Journal of Immunological Method.s, 128 (1990)51-58
Elsevier JIM 05499
An immunoassay for histamine based on monoclonal antibodies E. Hammar, A. Berglund, A. Hedin, A. Norrman, K. Rustas, U. Ytterstr~m and E. ~ k e r b l o m Pharmacia Diagnostics A B, S- 7~1 82 Uppsala, Sweden
(Received28 June1988,revisedreceived30 October1989,accepted20 November1989)
Monoclonal antibodies with high specificity for histamine as well as for 1-methylhistamine were obtained after immunization of mice with a conjugate where the histamine was coupled via its ring 1-nitrogen to dog serum albumin. An immunoassay was developed for the quantitation of histamine release from basophils and 1-methylhistamine release from mast cells after provocation. The test method is based on competitive inhibition between histamine and a labelled histamine conjugate for the antigen binding sites of the antibodies. The separation step is performed by the addition of solid phase bound anti-mouse subclass specific antibodies. The sensitivity of the assay is 2/~g/1 for histamine and 0.1/~g/l for 1-methylhistamine. No cross-reactivity was obtained with other metabolites of histamine or with histidine. Serotonin and dopamine were detectable, but only in doses (mg/l) well above the normal concentration found in the circulation. The immunoassay has been evaluated for its capacity to measure histamine release in vitro. A good correlation with the conventional fluorometric assay was obtained when histamine released from allergen stimulated leucocytes from allergic patients was tested. Urinary samples from patients undergoing hyposensitization showed a mean excretion of 1-methylhistamine at a level of 131 ismol MeHi/mol creatinine. The release of histamine and 1-methylhistamine in vivo was examined in plasma samples taken during a bronchial provocation test. A significant elevation above the basal analyte level occurred ten minutes after provocation. Key words: Radioimmunoassay;Enzymeimmunoassay;Histamine;1-Methylhistamine;Monoclonalantibody
Introduction Histamine stored by mast cells and basophils is an important mediator in allergic and inflammatory reactions. Several methods for the determination of histamine and its metabolites are available
Correspondence to: E. Hammar,PhasmaciaLEO Therapeutics AB, S-75182 Uppsala.Sweden. Abbreviations: EIA, enzyme-linkedimmanosorbentassay; RIA, radioimmunoassay;HPLC, high performance liquid chromatography;EDTA,ethylenediaminetetraaceticacid;BPT, bronchialprovocationtest; FEVI,forcedexpiratoryvolumein 1 s; OPT, fluorometricassay of histamine;MeHi, 1-methylhistamine.
but they all require expensive instrumentation and are complex to perform (Beaven et al., 1978; Hook et al., 1981; Keyzer et al., 1981; Granerus et al., 1984). There has been a need for many years to develop a simple histamine assay which could be used routineiy for the analysis of a large number of samples. An immunoassay is sensitive, rapid and convenient but requires a specific antibody against the compound to be measured. Efforts have been made to produce specific antibodies against histamine but it has proved to be difficult (Mita et al., 1984). Histamine has been coupled via its primary amino group to a carrier protein in order to obtain an immunogen for immunization.
Antibodies produced in this way show a high crossreactivity with histidine. Up till now only one abstract (Ha.vdik, 1983) and one patent (Buckler et al.) have been published describing specific radioimmunoassays for histamine. In addition several articles about immunoassays which require derivatization of the histamine before assay have been published (Guesdon et al., 1986,1988; Peyret et al., 1986; Delaage et al., 1988; McBride et al., 1988; Morel et al., 1988). We have produced a specific monoclonal antibody against histamine by coupling histamine via its 1-ring nitrogen to a carder protein. No crossreaction was found with histidine. As a consequence of this choice of cow piing position the antibody produced also reacted with 1-methylhistamine, one of the main metabolites of histamine. The test method used was based on a double antibody solid phase system and measured both histamine and 1-methylhistamine. The sensitivity was ten times higher for 1-methylhistamine than for histamine. Thus, in the in vitro situation, where the released histamine was not metabolised, the measured change in analyte level CH.CH.NHCCH.-.CH.SH
Materials and methods
Immunoassay reagents Immunogen. The hnmunogen III (Fig. 1) was prepared by coupling the histamine derivative 1( N-(3-thiopropionyl)-2-aminoethyl)-4-(2-aminoethyl)imidazole (Fig. 1, I) to iodoacetyl dog serum albumin (Fig. 1, II) at pH 7.5. The product was purified by size exclusion chromatography on Sephadex G-25 (Pharmacia AB). The degree of modification was calculated from the sulphur content (elemental analysis) and the protein content (amino acid analysis). Approximately 17 4-(2aminoethyl)-imidazolyl groups were coupled per DSA. The iodoacetyl DSA used for the coupling I N
was specific for histamine. However, in the in vivo situation, histamine', is rapidly degraded. Since our method is ten times more sensitive for 1-methylhistamine than for histamine, 1-methylhistamine is preferentially measured in plasma and urine samples.
CH2CN DSA -----9 O "an
0 DSA + I C H 2 C O O N , ~ 0
> [ICH2~NH~DSA O
Fig. 1. Reactionmechanismsinvolvedin the synthesisof the immunogen(IIl), the iodoacetyldog serum albumin(II) and the EIA tracer(VI).
53 was synthesized by acylation of DSA with N-succinimidyl iodoacetate (IV, Fig. 1) at pH 8.1. The product was purified on a Sephadex G-25 cohinm. The degree of modification was calculated from the iodine content and the protein content. The compounds I, II and III (Fig. 1) were synthesized as described by .~kkerblom and Norman (1986) and Berglund et al. (1986). Antibody and antisera. A mouse antihistamine monoclonal antibody of the IgG2a subclass (aHiMab) was prepared as described elsewhere (Turk, 1981). Briefly BALB/c mice were immunized by subcutaneous injections of 50 tLg of the histamine immunogen emulsified in complete Freund's adjuvant. Final injections of the immunogen were given intraperitoneally after 4 weeks, for 3 consecutive days and 3-4 days later the spleens of the mice were removed for hybridization. The myeloma celline SP2/0-Ag 14 was used for hybridization. 108 spleen cells were mixed with 1-3 x 107 myeloma cells in Dulbecco's medium. The cells were spun down to form a tight pellet and all of the supernatant was removed. 1 ml of 50% (w/v) polyethylene glycol was added slowly with gently mixing of the cells. After I min the cell suspension was diluted slowly with 5 ml of Dulbecco's medium. The cells were spun down, washed and distributed into five 96 well plates in Dulbecco's medium containing 10% fetal calf serum, 100 I U / m l penicillin and 100 /~g/ml streptomycin. The plates were placed into 7% CO2 incubation at 37 ° C. Half of the culture medium in each well was replaced by fresh medium containing hypoxanthine-aminopterin-thymidine(HAT) on days 1, 2, 3 and every 3 days thereafter for 2 weeks. The cultures were tested for antihistamine activity using the enzyme linked immunosorbent assay. The positive cultures detected by this assay were cloned repeatedly using the limiting dilution method and were then grown in cell culture. The mouse monoclonal antibodies were purified on an ion exchange column and then gel-filtered (Carlsson et al., 1985). Polyclonal antibodies were raised in rabbits using the monocloval mouse Ig(32a as immunogen. The gamma fraction of the rabbit anti-mouse IgG2a antiserum was prepared by Na2SO4 precipitation and then covalently coupled to micro agarose bead~
RIA tracer. The immtmogen was labelled with three 125I per DSA using a modification of the chloramine-T method (Hunter et al., 1962). EIA tracer. The enzyme conjugate (VI) was prepared by reacting 1-(N-(3-(2-pyridyldithio)propionyl)-2-aminoethyl)-4-(2-aminoethyl)-imidazole (V) with reduced fi-galactosidase in 0.2 M borate buffer pH 8.0 (Fig. 1) (~,kerblom et al.; Berglund et al.). Released pyridine-2-thione was measured at 343 nm (Carlsson et al.) and was used for the calculation of the degree of modification. Protein content was determined by amino acid analysis. About 15 4-(2-aminoethyl)imidazolylgroups were coupled per p-galactosidase. The conjugate was purified by size exclusion chromatography on Sephadex G-25 (Pharmacia AB). Standard Histamine diphosphate (Fhika) or 1-methylhistamine-dihydrochloride (CalbiochemBehring Corp.). Buffer and substrate solutions. Phosphatebuffered saline containing 0.3~ HSA, PBS-HSA, (0.15 M NaC1 containing 0.050 M phosphate buffer, pH 7.4). Substrate solution (O-nitrophenyl-fi-galactoside) was prepared from reagents in the Phadezym IgE PRIST kit (Pharmacia Diagnostics AB). Sample preparations Histamine release in vitro. Hepafinized venous blood was collected from allergic individuals. From each of them one part of the sample was used to prepare washed leucocytes (Lichtenstein et al., 1964). These preparations were resuspended in approximately one sixth of the initial volume. Whole blood as well as leucocyte preparations were stimulated at 37°C with dilutions of allergen. The supernatants obtained by centrifugation (800 × g in 10 rain at 4°C) were tested in the immunoassay, whereas only the histamine release from washed leucocytes could be tested in the conventional fluorometric assay. Provocation reaction Allergic patients on histamine-free diets underwent bronchial provocation tests (BFI') at Sahlgren's Hospital, Gothenburg, Sweden. The maximum fall in the FEV1 (forced expiratory volume in 1 s) was around 30~ after 15 min. Blood samples were collected in chilled EDTA tubes from a venous catheter before, and 5, 10, 20, 40 and 60 min after BPT. The
54 tubes were kept on melting ice and centrifuged within 30 min at 2 0 0 x g , for 10 mm at 4°C. Urine was collected in hourly fractions starting 1 h before to 2 h after inhalation of the allergen extracts. The plasma and urine samples were immediately frozen and tested by the immunoassay within 1 week. Generally, urine samples had to be diluted 1/50 while plasma samples were tested undiluted.
Material used for correlation and recovery studies. The urines, collected from patients receiving hyposensitization therapy, were stored at - 20 ° C without additives until the day of testing. The length of collection time (4 h), and the storage temperature ( - 2 0 °C) did not necessitate the inclusion of any additives to prevent bacterial growth. Otherwise acidification by the addition of 2 M HCI is recommended. The plasma samples studied comprised pools of either EDTA-plasma or citrated plasma which had been kept frozen at - 2 0 ° C for at least 1 year. Samples for use in recovery studies were prepared by the addition of a small volume of a known amount of 1-methylhistamine to buffer and plasma respectively.
pressed as a percentage of the zero response (B0). The calibration curve was constructed by plotting the response for the analyte as a percentage of the zero dose-response semilogarithmically versus the analyte concentration. The average response for each unknown sample was expressed as the percentage of the zero dose-response and the concentration of the sample could be read from the dose response curve and thereafter multiplied by the appropriate dilution factor. The formula used for the calculation of recoveries was Co) × 100~ CT
( CR --
C R = observed analyte concentration in the recovery sample. Co = Analyte concentration in the recovery sample originating from the matrix. C T = Theoretical analyte concentration in the recovery sample originating from the added amount of analyte.
Test methods Immunoassay.
Standard, urine samples, tracer and antibody solutions were diluted in PBS-HSA buffer. 100 /~1 of standard or sample, 100 /~1 of EIA or RIA tracer and 100 /~1 of aHiMab were mixed and incubated for 18 h at room temperature. Free and bound tracer were separated by adding 1000 /zl anti-mouse IgG2a-micro agarose beads. After 1 h on a shaker, the suspension was washed three times in 0.15 M NaCI containing 0.5% Tween 20. RIA procedure: the pellet was measured in a gamma-counter. EIA procedure: 200/~1 substrate were added to the pellet and after 2.5 h at 37 o C the enzymatic reaction was stopped by the addition of 1000 /~1 of 0.5 M Na2CO 3. After centrifugation, the supernatant was measured at 420 nm in a spectrophotometer. The conventional automated fluorometric assay of histamine was performed as previously described (Hook et al., 1981). Calculation of results. The average response of the zero standard (B0) was calculated. The response of each analyte concentration was ex-
Characterization of the test procedure The specificity of the Mab is demonstrated in Fig. 2. No detectable cross-reactivities with histidine or other metabolites of histamine (1-methylimidazole-4-acetic acid, imidazole-4-aeetic acid), except 1-methylhistamine, were observed. The detection limits were 10 -9 M for 1-methylhistamine, 10 -s M for histamine and 10 -s M for serotonin. The coefficients of variation were 7~ (within-assay), 4% (between-assay) and 8 - 9 ~ for the total assay precision. The overall immunoreactivity of the tracer reagents was 90~. Actual values for zero dose responses were approximately 8000 cpm for freshly made radioactive tracer, and an absorbance value of about 0.8 for the EIA conjugate.
Measurements of histamine release in vitro A correlation coefficient of 0.96 was obtained between the fluorometric assay and the immunoassay when samples prepared from washed leucocytes were tested (Fig. 3). A comparison between the two methods, when samples prepared
. . . .
50 40, 30, 20 10 10-+o ~C
0 ..... ~CH2NH2
0% ~ ._ ~'~4 . ~JQnnoooO 10-7 1O'S lO-S
H ~ C H = C O O H0 . . . . .
Imidazole-4 -acetic acid
~ C H = CH=NH2e ='ram H i s t a r l d n e
Corm. o f a n a l y t e (bO I)
H COOH A =,==mHistidine I~JCH=(~HNHz
Fig. 2. Dose-response curves for substances similar to histamine determined by the immunoassay.
COMPARISON BETWEEN TWO DIFFERENT METHODS OF STIMULATION AND DETECTION OF HISTAMINE
' 'Co¢¢.of hlstan~le 70. (ng/r~)
Cone. of histamine (ng/r~)
Ooo o o e oo2O°o o 0 o o._~_o 10
Number of saml0k)s Corr.¢o.eff. Eo~atlon of the curve
Method Patient RIA OPT
o~ o o
•®e~'.'~'~'~ee~=e=e= ~ e
10 OPT method 10~
Fig. 3. Correlation between blood histamine concentration determined by the enzyme-linked immunosorbent assay (EIA) and the fluorometriv method (OPT).
1(,4Co lO'aCo 10"=Co 104co
: Cone.of aBergen co
Fig. 4. Histamine response curves ('or allergen stimulation of basophils determined by the immunoassay (RIA) and the fluorometric assay (OPT). The allergens used for stimulation were birch (patient 1) and timothy (patients 2 and 3) extracts used at concentrations < 40 BU/ml (Co).
f r o m whole b l o o d were u s e d in t h e i m m u n o a s s a y , is ~ o w n in Fig. 4.
Measurements of 1-methylhistamine in plasma and urine Provocation reaction. A significant elevation o f p l a s m a 1 - m e t h y l h i s t a m i n e w a s o b s e r v e d 10 rain after B I T (Fig. 5). T h e u r i n a r y excretion o f 1m e t h y l h i s t a m i n e w a s f o u n d to b e o f t h e o r d e r o f ~ g / h . T h e actual s a m p l e c o n c e n t r a t i o n s were r o u g h l y 100 t i m e s h i g h e r t h a n t h e p l a s m a samples. A 15% increase o f u r i n a r y 1 - m e t h y l h i s t a m i n e w a s o b s e r v e d d u r i n g t h e first h o u r o f u r i n e collection after BPT, w h i c h indicates, w i t h o u t b e i n g significant, a n elevation of t h e a n a l y t e c o n c e n t r a t i o n . Correlation with other methods. C o r r e l a t i o n studies b e t w e e n t h e R I A a n d t h e H P L C m e t h o d ( G r a n e r u s et al., 1984) were p e r f o r m e d at S a h l g r e n ' s Hospital, G o t h e n b u r g , Sweden. T h e v a l u e s o b t a i n e d s h o w e d a g o o d correlation b e t w e e n the two t e c h n i q u e s (for n = 85, r = 0.98 a n d a (intercept o f regression line) ffi 6.6 a n d b (slope o f regression line) ffi 0.93.) T h e m e a n excretion level o f 1 - m e t h y l h i s t a m i n e i n t h e s e u r i n e s a m p l e s w a s 131 p m o l M e H i / m o l crea~e (or if e x t r a p o l a t e d f r o m 4 h collection time, 2 5 0 / ~ g M e H i / 2 4 h).
, ) Time after BPT 120 (min)
11 +~ 10
) Time after BPT (min)
~ "Vaneafter BPT 6 (~0 120" (rain) Fig. 5. The relationship between lung function (FEVt), plasma 1-methylhistamine and urinary 1-methylhistamine excretion after allergen challenge in a bronchial provocation test (BPT).
TABLE I COMPARISON BETWEEN TWO METHODS OF ANALYSING 1-METHYLHISTAMINE IN PLASMA SAMPLES, NAMELY MASS FRAGMENTOGRAPHY METHOD AND IMMUNOASSAY The table shows the values obtained before and after addition of analyte to plasma and buffer samples and the resulting percentage of recovery upon addition. Matrix
CT (nM) 1-methylhistamine
CR (riM) 1-methylhistamine
Recovery i (%)
Mass fragraentography method
1.3 = C o × 1 . 0 1
0 = Co × 1 . 0 1
Citrate plasma Citrate plasma Citrate plasma
2.3 ffiCo x 1.01 5.3 2.6
6.2 = Co × 1.01 8.3 8.5
Mass fragmeoto8raphy method
Key: C Rffiobserved 1-methylhistamine conc. in the recovery sample. CO ffi methylhJstamine conc. in the recovery ociginating from the matrix (which is 99% of the matrix value). C T ffi theoretical methylhistamine conc. in the recovery sample originating from the added amount of 1-methylhistamine. RecoveryI ffi ~ a - ~o~ .100%. c r
57 The reported normal ranges for urinary 1-methylhistamine are 40-160 /~mol M e H i / m o l creatinine and 60-280/~g MeHi/24 h (Keyzer et al., 1981,1984). On an average, 1-methylhistamine and histamine occurred in a ratio of four to one in the tested material, according to the HPLC method. In most cases, urine samples had to be diluted 1/50 in order to be measured with optimal accuracy with the immunoassay. Upon dilution, the dose response curves of the urine samples were parallel with the reference material. Comparable results were also obtained 'when plasma samples were tested with the RIA and the mass fragmentography methods (Keyzer et al., 1981) at the Bergschot Centre for Research, Breda, The Netherlands (see Table I). The non-specific factors which are known to affect immune reactions include detergents, ionic strength, protein concentration and pH. These factors vary in different samples. With the chosen buffer system, the effect of such variations were minimized. However, in comparison with the reference method there was a tendency towards for the analyte concentration in undiluted plasma samples to be higher in the immunoassay. This suggests that the matrix used for preparation of the standard solution requires adjustment. With existing reference material, parallel dose-response curves were obtainea with plasma samples starting from a dilution of 1/2. Sample recovery as measured in the immunoassay following addition of analyte to plasma or buffer was found to be good (Table I).
The immunoassay described here is easy to perform and hundreds of samples can easily be analysed within 2 days by one technician. The specificity of the Mab eliminates the need for any pretreatment of the samples. Serotonin, which is one of the two analytes that ¢rossreact with histamine, is of no importance since it occurs in biological samples at a concentration which is 10,000 times lower than the detection limit dose. In contrast, 1-methylhistamine, the other cross-reacting analyte, shows a ten-fold higher sensitivity of detection compared to histamine. When present
at a molar ratio of 2 MeHi:l Hi it leads to an analytical error of less than 5~ in the assay. In contrast to the fluorometric assay the high specificity of the Mab permits the immunoassay to be used for the analysis of samples prepared from allergen stimulated whole blood. When analysing mediator release from washed leucocytes, a good correlation was obtained between the immunoassay and the fluorometric assay. Although differing in basophil content, a good correlation was also found between the patterns of reactivity when fluorometric-tested leucocyte derived samples and immunoassay-tested whole blood derived samples from the same patient were compared (Fig. 3). Consequently, in this appfication, the observed changes in anaiyte level appear to be specific for histamine. The sensitivity and precision of the immunoassay make it suitable to use for the detection of fluctuations in the 1-methylhistamine level in a series of biological samples. The increase in 1methylhistamine level during a provocation study with standardized sampling was high enough to be measured. Elevated urinary 1-methylhistamine secretion was found during the first hour of collection after BPT, as has been reported by Keyzer et al. (1984). The immunoassay was found to be in accordance with established methods for measurements of 1-methylhistamine levels in urine and plasma samples. Using urine samples a good correlation was found between the immunoassay and the HPLC-method. In a smaller study on plasma samples, values of a similar order were obtained with the RIA and the mass fragmentography method. The distribution and concentration ranges of the metabolites make this sensi~.iveimmunoassay an excellent tool for quantitative measurements of histamine and 1-methylhistamine.
References ~kerblom, E. and Norrman,A. PharmaciaAB, Carrier-bound histamine,its manufactureand its use. InternationalPatent PublicationWO 86/04331,1986. Beaven,M.A.and Horakova,L. (1978)The enzymaticisotopic assay of histamine.In: M. RochaeSilvia(Ed.), Handbook of ExperimentalPharmacology,Vol.18/2. Springer,Berlin, p. 153.
Berglund, A., Akerblom, E. and Hedin, A. Phasmacia AB, An antibody preparation and processes for its manufacture and use. International Patent Publication WO 86/04420,1986. Buckler, R.T., Dailey, F., Ficalora, 1., Gavin, I. and Plankett, G. Miles AB. European Patent Publication 0 208 953 A1. Carlsson, J., Drevin, H. and Ax~n, 11. (1978) Protein thiolation and reversible protein - Protein conjugation. Biochem. J. 173, 723. Carlsson, M., Hedin, A., Inganiis M., H~fast, B. and Biotaberg, F. (1985) Purification of in vitro produced mouse monoclonal antibodies. A two-step procedure utilizing cation exchange chromatography and gel filtration. J. Immunal. Methods 79, 89. Delaage, M., Darmon, M. and Morel, A. (1988) Histamine assay: use of a monoclonal antibody for radioimmunoassay and enzyme immunuassay. Allerg. ImmunoL (Pads) 20, 9. Granerus, G. and Wass, U. (1984) Urinary excretion of histamine, methylidstamine (1-MeHi) and methylimidazolcacetic acid (MelmAA) in mastocytosis: comparison of new HPLC methods with other present methods. Agents Actions 14, 341. Guesdon, J.-L., Cbevrier, D., Mazi6, J.-C., David, B. and Avreameas, S. (1986) Monoclonal anti-histamine antibody. Preparation, characterization and application to enzyme immunoassay of histamine. J. lmmanol. Methods 87, 69. Guesdon, J.-L., Chevrier, D., Fadel, R. and Avrameas, S. (1988) Immanoenzyme assay for histamine. Allerg. Immunul. (Paris) 20, 9. Haydik, LB. (1983) Histamine determination by radio-immanoassay (RIA). J. Allergy Immunol. 71,152 (abstract). Hook, W.A. and Siraganian, R.P. (1981) Influence of anions, cations and osmoladty on IgE-mediated histamine release from human basophils. Immunology 43, 723. Hunter, W.M. and Greenwood, F.C. (1962) Preparation of
iodine-131 labelled human growth hormone of high specific activity. Nature 194, 495. Keyzer, J.J., Wolthers, B.G., Muskier, F.A.J., Kauffman, H.F. and Groen, A. (1981) Determination of N.methyihistamine in plasma and urine by isotope dilution mass fragmentography. Clin. Chim. Acta 113,165. Keyzer, I.I., Kauffman, H.F., De Monehy, I.G.R., Keyzer-Udding, I.I. and De Vries, K. (1984). Urinary N-methylhistamine during early and late allergen-induced bronchial-obstructive reactions. J. Allergy Clin. Immanol. 74, 240. Lichtenstein, L.M. and Osier, A.G. (1964) Studies on the mechanisms of hypersensitivity phenomena. IX. Histamine release from human leukocytes by ragweed pollen antigen. J. Exp. Med. 120, 507. McBride, P., Bradley, D. and Kaliner, M. (1988). Evaluation of a radioimmunoassay for histamine measurement in biologic fluids. J. Allergy Clin. Immunol. 82, 4. Mita, H., Yasueda, H. and Shida, T. (1984) An attempt to produce an antibody of hist,~nine and histamine derivatives. Agents Actions 14, 574. Morel, A. and Delaaga, M. Immunotech, Immunoassay for low mol. wt. amine esp. hsitamlne - after acylation to increase mol. wt. Then competitive binding reaction. EP - 161-195A. Morel, A. and Delaage, M. (1988) Immunotech, Immunoanalysis of histamine through a novel chemical derivatizadon. J. Allergy Clin. Immunol. 82, 4. Peyret, L.M., Moreau, P., Dullac, J. and Geffard, M. (1986) Antibodies to histamine. Specificity studies and radioimmunological assay. J. Immunol. Methods 90, 39. Turk, J.L. (1981) Production of Antibody-Producing Hybridomas in the Rodent System. In: G.J. H~lmmerling, U. Hiimmerling and J.F. Kearney (Eds.), Research Monographs in Immunology, Vol. 3. Elsevier, Amsterdam, p. 563.