The epitope of the antibody used in the REAADS VWF activity assay is quaternary

Abstract

The REAADS VWF activity assay is often assumed to be specific for the A1 domain, the portion of VWF that binds platelet GPIbα. We tested this assay on the A1A2A3 region of VWF with each domain expressed independently of one another and together in combination as a tri-domain. The monoclonal antibody used in this assay is found to be insensitive to the single A domains and does not recognize free A1 domains as it is often assumed. Rather, we find the assay to effectively recognize A1A2A3 with the domains together in their natural glycosylated sequence context. Furthermore, type 2M and 2B Von Willebrand Disease mutations differentially disrupt the sensitivity of the assay, indicating that mutational effects on the structure of A1 in the A1A2A3 context concomitantly disrupt the epitope of the antibody. The REAADS VWF activity assay therefore is conformationally sensitive to the native quaternary association of the A domains together and it is not specific to freely exposed A1 domains.

The REAADS Von Willebrand Factor (VWF) activity assay is currently marketed by Corgenix Medical Co, located in Broomfield, Colorado, USA. It is designed as an elegant sandwich ELISA assay in which a monoclonal murine anti-VWF antibody recognizes a functional epitope on the VWF molecule to assess VWF activity levels relative to a supplied calibrator. It is stated in the product insert that the antibody is ” specific for the portion of VWF which binds platelets.” Designed for a 96-well plate format, the REAADS VWF activity kit allows a rapid and reproducible determination of VWF activities that is presumed to be sensitive to exposure of the platelet GPIbα binding A1 domain of VWF.¹

Having recently used this assay in published [1, 2] and ongoing studies, we were interested in where the antibody actually interacts with the VWF molecule. The original 1985 publication by Goodall and colleagues states that the antibodies RFF-VIII:R/1 & 2 neutralize VWF activity to different extents [3]. In 1997, Murdock and colleagues described the epitope of RFF-VIII:R/1 [4]. It is stated that RFF-VIII:R/1”specifically recognizes the GPIbα binding domain on VWF.” However, the location of the antibody’s epitope is inferred (not directly determined) from its ability to inhibit ristocetin-induced platelet agglutination, the binding of VWF to GPIbα, and its ability to block botrocetin binding to VWF [4]. It is further stated that the antibody is conformationally specific, as it does not bind to reduced VWF [4, 5]. RFF-VIII:R/1 is also implemented in the latex particle-enhanced immunoturbidometric activity assay [6,7,8]. It is often assumed that the REAADS antibody recognizes free functional A1 domains in VWF in publications that use the turbidometric assay [6] or in assays where REAADS is used as a standard against other inhibitors, such as aptamers [9,10,11,12]. We have also previously made this assumption with respect to trauma, but noted a lack of statistical significance when REAADS is normalized to VWF:Ag [2].

Our laboratory’s focus is on the biophysical and functional characterization of the VWF A domains and on the impact of clinically relevant Von Willebrand Disease (VWD) mutations. We therefore routinely express the A domains as recombinant proteins – either as separate single A1, A2, and A3 domains or as A1A2A3 tri-domain constructs. With these tools, we questioned, “does REAADS really bind the A1 domain and report free functional A1 domains in VWF?”.

In 2020 [1], we used the REAADS assay on our A1A2A3-tri-domain (Gln1238—Gly1874) and on recombinant full-length multimeric VWF and were able to confirm the disruption of the antibody’s epitope by the type 2 VWD mutations F1369I (2 M), S1285F (2 M), V1314D (2B), and P1337L (2B). These mutations, known to misfold the A1 domain in VWF, reduced REAADS activity relative to wildtype. More recently, while exploring the use of the REAADS antibody as a conformational probe, we tested the binding of it to our bacterially-expressed A1 domain construct (Gln1238—Pro1471). To our surprise, the antibody was incapable of detecting 20 µM of the supplied protein, while it worked well on ∼ 5 µM A1A2A3.

To further investigate the binding abilities of the antibody, we tested a panel of nine recombinant protein constructs (Fig. 1), A1 (residues Gln1238—Pro1471, expressed in E.coli), A2 (residues Gly1481-Arg1668, E.coli), A3 (residues Ser1671—Gly1874, E.coli), A1 (residues Gln1238-Pro1480, HEK293 cells) and A2 (residues Pro1480—Pro1678, HEK293 cells), and A1A2A3 (residues Gln1238—Gly1874, HEK293 cells) with the afore-mentioned von Willebrand disease mutations F1369I, V1314D and P1337L. Again, the antibody was able to detect all four A1A2A3 tri-domain variants with variable activities – depending on the structural defects caused by the mutation. However, none of the single domains were detected efficiently, suggesting that the antibody does require the A1A2A3 tri-domain as the smallest recognizable unit. It does not effectively recognize separately expressed A domains unless they are supplied undiluted at very high concentrations of 10 – 20 µM. Even at these concentrations the sensitivity is substantially lower for single domains than for A1A2A3.

Fig. 1

REAADS VWF Activity assay results for nine different recombinant A domains expressed and purified regularly in our laboratory [1, 13,14,15]. Left panel shows specific VWF REAADS activity (% Activity/µM protein). A1 and A2 were expressed in both E. coli and in HEK293 cells to account for the absence or presence of glycosylation, respectively. A3 was expressed in E. coli only as there is no reported glycosylation in A3 within VWF. A1A2A3 variants were all expressed in HEK293 cells. Note the difference in the Y-axis scale for the single A domains relative to the tri-domains. Per micromolar, the REAADS activity for single domains is less than 1, whereas for A1A2A3, the activity is 2–3 orders of magnitude greater. For protein concentrations and dilution factors used please refer to Supporting Table 1. Right panel shows reverse phase HPLC chromatograms that demonstrate the purity of our protein preparations

Our results provide direct experimental evidence that the REAADS antibody RFF VIII:R/1 is conformationally sensitive in that it does not efficiently recognize individual A domains expressed in isolation, but rather detects the three A domains together in a quaternary association. Our results, however, do not exclude the possibility that RFF VIII:R/1 recognizes domain pairs in complex, such as A1A2, A2A3, or A1A3, but prior published evidence from our studies shows that A1, A2, and A3 are thermodynamically coupled in both function and stability through quaternary structure. Comparative differential scanning calorimetry studies of A1A2A3 and the single A domain constructs show linkage in the stability of the A1, A2, and A3 domains together as a tri-domain construct and their quaternary association gets disrupted if a mutation is present in either A1 or A2 resulting in enhance platelet adhesion to the A1 domain [13, 16]. The differential loss in REAADS activity caused by the VWD variants in A1A2A3 indicates that localized structural consequences also disrupt the normal native quaternary association of the A domains. REAADS activity in VWF, therefore, is not a measure of exposure of the GPIbα-binding A1 domain, but rather a metric for the accessibility of all three A-domains in their native conformational arrangement relative to each other.

Shear stress also alters quaternary interactions. Blood flow elongates VWF multimers inducing tension in the A domains that enables GPIbα to bind to A1 at shear stresses greater than 720 dynes/cm² [17]. Quaternary interactions between the A domains dissociated by tension would allow A1 to bind GPIbα under high shear. VWF extension is reversible [17]. Returning to stasis allows VWF multimers to relax which may restore quaternary associations of the A domains and favor the dissociation of GPIbα. In principle, binding of RFF VIII:R/1 to A1A2A3 within VWF multimers would favor relaxed conformations under stasis where the A domains have the opportunity to re-associate. Tension applied to VWF under high shear would favor a loss of recognition by RFF VIII:R/1. REAADS is performed at equilibrium under static conditions with relaxed multimers of plasma VWF. As such, in this context and the results of Figure 1, the use of the phrase “activity assay” is a misnomer; particularly since its recognition of both gain- and loss-of-function variants is reduced relative to WT.

The REAADS antibody, RFF-VIII:R/1, is used in clinical assays. The HemosIL VWF activity assay (VWF:Ab) is an automated latex particle-enhanced immunoturbidimetric von Willebrand factor activity assay [6]. The same latex particle agglutination principle has also been used with GPIbα in using WT GPIbα (HemosIL VWF:RCo VWF:GPIbR) and with the G233V / M239V GPIbα double mutant (Siemens Innovance VWF:GPIbM) [18, 19]. These agglutination assays compare well with the gold standard of (VWF:RCo) activity implemented as an automated platelet agglutination assay (Siemens BC Von Willebrand reagent) [7, 20]. Comparing the manually implemented REAADS ELISA to these automated turbidometric assays reveals that REAADS underperforms with proportional error and variability resulting in some systematic bias [20, 21]. Latex HemosIL assays utilizing RFF VIII:R/1 show a reduction in VWF activity in all subclasses of type 2 VWD patients [6, 7, 22], however, Trossaert et al. did note a normal discrepancy of the Latex HemosIL assay in patients with the type 2M G1324A substitution [7], a variant known to stabilize the A1 domain [23], which could be an issue that is common for stabilizing 2M variants [24].

While RFF-VIII:R/1 performs consistently well in automated clinical assays, the REAADS ELISA re-mains a cost-effective way of determining VWF ”activity”. RFF-VIII:R/1 does not specifically report on A1 exposure, but rather on the native quaternary arrangement of all three A-domains. This circumstance is important as it should influence the interpretation of assays utilizing RFF-VIII:R/1.