messenger rnas located in myelin sheath assembly sites

11
Messenger RNAs Located in Myelin Sheath Assembly Sites Robert M. Gould, Concetta M. Freund, Frank Palmer, and *Douglas L. Feinstein Department of Pharmacology, NYS Institute for Basic Research in Developmental Disabilities, Staten Island, New York, and *Department of Anesthesiology, University of Illinois, Chicago, Illinois, U.S.A. Abstract: The targeting of mRNAs to specific subcellular locations is believed to facilitate the rapid and selective incorporation of their protein products into complexes that may include membrane organelles. In oligodendro- cytes, mRNAs that encode myelin basic protein (MBP) and select myelin-associated oligodendrocytic basic pro- teins (MOBPs) locate in myelin sheath assembly sites (MSAS). To identify additional mRNAs located in MSAS, we used a combination of subcellular fractionation and suppression subtractive hybridization. More than 50% of the 1,080 cDNAs that were analyzed were derived from MBP or MOBP mRNAs, confirming that the method se- lected mRNAs enriched in MSAS. Of 90 other cDNAs identified, most represent one or more mRNAs enriched in rat brain myelin. Five cDNAs, which encode known proteins, were characterized for mRNA size(s), enrich- ment in myelin, and tissue and developmental expression patterns. Two of these, peptidylarginine deiminase and ferritin heavy chain, have recognized roles in myelination. The corresponding mRNAs were of different sizes than the previously identified mRNA, and they had tissue and development expression patterns that were indistin- guishable from those of MBP mRNA. Three other cDNAs recognize mRNAs whose proteins (SH3p13, KIF1A, and dynein light intermediate chain) are involved in membrane biogenesis. Although enriched in myelin, the tissue and developmental distribution patterns of these mRNAs dif- fered from those of MBP mRNA. Six other cDNAs, which did not share significant sequence homology to known mRNAs, were also examined. The corresponding mRNAs were highly enriched in myelin, and four had tissue and developmental distribution patterns indistinguishable from those of MBP mRNA. These studies demonstrate that MSAS contain a diverse population of mRNAs, whose locally synthesized proteins are placed to contrib- ute to myelin sheath assembly and maintenance. Char- acterization of these mRNAs and proteins will help pro- vide a comprehensive picture of myelin sheath assembly. Key Words: Dynein light intermediate chain—Ferritin heavy chain—Myelination—Kinesin— Peptidylarginine de- iminase —SH3p13—Suppression subtractive hybridization. J. Neurochem. 75, 1834 –1844 (2000). Individual oligodendrocytes assemble many myelin sheaths, some near the soma and some at considerable distances from the cell body. To efficiently incorporate myelin basic protein (MBP) and select myelin-associated oligodendrocytic basic proteins (MOBPs) into the sheaths, the corresponding mRNAs are transported to and translated in each myelin sheath assembly site (MSAS). These small, highly basic proteins, confined mainly to the major dense line (MDL), are believed to function in MDL formation and maintenance (Omlin et al., 1982; McLaurin et al., 1993; Yamamoto et al., 1994). Many oligodendrocytes in mutants that lack MBP (Privat et al., 1979; Ganser and Kirschner, 1980; Rosen- bluth, 1980; Matthieu et al., 1981) or MOBP (Yamamoto et al., 1999) carry out initial steps of myelination nor- mally. It therefore seems likely that additional proteins, some of which are synthesized in MSAS, participate in the generation of myelin sheaths. Like MBP and MOBP mRNAs, the mRNAs for these proteins would be ex- pected to be transported to MSAS. The goal of this study was to identify mRNAs that co-localize with MBP and MOBP mRNAs in MSAS. Further characterization of these mRNAs will surely broaden our understanding of how myelin sheaths are assembled. Previous studies have shown that the only mRNAs selectively enriched in rat brain myelin are located in MSAS (Colman et al., 1982; Gillespie et al., 1990a,b; Gould et al., 1999a). Subcellular fractionation was a logical first step to use to separate MSAS RNA from other neural mRNAs. However, additional steps were needed as contaminating non-MSAS mRNAs are also present, though not enriched, in the myelin fraction (Gould, 1998; Gould et al., 1999a). We used PCR-based suppression subtractive hybridization (SSH) in conjunc- tion with northern blot analysis to isolate and identify Received March 24, 2000; revised manuscript received June 7, 2000; accepted July 7, 2000. Address correspondence and reprint requests to Dr. R. M. Gould at Department of Pharmacology, NYS Institute for Basic Research in Developmental Disabilities, 1050 Forest Hill Rd., Staten Island, NY 10314, U.S.A. E-mail: [email protected] Abbreviations used: CNP, 29,39-cyclic nucleotide phosphodiesterase; DLIC, dynein light intermediate chain; ds, double-stranded; EST, ex- pressed sequence tag; FHC, ferritin heavy chain; LSS, low-speed supernatant; MBP, myelin basic protein; MDL, major dense line; MOBP, myelin-associated oligodendrocytic basic protein; MSAS, my- elin sheath assembly site(s); PAD, peptidylarginine deiminase; PLP, proteolipid protein; RACE, rapid amplification of cDNA ends; SSH, suppression subtractive hybridization. 1834 Journal of Neurochemistry Lippincott Williams & Wilkins, Inc., Philadelphia © 2000 International Society for Neurochemistry

Upload: robert-m-gould

Post on 06-Jul-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Messenger RNAs Located in Myelin Sheath Assembly Sites

Robert M. Gould, Concetta M. Freund, Frank Palmer, and *Douglas L. Feinstein

Department of Pharmacology, NYS Institute for Basic Research in Developmental Disabilities, Staten Island, New York, and*Department of Anesthesiology, University of Illinois, Chicago, Illinois, U.S.A.

Abstract: The targeting of mRNAs to specific subcellularlocations is believed to facilitate the rapid and selectiveincorporation of their protein products into complexesthat may include membrane organelles. In oligodendro-cytes, mRNAs that encode myelin basic protein (MBP)and select myelin-associated oligodendrocytic basic pro-teins (MOBPs) locate in myelin sheath assembly sites(MSAS). To identify additional mRNAs located in MSAS,we used a combination of subcellular fractionation andsuppression subtractive hybridization. More than 50% ofthe 1,080 cDNAs that were analyzed were derived fromMBP or MOBP mRNAs, confirming that the method se-lected mRNAs enriched in MSAS. Of 90 other cDNAsidentified, most represent one or more mRNAs enrichedin rat brain myelin. Five cDNAs, which encode knownproteins, were characterized for mRNA size(s), enrich-ment in myelin, and tissue and developmental expressionpatterns. Two of these, peptidylarginine deiminase andferritin heavy chain, have recognized roles in myelination.The corresponding mRNAs were of different sizes thanthe previously identified mRNA, and they had tissue anddevelopment expression patterns that were indistin-guishable from those of MBP mRNA. Three other cDNAsrecognize mRNAs whose proteins (SH3p13, KIF1A, anddynein light intermediate chain) are involved in membranebiogenesis. Although enriched in myelin, the tissue anddevelopmental distribution patterns of these mRNAs dif-fered from those of MBP mRNA. Six other cDNAs, whichdid not share significant sequence homology to knownmRNAs, were also examined. The corresponding mRNAswere highly enriched in myelin, and four had tissue anddevelopmental distribution patterns indistinguishablefrom those of MBP mRNA. These studies demonstratethat MSAS contain a diverse population of mRNAs,whose locally synthesized proteins are placed to contrib-ute to myelin sheath assembly and maintenance. Char-acterization of these mRNAs and proteins will help pro-vide a comprehensive picture of myelin sheath assembly.Key Words: Dynein light intermediate chain—Ferritinheavy chain—Myelination—Kinesin—Peptidylarginine de-iminase—SH3p13—Suppression subtractive hybridization.J. Neurochem. 75, 1834–1844 (2000).

Individual oligodendrocytes assemble many myelinsheaths, some near the soma and some at considerabledistances from the cell body. To efficiently incorporatemyelin basic protein (MBP) and select myelin-associated

oligodendrocytic basic proteins (MOBPs) into thesheaths, the corresponding mRNAs are transported toand translated in each myelin sheath assembly site(MSAS). These small, highly basic proteins, confinedmainly to the major dense line (MDL), are believed tofunction in MDL formation and maintenance (Omlinet al., 1982; McLaurin et al., 1993; Yamamoto et al.,1994). Many oligodendrocytes in mutants that lack MBP(Privat et al., 1979; Ganser and Kirschner, 1980; Rosen-bluth, 1980; Matthieu et al., 1981) or MOBP (Yamamotoet al., 1999) carry out initial steps of myelination nor-mally. It therefore seems likely that additional proteins,some of which are synthesized in MSAS, participate inthe generation of myelin sheaths. Like MBP and MOBPmRNAs, the mRNAs for these proteins would be ex-pected to be transported to MSAS. The goal of this studywas to identify mRNAs that co-localize with MBP andMOBP mRNAs in MSAS. Further characterization ofthese mRNAs will surely broaden our understanding ofhow myelin sheaths are assembled.

Previous studies have shown that the only mRNAsselectively enriched in rat brain myelin are located inMSAS (Colman et al., 1982; Gillespie et al., 1990a,b;Gould et al., 1999a). Subcellular fractionation was alogical first step to use to separate MSAS RNA fromother neural mRNAs. However, additional steps wereneeded as contaminating non-MSAS mRNAs are alsopresent, though not enriched, in the myelin fraction(Gould, 1998; Gould et al., 1999a). We used PCR-basedsuppression subtractive hybridization (SSH) in conjunc-tion with northern blot analysis to isolate and identify

Received March 24, 2000; revised manuscript received June 7, 2000;accepted July 7, 2000.

Address correspondence and reprint requests to Dr. R. M. Gould atDepartment of Pharmacology, NYS Institute for Basic Research inDevelopmental Disabilities, 1050 Forest Hill Rd., Staten Island, NY10314, U.S.A. E-mail: [email protected]

Abbreviations used:CNP, 29,39-cyclic nucleotide phosphodiesterase;DLIC, dynein light intermediate chain; ds, double-stranded; EST, ex-pressed sequence tag; FHC, ferritin heavy chain; LSS, low-speedsupernatant; MBP, myelin basic protein; MDL, major dense line;MOBP, myelin-associated oligodendrocytic basic protein; MSAS, my-elin sheath assembly site(s); PAD, peptidylarginine deiminase; PLP,proteolipid protein; RACE, rapid amplification of cDNA ends; SSH,suppression subtractive hybridization.

1834

Journal of NeurochemistryLippincott Williams & Wilkins, Inc., Philadelphia© 2000 International Society for Neurochemistry

MSAS mRNAs. In this study, we report the partialcharacterization of 11 cDNAs isolated by this methodwith respect to mRNA size, enrichment in myelin, tissuedistribution, and developmental appearance. These re-sults suggest that in addition to MBP and MOBPmRNAs, a number of low-abundance mRNAs are trans-ported to MSAS. This transport would allow the encodedproteins to more readily participate in myelin sheathassembly. Furthermore, sequence analysis of fiveMSAS-enriched cDNAs reveals strong similarity toknown mRNAs, suggesting new functions (posttransla-tional modification of MBP, iron metabolism, endocyto-sis, and microtubule-based vesicle transport) for MSAS.

MATERIALS AND METHODS

AnimalsCD rats (Charles River Breeding Laboratory, Wilmington,

MA, U.S.A.) were used for all studies. Handling and killing ofthe rats followed protocols that had received prior approvalfrom the institute’s Animal Welfare Committee.

Subcellular fractionation and preparation of mRNAThe subcellular fractionation methods are based on a recent

study (Gould et al., 1999a). As in this study, adult rat brains(cerebral hemispheres) were used as starting material. Freshlydissected brains were homogenized in 12.5 ml of ice-coldsucrose (both 0.32 and 0.85M sucrose were used) in a 15-mlDounce homogenizer. Solutions were buffered with 10 mMHEPES (pH 7.4) and 3 mM dithiothreitol (Gillespie et al.,1990a). Prior to addition of tissue, prime RNase inhibitor (30U; Five Prime3 Three Prime, Boulder, CO, U.S.A.) wasadded to the homogenization solution to prevent RNA degra-dation. After the samples were homogenized, they were cen-trifuged (1,000 rpm for 10 min) to pellet nuclei and cell debris.The low-speed supernatant (LSS) was removed and a portion(600ml) reserved. For samples homogenized in 0.32M sucrose,the remaining LSS was adjusted to 0.85M sucrose with added2.8M buffered sucrose. Density gradients were simplified fromthose used in the earlier study (Gould et al., 1999a). Thismodification allowed us to use only two tubes per brain. Thegradients were formed with 6 ml of 1.2M sucrose at thebottom, 6–7 ml of sample (at 0.85M sucrose) in the middle,and a 0.25M sucrose overlay. During centrifugation (100,000g for 3.5 h at 4°C in an SW28 swinging bucket rotor), myelinvesicles accumulate at the 0.25M sucrose/0.85M sucroseinterface. Material at this interface was collected, dispersed inan equal volume of 10 mM MgCl2, and pelleted (12,500 rpmfor 20 min). Total RNA was extracted from the reserved LSSand from the myelin pellets with Tri reagent (Molecular Re-search Center, Cincinnati, OH, U.S.A.). For the SSH studies,total RNA (100–150mg) from both LSS and myelin was usedto prepare mRNA with the Micro FastTrack kit (InvitrogenCorp., Carlsbad, CA, U.S.A.).

SSHThe PCR-Select cDNA Subtraction kit protocol (Clontech,

Palo Alto, CA, U.S.A.) was followed. In brief, driver (“non-specific”) and tester (containing mRNAs of interest) double-stranded (ds) cDNAs were synthesized from;1 mg of LSS andmyelin mRNAs, respectively, with reagents supplied in the kit.After LSS and myelin ds-cDNAs were digested withRsaIrestriction enzyme, the myelin (tester) sample was divided intotwo portions. Each portion was ligated with a different adapter.

For the first hybridization, each denatured adapter-ligated testersample was mixed individually with excess denatured drivercDNA to allow common sequences to cross-hybridize andleave some “tester-specific” cDNAs as single-stranded mole-cules. For the second hybridization, the two hybridization re-actions were combined along with additional denatured drivercDNA. During this hybridization, single-stranded molecules inthe one tester cDNA anneal to complementary single-strandedmolecules in the second tester cDNA. The resultant productcontains molecules with different adapters at each end. Thispopulation is referred to as the subtraction product cDNA. Taqpolymerase was used to catalyze filling of the ends of therenatured products with dNTP. Subtraction products were thenamplified with primers specific to each adapter. Moleculeshaving the same adapter at both ends are poorly amplified dueto suppressive effects (Diatchenko et al., 1996, 1999).

Three SSH studies were conducted (Table 1), two with LSSand myelin RNA prepared from samples homogenized with0.32 M sucrose and one with LSS and myelin RNA preparedfrom a sample homogenized in 0.85M sucrose. Products fromSSH1 were subcloned into pGEM T-Easy (Promega, Madison,WI, U.S.A.). Products from SSH2 and SSH3 were subclonedinto pCR 2.1 TOPO (Invitrogen).

PCR analysis of subtraction efficiencyWe used PCR analysis of unsubtracted and subtracted cDNA

populations to assess the efficiency of the SSH procedures, asrecommended (Diatchenko et al., 1996, 1999). The amplifiedunsubtracted tester control cDNA template (U; Fig. 1) wasprepared as recommended (Clontech manual). The subtractedtemplate (S; Fig. 1) was an aliquot of the cDNA obtained afterthe two rounds of hybridization and subsequent two rounds ofPCR amplification (Diatchenko et al., 1996, 1999). Primers(Table 1) were designed with a primer design program (www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) and madecommercially (Integrated DNA Technologies, Coralville, IA,U.S.A.). Primers were made against mRNAs that we knew wereenriched in myelin and against mRNAs expected to be removedand were designed so that the resulting PCR products did notcontain anRsaI site.

Identification of cDNAs derived from MBP andMOBP mRNAs

As the subtracted cDNA population was expected to beenriched in cDNAs derived from MBP and MOBP mRNAs,steps were taken to identify them and eliminate them fromsubsequent analyses. Clones that contained MBP or MOBPinserts were identified by colony hybridization and/or by hy-bridization to inserts released during digestion of miniprepsamples. Near full-length MBP was made with 59-GAA-GAGACCCTCACAGCGAC (position 36, accession no.K00512) and 59-TTCTGAGCTCCTCATCCCTG (position1,445) as primers and a template prepared from myelin mRNAwith the Marathon cDNA amplification kit (Clontech). Nearfull-length MOBP cDNA was prepared with 59-ATACCTG-CAGGGCAACAAAG (position 16, MOBP-81A, accessionno. X87900) and 59-AACGCTGTGCATGTGTTAGC (posi-tion 2,304) as primers and with the myelin template. The PCRproducts were purified (Wizard PCR purification kit; Promega),radiolabeled (Prime-a-Gene kit; Promega) with [a-32P]dCTP(Amersham-Pharmacia Biotech, Piscataway, NJ, U.S.A.), andpurified on MicroSpin S-300 columns (Amersham-PharmaciaBiotech). Approximately 109 dpm of labeled probe was usedfor each blot. Hybridization conditions were those used fornorthern blot studies.

J. Neurochem., Vol. 75, No. 5, 2000

1835mRNAs ENRICHED IN RAT BRAIN MYELIN

Northern blots: mRNA distributions in myelinfraction and in tissues

The northern blot procedure has been described (Gouldet al., 1999a). In brief, total RNA (10mg/lane, determined byabsorbance at 260 nm) was separated through 1% agarose gelsthat contained 2.2M formaldehyde, transferred to Nytran su-percharge membranes (Schleicher & Schuell, Keene, NH,U.S.A.), UV-cross-linked, prehybridized 1–4 h at 42°C, andhybridized overnight with cDNA probe at the same tempera-ture. The cDNA inserts used as probes were prepared byamplification of individual clones with primers directed againstthe two adapter sequences (nested primers 1 and 2R; Clontechkit). They were purified, radiolabeled, purified again, dena-tured, and added to blots. After hybridization, the blots werewashed two times in 23 saline/sodium phosphate/EDTA(SSPE) at room temperature (10 min each), one time in 23SSPE/2% sodium dodecyl sulfate at 65°C (30–45 min each),and then in 0.23 SSPE at room temperature for 10–15 min. For

the tissue and developmental distribution studies, total RNAwas prepared from frozen tissues with Tri reagent.

Sequencing and sequence analysisPlasmids were either sequenced at the Joseph Bay Paul

Center for Comparative Molecular Evolution (Woods Hole,MA, U.S.A.) or at the Molecular Resources Center (UMDNJ–NJMS, Newark, NJ, U.S.A.). Sequences were used as queriesfor Blast and Advanced Blast analyses over the Internet (www.ncbi.nlm.nih.gov/blast). Both nucleotide (blastn) and protein(blastx) searches were made of the nonredundant GenBankdatabase. Clones whose cDNA sequences did not match se-quences in this database were used as queries for blastnsearches against the existing expressed sequence tag (EST)databases. Matching ESTs (representative ESTs are listed foreach clone; see Table 4) were used as query for searching theUnigene cluster databases at NCBI (www.ncbi.nlm.nih.gov/UniGene) or the assembled human EST database (www.mips.biochem.mpg.de/proj/human/human_blast.html). ESTs, whichwere contained within an existing Unigene cluster, were alsoused as query for the EST assembler program (www.hercules.tigem.it/cgi-bin/uniestass.pl) to extract the longest contig. Thiscontig was used as a query for blastn and blastx searchesagainst the existing GenBank and Swissprot NR databases aswell as against dbEST to obtain additional ESTs.

RESULTS

Selection of cDNAs that represent mRNAs enrichedin rat brain myelin by SSH

Although the myelin fraction from rat brain is highlyenriched in MBP and select MOBP mRNAs, it alsocontains mRNAs derived from other neural sites (Col-man et al., 1982; Gould, 1998; Gould et al., 1999a). Weused SSH to select cDNAs that represent mRNAs en-

TABLE 1. Primer pairs used to test efficiency of SSH

mRNA Cellular/intracellularAccession

no. Fwd. Fwd. primer sequence Rev. Rev. primer sequenceSize(nt)

MBP OL processes M25889 93 CAGCAAGTACCATGGACCAT 293 ATGTTCTTGAAGAAGTGGAC 220MOBP OL processes X87900 76 TTGCCAGATGGGAGCTTGAA 283 GGCAGGCACAGCAGATCCAG 227CA II OL soma X58294 33 AAGAGCAACGGACCAGAGAA 374 CCCCATATTTGGTGTTCCAG 361CNP OL soma M18630 219 CAAGGGTCGTTCAAGTTAGGC 512 GGCCACGCAGGATGAACAGC 313Gelsolin OL soma J04953 409 AATGAGGTGGTGGTCCAGAG 891 GCTTGCCTTTCCAAACAAAG 502PLP OL soma X02809 122 TGTGTTTCTTTGGAGTGGCA 849 GGCCCATGAGTTTAAGGACG 857GFAP Astrocytes L27219 413 GGTGGAGAGGGACAATCTCA 1,116 ATGGTGATGCGGTTTTCTTC 723Glutamine

synthase Astrocytes X07921 302 ACCCGAGTGGAACTTTGATG 781 AAAACGGGCTACCCAGAGAT 499S-100b Astrocytes X01090 187 GGTGACCAAGCACAAGCTGAA 869 TCGTTTGCACAGAGGACAAG 702NF-L Neuronal soma M25638 81 AAAGGTGCACGAGGAAGAGA 442 TGTCTGCATTCTGCTTGTCC 381NSE Neuronal soma M11931 264 GTGGACCACATCAACAGCAC 869 AGGGTCAGCAGGAGACTTGA 625Ca/calmodulin

kinase II Neuronal dendrites AJ222796 5,643 AGCACTGGCTCTCAACCTGT 6,143 TTGCACCATTACCACCTTCA 501MAP-2 Neuronal dendrites X51842 3,410 CAAAGAGAAAGGTGGCAAAGC 3,917 TTACCACGGACTCAACCACA 527Cyclophilin General M25637 99 TTVGAGCTGTTTGCAGACAA 491 AGATGGTGATCTTCTTGCTG 412GDH General X02231 591 ACCACAGTCCATGCCATCAC 1,023 TCCACCACCCTGTTGCTGTA 452

Primers specific for mRNAs located in each major cellular/intracellular neural site were prepared (see Materials and Methods) and used to amplifyproducts in reactions with subtracted and unsubtracted templates. Messenger RNAs are grouped according to their cellular/intracellular location.Sample reaction results are shown in Fig. 1. Accession nos. for the cDNAs used to design PCR primers are listed, as are the start sites of the forward(Fwd.) and reverse (Rev.) primers, which are listed from 59 and 39 ends, respectively. The nos. are based on the sequences given by the accessionno. The reverse primer is amplified from the designated position. Product sizes are given as number of nucleotides (nt). CA II, carbonic anhydraseII; GFAP, glial fibrillary acidic protein; NF-L, low molecular weight neurofilament protein; NSE, neuron-specific enolase; MAP-2, microtubule-associated protein-2; GDH, glyceraldehyde 3-phosphate dehydrogenase.

FIG. 1. SSH selects cDNAs thatrepresent mRNAs enriched in ratbrain myelin. PCR reactions (30 mleach) were run with primers toMBP, MOBP, CNP, S-100b, andglyceraldehyde 3-phosphate de-hydrogenase (GDH) (Table 1) asdescribed in the Clontech PCR-Select cDNA subtraction kit usermanual with S (subtracted) and U(unsubtracted control) templates.Samples (5 ml/time point) weretaken after 18 (lanes 1), 23 (lanes2), 28 (lanes 3), and 33 (lanes 4)cycles. Products were of ex-pected sizes (Table 1).

J. Neurochem., Vol. 75, No. 5, 2000

1836 R. M. GOULD ET AL.

riched in myelin and to exclude cDNAs that representmRNAs present in myelin at low abundance relative toother sources. PCR analysis of subtraction efficiency (seeMaterials and Methods) was used to determine the ef-fectiveness of SSH through comparisons of the abun-dance of known cDNAs in subtracted and unsubtractedcontrol templates (Fig. 1) with a variety of neural celltype-specific primers (Table 1).

MBP and MOBP cDNAs served as positive controlsfor mRNAs located in MSAS. The MBP cDNA PCRproduct amplified equally well in both subtracted andunsubtracted reactions (Fig. 1). As the enrichment pro-cedure should allow MSAS mRNAs to be more rapidlyamplified in the subtracted sample, this result indicatesthat levels of MBP mRNA in reactions with the unsub-tracted control template are already at sufficiently highlevels to allow robust amplification after only 18 cycles.In contrast, the PCR product obtained with MOBP prim-ers was detected in the reaction with subtracted templatemuch sooner than in the reaction with unsubtracted tem-plate (18 vs. 28 cycles; Fig. 1). This difference suggestsa 400-fold enrichment of MOBP cDNA in the subtractedtemplate, because a 20-fold difference is estimated forproducts that are detected five cycles apart (Clontechmanual). The MOBP primers used amplify exon 3, anexon shared by all known MOBP mRNAs (Gould et al.,1999a; McCallion et al., 1999); hence, the PCR productreflects mainly MOBP-81A mRNA, the most abundantMOBP mRNA in rat brain (Yamamoto et al., 1994).These results demonstrate that SSH positively selectsDNAs derived from MOBP mRNAs and that, as ex-pected, MOBP levels are much lower than comparableMBP mRNA levels.

A wide spectrum of primers was used to examine theability of SSH to reduce or eliminate cDNAs that repre-sent low-abundance mRNAs in myelin from the subtrac-tion product. Representative results with primers for29,39-cyclic nucleotide phosphodiesterase (CNP; for oli-godendrocyte somal mRNA), S-100b primers (for astro-cyte mRNAs), and glyceraldehyde 3-phosphate dehydro-genase primers (for mRNAs with broad cellular distri-bution) are shown (Fig. 1). None of the reaction pairs

gave results similar to those obtained for the positive(MBP and MOBP) controls. Instead, these primers [aswell as primers for carbonic anhydrase II, gelsolin, pro-teolipid protein (PLP), glial fibrillary acidic protein, glu-tamine synthase, cyclophilin, andb-actin mRNAs]yielded PCR products that were detected after far fewercycles with the unsubtracted control template than withthe subtracted template. PCR reactions to amplifycDNAs derived from mRNAs located in neuronal soma(low molecular weight neurofilament protein and neu-ron-specific enolase) or in neuronal dendrites (calcium/calmodulin-dependent kinase II and microtubule-associ-ated protein-2) gave no detectable products in reactionwith either unsubtracted control or subtracted templates(data not shown). These results indicate that SSH iseffective in eliminating non-myelin-enriched cDNAsfrom the myelin-enriched cDNA population.

Identification of cDNAs that represent non-MBP/MOBP mRNAs enriched in rat brain myelin

In total, three different SSH experiments (1, 2, and 3)were performed, each with different rat brains (Table 2).SSH1 product was PCR amplified in three separate exper-iments (A, B, and C). Of 1,080 total clones analyzed,.60% (648) were found to be derived from MBP orMOBP mRNAs. Numerous duplicate clones were obtainedin SSH1C, most likely due to variability in the PCR am-plification. As well, many clones were derived from mito-chondrial DNA or repetitive DNA sequences. Of the 90distinct cDNAs that were unrelated to MBP, MOBP, mito-chondria, or repetitive DNA, 50 are classified as “known”because they matched sequences in the nonredundant Gen-Bank database. The other 40 are classed as “unknown”because 32 matched sequences in the EST database and 8did not match sequences in either database. Interestingly, asignificantly higher number of distinct cDNAs were ob-tained from the SSH2 product (54 total) than from the otherexperiments. This result suggests that homogenization inhypertonic (0.85M) sucrose traps a more varied populationof mRNAs in myelin than is trapped when samples arehomogenized in near-isotonic (0.32M) sucrose. We hadshown previously that higher amounts of total RNA are

TABLE 2. Identification of cDNAs obtained by SSH

SSH Homogenization Clones MBP/MOBP NR EST Novel

1A 0.32M sucrose 54 47 2 1 21B 229 195 5 6 01C 576 293 4 6 32 0.85M sucrose 159 80 36 17 13 0.32M sucrose 62 33 2 2 2Total 1,080 648 49 32 8

Three screenings (A–C) were performed with SSH1 and one each with SSH2 and SSH3. The numbers ofcDNAs (clones) analyzed for each screen and the numbers that were found MBP or MOBP positive byhybridization are given. Of those sequenced: NR (hits in the nonredundant GenBank database), EST (hits in thedbEST database), and Novel (no hits in either NR or dbEST databases). All database searches were conductedbetween 2/16/00 and 2/18/00. The differences between the total numbers of clones analyzed and thoserepresented by MBP/MOBP, NR, EST, and Novel are due to duplications. Most likely, the PCR conditions werenot optimal for SSH1C, as numerous sequences were replicated many times.

J. Neurochem., Vol. 75, No. 5, 2000

1837mRNAs ENRICHED IN RAT BRAIN MYELIN

trapped in myelin prepared from 0.85M sucrose homoge-nates than in myelin prepared from 0.32M sucrose homog-enates (Gould et al., 1999a). Among the cDNAs obtained inSSH2 were ones with sequence identity to glial fibrillaryacidic protein, glutamine synthase, gelsolin, 3-hydroxyl-3-methylglutaryl (HMG)-CoA reductase, and MVP17/rMal.When analyzed by northern blot and/or PCR efficiency,none of these mRNAs/cDNAs were enriched in myelin(data not shown). In this article, we focus on five cDNAsrelated to proteins in the nonredundant GenBank database(Figs. 2 and 4) and six cDNAs unrelated to any nonredun-dant proteins (Figs. 3 and 4). To date, we have analyzed 28cDNAs in the known group and 10 sequences in the un-known group by northern blot for enrichment in myelin.Seventeen (61%) of the knowns recognized mRNA speciesthat were enriched in myelin. All 10 of the unknownsrecognized mRNAs enriched in myelin.

Clones with similarity to known proteinsFive cDNAs were chosen for further study because the

respective proteins have recognized relationships to my-elination [peptidylarginine deiminase (PAD) and ferritinheavy chain (FHC)], endocytosis (SH3p13), or microtu-bule-based organelle transport [KIF1A and dynein lightintermediate chain (DLIC-2)]. In each case, the cDNA

sequence was 94–99% identical to the published se-quence (Table 3). These values are within the range ofsequence identity expected from one-pass sequencingand/or from animal/strain differences. In all cases, thecDNAs recognized mRNAs that were of different sizesfrom published mRNA sequences (Table 3). This resultindicates that the “myelin” mRNAs, with the exceptionof PAD (which is smaller), contain additional nucleotidesequences. In the case of SH3p13, the smaller size of thepublished sequence may indicate that it is incomplete;that is, it lacks the true 59 end.

Four of the five cDNAs hybridized to mRNAs that areenriched five- to 10-fold in myelin over the LSS startingmaterial. In addition to the PAD and SH3p13 mRNAs,the larger FHC and DLIC-2 mRNAs, designated FHCmand DLIC-2m, respectively (Fig. 2A, thick arrows), were

FIG. 3. Northern blot analysis with cDNA probes prepared toMBP, FHC, SH3p13, KIF1A, and DLIC-2 and to cDNAs 63OR23and 66NB08. Expression levels are compared for cerebral hemi-sphere total RNAs from 1-, 3-, 6-, 10-, 16-, 20-, 27-, and 60-dayrats. Exposure times are 3 h (MBP), 3 days (FHC), 3 days(SH3p13), 3 days (KIF1A), 3 days (DLIC-2), 14 days (63OR23),and 18 h (66NB08).

FIG. 2. Northern blot analysis with cDNA probes for MBP,MOBP, and five “known” SSH products. “Knowns” are highlyrelated to PAD-II, FHC, SH3p13, KIF1A, and DLIC-2 (Table 3). A:Expression levels are compared for LSS and myelin (M). B:Expression levels are compared among pancreas (P), cerebel-lum (C), brainstem (B), testis (T), heart (H), and kidney (K). Expo-sure times for LSS and M blots are 2.5 h (MBP), 14 h (MOBP),3.5 h (PAD), 7 days (FHC), 4 days (SH3p13), 20 h (KIF1A), and3.5 h (DLIC-2). Exposure times for B are the same except forMOBP (3.5 days). Thick arrows for FHC and DLC-2 point to mRNAs(designated m in text) enriched in myelin. Thin arrows point tomRNAs that are not or are only slightly enriched. Asterisks indicateenriched PADm and FHCm in tissue distribution blots.

J. Neurochem., Vol. 75, No. 5, 2000

1838 R. M. GOULD ET AL.

both five- to tenfold enriched in the myelin fraction.However, the major FHC mRNA (Fig. 2A, thin arrows)was only slightly enriched in myelin, and the majorDLIC-2 mRNA of 4.3 kb (Hughes et al., 1995) was notenriched in myelin (Fig. 2A, thin arrows). Finally,63OA01 cDNA (having sequence identity to mouseKIF1A) recognized two mRNAs that were enriched to alesser degree (two- to threefold) in myelin (Fig. 2A).

The distributions of these mRNAs were comparedamong RNAs from six tissues (Fig. 2B). The expressionof MBP and MOBP-81A mRNAs, included for compar-ison, are detected solely in neural tissues. Both mRNAsare expressed more strongly (roughly twofold) in whitematter-rich brainstem versus cerebellum. The myelin-enriched 4-kb PAD mRNA (PADm, indicated by aster-isk, Fig. 2B) is expressed differently from a larger-sizedmRNA. Like MBP and MOBP RNAs, PADm is ex-pressed only in neural tissues, with greater expression inbrainstem than in cerebellum. In contrast, the larger PADmRNA, detected in all tissues, is the same size (4.5 kb)as skeletal muscle PAD type II (Watanabe and Senshu,1989; Terakawa et al., 1991). The larger FHCm (indi-cated by asterisk, Fig. 2B) is expressed similarly to MBP,MOBP, and PADm mRNAs. The smaller FHC mRNA,identical in size (0.83 kb) to the known FHC (Wu et al.,1997), is expressed strongly in all tissues examined. Inneural tissues, SH3p13 is also expressed more strongly inbrainstem than in cerebellum. As previously shown

(Ringstad et al., 1997), its expression is strongest intestis. Unlike the other mRNAs, KIF1A and DLIC-2mare both expressed more strongly in cerebellum than inbrainstem. Like SH3p13, DLIC-2m is expressed moststrongly in testis.

Among these five mRNAs, only FHCm and PADm(not shown) mRNAs had the same developmental ex-pression pattern, as did the MBP mRNAs (Fig. 3). Incontrast, SH3p13, KIF1A, and DLIC-2m mRNAs weredetectable at earlier times. SH3p13, DLIC-2m, and thesmaller KIF1A appeared to be constitutively expressedthroughout development. The larger KIF1A mRNA wasexpressed more variably. These findings lend support tothe notion that unique FHCm and PADm mRNAs aretransported to MSAS at the same times as MBP andMOBP-81A mRNAs.

Clones with similarity to ESTsNearly 50% (40/89) of cDNAs identified in this study

are unrelated to cDNAs in the nonredundant GenBankdatabase (Table 1). Furthermore, many of those withidentities in the nonredundant GenBank databasematched cDNAs encoding poorly characterized proteinsand/or limited (20–50%) portions of known cDNAs/mRNAs. Therefore, to obtain a comprehensive picture ofthe role local protein synthesis plays in MSAS function,the proteins represented by uncharacterized cDNAs mustbe characterized. As a starting point, we selected sixcDNAs from the group of unknowns and analyzed themfor enrichment in myelin, tissue distribution, and devel-opmental expression patterns (Figs. 3 and 4). We plan toobtain full-length cDNA sequences for each of them anduse the sequences to identify protein motifs that will givehints as to the protein’s function(s). All the unknownsexamined in this study were obtained from SSH1. Fourof them were related to cDNAs in the GenBank dbESTdatabase (Table 4).

Clone 63NA25 is closely related to rat ESTs, whichare part of Unigene cluster Mm.34007, and it has se-quence similarity to yeast helicase protein. The Unigenecluster Rn.17993 is closely related (59/73 positions,80%) to the 39 untranslated region of human assembledESTs H55494S1 (Unigene Hs.154248), which encodesthe novel human KIAA0951 protein (Nagase et al.,1998). Clone 66NB08 shows significant homology tomouse EST AI553013 and human AA116649. The latterEST is part of Unigene cluster Mm.28626. The proteinencoded by this assembled mouse cluster is related to aGTP binding protein (59% identical of 109 residues).Clone 66NB08 may therefore be derived from an mRNAthat encodes a novel GTP binding protein.

Clone Sh3-53 is identical to rat EST AI785357 and ishighly related (97%) to several mouse ESTs. Both the ratEST and the associated mouse Unigene cluster are re-lated to a human assembled cluster (Unigene Hs.75400),which encodes the novel human KIAA0280 protein (Na-gase et al., 1998). Clone SH3-227 is related to mouseEST AA260950. The mouse EST is part of a Unigenecluster, which is highly related to a human cluster

FIG. 4. Northern blot analysis with cDNA probes for 63OR23,63NA25, 63NA65, 66NB08, SH3-227, and SH3-53. Four of thesesequences are related to sequences in the EST database (Table4). A: Expression levels are compared between LSS and myelin(M). B: Expression levels are compared among pancreas (P),cerebellum (C), brainstem (B), testis (T), heart (H), and kidney (K).Exposure times for both A and B are 14 days (63OR23), 7 days(63NA25), 10 days (63NA65), 1 day (66NB08), 10 days (SH3-227), and 6 days (SH3-53). Arrow points to SH3-53 mRNAenriched in myelin.

J. Neurochem., Vol. 75, No. 5, 2000

1839mRNAs ENRICHED IN RAT BRAIN MYELIN

Hs.11114. Neither sequence contains a long open read-ing frame.

Clone 63OR23 did not show similarity to any se-quences in the dbEST database queried by blastn but didshow similarity to ESTs contained in assembled humanEST contig H43534S1 (Unigene cluster Hs.227327) hav-ing 45 of 66 (68%) identities, which are present in the 39untranslated region of H4354S1. The open reading framein H4354S1 encodes the novel human KIAA0951 pro-tein (g458546), which is related to the rat brain-specificprotein neuroligin-3 (56% amino acid identity) (Nagaseet al., 1998). Neuroligin-3 is related to neuroligin-1, aneuronal cell surface protein that binds to a subset ofb-neurexins that are also localized to neurons (Ichtch-enko et al., 1996). Clone 63OR23 may therefore be aderivative of the 39 untranslated region of the mRNAencoding rat neuroligin-3 or a related protein.

Expression of mRNAs related to unknown cDNAsThe main mRNAs recognized by five of the six

cDNAs examined are all highly (five- to tenfold) enrichedin myelin (Fig. 4A). Three cDNAs (63OR23, 63NA25, andSH3-227) recognize a single mRNA only. The other threecDNAs recognize several mRNAs. Clones 63NA65 and66NB08 recognize larger low-abundance mRNAs that arealso enriched in myelin. SH3-53 recognizes a main largemRNA that is not enriched in myelin and several interme-diates, two of which appear to be enriched in myelin.

63OR23, 63NA25, and 66NB08 cDNAs recognize 2-,3.5-, and 3-kb mRNAs, respectively, that, like MBP andMOBP mRNAs, are expressed solely in neural tissues andpreferentially in brainstem (twofold over cerebellum; Fig.4B). All three have the same developmental expressionpattern as MBP mRNA (Fig. 3; 63NA25 is not shown).SH3-53 recognizes a low molecular weight mRNA (2 kb)that is also expressed with the same tissue (Fig. 4B) anddevelopmental (not shown) patterns as MBP mRNA.63NA65 cDNA recognizes three to four mRNAs that are allhighly enriched in myelin. The most abundant isoform isexpressed most strongly in testis, followed by cerebellum,pancreas, brainstem, and kidney. Its developmental expres-sion pattern mirrors that of DLIC-2m and SH3p13; expres-sion levels are unchanged throughout postnatal develop-ment. SH3-227 cDNA recognizes a 5-kb mRNA that isexpressed in all tissues except pancreas. It is expressed moststrongly in testis and slightly more strongly in heart andkidney than in brainstem and cerebellum. This mRNA isexpressed at similar levels throughout postnatal develop-ment (data not shown).

DISCUSSION

Subcellular fractionation and in situ hybridizationstudies locate both MBP and MOBP mRNAs in MSAS.Complementary kinetic studies demonstrate that MBP,synthesized from MSAS-located mRNA, enters myelin

TABLE 3. Relation of cDNAs to known mRNAs

cDNA mRNA Size (kb) Published size (kb) Accession no. Identity Region

63NA26 PAD 4 4.5 J05022 455/458 1,082–1,53963NA34 FHC .1 0.8 U58829 368/385 220–604SH4-23 SH3p13 1.8 1.5 AF227439 538/543 650–1,19163OA01 KIF1A 9 5.4 NM008440 594/628 236–862SH4-63 DLIC-2 6 4.3 U15138 549/567 2,002–2,568

The cDNAs from SSH1B (63NA26, 63NA34, and 63OA01) and from SSH2 (SH4-23 and SH4-63) havesequences identical within experimental error to published sequences for PAD, FHC, SH3p13, KIF1A, andDLIC-2. The size of the mRNAs in myelin (Size), recognized by the designated cDNA, is compared with thesize of the published cDNA (Published size). Accession nos., ratio of nucleotides identical to the stretch ofnucleotides recognized from the published sequence (identity), and the regions of identity are given.

TABLE 4. Relatedness of cDNAs to GenBank EST database sequences

cDNA No. EST EST UniCluster Longest contig Similarities

63NA25 11 AA819478 Rn. 17993 618 None25 AW121043 Mm. 34007 589 Q09775 RNA Helicase 50% (X)

66NB08 24 AI553013 Mm. 30075 1,344 MSE55 59% (N)116 AA116649 Mm. 28626 968 GTP BP 59%, P34280 (X)

SH3-53 2 AI785357 Mm. 38855 585 KIAA0280 92% (N) D8747022 AI315312 408/431

SH3-227 AA667378 NoneAA690780 None

131 AA260950 1,989 AL137597125 N90816 2,322 Same

No. EST is the number of hits with scores of less than e-12. EST is the accession no. of the EST with thehighest match. UniCluster is the accession no. of the closest matching sequence. Similarities are for proteinsrelated to the sequence of the longest contig.

J. Neurochem., Vol. 75, No. 5, 2000

1840 R. M. GOULD ET AL.

membrane with little delay, compared with protein (e.g.,PLP) synthesized from mRNAs located in oligodendro-cyte soma (Colman et al., 1982). MOBP-71, MOBP-81A, MOBP-99, and MOBP-169, synthesized fromMSAS-located mRNAs (Holz et al., 1996; Gould et al.,1999a), are also expected to be rapidly incorporated intomyelin. Using a combination of subcellular fractionationand SSH, we isolated cDNAs that represent 90 candidatemRNAs, many of which are expected to co-localize withMBP and MOBP mRNAs in MSAS. Here we describethe expression patterns of 11 of these mRNAs (Figs.2–4). One subset, PADm, FHCm, and mRNAs recog-nized by 63OR23, 66NB08, 63NA25, and SH3-53, havetissue (brainstem. cerebellum) and developmental(postnatal day@16 to adult) expression patterns that areindistinguishable from those of MBP and MOBPmRNAs. The combination of enrichment in myelin andtissue and developmental expression patterns indicatesthat PAD, FHC, and proteins derived from mRNAs rec-ognized by 63OR23, 66NB08, 63NA25, and SH3-53 arealso synthesized in MSAS.

A second subset of mRNAs, all of which co-localizewith MBP and MOBP mRNA in rat brain myelin (Figs.2A and 3A), has different tissue and developmentalexpression patterns. The mRNAs for SH3p13, KIF1A,and DLIC-2m and those that are recognized by 63NA65and SH3-227 are all expressed more strongly in cerebel-lum than in white matter-rich brainstem. Furthermore,some of them (DLIC-2m, SH3p13, and mRNAs recog-nized by 63NA65 and SH3-227) are expressed morestrongly in testis than in nervous system tissues. Noneare expressed coordinately with myelination. They are allexpressed in early postnatal development before oligo-dendrocytes develop. Currently, we are using in situhybridization to determine the locations of these mRNAsduring early postnatal development.

A new PAD mRNA in MSASPAD mRNA translation products are enzymes that

deiminate arginine residues contained within protein.The only known myelin protein to be so modified isMBP. In normal human white matter, roughly 20% ofMBP is the C-8 charge isomer; 6 of 19 arginine residuesare modified (Wood and Moscarello, 1989). BecauseC-8, the least cationic MBP charge isomer, is mostabundant during early development, it may play a role inearly myelination steps (Moscarello et al., 1994). It wasrecently shown that the PAD enzyme is highly enrichedin myelin (Pritzker et al., 1999), a property consistentwith synthesis in MSAS. A PAD protein, roughly thesize of PAD-II, was shown to be expressed in culturedoligodendrocytes (Akiyama et al., 1999). It seems rea-sonable that myelinating oligodendrocytes synthesizePAD-II (or a closely related protein) coordinately withits principal substrate, MBP. With both PAD and MBPsynthesis occurring in MSAS, production of C-8 wouldbe tightly regulated.

Although the sequence of the PAD mRNA concen-trated in rat brain myelin is highly similar to PAD-II

(Table 3), its overall size is smaller (Fig. 2B) and thusrepresents a new isoform. Sequences obtained from 39-and 59-rapid amplification of cDNA ends (RACE) matchthe 3,721 nucleotides (within 99%) at the 39 end ofPAD-II (data not shown). The 59 end of the mRNA inmyelin is clearly different. A cDNA probe made to the 59end of PAD-II (residues 225–725) did not detect PADmRNA enriched in myelin by northern blot (data notshown). These results suggest that the difference be-tween myelin-enriched PAD and PAD-II resides at the 59end of the mRNA, which could result in a PAD proteinhaving a distinct N-terminal amino acid sequence fromPAD-II. It has been suggested that variability in theN-terminal regions of known PAD isoforms confersspecificity onto the enzyme, whereas the more highlyconserved C terminus contains the catalytic domain (Ish-igami et al., 1998). Alternative splicing at the N terminuscould therefore provide substrate specificity for MBP.Whether or not PAD-II in myelin has specificity forMBP will require expression of the protein and enzy-matic analysis. It should be noted that the PAD in myelinis located appropriately to modify other proteins locatedin MSAS, including several MOBPs.

In addition to PAD modification, charge isomers of MBPare deamidated (Deibler et al., 1975), N-terminal acylated(Moscarello et al., 1992), arginine methylated (Kim et al.,1997), and phosphorylated (Miyamoto and Kakiuchi, 1975;Martenson et al., 1983). All these modifications (for review,see Moscarello, 1997) could take place with enzymes syn-thesized in MSAS. The high amounts of methyltransferase(Kim et al., 1997) and protein kinase (Ledeen, 1992) activ-ities in myelin fractions may indicate that these enzymes arealso located in MSAS.

FHC mRNA in MSASThe presence of FHC mRNA in MSAS is not entirely

unexpected. FHC is used for iron storage (Theil, 1990),and roles for iron in myelination are well established(Larkin and Rao, 1990; Connor and Menzies, 1996;Sanyal et al., 1996). However, it is not yet known if anyof the iron-dependent enzymes involved in the synthesisor degradation of myelin lipids (Connor and Menzies,1996) are located in MSAS. Studies to locate theseenzymes and their mRNAs to myelin are clearly war-ranted.

Somewhat surprising are the sizes of FHC mRNA(s)in myelin. Three to four FHC mRNAs are clearly en-riched in myelin. They are 1–2 kb in size, larger than theknown 0.83-kb FHC mRNA (Wu et al., 1997). Some ofthese large mRNAs may contain an added 39 untrans-lated region, recognized in human FHC (Percy et al.,1998). FHC mRNA with the added 39 untranslated re-gion, like our mRNAs, is enriched in adult tissues (Percyet al., 1998). Our result clearly shows that FHCm ex-pression from the novel mRNAs is up-regulated inMSAS coordinately with MBP and MOBP mRNAs andsuggests a role for iron-dependent events in MSAS.

J. Neurochem., Vol. 75, No. 5, 2000

1841mRNAs ENRICHED IN RAT BRAIN MYELIN

SH3p13 mRNA: implications for endocytosisin MSAS

SH3p13 is a binding partner of synaptojanin and dy-namin, two proteins that function in neurotransmitterrecycling (Ringstad et al., 1997). This protein is veryclosely related to SH4p4 (Sparks et al., 1996; Ringstadet al., 1997), a protein that possesses lysophosphatidicacid acyltransferase activity (Scales and Scheller, 1999;Schmidt et al., 1999). Acyltransferase activity wouldenable SH3p13 to alter the lipid composition of mem-branes in MSAS, possibly as a means to coordinateendocytosis and/or exocytosis (see below). It should benoted that SH3p13 mRNA is expressed differently fromMBP and MOBP mRNAs. It is more highly expressed incerebellum than in brainstem. Furthermore, high expres-sion is detected early in the postnatal brain. In situhybridization studies to locate SH3p13 mRNA in devel-oping and adult white matter and studies to identifyproline-rich proteins that associate with its SH3 domainmay help determine the role(s) of SH3p13 in myelina-tion.

Molecular motor protein mRNAs located in MSASThe present study (see also Gould et al., 1999b) iden-

tifies mRNAs to several microtubule-based motor pro-teins in MSAS, two with high sequence homology toKIF1A (Okada et al., 1995) and one with high sequencehomology to DLIC-2 (Hughes et al., 1995). The KIF1Aand DLIC-2 mRNAs enriched in myelin are clearlylarger than the known KIF1A and DLIC mRNAs (Table3). Using 39- and 59-RACE, we found that the DLIC-2enriched in myelin has the same 39 end as the publishedsequence. Characterization of more 59 sequences willenable us to find out if the encoded protein is larger thanthe known DLIC-2. In addition to these mRNAs, we alsoidentified mRNAs for a novel KIF2 and KIF5B that areenriched in myelin (Gould et al., 1999b). Thus, a varietyof motor proteins are likely to be synthesized in MSAS.In the recent study (Gould et al., 1999b), we used in situhybridization to show that KIF1A mRNA is located inoligodendrocyte processes in vivo and in granules simi-lar to those that transport MBP mRNA in cultured oli-godendrocytes (Ainger et al., 1997). Further studies areneeded to better define all the kinesin and dynein/dyn-actin family members synthesized in MSAS and to iden-tify the organelles and destinations transported by each.Also, additional sequence information will help to locateRNA transport sequences associated with these and othermRNAs identified in this study.

Possible roles for endocytosis and retrogradetransport in MSAS

Our current understanding of CNS myelination isbased mostly on the behavior of MBP and PLP. Asillustrated in Fig. 5B, PLP is transported from syntheticsites in oligodendrocyte soma to MSAS. Upon its arrivalin MSAS, membrane fusion places PLP in oligodendro-cyte plasma membrane as a prerequisite to exposingexternal domains that function in intraperiod line com-

FIG. 5. A schematic diagram that illustrates movements of proteinstransported to and synthesized in MSAS. A: Low-power view fororientation: The oligodendrocyte (OL) process viewed with transportvesicles, MSAS, compact myelin, and axon surface exposed in cut-away. B: An enlargement to show positions and movements of fourclasses of myelin proteins represented by PLP, uncharacterized mem-brane proteins (UPm) transported with PLP, and MBP and uncharac-terized cytoplasmic proteins (UPc) synthesized in MSAS. Polyribo-somes locate protein synthesis to MSAS. MBP, MOBPs, and otherproteins (for simplicity, all are represented by a U-shaped symbol)stabilize the MDL. Other proteins (represented by UPc, open square),which would include PAD and FHC, are also synthesized on polyribo-somes in MSAS but are not incorporated into myelin. UPm representsproteins that participate in myelin membrane compaction that do notbecome incorporated into myelin. C: Retrieval of vesicles enriched inproteins removed from MSAS (UPc and UPm) and transported backto oligodendrocyte soma. This process requires proteins such asSH3p13 and DLIC-2, whose levels would be controlled by local syn-thesis. IPL, intraperiod line.

J. Neurochem., Vol. 75, No. 5, 2000

1842 R. M. GOULD ET AL.

paction. SH3p13 may function in PLP vesicle fusion (seeabove). Because compaction occurs quite normally inmutant mice that lack PLP (Rosenbluth et al., 1996;Klugmann et al., 1997), other proteins co-transportedwith PLP (Fig. 5B, triangles) may contribute to intrap-eriod line formation.

In contrast to PLP and other myelin-related integralmyelin proteins (e.g., myelin-associated glycoprotein,myelin oligodendrocyte glycoprotein, myelin oligoden-drocyte specific protein/claudin), MBP is synthesized onmRNAs that are located in MSAS (for review, see Bro-phy et al., 1993; Kalwy and Smith, 1994). The presenceof MBP mRNA in MSAS allows MBP translation prod-uct (Fig. 5B, U-shaped symbols) ready access to sitesthat compact and become the MDL. Other myelin pro-teins including MOBP-81 (could also be represented asthe U-shaped MBP symbol in Fig. 5B) are also incorpo-rated into compact myelin membranes and contribute toMDL compaction (Yamamoto et al., 1994, 1999). Inaddition to MBP and MOBP, many other proteinspresent in low abundance (e.g., PAD and FHC) aresynthesized in MSAS (Fig. 5B, squares). We proposethat some of these proteins help coordinate myelin sheathassembly. For example, PAD modifies the charge onMBP, influencing its interactions with lipids (Boggset al., 1999) and likely its role in myelin membranecompaction. FHC isoforms influence iron availability inMSAS and possibly enzymes of lipid metabolism impor-tant for the formation of myelin membranes with properlipid composition.

Not all proteins (Fig. 5B and C, those represented bysquares and triangles) are destined to reside in the com-pact myelin membrane. In fact, the presence of some ofthese proteins in compact myelin could be disruptive;that is, they could destabilize the myelin lamellae. Wesuggest that mechanisms, which may include SH3p13 forendocytosis and kinesin and dynein for organelle trans-port, could function to remove these proteins, once theirrole in membrane assembly is completed and return themto oligodendrocyte soma (Fig. 5C).

Future studies needed to gain understanding ofuncharacterized mRNAs

In addition to the known cDNAs, which are enrichedin MSAS, many cDNAs (44% of those identified; Table1) potentially code for proteins whose function(s) is notyet established. Knowledge of the structural motifs con-tained within the putative proteins will surely enlarge ourappreciation of how assembly of myelin membranesoccurs in MSAS.

Acknowledgment: National Multiple Sclerosis Societygrants PP0502, PP0589, and RG2944-A-1 to R.M.G. and Na-tional Multiple Sclerosis Society grant RG 2578-B-5 to D.L.F.helped support this work. Ms. Mary Ellen Cafaro, Ms. SharonMathier, and Ms. Jill Gregory prepared the figures. Drs. HilaryMorrison and Mitchell Sogin provided invaluable assistancewith sequencing at the MBL. The technical support staff atClontech provided invaluable advice, which added to the suc-cess of this project.

REFERENCES

Ainger K., Avossa D., Diana A. S., Barry C., Barbarese E., and CarsonJ. H. (1997) Transport and localization elements in myelin basicprotein mRNA.J. Cell Biol. 138,1077–1087.

Akiyama K., Sakurai Y., Asou H., and Senshu T. (1999) Localizationof peptidylarginine deiminase type II in a stage-specific immatureoligodendrocyte from rat cerebral hemisphere.Neurosci. Lett.274,53–55.

Boggs J. M., Rangaraj G., Koshy K. M., Ackerley C., Wood D. D., andMoscarello M. A. (1999) Highly deiminated isoform of myelinbasic protein from multiple sclerosis brain causes fragmentation oflipid vesicles.J. Neurosci. Res.57, 529–535.

Brophy P. J., Boccaccio G. L., and Colman D. R. (1993) The distri-bution of myelin basic protein mRNAs within myelinating oligo-dendrocytes.Trends Neurosci.16, 515–521.

Colman D. R., Kreibich G., Frey A. B., and Sabatini D. D. (1982)Synthesis and incorporation of myelin polypeptides into CNSmyelin. J. Cell Biol. 95, 598–608.

Connor J. R. and Menzies S. L. (1996) Relationship of iron to oligo-dendrocytes and myelination.Glia 17, 83–93.

Deibler G. E., Martenson R. E., Kramer A. J., and Kies M. W. (1975)The contribution of phosphorylation and loss of COOH-terminalarginine to the microheterogeneity of myelin basic protein.J. Biol.Chem.250,7931–7938.

Diatchenko L., Lau Y.-F. C., Campbell A. P., Chenchik A., MooadamF., Huang B., Lukyanov S., Lukyanov K., Gurskaya N., SverdlovE. D., and Siebert P. D. (1996) Suppression subtractive hybrid-ization: a method for generating differentially regulated or tissue-specific cDNA probes and libraries.Proc. Natl. Acad. Sci. USA93, 6025–6030.

Diatchenko L., Lukyanov S., Lau Y. F. C., and Siebert P. D. (1999)Suppression subtractive hybridization: a versatile method for iden-tifying differentially expressed genes.Methods Enzymol.303,349–380.

Ganser A. L. and Kirschner D. A. (1980) Myelin structure in theabsence of basic protein in the shiverer mouse, inNeurologicalMutations Affecting Myelination(Baumann N., ed), pp. 171–176.North Holland Biomedical Press, Amsterdam.

Gillespie C. S., Bernier L., Brophy P. J., and Colman D. R. (1990a)Biosynthesis of the myelin 29,39-cyclic nucleotide 39-phosphodi-esterases.J. Neurochem.54, 656–661.

Gillespie C. S., Trapp B. D., Colman D. R., and Brophy P. J. (1990b)Distribution of myelin basic protein and P2 mRNAs in rabbitspinal cord oligodendrocytes.J. Neurochem.54, 1556–1561.

Gould R. M. (1998) Compartmentalization of brain mRNAs deter-mined by subcellular fractionation.J. Neurochem.70 (Suppl. 1),S53.

Gould R. M., Freund C., and Barbarese E. (1999a) Myelin-associatedoligodendrocytic basic protein (MOBP) mRNAs reside at differ-ent subcellular locations.J. Neurochem.73, 1913–1924.

Gould R., Freund C., Palmer F., Knapp P. E., Huang J., Morrison H.,and Feinstein D. L. (1999b) Messenger RNAs for kinesin anddynein are located in neural processes.Biol. Bull. 197,259–260.

Holz A., Schaeren-Wiemers N., Schaefer C., Pott U., Colello R. J., andSchwab M. E. (1996) Molecular and developmental characteriza-tion of novel cDNAs of the myelin-associated oligodendrocyticbasic protein.J. Neurosci.16, 467–477.

Hughes S. M., Vaughan K. T., Herskovits J. S., and Vallee R. B. (1995)Molecular analysis of a cytoplasmic dynein light chain interme-diate chain reveals homology to a family of ATPases.J. Cell Sci.108,17–24.

Ichtchenko K., Nguyen T., and Su¨dhof T. C. (1996) Structures, alter-native splicing, and neurexin binding of multiple neuroligins.J. Biol. Chem.271,2676–2682.

Ishigami A., Kuramoto M., Yamada M., Watanabe K., and Senshu T.(1998) Molecular cloning of two novel types of peptidylargininedeiminase cDNAs from retinoic acid-treated culture of a newbornrat keratinocyte cell line.FEBS Lett.433,113–118.

Kalwy S. A. and Smith R. (1994) Mechanisms of myelin basic proteinand proteolipid protein targeting in oligodendrocytes (rev.).Mol.Membr. Biol.11, 67–78.

J. Neurochem., Vol. 75, No. 5, 2000

1843mRNAs ENRICHED IN RAT BRAIN MYELIN

Kim S., Lim I. K., Park G. H., and Paik W. K. (1997) Biologicalmethylation of myelin basic protein: enzymology and biologicalsignificance.Int. J. Biochem. Cell. Biol.29, 743–751.

Klugmann M., Schwab M. H., Pu¨hlhofer A., Schneider A., Zimmer-mann F., Griffiths I. R., and Nave K. A. (1997) Assembly of CNSmyelin in the absence of proteolipid protein.Neuron18, 59–70.

Larkin E. C. and Rao G. A. (1990) Importance of fetal and neonataliron: adequacy for normal development of central nervous system,in Brain, Behavior and Iron in the Infant(Dobbing J., ed), pp.43–63. Springer-Verlag, London.

Ledeen R. W. (1992) Enzymes and receptors of myelin, inMyelin:Biology and Chemistry(Martenson R. E., ed), pp. 531–570. CRCPress, Boca Raton, Florida.

Martenson R. E., Law M. J., and Deibler G. E. (1983) Identification ofmultiple in vivo phosphorylation sites in rabbit brain myelin.J. Biol. Chem.258,930–937.

Matthieu J.-M., Ginalski-Winkelmann H., and Jacque C. (1981) Sim-ilarities and dissimilarities between two myelin deficient mutantmice, shiverer and mld.Brain Res.214,219–222.

McCallion A. S., Stewart G. J., Montague P., Griffiths I. R., and DaviesR. W. (1999) Splicing pattern, transcript start distribution, andDNA sequence of the mouse gene (Mobp) encoding myelin-associated oligodendrocytic basic protein.Mol. Cell. Neurosci.13,229–236.

McLaurin J., Ackerley C. A., and Moscarello M. A. (1993) Localiza-tion of basic proteins in human myelin.J. Neurosci. Res.35,618–628.

Miyamoto E. and Kakiuchi S. (1975) In vitro and in vivo phosphory-lation of myelin basic protein by exogenous and endogenous39,59-monophosphate-dependent protein kinases in brain.J. Biol.Chem.249,2769–2777.

Moscarello M. A. (1997) Myelin basic protein, the “executive” mole-cule of the myelin membrane, inCell Biology and Pathology ofMyelin (Juurlink B. H. J., Devon R. M., Doucette J. R., NazaraliA. J., Schreyer D. J., and Verge V. M. K., eds), pp. 13–25. PlenumPress, New York.

Moscarello M. A., Pang H., Pace-Asciak C. R., and Wood D. D. (1992)The N terminal of human myelin basic protein consists of C2, C4,C6 and C8 alkyl carboxylic acids.J. Biol. Chem.267,9779–9782.

Moscarello M. A., Wood D. D., Ackerley C., and Boulias C. (1994)Myelin in multiple sclerosis is developmentally immature.J. Clin.Invest.94, 146–154.

Nagase T., Ishikawa K., Miyajima N., Kotani H., Nomura N., andOhara O. (1998) Prediction of the coding sequences of unidenti-fied human genes IX. The complete sequences of the codingsequences of 100 new cDNA clones from brain which can codefor large proteins in vitro.DNA Res.5, 31–39.

Okada Y., Yamazaki Y., Sekine-Aizawa Y., and Hirokawa N. (1995)The neuron-specific kinesin superfamily protein KIF1A is aunique monomeric motor for anterograde axonal transport ofsynaptic vesicle precursors.Cell 81, 769–780.

Omlin F. X., Webster H. D., Palkovits C. G., and Cohen S. R. (1982)Immunocytochemical localization of basic protein in major denseline regions of central and peripheral myelin.J. Cell Biol. 95,242–248.

Percy M. E., Wong S., Bauer S., Liaghati-Nasseri N., Perry M. D.,Chauthaiwale V. M., Dhar M., and Joshi J. G. (1998) Iron me-

tabolism and human ferritin heavy chain cDNA from adult brainwith an elongated untranslated region: new findings and insights.Analyst123,41–50.

Pritzker L. B., Nguyen T. A., and Moscarello M. A. (1999) Thedevelopmental expression and activity of peptidylarginine deimi-nase in the mouse.Neurosci. Lett.266,161–164.

Privat A., Jacque C., Bourre J. M., Dupouey P., and Baumann N.(1979) Absence of the major dense line in myelin of the mutantmouse shiverer.Neurosci. Lett.12, 107–112.

Ringstad N., Nemoto Y., and De Camilli P. (1997) The SH3p4/SH3p8/SH3p13 protein family: binding partners for synaptojanin anddynamin via a grb-2-like src homology 3 domain.Proc. Natl.Acad. Sci. USA94, 8569–8574.

Rosenbluth J. (1980) Central myelin in the mutant mouse shiverer.J. Comp. Neurol.194,639–648.

Rosenbluth J., Stoffel W., and Schiff R. (1996) Myelin structure inproteolipid protein (PLP)-null mouse spinal cord.J. Comp. Neu-rol. 371,336–344.

Sanyal B., Polak P. E., and Szuchet S. (1996) Differential expression ofthe heavy-chain ferritin gene in non-adhered and adhered oligo-dendrocytes.J. Neurosci. Res.46, 187–197.

Scales S. J. and Scheller R. H. (1999) Lipid membranes shape up.Nature401,123–124.

Schmidt A., Wolde M., Thiele C., Fest W., Kratzin H., PodtelejnikovA. V., Witke W., Huttner W. B., and So¨ling H.-D. (1999) En-dophilin I mediates synaptic vesicle formation by transfer ofarachidonate to lysophosphatidic acid.Nature401,133–141.

Sparks A. B., Hoffman N. G., McConnell S. J., Fowlkes D. M., andKay B. K. (1996) Cloning of ligand targets: systematic isolation ofSH3 domain-containing proteins.Nat. Biotechnol.14, 741–744.

Terakawa H., Takahara H., and Sugawara K. (1991) Three types ofpeptidylarginine deiminase: characterization and tissue distribu-tion. J. Biochem. (Tokyo)110,661–666.

Theil E. C. (1990) The ferritin family of iron storage proteins.Adv.Enzymol.63, 421–449.

Watanabe K. and Senshu T. (1989) Isolation and characterization ofcDNA clones encoding rat skeletal muscle peptidylarginine de-iminase.J. Biol. Chem.264,15255–15260.

Wood D. D. and Moscarello M. A. (1989) The isolation, characteriza-tion, and lipid-aggregating properties of a citrulline-containingmyelin basic protein.J. Biol. Chem.264,5121–5127.

Wu C.-G., Groenink M., and Bosma A. (1997) Rat ferritin-H: cDNAcloning differential expression and localization during hepatocar-cinogenesis.Carcinogenesis18, 53–58.

Yamamoto Y., Mizuno R., Nishimura T., Ogawa Y., Yoshikawa H.,Fujimura H., Adachi E., Kishimoto T., Yanagihara T., and SakodaS. (1994) Cloning and expression of myelin-associated oligoden-drocytic basic protein. A novel basic protein constituting thecentral nervous system myelin.J. Biol. Chem.269,31725–31730.

Yamamoto Y., Yoshikawa H., Nagano S., Kondoh G., Sadahiro S.,Gotow T., Yanagihara T., and Sakoda S. (1999) Myelin-associatedoligodendrocytic basic protein is essential for normal arrangementof the radial component in central nervous system myelin.Eur.J. Neurosci.11, 847–855.

J. Neurochem., Vol. 75, No. 5, 2000

1844 R. M. GOULD ET AL.