Avidity of onconeural antibodies is of clinical relevance

Abstract

Onconeural antibodies are important in the detection of paraneoplastic neurological syndromes (PNS). The avidity of Hu, Yo, and CRMP5 antibodies from 100 patients was determined by immunoprecipitation (IP), and 13 of the Yo positive sera were also tested by surface plasmon resonance (SPR). There was a significant association between the results from IP and SPR. Yo antibodies had higher avidity than Hu and CRMP5 antibodies, and both high- and low-avidity antibodies were associated with tumors and PNS. High-avidity Yo antibodies were mainly associated with ovarian cancer, whereas high-avidity Hu and CRMP5 antibodies were mainly associated with small-cell lung cancer. Low-avidity CRMP5 and Yo antibodies were less often detected by a commercial line blot than high-avidity antibodies. The failure to detect low-avidity onconeural antibodies may result in under diagnosis of PNS.

Introduction

Approximately, 1 % of all cancers are associated with paraneoplastic neurological syndromes (PNS). Cancer types frequently associated with PNS are lung cancer, ovary cancer, breast cancer, testis cancer, lymphoma, plasmacytoma, and thymoma [1, 2]. PNS occurs due to an autoimmune cross-reaction that recognizes antigens expressed on cancer cells and similar epitopes found on cells of the nervous system [1, 2]. This immune response can suppress the growth of the cancer, but when activated lymphocytes and onconeural antibodies cross the blood–brain or blood–nerve barriers, it can lead to an autoimmune-driven damage to cells of the nervous system and thus to a PNS [1, 2].

Paraneoplastic neurological syndromes are frequently associated with onconeural antibodies that can be detected in the serum and cerebrospinal fluid of patients with underlying cancer. These antibodies are important markers for early diagnosis of cancer, since the symptoms of PNS often become visible before the cancer itself can be detected [1–3]. The well-characterized onconeural antibodies used in routine diagnostics are anti-Hu, anti-Yo, anti-CRMP5, anti-amphiphysin, anti-Ma, anti-Ri, and anti-Tr [4, 5]. Several different assays have been validated for detecting these antibodies in clinical service laboratories.

It has been shown previously that Yo and Hu antibodies have different avidity, that is, different overall binding strength to their antigens [6]. High-avidity antibodies may have pathogenic importance as they have been associated with onset and severity of autoimmune diseases [7, 8]. Antibody avidity may also influence the sensitivity of serological tests, for example, low-avidity acetylcholine-receptor antibodies are not always detected by radioimmunoassay, which is frequently used in the routine diagnosis of myasthenia gravis [9]. In the present study, we have determined the avidity of the three most common onconeural antibodies, namely anti-Hu, anti-Yo, and anti-CRMP5. The aim was to study whether high or low antibody avidity could predict differences in PNS or cancer type and whether the avidity could influence the sensitivity of the diagnostic methods commonly used to detect onconeural antibodies.

Materials and methods

Patients

Onconeural antibody avidity was measured in sera from 100 patients, 45 with anti-Hu, 31 with anti-Yo, and 24 with anti-CRMP5. Positive sera were ascertained in the diagnostic service activity of the Neurology Research Laboratory, Haukeland University Hospital, Bergen, Norway, having been submitted by neurologists in practice. The following onconeural antibodies had been analyzed by both immunoprecipitation and line blot assays (www.ravo.de) [10]: Hu, Yo, Ri, amphiphysin, CRMP5, and Ma1/2. Avidity had been determined previously in 35 of the Yo and Hu positive patient sera [6]. Hospital medical records for all patients were reviewed, but in 18 of the patients, we had no information on the PNS and/or cancer diagnosis.

Immunoprecipitation (IP)

In vitro transcription–translation was performed using the TNT-coupled reticulocyte lysate system (www.promega.com). cDNA coding for either HuD, CDR2, or CRMP5 was cloned into expression vectors containing either a T3 or a T7 promoter. These vectors have been described earlier [11–13]. 35S-methionine (www.gehealthcare.com)-labeled-recombinant onconeural CDR2, HuD, and CRMP5 proteins were made by in vitro transcription and translation of the selected cDNAs in the TNT T3 or T7 coupled reticulocyte lysate system (www.promega.com) [14]. IP was performed in Multiscreen 96-well filtration plates (MABVN0B50, www.millipore.com). Each well was preincubated with 200 μl of buffer A (150 mmo/l NaCl, 20 mmol/l Tris–HCl, and 0.01 % azide, pH 8.0) for 1 h at room temperature. Buffer A was then removed, and the wells were then blocked for 2 h with 200 μl blocking solution [1 % BSA (www.sigma.com) in Buffer A]. Thereafter, the wells were washed twice with 0.05 % Tween-20 in buffer A. After washing, 200 μl of buffer B (0.1 % BSA and 0.05 % Tween-20 in buffer A) was added to each well and the plates were stored at 4 °C in Buffer B. Triplicate samples of either 35S-labelled CDR2, HuD, or CRMP5 protein and patient sera diluted 1:100 to 1:1,000 in buffer B were incubated at 4 °C overnight. The dilution of each patient’s serum was optimized to yield a concentration of specific antibody high enough to give a signal, but sufficiently low not to impact avidity measurements. The next day, 50 μl of a 50 % (vol/vol) slurry of suspended protein A Sepharose (www.gelifesciences.com) in buffer B was added to each well of the multiscreen plates followed by addition of the immune complexes. The plates were incubated on a shaking platform at 4 °C for 45 min at 120 rpm and washed three times with buffer B using a vacuum manifold and incubated with 150 μl of buffer B on a shaking platform for 5 min at 4 °C. To determine the onconeural antibody avidity, 150 μl of 8 M urea was added to the wells (yielding a final concentration 7 M). The plates were then incubated at 4 °C on the shaking platform for another 30 min before the plates were washed twice as described above. The multiscreen wells were then dried overnight, and 20 μl scintillation fluids (MicrocintTM-0, Packard, CT, USA) were added to each well. The radioactivity was measured in a β-counter (Topcount NXT, Packard, USA) using 30 s per sample. The avidity index (AI) was calculated from the formula: AI = cpm (with urea)/cpm (without urea). All experiments were repeated in triplicate at least three times. The results were expressed as mean values of these. Low avidity was defined as antibodies with an index <30 %, medium avidity as index 30–50 %, and high avidity as index >50 % [6, 15].

Surface plasmon resonance (SPR)

The SPR analyses were carried out using a Biacore 3000 instrument and the GST Capture Kit (www.gehealthcare.com) [16]. CDR2 was expressed as glutathione S-transferase fusion protein (GST-CDR2) in Escherichia coli (BL-21, codon+) grown in LB medium containing ampicillin (100 μg/ml) at 37 °C. Eight hours after induction by 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) at 28 °C (A₆₀₀ nm = 0.7–0.8), the cells were harvested and disrupted by French press, and the supernatant was applied to a glutathione-Sepharose 4B (www.gehealthcare.com). GST-CDR2 was eluted with 50 mM Tris–HCl, pH 8.0, 5 % glycerol, 10-mM-reduced glutathione, followed by size-exclusion chromatography on a HiLoad Superdex 200 (www.gehealthcare.com) with 20 mM Na-Hepes, 200 mM NaCl, pH 7.0. GST-CDR2 was captured on anti-GST antibody immobilized covalently on a CM5 sensor chip, followed by injection of purified IgG from patient’s serum using a Melon Gel IgG Spin Purification Kit (www.piercenet.com). Melon Gel Resin binds and removes serum proteins, thereby allowing pure IgG to be collected in the flow-through fraction. In the controls, GST-CDR2 was replaced with buffer. At the end of the binding assay, GST-CDR2 was removed by injecting glycine–HCl, pH 2.0, and the chip was subsequently loaded with new GST-CDR2. The BIA evaluation 3.2 software (www.gehealthcare.com) was used for analysis of the sensorgrams and calculation of the dissociation constant (K _D).

Statistics

The nonparametric Mann–Whitney test was used to compare independent group of sampled data, and the Pearson’s chi-squared test was used to compare the results of the two avidity tests.

Results

The IP assay showed that 8/31 (26 %) Yo positive sera contained low-avidity antibodies, 6 (19 %) medium-avidity antibodies, and 17 (55 %) high-avidity antibodies. Among the 45 Hu positive sera, 16 (36 %) contained low-avidity antibodies, 11 (24 %) medium-avidity antibodies, and 18 (40 %) high-avidity antibodies. Among the 24 CRMP5 positive sera, 13 (54 %) contained low-avidity antibodies, 2 (8 %) medium-avidity antibodies, and 9 (38 %) high-avidity antibodies (Fig. 1). Yo antibodies had higher avidity than Hu and CRMP5 antibodies (p = 0.004).

Fig. 1.

Box plot of the avidity index for Yo, Hu, and CRMP5 antibodies

Thirteen Yo positive sera, with various avidity determined by the IP assay (3 with low avidity, 4 with medium avidity, and 6 with high avidity), were tested by SPR. There was a significant correlation between the results of the IP avidity index and SPR (K _D) (r = 0.75, p = 0.003; Pearson’s chi-squared test) (Fig. 2).

Fig. 2.

Correlation between the dissociation rate constants (K _D) obtained by binding kinetic analyses with surface plasmon resonance and the avidity index obtained by immunoprecipitation

Onconeural antibodies with high avidity were associated with female gender (p = 0.007), but not age (p = 0.920). The majority of the tumors among the patients with high-avidity Yo antibodies were ovarian cancer. The following tumors were present among the patients with low-avidity Yo antibodies: lung adenocarcinoma, Hodgkin’s lymphoma, colon cancer, and astrocytoma. The frequency of paraneoplastic cerebellar degeneration was similar in the low- and high- avidity Yo groups, except for the subgroup of male patients with paraneoplastic cerebellar degeneration as part of paraneoplastic encephalomyelitis. In patients with Hu or CRMP5 antibodies, small-cell lung cancer was associated with high-avidity antibodies, whereas in the low-avidity groups, a mixture of different cancers was seen. Paraneoplastic encephalomyelitis including limbic encephalitis and peripheral neuropathy was the main PNS in both the high and low Hu or CRMP5 antibody avidity groups.

Line blot positivity was associated with high onconeural antibody avidity: mean 0.57 ± 0.29 versus 0.26 ± 0.17; p < 0.001. This was significant for the CRMP5 (p < 0.001) and Yo antibody avidity (p = 0.003), but not for Hu antibody avidity (p = 0.120).

Discussion

Our findings that onconeural Yo antibodies have higher avidity than Hu and CRMP5 antibodies are in accordance with previous data from a small cohort of patients [6]. In both studies, urea treatment in the IP assay was used to break the antibody–antigen complexes. As urea is a chaotrope that can partially denature proteins, SPR was used to confirm Yo antibody avidity. SPR is a powerful technique that permits measurement of biomolecular interactions in real-time in a label-free environment [17]. We found a significant correlation between the SPR and IP avidity measurements suggesting that the heterogeneity in avidity seen in patients with Yo, Hu, or CRMP5 antibodies reflects differences among the patients and not the effect of urea.

Both high- and low-avidity onconeural antibodies were associated with tumors and PNS, although both the tumors and PNS were more heterogeneous in the low-avidity group. Interestingly, high-avidity Yo antibodies were mainly associated with ovarian cancer, whereas high-avidity Hu and CRMP5 antibodies were mainly associated with small-cell lung cancer. This is in line with previous results showing that these antibodies are associated with certain cancer types regardless of their avidity [10, 18]. We have previously not found any association between Yo antibodies and histological subtypes of ovarian cancer [12], but the influence of antibody avidity has not been studied. Furthermore, whether the higher avidity of the Yo antibodies compared to the Hu and CRMP5 antibodies implicates a more direct involvement in the pathogenesis of PNS, such as paraneoplastic cerebellar degeneration, is also unknown. Yo antibodies have been shown to induce Purkinje cell death in cerebellar slice culture [19], but the impact of antibody avidity in this context remains to be determined. The association between high-avidity onconeural antibodies and female gender, but not with age, is also confirmed in our data. Whether this is true also for other high-avidity autoantibodies is not known. Generally, most seropositive PNS patients are females, particularly Yo positive patients, which are strongly associated with gynecological cancer [10, 18].

It has been shown recently that the IP test is more sensitive for detecting onconeural antibodies than the commercial line blot test [10]. The present study shows that this difference in sensitivity is due partly to antibody avidity. In particular, CRMP5 and Yo antibodies with low-avidity were less often detected by line blot using recombinant proteins. One explanation for this is that low-avidity antibodies may bind more readily to the native form of the antigen in a fluid phase system, such as in the IP assay, than to the denatured form used in the line blot. In the present study, the patient sera were used at similar concentrations in the IP and line blot tests. However, the IP assay includes the use of radioactive labeling which may also increase its sensitivity.

The failure of conventional immune assays such as line blots to detect low-avidity onconeural antibodies, especially anti-CRMP5 and anti-Yo, demonstrates the need for more sensitive routine assays. This is of great clinical importance since antibody positivity supports the diagnosis of a neurological disorder as paraneoplastic. However, while antibody positivity defines the neurological disorder as paraneoplastic, it does not predict the type of neurological presentation, simply the type(s) of cancer we need to search for [3].