- Email: [email protected]
Preparation and characterisation of quillaja saponin with less heterogeneity than Quil-A
Vaccine 18 (2000) 2244±2249
www.elsevier.com/locate/vaccine
Preparation and characterisation of quillaja saponin with less heterogeneity than Quil-A Sùren Kamstrup a,*, Ricardo San Martin b, Alfredo Doberti b, Hans Grande c, Kristian Dalsgaard a, 1 a Danish Veterinary Institute for Virus Research, Lindholm, DK-4771, Kalvehave, Denmark Catholic University, Department of Chemical Engineering, Av. VicunÄa Mackenna 4860, Santiago, Chile c Licentech, Bernadottelaan 15, P.O. Box 8323, 3503 RH, Utrecht, The Netherlands
b
Received 4 January 1999; received in revised form 15 November 1999; accepted 2 December 1999
Abstract Immunisation against pathogens remains one of the most eective ways of preventing or reducing losses due to infectious diseases in animal husbandry. When inactivated vaccines are used, adjuvants are most often required to obtain satisfactory immune responses. One such type of adjuvant is saponin derived from the bark of Quillaja saponaria Molina, a tree of the rose family. A few dierent commercial sources exist, but due to the structural complexity and heterogeneity of these saponin preparations, it has been dicult to establish exactly which components are responsible for the adjuvant activity. By carefully selecting the bark source, we have succeeded in preparing a much less heterogeneous preparation of quillaja saponin. In this report we describe the preparation, in terms of structural complexity, hemolytic activity, adjuvant activity, and its ability to form ISCOM matrix. This new preparation could have implications for use per se, or as starting material for more eective preparation of pure substances. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Quil A; Adjuvant; Saponin; ISCOM
1. Introduction Since the original observation of Espinet [1], that certain saponins act as immunological enhancers Ð adjuvants Ð when given together with an antigen in a vaccine, considerable eort has been put into the identi®cation of the active components responsible for this eect. One of the major steps forward in this respect was the identi®cation of a puri®ed fraction of saponin from Quillaja saponaria Molina, which possessed the full adjuvant activity of the crude quillaja * Corresponding author. Tel: +45-55-86-02-00; fax: +45-55-86-0300. E-mail address: [email protected] (S. Kamstrup). 1 Present address: Vaccinology Group, University of Copenhagen, IMMI Panum 24.2.40, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark.
saponin [2]. Still, the substance denoted Quil A is a mixture of several dierent components, which can be clearly seen by e.g. reversed-phase HPLC separation of Quil A [3]. Quil A is the basis of the ISCOM technology [4], a very potent adjuvant with several unique features, e.g. the ability to stimulate both the humoral and cellular branch of the immune system (for a review, see [5]). The starting material for the production of Quil A is the bark from Quillaja saponaria Molina, a tree of the rose family which is indigenous to South America, most notably Chile. Bark from large areas is collected typically during spring and early summer, dried and stored, and then shipped for further processing. We have investigated, whether it would be possible to obtain a less heterogeneous product by: (1) selective use of bark from dierent locations or trees; (2) using freshly harvested bark (i.e. without the long-term sto-
0264-410X/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 4 - 4 1 0 X ( 9 9 ) 0 0 5 6 0 - 5
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
rage which may lead to oxidative or other changes in the product composition) as starting material for extraction of saponins. The resulting preparations of saponin have then been tested for compositional heterogeneity by reversed-phase HPLC. In this way we have identi®ed a less heterogeneous product, designated Quadri A, which has been tested for hemolytic activity, ISCOM matrix forming capacity and for adjuvant activity in mice. 2. Materials and methods 2.1. Selection and harvesting of bark Samples of quillaja bark were collected in the foothills of the Andes range (altitude 800±1000 m), close to Santiago. 30 natural (not planted) trees were examined. Each sample weighed about 50 g. The external part of the bark (cork), was peeled o. Each sample was air dried to about 10% moisture content and stored at room temperature before analysis. 2.2. Extraction of saponins For extraction of saponin, the bark was cut to squares of about 0.5 cm. 10 g of bark were extracted with 30 g of distilled water for 2 h at room temperature. The extract was collected, ®ltered through 0.2 micron ®lters, and kept frozen at ÿ188C, before analysis. All samples were subjected to a preliminary screening to determine its saponin composition by reversedphase HPLC. 2.3. Puri®cation of samples Based on the HPLC analysis, the most promising samples were puri®ed at a larger scale. For this, 200 g of bark were placed in a 4 l glass extractor, and contacted with 2 l of distilled water. The water was recirculated with a peristaltic pump at 150 ml/min. The extraction time was 1 h. The extract was removed, and fresh water was added to the system (same volume as that of extract), and the extraction was continued for 1 h. Both extracts were combined and ®ltered through glass wool. Small molecules, e.g. salts and sugars, were removed by cross ¯ow ultra®ltration (Minitan system, Millipore Corp., USA) and dia®ltration with distilled water. Polysulfone membranes with a nominal cut-o of 10,000 Dalton were employed. To minimise the amount of water used for dia®ltration, and hence reduce ®ltration time and membrane area, the extract was concentrated to about 1/3 of the initial volume. Dia®ltration was then started and the conductivity of the ®ltrate monitored with a conduc-
2245
tivity meter (model 1481-55, Cole Parmer, USA). The operation was stopped when the conductivity of the permeate was of the same range as that of distilled water. The puri®ed extract was then subjected to ionexchange following the procedure described by Dalsgaard [2]. 9 vols of dia®ltered extract were mixed with 1 vol of 1 M Tris±HCl buer, pH 7.5, and then loaded onto a column of DE-52 anionic cellulose (Whatman, USA). The column was 19.5 cm long with a diameter of 3.6 cm; 110 ml of packed bed was used in each experiment (50 g dry DE-52 cellulose). After loading, washing was performed with 0.1 M Tris±HCl buer until the absorbance measured at 280 nm in a Shimadzu spectrophotometer (model UV-160) was below 0.09. Then, the product was eluted with 0.2 M NaCl and collected in an automated sample collector. The collected sample was desalted using dia®ltration with a procedure that was identical to that used prior to ionexchange. The ®nal extracts were lyophilised in 1 l ¯asks, in a 4.5 l lyophiliser (Labconco, USA). 2.4. Analysis by HPLC Samples were analysed by reversed-phase chromatography on a Vydac C-4 column (250 4.6 mm column), essentially as desribed by [3], using a gradient of 30±45% acetonitril in water, with 0.15% tri¯uoracetic acid. Detection was done by monitoring absorbance at 210 nm. 2.5. Hemolytic activity and ISCOM matrix formation A stock solution of lipids containing 2% cholesterol and 2% phosphatidylcholine dissolved in 20% Mega10 was prepared by gently heating until dissolved. Saponin (®nal concentration 4 mg/ml) was mixed with various amounts of lipids to monitor the decrease in hemolytic activity resulting from complex formation between saponin and lipids. After 2 h stirring at RT, the samples were dialysed against water or phosphatebuered saline (PBS) at 58C overnight, and analysed for hemolytic activity and for ISCOM matrix formation by transmission electron microscopy using uranyl acetate as negative contrast stain. Hemolytic activity was measured by a radial diusion hemolysis assay in agarose. Brie¯y, washed and packed sheep erythrocytes were diluted in molten 1% agarose in 60 mM barbiturate buer containing 42 g/ml glucose, pH 8.8, and poured onto a 10 10 cm glass plate. After solidi®cation, holes were punched and samples were applied. The plates were stored horizontally at RT in a humidifying chamber and read after 24 h. Hemolysis was measured as the area of the clear zone where erythrocytes had lysed minus the area of the sample application well.
2246
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
2.6. Adjuvant activity of extracted saponins To test the potency of the extracted saponin as adjuvant, groups of 10 mice were immunised subcutaneously twice, 14 days apart with 25 mg of ovalbumin (Sigma, St Louis, USA), supplemented with dierent amounts of saponin, either our extracts or commercially available Quil-A (Superfos Biosector, Frederikssund, Denmark), in a total volume of 200 ml. A dose±response relationship was determined by serially diluting the amount of saponins, using amounts of 10, 5, 2, and 1 mg of saponin. A blank (antigen only) sample was also included. 14 days after the second immunisation, serum was obtained from all mice. Sera
Fig. 1. HPLC pro®les of dierent saponin preparations. Pro®les were obtained by reversed-phase chromatography as described in materials and methods. Fig. 1A shows an example of saponin from fresh bark with complex-type pro®le (pro®le 1), with predominantly doublet peaks. Fig. 1B shows the pro®le of commercial available Quil-A, and Fig. 1C shows pro®le 2 of saponin from fresh bark, with the two predominant components eluting essentially as single peaks, i.e. the saponin preparation denoted Quadri A.
from each group were pooled and analysed for antibody concentration using a standard ELISA protocol with ovalbumin coated onto 96-well microtiter plates in 50 mM carbonate buer, pH 9.6, reacting with serial dilutions of test serum and developing with peroxidase conjugated rabbit-anti-mouse immunoglobulins (DAKO, Roskilde, Denmark) and orto-phenylene diamine and hydrogen peroxide as chromogenic system. Titer is expressed as the dilution of serum giving an OD492 of 0.250. 3. Results A preliminary screening of bark from 30 trees in one site yielded two main HPLC pro®les: pro®le 1 consisted of a heterogeneous mixture of saponins, with the two major components appearing as double peaks (Fig. 1A). These samples have a similar saponin HPLC pro®le as Quil-A (Fig. 1B), in that the majority of components elute between 35±40% acetonitrile. Pro®le 2 also resembles Quil-A, but has 2 predominant peaks almost free of related saponins (Fig. 1C). It should be noted that dierent batches of the commercial product Quil-A also exhibits some variation with respect to relative proportions of the dierent components in the chromatographic separation (data not shown). Almost 50% of the trees analysed exhibited pro®le 1, while 50% exhibited pro®le 2. Apparently, there is no relation between the age of the tree, or freshness of the sample with the observed HPLC patterns. The same tree sampled in dierent parts (e.g. trunk, branches) showed the same HPLC pro®le. We
Fig. 2. Hemolytic activity of Quadri A and Quil-A. Saponin was mixed with dierent amounts of lipids (cholesterol and phosphatidylcholine), dialysed, and residual hemolytic activity was measured by radial diusion hemolysis in agarose gel. The hemolytic activity was measured as the area of the zone where erythrocytes had lysed. Quadri A (*) has an initially lower hemolytic activity than does Quil-A (Q), and shows a similar reduction pattern when complexed with increasing amounts of lipids. The lowest level of hemolysis occurs with a 4:1 ratio of saponin:lipid for both Quadri A and QuilA.
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
denote this product (with pro®le 2) Quadri A. The physical characteristics of Quadri A are identical to Quil-A, i.e. very soluble, foam forming, and micelle forming. The dry substance is a white to yellowish, glassy powder. When compared to Quil-A, the hemolytic activity of Quadri A is found to be slightly lower, and showing the same decrease after addition of lipids (Fig. 2), suggesting interaction between saponin and the lipids. Electron microscopy revealed that these complexes were ISCOM matrix (Fig. 3), which is the complex formed between saponin, cholesterol and phospholipid, but without antigen [6]. Ultracentrifugation in sucrose gradients [7,8] showed that this ISCOM matrix sedimented as a narrow band with sedimentation properties identical to ISCOM matrix prepared from Quil-A (data not shown). The quality of ISCOM matrix obtained depended on the correct ratio of lipid:saponin used, for both Quadri A and Quil-A we found an optimal ratio of 1:4 on a weight basis, determined by: (1) minimal hemolytic activity; (2) most homogeneous and complete ISCOM matrix particles (TEM); and (3) highest and most well-de®ned sedimentation coecient
2247
by sucrose gradient ultracentrifugation (data not shown). The activity of Quadri A as an adjuvant was studied in mice with ovalbumin as a model antigen. The antibody responses of mice showed similar eect of Quadri A as compared to Quil-A (Fig. 4). There is a clear dose±response relationship showing a slight increase in titer at 5 mg, and higher (approx. 10-fold) increase at 10 mg saponin. 4. Discussion This study has enabled us to identify trees which show a much more restricted composition of saponin than observed in commercially available Quil-A. The composition of saponin in these trees is consistent in dierent tissues sampled. Other trees from the same location showed more complex pro®les. These observations suggest that the dierences observed between individual trees is due to genetic factors, as neither soil, altitude or age of the trees or sampled tissue correlates with composition of saponins.
Fig. 3. Quadri A forms ISCOM-matrix when complexed with cholesterol and phosphatidylcholine. Complexes formed between Quadri A and lipids with minimal hemolysis (see Fig. 2) were examined by electron microscopy using conventional techniques. Characteristic cage-like complexes are abundant, showing clearly that Quadri A is capable of forming ISCOM matrix structures. Bar=100 nm.
2248
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
The commercial product used for comparison throughout this study is Quil-A (Quil-hyphen-A), which is a registered trademark of Superfos, Frederikssund, Denmark. This product is produced on the basis of the original puri®cation protocol by Dalsgaard [2]. All our evidence suggests, that the product denoted Quadri A posesses similar properties as Quil-A, both in terms of physical/chemical characteristics as well as biological eects. However, the HPLC pro®les shown in Fig. 1 show a markedly dierent composition, with pro®le two showing only two predominant components at approx. 26 and 29 min (Fig. 1C). Its seems likely, that the commercial product Quil-A has its heterogeneous composition due to mixing of large quantities of bark with the two dierent pro®les shown in Fig. 1A and C. Also, the longer storage time from harvest to processing for Quil-A may contribute to the heterogeneity observed with this product, if components are subject to e.g. oxidation or intramolecular rearrangements. Determination of such eects will need further studies, where the availability of Quadri A could be useful as a starting material due to its low heterogeneity at the time of harvest. The immunisation experiment in mice shows no dierence between the adjuvant activity of Quadri A and Quil-A, neither in eciency nor in dose±response relationship. Taken together, this further documents previous observations that the heterogeneity of Quil-A is not necessary for the adjuvant eect of Quil-A [3], and also not for the capacity to form ISCOM matrix [9]. We also showed similar kinetics of hemolysis inhibition by lipids (Fig. 2) and ISCOM matrix forming capacity, supporting that several features of Quil-A are recurring in Quadri A. It is possible, that all or some
Fig. 4. Adjuvant activity of Quadri A and Quil-A. Groups of 10 mice were immunised at day 0 and 14 with 25 mg of ovalbumin (OVA) and varying amounts of either Quadri A (*) or Quil-A (Q). Serum pools of all groups were obtained at day 28, and the amount of anti-OVA antibodies determined by ELISA. The two saponins show almost identical dose±response pro®les, indicating similar adjuvant activity of the two preparations.
of these properties of Quadri A may be ascribed to one or both of the two major components observed in Fig. 1C. Considering the potential use of Quadri A as an adjuvant, the lower hemolytic eect of Quadri A would be expected to have a bene®cial eect by decreasing toxicity. Initial experiments in mice have shown lower mortality after giving high doses (s.c.) of Quadri A, as compared to Quil-A. However, the decrease in hemolytic activity by complexing with lipids (ISCOM or ISCOM matrix formation) does not give a proportionate decrease in toxicity, which indicates that the hemolytic eect of the saponin is not solely responsible for its toxicity (data not shown). By its much more restricted composition, Quadri A could have dierent advantages over Quil-A. It may be more easily standardised per se, but from a regulatory point of view, would probably still be considered a biological. A main advantage of Quadri A is that it would be a suitable raw material for preparation of pure saponin components. The presence of fewer components will result in higher speci®c yields. During recent years, dierent pure components have been obtained from Quil A, enabling structural studies. The most well-described of these is denoted QS-21 [3], which has shown good promise for inclusion into both viral, bacterial, parasite, and anti-cancer vaccines [10± 16]. When the two main components of our preparation were puri®ed and used to spike saponin similar to the starting preparation of Kensil et al. [3], they were shown to coelute with the components corresponding to QS18 and QS21, respectively (data not shown). This suggests, but does not prove that they are the same substances, since dierences in e.g. monosaccharide composition in side chains may not resolve in dierent peaks during reversed-phase HPLC. We are currently gathering further evidence, that the second peak in our chromatogram is very similar, if not identical to QS21. These results will be published separately (manuscript in preparation). In conclusion, our studies have shown, that it is possible to obtain, by carefully selecting the source material, saponin of very de®ned composition, still possessing important properties of Quil-A such as adjuvant activity and ISCOM formation. Quadri A could thus be used both for the production of Quil-A, and for use as raw material for puri®cation of pure substances like e.g. QS-21. Acknowledgements This work was supported by the European communities contracts ``BIOT-CT91-0256'', ``BIO4-CT980096'', and ``BIO4-CT98-0505''.
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
References [1] Espinet EG. Nuevo tipo de vacuna antiaftosa a complejo glucovirico. Gac Vet 1951;74:1±13. [2] Dalsgaard K. Saponin adjuvants. 3. Isolation of a substance from Quillaja saponaria Molina with adjuvant activity in foodand-mouth disease vaccines. Archiv fuÈr die Gesamte Virusforschung 1974;44:243±54. [3] Kensil CR, Patel U, Lennick M, Marciani D. Separation and characterization of saponins with adjuvant activity from Quillaja saponaria Molina cortex. Journal of Immunology 1991;146:431±7. [4] Morein B, Sundquist B, HoÈglund S, Dalsgaard K, Osterhaus A. Iscom, a novel structure for antigenic presentation of membrane proteins from enveloped viruses. Nature 1984;308:457±60. [5] Sjùlander A, Cox JC, Barr G. ISCOMs: an adjuvant with multiple functions. Journal of Leucocyte Biology 1998;64:713±23. [6] LoÈvgren K, Morein B. The requirement for lipids for the formation of immunostimulating complexes (iscoms). Biotechnol Appl Biochem 1988;10:161±72. [7] Baxter-Gabbard KL. A simple method for the large-scale preparation of sucrose gradients. FEBS Letters 1972;20:117±9. [8] Beekman N, Kamstrup S, Dalsgaard K. Method for monitoring ultracentrifuge gradients using a high-performance liquid chromatographic diode array detector. Analytical Biochemistry 1995;228:168±9. [9] RoÈnnberg B, Fekadu M, Morein B. Adjuvant activity of nontoxic Quillaja saponaria Molina components for use in ISCOM matrix. Vaccine 1995;13:1375±82. [10] Hancock GE, Speelman DJ, Frenchick PJ, Mineo-Kuhn MM, Baggs RB, Hahn DJ. Formulation of the puri®ed fusion protein of respiratory syncytial virus with the saponin QS-21 induces protective immune responses in Balb/c mice that are similar to
[11]
[12]
[13]
[14]
[15]
[16]
2249
those generated by experimental infection. Vaccine 1995;13:391± 400. Coughlin RT, Fattom A, Chu C, White AC, Winston S. Adjuvant activity of QS-21 for experimental E. coli 018 polysaccharide vaccines. Vaccine 1995;13:17±21. Cleland JL, Barron L, Daugherty A, Eastman D, Kensil C, Lim A, Weissburg RP, Wrin T, Vennari J, Powell MF. Development of a single-shot vaccine for HIV-1. 3. Eect of adjuvant and immunization schedule on the duration of the humoral immune response to recombinant MN gp120. Journal of Pharmaceutical Sciences 1996;85:1350±7. Helling F, Zhang S, Shang A, Adluri S, Calves M, Koganty R, Longenecker BM, Yao TJ, Oettgen HF, Livingston PO. G(M2)KLH conjugate vaccine: Increased immunogenicity in melanoma patients after administration with immunological adjuvant QS-21. Cancer Research 1995;55:2783±8. Ma JN, Bulger PA, Davis DV, Perillipalmer B, Bedore DA, Kensil CR, Young EM, Hung CH, Seals JR, Pavia CS, Coughlin RT. Impact of the saponin adjuvant QS-21 and aluminium hydroxide on the immunogenicity of recombinant OspA and OspB of Borrelia burgdorferi. Vaccine 1994;12:925±32. Livingston PO, Adluri S, Helling F, Yao TJ, Kensil CR, Newman MJ, Marciani D. Phase 1 trial of immunological adjuvant QS-21 with a GM2 ganglioside-keyhole limpet haemocyanin conjugate vaccine in patients with malignant melanoma. Vaccine 1994;12:1275±80. Newman MJ, Munroe KJ, Anderson CA, Murphy CI, Panicali DL, Seals JR, Wu JY, Wyand MS, Kensil CR. Induction of antigen-speci®c killer T lymphocyte responses using subunit SIVmac251 gag and env vaccines containing QS-21 saponin adjuvant. AIDS Research and Human Retroviruses 1994;10:853±61.
www.elsevier.com/locate/vaccine
Preparation and characterisation of quillaja saponin with less heterogeneity than Quil-A Sùren Kamstrup a,*, Ricardo San Martin b, Alfredo Doberti b, Hans Grande c, Kristian Dalsgaard a, 1 a Danish Veterinary Institute for Virus Research, Lindholm, DK-4771, Kalvehave, Denmark Catholic University, Department of Chemical Engineering, Av. VicunÄa Mackenna 4860, Santiago, Chile c Licentech, Bernadottelaan 15, P.O. Box 8323, 3503 RH, Utrecht, The Netherlands
b
Received 4 January 1999; received in revised form 15 November 1999; accepted 2 December 1999
Abstract Immunisation against pathogens remains one of the most eective ways of preventing or reducing losses due to infectious diseases in animal husbandry. When inactivated vaccines are used, adjuvants are most often required to obtain satisfactory immune responses. One such type of adjuvant is saponin derived from the bark of Quillaja saponaria Molina, a tree of the rose family. A few dierent commercial sources exist, but due to the structural complexity and heterogeneity of these saponin preparations, it has been dicult to establish exactly which components are responsible for the adjuvant activity. By carefully selecting the bark source, we have succeeded in preparing a much less heterogeneous preparation of quillaja saponin. In this report we describe the preparation, in terms of structural complexity, hemolytic activity, adjuvant activity, and its ability to form ISCOM matrix. This new preparation could have implications for use per se, or as starting material for more eective preparation of pure substances. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Quil A; Adjuvant; Saponin; ISCOM
1. Introduction Since the original observation of Espinet [1], that certain saponins act as immunological enhancers Ð adjuvants Ð when given together with an antigen in a vaccine, considerable eort has been put into the identi®cation of the active components responsible for this eect. One of the major steps forward in this respect was the identi®cation of a puri®ed fraction of saponin from Quillaja saponaria Molina, which possessed the full adjuvant activity of the crude quillaja * Corresponding author. Tel: +45-55-86-02-00; fax: +45-55-86-0300. E-mail address: [email protected] (S. Kamstrup). 1 Present address: Vaccinology Group, University of Copenhagen, IMMI Panum 24.2.40, Blegdamsvej 3, DK-2200 Copenhagen N, Denmark.
saponin [2]. Still, the substance denoted Quil A is a mixture of several dierent components, which can be clearly seen by e.g. reversed-phase HPLC separation of Quil A [3]. Quil A is the basis of the ISCOM technology [4], a very potent adjuvant with several unique features, e.g. the ability to stimulate both the humoral and cellular branch of the immune system (for a review, see [5]). The starting material for the production of Quil A is the bark from Quillaja saponaria Molina, a tree of the rose family which is indigenous to South America, most notably Chile. Bark from large areas is collected typically during spring and early summer, dried and stored, and then shipped for further processing. We have investigated, whether it would be possible to obtain a less heterogeneous product by: (1) selective use of bark from dierent locations or trees; (2) using freshly harvested bark (i.e. without the long-term sto-
0264-410X/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 4 - 4 1 0 X ( 9 9 ) 0 0 5 6 0 - 5
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
rage which may lead to oxidative or other changes in the product composition) as starting material for extraction of saponins. The resulting preparations of saponin have then been tested for compositional heterogeneity by reversed-phase HPLC. In this way we have identi®ed a less heterogeneous product, designated Quadri A, which has been tested for hemolytic activity, ISCOM matrix forming capacity and for adjuvant activity in mice. 2. Materials and methods 2.1. Selection and harvesting of bark Samples of quillaja bark were collected in the foothills of the Andes range (altitude 800±1000 m), close to Santiago. 30 natural (not planted) trees were examined. Each sample weighed about 50 g. The external part of the bark (cork), was peeled o. Each sample was air dried to about 10% moisture content and stored at room temperature before analysis. 2.2. Extraction of saponins For extraction of saponin, the bark was cut to squares of about 0.5 cm. 10 g of bark were extracted with 30 g of distilled water for 2 h at room temperature. The extract was collected, ®ltered through 0.2 micron ®lters, and kept frozen at ÿ188C, before analysis. All samples were subjected to a preliminary screening to determine its saponin composition by reversedphase HPLC. 2.3. Puri®cation of samples Based on the HPLC analysis, the most promising samples were puri®ed at a larger scale. For this, 200 g of bark were placed in a 4 l glass extractor, and contacted with 2 l of distilled water. The water was recirculated with a peristaltic pump at 150 ml/min. The extraction time was 1 h. The extract was removed, and fresh water was added to the system (same volume as that of extract), and the extraction was continued for 1 h. Both extracts were combined and ®ltered through glass wool. Small molecules, e.g. salts and sugars, were removed by cross ¯ow ultra®ltration (Minitan system, Millipore Corp., USA) and dia®ltration with distilled water. Polysulfone membranes with a nominal cut-o of 10,000 Dalton were employed. To minimise the amount of water used for dia®ltration, and hence reduce ®ltration time and membrane area, the extract was concentrated to about 1/3 of the initial volume. Dia®ltration was then started and the conductivity of the ®ltrate monitored with a conduc-
2245
tivity meter (model 1481-55, Cole Parmer, USA). The operation was stopped when the conductivity of the permeate was of the same range as that of distilled water. The puri®ed extract was then subjected to ionexchange following the procedure described by Dalsgaard [2]. 9 vols of dia®ltered extract were mixed with 1 vol of 1 M Tris±HCl buer, pH 7.5, and then loaded onto a column of DE-52 anionic cellulose (Whatman, USA). The column was 19.5 cm long with a diameter of 3.6 cm; 110 ml of packed bed was used in each experiment (50 g dry DE-52 cellulose). After loading, washing was performed with 0.1 M Tris±HCl buer until the absorbance measured at 280 nm in a Shimadzu spectrophotometer (model UV-160) was below 0.09. Then, the product was eluted with 0.2 M NaCl and collected in an automated sample collector. The collected sample was desalted using dia®ltration with a procedure that was identical to that used prior to ionexchange. The ®nal extracts were lyophilised in 1 l ¯asks, in a 4.5 l lyophiliser (Labconco, USA). 2.4. Analysis by HPLC Samples were analysed by reversed-phase chromatography on a Vydac C-4 column (250 4.6 mm column), essentially as desribed by [3], using a gradient of 30±45% acetonitril in water, with 0.15% tri¯uoracetic acid. Detection was done by monitoring absorbance at 210 nm. 2.5. Hemolytic activity and ISCOM matrix formation A stock solution of lipids containing 2% cholesterol and 2% phosphatidylcholine dissolved in 20% Mega10 was prepared by gently heating until dissolved. Saponin (®nal concentration 4 mg/ml) was mixed with various amounts of lipids to monitor the decrease in hemolytic activity resulting from complex formation between saponin and lipids. After 2 h stirring at RT, the samples were dialysed against water or phosphatebuered saline (PBS) at 58C overnight, and analysed for hemolytic activity and for ISCOM matrix formation by transmission electron microscopy using uranyl acetate as negative contrast stain. Hemolytic activity was measured by a radial diusion hemolysis assay in agarose. Brie¯y, washed and packed sheep erythrocytes were diluted in molten 1% agarose in 60 mM barbiturate buer containing 42 g/ml glucose, pH 8.8, and poured onto a 10 10 cm glass plate. After solidi®cation, holes were punched and samples were applied. The plates were stored horizontally at RT in a humidifying chamber and read after 24 h. Hemolysis was measured as the area of the clear zone where erythrocytes had lysed minus the area of the sample application well.
2246
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
2.6. Adjuvant activity of extracted saponins To test the potency of the extracted saponin as adjuvant, groups of 10 mice were immunised subcutaneously twice, 14 days apart with 25 mg of ovalbumin (Sigma, St Louis, USA), supplemented with dierent amounts of saponin, either our extracts or commercially available Quil-A (Superfos Biosector, Frederikssund, Denmark), in a total volume of 200 ml. A dose±response relationship was determined by serially diluting the amount of saponins, using amounts of 10, 5, 2, and 1 mg of saponin. A blank (antigen only) sample was also included. 14 days after the second immunisation, serum was obtained from all mice. Sera
Fig. 1. HPLC pro®les of dierent saponin preparations. Pro®les were obtained by reversed-phase chromatography as described in materials and methods. Fig. 1A shows an example of saponin from fresh bark with complex-type pro®le (pro®le 1), with predominantly doublet peaks. Fig. 1B shows the pro®le of commercial available Quil-A, and Fig. 1C shows pro®le 2 of saponin from fresh bark, with the two predominant components eluting essentially as single peaks, i.e. the saponin preparation denoted Quadri A.
from each group were pooled and analysed for antibody concentration using a standard ELISA protocol with ovalbumin coated onto 96-well microtiter plates in 50 mM carbonate buer, pH 9.6, reacting with serial dilutions of test serum and developing with peroxidase conjugated rabbit-anti-mouse immunoglobulins (DAKO, Roskilde, Denmark) and orto-phenylene diamine and hydrogen peroxide as chromogenic system. Titer is expressed as the dilution of serum giving an OD492 of 0.250. 3. Results A preliminary screening of bark from 30 trees in one site yielded two main HPLC pro®les: pro®le 1 consisted of a heterogeneous mixture of saponins, with the two major components appearing as double peaks (Fig. 1A). These samples have a similar saponin HPLC pro®le as Quil-A (Fig. 1B), in that the majority of components elute between 35±40% acetonitrile. Pro®le 2 also resembles Quil-A, but has 2 predominant peaks almost free of related saponins (Fig. 1C). It should be noted that dierent batches of the commercial product Quil-A also exhibits some variation with respect to relative proportions of the dierent components in the chromatographic separation (data not shown). Almost 50% of the trees analysed exhibited pro®le 1, while 50% exhibited pro®le 2. Apparently, there is no relation between the age of the tree, or freshness of the sample with the observed HPLC patterns. The same tree sampled in dierent parts (e.g. trunk, branches) showed the same HPLC pro®le. We
Fig. 2. Hemolytic activity of Quadri A and Quil-A. Saponin was mixed with dierent amounts of lipids (cholesterol and phosphatidylcholine), dialysed, and residual hemolytic activity was measured by radial diusion hemolysis in agarose gel. The hemolytic activity was measured as the area of the zone where erythrocytes had lysed. Quadri A (*) has an initially lower hemolytic activity than does Quil-A (Q), and shows a similar reduction pattern when complexed with increasing amounts of lipids. The lowest level of hemolysis occurs with a 4:1 ratio of saponin:lipid for both Quadri A and QuilA.
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
denote this product (with pro®le 2) Quadri A. The physical characteristics of Quadri A are identical to Quil-A, i.e. very soluble, foam forming, and micelle forming. The dry substance is a white to yellowish, glassy powder. When compared to Quil-A, the hemolytic activity of Quadri A is found to be slightly lower, and showing the same decrease after addition of lipids (Fig. 2), suggesting interaction between saponin and the lipids. Electron microscopy revealed that these complexes were ISCOM matrix (Fig. 3), which is the complex formed between saponin, cholesterol and phospholipid, but without antigen [6]. Ultracentrifugation in sucrose gradients [7,8] showed that this ISCOM matrix sedimented as a narrow band with sedimentation properties identical to ISCOM matrix prepared from Quil-A (data not shown). The quality of ISCOM matrix obtained depended on the correct ratio of lipid:saponin used, for both Quadri A and Quil-A we found an optimal ratio of 1:4 on a weight basis, determined by: (1) minimal hemolytic activity; (2) most homogeneous and complete ISCOM matrix particles (TEM); and (3) highest and most well-de®ned sedimentation coecient
2247
by sucrose gradient ultracentrifugation (data not shown). The activity of Quadri A as an adjuvant was studied in mice with ovalbumin as a model antigen. The antibody responses of mice showed similar eect of Quadri A as compared to Quil-A (Fig. 4). There is a clear dose±response relationship showing a slight increase in titer at 5 mg, and higher (approx. 10-fold) increase at 10 mg saponin. 4. Discussion This study has enabled us to identify trees which show a much more restricted composition of saponin than observed in commercially available Quil-A. The composition of saponin in these trees is consistent in dierent tissues sampled. Other trees from the same location showed more complex pro®les. These observations suggest that the dierences observed between individual trees is due to genetic factors, as neither soil, altitude or age of the trees or sampled tissue correlates with composition of saponins.
Fig. 3. Quadri A forms ISCOM-matrix when complexed with cholesterol and phosphatidylcholine. Complexes formed between Quadri A and lipids with minimal hemolysis (see Fig. 2) were examined by electron microscopy using conventional techniques. Characteristic cage-like complexes are abundant, showing clearly that Quadri A is capable of forming ISCOM matrix structures. Bar=100 nm.
2248
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
The commercial product used for comparison throughout this study is Quil-A (Quil-hyphen-A), which is a registered trademark of Superfos, Frederikssund, Denmark. This product is produced on the basis of the original puri®cation protocol by Dalsgaard [2]. All our evidence suggests, that the product denoted Quadri A posesses similar properties as Quil-A, both in terms of physical/chemical characteristics as well as biological eects. However, the HPLC pro®les shown in Fig. 1 show a markedly dierent composition, with pro®le two showing only two predominant components at approx. 26 and 29 min (Fig. 1C). Its seems likely, that the commercial product Quil-A has its heterogeneous composition due to mixing of large quantities of bark with the two dierent pro®les shown in Fig. 1A and C. Also, the longer storage time from harvest to processing for Quil-A may contribute to the heterogeneity observed with this product, if components are subject to e.g. oxidation or intramolecular rearrangements. Determination of such eects will need further studies, where the availability of Quadri A could be useful as a starting material due to its low heterogeneity at the time of harvest. The immunisation experiment in mice shows no dierence between the adjuvant activity of Quadri A and Quil-A, neither in eciency nor in dose±response relationship. Taken together, this further documents previous observations that the heterogeneity of Quil-A is not necessary for the adjuvant eect of Quil-A [3], and also not for the capacity to form ISCOM matrix [9]. We also showed similar kinetics of hemolysis inhibition by lipids (Fig. 2) and ISCOM matrix forming capacity, supporting that several features of Quil-A are recurring in Quadri A. It is possible, that all or some
Fig. 4. Adjuvant activity of Quadri A and Quil-A. Groups of 10 mice were immunised at day 0 and 14 with 25 mg of ovalbumin (OVA) and varying amounts of either Quadri A (*) or Quil-A (Q). Serum pools of all groups were obtained at day 28, and the amount of anti-OVA antibodies determined by ELISA. The two saponins show almost identical dose±response pro®les, indicating similar adjuvant activity of the two preparations.
of these properties of Quadri A may be ascribed to one or both of the two major components observed in Fig. 1C. Considering the potential use of Quadri A as an adjuvant, the lower hemolytic eect of Quadri A would be expected to have a bene®cial eect by decreasing toxicity. Initial experiments in mice have shown lower mortality after giving high doses (s.c.) of Quadri A, as compared to Quil-A. However, the decrease in hemolytic activity by complexing with lipids (ISCOM or ISCOM matrix formation) does not give a proportionate decrease in toxicity, which indicates that the hemolytic eect of the saponin is not solely responsible for its toxicity (data not shown). By its much more restricted composition, Quadri A could have dierent advantages over Quil-A. It may be more easily standardised per se, but from a regulatory point of view, would probably still be considered a biological. A main advantage of Quadri A is that it would be a suitable raw material for preparation of pure saponin components. The presence of fewer components will result in higher speci®c yields. During recent years, dierent pure components have been obtained from Quil A, enabling structural studies. The most well-described of these is denoted QS-21 [3], which has shown good promise for inclusion into both viral, bacterial, parasite, and anti-cancer vaccines [10± 16]. When the two main components of our preparation were puri®ed and used to spike saponin similar to the starting preparation of Kensil et al. [3], they were shown to coelute with the components corresponding to QS18 and QS21, respectively (data not shown). This suggests, but does not prove that they are the same substances, since dierences in e.g. monosaccharide composition in side chains may not resolve in dierent peaks during reversed-phase HPLC. We are currently gathering further evidence, that the second peak in our chromatogram is very similar, if not identical to QS21. These results will be published separately (manuscript in preparation). In conclusion, our studies have shown, that it is possible to obtain, by carefully selecting the source material, saponin of very de®ned composition, still possessing important properties of Quil-A such as adjuvant activity and ISCOM formation. Quadri A could thus be used both for the production of Quil-A, and for use as raw material for puri®cation of pure substances like e.g. QS-21. Acknowledgements This work was supported by the European communities contracts ``BIOT-CT91-0256'', ``BIO4-CT980096'', and ``BIO4-CT98-0505''.
S. Kamstrup et al. / Vaccine 18 (2000) 2244±2249
References [1] Espinet EG. Nuevo tipo de vacuna antiaftosa a complejo glucovirico. Gac Vet 1951;74:1±13. [2] Dalsgaard K. Saponin adjuvants. 3. Isolation of a substance from Quillaja saponaria Molina with adjuvant activity in foodand-mouth disease vaccines. Archiv fuÈr die Gesamte Virusforschung 1974;44:243±54. [3] Kensil CR, Patel U, Lennick M, Marciani D. Separation and characterization of saponins with adjuvant activity from Quillaja saponaria Molina cortex. Journal of Immunology 1991;146:431±7. [4] Morein B, Sundquist B, HoÈglund S, Dalsgaard K, Osterhaus A. Iscom, a novel structure for antigenic presentation of membrane proteins from enveloped viruses. Nature 1984;308:457±60. [5] Sjùlander A, Cox JC, Barr G. ISCOMs: an adjuvant with multiple functions. Journal of Leucocyte Biology 1998;64:713±23. [6] LoÈvgren K, Morein B. The requirement for lipids for the formation of immunostimulating complexes (iscoms). Biotechnol Appl Biochem 1988;10:161±72. [7] Baxter-Gabbard KL. A simple method for the large-scale preparation of sucrose gradients. FEBS Letters 1972;20:117±9. [8] Beekman N, Kamstrup S, Dalsgaard K. Method for monitoring ultracentrifuge gradients using a high-performance liquid chromatographic diode array detector. Analytical Biochemistry 1995;228:168±9. [9] RoÈnnberg B, Fekadu M, Morein B. Adjuvant activity of nontoxic Quillaja saponaria Molina components for use in ISCOM matrix. Vaccine 1995;13:1375±82. [10] Hancock GE, Speelman DJ, Frenchick PJ, Mineo-Kuhn MM, Baggs RB, Hahn DJ. Formulation of the puri®ed fusion protein of respiratory syncytial virus with the saponin QS-21 induces protective immune responses in Balb/c mice that are similar to
[11]
[12]
[13]
[14]
[15]
[16]
2249
those generated by experimental infection. Vaccine 1995;13:391± 400. Coughlin RT, Fattom A, Chu C, White AC, Winston S. Adjuvant activity of QS-21 for experimental E. coli 018 polysaccharide vaccines. Vaccine 1995;13:17±21. Cleland JL, Barron L, Daugherty A, Eastman D, Kensil C, Lim A, Weissburg RP, Wrin T, Vennari J, Powell MF. Development of a single-shot vaccine for HIV-1. 3. Eect of adjuvant and immunization schedule on the duration of the humoral immune response to recombinant MN gp120. Journal of Pharmaceutical Sciences 1996;85:1350±7. Helling F, Zhang S, Shang A, Adluri S, Calves M, Koganty R, Longenecker BM, Yao TJ, Oettgen HF, Livingston PO. G(M2)KLH conjugate vaccine: Increased immunogenicity in melanoma patients after administration with immunological adjuvant QS-21. Cancer Research 1995;55:2783±8. Ma JN, Bulger PA, Davis DV, Perillipalmer B, Bedore DA, Kensil CR, Young EM, Hung CH, Seals JR, Pavia CS, Coughlin RT. Impact of the saponin adjuvant QS-21 and aluminium hydroxide on the immunogenicity of recombinant OspA and OspB of Borrelia burgdorferi. Vaccine 1994;12:925±32. Livingston PO, Adluri S, Helling F, Yao TJ, Kensil CR, Newman MJ, Marciani D. Phase 1 trial of immunological adjuvant QS-21 with a GM2 ganglioside-keyhole limpet haemocyanin conjugate vaccine in patients with malignant melanoma. Vaccine 1994;12:1275±80. Newman MJ, Munroe KJ, Anderson CA, Murphy CI, Panicali DL, Seals JR, Wu JY, Wyand MS, Kensil CR. Induction of antigen-speci®c killer T lymphocyte responses using subunit SIVmac251 gag and env vaccines containing QS-21 saponin adjuvant. AIDS Research and Human Retroviruses 1994;10:853±61.