University of Huddersfield Repository Solution conformation and flexibility of capsular polysaccharides from Neisseria meningitidis and glycoconjugates with the tetanus toxoid protein

The structural integrity of meningococcal native, micro-fluidized and activated capsular polysaccharides and their glycoconjugates – in the form most relevant to their potential use as vaccines (dilute solution) have been investigated with respect to their homogeneity, conformation and flexibility. Sedimentation velocity analysis showed that the polysaccharide size distributions were generally bimodal with some evidence for higher molar mass forms at higher concentration. Weight average molar masses M w where lower for activated polysaccharides. Conjugation with tetanus toxoid protein however greatly increased the molar mass and polydispersity of the final conjugates. Glycoconjugates had an approximately unimodal log-normal but broad and large molar mass profiles, confirmed by sedimentation equilibrium “SEDFIT MSTAR” analysis. Conformation analysis using HYDFIT (which globally combines sedimentation and viscosity data), “Conformation Zoning” and Wales-van Holde approaches showed a high degree of flexibility – at least as great as the unconjugated polysaccharides, and very different from the tetanus toxoid (TT) protein used for the conjugation. As with the recently published finding for Hib-TT complexes, it is the carbohydrate component that dictates the solution behaviour of these glycoconjugates, although the lower intrinsic viscosities suggest some degree of compaction of the carbohydrate chains around the protein. This present study investigates a recently FDA approved conjugate vaccine against serogroups A, C, W-135 and Y namely, Nimenrix ® produced by Pfizer Ltd., UK. The focus of this study has been sedimentation velocity in the analytical ultracentrifuge, together with sedimentation equilibrium and viscosity to examine the molecular integrity of the samples in terms of heterogeneity, molar mass distribution and conformational flexibility of purified native, micro-fluidized and activated (chemically modified to facilitate conjugation) capsular polysaccharides from N. meningitidis serogroups as well as meningococcal-tetanus toxoid (TT) conjugates. It follows our recent studies on (i) S. pneumoniae polysaccharides 24 ; (ii) the tetanus toxoid protein 25 and on the Hib-TT system 26 the latter study clearly demonstrating it was the carbohydrate polymer component rather than which dictated solution rather than the protein component for that glycoconjugate vaccine We now seek to establish if is also for the Men-TT system. (MenC), W-135 (MenW135), Y (MenY) and serogroup A sodium salt (MenA-Na + ) and calcium salt (MenA-Ca 2 + ). Micro-fluidized polysaccharides from serogroups A, C, W-135 and Y, activated polysaccharides from serogroups A and C, were also examined. Serogroup A conjugate without spacer (MenA without spacer), sodium salt conjugate with spacer (MenA-Na + with spacer), calcium salt conjugate with spacer (MenA-Ca 2 + with spacer) and serogroup C conjugate with spacer (MenC with spacer), conjugate without spacer (MenC without spacer) and serogroup W-135 conjugate without spacer (MenW135 without spacer) and serogroup Y conjugate without spacer (MenY without spacer), were also investigated.

serogroups, namely: A, B, C, W-135, X and Y 7,8 . Serogroup A meningococci are the major cause of the large, cyclic epidemics in Africa and Asia, while in industrialized nations 30-70% of the disease is caused by serogroup B organisms. Serogroup C meningococci are particularly associated with, usually, smaller-scale outbreaks worldwide [7][8][9] . Serogroup X is more restricted to parts of sub-Saharan Africa 9 . Serogroup Y meningococci are currently accounting for over 30% of cases in the USA. Serogroup W-135 meningococci was associated with large outbreaks among pilgrims to the Hajj in Saudi Arabia in 2000 and was responsible for the epidemic in Burkina Faso in 2002 6 . The polysaccharide capsules of N. meningitidis are important determinants of virulence. Mutants without capsular expression are serum sensitive (i.e. killed by complement, and non-pathogenic). These polysaccharides are large, unbranched structures made up of O-acetylated residues in the 3 position of D-mannosamine-6 phosphate linked (1→ 6). These are O-acetylated repeating units of N-acetylmannosamine, linked with α -(1→ 6) phosphodiester bonds in case of serogroup A: → α → → ---p ( 6) D Man NAc (3/4OAc) (1 OPO ) 3 (Ac = acetyl) or in case of serogroup C these are linear polymers made up of partly O-acetylated repeating units of sialic acid, linked with α -(2→ 9) glycosidic bonds: → α → ----( 9) D Neup5Ac(7/8OAc) (2 ) W-135 polysaccharide consists of partly O-acetylated alternating units of sialic acid and D-galactose, linked with α -(2→ 6) and α -(1→ 4) glycosidic bonds: D Gal (1 4) D Neu 5Ac(9OAc) (2 ) Neisseria meningitidis group Y polysaccharide consists of partly O-acetylated alternating units of sialic acid and D-glucose, linked with α -(2→ 6) and α -(1→ 4) glycosidic bonds: (1 4) D Neup5Ac(9OAc) (2 ) see e.g. [10][11][12][13] . The first successful capsular polysaccharide vaccines against groups A and C were developed in response to epidemics of meningitis among US military recruits 14,15 . However, it was found that the meningococcal polysaccharides proved to be poor immunogens in infants and fail to induce immunological memory in people of any age. Additionally, immune hypo-responsiveness is recorded after repeated vaccination with group C polysaccharide vaccine, which may cause difficulties for individuals who need long-term protection 16 . These disadvantages motivated extensive research to produce a polysaccharide-protein conjugate vaccine. Successes were subsequently achieved in the development of conjugate vaccines against Haemophilus influenzae type b (Hib) and Streptococcus pneumoniae thereby showing that the immunogenicity of polysaccharides could be improved by chemical conjugation to a protein carrier and eliciting a T-cell dependent anti-saccharide antibody response. The resulting polysaccharide -protein conjugate vaccines -are safe, immunogenic in young infants and induce long-term protection. In November 1999, meningococcal group C conjugate vaccine was introduced into routine immunisation in the UK 17 . Bivalent A plus C polysaccharide conjugate vaccines, have been assessed in clinical trials and were well tolerated and immunogenic in infants, toddlers, and adults [18][19][20] . Vaccine manufacturers have now developed conjugate vaccine combinations incorporating groups A, C, Y, and W-135 see e.g. 11,[21][22][23] . This present study investigates a recently FDA approved conjugate vaccine against serogroups A, C, W-135 and Y namely, Nimenrix ® produced by Pfizer Ltd., UK. The focus of this study has been sedimentation velocity in the analytical ultracentrifuge, together with sedimentation equilibrium and viscosity to examine the molecular integrity of the samples in terms of heterogeneity, molar mass distribution and conformational flexibility of purified native, micro-fluidized and activated (chemically modified to facilitate conjugation) capsular polysaccharides from N. meningitidis serogroups as well as meningococcal-tetanus toxoid (TT) conjugates. It follows our recent studies on (i) S. pneumoniae polysaccharides 24 ; (ii) the tetanus toxoid protein 25 and (iii) on the Hib-TT system 26 the latter study clearly demonstrating it was the carbohydrate polymer component rather than the protein which dictated the solution properties rather than the protein component for that glycoconjugate vaccine system. We now seek to establish if this is true also for the Men-TT system.

Results
Sedimentation coefficient distributions and molar mass. Analysis of the sedimentation coefficient distributions g*(s) vs s profiles in phosphate-chloride buffer (pH = 6.8, I = 0.10) showed that, in common with Hib polysaccharides 26 the native Men polysaccharides are bimodal systems with the high molar mass components ranging from 5% by mass in native MenY to 20% in native MenA (for both Na + and Ca 2+ salts). The high molar mass component was relatively less significant at low concentrations. The micro-fluidized (M-F) Men polysaccharides were characterized by lower sedimentation coefficients with the two components still present in M-F MenA. All other micro-fluidized Men polysaccharides appear to be mostly unimodal at different concentrations. The activated MenA and MenC polysaccharides were very nearly unimodal (Fig. 1). The bimodal structure of the polysaccharide distributions -particularly the native Men polysaccharides -are considered to be due to a higher degree of polymerization particularly for MenA Na + and Ca 2+ salts -see for example 11 . The g*(s) plots for all samples also showed the classical increase in the value of sedimentation coefficient s with the decrease in concentration, c, due to lowering non-ideality effects which vanish as c→ 0. After normalisation of s values measured in the buffer to standard solvent conditions (the viscosity and density of water at 20.0 °C) to give s 20,w , standard reciprocal plots 1/s 20,w vs c were used to obtain the non-ideality free s°2 0,w and concentration dependence Gralen values k s and are shown for comparison in Table 1.
Scientific RepoRts | 6:35588 | DOI: 10.1038/srep35588 Sedimentation equilibrium was then used to determine average molar masses using the SEDFIT-MSTAR and MFIT algorithms which give weight-average molar masses and z-average molar masses, respectively for the prime parameters (Table 1, see also Fig. 2). With the glycoconjugates generally a very low rotor speed was required to register a measurable distribution at equilibrium. In the case of the Men A conjugates without spacer it was not possible to get a measurable distribution, so molar masses were obtained from comparison of the sedimentation coefficient values with Men A conjugates with spacer and MHKS power law coefficients b of 0.4 and 0.5, in the relation s ~ M b , a range consistent with a macromolecule with a high degree of flexibility (see the consideration of conformation and flexibility below). Rotor speed = 45000 rpm. The plot has been normalized so that the main peak height is set to 1 for each concentration. The steady movement of the distribution to higher s (decreasing non-ideality) as the concentration is decreased can be clearly seen.    Table (and Tables 2 & 3) are due to the respective fits, taking into account other parameters such as error in the partial specific volume.
Sedimentation coefficient distributions could then be transformed into distributions of molar masses for the Men-TT conjugates using the Extended Fujita Approach of Harding, et al. 27 (Fig. 3) The transformation is as follows: b as referred to above is a conformation parameter that has already been estimated for number of polysaccharides in particular solvent conditions 27 and κ s can be found from equation (2) provided that at least one value of M (e.g. the weight average over the whole distribution, M w from sedimentation equilibrium) is known for one value of s (e.g. the weight average s value).

Intrinsic viscosity.
Intrinsic viscosity values resulting from three different extrapolation methods (due to Huggins, Kraemer and Solomon & Ciuta) are reported in Table 2. Consistent with the behaviour of the sedimentation coefficients, the intrinsic viscosity results for the polysaccharides are decreasing with the size reduction and activation of native polysaccharide. Intriguingly they are generally lower for the glycoconjugates even though the sedimentation coefficients and molar masses are much higher, although still an order of magnitude higher than    Translational frictional ratio. The translational frictional ratio, f/f o is a parameter which depends on conformation and molecular expansion through hydration effects 31 . It can be measured experimentally from the sedimentation coefficient and molar mass: where N A is Avogadro's number, f is the friction coefficient of the molecule and f o the corresponding value for a spherical particle of the same mass and (anhydrous) volume 31 . Values of f/f o increase with increasing chain stiffness, although they are also molar mass dependent. The values for the frictional ratio in Table 2 are also consistent with a flexible coil structure.
Persistence length L p . For a more quantitative estimate of chain flexibility we can use the persistence length L p , which has theoretical limits of 0 for a random coil and ∞ for a stiff rod. Practically the limits are ∼ 1-2 nm for a random coil and ∼ 200-300 nm for a very stiff rod shaped macromolecule. Several methods are available for estimation of L p using either intrinsic viscosity 32,33 or sedimentation coefficient 34,35 measurements. For example the relation 32,33 .
where φ is the Flory-Fox coefficient (2.86 × 10 23 mol −1 ) and A 0 and B 0 are tabulated coefficients, and the Yamakawa-Fujii equation 34 . Yamakawa and Fujii showed that A 2 can be considered as -ln(d/2L p ) and A 3 = 0.1382 if the L p is much higher than the chain diameter, d. Difficulties arise if the mass per unit length is not known, although both relations have now been built into an algorithm Multi-HYDFIT 35 which estimates the best estimates or best range of values of L p and M L based on minimization of a target function Δ. An estimate for the chain diameter d is also required but extensive simulations have shown that the results returned for L p are relatively insensitive to the value chosen for d which was fixed at an average of ∼ 0.8 nm see e.g. 36,37 M L and L p were treated as variables and the minimum value of the target function Δ was estimated on a 2D contour plot for each sample: an example is given in Fig. 4, and the values estimated for Men samples given in Table 3. All the values are consistent with flexible random coil structures, including the polysaccharides both prior and after conjugation.
These results are in support of suggesting highly flexible random coil structures. Molecular flexibility is increasing by size reduction and activation and more because of conjugation. The estimation for the mass per unit lengths within experimental errors are in good agreement with the predicted values from the repeating unit structures.

20,
A Conformation Zoning plot for native polysaccharides (Fig. 5) shows semi-flexible structures for native MenA in its Na + salt which becomes more flexible in the Ca 2+ salt. It also shows semi-flexible structures of native MenW135 and native MenY and more flexible random coil structure of native MenC. After the size reduction process "micro-fluidization" and the activation of the polysaccharides they are shown to be adopting semi-flexible structures. On the other hand, the zoning plot of meningococcal conjugates (Fig. 6) confirms the significant increase in the flexibility of the conjugates (Zone D -random coils) compared to their native, micro-fluidized and activated counterparts.

Concluding Remarks
The conformation analyses suggest semi-flexible random coil structures for MenA Na + salt, MenW135 and MenY polysaccharides and more flexible random coil structures for MenA Ca 2+ salt and MenC polysaccharides. The micro-fluidization significantly reduced the viscosity and molar mass. Conformation of those micro-fluidized polysaccharides has changed to less flexible structures. Similarly the conformation of the activated polysaccharides slightly changed to less semi-flexible structures, their molar masses have also decreased from the native polysaccharides. The conjugation process has altered the viscosity of the final conjugates and remarkably increased the molar masses of the final conjugates. However, the conformation of the conjugates has become more flexible than native, micro-fluidized and activated polysaccharides and that the conjugates are highly flexible chains or highly flexible random coil structures. The lower intrinsic viscosities compared with the native or activated polysaccharides is indicative of some degree of wrapping of the polysaccharide chains around the tetanus toxoid protein core, although, like with glyco-vaccines based on Hib 26 it seems it is the carbohydrate polymer component which dictates the hydrodynamic properties of these substances.

Methods
Sample preparation. Preparation of polysaccharides was similar to that described for Streptococcus pneumoniae 24 . Prior to their conjugation, polysaccharides undergo size reduction by micro-fluidization in a high pressure homogenizer, which significantly reduces the viscosity and is likely to enhance the polysaccharides reactivity during conjugation 22,40,41 . During the development of the quadrivalent conjugate vaccine used in this study, different conjugation chemistries have been used in coupling of the polysaccharides to the carrier protein using a spacer or direct linking them without a spacer. Hence, samples used in this study are native polysaccharides from serogroups C (MenC), W-135 (MenW135), Y (MenY) and serogroup A sodium salt (MenA-Na + ) and calcium salt (MenA-Ca 2+ ). Micro-fluidized polysaccharides from serogroups A, C, W-135 and Y, activated polysaccharides from serogroups A and C, were also examined. Serogroup A conjugate without spacer (MenA without spacer), sodium salt conjugate with spacer (MenA-Na + with spacer), calcium salt conjugate with spacer (MenA-Ca 2+ with spacer) and serogroup C conjugate with spacer (MenC with spacer), conjugate without spacer (MenC without spacer) and serogroup W-135 conjugate without spacer (MenW135 without spacer) and serogroup Y conjugate without spacer (MenY without spacer), were also investigated.
Polysaccharide-protein conjugation with a spacer. The covalent binding of the polysaccharides and the spacer adipic acid dihydrazide (ADH) is carried out using a coupling chemistry by which the polysaccharides are activated under controlled conditions by a suitable cyanylating agent, 1-cyano-4-dimethylamino-pyridinium tetrafluoroborate (CDAP) see e.g. 41,42 . CDAP reacts with the polysaccharide, exchanging a cyano group for hydroxyl hydrogen, hydroxyl groups being abundant on the polysaccharide, creating a highly reactive cyanoester. The activation is best done at pH 9-10, and in fact there is a strong pH dependence on CDAP polysaccharide (PS) activation efficiency 40 . The spacer reacts with the cyanylated-polysaccharide through its hydrazino groups, to form a stable isourea link between the spacer and the polysaccharide. The conjugation of the tetanus toxoid (TT) with PS-ADH consists of the covalent binding of the protein on the derivatized PS using 1-ethyl-3-(3 -dimethylaminopropyl) carbodiimide (EDC) 43,44 . A stable peptidic link is formed between the carboxylic group of the protein and the hydrazino group of the spacer (ADH).
Conjugation without a spacer. As an alternative to using a linker (spacer), direct linkage can be used.
Activation of polysaccharides with CDAP introduces a cyano group in the polysaccharides and dimethylaminopyridine (DMAP) is liberated. The cyano group reacts with NH 2 -groups of the protein during the subsequent coupling phase leading to binding of the polysaccharide to the TT protein by a way of an isourea link. Prior to their purification, the conjugates are filtered through a 10 μ m membrane in order to remove potential aggregates. The conjugates are then purified on a Sephacryl S400HR column to remove the by-products and unbound protein and/or polysaccharides. They are finally sterile filtered on a 0.22 μ m membrane. All samples were dissolved in phosphate-chloride buffer pH~ 6.8, I = 0.1 45 , at 20.0 °C. All solutions were then diluted to the appropriate concentrations required for the hydrodynamic characterisations.

Sedimentation velocity in the analytical ultracentrifuge. Sedimentation velocity experiments
were performed using a Beckman (Palo Alto, CA, USA) Optima XL-I analytical ultracentrifuge equipped with Rayleigh interference optics and an automatic on-line data capture system. Conventional 12.0 mm double-sector epoxy cells with sapphire windows were loaded with 400 μ L of different concentrations (0.1-2.0 mg mL −1 ) of each sample and a matching amount of the corresponding reference buffer in appropriate channels. Native, micro-fluidized and activated capsular polysaccharides from N. meningitidis were centrifuged at 45000 rpm at a temperature of 20.0 °C, while the glycoconjugates solutions were spun at 7000 rpm and the same temperature. Concentration profiles in the analytical ultracentrifuge cell were recorded using the Rayleigh interference optical system and converted to concentration (in units of fringe displacement relative to the meniscus, j) versus radial position, r at particular times t 46 . Data were analysed using the least squares boundary modelling referred to as the least squares ls-g*(s) model incorporated into the SEDFIT algorithm 47 . SEDFIT generates an apparent distribution of sedimentation coefficients in the form of g*(s) versus s T,b , where the (*) indicates that the distribution of sedimentation coefficients has not been corrected for diffusion effects, see e.g. 46 . The correction procedure requires assumptions about the friction coefficient not valid for continuous distributions of molecular weight/ sedimentation coefficient, although for large polysaccharides/glycoconjugates, these corrections will be small 46 . This was followed by the correction to standard solvent conditions-namely the density and viscosity of water at 20.0 °C -to yield s 20,W using the utility program SEDNTERP 48 . Note that to account for non-ideality (co-exclusion and backflow effects), the apparent sedimentation coefficient (s 20,w ) was calculated at a series of different cell loading concentration and extrapolated to infinite dilution using the Gralén relation 49 : where k s is the Gralén or concentration dependence coefficient.

Sedimentation equilibrium (SE) in the analytical ultracentrifuge.
Samples were prepared at a concentration of 0.3 mg/mL sufficiently low to minimize the effects of thermodynamic non-ideality. A volume of 1.0 ml of each sample was dialysed for 48 hours at the ambient temperature, each against 300 mL of the phosphate-chloride buffer. Sedimentation equilibrium experiments were also performed using the Beckman (Palo Alto, CA, USA) Optima XL-I analytical ultracentrifuge equipped with Rayleigh interference optics and an automatic on-line data capture system. The modified long (20.0 mm) optical path length double-sector titanium cells with sapphire windows were loaded with 0.15 mL of dialysed sample and a matching amount of reference buffer dialysate in appropriate channels. Samples were centrifuged at rotor speeds selected to give a sufficient fringe increment from meniscus to base 50 i.e. 4000, 9000, 9000 and 2000 rpm for native, micro-fluidized, activated polysaccharides and glycoconjugates, respectively. Using the Rayleigh interference optical system, scans were taken every hour and equilibrium was reached after approximately 48 hours. Record of the relative concentration distribution of the solute at equilibrium was analysed to give the weight average apparent molar mass M w,app using the SEDFIT-MSTAR algorithm 51 based on an earlier algorithm of Cölfen and Harding 52 and the M* function of Creeth and Harding 53 . The use of the long path length cells (20.0 mm) meant that low loading concentrations can be used to give a sufficient signal (~0.3 mg mL −1 ). At such low concentrations, non-ideality effects will be small and hence the apparent weight average molar mass will be approximately equal to the true weight average molar mass M w . M w from SEDFIT-MSTAR could then be combined with the weight average sedimentation coefficient to yield the molecular weight distribution 54 . We also estimate M z using the MFIT algorithm of Ang and Rowe 55 .

Viscometry. Dynamic viscosity measurements for Men native, Men-ADH capsular polysaccharides and the
Men-TT conjugate, were carried out using the automated micro-viscometer Anton Parr AMVn (Anton Parr, Graz, Austria) across a concentration series from 0.1-2.0 mg mL −1 . The rolling ball viscosity method measures the time of a steel ball needed to roll in the 1.6 mm diameter silanized glass capillary containing each sample. The experiment was performed at different reclining angles of 70° (n = 4 times), 60° (n = 4 times) and 50° (n = 6 times) under precise temperature control (20.00 ± 0.01) °C. Huggins 56 and Kraemer 57 extrapolations forms were performed to obtain the intrinsic viscosity, as described in ref 26. Intrinsic viscosities were also estimated using the Solomon -Ciutâ relation which represents a combination of the opposite trends of the Huggins and Kraemer relations allowing estimation of [η] without extrapolation 58,59 : Nonetheless as a check on consistency of data such estimates are still registered at different concentrations.