Glycopeptide antigen CSF114(Glc) was prepared by microwave-assisted solid phase peptide synthesis. The glycopeptide was purified to homogeneity by solid phase extraction and reverse phase high-pressure liquid chromatography (HPLC), and further characterized by mass spectrometry and analytical HPLC as described elsewhere [37].

Sensor chip CM5 and the running buffer HBS-EP+ 10× (0.1 M HEPES, 1.5 M NaCl, 30 mM EDTA and 0.5% v/v Surfactant P20; yielded pH 7.4 when diluted) were purchased from Biacore AB (GE Healthcare, Uppsala, Sweden). The amine coupling reagents N-hydroxysuccinimide (NHS), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), and 1 M ethanolamine hydrochloride-NaOH pH 8.5 were provided by Biacore AB (GE Healthcare). Sodium acetate was purchased from Carlo Erba (Milano, Italy). Sodium hydroxide was provided by Honeywell-Riedel deHaen (Seelze, Germany). All analyses were performed in a Biacore T100 instrument (GE Healthcare). All experiments were made at 25 °C using HBS-EP+ as running buffer.

One hundred and twenty one human serum samples were obtained for diagnostic purposes from patients and healthy blood donors who had given their informed consent. Each serum sample was centrifuged and supernatant aliquoted and stored at −20 °C until use. Patients' sera were obtained from a group of 61 relapsing—remitting MS (RR-MS) patients after a diagnostic lumbar puncture. Cerebrospinal fluid analysis and MRI examinations were performed for diagnostic purposes.

A stock solution of glycopeptide CSF114(Glc) was prepared in pure water (1 μg/μL) and stored at +4 °C. Immediately prior to immobilization procedure, peptide stock solution was diluted to a concentration of 10 μg/mL in 0.1 mM sodium acetate pH 5.5. Standard amine coupling procedure was employed for glycopeptide immobilization, essentially according to the Biacore procedures. The appropriate flow cell of the sensor chip surface was activated by injecting an EDC/NHS (50:50) mixture at a flow rate 10 μL/min during 420 s. CSF114(Glc) was injected at 10 μL/min using the aim of immobilization procedure to give a final immobilization level of 800 ± 100 Resonance Units (RU). Unreacted groups on sensor chip surface were blocked by injecting 60 s-pulses of 1 M ethanolamine at pH 8.5 until complete deactivation. Reference channel was activated and subsequently blocked with ethanolamine.

All analyses were performed in triplicate at a flow rate of 30 μL/min. Human serum samples were diluted 1:100 and/or 1:50 in running buffer. Samples were injected, at different injection times, in both active and control channels followed by 60 s of buffer injection to allow dissociation. Interaction of samples with sensor chip flow cells were monitored as separate sensorgrams and measurements were taken 15 s after the end of each injection. The antibody responses were measured in RU units as a signal difference between active channel and reference channel. After each measurement, surface was regenerated injecting two pulses of a solution 100 mM NaOH during 60 s.

The two panels of serum samples were tested in ELISA to check the presence of specific antibodies. Ninety six well activated polystyrene ELISA plates (NUNC Maxisorb; Sigma-Aldrich, Milan, Italy) were coated with 1 μg per 100 μL of glycopeptide CSF114(Glc) per well in pure carbonate buffer 0.05 M (pH 9.6) and incubated at +4 °C overnight. After three washes with saline solution containing 0.05% Tween 20, non-specific binding sites were blocked with fetal calf serum (FCS), 10% in saline Tween 20 (100 μL per well) at room temperature for 60 min. Sera diluted from 1:100 to 1:10,000 were applied at +4 °C overnight in saline solution/Tween 20/10% FCS. After three washes, 100 μL of alkaline phosphatase conjugated antihuman IgM (diluted 1:200 in saline/Tween 20/FCS) or IgG (diluted 1:8,000 in saline/Tween 20/FCS; Sigma-Aldrich) were added to each well. After 3 h incubation at room temperature and three washes, 100 μL of 1 mg·mL⁻¹ p-nitrophenyl phosphate (Sigma-Aldrich) in 10% diethanolamine substrate solution (pH 9.8) was added. After 30 min, the reaction was stopped with 1 M NaOH (50 μL), and the absorbance was read in a multichannel ELISA reader (Tecan Sunrise, Männedorf, Switzerland) at 405 nm. ELISA plates, coating conditions, reagent dilutions, buffers, and incubation times were preliminary tested [16]. Antibody levels are expressed as absorbance in arbitrary units at 405 nm (sample dilution 1:100).

To determine MS-specific antibody reactivity against the CSF114(Glc)-based biosensor in a Biacore T100 instrument 60 healthy blood donors' (BD) and 61 MS patients' sample sera were tested. For this purpose glycopeptide CSF114(Glc) was immobilized on the sensor chip surface following the amino coupling chemistry. Reproducible coupling conditions were initially optimized using the pH scouting procedure, as described in the instrument protocol, and 0.1 mM sodium acetate pH 5.5 was selected as immobilization buffer.

To establish a reproducible method for autoantibody detection, diluted serum samples were injected over the immobilized glycopeptide at different contact times (60, 120, 180 and 240 s). A 240 s injection was found to be optimum for increasing signal differences between positive and negative samples maintaining low signals in the reference channel. After each sample injection, surface was regenerated with two 60 s pulses of a solution 100 mM NaOH allowing the complete removal of specifically and unspecifically attached material from the surface. Following this protocol all further experiments were performed not over and above 100 measurements per channel.

Biosensor was used for the screening of high positive control sera at dilution 1:100 and 1:50. The analytical variability of the assay was checked repeating the same test (two sera, 15 runs each) or in different experiments (two sera, 12 runs performed once a week). The within-assay and between-assay coefficients of variation (SE/mean) were below 10% for sample dilution 1:100 and below 5% for dilution 1:50. Further serological analyses were performed at sample dilution 1:50, which presented lower signal average. One positive and one negative sample were used as controls each 15 measurements, verifying the stability of the probe upon a large number of cycles. For each measurement a sample volume of 150 μL was employed, thus small amount of 3 μL of patient serum was required for each assay.

Specific antibodies were detectable in some patients' sera. A typical sensorgram obtained when both healthy control and MS patient' sera were injected over the glycopeptide CSF114(Glc) is illustrated in Figure 1. Sensorgram show low association and very weak dissociation. All sample sera presented a similar sensorgram profile.

The column scatter of the data reported in Figure 2 summarizes all serological results obtained in the Biacore assays at sample dilution 1:50. The differences between the MS and BD mean values were significant (94.6 vs. 48.9 RU respectively), observing the higher values in MS subjects.

A receiver operating characteristic (ROC)-based analysis was employed comparing different cut-off values as sensitivity, specificity and likelihood ratios [38]. ROC curve for anti-CSF114(Glc) activity was constructed based on 61 cases with multiple sclerosis vs. 60 controls (Figure 3).

The panel of MS and BD sera was also tested against the synthetic peptide CSF114(Glc) following the previously validated ELISA procedure [16]. Both protocols have been evaluated and results compared.

Regarding the coating procedure, immobilization cost in terms of time (2 h aprox) and ligand quantities (1 μg for 100 measurements aprox) in SPR differs clearly from ELISA (coating overnight, 1 μg for each measurement). Moreover, one analysis in SPR can be performed in 10 min compared to at least one day in ELISA. On the other hand, ELISA offers the possibility to perform a large number of measurements simultaneously whereas the Biacore assay cannot analyse more than one sample contemporaneously. Methodologically speaking, SPR protocol appears in any case more convenient than ELISA.

Data obtained in ELISA were further compared with surface plasmon resonance results. Antibody responses were found to be significantly higher in MS than in healthy subjects both in SPR and ELISA. Specificity was similar for both methods, and sensitivity was increased from 26% for ELISA [16] to 36% for the SPR methodology.

Although ELISA exhibited overall results similar to those obtained from the SPR system, in some case only one of the assays was able to recognize specific antibodies. The SPR-based biosensor scored as positives nine MS patients (15%) further screened as negative in ELISA (Table 1, entry 4). Furthermore, the SPR biosensor was able to differenciate as positives seven cases in which ELISA recognized only IgG type antibodies (Table 1, entry 2), and two positive cases of IgM positive in ELISA (Table 1, entry 3). On the other hand, the SPR-biosensor did not detect specific antibody responses whereas the ELISA recognized ten (16%) positive cases to IgM and/or IgG (Table 1, entries 5–7). False positive cases (5%) scored in SPR-assay were similar to those observed in ELISA.

Briefly, immunoassays employed are able to detect glycopeptide CSF114(Glc)-antibody interactions presenting similar sensibility, but increasing sensitivity and recognizing slightly different patient populations, due to the ability of the SPR assay to detect low-affinity antibodies. Besides the two parameters sensitivity and specificity, also affinity can be considered as a parameter since the two methods differ in their underlying principle.