| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Clinical Immunology |
1 Institute of Clinical Chemistry and Pathobiochemistry, Klinikum rechts der Isar der Techni-schen Universität München, München, Germany.
2 Statens Serum Institut, Copenhagen, Denmark.
3 Institute of Clinical Chemistry and Laboratory Medicine, Klinikum der Johannes Gutenberg-Universität Mainz, Mainz, Germany.
4 Department of Internal Medicine I, Klinikum der Universität Regensburg, Germany.
aAddress correspondence to this author at: Institute of Clinical Chemistry and Pathobiochemistry, Klinikum rechts der Isar der Techni-schen Universität München, Ismaninger Str. 22, D-81675 München, Germany. Fax 49-89-4140-4875; e-mail luppa{at}klinchem.med.tum.de.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We developed a novel surface plasmon resonance (SPR) biosensor chip for studies of dsDNA and anti-dsDNA binding. A synthetic oligonucleotide was coupled to biotinylated human transferrin, hybridized with the complementary antistrand, and ligated with a human recombinant dsDNA fragment 233 bp in length. After surface immobilization of this antigenic construct, diluted sera from SLE patients and healthy donors were analyzed with the resulting SPR biosensor system.
Results: This SPR biosensor allowed specific detection of anti-dsDNA. In pilot experiments, sera from SLE patients were distinguished from control sera. We also confirmed the specificity of this biosensor by supplementing anti-dsDNA–positive sera with salmon sperm DNA, which blocked the surface binding of anti-dsDNA in a concentration-dependent manner.
Conclusions: An SPR biosensor monitors interactions in real time under homogeneous conditions, providing information about binding kinetics and affinities. Its applicability critically depends on the design of the solid-state surface of the sensor chips. Covalently immobilizing dsDNA as the antigen to the surface in a flow-through cell assured maximal stability for multiple serum injections and regeneration cycles. This technique, which adds a new analytic quality to existing methods, may be beneficial in the diagnosis and clinical monitoring of SLE.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
According to the SLE classification criteria of the American College of Rheumatology (4)(5), at least 4 of 11 criteria have to be met for disease diagnosis. There is a delay of 
2 years between the onset of symptoms and final diagnosis (6), reflecting the fact that reliably diagnosing SLE at an early stage is still a difficult task. Both the detection of antinuclear antibodies by immunofluorescence and findings of "antibodies to native DNA in abnormal titer" (American College of Rheumatology criterion 10.b) are of high diagnostic significance and are considered specific (7) and early(8) markers for SLE. Moreover, assaying for antibodies is feasible for follow-up, because rising titers of double-stranded DNA autoantibodies (anti-dsDNA) IgG are of great prognostic value for predicting disease flares, particularly for the development of glomerulonephritis (7)(9). In principle, anti-dsDNA antibodies in the sera of SLE patients may be of the 3 isotypes (IgG, IgM, and IgA) (10).
Three clinical laboratory methods are commonly used for the determination and quantification of anti-dsDNA: Crithidia luciliae indirect immunofluorescence, ELISA, and RIA (the Farr assay). Because of the great differences among these assay systems in methodology, DNA antigen source, steric DNA presentation, and reaction conditions, a comparison of these methods is limited, and analytic results have to be interpreted carefully in light of the assay’s characteristics and the patient’s condition (11)(12). The main discrepancies depend on whether IgG, IgM, or the collection of isotypes is detected and whether low-avidity antibodies are also detected. Furthermore, these assays are conducted under equilibrium conditions and so are unable to provide any information on binding characteristics. A positive correlation between avidity and disease severity has been demonstrated for several autoantibodies, however, including those against the glomerular basement membrane (13) and ß2-glycoprotein I (14). High-avidity anti-dsDNA IgG has been shown to be associated with glomerulonephritis (15). Avidity has been discussed repeatedly in the literature (16) but still remains a subject of controversy.
The aim of this study was to develop a novel biosensor device on the basis of surface plasmon resonance (SPR) that would allow label-free monitoring of the interaction between dsDNA and anti-dsDNA in real time. The change of mass concentration at the interface because of specific binding of anti-dsDNA to surface-immobilized dsDNA would be detected as changes in the refractive index via the SPR effect (17)(18).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Serum samples were leftover specimens that were not individually identifiable, in accordance with the Food and Drug Administration document, "Guidance on Informed Consent for In Vitro Diagnostic Device Studies Using Leftover Human Specimens That Are Not Individually Identifiable – Guidance for Sponsors, Institutional Review Boards, Clinical Investigators and FDA Staff" (OMB control no. 0910–0582, issued April 25, 2006; http://www.fda.gov/cdrh/oivd/guidance/1588.html). The investigator was not able to identify the source from the clinical information that accompanied the sample.
Thirty-nine serum samples from apparently healthy control individuals recruited from laboratory staff and medical students were age matched and were presumed to be free of any acute or chronic disease on the basis of a medical and clinical chemistry evaluation. Written informed consents were obtained from these individuals.
Blood samples were collected without anticoagulant and after clot formation were centrifuged at 1500g for 20 min. Serum was stored at –70 °C in aliquots until analysis.
chemicals and oligonucleotides
Human transferrin (hTf) (98% purity) and 4-nitrobenzaldehyde were purchased from Sigma-Aldrich. Succinimidyl 4-hydrazinonicotinate acetone hydrazone (SANH) and a 5'-aldehyde–modified (CHO) oligodeoxynucleotide (ODN) were from Solulink. BssSI and T4 DNA ligase were from New England Biolabs. The amine-coupling kit and the surfactant P20 were from Biacore. We purchased high-resolution agarose from Roth, deoxynucleoside triphosphates from Peqlab Biotechnologie, salmon sperm DNA from Fluka, and Taq DNA polymerase from Qbiogene. All nonaldehyde-modified ODNs were synthesized by MWG Biotech with standard phosphoramidite chemistry. The sequences are given in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol53/issue2).
antibodies
Rabbit horseradish peroxidase–conjugated polyclonal antibody to hTf was purchased from Acris Antibodies. F7–26, a control antibody against single-stranded DNA (ssDNA), was from Hölzel Diagnostika. A monoclonal antibody (Hyb 331-01) from a NZW x NZB F1 mouse strain with specificity toward both dsDNA and ssDNA ("anti-DNA monoclonal antibody") was developed at the Statens Serum Institut (19).
gel electrophoresis
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis and protein transfer were performed with the MiniProtean 3 Cell system (Bio-Rad). After immunoblotting, the nitrocellulose membrane was probed with the hTf-specific rabbit horseradish peroxidase–conjugated polyclonal antibody. Immunoreactive bands were resolved with the SuperSignal® enhanced-chemiluminescence reagent (Perbio Science) and HyperfilmTM (Amersham Biosciences).
immunoassays
We used the anti-dsDNA RIA reagent set from Trinity Biotech according to the manufacturer’s instructions. Values >7.0 x 103 IU/L were regarded as indicative of SLE. The FARRZYMETM Human High Avidity Anti-dsDNA enzyme immunoassay (EIA) reagent set from the binding site was used to detect IgG-specific high-avidity anti-dsDNA with a cutoff value of >30 x 103 IU/L.
apparatus
Biosensor measurements were made at 25 °C with a Biacore X instrument (Biacore). All measurements were performed on sensor chips SA (Biacore), which consist of a gold surface coated with carboxymethyldextran preimmobilized with streptavidin.
The online Data Supplement provides details on the preparation and immobilization of antigen as well as on the biosensor measurements (see online Data Supplement at http://www.clinchem.org/content/vol53/issue2).
statistical analysis of biosensor measurements
Imprecision was assessed by injecting a representative SLE serum sample (c-1) in quadruplicate on 4 different days. We calculated CVs for the maximum association levels for intraassay and among-day imprecision, diagnostic sensitivities and specificities, and likelihood ratios for detecting positive sera. The term "positive" refers to either a confirmed SLE diagnosis or a positive outcome in the Farr assay, which is used as a reference for anti-dsDNA determination.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). hTf was chosen because it contains many lysine residues, which feature primary amino groups available for coupling reactions. A 24-base ODN was synthesized with an incorporated CHO phosphoramidite, and hTf was modified with hydrazine moieties. CHO-ODN and hydrazine-hTf are chemically stable compounds. They react rapidly with each other, however, to form stable hydrazone bonds; therefore, we had to introduce hydrazine groups in the hTf protein after biotinylation via reaction with the bifunctional reagent SANH. Because we desired only 1:1 ODN-hTf conjugates, we optimized the process of SANH modification to obtain a molar-substitution ratio of 1.0. With a 10-fold molar excess of SANH, photometric quantification of the introduced hydrazones demonstrated a [hydrazones]/[protein] ratio of 0.83. To further ensure that only a single CHO-ODN was coupled to the protein, we added only 0.25 equivalents to the solution of modified hTf. Adding more SANH and ODN yielded heterogeneous products that were difficult to analyze. The purity of the conjugation product was demonstrated with sodium dodecyl sulfate–polyacrylamide gel electrophoresis and immunoblotting analysis. Only a single distinct band with an Mr greater than that of unprocessed hTf was visible in the enhanced-chemiluminescence blot (Fig. 2
). The aim of gel filtration was to remove unreacted CHO-ODN residues. Absorbance measurements at 260 and 280 nm indicated concentrations of 368 x 106 mol/L for hTf and 36.6 x 106 mol/L for the ODN; thus, 10% of the protein had been conjugated with the ODN.
|
|
dna preparation and ligation
We amplified a 231-bp fragment of the hemochromatosis (HFE)2
gene via the PCR with elongated primers that introduced a new restriction site (see online Data Supplement for sequences). The length of the DNA antigen can be adjusted for future investigations by simply changing the primer sequences. We selected BssSI as the restriction endonuclease because its nonpalindromic recognition site appeared nowhere else in the amplified sequence. The short fragments removed from the 2 ends of the strands during digestion (7 bp plus a 4-base overhang) left a much longer product (233 bp plus a 4-base overhang at each end) that facilitated the purification of the product. The asymmetric nature of the BssSI restriction site ruled out self-ligations during the ligation of ODN-hTf and dsDNA. The PCR products were ligated with different equivalents of ODN-hTf and subsequently analyzed by electrophoresis in a 2.5% agarose gel containing ethidium bromide (Fig. 3
). The appearance of 2 product bands with lower electrophoretic mobilities depended on the ODN-hTf concentration. The 2 bands resulted from the ligation of either 1 or 2 ODN-hTf molecules to the DNA antigen. In the absence of ODN-hTf or ligase, no product bands appeared. A 4-fold molar excess of ODN-hTf produced acceptable amounts of hTf-dsDNA.
|
determination of anti-dsdna in human sera
Sera from SLE patients were distinguishable from control sera of healthy individuals (Figs. 4A
and 4B
). Serum samples from patient i, however, had no dsDNA-specific reactivity, whereas Farr assay results were positive (Table 1
). Samples from patient i also tested negative in the high-avidity IgG-specific EIA (data not shown). Serum samples from patients c, e, and f showed faster dissociation and association in the sensorgram than sera from other SLE patients. This finding is exemplified for serum sample c-1 (Fig. 4A
). For patient b, the biosensor association levels did not follow the trend of concentration measurements produced in the Farr assay. However, an increase in dissociation rates during the course of the disease could be detected for patient b (see the decrease in residual binding for patient b in Table 1
).
|
|
statistical analysis of biosensor measurements
Intraassay imprecision was between 1.2% and 1.8%, whereas among-day variation was 8.0%.
The cutoff value, which reflects the lowest concentration of anti-dsDNA association regarded as positive, was set to 25 resonance units, calculated as the mean plus 2 SDs of the maximum association concentrations of control sera. We detected sera from patients with confirmed SLE diagnoses with 98.2% specificity at a sensitivity of 83.3%, which yielded a likelihood ratio of 47.5. The biosensor detected sera with positive results in the Farr assay with 88.1% specificity at a sensitivity of 87.5%, yielding a likelihood ratio of 7.3.
specificity
To evaluate possible cross-reaction with ssDNA regions, we prepared a 1:100 dilution of a control serum sample from a healthy donor (1 volume of serum sample diluted with 99 volumes of diluent) and supplemented it to a concentration of 0.01 g/L with either the F7–26 monoclonal antibody (specifically binds to ssDNA) as a negative control or anti-DNA monoclonal antibody as the positive control. Both were injected in the flow cell under the same conditions as described for the analysis of sera (see the online Data Supplement). The anti-DNA monoclonal antibody showed high association levels on the chip surface, whereas F7–26 failed to bind to the immobilized dsDNA (Fig. 4C
). To demonstrate the specificity of the anti-dsDNA assay with patient samples, we carried out competition experiments by preincubating a positive sample with salmon sperm DNA. Specific binding diminished with increasing DNA concentration (Fig. 5A
). We reproduced this competition result with 2 other positive serum samples (data not shown). We further assessed the antibody-antigen interaction by coinjecting salmon sperm DNA and running buffer in the dissociation phase (see online Data Supplement). A dose-dependent acceleration of complex dissociation was observed as soluble DNA competed with immobilized DNA for binding to anti-dsDNA (Fig. 5B
).
|
Binding was also successfully inhibited with both singlestranded ODN and double-stranded ODN (Fig. 5C
). This result reflects the fact that the ODNs (26 bases and 22 bp, respectively) were long enough to act as antigens for anti-dsDNA antibodies. In accordance with well-known observations that anti-dsDNA antibodies in SLE sera are directed against both dsDNA and ssDNA, binding was inhibited with the 2 conformations to approximately the same degree. This result indicated that the relevant autoantibodies were also reactive to ssDNA and were directed against the phosphodiester backbone independently of the sequence.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The pivotal step in the development of solid-state biosensor surfaces is the immobilization of one of the interaction partners (i.e., dsDNA, referred to here as ligand) on the gold support. Because the surfaces have to be regenerated many times by treatment with alkaline solutions, the use of stable—preferably covalent—immobilization strategies is a fundamental requirement.
Several methods have been developed for covalently immobilizing DNA on solid surfaces. These methods have been established for use with DNA microarrays; however, they are not transferable to the SPR system for several reasons. First, because it is desirable to immobilize the ligand under flow conditions, immobilization has to be completed quickly at temperatures between 4 °C and 40 °C. Incubation times of >100-min cannot readily be performed with the system. Second, standard gold chips for the SPR biosensor are coated with a carboxymethylated dextran layer. This coating forms a 3-dimensional matrix that enhances ligand immobilization capacity and flexibility. The negative charge on carboxylic moieties and DNA molecules at moderate pH thus produces strong electrostatic repulsion. Third, the use of organic aprotic solvents is severely limited because of the limited chemical resistance of the biosensor apparatus.
Our immobilization strategy meets all of these ambitious requirements through the use of 2 coupling chemistries: biotin/streptavidin for fast and efficient capturing of ligand at the surface of streptavidin-coated chips, followed by covalent amine coupling for the final stabilization and rigid surface presentation of the dsDNA.
A synthetic single-stranded ODN was covalently coupled to biotinylated hTf. We used aldehyde and hydrazine chemistry for this conjugation (20). hTf was used in excess to prevent high losses of CHO-ODN and formation of poorly defined side products. Product analysis revealed homogeneous 1:1 conjugation of hTf and ODN (see Fig. 2
).
Because the source of the DNA used and its preparation play a decisive role in the quality of the results obtained from an anti-dsDNA assay, we used human recombinant DNA amplified via the PCR. Immobilization provides a defined steric access to the DNA and avoids protein contamination. In this respect, this method contrasts with the C. luciliae indirect immunofluorescence technique, in which histone impurities may lead to cross-reactivity (21), and with the Farr assay, in which radiolysis may lead to DNA decomposition (22).
The most important feature of our SPR biosensor should be the ability to differentiate the sera from SLE patients and healthy control individuals, and we were able to demonstrate this capacity in pilot experiments with serum samples from healthy individuals, patients with non-SLE autoimmunopathies, and SLE patients. Use of a cutoff value of 25 resonance units in the SPR biosensor assay offers superb specificity (98%) at a high sensitivity (83%); thus, the results obtained with this method appear to approximate the performance of the Farr assay. Table 1
compares anti-dsDNA concentrations obtained with the Farr assay with both maximum association concentrations at the end of association phases in the SPR biosensor and the percentages of residual binding after 300 s of dissociation. As expected, there is no simple correlation between concentrations obtained with the Farr assay and binding values obtained in the biosensor. This finding is due to the fundamentally different methodologies of the 2 assay systems. In contrast to results derived with the Farr assay, biosensor association levels depend not only on antibody concentrations but also on their affinities. Our initial data demonstrate that the SPR biosensor–based analytic method gives more detailed information about the anti-dsDNA in SLE sera.
Patient i, who tested positive in the Farr assay and negative in the high-avidity IgG-specific EIA, tested negative with our biosensor system. The Farr assay and the EIA method use a high salt concentration (Farr assay) or stringent washing steps (EIA) in the first analytic phase to differentiate high-avidity antibodies. In the second step, however, all bound antibodies contribute equally to the quantification. The biosensor allows further differentiation by means of a direct impact of antibody avidity on the respective sensorgram. In particular, anti-dsDNA from patient i caused low association levels in our system. The antibodies were avid enough, however, to produce a small signal in the Farr assay but not in the EIA.
The different shapes of the binding curves, which were caused by different binding kinetics, suggest that varying combinations of anti-dsDNA isotypes are present in the sera. This point is exemplified in patient c (Fig. 4A
). Different shapes of the characteristic curves may indicate either IgG or IgM isotype predominance. Whereas IgM shows a delayed association phase and weak dissociation correlated with high residual binding, IgG exhibits faster association and a variable dissociation phase. Accordingly, the increase in dissociation rates during the course of the disease, as seen in the sera samples from patient b (Table 1
), hints at changing isotype patterns. Such changing patterns would be associated with alterations in avidities and antibody molecular masses and are a credible explanation for the lack of correlation of the Farr assay and biosensor assay data. Further investigations, especially comparisons of Farr assay and biosensor assay results with clinical records, are required to understand the significance of these differences.
The formation of ssDNA regions within the dsDNA molecule has to be considered. Autoantibodies against ssDNA are also used as markers for SLE, but their lack of specificity (23) makes their determination undesirable. Antibodies that are of high diagnostic or predictive value for SLE are dsDNA specific and of high affinity. Some assays use a single-stranded control to exclude cross-reactivity. One possibility for evaluating the reactivity of the surface toward ssDNA antibodies is to use the F7–26 monoclonal antibody as a negative control. F7–26 specifically binds to ssDNA and shows no reactivity toward dsDNA (24)(25). On the other hand, we used a monoclonal antibody derived from a lupus-prone mouse model as a positive control. This antibody reacts with both ssDNA and, with higher affinity, dsDNA (19). The superior specificity of the biosensor surface can be presumed from the data generated with these controls (Fig. 4
). We further established the specificity of the system in competition experiments with salmon sperm DNA and both single-stranded and double-stranded ODNs.
A major distinguishing feature of existing methods for measuring anti-dsDNA is the detection of different sets of isotypes. Although the Farr assay does not discriminate between isotypes, both the C. luciliae indirect immunofluorescence and the ELISA methods offer this possibility via the choice of the secondary antibody. Different isotypes are most likely to differ in clinical relevance. Some authors recommend IgG determination alone, whereas others state that the analysis of all isotypes may be superior in terms of specificity and sensitivity (7)(10)(22). Because our device monitors formation of the immune complex directly, there is no isotype differentiation for the most part; hence, we have developed a protocol for the removal of unwanted isotypes before analysis (to be reported elsewhere).
In summary, the SPR biosensor device is well suited for the detection of anti-dsDNA in the sera of SLE patients. In addition, the technique acquires information about binding kinetics and affinities of the specific autoantibodies through its capacity to monitor biospecific interactions in real time. Although this technique cannot readily differentiate between concentrations and affinities/avidities of antibodies in polyclonal sera, the association concentration, which represents the degree of immune complex formation within a given time period and thus is defined by both of these variables, is one of the most straightforward and accurate measures of antibodybinding intensity now available.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Human gene: HFE, hemochromatosis.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
