NMR Assays for Estimating the O-Acetyl Content of Meningococcal Polysaccharide Serogroup A in Quadrivalent Conjugate Vaccine Formulation

Abstract

The use of multivalent glycoconjugate vaccines has dramatically contributed to reduce the incidence of meningococcal infectious disease. The advanced structural characterization of polysaccharide conjugates leads to enhancements in the quality and control of the products. Here, we report a novel nuclear magnetic resonance (NMR) method to confirm the identity and structural conformity (e.g., O-acetyl content) of saccharide antigens that comprise a licensed tetravalent meningococcal serogroups A, C, W, and Y vaccine. For the first time, the NMR methodology is applied on a formulation (licensed vaccine) containing a large excess of excipient (i.e., sucrose) without analytical sample pretreatment. This work confirms the applicability of a rapid and easy NMR assay on a multivalent conjugate vaccine, which might be extended to other combination vaccines that are already licensed or in clinical development.

Introduction

Multivalent polysaccharide (PS) conjugate vaccines have been licensed to prevent bacterial infection caused by serogroup A, C, W, and Y of Neisseria meningitidis.¹ A pentavalent combination vaccine including the additional serogroup X is currently under clinical development and the results of a phase I clinical trial have been recently published.²

When different antigens are combined to form a multivalent vaccine, a panel of tests that is able to characterize the formulation in terms of potency, stability, identity, and strength has to be developed. Although tests might be in place for single components, it is important to determine if their combination gives rise to compatibility issues that might change some characteristics of single antigens (i.e., stability, immunogenicity). Physicochemical and immunological tests are commonly used to compare the biological activity of individual antigens before and after the combination, allowing for the quantification of any chemical/immunological interference. The structural integrity and stability of PS-based antigens is a critical quality attribute to be assessed after combination and is a common indicator of the product stability and vaccine potency. For instance, the generation of the unconjugated saccharide or the release of functional groups (i.e., O-acetylation, OAc) relevant for a functional PS epitope can dramatically affect the vaccine potency.

The meningococcal (Men) serogroups A, C, W, and Y capsular PS repeating units have the chemical structures →6)-α-d-ManpNAc(3/4OAc)-(1→OPO₃→; →9)-α-d-Neu5Ac(7/8OAc)-(2→; →6)-α-d-Galp-(1→4)-α-d-Neu5Ac(7/9OAc)-(2→; →6)-α-d-Glcp-(1→4)-α-d-Neu5Ac(7/9OAc)-(2→, respectively (Figure 1).³ MenA PS is partly O-acetylated at position C₃ or C₄, while all MenCWY PS contain Neu5Ac moiety in the relative repeating units that are partially O-acetylated at the glycerol chain of Neu5Ac (at position C₇ and C₈ for serogroup C and at position C₇ and C₉ for serogroup W and Y).³

Figure 1.

Structural representation of MenA-, MenC, MenW, and MenY-CRM₁₉₇ conjugates (R: H, acetyl). The protons used as the label for the identity test of each antigen are highlighted in blue.

Nuclear magnetic resonance (NMR) spectroscopy is currently used in the advanced characterization and quality control of existing and novel products.^4,5 NMR applications have been first developed and implemented for structure determination and for identity/integrity testing of the carbohydrate components. NMR provides a single generic tool that is able to replace a wide range of wet chemical/chromatographic approaches that have been largely used in the past whilst providing more precise data. NMR-based methods for the identification and quantification of known impurities that have previously been determined by indirect approaches, or not suspected to be present, have been developed,⁴⁻⁶ as well as for absolute quantification of PS content.^7,8

¹H, ¹³C, and ³¹P NMR methods have been applied for the characterization and purity evaluation (both product and process related impurities) of manufacturing intermediates starting from the purification process of native PSs, the activated intermediates before protein coupling and to the final monovalent bulk conjugates.^4,6 For multivalent vaccines, the complexity due to the combination of several PS antigens with common motifs in their repeating units which results in diffused overlapping of signals in the NMR profile, the presence of a large excess of the excipient for lyophilization (e.g., sucrose, mannitol) which dominates the spectrum, as well as the relatively low concentration of antigens as carbohydrate content (i.e. 1–10 μg/dose) have strongly limited the application of NMR methodologies. Recent advances in liquid-state NMR spectroscopy and, in particular, (a) higher peak to peak resolution at stronger magnetic field (proton frequency up to 1 GHz) and (b) cryoprobe technology (increased sensitivity of the probe coil which also results in shorter acquisition time) have provided insights into the challenges.

Herein, we report the development of a novel NMR-based assay applied to the tetravalent vaccine against meningococcal serogroup A, C, W, Y infection, licensed for human use in a two-vial presentation (a lyophilized monovalent MenA component with sucrose/potassium dihydrogen phosphate as excipient to be reconstituted with the liquid MenCWY component immediately before injection),⁹ and an investigational liquid formulation currently in clinical trial.^10,11

The commercial Menveo vaccine contains size-reduced oligosaccharides of serogroups A, C, W, and Y, conjugated to a nontoxic mutant of Corynebacterium diphtheriae toxin, termed “cross reacting material 197” (CRM₁₉₇). Each 0.5 mL dose of vaccine contains 10 μg of MenA oligosaccharide and 5 μg of each of MenC, MenY, and MenW oligosaccharides and 32.7–64.1 μg of CRM₁₉₇ protein.⁹

This methodology has confirmed the applicability of NMR spectroscopy on a final combination conjugate vaccine as an identity and structural conformity test (i.e., O-acetyl content of the MenA antigen) of the carbohydrate moieties.

Results and Discussion

Identity Test of MenACWY Antigens

In accordance with the strategy applied for the development of an identity test method successfully developed for the 23-valent pneumococcal PS-based combination vaccine (Pneumovax23, Merck), where one proton NMR signal as the “probe signal” for each antigen has to be compared with the reference¹² and also with the experience collected internally on the method developed for MenACWY monovalent purified PS antigens; other process intermediates and bulk glycoconjugates (recently included as the reference method in the USP monograph ⟨198⟩),¹⁸ where five signals for each antigen have been used for comparison with references, the feasibility of a new identity test has been investigated.⁶

As shown in Figure 1 (schematic representation of MenACWY-CRM₁₉₇ conjugates), the proton signals H₁^ManNAc (proton at position C₁ of ManNAc residues) for MenA, H_3eq (proton at position equatorial C₃ of the Neu5Ac residues O-acetylated at position C₈) for MenC, H₁^Gal (proton at position C₁ of the Gal residues) for MenW, H₁ (proton at position C₁ of the Glc residues) for MenY have been selected as probe signals. For MenC, the protons at C₃ equatorial position of Neu5Ac residues resonate at different chemical shifts, depending on the position (C₇ or C₈) of the OAc group (2.92 and 2.79 ppm for H_3eq^Neu5Ac-7OAc and H_3eq, respectively). As published in the literature,¹⁹ the H_3eq^Neu5Ac-7OAc signal includes both the 8OAc residues flanked at their reducing side by an 8OAc residue and by a 7OAc residue. Because H_3eq overlaps with the H_3eq^Neu5Ac signals of MenW and MenY, the H_3eq peak has been selected for identity of MenC.

The ¹H NMR profile of Menveo vaccine, acquired by using a diffusion filter pulse sequence to suppress the sucrose signals (Figure 2), shows the peaks of the four antigens used as identity probe signals for MenA, MenC, MenW, and MenY antigens (Table 1).

Figure 2.

¹H NMR spectra of the quadrivalent licensed Menveo vaccine. Labels to facilitate the peak assignment for MenA, MenC, MenW, and MenY antigens are shown.

Table 1. Chemical Shift (δ) of H₁^ManNAc, H_3eq, H₁^Gal MenW, H₁ Peaks Used as Identity Probe Signals for MenA, MenC, MenW, and MenY Antigens, Respectively.

	MenA	MenC	MenW	MenY
δ (ppm)	∼5.46	∼2.79	∼5.11	∼5.08

Quantification of O-Acetyl Content of the MenA Antigen

The quantification of the OAc level allows for the evaluation of the PS antigen conformity, which may also occur because of adverse events which could provoke the release of O-acetyl groups. The MenA plain PS is largely O-acetylated at position C₃ (>90%) and slightly at position C₄ (Figure 1).^6,13 The importance of OAc in MenA antigen immunogenicity has been confirmed in preclinical study. Specific antibodies elicited in post-immunization human sera mostly identified O-acetylated PS residues impacting the degree of antibody inhibition. Comparative immunogenicity studies in mice revealed that de-O-acetylated antigens resulted in a marked loss of immunogenicity (total IgG) and, most relevantly, in their ability to induce functional bactericidal antibodies.¹⁴ In a phase III partially blinded, controlled study [i.e., 1170 healthy subjects aged 18–25 years were randomized (1:1:1)], two lots of meningococcal ACWY-tetanus toxoid conjugate vaccine (MenACWY-TT) that differed in serogroup A PS OAc levels(68 vs 92% OAc), respectively, were evaluated in terms of immunogenicity (i.e., rabbit serum bactericidal activity) and safety in comparison to a licensed MenACWY PS vaccine (82% serogroup A OAc). The MenA-TT conjugates with the OAc level of 68 and 92% resulted in comparable vaccine immunogenicity, considering that, however, the vaccine population was seropositive at the time of vaccination.¹⁵ As a matter of fact, all the meningococcal serogroup A conjugates contained in licensed vaccines (MenAfriVac—Serum Indian Institute; Menveo—GSK Vaccines; Menactra—Sanofi Pasteur; Nimenrix—GSK Vaccines later taken over by Pfizer) are largely O-acetylated. However, no evidence has been collected to define the minimal OAc level affecting the immunogenicity of meningococcal serogroup A vaccines so far.¹⁵

High relevance of O-acetyl moieties of MenA PS could also be related to the presence of this decorative group within the saccharide ring, probably influencing and specifying the epitope in a unique arrangement that could not be easily mimicked in its absence.^16,17

For the monovalent MenA PS and conjugate, the OAc content can be estimated by integration of the H₂^ManNAc-3OAc peak (proton at position C₂ of ManNAc residues O-acetylated at C-3), H₂ peak (proton at position C₂ of ManNAc residues O-acetylated at C₄), and H₂^ManNAc-deOAc peak (proton at position C₂ of ManNAc residues without OAc), in comparison to H₁ ManNAc (proton at position C₁ of ManNAc residues). The total OAc level can be obtained by the sum of H₂ and H₂^ManNAc-4OAc peak integration as well as by integrating the H_3/4 peak.⁶ The same method already developed for the monovalent MenA conjugate cannot be applied to the MenACWY combination vaccine as the NMR profile exhibits large overlapping signals because ofthe presence of the four antigens (Figure 2). However, the H₁^ManNAc peak at approximately 5.5 ppm, which appears as a complex peak, is not affected by the presence of other antigens. As reported in Figure 3a, the ¹H–³¹P heteronuclear multiple-bond correlation (HMBC) NMR spectrum provides a clear assignment of different MenA residues, depending on different OAc patterns, terminal residues, and so forth.

Figure 3.

(a) ¹H–³¹P HMBC spectrum of the MenA-CRM₁₉₇ sample; (b) ¹H NMR spectral window where H₁^MAnAc of MenA-antigen resonates. The vertical bars show the integral regions identified for measuring the OAc and non OAc forms. Labels to facilitate the peak assignment are shown.

The H₁^ManNAc nuclear spin resonates at a slightly different frequency and correlates with the diester phosphate group (the monoester phosphate signal falls at higher chemical shift¹⁹), depending from the specific magnetic environment generated by (a) presence of the OAc group at position C₃ or C₄, (b) geminal with the di-ester phosphate group of the open ring unit at reducing end terminus (unit #2) (see Figure 1). The HMBC spectrum shows six ¹H signals (excluding the satellite peaks), originated by the scalar long–range correlation (H–C–O–P) with diester ³¹P signals (separated peaks for OAc, −5.2 ppm, and non OAc, −5.7 ppm, forms). As shown in Figure 3a, the proton at position C₁ of several ManNAc species O-acetylated at position C₃ (3OAc) or C₄ (4OAc) or non-O-acetylated (deOAc) resonate at different chemical shift values. From downfield to upfield, the signals H_1a, H_1a^4OAc, H_1a, H_1b^3OAc/H_1b, H_1c,d,e^deOAc, H_1b resonate in the sequence and some signals partially overlap each other.

As shown in Figure 3b, integral areas have been defined according to the assignment of different H₁^ManAc species and also considering “operative” reasons (i.e., the presence of a valley at ∼5.46 ppm) related to the overall peak shape. In essence, the OAc content has been calculated according to the formula % OAc = H₁/H₁ × 100, where H₁ (range 5.52–5.34 ppm) is the sum of H₁^OAc (range 5.52–5.46 ppm) and H₁ (range 5.46–5.34 ppm) peak integrals. Because the integration is performed in a restricted spectral window (5.52–5.34 ppm), the noise baseline does not affect the integral values significantly.

As compared to the OAc content measured on the monovalent MenA bulk conjugate by integrating the signals of protons at C₂ (H₂^3OAc, H₂, H₂^deOAc) and the protons at C₃/C₄ positions (H₃ + H₄^OAc) in comparison to protons at the C₁ position,⁶ the method applied by dissecting the H₁ provides a lower estimation (Table 2).

Table 2. % OAc Estimated on a Monovalent MenA-CRM₁₉₇ Sample by Integrating Different Signals: (a) % OAc = H₁^OAc/H₁ × 100, Where H₁ is the Sum of H₁ and H₁^deOAc Peak Integrals; (b) % OAc = (H₂ + H₂^4OAc/H₂) × 100, Where H₂ is the Sum of H₂, H₂^4OAc and H₂ Peak Integrals; (c) % OAc = (H₃^OAc + H₄/H₁) × 100, Where H₁ is the Sum of H₁^OAc and H₁ Peak Integrals (Figure 4).

	H1	H3/H4 vs H1	H2
% OAc	66	78	80

Because of the partial overlapping of the H₁ signals of OAc and non OAc forms, the integration strategy in the one-dimensional spectrum results in under estimation of approximately 10% of OAc content for the OAc level in the range of 70–80%. Essentially, some of the OAc species are grouped with the non OAc form in the integral measurement (Figure 3).

Figure 4.

¹H spectrum of the monovalent MenA-CRM₁₉₇ sample. Labels to facilitate the peak assignment are shown.

Conclusions

Novel applications for the identity test and O-acetyl content estimation of the MenA antigen in multivalent MenACWY combination vaccines have been developed.

NMR-based identity assay confirmed very high specificity and reproducibility and should be considered as a suitable alternative to commonly used non-specific and time-consuming colorimetric/chromatographic assays.

This NMR methodology can be readily adapted for use with other bacterial PS preparations. For instance, it has the potential to be applied on other licensed mono- and multivalent conjugate vaccines against Haemophilus influenzae type b (PedvaxHIB—Merck; ActHIB—Sanofi Aventis Pasteur; Hiberix—GSK Vaccines), meningococcal (Menjugate, Menitorix—GSK Vaccines; Menactra—Sanofi Pasteur; NeisVac-C, Nimenrix—Pfizer; Meningitec—Nuron Biotech; MenAfriVac—Serum Indian Institute), meningococcal/H. influenzae type b (MenHibrix—GSK Vaccines), and pneumococcal (Prevnar, Prevnar13—Pfizer; Synflorix—GSK Vaccines) infections.

A new method for the estimation of OAc content of the MenA antigen in Menveo-licensed vaccine, based on the integration of H₁ signals in one-dimensional spectra, has been developed. Notably, this NMR methodology can be also adopted for use with other vaccine preparations potentially including protein antigens.

Materials and Methods

The Menveo vaccine, as a two-vial presentation (monovalent lyophilized MenA-CRM₁₉₇ and trivalent liquid MenC-, MenW-, and MenY-CRM₁₉₇)⁹ as obtained from GSK Vaccines Manufacturing.

NMR Samples Preparations

To prepare NMR analytical samples, 10 vials of lyophilized MenA-CRM₁₉₇ were reconstituted with 10 vials of liquid MenCWY singularly. All the them were pooled in 8 mL glass vial (Wheaton) and finally dried under vacuum.

A monovalent MenA-CRM₁₉₇ sample was also prepared by drying under vacuum approximately 1 mg of conjugate as the saccharide content.

The lyophilized contents were solubilized in 0.6 mL of deuterium oxide (99.9 atom % deuterium; Aldrich), mixed to obtain a uniform concentration, and subsequently transferred to a 5 mm NMR tube (Bruker).

NMR Measurements

Proton NMR experiments were recorded on a Bruker AEON AVANCE III 600 MHz spectrometer equipped with a high-precision temperature controller using a 5 mm QCI CryoProbe (Bruker). TopSpin software (Bruker) was used for data acquisition and processing.

¹H NMR one dimensional spectra were collected at 600 MHz and 25 ± 0.1 °C, with 32k data points over a 12 ppm spectral width, accumulating a sufficient number of scans. The transmitter was set at the water frequency, which was used as reference signal (4.79 ppm).

To suppress the sucrose signals, which dominate the spectrum because of its large excess (i.e., >1500 molar fold) in comparison to the MenACWY antigens, a diffusion filter pulse sequence (ledbpgp2s1d) was applied by following an appropriate setting of the acquisition parameters (i.e., gradient pulses, delays). The spectra were weighted with 2.0 Hz line broadening and Fourier-transformed.

³¹P NMR one-dimensional spectra were recorded at 161.9 MHz at 25 ± 0.1 °C, with 32k data points over a 20 ppm spectral width, accumulating an appropriate number of scans. The spectra were weighted with 3.0 Hz line broadening and Fourier-transformed. Phosphoric acid (85%) in deuterium oxide was used as external standard (0 ppm).

All the ¹H and ³¹P NMR spectra were obtained in a quantitative manner by using a total recycle time to ensure a full recovery of each signal (5 × longitudinal relaxation time T₁).

The ¹H–³¹P two-dimensional HMBC NMR experiment was recorded on the monovalent MenA-CRM₁₉₇ sample by a standard pulse program. 4096 and 512 data points were collected in F₂ and F₁ dimension, respectively. A sufficient number of scans was accumulated prior to Fourier transformation to yield a digital resolution of 0.2 and 3.0 Hz per point in F₂ and F₁, respectively.

Author Contributions

The manuscript was written through contributions of all the authors. All the authors have given approval to the final version of the manuscript.

The authors declare the following competing financial interest(s): All the authors are employees of the GSK companies. Francesco Berti (corresponding author) is owner of patents on related topics. This study was sponsored by GSK Biologicals SA.

Notes

Menveo, Hiberix, Menitorix, MenHibrix, and Synflorix are trademarks of the GSK group of companies.

Notes

PedvaxHIB is a trademark of the Merck group of companies.

Notes

ActHIB, Menactra are trademarks of the Sanofi group of companies.

Notes

NeisVac-C, Nimenrix, Prevnar, and Prevnar13 are trademarks of the Pfizer group of companies.

Notes

Meningitec is a trademark of Nuron Biotech company.

Notes

MenAfriVac is a trademark of Serum Indian Institute company.