Original Article
Peptide Profiling Using Matrix-Assisted Laser Desorption/Ionization-Time-of-Flight Mass Spectrometry for Identification of Animal Fibers
2013 Volume 2 Issue 1 Pages A0023
Details
2013 Volume 2 Issue 1 Pages A0023
Identification of fibers for verification of their specific animal origin is necessary for maintaining quality and value in the clothing industry. In order to examine adulteration in animal fibers, there is a commercially accepted method of microscopy analysis. However, this method is subjective and time-consuming due to its reliance on an operator identifying magnified fibers from their scale image and other features. Therefore, alternative reliable identification methods are required. In this study, peptide analysis using matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOFMS) is presented and used to distinguish between cashmere, wool, mohair, yak, camel, angora, and alpaca in untreated and treated fibers (dyed, chlorinated wool). Typical m/z values for each specific type of animal fiber were identified. Predictive models that could identify seven types of animal fibers as well as 50% blended samples were successfully constructed using multivariate analyses such as PCA and PLS regression. This technique is therefore extremely useful for complementing the conventional tests for detecting adulteration in animal fiber fabrics and clothing.
Fibrous materials of an animal origin are widely used for fabrics and clothing because of their soft texture and warmth. In particular, cashmere, the so-called “fiber diamond,” is a valuable natural fiber and a high-quality material used in the textile industry. Accordingly, labeling fabrics and clothing composed of animal fibers is required for maintaining their quality and value.
In order to identify specific animal fibers, there is a commercially accepted method of testing that relies on microscopy-based comparison. However, the accuracy that can be achieved depends largely on the expertise of the operator in identifying different fibers from their visual/microscopic appearances. Recently, the cuticular scale characteristics that are the main diagnostic features used to differentiate between wool and specialty fibers have been modified due to processing methods such as chlorinating and stretching. In addition, the surface characteristics of fibers are further complicated by improvements in animal breeding quality. Therefore, it is extremely difficult to identify adulteration in animal fibers using solely microscopy images.
For these reasons, there is a requirement for more reliable identification methods. We previously established an objective test method for distinguishing between different types of animal fibers using a PCR-based DNA fingerprinting technique.1) In recent years, advanced and sensitive real-time PCR techniques for identifying animal fibers have been developed and commonly employed. However, it should be noted that the DNA copy is variable in number between individuals. In addition, the DNA copy cannot be detected in treated (dyed, chlorinated, or stretched) animal fibers.2)
It is therefore clear that more reliable identification methods that complement the microscopy and DNA techniques are required. We have previously studied the identification and qualification of cashmere and wool (untreated and dyed) fibers using matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOFMS),3) with a similar study focusing on feathers and down.4) Proteome analysis of cashmere,5) and identification of cashmere, wool, and yak fibers by peptide mass fingerprint analysis have also been conducted.6,7) To date, identification of animal fibers has been mostly applied to untreated cashmere, wool, and yak; however, in order to accurately differentiate cashmere from substituted fibers of lower value, it is necessary to widen the scope of materials included in the investigations. Moreover, no discussion has taken place concerning alternative methods for classifying animal fibers by targeted m/z values.
The purpose of the present paper is to identify the typical m/z values of cashmere, wool, mohair, yak, camel, angora, and alpaca using MALDI-TOFMS. Furthermore, predictive models able to classify seven different types of animal fibers, as well as 50% blended samples, were constructed with the entirety of the spectral data by means of multivariate methods such as PCA and PLS regression.
The samples used in this study consisted of both treated and untreated materials, in addition to some blended fiber materials. Pure animal fibers; 4 types of untreated cashmere, 4 types of treated cashmere, 4 types of untreated wool, 4 types of treated wool, 2 types of untreated mohair, 1 type of treated mohair, 1 type of untreated yak, 1 type of treated yak, 1 type of untreated camel, 1 type of treated camel, 2 types of untreated angora, 1 type of treated angora, 4 types of untreated alpaca, and 3 types of treated alpaca. Blended animal fibers (50 : 50); treated cashmere/treated wool, treated cashmere/treated yak, treated cashmere/treated camel, treated mohair/treated wool, treated camel/treated yak, and treated alpaca/treated wool. After interviewing an expert operator of the conventional microscopy testing,8) frequently troubling fibers and blended samples were identified for use in the study. Treated fibers were deep color dyed, and treated wool fibers were additionally chlorinated. The samples were washed with petroleum ether (first grade, Wako Pure Chemical Industries, Ltd., Osaka, Japan), and then air-dried prior to the analysis.
Sample preparationPrior to enzymatic digestion, the samples were converted to powder using a cryogenic sample crusher JFC-300 (Japan Analytical Industry, Tokyo, Japan). The samples were placed in liquid nitrogen for 10 min, followed by crushing for 5 min.
Samples of 0.1 mg crushed animal fibers were reduced with 100 μL of a 50 mM dithiothreitol solution (Sigma-Aldrich, St. Louis, MO, USA). The samples were then incubated with 10 μg/mL trypsin (proteomics grade, Promega, WI, USA) at 37°C for 16 h.
MALDI-TOFMS analysisTryptic digest samples were concentrated and purified using SPE C-TIP (Nikkyo Technos, Tokyo, Japan) according to the manufacturer’s instructions, prior to the MALDI analyses. α-Cyano-4-hydroxycinnamic acid (Shimadzu GLC, Tokyo, Japan) was employed as the ionization matrix. MALDI-TOFMS positive ion spectra were obtained in linear mode with delayed extraction, in the scanning range m/z 700 to 4,000 using an AXIMA-Performance (Shimadzu, Kyoto, Japan). Mass calibration was carried out using angiotensin II ([M+H]+ 1047.20), adrenocorticotropin hormone ([M+H]+ 2466.71), and insulin B chain ([M+H]+ 3496.95) purchased from Sigma-Aldrich, St. Louis, MO, USA.
Multivariate analysisMALDI mass spectra were preprocessed by baseline subtraction and smoothing, with subsequent preparation of a data matrix of m/z and peak intensities prior to carrying out multivariate analysis with The Unscrambler software (CAMO, Oslo, Norway). In addition, normalization of spectra to the total ion current was performed.
An auto-scaling method was applied when PCA and PLS regression were used. PCA data analysis works by extracting the useful information from a highly correlated data set to create a scatter plot, in this case, for each type of animal fiber. PLS was carried out to produce predicted models that could classify seven pure animal fiber types and 50% blended samples. In general, PLS gives more accurate results than the traditional multiple regression approach, allowing analysis of data with numerous strongly correlated variables such as those from spectra. In the present study, a NIPALS algorithm was applied.
Figure 1 shows the family tree of animal fibers that were analyzed using MALDI-TOFMS, including cashmere, wool, mohair, yak, camel, angora, and alpaca, both treated and untreated. The mass spectra that were acquired for each of the samples are shown in Fig. 2. Candidates for typical m/z of each animal fiber are summarized in Table 1. Chemically treated fibers such as dyed or chlorinated, or 50% blended fibers showed a reduced intensity of original species-specific peptide masses. Moreover, some m/z peaks were lost because of the alteration in the properties of keratin and the masses of digest peptides. However, the remaining typical m/z values (shown in Table 1) were sufficient for identifying the fiber origin. This demonstrates that the method could be used for the examination of fibrous materials, even after chemical treatment, which is of interest in the textile industry.
| Cashmere | Wool | Angora | Alpaca | Yak | Mohair | Camel |
|---|---|---|---|---|---|---|
| 3183 | 815 | 1132 | 1226 | 1638 | 2128 | 1226 |
| 3224 | 1860 | 1955 | 1464 | 2506 | 2910 | 2583 |
| 2744 | 2583 | 3183 | 3463 | |||
| 2753 | 2803 |
Further, the results suggest that the mass spectra relate quite well to the biological taxonomy. Typically, there were some common m/z values observed for cashmere and mohair, which both belong to the genus capra (members of the family bovidae), and between camel and alpaca, which belong to the family camelidae. It was found to be difficult to distinguish between species that were closely related such as cashmere and mohair. In addition, wool, which belongs to the genus ovis (members of the family bovidae), gave low intensity signals and some broadened peaks throughout the spectra. In contrast, angora, which belongs to the family leporidae, and yak, which belongs to the genus bos (member of the family bovidae), showed clear typical m/z values.
Multivariate analysis (PCA and PLS)The multivariate methods focused on not only the typical peaks, but also the low intensity signals and the coefficient peaks for each animal fiber. Therefore, high throughput analysis with no loss of information was produced from the mass spectra with many thousands of data points.
PCA is an extremely useful approach for reducing multidimensional data to lower dimensional, while retaining most of the information. Each data point in the score plot represents a spectrum (Fig. 3). As shown, a cluster of points attributed to a specific animal fiber indicates the capability to differentiate between them by means of multivariate analysis.
In addition to the PCA, PLS regression was performed using the spectral data as a set of predictor variables. This multivariate method is one of the best statistical approaches for prediction when there is multicollinearity and a much larger data set. PLS regression forms the score (factors or latent variables) as new independent variables, where the correlation between two variables (predictor variable and response variable) becomes more significant in the regression models. This allows us to build more accurate models than when using a traditional multiple regression approach.9) In this study, a predictive model for cashmere, wool, mohair, yak, camel, angora, and alpaca, which included four formed factors, was created (Fig. 4). The predictive model had an R2 value of 0.970, which was the coefficient of multiple determinations. The predictive strength of the model (Q2 value) was found to be 0.835, which was assessed by linear regression of the measured values obtained in the cross validation procedure. RMSE, the average difference between the measured and predicted values of the response variables, was calculated to be 0.365. The results show that the calibration model is efficient enough to produce accurate and reliable predictions. Cross validation was used in the modeling procedure for each excluded sample to evaluate the accuracy of the predicted model created from the remaining samples.10–12)
In the same way, a predictive model for the 50% blended animal fiber samples was built by means of the predictor variables. Five factors were used for six different blended samples. The model is presented in Fig. 5 (R2=0.933, Q2=0.674, RMSE=0.471). Although the predictive strength of the model is inferior to that for the pure animal fibers, it was still found to be good enough for practical use.
Thus, applying the PLS method to MALDI-TOFMS spectra was found to be an effective approach to identifying different animal fibers.
In this study, tryptic digest of cashmere, wool, mohair, yak, camel, angora, and alpaca in untreated and treated fibers (dyed, chlorinated wool) were analyzed using MALDI-TOFMS. Typical m/z values that were specific to each type of animal fiber, even after chemically processed, were proposed. Moreover, it was found that the biological taxonomy contributed to the difficulties involved in the identification of animal fibers using MALDI-TOFMS. In addition, reliable prediction models were created using mass spectra of pure animal fibers and 50% blended samples by means of multivariate methods such as PCA and PLS regression. This powerful multivariate analysis approach was used to evaluate the complicated spectral data.
This technique can therefore be extremely useful as a complementary method to the more conventional tests used for detecting adulteration in animal fiber fabrics and clothing.
The use of MALDI-TOFMS analysis for constructing predicted models via this multivariate analysis is expected to be applicable to other blended fibers or leather goods that have not yet been identified.
The authors would like to thank Associate Professor Takeshi Bamba and Project Researcher Mr. Toshiyuki Yamashita in the Department of Biotechnology at Osaka University for their valuable comments and suggestions for improving the quality of this paper.
