Detection of alkaline oligoclonal IgC in Cerebrospinal fluid
Electrophoresis 1990. / I , 813-818
Geoffrey Cowdrey Barry Gouldz John Rees' Gary Firth' 'Department of Biochemistry, HurstwoodPark Neurological Centre, Haywards Heath, Sussex *Departmentof Biochemistry, University of Surrey, Guildford, Surrey
8 13
The separation and detection of alkaline oligoclonal IgG bands in cerebrospinal fluid using immobilised pH gradients A method is describedfor the separation and detection ofhighly alkaline IgG bands in unconcentrated cerebrospinal fluid (CSF). These bands are frequently found in the cerebrospinal fluid of patients with inflammatory diseases of the central nervous system, particularly in the case of multiple sclerosis, and their detection is an important aid in clinical diagnosis. An isoelectric focusing technique using an immobilised pH gradient in polyacrylamide gel has been developed over the pH range 7- 10, producing a linear and stable pH gradient with excellent resolution. After electrofocusing, the protein patterns were blotted onto polyvinylidene difluoride membranes and visualised using anti-human IgG followed by an enzyme-labelled second antibody. Blotting could be carried out by capillary diffusion for up to 16 h duration without any loss in resolution. Using this method, highly alkaline intrathecal IgG bands were found in the cerebrospinal fluid of all of the 14 multiple sclerosis patients. There were also 2 patients with alkaline IgG bands in their cerebrospinal fluid who were not diagnosed as multiple sclerosis. By contrast, no alkaline IgG bands with an isoelectric point (pZ) greater than 8.6 were found in any of the serum samples studied ( n = 50) from patients with various neurological disorders including multiple sclerosis.
1 Introduction In certain inflammatory diseases involving the central nervous system (CNS) and particularly in the case ofmultiple sclerosis (MS), intrathecal synthesis of immunoglobulin occurs locally in the brain white matter by lymphocytes situated around areas of demyelination [ 11. Some of this immunoglobulin then finds its way into the circulating cerebrospinal fluid (CSF). The subsequent demonstration of the presence of intrathecal immunoglobulin as oligoclonal banding in the CSF is considered by many to be important as an aid to the diagnosis of MS [2]. Various electrophoretictechniques have been used for this purpose, differing predominantly in their resolving power and sensitivity [3-71. The best resolution has been obtained with isoelectric focusing (IEF) methods using agarose or polyacrylamide gel and carrier ampholytes (CA) to generate the pH gradient [8,91. The potential of IEF for the separation of CSF protein was first demonstrated in 1970 101 and was subsequently used to investigate CSF proteins in various neurological diseases [ 11, 121. These studies indicated thatin samplesof CSF takenfrom patients with MS, there were proteins present with abnormally high isoelectric points (pf), i.e. greater than pH 8.0. Further reports [ 13-151 have confirmed that IgG synthesized intrathecally is more alkaline than serum polyclonal IgG. Serum IgG focuses as multiple bands on IEF between pH 4.7 and 8.6 but in the CSF from patients with MS there were additional bands between pH 8.6 and 9.5 [161. It was assumed that these bands constituted oligoclonal IgG. Recent work [ 171 has confirmed that the synthesis of cathodic IgG is a feature of MS. A few patients also synthesize small amounts of anodic IgG [ 171.
It is also important to be able to distinguish between oligoclonal IgG bands that are due to intrathecal synthesis and those that are present as the result of blood-brain barrier membrane impairment and which have leaked into the CSF from the bloodstream. The criteria most frequently used to establish intrathecal synthesis is the demonstration of IgG bands in the CSF that are not present in the corresponding serum (8, 181. When CA are used for IEF the entire pH gradient slowly migrates towards the cathode: this is known as gradient drift 1191. It is most troublesome when proteins with high p l are analysed as in the case of the alkaline intrathecal IgG. As a consequence, the most alkaline proteins tend to bunch up at the cathodic end of the pH gradient and are either not well resolved or migrate into thecathode wick. Another problem of IEF is theuneven conductivity and buffering capacity produced by the focusing of the individual components of the CA solution into a series of ridges and troughs spread throughout the pH gradient. This in turn leads to the appearance of artefactual bands, even in the case of polyclonal antibody, which by all other electrophoresis methods separates as acontinuous diffuse zone. As a result, IEF patterns can lead to ambiguities in interpretation between the presence of both pathological protein bands and bands caused by artefacts. Despite these limitations, IEF using CA has been successfully employed by many workers to detect oligoclonal IgG bands in CSF and detection rates of 85-100 %have consistently been reported in cases of MS 19, 14-2 1I.
-
Abbreviations: CA, carrier ampholytes; CNS, central nervous system; CSF, cerebrospinal fluid; IEF, isoelectric focusing; IgG, immunoglobulin G ; IPG, immobilised pH gradient; MS; multiple sclerosis; p ~ sphosphate , buffered saline; PI, isoelectric point; PVDF, polyvinylidene difluoride; TEMED, N,N,N,'N',-tetramethylethylenediamine
0VCH Verlagsgesellschaft mbH, D-6940 Weinheirn, 1990
.
at high pH-values* However, we have used the recentiy 'Iltroduced technology of immobilised pH gradients (IPG) [22,231 to develop a method to detect the highly alkaline IgG bands present in the CSF. The method gives higher resolution and is less prone to artefacts. 01 73-0835/90/0909-08 13 %3.50+.25/0
8 14
G. Cowdrey el ai.
Electrophoresis 1990.11, 813-818
2 Materials and methods 2.1 Samples CSF and serum samples were collected from 50 patients who had been admitted to Hurstwood Park Neurological Centre, Haywards Heath, West Sussex, UK for investigation. All samples were stored at 4 "C with 0.1 % w/v sodium azide for up to 4 weeks prior to electrofocusing. The concentrations of IgG in the unconcentrated CSF and in the corresponding sera were measured by radial immunodiffusion. Serum samples, and if necessary CSF samples, were diluted with distilled water and 0.1 pg of IgG applied to the gel.
formers (7 x 2 x 0.25 mm) were placed at a distance of 25 mm from the anode side. The corresponding U-frame (0.5 mm) was attached to provide a mould into which the gel solutions were pumped, having been mixed together in the gradient former. Freshly prepared ammonium persulphate catalyst, (6 pL of 400 mg/mL in distilled water) was used to initiate polymerisation in each gel solution. The gel was allowed to stand undisturbed for a turther 5 rnin to allow any irregularities to smooth out and was then placed in an incubator at 50 OC for 1 h to polymerise. Immediately prior to use the gel mould was dismantled, leaving the plain glass plate with attached IPG gel ready for use.
2.6 Electrofocusing conditions
2.3 Materials Sheep anti-human IgG (heavy chain specific) and donkey anti-sheep alkaline phosphatase-labelled second antibody were from Guildhay Antisera Ltd. (Guildford, UK). Immobiline solutions, CA, electrofocusing wicks, Repel-Silane and IEF calibration set (pH 5-10.5) were from Pharmacia-LKB (Milton Keynes, UK). Nitroblue Tetrazolium, 5-bromo-4chloro-3-indolyl-phosphate toluidine salt, N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid (HEPES) buffer, trypsinogen, L-lactic acid dehydrogenase and cytochrome c (horse muscle) were from Sigma (Poole, Dorset, UK). Tween 20 was from Bio-Rad (Watford, Hertfordshire, UK). AuroDye forte colloidal gold was from Amersham International UK. Acrylamide, N,N'-methylenebisacrylamide, ammonium persulphate, N,N,N',N'-tetramethylethylenediamine (TEMED), sorbitol, Amberlite MB-1 mixed bed resin and all other chemicals were from BDH (Poole, Dorset, UK). Marvel UK dried skimmed milk was used as blocking agent.
2.4 Preparation of IPG gels Stock acrylamide solution was stored over Amberlite MB-1 mixed resin to remove acrylic acid. Starting solutions for an IPG gel, pH 7-10, (4 % acrylamide, 4 % N,N'-methylenebisacrylamide) were prepared essentially as described in LKB Application Note 324 available from Pharmacia-LKB. The volumes of Immobilines and other gel constituents are given in Table 1. In addition, Tween 20 was added to a final concentration of 0.05 % v/v together with CA mixture (pH 7-9 and 9- 11, 1:1 v/v) at a final concentration of 0.2 % v/v. Finally, 3 pL of TEMED were added to each gel solution and the pH adjusted to pH 7.0 by titration with 0.5 M formic acid.
2.5 Gel casting Gels were cast on plain glass plates (260 x 125 mm) which had been coated with Repel-Silane. A row of 22 Dymotape well Table 1. Composition of the light and dense gel solutions Light gel solution PH 7 Immobiline pK 3.6 Immobiline pK 7.0 Immobiline pK 8.5 Immohiline pK 9.3 Sorbitol(60 % w/v) Stock acrylamide Distilled water
220 pL 189 pL 175 pL -
1.0 mL 1.0 mL to 7.5 mL
Dense gel solution pH 10
162 pL 175 pL 170pL 3.8 mL 1.0 mL to 7.5 mL
The anode electrofocusing wick was impregnated with 0.25 M HEPES buffer and the cathode wick with 1.0 M sodium hydroxide. A prefocus run of 30 rnin was usedprior to sample application. The maximum settings on the Macrodrive 5 power supply were for the prefocus 1750 V, 50 mA and 10 W and for electrofocusing 5000 V, 10 mA and 10 W. A dry filter paper strip was applied on the gel adjacent to each electrode wick to drain away surface water. Electrofocusing was complete after 2 h.
2.7 Protein transfer Immunoblotting was carried out using Immobilion polyvinylidene difluoride (PVDF) membranes cut to the appropriate size. The membranes were wetted with methanol, immersed in distilled water and finally soaked in 0.01 M phosphate buffered saline(PBS),pH 7.2.Themembranes werelaidonthe gel surface without trapping any air bubbles and then covered with filter paper (Munktell, Pharmacia - LKB) moistened with PBS. A layer of dry filter papers was then placed on top, followed by a glass plate. A 2 kg weight was applied to the stack and capillary transfer of the proteins was carried out for 1 h. The filter papers were removed, the gel was submerged in distilled water and then the PVDF membranes were peeled off the gel.
2.8 Visualisation of proteins After blotting, any remaining protein binding sites on the membranes were blocked by immersion in a solution of 1 Yo w/v dried skimmed milk in PBS for 10 min with constant shaking at room temperature. The blocking solution was removed and replaced with sheep anti-human IgG diluted 1 in 1000 with 0.2 % w/v dried skimmed milk in PBS solution and incubated for 30 rnin with constant shaking. The primary antibody solution was removed and the membranes washed 3 times for 5 rnin in PBS. The membranes were then incubated for 30 min with donkey anti-sheep alkaline phosphatase-conjugated second antibody diluted 1 in 1000 with 0.2 % w/v dried skimmed milk in PBS solution. The membranes were washed 3 times for 5 rnin in PBS and stained in a solution ofO. 1 mL of 5-bromo-4-chloroindoxyl phosphate, 1 mL of 0.1 % Nitroblue Tetrazolium and 9 mL of alkaline phosphatase buffer solution. Stock solutions for the alkaline phosphatase reaction were prepared as follows: 5-bromo-4-chloroindoxyl phosphate, 5 mg/mL, in dimethylformamide; Nitroblue Tetrazolium, 1 mg/mL, in alkaline phosphatase buffer solution; and alkaline phosphatase buffer solution, 1.02 g of 2-amino-2-
E/ectrophoresis 1990, I / . 813-818
Detection of alkaline oligoclonal IgG in cerebrospinal fluid
8 15
Figure 1. Immunoblot with a PVDF membrane from anIPGgel(A)withoutCA and (B) containing 0.2 % v/v CA. Two pairs of matched serum (S) and C S F (C) from the same patient known to contain intrathecally synthesised alkaline oligoclonal IgG. Total amount of IgG in each sample is 0.1 pg. Detection with sheep antihuman IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with Nitroblue Tetrazolium.
methyl-1,3-propanediol, 1 mg MgC1,.2H20, in 100mLofdistilled water. Marker proteins were stained with colloidal gold solution. Staining took 30-60 min, the membranes were then washed in distilled water and dried at room temperature. The patterns were best viewed by transmitted light from a strong light source.
3.4 Transfer of proteins from gel to PVDF membrane Progressively more protein was transferred from the gel to the PVDF membrane with time (Fig. 3). The actual amounts of protein were not known but even after 20 min, blotting gave the characteristic banding pattern. Blotting was for 1 h, which enabled an experiment to be finished in a working day.
3 Results 3.1 IPG gels run with and without CA When IPG gels were prepared without the addition of any CA it was found that the alkaline oligoclonal IgG migrates towards the cathode in the form of streaks and no resolution was obtained (Fig. la), whereas the addition of CA at a concentration of 0.2 % v/v resulted in the formation of wellseparated and tightly resolved protein bands (Fig. lb). The additionofdifferent concentrationsofCAgreater than0.2 %v/v only led to the formation of artefactual bands (results not shown).
3.2 pH Gradient linearity The linearity of the pH gradient was checked by the use of p l marker proteins (Fig. 2) although the protein patterns were more complicated than suggested by the manufacturers. Linearity was confirmed by plotting the distances migrated by the marker proteins against their respective pl. The higher pH range was calibrated using trypsinogen (pl 9.3) and cytochrome c (pl 10.25). The use of plmarkers, run consistently with each experiment, also acts as an internal quality control of the pH gradient.
3.3 Stability of the pH gradient The stability ofthe PH gradient was demonstrated in (Fig. 3). There was no deterioration in resolution during the h n u n o blotting stage for up to 16 h.
Figure 2. Immunoblot with a PVDF membrane ofisoelectric point calibration marker proteinsrunon anIPGgel,pH 7-10.(A)Lacticdehydrogenase 2 pg and(B) highisoelectricpoint calibration set markerproteins, 3 pg, from Pharmacia (LKB). Staining is with colloidal gold solution.
8 16
G. Cowdrey ef a!.
Electrophoresis 1990. 11. 8 13-8 I8
Figure 3. Two samples of CSF containing alkaline oligoclonai IgG bands run simultaneously on an IPG gel, pH 7-10. Immunoblots with PVDF membranes taken after capillary blotting at (A) 20 min, (B)40 min, (C) 80 min and (D) I 6 h. Total amount of IgG in each sampleis0.1 bg. Detection with sheep anti-human IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with NitroblueTetrazolium.
3.5 Detection of intrathecal IgG synthesis Samples of CSF and serum taken from patients with MS which had been run on an IPG gel, pH 7-10, containing 0.2 % v/v CA clearly showed the presence of alkaline oligoclonal IgG bands in the CSF that were not present in the corresponding serum, indicating that the IgG had been synthesized intrathecally (Fig. 4).
3.6 Correlationof alkalineoligoclonalLgG detection with the diagnosis of MS
In this study, the CSF and corresponding serum from 50 patients admitted to Hurstwood Park Neurological Centre, Haywards Heath, West Sussex, UK for the investigation of various neurological diseases were analysed for the presence of alkaline oligoclonal IgG bands using an IPG, pH 7- 10. Of these, 16 had alkaline oligoclonal IgG bands in their CSF. Out of this group, 14 patients were diagnosed as having MS and all of them had alkaline oligoclonal IgG bands (intrathecal synthesis) in their CSF (Table 2). Diagnostic criteria for the classification of MS was according to Poser 121. There were 2 patients who had alkaline oligoclonal IgG bands in their CSF who were not diagnosed as having MS. The remaining 34 patients studied did not have MS either. Results of the CSF electrofocusing experiments were obtained without prior knowledge of the clinical diagnosis.
4 Discussion There are two good reasons to suggest that IPGgels are ideally suited for the detection of oligoclonal IgG bands in CSF. First, patients suffering from MS have abnormal IgG bands in their CSF, which are characteristically gamma globulins with cathodic electrophoretic mobility. This intrathecally synthesized IgG is predominantly more alkaline than serumTable 2. Comparison of the presence of alkaline oligoclonal IgG bands with the corresponding clinical diagnosis Figure 4. Immunoblot with a PVDF membrane from an IPG gel, pH 7- 10. containing 0.2 % v/v carrier ampholytes. Serum (S) and C S F (C) from two patients without any alkaline oligoclonal IgG bands, (A) and (C) and two patients with intrathecally synthesized alkaline oligoclonal IgG bands (B) and (D). Total amount of IgG applied in each sample is0.l pg. Detection is with sheep anti-human IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with Nitroblue Tetrazolium.
Clinical diagnosis Clinically definite MS Laboratory-supported definite MS Clinically probable MS Laboratory-supported probable MS
Number of Patients with patients alkaline IgG bands 6
6
2 3 3
2 3 3
Electroplzorrsis 1990. 1 1 . 813-818
derived IgG [ 12-16]. Secondly, IEF using an I P G achieves better resolution than p H gradients formed with conventional C A [17, 181 and consequently lower detection limits are attainable. A method has, therefore, been devised for the detection and separation of highly alkaline C S F IgG bands between p H 7 - 10. Initially, we experienced problems with focusing because it took a long time for complete focusing to occur and most samples had migrated as streaks with poor resolution. We found that addition of CA to the gel solutions prior to gel casting at a final concentration of 0.5-1.0 % v/v increased migration and the protein bands became well resolved. However, as the concentration of added C A was increased, more artefactual bands were formed. The best compromise was to add C A to a final concentration of 0.2 % v/v, in which case the focusing was completed in 2 h and the few artefacts produced (usually 5 ) were very faint. This avoided problems with interpretation because any faint artefacts were instantly recognisable with a characteristic p l as continuous bands across the width of the gel. In the niethod described, the distance between anode and cathode was 10 cm. Individual bands were therefore spread out over a long distance. This procedure of spreading the pattern out did not make the IgG bands fainter or less well resolved. On the contrary, the bands were extremely well focused, manifest as sharp, narrow bands only a few microns thick. The overall effect was to achieve the maximum resolution of alkaline bands, well separated and not bunched up at the cathode end of the p H gradient. Significantly, all of the samples from the MS group of patients had between 5 and 10 well-defined, highly alkaline bands with p l values in excess of p H 7.5. This was in marked contrast to the patterns of the corresponding sera in which no bands were detected with a p l greater than 8.6. It is unlikely that any IgG bands with a plgreater than about 9.5 exist in C S F becauseno bands were ever observed in the region of gel at this pH. Isoelectric point calibration markers (LDH and Pharmaciapl calibration set) were run with each gel as a quality control check of the pH gradient and also to assess the p l of the separated bands. There were no significant inter-gelvariations in the patterns of the marker proteins, indicating good reproducibility. Much of the previous work using I P G has used gels that had been washed and then dried before being rehydrated with appropriate buffer solutions immediately prior to use [ 181. By contrast, we found that better results were obtained if gels were not washed. This was because the ionic strength of an unwashed gel is initially much higher than in a washed one and so the sample proteins maintain perfectly straight tracks during separation whereas in washed gels lateral band spreading frequently occurs. Also, the sorbitol density gradient used to form the I P G during gel casting was maintained and subsequently acts as a conductivity quencher [241. This reduces electroendosmosis and also the formation of surface water. Surface water formation was a problem withtheonly previous paper which described the technique of I E F using I P G for the detection of oligoclonal IgG bands in CSF [25]. In our experience with this method, water formed at the electrode wicks which became saturated during focusing. As a consequence, (i) less current flowed through the gel, (ii) the focusing time increased, (iii) there was excessive heating at the edges of the gel causing it to burn, and (iv) water, collecting mainly at the anode wick, leaked around the side of the gel,
Detection of alkaline ollgoclonal IgC in cerebrospinal fluld
8I7
causing a short circuit between the electrodes. These problems were largely solved by applying lengths of filter paper, adjacent to the electrode wicks, to act as a drainage system. Other differences between this work, compared to the previous paper [241, include the use of unconcentrated C S F and a shorter IEF time (2 h). Initially, we attempted to transfer the separated protein patterns onto nitrocellulose membranes but it was not possible to remove the nitrocellulose from the gel surface after blotting. This problem was completely overcome by using PVDF membranes. In addition, these membranes took up and retained more protein than nitrocellulose (261. Immunoblotting by capillary transfer using PVDF membranes for 15 min gave satisfactory results. If the patterns were transferred overnight, or over a long period of time, there was no loss of definition to the patterns. This was probably because the focused bands were tightly held at their respective p l within the IPG and thus unable to diffuse and also immobilised when bound to the P V D F membrane. This study set out to identify intrathecally synthesized IgG in unconcentrated CSF by virtue ofits char,acteristically high p l [ 12-16]. There are some patients who also synthesize anodic IgG and it could be argued that awidepHgradient, e. g., 3- 10, is necessary if all oligoclonal IgG bands are to be detected. Nevertheless, in this study we have detected alkaline intrathecal oligoclonal IgG bands in all of 14 MS patients. There were, however, 2 patients with alkaline oligoclonal IgG bands in their C S F who had not been diagnosed as having MS. One was a patient with Guillain Barre Syndrome, whereas the diagnosis of the other was uncertain after routine neurological investigation. Furthermore, we have noticed that none of the serum samples showed oligoclonal IgG bands at p H values greater than about p H 8.6. If this finding is maintained then it may be possible to dispense with serum samples altogether and any IgG bands found in the CSF at a very alkaline pH could be identified as intrathecally synthesized. We have shown that IEF using IPG offers improved resolution and detection of alkaline oligoclonal IgG bands. The use of this technique in the investigation of neurological diseases is being continued. Further work is in progress to study the wider application of this electrophoretic technique and to establish the structural properties of oligoclonal IgG bands that produce their alkaline pls. Received April 1, I990
5 References [11 Thompson, E. J., Br. Med. Bull. 1977,33, 28-30 121 Poser, C. M., Paty, D. W., Scheinberg, L.,McDonald,I., Davis, F. A., Ebers, G. C., Johnson, K. P., Stibley, W. A., Silberberg, D. H. and Tourtellotte, W. W., Ann. Neurol. 1983, 13, 227-231. [31 Glasner, H., J . Neurol. 1978,218, 73-76. 141 Lowenthal, A,, van Sande, M. and Karcher, D., J. Neurochem. 1960, 6, 51-56. 151 Thompson, E. J., Kaufmann, P., Shortman, R. C., Rudge, P. andMcDonald, W. I., Brit. Med.J. 1979, 1, 16-17. 161 Walker. R. W. H., Keir, G., Johnson, M. H. and Thompson, E. J., J . Neuroirnmunol. 1983,4, 141-148. 171 Hershey, L. A. and Trotter, J . L., Ann. Neurol. 1980,8,426-434. 181 Kostulas, V. K., Link, H. and Lefvert, A. K., Arch. Neurol. 1987,44, 1041- 1044.
8 18
G. Cowdrey et a[.
[91 Fossard, C., Dale, G. and Latner, A. L., J. Clin. Pathol. 1970,23, 586-589. [lo] Kjellin,K.G.andVesterberg,O.,J. Neurol.Sci. 1974,23,199-213. 11 11 Siden, A. and Kjellin, K. G., J. Neurol. Sci. 1978,39, 131-146. I121 Delmotte, P. and Gonsette, R., J. NeuroI. 1977,215,27-37. 1131 Hosein, Z. Z. and Johnson, K. P., Neurology 1981,3I, 70-76. 1141 Zaffaroni, M., Caputo, D. and Cazzullo, C. L.,J. Neurol. 1983,229, 55-60. [151 Laurenzi, M. A. and Link, H.,J. Neurol. Neurosurg. Psychiat. 1979, 42,368-372. [161 Tourtellotte, W. W., Potvin, A. R., Booe, I., Baumhefner, R. W., Walsh, M. J., Dickstein, P., Ingram,T., Cowan, T., Shapshak, P. and Delmotte, P., Neurology 1982,32,261-266. 1 171 Righetti, P. G. and Drysdale, J. W., Ann. N . Y.Acad. Sci. 1973,209, 163- 186. 1181 Olsson, T., Kostulas, V. and Link, H., Clin. Chem. 1984, 30, 1246- 1249.
Electrophoresis 1990, 11, 813-818 1191 Ersmark, B. and Siden, A.,J. Neurol. 1984,231, 117-121. [201 Mattson, D. H., Roos, R. P. and Arnason, B. G. W., Ann. Neurol. 1981,9,34-41. 1211 Trbojevic-Cepe, M., Poljakovic, Z., Vrkic, N. and Bielen, I., J. Clin. Chem. Clin. Biochem. 1989,27,211-216. (221 Bjellqvist, B., Ek, K., Righetti, P. G., Giannazza, E., Gorg, A,, Westermeir, R. and Postel, W., J. Biochem. Biophys. Methods 1982, 6,317-339. I231 Righetti, P. G. and Giannazza, E., Methods Biochem. Analysis 1987, 32,215-278. [241 Mosher, R. A., Bier, M. and Righetti, P. G., Electrophoresis 1986, 7 , 128-133. [251 Pirttila, T., Frey, H., Mattila, K. and Siden, A., Electrophoresis 1988, 9,582-588. 1261 Nespolo, A., Bianchi, G., Salmaggi, A., Lazzaroni, M., Cerrato, D. and Tajoli, M., Electrophoresis 1989,10,34-40.
Electrophoresis 1990. / I , 813-818
Geoffrey Cowdrey Barry Gouldz John Rees' Gary Firth' 'Department of Biochemistry, HurstwoodPark Neurological Centre, Haywards Heath, Sussex *Departmentof Biochemistry, University of Surrey, Guildford, Surrey
8 13
The separation and detection of alkaline oligoclonal IgG bands in cerebrospinal fluid using immobilised pH gradients A method is describedfor the separation and detection ofhighly alkaline IgG bands in unconcentrated cerebrospinal fluid (CSF). These bands are frequently found in the cerebrospinal fluid of patients with inflammatory diseases of the central nervous system, particularly in the case of multiple sclerosis, and their detection is an important aid in clinical diagnosis. An isoelectric focusing technique using an immobilised pH gradient in polyacrylamide gel has been developed over the pH range 7- 10, producing a linear and stable pH gradient with excellent resolution. After electrofocusing, the protein patterns were blotted onto polyvinylidene difluoride membranes and visualised using anti-human IgG followed by an enzyme-labelled second antibody. Blotting could be carried out by capillary diffusion for up to 16 h duration without any loss in resolution. Using this method, highly alkaline intrathecal IgG bands were found in the cerebrospinal fluid of all of the 14 multiple sclerosis patients. There were also 2 patients with alkaline IgG bands in their cerebrospinal fluid who were not diagnosed as multiple sclerosis. By contrast, no alkaline IgG bands with an isoelectric point (pZ) greater than 8.6 were found in any of the serum samples studied ( n = 50) from patients with various neurological disorders including multiple sclerosis.
1 Introduction In certain inflammatory diseases involving the central nervous system (CNS) and particularly in the case ofmultiple sclerosis (MS), intrathecal synthesis of immunoglobulin occurs locally in the brain white matter by lymphocytes situated around areas of demyelination [ 11. Some of this immunoglobulin then finds its way into the circulating cerebrospinal fluid (CSF). The subsequent demonstration of the presence of intrathecal immunoglobulin as oligoclonal banding in the CSF is considered by many to be important as an aid to the diagnosis of MS [2]. Various electrophoretictechniques have been used for this purpose, differing predominantly in their resolving power and sensitivity [3-71. The best resolution has been obtained with isoelectric focusing (IEF) methods using agarose or polyacrylamide gel and carrier ampholytes (CA) to generate the pH gradient [8,91. The potential of IEF for the separation of CSF protein was first demonstrated in 1970 101 and was subsequently used to investigate CSF proteins in various neurological diseases [ 11, 121. These studies indicated thatin samplesof CSF takenfrom patients with MS, there were proteins present with abnormally high isoelectric points (pf), i.e. greater than pH 8.0. Further reports [ 13-151 have confirmed that IgG synthesized intrathecally is more alkaline than serum polyclonal IgG. Serum IgG focuses as multiple bands on IEF between pH 4.7 and 8.6 but in the CSF from patients with MS there were additional bands between pH 8.6 and 9.5 [161. It was assumed that these bands constituted oligoclonal IgG. Recent work [ 171 has confirmed that the synthesis of cathodic IgG is a feature of MS. A few patients also synthesize small amounts of anodic IgG [ 171.
It is also important to be able to distinguish between oligoclonal IgG bands that are due to intrathecal synthesis and those that are present as the result of blood-brain barrier membrane impairment and which have leaked into the CSF from the bloodstream. The criteria most frequently used to establish intrathecal synthesis is the demonstration of IgG bands in the CSF that are not present in the corresponding serum (8, 181. When CA are used for IEF the entire pH gradient slowly migrates towards the cathode: this is known as gradient drift 1191. It is most troublesome when proteins with high p l are analysed as in the case of the alkaline intrathecal IgG. As a consequence, the most alkaline proteins tend to bunch up at the cathodic end of the pH gradient and are either not well resolved or migrate into thecathode wick. Another problem of IEF is theuneven conductivity and buffering capacity produced by the focusing of the individual components of the CA solution into a series of ridges and troughs spread throughout the pH gradient. This in turn leads to the appearance of artefactual bands, even in the case of polyclonal antibody, which by all other electrophoresis methods separates as acontinuous diffuse zone. As a result, IEF patterns can lead to ambiguities in interpretation between the presence of both pathological protein bands and bands caused by artefacts. Despite these limitations, IEF using CA has been successfully employed by many workers to detect oligoclonal IgG bands in CSF and detection rates of 85-100 %have consistently been reported in cases of MS 19, 14-2 1I.
-
Abbreviations: CA, carrier ampholytes; CNS, central nervous system; CSF, cerebrospinal fluid; IEF, isoelectric focusing; IgG, immunoglobulin G ; IPG, immobilised pH gradient; MS; multiple sclerosis; p ~ sphosphate , buffered saline; PI, isoelectric point; PVDF, polyvinylidene difluoride; TEMED, N,N,N,'N',-tetramethylethylenediamine
0VCH Verlagsgesellschaft mbH, D-6940 Weinheirn, 1990
.
at high pH-values* However, we have used the recentiy 'Iltroduced technology of immobilised pH gradients (IPG) [22,231 to develop a method to detect the highly alkaline IgG bands present in the CSF. The method gives higher resolution and is less prone to artefacts. 01 73-0835/90/0909-08 13 %3.50+.25/0
8 14
G. Cowdrey el ai.
Electrophoresis 1990.11, 813-818
2 Materials and methods 2.1 Samples CSF and serum samples were collected from 50 patients who had been admitted to Hurstwood Park Neurological Centre, Haywards Heath, West Sussex, UK for investigation. All samples were stored at 4 "C with 0.1 % w/v sodium azide for up to 4 weeks prior to electrofocusing. The concentrations of IgG in the unconcentrated CSF and in the corresponding sera were measured by radial immunodiffusion. Serum samples, and if necessary CSF samples, were diluted with distilled water and 0.1 pg of IgG applied to the gel.
formers (7 x 2 x 0.25 mm) were placed at a distance of 25 mm from the anode side. The corresponding U-frame (0.5 mm) was attached to provide a mould into which the gel solutions were pumped, having been mixed together in the gradient former. Freshly prepared ammonium persulphate catalyst, (6 pL of 400 mg/mL in distilled water) was used to initiate polymerisation in each gel solution. The gel was allowed to stand undisturbed for a turther 5 rnin to allow any irregularities to smooth out and was then placed in an incubator at 50 OC for 1 h to polymerise. Immediately prior to use the gel mould was dismantled, leaving the plain glass plate with attached IPG gel ready for use.
2.6 Electrofocusing conditions
2.3 Materials Sheep anti-human IgG (heavy chain specific) and donkey anti-sheep alkaline phosphatase-labelled second antibody were from Guildhay Antisera Ltd. (Guildford, UK). Immobiline solutions, CA, electrofocusing wicks, Repel-Silane and IEF calibration set (pH 5-10.5) were from Pharmacia-LKB (Milton Keynes, UK). Nitroblue Tetrazolium, 5-bromo-4chloro-3-indolyl-phosphate toluidine salt, N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid (HEPES) buffer, trypsinogen, L-lactic acid dehydrogenase and cytochrome c (horse muscle) were from Sigma (Poole, Dorset, UK). Tween 20 was from Bio-Rad (Watford, Hertfordshire, UK). AuroDye forte colloidal gold was from Amersham International UK. Acrylamide, N,N'-methylenebisacrylamide, ammonium persulphate, N,N,N',N'-tetramethylethylenediamine (TEMED), sorbitol, Amberlite MB-1 mixed bed resin and all other chemicals were from BDH (Poole, Dorset, UK). Marvel UK dried skimmed milk was used as blocking agent.
2.4 Preparation of IPG gels Stock acrylamide solution was stored over Amberlite MB-1 mixed resin to remove acrylic acid. Starting solutions for an IPG gel, pH 7-10, (4 % acrylamide, 4 % N,N'-methylenebisacrylamide) were prepared essentially as described in LKB Application Note 324 available from Pharmacia-LKB. The volumes of Immobilines and other gel constituents are given in Table 1. In addition, Tween 20 was added to a final concentration of 0.05 % v/v together with CA mixture (pH 7-9 and 9- 11, 1:1 v/v) at a final concentration of 0.2 % v/v. Finally, 3 pL of TEMED were added to each gel solution and the pH adjusted to pH 7.0 by titration with 0.5 M formic acid.
2.5 Gel casting Gels were cast on plain glass plates (260 x 125 mm) which had been coated with Repel-Silane. A row of 22 Dymotape well Table 1. Composition of the light and dense gel solutions Light gel solution PH 7 Immobiline pK 3.6 Immobiline pK 7.0 Immobiline pK 8.5 Immohiline pK 9.3 Sorbitol(60 % w/v) Stock acrylamide Distilled water
220 pL 189 pL 175 pL -
1.0 mL 1.0 mL to 7.5 mL
Dense gel solution pH 10
162 pL 175 pL 170pL 3.8 mL 1.0 mL to 7.5 mL
The anode electrofocusing wick was impregnated with 0.25 M HEPES buffer and the cathode wick with 1.0 M sodium hydroxide. A prefocus run of 30 rnin was usedprior to sample application. The maximum settings on the Macrodrive 5 power supply were for the prefocus 1750 V, 50 mA and 10 W and for electrofocusing 5000 V, 10 mA and 10 W. A dry filter paper strip was applied on the gel adjacent to each electrode wick to drain away surface water. Electrofocusing was complete after 2 h.
2.7 Protein transfer Immunoblotting was carried out using Immobilion polyvinylidene difluoride (PVDF) membranes cut to the appropriate size. The membranes were wetted with methanol, immersed in distilled water and finally soaked in 0.01 M phosphate buffered saline(PBS),pH 7.2.Themembranes werelaidonthe gel surface without trapping any air bubbles and then covered with filter paper (Munktell, Pharmacia - LKB) moistened with PBS. A layer of dry filter papers was then placed on top, followed by a glass plate. A 2 kg weight was applied to the stack and capillary transfer of the proteins was carried out for 1 h. The filter papers were removed, the gel was submerged in distilled water and then the PVDF membranes were peeled off the gel.
2.8 Visualisation of proteins After blotting, any remaining protein binding sites on the membranes were blocked by immersion in a solution of 1 Yo w/v dried skimmed milk in PBS for 10 min with constant shaking at room temperature. The blocking solution was removed and replaced with sheep anti-human IgG diluted 1 in 1000 with 0.2 % w/v dried skimmed milk in PBS solution and incubated for 30 rnin with constant shaking. The primary antibody solution was removed and the membranes washed 3 times for 5 rnin in PBS. The membranes were then incubated for 30 min with donkey anti-sheep alkaline phosphatase-conjugated second antibody diluted 1 in 1000 with 0.2 % w/v dried skimmed milk in PBS solution. The membranes were washed 3 times for 5 rnin in PBS and stained in a solution ofO. 1 mL of 5-bromo-4-chloroindoxyl phosphate, 1 mL of 0.1 % Nitroblue Tetrazolium and 9 mL of alkaline phosphatase buffer solution. Stock solutions for the alkaline phosphatase reaction were prepared as follows: 5-bromo-4-chloroindoxyl phosphate, 5 mg/mL, in dimethylformamide; Nitroblue Tetrazolium, 1 mg/mL, in alkaline phosphatase buffer solution; and alkaline phosphatase buffer solution, 1.02 g of 2-amino-2-
E/ectrophoresis 1990, I / . 813-818
Detection of alkaline oligoclonal IgG in cerebrospinal fluid
8 15
Figure 1. Immunoblot with a PVDF membrane from anIPGgel(A)withoutCA and (B) containing 0.2 % v/v CA. Two pairs of matched serum (S) and C S F (C) from the same patient known to contain intrathecally synthesised alkaline oligoclonal IgG. Total amount of IgG in each sample is 0.1 pg. Detection with sheep antihuman IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with Nitroblue Tetrazolium.
methyl-1,3-propanediol, 1 mg MgC1,.2H20, in 100mLofdistilled water. Marker proteins were stained with colloidal gold solution. Staining took 30-60 min, the membranes were then washed in distilled water and dried at room temperature. The patterns were best viewed by transmitted light from a strong light source.
3.4 Transfer of proteins from gel to PVDF membrane Progressively more protein was transferred from the gel to the PVDF membrane with time (Fig. 3). The actual amounts of protein were not known but even after 20 min, blotting gave the characteristic banding pattern. Blotting was for 1 h, which enabled an experiment to be finished in a working day.
3 Results 3.1 IPG gels run with and without CA When IPG gels were prepared without the addition of any CA it was found that the alkaline oligoclonal IgG migrates towards the cathode in the form of streaks and no resolution was obtained (Fig. la), whereas the addition of CA at a concentration of 0.2 % v/v resulted in the formation of wellseparated and tightly resolved protein bands (Fig. lb). The additionofdifferent concentrationsofCAgreater than0.2 %v/v only led to the formation of artefactual bands (results not shown).
3.2 pH Gradient linearity The linearity of the pH gradient was checked by the use of p l marker proteins (Fig. 2) although the protein patterns were more complicated than suggested by the manufacturers. Linearity was confirmed by plotting the distances migrated by the marker proteins against their respective pl. The higher pH range was calibrated using trypsinogen (pl 9.3) and cytochrome c (pl 10.25). The use of plmarkers, run consistently with each experiment, also acts as an internal quality control of the pH gradient.
3.3 Stability of the pH gradient The stability ofthe PH gradient was demonstrated in (Fig. 3). There was no deterioration in resolution during the h n u n o blotting stage for up to 16 h.
Figure 2. Immunoblot with a PVDF membrane ofisoelectric point calibration marker proteinsrunon anIPGgel,pH 7-10.(A)Lacticdehydrogenase 2 pg and(B) highisoelectricpoint calibration set markerproteins, 3 pg, from Pharmacia (LKB). Staining is with colloidal gold solution.
8 16
G. Cowdrey ef a!.
Electrophoresis 1990. 11. 8 13-8 I8
Figure 3. Two samples of CSF containing alkaline oligoclonai IgG bands run simultaneously on an IPG gel, pH 7-10. Immunoblots with PVDF membranes taken after capillary blotting at (A) 20 min, (B)40 min, (C) 80 min and (D) I 6 h. Total amount of IgG in each sampleis0.1 bg. Detection with sheep anti-human IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with NitroblueTetrazolium.
3.5 Detection of intrathecal IgG synthesis Samples of CSF and serum taken from patients with MS which had been run on an IPG gel, pH 7-10, containing 0.2 % v/v CA clearly showed the presence of alkaline oligoclonal IgG bands in the CSF that were not present in the corresponding serum, indicating that the IgG had been synthesized intrathecally (Fig. 4).
3.6 Correlationof alkalineoligoclonalLgG detection with the diagnosis of MS
In this study, the CSF and corresponding serum from 50 patients admitted to Hurstwood Park Neurological Centre, Haywards Heath, West Sussex, UK for the investigation of various neurological diseases were analysed for the presence of alkaline oligoclonal IgG bands using an IPG, pH 7- 10. Of these, 16 had alkaline oligoclonal IgG bands in their CSF. Out of this group, 14 patients were diagnosed as having MS and all of them had alkaline oligoclonal IgG bands (intrathecal synthesis) in their CSF (Table 2). Diagnostic criteria for the classification of MS was according to Poser 121. There were 2 patients who had alkaline oligoclonal IgG bands in their CSF who were not diagnosed as having MS. The remaining 34 patients studied did not have MS either. Results of the CSF electrofocusing experiments were obtained without prior knowledge of the clinical diagnosis.
4 Discussion There are two good reasons to suggest that IPGgels are ideally suited for the detection of oligoclonal IgG bands in CSF. First, patients suffering from MS have abnormal IgG bands in their CSF, which are characteristically gamma globulins with cathodic electrophoretic mobility. This intrathecally synthesized IgG is predominantly more alkaline than serumTable 2. Comparison of the presence of alkaline oligoclonal IgG bands with the corresponding clinical diagnosis Figure 4. Immunoblot with a PVDF membrane from an IPG gel, pH 7- 10. containing 0.2 % v/v carrier ampholytes. Serum (S) and C S F (C) from two patients without any alkaline oligoclonal IgG bands, (A) and (C) and two patients with intrathecally synthesized alkaline oligoclonal IgG bands (B) and (D). Total amount of IgG applied in each sample is0.l pg. Detection is with sheep anti-human IgG, followed by donkey anti-sheep alkaline phosphatase conjugate and enhanced with Nitroblue Tetrazolium.
Clinical diagnosis Clinically definite MS Laboratory-supported definite MS Clinically probable MS Laboratory-supported probable MS
Number of Patients with patients alkaline IgG bands 6
6
2 3 3
2 3 3
Electroplzorrsis 1990. 1 1 . 813-818
derived IgG [ 12-16]. Secondly, IEF using an I P G achieves better resolution than p H gradients formed with conventional C A [17, 181 and consequently lower detection limits are attainable. A method has, therefore, been devised for the detection and separation of highly alkaline C S F IgG bands between p H 7 - 10. Initially, we experienced problems with focusing because it took a long time for complete focusing to occur and most samples had migrated as streaks with poor resolution. We found that addition of CA to the gel solutions prior to gel casting at a final concentration of 0.5-1.0 % v/v increased migration and the protein bands became well resolved. However, as the concentration of added C A was increased, more artefactual bands were formed. The best compromise was to add C A to a final concentration of 0.2 % v/v, in which case the focusing was completed in 2 h and the few artefacts produced (usually 5 ) were very faint. This avoided problems with interpretation because any faint artefacts were instantly recognisable with a characteristic p l as continuous bands across the width of the gel. In the niethod described, the distance between anode and cathode was 10 cm. Individual bands were therefore spread out over a long distance. This procedure of spreading the pattern out did not make the IgG bands fainter or less well resolved. On the contrary, the bands were extremely well focused, manifest as sharp, narrow bands only a few microns thick. The overall effect was to achieve the maximum resolution of alkaline bands, well separated and not bunched up at the cathode end of the p H gradient. Significantly, all of the samples from the MS group of patients had between 5 and 10 well-defined, highly alkaline bands with p l values in excess of p H 7.5. This was in marked contrast to the patterns of the corresponding sera in which no bands were detected with a p l greater than 8.6. It is unlikely that any IgG bands with a plgreater than about 9.5 exist in C S F becauseno bands were ever observed in the region of gel at this pH. Isoelectric point calibration markers (LDH and Pharmaciapl calibration set) were run with each gel as a quality control check of the pH gradient and also to assess the p l of the separated bands. There were no significant inter-gelvariations in the patterns of the marker proteins, indicating good reproducibility. Much of the previous work using I P G has used gels that had been washed and then dried before being rehydrated with appropriate buffer solutions immediately prior to use [ 181. By contrast, we found that better results were obtained if gels were not washed. This was because the ionic strength of an unwashed gel is initially much higher than in a washed one and so the sample proteins maintain perfectly straight tracks during separation whereas in washed gels lateral band spreading frequently occurs. Also, the sorbitol density gradient used to form the I P G during gel casting was maintained and subsequently acts as a conductivity quencher [241. This reduces electroendosmosis and also the formation of surface water. Surface water formation was a problem withtheonly previous paper which described the technique of I E F using I P G for the detection of oligoclonal IgG bands in CSF [25]. In our experience with this method, water formed at the electrode wicks which became saturated during focusing. As a consequence, (i) less current flowed through the gel, (ii) the focusing time increased, (iii) there was excessive heating at the edges of the gel causing it to burn, and (iv) water, collecting mainly at the anode wick, leaked around the side of the gel,
Detection of alkaline ollgoclonal IgC in cerebrospinal fluld
8I7
causing a short circuit between the electrodes. These problems were largely solved by applying lengths of filter paper, adjacent to the electrode wicks, to act as a drainage system. Other differences between this work, compared to the previous paper [241, include the use of unconcentrated C S F and a shorter IEF time (2 h). Initially, we attempted to transfer the separated protein patterns onto nitrocellulose membranes but it was not possible to remove the nitrocellulose from the gel surface after blotting. This problem was completely overcome by using PVDF membranes. In addition, these membranes took up and retained more protein than nitrocellulose (261. Immunoblotting by capillary transfer using PVDF membranes for 15 min gave satisfactory results. If the patterns were transferred overnight, or over a long period of time, there was no loss of definition to the patterns. This was probably because the focused bands were tightly held at their respective p l within the IPG and thus unable to diffuse and also immobilised when bound to the P V D F membrane. This study set out to identify intrathecally synthesized IgG in unconcentrated CSF by virtue ofits char,acteristically high p l [ 12-16]. There are some patients who also synthesize anodic IgG and it could be argued that awidepHgradient, e. g., 3- 10, is necessary if all oligoclonal IgG bands are to be detected. Nevertheless, in this study we have detected alkaline intrathecal oligoclonal IgG bands in all of 14 MS patients. There were, however, 2 patients with alkaline oligoclonal IgG bands in their C S F who had not been diagnosed as having MS. One was a patient with Guillain Barre Syndrome, whereas the diagnosis of the other was uncertain after routine neurological investigation. Furthermore, we have noticed that none of the serum samples showed oligoclonal IgG bands at p H values greater than about p H 8.6. If this finding is maintained then it may be possible to dispense with serum samples altogether and any IgG bands found in the CSF at a very alkaline pH could be identified as intrathecally synthesized. We have shown that IEF using IPG offers improved resolution and detection of alkaline oligoclonal IgG bands. The use of this technique in the investigation of neurological diseases is being continued. Further work is in progress to study the wider application of this electrophoretic technique and to establish the structural properties of oligoclonal IgG bands that produce their alkaline pls. Received April 1, I990
5 References [11 Thompson, E. J., Br. Med. Bull. 1977,33, 28-30 121 Poser, C. M., Paty, D. W., Scheinberg, L.,McDonald,I., Davis, F. A., Ebers, G. C., Johnson, K. P., Stibley, W. A., Silberberg, D. H. and Tourtellotte, W. W., Ann. Neurol. 1983, 13, 227-231. [31 Glasner, H., J . Neurol. 1978,218, 73-76. 141 Lowenthal, A,, van Sande, M. and Karcher, D., J. Neurochem. 1960, 6, 51-56. 151 Thompson, E. J., Kaufmann, P., Shortman, R. C., Rudge, P. andMcDonald, W. I., Brit. Med.J. 1979, 1, 16-17. 161 Walker. R. W. H., Keir, G., Johnson, M. H. and Thompson, E. J., J . Neuroirnmunol. 1983,4, 141-148. 171 Hershey, L. A. and Trotter, J . L., Ann. Neurol. 1980,8,426-434. 181 Kostulas, V. K., Link, H. and Lefvert, A. K., Arch. Neurol. 1987,44, 1041- 1044.
8 18
G. Cowdrey et a[.
[91 Fossard, C., Dale, G. and Latner, A. L., J. Clin. Pathol. 1970,23, 586-589. [lo] Kjellin,K.G.andVesterberg,O.,J. Neurol.Sci. 1974,23,199-213. 11 11 Siden, A. and Kjellin, K. G., J. Neurol. Sci. 1978,39, 131-146. I121 Delmotte, P. and Gonsette, R., J. NeuroI. 1977,215,27-37. 1131 Hosein, Z. Z. and Johnson, K. P., Neurology 1981,3I, 70-76. 1141 Zaffaroni, M., Caputo, D. and Cazzullo, C. L.,J. Neurol. 1983,229, 55-60. [151 Laurenzi, M. A. and Link, H.,J. Neurol. Neurosurg. Psychiat. 1979, 42,368-372. [161 Tourtellotte, W. W., Potvin, A. R., Booe, I., Baumhefner, R. W., Walsh, M. J., Dickstein, P., Ingram,T., Cowan, T., Shapshak, P. and Delmotte, P., Neurology 1982,32,261-266. 1 171 Righetti, P. G. and Drysdale, J. W., Ann. N . Y.Acad. Sci. 1973,209, 163- 186. 1181 Olsson, T., Kostulas, V. and Link, H., Clin. Chem. 1984, 30, 1246- 1249.
Electrophoresis 1990, 11, 813-818 1191 Ersmark, B. and Siden, A.,J. Neurol. 1984,231, 117-121. [201 Mattson, D. H., Roos, R. P. and Arnason, B. G. W., Ann. Neurol. 1981,9,34-41. 1211 Trbojevic-Cepe, M., Poljakovic, Z., Vrkic, N. and Bielen, I., J. Clin. Chem. Clin. Biochem. 1989,27,211-216. (221 Bjellqvist, B., Ek, K., Righetti, P. G., Giannazza, E., Gorg, A,, Westermeir, R. and Postel, W., J. Biochem. Biophys. Methods 1982, 6,317-339. I231 Righetti, P. G. and Giannazza, E., Methods Biochem. Analysis 1987, 32,215-278. [241 Mosher, R. A., Bier, M. and Righetti, P. G., Electrophoresis 1986, 7 , 128-133. [251 Pirttila, T., Frey, H., Mattila, K. and Siden, A., Electrophoresis 1988, 9,582-588. 1261 Nespolo, A., Bianchi, G., Salmaggi, A., Lazzaroni, M., Cerrato, D. and Tajoli, M., Electrophoresis 1989,10,34-40.
