In this study, recombinant BLyS/APRIL heterotrimers were produced by using a novel trimerization domain, purified, and extensively characterized. Their biologic activity was evaluated in vitro, and the effect of atacicept and other related soluble receptors on their activity was determined. A novel immunoassay was developed, and the occurrence of endogenous heterotrimers in healthy donors and patients with SLE and RA was demonstrated. Heterotrimer levels were compared in a subset of the patient samples with levels of homotrimeric BLyS and APRIL.
Several trimeric forms have been proposed for BLyS and APRIL. In principle, BLyS/APRIL heterotrimers could be any mix of stoichiometries (A2
B or AB2
). The heterotrimers produced for this study were predominantly A2
B, and their biologic activities were more similar to APRIL than to BLyS. The A2
B stoichiometry was obtained because of the characteristics of the two vectors that were used to express BLyS and APRIL in the production cell lines. In vivo
, BLyS/APRIL heterotrimers are likely to be formed stochastically in cells in which both BLyS and APRIL are generated, and the relative production of these proteins may influence the proportion of A2
B and AB2
heterotrimers that are endogenously formed. Furthermore, BLyS and APRIL appear to be differentially regulated [16
], and their levels may even be inversely correlated [23
], which is also suggested by a trend in our limited dataset of patient samples.
Differential TACI, BCMA, and BAFF-R receptor expression may favor biologic activity of the A2B versus the AB2 heterotrimers. The highest-affinity binding was observed between APRIL and BCMA-Ig, which was up to three orders of magnitude greater than observed for BLyS or heterotrimer binding to atacicept, BCMA-Ig, or BAFF-R-Ig. In the TACI-Jurkat assay, heterotrimer signaling was similar to that of APRIL homotrimers. The reduced potency of the heterotrimer ligands in the primary human B cell-proliferation assay compared with the homotrimeric BLyS may be explained by the predominant expression on circulating B cells of BAFF-R, to which our heterotrimers bind poorly owing to their predominantly A2B stoichiometry. Differences between receptor expression and/or B cell composition in the blood from different donors may explain the complex differences in dose-response curves observed in the primary B-cell assay.
Roschke et al
] also investigated the ability of soluble Ig fusion versions of TACI, BCMA, and BAFF-R to neutralize BLyS and BLyS/APRIL heterotrimers in a human B cell-proliferation assay. In that study, only TACI-Ig inhibited the BLyS/APRIL heterotrimer, and BCMA-Ig or BAFF-R-Ig was ineffective [6
]. In our studies, both atacicept and BCMA-Ig neutralized the activity of BLyS, APRIL, and the recombinant BLyS/APRIL heterotrimer in the TACI-Jurkat and human B cell-proliferation assays, whereas BAFF-R-Ig inhibited BLyS but exhibited little or no inhibition of heterotrimer or APRIL activity. This was expected and consistent with the observed binding of BLyS, but not APRIL, to BAFF-R. It could be hypothesized that the Roschke et al
. heterotrimers were composed of predominantly AB2
trimers; however, it is difficult to explain why these heterotrimers were not inhibited by soluble BAFF-R-Ig. This discrepancy would best be addressed by specifically purifying A2
B and AB2
heterotrimers, and assessing their activity in the presence of each of the three soluble receptors. We would speculate that native A2
B trimers are likely capable of binding to TACI and BCMA, whereas AB2
trimers should predominantly bind to TACI and possibly BAFF-R.
Soluble BLyS has been reported to form higher-order oligomers (for example, 60-mers) composed of multiple homotrimers, a cluster formation mediated by a flap-like region that is not present in APRIL [5
]. Although some early reports showed that BLyS existed in vivo
only in trimeric form [26
], a more recent study suggested that BLyS 60-mers may form naturally in vivo
and have biologic activity distinct from that of BLyS homotrimers [4
]. However, no reports have been published of native oligomeric APRIL. It has been proposed that signaling through TACI in mature B cells or plasmablasts requires higher-order BLyS oligomers or the cross-linking of APRIL through its binding to proteoglycans, whereas BAFF-R and TACI on primary B cells can bind and respond to all forms of BLyS [4
]. Our BLyS homotrimers signal less strongly than APRIL and the heterotrimers in the TACI-Jurkat assay, leading to a 20- to 25-fold difference in EC50
values between BLyS and APRIL or the heterotrimers. This finding appears to support the contention that BLyS oligomers may be required for optimal signaling through TACI. However, after extensive evaluation of the recombinant BLyS, APRIL, and heterotrimers using SEC-MALS and other techniques, we found no evidence for higher-order multimers or oligomerization of the ligands in this study (data not shown).
The inhibition of BLyS, APRIL, and the heterotrimers by atacicept is consistent with the observed effects of atacicept and/or murine TACI-Ig in preclinical and clinical studies. In mice and monkeys, atacicept reduces serum IgM levels and inhibits the IgM response to T-dependent antigen [28
]. It inhibits B-cell maturation and survival, age-related T-cell activation, and the T cell-independent marginal zone B-cell response, and significantly decreases levels of plasma cells in the spleen and bone marrow [28
]. However, atacicept does not reduce the numbers of B memory cells, which are active in long-term humoral immunity, as their survival is independent of BLyS or APRIL [31
]. These biologic changes in response to atacicept are associated with reduced disease scores and prolonged survival in SLE-prone mice [19
]. In Phase Ib studies, subcutaneous atacicept treatment reduced serum Ig, mature B-, and total B-cell levels in patients with RA or SLE [17
]. These actions were coupled with promising exploratory effects on disease-activity measures [17
]. Phase II/III trials are currently assessing the efficacy and tolerability of atacicept in patients with these conditions.
Our patient cohort data support and expand on a previous report showing higher serum levels of BLyS/APRIL heterotrimers in a limited sample of patients with autoimmune diseases (n = 15) compared with healthy controls (n = 6) [6
]. Roschke et al
] investigated whether BLyS/APRIL heterotrimers are elevated in patients with autoimmune diseases, and reported levels of up to ~230 ng/ml [6
]. In the Roschke et al
. study, a mAb reagent capable of immunoprecipitating the heterotrimers was identified, and an assay was performed to measure heterotrimers by using an ELISA strategy. Data were reported from two separate ELISA assays, using either the anti-heterotrimer mAb or an anti-BLyS pAb as capture antibodies to quantify the proteins in patient sera. In several cases, data from the two assays differed by an order of magnitude for the same sample. The authors postulated that the higher heterotrimer levels were detected with the pAb assay because of better capture abilities than the mAb-based assay, or a possible preference for either the A2
B or the AB2
forms of the heterotrimers. In contrast, the immunoassay described in the current study was designed to detect both the A2
B and AB2
forms of the heterotrimers, and heterotrimer levels were quantified by using laser detection of fluorescently labeled detection mAbs. The results from our assay suggest that the levels of heterotrimers in vivo
, even in very ill patients, are similar to or lower than those of the homotrimeric forms of BLyS and APRIL, with serum concentrations of native heterotrimers observed that were typically < 5 ng/ml. However, although heterotrimer levels are typically somewhat lower than those of the homotrimers, in certain patients, they may be found in similar or even greater concentrations. Larger studies with well-characterized assays are needed to assess accurately the relative levels of BLyS, APRIL, and heterotrimers, and to determine how common elevations are in their levels in clinical populations.
In the serum samples from patients with SLE and RA used in this study, levels of BLyS, APRIL, and heterotrimer were elevated in patients with SLE, compared with the sera of healthy donors. The detection of a single ligand (BLyS, APRIL, or heterotrimer) in more than one third of the samples may reflect specific control mechanisms for these TNF family members. The available data also suggest a trend toward correlation of BLyS and heterotrimer levels in patients with SLE, although this result is based on a very small number of samples with levels above the assay LOQ for both ligands.
It should be noted that, although our analysis shows that APRIL is detectable in a higher fraction of patients with SLE than in healthy controls, some controversy exists with regard to the role of APRIL in SLE. Several studies [6
] have shown that APRIL levels are elevated in patients with SLE. Koyama et al
] showed a trend between APRIL levels and anti-dsDNA Ab levels and a correlation with the British Isles Lupus Assessment Group (BILAG) index score of musculoskeletal disease. In contrast, Stohl et al
] reported an inverse correlation between APRIL and anti-dsDNA Ab levels and disease activity measured by SLEDAI score. Another recent study reported high APRIL levels in the sera of patients with SLE, but no correlation with SLEDAI score [23
]. The use of different APRIL assays may contribute to the disparate results currently in the literature, as some assays (including ours) show serum APRIL concentrations of 2-8 ng/ml (see, for example, Planelles 2004 (34)), whereas other assays yield much higher values, ≤ 2,500 ng/ml [14
]. A possible explanation for these discrepancies is the biochemical characteristics of the recombinant APRIL used to generate capture and detection reagents and to provide reference standards for each assay. In our experience, some commercially available forms of APRIL, when used as reference standards, yield inaccurate (high) or imprecise determinations of native APRIL levels in sera (unpublished observations).
The present analysis of our limited RA patient cohort shows that APRIL may be specifically elevated in RA, whereas BLyS or heterotrimer levels do not appear to be increased. Others have previously reported that levels of both BLyS and APRIL in patients with RA are higher in synovial fluid than in serum, suggesting that these ligands play an important role in the inflamed synovial compartment [15
]. It would be of interest to repeat these studies of matched RA serum and synovial fluid samples by using our APRIL assay.
Our analysis of the subset of serum samples from patients with SLE for which corresponding disease-activity data were available (that is, SLEDAI scores, anti-dsDNA Abs, C3 and C4 levels, and ESR) indicates that elevated heterotrimer levels may be associated with increasing SLE disease activity (Supplemental Tables 1 and 2 in Additional file 2
). Thus, further studies with larger group sizes are warranted to pursue this possible correlation of heterotrimers with SLE disease activity. Indeed, a larger dataset should be constructed to confirm data trends identified in this study, and as disease levels may fluctuate for individual patients, information on disease activity scores (at the time of blood draw) should be collected, along with the serum BLyS, APRIL, and heterotrimer levels.
In agreement with previous reports [15
], our data also show that BLyS is detectable in a higher fraction of patients with SLE than in healthy controls. BLyS levels also reportedly correlate with clinical disease activity in SLE [39
] and RA [25
]. Levels of BLyS have been shown to increase during anti-CD20 mAb-mediated B-cell depletion in patients with both SLE and RA, and to decline with B-cell repopulation in patients with SLE [25
]. Similar elevations in BLyS have also been observed in sera from patients with Sjögren's syndrome and non-Hodgkin lymphoma treated with an anti-CD20 mAb [41
]. In contrast, APRIL levels were reported to decrease during anti-CD20 mAb-mediated B-cell depletion in patients with SLE, whereas no significant changes in APRIL levels were observed in patients with RA undergoing anti-CD20 therapy [25
]. To address the discrepancies in reported APRIL data from various laboratories, these studies are currently being repeated using our validated APRIL ELISA assay.
Given that heterotrimers have similar binding properties and in vitro
activities to the BLyS and APRIL homotrimers, and may be present in sera in similar amounts to APRIL and BLyS, we postulate that they may likely play similar biologic roles to BLyS and APRIL in B-cell development and differentiation. As our recombinant A2
B heterotrimers behave much like APRIL in vitro
, we also speculate that AB2
heterotrimers would function more like BLyS, for example, playing a role in early B-cell survival and selection. BLyS exists in both soluble and transmembrane-bound forms, whereas APRIL is believed to exist only in a soluble form, with the exception of the TWEAK-APRIL fusion protein TWE-PRIL [43
]. We did not test whether BLyS/APRIL heterotrimers were expressed on the cell surface, but if so, their potential for exerting biologic effects would presumably be expanded.
Whether native heterotrimers play a biologic role distinct from their homotrimeric counterparts remains to be determined. Our data suggest that investigating forms of BLyS and APRIL other than the conventional homotrimers in patients with autoimmune diseases may help to elucidate the pathology of such disorders and may also reveal additional disease markers and targets for treatment.