The proopiomelanocortin (POMC) protein is a fascinat-ing molecule. It is the precursor of a number of peptide

hormones that have powerful effects on physiology,including b-endorphin, adrenocorticotrophin (ACTH)and a number of different melanocyte-stimulating hor-mones (MSHs). The full panoply of targets of POMCderivatives should be revealed in patients with a geneticdeficiency of POMC, two of whom have been described1.Both lack all POMC derivatives, except possibly gMSH,and both have adrenal insufficiency (through lack ofACTH), have red hair and are clinically obese. These lattertwo phenotypes, surprising though it may seem, are bothcaused by the lack of a 13-amino acid peptide, aMSH.Experiments initiated in the mouse have demonstratedthat aMSH acts through a number of different receptors(MC1R–MC5R) to elicit different effects. Pigment cellsexpress the melanocortin-1 receptor, MC1R, stimulationof which is necessary for the synthesis of the dark eumelanin that colours skin and hair. Absence of thereceptor results in the production of yellow or red melanin(mice mutant at Mc1r have yellow fur and most red-hairedhumans have variants in the MC1R gene2–4). MC4R isexpressed in the hypothalamus, where it appears to regu-late feeding behaviour, and mice or humans with one ortwo mutant MC4R genes are obese5–7. The effects ofaMSH at these receptors can be modulated by the en-dogenous antagonist proteins agouti and agouti-relatedprotein (AGRP). Agouti is normally produced in the hairfollicle, where it antagonizes MC1R on nearby melanocytes,resulting in the production of yellow phaeomelanin; nor-mal mouse hair is coloured with alternate bands of blackand yellow melanin, resulting from the transient expres-sion of agouti midway through the hair-growth cycle8,9.Some mouse mutants overexpress agouti throughout thebody, and these animals are both yellow and obese,because of antagonism of MC1R and MC4R (Ref. 10).AGRP is made in the hypothalamus and is presumed to bethe naturally occurring antagonist of MC4R, which modu-lates aMSH-regulated feeding behaviour; the overexpres-sion of AGRP in transgenic animals results in increasedweight, mimicking loss of MC4R (Refs 11, 12; Fig. 1).

Recently published work now brings a third proteininto the equation. Mice that are mutant at the mahogany(mg) locus are darkly pigmented because the agouti-induced yellow band on the hair is suppressed13. Themutation also suppresses the yellow fur and the obesity

caused by the overexpression of agouti in mutant ani-mals13,14. Furthermore, mahogany mice are abnormal intheir feeding behaviour, motor activity and metabolic rate,even in the absence of agouti overexpression14. Normalmice that are fed a diet containing .40% fat (appropri-ately named ‘Western chow’) gain weight when comparedwith mice on a diet of ,10% fat, despite an equal intakeof calories. By contrast, mahogany mice appear to beresistant to the effects of dietary fat15; in fact, mahoganymice eat more than normal mice but, nevertheless, do notgain weight, presumably because they have an increasedbasal metabolic rate and increased motor activity14.

Such an interesting gene product is a promising targetfor positional-cloning efforts and two groups have nowpublished their successes15,16. Both groups made largecrosses that enabled the mg locus to be localized within afew hundred kilobases. Sequencing of the clones that spanthe locus shows that the mutation lies within a gene-richregion; many of these candidates were examined in orderto identify the mg locus. The locus encodes a protein ofabout 150 kD and definitive evidence that this candidate isthe mahogany gene is a 5 bp deletion and frameshift in themg3J mutation15. Two other mutant alleles, mg and mgL,affect the transcription or splicing of the gene, and bothappear to be caused by insertions of about 5 kb into different introns16. The mg3J phenotype is more severe thanmg or mgL, in accordance with its probable loss of func-tion compared with the reduced level of mRNA in themilder phenotypes.

The mg gene is widely expressed and has multiple iso-forms; at least one of these forms encodes a protein with atransmembrane domain, and others encode proteins thatare apparently secreted. The intracellular part of the pro-tein does not contain any recognizable signalling motifs,which suggests that its action is entirely at the cell surface.Numerous motifs can be identified in the extracellularsequence, such as several EGF domains, a CUB domainand a C-type lectin domain. The first of these is frequentlyfound in proteins that interact with other proteins, whilethe latter two suggest that the protein interacts withcarbohydrate moieties. However, it is not easy to makefunctional predictions from these structural motifs. Rathermore revealing are homologues of this protein in otherspecies. There is a good homologue in C. elegans,although it seems to be missing some of the structural features, and no function has been ascribed to it.

The mahogany mouse mutationfurther links between pigmentation, obesityand the immune system

OutlookCOMMENTThe mahogany mouse mutation

TIG November 1999, volume 15, No. 110168-9525/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01824-7

Ian J. Jacksonian.jackson@hgu.mrc.ac.uk

MRC Human GeneticsUnit, Western GeneralHospital, Crewe Road,Edinburgh, UK EH4 2XU.

429

Completely different lines of experimentation have identified attractin, a protein that seems to have multipleroles in regulating physiological processes. It affects the balance between agonist and antagonist at receptorson melanocytes, modifies behaviour and basal metabolic rate, and mediates an interaction between activated T cells and macrophages. It may well be a target for development of drugs to treat obesity.

However, the human orthologue of the mahogany locusproduct has recently been identified as a protein termedattractin17. The identification of attractin was by a com-pletely different route; it was initally thought to be a sol-uble form of the T-cell antigen CD26 but, in fact, it has nosequence similarity to this protein. Nevertheless, it doesseem to have one property in common with CD26, whichis that it has dipeptidyl peptidase activity: it is able tocleave dipeptides from the N-terminus of target proteins18.Attractin was characterized as a serum protein, but clearlyit occurs in both a soluble and membrane-bound form.

If a deficiency of this protein results in agouti being lessable to antagonize MC1R, then what is the normal role ofattractin in the interactions between agouti, aMSH andMC1R? Bearing in mind that genetic evidence places theaction of the gene at, or upstream of, MC1R (Ref. 13),several models are possible. Membrane-bound attractincould be a low-affinity receptor for agouti, needed to con-centrate the antagonist (either on the melanocyte or onneighbouring cells) to affect MC1R. Alternatively, it couldsequester aMSH and reduce its effective concentration, sothat agouti is better able to act as an MC1R antagonist. Athird model is that it might act to desensitize MC1R following aMSH binding, so that signalling is effectivelyattenuated in its presence. Another interesting model,based on the suggested peptidase activity of attractin, isthat peptidyl cleavage of agouti could be necessary for itsfull activity. CD26 prefers to cleave a dipeptide ending inproline or alanine but, in fact, it has recently been shownto efficiently cleave a second dipeptide (following aglycine) from certain substrates19. Furthermore, the sub-strate specificity of attractin has not been characterized indetail, except to show that it is similar to CD26. A fewlines of evidence argue against cleavage-activation ofagouti. Gunn et al. used western blotting to demonstrate

that agouti protein (or at least the agouti protein over-expressed in the obese mutant hypothalamus) was notaltered in size in the mahogany mice16. But if the functionof attractin is simply to cleave a peptide from the N-termi-nus, then this would not have been detected by the assay.A more-critical problem for the model is that recombinantagouti works in vitro to bind to and antagonize MC1R(Ref. 9), without any known peptidase cleavage, althoughwhether the concentrations used in these experiments arephysiological is unknown, and of course the assay systemmight contain peptidases.

Clearly mahogany mutations cannot alter pigment syn-thesis without acting on the melanocyte, but is their sup-pression of obesity due to a similar effect on AGRP actingat MC4R? This is much less certain. Inhibition of MC4Rsignalling by overexpression of agouti leads to hyperphagy(over-eating), but the suppression of this phenotype bymahogany does not involve suppression of hyperphagy. Itis possible that the increased basal metabolic rate andhyperactivity caused by mahogany mutations nullify theeffects of the increased calorific intake. It is clear is thatattractin is playing a role in the CNS and that it wouldappear to be a valid target against which to test obesitydrugs. Perhaps a promising therapeutic approach might bethe use of specific dipeptidyl peptidase inhibitors, of whichseveral are available.

Finally, what do we make of the earlier studies onattractin in the context of the mahogany mutants?Attractin is present at some level on numerous differentcell types in the immune system, but is thought to accu-mulate rapidly on the surface of activated T cells fromwhere it can be released to the serum. It acts to cause thespreading of macrophages, which act as foci for the clus-tering of T cells. Duke-Cohan et al. propose that this facili-tates the response of T cells to antigen presented by

Outlook COMMENT The mahogany mouse mutation

TIG November 1999, volume 15, No. 11430

FIGURE 1. Melanocortin receptor agonists and antagonists

Proposed interactions between attractin, a-melanocyte-stimulating hormone (aMSH), melanocortin receptors (MC1R, MC4R), and agouti and agouti-related protein(AGRP), affecting pigmentation, inflammation and feeding behaviour.

In mammals, the maternal and paternal genomes are bothrequired for normal embryonic and postnatal develop-

ment. Their functional non-equivalence is mediated bygenomic imprinting, an epigenetic mechanism by whichthe expression of certain genes becomes dependent ontheir parental origin1. To date, some 30 imprinted geneshave been identified in humans and mice. Many of theseplay key roles in growth and differentiation, and imprint-ing is now recognized to be an important factor in severalinherited diseases and carcinogenesis in humans2,3.Although the precise mechanisms that allow cells to dis-tinguish the parental chromosomes at imprinted loci arepoorly understood, one epigenetic feature that is con-sistently associated with imprinting is CpG methylation4.Almost all imprinted genes have sequence elements thatare methylated only on one of the two parental alleles.

These are usually referred to as ‘differentially methylatedregions’ (DMRs). Experiments involving the targeteddeletion of the main mouse methyltransferase gene(Dnmt1), which led to loss of imprinting in mice that weredeficient in cytosine methylation5,6, imply that this methyl-ation mark is essential for the maintenance of imprinting.

Differentially methylated regionsMore recently, homologous-recombination experiments inthe mouse have demonstrated that individual DMRs are,indeed, important for the expression of imprinted genes.For example, a DMR located upstream of the mouse H19gene7 that is methylated on the paternal chromosome(Fig. 1) was shown to be essential for the imprintedexpression both of H19 and of the neighbouring, pater-nally expressed gene that encodes insulin-like growth

Genomic imprinting in mammalsan interplay between chromatin and DNAmethylation?

OutlookCOMMENTThe mahogany mouse mutation

TIG November 1999, volume 15, No. 110168-9525/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01822-3

Robert Feilrobert.feil@bbsrc.ac.uk

Sanjeev Khoslasanjeev.khosla@bbsrc.ac.uk

Programme inDevelopmental Genetics,The Babraham Institute,Cambridge, UK CB2 4AT.

431

macrophages17. Analysis of the immune responses of mg3J animals, which lack attractin, will no doubt help toelucidate its function. Is there any role for aMSH in theimmune system and might attractin be involved in that?Certainly aMSH has anti-inflammatory activity in ani-mals. At least two cell types involved in inflammationexpress MC1R: neutrophils, whose chemotaxis to sites

of inflammation might be inhibited by aMSH; andmacrophages, which are inhibited by aMSH from releas-ing nitric oxide, a mediator of inflammation20,21. Theseprocesses are rather different from those described thatinvolve attractin in vitro, but it will nevertheless be usefulto investigate the inflammatory response of mahoganymice.

References1 Krude, H. et al. (1998) Severe early-onset obesity, adrenal

insufficiency and red hair pigmentation caused by POMCmutations in humans. Nat. Genet. 19, 155–157

2 Robbins, L.S. et al. (1993) Pigmentation phenotypes of variantextension locus alleles result from point mutations that alterreceptor function. Cell 72, 827–834

3 Valverde, P. et al. (1995) Variants of the melanocyte-stimulating hormone-receptor gene are associated with redhair and fair skin in humans. Nat. Genet. 11, 328–330

4 Smith, R. et al. (1998) Melanocortin 1 receptor variants in anIrish population J. Invest. Dermatol. 111, 119–122

5 Huszar, D. et al. (1997) Targeted disruption of themelanocortin-4 receptor results in obesity in mice. Cell 88,131–141

6 Vaisse, C. et al. (1998) A frameshift mutation in human MC4Ris associated with a dominant form of obesity. Nat. Genet. 20,113–114

7 Yeo, G.S.H. et al. (1998) A frameshift mutation in MC4Rassociated with dominantly inherited obesity. Nat. Genet. 20,111–112

8 Bultman, S.P. et al. (1992) Molecular characterisation of the

mouse agouti locus. Cell 71, 1–209 Ollmann, M.M. et al. (1998) Interaction of agouti protein with

the melanocortin 1 receptor in vitro and in vivo. Genes Dev.12, 316–330

10 Duhl, D.M. et al. (1994) Neomorphic agouti mutations in obeseyellow mice. Nat. Genet. 8, 59–65

11 Ollmann, M.M. et al. (1997) Antagonism of centralmelanocortin receptors in vitro and in vivo by agouti-relatedprotein. Science 278, 135–138

12 Shutter, J.R. et al. (1997) Hypothalamic expression of ART, anovel gene related to agouti is up-regulated in obese anddiabetic mutant mice. Genes Dev. 11, 593–602

13 Miller, K.A. et al. (1997) Genetic studies of the mousemutations mahogany and mahoganoid. Genetics 146,1407–1415

14 Dinulescu, D.M. et al. (1998) Mahogany (mg) stimulatesfeeding and increases basal metabolic rate independent of itssuppression of agouti. Proc. Natl. Acad. Sci. U. S. A. 95,12707–12712

15 Nagle, D.L. et al. (1999) The mahogany protein is a receptorinvolved in suppression of obesity. Nature 398, 148–152

16 Gunn, T.M. et al. (1999) The mouse mahogany locus encodes a

transmembrane form of human attractin. Nature 398,152–156

17 Duke-Cohan, J.S. et al. (1998) Attractin (DPPT-L), a member ofthe CUB family of cell adhesion and guidance proteins, issecreted by activated human T lymphocytes and modulatesimmune cell interactions. Proc. Natl. Acad. Sci. U. S. A. 95,11336–11341

18 Duke-Cohan, J.S. et al. (1995) A novel form ofdipeptidylpeptidase IV found in human serum. J. Biol. Chem.270, 14107–14114

19 Proost, P. et al. (1999) Truncation of macrophage-derivedchemokine by CD26/dipeptidyl-peptidase IV beyond itspredicted cleavage site affects chemotactic activity andCC chemokine receptor 4 interaction. J. Biol. Chem. 274,3988–3993

20 Catania, A. et al. (1996) The neuropeptide aMSH has specificrecptors on neutrophils and reduces chemotaxis in vitro.Peptides 17, 675–679

21 Star, R.A. et al. (1995) Evidence of autocrine modulation ofmacrophage nitric-oxide synthase by alpha-melanocyte-stimulating hormone. Proc. Natl. Acad. Sci. U. S. A. 92,8016–8020

Most imprinted loci have key regulatory elements that are methylated on only one of the parental chromosomes.For several of these ‘differentially methylated regions’, recent studies establish that the unmethylatedchromosome has a specialized chromatin organization that is characterized by nuclease hypersensitivity. Thenovel data raise the question of whether specific proteins and associated chromatin features regulate theallele-specificity of DNA methylation at these imprinting control elements.