Anne Renström, PhD, Department of Occupational Medicine, Division of Respiratory Allergy, National Institute for Working Life, S-171 84 Solna, Sweden
Potential factors influencing antigen detection in immunoassays for measuring rat or mouse aeroallergen (i.e., assay setup, antigen specificities, standard extracts used, and antigen decay) were investigated in a three-country study (the UK, The Netherlands, Sweden). An inhibition enzyme immunoassay (EIA) setup gave nominal rat urinary allergen (RUA) sample values seven times higher than a sandwich EIA setup utilizing identical antibodies and standards. In immunoblotting experiments, pooled patient serum and polyclonal rabbit antibodies partly detected different rat antigens; monoclonal antibody specificity could not be determined. Immunoblot detection of mouse urinary antigens (MUA) by the polyclonal rabbit antibodies from all laboratories was similar. In both the RUA and the MUA assays, urinary antigen standards were detected with similar potency, except purified Rat n 1, which was an inefficient inhibitor in the RUA RAST inhibition. In the sandwich EIA RUA assays, a rat room-dust extract was detected with 700–800-fold less sensitivity than rat urine, whereas in the RAST RUA assay, dust inhibited equally with rat urine. Simulated decay did not decrease the potency of urinary antigen in any assay. Thus, assay setup and choice of detection antibodies strongly influence the nominal allergen levels. We recommend the use of standardized and characterized antibodies and standard extracts in sandwich EIAs to measure airborne rodent urinary allergens.
Abbreviations: NHLI: National Heart and Lung Institute, London, UK; WAU: Wageningen Agricultural University, Wageningen, The Netherlands; NIWL: National Institute for Working Life, Solna, Sweden; Ab: antibody; mAb: monoclonal antibody; EIA: enzyme immunoassay; RAST: radioallergosorbent test; RIA: radioimmunoassay; MUA: mouse urinary allergen; RUA: rat urinary allergen; BSA: bovine serum albumin; HSA: human serum albumin; PBS: phosphate-buffered saline; PTFE: polytetrafluoroethylene (Teflon); TBS: Tris-buffered saline.
Airborne allergens from sources such as cat, dog, and laboratory animals are collected on filters, eluted, and measured by a wide variety of methods. Patient serum ( 1), polyclonal antibodies ( 2), monospecific antibodies ( 3), or monoclonal antibodies (mAb) ( 4) have been used for allergen detection. Furthermore, the immunologic methods differ, including inhibition assays ( 5) or sandwich assays ( 6) utilizing radioactive ( 7), fluorescence ( 8), or enzymatic ( 9) visualization, and quantification systems. Thus, it is not surprising that reported allergen values vary greatly between studies on the same species. To our knowledge, there have been few efforts to compare or evaluate (5, 10) and none to standardize different methods. Recently, in a preliminary study, two methods for rat urinary aeroallergen measurement were compared ( 11). In 40 samples, exchanged between the UK and Sweden and analyzed in each country, the values obtained correlated (t2=0.72, P<0.0001), but differed by several orders of magnitude.
Thus, one goal of a European Union Concerted Action program, “Epidemiology of occupational allergic asthma and exposure to bioaerosols”, was to evaluate rodent aeroallergen measurement methods and the need for standardization. As described in a companion paper ( 12), a large variation in nominal allergen levels was found in triplicate samples, eluted and analyzed by the methods of the National Heart and Lung Institute (NHLI), the UK; Wageningen Agricultural University (WAU), The Netherlands; and the National Institute for Working Life (NIWL), Sweden, respectively. In the rat urinary allergen (RUA) assays, the nominal RAST inhibition (NHLI) levels were 2–3 orders of magnitude higher than in the polyclonal (WAU) and monoclonal (NIWL) sandwich enzyme immunoassay (EIA) methods, respectively. In the mouse urinary allergen (MUA) assays, in which all institutes used rabbit polyclonal antibodies, values were more similar, within an order of magnitude, despite one assay being a competitive inhibition RIA (NHLI) and the others sandwich EIAs (WAU and NIWL). In this paper, we explore some possible effects of the different assay setups (inhibition vs sandwich EIA), antibody specificities, and standard extracts, and of antigen decay, on the ability to detect antigen.
Material and methods
The three RUA and three MUA methods have been described previously (13–17), and an overview of these immunoassays is presented in the companion paper ( 12).
For experiment consistency, protein concentrations used in this study were quantified in one laboratory by the bicinchoninic acid (BCA) assay kit (Pierce, Rockford, IL, USA) with the included bovine serum albumin (BSA) as protein reference ( 18).
Preparation of antigens
Standard rat and mouse urinary antigens were prepared according to the methods of each institute (13–17).
Animal room-dust extracts were prepared from the exit ventilation filters of a rat room and a mouse room, respectively. Dust was dissolved in PBS, 0.15% v/v Kathon CG (Rohmand Haas, Hydrosupra Kemiservice, Helsingborg, Sweden) and rotated overnight at 4°C. The tubes were centrifuged to remove debris, and the supernatant was sterile filtered.
A rodent food extract was prepared by grinding food pellets (RMH-B, Hope Farms, Woerden, The Netherlands) to fine powder in a mortar and dissolving in PBS, 0.05% sodium azide. The suspension was vortexed (2×5 min), shaken (30 min), incubated in an ultrasonic bath (15 min), and again vortexed (5 min). After centrifugation, the supernatant was filtered through a Schleicher & Schuell S+S 15-cm red-band filter (Schleicher & Schuell, Dassel, Germany) and finally sterile filtered.
House-dust extracts were prepared from vacuum-cleaner dust from two homes, one home with four resident cats, and one in which cats had not lived for at least 5 years. Dust was mixed with 0.1 M ammonium bicarbonate at 4°C overnight. The extract was serially filtered and thereafter dialyzed against distilled water and freeze-dried.
Urinary antigens were broken down by a modified accelerated degradation protocol ( 19) meant to simulate aging of the proteins after excretion; i.e., dehydration, temperature change, oxidization, and ion-strength changes. Urine from young postpubertal male Sprague-Dawley rats and NMRI mice was collected in metabolic cages. The urine was centrifuged, sterile filtered, and dialyzed against PBS, and then Kathon 0.15% v/v was added. Aliquots of the rat and mouse urine preparations were either stored at 4°C or added to Polysorp microtiter plates (Nunc InterMed, Roskilde, Denmark) and allowed to dry out at 37°C for 8 days. The “aged” urine was reconstituted by adding MilliQUF-water (Millipore, Sundbyberg, Sweden), shaking for 5 s, incubating for 1 h, and shaking for 5 s, and was collected from the wells.
Comparison between sandwich and inhibition assay design
To study differences caused by assay design only, an inhibition assay was set up, using the RUA antibodies and standard extracts from WAU, and compared with the WAU EIA ( 15) values for 25 air samples collected and eluted by WAU. In the inhibition assay, EIA high-binding plates (Greiner, Nuertingen, Germany) were coated with RUA standard extract at 2 μg/ml in PBS. Briefly, the plates were washed and blocked with 0.5% gelatin before rabbit polyclonal Ab and air-sample eluates were added 1:1 and incubated for 2 h at 37°C. After washing, bound antibody was detected with peroxidase-labeled horse antirabbit Ig (1 h at 37°C) and, after washing, the substrate o-phenylenediamine (2 mg/ml) in 0.05 M citrate-phosphate buffer, pH 5.5, with 0.015% H2O2. The reaction was stopped after 30 min with 2 N HCl and the absorbance read at 492 nm.
Immunoblotting of MUA and RUA standards and animal room-dust extracts with detection antibodies from all institutes
To study antibody specificity to present antigens, nonreducing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) was performed mainly as described previously ( 20). The antigens (RUA and MUA standards, rat and mouse room-dust extracts, and human serum albumin [HSA, Behring Diagnostics, Inc, supplied by Hoechst UK Ltd, Hounslow, Middlesex, UK] as control) were electrophoresed in 15% acrylamide gels, and electroblotting was performed overnight onto nitrocellulose sheets (0.45-μm pore size, Bio-Rad, UK) at 0.1 mA. After washing in TBS buffer containing 0.1% v/v Tween 20 (with 3% casein for experiments using mAbs), the blots were incubated overnight with all RUA and MUA antibodies and buffer controls, as described in the legends to Figs. 2 and 3. After further washing, the filters were labeled overnight (label for pooled patient serum: 125I-labeled anti-IgE, 1:10 [Pharmacia & Upjohn, Milton Keynes, UK]; label for other Ab: 125I-labeled protein-A, 1:10 [Amersham International, Little Chalfont, UK]. The immunoblots were washed, dried, and exposed to radiographic film (Hyperfilm, Amersham International) at –70°C.
Comparison of assay binding to reference standards and dust extracts
The RUA and MUA standard extracts from all institutes, rat and mouse room-dust and animal food extracts, and extracts collected from a home with and a home without cats, respectively, were serially diluted and analyzed for binding in each RUA and MUA assay.
Assay detection of RUA and MUA subjected to an accelerated degradation protocol
Degraded or “aged” sample antigen may be detected differently from fresh (standard) antigen in the assays. Therefore, “aged” urine preparations were assayed in parallel with “fresh” urine in the RUA and MUA assays in serial dilutions.
Comparison between values using sandwich or inhibition assay design
The correlation between values was high, t2=0.95. However, nominal values for the 25 sample eluates were a median 83.4 (range 7.4–550) ng/ml using inhibition, and a median 11.2 (0.57–101) ng/ml using a sandwich EIA setup. The median ratio of inhibition value to sandwich value for each sample was 7.3 (geometric mean ratio=7.4, Fig. 1); however, the ratio decreased with increasing aeroallergen content.
Immunoblotting of MUA and RUA standards and dust extracts using detection antibodies from all institutes
Immunoblotting showed that antigen specificities of the RUA antibodies differed somewhat ( Fig. 2). The rat room-dust extract was bound diffusely by the pooled patient IgE, whereas the rabbit polyclonal Ab binding appeared distinct. Moreover, patient IgE bound to a 21-kDa antigen in the urine samples, which was not recognized by the rabbit polyclonals. No binding by the mAbs to RUA antigens could be demonstrated. The mouse urinary standards, also used to immunize rabbits, originated from mice of different ages, sex, and mouse strains, and showed slightly different antigen sizes in the blotting. However, all anti-MUA Ab preparations bound strongly to the 15–18-kDa MUA antigens ( Fig. 3) with similar pattern. The apparently more concentrated WAU and NIWL Ab also detected 44–50-kDa antigens. The NIWL Ab gave a very weak band at the HSA control (72 kDa).
Comparison of assay binding to reference standards and dust extracts
The comparisons of the three RUA methods' detection of standards and dust extracts are shown in Fig. 4A. The mean relative potency of the extracts was calculated as the mean of the inhibition (logit transformed as previously described ) or OD values from 2–8 dilutions (from the linear part of the curves), corrected for total protein concentration. Generally, the standards of the different institutes were detected similarly, with a slightly higher binding of the assay's “own” standard. In the mAb assay (NIWL), the binding reflected the proportion of Rat n 1.02 present in each standard extract. However, Rat n 1 (the NIWL standard) was an inefficient inhibitor in the RAST method, indicating that a significant proportion of the human serum IgE Ab was specific to other antigens present in the rat urine and dust. The most striking discrepancy between the RAST inhibition and the sandwich EIA methods was in the detection of rat room dust. The RAST inhibition method detected dust similarly to rat urine standards added at the same protein concentrations, while in the EIAs, dust had to be added at 690 (WAU) or 760 (NIWL) times higher concentrations than the urine standards to yield the same absorbance. Mouse urine (WAU standard, prepared from old/young/male/female mice) was detected in the WAU assay with 13 000-fold less sensitivity than rat urine. Neither RUA assay bound mouse room dust or rodent food, which may be present in air samples, or house dust. Furthermore, the RAST inhibition standard curve ranged over several orders of magnitude, whereas the EIA curves were steep.
In the three MUA assays, the “other” MUA standards were detected similarly to each assay's “own” standard extract (Fig. 4B), except the WAU standard, which was bound with 6.5-fold less sensitivity in the NHLI assay. Mouse room dust was detected with about 30-fold (NHLI) to 50-fold (WAU, NIWL) less sensitivity than the urine standards. Rat urine (WAU standard, young/old/male/female rat urine) was detected in all MUA assays, and the NHLI assay bound to rat room dust with 2700 times lower sensitivity than to mouse urine standard; the difference was great for the EIAs. At high relative concentrations, slight binding was observed to rodent food and the house dust from a cat home.
Assay detection of RUA and MUA subjected to an accelerated degradation protocol
Antigen “aging” affected the binding of antigen in the RUA and MUA assays only marginally. The mean percent inhibition or optical density for “aged” rat urine were 92% (NHLI), 99% (WAU), and 94% (NIWL) of the values obtained using “fresh” antigen; for “aged” mouse urine, the values were 90% (NHLI), 91% (WAU), and 98% (NIWL) of the “fresh” values, respectively.
In this three-country study comparing different methods for measuring rodent aeroallergens, large differences in values were found in parallel air samples, especially between RAST inhibition and sandwich EIA measurements. Differences between EIA assay values could partly be attributed to extraction practices, and partly to type of immunoassay (companion paper ). We have now examined how assay setup, type and specificities of the antibodies, standard extract differences, and antigen aging may affect the ability to detect antigen.
When the same antibody and standard were utilized in an inhibition and a sandwich EIA setup, respectively, inhibition gave sevenfold higher aeroallergen sample values. This may have been due to differences in availability of antigen epitopes in solution compared to the bound state, which changes the three-dimensional structures of the molecules ( 21). The decreasing ratio of inhibition setup to EIA setup with increasing allergen concentration was also observed in the air-sample comparisons ( 12); however, the reason for the decrease is not known. The inhibition standard curves are considerably less steep than the sandwich EIA curves. A small difference in inhibition will thus give a large difference in nominal allergen level, and larger measurement variability. This may contribute to the r2 values for air samples being smaller between inhibition and sandwich methods than the r2 values between sandwich methods (companion paper ). Similarly, in a comparison between EIA methods for measuring swine aeroallergens, a sandwich setup was found to give better reproducibility than an inhibition design ( 10).
Immunoblotting experiments showed several differences in the specificities of the antibodies used, especially in the binding to rat room dust. While the specificity of the mAbs could not be shown here, they have previously been shown to bind selectively to the Rat n 1 isoallergens ( 16). In the MUA blots, all rabbit Ab bound to 15–18-kDa proteins present in urine and dust (probably Mus m 1 , to which most mouse-allergic patients react ). The WAU and NIWL Ab also bound lightly to 44–50-kDa proteins whose identity is not known. The ability of the Mus m 1 affinity-purified Ab (NIWL) to detect these allergens suggests that these may be aggregate forms of Mus m 1.
In agreement with a previous comparison ( 11), all urine extracts were detected similarly in the RUA and MUA assays. However, purified Rat n 1 (used as standard in the mAb assay) was inefficient as an inhibitor in the RAST-inhibition assay, indicating that a significant proportion of the patient IgE is specific for other rat urinary proteins, as confirmed in the Western blots. In the mAb RUA assay, differences in detection reflected the relative amount of Rat n 1.02 (α2u-globulin) in each extract. All RUA and MUA assays were specific; binding to other allergens would be of little, if any, practical significance.
The most important difference between RUA assays was in their detection of the rat room-dust extract (containing material from bedding, food, feces, hair, etc.). When added at the same total protein concentrations as the urine reference (as measured with the Pierce BCA method), dust was 700–800-fold less potent in the EIAs, whereas it was detected with equal potency by the pooled patient serum in the RAST-inhibition assay. If the dust extract approximates air samples, this could explain a major part of the observed differences ( 12). Perhaps IgE binding to rat dander or hair allergens plays a part; food and mouse room-dust extracts were not detected. In contrast, all rabbit polyclonal MUA assays detected mouse room dust with about 30–50-fold less sensitivity than mouse urine. The antibodies used derive from different types of immunization: human IgE produced after occupational inhalation of dust, and Ab from rabbits or mice serially immunized with mouse or rat urinary proteins, respectively. Although the latter antibodies capture the urinary allergens, the clinical allergenic potency of an air sample may be underestimated.Swanson et al. ( 5) similarly observed, when measuring airborne mite allergens using mAbs, rabbit polyclonals, or human sera as detection antibodies, that human sera gave about 20-fold higher values.
Using the artificial “aging” protocol, we could not find evidence that differences in detection of dust might be due to antigen degradation. However, it is possible that the protocol used was not representative of natural antigen breakdown.
To summarize, we have shown that several factors contribute to differences in nominal levels of allergen measured in air samples. The most important is source and type of antibodies, whose specificities contributed to up to 800-fold of the differences. Assay design contributed, with inhibition giving about sevenfold higher values. We found only a minor effect of using urine standards from different sources. In the companion paper ( 12), we concluded that the use of Tween and BSA enhanced extraction efficiency up to about 10-fold and improved stability during storage.
By definition, the use of patient serum for allergen detection has clinical relevance. However, monoclonal sandwich assays specific for major allergens have important advantages, both for the purpose of standardization, and because of high sensitivity, specificity, and reproducibility ( 12). The disadvantage is that they may not detect other present allergens. Standardization of polyclonal EIA sandwich assays may be achieved by the acquisition of large amounts of reagents and an agreed assay protocol.
However, it is important to appreciate that what is measured by immunoassays is a marker of allergen load, not the actual dose of airborne allergen inhaled and potentially reacting with the immune system.
This study was supported by the European Union (contract no. BMH1-CT94-1446) and the Swedish Council for Work Life Research (no. 93:0450).
List of abbreviations. NHLI: National Heart and Lung Institute, London, UK; WAU: Wageningen Agricultural University, Wageningen, The Netherlands; NIWL: National Institute for Working Life, Solna, Sweden; Ab: antibody; mAb: monoclonal antibody; EIA: enzyme immunoassay; RAST: radioallergosorbent test; RIA: Radioimmunoassay; MUA: mouse urinary allergen; RUA: rat urinary allergen; BSA: bovine serum albumin; HSA: human serum albumin; PBS: phosphate-buffered saline; PTFE: polytetrafluoroethylene (Teflon); TBS: Tris-buffered saline.