Aiming at addressing the need for increased structural diversity and to cover a wide range of plant-oligosaccharide chain-length in microarray platforms developed to date, we report here an oligosaccharide-NGL microarray platform, comprised of naturally-derived hemicellulose-related oligosaccharides. The microarray of these sources further extended and complemented the previously constructed glucan microarray platform32. The microarray analysis with the different
carbohydrate-binding proteins investigated validated the glucan and hemicellulose oligosaccharide microarrays for their application to plant cell wall carbohydrate recognition. The
51
Figure 2.8. Validation and analysis of the new xyloglucan microarrays. Binding signals of
non-fucosylated- and fucosylated-xyloglucan-specific monoclonal antibodies LM24, LM25 and CCRC-M1 and α-fucose-specific lectin ALL are depicted as means of fluorescence intensities of duplicate spots at 5 fmol of oligosaccharide probe arrayed (with error bars) and are representative of at least two independent experiments. The different xyloglucan sources are indicated in the coloured panels. (A) The microarrays
included the xyloglucan-DAN-DHPA-NGLs probes 1 to 3, 7 and 8, as controls and the initially arrayed AO-NGLs 4 to 6, 9 and 10 (probes 1 to 10). (B) The new xyloglucan-AO-NGLs derived from the
deconvolution of xyloglucan oligosaccharide mixture from apple were also included for analysis (probes 11 to 21). Carbohydrate sequence information of the DAN-DHPA- and AO-NGL probes is in Table S2.5.
results showed robust binding signals and differing patterns with respect to oligosaccharide structural type, linkage and sources by the anti-plant carbohydrate antibodies, lectins and bacterial CBMs. The results were in overall consistent with data obtained previously on the specificity of these proteins using different techniques. Additionally, the wide range of oligosaccharide sequences covered in the microarrays allowed to broaden the knowledge on their carbohydrate binding and specificities.
52
Table 2.3. MALDI-MS analysis of the new xyloglucan-AO-NGL probes generated.
Xyloglucan probes DPa [M-H]-
calculatedb [M-H]- detectedc,d Main compositione
Xyl-Glucan DP4-AO 4 1353.77 1354.0 (1690.2) Hex3Xyl1 (Hex4Xyl2(Ac))
Xyl-Glucan DP5-AO 5 1485.81 1486.1 (1354.0) Hex3Xyl2 (Hex3Xyl1)
Xyl-Glucan DP6a-AO 6 1647.86 1648.1 Hex4Xyl2
Xyl-Glucan DP6b-AO 6 1689.86 1690.1 (1836.2) Hex4Xyl2(Ac) (Hex4Xyl2Fuc(Ac))
Xyl-Glucan DP7-AO 7 1779.90 1780.1 Hex4Xyl3
Xyl-Glucan DP8a-AO 8 1941.96 1942.3 (2089.4, 2147.4, 2293.5)
Hex5Xyl3 (Hex5Xyl3Fuc,
Hex6.Xyl3(Ac), Hex6Xyl3Fuc(Ac))
Xyl-Glucan DP8b-AO 8 1941.96 1942.2 (2105.3,
2251.4) Hex5Xyl3 (Hex6Xyl3, Hex5Xyl3Fuc) Xyl-Glucan DP9-AO 9 2104.01 2105.2 (2251.4) Hex6Xyl3 (Hex6Xyl3Fuc)
Xyl-Glucan DP11/12-AO 11 2454.12 2454.6 (2586.6) Hex7Xyl3Fuc(Ac) (Hex7Xyl4Fuc(Ac))
Xyl-Glucan DP13-AO 13 2706.21 2707.7 Hex8Xyl4Fuc
aDegree of polymerization (DP) for the major components in each fraction; bCalculated masses for major components are
given; cNegative-ion MALDI-MS was used for the analysis of the AO-NGLs and [M-H]- were detected; dWhere multiple
components were detected, relative intensities of ions greater than 20% are shown in brackets; eProposed composition
of the major components as detected by negative-ion MALDI-MS.
The binding recognition of linear α1,2-, α1,3-, α1,4- and α1,6-glucan-oligosaccharides was differentiated for α-glucan recognising proteins, where the glucan-binding property of human malectin was shown predominantly to linear α1,3-glucan oligosaccharides, exhibiting a chain-length dependency up to the tetrasaccharide. The additional binding to α1,4-, α1,6- and β1,3-linked glucose oligosaccharides highlights the plasticity of the malectin binding site to accommodate other glucose linkages dependent of their conformation and linkage. The ligand-specificity of CtCBM25Cthe_0956 was assigned to α1,4-linked glucose epitopes in linear or
mixed-linked glucan-oligosaccharides. The chain-length dependency observed for sequences longer than DP-4 points to a type B topology of this CBM’s binding site, able to accommodate a minimum of 4 α1,4-glucose units. The microarray analysis also enabled to discriminate specific binding towards linear β1,3-, β1,4- or mixed-linked β1,3-β1,4-glucans and the chain-length requirements of β-glucan recognising proteins. The broader binding patterns exhibited by
CmCBM6-2 emphasize the plasticity of its type B and C binding sites to a wide range of β-linked
hexoses. The analysis also showed the influence of the spacing and positioning of β1,3-linkages within mixed-linked β1,3-β1,4-glucan chains for binding by the antibody 400-3 and
CtCBM11Cthe_1472.
The carbohydrate-binding proteins analysed on the new hemicellulose-related microarrays developed, allowed not only to increase knowledge on the carbohydrate-specificities for the proteins analysed, but also to shed light into the sequence of some of the NGL-oligosaccharides derived from heterogeneous mixtures, for which sequence is still under full assignment.
The binding patterns of β1,4-linked xylan-specific antibodies LM10 and LM11 and CBM22-2Cthe_0912 highlighted the differences in the binding mode by these proteins, which can
53
relate to their binding sites topology. LM10 showed a specificity towards the non-reducing end of β-linked xylose oligosaccharides, although being able to accommodate to some extent α1,2-/1,3-arabinose terminal branches at this position, which points to a cavity-type antibody. LM11 on its turn required the internal β-xylose backbone accommodating longer epitopes of at least 3 xylose monomers, suggesting a groove-type antibody. The binding of CtCBM22-2Cthe_0912
to linear β-linked xylan oligosaccharides with a chain-length dependency from DP-4 up to DP-8 depicts the typical binding profile observed for a groove-type B CBM. Remarkably, the fact that both the anti-β1,4-xylan antibodies and CtCBM22-2Cthe_0912 bound strongly to the extent of the
mixed-linked β1,3-β1,4-xylose probes (probes 153 to 164), allows to infer that the β1,3-linkage, if present, might be positioned at the reducing end of the smaller DP probes, and hence not exposed to the proteins. On the one hand, in longer DP fractions, β1,3-linkages could be present every 4 to 5 xylose monomers144, exposing β1,4-linked xylose stretches available for binding. On
the other hand, binding to these sequences might also not be influenced by the presence of a β1,3-linkage. While the microarray analysis allows to infer about these oligosaccharide sequences, their defined sequence needs to be confirmed, recurring for example to diagnostic fragmentation method by negative-Ion ESI-CID-MS/MS32.
The microarray analysis with mannan-directed proteins showed the specific and different chain-length requirement for binding to linear β1,4-mannose sequences or the requirement of the α-galactose substitutions, highlighting proteins that selectively bind to linear mannans or galactomannans. While 400-4 showed a chain-length dependency of the β1,4-linked mannan backbone, pointing to a groove-type antibody, LM21 bound to shorter chains, pointing to recognition of the non-reducing end of the β1,4-linked mannans and to a cavity-type antibody. The assignment of the specificity for CtCBM35Cthe_2811 to the β1,4-linked mannose backbone of
mannans with a chain-length of at least 4 residues is in agreement with a type B CBM, as generally observed for family 35141. In contrast with these proteins, the antibody CCRC-M7
requires the α-galactose substitutions of β1,4-mannose backbone for binding. The lack of binding to the di-galactosyl-mannopentaose probe (probe 195) and the higher binding intensity to probe 196 (DP-8), may indicate that, for recognition to occur, CCRC-M70 requires either more than 2 α-galactose residues or a β1,4-mannose backbone of more than 5 residues. The integration of the microarray data showed that probes from the e-series are composed mainly by linear β1,4-mannose up to the DP-9 probe, and that the m-series comprise sequences with a higher ratio of α-galactose substitutions. Using the same rationale, the binding profiles to the longer
e-series probes of DP-9 to DP-11, points to longer DPs of the β1,4-mannose backbone and the
presence of at least 2 or higher α-galactose substitutions in these sequences. Sequencing of these galactomannan-derived oligosaccharide sequences is under way using the established ESI-MS/MS method and NMR analysis32.
The analysis with the xyloglucan microarrays revealed the different binding features and epitopes for 3 anti-xyloglucan antibodies: LM24 towards β1,2-linked galactose substituted xyloglucans;
54
LM25 towards α1,6-linked xylose substituted xyloglucans, exhibiting less tolerance for highly substituted β1,2-linked galactose or α1,2-linked fucose xyloglucans; and CCRC-M1 towards fucosylated xyloglucans requiring a single α-Fuc-(1,2)-β-Gal as the minimum recognition epitope. Deconvolution of the detailed oligosaccharide epitopes recognized by these antibodies is detrimental for their application as research tools for detection and characterization of specific carbohydrate sequences present in plant polysaccharides.
Aiming at diversifying the microarrays with sequence-defined NGL-oligosaccharides, the deconvolution method with conjugation to the bi-functional DAN, followed by derivatization to an aldehyde-functionalized lipid, was attempted to separate the neutral xyloglucan oligosaccharides and to achieve a relatively homogenous population of DAN-DHPA-NGL probes. This strategy showed potential to enrich each of the tamarind xyloglucan oligosaccharide fractions. However, application of the method to more heterogeneous oligosaccharide mixtures, like the fucosylated-xyloglucans from apple, requires further optimisation in order to improve the conjugation yields and chromatographic resolution. Additionally, as the purified oligosaccharides from complex plant polysaccharides are frequently obtained in limited amounts, this precludes the use of conventional NMR method for sequencing, which is currently ongoing for xyloglucan fractions, as well as for galactomannans from the m-series and e-series.