The extraction and purification of biosilica from diatom cells is necessary for the selective investigation of the silica and its tightly associated biomolecules. To this end, the studied C. cryptica cells were treated with an SDS/EDTA containing buffer solution. After this lysis buffer treatment, only tightly silica-associated organic material remains. A comparably massive insoluble organic matrix is characteristic for C. cryptica.32 It makes up about 25–40 wt.-% of the biosilica as could be determined by measuring the mass difference between the purified biosilica and the organic matrix remaining after de-silicfication of the cell walls. This amount of organic matrix is up to ten times higher than the values reported previously for other diatom species like S. turris or T. pseudonana.17,33
The solid-state 13C cross polarization (CP)/MAS NMR spectrum of isotope-enriched diatom biosilica reveals various, strongly overlapping and relatively broad signals (Fig. 1) because the silica-associated organic matrix is a mixture of various components (see above). Nevertheless, different functional groups can be identified by their characteristic 13C chemical shifts. The intense signal at 175 ppm is due to the carbonyl groups of proteins and carbohydrates. The signals between 120–140 ppm are characteristic for aromatic carbons especially in the aromatic amino acids of silica-attached proteins. An amino acid analysis was performed additionally to characterize the overall composition of biosilica-attached proteins (cf. ESI, Figure S12). Most alkyl signals appear in a chemical shift range of 10–40 ppm. N- and O-alkyl signals are found between 40–90 ppm mainly due to the presence of long-chain polyamines (LCPAs) and carbohydrates. One prominent signal group appears around 105 ± 5 ppm. This region is characteristic for C-1 atoms of polysaccharides including chitin. In general, the recorded spectrum shows similarity to the spectra of biosilica from other diatom species.17 However, the massive organic matrix of the studied species results in high signal intensities for different organic compounds, and especially for the carbohydrates.
The directly excited 29Si MAS NMR spectrum of isotope-enriched diatom biosilica shows three characteristic signals (Figure S1). They are assigned to Si(OSi)n(OH)4−n moieties with n = 2,3,4, denoted as Qn-groups.34 The strong signal intensity of Q3- and especially Q4-groups confirms the high condensation state of the silica phase, as it is characteristic for biosilica and other amorphous silica compounds.17 In addition, a 15N CP/MAS NMR spectrum was recorded (Figure S2). An intense, broad signal at 120 ppm is assigned to amide nitrogen atoms characteristic for proteins. The signals between 20–60 ppm are due to amines in LCPAs and the nitrogen atoms of lysine side chains.
As also indicated in the above-discussed 13C NMR spectrum, chitin is present in the biosilica extracted from C. cryptica as reported previously27 similar like in T. pseudonana.31 The presence of silica-associated chitin was shown in C. cryptica using a GFP (green fluorescent protein)-tagged fusion protein with a chitin-binding domain.27 To isolate chitin, the biosilica was dissolved and the remaining insoluble organic matrix (Figure S9) was treated with NaOH. This treatment removes the other biomolecules whereas the alkali-resistant chitin remains unchanged. The recorded 13C NMR spectrum shows all characteristic NMR signals for chitin (Fig. 2).35,36 Small but significant differences in chemical shift and line width of several signals confirm the presence of two chitin conformations in the sample (Table 1, Figure S3 and S4). The extremely narrow signals and their chemical shifts are characteristic for the highly ordered β-chitin. It is well-known that C. cryptica forms extracellular β-chitin fibers.28–30 Minor amounts of such chitin fibers were indeed found in our samples (Figure S8 and S9) and are probably responsible for the set of narrow 13C NMR signals. In addition, a second set of signals is observed. These signals exhibit clearly different chemical shifts especially for C-1 and C-4 (cf. inserts in Fig. 2). The C-1 signal (around 105 ppm) splits into two signals. The narrow signal at higher chemical shift is due to β-chitin. The broader signal exhibits a lower chemical shift indicative for a conformation similar to α-chitin. Both signals split into doublets due to 13C-13C spin-spin coupling in this isotope-enriched sample (cf. Figure S4). The coupling constant amounts to 44 Hz for both C-1 signals. The C-4 signal splits into a triplet with 36 Hz coupling constant for both chitin conformations. In summary, the second set of signals differs from β-chitin and exhibits chemical shifts and linewidths similar like α-chitin. However, the partial transformation of β-chitin fibers into another conformation by the used extraction procedure can be ruled out.37 Significantly higher NaOH concentrations and temperatures would be necessary to modify β-chitin. The α-chitin-like set of signals is thus attributed to the silica-associated chitin in analogy to the chitin-based meshworks observed for T. pseudonana previously.31
Table 1
Chemical shifts for the two observed chitin conformations in the 13C NMR spectrum. For comparison, previously published reference data are provided.33 Estimated shifts for strongly overlapping signals with supposedly larger experimental errors are marked with *.
| | δ (13C) / ppm |
| | C = O | C-1 | C-4 | C-5 | C-3 | C-6 | C-2 | CH3 |
C. cryptica | “α-like” chitin | 174.2 | 104.2 | 81.9 | 75.8* | 73.5* | 61.4 | 55.4 | 23.9* |
| β-chitin fibers | 175.0 | 105.3 | 84.4 | 75.4 | 73.0 | 59.8 | 55.8 | 23.4 |
References33 | α-chitin | 174.2 | 104.4 | 83.3 | 76.0 | 73.6 | 61.2 | 55.3 | 23.1 |
| β-chitin | 175.8 | 105.3 | 84.4 | 75.4 | 73.0 | 59.8 | 55.6 | 22.7 |
Some additional signals with relatively low signal intensity can be observed, e.g., at 34 ppm or 128 ppm. These signals remaining after NaOH treatment indicate biomolecules cross-linked to the chitin, most likely proteins. Such cross-linking between chitin and other biomolecules is well-known in nature and was also found for chitin in different biominerals previously.38–41
A monosaccharide analysis was performed to further characterize the carbohydrate composition of the silica-associated organic matrix using gas chromatography coupled with mass spectrometry (GC-MS, Table 2). Note that these monosaccharides will mainly be present as polysaccharides in the native samples. Sample pretreatment for analysis cleaves polysaccharides into the monomer units. Mannose and glucose are the main monosaccharide components in C. cryptica and its insoluble organic matrix. This is similar to previous observations on other diatom species where high amounts of both monosaccharides were found as well.42,43 For the isolated insoluble organic matrix, the monosaccharide glucosamine shows a similarly high concentration as mannose and glucose. Note, that N-acetyl-glucosamine, the monosaccharide unit of chitin, is not stable under the chosen hydrolysis conditions. It is de-acetylated and contributes to the glucosamine signal. To prove the presence of N-acetyl-glucosamine, chitinase from Trichoderma viride is used for bond cleavage of chitin instead of hydrochloric acid. The detection of the N-acetyl-glucosamine as monosaccharide unit of chitin was performed with GC-MS (Figure S10). The presence of high amounts of N-acetyl-glucosamine is detected which supports the above-discussed NMR results. This observation is remarkable because significant amounts of glucosamine and N-acetyl-glucosamine are unusual in diatom biosilica and were just found for the diatom species Triceratium dubium before.42
Table 2
Monosaccharides in C. cryptica biosilica and its insoluble organic matrix determined using GCMS. The recorded signal intensities are related to the signal intensity of the internal standard myo-inositol: – = not present; + = 1.0–5.0%; ++ = 5.1–10.0%; +++ = 10.1–20.0%; ++++ = 20.1–50.0%.
Monosaccharide | C. cryptica biosilica | C. cryptica insoluble organic matrix |
Rhamnose | ++ | + |
Ribose | +++ | +++ |
Fucose | +++ | + |
Xylose | +++ | +++ |
Mannose | ++++ | +++ |
Glucuronolactone | ++ | − |
Galactose | +++ | ++ |
Glucosamine | +++ | ++++ |
Glucose | ++++ | ++ |
Moreover, some additional monosaccharides can be observed at high retention times especially for the insoluble organic matrix. These monosaccharides are pyranoses which cannot be assigned to a specific monosaccharide with our detection method and spectra library. Phosphorylation of these monosaccharides as it was observed for S. turris previously43 can be excluded here by the recorded mass spectra. As an example, one of those mass spectra is shown in Figure S11. The mass spectra show some indication for cross-linking of the monosaccharide components. This meets the NMR results very nicely, where even after NaOH treatment some additional signals confirm the presence of chitin-bound biomolecules different from N-acetyl-glucosamine (cf. Figure 2).
Two-dimensional (2D) NMR experiments can visualize homonuclear as well as heteronuclear correlations. Proton-driven spin-diffusion (PDSD) experiments show 13C-13C correlations between neighboring carbon atoms. PDSD experiments were recorded at mixing times of 30 ms and 120 ms (Fig. 3). The application of a longer mixing time visualizes correlations over longer distances. For example, correlations extended over a whole monosaccharide unit are then observable. In the chemical shift region typical for most proteins, numerous correlations between different amino acids are found. A high amount of proteins is obviously associated to the biosilica. Following the results of the NMR experiments on the insoluble organic matrix, a closer look to the chemical shifts characteristic for carbohydrates clearly shows the presence of the characteristic chitin spin system (e.g. for C-1 at 105 ppm) in the biosilica sample. The occurrence of two signals with only minor differences in the chemical shift for example at the carbon positions C-1 and C-4 confirms the presence of two chitin conformations discussed above (cf. Table 1). Since the presence of a chitin-based meshwork is already known for T. pseudonana, the chitin spin systems for both species are compared in the ESI (Figure S5).
Moreover, a correlation between the signals at 83 ppm and 78 ppm is observed which cannot be assigned to a known spin system. However, this correlation within the chemical shift range of carbohydrates supports the possibility of cross-linking between different carbohydrates or other biomolecules as discussed in the chitin spectrum above. In studies of T. pseudonana carbohydrates cross-linked to the protein silaffin-2 were found previously.44,45 Taken together, the discussed experiments as well as a comparison to literature show strong indication for carbohydrate cross-linking which may be a topic of future analyses.
Information on possible silica-biomolecule contacts can be obtained from the 13C{29Si} REDOR experiment. Figure 4 displays 13C spectra acquired at 10 ms echo time. One spectrum represents the overall signal intensity S0 while the other spectrum shows the reduced signal intensity Sr acquired with 29Si dephasing. For two signals at 43 ppm and at 56 ppm, remarkable differences between both spectra, i.e., high REDOR fractions are observed. The signal at 43 ppm can be assigned to LCPAs and/or lysine sidechains. The lysine can significantly contribute to this signal especially because we used a CP-transfer from neighboring 1H nuclei.20 However, inspection of the results of the amino acid analysis (cf. Figure S12) reveals only a small amount of lysine present in the sample in comparison to other amino acids. The relatively low overall amount of lysine gives rise to the assumption that LCPAs may be the major component contributing to this signal. For the signal at 56 ppm, overlap of different signals is likely. The C-2 position of carbohydrates and the C-α atoms of proteins can be found at this shift. However, LCPAs can also contribute to the 56 ppm-signal significantly. This is further confirmed in 15N-13C correlation experiments which are discussed in a related paper published in the same journal issue. The analysis of the REDOR data for all echo times between 2 and 20 ms yields the REDOR curves shown in Fig. 4 and Figure S7. These curves allow to calculate heteronuclear dipolar second moments (Eq. 1). For the signal at 43 ppm, data up to 10 ms echo time are used for the first-order approximation resulting in a second moment of 1.3·104 s− 2. For the signal at 56 ppm, data points up to 14 ms echo time can be used for the first-order approximation providing a second moment of 0.8·104 s− 2. These observations indicate a close proximity between 29Si spins and the 13C spins at 43 and 56 ppm. These second moments are smaller but comparable to the values determined for synthetic silica-LCPA composites containing one defined LCPA. A second moment of ca. 2·105 s− 2 is then detected.21 The smaller second moment calculated for the 56 ppm signal is not necessarily due to a weaker dipolar coupling between 13C spins and neighboring 29Si spins. As already mentioned, an overlap of different signals at this chemical shift is very likely. Since some of the corresponding compounds may be far away from silica, this would lead to an underestimation of the calculated second moment. That means, the calculated second moments are probably underestimated for both signals at 43 and 56 ppm.
For the amorphous biosilica and its unknown spin system geometry, distance calculation is not possible. However, a comparison with former results shows similarities between the different diatom species. High REDOR fractions and thus a proximity between 13C and 29Si were observed in the same typical chemical shift range of LCPAs for T. pseudonana biosilica.19 Higher REDOR fractions at smaller echo times were observed for T. pseudonana than for C. cryptica. These differences are probably due to the different organic matrix of both species. The massive organic matrix of C. cryptica biosilica can cause stronger signal overlap and thus lead to a more pronounced underestimation of the heteronuclear dipolar second moment. However, the occurrence of REDOR effects in the same chemical shift range for the different diatom species indicates that the close contact between LCPAs and silica may be a general property of diatom biosilica.