Neurotoxicologyand Teratology, Vol. 13, pp. 275-281. Pergamon Press plc, 1991. Printed in the U.S.A.
0892-0362/91 $3.00 + .00
Quantification of Glial Fibrillary Acidic Protein: Comparison of Slot-Immunobinding Assays With a Novel Sandwich ELISA J A M E S P. O ' C A L L A G H A N
Neurotoxicology Division (MD-74B), Health Effects Research Laboratory U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 R e c e i v e d 10 January 1991 O'CALLAGHAN, J. P. Quantification of glial fibriUary acidic protein: Comparison of slot-immunobinding assays with a novel sandwich ELISA. NEUROTOXICOL TERATOL 13(3) 275-281, 1991,--Detailed protocols are presented for assaying glial fibrillary acidic protein (GFAP), an astrocyte localized protein which serves as a quantitative marker of toxicant-induced injury to the central nervous system. Two different solid-phase assay procedures are described: 1) a nitrocellulose based slot-immunobinding assay and 2) a novel microtiter plate based sandwich ELISA. The performance of both assays was assessed by measuring the content of GFAP in homogenates of specific regions of the rat brain and in homogenates of brain regions damaged by the prototype neurotoxicants, trimethyltin (TMT) and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). Both procedures gave similar results that were consistent with previously published observations. By comparing the simplicity, cost effectiveness, safety and speed of the two methods, it appears likely that the sandwich ELISA has several advantages over slot-immunobinding assays. Glial fibrillary acidic protein
Slot-immunobinding assays
ELISA
DIVERSE injuries to the central nervous system result in proliferation and hypertrophy of astrocytes (5,9). The hallmark of this response ~is enhanced expression of the major intermediate filament protein of astrocytes, glial fibrillary acidic protein (GFAP) [for a review, see (5)]. These observations suggested that GFAP may be a useful biochemical indicator of neurotoxicity (10). For the past several years we have investigated this possibility by measuring the content of GFAP in brain samples obtained from experimental animals that had been exposed to prototype neurotoxicants. In aggregate, the results of these studies show that a large variety of toxic insults cause an increase in GFAP; moreover, these effects are dose-related, they correspond to regions of neural damage and they can be demonstrated in the absence of overt pathology [for reviews, see (13, 14, 19)]. Validation of GFAP as an indicator of neurotoxicity required a method for assaying the amount of this protein in nervous tissue. Thus, as an initial step, we developed a detergent-based, solid-phase radioimmunoassay for GFAP (11). The original protocol chosen was essentially the "dot-immunobinding" procedure described by Jahn et al. (7). Over the past few years, we have made several improvements to the original assay, most notably, the replacement of manual "dotting" with template-guided spotting using a slot-blot manifold (2). The specifics of these modifications have been described in a number of publications (2, 12, 17). H o w e v e r , a complete protocol for the G F A P slotimmunobinding assay has not been published apart from the descriptions that have already appeared in the Method sections of the aforementioned publications. Therefore, the first purpose of
this paper is to present a detailed protocol for the GFAP assay that incorporates all the improvements developed to date. A description of the slot-immunobinding assay also will appear as part of the U.S. EPA Neurotoxicity Testing Guidelines (22). The currently used assay for GFAP, while simple to perform and useful for assaying large numbers of samples, has a number of drawbacks. These include the requirement for using a radioactive detection reagent (1251 Protein A), the cost associated with the amount of reagents used, and the time needed to perform the assay. With these problems in mind, a microtiter plate based, sandwich format, enzyme-linked immunosorbant assay (ELISA) for GFAP has been developed. The second purpose of this paper is to provide a protocol for this sandwich ELISA and compare its performance to the slot-immunobinding method. METHOD
Materials Miscellaneous. Trimethyltin hydroxide of defined potency [bottle No. 3; see (15)] was obtained from K & K Laboratories (a division of ICN Pharmaceuticals, Plainview, NY). 1-Methyl-4phenyl-l,2,3,6-tetrahydropyridine hydrochloride was obtained from Aldrich (Milwaukee, WI). Sodium dodecyl sulfate (SDS) (for preparing tissue homogenates and for gel electrophoresis) and all other electrophoresis reagents were purchased from Bio Rad Laboratories (Richmond, CA). BCA protein assay reagent was purchased from Pierce Chemical Co. (Rockford, IL). Bovine se-
275
276
rum albumin was obtained from Sigma Chemical Company (St. Louis, MO). Slot-immunobinding assay. Nitrocellulose paper (0.1 or 0.2 ~xm porosity), in roll form, was purchased from Schleicher and Schuell (Keene, NH) as was the slot-blot manifold (Minifold II). Isopropanol, acetic acid and all salts used in the sample buffer were from Fisher Scientific (Fair Lawn, NJ). HEPES was obtained from Calbiochem (La Jolla, CA). Triton X-100 and gelatin of EIA purity were obtained from Bio Rad Laboratories (Richmond, CA). Rabbit anti-bovine GFAP (Cat. No. Z 334, lot No. 119) and rabbit anti-mouse immunoglobulins (Cat. No. Z 109, lot No. 069) were obtained from Dako Corporation (Carpenteria, CA). Monoclonal anti-porcine GFAP (Cat. No. 814-369) was purchased from Boehringer Mannheim (Indianapolis, IN). [125I]rProtein A (2-10 ixCi/Ixg; 1 Ci=37 GBq) was purchased from New England Nuclear (Boston, MA). Sandwich ELISA. Flat-bottomed microtiter plates (Immulon 2) were purchased from Dynatech Laboratories (Chantilly, VA). Phosphate-buffered saline was made from reagents obtained from Fisher Scientific (Fair Lawn, NJ). Triton X-100 was purchased from Bio Rad. Tween 20 was obtained from Sigma Chemical Co. (St. Louis, MO). Carnation brand nonfat dry milk was purchased from a local grocer. Anti-GFAP antibodies were the same as those employed for the slot-immunobindingassay. Alkaline phosphatase conjugated rabbit anti-mouse IgG (Cat. No. 315-055-003) was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). An alkaline phosphatase substrate kit (P-nitrophenylphosphate) was purchased from Bio Rad Laboratories (Richmond, CA).
Purification of GFAP The GFAP standard was prepared from rat spinal cord according to the method of Eng and DeArmond (6). Purification of GFAP to homogeneity was achieved by preparative polyacrylamide gel electrophoresis (model 1100PG, Bethesda Research Laboratories, Gaithersburg, MD). Standard purity was verified by analytical SDS polyacrylamide gel electrophoresis. The concentration of SDS in the resolving gel was 10% and the GFAP band was visualized by staining with Coomassie Blue R250.
GFAP Immunoblots The specificity of the GFAP antibodies was verified by immunoblotting (21) using the modification of Burnette (3).
Toxicant Administration Damage to striatal dopaminergic fibers was induced by a single SC injection of MPTP (12.5 mg/kg) to 4-6-week-old C57BL/6J female mice (Jackson Laboratories, Bar Harbor, ME) [see (17,18)]. Damage to hippocampal pyramidal neurons was induced by a single IV injection of TMT (8.0 mg/kg) to 6-weekold Long-Evans male rats (Charles River Laboratories, Raleigh, NC) (2). The dosages of both toxicants are expressed as the free base. Mice and rats were killed 2 and 21 days postdosing, respectively; the increase in GFAP caused by MPTP and TMT is maximal at these time points.
Tissue Preparation Following decapitation, brains immediately were removed and placed on a cold plate (Model TCP-2, Thermoelectrics Unlimited, Wilmington, DE). With the aid of curved forceps, striatum
O'CALLAGHAN
was dissected from mice and hippocampus was dissected from rats. Brain parts were weighed, homogenized with a sonic probe (Model XL 2005, Heat Systems, Inc., Farmingdale, NY) in 10 volumes of hot (90-95°C) 1% (w/v) SDS and stored frozen at -70°C before radioimmunoassay. Tissue can be stored frozen prior to homogenization as long as the eventual homogenization step is carried out in hot SDS. In lieu of sonification, homogenization can be achieved with a motor-driven pestle and homogenizing vessel.
Total Protein Assay Total protein in the SDS homogenates was assayed by the method of Smith et al. (20). Bovine serum albumin was used as the standard.
Preparation of Standard Curve Both the slot-immunobinding procedure and the sandwich ELISA can be performed without a pure standard. Standard curves are simply constructed from dilutions of a single control sample. By comparing the immunoreactivity of individual samples (both control and treated groups) with that of the sample used to generate the standard curve, the relative immunoreactivity of each sample is obtained. The immunoreactivity of the control group is normalized to 100% and all data are then expressed as a percentage of control. This approach not only avoids the need to purify GFAP, it serves to normalize differences in antigen-antibody binding that are often observed when comparing assays of pure antigen to assays of antigen contained in a heterogenous mixture, such as a brain homogenate. In my laboratory, a standard was prepared by pooling several homogenates of hippocampus, aliquots of which are stored frozen for use on the day of assay. The concentration of GFAP in this homogenate was determined by comparing the immunoreactivity of a single aliquot with that of purified GFAP.
Slot-Immunobinding Protocol A schematic representation of the GFAP slot-immunobinding assays is shown in Fig. 1. These assays are based on the propensity of proteins to bind to nitrocellulose paper [see Jahn et al., (7)]. Following sample application through a slotted template, unbound sites are blocked and the nitrocellulose sheets then are exposed to polyclonal or monoclonal anti-GFAP antibodies. In the standard assay, the amount of bound antibody and, therefore, the amount of GFAP is determined by binding 125I Protein A to the anti-GFAP antibodies. In the monoclonal antibody based assay, Protein A binding is achieved via a link antibody. This is necessary because the monoclonal anti-GFAP we use is an IgG1 subclass with low inherent affinity for Protein A. In either assay the amount of bound Protein A is quantified by gamma spectrometry. The step-by-step procedure follows: 1. Dilute 1% SDS samples in sample buffer (120 mM KC1/20 mM NaC1/2 mM NaHCO3/2 mM MgC12/5 mM HEPES, pH 7.4/0.7% Triton X-100/0.2% NaN3) to a concentration of 0.25 mg/ml (dilutions may be stored frozen at -70°C). Prepare standard curve samples in sample buffer at dilutions between 1.0 and 10 p~g total protein/20 txl (e.g., 1.25, 2.5, 3.75, 5.0, 6.25, 7.5, 8.75 and 10 p,g/20 p~l). 2. Immerse nitrocellulose sheets (0.1-0.2 p~m porosity) in distilled water for 5 minutes and hang up to dry. 3. Place dried nitrocellulose sheets over 2 pieces of filter paper
IMMUNOASSAYS FOR GFAP
277
Slot-lmmunobinding Standard
ELISA
Monoclonals
Direct
Sandwich ..,
~25[e
• .'" ••
~
Alkaline Phos hatase
Alkaline
S
p
/
V
Nitrocellulose Sheets
Microtiter Plate
Wells
125~=1251
Protein A
,~ = Rabbit ~ G F A P
~, = M o u s e c~ G F A P ?, = G o a t c( R a b b i t I~G ~. = R a b b i t c~ M o u s e I~ ,~ = R a b b i t (~ G F A P
FIG. 1. Schematic diagram of the standard (rabbit polyclonalanti-GFAP) and the monoclonal anti-GFAP based slot immunobindingassays for GFAP. and clamp all three sheets between the plastic blocks of a slotblot manifold (Minifold II or equivalent). Do not use manifolds with " O " rings as they will leak one sample into another due to the detergent in the samples. Do not wet the nitrocellulose or filter paper prior to sample application or the assay will not work. 4. Pipette 20 Vd aliquots of samples and standards into the slots of the slot-blot manifold. Typically, standard curves are run in duplicate; one curve is spotted at the top of the nitrocellulose sheet, while the second curve is spotted at the bottom. The test samples are always spotted as singlets because duplicates rarely vary by more than 5%. Wait 15 minutes after application of the last sample prior to unclamping the manifold. This permits uniform binding of the samples to the nitrocellulose sheets. Immediately after unclamping the nitrocellulose sheets, use a soft pencil to create a grid that brackets the spotted samples; this permits the immunoreactive "slots" to be identified at the end of the assay. 5. Air-, blow-, or oven-dry (60°C) the spotted nitrocellulose sheets. 6. Incubation of spotted sheets: All steps are done at room temperature on a flat reciprocating shaker (one complete excursion every 2-3 seconds) in plastic trays (e.g., Rubbermaid No. 2915) that approximate the size of the spotted sheets. For best results do not use rocking or orbital shakers. Orient the incubation trays parallel to the direction of the moving platform. Perpendicular orientation does not provide adequate exposure of the spotted samples to the solutions of assay reagents. Perform the following steps in enough solution to cover the nitrocellulose sheets to a depth of 1.0 cm: a.
b.
c.
Fix the proteins by laying the spotted sheets onto a solution of 25% (v/v) isopropanol, 10% (v/v) acetic acid. Do not pour the fixing solution onto the spotted sheets; this wrinkles the sheets and results in uneven exposure to subsequent assay reagents. Incubate 15 minutes. Discard fix, wash 4--5 times in deionized water to remove residual fix and incubate in Tris-buffered saline (TBS) (200 mM NaC1, 50 mM Tris-Cl, pH 7.4) for 5 minutes. Discard TBS and incubate 1 hour in blocking solution [0.5%
= M o u s e (x G F A P ]., = R a b b i t ~ Mouse I=G
FIG. 2. Schematic diagram of direct and sandwichELISAs for GFAP.
d.
e. f. g.
h.
gelatin (w/v) in TBS]. Discard blocking solution and incubate in antibody solution (polyclonal or monoclonal anti-GFAP, 1:500, in blocking solution containing 0.1% Triton X-100) for 2 hours. Discard antibody solution and wash 5 × 5 minutes in TBS. Discard TBS and incubate in blocking solution for 30 minutes. If a polyclonal antibody was used in step 4, discard blocking solution and incubate in Protein A solution for 1 hour. Protein A solution contains t2Sl-labeled Protein A diluted in blocking solution containing 0.1% Triton X-100 to 2000 CPM/10 ixl. If a monoclonal was used in step 4, follow step 6 with incubation in antibody solution containing rabbit anti-mouse immunoglobulins, 1:500 for 1 hour; then repeat steps 5 and 6 before incubating in Protein A. Recover Protein A solution (it can be saved and reused 1-2 times) and wash in TBS/0.1% Triton X-100, 4 × 5 minutes each wash followed by 4 × for several hours each wash. An overnight wash in a larger volume can be used to replace the last 4 washes.
7. Air-dry the nitrocellulose sheets, cut out the slots delineated by the penciled grid and count radioactivity in a gamma counter.
Sandwich ELISA Protocol
A schematic representation of direct and sandwich ELISAs for GFAP is shown in Fig. 2. The direct ELISA (Fig. 2, left) is based on the binding of homogenate proteins to the plastic wells of microtiter plates. After blocking unbound sites, the GFAP bound to the wells is detected by an antibody directed against GFAP. This antibody in turn is bound by an enzyme linked antibody directed toward the first antibody. Quantification is achieved by addition of a substrate for the antibody bound enzyme followed by spectrophotometry of the colored reaction product as determined in a microtiter plate reader. The speed, simplicity and cost-effective nature of direct ELISAs would appear to make them the method of choice for assaying GFAP. Unfortunately, we have not found them to be reliable for assaying GFAP in detergentbased samples or aqueous extracts of brain homogenates (see the Results section).
278
O'CALLAGHAN
The sandwich ELISA (Fig. 2, right) is based on binding of anti-GFAP antibody to the plastic wells of microtiter plates. This antibody, rather than the wells of the plate, serves as the receptor or "capture" reagent for binding GFAP. After blocking unbound sites, GFAP in brain homogenates is bound to the capture antibody coating the wells. The GFAP bound to the capture antibody then is bound by a second antibody (from another species). This second antibody is then bound by an enzyme-linked antibody directed toward it but not toward the capture antibody. Quantification is achieved as with the direct ELISA. The step-bystep procedure follows: 1. Dilute 1% SDS samples in sample buffer [PBS (137 mM NaC1/2.7 mM KC1/I.4 mM KH2PO4/8.0 mM Na2HPO4.7 H20/pH 7.4) containing 0.5% Triton X-100] to a concentration of 10 Ixg/ml. Samples high in GFAP (e.g., cerebellum) may need to be diluted to 5 Ixg/ml to avoid readings above the linear portion of the standard curve. Likewise, samples low in GFAP (e.g., striatum) may need to be diluted to only 20 txg/ml to avoid readings below the linear portion of the standard curve. Prepare standard curve samples in sample buffer at dilutions between 0.05 and 10 i~g/100 ILl (e.g., 0.05, 0.1, 0.25, 0.5, 1.0, 5.0, and 10 Ixg/100 pAL 2. Coat Immulon-2 flat-bottom microtiter plates with an lgG fraction of a polyclonal anti-GFAP, 1.0 jxg total immunoglobulin protein per 100 jxl, per well, diluted in PBS. Incubate 1-2 h at 37°C and then overnight at 4°C. All other incubations are at room temperature with moderate shaking on a reciprocating shaker. 3. Wash 4 x with PBS, 200 ixl/well. 4. Block 1 h with BLOTTO (8) [5% (w/v) nonfat dry milk in PBS], 100 ~l/well. 5. Incubate standards and samples in duplicate for 1 h, 100 pA/ well. 6. Wash 4 × with PBS containing 0.5% Triton X-100, 200 Ixl/ well. 7. Incubate 1 h with monoclonal anti-GFAP (Boehringer Mannheim), 1:500, made up in BLOTTO containing 0.5% Triton X-100, 100 I,d/well. 8. Wash 4 × with PBS containing 0.5% Triton X-100, 200 tzl/ well. 9. Incubate 30 minutes with alkaline phosphatase conjugated anti-mouse IgG (Jackson ImmunoResearch), 1:3000, made up in BLOTTO containing 0.5% Triton X-100, 100 jxl/well. 10. Wash 4 × with PBS containing 0.5% Triton X-100,200 ~zl/ well. 11. Add 100 Ixl of P-nitrophenylphosphate substrate (Bio Rad) and stop reaction 10 minutes later with 100 i~1 of 0.4 N NaOH. 12. Read at 405 nm. We used a Molecular Devices UV Max Microplate Reader (Menlo Park, CA) coupled to a Macintosh computer running a Soft Max (Molecular Devices, Menlo Park, CA) program. RESULTS
Standard Purity and Antibody Specificity Protein staining of GFAP purified by preparative PAGE revealed a band of approximately 50,000 daltons (Fig. 3, left side). Immunoblots of this band using polyclonal anti-GFAP antibodies verified that it was authentic GFAP (data not shown). This preparation of GFAP was used as the standard in a slot-immunobinding assay (standard format) to determine the concentration of GFAP in a homogenate of rat hippocampus. Aliquots of this homogenate were then used as a standard in all subsequent assays
GFAP Standard
GFAP Immunoblots Polyclonal Monoclonal
I 0
×
100
50
W
l
m
0 G)
m
0
20 FIG. 3. Protein staining of the GFAP standard and GFAP immunoblots of total hippocampalhomogenateproteinusingpolyclonaland monoclonal anti-GFAP. The GFAP standard (approximately0.25 Ixg) and the hippocampal homogenate(10 ixg)were separatedby SDS polyacrylamide(10%) gel electrophoresis. The standard was stainedwith Coomassie Blue R250, The homogenate proteins were transferred to nitrocellulose paper and probed with the appropriate antibody followingthe procedures described in the Method section.
(slot-immunobindingand ELISA). Immunoblots of this hippocampus homogenate showed that both polyclonal and monoclonal anti-GFAP detected only a single 50,000 dalton band (Fig. 3, right side). From time to time, however, I have seen evidence of slight cross reactivity with a 60-70 kilodalton band (neurofilament 68?) using polyclonal anti-GFAP.
Slot-lmmunobinding Assay Standard Curves The amount of radioactivity bound in the standard and monoclonal-based slot-immunobinding assays, as a function of total homogenate protein and of GFAP contained in the homogenate, is shown in Fig. 4. Both polyclonal- and monoclonal-antibody based assays showed a linear relationship between bound counts and the amount of total protein (and GFAP) spotted, The assay is linear up to at least 15 Ixg total protein (--60 ng GFAP). Spotting greater than 15 jxg total protein exceeds the binding capacity of the nitrocellulose and results in a loss of linearity. At the antibody dilutions used (1:500), the amount of radioactivity bound in the presence of 1.25 Ixg total protein was 8 and 2 times blank for polyclonal and monoclonal antibodies, respectively. The polyclonalbased assay is at least sensitive enough to detect 1 ng GFAP in 0.5 Ixg total hippocampal protein. As has been shown for other antibodies (7), decreasing the concentration of the GFAP antibodies (monoclonal or polyclonal) decreases the sensitivity of the assay (data not shown).
Direct ELISA Standard Curves The SDS-solubilized hippocampal homogenate standard, when diluted in PBS or even low concentrations of Tween 20 or Triton X-100, appeared to bind to the microtiter plate wells (data not shown). In my hands, however, the binding curves were never reproducible. Moreover, samples containing large toxicant-in-
IMMUNOASSAYS FOR GFAP
279
20
i
'
'
P;lyclonalAnti- ' G ~
i
i
Triton X-100/=
1.4
*
1.2
15
E ¢o v
o
5
i
__
i
Tween 20
~, 0.5°/' I 0.75%
1.0 0.8 0.6 0.4 0.2 0.0
~ 1~0
I
I
t
I
I
I
I
I
I
[
I
i I I I I I O ,ID 0.4 ~Without Capture Antibody_ _Without Capture Antibody.
0892-0362/91 $3.00 + .00
Quantification of Glial Fibrillary Acidic Protein: Comparison of Slot-Immunobinding Assays With a Novel Sandwich ELISA J A M E S P. O ' C A L L A G H A N
Neurotoxicology Division (MD-74B), Health Effects Research Laboratory U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 R e c e i v e d 10 January 1991 O'CALLAGHAN, J. P. Quantification of glial fibriUary acidic protein: Comparison of slot-immunobinding assays with a novel sandwich ELISA. NEUROTOXICOL TERATOL 13(3) 275-281, 1991,--Detailed protocols are presented for assaying glial fibrillary acidic protein (GFAP), an astrocyte localized protein which serves as a quantitative marker of toxicant-induced injury to the central nervous system. Two different solid-phase assay procedures are described: 1) a nitrocellulose based slot-immunobinding assay and 2) a novel microtiter plate based sandwich ELISA. The performance of both assays was assessed by measuring the content of GFAP in homogenates of specific regions of the rat brain and in homogenates of brain regions damaged by the prototype neurotoxicants, trimethyltin (TMT) and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). Both procedures gave similar results that were consistent with previously published observations. By comparing the simplicity, cost effectiveness, safety and speed of the two methods, it appears likely that the sandwich ELISA has several advantages over slot-immunobinding assays. Glial fibrillary acidic protein
Slot-immunobinding assays
ELISA
DIVERSE injuries to the central nervous system result in proliferation and hypertrophy of astrocytes (5,9). The hallmark of this response ~is enhanced expression of the major intermediate filament protein of astrocytes, glial fibrillary acidic protein (GFAP) [for a review, see (5)]. These observations suggested that GFAP may be a useful biochemical indicator of neurotoxicity (10). For the past several years we have investigated this possibility by measuring the content of GFAP in brain samples obtained from experimental animals that had been exposed to prototype neurotoxicants. In aggregate, the results of these studies show that a large variety of toxic insults cause an increase in GFAP; moreover, these effects are dose-related, they correspond to regions of neural damage and they can be demonstrated in the absence of overt pathology [for reviews, see (13, 14, 19)]. Validation of GFAP as an indicator of neurotoxicity required a method for assaying the amount of this protein in nervous tissue. Thus, as an initial step, we developed a detergent-based, solid-phase radioimmunoassay for GFAP (11). The original protocol chosen was essentially the "dot-immunobinding" procedure described by Jahn et al. (7). Over the past few years, we have made several improvements to the original assay, most notably, the replacement of manual "dotting" with template-guided spotting using a slot-blot manifold (2). The specifics of these modifications have been described in a number of publications (2, 12, 17). H o w e v e r , a complete protocol for the G F A P slotimmunobinding assay has not been published apart from the descriptions that have already appeared in the Method sections of the aforementioned publications. Therefore, the first purpose of
this paper is to present a detailed protocol for the GFAP assay that incorporates all the improvements developed to date. A description of the slot-immunobinding assay also will appear as part of the U.S. EPA Neurotoxicity Testing Guidelines (22). The currently used assay for GFAP, while simple to perform and useful for assaying large numbers of samples, has a number of drawbacks. These include the requirement for using a radioactive detection reagent (1251 Protein A), the cost associated with the amount of reagents used, and the time needed to perform the assay. With these problems in mind, a microtiter plate based, sandwich format, enzyme-linked immunosorbant assay (ELISA) for GFAP has been developed. The second purpose of this paper is to provide a protocol for this sandwich ELISA and compare its performance to the slot-immunobinding method. METHOD
Materials Miscellaneous. Trimethyltin hydroxide of defined potency [bottle No. 3; see (15)] was obtained from K & K Laboratories (a division of ICN Pharmaceuticals, Plainview, NY). 1-Methyl-4phenyl-l,2,3,6-tetrahydropyridine hydrochloride was obtained from Aldrich (Milwaukee, WI). Sodium dodecyl sulfate (SDS) (for preparing tissue homogenates and for gel electrophoresis) and all other electrophoresis reagents were purchased from Bio Rad Laboratories (Richmond, CA). BCA protein assay reagent was purchased from Pierce Chemical Co. (Rockford, IL). Bovine se-
275
276
rum albumin was obtained from Sigma Chemical Company (St. Louis, MO). Slot-immunobinding assay. Nitrocellulose paper (0.1 or 0.2 ~xm porosity), in roll form, was purchased from Schleicher and Schuell (Keene, NH) as was the slot-blot manifold (Minifold II). Isopropanol, acetic acid and all salts used in the sample buffer were from Fisher Scientific (Fair Lawn, NJ). HEPES was obtained from Calbiochem (La Jolla, CA). Triton X-100 and gelatin of EIA purity were obtained from Bio Rad Laboratories (Richmond, CA). Rabbit anti-bovine GFAP (Cat. No. Z 334, lot No. 119) and rabbit anti-mouse immunoglobulins (Cat. No. Z 109, lot No. 069) were obtained from Dako Corporation (Carpenteria, CA). Monoclonal anti-porcine GFAP (Cat. No. 814-369) was purchased from Boehringer Mannheim (Indianapolis, IN). [125I]rProtein A (2-10 ixCi/Ixg; 1 Ci=37 GBq) was purchased from New England Nuclear (Boston, MA). Sandwich ELISA. Flat-bottomed microtiter plates (Immulon 2) were purchased from Dynatech Laboratories (Chantilly, VA). Phosphate-buffered saline was made from reagents obtained from Fisher Scientific (Fair Lawn, NJ). Triton X-100 was purchased from Bio Rad. Tween 20 was obtained from Sigma Chemical Co. (St. Louis, MO). Carnation brand nonfat dry milk was purchased from a local grocer. Anti-GFAP antibodies were the same as those employed for the slot-immunobindingassay. Alkaline phosphatase conjugated rabbit anti-mouse IgG (Cat. No. 315-055-003) was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). An alkaline phosphatase substrate kit (P-nitrophenylphosphate) was purchased from Bio Rad Laboratories (Richmond, CA).
Purification of GFAP The GFAP standard was prepared from rat spinal cord according to the method of Eng and DeArmond (6). Purification of GFAP to homogeneity was achieved by preparative polyacrylamide gel electrophoresis (model 1100PG, Bethesda Research Laboratories, Gaithersburg, MD). Standard purity was verified by analytical SDS polyacrylamide gel electrophoresis. The concentration of SDS in the resolving gel was 10% and the GFAP band was visualized by staining with Coomassie Blue R250.
GFAP Immunoblots The specificity of the GFAP antibodies was verified by immunoblotting (21) using the modification of Burnette (3).
Toxicant Administration Damage to striatal dopaminergic fibers was induced by a single SC injection of MPTP (12.5 mg/kg) to 4-6-week-old C57BL/6J female mice (Jackson Laboratories, Bar Harbor, ME) [see (17,18)]. Damage to hippocampal pyramidal neurons was induced by a single IV injection of TMT (8.0 mg/kg) to 6-weekold Long-Evans male rats (Charles River Laboratories, Raleigh, NC) (2). The dosages of both toxicants are expressed as the free base. Mice and rats were killed 2 and 21 days postdosing, respectively; the increase in GFAP caused by MPTP and TMT is maximal at these time points.
Tissue Preparation Following decapitation, brains immediately were removed and placed on a cold plate (Model TCP-2, Thermoelectrics Unlimited, Wilmington, DE). With the aid of curved forceps, striatum
O'CALLAGHAN
was dissected from mice and hippocampus was dissected from rats. Brain parts were weighed, homogenized with a sonic probe (Model XL 2005, Heat Systems, Inc., Farmingdale, NY) in 10 volumes of hot (90-95°C) 1% (w/v) SDS and stored frozen at -70°C before radioimmunoassay. Tissue can be stored frozen prior to homogenization as long as the eventual homogenization step is carried out in hot SDS. In lieu of sonification, homogenization can be achieved with a motor-driven pestle and homogenizing vessel.
Total Protein Assay Total protein in the SDS homogenates was assayed by the method of Smith et al. (20). Bovine serum albumin was used as the standard.
Preparation of Standard Curve Both the slot-immunobinding procedure and the sandwich ELISA can be performed without a pure standard. Standard curves are simply constructed from dilutions of a single control sample. By comparing the immunoreactivity of individual samples (both control and treated groups) with that of the sample used to generate the standard curve, the relative immunoreactivity of each sample is obtained. The immunoreactivity of the control group is normalized to 100% and all data are then expressed as a percentage of control. This approach not only avoids the need to purify GFAP, it serves to normalize differences in antigen-antibody binding that are often observed when comparing assays of pure antigen to assays of antigen contained in a heterogenous mixture, such as a brain homogenate. In my laboratory, a standard was prepared by pooling several homogenates of hippocampus, aliquots of which are stored frozen for use on the day of assay. The concentration of GFAP in this homogenate was determined by comparing the immunoreactivity of a single aliquot with that of purified GFAP.
Slot-Immunobinding Protocol A schematic representation of the GFAP slot-immunobinding assays is shown in Fig. 1. These assays are based on the propensity of proteins to bind to nitrocellulose paper [see Jahn et al., (7)]. Following sample application through a slotted template, unbound sites are blocked and the nitrocellulose sheets then are exposed to polyclonal or monoclonal anti-GFAP antibodies. In the standard assay, the amount of bound antibody and, therefore, the amount of GFAP is determined by binding 125I Protein A to the anti-GFAP antibodies. In the monoclonal antibody based assay, Protein A binding is achieved via a link antibody. This is necessary because the monoclonal anti-GFAP we use is an IgG1 subclass with low inherent affinity for Protein A. In either assay the amount of bound Protein A is quantified by gamma spectrometry. The step-by-step procedure follows: 1. Dilute 1% SDS samples in sample buffer (120 mM KC1/20 mM NaC1/2 mM NaHCO3/2 mM MgC12/5 mM HEPES, pH 7.4/0.7% Triton X-100/0.2% NaN3) to a concentration of 0.25 mg/ml (dilutions may be stored frozen at -70°C). Prepare standard curve samples in sample buffer at dilutions between 1.0 and 10 p~g total protein/20 txl (e.g., 1.25, 2.5, 3.75, 5.0, 6.25, 7.5, 8.75 and 10 p,g/20 p~l). 2. Immerse nitrocellulose sheets (0.1-0.2 p~m porosity) in distilled water for 5 minutes and hang up to dry. 3. Place dried nitrocellulose sheets over 2 pieces of filter paper
IMMUNOASSAYS FOR GFAP
277
Slot-lmmunobinding Standard
ELISA
Monoclonals
Direct
Sandwich ..,
~25[e
• .'" ••
~
Alkaline Phos hatase
Alkaline
S
p
/
V
Nitrocellulose Sheets
Microtiter Plate
Wells
125~=1251
Protein A
,~ = Rabbit ~ G F A P
~, = M o u s e c~ G F A P ?, = G o a t c( R a b b i t I~G ~. = R a b b i t c~ M o u s e I~ ,~ = R a b b i t (~ G F A P
FIG. 1. Schematic diagram of the standard (rabbit polyclonalanti-GFAP) and the monoclonal anti-GFAP based slot immunobindingassays for GFAP. and clamp all three sheets between the plastic blocks of a slotblot manifold (Minifold II or equivalent). Do not use manifolds with " O " rings as they will leak one sample into another due to the detergent in the samples. Do not wet the nitrocellulose or filter paper prior to sample application or the assay will not work. 4. Pipette 20 Vd aliquots of samples and standards into the slots of the slot-blot manifold. Typically, standard curves are run in duplicate; one curve is spotted at the top of the nitrocellulose sheet, while the second curve is spotted at the bottom. The test samples are always spotted as singlets because duplicates rarely vary by more than 5%. Wait 15 minutes after application of the last sample prior to unclamping the manifold. This permits uniform binding of the samples to the nitrocellulose sheets. Immediately after unclamping the nitrocellulose sheets, use a soft pencil to create a grid that brackets the spotted samples; this permits the immunoreactive "slots" to be identified at the end of the assay. 5. Air-, blow-, or oven-dry (60°C) the spotted nitrocellulose sheets. 6. Incubation of spotted sheets: All steps are done at room temperature on a flat reciprocating shaker (one complete excursion every 2-3 seconds) in plastic trays (e.g., Rubbermaid No. 2915) that approximate the size of the spotted sheets. For best results do not use rocking or orbital shakers. Orient the incubation trays parallel to the direction of the moving platform. Perpendicular orientation does not provide adequate exposure of the spotted samples to the solutions of assay reagents. Perform the following steps in enough solution to cover the nitrocellulose sheets to a depth of 1.0 cm: a.
b.
c.
Fix the proteins by laying the spotted sheets onto a solution of 25% (v/v) isopropanol, 10% (v/v) acetic acid. Do not pour the fixing solution onto the spotted sheets; this wrinkles the sheets and results in uneven exposure to subsequent assay reagents. Incubate 15 minutes. Discard fix, wash 4--5 times in deionized water to remove residual fix and incubate in Tris-buffered saline (TBS) (200 mM NaC1, 50 mM Tris-Cl, pH 7.4) for 5 minutes. Discard TBS and incubate 1 hour in blocking solution [0.5%
= M o u s e (x G F A P ]., = R a b b i t ~ Mouse I=G
FIG. 2. Schematic diagram of direct and sandwichELISAs for GFAP.
d.
e. f. g.
h.
gelatin (w/v) in TBS]. Discard blocking solution and incubate in antibody solution (polyclonal or monoclonal anti-GFAP, 1:500, in blocking solution containing 0.1% Triton X-100) for 2 hours. Discard antibody solution and wash 5 × 5 minutes in TBS. Discard TBS and incubate in blocking solution for 30 minutes. If a polyclonal antibody was used in step 4, discard blocking solution and incubate in Protein A solution for 1 hour. Protein A solution contains t2Sl-labeled Protein A diluted in blocking solution containing 0.1% Triton X-100 to 2000 CPM/10 ixl. If a monoclonal was used in step 4, follow step 6 with incubation in antibody solution containing rabbit anti-mouse immunoglobulins, 1:500 for 1 hour; then repeat steps 5 and 6 before incubating in Protein A. Recover Protein A solution (it can be saved and reused 1-2 times) and wash in TBS/0.1% Triton X-100, 4 × 5 minutes each wash followed by 4 × for several hours each wash. An overnight wash in a larger volume can be used to replace the last 4 washes.
7. Air-dry the nitrocellulose sheets, cut out the slots delineated by the penciled grid and count radioactivity in a gamma counter.
Sandwich ELISA Protocol
A schematic representation of direct and sandwich ELISAs for GFAP is shown in Fig. 2. The direct ELISA (Fig. 2, left) is based on the binding of homogenate proteins to the plastic wells of microtiter plates. After blocking unbound sites, the GFAP bound to the wells is detected by an antibody directed against GFAP. This antibody in turn is bound by an enzyme linked antibody directed toward the first antibody. Quantification is achieved by addition of a substrate for the antibody bound enzyme followed by spectrophotometry of the colored reaction product as determined in a microtiter plate reader. The speed, simplicity and cost-effective nature of direct ELISAs would appear to make them the method of choice for assaying GFAP. Unfortunately, we have not found them to be reliable for assaying GFAP in detergentbased samples or aqueous extracts of brain homogenates (see the Results section).
278
O'CALLAGHAN
The sandwich ELISA (Fig. 2, right) is based on binding of anti-GFAP antibody to the plastic wells of microtiter plates. This antibody, rather than the wells of the plate, serves as the receptor or "capture" reagent for binding GFAP. After blocking unbound sites, GFAP in brain homogenates is bound to the capture antibody coating the wells. The GFAP bound to the capture antibody then is bound by a second antibody (from another species). This second antibody is then bound by an enzyme-linked antibody directed toward it but not toward the capture antibody. Quantification is achieved as with the direct ELISA. The step-bystep procedure follows: 1. Dilute 1% SDS samples in sample buffer [PBS (137 mM NaC1/2.7 mM KC1/I.4 mM KH2PO4/8.0 mM Na2HPO4.7 H20/pH 7.4) containing 0.5% Triton X-100] to a concentration of 10 Ixg/ml. Samples high in GFAP (e.g., cerebellum) may need to be diluted to 5 Ixg/ml to avoid readings above the linear portion of the standard curve. Likewise, samples low in GFAP (e.g., striatum) may need to be diluted to only 20 txg/ml to avoid readings below the linear portion of the standard curve. Prepare standard curve samples in sample buffer at dilutions between 0.05 and 10 i~g/100 ILl (e.g., 0.05, 0.1, 0.25, 0.5, 1.0, 5.0, and 10 Ixg/100 pAL 2. Coat Immulon-2 flat-bottom microtiter plates with an lgG fraction of a polyclonal anti-GFAP, 1.0 jxg total immunoglobulin protein per 100 jxl, per well, diluted in PBS. Incubate 1-2 h at 37°C and then overnight at 4°C. All other incubations are at room temperature with moderate shaking on a reciprocating shaker. 3. Wash 4 x with PBS, 200 ixl/well. 4. Block 1 h with BLOTTO (8) [5% (w/v) nonfat dry milk in PBS], 100 ~l/well. 5. Incubate standards and samples in duplicate for 1 h, 100 pA/ well. 6. Wash 4 × with PBS containing 0.5% Triton X-100, 200 Ixl/ well. 7. Incubate 1 h with monoclonal anti-GFAP (Boehringer Mannheim), 1:500, made up in BLOTTO containing 0.5% Triton X-100, 100 I,d/well. 8. Wash 4 × with PBS containing 0.5% Triton X-100, 200 tzl/ well. 9. Incubate 30 minutes with alkaline phosphatase conjugated anti-mouse IgG (Jackson ImmunoResearch), 1:3000, made up in BLOTTO containing 0.5% Triton X-100, 100 jxl/well. 10. Wash 4 × with PBS containing 0.5% Triton X-100,200 ~zl/ well. 11. Add 100 Ixl of P-nitrophenylphosphate substrate (Bio Rad) and stop reaction 10 minutes later with 100 i~1 of 0.4 N NaOH. 12. Read at 405 nm. We used a Molecular Devices UV Max Microplate Reader (Menlo Park, CA) coupled to a Macintosh computer running a Soft Max (Molecular Devices, Menlo Park, CA) program. RESULTS
Standard Purity and Antibody Specificity Protein staining of GFAP purified by preparative PAGE revealed a band of approximately 50,000 daltons (Fig. 3, left side). Immunoblots of this band using polyclonal anti-GFAP antibodies verified that it was authentic GFAP (data not shown). This preparation of GFAP was used as the standard in a slot-immunobinding assay (standard format) to determine the concentration of GFAP in a homogenate of rat hippocampus. Aliquots of this homogenate were then used as a standard in all subsequent assays
GFAP Standard
GFAP Immunoblots Polyclonal Monoclonal
I 0
×
100
50
W
l
m
0 G)
m
0
20 FIG. 3. Protein staining of the GFAP standard and GFAP immunoblots of total hippocampalhomogenateproteinusingpolyclonaland monoclonal anti-GFAP. The GFAP standard (approximately0.25 Ixg) and the hippocampal homogenate(10 ixg)were separatedby SDS polyacrylamide(10%) gel electrophoresis. The standard was stainedwith Coomassie Blue R250, The homogenate proteins were transferred to nitrocellulose paper and probed with the appropriate antibody followingthe procedures described in the Method section.
(slot-immunobindingand ELISA). Immunoblots of this hippocampus homogenate showed that both polyclonal and monoclonal anti-GFAP detected only a single 50,000 dalton band (Fig. 3, right side). From time to time, however, I have seen evidence of slight cross reactivity with a 60-70 kilodalton band (neurofilament 68?) using polyclonal anti-GFAP.
Slot-lmmunobinding Assay Standard Curves The amount of radioactivity bound in the standard and monoclonal-based slot-immunobinding assays, as a function of total homogenate protein and of GFAP contained in the homogenate, is shown in Fig. 4. Both polyclonal- and monoclonal-antibody based assays showed a linear relationship between bound counts and the amount of total protein (and GFAP) spotted, The assay is linear up to at least 15 Ixg total protein (--60 ng GFAP). Spotting greater than 15 jxg total protein exceeds the binding capacity of the nitrocellulose and results in a loss of linearity. At the antibody dilutions used (1:500), the amount of radioactivity bound in the presence of 1.25 Ixg total protein was 8 and 2 times blank for polyclonal and monoclonal antibodies, respectively. The polyclonalbased assay is at least sensitive enough to detect 1 ng GFAP in 0.5 Ixg total hippocampal protein. As has been shown for other antibodies (7), decreasing the concentration of the GFAP antibodies (monoclonal or polyclonal) decreases the sensitivity of the assay (data not shown).
Direct ELISA Standard Curves The SDS-solubilized hippocampal homogenate standard, when diluted in PBS or even low concentrations of Tween 20 or Triton X-100, appeared to bind to the microtiter plate wells (data not shown). In my hands, however, the binding curves were never reproducible. Moreover, samples containing large toxicant-in-
IMMUNOASSAYS FOR GFAP
279
20
i
'
'
P;lyclonalAnti- ' G ~
i
i
Triton X-100/=
1.4
*
1.2
15
E ¢o v
o
5
i
__
i
Tween 20
~, 0.5°/' I 0.75%
1.0 0.8 0.6 0.4 0.2 0.0
~ 1~0
I
I
t
I
I
I
I
I
I
[
I
i I I I I I O ,ID 0.4 ~Without Capture Antibody_ _Without Capture Antibody.
