2.1. Purified LPS from Lcr Is Low Molecular Weight
To get an estimation of the size of the LPS from Lcr, and specifically, whether it synthesizes a complete high molecular weight
O-polysaccharide, or only a truncated low molecular weight LPS/LOS, we compared its DOC-PAGE profile (
Figure 1, Lane 6) with PAGE profiles of well-characterized S- and R- type lipopolysaccharides. We used
Salmonella enterica serovar Minnesota S-strain (
Figure 1, Lanes 1 and 7);
E. coli EH100 (Ra mutant) (
Figure 1, Lane 2),
S. minnesota R5 (Rc mutant) (
Figure 1, Lane 3);
S. minnesota R7 (Rd mutant) (
Figure 1, Lane 4); and
S. enterica Re 595 (Re mutant) (
Figure 1, Lane 5) as standards. The comparative analysis demonstrated that most of the Lcr LPS resolved in two distinct LMW bands at the bottom of the gel. The upper LMW band ran similarly to the LMW band of the Ra mutant of
E. coli EH100. This mutant produces LPS deprived of the O-chain, with a lipid A and full core oligosaccharide (approximate
MW of 3.9 kDa; compare Lane 6 with Lane 2). The location of the lower band was consistent with the LMW bands of the Rd-Re mutants of
S. minnesota (approximate maximum
MW 3–3.165 kDa; compare Lane 6 with Lanes 4 and 5, respectively). In addition to the two LMW bands, we observed a very faint ladder pattern in the upper part of the PAGE (
Figure 1, Lane 6)
Figure 1, Panel B shows the same gel after overexposure to the silver stain. The overexposure allowed for better visualization of the higher molecular weight ladder pattern seen faintly in Panel A. This suggests the presence in trace amounts of LPS with a higher molecular weight
O-polysaccharide or other contaminating polysaccharides.
To further determine the molecular weight of the Lcr LPS, we performed size exclusion chromatography (SEC) of the
O-polysaccharides from Lcr and the
E. coli EH100 Ra mutant after mild acid hydrolysis and removal of the lipid A. These results can be seen in
Figure 2. As with the DOC-PAGE gel (
Figure 1), the SEC analysis shows the two released polysaccharides from both species to be of similar molecular weight. The released polysaccharide from Lcr has a slightly larger range of molecular weights, with most of the sample coming out as a sharp, early eluting peak in the column void volume (between 6 and 8 min,
Figure 2) and the rest of the released polysaccharide eluting between 8 and 12 min. The
E. coli EH100 core has an early eluting peak that comes out slightly later than the void volume of the column and ends slightly earlier than the Lcr polysaccharide. This analysis is consistent with the DOC-PAGE gel showing the two main Lcr bands having a slightly larger range than the
E. coli EH100 Ra mutant. The SEC analysis confirms the size of the Lcr oligosaccharide is small, with the LMW portion of the released
O-polysaccharide smaller than the 3.9 kDa
E. coli EH100 Ra mutant, but the HMW portion of the released
O-polysaccharide larger than the void volume of the column.
Taken together, the results of the DOC-PAGE gel and the SEC analysis revealed that the molecular weight of the Lcr LPS is relatively small, with some portion of the LPS larger in range and molecular weight than that of 3.9 kDa molecular weight of the E. coli EH100 Ra mutant, and without significant amounts of larger molecular weight O-polysaccharide.
2.7. NMR Analysis of the O-polysaccharide from Lcr
The 1-D proton NMR spectrum of the polysaccharide released from Lcr LPS after removal of lipid A by mild acid hydrolysis is shown in
Figure 7. The anomeric region of the spectrum featured 7 major and about 10–12 minor resonances. The proton NMR spectrum was also characterized by the presence of a complex region between 4.4 and 3.5 ppm due to the remaining glycosyl ring protons, N-acetyl methyl signals at 2.04 ppm, a few smaller signals between 2.2 and 1.8 ppm, corresponding to the H-3 protons of Kdo, and a strong multiplet signal around 1.28 ppm, belonging to H-6 of several Rha residues.
The residues belonging to the major anomeric signals were labeled with letters and assigned using a set of 2-dimensional 1H-1H- and 1H-13C-NMR correlation experiments, including 1H-1H-COSY, TOCSY, and NOESY and 1H-13C-HSQC and HMBC.
Residues A, B, C, E, and G contained anomeric protons resonances of 5.50, 5.42, 5.40, 5.33, and 5.22 ppm (
Table 5), respectively. According to the HSQC spectrum, these anomeric protons were attached to carbons resonating between 109.2 and 111.0 ppm (
Table 5), indicating that all of these residues were in the furanose ring form [
24]. This was further confirmed by HMBC correlations between H-1 and C-4 for each residue (
Figure 8 and
Figure 9) and linkage analysis (
Figure 5) which found Rib to be in the furanose form. The
3J
H1-H2 coupling constant for each ribofuranosyl (Ribf) residue was too small to measure, indicating β-anomeric configurations [
25]. Residues A, B, and C had carbon-2 resonances of 84.0, 83.7, and 83.7 ppm respectively (
Table 5), which is 5.9 and 5.6 ppm downfield from C-2 of unsubstituted β-Ribf [
26], identifying these as 2-linked-β-Ribf-1→ residues. Conversely, C-2 of residues E and G appeared slightly upfield of referenced unsubstituted β-Ribf, and their C-3 resonances appeared at 81.9 and 82.3 ppm (
Table 5), 8.6 and 9.0 ppm downfield of unsubstituted β-Ribf, indicated that residues E and G were 3- β-Ribf-1.
The chemical shifts of residues P (H-1 at 4.497 ppm) and Q (H-1 at 4.493 ppm) were very similar. The anomeric signals of these two residues could only be distinguished in the 2D NOESY and HMBC spectra (by their correlations to different neighboring residues, see below) but were not resolved in the proton spectrum. Both residues were characterized by a TOCSY pattern with 3 cross-peaks correlated with H-1. The COSY spectrum revealed the sequence of these peaks within the monosaccharide ring and assigned H-2 and H-3 at 3.57 and 3.77 ppm for both residues P and Q, and H-4 at 4.03 ppm for residue P or 4.05 ppm for residue Q, respectively. This peak pattern is consistent with these residues having a β-galactopyranose [
24] configuration. Compared with monomeric β-Gal
p [
27], the chemical shift of C-4 of both P and Q displayed a strong α-effect of about +7 ppm, whereas C-3 exhibited a weak β-effect of −0.2 ppm. This suggested the two residues were 4-β-Galp-1→.
The sequence of the monosaccharide residues A, B, C, E, G, P, and Q was determined by inspecting the
1H-
1H-NOESY and
1H-
13C-HMBC NMR spectra for inter-residue correlations. Here, two different polymers were observed. The simpler of the two consisted of a disaccharide repeating unit of residues E and Q. The HMBC showed clear correlations between the anomeric proton of Q at 4.49 ppm and the C-3 of E at 81.9 ppm. Similarly, the anomeric proton of E at 5.33 ppm was correlated to the C-4 of Q at 78.5 ppm. The NOESY spectrum showed equivalent results, with correlations between the anomeric proton of residue E at 5.33 and H-4 of residue Q at 4.05 ppm. A similar correlation was present between the anomeric proton of Q at 4.49 ppm and the H-3 of residue E at 4.21 ppm (
Figure 9). No other inter-residue correlations were found connecting these two residues with any of the other major residues in the sample, although the terminal β-Gal residue W was found at the non-reducing end of this chain. Taken together, these results led to the determination that these two residues were part of a disaccharide repeat with the sequence 3-β-Ribf-1→4-β-Galp-1→ (E-Q).
Connectivities between the remaining major residues were further examined. The HMBC spectrum showed inter-residue connections between H-1 of B and C-2 of A (5.42/84.0 ppm), between H-1 of A and C-4 of P (5.50/78.5 ppm), between H-1 of P and C-3 of G (4.50/82.3 ppm), between H-1 of G and C-2 of C (5.22/83.7 ppm), between H-1 of C and C-2 of B (5.41/83.7 ppm). The NOESY spectrum confirmed these connectivities with cross-peaks correlating H-1 of B with H-2 of A (5.42/4.22 ppm), H-1 of A with H-4 of P (5.50/4.03 ppm), H-1 of P with H-3 of G (4.50/4.31 ppm), H-1 of G with H-2 of C (5.22/4.21 ppm), and H-1 of C with H-2 of B (5.41/4.20 ppm). These correlations indicated the pentasaccharide repeat sequence B-A-P-G-C or [2-β-Ribf-1→2-β-Ribf-1→4-β-Galp-1→3-β-Ribf-1→2-β-Ribf-1→]n. Two minor terminal O-acetylated ribose residues, S and T make up the non-reducing end of this polymer, S being acetylated on O-3 and T on O-2.
Among the minor anomeric signals were two more that corresponded to the residues of a disaccharide repeating unit. Residue M with an anomeric proton chemical shift of 4.91 ppm was identified as α-rhamnopyranose by its downfield H-2 (4.08 ppm) and H-5 (4.05 ppm) and upfield H-6 (1.26 ppm). An α-effect of +5.3 ppm and a β-effect of −3.6 ppm [
28] indicated glycosylation on O-3. Residue L was assigned as α-GlcNAc because of its
3J
H1,H2 coupling constant of 3.8 Hz and its upfield H-4 (3.68 ppm) and C-2 (56.6 ppm) resonances. An α-effect of +6.9 ppm of C-4 (80.1 ppm) relative to H-4 of unsubstituted GlcNAc [
28] indicated glycosylation in this position. The connections between residues M and L were established by HMBC cross-peaks at 4.91/80.1 ppm and 5.01/78.2 ppm and by NOESY cross-peaks at 4.91/3.68 ppm and 5.01/3.76 ppm, showing a disaccharide repeat of 3-α-Rhap-1→4-α-GlcpNAc-1→ (M-L). This repeating unit structure is identical to that of the
O-polysaccharide of the LPS from
Actinobacillus actinomycetemcomitans (OMZ 534) serotype e [
29].
The residue labeled R contained a pair of resonances at 1.86 and 2.07 ppm arising from a methylene group, as evidenced by its negative intensity in the multiplicity-edited HSQC spectrum. These peaks are characteristic of the Kdo H-3 protons consistent with the observation of Kdo during glycosyl composition analysis (
Figure 3). This residue links the core oligosaccharide to lipid A in the LPS. Downfield C-1 (179.2 ppm) and upfield C-2 (99.2) chemical shifts indicated α-anomeric configuration of this residue [
30,
31], and a downfield displacement of the C-5 chemical shift (+9.4 ppm) [
32] gave evidence of it being glycosylated on O-5.
The anomeric peaks of residues O and N contained proton anomeric resonances at 4.51 and 4.65 respectively. The TOCSY spectrum showed three cross-peaks each to these anomeric protons. Following the correlations in the COSY spectrum gave H-2, H-3, and H-4 chemical shifts of 3.67, 3.72, and 4.00 ppm for residue O and 3.60, 3.78, and 4.02 ppm for residue N. These peak patterns are consistent with these residues having a β-galactopyranose [
24] configuration. Compared with unsubstituted β-Gal
p [
27], the chemical shift of C-4 of residue N and C-3 of residue O displayed strong α-effects of about 9.0 and 8.8 ppm respectively, suggesting residue N was 4-β-Galp-1→ and residue O was 3-β-Galp-1→.
The anomeric signal of residue F resonated at 5.32/111.1 in the HSQC spectrum, indicating furanose form. Since ribose was the only furanose found in the linkage, residue F could be identified as Ribf. The lack of a measurable
3J
H1-H2 coupling constant indicated it was in the β-anomeric configuration [
25]. The chemical shift of C-3 for this residue was 82.0 ppm, a downfield displacement of over 8 ppm from unsubstituted β-Ribf, identifying residue F as 3-β-Ribf-1→.
The TOCSY spectrum allowed assignment of residues K and J as rhamnosyl residues based on the cross-peaks of H-5, H-4, H-3, and H-2 with H-6 at 1.28 and 1.30 ppm, respectively. These latter, upfield proton chemical shifts are characteristic of the C-6 methyl of 6-deoxyhexose residues, and rhamnose is the only 6-deoxyhexose residue present in these samples, as observed in the composition and linkage analyses. H-5 of both of these residues resonated above 3.8 ppm and C-5 around 72 ppm, values are consistent with K and J having an α-anomeric configuration [
28]. The chemical shifts of C-3 of residues K and J at 80.9 and 82.7 ppm respectively, showed large α-effects of 9.9 and 11.7 ppm from unsubstituted α-Rhap, identifying residues K and J as 3-α-Rhap-1→.
The H-1 of residue I resonated at 5.15 ppm. The TOCSY spectrum showed four cross-peaks correlated to the anomeric proton of this residue. Sequencing these peaks in the COSY spectrum identified H-2, H-3, H-4, and H-5 as having chemical shifts of 3.59, 3.94, 3.71, and 4.24 ppm, a pattern consistent with α-Glcp. A 3JH1-H2 coupling constant of 4 Hz confirmed α-anomeric configuration. The chemical shift of C-4 for this residue was 80.6 ppm, which is a downfield displacement of over 9 ppm from unsubstituted α-glucopyranose, identifying residue I as 4-α-Glcp-1→.
The HMBC spectrum showed correlations between H-1 of residue I and C-5 of R (5.15/78.1 ppm), between H-1 of O and C-4 of I (4.51/80.6), between H-1 of K and C-3 of O (5.02/82.8 ppm), and between H-1of J and C-3 of K (5.05/80.9 ppm). The NOESY spectrum showed these same four residues to have corresponding correlations between their anomeric protons and ring protons H-5 on residue R at 4.09 ppm, H-4 of residue I at 3.71 ppm, H-3 of residue O at 3.72 ppm, and H-3 of residue K at 3.91 ppm respectively. These correlations indicated the residue sequence J-K-O-I-R or 3-α-Rhap-1→3-α-Rhap-1→3-β-Galp-1→4-α-Glcp-1→5-α-Kdop.
The HMBC and NOESY spectra furthermore showed correlations between H-1 of residue N and C-3 and H-3 of J (4.65/82.7 ppm and 4.65/4.00 ppm). A correlation between H-1 of F and H-4 of N (5.32/4.02 ppm) was also found, allowing the conclusion that the disaccharide 3-β-Ribf-1→4-β-Galp-1→ is attached to the non-reducing end of the core pentasaccharide, forming the sequence F-N-J-K-O-I-R. It is likely that the E-Q- repeat is attached to the non-reducing end of F, although the similarity of the chemical shifts of F with those of E and the associated peak overlap and, thus, precluded confirming this linkage. The complete core oligosaccharide consists of the pentasaccharide polymer 3-α-Rhap-1→3-α-Rhap-1→3-β-Galp-1→4-α-Glcp-1→5-α-Kdop-1 with residues F and N comprising the first polysaccharide disaccharide repeat linking the O-polysaccharide to the core.
Two additional non-reducing end residues, U and V, constitute the termini of core structures, truncated at the sixth (i.e., U replacing N) or fifth (i.e., V replacing J) positions, respectively, from Kdo. These, together with the three other terminal residues, S, T, and W, helped define the different LPS/LOS species isolated from Lcr. No reducing end residues other than Kdo were detected, indicating that the five non-reducing end sugars each represented an O-chain-core lipopolysaccharide (LPS) or core only lipooligosaccharide (LOS) and that each corresponded to a species that included one or more of the different sequences described above. Thus, structures 1 and 2 contained the F-N-J-K-O-I-R core, a number of E-Q repeats, and several B-A-P-G-C units. Structure 3 comprised the F-N-J-K-O-I-R core and a number of E-Q repeats, and structures 4 and 5 consisted of different lengths of core only (
Figure 10). The number of O-chain repeats B-A-P-G-C and E-Q was estimated from the average intensity of their anomeric proton signals.
The intensity of the terminal residues S, T, U, V, and W was used to estimate the molar amounts of the five structures present. Then, the expected intensities of all residues were back-calculated using the numbers of residues in, together with the relative molar amounts of each of the five structures (
Table 6). A comparison of the calculated with the observed intensities gave reasonably good agreement, confirming the accuracy of the proposed structures. Knowledge of the monosaccharide composition thus obtained of each structure also gave their molecular weights and, together with the molecular weight determined above of the lipid A (~1.8 kDa), the molecular weights of the corresponding LPS/LOS species (
Figure 10). The five structures fall into three size categories: LPS structures 1 and 2 are fairly large (~11 kDa), LOS structures 4 and 5 are small (~3 kDa), and structure 3 is of intermediate size (~5 kDa). The abundance ratio of the three categories is about 2:2:1. These observations agree with the DOC-PAGE analysis (
Figure 1), which showed two intense bands near the bottom and a weaker band between them. These findings also agree with the SEC results from the delipidated polysaccharides (
Figure 2). The gel also showed a very faint ladder of much higher molecular weight. This most likely corresponds to the 3-α-Rhap-1→4-α-GlcpNAc-1→ (M-L) polysaccharide, of which no terminal residues were found as would be expected for a very large polysaccharide.