitro and in vivoby agouti-related protein. Science 278:135–138.

Olney JW. 1969. Brain lesions, obesity, and other disturbances in micetreated with monosodium glutamate. Science 164:719–721.

Parker CR Jr., Porter JC. 1979. Subcellular localization of immunoreactivealpha-melanocyte stimulating hormone in human brain. Brain Res Bull4:535–538.

Pelletier G, Desy L, Kerkerian L, Cote J. 1984. Immunocytochemicallocalization of neuropeptide Y (NPY) in the human hypothalamus. CellTissue Res 238:203–205.

Pelletier G, Guy J, Desy L, Li S, Eberle AN, Vaudry H. 1987. Melanin-concentrating hormone (MCH) is colocalized with alpha-melanocyte-stimulating hormone (alpha-MSH) in the rat but not in the humanhypothalamus. Brain Res 423:247–253.

Pelleymounter MA, Cullen MJ, Baker MB, Hecht R, Winters D, Boone T,Collins F. 1995. Effects of the obese gene product on body weightregulation in ob/ob mice. Science 269:540–543.

Qu D, Ludwig DS, Gammeltoft S, Piper M, Pelleymounter MA, Cullen MJ,Mathes WF, Przypek R, Kanarek R, Maratos-Flier E. 1996. A role formelanin-concentrating hormone in the central regulation of feedingbehaviour. Nature 380:243–247.

Rossi M, Choi SJ, O’Shea D, Miyoshi T, Ghatei MA, Bloom SR. 1997.Melanin-concentrating hormone acutely stimulates feeding, but chronicadministration has no effect on body weight. Endocrinology 138:351–355.

Sakurai T, Amemiya A, Ishii M, Matsuzaki I, Chemelli RM, Tanaka H,Williams SC, Richardson JA, Kozlowski GP, Wilson S, et al. 1998.Orexins and orexin receptors: A family of hypothalamic neuropeptidesand G protein-coupled receptors that regulate feeding behavior. Cell92:573–585.

Saper CB. 1985. Organization of cerebral cortical afferent systems in therat. II. Hypothalamocortical projections. J Comp Neurol 237:21–46.

Saper CB, Akil H, Watson SJ. 1986. Lateral hypothalamic innervation ofthe cerebral cortex: immunoreactive staining for a peptide resemblingbut immunochemically distinct from pituitary/arcuate alpha-melano-cyte stimulating hormone. Brain Res Bull 16:107–120.

Sawchenko PE, Swanson LW, Grzanna R, Howe PR, Bloom SR, Polak JM.1985. Colocalization of neuropeptide Y immunoreactivity in brainstemcatecholaminergic neurons that project to the paraventricular nucleusof the hypothalamus. J Comp Neurol 241:138–153.

Schwartz MW, Seeley RJ, Campfield LA, Burn P, Baskin DG. 1996.Identification of targets of leptin action in rat hypothalamus. J ClinInvest 98:1101–1106.

Schwartz MW, Seeley RJ, Woods SC, Weigle DS, Campfield LA, P. Burn P,Baskin DG. 1997. Leptin increases hypothalamic pro-opiomelanocortinmRNA expression in the rostral arcuate nucleus. Diabetes 46:2119–2123.

Seeley RJ, Yagaloff KA, Fisher SL, Burn P, Thiele TE, van Dijk G, BaskinDG, Schwartz MW. 1997. Melanocortin receptors in leptin effects.Nature 390:349.


Sekiya K, Ghatei MA, Lacoumenta S, Burnet PW, Zamir N, J. Burrin JM,Polak JM, Bloom SR. 1988. The distribution of melanin-concentratinghormone-like immunoreactivity in the central nervous system of rat,guinea-pig, pig and man. Neuroscience 25:925–930.

Sherin JE, Shiromani PJ, McCarley RW, Saper CB. 1996. Activation ofventrolateral preoptic neurons during sleep. Science 271:216–219.

Shutter JR, Graham M, Kinsey AC, Scully S, Luthy R, Stark KL. 1997.Hypothalamic expression of ART, a novel gene related to agouti, isup-regulated in obese and diabetic mutant mice. Genes Dev 11:593–602.

Simmons DM, Arriza JL, Swanson LW. 1989. A complete protocol for in situhybridization of messenger RNAs in brain and other tissues withradiolabelled single stranded RNA probes. J Histotech 12:169–181.

Spiegelman BM, Flier JS. 1996. Adipogenesis and obesity: Rounding outthe big picture. Cell 87:377–389.

Stanley BG, Magdalin W, Seirafi A, Thomas WJ, Leibowitz SF. 1993. Theperifornical area: the major focus of (a) patchily distributed hypotha-lamic neuropeptide Y-sensitive feeding system(s). Brain Res 604:304–317.

Stephens TW, Basinski M, Bristow PK, Bue-Valleskey JM, Burgett SG,Craft L, Hale J, Hoffmann J, Hsiung HM, Kriauciunas A, MaKellar W,Roseck PR Jr, Schoner B, Smith D, Tinsley FC, Zhang XY, Heiman M.1995. The role of neuropeptide Y in the antiobesity action of the obesegene product. Nature 377:530–532.

Stricker EM. 1984. Biological bases of hunger and satiety: Therapeuticimplications. Nutr Rev 42:333–340.

Stricker EM, Verbalis JG.1990. Control of appetite and satiety: Insightsfrom biologic and behavioral studies. Nutr Rev 48:49–56.

Takahashi K, Suzuki H, Totsune K, Murakami O, Satoh F, Sone M, SasanoH, Mouri T, Shibahara S. 1995. Melanin-concentrating hormone inhuman and rat. Neuroendocrinology 61:493–498.

Tatro JB. 1990. Melanotropin receptors in the brain are differentiallydistributed and recognize both corticotropin and alpha-melanocytestimulating hormone. Brain Res 536:124–132.

Tatro JB, Reichlin S. 1987. Specific receptors for alpha-melanocyte-stimulating hormone are widely distributed in tissues of rodents.Endocrinology 121:1900–1907.

Thornton JE, Cheung CC, Clifton DK, Steiner RA. 1997. Regulation ofhypothalamic proopiomelanocortin mRNA by leptin in ob/ob mice.Endocrinology 138:5063–5066.

Tsujii S, Bray GA. 1989. Acetylation alters the feeding response to MSHand beta-endorphin. Brain Res Bull 23:165–169.

Viale A, Zhixing Y, Breton C, Pedeutour F, Coquerel A, Jordan D, Nahon JL.1997. The melanin-concentrating hormone gene in human: Flankingre-gion analysis, fine chromosome mapping, and tissue-specific expres-sion. Brain Res Mol Brain Res 46:243–255.

Walter A, Mai JK, Jimenez-Hartel W. 1990. Mapping of neuropeptide Y-likeimmunoreactivity in the human forebrain. Brain Res Bull 24:297–311.

Watson SJ, Akil H. 1979. The presence of two alpha-MSH positive cellgroups in rat hypothalamus. Eur J Pharmacol 58:101–103.

Watson SJ, Akil H, Richard CW III, Barchas JD. 1978. Evidence for twoseparate opiate peptide neuronal systems. Nature 275:226–228.

Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. 1994.Positional cloning of the mouse obese gene and its human homologue.Nature 372:425–432.


chemically defined projections linking the mediobasal hypothalamus and the lateral hypothalamic area

Documents