Lipid A fractions analyzed by a technique involving thin-layer chromatography and enzyme-linked immunosorbent assay

Transcription

1 Eur. J. Biochem. 138, (1984) 0 FEBS 1984 Lipid A fractions analyzed by a technique involving thin-layer chromatography and enzyme-linked immunosorbent assay Inger MATTSBY-BALTZER and Carl R. ALVING Department of Membrane Biochemistry, Walter Reed Army Institute of Research, Washington DC (Received July S/September 30, 1983) - EJB A modified enzyme-linked immunosorbent assay (ELISA), with alkaline phosphatase as enzyme, was used for the study of antigenicity of lipid A fractions directly on thin-layer chromatographic (TLC) plates. For visualization a gel slab containing the enzyme substrate was placed on the plate containing enzyme-conjugated antibodies. The plate was read by a thin-layer chromatogram spectrophotometer. The immunoassay was both highly specific and quite sensitive. Sensitivity was superior to levels obtained by staining the plate with molybdenum blue (for phosphate) or orcinol (for carbohydrate). Fractions of lipid A from Escherichia coli0111, Shigellaflexneri or Salmonella minnesota R595, after being separated by thin-layer chromatography, were analyzed using rabbit anti-(lipid A) serum. Patterns obtained by scanning the same plates for phosphate staining and for the TLC-ELISA corresponded well. For comparison with TLC-ELISA, an inhibition assay was run using a tube ELISA. The tube ELISA, run in aqueous medium, showed that fractions 6-8 (those having the highest R, values) had the least activities. In contrast, TLC- ELISA did not detect large differences between fractions 2-7. This discrepancy probably reflected limited aqueous solubility of fractions 6 and 7. We conclude that TLC-ELISA might reveal antigenic activities of lipids that could be missed by other methods. The data suggested that all fractions, except for fraction 8, were similar in their antigenicity by TLC-ELISA. Differences in antigenicity between the fractions occurred when the fractions were tested in free form in an aqueous environment and these differences possibly could have been due to different solubilities of individual fractions. Lipid A, the lipid part of bacterial lipopolysaccharide, has been shown to be heterogeneous [l However, despite the heterogeneity observed by techniques such as thin-layer chromatography, the basic structure of lipid A is thought to be similar for many bacterial strains [13]. Investigation on the antigenicity of eight purified lipid A fractions from Shigella flexneri indicated that the fractions having the highest mobilities on thin-layer chromatography (TLC) (fractions 6-8) were not antigenic when presented on liposomes [12]. Recently we found that several bacterial strains contained lipid A fractions which were similar to those found in Sh.flexneri [14]. From these studies the question arose whether lipid A fractions having similar R, values, but from different bacterial strains, would have similar abilities to react with anti-(lipid A) antibodies. In the present investigation a modified ELISA applied directly on the TLC plate (TLC-ELISA) has been developed in order to study the antigenicity of lipid A fractions without requiring previous purification steps. This assay is analogous to the immuno-tlc method developed by Magnani and coworkers [15] except that an ELISA technique replaces radioautography. A comparison is made between the antigenic characteristics of lipid A fractions adsorbed to a solid phase (TLC- ELISA) and the immuno-inhibitory capacity of purified lipid A fractions in aqueous solution in a previously modified ELISA performed in test tubes [16]. Abbreviations. ELISA, Enzyme-linked immunosorbent assay; TLC, thin-layer chromatography. MATERIALS AND METHODS Bacteria Escherichia coli EHlOO was grown on nutrient agar broth and harvested the next day. Lipopolysaccharide Lipopolysaccharides prepared by the Westphal method from Shigella flexneri, E. coli 01 11, and Salmonella minnesota R595 were purchased from Difco laboratories (Detroit, MI). E. coli EHlOO lipopolysaccharide was prepared according to the phenol chloroform petroleum ether procedure [17]. Lipid A The lipopolysaccharide preparations were boiled for 2 h with 1 % acetic acid. After washing three times with distilled water the precipitates were lyophilized. A stock solution of lipid A from E. coli EHlOO was solubilized with 0.5% triethylamine and prepared for ELISA performed in tubes. Lipid A was further purified by EDTA treatment followed by chloroform extraction [12]. Eight fractions of lipid A from Sh. flexneri, earlier described [12], were isolated by preparative thin-layer chromatography [12]. Lipid A fractions added to diluted antiserum for the inhibition assay were first solubilized with 0.5 % triethylamine. A glycine saline buffer (ph 8.6, 0.5 M) was used in this step. Lipid A that was not treated with EDTA followed by chloroform extraction will be referred to as crude lipid A.

2 334 Antiserum Rabbit anti-(lipid A) serum was produced as earlier described [12,16]. Acid-treated bacterial cells (S. minnesota R595) coated with crude lipid A [18] or crude lipid A from E. coli EHlOO coated on human erythrocytes [I61 was used as the antigen for immunization. Thin-layer chromatography (TLC) Lipid A from different sources was run on a water-resistant TLC plate (Whatman K6) with chloroform methanol water, (130:45:7, v/v) as solvent. The plate was stained by spraying with orcinol reagent [I91 for detection of carbohydrate or molybdenum blue for phosphate [20]. Enzyme-linked immunoso rbent assay (EL ISA) Tube assay. The procedure in tubes was performed according to Mattsby-Baltzer et al. [16]. The lipid A concentration used for coating the tubes was 5 pg/ml. Anti-(rabbit IgG) serum (Nordic, Tilburg, Netherlands) was conjugated by diglutaraldehyde with alkaline phosphatase (Sigma Chemical Co., St Louis, MO). A commercially available anti-(rabbit 1gG)- alkaline-phosphatase conjugate was also used (AMF Immuno- Reagents, Inc., Seguin, TX). Different amounts of crude lipid A or purified fractions of lipid A were added to the antiserum that had been diluted 1 : and the mixtures were incubated for 15 min at 37 C. Thereafter the solutions were stored overnight at 4 C prior to analysis by ELISA. Antigenic activity was expressed as the amount needed for 50% inhibition of the antibody activity. TLC assay. Purified lipid A preparations were run on TLC plates. After drying, the plate was kept in a tank with 0.25 % bovine serum albumin in NaC1/Pi for 1 h. The plate was washed three times with NaC1/Pi containing 0.05 %Tween 20 (NaCl/P,- Tween) by keeping it in buffer solution for 5-10 min. The plate was placed in a humid chamber and antiserum, diluted with NaC1/Pi-Tween (1:100, v/v), was applied on the plate and allowed to remain for 5 h. After washing three times with NaC1/Pi-Tween, the anti-(rabbit IgG) conjugate (diluted with NaC1/Pi-Tween 1 : 100, v/v) was added in the same manner as described above. The plate was next placed in the humid chamber overnight at room temperature. The next day the plate was washed again three times with NaC1/Pi-Tween. In the meantime a gel slab was prepared by mixing equal amounts of 1 % agarose and the substrate solution (2 mg p-nitrophenylphosphate/ml of 1 M diethanolamine buffer, ph 9.8) and pouring it onto a glass plate of suitable size. It was important to use a horizontal table in order to obtain a homogenous thickness of gel. The gel was allowed to solidify and thereafter gently placed on the surface of the TLC plate after the plate had been blotted with paper along the edges. After min the plate was read by a KM3 thin-layer chromatogram spectrophotometer (Zeiss) using absorbance of reflected light at 405 nm. The gel was then removed and the plate was washed three times in 0.05 M glycine/hcl buffer, ph 2.6, followed by one wash in distilled water. After drying, the plate could be stained for phosphate. The specificity of the technique was tested by changing the anti-(lipid A) serum to NaCl/P, and the lipid A to cholesterol or phosphatidylethanolamine. No reaction above the background activity was seen after making either of the changes. One anti- (lipid A) serum was also absorbed with liposomes containing lipid A. Briefly, liposomes containing 25 nmol lipid A/pmol Amount of lipid A phosphate (nmol) Fig. 1. Comparison of the amount of lipid A phosphate added to the TLC plate. Different lipid A amounts were analyzed by the TLC-ELISA assay and plotted against the square of the area enclosed by the curve as obtained by measuring the absorbance at 405nm with a TLC scanning spectrophotometer. Each point represent the mean of three estimations phosphatidylcholine and prepared as described earlier [ 121 were added to the antiserum in a 1 : 1 (v/v) ratio and incubated at room temperature for 1 h. The liposomes were removed by centrifugation [21] and the adsorption was repeated once more. In order to compare the sensitivity of the TLC-ELISA with the sensitivity obtained by visualizing with carbohydrate staining, lipid A was assayed on a TLC plate in amounts ranging over nmol lipid A phosphate, corresponding to a mass range of approximately pg lipid A. Each sample was applied in 5 pl and the area of each was approximately the same. The plates were not developed in a solvent system, but were either stained directly for carbohydrate or assayed by the TLC-ELISA. The sensitivity of TLC-ELISA was approximately 10-times higher than the sensitivity obtained by carbohydrate staining, and at least 30-times higher than the sensitivity obtained by phosphate staining. The squares of the areas of the TLC-ELISA peaks obtained by scanning at 405 nm as a function of the amounts of lipid A phosphate applied to the TLC plate are shown in Fig. 1. When 3 nmol or more of lipid A were applied on the TLC plate, no further increases of color intensity were obtained. RESULTS Comparison of different lipid A preparations Chromatograms of lipid A preparations from Shigella jlexneri, Escherichia coli and Salmonella minnesota R595 were compred by TLC-ELISA, using anti-(lipid A) sera, and by phosphate staining (Fig. 2). In each case seven or eight major bands were found. Bands with R, values corresponding to the eight bands of Sh.flexnerifound earlier [I21 were designated by the same numbers previously applied to Sh. Jlexneri. Band 1, which is often present in both Sh. flexneri and S. minnesota R595, was absent, or present in a very low amount, in the lipid A used in this study. Band 1 was variable among lipid A preparations made from different batches of the same lipopolysaccharide type. The anti-(lipid A) serum gave an ELISA profile which was in accordance with the phosphate-stained lipid A preparations,

3 335 ((11 E. coli 0111 lb) 5. minnesoto R595 lcl 5h. flexneri t I Band r 3 L RF RF RF Fig. 2. ScanningdiagramsoflipidA (300nmol ofeach) frome.colioll1 (a), S.minnesota R.595 (b), andsh.flexneri (c). The various preparations were separated on a TLC plate using a solvent system of chloroform/methanol/water (130 :45: 7, v/v). The plate was assayed and read at 405 nm (ELISA) after a I-h reaction time with the substrate, using the TLC spectrophotometer. Thereafter the substrate-containing gel wds removed and the plate washed three times in 0.05 M glycine/hcl buffer, ph 2.6, in order to remove antibodies. The dried plate was stained with phosphate reagent and read at 700 nm. Anti-(lipid A) (-); phosphate stain (-----) U E 50 1 had the lowest activity of all the bands by ELISA assay. Therefore there was a marked discrepancy between the apparent amount of lipid A and the antigenic activity of band 8 of E. coli The lipid A of S. minnesota R595 also contained a band 9 and it is not clear which of the bands accounted for the antigenic activity (Fig.2b). Band 8 of Sh. JZexneri lipid A seemed to have a somewhat stronger antigenic activity than the corresponding band of E. coli (Fig. 2c). Using anti-(lipid A) serum adsorbed with liposomes containing lipid A, little or no activity of the separated lipid A from E. colio111 could be detected above background (Fig. 3 b). The scanning profile of lipid A, as observed with phosphate stain, differed somewhat from that in Fig. 2 because lipid A prepared from different batches of the same lipopolysaccharide type may vary slightly in the amount of each fraction. Changing the anti- (lipid A) serum to normal rabbit serum usually eliminated the TLC-ELISA bands observed with antiserum (not shown). However, in some cases a very low level of anti-(lipid A) activity (against fractions 5, 6 andjor 7) was also detected in normal rabbit serum. Naturally occurring antibodies against lipid A are often found both in normal rabbit and normal human sera [16, 21, 231. Fig. 3. Scanning diagram of lipid A (300 nmol ofphosphate) from E. coli 01 I I as assayed in TLC-ELISA. The lipid A preparation was separated on a TLC plate using the solvent system of chloroform/methanol/water (130:45:7, v/v). The anti-(lipid A) serum (-) was absorbed with liposomes lacking lipid A (a) or liposomes containing lipid A (b). The plates were thereafter washed three times in glycine/hcl buffer. After drying the plates were stained for phosphate (----) RF Antigenicity of free lipid A fractions in aqueous medium Fractions of lipid A from Sh. jlexneri were tested for the ability to inhibit the lipid A/anti-(lipid A) reaction by a tube ELISA (Table 1). Fractions 1, 2 and 3 showed an inhibitory activity approximately comparable to a control consisting of crude lipid A from E. coli EH100. Fractions 7 and 8 could not inhibit the reaction by 50 % at the concentrations used. i.e. six or seven peaks were seen on the diagram. Bands 7,8 and 9 appeared very close to each other on the TLC plate and the peaks of these bands therefore were not well-resolved by either anti-(lipid A) antibody activity or by phosphate stain (Fig. 2). Careful examination of Fig.2a and 3a also shows that although band 8 in E. cozi0111 was present in a high amount, it DISCUSSION In the present study a modified immunosorbent technique utilizing the TLC plate was developed in order to analyse a relatively insoluble and extremely complex antigen preparation. A substrate-containing gel was used for the visualization step, which allowed the plate to be read by a chromatogram

4 336 Table 1. InhibitionofthelipidAanti-(IipidA) reaction in tubeelisa by dij'ferent fractions of lipid A from Sh. flexneri as analysed with anti- (rabbit IgG) conjugate The amounts of the different fractions were calculated from 0.2 pmol lipid A phosphate, which is equivalent to approxiniately 1 mg of lipid A. 50 % inhibition was defined as the decrease of the absorbance at 405 nm to half the original absorbance obtained with uninhibited serum Fraction crude lipid A Amount needed for 50 % inhibition Pg > 150 > spectrophotometer. The procedure of using a substratecontaining gel in ELISA for detecting antibody activity is not new [24]. It has been used for detecting antibodies in a diffusion-in-gel ELISA technique. The use of TLC plates for identification of insoluble lipids by reaction with antibodies has also been reported earlier [15]. Magnani et al. [15] used monoclonal antibodies for detecting monosialoganglioside on a TLC plate but the visualization step included the use of radiolabeled antibodies. Our procedure excludes the use of radiolabeled material and has the additional advantage that the time for developing the plate can be decreased. We also found that it was possible to scrape off the silica gel containing the bands and measure the enzyme activity by the means of color intensity in tubes with silica gel plus substratebuffer added. One advantage of using a TLC spectrophotometer was that the antibodies could be removed by acidic buffer from the plate, and the bands could be stained for phosphate and scanned once more. This latter technique gave a more exact comparison between antibody-reactive bands and phosphatestainable bands, and this multistain capability may be useful for identification of certain complex antigen preparations. In this study extremely complex lipid A antigen preparations with up to eight major bands were analysed with a TLC-ELISA technique. The antigenicity of band 7 was quantitatively comparable with the other bands in the lipid A preparations from all the three bacterial strains. This observation differed from the results of tube ELISA assays where fractions 7 and 8 could not inhibit the assay to 50 % by the amounts tested. For future applications of the TLC-ELISA technique with lipid A (or with other glycolipids) it should be pointed out that rabbit serum and also human serum normally may contain anti-(lipid A) or other anti-glycolipid antibodies [16,22,23,25, 261. Such 'natural antibodies' may not react with all fractions to the same degree as antibodies obtained by immunization. In the present study, natural anti-(lipid A) antibodies were sometimes observed in normal rabbit serum. The antigenic determinant in the lipid A structure, which consists of a phosphorylated diglucosamine with amide-linked and ester-linked fatty acids, has been found not to be affected by removal of ester-linked fatty acids or phosphate groups [ One study reported that glucosamine with an amid- ated fatty acid (not necessarily 3-hydroxymyristic acid) was inhibitory in the lipid A-anti-(lipid A) hemolysis system [30]. Changing the amidated fatty acid to an esterified group completely abolished the inhibitory activity of the compound. In another study the amide-linked 3-hydroxymyristic acid was found to exert inhibitory activity when analysed in the lipid A/ anti-(lipid A) hemolysis system [29]. Although the smallest group in the lipid A structure remaining antigenic may vary, possibly depending on differences on the antisera used, the antigenic determinant(s) is (are) found in the lipid A backbone, as demonstrated by the cross-reactivity between different bacterial genera having similar chemical composition of lipid A [27, 311. In the present study it was shown by tube ELISA that, of the eight purified lipid A fractions, fraction 6 was approximately 25-times less effective in inhibiting the lipid A anti-(lipid A) ELISA compared to fractions 1,2 or 3. Fractions 7 and 8 could not inhibit the assay to 50% at the concentrations used (Table 1). This result is compatible with that previously observed with another anti-(lipid A) assay, liposome glucose release assay [12]. The apparently low inhibitory activity of high-numbered fractions may have been caused, at least partially, by low solubility of these fractions, even when triethylamine was used. It is also possible that the low reactivity of fraction 8 was caused by the fact that this fraction actually consists virtually entirely of phosphatidylethanolamine. This latter fact has been established both by TLC [14] and NMR (Baltzer, L. and Mattsby-Baltzer, I., unpublished results). However, it is likely that phosphatidylethanolamine can be antigenic [32] and it is possible that the anti-phosphatidylethanolamine titer in the antiserum was different from the anti- (lipid A) titer. In summary, the TLC-ELISA technique that we have utilized for studying qualitative differences between various fractions of lipid A has the advantage of decreasing or overcoming the problems of poor (or absent) solubility of lipids in aqueous solutions. It also has the advantages of rapidity, of not requiring isotopes and of not requiring purification of the material before analysis. As shown by the lipid A study, it is suitable for use with complex mixtures of antigens. One of the authors (I. M-B.) was supported by a research associate-ship from the National Research Council, USA. REFERENCES 1. Ng, A. K., Butler, C., Chen, C. L. & Nowotny, A. (1976) J. Bacteriol. 126, Hase, S. & Rietschel, E. Th. (1977) Eur. J. Biochem. 75, Lehmann, V. & Rupprecht, E. (1977) Eur. J. Biochem. 81, Muhlradt, P. F., Wray, V. & Lehmann, V. (1977) Eur. J. Biochem. 81, Rosner, M. R., Khorana, H. G. & Satterthwait, A. C. (1979) J. Bwl. Chern. 254, Drewry, D. T., Lomax, J. A., Gray, G. W. & Wilkinson, S. G. (1973) Biochem. J. 133, Kasai, N. &Yamano, A. (1964) Jap. J. Exp. Med. 34, Kasai, N. (1966) Ann. N. Y. Acad. Sci. 133, Chang, C. M. & Nowotny, A. (1975) Immunochemistry, 12, Schiller, J. G., Ribovich, R., Feingold, D. S. & Youngner, J. A. (1976) Infect. Immunol. 14, Takayama, K., Ribi, E. & Cantrell, J. L. (1981) Cancer Res. 41, Banerji, B. & Alving, C. R. (1979) J. Immunol. 123, Galanos, C., Luderitz, O., Rietschel, E. T. & Westphal, 0. (1977) Int. Rev. Biochem. 14,

5 Mattsby-Baltzer, I., Gemski, P. &Alving, C. R. (1982) Fed. Proc. 41, Magnani, J. L., Brockhaus, M., Smith, D. F., Ginsburg, V., Blaszczyk, M., Mitchell, K. F., Steplewski, Z. & Koprowski, H. (1981) Science (Wash. DC) 212, Mattsby-Baltzer, I. & Kaijser, B. (1979) Infect. Immunol. 23, Galanos, C., Liideritz, 0. & Westphal, 0. (1969) Eur. J. Biochem. 9, Galanos, C., Liideritz, 0. & Westphal, 0. (1971) Eur. J. Biochem 24, Svennerholm, L (1956) J. Neurochem. 1, Dittmer, J. C. & Lester, R. L. (1964) J. Lipid Res. 5, Alving, C. R., Schichijo, S. & Mattsby-Baltzer, I. (1983) in Liposome Technology (Gregoriadis, G., ed.) CRC Press, Boca Raton FL, in the press. 22. Schuster, B. G., Neidig, M., Alving, B. M. & Alving, C. R. (1979) J. Immunol. 122, Mattsby-Baltzer, I., Claesson, I., Hanson, L. A., Jodal, U., Kaijser, B., Lindberg, U. & Peterson, H. (1981) J. Infect. Dis. 144, Elwing, H., Lange, S. & Nygren, H. (1980) J. Immunol. Method, 39, Richards, R. & Alving, C. R. (2980) Am. Chem. SOC. Symp. Ser. 128, Mattsby-Baltzer, I. & Alving, C. R. (1984) Rev. Infect. Dis., in the press. 27. Galanos, C., Freudenberg, M., Hase, S., Jay, F. & Ruschmann, E. (1977) Microbiology (Wash. DC), Sidorzyk, Z., Rozalski, A., Deka, M. & Kotelko, K. (1978) Arch. Imrnunol. Ther. Exp. 26, Lugowski, C. & Romanovska, E. (1974) Eur. J. Biochem. 48, Gorbach, V. I., Krasikova, I. N., Lucyanov, P. A., Razmakhuina, 0. Y., Solov eva, T. F. & Ovodov, Y. S. (1979) Eur. J. Biochem. 98, Johns, M. A,, Bruins, S. C., & McCabe, W. R. (1977) Infect. Immunol. 17, Alving,C. R. (1977)in TheAntigens(Sela,M.,ed.)volIV,pp.48-51, Academic Press, New York. I. Mattsby-Baltzer, Institutionen for Medicinsk Mikrobiologi, Goteborgs Universitet, Fack, S Goteborg. Sweden C. R. Alving, Department of Biochemistry, Walter Reed Army Institute of Research, Washington, District of Columbia, USA 20307