Sugar composition, methylation, NMR, and MALDI-TOF (MS) analysis support the conclusion that the CE367 O-chain polysaccharide differs from that of CE3 in one primary feature. It is missing the TOM-Fuc residue and instead is terminated by a GlcA residue. All other known structural features of the wild-type LPS—including the sugars, their linkages, the variable methylation at O-2 of the 3- and 3,4-linked Fuc residues, the methyl esterification of GlcA, and the O-acetyl content—appear to be identical in mutant CE367. It has been speculated that the addition of the terminal TOM-Fuc residue might be an essential part of the regulation of O-chain length because it necessarily prevents the addition of further repeating units (18
). However, this appears not to be the case. The mutant has the same remarkable uniformity in the number of repeating units as that exhibited by the wild type. With regard to LPS synthesis, it is also interesting that the absence of TOM-Fuc and DOM-Fuc does not seem to decrease the relative number of LPS molecules carrying the O chain (LPS I), whereas the absence of 2-O
-methylation of the internal Fuc residues in strain CE395 leads to a 50% decrease in the relative amount of LPS I (35
One possible function of this residue might be as a structural feature involved in liganding to another biological molecule. An artificial illustration is in the binding of antibodies to the LPS. The lpe3
locus, which is required for the presence of TOM-Fuc in the O chain, is required also for maturation of the R. etli
CE3 O chain into a structure that is recognized by three rat MAbs and the majority of the antibodies of polyclonal sera developed in rabbits against the wild-type LPS. Previous studies have indicated that the epitopes recognized by MAbs JIM27, JIM28, and JIM29 probably overlap but are not identical (40
). Mutations in the lpe3
locus are unique in eliminating the binding of all three of these antibodies. The TOM-Fuc residue may be shared spatially among the epitopes, or it may affect the structure of one or more of the epitopes by a conformation effect at a distance.
Of these three antibodies, JIM28 and its requirements for binding have been studied the most extensively. The absence of TOM-Fuc in CE367 is the third documented change in the O chain that is correlated with decreased binding of this antibody. Loss of reactivity after growth in anthocyanins and at low pH is correlated with increased 2-O
-methylation of the internal fucose residues (5
; J. Box and K. D. Noel, unpublished data), and conversely, the absence of this 2-O
-methylation of internal fucoses in mutant strain CE395 (35
) is correlated with increased binding of the antibody (J. Box, V. J. Neumann, and K. D. Noel, unpublished data). Treatment of the LPS at pH 12, which should have eliminated both O-acetylation and the methyl esterified to GlcA in the O chain (18
), also eliminated JIM28 binding (Fig. C). If this base-sensitive feature, the TOM-Fuc residue, and the 2-hydroxyl of an internal fucose residue were all in contact with the antibody, the binding site would have spanned at least the last three residues of the O chain (Fig. ). X-ray crystallographic studies have documented two types of structural interactions of antibodies with LPS oligosaccharides (13
), burying of the terminal sugar residue in a hydrophobic cavity at the antibody binding site (43
), and an extended “groove” in the antibody that interacts by hydrophobic stacking and hydrogen bonding with at least three sugar residues in a branched repeating unit (12
). The hypothetical interaction of the R. etli
CE3 O chain might have features in common with both of these models, with TOM-Fuc fitting into a hydrophobic pocket of an extended groove on the antibody.
Although the sequences of the putative lpe3
ORFs did not provide strong matches with sequences of known function in the database, the weak matches were consistent with hypothetical roles in the methylation of fucose. In concert with this role, preliminary sequencing of DNA upstream of lpe3
has revealed that ORF1 has greater sequence similarity than LpeA to NoeI (the 2-O
-methylase of fucose in Nod factors [26
]). It also has uncovered an ORF with great similarity to fucose synthetases and another with sequence similarity to putative glycosyltransferases (J. Box, D. M. Duelli, and K. D. Noel, unpublished data). Hence, lpe3
may be part of a larger locus that specifies both the methylation of fucose and its addition to the end of the O antigen. In regard to the specific biochemistry involved, two findings from the LPS structural analysis should be noted. One, not only the methyl groups of TOM-Fuc but also the fucose itself are missing from CE367 LPS. Two, other fucose residues and their variable 2-O
-methylations are not affected. If this locus is only involved in the 2,3,4-tri-O
-methylation, the absence of unmethylated terminal fucose would favor the idea that methylation occurs at the level of GDP-Fuc, before transfer to the O chain.
Genetic analysis suggests that the lpe3
genes are expressed in at least two separate transcriptional units, one ending in lpeA
and the other including ORF2. At least one downstream gene was required for complementation of the insertion mutation in ORF2, perhaps because this insertion has polar effects on expression of ORF3 or ORF4. Indeed, a kanamycin resistance cassette that is believed to be very similar to the one used in this study has strongly polar effects, regardless of the orientation of the insertion (3
). Similarly, the insertion in ORF3 (lpe-462
::Km) may be affecting the expression of ORF4. In any case, it also remains to be seen whether the lpe
genes are expressed in R. etli
as the polypeptides inferred from ORF analysis. Hence, a conservative interpretation of the genetic data is that at least two lpe
genes are required for epitope synthesis: lpeA
and a gene downstream of ORF2.
The results of this study provide no support for the possibility that CE3 responds to low pH or the presence of anthocyanins by repressing lpeA
transcription. Such results do not rule out the regulation of the lpe
gene products in response to such conditions, but structural analysis of the LPS produced during growth at low pH lends no support to that idea either. Although TOM-Fuc is absent, DOM-Fuc is present (5
). On the other hand, recent results indicate that one effect of growth in seed exudate is the absence of TOM-Fuc and DOM-Fuc from the synthesized O antigen (J. Box and K. D. Noel, unpublished data). In that case, another gene responsible for adding the terminal residue, such as the gene for the glycosyltransferase, may be repressed or the effect may be posttranscriptional.
The biological functions of the TOM-Fuc residue and the lpe3 locus remain obscure. Recent work in progress indicates that another decoration of the R. etli CE3 LPS, the variable 2-O-methylation of the internal fucose residues of the O chain, is relatively unimportant when the bacteria are inoculated on roots but becomes important when the bacteria are inoculated directly onto P. vulgaris seeds (J. Box and K. D. Noel, unpublished data). The benefit in this latter case may derive from coping with toxic compounds released from the seed. In this connection, it may be significant that all of the known antigenic modifications of the LPS in R. leguminosarum and R. etli are induced by what could be considered stress conditions. The results of this study indicate that the terminal residue of the O antigen and its three O-methyl groups have little effect, if any, on the symbiotic proficiency of this bacterium, even when the bacteria are inoculated directly on the host seed. Indeed, if loss of this structure is one of the LPS changes induced by the plant, this result is as expected. The benefit of having the TOM-Fuc residue under other conditions may be related to surviving or thriving under conditions that have been common during the evolution of this species but which thus far have not been tested in the laboratory.