Immunochemical studies of Shigella flexneri 2a and 6, and Shigella dysenteriae type 1 O-specific polysaccharide-core fragments and their protein conjugates as vaccine candidates

Abstract

There is no licensed vaccine for the prevention of shigellosis. Our approach to the development of Shigella vaccine is based on inducing serum IgG antibodies to the O-specific polysaccharide (O-SP) domain of their lipopolysaccharides (LPS). We have shown that low molecular mass O-SP-core (O-SPC) fragments isolated from Shigella sonnei LPS conjugated to proteins induced significantly higher antibody levels in mice than the full length O-SP conjugates. This finding is now extended to the O-SPC of S. flexneri 2a and 6, and S. dysenteriae type 1. The structures of O-SPC, containing core plus 1–4 O-SP repeat units (RU), were analyzed by NMR and mass spectroscopy. The first RUs attached to the cores of S. flexneri 2a and 6 LPS were different from the following RUs in their O-acetylation and/or glucosylation. Conjugates of core plus more than 1 RUs were necessary to induce LPS antibodies in mice. The resulting antibody levels were comparable to those induced by the full length O-SP conjugates. In S. dysenteriae type 1, the first RU was identical to the following RUs, with the exception that the GlcNAc was bound to the core in the β-configuration, while in all other RUs the GlcNAc was present in the α-configuration. In spite of this difference, conjugates of S. dysenteriae type 1 core with 1, 2, or 3 RUs induced LPS antibodies in mice with levels statistically higher than those of the full size O-SP conjugates. O-SPC conjugates are easy to prepare, characterize, and standardize, and their clinical evaluation is planned.

Introduction

Shigella is a major cause of diarrhea and dysentery worldwide, affecting millions of individuals, especially young children, and causing over a million deaths annually [1–3]. The infectious dose of Shigellae is very low, and ingestion of just 10 organisms may cause severe diarrhea and dehydration. The most dangerous form of the disease, dysentery, is characterized by bloody and mucous diarrhea, abdominal cramps, rectal pain, and fever; it is also a major cause of stunted growth [4].

Shigellae are divided into four groups: Group A - Shigella dysenteriae, Group B - Shigella flexneri, Group C - Shigella boydii, and Group D - Shigella sonnei [5]. The groups are further divided into serotypes based upon the structure of the O-SP)of their LPS) The Shigella O-SP is an essential virulence factor and a protective antigen [6].

Despite the discovery of the Shiga bacillus (S. dysenteriae type 1) in Japan over a century ago, there is still no licensed vaccine for shigellosis [2]. S. sonnei is the most common type in industrialized countries, as is S. flexneri 2a in developing countries. The most severe disease is caused by S. dysenteriae type 1, which is endemic and epidemic in developing countries. Experimental vaccines composed of protein conjugates of the O-SPs of S. dysenteriae type 1, S. sonnei, and S. flexneri 2a were safe, and induced specific LPS antibodies in young adults. The S. sonnei conjugate was > 70 % protective in young adults and in 3–4 year olds [3,6]. More immunogenic vaccine candidates were prepared using synthetic S. dysenteriae type 1 saccharides containing 2, 3, or 4 O-SP RU and low molecular mass O-SP-core (O-SPC) fragments containing an average of 3–4 RUs of S. sonnei LPS bound to carrier proteins [7,8]. Now, we examine if the latter approach can be extended to other Shigellae: S. flexneri types 2a and 6, and S. dysenteriae type 1.

Material and Methods

Bacteria and isolation of LPS

S. flexneri 2a strain 2457T10 and S. dysenteriae type 1 strain 1617 were from Dr. Sam Formal (WRAIR), S. flexneri 6 was a clinical isolate from a child with shigellosis in Israel [3]. All strains were cultivated as described [9]. LPSs were extracted by the hot phenol method and purified as described [10]. The average yields of LPS for each strain were 1.3–1.5% of wet cell mass.

Isolation of oligosaccharides

LPS (200 mg) was treated with 1% acetic acid at 100°C for 1.5 h. Lipid A was removed by ultracentrifugation at 35,000 rpm for 5 h at 4°C, and the soluble product subjected to gel chromatography on a BioGel P-10 column (1×100 cm) in pyridine/acetic acid/water buffer (4/8/988 mL), monitored with a Knauer differential refractometer. Fractions containing core plus one RU (core-1RU) were further purified by Hi-Trap Q anion-exchange chromatography using a 0–1 M NaCl gradient in water. The ratios of oligosaccharides obtained after gel filtration were as follows: for S. flexnerii 6: full length O-SP comprised 60.8%; F2, 3.6%; F3, 6.3% and F4, 14%; for S. flexnerii 2a: full length O-SP comprised 57.8%; F2, 7.3 %; F3, 7.4% and F4, 9.0%; for S. dysenteriae type 1: full length O-SP comprised 66.1%; F2 5.8 %; F3, 11.7% and F4, 15.5%.

Analytical methods

Protein concentration was measured by the method of Lowry [11], sugar concentration by the anthrone assay [12]. SDS-PAGE used 14% gels according to the manufacturer's instructions (Bio-Rad, Hercules, CA). Immunodiffusion was performed in 1% agarose in PBS.

Methylation and monosaccharide analysis

Methylation was performed using the DMSO-NaOH-MeI procedure [13]. Methylated and non-methylated oligosaccharides were hydrolyzed with 3M TFA at 120°C for 3h and converted into alditol acetates [13]. Samples were analyzed by GLC-MS using an Agilent 6850 chromatograph equipped with DB-17 (30m × 0.25mm) fused-silica column using a temperature gradient of 180°C →240°C at 2°C/min or by a Varian Saturn 2000 GC-MS using the same column.

NMR Spectroscopy

¹H and ¹³C NMR spectra were recorded using a Varian Inova 600 MHz spectrometer for samples in D₂O at 25–35 °C with acetone as an internal reference (2.225 ppm for ¹H and 31.5 ppm for ¹³C) and standard pulse sequences for double quantum filtered COSY, TOCSY (mixing time 120 ms), NOESY (mixing time 400 ms), gradient HSQC, and HMBC (100 ms long range transfer delay).

Mass spectrometry

Capillary electrophoresis - mass spectrometry (CE-MS) used a 4000 QTRAP mass spectrometer (Applied Biosystems/MDS Sciex, Toronto, ON, Canada) with a Prince CE system (Prince Technologies, Amsterdam, The Netherlands) with a 90 cm length, bare fused-silica capillary, and 15 mM ammonium acetate in deionized water, pH 9.0, in the injection module. A sheath solution (isopropanol-methanol, 2:1 v/v) was delivered at a flow rate of 1.0 µL/min. A −5 kV electrospray ionization voltage was used for negative ion detection modes.

MALDI-TOF mass spectra of the derivatized proteins and of the conjugates were obtained with an OmniFlex MALDI-TOF instrument (Bruker Daltonics, Billerica, MA) operated in the linear mode. Samples for analysis were desalted, and a 1 µL aliquot, mixed with 20 µL of sinnapinic acid matrix made in 30% CH₃CN and 0.1% TFA. Of this mixture, 1 µL was applied to and dried on the sample stage, then placed in the mass spectrometer.

Preparation of conjugates

Conjugates were prepared similarly to those of S. sonnei O-SPC [8]. To 15 mg of BSA (Sigma, St. Louis, MO) in 2.2 mL Buffer A (PBS, 0.1% glycerol, 5 mM EDTA, pH 7.2), 4 mg of SBAP (Pierce, Rockford, IL) in 40 µL DMSO were added and reacted with mixing for 1.5 h, pH 7.2, at room temp. Next, the reaction mixture was applied to a Sephadex G-50 column (1 × 50 cm) in 0.2 M NaCl, and the void volume fractions (BSA-Br) concentrated using an Amicon Ultra-15 centrifuge filter device (Millipore, Billerica, MA) to 2.6 mL (13 mg recovered); 0.2 mL was removed for analysis. An average of 30 linkers per BSA molecule was introduced under these conditions. as assayed by MALDI-TOF. To 12 mg of BSA-Br in 2.4 mL Buffer A, 10 mg O-(3-thiopropyl)hydroxylamine, prepared in our laboratory [14], dissolved in 300 µL of 1M K₂HPO₄, was added and reacted with mixing for 3 h, pH 7.2, at room temp. The reaction mixture was then passed through the Sephadex G-50 column and the void volume fractions (BSA-ONH₂) concentrated to 2.6 mL as above, and 0.2 mL removed for analysis. BSA-ONH₂, 10 mg, was reacted with 25 mg O-SPC in 3 mL of Buffer A, overnight, pH 5.6, at room temp, with mixing. The solution was then passed through Sepharose G-75 (1 × 100 cm) in 0.2 M NaCl and the void volume fractions collected and analyzed for sugar, protein and molecular mass by MALDI-TOF and SDS-PAGE. The following conjugates obtained in this way were designated: 1: BSA/S. flexneri 2a O-SPC-F2; 2: BSA/S. flexneri 2a O-SPC-F3, 3: BSA/S. flexneri 6 O-SPC-F3, 4: BSA/S. flexneri 6 O-SPC-F4, 5: BSA/S. dysenteriae type 1 O-SPC-F2, 6: BSA/S. dysenteriae type 1 O-SPC-F3, 7: BSA/S. dysenteriae type 1 O-SPC-F4. A conjugate of full length S. flexneri 6 O-SP bound to BSA was prepared as described for S. flexneri 2a [9].

Reactivity of O-(3-thiopropyl)hydroxylamine with O-acetyl groups of O-SP

This study was conducted by 1D ¹H NMR at 500 MHz by using a Bruker DRX-500 spectrometer equipped with a 5 mm TXI cryoprobe. A spectral width of 11 ppm was used, together with 32,768 point data sets, forward linear predicted to 65,536 points, a 30° pulse (2.7 µs), and a recycle time of 6 s. S. flexneri 6 O-SP (5.3 mg) was dissolved in 0.5 mL of 99.96 atom% D₂O, pH was 6.0, and an initial ¹H spectrum was run to define the OAc/NAc pattern. The Ac resonances fell into two groups: 2.196, 2.172, 2.159, 2.156, and 2.149 ppm, which are assigned as OAc, and 2.085, 2.079, 2.074, 2.031, and 2.027 ppm which are assumed to be NAc. It was reasoned that if O-(3-thiopropyl)hydroxylamine reacted with OAc on the O-SP, then signal(s) would disappear from the OAc group and appear in the NAc group, or at least some changes in the relative intensities of the signals would be observed. An aliquot of 1 mg of O-(3-thiopropyl)hydroxylamine solution (corresponding to about a 1:1 molar ratio of O-Ac groups to O-(3-thiopropyl)hydroxylamine, and brought to pH 5.7 to reproduce the conjugation conditions) was added to the O-SP solution. ¹H spectra were run at 5–13 min, 31–37 min, 60–66 min, 120–146 min, and 22 h 2 min-22 h 28 min, and the spectra compared with that of the O-SP by overlaying the spectra in dual display mode. No change in the relative intensities of the OAc and NAc groups was observed, and no changes in other parts of the spectra were observed that would support an on-going chemical reaction.

Immunization

Five-six weeks-old female NIH Swiss Webster mice were injected s.c. two or three times at two week intervals with 2.5 µg saccharide as a conjugate, in 0.1 mL PBS. Groups of 10 mice were exsanguinated seven days after the second or the third injections [15]. Controls received PBS.

Antibodies

Serum IgG antibodies were measured by ELISA using S. flexneri 2a and 6 and S. dysenteriae type 1 LPS as coating antigens. The results were computed using an ELISA data processing program provided by the Biostatistics and Information Management Branch, CDC [16]. Antibody levels were calculated relative to a standard, a pool of high antibody level sera induced by the O-SP conjugates, assigned a value of 100 EU. A polyclonal rabbit antiserum, obtained by immunizing rabbits with multiple i.v. injections of heat-killed S. flexneri 2a or S. dysenteriae type 1 bacteria and mice serum raised against S. flexneri 6, were used in the immunodiffusion assays.

Statistics

Data are presented as GM values of groups of 10 mice. An unpaired t test was used to compare GMs between groups.

Results

Fragments of core and core with one or several RUs for each Shigella type separated by gel chromatography are shown in Fig. 1.

Figure 1.

BioGel P-10 gel filtration profiles of 2% acetic acid hydrolyzed Shigellae LPSs. F1 – O-SP; more then 10 RU; F2, F3 – core + avr. 2–5 O-SP RU; F4 - core + avr. 1 O-SP RU.

Structures of the core fragments

The inner part of the core of all compounds had the common enterobacterial structure (Kdo+3Hep, or Kdo+3Hep+GlcN). Several variants differing in phosphorylation of O-4 of Hep F and in the presence of GlcN N on O-7 of Hep G were observed (Fig. 2). As published for E. coli [17], Hep F was phosphorylated when GlcN N was absent (S. dysenteriae type 1 and S. flexneri 2a) or Hep G was substituted with GlcN N when Hep F had no phosphate (S. flexneri 6). In S. sonnei both the GlcN on Hep G and the phosphate on Hep F O-4 were described [8,18]. NMR spectral assignments for the inner part of the core were in agreement with published data for related structures, and are not shown here. The ¹H and ¹³C signal assignments of the outer part of the cores are shown in Tables 1–3. Monosaccharides were identified on the basis of vicinal proton coupling constants and ¹³C NMR chemical shifts. Anomeric configurations were deduced from the J_1,2 coupling constants, from the chemical shifts of H-1 and C-1, C-3 and C-5 signals, and from the intraresiduel NOEs. Linkages between monosaccharides were identified on the basis of NOE and HMBC correlations. In addition to transglycosidic NOE correlations between H-1 and a proton at the linkage site, all cores showed correlations from the M1 and L1 protons to Glc H H-4, indicating tight packing of the outer part of the core. The core of S. flexneri 2a conformed to that of the E. coli R3 type, S. flexneri 6 to the R1 type and S. dysenteriae type 1 to the R4 type.

Figure 2.

Structures of Shigellae O-SPC fragments of core plus one O-SP repeat unit (RU).

Table 1.

NMR data for S. flexneri 2a core plus one O-SP repeat unit.

Unit		1	2	3	4	5	6a/b
Rha T0 no Ac	H	4.96	4.08	3.80	3.44	3.70	1.26
	C	103.5	71.3	71.7	73.1	70.3	17.8
Rha T2 2-O-Ac	H	5.01	5.24	3.99	3.49	3.78	1.29
	C	100.4	73.5	70.7	73.4	70.4	17.8
Rha T3 3-O-Ac	H	4.98	4.23	5.00	3.65	3.82	1.29
	C	103.3	71.0	74.7	72.4	70.4	17.8
Rha T4 4-O-Ac	H	5.01	4.13	4.01	4.89	3.87	1.16
	C	103.3	71.2	70.7	75.3	68.1	17.8
Rha Z	H	5.16	4.06	3.92	3.49	3.73	1.31
	C	102.0	79.2	71.2	73.3	70.4	17.9
Rha Y	H	4.84	3.82	3.79	3.55	4.03	1.25
	C	102.7	71.8	78.2	72.9	70.3	17.6
GlcNAc X	H	4.59	3.95	3.62	3.57	3.49	3.75/3.94
	C	102.5	56.7	83.2	69.6	77.3	61.7
Glc L	H	5.52	3.79	4.09	3.57	3.80	3.71/3.87
	C	91.6	75.6	72.0	81.4	73.4	61.8
Glc M	H	5.21	3.55	3.75	3.45	3.99	3.75/3.80
	C	97.2	70.8	74.5	70.6	73.1	61.8
Gal K	H	5.90	4.11	4.25	4.30	4.29	3.77/3.79
	C	95.1	68.7	71.2	65.3	71.1	61.7
GlcNAc P	1H	5.11	4.02	3.80	3.57	4.05	3.79/3.93
	13C	93.5	54.7	72.3	71.3	73.2	61.9
Glc H	H	5.20	3.68	4.14	3.82	3.90	3.78/3.89
	C	102.5	71.4	76.2	71.8	73.5	60.8

Table 3.

NMR data for S. dysenteriae type 1 core plus one O-SP repeat unit.

Unit		1	2	3	4	5	6a/b
Rha T	H	5.08	4.07	3.85	3.47	3.85	1.32
	C	103.5	71.4	71.4	73.3	70.3	17.8
Rha Z	H	5.06	4.17	3.87	3.56	3.88	1.32
	C	102.8	70.8	79.2	72.6	70.5	17.8
Gal Y	H	5.60	3.95	3.93	4.10	3.90	3.77/3.77
	C	98.7	75.0	70.1	70.5	72.2	61.8
GlcNAc X	H	4.65	3.84	3.84	3.71	3.49	3.77/3.93
	C	102.2	55.3	78.4	72.6	77.0	61.7
Gal P	H	4.47	3.56	3.65	3.91	3.81	3.89/3.96
	C	104.3	72.1	73.6	69.3	74.0	68.5
Gal M	H	5.27	3.86	3.97	4.02	4.17	3.76/3.76
	C	97.1	69.5	70.6	71.0	72.5	62.5
Gal L	H	5.58	4.02	4.12	4.07	4.10	3.77/3.77
	C	93.7	73.5	69.0	70.5	72.4	62.4
Glc K	H	5.76	3.78	4.03	3.73	4.18	3.88/4.01
	C	96.1	75.1	71.2	79.3	71.3	60.9
Glc H	H	5.21	3.68	4.08	3.77	3.92	3.82/3.89
	C	102.6	71.4	77.8	71.5	73.6	61.0

Structure and immunogenicity of core and one O-SP RU (core-1RU)

In all cases, the RU was attached to the core by its aminosugar (β-GlcNAc in S. dysenteriae type 1 and S. flexneri 2a, and β-GalNAc in S. flexneri 6). Mass spectroscopy analyses of all compounds showed significant heterogeneity due to variable phosphorylation, acetylation, and the presence of EtN, GlcN (N) side chain, and anhydro-Kdo.

S. flexneri 2a

Assignment of the S. flexneri 2a core with one RU spectrum (F4, Figure 1) revealed the attachment of the O-chain to the O-4 of core Glc L. This was confirmed by the results of methylation analysis, which showed the presence of 2,4-disubstituted Glc; all other expected methylated sugars were also identified. The lack of a Glc side chain in the first RU was also confirmed [19,20]. A terminal Rha was present in four variants: non-acetylated, and mono-acetylated at O-2, O-3 or O-4 (Fig. 3). Acetylation was deduced from the downfield shift of the proton signal at the acetylation site. No acetylation of GlcNAc X was observed. Another difference of the first RU from the following RUs, was that GlcNAc was partly acetylated at O-6. ESI-MS of the core-1RU showed the main species having a mass of 2511 amu, corresponding to Rha₃Hex₄Hep₃GlcN₂Kdo₁P₂Ac₃. Higher mass ions were observed at 2553 amu (+EtN or Ac), 2591.4, and 2634.6 amu.

Figure 3.

Fragment of COSY (green-cyan) and TOCSY (red) spectra of the core-RU from the LPSs of S. flexneri O2a (top) and S. flexneri O6 (bottom), illustrating O-acetylation of the terminal rhamnose residues (T). Proton signals at the acetylation site are shifted to low field.

This fraction (core-1RU) conjugated to BSA at an average of 10 saccharide chains per protein, did not react with rabbit anti-S. flexneri 2a serum by immunodiffiusion, and was not used for mice immunization.

S. flexneri 6

Complete assignment of core-1RU for S. flexneri 6 is presented in Table 2. Attachment of the RU GlcNAc X to core O-3 of Glc P followed from NOE data, and was further confirmed by methylation analyses, where the terminal Glc of the core was replaced by 3-substituted Glc in the core-1RU. As in S. flexneri 2a, the terminal Rha residue was present in four variants: non-acetylated, and mono-acetylated at O-2, O-3 or O-4 (Fig. 3). The most intense peak in the mass spectrum corresponded to Rha₂Hex₅Hep₃GalA₁GlcN₂Kdo1P₁Ac₂ with 2582.4 amu, and a second one to +PEtN = 2702.8 amu. In this spectrum, one Ac is N-acetyl group of GalNAc (X) and another is O-acetyl group of Rha (T); a minor peak of non-O-acetylated variant was also observed at 2540.1 amu.

Table 2.

NMR data for S. flexneri 6 core plus one O-SP repeat unit.

Unit	Atom	1	2	3	4	5	6 (a/b)
Rha T0 no Ac	H	4.96	4.06	3.88	3.43	3.76	1.26
	C	103.3	71.3	71.7	73.4	70.4	17.8
Rha T2 2-O-Ac	H	5.02	5.22	3.97	3.49	3.48	1.29
	C	100.5	73.6	70.7	73.4	70.4	17.8
Rha T3 3-O-Ac	H	4.98	4.20	4.99	3.64	3.82	1.30
	C	103.0	69.4	74.7	73.4	70.4	17.8
Rha T4 4-O-Ac	H	5.00	4.11	3.99	4.88	3.87	1.16
	C	103.2	71.2	70.7	75.3	68.1	17.8
Rha Z	H	5.37–5.42	4.08–4.10	3.88	3.44;3.49*	3.65	1.26
	C	101.0	80.9	70.9	73.2	70.6	17.8
GalA Y	H	4.51	3.63	3.84	4.35	4.33
	C	105.7	71.1	74.0	77.4	74.5	172.0
GalNAc X	H	4.79	4.08	3.91	4.26	3.74	3.83/3.83
	C	103.0	52.9	81.6	69.0	76.1	62.3
Gal L	H	5.62	3.99	4.22	4.01	4.16	3.74/3.74
	C	92.6	73.6	69.2	70.5	72.5	62.6
Gal M	H	5.33	3.86	4.00	3.96	4.16	3.74/3.74
	C	97.0	69.5	70.5	70.7	72.5	62.6
Glc K	H	5.84	3.89	4.20	3.59	4.12	3.81/3.87
	C	95.8	73.7	79.1	69.0	72.5	61.7
Glc H	H	5.28	3.69	4.06	3.80	3.86	3.86/3.91
	C	101.7	71.7	77.5	71.7	73.7	61.2
Glc P	H	4.76	3.42	3.74	3.53	3.47	3.75/3.92
	C	103.6	73.9	86.0	69.6	76.6	62.0

This fraction (core-1RU) bound to BSA at an average of 12 saccharide chains per BSA molecule, reacted with anti-S. flexneri type 6 serum by immunodiffiusion, but induced only low levels of anti-LPS antibodies in mice (Table 5).

Table 5.

Composition and Geometric Mean of serum IgG anti-S. flexneri 6 LPS induced by O-SP and O-SPC bound to BSA.

Conjugate	Mol. mass of conjugate [kDa]	Mol. mass of O-SPC [Da]	No of chains per BSA	BSA: Sugar ratio [wt:wt]	IgG anti- LPS [EU]1
BSA/F1: core + 15–20 RU2	na3	>10000	na	1:0.4	21.3
BSA/F2: core + 7 RU	96	6607	3.5	1:0.4	23.7
BSA/F3: core + 2.5 RU	106	3588	9	1:0.5	1.6
BSA/F4: core + 1 RU	106	2581	12	1:0.5	1.2

¹EU – ELISA units.

²The number of O-SP repeat units (RU) is an average for the given fraction.

³na – not analyzed.

Mice (10 per group) were injected with 2.5 µg of saccharide as a conjugate per mouse, 3 times, 2 weeks apart and bled one week after the 3^rd injection.

S. dysenteriae type 1

Assignment of core-1RU of S. dysenteriae type 1 is presented in Table 3. The first RU was identical to the following RUs with the exception that GlcNAc was bound to the core in the β-configuration, while in all other RUs it was present in the α-configuration. The attachment of the first O-SP β-GlcNAc to O-6 of core Gal followed from NOE and HMBC data, and was confirmed by methylation analysis.

Core-1RU bound to BSA at an average of 23 oligosaccharide chains per protein reacted with anti- S. dysenteriae type 1 serum by immunodiffiusion, and when injected into mice induced LPS antibodies (Table 6).

Table 6.

Composition and Geometric Mean of serum IgG anti-S. dysenteriae type 1 LPS induced by O-SPC conjugates bound to BSA.

Conjugate	Mol. mass of conjugate [kDa]	Mol. Mass of O-SPC [Da]	No of chains per BSA	BSA: Sugar ratio [wt:wt]	IgG anti- LPS [EU]1
BSA/F2: core + 3 RU2	130	3550	16	1:0.9	10.0
BSA/F3: core + 2 RU	133	2892	21	1:0.9	7.6
BSA/F4: core + 1 RU	125	2235	23	1:0.8	21.9

¹EU – ELISA units.

²The number of O-SP repeat units (RU) is an average for the given fraction.

³na – not analyzed.

Mice (10 per group) were injected with 2.5 µg of saccharide as a conjugate per mouse, 3 times, 2 weeks apart and bled one week after the 3^rd injection.

Structure and immunogenicity of core and several RUs (O-SPC)

Longer fractions (F2 and F3, Figure 1) were too heterogeneous in size and substitutions to interpret their mass spectra, therefore integration of signals in ¹H-NMR spectra was used to calculate an average number of O-SP repeats in an O-SPC chain.

S. flexneri 2a

For F2 and F3, integrations of the H-6 proton signals for Rha methyl groups in ¹H-NMR spectra (1.16–1.31 ppm) relative to the anomeric signals of core α-Gal K (5.90 ppm) and α-Glc L (5.52 ppm) were used to calculate the average number of O-SP repeats (Fig. 4). F3 contained core plus an average of 2 RU and F2 contained core with an average of 3 RU (Figure 1). Both fractions were bound to BSA at an average of 8 chains per BSA molecule. These conjugates reacted with anti-S. flexneri 2a rabbit serum by immunodiffiusion and induced LPS antibodies in mice with booster responses (Table 4) .

Figure 4.

Integration of the proton signals of S.flexneri 2a 1 O-SPC fractions for the determination of the average number of O-SP repeat units (RU). Rha methyl signals in ¹H-NMR spectra (1.16–1.31 ppm) in the O-SP relative to the anomeric signals of the core α-Gal K (5.90 ppm) and α-Glc L (5.52 ppm) were used. F1-O-SP; F2-core-3RU; F3-core-2RU; F4-core-1RU; F5-core.

Table 4.

Composition and Geometric Mean of serum IgG anti-S.flexneri 2a LPS induced by O-SPC bound to BSA.

Conjugate	Mol. Mass of conjugate [kDa]1	Mol. mass of O-SPC [Da]	No of chains per BSA	BSA: Sugar ratio [wt:wt]	IgG anti-LPS [EU]2
					2nd inj.	3rd inj.
BSA/F2: core + 3 RU3	103	4202	8	1:0.5	0.1	10
BSA/F3: core + 2 RU	101	3356	8	1:0.5	0.2	11

¹EU – ELISA units.

²Mass of O-SPC is calculated assuming one Glc per RU, except the first one, and one O-acetyl group. The number of O-SP repeat units (RU) is an average for the given fraction.

Mice (10 per group) were injected with 2.5 µg of saccharide as a conjugate per mouse, 2 or 3 times, 2 weeks apart and bled one week after the last two injections.

S. flexneri 6

Similarly to S. flexneri 2a, integrations of the Rha methyl signals in ¹H-NMR spectra (1.16–1.30 ppm) relative to the anomeric signals of core α-Glc K (5.84 ppm) and α-Gal L (5.62 ppm) were used to calculate the average number of O-SP repeats. F3 contained an average of 2.5 RU and was bound to BSA at an average of 9 chains per BSA molecule (Table 5). This conjugate was antigenic, but similarly to the F4 conjugate (core-1RU) was poorly immunogenic. The F2 conjugate composed of about 7 RU plus core bound at an average density of 3.5 chains per protein was a better immunogen (p<0.01). It was not more immunogenic than the full length O-SP bound to BSA by multiple attachments using ADH as a linker (Table 5).

S. dysenteriae type 1

To estimate the number of RU in S. dysenteriae type 1 fractions, integration of the anomeric signal of the terminal Rha T (5.08 ppm) relative to the substituted T (5.11 ppm) was used (Fig. 5). F2 contained core plus an average of 3 RUs, and F3 - 2 RUs. 2D NMR of F3 confirmed the β-configuration (H/C-1: 4.65/102.1; H/C-2: 3.84/55.3 ppm) of GlcNAc in the first RU, linked to core. The GlcNAc in all the following RUs of the O-SP was in the α-configuration (H/C-1: 5.05/95.2; H/C-2: 4.15/53.2 ppm). Separate signals for Gal Y, linked to β- (in the first RU) and α-GlcNAc (in the second RU) were also observed with corresponding NOE correlations to H-3 of GlcNAc. Both fractions bound to BSA were antigenic and induced similar antibody levels in mice (Table 6).

Figure 5.

Integration of the anomeric signals of S. dysenteriae 1 O-SPC fractions for the determination of the average number of O-SP repeat units (RU). The anomeric signal of the terminal Rha T (5.08 ppm) relative to that of the substituted Rha T (5.11 ppm) was used. F1-O-SP; F2-core-3RU; F3-core-2RU; F4-core-1RU.

Discussion

The LPS of Gram-negative bacteria may be present in three forms: “rough” consisting of Lipid A and a core only, “semi-rough” consisting of Lipid A, core and a small number of O-SP RUs (O-SPC), or as “smooth” with 10–20 RUs (O-SP). The LPS molecule can also undergo bacteriophage-encoded modification like glucosylation or O-acetylation [19]. Most of the S. flexneri LPS cores conformed to the E. coli R3 type [21,22]. We confirmed the presence of R3 type core in S. flexneri 2a and demonstrated the linkage and the unique structure (lack of Glc side chain and the pattern of O-acetylation) of the first O-SP RU (Figure 2). The structure of the S. flexneri 6 core, including the saccharide to which the O-chain is attached, but without indication of the exact attachment position, has been reported [23]. Now, we have confirmed that the S. flexneri 6 core conforms to the E. coli R1 type and have shown that the O-SP is attached to the core through its β-GalNAc (Figure 2). The core of S. dysenteriae type 1 conforms to that of E. coli R4 type. The first O-SP RU was identical to the other RUs, with the exception that the GlcNAc was bound to the core in the β-configuration, while in all other RUs, the GlcNAc was present in the α-configuration [24]. For S. sonnei, the structure of the LPS R1 type core and the linkage of the O-SP have been reported [18].

Investigational vaccines composed of protein conjugates of the O-SP of S. dysenteriae type 1, S. sonnei, and S. flexneri 2a were safe and elicited specific LPS antibodies in young adults [9]. The S. sonnei O-SP conjugate had > 70% efficacy in Israeli army recruits and in children older than three years of age [3,6]. As shigellosis peaks in 1–4 year old children, a more immunogenic vaccine is needed. The immunogenicity of S. dysenteriae type 1 and S. sonnei conjugates was significantly improved by using low molecular mass saccharides, whether synthesized or isolated from LPS, bound by their reducing ends to protein carriers at defined densities [7,8]. Now, we examined if a similar approach could be adapted to other Shigella LPS. The first RU of S. flexneri 2a O-SP was different from the following RUs of the O-SP because it lacked a Glc side chain, there was no O-Ac on GlcNAc, and its non-reducing terminal Rha was heterogeneously O-acetylated at all positions [25,26]. Probably, because of these modifications, core-1RU conjugated to protein was not antigenic using anti-LPS serum. Longer O-SPC with an average 2 or 3 O-SP RUs were antigenic and induced LPS antibodies in mice with booster responses. Antibody levels were comparable, but not higher than those induced by O-SP conjugates used in clinical trials [9]. Synthetic protein-carbohydrate conjugates of S. flexneri 2a O-SP fragments were shown to be immunogenic in mice but it is difficult to compare antibody levels between laboratories [27]. Similar to S. flexneri 2a, the O-acetylation of the first RU of S. flexneri 6 was different from the following RUs: the terminal Rha was heterogeneously O-acetylated at all positions, while in the latter RUs this Rha was reported to be O-acetylated on position O-3 only [28]. That could be part of the reason why O-SPC with an average of 1 or 2 RU induced only low antibody levels. Conjugates of O-SPC with an average of 7 RUs and of full length O-SP induced similar and higher antibody levels than those containing 1 or 2 RUs. In the case of S. dysenteriae type 1, all RUs were identical with the exception of the β-configuration of the GlcNAc bound to the core (the α-configuration is present in all other RUs). In this case, the core with even 1 RU was antigenic, and also immunogenic in mice. Antibody levels induced by these conjugates containing 1, 2 or 3 RU were similar to those of the synthetic saccharide conjugates, and statistically higher than those induced by the O-SP conjugates [7]. Tailoring of these conjugates to produce the most immunogenic constructs and characterization of the epitopes recognized by the O-SPC induced antibodies are planned.

The O-SPC conjugates, resembling synthetic conjugates, induced similar (S. flexnerii 2a and 6) or higher (S. sonnei and S. dysenteriae type 1) antibody levels in mice than the full length O-SP conjugates. They are easy to prepare, characterize and standardize. The conjugation method can be applied to other Gram-negative bacteria.

Abbreviations

COSY

correlation spectroscopy

TOCSY

total correlation spectroscopy

NOESY

nuclear Overhauser enhancement spectroscopy

HSQC

heteronuclear single quantum coherence

HMBC

heteronuclear multiple bond connectivity

LPS

lipopolysaccharide

O-SP

O-specific polysaccharide

O-SPC

low molecular mass O-SP-core fragments

GlcN

glucosamine

GlcNAc

N-acetylglucosamine

Hep

L-glycero-D-manno-heptose

Kdo

3-deoxy-D-manno-octulosonic acid

RU

repeat unit of O-SP

ES-MS

electrospray mass spectrometry

GLC-MS

gas liquid chromatography mass spectrometry

GM

geometric mean

SBAP

N-succinimidoyl 3-(bromoacetamido) propionate

TFA

trifluoroacetic acid

PBS

phosphate buffer saline, pH 7.4

s.c.

subcutaneously

EU

Elisa units

GM

geometric mean