Two commercial preparations of IVIg were tested, namely GamunexTM from Talecris Biotherapeutics ltd. (Toronto, ON, Canada) and Gammagard from Baxter Corporation, (Mississauga, ON, Canada). Both IVIg preparations were similar in terms of their proportions of IgG1, IgG2, IgG3 and IgG4, which represents 64% ± 3%, 28% ± 4%, 4% ± 2% and 4% ± 1% of the total IgG content, respectively (data not shown and [26] ).

Reactivity of IgG in IVIg was evaluated by ELISA using a selection of antigens constituted of human transferrin, whole molecules and Fc fragments of murine IgG (Jackson Immuno Research Laboratories, West Grove, PA, USA), bovine actin and histone, salmon dsDNA and human fibronectin (Sigma-Aldrich, Oakville, ON, Canada), human ferritin and thyroglobulin (Calbiochem, San Diego, USA) [25] . Bovine histone and salmon dsDNA were adsorbed at 50 μg/ml in 100 mM citrate buffer pH 6.0 [27] . All other antigens were adsorbed at 10 µg/ mL in 100 mM carbonate buffer pH 9.6 [25] . Bound human IgG were revealed with peroxidase-conjugated goat anti-human IgG gamma-chain specific antibodies (Jackson ImmunoResearch) or peroxidase-conjugated mouse antibodies specific to human IgG1, IgG2, IgG3 and IgG4 subclasses (Southern Biotech, Birmingham, Alabama, USA). A solution of 10 mM Phosphate buffered saline pH 7.0 containing 5% fetal bovine serum (Invitrogen, Burlington, ON, Canada) was used as blocking buffer and as a diluent for IVIg as well as all peroxidase-conju- gated antibodies. Non-specific human IgG bindings were determined by using wells coated with 100 mM carbonate buffer alone or containing 0.001% FBS. When indicated, the strength of IVIg reactivity was determined using two-fold serial dilutions ranging from 250 µg/ml to 8 µg/mL and revealed according to each IgG subclasses as described above. All ELISA analyses were done in triplicate.

In a first step, the reactivity pattern was determined for IVIg using a final concentration of 25 µg/mL of total IgG by a direct detection in ELISA (Figure 1). Background for IVIg binding was 0.2 ± 0.1 (O.D.) as determined by controls done with diluent (FBS) and uncoated wells (CTL-neg). Both IVIg preparations were tested simultaneously and showed similar capacity to bind dsDNA, actin, ferritin, thyroglobulin and transferrin. In this ELISA system, the total IgG anti-species cross-reactivity of IVIg for murine IgG molecules was comparable to that of transferrin, thyroglobulin and ferritin antigens Analysis done with both IVIg adjusted to 250 µg/mL resulted in O.D. varying from 1.0 to more than 3.5 in most cases (data not shown) indicating that these reactivity

patterns were proportional to the amount of human IgG in tested samples.

To compensate for the lower concentration of IgG3 and IgG4, IVIg-A and IVIg-B were adjusted to 250 µg/mL and their reactivity patterns were determined for each IgG subclasses (Figure 2). Background reactivity for IgG1, IgG3 and IgG4 was higher for IVIg-A when compared to IVIg-B, resulting in a similar level of specific binding for most selected antigens. As expected, IgG2 was reactive only for mouse Fc-IgG, which is known to have several glycosylation sites. IgG1, IgG3 and IgG4 showed comparable pattern of reactivity and even though IgG1 is about 10-fold more abundant than IgG3 and IgG4, their level of reactivity was also similar. In fact, IgG3 and IgG4 were characterized by a higher level of reactivity for bovine actin, salmon dsDNA, as well as human ferritin and thyroglobulin.

Based on the above observations, titration of IVIg reactivity was done to evaluate the strength of IgG1, IgG2, IgG3 and IgG4 binding on human ferritin and thyroglobulin, which are targets naturally present in human blood [18] . Serial dilution concentrations were established in function of the total IgG content while reactivity was detected using antibody specific for each IgG subclass. Bovine actin reactivity as well as background was also assessed in parallel using the same serial dilutions of IgG in both IVIg preparations. These results are representative of four independent experimentations using IVIg-A and three using IVIg-B. The relative reactivities of IgG subclasses for actin, ferritin and thyroglobulin were similar to those observed above for both IVIg (Figure 3) and were all directly proportional to their final concentration. IVIg-A was showing high IgG1 background reactivity as above. Overall, the titration for thyroglobulin reactivity corroborated that the high level of anti-thyrog- lobulin antibodies, present in both IVIg, is mainly confined to the IgG4 subclass.