Complex alternative splicing of the myelin oligodendrocyteglycoprotein gene is unique to human and non-human primates

Cecile Delarasse,*,�,�� Bruno Della Gaspera,*,�,�� Chuan Wei Lu,§ Francois Lachapelle,�,��Antoinette Gelot,§§ Diana Rodriguez,�,��,¶ Andre Dautigny,�,�� Claude Genain§ andDanielle Pham-Dinh�,��,¶

�INSERM UMR 546, Paris, France

��Universite Pierre et Marie Curie Paris 6, Paris, France, France

�CNRS UMR 7060, Paris, France

��Universite de Paris 5, Paris, France

§University of California, Department of Neurology, San Francisco, California, USA

§§AP-HP, Unite de Neuropathologie, Hopital Trousseau, Paris, France

¶AP-HP, Service de Neuropediatrie, Hopital Trousseau, Paris, France

Abstract

Myelin/oligodendrocyte glycoprotein (MOG) is a minor integral

membrane protein specific to CNS myelin, encoded by a gene

located in the major histocompatibility complex. MOG is an

highly encephalitogenic autoantigen and a target for auto-

aggressive immune responses in CNS inflammatory demye-

linating diseases. We performed transcriptomic analyses for a

gene expressed only in mammalian CNS, myelin oligodend-

rocyte glycoprotein (MOG). Complex splicing patterns were

exclusively found in primates and not in mice, unlike patterns

found for other myelin protein genes. In addition to those

shared with rodents, these multiple MOG isoforms likely

support functions unique to the primate order, in particular

maintenance of myelin structure, intracellular signaling, and

modulation of CNS autoimmunity via exposure of specific

MOG determinants. Developmentally, in human brain the

splice variants of MOG appear at a late stage compared to

the major isoform, coincidental with myelination and myelin

maturation, unlike other myelin proteins. These findings are

discussed within the framework of a biological basis for phe-

notype diversity in recent mammalian evolution and for the

notoriously variable clinical expression of diseases such as

multiple sclerosis.

Keywords: alternative splicing, evolution, experimental

autoimmune encephalomyelitis, myelin/oligodendrocyte glyco-

protein, multiple sclerosis, myelin.

J. Neurochem. (2006) 98, 1707–1717.

Bioinformatic analysis of expressed sequence tag and mRNAsequences has revealed an unexpected high frequency andcomplexity of alternative splicing. Moreover, whereas mostgene structures and constitutive exons in major transcriptisoforms are highly similar, alternatively spliced formsfrequently lack conservation across species; this phenom-enon may be one evolutionary pathway to generate species-specific transcriptome diversity (Modrek and Lee 2003;Nurtdinov et al. 2003). This hypothesis can be tested byassessing alternative splicing patterns by sequencing full-length mRNAs as well as directed RT-PCR-based analysesusing primers encompassing alternative splicing sites.

Myelin is a relatively recent invention of evolution thatappeared about 400million years ago, and it is a unique feature

of vertebrates that is essential for fast conduction of nervepotentials along axons of the peripheral and central nervoussystems (reviewed in ref. (Colman et al. 1996)). Myelin is a

Received November 21, 2005; revised manuscript received March 17,2006; accepted May 31, 2006.Address correspondence and reprint requests to Danielle Pham-Dinh

INSERM UMR 546, Universite de Paris VI, 105 Boulevard de l’Hopital,75013 Paris, France. E-mail:danielle.pham-dinh@chups.jussieu.fr*Both authors contributed equally to this work.Abbreviations used: BT, butyrophillin; EAE, experimental autoim-

mune encephalomyelitis; Golli, gene expressed in the oligodendrocytelineage; MBP, myelin basic protein; MHC, major histocompatibilitycomplex; MOG, myelin oligodendrocyte glycoprotein; MS, multiplesclerosis; PLP, proteolipid proteins.

Journal of Neurochemistry, 2006, 98, 1707–1717 doi:10.1111/j.1471-4159.2006.04053.x

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717 1707

complex multilamellar structure synthesized by specializedcells, and comprises various cytoplasmic or transmembraneproteins with highly organized interactions with lipids andglycolipids responsible for the integrity andmaintenance of itsstructure and function. Most myelin proteins exist as proteinfamilies comprising several isoforms that result from alter-native splicing from a unique gene (Baumann and Pham-Dinh2001; Campagnoni and Skoff 2001). In CNS myelin, 90% ofthe protein content belongs to two major protein families: themyelin basic proteins (MBP), and proteolipid proteins (PLP).Quantitatively minor components include glycoproteins, suchasmyelin associated glycoprotein andmyelin oligodendrocyteglycoprotein (MOG), a 26–28-kDa glycoprotein only presentin mammals (Birling et al. 1993), in contrast to the moreancestral MBP and PLP (Baumann and Pham-Dinh 2001;Campagnoni and Skoff 2001).

MOG is encoded by a gene located in the majorhistocompatibility complex region (MHC) (Pham-Dinh et al.1993) and is the smallest representative in the immunoglob-ulin protein superfamily. MOG is the focus of considerableinterest due to the highly immunogenic properties of itssurface-exposed, IgV-like extracellular domain, and itspotential role as the target for pathogenic (e.g. disease-inducing) antibody responses in CNS autoimmune demye-lination (Pham-Dinh et al. 1993; Bernard et al. 1997;Iglesias et al. 2001; Delarasse et al. 2003; Pham-Dinh et al.2004). The sequence of MOG is highly conserved betweenspecies, both at the nucleotide and amino acid levels, asobserved for other myelin genes. However, preliminarystudies indicate that alternative splicing patterns of the MOGgene are likely to be more complex in humans compared tomice (Pham-Dinh et al. 1995; Ballenthin and Gardinier1996). Exon 2, which encodes the IgV-like domain for MOG,appears to have undergone adaptive ‘shuffling’ processesduring evolution, as it is shared with a number of other genesof the BT/B7 family (Linsley et al. 1994; Henry et al. 1999).These evolutionary events themselves provide the potentialfor differential expression of MOG and BT/B7 isoforms asan additional source of phenotypic diversity among species.The expression of MOG isoforms, similar to substitutions ingene nucleotide sequences that may result in functionallydifferent proteins, can underlie different biological roles forthis protein in different species and within a single species inhealth and disease.

We report the first complete analysis of the alternativesplicing of the MOG gene in five representative species fromrodent to human and find that the most complex splicingpatterns are exclusive to higher mammals including humanand non-human primates, unlike for other myelin proteins.Because structural features of the expressed MOG proteinsand their topographical expression control the exposure ofmajor immunogenic determinants within CNS (Mesleh et al.2002; Breithaupt et al. 2003; Clements et al. 2003; vonBudingen et al. 2004), these findings underscore the com-

plex splicing patterns of MOG as a novel factor that could becritical to phenotypic expression of CNS-directed autoim-munity in higher mammals.

Experimental procedures

Brain samples

Total brain RNA from adult C57BL/6 mouse, bovine and human was

obtained from Clontech (Biosciences, Saint-Quentin-en-Yvelines

France); sample of brains (frontal lobe) from a 21-week-old human

fetus was obtained at abortion, and 40-day- old and 2-year-old

children from postmortem autopsies. Consent was given to obtain the

human fetal and infant brain samples; adult brains of Macacafascicularis were obtained from two animals bred at the INRA

monkey colony (Jouy-en-Josas, France), two brains of adult common

marmoset (Callithrix jacchus) from the New England Regional

Primate Research Center, Southborough, MA, USA.

RT-PCR amplification, cloning and sequencing of MOG

isoforms

Brain mRNAs were reverse-transcribed following a-3¢ RACE (rapid

amplification of cDNA ends) protocol using a dT18-adapter primer,

5¢-GATGACTCGAGTCGACTCAG(T)18)3¢. To complete our

panel of MOG cDNAs, the major full-length MOG mRNAs from

macaque and marmoset brains were amplified, cloned and

sequenced, as previously described (Pham-Dinh et al. 1993). Thenumber and relative abundance of MOG splicing variants were first

evaluated in human, macaque, marmoset, bovine and mouse, using a

first round of PCR with species-specific sense primers located in

exon 1 and reverse primers located in exon 8 part B (sequences in

Table S1 in Supplementary material.

To clone the corresponding cDNAs, we increased the represen-

tation of minor splicing variants using a previously published

strategy (Pham-Dinh et al. 1995). In brief, the 1-08B PCR products,

comprising potentially all splicing MOG variants (containing

alternate a- or b 3¢-end), were successively cut with exon-specific

restriction enzymes, then nested PCRs were performed. BclIrestriction cutting was used (cutting site at nt 671 in exon 6B) to

amplify forms without exon 6B. To amplify forms without exon 2,

BamHI (cutting site nt 338 in exon 2) was used for all species exceptbovine, for which HaeII was used. To amplify only a forms, a

second run of amplification was performed using 1-08A primer sets.

EcoRI was used (cutting site at nt 886 in exon 8A) to selectively

amplify the b forms (in which exon 8 part A is spliced out). Further

amplifications were done using the 1-08B primer sets.

The final nested PCR products were subcloned shotgun in

pGEM-T easy vector (Promega Biotech, Madison, WI, USA). PCR

was used to select clones of interest on positive colonies, and direct

sequencing on the purified PCR products was performed to assess

exon composition and boundaries in the splicing variants (Genome-

Express, Meyran, France).

To perform PCR during the exponential phase of amplification,

the number of cycles was fixed to 30.

To analyze the developmental expression of MOG splicing

variants in human brains, RT-PCR were performed by standard

procedures for 30 cycles using mRNA from fetal 21-week-old,

postnatal 40-day- and 2-year-old, and adult brain (frontal lobes)

1708 C. Delarasse et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717� 2006 The Authors

samples. The PCR products were analyzed by Southern blotting

using exon-specific primers. Transcripts specific for exon 8A were

amplified with Hex1/Hex 08A primer pairs and hybridized with

labelled Hex3 primer. Transcripts specific for exon 8B were

amplified with Hex1/Hex 08B primer pairs and hybridized with

labelled Hex3 primer. Transcripts specific for exon Alu were

amplified with Hex1/Hex 0Alu primer pairs and hybridized with

labeled Alu primer. To avoid formation of heteroduplex, RT-PCR

products were run in denaturing conditions in NaOH buffer at 4�C,transferred on nylon membrane and probed with [a-32P] dCTP 3¢endlabeled exon-specific primers (Table S1). All amplified isoforms

detected by hybridization were assessed by cloning and sequencing.

Results

Sequencing of macaque and marmoset MOG cDNAs (NCBInumbers: AF399846, AY566834, respectively) confirmedthe high degree of conservation previously observed for thepredicted extracellular domain of MOG as compared tohuman and rodent sequences (this work and ref. (Pham-Dinh2004)) (Fig. 1). By contrast, the full set of CNS MOGtranscripts amplified using a single round of PCR with themost distal MOG primer sets (primer sequences in Supple-mentary Table S1) revealed multiple transcripts with clear-

cut size and quantitative differences (Fig. 2b). In addition tothe major transcript present in all species, several other bandswere detected in human, bovine, C. jacchus, and M.fascicularis brain amplified transcripts.

Alternative splicing of the human MOG gene

The splicing sites used in the primary transcript of theMOG gene led to discovery of 15 MOG isoforms in thehuman brain (recapitulated in Fig. 3a). The correspondingputative MOG protein isoforms are also schematicallydepicted (Fig. 3b). Consensus motifs for splicing weredetected at all intron–exon boundaries (Table S2 inSupplementary material), and specific nested PCR experi-ments are provided in Fig. 2c–e. Primer sets encompassingpotential splicing sites were used to specifically explore the5¢ part of the gene that was incompletely characterized inprevious studies (Pham-Dinh et al. 1995; Ballenthin andGardinier 1996) (Fig. 2a): three isoforms lacking exon 2(encoding the IgV-like domain) were found (Figs 2c,d),differing by their exon composition in the 3¢ end of thecDNA: a4 transcript containing exon 6B, a5 transcriptdevoid of any exon 6, and a6 transcript containing exon 6A (Fig. 3a; NCBI numbers: AY786325, AY566846, and

Fig. 1 Comparative sequences of the mature MOG protein between

mammalian species. Note high conservation in the extracellular do-

main, the second hydrophobic transmembrane domain (HR2) and the

carboxy-terminal cytoplasmic tail. MOG domains are encoded as fol-

lows: exon 2, aa 1–116 (most of the Ig-like domain); exon 3, aa 117–

154 (HR1) (della Gaspera et al. 1998); exons 4 and 5, aa 155–161 and

aa 162–168, respectively (intracellular loop); exon 6B, aa 169–207

(HR2); exons 7 and 8A encode aa 208–214, and 215–218, respect-

ively (intracellular C-terminal tail). *Asparagine 31 as a glycosylation

site.

Alternative splicing of the MOG gene across species 1709

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717

AY566847, respectively). A previously published sequence(Ballenthin and Gardinier 1996) of 138 bp located in intron2 (position 8583–8720 according to numbering in thehuman MOG gene (Roth et al. 1995)) was also found(Fig. 2d). This sequence belongs to the Alu repeat familyand has been designated Alu-10 exon, as it is the 10th Alusequence found within the gene (Roth et al. 1995). Wefurther searched for the presence of the Alu sequence inadditional transcripts by performing PCRs using an anti-sense primer derived from the Alu-10 sequence (Fig. 2e).

Two bands were obtained, cloned and sequenced. Surpris-ingly, four different sequences were found:

(i) an isoform comprising exons 1, 2 and Alu-10,designated Ig-Alu-10;

(ii) a sequence lacking exon 2, designated Alu-10;(iii) and (iv) two additional forms similar to (i) and (ii) but

presenting a deletion of 20 bp in the 5¢ end of exon Alu-10,indicating the occurrence of a further alternative splicingevent in this exon by use of an additional acceptor splicingsite. Thus, the full Alu-10 exon was designated Alu-10A, and

1710 C. Delarasse et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717� 2006 The Authors

the partly excised one, Alu-10B (NCBI numbers: AY566848(Ig-Alu-10B), AY566849 (Alu-10 A), AY566850 (Alu-10B).

Only one band corresponding to the splicing of exon 1 toexon 2 was found when this boundary was explored, whichindicates that there is no additional exon encoded bysequences located within intron 1. By contrast, amplificationof transcripts from exon 2–3 made it possible to find anadditional new splicing variant comprising a 163-bpsequence found at the junction between the classic exons 2and 3 (not shown). This sequence, which is located withinintron 2 upstream to exon Alu-10 (position 7923–8092 in ref.(Roth et al. 1995)), was designated exon 2¢ (NCBI number:AY566851) (recapitulated in Fig. 3a).

Identification of alternative MOG transcripts in other

mammalian species

The exon composition of alternative variants in the non-human mammalian species was first explored using species-specific 1-08B primer sets (Fig. 2b, and Table S1 inSupplementary material) corresponding to the non-selectiveamplification of a and b isoforms (thus including thesequences encoding the major a-specific carboxy-terminusand the alternative b-carboxy-terminus, respectively). In asecond step, nested PCRs were performed with the 1-08Aprimer sets, which amplifies only a forms, after Bcl1restriction cutting of exon 6B (contained in a1 majorisoforms), thus permitting selective amplification of minora transcripts (not shown). All these mRNA variants were

Fig. 3 Developmental analysis of MOG transcript expression in the

human brain. Transcripts specific for exon 8A, coding for the 3¢ end of

the major a1 MOG isoform, were amplified with Hex1/08A primer pairs

and hybridized with labeled Hex3 primer. Transcripts specific for exon

8B, specific for the 3¢ end of the MOG b isoforms, were amplified with

Hex1/08B primer pairs and hybridized with labeled Hex3 primer.

Transcripts specific for exon Alu were amplified with Hex1/Hex0Alu

primer pairs and hybridized with labeled Alu primer. All the relatively

major MOG isoforms were detectable using RT-PCR without re-

amplification. Only the a1 MOG isoform is detectable in all samples at

all times tested; splicing variants appear only at 2 years in childhood.

Fig. 2 Diversity of MOG alternate splicing. (a) Schematic represen-

tation of the MOG gene and position of PCR primers (primer se-

quences in Table S1). The MOG gene with its encoded domains is

depicted: Exon 1; leader peptide sequence (L); exon 2, extracellular

IgV-like domain; exon 3, hydrophobic transmembrane domain (HR1);

exon 4, 5, 7 and 8, intracellular cytoplasmic regions (IC); exon 6 (6B

according, to the new designation issued from alternative splicing in

ref. (Pham-Dinh et al. 1995) encodes a second hydrophobic region

(HR2). Alternate exons are colored. (b) Splicing variants RT-PCR

amplified from CNS of different mammalian species. Agarose gel

electrophoresis of RT-PCR products after 40 cycles of amplification.

RT-PCR amplification of MOG transcripts from different mammalian

species was obtained with Ms1/Ms08B, Bex1/Bex08B, Hex1/Mr08B,

Hex1/Mc08B and Hex1/Hex08B primer pairs for mouse, bovine,

marmoset, macaque and human MOG, respectively. Aliquots of

100 lL from RT-PCR mixture were separated by agarose gel elec-

trophoresis. DNA bands were visualized by ethidium bromide staining

of 1.4% agarose gel. The major band in each lane corresponds to the

a1 MOG major isoform (about 830 nucleotides), according to the

position of the species-specific primers used. Minor forms seen under

the major form are not found in mouse. (c, d, e) PCR amplifications of

human MOG cDNAs. All nested PCR experiments are from a first PCR

with primer set Hex1/Hex03AT (primer Hex1, located in exon 1; primer

Hex03AT, at the end of the non-coding part). PCR products are des-

cribed by exon composition. Primers pairs are positioned below the

schematic exon composition in a and indicated above each gel in c, d,

e. Exon sizes is indicated above exons when they are described for

the first time. The size of 5¢- and 3¢ exons is reduced due to the use of

internal primers. M, molecular weight marker. (c) Hex1/Hex08A primer

set revealed the previously unknown MOG isoform a4 (479 bp), the

major MOG isoform a1 (827 bp, all constitutive exons), and the iso-

form lacking exon 6B, previously designated as a2 (713 bp) (Pham-

Dinh et al. 1995). D denotes deleted exon. (d) Hex1/Hex05 primer set:

additional minor transcripts besides the expected major isoform a1 at

592 bp: a 244-bp band corresponding to splicing variants where the

348-bp sequence of exon 2 is skipped, as found in a4, a5 and a6

transcripts; and a 730-bp band containing a previously published se-

quence of 138 bp (Pham-Dinh et al. 1995) located in intron 2 (position

8583–8720 according to numbering in della Gaspera et al. (1998). The

latter sequence belongs to the Alu repeat family and has been des-

ignated the Alu-10 exon. (e) Antisense primer derived from the Alu-10

sequence (Hex0Alu). Two bands and four different sequences were

obtained: a 574-bp isoform comprising exons 1, 2 and Alu-10 (MOG

Ig-Alu-10); a 226-bp sequence lacking exon 2 (MOG Alu-10); and two

similar forms with a deletion of 20 bp in the 5¢ end of exon Alu-10. The

full Alu-10 exon was designated Alu-10A, and the partly excised one,

Alu-10B.

Alternative splicing of the MOG gene across species 1711

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717

further cloned and sequenced to fully assess exon organiza-tion of the corresponding cDNAs (summarized in Table 1).

MouseWe observed only one band corresponding to the majorisoform, a1, after amplification with the mouse-specific 1-08B primer set (Fig. 2b). However, another form corres-ponding to the minor human a2 (lacking exon 6B) wasfurther detected using 1-08A a isoform-specific nestedamplifications (Table 1) (NCBI number: AY566830). Neitherb nor other a forms were found after nested amplificationfollowing restriction cutting of PCR products using EcoRI orBamHI.

BovineSequencing of the lower molecular weight band detected inagarose gels (Fig. 2b), showed that this variant lacks exon 2and may correspond to the a4 human MOG isoform(Fig. 4a). After restricted amplification of the major formby Bcl1 restriction cutting, two additional bands weredetected corresponding to human a2 and a5 isoforms

(Fig. 4a). Further sequencing demonstrated three variants,a2, a4, a5 (Table 1) (NCBI numbers: AY566831,AY566832, AY566833, respectively). No b forms werefound after cutting by EcoRI in bovine brain PCRproducts.

MarmosetSequencing of the band visible on agarose gels below themajor form a1 (Fig. 2b; NCBI number: AY566834) showedan a4 isoform lacking exon 2 (Fig. 3a; NCBI number:AY566835), and also an unexpected isoform lacking 154 on348 bp at the 5¢ end of exon 2. In this variant, exon 1 isspliced to nucleotide 273 in exon 2, due to a crypticconsensus ‘ag’ dinucleotide acceptor site at position 271–272. By analogy to the a4 form, which lacks exon 2, thisform was designated a4Cj. Using the exon-specific restric-tion cutting strategy, another form lacking exon 3, designateda7Cj, was detected (Fig. 3a, Table 1). Neither a4Cj or a7Cjforms were observed in the other species studied and thusappear to be unique to marmosets (NCBI numbers:AY566836, AY566837, respectively).

Table 1 Exon composition of alternative

splicing variants through species

Presence of the isoform in the species is indicated by filled cells; alternative exons are high-

lighted in white type on gray shade, and isoforms unique to marmoset in gray shade, left.

*Partial deletion in exon 2.

1712 C. Delarasse et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717� 2006 The Authors

(a) (b)

Fig. 4 (a)Schematic representation of exon composition of MOG

splicing variants across species. Exon numbering according to Pham-

Dinh et al. (1995). Open boxes, non-coding regions. Black boxes,

constitutive coding exons; alternative exons: yellow, 2¢; red, Alu A;

blue, Alu B; green, 6A. Previously published human isoforms: MOG

a1, a2, a3, b1, b2, b3, b4, Ig-Alu-10A, in black type; newly discovered

isoforms: MOG a4, a5, a6, Iglu10B, Alu-10A, Alu-10B, Ig-2¢, in red

type and framed; **a4Cj and a7Cj, the two marmoset-specific MOG

isoforms. Bovine and mouse MOG transcripts are not represented

here as these isoforms do not differ from those found in primates

(please refer to Table 1 for mouse/bovine transcript composition).

Three families of mRNAs are defined, based on use of one of two

alternative acceptor splice sites in exon 8 in the 3¢ part of the gene (a

and b), or the presence of a premature stop codon introduced by

exonization (spliced-mediated) of Alu-10 or 2’ sequences from intron 2

(Alu forms). *Alternative in-frame stop codons. (b) Schematic topolo-

gical models of the putative MOG protein isoforms. Exon 1 (leader

peptide) is not represented. White boxes, constitutive exons contained

in the major MOG protein. Striped boxes, C-terminal protein domain

encoded by the alternative 8B exon, defining the b isoform family.

Colored boxes indicate domains encoded by alternative exons. The

highly hydrophobic exon 6B-encoded domain is represented as semi-

embedded in the membrane in the forms containing exon 3 (encoding

the transmembrane domain), whereas it is tentatively drawn spanning

the membrane when exon 3 is skipped, as in a7Cj. Exons 4 and 5 in

a7Cj are tentatively represented as extracellular. N-glycosylation site

is indicated by an asterisk above the asparagine 31 (N 31).

Alternative splicing of the MOG gene across species 1713

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717

MacaqueThe complex expression pattern detected in this speciescomprises four faint bands in addition to the major form(Fig. 2b) and appears virtually identical to that of humans.Sequencing revealed one form without exon 2 (a4 isoform),and another containing exon 8B in place of exon 8A, thuscorresponding to a macaque b1 isoform. Using a-specificamplification, we detected a2, a3, a4, a5 and a6. Using thenested amplification strategy followed by EcoRI cutting,we also found b2 and b3 forms. Thus the human andM. fascicularis brains both exhibit a1, a2, a3, a4, a5, a6, aswell as b1, b2, and b3 MOG transcripts (recapitulated inTable 1; NCBI numbers: AF399846, AY566838, AY566839,AY566840, AY566841, AY566842, AY566843, AY566844,and AY566845, respectively).

We searched for the Alu and 2¢ exon sequences in non-human primates, C. jacchus and M. fascicularis, using, as forhuman, a short PCR amplification from exon 2–3 (notshown). Only exon 2¢ was found in the M. fascicularis(NCBI number: AY786326). By comparison with all otherspecies studied here, the closest to humans is macaque,which is also the only other mammal to express b forms ofMOG (recapitulated in Fig. 3).

Expression of the MOG spliced variants during

development of human CNS

To study the MOG alternative transcripts regulation duringmyelinogenesis, we performed RT-PCR analysis of theMOG mRNA variants in human brain tissues (frontal lobe).As shown on Fig. 3 the major a1 mRNA isoform is presentat all time points, albeit weakly in the fetus and 40 days afterbirth, at which time the other variants were undetectable inthe experimental conditions we used. By contrast, in a 2-year-old brain sample, all isoforms are clearly expressed,although at apparently lower levels than that of the majorform.

Discussion

This report is the first to comprehensively characterizediversity generated by splicing of the primary transcript fromthe MOG gene during recent evolution. The mature MOGprotein is encoded by the major MOG transcript a1.Although the pattern of transcription appears similar to thatof major structural myelin protein genes (Baumann andPham-Dinh 2001; Campagnoni and Skoff 2001) such asMBP and PLP, MOG is unique in that:

(i) It is only found in mammalian genomes, where it islocated within the MHC (Pham-Dinh et al. 1993; Pham-Dinh2004).

(ii) Whereas other myelin genes tend to be ubiquitouslyexpressed (Baumann and Pham-Dinh 2001; Campagnoni andSkoff 2001), expression of the MOG protein is restricted tomyelin (Colman 1996).

(iii) When comparing rodents to higher mammals, MOGis overall less conserved than other myelin proteins (Popotet al. 1991), especially within the transmembrane domain(Fig. 1).

Alternative splicings of major and minor transcripts ofMOG exhibit a striking increasing complexity that appears toparallel evolutionary events, as the most complex patterns arefound in human and non-human primates. Fifteen alternativesplicing variants using combinations of 11 exons are presentin human, 10 in Old World primates (M. fascicularis), andonly two in mice (when including the major form a1). The bvariants are only found in macaques and humans, and solubleforms are not expressed in any non-primate species(Table 1). This phenomenon has not been reported foranother well studied myelin gene, the Golli/MBP gene, asGolli/MBP isoforms are well conserved between human andmouse (Pribyl et al. 1993; Campagnoni and Skoff 2001).Moreover, contrary to both the MOG and Golli/MBP genes,the PLP gene encodes two mouse-specific isoforms, whichhave not yet been found in human (Campagnoni and Skoff2001), in addition to the strictly conserved PLP and DM20isoforms found in most vertebrate species.

In all the species examined here, several alternatingvariants show splicing out of sequences encoding the secondhydrophobic domain of MOG (exon 6B), or the carboxy-terminus (exon 8A). These sequences may be replaced withhydrophilic exon 6A, or a new carboxy-terminus (exon 8B)(Fig. 4; Table 1). The resulting variant proteins, especiallythose containing exon 8B (b forms) could underlie distinctMOG–cytoskeleton interactions and transmembrane signa-ling pathways. It has recently been reported that interactionbetween myelin proteins and the myelin membrane lipids, asopposed to proteins themselves, may be essential formembrane stability and integrity (Hu et al. 2004), thus bothMOG/lipid interactions and signaling pathways throughMOG could potentially differ between rodents and primatesincluding man. We note that, similar to MOG, the evolu-tionary-related butyrophilin (BT) and bg antigen members ofthe B7/BT family genes (Linsley et al. 1994), are alsoalternatively spliced in the short exons encoding heptadsforming the cytoplasmic part of their molecule (Henry et al.1999).

In addition to myelin maintenance and function, theintrinsic alternative splicing machinery may operate toregulate the expression of immunologically relevant domainsof MOG in primates. In the 5¢ part of the gene, MOG splicingvariants that lack exon 2 (exposed, encephalitogenic IgV-likedomain) are found in all species except mouse (Table 1). Theimmune system may ignore these isoforms during develop-ment, and thus truncated MOG proteins likely play a role inmaintenance of central and peripheral tolerance and/or ininflammatory and demyelinating diseases (Bernard et al.1997; Iglesias et al. 2001). Expression of these isoforms inthe CNS can now be examined in physiological and

1714 C. Delarasse et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717� 2006 The Authors

pathological conditions as it may regulate the balancebetween tolerance and autoimmunity. Similarly, primate-specific exons encode for Ig-like domain isoforms resultingfrom exonization of premature stop codon-containing exons,such as Alu or 2¢ sequences, downstream of exon 2sequences. These transcripts correspond to species-specificprotein motifs that are predicted to be soluble rather thanmembrane-bound (Fig. 4b). The expression of such solubleMOG proteins has not yet been investigated, but may have aprofound impact on immune homeostasis or dysregulation.

Quite unexpectedly and remarkably, C. jacchus marmosetspresent a pattern of MOG alternative splicing with only fourisoforms, including the two variants unique to this species,a4Cj and a7Cj (Table 1; Fig. 4). These New World primatesevolved from a split of the Anthropoidea into Platirrhini(New World primates) and Catarrhini (Old World primates)about 35–40 million years ago, whereas M. fascicularis andHomo sapiens diverged at 35 million years in the Old Worldprimate tree. Both M. fascicularis and C. jacchus evolvedwith unique characteristics such as limited variability in MHCclass I and II genes and, for marmosets, bone marrowchimerism (Watkins et al. 1990; Cadavid et al. 1997; Uccelliet al. 1997; Antunes et al. 1998), but differentiation of otherimmune system genes is quite similar to man in complexity(Uccelli et al. 1997; von Budingen et al. 2001). It would beof interest to ask whether the transcriptional organization ofMOG has evolved similarly in other NewWorld primates thatshow evolutionary instability of the MHC class I loci(Cadavid et al. 1997), or whether it is restricted to C. jacchus.

The marmoset isoform a4Cj may be regrouped withisoforms where exon 2 is spliced out, resulting in truncatedproteins lacking the IgV-like domain found in all speciesexcept mouse. While a4Cj is most likely soluble, a7Cj isunique in that it lacks exon 3 (first transmembrane domain),and would thus encode for an isoform with an extracellulardomain extended by the hydrophilic peptides encoded byexons 4 and 5. These proteins will have a distinct tertiarystructure, and/or may be less tightly associated with themyelin membrane than the major MOG isoform. It isnoteworthy that the most complex neuropathological patternsof experimental inflammatory demyelination can be repro-duced in C. jacchus and are highly reminiscent of humanmultiple sclerosis (MS). Similarly to a mouse model of MS(Marta et al. 2005 and references therein), the demyelinatingcomponent of lesions in C. jacchus is produced by asubgroup of anti-MOG-antibodies that are pathogenicbecause they bind to conformational determinants of theIgV-like domain, and not its linear domains located in theexposed random loops (von Budingen et al. 2004). Theserecent findings in the C. jacchus system could illustrate howdifferential expression of the spliced variants of MOG,through presentation of specific antigenic determinantswithin CNS, may influence susceptibility to disease and theexpression of phenotype in primates including man. This

phenomenon has recently been demonstrated using an anti-MOG TCR transgenic mouse model, in which a largeproportion of mice spontaneously develop an isolated opticneuritis without any evidence of EAE, in accordance with thehigh level of expression of the MOG antigen target in theoptic nerve vs. the spinal cord (Bettelli et al. 2003). Inanother context, the targeted ablation of the golli-s (geneexpressed in the oligodendrocyte lineage), but not the classicMBPs (myelin basic protein), resulted in a distinct phenotypeunlike that of knock-outs of the classic MBPs (Jacobs et al.2005). On this basis, the topographical distribution andquantitative expression of MOG variants that contain deter-minants exposed to pathogenic antibodies might have thepotential to influence the location, severity, and density ofdemyelinating lesions along the neuraxis in disorders likeMS.

The human MOG gene contains as many as 14 Alusequences concentrated in three introns (Roth et al. 1995).The Alu-10 exon belongs to the J-subclass of Alu element,one of the oldest Alu subfamilies, and is located in theantisense orientation (position 8568–8863; 296 bp length),the most prone to be exonized. We note that the genomicsequence surrounding the acceptor site of the Alu-10 exon inthe MOG gene (Table S2 in Supplementary material), issimilar to that previously determined as necessary forexonization for the Alu-J family subtype (Lev-Maor et al.2003). The four MOG Alu-containing isoforms describedhere were only found in human, and provide additionalexamples of the potential of Alu elements for enriching thetranscriptome (Kreahling and Graveley 2004). Moreover, iftranslation occurs as expected for these polyadenylatedRNAs, the exonization of Alu and 2¢ sequences specific tohigher primates may contribute to distinguish the primateproteome from that of other mammals. It may be relevant tofast evolutionary changes, as a ready-to-use source of novelprotein-coding sequences (Modrek and Lee 2003; Nurtdinovet al. 2003). Alu-10 and 2¢ sequences of the MOG gene arelocated in similar positions in the chimpanzee and humangenomes (http://genome.ucsc.edu/cgi-bin/hgBlat).

We studied the regulation of MOG alternative splicingduring development of the human brain. The major MOG a1mRNA isoform is present at a weak level as early as the 21stweek of pregnancy and 40 days after birth (Fig. 3). Adramatic change appears in the 2-year-old brain, the level ofthe major MOG a1 mRNA isoform becoming more abundantand the other isoforms detectable. Although under ourexperimental conditions we could not perform adequate andreliable quantification, the qualitative shift of expression ofall MOG mRNA isoforms at age 2 correlates quite well withthe stage of maturation of myelin structures (Brody et al.1987; Kinney et al. 1988) and acquisition of coordinatedCNS activity, or even the development of language (Pujolet al. 2006), at this age in humans. These results are in partreminiscent of the expression shift of the two isoformsderived from the PLP gene during human brain development,

Alternative splicing of the MOG gene across species 1715

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717

as DM20 is predominant during fetal life, whereas PLP ismore abundantly expressed from the second year of age,where it constitutes 50% of the myelin protein content. Inthis context, the detection of MOG splicing variants duringlate development implies that their role might be specific tomature myelin.

We also tested the hypothesis that some ‘developmental’isoforms could be re-expressed during the remyelinatingprocess in MS. Such a process – developmental re-expres-sion during remyelination – has been described for MBP (re-expression of exon 2-containing isoform) (Voskuhl et al.1993; Segal et al. 1994) and PLP (increased expression ofDM20 isoform) (Mathisen et al. 2001). Analysis by RT-PCRof several samples of normal appearing and diseased whitematter failed to demonstrate different patterns of splicing forMOG (not shown). Our interpretation of these results isobviously limited, but is in line with the finding that themajor a1 MOG isoform is the first form to appear duringhuman gestation.

In summary, the complexity of alternative splicing patternof the MOG gene seems to be correlated with position ofmammals in the phylogenic evolutionary tree, a phenomenonthat is not observed for other myelin genes. One explanationmay be that the MOG gene is located in the MHC, a mostunstable genomic region evolving the most rapidly acrossspecies because of its prime importance in defining andregulating self-tolerance and autoimmunity. The recentlyemerged MOG gene appears to represent an importantmilestone in mammalian evolution in the context of itsmyelin-restricted expression, and may be at a crossroadslinking CNS biology and autoimmunity. The presence ofsoluble isoforms and b isoforms that are unique to primates isintriguing in view of the putative roles of MOG as astructural myelin protein, a receptor for signaling (Johns andBernard 1999) and an antigenic target for demyelinatingantibodies with similar antigenic determinants in humans andprimates (von Budingen et al. 2002; Pham-Dinh 2004).Finally, our data can be placed in context with the recent andsomewhat surprising discovery that the human genomecontains only about 25–30 000 genes, most of them sharedwith other species including mouse (Lander et al. 2001;Venter et al. 2001). How did phenotypic variability betweenspecies then emerge? Possibly, alternative splicing plays aunique role in increasing the rate of evolutionary changes byintroducing species-specific exons in the transcriptomes. Thegenome of humans and other mammals might differ in theirexpression of species-specific groups of alternative exonsdue to the creation or loss of exons during recent evolution(Modrek and Lee 2003; Nurtdinov et al. 2003).

Acknowledgements

Supported by the European Leukodystrophy Association (ELA) and

Association de Recherche sur la Sclerose en Plaques (DPD), the

National Multiple Sclerosis Society (CPG). CD received fellowships

from the Fondation de la Recherche Medicale and Association de

Recherche sur la Sclerose en Plaques. We thank Pierre Pontarotti

and Richard Miles for helpful discussions.

References

Antunes S. G., de Groot N. G., Brok H., Doxiadis G., Menezes A. A.,Otting N. and Bontrop R. E. (1998) The common marmoset: a newworld primate species with limited Mhc class II variability. Proc.Natl Acad. Sci. U S A 95, 11 745–11 750.

Ballenthin P. A. and Gardinier M. V. (1996) Myelin/oligodendrocyteglycoprotein is alternatively spliced in humans but not mice.J. Neurosci. Res. 46, 271–281.

Baumann N. and Pham-Dinh D. (2001) Biology of oligodendrocyte andmyelin in the mammalian central nervous system. Physiol. Rev. 81,871–927.

Bernard C. C., Johns T. G., Slavin A., Ichikawa M., Ewing C., Liu J. andBettadapura J. (1997) Myelin oligodendrocyte glycoprotein: anovel candidate autoantigen in multiple sclerosis. J. Mol Med. 75,77–88.

Bettelli E., Pagany M., Weiner H. L., Linington C., Sobel R. A. andKuchroo V. K. (2003) Myelin oligodendrocyte glycoprotein-spe-cific T cell receptor transgenic mice develop spontaneous autoim-mune optic neuritis. J. Exp Med. 197, 1073–1081.

Birling M. C., Roussel G., Nussbaum F. and Nussbaum J. L. (1993)Biochemical and immunohistochemical studies with specificpolyclonal antibodies directed against bovine myelin/oligodend-rocyte glycoprotein. Neurochem. Res. 18, 937–945.

Breithaupt C., Schubart A., Zander H., Skerra A., Huber R., Linington C.and Jacob U. (2003) Structural insights into the antigenicity ofmyelin oligodendrocyte glycoprotein. Proc. Natl Acad. Sci. U S A100, 9446–9451.

Brody B. A., Kinney H. C., Kloman A. S. and Gilles F. H. (1987)Sequence of central nervous system myelination in human infancy.I. An autopsy study of myelination. J. Neuropathol. Exp. Neurol.46, 283–301.

von Budingen H. C., Hauser S. L., Fuhrmann A., Nabavi C. B., Lee J. I.and Genain C. P. (2002) Molecular characterization of antibodyspecificities against myelin/oligodendrocyte glycoprotein in auto-immune demyelination. Proc. Natl Acad. Sci. U S A 99, 8207–8212.

von Budingen H. C., Hauser S. L., Ouallet J. C., Tanuma N., Menge T.and Genain C. P. (2004) Frontline: Epitope recognition on themyelin/oligodendrocyte glycoprotein differentially influences dis-ease phenotype and antibody effector functions in autoimmunedemyelination. Eur. J. Immunol. 34, 2072–2083.

von Budingen H. C., Tanuma N., Villoslada P., Ouallet J. C., Hauser S.L. and Genain C. P. (2001) Immune responses against the myelin/oligodendrocyte glycoprotein in experimental autoimmune demy-elination. J. Clin. Immunol. 21, 155–170.

Cadavid L. F., Shufflebotham C., RuiZ. F. J., Yeager M., Hughes A. L.and Watkins D. I. (1997) Evolutionary instability of the majorhistocompatibility complex class I loci in New World primates.Proc. Natl Acad. Sci. U S A 94, 14 536–14 541.

Campagnoni A. T. and Skoff R. P. (2001) The pathobiology of myelinmutants reveal novel biological functions of the MBP and PLPgenes. Brain Pathol. 11, 74–91.

Clements C. S., Reid H. H., Beddoe T., Tynan F. E., Perugini M. A.,Johns T. G., Bernard C. C. and Rossjohn J. (2003) The crystalstructure of myelin oligodendrocyte glycoprotein, a key autoanti-gen in multiple sclerosis. Proc. Natl Acad. Sci. U S A 100, 11 059–11 064.

1716 C. Delarasse et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717� 2006 The Authors

Colman D. R., Doyle J. P., D’Urso D., Kitagawa K., Pedreza L., Yool D.,Yoshida M. and Fannon A. M. (1996) Glial Cell Development:Basic Principles and Clinical Relevance, Vol. 1, pp. 85–100. BIOSScientific Publishers Ltd, Oxford.

Delarasse C., Daubas P., Mars L. T. et al. (2003) Myelin/oligodendro-cyte glycoprotein-deficient (MOG-deficient) mice reveal lack ofimmune tolerance to MOG in wild-type mice. J. Clin. Invest 112,544–553.

della Gaspera B., Pham-Dinh D., Roussel G., Nussbaum J. L. andDautigny A. (1998) Membrane topology of the myelin/oligo-dendrocyte glycoprotein. Eur. J. Biochem. 258, 478–484.

Henry J., Miller M. M. and Pontarotti P. (1999) Structure and evolutionof the extended B7 family. Immunol. Today 20, 285–288.

Hu Y., Doudevski I., Wood D., Moscarello M., Husted C., Genain C.,Zasadzinski J. A. and Israelachvili J. (2004) Synergistic interac-tions of lipids and myelin basic protein. Proc. Natl Acad. Sci. U S A101, 13 466–13 471.

Iglesias A., Bauer J., Litzenburger T., Schubart A. and Linington C.(2001) T- and B-cell responses to myelin oligodendrocyte glyco-protein in experimental autoimmune encephalomyelitis and mul-tiple sclerosis. Glia 36, 220–234.

Jacobs E. C., Pribyl T. M., Kampf K. et al. (2005) Region-specificmyelin pathology in mice lacking the golli products of the myelinbasic protein gene. J. Neurosci. 25, 7004–7013.

Johns T. G. and Bernard C. C. (1999) The structure and function ofmyelin oligodendrocyte glycoprotein. J. Neurochem. 72, 1–9.

Kinney H. C., Brody B. A., Kloman A. S. and Gilles F. H. (1988)Sequence of central nervous system myelination in human infancy.II. Patterns of myelination in autopsied infants. J. Neuropathol.Exp. Neurol. 47, 217–234.

Kreahling J. and Graveley B. R. (2004) The origins and implications ofalternative splicing. Trends Genet. 20, 1–4.

Lander E. S., Linton L. M., Birren B. et al. (2001) Initial sequencing andanalysis of the human genome. Nature 409, 860–921.

Lev-Maor G., Sorek R., Shomron N. and Ast G. (2003) The birth of analternatively spliced exon: 3¢ splice-site selection in Alu exons.Science 300, 1288–1291.

Linsley P. S., Peach R., Gladstone P. and Bajorath J. (1994) Extendingthe B7 (CD80) gene family. Protein Sci. 3, 1341–1343.

Marta C. B., Oliver A. R., Sweet R. A., Pfeiffer S. E. and Ruddle N. H.(2005) Pathogenic myelin oligodendrocyte glycoprotein antibodiesrecognize glycosylated epitopes and perturb oligodendrocytephysiology. Proc. Natl Acad. Sci. U S A 102, 13 992–13 997.

Mathisen P. M., Kawczak J. A. YuM., Johnson J. M. and Tuohy V. K.(2001) Differential DM20 mRNA expression distinguishes twodistinct patterns of spontaneous recovery from murine autoimmuneencephalomyelitis. J. Neurosci. Res. 64, 542–551.

Mesleh M. F., Belmar N., Lu C. W., Krishnan V. V., Maxwell R. S.,Genain C. P. and Cosman M. (2002) Marmoset fine B cell and Tcell epitope specificities mapped onto a homology model of theextracellular domain of human myelin oligodendrocyte glycopro-tein. Neurobiol. Dis. 9, 160–172.

Modrek B. and Lee C. J. (2003) Alternative splicing in the human,mouse and rat genomes is associated with an increased frequencyof exon creation and/or loss. Nat. Genet. 34, 177–180.

Nurtdinov R. N., Artamonova I. I., Mironov A. A. and Gelfand M. S.(2003) Low conservation of alternative splicing patterns in thehuman and mouse genomes. Hum. Mol. Genet. 12, 1313–1320.

Pham-Dinh D., Mattei M. G., Nussbaum J. L., Roussel G., Pontarotti P.,Roeckel N., Mather I. H., Artzt K., Lindahl K. F. and Dautigny A.(1993) Myelin/oligodendrocyte glycoprotein is a member of asubset of the immunoglobulin superfamily encoded within themajor histocompatibility complex. Proc. Natl Acad. Sci. U S A 90,7990–7994.

Pham-Dinh D., Della Gaspera B., Kerlero de Rosbo N. and Dautigny A.(1995) Structure of the human myelin/oligodendrocyte glycopro-tein gene and multiple alternative spliced isoforms. Genomics 29,345–352.

Pham-Dinh D., Dautigny A. and Linington C. (2004) Myelin oligo-dendrocyte glycoprotein gene, in Myelin Biology and Disorders(Lazzarini R. A., ed.), Vol. 1, pp. 469–497. Elsevier Science, NewYork.

Popot J. L., Pham Dinh D. and Dautigny A. (1991) Major myelinproteolipid: the 4-alpha-helix topology. J. Membr. Biol. 120,233–246.

Pribyl T. M., Campagnoni C. W., Kampf K., Kashima T., Handley V. W.,McMahon J. and Campagnoni A. T. (1993) The human myelinbasic protein gene is included within a 179-kilobase transcriptionunit: expression in the immune and central nervous systems. Proc.Natl Acad. Sci. U S A 90, 10 695–10 699.

Pujol J., Soriano-Mas C., OrtiZ. H., Sebastian-Galles N., Losilla J. M.and Deus J. (2006) Myelination of language-related areas in thedeveloping brain. Neurology 66, 339–343.

Roth M. P., Malfroy L., Offer C., Sevin J., Enault G., Borot N.,Pontarotti P. and Coppin H. (1995) The human myelin oligoden-drocyte glycoprotein (MOG) gene: complete nucleotide sequenceand structural characterization. Genomics 28, 241–250.

Segal B. M., Raine C. S., McFarlin D. E., Voskuhl R. R. and McFarlandH. F. (1994) Experimental allergic encephalomyelitis induced bythe peptide encoded by exon 2 of the MBP gene, a peptideimplicated in remyelination. J. Neuroimmunol. 51, 7–19.

Uccelli A., Oksenberg J. R., Jeong M. C., Genain C. P., Rombos T.,Jaeger E. E., Giunti D., Lanchbury J. S. and Hauser S. L.(1997) Characterization of the TCRB chain repertoire in theNew World monkey Callithrix jacchus. J. Immunol. 158, 1201–1207.

Venter J. C., Adams M. D., Myers E. W. et al. (2001) The sequence ofthe human genome. Science 291, 1304–1351.

Voskuhl R. R., McFarlin D. E., Tranquill L. R., Deibler G., Stone R.,Maloni H. and McFarland H. F. (1993) A novel candidate auto-antigen in a multiplex family with multiple sclerosis: prevalence ofT-lymphocytes specific for an MBP epitope unique to myelination.J. Neuroimmunol. 46, 137–144.

Watkins D. I., Chen Z. W., Hughes A. L., Hodi F. S. and Letvin N. L.(1990) Genetically distinct cell populations in naturally occurringbone marrow-chimeric primates express similar MHC class I geneproducts. J. Immunol. 144, 3726–3735.

Alternative splicing of the MOG gene across species 1717

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 98, 1707–1717