Introduction

In recent years, scientific research on dietary fibre (DF) has seen significant growth, driven by increasing consumer awareness of the importance of a healthier diet. DF represents an essential component of the human diet, playing a crucial role in maintaining digestive health and preventing various chronic diseases. Its intake is associated with benefits such as improving bowel function, regulating blood glucose and cholesterol levels, and promoting satiety1. Additionally, certain types of DF have a prebiotic effect, promoting the growth and activity of beneficial gut bacteria2. Understanding the composition and structure of DF is therefore critical for taking full advantage of its health benefits and developing innovative functional products.

DF can be classified into different types, including mainly pectin, hemicellulose, cellulose, lignin, and resistant starch, each with distinct structural properties and physiological effects3. Soluble DF, like pectin and some hemicelluloses, contribute to gel formation and slow digestion, aiding in blood cholesterol control, while insoluble DF, such as cellulose and lignin, enhance stool bulk and promote regularity4. The various types of DF differ primarily in their monosaccharide compositions. For instance, pectin is rich in galacturonic acid, hemicellulose contains a mix of monosaccharides like xylose, mannose, and arabinose, while cellulose contains mainly glucose. These compositional differences significantly influence characteristics such as solubility, viscosity and fermentability, which are crucial aspects for applications in the food, nutraceutical, and pharmaceutical industries4. Therefore, determining the monosaccharide composition not only enables the identification of specific DF components but also serves as a diagnostic fingerprint to infer the origin, type, and potential physiological function of the fibres. However, accurate quantification of DF and its monosaccharide composition is still a significant challenge due to the structural complexity of these substances5.

Traditionally, DF content has been quantified using gravimetric methods, which involve isolating and weighing the DF fraction after removing starch and proteins. The Van Soest method, introduced in the 1960s, employs detergent solutions to fractionate plant cell walls into neutral detergent fibre (NDF) and acid detergent fibre (ADF). NDF includes hemicellulose, cellulose, and lignin, representing the total DF content, while ADF comprises cellulose and lignin, excluding hemicellulose6. Similarly, the Association of Official Analytical Chemists (AOAC) has developed official methods for DF analysis, such as AOAC 991.43, which utilises enzymatic-gravimetric techniques to determine total, soluble, and insoluble DF in foods7,8. It involves enzymatic digestion followed by gravimetric measurement, offering standardised approaches for DF quantification. These gravimetric methods provide a measure of DF fractions but do not offer detailed information about the DF’s monosaccharide composition or molecular structure. To gain such insights, more advanced analytical techniques are required.

Chromatography coupled with mass spectrometry (e.g. GC-MS or LC-MS) has become a key tool for identifying and quantifying specific constituents of DF. Complex polysaccharides are analysed by applying an acid hydrolysis step to break them down into monosaccharides before further derivatisation and analysis9. This method provides the ability to quantify the DF and, simultaneously, to determine its qualitative composition. Importantly, the methods for monosaccharide analysis using GC-MS and LC-MS are often based on hydrolysis in trifluoroacetic acid (TFA), as it produces the monomers while preventing the degradation of the carbohydrate that can occur when using sulfuric acid (H2SO4). However, TFA hydrolysis may not be as exhaustive as Hâ‚‚SOâ‚„ hydrolysis, potentially leading to incomplete breakdown of certain DF components10. Furthermore, this analysis offers greater insight into DF composition but loses the separation of the different fractions by gravimetric methods. For this reason, in our previous study, we integrated several existing protocols to develop a more comprehensive method. We designed a sequential procedure involving the hydrolysis of DF fractions obtained through the AOAC gravimetric method, first with TFA and subsequently with Hâ‚‚SOâ‚„, to enable the separation of the different components. The hydrolysed fractions were then analysed and quantified by GC-MS. Prior to GC-MS analysis, a neutralisation step of Hâ‚‚SOâ‚„ is required, adding an additional level of complexity to the procedure11.

In this context, proton nuclear magnetic resonance (1H NMR) spectroscopy represents a high-throughput method and offers numerous advantages. NMR relies on the magnetic properties of atomic nuclei and provides a detailed view of molecular structure without the need for derivatisation or separation as in chromatography. Compared to traditional chromatography techniques, the NMR approach reduces analysis time and provides absolute quantitative data without the need for reference compounds or calibration curves12. Additionally, NMR can directly analyse DF acid-hydrolysed fractions without the need for a neutralisation step, making it particularly suitable for characterising DF components without the interference of H2SO4. NMR offers a highly versatile and robust quantitative approach that can provide detailed information on structure and composition9. Beyond monosaccharide composition, NMR also enables the quantification of specific modifications within DF, such as degrees of methylation and acetylation, by direct measurement of the resulting methanol and acetic acid after hydrolysis. As an example, in pectin, the esterification levels significantly influence their functional properties, including gelling behaviour and interaction with other biomolecules13. Furthermore, NMR can monitor degradation products formed during the hydrolysis of polysaccharides into monosaccharides14. This capability is essential for understanding the breakdown processes of DF and assessing their stability and functionality under various conditions. It also allows for the eventual identification of unknown carbohydrates.

While 1D ¹H NMR offers many practical advantages, more complex ¹H NMR techniques have also been applied to carbohydrate mixture analysis. In particular, 2D NMR methods such as HSQC, HMQC, and TOCSY provide improved spectral resolution and facilitate structural assignment in highly complex matrices9,12. However, these techniques generally require longer acquisition times, more complex data processing, and larger sample amounts, which may limit their applicability for routine quantitative analysis of DF. In contrast, our focus on ¹H NMR aims to balance analytical simplicity with sufficient structural information, making it more suitable for high-throughput applications while still enabling the detection of relevant modifications and degradation products.

These features make the 1H NMR technique particularly advantageous for DF characterisation, where chemical composition, carbohydrate esterification, and molecular arrangement are critical for understanding material properties and functionality. Despite these advantages, to date, NMR is mainly used to analyse fibre samples without prior treatment. Some studies have applied ¹H NMR directly to unfractionated food samples such as honey and beer15,16. While these methods benefit from minimal sample preparation and can rapidly quantify major carbohydrates, they face limitations when applied to complex mixtures. These include severe spectral overlap and an inability to assign monosaccharides to specific polysaccharide sources. In such untreated matrices, the presence of diverse oligo- and polysaccharides complicates both identification and quantification, and the lack of a controlled hydrolysis step prevents targeting fibre-specific structural information. Furthermore, when fibres are insoluble, this requires the use of solid-state NMR, which, although effective for studying interactions among polysaccharides like cellulose, hemicellulose, and lignin, has limitations in resolution, sensitivity, and also cost compared to solution-state NMR12. Several studies, including those by Merkx and colleagues17, have applied 1H NMR to investigate the monosaccharide composition of fibres following acid hydrolysis. However, even in these cases, the approach typically involves hydrolysing the entire sample without prior separation of the different DF fractions, thus limiting the possibility of distinguishing the composition of individual fibre components. Our work advances this field by integrating ¹H NMR spectroscopy with the fractionation and sequential hydrolysis strategy previously developed in our GC-MS-based study11, applying it to gravimetrically separated soluble and insoluble DF fractions to target mainly pectin, hemicellulose, and cellulose. This NMR-based approach not only retains the advantage of fraction-specific resolution but also provides deeper structural insights compared to GC-MS. As a result, it allows for a more refined and informative characterisation of the structural complexity and distribution of DF components within food matrices, which was not addressed in previous NMR-based methods.

Specifically, we developed a method utilising 1H NMR spectroscopy for the quantification and characterisation of monosaccharides in hydrolysed DF fractions. Qualitative and quantitative analysis was conducted on the hydrolysed DF standard and DF fractions of different food matrices obtained from the AOAC 991.43 gravimetric method, without the use of any sample derivatisation and neutralisation procedure. The validity and reliability of the method were assessed using standard validation procedures and by comparison with GC-MS data, to evaluate its potential as an alternative to conventional analytical techniques.

Results and discussion

Signal identification (specificity)

A spectral library of the monosaccharides of interest was created using the pure standards of glucose (Glc), galactose (Gal), mannose (Man), rhamnose (Rha), xylose (Xyl), fucose (Fuc), arabinose (Ara), rhamnose (Rha), ribose (Rib), glucuronic acid (GlcA), and galacturonic acid (GalA) dissolved in 2 N TFA in D2O. The monosaccharides were first analysed individually to accurately determine their respective chemical shift values (ppm) and then in a mixture to assess potential spectral overlaps (Fig. 1).

Fig. 1: 1D 1H-NOESY NMR spectrum (600 MHz) of the anomeric region of standard monosaccharides.
figure 1

The doublet at 4.76 ppm corresponds to the β H5 proton near the COOH group of galacturonic acid, which, at acidic pH, shifts into the anomeric region; thus, this signal was not used for quantification.

Full size image

The internal standard (TSP) was used to calculate the absolute concentrations of monosaccharides.

The complete signal assignment of the α and β anomeric protons of the monosaccharides identified and subsequently used for quantification is shown in Table 1. Moreover, the chemical shifts of the acetate and methanol signals were confirmed by spiking the samples with pure acetate and methanol, since in the literature their chemical shifts are reported only under neutral pH conditions. Finally, the identification of degradation products was validated by comparison with data reported in the literature14,18.

Table 1 Characteristics of 1H NMR anomeric signals of monosaccharides, their esterification, and degradation products
Full size table

Compared with a neutral aqueous solution, the solution in 2 N TFA (pH 0.5, measured with a Mettler Toledo pH metre, Greifensee, Switzerland) allows better separation of the anomeric signals of the different monosaccharides and avoids overlap with the solvent (water) signal. This ensures good specificity of the signals. However, some partial overlaps still occur, particularly between the α-glucose with β-arabinose, the α-xylose with α mannose, the β-glucose with β-galactose, and β-xylose with β-galactose and β-fucose. In these cases, only one of the two peaks of the doublet was considered, and its intensity was multiplied by two. Furthermore, the β-glucuronic acid signal overlaps with that of its lactone. In this case, only one peak of the β-glucuronic acid doublet is considered, and its intensity is multiplied by two, as for the other overlapping signals, while for the lactone, the entire integral is taken and corrected by subtracting half of the β-glucuronic acid contribution. Glucuronic acid and its lactone are then considered together as a single compound.

Fructose has no proton signals in the typical anomeric region (4.5–5.5 ppm), instead its anomeric protons resonate in the same region as the non-anomeric protons of the other monosaccharides (3.2–4.5 ppm). Therefore, for quantification purposes, only the proton signals that did not overlap with those of other monosaccharides were considered, specifically the signal at 4.13 ppm that contains the protons H3 and H4 of the β-D-fructofuranose, as reported by Barclay and colleagues19. To estimate the relative percentage of this anomer, we used pure solutions of fructose at the concentration of 100, 500, and 1000 ppm, and they were monitored over time, with measurements taken every hour for the first 5 h, then at 12, and finally at 24 h, to estimate the time required for the sample to reach a stable state of mutarotation. The results showed that the value remained constant over time, stabilising at 22.6 ± 0.4% of the total fructose (Supplementary Table S1). This finding is consistent with the literature19,20. Consequently, this percentage was used to calculate the total ¹H NMR integral corresponding to fructose.

Ribose exhibits a more pronounced furanosidic form compared to other aldopentoses21; however, the signals of the furanosidic anomers are either too close to the residual water peak or partially overlapped, making their reliable integration challenging. Therefore, for quantification purposes, only the proton signals of the pyranosidic forms were considered, which correspond to 60.2 ± 1.8% of the total ribose and remained stable over the monitored time frame (up to 24 h). This stable proportion was therefore used to extrapolate the total ¹H NMR integral corresponding to ribose, ensuring accurate quantification despite the limited detectability of the furanosidic signals.

Relaxation delay evaluation

The longitudinal relaxation time (T₁) is the time constant that describes the return of nuclear magnetisation to equilibrium along the z-axis after excitation by a radiofrequency pulse. For accurate quantitative NMR analysis, it is essential that all nuclei relax fully within the acquisition cycle. The recycle delay (d₁), representing the time between successive scans, must therefore be chosen appropriately. According to established qNMR principles, the total relaxation time (d₁ + acquisition time) should be at least five times the longest T₁ value among the observed signals to ensure reliable quantification20.

It should be considered that relaxation times are not absolute values but depend on several factors, including sample composition and pH. In acidic environments, such as those of the samples analysed in this study, relaxation times may be prolonged compared to neutral conditions. On the other hand, the relatively long acquisition time used here (9.7 s) contributes to the effective relaxation period available to the nuclei, partly compensating for the shortened recycle delay. Therefore, the total recovery time relevant for quantification is better represented by the sum of d₁ and acquisition time (14.7 s).

To evaluate the influence of relaxation delay on quantification accuracy, different relaxation delays (5, 8, 15, 20, 25, and 30 s) were tested using both standard monosaccharide mixtures and a representative food sample. The integrals of selected anomeric proton signals were monitored at each delay time and normalised to the TSP signal. In this context, the T₁ of TSP was also evaluated by comparing its signal integral at different relaxation delays (Supplementary Table S2), showing that the TSP signal is fully relaxed already at 5 s.

The standard solution of monosaccharides was analysed using different relaxation times. Since the TSP reference did not show variations across delays, it could be used confidently for normalisation of monosaccharide integrals. A relaxation time of 5 s proved sufficient for accurate quantification of the monosaccharides, as the results did not differ from those obtained with longer relaxation times (Supplementary Table S2). The same procedure was applied to a fibre sample (SDF from orange by-products) in order to evaluate both the monosaccharides and the degradation products. In this case, relaxation times were tested until the signal of the compound requiring the longest relaxation time was completely stabilised. Once again, the monosaccharides were quantified accurately with a relaxation time of 5 s. Acetate, methanol, and furfural could also be reliably quantified with a 5 s relaxation time. The main issue concerned formic acid, which requires a relaxation time of at least 15 s to achieve complete quantification. With a relaxation time of 5 s, only about 80% of the formic acid signal was detected. Since longer relaxation times considerably extend the overall analysis duration, a relaxation time of 5 s was used, while acknowledging this limitation in the case of formic acid (Supplementary Table S3).

Method validation

Determination of LOQ and LOD

The instrumental limit of detection and limit of quantification were calculated utilising the S/N ratio methods. LOD and LOQ were therefore calculated as the concentrations of each monosaccharide producing a recognisable peak with an S/N ratio of, respectively, 3.3 and 10. Each monosaccharide had its own LOQ and LOD. To ensure the accurate quantification of all monosaccharides, the highest LOQ and LOD values among those determined were adopted, ensuring that no compound fell below the limit of quantification or detection.

The limit of detection (S/N ratio of 3.3) and the limit of quantification (S/N ratio of 10) values, with the conditions of this method, corresponded respectively to a concentration of 100 ppm and 25 ppm of each monosaccharide.

Determination of linearity of the 1H NMR method

The linearity was tested in the concentration range of 100–1500 ppm, and the linear regression equations and the correlation coefficient (R2) for each standard monosaccharide are reported in Table 2, demonstrating a good linearity for all the tested standards with R2>0.99.

Table 2 Linear regression equations and the correlation coefficient (R2) of the different monosaccharide standards tested
Full size table

Determination of accuracy and precision of the 1H NMR method

1H NMR analysis of the sample was repeated six times, and for each spectrum obtained, the signals of the substances identified were integrated. The results are reported in Table 3 and compared with accepted real values (weights). In the same table, the NMR precision (standard deviation, SD %), the BIAS (%), and the accuracy for each substance are reported as a percentage.

Table 3 Quantitative NMR performance for monosaccharide standards: reference value (determined by weighting), average concentration by NMR measurements and associated precision (standard deviation, SD %), bias (%), accuracy (%), and recovery (%)
Full size table

The accuracy is acceptable, as it is between 95 and 105%. In the case of glucose, this value was slightly higher (110%), in agreement with Caligiani and colleagues20.

Determination of repeatability and reproducibility of the 1H NMR method

Repeatability was calculated as intraday precision on a monosaccharide standard mixture solution at three concentration levels (100, 500, and 1000 ppm). For each concentration, three NMR tubes were independently prepared from the same stock solution using the same procedure and equipment, in the same laboratory, and were analysed six times within 1 day. The results of precision, expressed in Table 4, were compared with Horwitz's predicted intra-laboratory precision (PRSD), calculated as 0.66 × 2 × C (− 0.1505), where C is the concentration level expressed as a mass fraction. Considering that a commonly used criterion is that satisfactory precision produces results with a HORRAT value of less than 2, our results indicate satisfactory precision (ICH guideline Q2(R1)).

Table 4 Intra-day and inter-day repeatability of the method, calculated on monosaccharide standard mixtures at different concentration levels
Full size table

The repeatability results obtained show that the Horwitz equation is always satisfied, even in the presence of different concentrations of monosaccharides, concluding that the method has a good intraday precision in the range of concentration considered.

In order to determine the inter-day reproducibility, the same NMR tubes, after the first measurement, were kept in the fridge and re-analysed by 1H NMR spectrometer after 3, 10, 20 (Supplementary Table S4), and 30 days (Table 4). On each day, three independent spectra were acquired, one for each tube. The Horwitz value remained always below 2, demonstrating good reproducibility. In addition, no increases in this value are observed between day 3 and day 30, confirming the excellent stability over time.

Matrix effect evaluation

Further, standard fibres were spiked with standard monosaccharides to assess potential quantification interference by the matrix.

The matrix effect was evaluated by comparing the hydrolysates of pectin, xylan, and cellulose, unspiked and spiked with 100 and 500 ppm of the solution containing the mixture of monosaccharides. The recovery of each monosaccharide (considering only those with a concentration higher than 100 ppm (LOQ) in the original sample (data shown below in Table 6) was calculated as the difference between the concentration determined in the spiked sample and the added amount of standard monosaccharides (based on weighing) (Table 5).

Table 5 Evaluation of matrix effect based on monosaccharide recovery in spiked samples
Full size table

The measured peak recovery ranged from 91.2% to 104.5%, suggesting minimal interference from the sample matrix.

Method applications

Method application in standard dietary fibres

The quantitative 1H NMR method was applied to determine the monosaccharides in standard DF, pectin, xylan, and cellulose, subjected to acid hydrolysis, as described in the ‘Methods’ section. The fibre standards were of known purity and composition, so the values obtained were compared with those stated on the label and also compared with the data obtained from GC–MS traditional method on the same compounds.

According to the manufacturer’s specifications, the pectin from citrus peel contains over 83% fibre, with mineral content <7%, moisture content <10%, and galacturonic acid accounting >74% of the dry matter (DM). The experimental data obtained were consistent with these values. Specifically, the fibre content was measured at 82.1 ± 1.5% on the sample, with a yield of 98.6 ± 1.5%. As shown in Table 6, galacturonic acid made up 74.4 ± 0.8% of the DM pectin. The degree of methylation, reported to be approximately 64% by the manufacturer, was found to be 61 ± 1.4% in our analysis.

Table 6 Monosaccharide composition and yield (g 100-1 g DF) of hydrolysed pectin, xylan, and cellulose standard obtained by 1H NMR and GC-MS
Full size table

Finally, these results were compared with those obtained via GC-MS, showing comparable values with no statistically significant differences in terms of monosaccharide distribution (Table 6). However, the pectin yield obtained through GC-MS was lower than that measured by NMR. This discrepancy could be attributed to the derivatisation process used in GC-MS analysis, which is particularly limiting in the case of the uronic acids. In addition to derivatization, uronic acids also require a prior delactonisation step, as they can form lactones in acidic environments through intramolecular esterification between the carboxylic group and a hydroxyl group. These lactones are less reactive or inaccessible to derivatising agents and may not reflect the actual concentration of uronic acids unless properly delactonised22,23. This additional complexity, combined with the fact that pectin, unlike other polysaccharides, undergoes a slower and different hydrolysis process into monosaccharides, further complicates accurate quantification by GC-MS17,24. As a result, these analytical limitations can lead to an underestimation of uronic acid content when using GC-MS.

One of the main advantages of NMR over other analytical techniques is its ability to simultaneously detect monosaccharides, oligo and polysaccharides within the same spectrum, allowing for direct monitoring of the hydrolysis state12,17. In our experiment, the 2 h hydrolysis protocol resulted in complete hydrolysis of all polysaccharides except for pectin, which is known to require longer treatment times17. To account for this, we included in the quantification of pectin the fraction that was not fully hydrolysed, by integrating the group signals in the 5.08–5.16 ppm region (Supplementary Fig. S1), according to Merkx17, and considering two protons (H1 and H5) for signal integration25. This fraction accounted for 20.8% of the total pectin signal. It is known that, in the case of pectin, the formation of monosaccharides occurs mainly during the second hour of hydrolysis, with complete degradation (of the remaining ~10%) achieved after 3 h of hydrolysis with H2SO4, as described by Merkx17. In our case, the observed pectin signal was slightly higher, likely due to the use of TFA instead of H2SO4, and the hydrolysis time was limited to 2 h, as longer durations would lead to the degradation of other carbohydrates, which are more sensitive to prolonged treatment.

According to the manufacturer’s specifications, the xylan from Beechwood contains over 95% DM fibre, with xylose accounting 84% of the DM fibre, 5.7% other monosaccharides, and glucuronic acid 10.3%. Furthermore, 13% of the xylan was substituted with O-methyl glucuronic acid. The total DF was 94.2 ± 1.2%, in line with the 95% stated. Regarding the distribution of monosaccharides, xylose was found to be 77.1 ± 0.3%, 11.0% other monosaccharides, and 6.8% of unknown compounds probably related to O-methyl glucuronic acid. Furthermore, degradation products related to carbohydrates were found in a percentage of 5.1 ± 0.4%, including mainly furfural (95%) and formic acid (5%). Last, the levels of methylation and acetylation were found to be very low (2%, and 0.1%, respectively), consistent with the product specifications, which stated that it should not be substituted with methylated or acetylated xylose.

Finally, these results were compared with those obtained by GC-MS, showing comparable values with no statistically significant differences in terms of monosaccharide distribution, except for rhamnose (Table 6). GC-MS yield resulted lower (75.0 ± 10.0%) compared to that obtained from 1H NMR likely due to the impossibility of quantifying the degraded portion.

Lastly, standard cellulose from softwood tree pulp was analysed after the hydrolysis in H2SO4. The sample consisted almost exclusively of cellulose, as seen from both NMR (99.6%) and GC-MS (100%) data, with a yield of 101.8 ± 8.1%, as determined by NMR (Table 6).

Even in complex matrices, such as DF samples, this 1H NMR method proves to be effective for the quantification of constituent monosaccharides. Compared to GC-MS, this technique offers significant advantages, including the ability to detect and quantify partially hydrolysed fractions, as observed in the case of pectin, as well as potential degradation products. These components, which are not detected or considered in GC-MS analyses, can instead be included in NMR quantification, leading to a more comprehensive and realistic estimation of carbohydrate composition. It is worth noting that some monosaccharides, such as fructose and ribose, were detected only in certain samples and sometimes by one method but not the other. This occasional discrepancy is primarily due to their low abundance, which makes them more susceptible to signal overlap, co-elution or noise. Nonetheless, their presence was reported when confidently identified, as they can still provide useful information for the overall carbohydrate profiling without affecting the robustness of either analytical approach.

Method application in dietary fibre samples

The 1H NMR method was applied to different food matrices from which SDF and IDF were extracted and subsequently hydrolysed using the protocol described in our previous work11. Samples were chosen because they represent different types of dietary fibres. In particular, hazelnut shells are rich in hemicellulose, especially xylan, and cellulose14; fruit peels are rich in pectin26,27; and algae are a source of unique fibres such as ulvans in Ulva28.

DF obtained from orange by-products, which consisted of peel, albedo, and residual part of pulp, was analysed for its monosaccharide composition using NMR (Fig. 2) and once again compared with the GC-MS method (Supplementary Table S5). The SDF fraction of orange waste consisted mainly of galacturonic acid (40%), arabinose (20%), galactose (17%), and rhamnose (8%), supporting the presence of arabinogalactans, homogalacturonan, and rhamnogalacturonan I29,30 (Fig. 2). Methylation and acetylation were 6.4% and 5.1% of SDF, respectively. The recovery of total monosaccharides calculated based on the grams of extracted fibre and adjusted for residual protein and minerals was found to be 101.2 ± 0.1%, thus confirming the accuracy of the method. The portion of IDF fraction hydrolysable in TFA, typically associated with hemicellulose, exhibited a low amount of xylose (9%), despite xylose usually being the main constituent of this fraction. Instead, higher amounts of arabinose (26%), galactose (19%), galacturonic acid (15%), glucose (11%), and rhamnose (10%) were found. This composition suggests that this fraction may consist of insoluble and particularly resistant pectin instead of hemicellulose, such as rhamnogalacturonan-II. In this portion of IDF, the percentages of methylation and acetylation were 3 and 11%, respectively. The cellulose fraction was composed mainly of glucose (77%), which is indeed the major constituent of cellulose. In cellulose, finally, the percentages of acetylation and methylation were both under 2%. Based on the knowledge of monosaccharide composition, it was possible to make a more correct distinction of the fractions present in the IDF, which were insoluble pectins (46%), hemicellulose (5%), cellulose (38%), and lignin (11%), in agreement with the literature, where citrus by-products show low hemicellulose and high cellulose content31,32.

Fig. 2
figure 2

SDF and IDF composition of orange by-products obtained using the AOAC method and quantified by 1H NMR. d.p. are degradation products.

Full size image

The protocol was also tested on hazelnut shells. These results agreed with our previously reported data obtained by GC-MS11 (Fig. 3 and Supplementary Table S6). The high content of galacturonic acid (56%) in the SDF from hazelnut shells indicated the presence of pectin as the main component. This is further supported by the detection of galactose (5%), arabinose (3%), xylose (3%), and rhamnose (2%), reflecting the neutral monosaccharide profile typical of pectin23. In addition, the presence of mannose (20%) and glucose (3%) suggested the presence of non-pectic polysaccharides such as glucomannans or galactoglucomannans. The degree of acetylation and methylation are relatively low, both about 5% of the SDF. The IDF was composed of hemicellulose (28%), cellulose (23%), and lignin (49%), complying with the composition reported in the literature (40–51% of lignin, followed by hemicellulose with 13–32%, and cellulose with 17–29%)33,34. The hemicellulose portion, rich in xylose (71%), showed an acetylation of 52% and methylation of 4.5%.

Fig. 3
figure 3

SDF and IDF composition of hazelnut shells obtained using the AOAC method and quantified by 1H NMR. d.p are degradation products.

Full size image

Finally, the analyses were conducted on Ulva algae (Fig. 4 and Supplementary Table S7). The SDF fraction of Ulva consisted mainly of rhamnose (50%), unknown compounds (4.62, 4.81, 5.37, and 5.44 ppm), (13%) glucuronic acid (12%), xylose (6%), and mannose (6%) with a recovery of total monosaccharides calculated based on the grams of extracted fibre and adjusted for the amount of residual protein and minerals, of 98.5 ± 1.8%. In agreement with our determination, the SDF of Ulva is largely made up of ulvans whose main monosaccharides are rhamnose (17 − 45%), uronic acid (6−19%), xylose (2 − 12%), rhamnose and xylose sulfate (16 − 23%), and iduronic acid (1−9%)35. The latter, in particular, might represent a significant part of the portion of monosaccharides that we did not recognise with the standard monosaccharides.

Fig. 4
figure 4

SDF and IDF composition of Ulva algae obtained using the AOAC method and quantified by 1H NMR. d.p. are degradation products.

Full size image

The portion of IDF hydrolysed in TFA consists mainly of glucose (34%), rhamnose (27%), and xylose (21%). The rhamnose residue could indicate a portion of ulvans bound to the cell wall that has therefore not been separated from the insoluble portion with the AOAC methods. The presence of xylose and glucose, on the other hand, indicates hemicellulose, in particular, xyloglucans, already identified in this species of algae36. The fraction hydrolysed with Hâ‚‚SOâ‚„ showed an almost exclusive presence of glucose (88%), confirming the presence of cellulose as the main polysaccharide component. A small amount of residual hemicellulose was also detected, as evidenced by the minor presence of xylose. The IDF fraction was composed of 47% hemicellulose, 21% cellulose, and 32% lignin and other insoluble compounds. Among these, hemicellulose was the most abundant component, with a content approximately twice that of cellulose, in agreement with the literature37.

Furthermore, none of the three fractions of Ulva exhibited either acetylation or methylation, which is consistent with the fact that algae polysaccharides are predominantly sulphated35.

Correlation with the traditional method of analysis on dietary fibre

The quantification of each monosaccharide obtained by 1H NMR from the three reference standards (pectin, xylan, and cellulose), as well as from the three DF fractions of orange by-products, hazelnut shells, and algae, was compared with that obtained through the traditional method by GC–MS, on the same samples. For each standard and food matrix, two replicates were analysed by GC–MS and compared with two of the three corresponding NMR replicates, resulting in a total of 24 paired measurements. Pearson correlation coefficients (r) and their statistical significance (p) were calculated to assess the agreement between the two methods, as shown in Table 7.

Table 7 Pearson correlation between data obtained from GC–MS and 1H NMR
Full size table

High correlations were obtained for glucose (r = 0.993), galactose (r = 0.963), arabinose (r = 0.976), rhamnose (r = 0.982), mannose (r = 0.932), xylose (r = 0.987), and galacturonic acid (r = 0.960). Good correlations were also found for glucuronic acid (r = 0.718), whereas for fructose, ribose, and fucose, the correlation was not significant (p>0.05). These monosaccharides are rarely present in the analysed samples, and if present, they were less than 2%, which likely contributes to the reduced correlation. At such low levels, all variations in instrumental sensitivity, background noise, and detection limits can significantly affect the accuracy and reproducibility of quantification. As a result, these minor components are more prone to analytical discrepancies, which may explain the weaker correlation observed.

Conclusions

A 1H NMR method has been developed for the quantification of monosaccharides in DF after its fractionation and acid hydrolysis. This 1H NMR method can be used satisfactorily for the quantification of the various monosaccharides due to the positive results of the validation tests and the good correlation with the commonly employed GC-MS technique.

The advantages of 1H NMR are its rapid sample preparation and data acquisition, as well as its ability to simultaneously detect degradation products and carbohydrate esterification, such as acetylation and methylation, alongside the monosaccharide composition. This provides a more comprehensive structural insight into fibre fractions, which would otherwise require time-consuming derivatisation and different preparation steps for GC-MS analysis. These features support the use of 1H NMR in both quantitative and exploratory applications.

By combining accurate quantification with structural information, the 1H NMR method not only simplifies the analysis of monosaccharides but also offers a more complete view of DF composition. Such a comprehensive analysis is essential for advancing the characterisation of these materials and unlocking their potential in food and feed applications, where specific structural features influence nutritional and functional properties.

Methods

Chemicals and reagents

Barium hydroxide (Ba(OH)2), cellulose powder from softwood tree pulp (C6288), D-(+)-glucose (>99%), D-(−)-fructose (>99%), D-(−)-arabinose (>98%), D-(+)-galactose (>99%), D-(+)-mannose (>99%), L-(+)-rhamnose monohydrate (>99%), D-(-)-ribose (>99%), D-(+)-xylose, (>99%), L-(−)-fucose (>99%), D-(+)-galacturonic acid monohydrate (>97%), D-(+)-glucuronic acid (>98%), deuterated water (D2O), N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA), 2-(N-morpholino)ethanesulfonic acid (MES), pectin from citrus peel (P9135), trifluoroacetic acid (98% TFA), tris(hydroxymethyl)aminomethane (TRIS), 3-(Trimethylsilyl)propionic acid-D4 sodium salt (TSP), sulphuric acid (96% H2SO4), ultrapure water obtained with Milli-Q® system were purchased from Merck (Merk, Darmstadt, Germany). Xylan from Beechwood (P-XYLNBE>95%) and the total dietary fibre kit (K-TDFR, 200 A) containing α-amylase, amyloglucosidase, and protease were purchased from Megazyme (Megazyme, Wicklow, Ireland).

Dietary fibre samples

Hazelnut (Corylus avellana L.) shells and orange by-products (mixture of peel, albedo, and residual part of pulp) were provided by food companies. Ulva Lactuca algae were collected from Sacca Di Goro (Ferrara, Italy).

The samples were ground using the IKA® A11 basic analytical mill (IKA®-Werke GmbH & Co, Staufen, FR, Germany) and were stored at −20 °C until the beginning of the analyses.

Dietary fibre fraction isolation and hydrolysis

The full procedure for isolation and hydrolysis to monosaccharides of the dietary fibre fractions from food matrices is described in our previous work11. Briefly, the Insoluble Dietary Fibre (IDF) and Soluble Dietary Fibre (SDF) were obtained following the AOAC 991.43 method, with the only difference being that crucible filtration with Celite was not performed. Instead, the resulting pellet was centrifuged and separated into IDF (pellet) and SDF (supernatant), thus avoiding contact between the fibre and Celite. IDF was washed with water, ethanol, and acetone (as described in the AOAC method), and SDF was precipitated overnight with ethanol, and both fractions were dried in an oven at 105 °C for the analysis. The residual protein and mineral content were calculated on both SDF and IDF fractions by Kjeldahl and mineralisation analyses, respectively.

For fibre characterisation, 25 mg of each fraction were hydrolysed with 3 mL trifluoroacetic acid (2 N TFA) at 110 °C for 2 h to fully break down SDF and partially hydrolyse IDF (targeting hemicellulose). The remaining IDF pellet (cellulose and lignin) underwent further 2 h hydrolysis with sulfuric acid (12 M H₂SO₄ at 50 °C for 1 h and then 2 M H₂SO₄ at 100 °C for another hour) to degrade cellulose. The supernatants were stored in a freezer until the GC-MS and 1H NMR analyses. Lignin content was finally determined by difference from the IDF, after quantifying hemicellulose and cellulose.

GC-MS analysis

The supernatants were derivatised and then analysed for monosaccharide content via GC–MS. Cellulose monosaccharides were also neutralised with Ba(OH)₂ to remove H2SO4 before the derivatisation, as described in our previous work11.

1H NMR analysis

Sample preparation

The sample, hydrolysed in TFA or H2SO4 (950 μL), was prepared to the desired concentration by the addition of deuterated water (D₂O) and placed directly into NMR tubes, minimising sample preparation, in contrast to GC-MS analysis, which requires the long step of derivatisation and, in the case of H2SO4 solution, the neutralisation step. TSP-D4 (50 μL, 100 ppm), as the internal standard, was added for the quantification of monosaccharides.

1H NMR conditions

1H NMR analysis was conducted using a JEOL JNM-ECZR 600 MHz NMR spectrometer. 1D 1H NOESY experiments were performed following the procedure and calculations described by Musio and colleagues38. Specifically, pulse programme: noesy_abs; size of FID (x_point = 131,072); spectral width (x_sweep = 24 ppm); transmitter offset (x_offset = 4.85 ppm, representing the water resonance during acquisition, equivalent to 5.42 ppm after referencing to TSP = 0 ppm); 90° hard pulse (x_pulse = 9.68 μs); transient scans (x_prescans = 2); number of transients (scans = 64); acquisition time (x_acq_duration = 9.7 s); mixing time (mix_time = 0.010 s); recycle delay (relaxation_delay = 5 s) (this value was established based on preliminary measurements and several optimisation experiments on standard and real monosaccharide samples to ensure complete relaxation of all nuclei, see Supplementary Table S1 and S2); irr_mode = presaturation; irr-offset = 4.85 ppm; irr attenuation = 68 db); the receiver gain optimisation was determined automatically. The internal temperature of the probe was set at 25 °C.

1H NMR processing

1H NMR spectra were processed and analysed with Delta software v5.3.3 (JEOL RESONANCE Inc., Japan). Baseline correction was applied to all samples, and then an automatic integration method was created based on the shifts (ppm) of each compound (Fig. 1 and Table 1), allowing integration with pre-defined and consistent ranges across all samples. Minor manual adjustments were performed only when necessary, particularly in cases of peak overlap due to matrix effects or increased baseline noise, to correct integration inaccuracies. Different spectral regions were integrated and used for quantification: anomeric protons of monosaccharides (4.50–5.40 ppm, Fig. 1), and degradation products, specifically formic acid (8.25 ppm), hydroxymethylfurfural (9.43 ppm), and furfural (9.48 ppm)18. Integrals were normalised to the integral of TSP, added to each sample in an exactly known amount, and the values were converted to mg L−1 (ppm), according to the equation below20:

$$\frac{{{Int}}_{x}\times {EWx}\,}{{m}_{x}}=\frac{{{Int}}_{{TSP}}\times {{EW}}_{{TSP}}\,}{{m}_{{TSP}}}$$

Where:

  • Intx is the integral of analyte;

  • IntTSP the integral of internal standard (Integral of TSP = 1 because it is normalised);

  • mx is the unknown mass of the analyte expressed as mg.

  • mTSP is the mass of internal standard expressed as mg (0.1 mg);

  • EWx the equivalent weight of analyte. EWx = MW/no. hydrogens in the signal;

  • EWTSP is the equivalent weight of internal standard. EWx = MW/no. hydrogens in the signal.

The analyte concentration was calculated by dividing the mass obtained (mx) by the known volume (mL) of analyte used in sample preparation, yielding a final value expressed in ppm.

Furthermore, the acetylation and methylation degree was determined by the integration of the signals corresponding to acetic acid (2.10 ppm) and methanol (3.37 ppm) released after hydrolysis. The integrals were first converted to mass value using the previous formula. These masses were then used to calculate the number of moles of each compound. Finally, the degree of substitution was expressed as a percentage by dividing the number of moles of each substituent (acetyl or methyl) by the total number of moles of monosaccharide units in the sample and multiplying by 100, in accordance with the equation described by Müller-Maatsch and colleagues13.

Validation of 1H NMR method for monosaccharide composition in dietary fibre

The method was validated using a mixture of pure standards representing the major monosaccharides found in dietary fibre (glucose, fructose, arabinose, galactose, mannose, rhamnose, ribose, xylose, fucose, galacturonic acid, and glucuronic acid). TSP (50 μL, 100 ppm), used as the internal standard, was added for the quantification of monosaccharides. The weighing of the standards was performed on a Sartorius CP225D analytical balance (Sartorius, Goettingen, Germany) with a readability of 0.1 mg.

The validation of 1H NMR was conducted according to ICH guideline Q2 (R1) and the established protocols (International Conference on Harmonisation39). The parameters of validation, including specificity, linearity, LOD, LOQ, accuracy, precision, repeatability, and reproducibility, were studied.

Specificity

Specificity was evaluated in the anomeric region (4.50–5.40 ppm), used for the quantification of monosaccharides, by verifying the absence of signal overlap between the different monosaccharides and the residual solvent signal (water).

Limit of detection and limit of quantitation

The limit of detection (LOD) and limit of quantification (LOQ) were calculated utilising the signal-to-noise ratio (S/N) methods, based on the determination of the peak-to-peak noise40. S/N determination was performed by comparing signals of anomeric protons measured from samples containing low known concentrations of analytes (25, 50, and 100 ppm) with those obtained from blank samples, considering the same range of chemical shift (ppm) for each analyte in both sample and blank. LOD and LOQ were calculated as the concentrations of each monosaccharide producing a recognisable peak with an S/N ratio of 3.3 and 10, respectively. LOD and LOQ were determined in a pure standard solution of monosaccharides.

Linearity

The linearity of the method was checked at a range between 100 (based on the LOQ) and 1500 ppm of each monosaccharide, and 7 concentration levels (100, 150, 250, 500, 750, 1000, and 1500 ppm) were considered. A total of three replicates were performed for each concentration. Regression was performed on the ratio of peak integrals of analyte and internal standard versus analyte concentrations added. The goodness of fit was determined by means of the Mandel test at the 99% significance.

Accuracy and precision

Accuracy and precision were calculated by repeating the 1H NMR analysis for three different concentrations six times each on the same sample. The results were compared with the accepted true values (weights). The precision of the analytical procedure was expressed as the standard deviation (SD).

Accuracy was determined by the following equation:

$${{\rm{Accuracy}}}=({{\rm{p}}}^{2}+{{\rm{b}}}^{2})^{1/2}$$

Where p is the precision (expressed as SD), and b is the bias, defined as the difference between the NMR-determined concentration and the accepted true concentration20.

Repeatability and reproducibility

Intraday repeatability of the method was evaluated by preparing three independent samples for each concentration level (100, 500, and 1000 ppm), and each sample was analysed six times on the same day to assess intra-day repeatability. The results were compared with the predicted intra-laboratory precision (PRSDr) by the calculation of the Horwitz ratio (HorRatr). To estimate the predicted PRSDr, the original equation, used to predict between-laboratory variability (RSDR), was adjusted by applying a factor of 0.66, as RSDr is typically two-thirds of RSDR. This approach, as reported by Horwitz41, allows estimation of repeatability under single-laboratory conditions.

$${{{\rm{P}}}{{\rm{R}}}{{\rm{S}}}{{\rm{D}}}}_{{{\rm{r}}}}=0.66\times 2\times {{\rm{C}}}^{(-0.1505)}$$
$${{\rm{HorRat}}}_{{\rm{r}}}={{{\rm{RSD}}}}_{{{\rm{r}}}}/{{{\rm{PRSD}}}}_{{{\rm{R}}}}$$

Where: C is the concentration in mass fraction, RSDr is the relative standard deviation under repeatability conditions (intra-laboratory), and PRSDR is the predicted relative standard deviation under reproducibility conditions, calculated from the Horwitz equation.

To determine the inter-day reproducibility, the samples, after the first measurement, were kept at +4–6 °C, and re-analysed by 1H NMR after 3, 10, 20, and 30 days. On each day, three repetitions of the experiment were performed, and the same formula was applied.

Validation on standard polysaccharide

As an additional validation of the 1H NMR method for carbohydrate analysis, three standard polysaccharides, pectin, xylan, and cellulose, were hydrolysed in 2 N TFA for the first two, and H2SO4 for the third, as described in the ‘Methods’ section. The resulting monosaccharide solutions were analysed by 1H NMR. The experimentally obtained results were then compared with the theoretical value (product specifications) and the results from conventional analysis by GC-MS.

Matrix effect evaluation

The method was also tested in a matrix, using the hydrolysate of pectin, xylan, and cellulose standards, spiked with two different amounts (100 and 500 ppm) of the standard monosaccharide mixture. The full procedure, including hydrolysis and analysis, was carried out in triplicate on independently prepared samples.

Quantitative analysis of monosaccharide components in real dietary fibre samples by 1H NMR

After the validation, the quantitative method developed was applied for the determination of total monosaccharides in soluble and insoluble dietary fibre real samples. Specifically, hazelnut shells, orange by-products, and algae were analysed.

Statistical analysis

All statistical analyses were performed using IBM SPSS v.23.0 software (SPSS Italia, Bologna, Italy). Experimental data were expressed as the mean of three replicates ± standard deviation. Differences in monosaccharides obtained using GC-MS and NMR methods were evaluated using a t-test. In addition, the Pearson correlation between the two methods was calculated. Significant differences were compared at the level of p < 0.05.