Evaluation of ring-5 structures of guaiacyl lignin in Ginkgo biloba L. using solid- and liquid-state ¹³C NMR difference spectroscopy

Preparation of ginkgo lignin specifically ¹³C-enriched at G5

The lignin in ginkgo xylem cell walls was specifically ¹³C-enriched at the G5 position by using a previously described method (Terashima et al. 2002) with some modifications. UE-coniferin and [G5-¹³C]-coniferin were synthesized according to a previously reported method (Terashima et al. 2003). Cut ginkgo shoots (3 years old, 20–25 cm in axial length) were placed into small vials containing an aqueous solution of either [G5-¹³C]-coniferin or UE-coniferin. After feeding the solution (400 mg of coniferin in 100 ml of H₂O per shoot) in a growth chamber for 5 days under 14 h at 27°C:10 h at 20°C as the light:dark period, the shoots were grown for an additional 3 weeks in flasks containing tap water. Then, the bark was removed, and the newly formed xylem was collected and extracted with acetone for 4 h and hot water for 1 h, and the xylem sample was stored in a desiccator.

Structural unit composition and content of lignin

To estimate the structural unit composition of the ginkgo lignin, thioacidolysis (Nimz 1974; Lapierre et al. 1985) was performed according to the method reported by Roland et al. (1992). Approximately 3 mg of the extracted wood sample was placed in 5 ml of a dioxane/ethanethiol mixture (8.75:1, vol./vol.) containing 0.2 M (C₂H₅)₂OBF₃ in a test tube. As an internal standard for gas chromatography-mass spectrometry (GC-MS) measurements, 1.0 ml of 0.1 mg/ml docosane/dioxane solution was added to the tube. The thioacidolysis was conducted at 100°C for 4 h. Then, the reaction mixture was cooled on ice and 25 ml of 0.4 M NaHCO₃ aqueous solution (aq.) was added to stop the reaction. The pH of the mixture was adjusted to 2–3 with HCl aq. (1:3. vol./vol.), and the mixture was extracted using 3 ml of CH₂Cl₂ 3 times. The thus obtained organic extract was dried over Na₂SO₄ and concentrated under reduced pressure at 40°C. The final solution was dissolved in 0.3 ml of CH₂Cl₂ and stored at 4°C until use. The silylated thioacidolysis product was analyzed by GC-MS (Trace 1300GC/ITQ 900, Thermo Fisher Scientific K.K., Tokyo, Japan) under the following conditions: capillary column, Rtx-1ms (30 m×0.32 mm ID, 0.25-μm film thickness); temperature program, 180–230°C at 2°C/min; carrier gas, He (1.5 ml/min). The monomeric products were assigned and evaluated by GC according to Lapierre et al. (1991).

The lignin content was estimated by the acetyl bromide method (Iiyama and Wallis 1990). The ball-milled ginkgo wood (see following text) was added into a mixture of 2.5 ml of 25% acetyl bromide/acetic acid solution. After the addition of 0.1 ml of 60% perchloric acid, the solution was stirred at 70°C for 30 min. Then, the mixed solution was cooled with ice water. The solution was poured into a mixture of 5 ml of 2 M NaOH aqueous solution and 13 ml of acetic acid. The whole solution was diluted to 50 ml using a volumetric flask. The absorbance at 280 nm was measured using a ultraviolet-visible (UV-Vis) spectrophotometer (V-530, JASCO, Tokyo, Japan), and the lignin content in the solution was evaluated. The resulting lignin content was 34.8 wt% for the G5-¹³C-enriched sample.

Determination of the ¹³C abundance in the newly formed xylem

The ¹³C incorporation in the ball-milled ginkgo wood containing G5-¹³C-enriched lignin was evaluated using isotope ratio mass spectrometry (IR-MS; Delta Plus, ThermoFinnigan, San Jose, CA, USA) connected to an elemental analyzer (NC2500, ThermoFinnigan, San Jose, CA, USA). ¹³C/¹²C at the G5 position, (¹³C/¹²C)_G5, was calculated using the following equation:

(13C/12C)sample=0.348×1/10×(13C/12C)G5+(1–0.348×1/10)×(13C/12C)control

where 0.348 is the resultant lignin content evaluated by the acetyl bromide method as described earlier, 1/10 is the specific carbon ratio at the position concerning the guaiacyl unit having a C₁₀ structure, (¹³C/¹²C)_sample is the obtained value for the G5-¹³C sample, and (¹³C/¹²C)_control is for the UE sample. The experiments were performed in triplicate, and the average value was used to estimate the ¹³C ratio (%).

Sample preparation for NMR measurements

The ¹³C-enriched and UE ginkgo xylem samples were ground in a Wiley mill and passed through a 24-mesh screen and then sequentially extracted with acetone for 4 h and hot water for 1 h. Wood meal (1 g) and zirconia balls (5-mm diameter, 100 g) were placed into a zirconia jar. The wood meal was finely ball-milled using a Fritsch Pulverisette 6 planetary ball-mill (FRITSCH, Idar-Oberstein, Germany) at 24.7 g for 6 h (grinding for 2 min, waiting for 2 min, 15 cycles×12 sets with 20-min intervals).

The EL samples were prepared as follows: Finely milled wood (1 g) was immersed in 5 ml of 7 wt% LiCl/N,N-dimethylacetamide and allowed to stand at 40°C for 24 h, and then 25 ml of pH 4.8 acetate buffer and 50 mg of meicelase (cellulase originating from Trichoderma viride, Meiji Seika Pharma Co. Ltd., Tokyo, Japan) were added to the resultant swollen mixture. The mixture was shaken at 40°C for 72 h. During the shaking period, additional 50-mg portions of meicelase were added at 24 h and 48 h after the start of the reaction. The resultant EL samples were washed with H₂O by centrifugation and freeze-dried.

The EL samples were acetylated as follows (Lu and Ralph 2003): 100 mg of EL was added to a mixture of 2 ml of dimethylsulfoxide (DMSO) and 1 ml of N-methylimidazole and stirred at 25°C for 3 h under an Ar atmosphere. Then, 0.6 ml of acetic anhydride was added into the solution, and the mixture was stirred at 25°C for 24 h. The resultant homogeneous solution was poured into excess ice-cold H₂O. The acetylated EL (ELAc) samples were washed with H₂O repeatedly and dried in vacuo at 40°C for 48 h.

Hereafter, the samples containing ¹³C-enriched lignin are described as G5-EL or G5-ELAc, and the samples containing UE lignin are described as UE-EL or UE-ELAc.

Solid-state ¹³C CP/MAS NMR

Solid-state ¹³C cross polarization/magic angle spinning (CP/MAS) NMR measurements were conducted using a JEOL ECA-700 spectrometer (JEOL, Ltd., Tokyo, Japan). Dried samples were filled into 4-mm zirconia sample tubes and measured under the following conditions: ¹³C resonance frequency, 175 MHz; magic angle spinning rate, 10.0 kHz; scan number, 17 000; acquisition time, 14.54 ms; repetition time, 5.0 s; contact time, 3, 4, 5, 6 or 7 ms. The obtained free induction decays (FIDs) were multiplied with an exponential window function using a line broadening of 200 Hz prior to Fourier transform. The chemical shifts were calibrated using the methoxy carbon as an internal standard (56.0 ppm). Phase and baseline corrections were performed using the Delta software (v5.0.5.1, JEOL). The room temperature was 293 K, and the actual sample temperature in the 10.0 kHz spinning sample tube was estimated to be 317.5 K (r.t.+24.5 K) using samarium acetate (Campbell et al. 1986).

Liquid-state quantitative ¹³C NMR

Liquid-state quantitative ¹³C NMR measurements were carried out on a Bruker Avance 600 spectrometer equipped with a cryo probe (Bruker BioSpin GmbH, Rheinstetten, Germany) using an inverse-gated decoupling mode. The acetylated samples were soluble in CDCl₃. The measurement conditions were as follows: ¹³C resonance frequency, 150 MHz; sample concentration, 100 mg/ml; solvent, CDCl₃ containing 0.01 M Cr(III) acetylacetonate as a relaxation reagent (Gansow et al. 1972; Rokhin et al. 1994; Xia et al. 2001); scan number, 43 000; acquisition time, 1.4 s; pulse delay, 1.7 s; temperature, 295 K.

NMR data processing

For solid-state NMR, five spectra were obtained for each EL sample with contact times of 3, 4, 5, 6 and 7 ms. The spectrum of the UE sample was subtracted from the spectrum of the G5-enriched sample after normalization using the methoxy carbon signal. Gaussian peak fitting by the least squares method was carried out using the solver function in Excel 2010 software (Microsoft, Redmond, WA, USA). The resultant peak area variation was plotted vs. the contact time, and linear approximation was calculated using Excel 2010 software.

For liquid-state NMR, the obtained quantitative ¹³C NMR spectra were exported to ASCII data using the Bruker Topspin software (ver. 4.0, Bruker BioSpin GmbH, Rheinstetten, Germany). The spectrum of UE-ELAc was subtracted from the spectrum of G5-ELAc after normalization using the methoxy carbon signal. The peak areas were integrated using Excel 2010 software.

Preparation of specifically ¹³C-enriched ginkgo lignin

The preparation and analyses of the specifically ¹³C-enriched ginkgo lignin are schematically illustrated in Figure 1. The ¹³C abundance was estimated by IR-MS, and the resultant ¹³C-enrichment ratio at G5 was 2.72%, which is approximately 2.5 times the natural abundance.

Figure 1:

Schematic illustration of the preparation and analyses of specifically ¹³C-enriched ginkgo lignin.

Solid-state NMR measurements and signal assignments of the lignin G5 structures

The solid-state ¹³C CP/MAS NMR spectra of UE-wood meal, G5-EL and UE-EL as well as a difference spectrum (between G5-EL and UE-EL) are shown in Figure 2. The aromatic carbon signals in the spectrum before enzymatic saccharification (Figure 2a) were relatively weak. After the enzymatic saccharification (Figure 2b), these signals were enhanced. The signals in the range from 50 to 110 ppm derived from residual polysaccharides were still detected, probably because of the imperfect removal by the present saccharification procedures.

Figure 2:

Solid-state ¹³C CP/MAS NMR spectra of (a) UE-wood meal, (b) UE-EL, (c) G5-EL, and (d) the difference between (b) and (c).

The arrows indicate the methoxy carbon at 56 ppm. The gray line is the ±0 line resulting from spectral subtraction. Contact time=5 ms.

Although the spectra of UE-EL (Figure 2b) and G5-EL (Figure 2c) are similar, the difference spectrum (Figure 2d) indicates that there are several enhanced signals assignable exclusively to the G5 carbon between 110 and 160 ppm. The gray line is the ±0 line, and the arrows indicate the methoxy signals. The downfield signals (>160 ppm) may be spinning side bands derived from the aromatic carbons. In this study, the magic angle spinning rate was at 10 kHz, and the spinning side band positions should be ±57.1 ppm. At this point, the upfield signals (<50 ppm) showed a nearly flat baseline in the difference spectrum, and we believe that the spinning side bands of the residual polysaccharides are negligible in the aromatic region, although the spinning side bands of the aromatic carbons should overlap in the polysaccharide region. The possible structures assigned exclusively to the G5-¹³C signals in the 110–160 ppm region are uncondensed guaiacyl rings with β-O-4, α-O-4, β-β and β-1 linkages, and condensed structures with β-5, 5-5 and 4-O-5 linkages (Figure 3).

Figure 3:

(a) Solid-state ¹³C CP/MAS NMR difference spectrum between UE-EL and G5-EL, (b) the result of Gaussian peak fitting of (a), and (c) the solid-state ¹³C CP/MAS NMR difference spectrum between UE-ELAc and G5-ELAc.

The arrows in (b) and (c) suggest the peak maxima in (b). The dashed arrow in (c) indicates the missing carbon. Contact time=5 ms.

These signals were separated into 6 G5-¹³C signals (signals a, b, c, d, e and f) and a polysaccharide signal (signal s, corresponding mainly to the polysaccharide C1) by Gaussian peak fitting (Figure 3b) (Atalla et al. 1980). The difference spectrum between the G5-¹³C-enriched and UE dehydrogenated polymers showed no signals in this region (near s) (Parkås et al. 2004). The plotted curve with circles is the fitting total. For these signals, the signals were assigned according to previous reports using solid- and liquid-state NMR measurements (Nimz et al. 1981; Haw and Schultz 1985; Hatcher 1987; Hatfield et al. 1987; Drumond et al. 1989; Xie and Terashima 1991; Hawkes et al. 1993; Alves et al. 2000; Parkås et al. 2004; Ralph and Ralph 2009, 2010; Li et al. 2016; Yue et al. 2016).

Signals a and b were assigned to the uncondensed G5 carbons (C4-OH or C4-OR). Signal c was determined to be an overlapping signal of the G5 carbons in β-5 [(c)-1] and 5-5 linkages with a neighboring C4-OH [5-5 (OH), (c)-2] moiety. Signal d was assigned to the G5 carbons involved in 5-5 linkages with a neighboring C4-OR [5-5 (OR), including 5-5 in the dibenzodioxocine unit]. Signals e and f were assigned to the G5 carbons of the 4-O-5 linkages having a neighboring C4-OH [4-O-5 (OH)] or C4-OR [4-O-5 (OR)] unit.

The thioacidolysis and GC-MS measurements showed that the obtained sample did not produce thioethylated H (p-hydroxyphenyl)-unit monomers (data not shown); however, previous studies suggested the presence of condensed H-units in the compound middle lamella and cell corner regions (Terashima and Fukushima 1988; Saito and Fukushima 2005) in gymnosperm lignin. Thus, although the neighboring unit in Figure 3 might be a guaiacyl and/or p-hydroxyphenyl moiety in the gymnosperm lignin, the difference should not affect these signal assignments.

To clarify the signal assignments, the difference spectrum between the acetylated G5-EL (G5-ELAc) and acetylated UE-EL (UE-ELAc) samples is shown in Figure 3c. Comparing these difference spectra before (Figure 3b) and after (Figure 3c) acetylation, we found that signal b at 116 ppm disappeared in Figure 3c. From this result, we determined that signal b was derived from the uncondensed G5 carbon having a neighboring C4-OH group [uncondensed (OH)]. Similarly, the peak shape of signal c changed slightly after acetylation, as indicated by the arrow. Here, we found that the peak shape of signal d was seldom changed by acetylation, indicating that this signal corresponds to 5-5 linkages (OR). From these results, we concluded that the overlapping signal in c was 5-5 (OH). Additionally, we surmised that signal e should originate from 4-O-5 (OH) because of the change in the peak shape upon acetylation.

As a consequence of the aforementioned discussion, the G5 carbon signals in the solid-state ¹³C NMR spectrum were assigned, and these assignments along with previously reported assignments are listed in Table 1 (Nimz et al. 1981; Haw and Schultz 1985; Hatcher 1987; Hatfield et al. 1987; Drumond et al. 1989; Xie and Terashima 1991; Hawkes et al. 1993; Alves et al. 2000; Parkås et al. 2004; Ralph and Ralph 2009, 2010; Li et al. 2016; Yue et al. 2016, see Supplementary Table S1 for details). Although the sample state was different, these signal assignments agreed in principle with those of the liquid-state ¹³C NMR spectrum (see following text). After acetylation, the signals were shifted, overlapped, broadened and more complex. Some signals were not determined after acetylation. Therefore, in this study, we focused on the results obtained from the G5-EL and UE-EL samples.

Evaluation of the G5 carbons by solid-state NMR

In the solid-state ¹³C CP/MAS NMR measurements, the signal intensity is not proportional to the molar abundance. For the quantitative evaluation by solid-state ¹³C NMR, the direct polarization method can be used if the signal intensity is sufficiently strong. However, in the cases of lignin, the signals were usually too weak to conduct an adequate number of scans within practical measurement times. To quantitatively evaluate the lignin G5 carbon signals using solid-state NMR, in this study, we used the contact time extrapolation method (Ottenbourgs et al. 1998). The following section will briefly explain this method.

In solid-state NMR, the ¹³C signal sensitivity was compensated by the CP/MAS method. In the CP/MAS sequence, the ¹H magnetization was transferred to ¹³C, and the transfer efficiency strongly depends on the environments of the proton and the carbon. Therefore, the obtained signals are not quantitative. The simplified evolution of carbon magnetization as a function of the contact time can be expressed as eq. (1).

where M is the experimentally obtained magnetization value (peak area), M₀ is the theoretical maximum magnetization value, T_CT is the contact time, T_CH is the cross-polarization time constant and T_1ρ^H is the proton relaxation time constant in the rotating frame. When T_CH<<T_CT, the term [1−exp(−T_CT/T_CH)] can be approximated to 1, and eq. (1) can be simplified to eq. (2).

Here, T_1ρ^H is the intrinsic constant of the structure, and we can draw a line in proportion to T_CT. As a result, the extrapolated intercept (T_CT=0) is the theoretical maximum magnetization value, and we can evaluate the signal quantitatively.

The T_CH of the aromatic signals of gymnosperm lignin were previously reported as T_CH≤0.46 ms (Hatfield et al. 1987). If T_CT>5T_CH, the term [1−exp(−T_CT/T_CH)] is larger than 0.99 and can be approximated to 1. Considering these points, we used the measurement parameters of contact times of 3, 4, 5, 6 and 7 ms in solid-state ¹³C CP/MAS NMR measurements. The obtained ¹³C NMR spectra were processed in the same way as described earlier. For example, the spectrum of UE-EL with T_CT=5 was subtracted from the spectrum of G5-EL with the same T_CT (=5), and the peak areas were evaluated by Gaussian fittings as displayed in Figure 3b. We obtained five data sets measured using different T_CT values; the obtained values were plotted vs. contact time, and the approximated results are shown in Figure 4.

Figure 4:

Peak area variations in G5 carbon signals depending on the contact time by solid-state ¹³C CP/MAS NMR measurements.

The plots (a, b, c, d, e and f) correspond to the signals shown in Figure 3b.

The slope of the lines obtained for the six signals varied from −0.0143 to −0.0573. The values correlated with the relaxation rate of the G5 carbons, allowing us to estimate the mobility in the structures. The order of the absolute values was as follows: b≤e≤a≤c≈f≤d, i.e. uncondensed (OH)≤4-O-5 (OH)≤uncondensed (OR)≤β-5 and 5-5 (OH)≈4-O-5 (OR)≤5-5 (OR). Although some coefficients of correlation (R²) were relatively low, probably because of the signal overlapping and the procedure accuracy of the difference spectrum calculation, the absolute values tend to increase when the G5 carbon gains a new connection in the structural unit. To discuss the structural mobility of lignin in more detail, further experiments should be conducted; nevertheless, we have attempted a quantitative discussion of lignin G5 structures based on these data.

The peak area ratios to the total aromatic area in T_CT=0 and 5 ms are summarized in Table 2. T_CT=5 ms is one of the usual measurement conditions in solid-state CP/MAS NMR, and the values in T_CT=0 ms are the result of contact time extrapolation. As a result of the extrapolation (T_CT=0), the ratio of uncondensed G5 carbons was 13+15=28%, G5 carbons having a carbon-carbon bond (β-5 and 5-5) was 26+27=53%, and G5 carbons having an ether linkage (4-O-5) was 5+14=19%. These values are slightly different from the values obtained in T_CT=5 ms. The G5 structures resulting in a large slope, such as 5-5 (OR), might be underestimated in the usual measurement conditions for solid-state NMR. These results will be discussed further with the results of the quantitative liquid-state ¹³C NMR measurements.

Quantitative evaluation of the G5 carbons by liquid-state NMR

The liquid-state quantitative ¹³C NMR spectra of G5-ELAc and UE-ELAc and their difference spectrum in the aromatic region are shown in Figure 5. The ¹³C signals in Figure 5c were sharp and better resolved in comparison with the solid-state NMR signals. The signals were assigned to the G5 carbon structures according to previous reports and are listed in Table 1. As mentioned earlier, the signal assignments agreed in principle with those of solid-state ¹³C NMR.

Figure 5:

Liquid-state quantitative ¹³C NMR spectra of (a) UE-ELAc and (b) G5-ELAc as well as (c) the difference between (a) and (b).

The signal assignments correspond to the structures shown in Figure 3 as follows: A′, [a]; B′, [b]; C′, [c]-1; C″, [c]-2; D′, [d]; E′, [e]; and F′, [f].

As discussed earlier, we used the solid-state ¹³C NMR spectra of EL due to the better signal resolution. Recent studies have reported that ball-milled cell walls or ELs can dissolve or swell in DMSO-containing solvent systems, and it is possible to acquire NMR spectra without acetylation (Lu and Ralph 2003; Kim et al. 2008; Kim and Ralph 2010). Based on these points, it might be possible to measure EL/DMSO by liquid-state NMR. In this study, however, we used the ELAc/CDCl₃ system to quantify the G5 structures. The major reasons are the small amount of sample available, the lower viscosity of this system and the better solubility of the sample in the ELAc/CDCl₃ system than in the EL/DMSO system.

The resultant peak area ratios are summarized in Table 2. The ratio of uncondensed G5 carbons was 17+23=40%, G5 carbons having carbon-carbon bonds (β-5 and 5-5) was 18+12+15=45%, and G5 carbons having an ether linkage (4-O-5) was 2+13=15%. According to recent studies evaluating lignin structures (Chen 1996; Capanema et al. 2004; Zhang and Gellerstedt 2007; Sette et al. 2011), nearly half of the interunit linkages in lignin are uncondensed structures (β-O-4, β-β and β-1), and other typical linkages are as follows: β-5≈10%, 5-5≈25% and 4-O-5<10%, as listed in Table 2 (see Supplementary Table S2 for details).

Discussion of condensed G5 structures in lignin as evaluated by NMR

Liquid-state NMR spectroscopy provides high-resolution spectra suitable for both qualitative and quantitative studies. In particular, recent progress using multidimensional NMR techniques has revealed the precise structures of plant lignin. The sensitivity of NMR for quaternary carbons is quite low, and they are difficult to resolve and quantify from ¹H-¹³C 2D NMR spectra. To determine these carbons by ¹³C NMR spectroscopy, many researchers have synthesized fine model compounds and assigned their chemical shifts. However, their actual abundance and structural variety in native lignin have remained unclear.

To address these points, in this study, we used selective ¹³C-enrichment and ¹³C NMR difference spectra to focus on the G5 carbons in lignin. The obtained ¹³C-enriched EL was examined by solid-state ¹³C CP/MAS NMR measurements with the contact time extrapolation method. In comparison with the results of the liquid-state quantitative ¹³C NMR measurements, most of the signal assignments were consistent, and the G5 carbon signals were successfully specifically assigned. For the quantity, the ratio of G5 carbons in the condensed structures was found to be larger by solid-state CP/MAS NMR than by liquid-state NMR spectroscopy. Using the solid-state CP/MAS NMR contact time extrapolation method, although the quantification quality might not be as good as that of liquid-state NMR, we would like to note that the condensed G5-structures such as 5-5 (OR) might be underestimated using the general measurement conditions because of the entangled structure (Table 1).

Here, it should be mentioned that the resultant value of the condensed G5 ratio by liquid-state NMR was somewhat higher than the previously reported values for softwoods, as suggested in Table 2. The difference might be related to variations in the species, the sample preparation procedure, and so on. With this sample preparation procedure, excessive ball milling may alter the cell wall structure and decrease the yield of thioethylated monomeric compounds by thioacidolysis and GC-MS measurements (Shipponen et al. 2014). This report suggests that ball milling may increase the ratio of condensed structures in lignin. In this study, we regarded the solubility of the samples in CDCl₃ for liquid-state NMR as important for comparing the samples using solid- and liquid-state NMR, and thus, the milling step was of great importance. For the precise quantification of the structure of lignin in its natural state using solid-state NMR, further studies using moderately prepared samples and molecular dynamics discussions are required.