JOUI7IJlI of Immunological Methods, 149 (1992) 63-68 ~ 1992 Elsevier Science Publishers B.V. All rights reserved 0022-1759/92/$05.00
63
JIM 06251
A quantitative immunocytochemical method for the measurement of islet cell cytoplasmic antibodies Michael O. Marshall and Poul E. H0yer Department Endocrinology and Immunology, Novo Nordisk A / S, Bagsvaerd, Denmark, and Department of Medical Anatomy A, Panum Institute, Denmark (Received 3 April 1991, revised received 6 December 1991, accepted 13 December 1991)
A quantitative immunocytochemical method for the measurement of islet cell cytoplasmic antibodies has been developed. The method employs human or rat pancreas, a protein A-peroxidase / diaminobenzidine secondary antibody system and an independent measurement of islet total and exocrine mean integrated absorbance by scanning microdensitometry. Specific islet cell cytoplasmic antibody binding Oslet total-exocrine mean integrated absorbance) was dependent on serum dilution and substrate reaction time. The detection limit was approximately 5 JOF units. Specific islet cell cytoplasmic antibody binding values with human and rat pancreas were similar. Specific islet cell cytoplasmic antibody binding (human pancreas) was greater (p < 0.(01) in sera from patients with newly diagnosed insulin dependent diabetes mellitus (0.119 ± 0.086, n = 29) compared to normal sera (0.003 ± 0.008, n = 29). Thus, the method has been validated and may be useful for measuring the blocking effect of potential antigens on specific islet cell cytoplasmic antibody. Key words: Immunocytochemistry; Islet cell; Cytoplasmic; (Antibody)
Introduction
ICA is frequently present in the sera of patients with 100M both at diagnosis and in the pre-diabetic phase (Bottazzo et aI., 1974; Lendrum et aI., 1976; Gorsuch et aI., 1981; Srikanta et aI., 1985a). The traditional method of detecting these antibodies is by indirect immunofluorescence on cryostat sections of human
Correspondence to: M.O. Marshall, Diabetes Research, Novo Nordisk A/S, 2880 Bagsvaerd, Denmark. Abbreviations: ICA, islet cell cytoplasmic antibody; 100M, insulin dependent diabetes mellitus; BSA, bovine serum albumin; FlTC, fluorescein isothiocyanate; PBS, phosphatebuffered saline.
pancreas (Bottazzo et aI., 1974), in which ICA positivity is based on a subjective judgement of the relative intensities of islet and exocrine fluorescence. The method can be semiquantitated by determining the end point titre of a serum against a JOF standard (Bonifacio et aI., 1988). However, for some experimental studies, in particular when testing the blocking of leA antibody binding by potential auto antigens, a method is required which can quantify the amount of antibody specifically bound to islets. The method of Srikanta et al. (1985b) uses human pancreas and fluorescein conjugated protein A as secondary antibody and photometric measurement of fluorescence intensity. We describe here an alternative, quantitative immunocytochemical method employing either human or rat pancreas, peroxidase-conjugated
64
protein A as secondary antibbdy and measurement of light absorbance with a microdensitometer. Materials and methods
Materials
Peroxidase-conjugated protein A and FITCconjugated rabbit anti-human IgG were purchased from Amersham (Little Chalfont, England) and Dako (Glostrup, Denmark), respectively, and diaminobenzidine was purchased from Sigma (St. Louis, MO, U~A). R2D6 (mouse monoclonal anti-islet ganglioside) was a kind gift from Dr. R. Alexandro, University of Miami, USA. Eukitt was purchased from O. Kindler, Freiburg, Germany.
Sera
Sera from 29 newly diagnosed 100M patients (mean age 19 years, range 4-38 years; eight female, 21 male), were obtained from Hvid0re and Glostrup Hospitals, Denmark, and sera from 30 normal individuals (mean age 25 years, range 7-63 years; 12 female, 18 male) were obtained from laboratory personnel. Sera were absorbed overnight with rat liver acetone powder (20 mg/mI) prior to assay. The JDF standard serum was a gift from Dr. J. Ludvigsson, Linkoping Hospital, Sweden.
Quantitative immunocytochemical method for ICA 4 JLm thick cryostat sections of human pancreas (blood group 0) or Wistar Furth rat pancreas were air dried (30 min), acetone fixed (5 min), and then incubated at 200C for 40 min in a moist chamber with rat liver absorbed sera diluted 1/2 in PBS-l % BSA. After washing with PBS (3 x 5 min), the sections were incubated with peroxidase-conjugated protein A 0/100) for 40 min and again washed. They were then immersed in 250 ml of a solution of diaminobenzidine (I mg/mI), H 202 (0.02%) in PBS for 10 min washed and dehydrated by sequential immersions in ethanol (96%, 3 min; 2 x 99%, 3 min), and dried before mounting with Eukitt medium. Each serum was incubated in duplicate on consecutive sections.
Quantitation of IgG binding. A minimum of 20 areas (707 JLm 2 per area) randomly selected from a minimum of six islets from two consecutive sections on coded slides were scanned using a Vickers M85A scanning and integrating microdensitometer (machine settings: X 40 objective, light of 460 nm, scanning spot diameter of 0.5 JLm). For each area the individual spot readings were integrated and a mean integrated absorbance (MIA) ± SEM over the 20 areas calculated by reference to a standard calibration graph constructed by measuring neutral density filters with known absorbance values. An equal number of similar sized areas were scanned in the exocrine tissue. The exocrine MIA (± SEM) value was taken as a measure of non specific IgG binding to both exocrine and islet tissue. Specific ICA was expressed as islet specific MIA (± SED), calculated as follows: islet specific MIA - islet total MIA - exocrine MIA. Indirect immunofluorescence method for ICA Undiluted sera were tested for ICA positivity using FITC-conjugated rabbit anti-human IgG as secondary antibody, using the same human pancreas and incubation times as described above. Results of the ICA proficiency test no. 2 were: validity (97%), consistency (95%), sensitivity (94%), specificity (100%).
JDF standard curve
Standard sera were prepared by diluting the JDF standard serum (80 JDF units) to 2, 4, 10, 20 and 40 JDF units in normal serum.
Statistics
JDF standard curves were estimated using non-linear regression analysis and diabetic and normal sera were compared using the t test. Human and rat pancreatic samples were compared using the error-in-variables regression method, which takes into account the random variation of the results. Results
Time course of peroxidase reaction The rate of formation of the diaminobenzidine reaction product in both islet and exocrine tissue
0.'
',1- - - -,i'
- - - i- - - - - -i- - - - -
c
i
0.1
.,.0- _ _ _ -
-
-
-0- -
-
-
-
_0-. -
0'
-
-
..... 1
-A
i
OA CO••
i
--0
was initially linear and then decreased until 10 min; after this the product concentration remained constant. An experiment illustrating these kinetics is shown in Fig. 1. The ratio of islet to exocrine remained constant over the whole time period. MIA values are normally measured during the linear phase of an enzyme reaction. However, for peroxidase activity any point on the curve may be selected (H0yer et aI., 1991). In this study a reaction time of 10 min was used for convenience. There was no difference in product formation when the sections were immersed in diaminobenzidine as described in the materials and methods section or when the diaminobenzidine was applied as a drop to the sections (not shown).
Absorbance in islet and exocrine tissue as a function of serum dilution
The dilution curves for three leA positive sera and monoclonal antibody R2D6 are shown in Fig. 2. For sera 1 and 2 the islet specific MIA and exocrine MIA both decreased on serum dilution. Likewise, the binding of R2D6 to the plasma membrane of beta-cells decreased with antibody dilution (Fig. 2). However, for serum 3 the islet specific MIA first increased on diluting to 1/16 and then decreased. Serum 3 was the only one of 12 sera tested showing this prozone-like effect.
JDF standardization and interassay variation With both human and rat pancreas islet specific MIA was proportional to the JDF value of
.....
: ~~~,
" "-
~-
......
0.1
c
i
0 ..
OA 0.1
0 ••
0.1
0.0
--
0.1
-, o •
----
~
0.0 '---~----o • 4 I 'I II 14,.1
0 ..
0.1
,,
0.1
0.0 L-~~~---=-'::::--:....1... o • • I 11 II 1.,.1 . . . . dIuIIoII
(Illinutea I
Fig. 1. Development of the diaminobenzidine reaction product as a function of time. Islet total (l»; islet-specific ( ... ); exocrine (0). The results are from a sil18le experiment with rat pancreas and tile variations are the SEM (islet, exocrine) or SED (islet-specific) of each MIA value.
OA
o.J
0.1
0.0 L~---,-~~-,------'-~""""'~-"--~--' 40 10 eo .0 10 10 o
RMotIon tIIIIe
......
0.1
0.1
0.4
6S
0.0
4 1 ,. II .4 1.1 ..... cIIuIIon
~.......;=-:o:::....;=..",,"--,z--
0
400
laoO
•• 00
An-,. _ _
Fig. 2. Mean integrated absorbance as a function of serum dilution. Symbols as in Fig. 1. Sera were diluted in PBS-l % BSA, and assayed on rat pancreas. R2D6 binding was assayed using peroxidase-conjugated anti-mouse Ig (1/100). The results for each serum are from a single experiment and the variations refer to the SEM (total islet, exocrine) or SED (islet-specific) of each MIA value.
the standard serum over a range of 5-80 JDF units, with the greatest discrimination between 40 and 80 units (Fig. 3). The standard curves of the human and rat systems were identical. The mean 0.2
0.0
b===::::::::~....J---L--.L.--L----'
(01
I.'
1
10
10
40
10
JDF unlta
Fig. 3. Islet-specific mean integrated absorbance as a function of concentration of JDF standard serum. Human pancreas (0); rat pancreas (.). Each curve best fit is the mean ( ± SEM) of four experiments and was calculated from the formula islet-specific MIA = ao + at exp(~ log 2 JDF unit) where ao - mean MIA of the 0 JDF standard, at'" mean MIA of the 80 JDF standard, and fJ - rate of change o~ MIA with JDF unit.
66
assay detection limit from four standard curves (99% upper limit of the zero JDF serum) was 4.2 lOF units (range 2.5-6.2 lDF units) for human pancreas and 5.2 JOF units (range 2.9-8.3 lOF units) for rat pancreas. The inter assay variation for a control serum with 20 lDF units was 8% (n = 6) with human pancreas and 7% (n = 6) with rat pancreas. Comparison of diabetic and normal sera Islet total and islet specific MIA values with human and rat pancreas are shown in Fig. 4. There was considerable overlap in islet total MIA values between the diabetic and normal groups. However, when the islet-specific MIA values were compared the two groups were clearly differentiated. For example, islet-specific MIA (human HUMAN PANCREAS IMt .,.oIfio MIA
0 ••
0 •• 0." 0.2
+-r •
•
I
0."
+ f
0.1 0
0
................. It
-
-.-
-
-+
It 10 I*MtIM ........
10
RAT PANCREAS ....t .,.oIfio MI A
....t total MIA
0••
0 .• 0."
-+-
I
0.2 0 10
+••
,.
I*NIIM ........
0.4 0.2
0
TABLE I ICA POSITIVITY IN SERA FROM NEWLY DIAGNOSED IDDM PATIENTS Human pancreas ICM
+ IOF
+
5 1
_T___ •
+ 3
2 1
5 18
n-29
]4
n-20
Key: ICM - quantitative immunocytochemical method; IDF - indirect immunofluorescence method; +, positive; -. nelative.
pancreas) was O. 119 ± 0.086 (mean ± SD) in the diabetic group and 0.003 ± 0.008 in the normal group (p < 0.001). ICA positivity in the diabetic group (islet-specific MIA values > mean + 3 SO of normals) was 79% and 85% with human and rat pancreas respectively; the corresponding positivity in the normal groups was 3% and 0%, respectively. W,hen the quantitative immunocytochemical and immunofluorescence methods were compared there was agreement regarding ICA positivity for 79% and 80% of the sera tested with human and rat pancreas respectively (Table I). Those sera positive with the quantitative immunocytochemical method but negative with the
I
004
MIA .... -O.OhO.'7MIA_ r • 0.84
• •
0.'
i
• •••
Rat pancreas ICM
•i
0 ••
1
0.1
-+---
1. ................ 10
Fig. 4. Comparison of sera from newly diagnosed diabetic patients and normals. Each serum was assayed once on both human and rat pancreas. - - refers to the mean of each group and ...... to the mean + 3 SO upper limit of normal range. The distnbution of MIA values in each group wu found 10 be normal. Group differences were analysed by Student's t test.
i
0
o
0.1
0••
0.'
004
....t epeoIfto MIA IIum8n ,.,.... Fig. 5. Comparison of human and rat pancreas. 20 sera from newly diagnosed 100M patients were assayed on both human and rat pancreas. The linear regression wu estimated usina the error invariable. rearcssion.
67
immunoflourescence method had low isletspecific MIA values (data not shown). Comparison of human and rat pancreas Islet total MIA values (p < 0.0001} from nor-
mal sera were greater using rat rather than human pancreas (Fig. 4). However, the islet-specific MIA values obtained using pancreatic material from the two species were closely correlated (r = 0.84, p < 0.001) and the estimated regression line was not significantly different from the identity line (p > 0.10) (Fig. 5). Discussion
The immunocytochemical method for leA described here permits the independent measurement of the polymerised oxidation product of diaminobenzidine in islet and exocrine pancreatic tissue. Assuming proportionality between absorbance and the amount of antibody bound, the islet-specific MIA values represent the specific antibody binding to islet antigens (i.e., specific leA). In this calculation the IgG non specific binding to islets is taken to be the same as that of the measured IgG binding to exocrine tissue. This is certainly valid for the normal sera we have studied since the islet-specific MIA values were zero (Fig. 4), and it is assumed to be valid also for sera from diabetics. It is clear that a considerable proportion of the total IgG binding to islets is non specific as seen from the pronounced exocrine MIA values (Fig. 2). Thus it is only when the islet-specific MIA values are compared that the diabetic and normal groups can be differentiated (Fig. 4). The method was validated by (0 our findings of the increase in islet-specific MIA values with increasing concentrations of leA positive serum concentration (Fig. 2, sera 1 and 2; Fig. 3); (ii) the leA positivity in all but one sera found positive by the indirect immunofluorescence method (Table 0; and (iii) the absence of leA in 29/30 normal sera (Fig. 4). With regard to serum dilution, a prozone-like effect was occasionally detected (1/12 sera) and this is consistent with previous findings (Olsson et al., 1987; Colman et al., 1988). It is clear that a working dilution of
1/2 is adequate to detect the presence of specific leA in most 100M sera. However, to detect a prozone-like effect it is advisable to determine either a complete dilution curve or more simply test at different dilutions, e.g. 1/2, 1/16 and 1/28. The explanation for the single prozone-like effect observed here is not known, but it cannot be ruled out that with this serum it is due to a nontypical higher background in exocrine compared with islet tissue at low dilution. The JOF standard curves for human and rat pancreas were identical with greatest discrimination within the range 40-80 JOF units. The low assay detection limit of around 5 JOF units resulted in the detection of several weak leA positive sera tested at 1/2 dilution which were not detected by the indirect immunofluorescence method on undiluted sera (Table O. At the same time the number of leA positives in normal sera were low (1/30 human pancreas; 0/19 rat pancreas, Fig. 4). Although a more precise determination of the upper limit of the normal range might have given other positivity values our results suggest that the quantitative immunocytochemical method has a lower assay detection limit compared to the immunofluorescence method. The interassay variation of the method was reasonably low « 10%) although even better reproducibility should be possible by JOF standardization. In the present study, human and rat pancreas were equally effective since the JOF standard curves were identical (Fig. 3) and there was a close correlation between the rat and human pancreas for sera over a wide range of isletspecific MIA values (Fig. 5). However, more studies need to be performed before it can be concluded that human and rat pancreatic islets contain equivalent amounts of the leA antigens. On this question Yamashiro et al. (1990) have observed a greater leA binding to human pancreatic islets compared to rat pancreatic islets for some sera. In conclusion, a sensitive immunocytochemical method to quantify the specific binding of leA antibodies to pancreatic islet antigens has been described. This method may be useful for studies of the blocking of specific leA binding by potential islet antigens.
68
Acknowledgements
We are grateful to Ann Dyreborg Jensen and Steen Kryger for skilled technical assistance and John Villumsen for the statistical analysis. Financial support from Direkt0r Leo Nielsen og Hustru Karen Margrethe Nielsens Legat for Lregevidenskabelig Grundforskning is gratefully acknowledged. References Bonifacio, E., Lemmark, A. and Dawkins, R.L. (1988) Serum exchange and use of dilutions have improved precision of measurement of islet cell antibodies. 1. Immunol. Methods Hl6,83. Bottazzo, G.F., Florin-Christensen, A. and Doniach, D. (1974) Islet-cell antibodies in diabetes mellitus with autoimmune polyendocrine deficiencies. Lancet ii, 1279. Colman, P.G., Di Mario, U., Rabizadeh, A., Dotta, F., Anastasi, E. and Eisenbarth, G.S. (l988b) A prozone phenomenon interferes in islet cell antibody detection: Direct comparison of two methods in subjects at risk of diabetes and in insulin dependent diabetics at onset. 1. Autoimmun. 1,109.
Gorsuch, A.N., Spencer, K.M., Lister, I., McNally, I.M., Dean, B.M., Bottazzo, G.F. and Cudworth, A.G. (1981) Evidence for a long prediabetic period in type I (insulin-dependent) diabetes mellitus. Lancet ii, 1363. H0)'er, P.E., Kayser, L., Barer, M.R. and Lyon, H. (1991) Quantitation in histochemistry. In: H. Lyon (Ed.), Theory and Strategy in Histochemistry. A Guide to the Selection and Understanding of Techniques. Springer Verlag, Berlin, p.397. Lendrum, R., Walker, G., Cudworth, A.G., Theophanides, Pyke, D.A., Bloom, A. and Gamble, D.R. (1976) Islet-cell antibodies in diabetes mellitus. Lancet ii, 1273. Olsson, M.L., Sundkvist, G. and Lemmark, A. (1987) Prolonged incubation in the two-colour immunonuorescence test increases the prevalence and titres of islet cell antibodies in Type 1 (insulin-dependent) diabetes mellitus. Diabetologia 30, 327. Srikanta S., Ganda, O.P., Rabizadeh, A., Soeldner, I.S. and Eisenbarth, G.S. (1985a) First-degree relatives of patients with Type I diabetes mellitus. Islet-cell antibodies and abnormal insulin secretion. New Engl. 1. Med. 313, 461. Srikanta, S., Rabizadeh, A., Omar, M.A.K. and Eisenbarth, G.S. (l985b) Assay for islet cell antibodies. Protein Amonoclonal antibody method. Diabetes 34, 300. Yamashiro, Y., Taniguchi, H., Baba, S., Taniguchi, T., Aono, S., Isshiki, G. and Jinnouchi, T. (1990) Heterogeneity of human islet cell antibodies in terms of cross-species reactivity. Diabetes ~esearch and Clinical Practice 8, 13.
c..
63
JIM 06251
A quantitative immunocytochemical method for the measurement of islet cell cytoplasmic antibodies Michael O. Marshall and Poul E. H0yer Department Endocrinology and Immunology, Novo Nordisk A / S, Bagsvaerd, Denmark, and Department of Medical Anatomy A, Panum Institute, Denmark (Received 3 April 1991, revised received 6 December 1991, accepted 13 December 1991)
A quantitative immunocytochemical method for the measurement of islet cell cytoplasmic antibodies has been developed. The method employs human or rat pancreas, a protein A-peroxidase / diaminobenzidine secondary antibody system and an independent measurement of islet total and exocrine mean integrated absorbance by scanning microdensitometry. Specific islet cell cytoplasmic antibody binding Oslet total-exocrine mean integrated absorbance) was dependent on serum dilution and substrate reaction time. The detection limit was approximately 5 JOF units. Specific islet cell cytoplasmic antibody binding values with human and rat pancreas were similar. Specific islet cell cytoplasmic antibody binding (human pancreas) was greater (p < 0.(01) in sera from patients with newly diagnosed insulin dependent diabetes mellitus (0.119 ± 0.086, n = 29) compared to normal sera (0.003 ± 0.008, n = 29). Thus, the method has been validated and may be useful for measuring the blocking effect of potential antigens on specific islet cell cytoplasmic antibody. Key words: Immunocytochemistry; Islet cell; Cytoplasmic; (Antibody)
Introduction
ICA is frequently present in the sera of patients with 100M both at diagnosis and in the pre-diabetic phase (Bottazzo et aI., 1974; Lendrum et aI., 1976; Gorsuch et aI., 1981; Srikanta et aI., 1985a). The traditional method of detecting these antibodies is by indirect immunofluorescence on cryostat sections of human
Correspondence to: M.O. Marshall, Diabetes Research, Novo Nordisk A/S, 2880 Bagsvaerd, Denmark. Abbreviations: ICA, islet cell cytoplasmic antibody; 100M, insulin dependent diabetes mellitus; BSA, bovine serum albumin; FlTC, fluorescein isothiocyanate; PBS, phosphatebuffered saline.
pancreas (Bottazzo et aI., 1974), in which ICA positivity is based on a subjective judgement of the relative intensities of islet and exocrine fluorescence. The method can be semiquantitated by determining the end point titre of a serum against a JOF standard (Bonifacio et aI., 1988). However, for some experimental studies, in particular when testing the blocking of leA antibody binding by potential auto antigens, a method is required which can quantify the amount of antibody specifically bound to islets. The method of Srikanta et al. (1985b) uses human pancreas and fluorescein conjugated protein A as secondary antibody and photometric measurement of fluorescence intensity. We describe here an alternative, quantitative immunocytochemical method employing either human or rat pancreas, peroxidase-conjugated
64
protein A as secondary antibbdy and measurement of light absorbance with a microdensitometer. Materials and methods
Materials
Peroxidase-conjugated protein A and FITCconjugated rabbit anti-human IgG were purchased from Amersham (Little Chalfont, England) and Dako (Glostrup, Denmark), respectively, and diaminobenzidine was purchased from Sigma (St. Louis, MO, U~A). R2D6 (mouse monoclonal anti-islet ganglioside) was a kind gift from Dr. R. Alexandro, University of Miami, USA. Eukitt was purchased from O. Kindler, Freiburg, Germany.
Sera
Sera from 29 newly diagnosed 100M patients (mean age 19 years, range 4-38 years; eight female, 21 male), were obtained from Hvid0re and Glostrup Hospitals, Denmark, and sera from 30 normal individuals (mean age 25 years, range 7-63 years; 12 female, 18 male) were obtained from laboratory personnel. Sera were absorbed overnight with rat liver acetone powder (20 mg/mI) prior to assay. The JDF standard serum was a gift from Dr. J. Ludvigsson, Linkoping Hospital, Sweden.
Quantitative immunocytochemical method for ICA 4 JLm thick cryostat sections of human pancreas (blood group 0) or Wistar Furth rat pancreas were air dried (30 min), acetone fixed (5 min), and then incubated at 200C for 40 min in a moist chamber with rat liver absorbed sera diluted 1/2 in PBS-l % BSA. After washing with PBS (3 x 5 min), the sections were incubated with peroxidase-conjugated protein A 0/100) for 40 min and again washed. They were then immersed in 250 ml of a solution of diaminobenzidine (I mg/mI), H 202 (0.02%) in PBS for 10 min washed and dehydrated by sequential immersions in ethanol (96%, 3 min; 2 x 99%, 3 min), and dried before mounting with Eukitt medium. Each serum was incubated in duplicate on consecutive sections.
Quantitation of IgG binding. A minimum of 20 areas (707 JLm 2 per area) randomly selected from a minimum of six islets from two consecutive sections on coded slides were scanned using a Vickers M85A scanning and integrating microdensitometer (machine settings: X 40 objective, light of 460 nm, scanning spot diameter of 0.5 JLm). For each area the individual spot readings were integrated and a mean integrated absorbance (MIA) ± SEM over the 20 areas calculated by reference to a standard calibration graph constructed by measuring neutral density filters with known absorbance values. An equal number of similar sized areas were scanned in the exocrine tissue. The exocrine MIA (± SEM) value was taken as a measure of non specific IgG binding to both exocrine and islet tissue. Specific ICA was expressed as islet specific MIA (± SED), calculated as follows: islet specific MIA - islet total MIA - exocrine MIA. Indirect immunofluorescence method for ICA Undiluted sera were tested for ICA positivity using FITC-conjugated rabbit anti-human IgG as secondary antibody, using the same human pancreas and incubation times as described above. Results of the ICA proficiency test no. 2 were: validity (97%), consistency (95%), sensitivity (94%), specificity (100%).
JDF standard curve
Standard sera were prepared by diluting the JDF standard serum (80 JDF units) to 2, 4, 10, 20 and 40 JDF units in normal serum.
Statistics
JDF standard curves were estimated using non-linear regression analysis and diabetic and normal sera were compared using the t test. Human and rat pancreatic samples were compared using the error-in-variables regression method, which takes into account the random variation of the results. Results
Time course of peroxidase reaction The rate of formation of the diaminobenzidine reaction product in both islet and exocrine tissue
0.'
',1- - - -,i'
- - - i- - - - - -i- - - - -
c
i
0.1
.,.0- _ _ _ -
-
-
-0- -
-
-
-
_0-. -
0'
-
-
..... 1
-A
i
OA CO••
i
--0
was initially linear and then decreased until 10 min; after this the product concentration remained constant. An experiment illustrating these kinetics is shown in Fig. 1. The ratio of islet to exocrine remained constant over the whole time period. MIA values are normally measured during the linear phase of an enzyme reaction. However, for peroxidase activity any point on the curve may be selected (H0yer et aI., 1991). In this study a reaction time of 10 min was used for convenience. There was no difference in product formation when the sections were immersed in diaminobenzidine as described in the materials and methods section or when the diaminobenzidine was applied as a drop to the sections (not shown).
Absorbance in islet and exocrine tissue as a function of serum dilution
The dilution curves for three leA positive sera and monoclonal antibody R2D6 are shown in Fig. 2. For sera 1 and 2 the islet specific MIA and exocrine MIA both decreased on serum dilution. Likewise, the binding of R2D6 to the plasma membrane of beta-cells decreased with antibody dilution (Fig. 2). However, for serum 3 the islet specific MIA first increased on diluting to 1/16 and then decreased. Serum 3 was the only one of 12 sera tested showing this prozone-like effect.
JDF standardization and interassay variation With both human and rat pancreas islet specific MIA was proportional to the JDF value of
.....
: ~~~,
" "-
~-
......
0.1
c
i
0 ..
OA 0.1
0 ••
0.1
0.0
--
0.1
-, o •
----
~
0.0 '---~----o • 4 I 'I II 14,.1
0 ..
0.1
,,
0.1
0.0 L-~~~---=-'::::--:....1... o • • I 11 II 1.,.1 . . . . dIuIIoII
(Illinutea I
Fig. 1. Development of the diaminobenzidine reaction product as a function of time. Islet total (l»; islet-specific ( ... ); exocrine (0). The results are from a sil18le experiment with rat pancreas and tile variations are the SEM (islet, exocrine) or SED (islet-specific) of each MIA value.
OA
o.J
0.1
0.0 L~---,-~~-,------'-~""""'~-"--~--' 40 10 eo .0 10 10 o
RMotIon tIIIIe
......
0.1
0.1
0.4
6S
0.0
4 1 ,. II .4 1.1 ..... cIIuIIon
~.......;=-:o:::....;=..",,"--,z--
0
400
laoO
•• 00
An-,. _ _
Fig. 2. Mean integrated absorbance as a function of serum dilution. Symbols as in Fig. 1. Sera were diluted in PBS-l % BSA, and assayed on rat pancreas. R2D6 binding was assayed using peroxidase-conjugated anti-mouse Ig (1/100). The results for each serum are from a single experiment and the variations refer to the SEM (total islet, exocrine) or SED (islet-specific) of each MIA value.
the standard serum over a range of 5-80 JDF units, with the greatest discrimination between 40 and 80 units (Fig. 3). The standard curves of the human and rat systems were identical. The mean 0.2
0.0
b===::::::::~....J---L--.L.--L----'
(01
I.'
1
10
10
40
10
JDF unlta
Fig. 3. Islet-specific mean integrated absorbance as a function of concentration of JDF standard serum. Human pancreas (0); rat pancreas (.). Each curve best fit is the mean ( ± SEM) of four experiments and was calculated from the formula islet-specific MIA = ao + at exp(~ log 2 JDF unit) where ao - mean MIA of the 0 JDF standard, at'" mean MIA of the 80 JDF standard, and fJ - rate of change o~ MIA with JDF unit.
66
assay detection limit from four standard curves (99% upper limit of the zero JDF serum) was 4.2 lOF units (range 2.5-6.2 lDF units) for human pancreas and 5.2 JOF units (range 2.9-8.3 lOF units) for rat pancreas. The inter assay variation for a control serum with 20 lDF units was 8% (n = 6) with human pancreas and 7% (n = 6) with rat pancreas. Comparison of diabetic and normal sera Islet total and islet specific MIA values with human and rat pancreas are shown in Fig. 4. There was considerable overlap in islet total MIA values between the diabetic and normal groups. However, when the islet-specific MIA values were compared the two groups were clearly differentiated. For example, islet-specific MIA (human HUMAN PANCREAS IMt .,.oIfio MIA
0 ••
0 •• 0." 0.2
+-r •
•
I
0."
+ f
0.1 0
0
................. It
-
-.-
-
-+
It 10 I*MtIM ........
10
RAT PANCREAS ....t .,.oIfio MI A
....t total MIA
0••
0 .• 0."
-+-
I
0.2 0 10
+••
,.
I*NIIM ........
0.4 0.2
0
TABLE I ICA POSITIVITY IN SERA FROM NEWLY DIAGNOSED IDDM PATIENTS Human pancreas ICM
+ IOF
+
5 1
_T___ •
+ 3
2 1
5 18
n-29
]4
n-20
Key: ICM - quantitative immunocytochemical method; IDF - indirect immunofluorescence method; +, positive; -. nelative.
pancreas) was O. 119 ± 0.086 (mean ± SD) in the diabetic group and 0.003 ± 0.008 in the normal group (p < 0.001). ICA positivity in the diabetic group (islet-specific MIA values > mean + 3 SO of normals) was 79% and 85% with human and rat pancreas respectively; the corresponding positivity in the normal groups was 3% and 0%, respectively. W,hen the quantitative immunocytochemical and immunofluorescence methods were compared there was agreement regarding ICA positivity for 79% and 80% of the sera tested with human and rat pancreas respectively (Table I). Those sera positive with the quantitative immunocytochemical method but negative with the
I
004
MIA .... -O.OhO.'7MIA_ r • 0.84
• •
0.'
i
• •••
Rat pancreas ICM
•i
0 ••
1
0.1
-+---
1. ................ 10
Fig. 4. Comparison of sera from newly diagnosed diabetic patients and normals. Each serum was assayed once on both human and rat pancreas. - - refers to the mean of each group and ...... to the mean + 3 SO upper limit of normal range. The distnbution of MIA values in each group wu found 10 be normal. Group differences were analysed by Student's t test.
i
0
o
0.1
0••
0.'
004
....t epeoIfto MIA IIum8n ,.,.... Fig. 5. Comparison of human and rat pancreas. 20 sera from newly diagnosed 100M patients were assayed on both human and rat pancreas. The linear regression wu estimated usina the error invariable. rearcssion.
67
immunoflourescence method had low isletspecific MIA values (data not shown). Comparison of human and rat pancreas Islet total MIA values (p < 0.0001} from nor-
mal sera were greater using rat rather than human pancreas (Fig. 4). However, the islet-specific MIA values obtained using pancreatic material from the two species were closely correlated (r = 0.84, p < 0.001) and the estimated regression line was not significantly different from the identity line (p > 0.10) (Fig. 5). Discussion
The immunocytochemical method for leA described here permits the independent measurement of the polymerised oxidation product of diaminobenzidine in islet and exocrine pancreatic tissue. Assuming proportionality between absorbance and the amount of antibody bound, the islet-specific MIA values represent the specific antibody binding to islet antigens (i.e., specific leA). In this calculation the IgG non specific binding to islets is taken to be the same as that of the measured IgG binding to exocrine tissue. This is certainly valid for the normal sera we have studied since the islet-specific MIA values were zero (Fig. 4), and it is assumed to be valid also for sera from diabetics. It is clear that a considerable proportion of the total IgG binding to islets is non specific as seen from the pronounced exocrine MIA values (Fig. 2). Thus it is only when the islet-specific MIA values are compared that the diabetic and normal groups can be differentiated (Fig. 4). The method was validated by (0 our findings of the increase in islet-specific MIA values with increasing concentrations of leA positive serum concentration (Fig. 2, sera 1 and 2; Fig. 3); (ii) the leA positivity in all but one sera found positive by the indirect immunofluorescence method (Table 0; and (iii) the absence of leA in 29/30 normal sera (Fig. 4). With regard to serum dilution, a prozone-like effect was occasionally detected (1/12 sera) and this is consistent with previous findings (Olsson et al., 1987; Colman et al., 1988). It is clear that a working dilution of
1/2 is adequate to detect the presence of specific leA in most 100M sera. However, to detect a prozone-like effect it is advisable to determine either a complete dilution curve or more simply test at different dilutions, e.g. 1/2, 1/16 and 1/28. The explanation for the single prozone-like effect observed here is not known, but it cannot be ruled out that with this serum it is due to a nontypical higher background in exocrine compared with islet tissue at low dilution. The JOF standard curves for human and rat pancreas were identical with greatest discrimination within the range 40-80 JOF units. The low assay detection limit of around 5 JOF units resulted in the detection of several weak leA positive sera tested at 1/2 dilution which were not detected by the indirect immunofluorescence method on undiluted sera (Table O. At the same time the number of leA positives in normal sera were low (1/30 human pancreas; 0/19 rat pancreas, Fig. 4). Although a more precise determination of the upper limit of the normal range might have given other positivity values our results suggest that the quantitative immunocytochemical method has a lower assay detection limit compared to the immunofluorescence method. The interassay variation of the method was reasonably low « 10%) although even better reproducibility should be possible by JOF standardization. In the present study, human and rat pancreas were equally effective since the JOF standard curves were identical (Fig. 3) and there was a close correlation between the rat and human pancreas for sera over a wide range of isletspecific MIA values (Fig. 5). However, more studies need to be performed before it can be concluded that human and rat pancreatic islets contain equivalent amounts of the leA antigens. On this question Yamashiro et al. (1990) have observed a greater leA binding to human pancreatic islets compared to rat pancreatic islets for some sera. In conclusion, a sensitive immunocytochemical method to quantify the specific binding of leA antibodies to pancreatic islet antigens has been described. This method may be useful for studies of the blocking of specific leA binding by potential islet antigens.
68
Acknowledgements
We are grateful to Ann Dyreborg Jensen and Steen Kryger for skilled technical assistance and John Villumsen for the statistical analysis. Financial support from Direkt0r Leo Nielsen og Hustru Karen Margrethe Nielsens Legat for Lregevidenskabelig Grundforskning is gratefully acknowledged. References Bonifacio, E., Lemmark, A. and Dawkins, R.L. (1988) Serum exchange and use of dilutions have improved precision of measurement of islet cell antibodies. 1. Immunol. Methods Hl6,83. Bottazzo, G.F., Florin-Christensen, A. and Doniach, D. (1974) Islet-cell antibodies in diabetes mellitus with autoimmune polyendocrine deficiencies. Lancet ii, 1279. Colman, P.G., Di Mario, U., Rabizadeh, A., Dotta, F., Anastasi, E. and Eisenbarth, G.S. (l988b) A prozone phenomenon interferes in islet cell antibody detection: Direct comparison of two methods in subjects at risk of diabetes and in insulin dependent diabetics at onset. 1. Autoimmun. 1,109.
Gorsuch, A.N., Spencer, K.M., Lister, I., McNally, I.M., Dean, B.M., Bottazzo, G.F. and Cudworth, A.G. (1981) Evidence for a long prediabetic period in type I (insulin-dependent) diabetes mellitus. Lancet ii, 1363. H0)'er, P.E., Kayser, L., Barer, M.R. and Lyon, H. (1991) Quantitation in histochemistry. In: H. Lyon (Ed.), Theory and Strategy in Histochemistry. A Guide to the Selection and Understanding of Techniques. Springer Verlag, Berlin, p.397. Lendrum, R., Walker, G., Cudworth, A.G., Theophanides, Pyke, D.A., Bloom, A. and Gamble, D.R. (1976) Islet-cell antibodies in diabetes mellitus. Lancet ii, 1273. Olsson, M.L., Sundkvist, G. and Lemmark, A. (1987) Prolonged incubation in the two-colour immunonuorescence test increases the prevalence and titres of islet cell antibodies in Type 1 (insulin-dependent) diabetes mellitus. Diabetologia 30, 327. Srikanta S., Ganda, O.P., Rabizadeh, A., Soeldner, I.S. and Eisenbarth, G.S. (1985a) First-degree relatives of patients with Type I diabetes mellitus. Islet-cell antibodies and abnormal insulin secretion. New Engl. 1. Med. 313, 461. Srikanta, S., Rabizadeh, A., Omar, M.A.K. and Eisenbarth, G.S. (l985b) Assay for islet cell antibodies. Protein Amonoclonal antibody method. Diabetes 34, 300. Yamashiro, Y., Taniguchi, H., Baba, S., Taniguchi, T., Aono, S., Isshiki, G. and Jinnouchi, T. (1990) Heterogeneity of human islet cell antibodies in terms of cross-species reactivity. Diabetes ~esearch and Clinical Practice 8, 13.
c..
