First Evidence for a Covalent Linkage between Enterobacterial Common Antigen and Lipopolysaccharide in Shigella sonnei Phase II ECA_LPS

Background: Enterobacterial common antigen (ECA) is a surface antigen of all enteric bacteria.

Results: ECA polysaccharide substitutes the outer core region of Shigella sonnei lipopolysaccharide (LPS).

Conclusion: First structural evidence for the existence of ECA covalently associated with LPS (ECA_LPS).

Significance: ECA_LPS is the only immunogenic form of ECA and could be a target for therapeutic strategies against nosocomial infections.

Abstract

Enterobacterial common antigen (ECA) is expressed by Gram-negative bacteria belonging to Enterobacteriaceae, including emerging drug-resistant pathogens such as Escherichia coli, Klebsiella pneumoniae, and Proteus spp. Recent studies have indicated the importance of ECA for cell envelope integrity, flagellum expression, and resistance of enteric bacteria to acetic acid and bile salts. ECA, a heteropolysaccharide built from the trisaccharide repeating unit, →3)-α-d-Fucp4NAc-(1→4)-β-d-ManpNAcA-(1→4)-α-d-GlcpNAc-(1→, occurs as a cyclic form (ECA_CYC), a phosphatidylglycerol (PG)-linked form (ECA_PG), and an endotoxin/lipopolysaccharide (LPS)-associated form (ECA_LPS). Since the discovery of ECA in 1962, the structures of ECA_PG and ECA_CYC have been completely elucidated. However, no direct evidence has been presented to support a covalent linkage between ECA and LPS; only serological indications of co-association have been reported. This is paradoxical, given that ECA was first identified based on the capacity of immunogenic ECA_LPS to elicit antibodies cross-reactive with enterobacteria. Using a simple isolation protocol supported by serological tracking of ECA epitopes and NMR spectroscopy and mass spectrometry, we have succeeded in the first detection, isolation, and complete structural analysis of poly- and oligosaccharides of Shigella sonnei phase II ECA_LPS. ECA_LPS consists of the core oligosaccharide substituted with one to four repeating units of ECA at the position occupied by the O-antigen in the case of smooth S. sonnei phase I. These data represent the first structural evidence for the existence of ECA_LPS in the half-century since it was first discovered and provide insights that could prove helpful in further structural analyses and screening of ECA_LPS among Enterobacteriaceae species.

Introduction

Enterobacterial common antigen (ECA)² is a surface antigen present in Gram-negative bacteria belonging to the Enterobacteriaceae family, including Escherichia coli, Salmonella spp., Shigella spp., Klebsiella spp., Enterobacter spp., Citrobacter spp., Serratia spp., Proteus spp., Yersinia spp., and Plesiomonas shigelloides (1). These bacteria are responsible for healthcare-associated infections, such as intestinal infections and nosocomial infections (e.g. sepsis). Drug resistance in a few members of this family, notably E. coli, Klebsiella spp. and Proteus spp., is an increasing global problem. This stresses the need for new cross-protective vaccines or therapeutic strategies against Gram-negative bacteria. Some of these could be based on ECA.

ECA is a heteropolysaccharide built from the trisaccharide repeating unit, →3)-α-d-Fucp4NAc-(1→4)-β-d-ManpNAcA-(1→4)-α-d-GlcpNAc-(1→ (2–4), modified with O-acetyl groups. The biological significance of ECA is not fully understood and requires further study. However, Barua et al. (5) showed the importance of ECA together with the O-specific polysaccharide of endotoxin (lipopolysaccharide (LPS)) for the resistance of Shiga toxigenic E. coli O157:H7 and Salmonella to acetic acid and bile salts (6). In the case of Serratia marcescens, mutations that disrupt the integrity of the ECA biosynthetic pathway severely diminish bacterial motility by influencing the transcription of genes involved in envelope integrity and flagellum expression (7, 8).

ECA occurs in three forms. The first two forms, a free, cyclic polysaccharide (ECA_CYC) localized to the periplasm (2, 3) and a linear, phosphatidylglycerol (PG)-linked polysaccharide located on the cell surface (ECA_PG) (4, 9–12), are completely characterized and are nonimmunogenic. The third form is poorly understood, but is thought to be composed of ECA and LPS. LPS is the main surface antigen of Gram-negative bacteria and is a potent virulence factor. It consists of three main regions: the O-specific polysaccharide (smooth strains only), the core oligosaccharide (OS), and lipid A. To date, the nature of the association between ECA and LPS has not been elucidated and remains the last missing link to understanding ECA diversity. Additionally, ECA_LPS is the only immunogenic form of ECA capable of eliciting cross-reactive anti-ECA antibodies (13, 14).

Paradoxically, given this structural uncertainty, the history of ECA began with the discovery of ECA_LPS. The presence of ECA was originally inferred from the extent of serological cross-reactions in hemagglutination assays between the sera of patients and various E. coli O-serotypes during studies of urinary tract E. coli infections in 1962 by Kunin et al. (15). A few E. coli strains (serotype O14, O54, O124, and O144) elicited highly cross-reactive antibodies in rabbits (13, 14) that could be removed from anti-O14 serum by absorption with extracts of any strain of E. coli, while retaining the homological reactivity of the serum. Thus, it was concluded that anti-O14 serum contained antibodies against an antigen that is common for all Enterobacteriaceae. This suggested that, unlike most E. coli strains, a few, especially O14, express this antigen in an immunogenic form, the presumptive ECA_LPS, capable of eliciting cross-reactive anti-ECA antibodies (1, 13–15). Following this initial discovery of ECA, ECA_PG was identified (4, 9, 10), and its chemical structure and covalent linkage to PG were ultimately elucidated for E. coli and Salmonella typhimurium (11, 12). The identification of ECA_CYC and its structural analysis were completed even earlier for Shigella sonnei phase I (2, 3). Therefore, ECA_PG and ECA_CYC seem to be well characterized with respect to their occurrence (11), structural variability (16), and the genetic basis for their biosynthesis (17–19).

To date, no direct evidence for the existence of a covalent linkage between ECA and LPS has been reported. Since its initial discovery, ECA_LPS in nature has usually been identified indirectly using serological approaches similar to those used to establish the immunogenic properties of E. coli O14 ECA_LPS (1, 13–15). The only indication for the existence of ECA_LPS, apart from its immunogenicity, has been co-migration of LPS/LOS (lipooligosaccharide) with ECA during SDS-PAGE, as detected by specific anti-ECA antibodies/sera. In subsequent studies, Kunin et al. (9) and Mayer et al. (20) identified the presence of such an immunogenic form of ECA in rough mutants of E. coli Ra, R1, R4, and K-12 that express a complete core OS region of LOS. More recent and extensive serological studies have addressed the occurrence of ECA_LPS in Yersinia enterocolitica (21–24) and Proteus mirabilis. ECA immunogenicity has also been demonstrated for several wild-type strains and mutants of Y. enterocolitica O:3 and O:9 as well as P. mirabilis (25). The authors of these latter studies utilized polyclonal and monoclonal antibodies against different forms of ECA to screen SDS-PAGE-separated LPS/LOS isolated from a broad range of wild-type and mutant bacteria to show the coexistence of ECA and LPS epitopes. Recent serological studies of Y. enterocolitica LPS pointed to the core OS of LPS as the probable location of the ECA and its coexistence with O-specific polysaccharide (21). The only attempted structural analyses of ECA_LPS were reported for E. coli strain F470 (R1 core type) (26) and S. sonnei phase II (27). These analytical strategies utilized chromatography to separate different fractions of degraded LPS and characterize them for the presence of ECA constituents. The detection of ECA constituents was qualitative and was based only on sugar and methylation analyses and isotope labeling of ECA constituents (26) or NMR spectroscopy (27). Both studies reported the coexistence of the core OS and ECA constituents, but lacked direct evidence for a covalent linkage between them.

To resolve the ECA_LPS structure, we sought to isolate ECA_LPS and perform a detailed structural analysis using the sensitive techniques of NMR spectroscopy and mass spectrometry. For this investigation, we used S. sonnei phase II, a species that is the causative agent of dysentery. This mutant is well characterized with respect to the structure of the core OS and ECA_CYC, expressing LOS with the core OS of E. coli R1 type and devoid of O-specific polysaccharide (PS) (2, 28, 29). The data presented here, coming a half-century after the initial discovery of ECA, represent the first structural evidence for the existence of ECA_LPS and provide insights that could prove helpful in further structural analyses and screening of ECA_LPS among Enterobacteriaceae species.

EXPERIMENTAL PROCEDURES

Bacteria and Sera

S. sonnei phase II was obtained from the Polish Collection of Microorganisms at the Ludwik Hirszfeld Institute of Immunology and Experimental Therapy, Polish Academy of Sciences (Wroclaw, Poland). The bacteria were grown to logarithmic phase in liquid Davis medium enriched with glucose, yeast extract, and casein hydrolysate; killed with 0.5% phenol; and centrifuged using a CEPA laboratory flow centrifuge. Rabbit polyclonal anti-ECA_CYC and anti-E. coli R1 core OS sera were prepared as part of a previously published study (30, 31).

Preparation of LOS and Oligosaccharides

LOS was extracted from bacterial cells using a hot phenol/water method (32). Poly- and oligosaccharides were obtained by mild acid hydrolysis of LOS (1.5% acetic acid, 100 °C, 30 min). The resulting mixture was centrifuged (40,000 × g), and the supernatant was collected, lyophilized, and fractionated on a Bio-Gel P-10 column (1.6 × 100 cm; Bio-Rad) equilibrated with 0.05 m pyridine/acetic acid buffer (pH 5.6). ECA was detected by dot-blotting of eluted fractions with anti-ECA_CYC serum (30).

Dot-blotting

Fractions eluted from Bio-Gel P-10 were spotted onto a dried nitrocellulose membrane. The membrane was air-dried, and sample spotting was repeated two times. The membrane was blocked with 2% casein in TBS (10 mm Tris pH 7.4, 150 mm NaCl) for 1 h at 37 °C, and then incubated overnight with a solution of polyclonal rabbit anti-ECA_CYC serum in TBS (200-fold dilution). The membrane was washed three times with TBS prior to incubation with 2000-fold diluted goat anti-rabbit IgG conjugated with alkaline phosphatase (Bio-Rad). After washing with TBS, the membrane was stained with 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium in 0.05 m Tris-HCl buffer (pH 9.6) containing 5 mm MgCl₂. The reaction was stopped by washing with an ethanol/water/acetic acid solution (4.5:4.5:1) followed by washing with water and air drying.

Electrospray Mass Spectrometry

MS and MS² experiments were carried out on an amaZon SL ion trap (IT) mass spectrometer (Bruker Daltonics) in both positive-ion and negative-ion modes. The samples were dissolved in acetonitrile/water/formic acid solution (50:50:0.5; 50 μg/ml). Source parameters were as follows: sample flow, 3 μl/min; ion source temperature, 200 °C; nitrogen flow, 5 liters/min at a pressure of 8 psi. Spectra were scanned in the 200–2000 m/z range. The system was calibrated in positive-ion mode using ESI-L tuning mix (Agilent Technologies) before acquisitions. MS² experiments were performed using an isolation width of 4 m/z, an amplitude value of 0.35, and a SmartFrag mode of 60–80%. The fragment ions structures were determined using GlycoWorkbench software (33).

NMR Spectroscopy

All NMR spectra were obtained using an Avance III 600 MHz (Bruker BioSpin) spectrometer equipped with a QCI-P cryoprobe. NMR spectra of isolated oligosaccharides were obtained in ²H₂O using acetone as an internal reference (δ_H 2.225 ppm; δ_C 31.05 ppm). Oligosaccharides (1.8 mg) were repeatedly exchanged with ²H₂O (99.95%) with intermediate lyophilization. The data were acquired and processed with standard Bruker software (TopSpin 3.1) and assigned using SPARKY (34). The signals were assigned based on two-dimensional experiments using correlation spectroscopy (COSY), clean total correlation spectroscopy (TOCSY), nuclear Overhauser effect spectroscopy (NOESY), heteronuclear multiple-bond correlation (HMBC), heteronuclear single-quantum coherence-distortionless enhancement by polarization transfer (HSQC-DEPT), and HSQC-TOCSY. The mixing times in clean-TOCSY experiments were 30, 60, and 100 ms. The delay time in HMBC was 60 ms, and the mixing time for NOESY was 200 ms.

Compositional Analysis

Methylation analyses were performed according to the method described by Ciucanu and Kerek (35). Partially methylated alditol acetates were analyzed by gas chromatography (GC)-MS using a Thermo Scientific TSQ system with an RX5 fused silica capillary column (0.2 mm × 30 cm) and a temperature program of 150–270 °C at 12 °C min⁻¹.

RESULTS

Isolation and Purification of S. sonnei Phase II ECA_LPS

The LOS of S. sonnei phase II was extracted from bacterial cells by a hot phenol/water method and was obtained with a yield of 2.11% of dry bacterial mass. Poly- and oligosaccharides (dLOS) were released from lipid A by mild acid hydrolysis of LOS (250 mg) in 1.5% acetic acid and separated by gel filtration on Bio-Gel P-10 (Fig. 1). Five fractions were eluted and analyzed for the presence of ECA epitopes by dot-blotting using polyclonal rabbit anti-ECA_CYC and anti-E. coli R1 core OS sera. Positive reactions with anti-ECA_CYC serum were observed for fractions I (0.65 mg), II (1.2 mg), III (1.5 mg), and IV (1.8 mg) (Fig. 1), indicating the presence of ECA epitopes, whereas reactions with anti-E. coli R1 core OS serum were positive only for fractions III, IV, and V (55 mg). All fractions were analyzed by electrospray ionization-ion trap mass spectrometry (ESI-IT-MSⁿ). Electrospray ionization-ion trap mass spectrometry (ESI-IT-MSⁿ) analyses confirmed dot-blotting results and the presence of different forms of ECA. Fraction I was identified as linear polymer of ECA subunit derived from ECA_PG and devoid of PG moiety (data not shown). Fraction II was identified as ECA_CYC (data not shown) as a result of co-purification of this form together with LPS. Fractions reacted with both sera (III, IV) were analyzed by ESI-IT-MSⁿ. Analyses confirmed dot-blotting results and the presence of poly- and oligosaccharides derived from ECA_LOS built of the core OS of LOS (dLOS) substituted with repeating units of ECA ([ECA]_n-dLOS; n ≥ 1). Fraction III consisted of [ECA]_2–4-dLOS containing two to four ECA repeating units (Fig. 2A), and fraction IV consisted of [ECA]-dLOS containing one repeating unit of ECA (Fig. 2B) were identified. According to previously published data (28), fraction V consisted of different unsubstituted core OS glycoforms of E. coli R1 type. The final yield of [ECA]_n-dLOS was 3% of released poly- and oligosaccharides (fractions III and IV).

FIGURE 1.

Elution profile of poly- and oligosaccharides derived from S. sonnei phase II LOS separated by size-exclusion chromatography. The oligosaccharides released by mild acid hydrolysis were fractionated on Bio-Gel P-10 in 0.05 m pyridine/acetic acid buffer, pH 5.6, and detected using a Knauer refractometric detector. Eluted fractions were tested for reactivity with anti-ECA_CYC (A) and anti-E. coli R1 core OS (B) sera. The + marks indicate positive reaction with serum, whereas the − marks indicate that no reaction was observed. Fractions reacted with both sera were collected and analyzed by ESI-IT-MSⁿ and NMR spectroscopy.

FIGURE 2.

ESI-IT mass spectra of [ECA]_n-dLOS fractions. The spectra were obtained for acetonitrile/water/formic acid solutions (50:50:0.5; 50 μg/ml). The charges of ions are indicated above the m/z values. A, the negative-ion mode spectrum of fraction III contained R1-type core oligosaccharides substituted by two, three, or four ECA trisaccharide repeating units. Populations differed by phosphate (P) and phosphoethanolamine (PEtn) group substitutions in the inner core OS. B, the positive-ion mode spectrum of fraction IV contained core OS substituted by one ECA repeating unit. A small amount of unsubstituted core (dLOS) was also detected. C, positive-ion mode fragmentation spectrum of the triple-charged ion at m/z 839.21. The marks above the signals refer to fragment ions derived from the inset structure. The asterisk denotes uninterpreted ions.

Under standard conditions, a basic methylation analysis of [ECA]-dLOS indicated the presence of partially methylated alditol acetate of one constituent of ECA, 3-substituted-GlcpNAc, as expected based on the susceptibility of →3)-α-d-Fucp4NAc-1→ to acidic degradation and the nonvolatile character of →4)-β-d-ManpNAcA-(1→ derivative. The following derivatives of constituents of the R1 core OS, in addition to nonvolatile residues substituted with a phosphate (P) or phosphoethanolamine (PEtn) group, were also identified: terminal Glcp, terminal Galp, 3-substituted Glcp, 2-substituted Galp, 2,3-disubstituted Glcp, and terminal Hepp. In comparison with unsubstituted core OS, molar ratios of terminal Glc, terminal Gal, →3)-Glc, and →2)-Gal were 0.4:1:1.4:1.2, showing decreasing amounts of terminal Glc and increasing amounts of →3)-Glc. These data provide an initial indication of the position of substitution of the core OS by ECA at →3)-Glc that would prove to be in agreement with subsequent studies.

Mass Spectrometry Analysis of ECA_LOS

A detailed analysis of the molecular mass of [ECA]_n-dLOS fractions indicated the presence of a covalent linkage between ECA and different glycoforms of the core OS (Fig. 2, A and B). Glycoforms substituted with two, three, and four repeating units of ECA were identified in fraction III (Fig. 2A) and were represented by prevailing triple-charged ions, [M-3H]³⁻, at m/z 1039.63 ([ECA]₂-dLOS), m/z 1242.01 ([ECA]₃-dLOS), and m/z 1444.39 ([ECA]₄-dLOS). A detailed summary of the interpretations of observed triple-charged ions is presented in Table 1. The observed heterogeneity of fraction III was related to the presence of P and PEtn, identified previously in the core OS of E. coli R1 type (28). Moreover, ions indicating nonstoichiometric O-acetylation, m/z 1458.39 and 1499.05 corresponding to glycoforms constituting of at least four repeating units of ECA, were also observed (Table 1).

TABLE 1.

Interpretation of ESI-IT mass spectra of [ECA]_2–4-dLOS (fraction III) isolated from S. sonnei phase II LOS

[ECA], trisaccharide repeating unit: →3)-α-d-Fucp4NAc-(1→4)-β-d-ManpNAcA-(1→4)-α-d-GlcpNAc (1→; OAc, O-acetyl group.

Oligosaccharide structure	Calculated mass (monoisotopic) (Da)	Observed ion (m/z)	Interpretation of the ion	Calculated mass of ion (monoisotopic) (m/z)
[ECA]2-Glc-Gal2-Glc2-P3-Etn-Hep3-Kdo	3121.91	1039.63	[M-3H]3−	1039.63
		1046.96	[M-3H,Na]3−	1046.96
[ECA]2-Glc-Gal2-Glc2-P4-Etn-Hep3-Kdo	3201.88	1066.28	[M-3H]3−	1066.28
[ECA]2-Glc-Gal2-Glc2-P4-Etn2-Hep3-Kdo	3244.92	1080.62	[M-3H]3−	1080.63
[ECA]2-Glc-Gal2-Glc2-P5-Etn2-Hep3-Kdo	3324.88	1107.28	[M-3H]3−	1107.29
[ECA]3-Glc-Gal2-Glc2-P2-Hep3-Kdo	3606.12	1201.01	[M-3H]3−	1201.03
[ECA]3-Glc-Gal2-Glc2-P3-Hep3-Kdo	3686.09	1227.67	[M-3H]3−	1227.69
[ECA]3-Glc-Gal2-Glc2-P3-Etn-Hep3-Kdo	3729.13	1242.01	[M-3H]3−	1242.04
[ECA]3-Glc-Gal2-Glc2-P3-Etn-Hep3-Kdo		1249.34	[M-3H,Na]3−	1249.36
[ECA]3-Glc-Gal2-Glc2-P4-Etn-Hep3-Kdo	3809.10	1268.66	[M-3H]3−	1268.69
[ECA]3-Glc-Gal2-Glc2-P4-Etn2-Hep3-Kdo	3852.14	1283.00	[M-3H]3−	1283.04
[ECA]3-Glc-Gal2-Glc2-P5-Etn2-Hep3-Kdo	3932.11	1309.66	[M-3H]3−	1309.69
[ECA]4-Glc-Gal2-Glc2-P2-Hep3-Kdo	4213.35	1403.39	[M-3H]3−	1403.44
[ECA]4-Glc-Gal2-Glc2-P3-Hep3-Kdo	4293.31	1430.05	[M-3H]3−	1430.10
[ECA]4-Glc-Gal2-Glc2-P3-Etn-Hep3-Kdo	4336.35	1444.39	[M-3H]3−	1444.44
[ECA]4-OAc-Glc-Gal2-Glc2-P3-Etn-Hep3-Kdo	4378.36	1458.39	[M-3H]3−	1458.45
[ECA]4-Glc-Gal2-Glc2-P4-Etn-Hep3-Kdo	4416.32	1471.04	[M-3H]3−	1471.10
[ECA]4-Glc-Gal2-Glc2-P4-Etn2-Hep3-Kdo	4459.36	1485.35	[M-3H]3−	1485.45
[ECA]4-OAc-Glc-Gal2-Glc2-P4-Etn2-Hep3-Kdo	4501.37	1499.05	[M-3H]3−	1499.45

The presence of a linkage between ECA and LOS was further confirmed by a positive-ion mode ESI-MS² analysis of the low molecular mass fraction of [ECA]-dLOS (fraction IV) containing only one repeating unit of ECA linked to the core OS (Fig. 2B). This linkage is represented by the prevailing triple-charged ion, [M+3H]³⁺, at m/z 839.21 attributable to a dodecasaccharide substituted with P and PPEtn (Fig. 2C, inset structure). The remaining ions reflected heterogeneity previously observed for [ECA]_2–4-dLOS and were related to P and PEtn nonstoichiometric substituents in the core OS region and the presence of sodium and potassium adducts. In addition, a small amount of unsubstituted core oligosaccharides (dLOS) was detected. An ESI-MS² fragmentation analysis of the ion at m/z 839.21 ([M+3H]³⁺; Fig. 2C) showed a broad pattern of double- and single-charged fragment ions of Y or B type (Fig. 2C, inset structure with graphical representation of fragmentation), according to the nomenclature of Domon and Costello (36). The ion at m/z 608.26 corresponded to singly protonated ECA trisaccharide (B_3α′), whereas B_4α′ ions at m/z 385.67 ([M+2H]²⁺) and m/z 770.30 ([M+H]¹⁺) matched the ECA repeating unit bound to an additional hexose, suggesting that the ECA chain is linked to terminal hexose of the outer core OS.

Identification of the Linkage between ECA and LOS by NMR Spectroscopy

Fraction IV containing [ECA]-dLOS was further investigated using one-dimensional and two-dimensional ¹H,¹³C,³²P NMR spectroscopy. The ¹H (Fig. 3), ¹H,¹³C HSQC-DEPT, and HMBC NMR spectra (Fig. 4) contained mainly signals for 11 anomeric protons and carbons and the Kdo spin system confirming the dodecasaccharide structure of [ECA]-dLOS (Fig. 3); signals for three acetamide groups, the carboxylic group of ManpNAcA and the methyl group of FucpNAc, characteristic ECA residues, were also identified. Capital letters shown in the structure are used throughout the text and in tables (Tables 2 and 3) and figures (Figs. 3, 4, and 5) to refer to the corresponding sugars. Because the ECA repeating unit and R1 core OS structures were previously described, the chemical shift values were compared with published NMR data (37, 38).

FIGURE 3.

¹H NMR spectrum and structure (top) of [ECA]-dLOS isolated from S. sonnei phase II LOS. The ¹H NMR spectrum was obtained for a ²H₂O solution at 600 MHz (30 °C). Residual water was removed by processing. The capital letters refer to carbohydrate residues of [ECA]-dLOS, as shown in the inset structure. The Arabic numerals refer to protons in respective residues. Letters with a prime sign denote residues of trace amounts of the core OS devoid of ECA. The ECA repeating unit is framed with a dashed box.

FIGURE 4.

Selected portions of ¹H,¹³C HSQC-DEPT and HMBC spectra of [ECA]-dLOS dodecasaccharide isolated from S. sonnei phase II LOS. The spectra were obtained for a ²H₂O solution at 600 MHz (30 °C). The capital letters refer to carbohydrate residues, as shown in Fig. 3 and Tables 2 and 3. The numbers refer to protons and carbons in the respective residues. Letters with a prime sign denote residues of trace amounts of the core OS devoid of the ECA. Signals indicating the linkage between ECA and dLOS are shown in bold.

TABLE 2.

¹H and ¹³C NMR chemical shifts of fraction IV containing core OS substituted by ECA trisaccharide ([ECA]-dLOS) isolated from S. sonnei phase II LOS

Residue	Description	Chemical shift (ppm)
		H-1/C-1	H-2/C-2	H-3(H3ax,eq)/C-3	H-4/C-4	H-5/C-5	H-6a, H-6b/C-6	H-7a, H-7b/C-7 (NHAc)	H-8a, H-8b/C-8 [C(O)]
A	→5)-α-Kdop	NDa	−/96.3	(1.90, 2.25)/34.1	4.11/66.3	4.17/73.3	3.69/69.7	3.80/72.6	3.47, 3.93/64.7
B	→3)-l-α-d-Hepp4PPEtn-(1→	5.20/100.1	4.01/71.6	4.08/78.5	4.61/72.3	4.22/72.0	4.10/69.3	3.72, 3.72/63.8
C	→3,7)-l-α-d-Hepp4P-(1→	5.10/103.5	4.38/70.6	4.12/79.8	4.40/69.4	3.80/73.2	4.23/68.5	3.58, 3.75/68.4
D	l-α-d-Hepp-(1→	4.98/100.2	3.93/70.7	3.87/71.4	3.84/66.9	3.61/71.9	4.04/69.5	3.65, 3.72/63.7
E	→3)-α-d-Glcp-(1→	5.20/102.0	3.66/71.0	4.07/76.7	3.77/71.2	3.91/73.1	3.79,3.92/60.5
F	→2,3)-α-d-Glcp-(1→	5.80/95.3	3.87/73.3	4.17/78.7	3.56/68.7	4.10/71.9	3.78,3.95/61.0
F′b	→2,3)-α-d-Glcp-(1→	5.81/95.1	3.88/73.3	4.19/78.8	3.57/68.7	4.11/72.0	3.79,3.96/61.0
G	→2)-α-d-Galp-(1→	5.61/92.1	3.98/73.2	4.19/68.9	3.98/70.7	4.13/72.0	3.74,3.74/61.9
H	α-d-Galp-(1→	5.31/96.6	3.85/69.0	3.95/70.1	3.99/70.1	4.13/72.0	3.75,3.75/61.9
I	→3)-β-d-Glcp-(1→	4.73/103.1	3.39/73.6	3.68/85.4	3.49/68.9	3.44/76.3	3.72,3.89/61.4
I′c	β-d-Glcp-(1→	4.75/103.1	3.33/73.9	3.51/76.6	3.40/70.4	3.45/76.6	3.73,3.91/61.4
J	→4)-α-d-GlcpNAc-(1→	4.78/102.3	3.75/56.3	3.74/72.7	3.68/79.5	3.54/75.2	3.86,3.70/60.9	(2.03/23.0)	[175.5]
K	→4)-β-d-ManpNAcA-(1→	4.93/99.7	4.49/54.2	4.07/73.2	3.82/74.8	3.86/77.2	−/175.1	(2.07/22.6)	[176.2]
L	α-d-Fucp4NAc-(1→	5.35/99.5	3.64/69.3	4.00/69.1	4.20/54.6	4.18/66.5	1.06/16.2	(2.07/22.6)	[176.3]
PPEtn		4.20/63.1	3.29/40.7

^a ND, not determined.

^b Residue F′ is a variant of residue F present in the core OS that is devoid of ECA trisaccharide.

^c Residue I′ is a terminal residue I present in the core OS that is devoid of ECA trisaccharide.

TABLE 3.

Selected inter-residue NOE and ³J_H,C connectivities from the anomeric atoms of [ECA]-dLOS dodecasaccharide isolated from S. sonnei phase II LOS

Data indicating the covalent linkage between ECA and LOS are shown in bold.

Residue	Description	Atom δH/δC	Connectivity to		Inter-residue atom/residue
			δC	δH
		ppm
B	→3)-l-α-d-Hepp4PPEtn-(1→	5.20/100.1	–	4.17a	H-5 of A
C	→3,7)-l-α-d-Hepp4P-(1→	5.10/103.5	78.5	4.08a	C-3, H-3 of B
D	l-α-d-Hepp-(1→	4.98/100.2	68.5	3.59/3.74a	C-7, H-7a, H-7b of C
E	→3)-α-d-Glcp-(1→	5.20/102.0		4.12	H-3 of C
F	→2,3)-α-d-Glcp-(1→	5.80/95.3		4.07	H-3 of E
F′	→2,3)-α-d-Glcp-(1→b	5.81/95.1		4.07	H-3 of E
G	→2)-α-d-Galp-(1→	5.61/92.1		3.87a	H-2 of F
H	α-d-Galp-(1→	5.31/96.6		3.97a	H-2 of G
I	→3)-β-d-Glcp-(1→	4.73/103.1	78.7	4.17	H-3 of F
I′	β-d-Glcp-(1→c	4.75/103.1	78.8	4.19	H-3 of F′
J	→4)-α-d-GlcpNAc-(1→	4.78/102.3	85.3	3.68	H-3 of I
K	→4)-β-d-ManpNAcA-(1→	4.93/99.7	79.4	3.69a	H-4 of J
L	α-d-Fucp4NAc-(1→	5.35/99.5	74.7	3.81	H-4 of K

^a Value represents NOE connectivities only.

^b Residue F′ is a variant of residue F present in the core OS that is devoid of ECA trisaccharide.

^c Residue I′ is a terminal residue I present in the core OS that is devoid of ECA trisaccharide.

FIGURE 5.

A selected portion of the NOESY spectrum containing the anomeric protons region of [ECA]-dLOS dodecasaccharide isolated from S. sonnei phase II LOS. The cross-peaks are labeled as explained in the legend for Fig. 2. The signal indicating the linkage between ECA and dLOS is shown in bold.

Residues A–I were identified as constituents of the core OS, and residues J–L were identified as constituents of ECA (Fig. 3, inset structure). Residue A, with characteristic deoxy proton signals at δ 1.90 ppm (H-3ax) and 2.25 ppm (H-3eq) together with a high chemical shift of the C-5 signal (δ 73.3 ppm), was identified as 5-substituted 3-deoxy-α-d-manno-oct-2-ulosonic acid [→5)-α-Kdop]. Residue B, with a J_H1,C1 of ∼174 Hz, small vicinal couplings between H-1, H-2, and H-3, and a high C-3 signal (δ 78.5 ppm), was identified as 3-substituted l-glycero-d-manno-Hepp4PPEtn. ¹H,³¹P HMBC experiments revealed connectivity between the pyrophosphate monoester signal (δ_P 11.6 ppm) and H-4 (δ 4.61 ppm) of residue B, suggesting a PPEtn substitution. Residue C, with a J_H1,C1 of ∼171 Hz, was assigned as 3,7-disubstituted l-glycero-d-manno-Hepp4P based on chemical shifts reported previously (38), small vicinal couplings of H-1 and H-2, and relatively high chemical shifts of C-3 (δ 79.8 ppm) and C-7 (δ 68.4 ppm) signals. ¹H,³¹P HMBC spectra showed connectivity between P (δ_P 0.8 ppm) and H-4 (δ 4.40 ppm) of residue C. Residue D, with a J_H1,C1 of ∼172 Hz, was assigned as terminal l-glycero-d-manno-Hepp based on ¹H,¹³C chemical shift values and small vicinal couplings between H-1, H-2, and the C-6 signal at δ 69.5, as previously shown for the monosaccharide l-α-d-Hepp (38). Residue E was recognized as 3-substituted α-d-Glcp based on the relatively large chemical shift of C-3 signal (δ 76.7 ppm) and chemical shifts similar to those reported previously (38). Residue F, with a J_H1,C1 of ∼175 Hz, was assigned as 2,3-disubstituted α-d-Glcp based on relatively high chemical shifts of C-2 (δ 73.3 ppm) and C-3 (δ 78.7 ppm) signals. Residue F′ was assigned as a residue F variant present in unsubstituted core OS (Table 2). Residue G, with a J_H1,C1 of ∼172 Hz, was identified as 2-substituted α-d-Galp based on the relatively large chemical shift in C-2 (δ 73.2 ppm). Residue H, with a J_H1,C1 of ∼178 Hz, was assigned as terminal α-d-Galp based on the large couplings between H-1, H-2, and H-3; small vicinal couplings among H-3, H-4, and H-5; and chemical shifts similar to those reported previously (38). Residue I was recognized as 3-substituted β-d-Glcp based on ¹H,¹³C chemical shifts and the very high chemical shift of its C-3 signal (δ 85.4 ppm) as compared with that of residue I′ (δ 76.6 ppm), identified as terminal β-d-Glcp, which was present as a unsubstituted variant of residue I in the core OS devoid of ECA. This latter finding affirms results obtained by ESI-MS² analysis (above), which detected a small amount of unsubstituted core oligosaccharides (dLOS). Residue J was assigned as 4-substituted α-d-GlcpNAc based on its characteristic C-2 signal (δ 56.3 ppm) and the downfield chemical shift of its C-4 signal (δ 79.5 ppm). Residue K was assigned as 4-substituted β-d-ManpNAcA based on its characteristic C-2 (δ 54.2 ppm) and C-6 (δ 175.1 ppm) signals and the relatively high chemical shift of C-4 (δ 74.8 ppm). Residue L was recognized as terminal α-d-Fucp4NAc based on the characteristic exocyclic methyl group signals (δ 1.06/16.2 ppm) and chemical shift of the C-4 signal (δ 54.6 ppm).

Each disaccharide element was identified by HMBC (Table 3, Fig. 4) and NOESY experiments (Table 3, Fig. 5), providing a complete sequence of dodecasaccharide [ECA]-dLOS (Fig. 3, inset structure). The HMBC spectra showed cross-peak signals between the anomeric proton and the carbon at the linkage position and between the anomeric carbon and proton at the linkage position (Fig. 4). Thus, the combined results suggest a dodecasaccharide structure of S. sonnei phase II [ECA]-dLOS containing one repeating unit of ECA linked to the LOS via an α(1→3) linkage between →4)-α-d-GlcpNAc-(1→ of ECA (residue J) and →3)-β-d-Glcp (residue I) of the outer core OS (Fig. 3, inset structure). Additionally, biological repeating units of ECA with →4)-α-d-GlcpNAc-(1→, as a first residue, and glycoforms with two, three, and four repeating units together with some O-acetyl group modifications, were also identified.

DISCUSSION

Lipopolysaccharides, the main surface antigens of Gram-negative bacteria, are virulence factors belonging to pathogen-associated molecular patterns recognized by the innate immune system that are involved in the development of sepsis and septic shock. The structures of lipopolysaccharides are able to undergo modifications to evade bactericidal factors of innate and adaptive immunity pathways, yet retain some common epitopes. Among these common epitopes are inner core motifs built of Hep and Kdo, which are considered to be potential candidates for therapeutic strategies against infections caused by enteric bacteria (39). In the case of Enterobacteriaceae, it is thought that a small part of LPS is decorated with ECA, a population referred to as ECA_LPS. ECA is shared among all species of Enterobacteriaceae and may be considered a potential target for the design of cross-reactive and bactericidal antibodies. Because of its coexistence with LPS, ECA_LPS is the only form of ECA capable of eliciting anti-ECA antibodies upon immunization of animals with ECA_LPS-positive strains. Of all the forms of ECA, ECA_LPS was the last to await definitive evidence for its presence.

We present herein the first structural evidence for the existence of ECA_LPS through a structural analysis of S. sonnei phase II LOS preparations. More precisely, we have identified the covalent linkage between ECA and LPS. S. sonnei phase II expressing the rough form of endotoxin, LOS with the core OS of E. coli R1-type and devoid of O-specific PS. We have found that this ECA_LPS is a molecule consisting of the LOS substituted with one to four repeating units of ECA in the outer core region. It coexists as ∼3% of the total amount of poly- and oligosaccharides released after mild acid hydrolysis of S. sonnei phase II LOS, indicating that unsubstituted core oligosaccharides prevail on the bacterial surface. In identifying the covalent linkage between →4)-α-d-GlcpNAc-(1→ of ECA and →3)-β-d-Glcp of the outer core OS, we have also revealed a biological repeating unit of ECA_LPS identical to that present in ECA_PG, an expected outcome given the biosynthesis of ECA (12, 17–19, 40). The number of repeating units was detected using ESI-IT-MS analysis. Because this MS technique has a limited m/z detection range, we attempted to search for longer polymers of ECA_LPS using MALDI-TOF MS. Unfortunately, this technique did not produce good quality spectra, even for the shortest glycoform, [ECA]-dLOS. However, the possibility that longer polymers of ECA_LPS are present cannot be excluded. Among the modifications reported previously for ECA_CYC, including O-acetylation and a nonstoichiometric lack of N-acetyl groups in the case of →4)-α-d-GlcpNAc-(1→ residue (41), only O-acetylation was observed for S. sonnei ECA_LPS. Moreover, one O-acetyl group was identified only in fraction III, consisting of the core OS substituted with four repeating units of ECA.

Notably, in the structure presented here, ECA occupies the position used to be substituted with the O-specific PS, →4)-α-l-AltpNAcA-(1→3)-β-d-FucpNAc4N-(1→ (Alt, altrose) in the case of smooth S. sonnei phase I (28). This finding may exclude the simultaneous presence of the O-antigen and ECA within the same glycoform of the core OS. However, the coexistence of core OS glycoforms with ECA or O-specific polysaccharides on the surface of smooth Enterobacteriaceae is possible because most bacteria synthesize significant amounts of unsubstituted core OS.

Our results are in agreement with indirect observations of the polymeric character of ECA_LPS made for Y. enterocolitica LPS/LOS using serological methods and ECA_PG- and ECA_LPS-specific antibodies. This species has been extensively investigated in studies designed to identify ECA_LPS (21–24). In these studies, researchers used two types of antibodies to track different forms of ECA: (i) a monoclonal anti-ECA antibody specific for ECA_PG and (ii) a monovalent anti-ECA_LPS antiserum produced by absorption of serum obtained against ECA-positive strain with ECA-negative bacteria (specific for ECA_LPS). Cross-reactivity of absorbed anti-ECA sera provided preliminary evidence for the presence of low molecular mass forms of ECA (ECA_LPS) in Y. enterocolitica O:3 ECA_LPS. This finding was supported by more recent, extensive serological studies on Y. enterocolitica smooth and rough mutants (21). In this latter study, the authors used high-resolution SDS-PAGE to separate LPS and LOS preparations for subsequent immunoblotting with the above described detection antibodies and concluded that both low molecular mass (one to three repeating units) and high molecular mass forms of ECA_LPS were present. However, further structural analyses are necessary to confirm serological results obtained for Y. enterocolitica.

It should be emphasized that there are obvious inherent difficulties in the structural analysis of ECA_LPS, including the very low amounts of such glycoforms and their contamination with other forms of ECA (i.e. ECA_CYC and ECA_PG). The studies presented herein provide important insights that could prove helpful in further structural analyses and screening of ECA_LPS among Enterobacteriaceae species.

Moreover, we are going to use the delipidated form of this antigen ([ECA]_n-dLOS) to prepare a neoglycoconjugate with tetanus toxoid to elicit antienterobacterial antibodies broadly cross-reactive with all forms of ECA and core OS epitopes. We hope to find them to be a protective and bactericidal solution against nosocomial infections of Gram-negative bacteria such as Klebsiella, E. coli, Enterobacter, Proteus spp., or Serratia spp. Because a small part of the identified antigens were O-acetylated, we are going to remove all O-acetyl groups from the [ECA]_n-dLOS preparation to produce a glycoconjugate vaccine that elicits the production of antibodies aimed at epitopes present in enterobacterial isolates with and without O-acetyl groups in their ECA, thus generating a wider ranging response and higher serum bactericidal antibody titers. It is known that O-acetyl groups influence antigenicity and immunogenicity of poly- and oligosaccharide antigens (30, 42–46).