Investigations of Native and Resistant Starches and Their Mixtures Using Near-Infrared Spectroscopy
- First Online:
- Cite this article as:
- Hódsági, M., Gergely, S., Gelencsér, T. et al. Food Bioprocess Technol (2012) 5: 401. doi:10.1007/s11947-010-0491-5
- 203 Views
Resistant starches (RS) play important roles in our nutrition; therefore, the investigation of these starches is notably important. In our study, two native starches (maize and wheat) and two resistant starches (Hi-maize™260, high amylose maize starch as RS2 and Fibersym™70, phosphorylated wheat starch as RS4) were investigated as is and in their physical mixtures (samples containing 20%, 40%, 60% and 80% RS) using near-infrared (NIR) spectroscopy. The aim of our study was to examine the spectra of resistant starches and to differentiate the resistant starch components in different ratios by NIR spectroscopy. The differences of samples were presented in two characteristic absorption bands for carbohydrate: carbohydrate II (2,080–2,130 nm) and carbohydrate III (2,275–2,290 nm) regions. Additionally, principal component analysis (PCA) for all samples was carried out. It was shown that the increasing amount of amylose can be sensitively followed up in carbohydrate II region. The phosphorylated RS4 is not so characteristic probably due to the reduced mobility of amorphous chains; however, the RS4 addition can be observed. Additionally, it was proven that the carbohydrate III region is sensitive for the changes of amylose–amylopectin ratio as well. The wheat-based RS4 addition causes linear changes in maize-based mixtures; thus the botanical origin is determining in this region. The global PCA analysis justified that the RS2 addition can be sensitively followed up independent on the medium; however, the increasing amount of RS4 cannot be detected in the PCA plot. The loading spectra of PC1 component attribute great significance to the carbohydrate III region.
KeywordsNIR spectroscopyHigh amylose resistant starchPhospho-ester-bounded resistant starchPeak assignmentCarbohydrate regionsPCA
Resistant starches (RS) are broadly investigated and applied due to their functional properties and health benefits in the human metabolism. Namely, resistant starches resist the digestion of amylolytic enzymes (Englyst et al. 1992) in the human small intestine, but in the large intestine they are fermented by colonic microflora. These fermentations result in short chain fatty acids (SCFAs) (Nugent 2005) which can help in the prevention of the development of abnormal colonic cell population. Additionally, resistant starches have great impact on the glucose and lipid metabolism (Nugent 2005).
Resistant starches can be divided into four groups (Sajilata et al. 2006). The first type (RS1) is physically intact for the digestive enzymes for example whole or partially milled grains. The second type (RS2) means physicochemically indigestible starch occurring for instance in green banana and high-amylose maize. RS3 means starch after recrystallization and retrogradation occurring, e.g. in cooked and cooled potato, while RS4 is chemically modified starch derived after cross-linking reactions for example phospho-ester-bounded starches.
Resistant starches can be used as healthy food additives in starch-based foods (Sajilata et al. 2006; Gelencsér et al. 2008a); therefore, the complex knowledge of their properties is remarkable important. Additionally, basic research of resistant starches can contribute to the understanding of the causes of resistance (Htoon et al. 2009; Sajilata et al. 2006) which can provide appreciable information for food development.
There are different methods which are commonly used in the RS analysis. RS content of starches can be determined using the enzymatic Association of Official Analytical Chemists (AOAC) Method 2002.02 (Megazyme 2004). Additionally, next to the measurement of the digestibility properties of these starches, the investigations of their rheological, thermal properties can be highly informative for the food industry (Gelencsér et al. 2008b). Investigating the physicochemical and structural properties of the starches can provide notable information about the digestibility (Ao et al. 2007) and about the specific applicability of starches in food production (Aina et al. 2009; Li et al. 2010). The physicochemical and chemical characteristics of starches can be analysed using near-infrared (NIR) spectroscopy as well. NIR spectroscopy is used for the analysis of starch bonds vibrations and damage (Gergely and Salgó 2005; Osborne and Fearn 1986) and it is also possible to evaluate the structural changes in retrograded resistant starch during storage (Garcia-Rosas et al. 2009). However, the spectra of resistant starches have not been analysed yet.
The primary aim of our study was to investigate whether resistant starches could be sensitively detected by NIR spectroscopy in native starch (maize and wheat) mixtures (native starch was replaced in 20%, 40%, 60% and 80% by RS) qualitatively. Definitely, an RS2 and an RS4 starch were investigated due to their higher resistance properties in contrast to RS3 (Gelencsér et al. 2008b). Using these wheat- or maize-based model systems as these are usually applied in starch-based food products (pasta and bread), we aimed to determine the effects of RS addition and of the matrices on the whole NIR spectra and the characteristic starch vibration regions of samples. It was also investigated whether NIR spectra have any information concerning the cause of resistance.
Materials and Methods
Two native starches (maize (S4126) and wheat (S5127) starch from Sigma-Aldrich Co., St. Louis, USA) and two resistant starches (Hi-maize™260, National Starch and Chemical GmbH, Hamburg, Germany and Fibersym™70, Loryma GmbH, Zwingenberg, Germany) were investigated. The maize-based Hi-maize™260 belongs to RS2; it has 60% total dietary fibre according to the AOAC Method 991.43. Fibersym™70 is chemically a modified phosphate wheat starch (RS4) which contains 70% total dietary fibre according to the AOAC Method 991.43. The native and resistant starches were measured as is and in their physical mixtures as well. The mixtures were made by using a single native starch and a single resistant starch in the ratio of 20%, 40%, 60% and 80% (w/w), respectively. Maize starch was marked as M, wheat starch as W, Hi-maize™260 as H and Fibersym™70 as F. The number next to the two capitals means the ratio of the resistant starch in the mixtures, for example MH40 notes 40% Hi-maize™260 and 60% native maize starch.
Scanning, Proccessing and Analysis of NIR Spectra
Ten grammes of powdery starch samples were packed into the standard sample cups (internal diameter, 55 mm; depth, 10 mm), and they were stamped until no further compression was observed. The standard sample cups were equipped with quartz window and threaded back. The starch samples were scanned from 1,100 to 2,498 nm using NIRSystems Model 6,500 monochromator system (Foss NIRSystems, Inc., Silver Spring, MD, USA) fitted with a rapid content analyzer module in reflectance mode. Data were collected in every 2 nm using Vision 3.20 SP5 software (Foss NIRSystems, Inc., Silver Spring, MD, USA) and stored as the average of 32 scans for each sample. Four spectra were recorded rotating the standard sample cup to 0°, 90°, 180° and 270° position in all cases.
The raw spectra were processed with different mathematical treatments. Standard normal variate (SNV) and then second derivation of spectra were calculated.
SNV is a scatter correction method (Foss NIRSystems 2000) used to normalize spectra when the effective pathlength varies among samples in a data set. Such pathlength variation can occur when measuring the spectra of powdery samples as in this study because particle sizes as well as colour vary between samples. SNV is calculated as follows, each spectrum is mean centred then divided by its standard deviation. The usage of SNV was previously tested and found essential.
After the SNV, the spectra were treated using second derivation. The second derivative calculation (Foss NIRSystems 2000) begins by identifying three segments at one end of the spectrum, each separated from other by a gap. Average absorbance values are calculated for the first, second and third segments (A, B and C, respectively). The second derivative value computed as A − 2B + C is assigned to the midpoint of the second segment. Then the whole sequence of three segments and two gaps are shifted one data point and the calculations are repeated until a second derivative value has been calculated for all data points in the spectrum.
Additionally, the second derivation was previously optimized. The specified segment size was 2, 6, 10 and 14 nm (1, 3, 5 and 7 points per segment, respectively). The optimal segment size was 10 nm for keeping the signal-to-noise ratio along the spectra (data not shown).
The mathematical treatments helped to eliminate the baseline shift and sloping background absorption, which arises from the physical nature of the sample. Additionally, the overlapped peaks can be resolved thus conducing to the assignment of a signal to certain functional groups.
We analysed the whole spectra and the four most characteristic absorption bands for carbohydrates (Gergely and Salgó 2005; Osborne and Fearn 1986; Williams 2001), namely carbohydrate I (1,575–1,590 nm, the first overtone of O–H stretching), carbohydrate II (2,080–2,130 nm, the combination of O–H bending and C–O stretching), carbohydrate III (2,275–2,290 nm, the combination of O–H stretching and C–C stretching) and carbohydrate IV (2,310–2,335 nm, the combination of C–H bending and C–H stretching). The carbohydrate II and III bands provided the most comparable and changeable regions of the samples; therefore, they are shown in this paper. The reproducibility of the spectral acquisition was good; the observed differences derived from the chemical features of the starches. The parallel scanning of the samples were averaged and the changes of average spectra were followed.
Variance in the intensity values of the local minimums reveal both quality and quantity differences. However, shift in the local minimums indicate bigger differences in quality due to the distinct peaks and wavelength absorption bands of starches.
The whole pretreated spectra (SNV + second derivation) were analysed using principal components analysis and applying Statistica 9.1 software (StatSoft, Inc., Tulsa, 2010).
Results and Discussion
The Analysis of the Carbohydrate II Region
The decreasing amylopectin concentration of the samples can be followed up in the spectra of the MH (Fig. 1a) and WH series (Fig. 1b) around 2,120 nm. The shapes of the spectra (one or two peaks) change with the amount of high-amylose maize starch (H). If less than 50% amylopectin is in the mixtures (60%, 40%, 20% or 0% of native starch containing samples) any local intensity minimums cannot be detected for the amylopectin around 2,120 nm.
Additionally, in MH series (Fig. 1a), the effects of the addition of H starch can be clearly noticed by investigating the changes of the wavelength minimums in their position (shift) and in their intensity values around the peak of amylose (around 2,096 nm). Similarly to MH series, a shift in the wavelength minimum at amylose peak can be observed by comparing W and H starches in WH samples (Fig. 1b). The effects of the addition of the characteristic H starch can be better followed up in maize-based mixtures (Fig. 1a) than in wheat-based ones (Fig. 1b) indicating that the medium is also a determining factor in this wavelength region. The differences between maize (Fig. 1a) and wheat (Fig. 1b) starches can derive from differences in their amylose and amylopectin molecules; from the different degree of polymerization of amylose molecules (Jane 2009; Shelton and Lee 2000; Tester et al. 2004) and of amylopectin molecules (Jane 2009; Tester et al. 2004). The degree of polymerization of amylopectin molecules is higher in maize (Tester et al. 2004) than in wheat starch (Jane 2009) indicating the more characteristic amylopectin peak of maize.
In case of MF mixtures (Fig. 1c), M (−0.0428) and F starch (RS4, −0.0408) differ significantly from each other investigating the intensity differences at 2,120 nm (amylopectin peak). The addition of F starch caused a linear increase in the intensity values at this peak. In case of WF series (Fig. 1d), the values of the amylose peak minimums of the samples around 2,096 nm differ significantly from each other by increasing the RS concentration except the WF60 (−0.0474) from WF80 (−0.476). Moreover, a linear increase can be detected in the intensity values with the addition of F starch. F starch is less characteristic in this region; however; it can be observed (Fig. 1c, d) that it has less intense amylopectin peak than M and W starches have probably due to the reduced mobility of amorphous chains of phospho-ester bounded starch (Chung et al. 2004). Otherwise, the medium is an influencing factor in case of F starch containing samples; the differences are relative in the function of the medium.
The Analysis of the Carbohydrate III Region
The minimum intensity values of the peak change linearly and significantly with the composition in case of MH (Fig. 2a), WH (Fig. 2b) and MF mixtures (Fig. 2c); however, in MF mixtures the position of the minimum remained the same. In MF mixtures, only dilution effect can be determined by adding RS owing to lower minimum value of native M (−0.2229) than resistant F starch (−0.2066).
It is interesting that the WF (Fig. 2d) samples do not show great variability, the phosphorylated structure of F starch (Seib and Woo 1997; Woo et al. 2008) does not differ notably from the wheat starch in these bonds which can be induced probably by the same botanical origin.
After peak assignments, the whole spectra were analysed using multi-component data-reduction method (principal component analysis) to further investigate the chemical differences of the samples.
The Multivariate Data Analysis (PCA) of NIR Spectra
The intra-series variability of MF and WF samples were not so remarkable compared to the MH and WH series. In WF mixtures, the differences between the samples are bigger than in MF series; though, the RS4 addition cannot be detected along the PC1 or PC2 components. However, it can be stated that each native and resistant starch can be distinguished from each other.
Figure 3b and c demonstrate the loading spectra of PC1 and PC2 of all samples. They show which wavelength values are the most characteristic. PC1 component describes the great part of the total variance (80.6%); however, PC2 is also remarkable (10.0%). One of the most intensive peaks is around 2,278 nm (Fig. 3b) which is in the carbohydrate III region. In this region the increasing amylose concentration caused a wavelength shift in the minimum values of the spectra (see Fig. 2).
In case of the loading spectrum of PC2 (Fig. 3c), the first overtone of O–H stretching vibrations (at 1,406 and 1,434 nm; commonly found in glucose units) is the most intensive.
Resistant starches have great interest broadly due to their health benefits and functional properties; therefore, the investigations of their specific characteristics are notably important using different methods. The NIR spectra of resistant starches have not been published yet; therefore, our research has great significance.
In the present study the appropriateness of near infrared spectroscopy in the distinction of different native and resistant starches and their mixtures was evaluated. The analysis of the two most characteristic absorption regions for carbohydrates (carbohydrate II, 2,080–2,130-nm regions; carbohydrate III, 2,275–2,290-nm regions) was presented in this paper. Finally, a global PCA containing all samples was also carried out.
On the basis of our results, it was proven that the changes of the amylose–amylopectin ratio can be sensitively followed up in carbohydrate II region. The phospho-ester bonds containing RS4 is not so characteristic probably due to reduced mobility of amorphous chains; however, the RS4 addition can be detected. It was also shown that the carbohydrate III region is sensitive for the differences of amylose–amylopectin ratio as well. The addition of wheat-based RS4 can be differentiated only in maize-based mixtures due to the different botanical origin.
The global PCA which was carried out based on all samples presents that the RS2 addition can be sensitively followed up independent on the medium. The RS4 containing mixtures show not so high variability; RS4 addition cannot be detected. The loading spectra of PC1 attribute great significance to the carbohydrate III region.
In summary, it can be stated that the Hi-maize™260 resistant starch was very characteristic due to its high amylose concentration while the other resistant starch (Fibersym™70) was characteristic as well due to its phosphorylated structure; however, this property relied on weaker spectral signals. The differences of amylose–amylopectin ratio can be observed in the region of 2,275–2,290 nm as the most sensitive.
This study was supported by DiOGenes which is the acronym of the project ‘Diet, Obesity and Genes’ supported by the European Community (contract no. FOOD-CT-2005-513946). The members of the project are listed on the web-site of the project: http://www.diogenes-eu.org. This work is also connected to the scientific programme of the ‘Development of quality-oriented and harmonized R + D + I strategy and functional model at BME’ project. Lastly, this project was supported by the New Hungary Development Plan (Project ID: TÁMOP-4.2.1/B-09/1/KMR-2010-0002).