Lipopolysaccharide-Linked Enterobacterial Common Antigen (ECALPS) Occurs in Rough Strains of Escherichia coli R1, R2, and R4

Enterobacterial common antigen (ECA) is a conserved surface antigen characteristic for Enterobacteriaceae. It is consisting of trisaccharide repeating unit, →3)-α-d-Fucp4NAc-(1→4)-β-d-ManpNAcA-(1→4)-α-d-GlcpNAc-(1→, where prevailing forms include ECA linked to phosphatidylglycerol (ECAPG) and cyclic ECA (ECACYC). Lipopolysaccharide (LPS)-associated form (ECALPS) has been proved to date only for rough Shigella sonnei phase II. Depending on the structure organization, ECA constitutes surface antigen (ECAPG and ECALPS) or maintains the outer membrane permeability barrier (ECACYC). The existence of LPS was hypothesized in the 1960–80s on the basis of serological observations. Only a few Escherichia coli strains (i.e., R1, R2, R3, R4, and K-12) have led to the generation of anti-ECA antibodies upon immunization, excluding ECAPG as an immunogen and conjecturing ECALPS as the only immunogenic form. Here, we presented a structural survey of ECALPS in E. coli R1, R2, R3, and R4 to correlate previous serological observations with the presence of ECALPS. The low yields of ECALPS were identified in the R1, R2, and R4 strains, where ECA occupied outer core residues of LPS that used to be substituted by O-specific polysaccharide in the case of smooth LPS. Previously published observations and hypotheses regarding the immunogenicity and biosynthesis of ECALPS were discussed and correlated with presented herein structural data.


Introduction
Enterobacterial common antigen (ECA) is a common surface antigen present in Gram-negative bacteria belonging to the Enterobacteriaceae family [1]. The ECA, a heteropolysaccharide built of the trisaccharide repeating unit, →3)-α-d-Fucp4NAc-(1→4)-β-d-ManpNAcA-(1→4)-α-d-GlcpNAc-(1→ [2], occurs as a cyclic form (ECA CYC ), a phosphatidylglycerol (PG)-linked form (ECA PG ), and lipopolysaccharide (LPS)-associated form (ECA LPS ). LPS is the main surface antigen of Gram-negative bacteria that typically comprises of three structural components: lipid A, core oligosaccharide (OS), and the O-specific polysaccharide (O-PS; O-antigen determining O serotype) [3]. ECA PG represents a major form of ECA and, together with LPS, is located on the cell surface, contributing to antigenicity, outer membrane integrity, and permeability, and finally, to viability and virulence of bacteria. Contrary to ECA PG and ECA LPS presented on the cell surface, ECA CYC is located in the periplasm and has been recently pointed out as an important factor maintaining the outer membrane permeability barrier [4]. coli R1, R2, R3, and R4 lipooligosaccharides (LOS). The ECA stands for enterobacterial common antigen. The sugar residue of the ligation between core OS and ECA is colored in red. The ECA repeating unit is colored in green. R stands for nonstoichiometric substituents, such as a phosphate group (P), a pyrophosphate group (PP), pyrophosphorylethanolamine (PPEtn). The symbol * indicates heterogeneity of the core OS regarding terminal sugar residues. Symbol ** indicates ECA ligation sites assumed by the mass spectrometry only. Symbol *** indicates an inverted anomeric configuration of the D-GlcpNAc in the first ECA unit linked to the core OS, whereas an αconfiguration is characteristic for the polymeric chain. The R1, R2, R3, and R4 core OS structures are presented according to published data [9][10][11]. Kunin et al. predicted the existence of ECALPS by broad cross-reactivity of the anti-E. coli O14 rabbit serum with various rough and smooth E. coli strains. It was further demonstrated that only a few E. coli strains (serotype O14, O54, O124, and O144) elicited highly cross-reactive anti-ECA antibodies upon immunization of rabbits [6,12]. E. coli O14 was proven later on to be rough and synthesize LOS characterized by the R4 core chemotype [13,14]. Further studies demonstrated that ECALPS-dependent immunogenicity generally was limited only to a few rough enterobacteria with Figure 1. Structures of the core oligosaccharides (OS) and [ECA]-core OS in S. sonnei phase II and E. coli R1, R2, R3, and R4 lipooligosaccharides (LOS). The ECA stands for enterobacterial common antigen. The sugar residue of the ligation between core OS and ECA is colored in red. The ECA repeating unit is colored in green. R stands for nonstoichiometric substituents, such as a phosphate group (P), a pyrophosphate group (PP), pyrophosphorylethanolamine (PPEtn). The symbol * indicates heterogeneity of the core OS regarding terminal sugar residues. Symbol ** indicates ECA ligation sites assumed by the mass spectrometry only. Symbol *** indicates an inverted anomeric configuration of the D-GlcpNAc in the first ECA unit linked to the core OS, whereas an α-configuration is characteristic for the polymeric chain. The R1, R2, R3, and R4 core OS structures are presented according to published data [9][10][11]. Kunin et al. predicted the existence of ECA LPS by broad cross-reactivity of the anti-E. coli O14 rabbit serum with various rough and smooth E. coli strains. It was further demonstrated that only a few E. coli strains (serotype O14, O54, O124, and O144) elicited highly cross-reactive anti-ECA antibodies upon immunization of rabbits [6,12]. E. coli O14 was proven later on to be rough and synthesize LOS characterized by the R4 core chemotype [13,14]. Further studies demonstrated that ECA LPS -dependent immunogenicity generally was limited only to a few rough enterobacteria with complete core OS, such as, for example, E. coli R1, R4, and K-12 [15], E. coli R2 and R3 [15][16][17], Proteus mirabilis [18], S. sonnei phase II [19], and Yersinia enterocolitica O:3 and O:9 [20][21][22][23].
Nowadays, LPS (including O-PS) and ECA biosynthesis pathways are, to a great extent, well-described and share some similarities. Lipid A-core OS part is assembled on the cytoplasmic side of the inner membrane and translocated across the inner membrane. The O-PS is synthesized in a separate pathway, whereas both O-PS and ECA polysaccharides are produced via the Wzx/Wzy-dependent assembly pathway [3,[24][25][26]. Completed O-PS is finally ligated to the lipid A-core OS by a WaaL ligase to form a mature LPS molecule ready for transport to the outer membrane [3]. Since repeating units of both polymers are assembled on the same lipid carrier-undecaprenyl pyrophosphate (Und-PP)-and undergone similar processing [25], key features of ECA LPS structures are predicted partially on the basis of biosynthesis pathway analyses: (i) ECA LPS can only be observed in strains incapable of producing the O-PS [3,25]; (ii) ECA is ligated to the core OS in the position used to be occupied by O-PS, and (iii) an inverted anomeric configuration of the D-GlcpNAc residue in the first ECA repeating unit linked to the core OS has to be observed, whereas an α-configuration is a characteristic for polymeric chain (Figure 1) [3]. All these presumptions were positively verified by our single case study on S. sonnei phase II ECA LPS [7,8].
Successful identification of ECA LPS in S. sonnei phase II prompted us to further survey for ECA LPS in rough mutants of prototype E. coli R1, R2, R3, and R4 strains to correlate serological observations that were made before concerning the presence or absence of ECA LPS . E. coli O39 strain (PCM 209) characterized as the R1 core OS chemotype was selected for the purpose of seeking ECA LPS presence in a smooth bacterium. ECA LPS was identified in E. coli R1, R2, and R4, where it occupied the outer core region in the position that used to be substituted by O-PS in smooth LPS. No ECA LPS -derived fragments were found in E. coli R3. Several new examples of ECA LPS presented herein were the preconditions for the further evaluation of biosynthesis and immunogenicity of this form of ECA. Finally, presented herein structural findings regarding ECA LPS were discussed in relation to its origin (biosynthesis) and immunogenicity (serology).

Isolation and
Purification of E. coli R1, R2, R3, R4, and O39 ECA LPS -Derived Poly-and Oligosaccharides Poly-and oligosaccharides of ECA LPS were searched for among products of E. coli LOS hydrolysis. Lipooligosaccharides (LOS) were extracted from bacterial cells by a hot phenol/water method. Purified LOS preparations were delipidated by mild acid hydrolysis and separated on Bio-Gel P-10. Alternative methods of poly-and oligosaccharide fractionation (gel chromatography; See Section 4.2) did not give better separation towards the identification of fragments of ECA LPS than Bio-Gel P-10 (data not shown). For all selected strains, including E. coli PCM 209 strain, Bio-Gel P-10 elution profiles were similar with profiles previously described for S. sonnei phase II [7,8], yielding from one to seven pooled poly-and oligosaccharide fractions and suggesting a rough character of all LOS preparations, as was shown for E. coli R1 (Figure 2a, inset Bio-Gel P-10 profile).  L,D-manno-Heptose; Kdo; NAc, N-acetyl group [27]. The interpretation of ions is shown in Table 1. The most informative ions are colored in red. The mark # stands for non-interpreted ions.   L,D-manno-Heptose; Kdo; NAc, N-acetyl group [27]. The interpretation of ions is in Table 1. The most informative ions are colored in red. The mark # stands for non-interpreted L,D-manno-Heptose; Kdo; NAc, N-acetyl group [27]. The interpretation of ions is Table 1. The most informative ions are colored in red. The mark # stands for non-interpreted  Kdo; NAc, N-acetyl group [27]. The interpretation of ions is ble 1. The most informative ions are colored in red. The mark # stands for non-interpreted Kdo; NAc, N-acetyl group [27]. The interpretation of ions is shown in Table 1. The most informative ions are colored in red. The mark # stands for non-interpreted ions.
Contrary to O39 serotype prediction for PCM 209 strain, the strain also appeared to be rough and was characterized by LPS core OS of the R1 chemotype. All fractions were analyzed by electrospray ionization-ion trap (ESI-IT) or matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry (MS). The aim of MS analyses was the identification of core OS fractions substituted by at least one ECA repeating unit ([ECA] n -core). MS spectra interpretations were based on the previously published structures of ECA [2,28] and core OS of chemotypes R1 and R2 [9,10], R3 and R4 [10,11] (Figure 1).
The fraction 4 consisted of [ECA] 2-3 -core OS and was also defined by the heterogeneity of substitution with P and PEtn ( Figure 2b, Table 1). Fraction 5 consisted mainly of [ECA]-core OS with a trace amount of [ECA] 2 -core. Two glycoforms were observed regarding the core OS region differing by the presence or the lack of the terminal GlcN residue in the inner core region (Figure 2c, Table 1, Figure 1). Interpretation of ions observed for fraction 6 was in agreement with the known structure of the free core OS of E. coli R1 (data not shown) [9,10] Table 1), was selected for ESI-MS 2 . The MS 2 fragmentation showed a pattern of single, triple, and double-charged fragment ions of Y i , B i , and Z i type, confirming the elucidated sequence of sugar residues, according to the nomenclature of Domon and Costello [29]. The profile of fragment ions matched the profile that was reported for the identical structure published previously for S. sonnei phase II [7,8]. Briefly, the ion at m/z 608.30 (1+) corresponded to singly protonated ECA trisaccharide (B3α'), whereas B4α' ions at m/z 385.66 (2+) and m/z 770.35 (1+) matched the ECA repeating unit bound to an additional hexose, suggesting that the ECA chain was linked to terminal hexose of the outer core OS, what was additionally supported by other fragment ions.
Moreover, interpreted NMR spectra (Table 2) indicated the identity of E. coli R1 [ECA]-core glycoform with the structure isolated from S. sonnei phase II characterized by R1 core chemotype ( Figure 3c) [7,8]. Thus, the combined results also proved that ECA was linked to the LOS via the β(1→3) linkage between →4)-β-d-GlcpNAc-(1→ of ECA (residue J) and →3)-β-d-Glcp (residue I) of the outer core OS. Similarly to [ECA]-core OS of S. sonnei phase II, an inverted anomeric configuration of the D-GlcpNAc residue in the first ECA repeating unit linked to the core OS was observed, whereas an α-configuration was characteristic for the subsequent ECA repeating units (Figures 1 and 3, inset structure). To track the presence of ECA LPS in the strain E. coli PCM 209 serotyped as O39 and characterized by R1 core OS, we performed identical preparation and analytical protocol, as for rough strains. Contrary to O-serotype designation suggesting smooth morphology, the strain turned out to be rough. The elution profile of Bio-Gel P-10 separation (data not shown) was identical to E. coli R1 and S. sonnei phase II [7,8]. The fraction 5 was analyzed by ESI-MS n (data not shown), where similar to E. coli R1 ion profile was observed with the ions attributed to [ECA]-core OS glycoforms. The presence of [ECA]-core in E. coli PCM 209 was further verified by 1D 1 H (Figure 3b) and 2D NMR spectra (data not shown). The close similarity of both ECA-core glycoforms was demonstrated herein only by simply comparing the 1 H NMR spectra of fractions 5 obtained for E. coli PCM 209 (R1 core chemotype), R1, and S. sonnei phase II (Figure 3a-c).
Furthermore, the identical analytical protocol was used to identify and analyze E. coli R2 [ECA]-core OS glycoforms. It was demonstrated that E. coli R2 also synthetized ECA LPS . The ESI-MS profiles for isolated fractions revealed the presence of various glycoforms of R2 core OS [9,10] substituted with ECA repeating units (Figure 4a-c; Table 3).   [7,8]. The ECA repeating unit is colored in red. The capital letters refer to carbohydrate residues of the [ECA]-core OS, as shown in the inset structure. Letters with a prime sign denote residues of trace amounts of the core OS devoid of ECA. Chemical shift assignment for E. coli R1 is present in Table 2.
Moreover, interpreted NMR spectra (Table 2) indicated the identity of E. coli R1 [ECA]-core glycoform with the structure isolated from S. sonnei phase II characterized by R1 core chemotype (Figure 3c) [7,8]. Thus, the combined results also proved that ECA was linked to the LOS via the β(1→3) linkage between →4)-β-D-GlcpNAc-(1→ of ECA (residue J) and →3)-β-D-Glcp (residue I) of the outer core OS. Similarly to [ECA]-core OS of S. sonnei phase II, an inverted anomeric configuration of the D-GlcpNAc residue in the first ECA repeating unit linked to the core OS was observed, whereas an α-configuration was characteristic for the subsequent ECA repeating units (Figures 1 and 3, inset structure). To track the presence of ECALPS in the strain E. coli PCM 209 serotyped as O39 and characterized by R1 core OS, we performed identical preparation and analytical protocol, as for rough strains. Contrary to O-serotype designation suggesting smooth morphology, the strain turned out to be rough. The elution profile of Bio-Gel P-10 separation (data not shown) was identical to E. coli R1 and S. sonnei phase II [7,8]. The fraction 5 was analyzed by ESI-MS n (data not shown), where similar to E. coli R1 ion profile was observed with the ions attributed to [ECA]-core OS glycoforms. The  [7,8]. The ECA repeating unit is colored in red. The capital letters refer to carbohydrate residues of the [ECA]-core OS, as shown in the inset structure. Letters with a prime sign denote residues of trace amounts of the core OS devoid of ECA. Chemical shift assignment for E. coli R1 is present in Table 2.   Table 3. The most informative ions are colored in red. The mark # stands for non-interpreted ions.  Table 3.
The most informative ions are colored in red. The mark # stands for non-interpreted ions.   (Figure 4a-c, Table 3). Additionally, linear [ECA] 5 or [ECA] 4 were identified in the fractions 3-4 and the unsubstituted core OS glycoforms in the fraction 5. Interpretation of ions observed for fraction 6 was in agreement with the known structure of the free core OS of E. coli R2 (data not shown) (Figure 1) [9,10].
The ion at m/z 1278.88 (2+) detected in positive-ion mode (Figure 4d) was used to confirm the linkage between ECA and core OS by ESI-MS 2 (Figure 4e). The ion corresponded to the [ECA]-Glc 3 -GlcNAc-Gal-Hep 3 -Kdo-P 3 -Etn glycoform. The pattern of fragment ions (Y i , B i , Z i , X i , A i ) not only supported previously published carbohydrate sequences for ECA and R2 core OS but also demonstrated the linkage between R2 core OS and one repeating unit of ECA (Figure 4e, inset structure).
Two high-intensity ions were identified as a result of the glycoform fragmentation into ECA trisaccharide [B3α' at m/z 608.15 (1+)] and R2 core OS [Y6α at m/z 1949.50 (1+)]. The presence of ECA within the outer core region (the region of the terminal disaccharide α-d-GlcpNAc-(1→2)-α-d-Glcp present in Figure 1) was supported by B5α ion at m/z 1135.63 (1+) and Y5α at m/z 1584.35 (1+). Further in-depth fragment ions interpretation allowed to discriminate between the distal core OS GlcNAc and Glc as equally possible residues to be substituted by ECA repeating unit. The ion Y6α"/B4α at m/z 770.22 (1+) built up of ECA-Glc fragment pointed out specifically Glc as a residue substituted by ECA. This structure was additionally supported by the ion at m/z 449.17 (1+) attributed to the fragment 1,3 A 4 /Z 7α of outer core Glc substituted by GlcNAc and ECA, which most probably indicated the position 3 of the →2)-α-d-Glcp as a place of substitution by ECA (Figure 1). Further analysis of the linkage between ECA and core OS by NMR spectroscopy was not possible due to the high heterogeneity of fraction 5 (Figure 4c,d) that resulted in the complexity of NMR spectra.
For E. coli R4, the Bio-Gel P-10-based fractionation of poly-and oligosaccharides gave similar elution profile as for E. coli R1 and R2, which was characterized by 1-7 fractions (data not shown). Interpretation of the ions observed for fraction 6 was in agreement with previously published structures of the core OS of R4 chemotype (Figure 1) [10,11]. As E. coli R4 synthesized trace amounts of ECA LPS , fractions within the region corresponding to fraction 5 were not pooled, and every single fraction (collection tubes 25-28) was analyzed by MALDI-TOF MS. An interpretation of four representative fractions are presented in Figure 5a Table 4).    Table 4). The fractions 27 and 28 had a similar composition to fraction 26 (Figure 5c,d, Table 4 The ESI-MS 2 fragmentation pattern confirmed the linkage between ECA and core OS since a variety of fragments were identified containing both ECA and core OS residues (Figure 5f). Observed fragment ions of Y i , Z i , B i , C i , X i , and A i (mostly single charged ions) supported the sugar sequence of the ECA and core OS region and were interpreted according to the nomenclature of Domon and Costello [29]. Variety of ions were identified as fragments of ECA part, i.e., the ions at m/z 236. 16

ECA LPS is Absent in Rough E. coli R3 Lipooligosaccharide Preparation
The Bio-Gel P-10 elution profile of E. coli R3 poly-and oligosaccharides was similar to R1, R2, and R4 and also revealed the rough character of LOS R3 (1-7 fractions, data not shown). The MALDI-TOF mass spectrum of the fraction 5 did not show the presence of any ECA LPS -derived glycoforms ( Figure 6).
Observed ions were limited to the core OS of R3 chemotype, according to previously published structures [10,11]. Three core OS glycoforms were identified and marked in colors in Figure 6 characterized by general schemes of the core-P n -Etn n , core-GlcN-P n -Etn n , and core-GlcN-P n -Etn n , where the core expression stands for Glc 3 -Gal-GlcNAc-Hep 3 -Kdo core OS backbone. The core-GlcN-P n -Etn n glycoforms containing free GlcN were not reported before and were positively verified by NMR analysis of the fraction 5 (unpublished results). In the region of higher values of m/z, only trace amounts of ECA linear polymers were detected instead of searched core OS substituted with ECA (Table 5). No ECA LPS -derived polymers were identified in fractions 4, 3, 2, and 1.  Figure 6. Negative-ion mode MALDI-TOF mass spectrum of fraction 5 isolated from E. coli R3 LOS preparation. The interpretation of ions is shown in Table 5. The core stands for Glc3-Gal-GlcNAc-Hep3-Kdo oligosaccharide. Colors green, red, and blue refer to three R3 core OS glycoforms ( Figure  1).   Table 5. The core stands for Glc 3 -Gal-GlcNAc-Hep 3 -Kdo oligosaccharide. Colors green, red, and blue refer to three R3 core OS glycoforms (Figure 1).

Discussion
The story of ECA began with the discovery of ECA LPS since the presence of ECA LPS was originally inferred from serological cross-reactions in hemagglutination assays between patients' sera and various E. coli O-serotypes during studies of urinary tract E. coli infections [5]. A few E. coli strains (serotypes O14, O54, O124, and O144) elicited highly cross-reactive antibodies in rabbits that could be removed from anti-O14 serum by absorption with extracts of any E. coli strain while retaining homological reactivity of the serum [6,12]. Described cross-reactivity was not related to O or K antigens. Thus, it was concluded that anti-O14 serum contained antibodies binding an antigen that had to be common for all Enterobacteriaceae, finally identified as ECA PG . Following this initial discovery of ECA, ECA PG was identified, and its chemical structure and linkage to PG were elucidated [30]. Observed cross-reactivity suggested the presence of ECA as a non-immunogenic and an antigenic (ECA PG ) and an immunogenic (ECA LPS ) form. Ultimately, E. coli O14 proved to be rough strain expressing presumptive immunogenic ECA LPS and characterized by the R4 chemotype of the core OS. Its roughness was masked by capsular antigen [31]. Finally, a presence of ECA LPS in E. coli was suggested only for rough mutants expressing LOS with complete core OS, such as R1, R2, R4, and K-12 [15,17]. Conclusions described above were further supported by Whang et al. [16], who studied pairs of smooth parent strain and its corresponding rough mutant in case of E. coli R1 (strain F470), R2 (strain 576), R4 (O14), and R3 (strain F653), possessing complete core regions. Only rough and viable counterparts were able to elicit significant antibody response upon intravenous injection into rabbits. For heat-killed bacterial cells (100 • C, 1 h) used as an immunogen, only R1 and R4 (O14) were immunogenic.
Presented herein studies completed described serological and biochemical observation for E. coli with strong evidence supported by structural analysis. We utilized the protocol developed for S. sonnei phase II ECA LPS to searching for this immunogenic form in the prototype E. coli R1, R2, R3, and R4. Analyzed strains were prototype strains that were used during pioneering studies on R1 (strain F470), R2 (strain F576), R3 (strain F653), and R4 (strain 2513) core OS structures [9][10][11]. The polyand oligosaccharides comprising core OS substituted by ECA repeating units were identified in R1, R2, and R4. No ECA LPS -derived fragments were identified for E. coli R3 fractions. For rough E. coli R1 and PCM 209 (serotyped at the time of deposition as O39), the ESI-MS 2 analysis supported by NMR analysis indicated that ECA occupied outer core residue that used to be substituted by O-PS in the case of smooth LPS. The same was only assumed with a high probability for R2 and R4 by MS since NMR analysis was impossible due to insufficient amounts of homogenous [ECA]-core samples (Figure 1). Structural results were in agreement with the previous hypothesis that the linking of ECA to the core OS of LPS was under genetic control [32,33]. The first indication of this phenomenon was based on the different behavior of two E. coli R1 mutants, F470 and F614, where the latter one was not immunogenic regarding ECA. Then, Schmidt et al. suggested the E. coli R1 probable connections between rfaL gene functionality and the expression of ECA immunogenicity in R4 and R1 strains. The rfaL gene (nowadays, referred to as WaaL in E. coli) encodes O-PS ligase responsible for glycosidic bond formation between O-PS and core OS of LPS [32,33].
Nowadays, LPS (O-PS) and ECA biosynthesis pathways are well described and share some similarities. The lipid A-core OS part is assembled on the cytoplasmic side of the inner membrane and translocated across the inner membrane. The O-PS part is synthesized in a separate pathway, whereas both O-PS and ECA polysaccharide are produced via the Wzx/Wzy-dependent assembly pathway [3,[24][25][26]. Single repeating units of both polymers are assembled on the same lipid carrier-undecaprenyl pyrophosphate (Und-PP) [34]. For E. coli, the WecA enzyme is responsible for linkage formation between UndPP and sugar intermediate to promote subsequent repeating unit assembly. The same mechanisms of chain elongation are also used for both polymers [25]. Completed Und-PP-O-PS is finally ligated to lipid A-core OS by WaaL ligase to form a mature LPS molecule ready for transport to the outer membrane [3]. In the vast majority of bacterial species, the ECA present in the outer membrane is found covalently linked to the phosphatidylglycerol (ECA PG ). ECA molecules bound to the lipid A can only be observed in bacterial strains incapable of producing the O-PS [1]. ECA LPS occurrence used to be explained by key similarities between LPS and ECA biosynthesis described above, including Und-P carrier for precursor sugars, repeating units, and mature polymers [25,30].
A comparison of the biological repeating unit of the ECA and E. coli O-PS revealed one common feature for both polymers. Most of E. coli O antigens are synthesized via the Wzx/Wzy pathway, and d-GlcNAc serves as the initial sugar at the reducing end of O-PS that is attached to the core [35]. WecA-like transferases are specific for GlcNAc and initiate (by the formation of Und-PP-GlcNAc) the synthesis of ECA and most of E. coli O-PS [3]. In agreement with S. sonnei phase II ECA LPS structure,