- Email: [email protected]
Gangliosides and allied glycosphingolipids in human peripheral nerve and spinal cord
ELSEVIER
et Biodwsica Acta Biochimica et Biophysics Acta 1214 (1994) 11.5-123
~angliosides and allied glycosphingolipids in human peripheral nerve and spinal cord Lars Svennerholm a,*, Kerstin Bostrijm b, Pam Fredman a, Birgitta Jungbjer ‘, Annika Lekman a, Jan-Eric M%nsson a, Britt-Marie Rynmark a aDepurtment ofChical neuroscience, Section of P~ch~i~
and ~eur~~e~~t~~ Universes ofGiiteborg, ~~lnda~ hospital, S-431 80 ~~l~d~~, Sweden h Department of Forensic Medicine, University of Gijteborg, G&eborg, Sweden Received 16 March 1994
Abstract Glycosphingolipids were determined in human spinal cord, cauda equina and femoral nerve of 10 subjects aged 20-70 years and in dorsal and ventral roots of four subjects aged 17-60 years. Myelin was isolated from corresponding tissue. Axons were isolated from the four specimens of dorsal and ventral roots. The concentration (mean and standard error of mean) of gangliosides in spinal cord was 0.80 f 0.03 pmol sialic acid/g fresh tissue, in cauda equina 0.40 f 0.02 @mot/g and in femoral
nerve 0.23 & 0.01 pmol/g. In spinal cord only trace amounts of glycosphingoiipids of the lacto series were found, and the ganglioside pattern differed from that in cerebral white matter by a relativeiy high proportion of GD3 and a low proportion of GDla. The ganglioside patterns were almost identical in cauda equina and femoral nerve - the major ganglioside being 3’-LMl, 0.07 and 0.04 kmol/g respectively. Another ganglioside of the lacto series, 3’-HexLMl, was 25% of 3’-LMl. Peripheral nerve also contained three acidic glycosphingolipids in addition to sulfatide - LKl and HexLKl belonging to the glycosphingolipid iacto series and containing glucuronyl-3-sulfate instead of sialic acid, and inositolphospho~l galactosylceramide. The dorsal (sensory and ventral (motor) roots had the same major membrane lipid composition but the ganglioside concentration was 30% higher in sensory than motor nerve and myelin. The patterns of gangliotetraose gangliosides were, however, the same in motor and sensory myelin and axons. The ceramide composition of the gangliosides is also reported. Key words: Peripheral nerve; Spinal cord; My∈ Axon; Ganglioside;
Abbreviations: The gangliosides and allied glycolipids have been designated according to the principles suggested by Svennerholm (Eur. J. Biocbem. (1977) 79, 11-21). GM3, I13NeuAc-LacCer; GD3, I13(NeuAc),-LacCer; GM2, I13NeuAc-GgOsesCer; GD2, I13(NeuAcJz-GgOsesCer; GMl, I13NeuAc-CgOse,Cer; GDla, IV3NeuAc,I13NeuAc-GgOse,Cer; GDlb, I13(NeuAc),-GgOse,Cer; GTlb, IV3NeuAc,I13(NeuAc),-GgOse,Cer; GQlb, IV3(NeuAc), I13(NeuAc),-Gg Ose,Cer; 3’-LMZ, IV3NeuAc-nLc~se~~r; 3’HexLMl, V13NeuA~-n~s~~Cer; LDl, (NeuAc),-nLcOse,Cer; 3’,8’-LDl, IV3(NeuAc),-nLcOse,Cer; LK1, IV3glucuronyI-3-sulfatenLcOse,Cer; HexLKl, V13glucuronyI-3-sulfate-nkcOse&er; gal, galactose; NeuAc, N-acetylneuraminic acid; BSA, bovine serum albumin; CNS, central nervous system; CSF, cerebrospinal fluid; GBS, Guillain-Barr6 syndrome; IPGC, inositolphosphoryl galactosylceramide; PNS, peripheral nervous system. * Corresponding author. Fax: + 46 31 862422. 00052760/94/$07.00 c
0 1994 Elsevier Science B.V. AI1 rights reserved
Acidic glycosphingolipid
1. Im~r~u~tion Results from several studies have demonstrated that some neuropathy syndromes are associated with increased titers of serum antibodies against carbohydrate epitopes of glycolipids and glycoproteins [l-6]. Among patients with IgM par~protei~emia who develop a slowly progressive demyelinating sensorimotor neuropathy, approx. 50% have serum antibodies against myelin-associated glycoprotein (MAGI and glycolipids with terminal glucuronyl-3-sulfate linked to galactose [4,7,8]. Patients with m~itifo~al motor neuropathy and some patients with lower motor syndromes (for review
116
L. Srennerholm
et al. / Biochimica
see Ref. [9]) have antibodies directed against glycolipids with a terminal galactosyl ~1-4 N-acetylgalactosaminyl moiety which occurs in the gangliosides GM1 and GDlb and the neutral glycolipid GAl. Several studies have demonstrated antibodies to peripheral nerve tissue and myelin in Guillain-Barre syndrome and in chronic inflammatory demyelinating polyradiculoneuropathy. The target antigens for these antibodies have not been identified, but two recent studies have reported a high frequency of anti-GM1 antibody [lO,ll]. We have not been able to confirm these findings in Scandinavian and German patients, but have found increased titers of 3’-LMl and sulfatide, the two major acidic glycolipids of peripheral nerve myelin [12,13]. There is thus evidence that some patients with peripheral neuropathies have humoral antibodies directed against carbohydrate epitopes. One major task is now to search for a causal relationship between the content and localization of the antigen recognized by the antibody and the patient’s clinical picture. In a previous study, the major membrane lipid composition of peripheral nerve (cauda equina) was examined and compared with that of spinal cord [14]. The study showed a lower proportion of cerebroside, a slightly lower proportion of cholesterol but a significantly higher proportion of sphingomyelin in cauda equina than in spinal cord. In the present study the gangliosides and allied acidic glycolipids of peripheral nerve have been isolated and characterized. The same acidic glycolipids have also been assayed in dorsal and ventral spinal roots and in their myelin and axons, to study whether a different composition of the glycolipids in sensory and motor nerves could explain the variations in motor and sensory expression of various forms of polyneuropathy.
2. Material and methods 2.1. Tissue specimens
The lower thoracal and lumbosacral portions of spinal cord and cauda equina were obtained from the Department of Forensic Medicine, from 17 subjects aged 17-91 years who had died from accidents or acute circulation failure with no signs of neurological disease. The dissection of the spinal roots (cauda equina) and spinal cord and the isolation of myelin with three different methods [15-171 were described in the preceding paper [14]. Myelin of spinal cord and cauda equina was isolated from 13 subjects. The dorsal (motor) and ventral (sensory) spinal roots of four subjects 17-60 years of age were dissected separately. Myelin and axons were isolated essentially as described by Micko and Schlaepfer [18].
et Biophysics Acta 1214 (1994) 115-123
The study was approved by the Ethics Committee for Medical Research of the University of Goteborg. 2.2. Characterization of myelin Protein profiles of myelin and axonal fractions were analyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 20% homogeneous gels (43 X 50 + 0.45 mm) in a Phast-System from Pharmacia according to the program given in the manual (Phast-SystemTM Separation Technique File No. 220). 2 pg of protein was applied to the gel. The gels were stained with Coomassie Brilliant Blue (Phast-SystemTM Developmen t File 200). The protein patterns of motor and sensory myelin did not differ, the major protein being P,. The axonal fraction had a significantly different protein pattern, and P, was only weakly discernible. 2.3. Chemicals Silica gel (230-400 mesh), TLC silica gel 60 and high performance thin-layer plates (HPTLC) were purchased from Merck (Darmstadt, Germany). TLCELISA was performed on precoated plastic sheets from PolygramR Sil G (Marchery and Nagel, Duren, Germany) or glass plates, SI-HPF Silica Gel TLC Plate (J.T. Baker, Philipsburg, NJ, USA). Sephadex G-25 fine and DEAE-Sepharose were from Pharmacia (Uppsala, Sweden). The anion exchange resin, SpherosilDEAE-dextran and cholera toxin B-subunit (CT-B) were gifts from Institut Merieux (Lyon, France). Ganglioside and neutral glycolipid isolated from human organs and characterized in our laboratory using component analysis, permethylation and assay of methylated sugars and FAB-MS were used as standards. Gangliosides of the ganglio series were also obtained from Fidia Research Laboratories (Abano Terme, Italy). Gangliosides GM1 and GTlb used as internal standards were labeled by tritiation of the sphingosine double bond [ 191. The specificity of the anti-ganglioside murine monoclonal antibodies produced in our laboratory and used in this study has recently been reported [20]. In addition to these antibodies, a new anti-3’-LMl ganglioside IgM murine monoclonal antibody @L-l), specific to the terminal epitope NeuAqx2-3GalPl-4GlcNAc, was used for the quantitative determination of 3’-LMl and 3’-HexLMl. HNK-1 murine monoclonal antibody was produced from a HNK-1 hybridoma obtained from American Type Culture Collection (Rockville, MD, USA). It was used for the quantitative assay of LKl and HexLKl. Alkaline phosphatase-conjugated goat antimouse IgM and IgG (heavy and light chains) were purchased from Jackson Immunoresearch Laboratory (West Grove, PA, USA).
L. Suennerholm et al. / Biochimica et Biophysics Acta 1214 (I 994) 115-123
2.4. Lipid extraction and isolation of total ganglioside fraction
Fresh tissue (0.5-10 g) was homogenized with 3 vol. of water in a Potter-Elvehjem homogenizer and myelin (10 mg) with 0.7 ml of water. Chloroform/methanol (1: 2, v/v) was then added to the suspension to give a final ratio of chloroform/ methanol/ water (4 : 8 : 3, v/v) [21]. The lipids were extracted for 1 h at room temperature, the extraction being facilitated by repeated mixing on a cyclomixer, and a clear supernatant was obtained by low speed centrifugation. The pellet was reextracted with the same solvent mixture and volume and under the same conditions, and the two supernatants were then combined and evaporated. The crude lipid extract of 5-10 g tissue was dissolved into 20 ml of chloroform/ methanol/ water (6.5 : 25 : 4, v/v) and put onto a column with 10 g of silica gel slurried in chloroform. The bulk of lipids was eluted with 80 ml of chloroform/ methanol/ water (65 : 25 : 4, v/v) and gangliosides and allied glycolipids with 150 ml of chloroform/ methanol/ water (3 : 6 : 2, v/v). When the tissue amount was lower than 5 g the size of the silica gel column and the volumes of eluting solvents were correspondingly reduced. The ganglioside fraction was evaporated, dissolved in water and dialyzed against running tap water for 3 days. The fractions of whole cauda equina and myelin of cauda equina were divided into two equal portions, one of which was treated with 0.1 M NaOH in methanol/water (1: 1, v/v) at room temperature overnight. After neutralization with acetic acid, the fraction was dialyzed and treated as the nonsaponified fraction. The two ganglioside fractions were dissolved into chloroform/ methanol/ water (60 : 30 : 4.5, v/v) and chromatographed on SpherosilDEAE-dextran [22]. Monosialogangliosides were eluted with 10 vol. of 0.02 M potassium acetate in methanol and the oligosialogangliosides with 10 vol. of 0.5 M potassium acetate in methanol. The resin was finally eluted with 10 vol. of 1.0 M potassium acetate in methanol to recover LKl and HexLKl. 3’-LMl and 3’-HexLMl were isolated from the monosialoganglioside fraction by a combination of silica gel column chromatography by elution with chloroform/ methanol/ water (60 : 30 : 5, v/v) and by preparative TLC with chloroform/ methanol/ 2.5 M ammonia (50 : 40 : 10, v/v). The corresponding neutral glycosphingolipids were obtained by sialidase hydrolysis, and separated from GM1 and GM2 on Spherosil-DEAEdextran. LKl and HexLKl were isolated from the oligosialoganglioside fraction by sialidase hydrolysis, which converted all gangliosides of the ganglio series to GM2 and GMl. After dialysis against distilled water GM2 and GM1 were eluted from a small SpherosilDEAE-dextran column with 0.02 M potassium acetate in methanol, and LKl and HexLKl were eluted with
117
1.0 M potassium acetate in methanol. Inositolphosphoryl-2(3)galactosylceramide was determined on the monosialoganglioside fraction by thin-layer chromatography. 2.5. Analytical methods Quantitative composition of the carbohydrate moieties was determined as alditol acetates by GLC with mannose as the internal standard [24]. Ganglioside sialic acid was determined on the total ganglioside fraction and isolated individual gangliosides using the resorcinol method [25]. Sphingosine was assayed by a modification of the methyl orange method of Lauter and Trams [26], and used for the quantification of glycolipids other than gangliosides. The fatty acid and sphingosine patterns were determined by GLC [27]. The glycolipids were further characterized by permethylation, and by analyses of partially methylated acetates [28]. FAB-MS was performed on a model VG 7070 E mass spectrometer equipped with a fast-atom gun 1291. 2.6. Silica gel thin-layer chromatography TLC was performed on precoated TLC and HPTLC plates, which were developed with chloroform/ methanol/0.25% aqueous KC1 (50: 40: 10, v/v) or chloroform/ methanol/ 2.5 M ammonia (50 : 40 : 10, v/v). Neutral glycolipids (after sialidase hydrolysis) were also developed with chloroform/ methanol/ water (65:25 :4, v/v). Th e plates were stained with resorcinol-HCl reagent [25] for gangliosides and densitometric scanning of the plates was performed at 620 nm on a CAMAG TLC scanner. The plates were also stained with orcinol-H,SO, and scanned at 515 nm for neutral glycosphingolipids and gangliosides [30]. 2.7. Sialidase treatment Isolated gangliosides and mono- and oligosialoganglioside fractions, corresponding to 5-10 nmol of sialic acid, were treated with sialidase from V. cholerae [20]. The hydrolyzed samples were chromatographed on HPTLC plates with the solvent systems described above. One plate was stained with resorcinol-HCl reagent and the other with orcinol-H,SO,. 2.8. TLC immunostaining Ganglioside and allied glycosphingolipids were also characterized and quantified with the murine monoclonal antibodies, using the procedure recently described [201. Gangliosides of the ganglio series were also assayed with cholera toxin-B subunit (CT-B) after sialidase hydrolysis [20].
L. Swnnerholm et al. /Biochimica et Biophysics Acta 1214 (1994) 115-123
118 Table 1 Composition
of gangliosides
Glycospingolipid
and allied glycosphingolipids
Spinal cord (n = 10) glycosphingolipid (nmol/g wet wt.)
GM4 GM3 GM2 GM1 3’-LMl 3’.HexLM 1 GD3 GD2 LDl GDla GDlb GTlb GQlb LKl HexLKl IPGC*
in adult human Cauda
o/c sialic acid
S.D.
64
18
(8)
n.d.
25 16 110 n.d. n.d. 60 8 n.d. 28 105 40 14 1.8 trace trace
2 4 20
(3) (2) (14)
6 1
(15) (2)
3 9 3 2
(7) (26) (15) (7)
24 4 36 68 16 18 4 8 30 33 17 4 64 15 154
and relative
equina
mean
S.D.
proportion
of ganglioside
Two monosialo gangliosides were detected with monoclonal antibody SL-1 with the same TLC migration as reference 3’-LMl and 3’-HexLMl. Sialidase hydrolysis of the fastest migrating ganglioside gave a
and allied glycosphingolipids
in adult human
motor
range
GM3 GM2 GM1 3’-LM 1 3’-HexLM 1 GD3 GD2 LDl GDla GDlb GTlb GQlb IPGC LKI HexLKl
20 4 34 66 14 18 5 9 24 28 16 2 198 54 16
16- 23 34 25- 39 61- 69 13- 14 IS- 20 45 8- 10 18- 32 26- 30 14- 17 l2 184-208 50- 59 14- 17
Total ganglioside
356
337-368
glycosphingolipid (nmol/g wet wt.)
(6) (1) (9) (17) (4) (9) (2) (4) (15) (16) (13) (3)
22 5 19 38 IO 7 3 4 19 17 12 3 39 10 70
mean
expressed
in nmol glycosphingolipid
and relative
% sialic acid
(6) (1) (10) (19) (4) (10) (3) (5) (13) (16) (13) (2)
proportion
c/r sialic acid
S.D.
5 1 3 4 2 2 1 1 3 2 1
(9) (2) (8) (16) (4) (6) (2) (3) (16) (14) (16) (4)
1 12 3 27
sialic acid (in parentheses).
and sensory
nerve nerve (n = 4)
glycosphingolipid (nmol/g wet weight) mean
range
25 5 34 85 20 23 8 15 31 38 19 2 279 72 20
2l- 31 45 2% 38 70- 94 16- 26 l6- 32 6Y 1218 2% 36 30- 44 IS- 22 23 236-310 70- 76 20- 21
460
412-545
sialic acid (nmol/g) Values
nerve (n = 10)
% sialic acid
Sensory
Motor nerve (n = 4) glycosphingolipid (nmol/g wet weight) mean
Femoral
nerve
product with the same TLC migration as reference LAl. FAB-MS analysis of the permethylated ganglioside showed the most prominent molecular ion at m/z 1910. This ion corresponds to the composition NeuAc* HexNAc* Hex: Cer with the d18 : l/ 24 : 0 ceramide. Ions of equal intensity at m/z 825 and 793 (825 - 32) were derived from the terminal sequence NeuAc-Hex-HexNAc with a Hexl-4HexNAc linkage. Ions at m/z 376 and 344 (376 - 32) correspond to
3.1. Characterization of gangliosides and allied acidic glycolipids of peripheral nerve
of gangliosides
and femoral
n.d. 3 1 7 9 2 5 1 2 4 5 3 1 10 4 33
3. Results
Tsble 2 Composition
equina
(n = 10)
glycosphingolipid (nmol/g wet wt.)
mean
Values expressed in nmol glycosphingolipid * Inositolphosphoryl galactosylceramide.
spinal cord, cauda
of ganglioside
sialic acid (in parentheses).
o/c sialic acid
(5) (1) (7) (18) (4) (10) (3) (7) (13) (17) (12) (2)
119
L. Suennerholmet al. /Biochimica et BiophysicsActa 1214 (1994) 115-123
3.2. Concentrative of gangl~o~ides and Alfred acidic gly-
terminal NeuAc and at m/z 1029 to NeuAc-HexHexNAc-Hex. The most prominent ion of the ceramide portion was at m/z 660 representing d18: l/24: 0, Analysis of the partially methylat~d alditol acetates from the ganglioside showed the occurrence of 2,4,6Me,-Ga1,2,3,6-Me,-Glc and 3,6-Me,-GlcNAcMe. No 4,6-Me*-Gl~NA~Me corresponding to the Gal l3GlcNAc linkage was detected. The results show that the structure of the fast-migrating SL-1 positive ganglioside is 3’-LMl (XV3NeuAc-nLcOse,Cer). Sialidase hydrolysis of the slowly migrating SL-1 positive ganglioside gave a product with the same TLC migration as reference nI_cOse,Cer. FAB-MS of the permethylated ganglioside showed that the most prominent molecular ion at m/z 2359 corresponded to the composition NeuAc* HexNAcz Hex: Cer with the d18: l/24: 0 ceramide. Fragment ions from the carbohydrate portion of ganglioside were found at m/z 376 and 344 (376 - 32), 580, 825 and 793 (825 - 321, 1029, 1274 and 1242 (1274 - 32) and 1478 representing NeuAc, NeuAc-Hex, NeuAc-Hex-HexNAc, NeuAcHex-HexNAc-Hex, NeuAc-Hex-HexNAc-Hex-HexNAc and NeuAc-Hex-HexNAc-Hex-HexNAc-Hex. Methylation analysis of ganglioside gave 2,4,6-Me,Gal, 2,3,6-Me,-Glc and 3,6-Me,-GlcNAcMe. Again, no 4,6-Me,-GlcNAcMe was detected, which excluded the occurrence of a Gal 1-3GlcNAc linkage. The structural analyses show that the slowly migrating SL-1 positive ganglioside is 3’-hexLM1 (IV”NeuAcnI_.cOse,Cer).
cosphingolipids in spinal cord and peripheral nerve
The concentrations of gangliosides and allied acidic glycosphingolipids are given in Table 1. The ganglioside concentration was highest in spinal cord, mean value F S.E.M. 0.80 rt 0.03 pmol/g fresh weight, significantly lower in cauda equina 0.40 i: 0.02 pmol/g and lowest in femoral nerve 0.23 k 0.01 pmol/g. The ganglioside patterns were virtually identical in cauda equina and femoral nerve, the only small difference was a larger proportion of GD3 and a smaller of GM3 in cauda equina than in femoral nerve. The major ganglioside in peripheral nerve was 3’-LMl (Tables 2 and 3). The pattern differed significantly from that in spinal cord, which mostly contained GM4 and none of the two gangliosides of the lactoseries, 3’-LMl and 3’-HexLMl. The spinal cord ganglioside pattern was further characterized by a large proportion of GD3 and a relatively small proportion of GDla, which gave a dominance of b-series gangliosides. Peripheral nerve also contained three acidic glycosphingolipids which only occurred in trace amounts in spinal cord. Two of them, LKl and I-IexLKl, belong to the giycosphingolipid lacto series and contain glucuronyl
3-sulfate
rather
than sialic acid. Their
concen-
trations were of the same magnitude as those of 3’-LMl and 3’-HexLMl in both cauda equina and femoral nerve (Table 1). Another characteristic component in peripheral
nerve
was ~~ositolphospho~l
galactosylce-
Table 3 Major gangliosides and allied acidic giycosphin~oIipids of spinal cord and peripheral nerve myelin Glycosphingolipid
Spinal cord (n = 4) glycosphingolipid (nmoi/g) mean
Cauda equina (n = 4) % sialic acid
SD.
~lycosphingaiipid (nmol/g)
% sialic acid
mean
S.D.
178 101 19
48 48 3
(8) (4) (1)
n.d. 45 -
7 _
(2)
GM1 3’-LM 1 3’-HexLMl GD3 CD2 LDI GDla GDlb GTlb GQlb LKl
351 n.d. n.d. 163 4 n.d. 65 325 128 42
52
(1.5)
_ 39 1
115) (2)
14 57 18 10
(6) (28) (16) (7)
138 420 60 71 12 23 143 120 71 14 112
17 93 21 41 2 7 28 20 10 2 17
(9) (17) (4) (6) (1) (2) (21) (17) (16) (4)
Total ganglioside sialic acid (nmol/g)
2 360
GM4 GM3 GM2
1750
Values expressed in nmol glycosph~ngolipid per g dry weight of myelin and relative proportion of ganglioside sialic acid (in parentheses).
L. St~ennerholm et al. /Biochimica et Biophysics Acta 1214 (1994) 115-123
120 Table 4 Lipid composition nerves (n = 4)
of myelin
and
axons
from
motor
Motor nerve
Sensory
and
sensory
nerve
myelin
axons
myelin
axons
Proteins a Phospholipids h Cholesterol ” Cerebroside h Sulfatide h Ganglioside sialic acid h Protein-bound sialic acid ’
20 558 502 111 53 1.7 2.8
80 105 80 21 12 0.9 3.8
21 560 483 107 49 2.1 3.3
82 113 86 21 12 1.0 4.4
Ganglioside GDla GDlb GTlb GQlb 3’-LMl 3’Hex-LMI LKl HexLKl
194 117 160 62 9 385 51 193 63
163 113 97 43 I 60 9 70 25
224 155 205 99 15 413 59 199 72
162 121 109 48 8 93 6 80 28
pattern
‘GM1
” Proteins expressed as weight percentage. ” Concentrations in pmol/g dry myelin or g dry axon. ’ Individual glycosphingolipids in nmol/g dry myelin or g dry axon.
ramide (IPGC) which we recently discovered and characterized [39]. The concentration of this glycosphingolipid was more abundant than we had assumed from the preparative isolation of the compound. Trace amounts of these three acidic lipids, LKl, HexLKl and IPGC, were found in spinal cord, which does not prove that they exist in this tissue - it is not possible to exclude the possibility that the low values found for these glycosphingolipids are attributable to an admixture with spinal roots from dissection of the spinal cord. The spinal roots from four cases were dissected in motor and sensory nerves. The concentration of major lipids did not differ significantly between motor and sensory nerves (Table 4). The concentration of gangliosides was approx. 30% higher in sensory than in motor nerves, 0.46 pmol and 0.36 pmol ganglioside sialic acid/g fresh weight, respectively, but although the ganglioside patterns were similar the proportion of GM1 was slightly lower in sensory than in motor nerve. The other important issue was to determine what proportion of the gangliosides was localized to the myelin and what proportion was present in the axons. Our aim was thus not to isolate a “pure” myelin fraction but to obtain a large yield of all forms of myelin, and our criterion on the purity of our myelin was that it did not contain detectable quantities of axonal proteins. Three different procedures - previously shown to give similar results for the major lipids [ 141 - were used for the isolation of myelin from spinal cord and cauda equina [15-171. All three methods gave similar results, but the results in Table 3 are from
spinal cord and cauda equina isolated with Cammer’s method [16] analyzed on the same occasion with the same batches of monoclonal antibodies. In spinal cord, the percentage composition of myelin gangliosides was almost identical to that in intact spinal cord. There was slightly more GM4 and slightly less GD3 and GDla. The latter two gangliosides were also enriched in the fine particles, which did not sediment when the myelin was centrifuged after the osmotic shock, with a relatively low g number (12,000 X g>. Cauda equina myelin did not differ significantly in its composition from intact cauda equina. In both spinal cord and cauda equina myelin, the proportion of GDlb myelin was slightly higher than in the original tissue. The concentration of gangliosides was 30% higher in sensory than in motor myelin (Table 4), but the proportions of the major gangliosides were similar. The finding of a slightly higher proportion of GM1 ganglioside in motor nerve also occurred in the myelin, but the content of GM1 per g myelin was slightly higher in sensory than in motor nerve. The ganglioside concentration was almost the same in sensory and motor axons, and the proportions of major gangliosides were almost identical. Another interesting observation was that the concentration of sialoglycoprotein sialic acid was 4-times higher than that of ganglioside sialic acid in both motor and sensory axons. The ceramide composition of gangliosides in spinal cord (Table 5) and cauda equina (Table 6) was determined on pooled samples from 10 cases. The gangliosides of the gangliotetraose series in spinal cord had
Table 5 Ceramide Fatty acid
composition
of major gangliosides
spinal cord
Ganglioside GM4 normal
16:0 18:0 2O:O 22:o 23:0 24:0 24:l 25:0 25:l 26:O 26:l
in human
CD3
GM1
GDla + GDlb + GTlb
15 56 2 3 4 3 13 1 1 iI 1
4 91 1
10 71 9 3 2 2 3
4 93 trace 3
4 62 2 32
2 36 5 57
hydroxy
16 58
I
6
2 4 13 2 1
15 44 21 7 3 2 2
o/o hydroxy fatty acids 15 Sphingosine d18:O d18: 1 d20:O d20: 1
3 94 trace 3
All values given as molar sphingosines.
percentage
of total fatty acids or of total
L. Svennerholm et al. /Biochimica Table 6 Ceramide Fatty acid
composition Ganglioside normal
of major gangliosides GM3 hydroxy
16:0 18:O 20:o 22:o 23:0 24:0 24:l 25:0 25:l 2610 26:l
13 10 6 15 9 31 11 3
% hydroxy
fatty acids 10
Sphingosine d18:O d18:l d20:O d20: 1
18 19 50 7 4 3
Gangliotetraose series GDla +GDlb + GTlb
cauda
Lactotetraose 3’-LMl
equina series
LKl
5 34 16 12 4 12 I
13 15 15 14 7 22 17 3
30 28 5 8 4 12 9
19 57 2 22
24 72 1 3
20 77 1 2
21 70 3 6
All values given as molar sphingosines.
in human
percentage
of total fatty acids or of total
the characteristic ceramide pattern of brain gangliosides, with predominantly stearic acid and a large proportion of d20:l sphingosine. GM4 and GD3 had the same fatty acid patterns of normal fatty acids and sphingosines with a negligible proportion of d20 sphingosines and a significant proportion of C,, fatty acids. GM4 also contained 15% hydroxy fatty acids. The ceramide composition of cauda equina gangliosides differed from that of spinal cord gangliosides in several respects. The fatty acid composition of the gangliotetraose gangliosides was characterized by a relatively large proportion of 20 : 0 and 22 : 0 fatty acids and only 34% stearic acid. LMl contained only 15% stearic acid and 39% C,,. The gangliosides of the gangliotetraose series contained 24% d20 sphingosines, compared with only 4% in 3’-LMl.
4. Discussion There have been relatively few studies of the ganglioside composition of human spinal cord 131,321.This may be due to its being considered a relatively unimportant region of the nervous system, or to the difficulty in separating grey and white matter areas. With the increasing interest in the potential role of glycosphingolipid antigens in the pathogenesis of neuropathies, it has been of importance to elucidate the occurrence of these antigens in spinal cord. The first
et Biophysics Acta 1214 (1994) 115-123
121
comprehensive study of spinal cord gangliosides [31] showed that human spinal cord contained less than 0.30 pmol ganglioside sialic acid/g wet weight, which was about one-third of the concentration found in cerebral white matter at that time. In a more recent study [32], the concentration was reported to be 0.55 pmol ganglioside sialic acid/g. The ganglioside pattern also differed from that in cerebral white matter, with a lower proportion of GDla and higher proportions of GM3 and GD3. Our study has shown a definitively higher ganglioside concentration: 0.80 * 0.03 pmol/g fresh weight. The pattern found by us (Table 1) also differs from that found in previous studies [31,32], with ours having significantly higher proportions of the less solvent soluble GDlb, GTlb and GQlb gangliosides, and consequently lower proportions of GM4, GM3 and GMl. A possible explanation for the lower ganglioside yield and the lower proportion of b-series gangliosides in the previous studies may be that the solvent mixture used did not extract all gangliosides and the least soluble b-series gangliosides were lost on the Sephadex column used for their isolation [22]. The major difference between our study and the previous study is the ganglioside pattern of spinal cord myelin. In the other study [31] GM4 and GM1 constituted more than 60% of total ganglioside sialic acid, as compared with only 24% in ours. The ganglioside pattern found by us did not differ from that of intact spinal cord, and our yield of ganglioside sialic acid was almost twice that found in the previous study [31]. It is not easy to find a simple explanation of the varying results of the two studies. A critical factor in the isolation of myelin is the clearance between the pestle and the homogenizer vessel. If the pestle is too tight, the myelin is markedly fragmented, and the risk of isolating only a minor subpopulation of myelin is considerable. We have used a gentle homogenization procedure to obtain a high yield of myelin. It is also reasonable to assume that most of the cord gangliosides are distributed in the myelin. The motor neurons have sparse numbers of dendrites and the neuronal cell bodies have previously been shown to be poor in gangliosides 1331. An unexpected high proportion of gangliosides of the b-series might be attributable to an attachment of axonal gangliosides to the myelin during the isolation, since Schnaar and colleagues [34] have shown myelin to contain receptors for b-series gangliosides. Gangliosides of the gangliotetraose series of brain are characterized by a high proportion of stearic acid (18:O) and a large proportion of d20: 1 sphingosine which is larger in the oligosialogangliosides than in GM1 135,361. GM1 had a typical cerebral ceramide composition, while the oligosialogangliosides contained a larger proportion of long-chain fatty acids than fore-
brain [36]. The proportion of d20 : 1 sphingosine in cord gangliosides was of the same magnitude as in forebrain. The ceramide composition of CD3 supports the view that it is a major myelin ganglioside. The normal fatty acid and sphingosine patterns are identical for GM4 and GD3, which suggests that they are synthesized from a common ceramide pool. The ganglioside concentration in cauda equina, motor and sensory roots was found to be about half that in spinal cord and one-third to one-fourth that in cerebral white matter. The ganglioside patterns in cauda equina and femoral nerve did not differ from those reported recently [12]. In addition to the gangliosides, cauda equina and femoral nerve contained similar amounts of the two sulfoglucurony1 glycolipids LKl and HexLKl as of 3’-LMl and 3’-HexLMl. Yu and colleagues [37,38] have previously reported figures for the concentration of the two sulfoglucuronyl glycolipids in pg/mg proteins and their most recent results are of similar magnitude as ours. Inositoiphosphoryl galactosylceramide was isolated at this time under strictly quantitative conditions, and the concentration found was approximately 5-times higher than previously reported [39] which means that it is the most common acidic phospholipid, second to sulfatide in peripheral nerve. It is, of course, important to know whether the glycosphingolipid antigens differ in motor and sensory nerves when so many neuropathies differ in their motor and sensory signs and symptoms. The spinal roots allow for separation in motor and sensory nerves. When our study of motor and sensory nerve had been completed, a similar study was published by Ogawa-Got~ et al. [40]. They reported the same ganglioside concentration in motor and sensory nerve roots, which was only 50% of that found by us. The proportions of b-series gangliosides (GDlb, GTlb and GQlb) were lower in their study, which, we propose, is attributable to a loss of gangliosides at their isolatian as discussed above for spinal cord gangliosides. It is particularly pertinent to examine what proportion of the glycosphingolipid antigens are distributed in axons and myelin. In both motor and sensory nerve roots, the yield of myelin was approx. IO-times larger than that of axons. The ganglioside concentration was twice as large in myelin as in axons, which means that 95% of the gangliosides of peripheral nerve are localized to myelin or Schwann ceI1 membranes. The proportions of gangliosides and the sulfoglucuronyl glycolipids LKl and HexLKI were the same in motor and sensory nerve myelin and axons. Our results are in contrast with the results reported by Ogawa-Goto et al. [40] who found practically no GM1 ganglioside in sensory nerve. It is not possible to suggest any explanation of the different resuIts. We have used the same method for the isolation
of myelin,
and there was no significant
difference in the concentration of phospholipids, cholesterol and cerebroside between the two studies. The ceramide composition of the gangliotetraose gangliosides in peripheral nerve had characteristics of both CNS and extraneuronal organs 136,411. We have also confirmed the finding of Ogawa-Goto et al. [42] that sensory nerve roots have much larger proportions of long-chain fatty acids in the gangliotetraose gangliosides than motor nerve roots. The low stearic acid proportion of 3’-LMl commented on in our first report on this new ganglioside in 1972 [41] was confirmed in the present study. The present study can be summarized as having demonstrated a significantly higher concentration of gly~osphingolipid antigens than previously expected in PNS myelin, which might be a target for immunoglobulins. The glycolipid antigens of peripheral nerve differ from those of CNS (spinal cord) in an important respect. PNS myelin contains large proportions of lactotetraose and lactohexaose gangliosides and sulfoglucuronyl glycosphiugolipids, which occur onfy in negligible amounts in CNS myelin.
Acknowledgements
We wish to thank Ulrika Svensson for expert secretarial assistance. The cost of this work was partially defrayed by grants from the Swedish Medical Research Council (03X-627), the Elsa and Eivind Kson Sylvans Foundation and Fidia Research Laboratories, Abano Terme, Italy.
References A.A., Quarles, R.R., Dalakas, M.C. and Brady, R.Q. (198% Proc. Natl. Acad. Sci. USA 82, 6697-6700.
III &as,
121Latov, N., Hays, A., Yu, R.K., Ito, H. and Thomas, F.P. (1988) Neurochem. Pathol. 8, 1781-1787. [31 Steck, A.J., Murray, N., Delagi, K., Brouet, J.-C. and Seligman,
M. (1987) Ann. Neural. 22, 764-767. E., Francomana, E., Daverio, R., Barbieri, S., Marmiroli, P., Manfredini. E., Carpo, M., Moggio, M., Legname. Cr., Baldini, L. and Scarlato. G. (1989) Ann. Neuroi. 26, 543-550. Pestronk, A., Chandry, V., Feldman, E.L., Griffin, J.W., Cornblath, D.R., Denys, E.H., Glasberg, M., Kunel, R.W., Olney, R.K. and Yet?, W.C. (1990) Ann. Neurol. 27, 316-326. Sky, M.E., Evans, V.A., Lublin, F.D., Knobler, R.L., HeimanPattersson, T., Tahmousk, A.J., Parry. G.. Schick, P. and DeRyk, T.G. (1989) Ann. Neural. 25, 51 l--513. Vital, A., Vital, C.. Julien, J., Baquey, A. and Steck, A.J. (1989) Acta Neuropathol. 79, 160-167. Hauttecoeur, B., Schmitt, C., Dubois, C., Danou, F. and Brouet, J.C. (1990) Clin. Exp. Immunol. HO,381-185. Pestronk, A. (1991) Muscle Nerve 14, 927-936. Oomes, P.G., van der Meche, F.G.A., Jacobs, B.C., Hazenberg, M.P. and Banffer, R. (1991) Peripheral Nerve Study Group Meeting, Ardenhouse, New York, 1991 (Abstract).
[41 Nobile-Oraaio,
151
[61
171 [81 [91 II01
L. Suennerholtn et al. /Biochimica [ll]
Walsh, F.S., Cronin, M., Koblar, S., Doherty, P., Winer, Leon, A. and Hughes, R.A.C. (1991) J. Neuroimmunol.
J., 34,
43-51. [12] Svennerholm, L. and Fredman, P. (1990) Ann. Neural. 27, Suppl., S36-S40. [13] Fredman, P., Vedeler, C.A., Aarli, J.A. and Svennerholm, L. (1991) J. Neural. 238, 75-79. [14] Svennerholm, L., Bostrom, K., Fredman, P., Jungbjer, B., MLnsson, J.-E. and Rynmark, B.-M. (1992) Biochim. Biophys. Acta, 1128, l-7. [15] Norton, W.T. and Poduslo, S.E. (1973) J. Neurochem. 21, 749757. [16] Cammer, W. (1979) J. Neurochem. 32, 651-654. [17] Chou, K.H., Nolan, C.E. and Jungalwala, F.B. (1985) J. Neurochem. 44, 1898-1912. [18] Micko, S. and Schlaepfer, W.W. (1978) J. Neurochem. 30, 1041-1049. [19] Schwarzmann, G. (1978) Biochim. Biophys. Acta 529, 106-114. [20] Svennerholm, L., Bostrom, K., Fredman, P., MLnsson, J.-E., Rosengren, B. and Rynmark, B.-M. (1989) Biochim. Biophys. Acta 1005, 109-117. [21] Svennerholm, L. and Fredman, P. (1980) Biochim. Biophys. Acta 617, 97-109. [22] Fredman, P., Nilsson, O., Tayot, J.-L. and Svennerholm, L. (1980) Biochim. Biophys. Acta 618, 42-52. [23] M&son, J.-E., Rynmark, B.-M. and Svennerholm, L. (1991) FEBS Lett. 280, 251-253. [24] Helm, M., Minsson, J.-E., Vanier, M.-T. and Svennerholm, L. (1972) Biochim. Biophys. Acta 280, 356-364. [25] Svennerholm, L. (1957) Biochim. Biophys. Acta 24, 604-611. [26] Lauter, C.J. and Trams, E.G. (1962) J. Lipid Res. 3, 136-138. 1271 M?insson, J.-E., Vanier, M.-T. and Svennerholm, L. (1978) J. Neurochem. 30, 273-275.
et Biophysics Acta 1214 (1994) 115-123
123
[28] M%rsson, J.-E., MO, H., Egge, H. and Svennerholm, L. (1986) FEBS Lett. 196, 259-262. [29] MLnsson, J.-E., Fredman, P., Bigner, D.D., Molin, K., Rosengren, B., Friedman, H.A. and Svennerholm, L. (1986) FEBS Lett. 201, 109-l 13. [30] Svennerholm, L., Minsson, J.-E. and Li, Y.-T. (1973) J. Biol. Chem. 248, 740-742. [31] Ueno, K., Ando, S. and Yu, R.K. (1978) J. Lipid Res. 19, 863-871. [32] Yu, R.K., Ueno, K., Glaser, G.H. and Tourtelotte, W.W. (1982) J. Neurochem. 39, 464-477. [33] Hamberger, A. and Svennerholm, L. (1971) J. Neurochem. 18, 1821-1829. [34] Tiemayer, M., Swank-Hill, P. and Schnaar, R.L. (1990) J. Biol. Chem. 265, 11990-11999. [35] Schengrund, C.L. and Garrigan, O.N. (1969) Lipids 4, 488-495. [36] Minsson, J.-E., Vanier, M.T. and Svennerholm, L. (1978) J. Neurochem. 30, 273-275. [37] Kohriyama, T., Kusunoki, S., Ariga, T., Yoshino, J.E., de Vries, G.H., Latov, N. and Yu, R.K. (1987) J. Neurochem. 48, 15161522. 1381 Yu, R.K., Ariga, T., Kohriyama, T., Kusunoki, S., Maeda, Y. and Miyatani, N. (1990) Ann. Neurol. 27, Suppl., S30-S35. 1391 M%rsson, J.-E., Rynmark, B.-M. and Svennerholm, L. (1991) FEBS Lett. 280, 251-253. [40] Ogawa-Goto, K., Funamoto, N., Outa, Y., Abe, T. and Nagashima, K. (1992) J. Neurochem. 59, 1844-1859. [41] Svennerholm, L., Bruce, A., M&sson, J.-E., Rynmark, B.-M. and Vanier, M.T. (1972) Biochim. Biophys. Acta 280, 626-636. [42] Ogawa-Goto, K., Funamoto, N., Abe, T. and Nagashima, K. (1990) J. Neurochem. 55, 1486-1493.
et Biodwsica Acta Biochimica et Biophysics Acta 1214 (1994) 11.5-123
~angliosides and allied glycosphingolipids in human peripheral nerve and spinal cord Lars Svennerholm a,*, Kerstin Bostrijm b, Pam Fredman a, Birgitta Jungbjer ‘, Annika Lekman a, Jan-Eric M%nsson a, Britt-Marie Rynmark a aDepurtment ofChical neuroscience, Section of P~ch~i~
and ~eur~~e~~t~~ Universes ofGiiteborg, ~~lnda~ hospital, S-431 80 ~~l~d~~, Sweden h Department of Forensic Medicine, University of Gijteborg, G&eborg, Sweden Received 16 March 1994
Abstract Glycosphingolipids were determined in human spinal cord, cauda equina and femoral nerve of 10 subjects aged 20-70 years and in dorsal and ventral roots of four subjects aged 17-60 years. Myelin was isolated from corresponding tissue. Axons were isolated from the four specimens of dorsal and ventral roots. The concentration (mean and standard error of mean) of gangliosides in spinal cord was 0.80 f 0.03 pmol sialic acid/g fresh tissue, in cauda equina 0.40 f 0.02 @mot/g and in femoral
nerve 0.23 & 0.01 pmol/g. In spinal cord only trace amounts of glycosphingoiipids of the lacto series were found, and the ganglioside pattern differed from that in cerebral white matter by a relativeiy high proportion of GD3 and a low proportion of GDla. The ganglioside patterns were almost identical in cauda equina and femoral nerve - the major ganglioside being 3’-LMl, 0.07 and 0.04 kmol/g respectively. Another ganglioside of the lacto series, 3’-HexLMl, was 25% of 3’-LMl. Peripheral nerve also contained three acidic glycosphingolipids in addition to sulfatide - LKl and HexLKl belonging to the glycosphingolipid iacto series and containing glucuronyl-3-sulfate instead of sialic acid, and inositolphospho~l galactosylceramide. The dorsal (sensory and ventral (motor) roots had the same major membrane lipid composition but the ganglioside concentration was 30% higher in sensory than motor nerve and myelin. The patterns of gangliotetraose gangliosides were, however, the same in motor and sensory myelin and axons. The ceramide composition of the gangliosides is also reported. Key words: Peripheral nerve; Spinal cord; My∈ Axon; Ganglioside;
Abbreviations: The gangliosides and allied glycolipids have been designated according to the principles suggested by Svennerholm (Eur. J. Biocbem. (1977) 79, 11-21). GM3, I13NeuAc-LacCer; GD3, I13(NeuAc),-LacCer; GM2, I13NeuAc-GgOsesCer; GD2, I13(NeuAcJz-GgOsesCer; GMl, I13NeuAc-CgOse,Cer; GDla, IV3NeuAc,I13NeuAc-GgOse,Cer; GDlb, I13(NeuAc),-GgOse,Cer; GTlb, IV3NeuAc,I13(NeuAc),-GgOse,Cer; GQlb, IV3(NeuAc), I13(NeuAc),-Gg Ose,Cer; 3’-LMZ, IV3NeuAc-nLc~se~~r; 3’HexLMl, V13NeuA~-n~s~~Cer; LDl, (NeuAc),-nLcOse,Cer; 3’,8’-LDl, IV3(NeuAc),-nLcOse,Cer; LK1, IV3glucuronyI-3-sulfatenLcOse,Cer; HexLKl, V13glucuronyI-3-sulfate-nkcOse&er; gal, galactose; NeuAc, N-acetylneuraminic acid; BSA, bovine serum albumin; CNS, central nervous system; CSF, cerebrospinal fluid; GBS, Guillain-Barr6 syndrome; IPGC, inositolphosphoryl galactosylceramide; PNS, peripheral nervous system. * Corresponding author. Fax: + 46 31 862422. 00052760/94/$07.00 c
0 1994 Elsevier Science B.V. AI1 rights reserved
Acidic glycosphingolipid
1. Im~r~u~tion Results from several studies have demonstrated that some neuropathy syndromes are associated with increased titers of serum antibodies against carbohydrate epitopes of glycolipids and glycoproteins [l-6]. Among patients with IgM par~protei~emia who develop a slowly progressive demyelinating sensorimotor neuropathy, approx. 50% have serum antibodies against myelin-associated glycoprotein (MAGI and glycolipids with terminal glucuronyl-3-sulfate linked to galactose [4,7,8]. Patients with m~itifo~al motor neuropathy and some patients with lower motor syndromes (for review
116
L. Srennerholm
et al. / Biochimica
see Ref. [9]) have antibodies directed against glycolipids with a terminal galactosyl ~1-4 N-acetylgalactosaminyl moiety which occurs in the gangliosides GM1 and GDlb and the neutral glycolipid GAl. Several studies have demonstrated antibodies to peripheral nerve tissue and myelin in Guillain-Barre syndrome and in chronic inflammatory demyelinating polyradiculoneuropathy. The target antigens for these antibodies have not been identified, but two recent studies have reported a high frequency of anti-GM1 antibody [lO,ll]. We have not been able to confirm these findings in Scandinavian and German patients, but have found increased titers of 3’-LMl and sulfatide, the two major acidic glycolipids of peripheral nerve myelin [12,13]. There is thus evidence that some patients with peripheral neuropathies have humoral antibodies directed against carbohydrate epitopes. One major task is now to search for a causal relationship between the content and localization of the antigen recognized by the antibody and the patient’s clinical picture. In a previous study, the major membrane lipid composition of peripheral nerve (cauda equina) was examined and compared with that of spinal cord [14]. The study showed a lower proportion of cerebroside, a slightly lower proportion of cholesterol but a significantly higher proportion of sphingomyelin in cauda equina than in spinal cord. In the present study the gangliosides and allied acidic glycolipids of peripheral nerve have been isolated and characterized. The same acidic glycolipids have also been assayed in dorsal and ventral spinal roots and in their myelin and axons, to study whether a different composition of the glycolipids in sensory and motor nerves could explain the variations in motor and sensory expression of various forms of polyneuropathy.
2. Material and methods 2.1. Tissue specimens
The lower thoracal and lumbosacral portions of spinal cord and cauda equina were obtained from the Department of Forensic Medicine, from 17 subjects aged 17-91 years who had died from accidents or acute circulation failure with no signs of neurological disease. The dissection of the spinal roots (cauda equina) and spinal cord and the isolation of myelin with three different methods [15-171 were described in the preceding paper [14]. Myelin of spinal cord and cauda equina was isolated from 13 subjects. The dorsal (motor) and ventral (sensory) spinal roots of four subjects 17-60 years of age were dissected separately. Myelin and axons were isolated essentially as described by Micko and Schlaepfer [18].
et Biophysics Acta 1214 (1994) 115-123
The study was approved by the Ethics Committee for Medical Research of the University of Goteborg. 2.2. Characterization of myelin Protein profiles of myelin and axonal fractions were analyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 20% homogeneous gels (43 X 50 + 0.45 mm) in a Phast-System from Pharmacia according to the program given in the manual (Phast-SystemTM Separation Technique File No. 220). 2 pg of protein was applied to the gel. The gels were stained with Coomassie Brilliant Blue (Phast-SystemTM Developmen t File 200). The protein patterns of motor and sensory myelin did not differ, the major protein being P,. The axonal fraction had a significantly different protein pattern, and P, was only weakly discernible. 2.3. Chemicals Silica gel (230-400 mesh), TLC silica gel 60 and high performance thin-layer plates (HPTLC) were purchased from Merck (Darmstadt, Germany). TLCELISA was performed on precoated plastic sheets from PolygramR Sil G (Marchery and Nagel, Duren, Germany) or glass plates, SI-HPF Silica Gel TLC Plate (J.T. Baker, Philipsburg, NJ, USA). Sephadex G-25 fine and DEAE-Sepharose were from Pharmacia (Uppsala, Sweden). The anion exchange resin, SpherosilDEAE-dextran and cholera toxin B-subunit (CT-B) were gifts from Institut Merieux (Lyon, France). Ganglioside and neutral glycolipid isolated from human organs and characterized in our laboratory using component analysis, permethylation and assay of methylated sugars and FAB-MS were used as standards. Gangliosides of the ganglio series were also obtained from Fidia Research Laboratories (Abano Terme, Italy). Gangliosides GM1 and GTlb used as internal standards were labeled by tritiation of the sphingosine double bond [ 191. The specificity of the anti-ganglioside murine monoclonal antibodies produced in our laboratory and used in this study has recently been reported [20]. In addition to these antibodies, a new anti-3’-LMl ganglioside IgM murine monoclonal antibody @L-l), specific to the terminal epitope NeuAqx2-3GalPl-4GlcNAc, was used for the quantitative determination of 3’-LMl and 3’-HexLMl. HNK-1 murine monoclonal antibody was produced from a HNK-1 hybridoma obtained from American Type Culture Collection (Rockville, MD, USA). It was used for the quantitative assay of LKl and HexLKl. Alkaline phosphatase-conjugated goat antimouse IgM and IgG (heavy and light chains) were purchased from Jackson Immunoresearch Laboratory (West Grove, PA, USA).
L. Suennerholm et al. / Biochimica et Biophysics Acta 1214 (I 994) 115-123
2.4. Lipid extraction and isolation of total ganglioside fraction
Fresh tissue (0.5-10 g) was homogenized with 3 vol. of water in a Potter-Elvehjem homogenizer and myelin (10 mg) with 0.7 ml of water. Chloroform/methanol (1: 2, v/v) was then added to the suspension to give a final ratio of chloroform/ methanol/ water (4 : 8 : 3, v/v) [21]. The lipids were extracted for 1 h at room temperature, the extraction being facilitated by repeated mixing on a cyclomixer, and a clear supernatant was obtained by low speed centrifugation. The pellet was reextracted with the same solvent mixture and volume and under the same conditions, and the two supernatants were then combined and evaporated. The crude lipid extract of 5-10 g tissue was dissolved into 20 ml of chloroform/ methanol/ water (6.5 : 25 : 4, v/v) and put onto a column with 10 g of silica gel slurried in chloroform. The bulk of lipids was eluted with 80 ml of chloroform/ methanol/ water (65 : 25 : 4, v/v) and gangliosides and allied glycolipids with 150 ml of chloroform/ methanol/ water (3 : 6 : 2, v/v). When the tissue amount was lower than 5 g the size of the silica gel column and the volumes of eluting solvents were correspondingly reduced. The ganglioside fraction was evaporated, dissolved in water and dialyzed against running tap water for 3 days. The fractions of whole cauda equina and myelin of cauda equina were divided into two equal portions, one of which was treated with 0.1 M NaOH in methanol/water (1: 1, v/v) at room temperature overnight. After neutralization with acetic acid, the fraction was dialyzed and treated as the nonsaponified fraction. The two ganglioside fractions were dissolved into chloroform/ methanol/ water (60 : 30 : 4.5, v/v) and chromatographed on SpherosilDEAE-dextran [22]. Monosialogangliosides were eluted with 10 vol. of 0.02 M potassium acetate in methanol and the oligosialogangliosides with 10 vol. of 0.5 M potassium acetate in methanol. The resin was finally eluted with 10 vol. of 1.0 M potassium acetate in methanol to recover LKl and HexLKl. 3’-LMl and 3’-HexLMl were isolated from the monosialoganglioside fraction by a combination of silica gel column chromatography by elution with chloroform/ methanol/ water (60 : 30 : 5, v/v) and by preparative TLC with chloroform/ methanol/ 2.5 M ammonia (50 : 40 : 10, v/v). The corresponding neutral glycosphingolipids were obtained by sialidase hydrolysis, and separated from GM1 and GM2 on Spherosil-DEAEdextran. LKl and HexLKl were isolated from the oligosialoganglioside fraction by sialidase hydrolysis, which converted all gangliosides of the ganglio series to GM2 and GMl. After dialysis against distilled water GM2 and GM1 were eluted from a small SpherosilDEAE-dextran column with 0.02 M potassium acetate in methanol, and LKl and HexLKl were eluted with
117
1.0 M potassium acetate in methanol. Inositolphosphoryl-2(3)galactosylceramide was determined on the monosialoganglioside fraction by thin-layer chromatography. 2.5. Analytical methods Quantitative composition of the carbohydrate moieties was determined as alditol acetates by GLC with mannose as the internal standard [24]. Ganglioside sialic acid was determined on the total ganglioside fraction and isolated individual gangliosides using the resorcinol method [25]. Sphingosine was assayed by a modification of the methyl orange method of Lauter and Trams [26], and used for the quantification of glycolipids other than gangliosides. The fatty acid and sphingosine patterns were determined by GLC [27]. The glycolipids were further characterized by permethylation, and by analyses of partially methylated acetates [28]. FAB-MS was performed on a model VG 7070 E mass spectrometer equipped with a fast-atom gun 1291. 2.6. Silica gel thin-layer chromatography TLC was performed on precoated TLC and HPTLC plates, which were developed with chloroform/ methanol/0.25% aqueous KC1 (50: 40: 10, v/v) or chloroform/ methanol/ 2.5 M ammonia (50 : 40 : 10, v/v). Neutral glycolipids (after sialidase hydrolysis) were also developed with chloroform/ methanol/ water (65:25 :4, v/v). Th e plates were stained with resorcinol-HCl reagent [25] for gangliosides and densitometric scanning of the plates was performed at 620 nm on a CAMAG TLC scanner. The plates were also stained with orcinol-H,SO, and scanned at 515 nm for neutral glycosphingolipids and gangliosides [30]. 2.7. Sialidase treatment Isolated gangliosides and mono- and oligosialoganglioside fractions, corresponding to 5-10 nmol of sialic acid, were treated with sialidase from V. cholerae [20]. The hydrolyzed samples were chromatographed on HPTLC plates with the solvent systems described above. One plate was stained with resorcinol-HCl reagent and the other with orcinol-H,SO,. 2.8. TLC immunostaining Ganglioside and allied glycosphingolipids were also characterized and quantified with the murine monoclonal antibodies, using the procedure recently described [201. Gangliosides of the ganglio series were also assayed with cholera toxin-B subunit (CT-B) after sialidase hydrolysis [20].
L. Swnnerholm et al. /Biochimica et Biophysics Acta 1214 (1994) 115-123
118 Table 1 Composition
of gangliosides
Glycospingolipid
and allied glycosphingolipids
Spinal cord (n = 10) glycosphingolipid (nmol/g wet wt.)
GM4 GM3 GM2 GM1 3’-LMl 3’.HexLM 1 GD3 GD2 LDl GDla GDlb GTlb GQlb LKl HexLKl IPGC*
in adult human Cauda
o/c sialic acid
S.D.
64
18
(8)
n.d.
25 16 110 n.d. n.d. 60 8 n.d. 28 105 40 14 1.8 trace trace
2 4 20
(3) (2) (14)
6 1
(15) (2)
3 9 3 2
(7) (26) (15) (7)
24 4 36 68 16 18 4 8 30 33 17 4 64 15 154
and relative
equina
mean
S.D.
proportion
of ganglioside
Two monosialo gangliosides were detected with monoclonal antibody SL-1 with the same TLC migration as reference 3’-LMl and 3’-HexLMl. Sialidase hydrolysis of the fastest migrating ganglioside gave a
and allied glycosphingolipids
in adult human
motor
range
GM3 GM2 GM1 3’-LM 1 3’-HexLM 1 GD3 GD2 LDl GDla GDlb GTlb GQlb IPGC LKI HexLKl
20 4 34 66 14 18 5 9 24 28 16 2 198 54 16
16- 23 34 25- 39 61- 69 13- 14 IS- 20 45 8- 10 18- 32 26- 30 14- 17 l2 184-208 50- 59 14- 17
Total ganglioside
356
337-368
glycosphingolipid (nmol/g wet wt.)
(6) (1) (9) (17) (4) (9) (2) (4) (15) (16) (13) (3)
22 5 19 38 IO 7 3 4 19 17 12 3 39 10 70
mean
expressed
in nmol glycosphingolipid
and relative
% sialic acid
(6) (1) (10) (19) (4) (10) (3) (5) (13) (16) (13) (2)
proportion
c/r sialic acid
S.D.
5 1 3 4 2 2 1 1 3 2 1
(9) (2) (8) (16) (4) (6) (2) (3) (16) (14) (16) (4)
1 12 3 27
sialic acid (in parentheses).
and sensory
nerve nerve (n = 4)
glycosphingolipid (nmol/g wet weight) mean
range
25 5 34 85 20 23 8 15 31 38 19 2 279 72 20
2l- 31 45 2% 38 70- 94 16- 26 l6- 32 6Y 1218 2% 36 30- 44 IS- 22 23 236-310 70- 76 20- 21
460
412-545
sialic acid (nmol/g) Values
nerve (n = 10)
% sialic acid
Sensory
Motor nerve (n = 4) glycosphingolipid (nmol/g wet weight) mean
Femoral
nerve
product with the same TLC migration as reference LAl. FAB-MS analysis of the permethylated ganglioside showed the most prominent molecular ion at m/z 1910. This ion corresponds to the composition NeuAc* HexNAc* Hex: Cer with the d18 : l/ 24 : 0 ceramide. Ions of equal intensity at m/z 825 and 793 (825 - 32) were derived from the terminal sequence NeuAc-Hex-HexNAc with a Hexl-4HexNAc linkage. Ions at m/z 376 and 344 (376 - 32) correspond to
3.1. Characterization of gangliosides and allied acidic glycolipids of peripheral nerve
of gangliosides
and femoral
n.d. 3 1 7 9 2 5 1 2 4 5 3 1 10 4 33
3. Results
Tsble 2 Composition
equina
(n = 10)
glycosphingolipid (nmol/g wet wt.)
mean
Values expressed in nmol glycosphingolipid * Inositolphosphoryl galactosylceramide.
spinal cord, cauda
of ganglioside
sialic acid (in parentheses).
o/c sialic acid
(5) (1) (7) (18) (4) (10) (3) (7) (13) (17) (12) (2)
119
L. Suennerholmet al. /Biochimica et BiophysicsActa 1214 (1994) 115-123
3.2. Concentrative of gangl~o~ides and Alfred acidic gly-
terminal NeuAc and at m/z 1029 to NeuAc-HexHexNAc-Hex. The most prominent ion of the ceramide portion was at m/z 660 representing d18: l/24: 0, Analysis of the partially methylat~d alditol acetates from the ganglioside showed the occurrence of 2,4,6Me,-Ga1,2,3,6-Me,-Glc and 3,6-Me,-GlcNAcMe. No 4,6-Me*-Gl~NA~Me corresponding to the Gal l3GlcNAc linkage was detected. The results show that the structure of the fast-migrating SL-1 positive ganglioside is 3’-LMl (XV3NeuAc-nLcOse,Cer). Sialidase hydrolysis of the slowly migrating SL-1 positive ganglioside gave a product with the same TLC migration as reference nI_cOse,Cer. FAB-MS of the permethylated ganglioside showed that the most prominent molecular ion at m/z 2359 corresponded to the composition NeuAc* HexNAcz Hex: Cer with the d18: l/24: 0 ceramide. Fragment ions from the carbohydrate portion of ganglioside were found at m/z 376 and 344 (376 - 32), 580, 825 and 793 (825 - 321, 1029, 1274 and 1242 (1274 - 32) and 1478 representing NeuAc, NeuAc-Hex, NeuAc-Hex-HexNAc, NeuAcHex-HexNAc-Hex, NeuAc-Hex-HexNAc-Hex-HexNAc and NeuAc-Hex-HexNAc-Hex-HexNAc-Hex. Methylation analysis of ganglioside gave 2,4,6-Me,Gal, 2,3,6-Me,-Glc and 3,6-Me,-GlcNAcMe. Again, no 4,6-Me,-GlcNAcMe was detected, which excluded the occurrence of a Gal 1-3GlcNAc linkage. The structural analyses show that the slowly migrating SL-1 positive ganglioside is 3’-hexLM1 (IV”NeuAcnI_.cOse,Cer).
cosphingolipids in spinal cord and peripheral nerve
The concentrations of gangliosides and allied acidic glycosphingolipids are given in Table 1. The ganglioside concentration was highest in spinal cord, mean value F S.E.M. 0.80 rt 0.03 pmol/g fresh weight, significantly lower in cauda equina 0.40 i: 0.02 pmol/g and lowest in femoral nerve 0.23 k 0.01 pmol/g. The ganglioside patterns were virtually identical in cauda equina and femoral nerve, the only small difference was a larger proportion of GD3 and a smaller of GM3 in cauda equina than in femoral nerve. The major ganglioside in peripheral nerve was 3’-LMl (Tables 2 and 3). The pattern differed significantly from that in spinal cord, which mostly contained GM4 and none of the two gangliosides of the lactoseries, 3’-LMl and 3’-HexLMl. The spinal cord ganglioside pattern was further characterized by a large proportion of GD3 and a relatively small proportion of GDla, which gave a dominance of b-series gangliosides. Peripheral nerve also contained three acidic glycosphingolipids which only occurred in trace amounts in spinal cord. Two of them, LKl and I-IexLKl, belong to the giycosphingolipid lacto series and contain glucuronyl
3-sulfate
rather
than sialic acid. Their
concen-
trations were of the same magnitude as those of 3’-LMl and 3’-HexLMl in both cauda equina and femoral nerve (Table 1). Another characteristic component in peripheral
nerve
was ~~ositolphospho~l
galactosylce-
Table 3 Major gangliosides and allied acidic giycosphin~oIipids of spinal cord and peripheral nerve myelin Glycosphingolipid
Spinal cord (n = 4) glycosphingolipid (nmoi/g) mean
Cauda equina (n = 4) % sialic acid
SD.
~lycosphingaiipid (nmol/g)
% sialic acid
mean
S.D.
178 101 19
48 48 3
(8) (4) (1)
n.d. 45 -
7 _
(2)
GM1 3’-LM 1 3’-HexLMl GD3 CD2 LDI GDla GDlb GTlb GQlb LKl
351 n.d. n.d. 163 4 n.d. 65 325 128 42
52
(1.5)
_ 39 1
115) (2)
14 57 18 10
(6) (28) (16) (7)
138 420 60 71 12 23 143 120 71 14 112
17 93 21 41 2 7 28 20 10 2 17
(9) (17) (4) (6) (1) (2) (21) (17) (16) (4)
Total ganglioside sialic acid (nmol/g)
2 360
GM4 GM3 GM2
1750
Values expressed in nmol glycosph~ngolipid per g dry weight of myelin and relative proportion of ganglioside sialic acid (in parentheses).
L. St~ennerholm et al. /Biochimica et Biophysics Acta 1214 (1994) 115-123
120 Table 4 Lipid composition nerves (n = 4)
of myelin
and
axons
from
motor
Motor nerve
Sensory
and
sensory
nerve
myelin
axons
myelin
axons
Proteins a Phospholipids h Cholesterol ” Cerebroside h Sulfatide h Ganglioside sialic acid h Protein-bound sialic acid ’
20 558 502 111 53 1.7 2.8
80 105 80 21 12 0.9 3.8
21 560 483 107 49 2.1 3.3
82 113 86 21 12 1.0 4.4
Ganglioside GDla GDlb GTlb GQlb 3’-LMl 3’Hex-LMI LKl HexLKl
194 117 160 62 9 385 51 193 63
163 113 97 43 I 60 9 70 25
224 155 205 99 15 413 59 199 72
162 121 109 48 8 93 6 80 28
pattern
‘GM1
” Proteins expressed as weight percentage. ” Concentrations in pmol/g dry myelin or g dry axon. ’ Individual glycosphingolipids in nmol/g dry myelin or g dry axon.
ramide (IPGC) which we recently discovered and characterized [39]. The concentration of this glycosphingolipid was more abundant than we had assumed from the preparative isolation of the compound. Trace amounts of these three acidic lipids, LKl, HexLKl and IPGC, were found in spinal cord, which does not prove that they exist in this tissue - it is not possible to exclude the possibility that the low values found for these glycosphingolipids are attributable to an admixture with spinal roots from dissection of the spinal cord. The spinal roots from four cases were dissected in motor and sensory nerves. The concentration of major lipids did not differ significantly between motor and sensory nerves (Table 4). The concentration of gangliosides was approx. 30% higher in sensory than in motor nerves, 0.46 pmol and 0.36 pmol ganglioside sialic acid/g fresh weight, respectively, but although the ganglioside patterns were similar the proportion of GM1 was slightly lower in sensory than in motor nerve. The other important issue was to determine what proportion of the gangliosides was localized to the myelin and what proportion was present in the axons. Our aim was thus not to isolate a “pure” myelin fraction but to obtain a large yield of all forms of myelin, and our criterion on the purity of our myelin was that it did not contain detectable quantities of axonal proteins. Three different procedures - previously shown to give similar results for the major lipids [ 141 - were used for the isolation of myelin from spinal cord and cauda equina [15-171. All three methods gave similar results, but the results in Table 3 are from
spinal cord and cauda equina isolated with Cammer’s method [16] analyzed on the same occasion with the same batches of monoclonal antibodies. In spinal cord, the percentage composition of myelin gangliosides was almost identical to that in intact spinal cord. There was slightly more GM4 and slightly less GD3 and GDla. The latter two gangliosides were also enriched in the fine particles, which did not sediment when the myelin was centrifuged after the osmotic shock, with a relatively low g number (12,000 X g>. Cauda equina myelin did not differ significantly in its composition from intact cauda equina. In both spinal cord and cauda equina myelin, the proportion of GDlb myelin was slightly higher than in the original tissue. The concentration of gangliosides was 30% higher in sensory than in motor myelin (Table 4), but the proportions of the major gangliosides were similar. The finding of a slightly higher proportion of GM1 ganglioside in motor nerve also occurred in the myelin, but the content of GM1 per g myelin was slightly higher in sensory than in motor nerve. The ganglioside concentration was almost the same in sensory and motor axons, and the proportions of major gangliosides were almost identical. Another interesting observation was that the concentration of sialoglycoprotein sialic acid was 4-times higher than that of ganglioside sialic acid in both motor and sensory axons. The ceramide composition of gangliosides in spinal cord (Table 5) and cauda equina (Table 6) was determined on pooled samples from 10 cases. The gangliosides of the gangliotetraose series in spinal cord had
Table 5 Ceramide Fatty acid
composition
of major gangliosides
spinal cord
Ganglioside GM4 normal
16:0 18:0 2O:O 22:o 23:0 24:0 24:l 25:0 25:l 26:O 26:l
in human
CD3
GM1
GDla + GDlb + GTlb
15 56 2 3 4 3 13 1 1 iI 1
4 91 1
10 71 9 3 2 2 3
4 93 trace 3
4 62 2 32
2 36 5 57
hydroxy
16 58
I
6
2 4 13 2 1
15 44 21 7 3 2 2
o/o hydroxy fatty acids 15 Sphingosine d18:O d18: 1 d20:O d20: 1
3 94 trace 3
All values given as molar sphingosines.
percentage
of total fatty acids or of total
L. Svennerholm et al. /Biochimica Table 6 Ceramide Fatty acid
composition Ganglioside normal
of major gangliosides GM3 hydroxy
16:0 18:O 20:o 22:o 23:0 24:0 24:l 25:0 25:l 2610 26:l
13 10 6 15 9 31 11 3
% hydroxy
fatty acids 10
Sphingosine d18:O d18:l d20:O d20: 1
18 19 50 7 4 3
Gangliotetraose series GDla +GDlb + GTlb
cauda
Lactotetraose 3’-LMl
equina series
LKl
5 34 16 12 4 12 I
13 15 15 14 7 22 17 3
30 28 5 8 4 12 9
19 57 2 22
24 72 1 3
20 77 1 2
21 70 3 6
All values given as molar sphingosines.
in human
percentage
of total fatty acids or of total
the characteristic ceramide pattern of brain gangliosides, with predominantly stearic acid and a large proportion of d20:l sphingosine. GM4 and GD3 had the same fatty acid patterns of normal fatty acids and sphingosines with a negligible proportion of d20 sphingosines and a significant proportion of C,, fatty acids. GM4 also contained 15% hydroxy fatty acids. The ceramide composition of cauda equina gangliosides differed from that of spinal cord gangliosides in several respects. The fatty acid composition of the gangliotetraose gangliosides was characterized by a relatively large proportion of 20 : 0 and 22 : 0 fatty acids and only 34% stearic acid. LMl contained only 15% stearic acid and 39% C,,. The gangliosides of the gangliotetraose series contained 24% d20 sphingosines, compared with only 4% in 3’-LMl.
4. Discussion There have been relatively few studies of the ganglioside composition of human spinal cord 131,321.This may be due to its being considered a relatively unimportant region of the nervous system, or to the difficulty in separating grey and white matter areas. With the increasing interest in the potential role of glycosphingolipid antigens in the pathogenesis of neuropathies, it has been of importance to elucidate the occurrence of these antigens in spinal cord. The first
et Biophysics Acta 1214 (1994) 115-123
121
comprehensive study of spinal cord gangliosides [31] showed that human spinal cord contained less than 0.30 pmol ganglioside sialic acid/g wet weight, which was about one-third of the concentration found in cerebral white matter at that time. In a more recent study [32], the concentration was reported to be 0.55 pmol ganglioside sialic acid/g. The ganglioside pattern also differed from that in cerebral white matter, with a lower proportion of GDla and higher proportions of GM3 and GD3. Our study has shown a definitively higher ganglioside concentration: 0.80 * 0.03 pmol/g fresh weight. The pattern found by us (Table 1) also differs from that found in previous studies [31,32], with ours having significantly higher proportions of the less solvent soluble GDlb, GTlb and GQlb gangliosides, and consequently lower proportions of GM4, GM3 and GMl. A possible explanation for the lower ganglioside yield and the lower proportion of b-series gangliosides in the previous studies may be that the solvent mixture used did not extract all gangliosides and the least soluble b-series gangliosides were lost on the Sephadex column used for their isolation [22]. The major difference between our study and the previous study is the ganglioside pattern of spinal cord myelin. In the other study [31] GM4 and GM1 constituted more than 60% of total ganglioside sialic acid, as compared with only 24% in ours. The ganglioside pattern found by us did not differ from that of intact spinal cord, and our yield of ganglioside sialic acid was almost twice that found in the previous study [31]. It is not easy to find a simple explanation of the varying results of the two studies. A critical factor in the isolation of myelin is the clearance between the pestle and the homogenizer vessel. If the pestle is too tight, the myelin is markedly fragmented, and the risk of isolating only a minor subpopulation of myelin is considerable. We have used a gentle homogenization procedure to obtain a high yield of myelin. It is also reasonable to assume that most of the cord gangliosides are distributed in the myelin. The motor neurons have sparse numbers of dendrites and the neuronal cell bodies have previously been shown to be poor in gangliosides 1331. An unexpected high proportion of gangliosides of the b-series might be attributable to an attachment of axonal gangliosides to the myelin during the isolation, since Schnaar and colleagues [34] have shown myelin to contain receptors for b-series gangliosides. Gangliosides of the gangliotetraose series of brain are characterized by a high proportion of stearic acid (18:O) and a large proportion of d20: 1 sphingosine which is larger in the oligosialogangliosides than in GM1 135,361. GM1 had a typical cerebral ceramide composition, while the oligosialogangliosides contained a larger proportion of long-chain fatty acids than fore-
brain [36]. The proportion of d20 : 1 sphingosine in cord gangliosides was of the same magnitude as in forebrain. The ceramide composition of CD3 supports the view that it is a major myelin ganglioside. The normal fatty acid and sphingosine patterns are identical for GM4 and GD3, which suggests that they are synthesized from a common ceramide pool. The ganglioside concentration in cauda equina, motor and sensory roots was found to be about half that in spinal cord and one-third to one-fourth that in cerebral white matter. The ganglioside patterns in cauda equina and femoral nerve did not differ from those reported recently [12]. In addition to the gangliosides, cauda equina and femoral nerve contained similar amounts of the two sulfoglucurony1 glycolipids LKl and HexLKl as of 3’-LMl and 3’-HexLMl. Yu and colleagues [37,38] have previously reported figures for the concentration of the two sulfoglucuronyl glycolipids in pg/mg proteins and their most recent results are of similar magnitude as ours. Inositoiphosphoryl galactosylceramide was isolated at this time under strictly quantitative conditions, and the concentration found was approximately 5-times higher than previously reported [39] which means that it is the most common acidic phospholipid, second to sulfatide in peripheral nerve. It is, of course, important to know whether the glycosphingolipid antigens differ in motor and sensory nerves when so many neuropathies differ in their motor and sensory signs and symptoms. The spinal roots allow for separation in motor and sensory nerves. When our study of motor and sensory nerve had been completed, a similar study was published by Ogawa-Got~ et al. [40]. They reported the same ganglioside concentration in motor and sensory nerve roots, which was only 50% of that found by us. The proportions of b-series gangliosides (GDlb, GTlb and GQlb) were lower in their study, which, we propose, is attributable to a loss of gangliosides at their isolatian as discussed above for spinal cord gangliosides. It is particularly pertinent to examine what proportion of the glycosphingolipid antigens are distributed in axons and myelin. In both motor and sensory nerve roots, the yield of myelin was approx. IO-times larger than that of axons. The ganglioside concentration was twice as large in myelin as in axons, which means that 95% of the gangliosides of peripheral nerve are localized to myelin or Schwann ceI1 membranes. The proportions of gangliosides and the sulfoglucuronyl glycolipids LKl and HexLKI were the same in motor and sensory nerve myelin and axons. Our results are in contrast with the results reported by Ogawa-Goto et al. [40] who found practically no GM1 ganglioside in sensory nerve. It is not possible to suggest any explanation of the different resuIts. We have used the same method for the isolation
of myelin,
and there was no significant
difference in the concentration of phospholipids, cholesterol and cerebroside between the two studies. The ceramide composition of the gangliotetraose gangliosides in peripheral nerve had characteristics of both CNS and extraneuronal organs 136,411. We have also confirmed the finding of Ogawa-Goto et al. [42] that sensory nerve roots have much larger proportions of long-chain fatty acids in the gangliotetraose gangliosides than motor nerve roots. The low stearic acid proportion of 3’-LMl commented on in our first report on this new ganglioside in 1972 [41] was confirmed in the present study. The present study can be summarized as having demonstrated a significantly higher concentration of gly~osphingolipid antigens than previously expected in PNS myelin, which might be a target for immunoglobulins. The glycolipid antigens of peripheral nerve differ from those of CNS (spinal cord) in an important respect. PNS myelin contains large proportions of lactotetraose and lactohexaose gangliosides and sulfoglucuronyl glycosphiugolipids, which occur onfy in negligible amounts in CNS myelin.
Acknowledgements
We wish to thank Ulrika Svensson for expert secretarial assistance. The cost of this work was partially defrayed by grants from the Swedish Medical Research Council (03X-627), the Elsa and Eivind Kson Sylvans Foundation and Fidia Research Laboratories, Abano Terme, Italy.
References A.A., Quarles, R.R., Dalakas, M.C. and Brady, R.Q. (198% Proc. Natl. Acad. Sci. USA 82, 6697-6700.
III &as,
121Latov, N., Hays, A., Yu, R.K., Ito, H. and Thomas, F.P. (1988) Neurochem. Pathol. 8, 1781-1787. [31 Steck, A.J., Murray, N., Delagi, K., Brouet, J.-C. and Seligman,
M. (1987) Ann. Neural. 22, 764-767. E., Francomana, E., Daverio, R., Barbieri, S., Marmiroli, P., Manfredini. E., Carpo, M., Moggio, M., Legname. Cr., Baldini, L. and Scarlato. G. (1989) Ann. Neuroi. 26, 543-550. Pestronk, A., Chandry, V., Feldman, E.L., Griffin, J.W., Cornblath, D.R., Denys, E.H., Glasberg, M., Kunel, R.W., Olney, R.K. and Yet?, W.C. (1990) Ann. Neurol. 27, 316-326. Sky, M.E., Evans, V.A., Lublin, F.D., Knobler, R.L., HeimanPattersson, T., Tahmousk, A.J., Parry. G.. Schick, P. and DeRyk, T.G. (1989) Ann. Neural. 25, 51 l--513. Vital, A., Vital, C.. Julien, J., Baquey, A. and Steck, A.J. (1989) Acta Neuropathol. 79, 160-167. Hauttecoeur, B., Schmitt, C., Dubois, C., Danou, F. and Brouet, J.C. (1990) Clin. Exp. Immunol. HO,381-185. Pestronk, A. (1991) Muscle Nerve 14, 927-936. Oomes, P.G., van der Meche, F.G.A., Jacobs, B.C., Hazenberg, M.P. and Banffer, R. (1991) Peripheral Nerve Study Group Meeting, Ardenhouse, New York, 1991 (Abstract).
[41 Nobile-Oraaio,
151
[61
171 [81 [91 II01
L. Suennerholtn et al. /Biochimica [ll]
Walsh, F.S., Cronin, M., Koblar, S., Doherty, P., Winer, Leon, A. and Hughes, R.A.C. (1991) J. Neuroimmunol.
J., 34,
43-51. [12] Svennerholm, L. and Fredman, P. (1990) Ann. Neural. 27, Suppl., S36-S40. [13] Fredman, P., Vedeler, C.A., Aarli, J.A. and Svennerholm, L. (1991) J. Neural. 238, 75-79. [14] Svennerholm, L., Bostrom, K., Fredman, P., Jungbjer, B., MLnsson, J.-E. and Rynmark, B.-M. (1992) Biochim. Biophys. Acta, 1128, l-7. [15] Norton, W.T. and Poduslo, S.E. (1973) J. Neurochem. 21, 749757. [16] Cammer, W. (1979) J. Neurochem. 32, 651-654. [17] Chou, K.H., Nolan, C.E. and Jungalwala, F.B. (1985) J. Neurochem. 44, 1898-1912. [18] Micko, S. and Schlaepfer, W.W. (1978) J. Neurochem. 30, 1041-1049. [19] Schwarzmann, G. (1978) Biochim. Biophys. Acta 529, 106-114. [20] Svennerholm, L., Bostrom, K., Fredman, P., MLnsson, J.-E., Rosengren, B. and Rynmark, B.-M. (1989) Biochim. Biophys. Acta 1005, 109-117. [21] Svennerholm, L. and Fredman, P. (1980) Biochim. Biophys. Acta 617, 97-109. [22] Fredman, P., Nilsson, O., Tayot, J.-L. and Svennerholm, L. (1980) Biochim. Biophys. Acta 618, 42-52. [23] M&son, J.-E., Rynmark, B.-M. and Svennerholm, L. (1991) FEBS Lett. 280, 251-253. [24] Helm, M., Minsson, J.-E., Vanier, M.-T. and Svennerholm, L. (1972) Biochim. Biophys. Acta 280, 356-364. [25] Svennerholm, L. (1957) Biochim. Biophys. Acta 24, 604-611. [26] Lauter, C.J. and Trams, E.G. (1962) J. Lipid Res. 3, 136-138. 1271 M?insson, J.-E., Vanier, M.-T. and Svennerholm, L. (1978) J. Neurochem. 30, 273-275.
et Biophysics Acta 1214 (1994) 115-123
123
[28] M%rsson, J.-E., MO, H., Egge, H. and Svennerholm, L. (1986) FEBS Lett. 196, 259-262. [29] MLnsson, J.-E., Fredman, P., Bigner, D.D., Molin, K., Rosengren, B., Friedman, H.A. and Svennerholm, L. (1986) FEBS Lett. 201, 109-l 13. [30] Svennerholm, L., Minsson, J.-E. and Li, Y.-T. (1973) J. Biol. Chem. 248, 740-742. [31] Ueno, K., Ando, S. and Yu, R.K. (1978) J. Lipid Res. 19, 863-871. [32] Yu, R.K., Ueno, K., Glaser, G.H. and Tourtelotte, W.W. (1982) J. Neurochem. 39, 464-477. [33] Hamberger, A. and Svennerholm, L. (1971) J. Neurochem. 18, 1821-1829. [34] Tiemayer, M., Swank-Hill, P. and Schnaar, R.L. (1990) J. Biol. Chem. 265, 11990-11999. [35] Schengrund, C.L. and Garrigan, O.N. (1969) Lipids 4, 488-495. [36] Minsson, J.-E., Vanier, M.T. and Svennerholm, L. (1978) J. Neurochem. 30, 273-275. [37] Kohriyama, T., Kusunoki, S., Ariga, T., Yoshino, J.E., de Vries, G.H., Latov, N. and Yu, R.K. (1987) J. Neurochem. 48, 15161522. 1381 Yu, R.K., Ariga, T., Kohriyama, T., Kusunoki, S., Maeda, Y. and Miyatani, N. (1990) Ann. Neurol. 27, Suppl., S30-S35. 1391 M%rsson, J.-E., Rynmark, B.-M. and Svennerholm, L. (1991) FEBS Lett. 280, 251-253. [40] Ogawa-Goto, K., Funamoto, N., Outa, Y., Abe, T. and Nagashima, K. (1992) J. Neurochem. 59, 1844-1859. [41] Svennerholm, L., Bruce, A., M&sson, J.-E., Rynmark, B.-M. and Vanier, M.T. (1972) Biochim. Biophys. Acta 280, 626-636. [42] Ogawa-Goto, K., Funamoto, N., Abe, T. and Nagashima, K. (1990) J. Neurochem. 55, 1486-1493.