To leverage the full potential of Luminex technology and avoid the need for sample pre-treatment, we have designed a configuration that selectively captures and quantifies only the conjugated saccharide of each antigen. This aim can be achieved by coupling each bead with monoclonal antibodies which are specific to MenACWY antigens, serving as serotype-specific capture elements, so that each bead corresponds to a distinct antigen. The conjugate can then be identified using polyclonal antibodies against the carrier protein, i.e., CRM197, ensuring specific detection of the conjugate, so well distinct from the free saccharide, as shown in Fig. 1. The subsequent paragraphs will illustrate the development and the results of the challenge that was performed on the method.

Sandwich configuration against conjugated saccharide in glycoconjugate vaccines. Image created by BioRender.com

During coupling, the major parameter that necessitated optimization was the amount of mAb coupled on the beads. Therefore, for each serotype, the mAb amount was doubled, ranging from 0.5 µg to 16.0 µg of mAb for millions (mln) of beads. After this screening, the optimal amount of coupled mAbs resulted in 4.0, 1.0, 2.0, and 4.0 µg for MenA, MenC, MenW, and MenY, respectively. These values were decided based on the ability to achieve the highest and most consistent MFI, lowest blanks, and ultimately the highest S/N ratio. Subsequently, an evaluation was conducted of the initial sample concentration, the titration step, Ab-I, and Ab-II dilution. In this step of the optimization process, both whole IgG-specific and Fc-specific Ab-II have been taken into consideration. The second led to a reduction in unspecific binding, thereby translating into an enhancement of the S/N ratio. Moreover, a great improvement of the latter has been obtained after purification with protein G columns of anti-CRM197 whole serum. This purification step has been necessary to improve linearity, LOD/LOQ, and have the lowest blank values, thereby minimizing nonspecific interactions. After setting up all the coupling and assay steps, we optimized the layout to find a common condition for analyzing concomitantly as many antigens as possible. Initially, a triplex + singleplex format was developed (Fig.  2A). For this setup, we used a two-fold dilution step to analyze MenACW and MenY separately across two different plates. However, given the higher resources and time demands of this format, improvements were made to find a unique condition in tetraplex (Fig. 2B) configuration that would allow all four serotypes to achieve a linear regression curve. A longer dilution step and a starting sample concentration were selected, enabling at least five points of calibration, also for MenA, whose linear range is achieved only after further dilution points compared to the other serotypes.

A Triplex + singleplex (3 + 1) assay format in which MenACW and MenY conjugates are analyzed separately in two plates with 2 different starting sample dilutions. B Tetraplex assay format in which all the antigens are analyzed in the same plate using a common starting dilution and titration step. Picture created on Microsoft Excel365

For an immune-based assay, the specificity of the binding between antibody and ligand is crucial. In this case, the main concern was related to possible cross-reactivity among MenW and MenY conjugates and their beads-mAb complex, due to their strictly similar saccharide repeating units [MenW: → 6) -α-D-Galp-(α1–4)–α-D-Neu5NAc–(α2 →; MenY: → 6) α-D-Glcp-(α1–4)–α-D-Neu5NAc–(α2 →], which differ only at the stereogenic center in C4. Thus, to demonstrate the absence of any cross-reactivity among serotypes, we prepared and analyzed ad hoc formulations, taking out one conjugate at a time, starting from the singular drug substances orthogonally quantified with HPAEC-PAD. All the beads were used in this analysis to verify that the missing conjugated antigen was not detected. The results are expressed as %, of recovery, so the ratio between the µg/mL of conjugated saccharide quantified in each ad hoc formulation and the one in the complete mixture with all the antigens. The results in Table S1 in the Supplementary Material show that recoveries are between 82 and 120%, confirming that there is no cross-reactivity between the conjugates. Consequently, we proceeded to evaluate the specificity against the free saccharides. High free OS levels were evaluated to assess the potential impact on the quantification of conjugated saccharides. This evaluation is critical due to the potential interference arising from saccharide-specific capture facilitated by beads conjugated with monoclonal antibodies specific to serotypes. To address this, we quantified conjugated saccharide in the absence and presence of a spike solution of MenACWY oligosaccharides (OS). Moreover, even the spike solution has been analyzed alone on the same plate as the other two samples. The oligosaccharide mixture was prepared at reduced concentrations, specifically at 10% and 20% of the target concentration (data not shown), as well as at a concentration equivalent to the saccharide titers present in the vaccine (MenA OS: 20.0 µg/mL; MenC OS: 10.0 µg/mL; MenW OS: 10.0 µg/mL; MenY OS: 10.0 µg/mL), which serves to simulate an extreme condition as reported in Fig. 3.

Assessment of no impact by free saccharide in CS quantification analyzing samples reported above. Results are expressed as µg/mL of conjugated saccharide

In the same way, we investigated a potential impact on antigens quantification by free CRM197, which is part of the conjugate detection by anti-CRM polyclonal antibodies. Therefore, we analyzed a vaccine solution at target level, a solution of free CRM197 (100 µg/mL), and a vaccine sample at target level mixed with the spike of CRM197. The results in Fig. 4 confirm that CRM197 does not interfere with conjugated antigens quantification using this assay configuration.

Assessment of no impact by free protein in CS quantification analyzing samples reported above. Results are expressed as µg/mL of conjugated saccharide

After developing and optimizing the method, we tested the classical analytical performances in a feasibility study conducted at five concentration levels (L1 = 50%; L2 = 75%; L3 = 100%; L4 = 125%; L5 = 150%), which were prepared by diluting the lyophilized vaccine with different volumes to reach the theoretical concentration. These were analyzed and quantified randomly in five analytical sessions. The number of sessions was deemed sufficient to assess preliminary method performances in terms of precision, linearity, and accuracy for each serotype and level. Except for the target concentration (L3), all accuracy levels were analyzed in a single replicate (duplicate wells) in each session. Thus, in the end, this feasibility study was based on five determinations for L1, L2, L4, and L5, and ten determinations for L3.

For conjugated saccharide of each serotype, we reported the lower limit of detection and lower limit of quantification expressed as pg/mL in the well as shown in Table 1. These in-well concentrations of conjugated saccharide correspond to the calculated LOD (LOD = μ MFI (blank) + 3σ MFI (blank)) and LOQ (LLOQ = μ MFI (blank) + 10σ MFI (blank)) starting from the average (µ) and standard deviation (σ) of blank MFI.

The inter-assay precision (intermediate) has been evaluated with one replicate at all five levels across the five independent analytical sessions, as shown in Figure S1 of the supplementary material. Although the limited number of analytical sessions, a final coefficient of variation (CV%) widely under 20% was obtained. As previously cited, for each session, two analytical replicates of L3 (100%) were independently prepared and analyzed, with a resulting intra-assay precision (repeatability) of 9%, 7%, 6%, 6% for MenA, MenC, MenW, and MenY, respectively.

The method resulted in a good linearity, as reported in Figure S2 of the supplementary material, which shows the regression lines derived by plotting the theoretical (Th.) antigen amount as the dependent variable (Y) and the observed (Obs.) concentrations as the independent variable (X). The latter values were obtained by calculating the average (Avg.) of all sessions. Each serotype presents an R2 ≥ 0.99 and a slope between 0.88 and 1.07, indicating a good correlation between the theoretical and observed concentrations investigated from 50 to 150% of the concentration range of each conjugate.

As illustrated in Fig. 5, the accuracy was reported as an average bias for the five analytical sessions, with the upper and lower limits calculated at 90% confidence intervals. The FDA guidelines recommend a ± 20% of accuracy as the limit for validating immune-based assays [23]. In this context, the method is accurate for all levels and serotypes, except for the lower levels of MenY—L1 and L2—which have slightly larger intervals (122% and 121%). However, this slight overestimation might be dependent on the reference standard titer based on the glucose content, assessed via the orthogonal HPAEC-PAD method.

Upper and lower accuracy limits of each serotype and level, calculated with the 90% confidence interval

To evaluate the ability of this multiplex bioassay to monitor the degradation of conjugated saccharides, we employed both chemically induced and long-term stress conditions. Initially, we conducted mild hydrolysis using 1.0 mM trifluoroacetic acid (TFA) at 30 °C for 2 and 4 h, aiming to partially degrade the conjugated saccharides (CS) of the four antigens. Resulting data reported in Figure S3 of the supporting material show a steeper degradation pattern for MenA serotype, containing a phosphodiester bridge in its repeating unit, which makes this conjugated saccharide more susceptible to hydrolysis [24]. On the other hand, the other conjugates are hydrolyzed the same but with a flatter trend with respect to MenA as expected. Consequently, we applied a thermal stress to samples at 55 °C for 1, 2, 5, and 6 days, as done in a previous study on the same vaccine for the validation of an HPAEC-PAD method targeting MenA-CRM197 conjugated saccharide [16]. Results graphed in Fig. 6 showed a rapid decrease in CS content. However, a mismatch emerged between Luminex and HPAEC-PAD comparing the results for the more stressful conditions (5 and 6 days).

Stability study at 55 °C for 1–2-5–6 days

As demonstrated by Pecetta S. et al. [25], the transition midpoint temperature (Tm) of CRM197 is 46.3 °C, and is even lower in glycoconjugates, depending on the degree of glycosylation, as was studied for MenA-CRM197 conjugates [25]. Therefore, a possible reason for this underestimation could be related to the high temperature used, which was above the midpoint temperature of the carrier protein. The consequent misfolding of the carrier protein at this temperature should have interfered in both capture and detection. This suggestion has also been confirmed by fluorescence analysis observing emission spectra at 280 nm aimed at assessing protein folding (Data not shown). Consequently, we carried out a long-term stability study by incubating the reconstituted vaccine sample at 25 °C for 1, 2, and 4 weeks. As shown in Fig. 7 all the time points were analyzed by Luminex and the previously cited validated HPAEC-PAD method targeting CS content [16]. Results confirmed a good alignment between the two methods, further confirming the suitability of the Luminex assay to assess glycoconjugate content as compared to the current standard of care. This is indirectly confirming the evidence of correlation between Conjugate Saccharide quantification vs Total – Free Saccharide quantification previously shown by Rech et al. Additionally, this immunoassay was able to discriminate not only antigen integrity but also that of its carrier.

Stability study at 25 °C up to one month of MenA, MenC, MenW, and MenY respectively. Results were expressed as conjugated saccharide concentration