Structural correlates of carrier protein recognition in tetanus toxoid-conjugated bacterial polysaccharide vaccines.

An analysis of structure-antibody recognition relationships in nine licenced polysaccharide-tetanus toxoid (TT) conjugate vaccines was performed. The panel of conjugates used included vaccine components to protect against disease caused by Haemophilus influenzae type b, Neisseria meningitidis groups A, C, W and Y and Streptococcus pneumoniae serotype 18C. Conformation and structural analysis included size exclusion chromatography with multi-angle light scattering to determine size, and intrinsic fluorescence spectroscopy and fluorescence quenching to evaluate the protein folding and exposure of Trp residues. A capture ELISA measured the recognition of TT epitopes in the conjugates, using four rat monoclonal antibodies: 2 localised to the HC domain, and 2 of which were holotoxoid conformation-dependent. The conjugates had a wide range of average molecular masses ranging from 1.8×10(6) g/mol to larger than 20×10(6) g/mol. The panel of conjugates were found to be well folded, and did not have spectral features typical of aggregated TT. A partial correlation was found between molecular mass and epitope recognition. Recognition of the epitopes either on the HC domain or the whole toxoid was not necessarily hampered by the size of the molecule. Correlation was also found between the accessibility of Trp side chains and polysaccharide loading, suggesting also that a higher level of conjugated PS does not necessarily interfere with toxoid accessibility. There were different levels of carrier protein Trp side-chain and epitope accessibility that were localised to the HC domain; these were related to the saccharide type, despite the conjugates being independently manufactured. These findings extend our understanding of the molecular basis for carrier protein recognition in TT conjugate vaccines. Crown

Introduction

The global burden of bacterial meningitis is primarily due to invasive infection by Neisseria meningitidis and Streptococcus pneumonia. Haemophilus influenzae serogroup B (Hib) accounted for many cases of bacterial meningitis in the developed world prior to the introduction of the Hib conjugate vaccine in 1987. Hib vaccines have reduced incidence of disease attributed to Hib by 80% or more, dependent on vaccine uptake [1,2]. Monovalent meningococcal group C (MenC) vaccines, licenced in 1999–2000, have reduced the incidence of invasive meningococcal disease caused by MenC by over 90% in the UK [3,4]. There are currently three licensed tetravalent meningococcal conjugate vaccines, which also offer protection from serotypes A, W and Y [5], and pneumococcal conjugate vaccines can protect against up to 13 disease-causing serotypes [6,7]. The significant mortality rates and long-term sequelae following infection by encapsulated bacteria have made such vaccination strategies highly sought after worldwide.

Conjugate vaccines have purified oligo- or polysaccharide (PS) covalently linked to a carrier protein, e.g. tetanus toxoid (TT), in a process known as conjugation. A conjugate vaccine elicits a T-cell dependent antibody response, leading to high-avidity, circulating antibodies and the establishment of immune memory in infants and other at-risk groups, which are not evoked by plain PS vaccines [4]. The failure of plain PS vaccines to elicit IgG memory in mice has led to the belief that elicitation of T-cell help by glycoconjugates was attributable to MHC Class II presentation of peptides to the T-cell receptor. Carbohydrates fail to directly bind MHC Class II receptor molecules and are not presented to T-cells, and are, therefore, truly ‘T-cell independent’ [8]. The 2011 study carried out by Avci et al. [8] has demonstrated that MHC Class II-presented glycopeptides elicit T-cell help; glycoconjugated carbohydrates are processed into smaller glycans which are presented to the T-cell receptor on the APC surface. Carbohydrate epitope presentation to CD4+ cells plays a vital role in inducing polysaccharide-specific adaptive immune responses. A separate study suggested that the carbohydrate component of a pneumococcal glycoconjugate is presented to the APC surface and co-localises with the MHC class II protein [9].

Glycoconjugate vaccines vary immensely due to biological variations such as polysaccharide type and chemical variations such as conjugation chemistry. These factors as well as the choice of carrier protein can provide glycoconjugate vaccines varying in terms of both size and structure. The size of the conjugate can depend on the oligomeric state (and monomeric size) of the carrier protein, the chain-length of the PS, the saccharide-to-protein loading and the conjugation chemistry used [2,10]. Previous studies have suggested that the immunogenicity of conjugate vaccines is partly dependent on their PS chain length and structural properties [11-13], as well as the intrinsic properties of the carrier protein, but studies have not been done to survey the protein epitope accessibility. In this study, a comparison of the protein structural and antibody recognition features of a panel of polysaccharide-tetanus toxoid conjugate vaccines has been undertaken to determine if the accessibility of the exposed TT epitopes is affected by high PS loading in polysaccharide-TT conjugates.

Materials and methods

Vaccines

A panel of nine glycoconjugates manufactured with TT as carrier protein by a variety of manufacturers was obtained. The panel included Hib-TT-A and Hib-TT-B (coded as described by Ho et al. [14]; two MenC-TT conjugates and two MenA-TT conjugates (arbitrary codes were assigned); and, one of each of the following conjugates; MenW-TT, MenY-TT and Pneumo 18C-TT. The bulk purified carrier protein conjugated to MenC-TT (2), MenW-TT and MenY-TT was also included in the panel.

Prior to analysis, samples of bulk intermediate polysaccharide-protein conjugates supplied by vaccine manufacturers were dialysed at 4 °C with three changes of phosphate buffered saline (PBS ‘A’) (10.1 mM Na2HPO4, 1.84 mM KH2PO4, 171 mM NaCl, 3 mM KCl pH 7.3–7.5) for 24–26 h using dialysis membranes with a 10 kDa-molecular-mass cut-off pore size (Spectra/Por® 7 Dialysis Membrane, Spectrum Laboratories Inc., Rancho Dominguez, CA, USA).

Polysaccharide loading ratios of mol polysaccharide repeating unit: mol tetanus toxoid monomeric unit were converted from the g PS:g ratios determined by the manufacturers.

Monoclonal antibodies

Four rat IgG monoclonal antibodies, TT04, TT07, TT08 and TT10, against tetanus toxoid were produced from rat hybridoma cell lines, provided by Wellcome Research Laboratories (Beckenham, U.K.). The hybridomas were prepared [15] and the antibodies characterised [15,16] as described previously.

Determination of protein content

For TT conjugates, protein concentration was determined using a molar absorption coefficient, A278 nm, 0.1% of 1.229 cm−1 mg−1 ml for TT [17]. Samples were analysed on a Perkin–Elmer Lambda 800 UV-VIS spectrophotometer. A data increment of 0.50 nm was used, scan speed was 125 nm/min, slit width was 1.00 nm and integration time was 0.24 s. A cell volume of 1.0 ml with 10 mm path-length was used. Following determination of protein concentration, the samples were diluted in PBS ‘A’ to 500 μg protein/ml, where possible, for SEC-MALS and 6–25 μg protein/ml for fluorescence spectroscopy.

Determination of dn/dc

The specific refractive increment, dn/dc, was measured using an interferometric refractometer (Optilab rEX; Wyatt Technology Corp. Santa Barbara, CA, U.S.A.) which had been calibrated with five different concentrations of BSA in PBS ‘A’, ranging from 0.25 to 2.0 mg/ml. A series of five different sample concentrations ranging from 0.05 to 1.0 total mg/ml [(mg of saccharide + mg of protein)/ml, for conjugates], was prepared from dialysed sample and syringed through the refractometer, which had been equilibrated with filtered dialysis buffer, starting with the lowest concentration. For conjugate samples, the saccharide and protein concentrations were added together to give the sample concentration, c (mg/ml). dn/dc values (ml/g) were calculated using the software supplied by the manufacturer (DNDC™; Wyatt Technology Corp.). Percentage errors were determined by the Astra software as a combination of noise for each detector combined with the quality of the line projected [18].

Molecular sizing by SEC/MALS

The HPLC-SEC/MALS system consisted of a gradient pump (Thermo Fisher (Dionex) UK Ltd., Hemel Hempstead, UK) and autosampler fitted with a 200 μl injection loop. A Tosoh Bioscience TSK PWXL guard column and TSK gel G6000PWXL + G5000PWXL analytical columns were connected in series. Chromatographic signals were collected by an ICS series multi-wavelength UV detector (Thermo Fisher UK Ltd.) an interferometric refractometer (Wyatt Technology Corp., Santa Barbara, USA) and a DAWN-EOS 18-angle angle light scattering detector (Wyatt Technology Corp.). PBS ‘A’ ultra-filtered through 0.10 μm Millipore filter was used as eluent at a flow rate of 0.25 ml/min. Fifty μg protein (500 μg/ml where possible) was injected per sample and total mg protein + mg saccharide (by calculation) were used for subsequent calculations.

Data from 11 detectors between the angles of 57.0° and 141.0° were used and weight-average molar mass and % recoveries were determined using Astra for Windows 5.3.4™ software from Wyatt Technology Corp. The dn/dc values determined for each conjugate/carrier protein (Table 1) were used in calculations, with the exception of two different MenA-TT conjugates which gave unusually low dn/dc values. For these a dn/dc value of 0.191 ml/g was used [19]. The weight-average molecular mass (Mw) and polydispersity (Mw/Mn) were determined by the Zimm method and were obtained for high molar mass (peak 1), lower molar mass (peak 2) and total combined peak data (peak 3), also used for the purpose of determining the % recovery. Errors are calculated by the Astra software in the same way as described for the refractive index calculation.

Table 1

SEC/MALLS-determined molar mass values of TT conjugates.

Sample	PS/protein,a mol:mol	dn/dc,b ml/g	Peak 1	Peak 2	Total Peak 3
			Mean Mw,d g/mol ×106	Mean Mw, g/mol ×106	Mean Mw, g/mol ×106	Polydispersity, Mw/Mn
TT	No PS present	0.155 (3%)	0.216 (2%)	0.130 (1%)	0.155 (2%)	1.0
Hib-TT-A	193	0.174 (1%)	18.3 (1%)	5.54 (0%)	7.16 (1%)	1.4
Hib-TT-B	178	0.189 (0%)	15.3 (1%)	4.39 (1%)	2.60 (1%)	1.5
MenA-TT (1)	575	0.191c	4.97 (5%)	0.929 (4%)	2.76 (5%)	2.3
MenA-TT (2)	209	0.191c	8.81 (6%)	3.03 (6%)	5.05 (6%)	1.4
MenC-TT (1)	414	0.166 (1%)	2.90 (0%)	0.829 (1%)	1.84 (1%)	1.6
MenC-TT (2)	517	0.212 (3%)	9.22 (1%)	3.87 (1%)	5.04 (1%)	1.2
MenW-TT	284	0.196 (1%)	21.6 (5%)	7.93 (1%)	11.5 (2%)	1.3
MenY-TT	302	0.209 (1%)	19.0 (4%)	10.5 (1%)	13.5 (2%)	1.1
Pneumo 18C-TT	84	0.153 (2%)	17.9 (1%)	6.54 (1%)	7.81 (1%)	1.6

PS: protein (mol:mol) values were converted from g:g values as quoted by the manufacturer.

Values in parentheses are the percent error for refractive index measurements and are ±values.

A dn/dc value of 0.191 ml/g [19] was used for the MenA-TT conjugates.

Values in parentheses are the percent error for molar mass measurements and are ±values.

Fluorescence spectroscopy

Intrinsic fluorescence spectra were obtained on a Horiba Jobin Yvon FluoroMax-4 spectrofluorometer in 1 ml, 1 cm path-length quartz cells. Excitation wavelengths (λex) of both 280 and 295 nm were used, with a band pass of 4.25 nm for excitation and emission monochromators and data was collected at increments of 0.5 nm. Emission spectra were collected using a wavelength range from 260 to 550 nm (280 nm excitation) or from 275 to 550 nm (295 nm excitation). Fluorescence spectra were corrected by subtracting the corresponding base-line spectra of buffer alone (PBS ‘A’). Samples of 6–25 μg protein/ml in PBS ‘A’ were used. Fluorescence emission wavelength maxima (Fmax) values were determined using the FluorEssence software and are accurate to ±0.5 nm.

Fluorescence quenching

Fluorescence quenching was carried out from measurements obtained using an excitation wavelength of 295 nm. The same parameters were used as for the intrinsic fluorescence measurements. Aliquots of a 5 M stock acrylamide (Sigma–Aldrich, UK, A9099) was added at concentrations ranging from 0.05 M to 0.83 M to initial 1 ml samples of TT or conjugate (12.5 μg protein/ml) in PBS ‘A’ and emission data was collected between 305 and 420 nm. A Peltier device was used to keep the solution at 25 °C. Spectra were corrected by subtracting the corresponding base-line spectra of PBS for each acrylamide concentration, and for volume adjustments. The Fo/F ratio was calculated where Fo is the height at the maximal fluorescence signal (cps) at Fmax for the sample with no acrylamide addition, and F is the height at Fmax for the sample at each acrylamide concentration. Stern–Volmer plots were used to determine the linearity of each quenching curve, with Fo/F ratio plotted against quencher concentration. The slope, equal to the Stern–Volmer quenching constant (KSV), Y-intercept and R2 were also determined.

ELISA

A capture ELISA was used to detect of the recognition of TT epitopes in the conjugates by rat monoclonal anti-tetanus toxoid antibodies, TT04, TT07, TT08, TT10 [15,16] as per the method used by Coombes et al. [20]. ELISA plates were coated overnight at 4 °C with 100 μl/well of an anti-tetanus monoclonal antibody (5 μg/ml), diluted in carbonate buffer (0.05 M, pH 9.6). The plates were washed (3×) by immersion in phosphate buffered saline (pH 7.4) containing 0.05% (vol/vol) Tween 20 (PBST), then blocked with 150 μl/well of PBST containing 2.5% (wt/vol) dried skimmed milk powder (PBSTM) for 1 h at 37 °C. Following a second wash in PBST, serial two-fold dilutions of the tetanus toxoid or bulk conjugate samples, at the same protein concentration, in PBSTM were prepared in the plate (final volume 100 μl ranging from 1.5 μg TT to 0.01 μg TT) and the plates were incubated at 37 °C for 2 h. Plates were washed as described previously and 100 μl/well of guinea pig anti-tetanus IgG, diluted in PBSTM, was added for 2 h at 37 °C. After further washing, antigen-specific IgG antibodies were detected using 100 μl/well of HRP-conjugated goat anti-guinea pig IgG diluted 1/2000 in PBSTM. After a further incubation of 1 h at 37 °C and a final wash, 100 μl/well of substrate solution containing 0.5 mg/ml ABTS and 0.008% (vol/vol) hydrogen peroxide in 0.05 M citric acid buffer (pH 4) were added. The reaction was allowed to develop at room temperature for up to 30 min and the optical density was measured at 405 nm using a Multiskan ELISA plate reader (Thermo Life Sciences, UK). Epitope recognition was calculated relative to the reference toxoid by parallel line analysis (log optical density vs. log dose), using a minimum of 3 sequential points from the linear section of the dose response curve for the reference and test sample, and expressed in Lf/ml using Combistats software, which was converted to μg TT/ml. Analysis of variance was used to test for any significant deviation from linearity or parallelism of the dose response relationship (p < 0.01).

Results

Molecular sizing and dn/dc determination of tetanus toxoid conjugates

A wide range of sizes were found for the conjugates, with weight-average molecular mass (Mw) values for the total combined peak area ranging from 1.8 × 106 g/mol to greater than 20.0 × 106 g/mol (Table 1 and Fig. 1). All of the TT conjugates were polydisperse as previously observed [19,21-23] with Mw/Mn up to 2.3.

Fig. 1

Size exclusion chromatograms showing RI signal (), 90° LS signal () and UV 280 nm signal (): MenC-TT (1) (A), MenC-TT (2) (B), MenW-TT (C), MenY-TT (D), MenA-TT (1) (E), MenA-TT (2) (F), Pneumo 18C-TT (G), TT (H), Hib-TT(A) (I) and Hib-TT(B) (J). The vertical bars indicate the integration limits for peak 1, peak 2 and peak 3.

The TT sample was found to have a Mw of 155,000 g/mol, which is in agreement with previously determined SEC/MALS [19,23,24] and sedimentation equilibrium [24] values. A peak due to dimerization was seen, as expected [24,25]. Following conjugation there were increases in the molecular mass of the samples, as expected. All conjugates eluted with broad asymmetric peak(s), as expected for the polydisperse samples (Fig. 1). Significantly different sizes (Mw) were determined for the high and low molar mass peak areas, Peaks 1 and 2, respectively (Table 1), suggesting the presence of conjugate oligomers. The elution time of the conjugates was related to their SEC/MALS-determined sizes (Fig. S1), except for Hib-TT-A, which eluted relatively later. MenW and MenY conjugates were the largest conjugates, and MenC-TT (1) and MenA-TT (1) were the smallest.

The calculated recoveries based on the dn/dc values and RI signals were generally between 50 and 80%, with the larger conjugates generally having the lower recoveries than the smaller conjugates. Aggregates of up to 109–1010 g/mol, the limit of the MALS, are easily detected by the light scattering signal, even at low concentration when UV and RI signals are diminished. The chromatography and column conditions do allow for the detection of aggregates at Vo, but these were not significant (>5%) in the samples evaluated. There was no overall correlation between the mass (Mw) and PS loading (Fig. 2A).

Fig. 2

Relationship between PS/protein loading and mean molar mass (A) and Stern–Volmer quenching constant and PS/protein loading (B) for TT conjugates.

Tetanus toxoid conformation in conjugates

Tetanus toxoid conformation is considered to be a good characteristic of vaccine quality [26], although there is no evidence suggesting that maintenance of native-like conformation is a requirement for its immunogenicity and protection. Intrinsic fluorescence spectroscopy has been used to differentiate between TT conformers through the spectral features of its aromatic amino acid side-chains [14,27-29]. Of the total 1315 amino acids that the parent toxin contains, there are 56 phenylalanine residues, 79 tyrosine (Tyr) residues and 13 tryptophan (Trp) residues [30]. An excitation wavelength of 280 nm gives an emission spectrum which is sensitive to the solvent exposure of side-chains of Trp and to a lesser extent Tyr, while λex of 295 nm is used to obtain spectra almost exclusively due to Trp. Changes in the conformation of a protein arising from Trp side-chain movements can be tracked by red-shifts, i.e. to longer λmax, upon protein unfolding and exposure of the side-chains; and by blue-shifts, i.e. to shorter λmax upon aggregation and internalisation of the residues, which are usually accompanied by a decrease in fluorescence intensity.

The TT carrier protein used in this study, had Fmax values of 330.5 nm (λex = 280 nm) and 334.0 (λex = 295 nm), which were within 3.5 nm of those previously published [14,27,28] toxoids produced by different manufacturers (duration and concentration of formaldehyde) or manufacturing sites (Table 2). Not more than 1 nm difference was seen between different batches of the same toxoid.

Table 2

Intrinsic fluorescence maximum values and quenching of TT conjugates.

Sample	Fmax(λex = 280 nm)	Fmax(λex = 295 nm)	KSV	Intercept	R2	Curve type
TTa	330.5	334.0	2.25	1.08	0.993	Linear
Hib-TT-A	331.0	335.5	2.75	1.09	0.990	Linear
Hib-TT-B	331.0	332.5	2.15	1.09	0.992	Linear
MenA-TT (1)	329.0	335.5	2.78	1.16	0.978	Upward
MenA-TT (2)	329.5	335.5	2.48	1.01	1.00	Linear
MenC-TT (1)	331.5	335.5	4.24	1.32	0.877	Downward
MenC-TT (2)	330.5	333.5	3.01	0.96	0.993	Linear
MenW-TT	330.0	333.5	2.69	1.07	0.988	Linear
MenY-TT	332.0	335.5	2.87	1.10	0.985	Linear
Pneumo 18C-TT	329.0	334.0	2.26	1.10	0.977	Downward
			Ref.b
TT (Hib-TT-A)	329	333	14
TT (Hib-TT-B)	327	331	14
TT (MenA-TT 1)	328	333
TT (MenC-TT 1)	328	333	14

TT used in the production of MenC (2), MenW, MenY and Pneumo 18C conjugates.

The Fmax values obtained in separate experiments.

The conjugate panel had Fmax values ranging from 329 to 332 nm when excited at 280 nm, and 332.5–334 nm, when excited at 295 nm (Table 2 and Fig. S2). There was a 1.5 nm red-shift in Fmax of TT residues following conjugation to MenC (2) and MenY, which may indicate that some protein unfolding had occurred from conjugation. There was no significant change in TT fluorescence emission following conjugation of MenW, however.

Comparison of the intrinsic fluorescence data of the toxoids with their ‘matching’ conjugate, shows that overall, there is little difference between the Fmax of TT prior to and following conjugation, relative to changes observed in chaotropic agent-induced unfolding experiments [28,29], indicating that conjugation of the TT monomer to the saccharide chains results in very little change in its conformation, as observed previously by optical spectroscopic studies for both Hib-TT and MenC-TT (1) conjugates [14]. Following conjugation there is a red-shift for all conjugates, with the exception of a 1 nm blue-shift (280 nm excitation only) for MenA-TT (2), and a 0.5 nm blue-shift (295 nm excitation only) for MenC-TT (2), W and Y. Fluorescence emission spectra recorded at 280 and 295 nm are dependent on different groups of amino acids, so it would not be expected that the changes would be identical.

This structural stability has been attributed to the formaldehyde-induced cross-linking that occurs during the detoxification stage of its manufacture. There was no spectral evidence of aggregated TT, as characterised by blue-shifted Fmax and lower intensity spectra (Fig. S2).

Fluorescence quenching was used to determine the degree of accessibility of Trp residues to the uncharged quencher, acrylamide. Using the Stern–Volmer relationship expressed as KSV, which relates fluorescence yield of the sample to the concentration of the added quencher, a linear relationship with an intercept of 1 was typically found. All the conjugates displayed a linear relationship (R2 ≥ 0.98) except MenA-TT (1), which had upward curvature, and MenC-TT (1) and Pneumo 18C-TT, which had downward curvatures in Stern–Volmer analysis (Figs. S3 and S4). The conjugates displaying linear curve types had similar gradients KSV of 2.15 to 3.01 (Table 2) indicating that the chromophores are equally accessible to the quencher in those conjugated molecules. Both MenC-TT conjugates have comparatively high KSV values indicating that Trp residues are more exposed to the quencher in these conjugates. The deviation from linearity seen with three of the conjugates could indicate that the quencher causes a perturbation in structure with increasing concentration, leading to a change in the Trp accessibility, and a positive or negative co-operativity arising from internal energy transfer (Trp-to-Trp fluorescence quenching and emission). Some acrylamide solutions contain acrylic acid as a contaminant; however it is unlikely that this is the cause as the acrylamide reagent used had a purity of 99.9%. A weak correlation was found between PS loading and Stern–Volmer quenching constant (r = 0.54), which increased (r = 0.71), when the 3 conjugates demonstrating non-linear behaviour were excluded (Fig. 2B).

Epitope recognition by monoclonal Abs to tetanus toxoid

The accessibility of epitopes on the conjugated TT molecules were evaluated with four rat monoclonal IgG antibodies made against toxoid, which are capable of recognising toxin and toxoid. The epitopes defined by mAbs, TT04 and TT10, are linear epitopes located on the C fragment of the heavy chain of tetanus toxin, also referred to the cell-binding, or HC domain, with TT10 being able to neutralise toxicity in mice [15,16]. Epitopes defined by TT07 and TT08 are dependent on the tertiary structure of the intact toxin and are conformational epitopes [15].

The TT carrier protein alone was recognised by all mAbs with the highest epitope recognition by the ‘conformational’ antibodies, TT07 and TT08 (Table 3). The epitope recognition of the conjugated TTs was 10–50-fold lower. Interestingly, the Hib conjugates generated similar results to each other, as did the MenC conjugates, and the structurally similar MenW and Y conjugates, with similar binding by TT04 and TT10 to the conjugate pairs. The size-variant MenA conjugates were not comparable; the smaller MenA-TT (1) showed a higher degree of epitope recognition than MenA-TT (2) particularly for the linear epitopes. The phosphodiester-containing Hib and MenA (2) – conjugated TT samples gave the lowest binding, and could have been affected by the phosphate in the assay buffer.

Table 3

TT epitope recognition determined by monoclonal antibodies.

mAb sample	TT content, μg/ml (95% C.I.)
	TT04a	TT07	TT08	TT010
TT	299.51 (248.06, 362.17)	972.15 (850.63, 1114.10)	445.58 (428.40, 463.34)	255.46 (231.54, 282.15)
Hib-TT-A	1.15 (0.95, 1.38)	1.09 (0.95, 1.26)	NP	2.73 (2.44, 3.04)
Hib-TT-B	1.55 (1.28, 1.87)	11.12 (10.00, 12.36)	NP	4.23 (3.79, 4.72)
MenA-TT (1)	32.01 (26.52, 38.74)	10.92 (9.70, 12.32)	99.76 (95.93, 103.73)	14.31 (12.84, 15.95)
MenA-TT (2)	1.89 (1.51, 2.33)	9.70 (8.73, 10.79)	NP	4.26 (3.86, 4.70)
MenC-TT (1)	8.07 (6.68, 9.80)	22.26 (20.02, 24.74)	14.01 (13.56, 14.57)	6.81 (6.12, 7.60)
MenC-TT (2)	7.92 (6.56, 9.59)	46.61 (40.97, 53.07)	34.55 (33.19, 35.95)	8.85 (7.92, 9.87)
MenW-TT	16.21 (14.94, 17.63)	33.76 (24.82, 45.59)	156.40 (130.52, 191.37)	15.99 (14.80, 17.27)
MenY-TT	18.67 (17.14, 20.35)	108.02 (80.10, 128.76)	97.73 (82.13, 116.86)	12.84 (11.92, 13.83)
Pneumo 18C-TT	22.93 (21.06, 24.97)	83.04 (62.89, 112.54)	73.64 (61.83, 87.65)	20.35 (18.93, 21.85)

NP: Not parallel and statistically invalid.

Values in parentheses are statistical error using a 95% confidence interval.

Only a partial correlation, at best, was found between conjugate size (Mw) and epitope binding by the linear HC mAbs, TT04 (r = 0.23) and T10 (r = 0.45), although the binding of these mAbs did show an increased correlation with PS loading (r = 0.52 and 0.68 for TT04 and TT10, respectively), when Pneumo 18C-TT was excluded. A positive correlation between size and the ability of conformational mAbs to bind conjugated TTs was seen for TT07 (r = 0.67) and TT08 (r = 0.68) (Fig. 3A and B).

Fig. 3

Relationship between epitope binding to TT conjugates and conjugate size. Conjugate size is defined by panel (A) Molar Mass (Mw) and panel (B) PS/protein loading. The monoclonal antibodies used were: TT04 (), TT10 (), TT07 () and TT08 (▴). Conjugates in panel (A) are 1, MenC-TT(1); 2, Hib-TT(B); 3, MenA-TT(1); 4, MenC-TT (2); 5, MenA-TT (2); 6, Hib-TT(A); 7, Pneumo 18C-TT; 8, MenW-TT; and 9, MenY-TT.

Discussion

Polysaccharide-protein conjugate vaccines that protect against carriage and infection of encapsulated bacteria are regarded as well-defined, purified subunit vaccines. Yet, little has yet been reported on the accessibility of key protein epitopes that might serve as the glycopeptide epitopes presented to antigen-presenting cells in its role as a carrier protein, or even as protective toxin neutralising epitopes. In this study, the structural features of nine different polysaccharide – tetanus toxoid conjugates comprising 6 different PS structural types, all components of commercially licenced vaccines, were compared using physico-chemical methods. An evaluation of their size, carrier protein conformation, and epitope binding was performed in order to correlate structural features with epitope recognition in the context of conjugated tetanus toxoid.

All TT conjugates were found to be polydisperse with respect to size, and contained subpopulations ranging from 1 to > 20 × 106 g/mol. The conjugated toxoids were found to be well-folded, and did not have spectral features typical of aggregated TT.

The structural conformation of the TTs was evaluated through intrinsic protein fluorescence which uses primarily Trp side-chains as markers. In order to localise the Trp side-chains with respect to the three-dimensional structure, the L-chain (catalytic domain) and the HC structures were used, as a crystal structure for full-length tetanus toxin is not available. The L-chain contains only a single Trp is at position 43 of the sequence which forms part of β-strand with regards to its secondary structure (Fig. 4A) [31], and it is unlikely to contribute greatly to the fluorescence signal observed. The H-chain contains the remaining 12 Trp residues from position 615 to 1312, with 9 of these being in the HC domain. The majority of these Trp residues are well buried in the hydrophobic core of the HC domain (Fig. 4B) [32], indicating that the majority fluorescence signal comes from this part of the tetanus toxin/toxoid. The HC domain is a key part of the molecule with respect to immunogenicity in that it contains the toxin neutralising epitopes of the molecule [15,33] and potent T-cell universal epitopes [34]. The Hc domain also contains the neuronal cell ganglioside binding sites for receptor binding activity [32] and forms stable dimers via an intermolecular bond [35].

Fig. 4

Ribbon backbone structures of tetanus toxoid domains. (A) Catalytic domain of tetanus toxin (PDB accession 1YVG). The single Trp in the structure is highlighted [31]. (B) HC domain of tetanus toxin (PDB accession 1DLL). All Trp residues are well buried, with the exception of one (Trp1289) [32].

Correlation was found between the PS loading (moles saccharide: moles TT) with acrylamide quenching of TT, suggesting that high ratios of negatively charged conjugated PS (ranging from 100 to 600 moles PS/mol TT) did not interfere with tetanus toxoid tryptophan amino acid side-chain interaction. Likewise, the recognition of epitopes on the Hc domain or the holotoxoid was not necessarily hampered by the size of the molecule or the extent of PS loading; in fact, positive correlations were found between epitope recognition and molecular mass and/or PS loading. Considering the diversity of conjugation chemistries used to manufacture these molecules, it was intriguing that ‘serogroup’-specific patterns were identified. All the TT conjugates studied here were manufactured with TT that had been purified of aggregates prior to conjugation.

It should be recognised that the conjugates did, in fact, have significantly lower binding to the mAbs than did the carrier protein alone, and this as well as the specific conjugation sites may play a role in their immune response. The conformational mAbs gave strongest evidence that larger conjugates which are likely to contain the networks of oligomeric carrier proteins described by Egan [36], actually have more accessible carrier protein. It has been hypothesised that the asymmetric, elongated structure of TT found by Abdehameed et al. [24] may provide the necessary surface area for conjugation. Such an elongated surface structure was also found by small angle X-ray scattering of the HC domain [35]. The mAbs recognising the linear epitopes of the HC domain, one of which defines a protective neutralising epitope, also showed a preference for binding TT molecules with higher PS loading.

Relatively few models for proteolytic processing of conjugate vaccine molecules following uptake by antigen presenting cells have been substantiated. Our understanding of what defines effective glycoprotein T-cell immunogens to provide the initial, critical stimulation of T- and B-cell responses are based on theoretical knowledge of the lower threshold limits for carbohydrate repeating units [37] and polypeptide compositional binding to MHC [38], as well as the identity of naturally processed peptides of TT [34] and potential conjugation sites therein. Peptide mapping for identification of glycopeptides in a formaldehyde-toxoided molecule presents considerable challenge, so simpler systems are required.

This study has shown the wide range of sizes and PS/protein loadings for TT conjugates used in licensed vaccines which have been shown to be safe and efficacious. The correlations identified here could provide in vitro tools for further development and studies on glycoconjugates.