HRGPs and low molecular weight mannose-rich proteins. We first extracted C. reinhardtii’s cell wall and verified if its composition was identical to the native one. It was extracted with chaotropic agents, then slowly reconstituted after dialysis - a milder protocol than the one originally proposed for C. reinhardtii17,18, and that preserves more material, compatible with NMR studies (Fig. 1a). Scanning electron microscopy (SEM) (Fig. 1b) and transmission electron microscopy (TEM) (Fig. 1c) respectively show the intact microalga and the characteristic layered organization of the cell wall described in the literature. Cells were also undamaged after wall extraction, as demonstrated by the intact membrane bilayer and internal organelles such as thylakoids, partially visible in Fig. 1d. This was further confirmed by flow cytometry analysis, which revealed that about 80% of the cells remained alive and metabolically active after wall extraction. The composition of the cell wall extract was also comparable to the native one, as shown by Magic-Angle Spinning Dynamic Nuclear Polarization (MAS-DNP22,23) of intact microalgae. This ssNMR technique provides a tremendous increase in sensitivity (ca. 45 times signal enhancement as shown in Fig. 1e) but also a loss in dynamics and resolution (Supplementary Fig. 1) due to the low temperature required for this method. Both the MAS-DNP enhanced 1D and 2D 13C-13C correlation spectra of whole C. reinhardtii cell and cell wall extracts are very similar (Fig. 1e-f), confirming that the cell wall structure was preserved in the extracts, and that the polarizing radical does not diffuse through the membrane, but preferentially locates in the cell wall. The remaining differences can be ascribed to the fact that only the W4 and W6 layers were removed when extracting the cell wall, leaving the fibrillar W1, W2 and W7 layers attached to the cell surface (Fig. 1a). This was also confirmed by the similarities between the MAS-DNP spectra of the strain described as cell-wall deficient (cw15) and the wild-type strain after cell-wall extraction (Fig. 1g).
We then determined the complete glycan and amino acid profile of C. reinhardtii’s cell wall extracts. Using fully relaxed 1D 13C ssNMR spectra, we compared the integrals of the carbohydrate C1 and amino acid CO resonances (Fig. 2a) and obtained a 1:3 glycan/protein molar ratio, indicating that a significant part of the cell wall is composed of glycans, in agreement with the literature24. The composition of the cell wall extracts was then monitored using SDS-PAGE and compared to the literature13. The amino acid analysis of the SDS-PAGE bands confirms that the high molecular weight proteins in the cell wall extracts are HRGPs. We found a higher abundance of low molecular weight proteins (LWGPs) following our milder extraction protocol (Fig. 2b). A chemical deglycosylation revealed, by SDS-PAGE, a glycosylation level (Fig. 2b) similar to that found by ssNMR (Fig. 2a).
We performed a quantitative analysis of the saccharide content and bonds of C. reinhardtii’s cell wall using our protocol25 based on 2D 13C-13C ssNMR spectra (Fig. 2c for glycan zoom-in and Supplementary Fig. 2 for complete 2D refocused J-INADEQUATE spectrum, Supplementary Tables 1–2 for glycan assignments and Supplementary Tables 3–4 for quantification), and MS data (Fig. 2d and Supplementary Table 5 for glycan linkages and Fig. 2e and Supplementary Table 4 for quantification). By ssNMR, in good agreement with GC-MS, we found a majority of mannose (52%), followed by galactose (15%), arabinose (15%), N-acetylated glucose (11%), non-modified glucose (3%), rhamnose (3%) and xylose (3%), as reported in Fig. 2e (see also Supplementary Table 4). Interestingly, a significant proportion of glucose is N-acetylated (N-acetylglucosamine, GlcNAc), as previously but not quantitatively reported in Chlamydomonas cells26. Moreover, a high number of anomeric carbons mainly ascribed to mannose and galactose, have very low chemical shift values and can therefore be assigned to non-reducing terminal groups of oligosaccharides (blue bars on Fig. 2e). These results also confirm the high proportion of non-reducing terminal groups detected by GC-MS (see t-X glycans on Fig. 2d).
GC-MS analysis also revealed glycan linkages (Supplementary Table 5), further demonstrating the high variety of glycosidic bonds in C. reinhardtii’s cell wall, even in mannose units. We also used GC-MS to evaluate the efficiency of our cell wall extraction protocol by comparing the glycan composition of whole cells (dominated by starch glucose), cells after wall extraction, and the extracted cell wall (Tables 1 and Supplementary Table 6). We calculated the contribution of glycans in the whole cells that originate from the cell wall, and found a similar proportionality factor for each glycan, confirming that almost no sugar units were lost or altered during the extraction and reconstitution, with an uncertainty for xylose, which proportion is close to ssNMR and MS detection limits. This proportionality factor of ∼20 indicates that about 20% of the whole cell glycans are contained in the cell wall.
Table 1
Glycan content determined by GC-MS.
| Whole cell | Whole cell after wall extraction | Cell wall extract | Proportionality factor |
Man | 8.7 | 1.6 | 50.4 | 17 |
Ara | 8.0 | 5.3 | 20.1 | 22 |
Gal | 4.6 | 2.5 | 18.7 | 15 |
Rha | 1.2 | 1.0 | 2.2 | 20 |
Glc | 75.0 | 88.5 | 1.5 | 18 |
Xyl | 1.3 | 0.8 | 1.5 | - |
GlcNAc | 1.4 | 0.3 | 5.6 | 26 |
Glycan contents (in molar %) of cells before and after wall extraction, and in the extracted cell wall. The proportionality factor corresponds to the cell wall contribution (molar %) in the whole cell glycans. See Supplementary Table 6 for glycan quantification in other samples.
A striking finding is the high abundance of mannose (Man) and N-acetylated glucose (GlcNAc) in C. reinhardtii’s cell wall, which has been debated in the literature24,27. While arabinose and galactose are generally presented as the most abundant glycans in HRGPs11,28,29, early work showed that a water-soluble fraction of Chlamydomonas cell wall was rich in mannose30 and, more recently, that GP3 in the cell wall was heavily mannosylated31. The presence of hydroxyproline-rich glycoproteins interacting with mannose-rich lectin-like domains involved in the cell wall of C. reinhardtii has also been demonstrated in cell-differentiated zygotes16. Finally, recent work reported oligo-mannose glycans and 6-O-methylmannose in soluble and membrane-bound proteins of C. reinhardtii27, and two putative mannosyltransferases, consistent with the presence of mannosylation32.
We then assigned (Supplementary Table 7) and quantified the protein content using 2D 13C refocused INADEQUATE spectra and HPLC methods (Fig. 2f and Supplementary Fig. 3, Table 2 and Supplementary Tables 8–9). The most abundant amino acid residues in the cell wall extract are alanines (13%), followed by glutamines and glutamates (9% each), glycines (8%), asparagines and aspartates (6% each). No tryptophan, cysteine, histidine or hydroxyproline residues were detected, their abundance being most likely below the 2% sensitivity threshold of ssNMR. Hydroxyproline (Hyp) was however detected using 2D 13C-13C-PDSD and MAS-DNP low temperature (see below and Supplementary Fig. 4) but could not be quantified. The relative abundance of the amino acids in the cell wall extract determined by HPLC is more sensitive, reproducible throughout different samples (Supplementary Table 10), but consistent with the ssNMR results (Table 2). No cysteine or tryptophan residues were detected by either technique.
Table 2
Amino acid composition of the overall cell wall and specific proteins.
| | | HRGPs | LWGPs |
| Total cell wall | GP1/1.5 | GP2 | GP3 | GP4 | GP5 | GP6 | GP7 |
| %mole | 298/280 kDa | 165 kDa | 140 kDa | ~ 50 kDa | ~ 37 kDa | ~ 25 − 20 kDa | < 20 kDa |
| ssNMR | HPLC | %mole | %mole | %mole | %mole | %mole | %mole | %mole |
Gln | 9.1 | 19.0 (0.3) | 7.2 | 5.6 | 19.5 | 17.3 | 18.7 | 23.0 | 15.3 |
Glu | 8.8 |
Asn | 6.4 | 14.8 (0.3) | 5.6/4.7/9.9 | 7.0/8.1 | 2.5/1.6 | 10.2 | 14.3 | 19.5 | 12.9 |
Asp | 5.9 |
Ala | 13.1 | 10.9 (0.3) | 9.4/10.1/8.4 | 10.5/8.5 | 15.4/10.2 | 8 | 9.9 | 11.9 | 8.7 |
Gly | 8.5 | 8.0 (0.3) | 11.6/6.6/23 | 7.8/8.5 | 10.7/11.0 | 8.2 | 10.3 | 6.5 | 10.8 |
Leu | 8.0 | 5.5 (0.3) | 3.7/2.9/5.1 | 7.5/7.0 | 0.2/7.0 | 10.2 | 4.8 | 4.6 | 8.4 |
Lys | 6.2 | 5.3 (0.1) | 2.7/3.6/4.2 | 3.0/3.0 | 7.1/4.5 | 5.8 | 6.5 | 3.4 | 8.2 |
Ser | 6.2 | 5.0 (0.2) | 14.4/15.8/10.7 | 7.0/8.3 | 8.1/9.3 | 3.3 | 5 | 3.2 | 5.3 |
Pro | 5.8 | 4.8 (0.2) | 3.4/2.7/3.2 | 5.5/7.7 | 0.7/4.5 | 6.6 | 6.1 | 4.3 | 2.7 |
Val | 5.7 | 5.6 (0.3) | 3.4/4.4/3.0 | 7.3/5.7 | 5.3/5.9 | 9.7 | 4.1 | 4.3 | 5.1 |
Arg | 5.1 | 3.5 (0.3) | 1.9/1.7/2.7 | 3.4/3.0 | 3.6/2.1 | 3.8 | 4 | 2.4 | 5.5 |
Thr | 4.3 | 4.7 (0.4) | 5.3/4.2/4.3 | 6.7/6.6 | 7.7/7.4 | 2.5 | 3.9 | 3.8 | 4.3 |
Met | 2.5 | 1.6 (0.2) | 1.1/0.8/0.5 | 1.6/1.0 | 1.4/0.9 | 1.6 | 2.7 | 1.3 | 1.9 |
Ile | 2.2 | 3.1 (0.3) | 3.2/3.3/2.3 | 4.2/1.3 | 2.2/3.2 | 5.9 | 3.1 | 4.0 | 2.9 |
Tyr | 1.5 | 2.4 (0.2) | 1.9/0.9/1.4 | 3.3/3.6 | 2.6/1.2 | 2 | 3.7 | 2.4 | 3 |
Phe | 0.7 | 2.8 (0.4) | 2.2/1.3/2.3 | 4.4/3.8 | 6.4/3.6 | 3.1 | 1.5 | 2.8 | 3.5 |
His | NA | 2.0 (0.1) | 0.6/0.8/1.5 | 0.8/0.2 | 1.0/1.0 | 1.4 | 1.3 | 1.5 | 1.4 |
Hyp | NA | 1.0 (0.1) | 22.4/32.3/15.5 | 14.4/14.7 | 6.6/5.7 | 0.4 | 0.1 | 1.1 | 0.1 |
% SDS-page | | | 5.3 | 7.2 | 9.3 | 10.6 | 24.3 | 35.1 | 8.3 |
| | | 21.8 | 78.2 |
Overall amino acid composition of the cell wall determined by ssNMR and HPLC (values in brackets correspond to standard deviation of three different samples) (see Supplementary Table 10 for replicates), and of specific proteins separated by SDS-PAGE and analyzed by HPLC. Published values are indicated for comparison and underlined18,31. Relative protein proportions (bottom two lines) were evaluated on Coomassie blue stained SDS-PAGE. Amino acids that can be glycosylated are highlighted in orange.
The very low abundance of Hyp detected in whole C. reinhardtii cell walls is surprising, considering that this residue has been described as constituting up to one third of the total amino acid content in HRGPs12,13,21,29,33,34. This result is independent of the strain type (Supplementary Tables 11, Supplementary Fig. 5), growth conditions or cell sexual differentiation (Fig. 2b). While the reported HRGPs are present in our cell wall extracts and contain the expected high Hyp abundance, they only constitute about 20% of the total protein content, based on SDS-PAGE image treatment (Fig. 2b and Table 2). We have identified four new groups of low molecular weight glycoproteins (LWGPs) that we will refer to as GP4 to GP7, which contribute to almost 80% of the amino acids content in our reconstituted cell wall, and have a very low Hyp content. The molecular weights of GP4 (~ 50 kDa) and GP5 (~ 37 kDa) correspond to those reported for the “14-3-3” protein fraction, which relative proportion was shown to vary with the cell state19,20. These two groups of proteins contribute ca. 35% of the amino acids, while the two lighter protein bands, GP6 (~ 25 kDa) and GP7 (< 20 kDa), account for the ca. 45% left. The presence of a greater variety of proteins in our cell wall extracts, without any significant cell damage, confirms that our protocol is milder, but is also an indication that its protein composition is closer to that of the native microalga.
We performed a chemical deglycosylation to remove all glycan units without degrading the proteins. Chemical deglycosylation was preferred over enzymatic treatment since it allows the removal of both O- and N-linked glycosylations35. Interestingly, the deglycosylation of the cell wall extract led to a shift of all SDS-PAGE protein bands, proving that even LWGPs are glycosylated (Fig. 2b, lane 5 vs. 6) with levels of glycosylation consistent with those determined by ssNMR.
Table 2 also reveals that GP1 to GP3 are enriched in Hyp, while the lower weight GP4 to GP7 are enriched in Gln/Glu and Asn/Asp. Hydroxyproline represents only 1% of the total amino acid content in our extracts, which are dominated by Gln/Glu, Asn/Asp, Ala and Gly. Interestingly, with the exception of Gln/Glu, these abundant residues have been reported to play an important role in plant cell walls36. Asparagine is known to be involved in N-linked glycosylation, while alanine-rich and hydroxyproline-containing cell walls, possibly associated with arabinogalactan, have been reported in plants and fungi37. Also, glycine-rich proteins are involved in the formation of β-pleated sheet structures that are important for cell mechanical resistance in plant cell walls38. Altogether, C. reinhardtii probably shares some structural features with plants in terms of amino acid composition of cell wall proteins.
β-sheet domains, protein-glycan and inter-glycan contacts. The protein secondary structure was probed by analysis of the 13C chemical shifts of amino acids. We could find no evidence of α-helices, but we identified almost half of the residues in β-sheet environments, mostly Ile, Lys, Thr and Val. The remaining chemical shifts were compatible with either random coils or polyproline II (PPII) helices39, mostly for Asn/Asp, Gly, Pro and Ser (blue bars on Fig. 2f and Supplementary Table 7). No correlation could be established between the hydrophobicity or charge of the residues, and their secondary structure propensity. Our results do not clearly indicate the presence of PPII that have been reported in HRGPs40,41, but their presence cannot be ruled out.
We then explored the role of glycans in the structuration of cell wall proteins by recording the ssNMR spectra of deglycosylated cell walls. While the complete deglycosylation resulted in a total disappearance of the glycan signals in the 2D-INADEQUATE (Fig. 3a) and DARR spectra (Supplementary Fig. 6), the amino acid chemical shifts remained almost identical, indicating no change in secondary structure of the proteins. However, Thr signals were completely lost following deglycosylation, which could be explained by an increase in Thr mobility after deglycosylation, going from a “slow motion regime” to the ssNMR unfavorable “intermediate regime” (Fig. 3a and Supplementary Fig. 6). This is a strong indication that Thr residues are heavily glycosylated to unidentified glycans in C. reinhardtii cell walls - another information that has never been reported before. Glycans thus appear to have almost no influence on the protein structure and dynamics and, if they are involved in crosslinks, these would mostly occur within glycans, or in isolated regions of the proteinated content of the cell wall. The lack of influence of glycans on the proteins structure moreover suggests a model in which glycans are spatially isolated from the amino acids. This new information must be considered in building a new cell wall model.
We pursued the examination glycan connectivities using dipolar-based ssNMR methods. Multiple through-space glycan-glycan contacts were revealed by the 13C-13C CP-PDSD spectrum of C. reinhardtii’s cell wall (Fig. 3b). For example, we detected inter-mannose contacts between their C2/3 (76 ppm) and C1 of units 9, 13 (94.5 ppm), 2, 10 and 12 (97.2 ppm). We also identified inter-glycan contacts between mannose’s C1 (94.5 and 97.2 ppm) and arabinose’s C1 (109.8 ppm), C2 (82.2 ppm) and C3/4 (77.4 ppm). These units are probably not covalently linked24, but are spatially close.
These room temperature spin-diffusion driven experiments (PDSD and DARR) did not allow detecting specific glycan-protein contacts (Supplementary Fig. 7), probably because of fast relaxation or phase separation in dynamically distinct domains. However, the increased sensitivity provided by hyperpolarization and low temperature enabled the detection of important such contacts by MAS-DNP (Fig. 3c). As C1 carbons have a relatively large chemical shift value and dispersion, they were used to identify glycans in the MAS-DNP spectrum. Since the spectral resolution is reduced by line-broadening at cryogenic temperature, we used the assignment obtained at room temperature, assuming no major chemical shift differences42,43. The nature of these experiments, based on inter-spin dipolar coupling, only provide information on the spatial distribution, through-space contacts, but does not distinguish glycosylation via covalent bonds. We found contacts between Thr (Cγ at ~ 21 ppm) and terminal mannose units 10 or 12 (C1 ~ 95.3 ppm). We also observed contacts between all Hyp carbons (sidechain at ~ 35, 50, 59 and 73 ppm) and both arabinose (C1 ~ 110 ppm) and the terminal mannose units 9 or 13 (C1 at ~ 94.8 ppm). These connectivities might result from the specific geometry of the Man-Hyp contact or reflect a tighter binding with reduced mobility. Other intense glycan C1-amino-acid contacts could be detected for glucose or rhamnose (C1 at ~ 103.5 ppm), or mannose/galactose (C1 at ~ 105.8 ppm), but the limited spectral resolution did not allow determining the amino acid residues involved. Conversely, unidentified glycan C1 atoms showed intense contact with asparagine carbonyl by MAS-DNP PDSD with a long (1.5 s) mixing time (Fig. 3c, bottom).
The sensitivity enhancement provided by MAS-DNP also allowed us to record a 15N-13C ssNMR spectrum (Fig. 3d and Supplementary Fig. 7), and to identify backbone nitrogen signals of amino acids, such as Hyp and Thr (~ 118 ppm) and Asn (~ 112 ppm). The one-bond NCα spectrum revealed a likely different unidentified glycan C1 signal with an intense contact with Asn backbone nitrogen (Fig. 3d) - a contact that was not detected at room temperature (Fig. 3d, purple spectrum). Using an additional 13C-13C mixing time for NCαCx detection, we observed connections between Hyp/Thr backbone nitrogens and several glycans’ C1, including that of arabinose (Supplementary Fig. 8). In summary, we detected unambiguous contacts between Hyp and arabinose, Hyp and mannose, Thr and mannose, as well as Asn with unidentified glycan C1s. The identified glycan-protein contacts and their relative intensities, mostly identified using MAS-DNP DARR spectra, are summarized in Fig. 3e and Supplementary Table 12.
The glycosylation sites can either be sparse and occupied by long oligosaccharides, or dense and occupied with short ones. Considering the large number of non-reducing terminal groups detected in our spectra, the presence of large molecular weight polysaccharides is not favored. The abundance of non-reducing terminal groups, revealed by conventional ssNMR experiments, indicates that most glycosylated amino acids are bound to short oligosaccharides. In C. reinhardtii’s cell wall, the three potential O-glycosylated residues, Ser, Thr and Hyp, represent approximately 10% of the overall amino acid content. However, no glycan contacts were detected with serine residues either by ssNMR and MAS-DNP, therefore few serines, if any, are involved in glycosylation. The percentage of amino acids available for glycosylation is thus ca. 6%, which lead on average to the presence of 3 to 5 sugar units per glycosylated amino acid in the cell wall extract. This number can be reduced to ~ 2 if the N-linked glycosylation of Asn is considered (which represents 6% amino acids).
To check the N-linked glycosylation hypothesis, we employed GC-MS analysis of partially methylated alditol acetates identifying mannose linkages (1,2 and 1,3) compatible with mannose oligosaccharides attached to proteins via N-linked glycosylation24. We note that in addition to favorable linkages, mannose-rich oligosaccharides are more likely to be present in N-glycosylated proteins44
These facts have two important implications. First, the high abundance of mannose can be explained by its role in glycosylation. Second, C. reinhardtii’s cell wall architecture is comprising of O-glycosylation, as previously demonstrated28,45, but also N-glycosylation seems to play an important role, at least in the abundant lighter glycoproteins detected. To combine those two consequences, we thereby hypothesize that at least some of the LWGPs GP4 to GP7 are heavily N-glycosylated via their Asn residues decorated with N-acetylglucosamine (GlcNAc) that are linked to short mannose-rich oligosaccharides. Our results suggest a model in which glycans are spatially isolated from the amino acid residues, and we postulate the presence of abundant but short mannose-rich oligosaccharides. We confirmed the O-glycosylation of Hyp to arabinose by MAS-DNP, in agreement with previous literature on HRGP45,46. Two new glycosylation bonds were revealed in LWGPs: an N-glycosylation of Asn with GlcNAc and oligo-manoside, and an O-glycosylation of Thr to unidentified glycans. Finally, we also report new potential intermolecular, non-covalent, contacts between Hyp and Man/Gal, and between Thr and Man.
Rigid/hydrated glycans vs. mobile/less hydrated proteins. Hydration heterogeneity in the cell wall is an essential feature for land plants47. This has been poorly investigated in microalgae and here we explored the water distribution in C. reinhardtii’s cell wall. To determine the amino acids or glycans tightly bound to water, we carried out a control and a water-edited 2D 13C-13C DARR experiments48 (Supplementary Fig. 9). The results are analyzed in terms of water hydration and are shown in Fig. 4a and Supplementary Table 13. The hydration calculations (see experimental section) indicate that water is more tightly bound to glycans than to amino acid residues. Mannose and galactose with low C1 chemical shifts, probably corresponding to terminal units, are more hydrated than other glycans. This superior hydration of glycans correlates with faster MAS-DNP hyperpolarization build-up (Supplementary Fig. 10 and Supplementary Table 14), which may indicate proximity with the biradicals dispersed in the water phase. Moreover, the residues that can be glycosylated, namely Hyp, Thr and, to a lesser extent, Asn and Ser, are hydrated to much higher levels close to those of glycans’. However, acetylated glucose units (GlcNAc) are less hydrated, suggesting that they are involved in structural domains where water is either absent or weakly bound. This is compatible with their implication in N-glycosylation of asparagines, which are also less hydrated than other glycosylated amino acids. Since we found that serines are hardly hydrated, we confirm that they are probably not involved in glycosylation in C. reinhardtii cell wall (Fig. 4a).
Altogether, these results help us formalize the macromolecular organization in the cell wall, with “dry” globular protein cores surrounded by hydrated layers containing glycans (Fig. 4a). Both regions would be linked through glycosylation processes, implying that the protein surfaces would contain glycosylated amino acids, mostly Hyp, Thr and Asn. The significant hydration of glycans is probably an integral part of the high plasticity of the cell wall, as observed in higher plants49,50. Water can be understood as playing a double role, first as a plasticizer, but also providing the flexibility to ensure glycan interactions that are necessary for the bridging of structural proteins51. In higher plants, it is for example required to form nanostructures in secondary cell walls, or at the hemicellulose-cellulose interface52.
We then probed the dynamics properties of the chemical site by using two different 1H-to-13C polarization transfer schemes by 1D ssNMR, the Cross-Polarization (CP) that enhances the sensitivity of the rigid molecular segments (CP), and the Insensitive nuclei enhancement by polarization transfer (INEPT)25,53 for dynamic segments. Despite the low resolution of the 1D spectra, especially in the carbohydrate region, 44 distinct carbon resonances could be identified (Supplementary Fig. 11), i.e., 20 from glycans and 24 from amino acids (Supplementary Table 15). Among glycans, those with low C1 chemical shifts, involved in terminal groups, were particularly favored by INEPT, for example mannose (Man11/2/10), xylose (Xyl11/2), and galactose (Gal11). Terminal mannose and galactose units are therefore both highly hydrated and very mobile as compared to other polysaccharides, suggesting their presence at the water-amino acid interface. Even among relatively rigid segments, differences could be observed by monitoring the 13C CP build-up and decay in the cell wall (Supplementary Fig. 12a and Supplementary Table 16). The CP build-up was faster for glycans, indicating a strong dipolar-coupling network, comparable to crystalline starch (Supplementary Fig. 12b and Supplementary Table 17). The CP decay on the other hand was slower for glycans, indicating less motion on the millisecond regime. Polarization transfer efficiency also explains why amino acids signals, including Hyp cross peaks, were preferentially detected by CP-PDSD, or at low temperatures (< -160°C) (Supplementary Fig. 2), most likely due to their unfavorable dynamics or relaxation rates54.
Motional differences could be more quantitatively determined by measuring 1H and 13C longitudinal relaxation times in the laboratory (T1) and the rotating frame (T1ρ). While T1 probes motions with correlation times on the order of ns, T1ρ is sensitive to slower (ms) motions. No differences were observed in the 1H T1 values (Supplementary Fig. 13), indicating that 1H polarization is homogenized throughout the sample by spin diffusion, implying up to several hundreds of nanometer homogeneity. On the other hand, carbohydrate 13C T1 as well as 1H and 13C T1ρ values were slightly longer than those of amino acids, showing that glycans are less mobile than proteins, both on nanosecond and millisecond scales. As an example of these clear discrepancies between glycan and proteins, 13C T1 are shown in Fig. 4b, while 1H and 13C T1ρ values are reported in Supplementary Fig. 13. Comparing relaxation times to published values measured in other contexts permits to refine our model. For example, since 13C T1 values are below 1.5 s on the 14.1 T magnet (600 MHz), glycans are unlikely present in large crystalline regions55,56. Also, even if 1H T1ρ values measured under similar conditions (60 kHz effective spin-lock) did not reach the 70 ms of crystalline cellulose57, the average value of 30 ms measured here falls into the range of amorphous or surface domains of cellulose, or that of semi-mobile hemicellulose such as xylan58,59 involved in an organized and significantly rigid biopolymer. On the other hand, the more dynamic amino acids in our sample, with 5–10 ms 1H T1ρ, confirm the predominance of motion on the ms scale, leading to a biomaterial similar to fungal cell wall60. We note that by using long mixing times as well as T1ρ filters, the glycan and protein parts of the cell wall can be selectively detected (Supplementary Fig. 14).
In conclusion, proteins and glycans differ in both dynamics and hydration, with mobile protein domains, more rigid but hydrated glycan domains, and glycosylated proteins at the interface. These two families of molecules are spatially segregated, even if it is in small nanometer-scale regions that are chemically and dynamically heterogeneous, similarly to layered semi-crystalline polymers or plant cell walls61.