Characterization of edible bird's nest by peptide fingerprinting ...

© The Author 2017. Published by Oxford University Press on behalf of Zhejiang University Press. 83

Food Quality and Safety, 2017, 1, 83–92doi:10.1093/fqs/fyx002

Research Paper

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected].

Research Paper

Characterization of edible bird’s nest by peptide fingerprinting with principal component analysisChun-Fai Wong*, Gallant Kar-Lun Chan*, Ming-Lu Zhang*, Ping Yao*, Huang-Quan Lin*, Tina Ting-Xia Dong*,**, Geng Li***, Xiao-Ping Lai**** and Karl Wah-Keung Tsim*,**

*Division of Life Science and Center for Chinese Medicine R&D, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, **HKUST Shenzhen Research Institute, Hi-Tech Park, Nanshan, ***School of Chinese Herbal Medicine, Guangzhou University of Chinese Medicine, and ****Dongguan Mathematical Engineering Academy of Chinese Medicine, Guangzhou University of Chinese Medicine, China

Correspondence to: Karl Wah-Keung Tsim, Division of Life Science, The Hong Kong University of Science and Technology, Clear Water Bay Road, Kowloon, Hong Kong SAR, China. E-mail: [email protected]

Received 8 November 2016; Revised 27 November 2016; Editorial Decision 30 December 2016.

Abstract

OBJECTIVES: Proteins are the major component and play a key role in nutritious and therapeutic functions of edible bird’s nest (EBN); however, limited studies have been conducted on the protein due to difficulties in extraction, isolation as well as identification. This study aimed to provide comprehensive information for the quality evaluation of EBN peptides, which would be a valuable reference for further study on EBN proteins.METHODS: Here, we developed a quality control method using high performance liquid chromatography (HPLC) peptide fingerprints deriving from EBN being digested with simulated gastric fluid. The characteristic peptide peaks were collected and identified by LC-MS/MS. RESULTS: The characteristic peptide peaks, corresponding to the protein fragments of acidic mammalian chitinase-like, lysyl oxidase, and Mucin-5AC-like, were identified and quantified. Interestingly, the principal component analysis indicated that the fingerprints were able to discriminate colour of EBN (white/red) and production sites (cave/house) of White EBN on the same weight basis. As proposed by the model developed in this study, Muc-5AC-like and AMCase-like proteins were the markers with the highest discriminative power. CONCLUSIONS: The overall findings suggest that HPLC peptide fingerprints were able to clearly demonstrate peptide profile differences between genuine and adulterated EBN samples; and classify EBN samples by its color and production site. In addition, the protein identification results suggested that Muc-5AC-like protein was the major protein in EBN.

Key words: Edible bird’s nest; Peptide fingerprint; Mucin-5AC-like; Acidic mammalian chitinase-like; Principal component analysis.

Introduction

Edible bird’s nest (EBN), or cubilose, is a health food supplement

originated from salivary secretion by specific swiftlets, mainly from

Aerodramus fuciphagus and Aerodramus maximus (Kang et al., 1991),

which has been proven to have nutritious and therapeutic values, such as anti-influenza viruses, antioxidant, skin lightening, bone strength improvement, anti-inflammatory, and epidermal growth enhancement (Kong et al., 1987; Kong et al., 1989; Guo et al., 2006; Aswir and Wan Nazaimoon, 2011; Matsukawa et al., 2011; Yew et al., 2014;

Dow

nloaded from https://academ

ic.oup.com/fqs/article/1/1/83/4791723 by guest on 29 January 2022

mailto:[email protected]?subject=

84 C.-F. Wong et al., 2017, Vol. 1, No. 1

Chan et al., 2015). Southeast Asian countries, including Indonesia, Malaysia, Vietnam, and Thailand, are the major exporting countries of EBN. Human consumption and medicinal application of EBN could be dated back to the Tang dynasty (618–907 A.D.) and the Sung dynasty (960–1279 A.D.) in China (Koon and Cranbrook, 2002).

Although EBN has been served as an esteemed food in Chinese community for over 1000 years, limited research has been conducted on EBN and its proteins. Protein is a major part of EBN accounting for 50% of EBN dried weight on average (Jiangsu New Medicine College, 1977); it is conjectured to be a key factor of its nourishing and/or medicinal functions. The epidermal growth factor (EGF)-like pep-tide was partially purified with Bio-Gel P-10 columns from aqueous extracts of EBN that stimulated cell division and growth and enhanced tissue growth and regeneration (Kong et al., 1987; Kong et al., 1989). Two major bands (~106 kDa and ~128 kDa) were identified by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) as ‘sialo-glycoprotein’. Nevertheless, no satisfactory result was obtained from protein identification studies that include N-terminal sequence determination (Edman degradation), matrix-assisted laser desorption ionization–tandem time of flight (MALDI–TOF/TOF), and liquid chromatography–tandem mass spectrometry (LC–MS/MS) (Zhang et al., 2012). Acidic mammalian chitinase-like (AMCase-like) protein fragments from Meleagris gallopavo and an allergen homologous to ovo-inhibitor have been identified by 2-DE assays followed by MALDI–TOF/TOF/MS analysis in EBN extract (Liu et al., 2012). In addition, a microbial nitrate reductase, converting nitrate to nitrite and playing a role in the colour change of White and Red EBN, was identi-fied by mass spectroscopy (Chan et al., 2013b). Nonetheless, it remains unclear whether those identified proteins could accurately represent the majority of EBN protein. The difficulties encountered in research of EBN proteins are: (i) extracting and purifying proteins; and (ii) lack-ing of full Aerodramus genome sequence.

Owing to the limited supply and labour-intensive cleaning process, EBN is always expensive with current prices ranging from USD 500 to 15 000/kg. Driven by the lucrative return, various materials, includ-ing Tremella fungus, fried porcine skin, carrageenan, agar, and gelatin, which are almost indistinguishable from the genuine samples by visual inspection, were commonly adulterated into EBN in order to increase the net weight (Ma and Liu, 2012). Some businesses have been known to mix low-quality EBN into high-quality EBN and selling that at a high price. Occasionally, consumers have been counterfeited into pur-chasing lower priced house EBN at a premium price associated with cave EBN. About 40 publications are found in PubMed today, and nearly one-third of the publications are published in the last 5 years.

Besides, most of the publications still retained in elucidating chemi-cal composition as the quality control parameters: since no official method has been established for quality surveillance of EBN (Deng et al., 2006; Wang et al., 2006; Wu et al., 2010; Chan et al., 2013a).

Here, we attempt to find a key to open these proteome barri-ers by high performance liquid chromatography (HPLC) peptide fingerprinting. HPLC fingerprinting is one promising tool widely used in the modern standardization of herbal extracts (Department of Health, Hong Kong, 2010; Chinese Pharmacopoeia Commission, China, 2015), which could be applied to EBN as a robust technique in qualitative and quantitative controls. Firstly, an over-stewing method was developed to extract most of the EBN protein. Secondly, simulated gastric fluid (SGF) was used to digest EBN protein fully into peptides that can be separated by HPLC according to their polarity. Thirdly, according to the most relevant NCBI protein data-base, the characteristic peaks in chromatograms were identified and quantified. In addition, principal component analysis (PCA) and hierarchical cluster analysis (HCA) were adopted to reveal the relationships of factors within the data, including colour, country of origin, and production site of EBN. The results therefore contributed to the authentication and classification of EBN. This study aimed to provide comprehensive information for the quality evaluation of EBN protein at the peptide level, which would be a valuable refer-ence for further study on EBN proteins.

Materials and Methods

Chemicals and EBNPepsin, SGF without enzyme (contains 0.07 M hydrochloric acid and 0.1 M sodium chloride), and trifluroacetic acid were purchased from Sigma/Aldrich (St Louis, MO). LC-MS-grade acetonitrile was obtained from JT Baker (Center Valley, PA). Aprotinin, a monomeric globular polypeptide with a molecular weight of 6512, was obtained from GE Healthcare (Buckinghamshire, UK). Twenty-five batches of EBN, including different colour, country of origin, and production site, from different commercial bands were randomly purchased in Hong Kong market, and all samples were in standard ‘cup’ grade. Samples were labelled and stored at room temperature upon arrival. The sample information was listed in Table 1.

Extraction and digestion of EBNA cup (a common market size) of dried EBN was accurately weighed and soaked in a 100-fold volume of water (w/v) for

Table 1. Information of edible bird’s nest (EBN) samples.

Sample code Colour Country of origin* Production site* Sample code Colour Country of origin* Production site*

1 White Indonesia House 14 White Thailand Cave2 White Vietnam Cave 15 Yellow Indonesia House3 White Thailand Cave 16 Yellow Indonesia House4 White Indonesia House 17 Yellow Indonesia House5 White Indonesia House 18 Yellow Indonesia House6 White Malaysia House 19 Red Indonesia House7 White Thailand Cave 20 Red Indonesia House8 White Vietnam Cave 21 Red Indonesia House9 White Thailand Cave 22 Red Indonesia House10 White Malaysia House 23 Red Indonesia House11 White Indonesia House 24 Red Indonesia House12 White Vietnam Cave 25 Red Malaysia Cave13 White Malaysia House

*The country of origin and production site of EBN were provided by the merchants.

Dow



Characterization of edible bird’s nest, 2017, Vol. 1, No. 1 85

3–15 h (3 h for White EBN; 10 h for Yellow EBN; 15 h for Red EBN), as described previously (Chan et al., 2013b). After discard-ing the soaked water to remove contaminants, the EBN was rinsed with water three times. The soaked EBN was put into a stewing pot with 30 volumes of water. The soaked EBN was stewed for 8–40 h at 98 ± 2°C until it was completely molten (8 h for White EBN; 16 h for Yellow; 40 h for Red EBN). The sample was dia-lyzed overnight against water in a dialysis bag with 2000 cut-off molecular weight and then filtered, freeze-dried, and stored at 4°C until use. The EBN lyophilized powder was dissolved and digested with a SGF (pH 2), consisting of pepsin and SGF without enzyme, for 1 h at 37°C. The digest was neutralized with sodium hydroxide and then filtered by 0.45 µm hydrophilic filter. The total protein content of EBN was determined by Kjeldahl method (Codex Alimentarius International Food Standards, 1999), and the extracted protein was determined by the Bradford method (Bio-Rad, Herculues, CA).

SDS–PAGE analysisThe EBN lyophilized powder was suspended in 1 ml Milli-Q water. The samples were treated with direct lysis buffer (0.125 M Tris–Cl, pH 6.8, 4% SDS, 20% glycerol, 2% 2-mercaptoethanol, and 0.02% bromophenol blue) at 95°C for 5 min, and subsequently 10 μg of extracted proteins of EBN were loaded onto a 15% SDS polyacryla-mide gel and run at 60 V for electrophoresis. After separation of protein, the gel was stained by Coomassie blue reagent and then destained by a solution consisting of water, methanol, and acetic acid. Gel photos were captured by a scanner.

HPLC conditionsThe chromatographic separation was performed on an Agilent HPLC 1200 series system (Agilent, Waldbronn, Germany), which was equipped with a diode array detector (DAD), a degasser, a binary pump, an autosampler, and a thermo-stated column compart-ment. The EBN samples were separated by a SymmetrySheild™ RP C18 column (5 mm i.d., 250 mm × 4.6 mm). The mobile phase was composed of 0.1% trifluroacetic acid in water (A) and acetonitrile (B) using the following gradient program: 0–10 min, linear gradi-ent 5–15% (B); 10–30 min, linear gradient 15–30% (B); 30–50 min, isocratic gradient 30% (B); 50–60 min, linear gradient 30–50% (B); a pre-equilibration period of 20 min was used between each run. The flow rate was 0.6 ml/min; the column temperature was 25°C; and the injection volume was 10 µl. The DAD wavelength was set to 214 nm since the peptide bond showed the highest absorbance at this wavelength.

Protein identification by LC–MS/MSThe eluent at 9–40 min from the peptide fingerprint was collected by a Gilson FC203B fraction collector at 1 min per fraction. Five fractions representing peaks A, B, C, D, and E were collected and subjected to protein identification. The samples were desalted, con-centrated, and purified by ZipTip and then analysed by a Thermo Scientific LTQ Velos Dual-Pressure Ion Trap Mass Spectrometer cou-pled with a Thermo Accela 600 pump, an Accela autosampler LC, and ETD source (spray voltage 1.6 kV, capillary temperature 250°C). A Thermo Scientific BioBasic-18 column was used (150.0 × 0.1 mm, 5 μm). Formic acid at 0.1% in MS-grade water and 0.1% formic acid in MS-grade acetonitrile were used as mobile phase A and B, respectively; with a flow rate of 0.15 ml/min. Mascot Daemon (ver-sion 2.3.0) was used as a sequence database searching engine. The

parameters were set as follows: (i) pepsin A as the digestion enzyme, (ii) allowing absence of two internal cleavage sites, (iii) oxidation of methionine as a variable modification, (iv) ±1 Da for peptide mass tolerance, and (v) ±0.8 Da for fragment mass tolerance. The false discovery rate was below 5% by evaluating the number of matched proteins in the search of real database and decoy database. The peptide sequences were searched in the NCBI library and Chaetura pelagica (Chimney swift) database.

Data analysisPCA is an unsupervised statistical model that transforms a set of observations of possibly correlated variables into a smaller group of linearly uncorrelated variables, thereby avoiding subjective deci-sions, making data easy to explore, and allowing for the visualiza-tion of significant differences in complex data sets with many factors (Abdi and William, 2010). Only those characteristic peaks having higher than their respectively limit of quantification (LOQ) values were quantified and used for subsequent data analysis. Five peaks (A, B, C, D, and E) could reach the LOQ requirement, and they were selected and quantified with reference to the spiked protein apro-tinin. PCA was performed using SIMCA 13 (Umetrics, Sweden) with the parameter set to ‘PCA-X’. The relative amount of peaks A–E from the chromatographic fingerprints was used for PCA, and the score plot, loading plot, and dendrogram were examined in order to reveal possible relationship between peptide mapping and EBN.

Statistical tests were done by using one-way analysis of variance (ANOVA) provided in GraphPad Prism 6.0. Statistically significant changes were classed as [*] where P < 0.05, [**] where P < 0.01, and [***] where P < 0.001.

Results

Optimization of protein digestion on EBNTwenty-five EBN samples collected from a Hong Kong market (Table 1) were extracted by the over-stewing method in order to ensure the complete solubility of EBN protein. To calculate the total amount of protein in EBN (soluble or insoluble), nine crude EBN samples, including different colour, country of origin, and production site, were sent to a food laboratory for protein analysis by Kjeldahl method (Codex Alimentarius International Food Standards, 1999), a well-established test to investigate the total protein amount of food. The average protein content of three White EBN was 48.5 ± 5.1% (Sample 1, 2, 6); Yellow EBN was 55.7 ± 2.0% (Sample 15, 16, 17); Red EBN was 53.1 ± 3.2% (Sample 19, 21, 25) of the dried weight that was very close to the literature value, i.e. 50% (Jiangsu New Medicine College, 1977). Thus, the reported value was adopted as the denominator in calculating the extraction rate for simplicity. By determining the extracted protein, 83.7 ± 10.0% (Sample 1–14) and 81.5 ± 10.9% (Sample 15–18) of proteins were extracted among White and Yellow EBN samples, respectively; whereas 65.6 ± 7.3% (Sample 19–25) were found in Red EBN samples. The low extrac-tion efficiency revealed in Red EBN was consistent with historical wisdom of subjecting this EBN to a longer stewing time (Chan et al., 2013b).

The extracted proteins of White EBN (Sample 1) were separated by SDS polyacrylamide gel, and most of them were stacked at the top of gel, which suggested that most of the soluble EBN proteins were over 200 kDa (Figure 1A). After SGF digestion, the size of White EBN proteins was decreased in a time-dependent manner. The pep-sin from SGF served as a control at ~35 kDa. The EBN digest was

Dow




Figure 2. HPLC chromatogram of EBN digests and its fake products. Fifty milligrams of EBN, or same amount of fake products extract, was digested in 5 ml SGF for 1 h and then neutralized. Ten microliters of the digest was injected into HPLC system. (A) The chromatographic patterns of White EBN, Yellow EBN, and Red EBN were revealed, and similar peaks B–E were identical. The common fake products, including agar, Tremella fungus, gelatin, pig skin, and carrageenan, showed very different pattern. (B) Aprotinin was spiked into the EBN digest for HPLC fingerprinting. A calibration curve of aprotinin protein content was shown in the insert. The peak content in the fingerprint was quantified with reference to aprotinin. n = 4. EBN, edible bird’s nest; HPLC, high performance liquid chromatography; SGF, simulated gastric fluid.

Figure 1. Digestion of White EBN with SGF. (A) Fifty milligrams White EBN extract (Sample 1) was digested in 5 ml SGF for 2 h. The digest was collected every 15 min and then neutralized immediately by sodium hydroxide. EBN lane represents EBN extract (indicated) without digestion; SGF lane with pepsin (indicated) served as blank control. The digested EBN protein (10 μl) was separated and analysed by 15% SDS–PAGE followed by Coomassie blue staining. (B) The protein contents of above different time of SGF digests were analysed by Bradford protein assay. Values are expressed in mg/ml, mean ± SEM, n = 4. Statistically significant is as [***] P < 0.005. (C) EBN digest was collected at different time points and analysed by HPLC, as shown in Figure 2A. The area of peaks B, C, D, E and the total peaks area at different time points in peptide fingerprint indicated the digestion was completed after 60 min. n = 4. EBN, edible bird’s nest; HPLC, high performance liquid chromatography; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; SGF, simulated gastric fluid.

Dow




collected every 15 min and analysed by SDS–PAGE. The protein con-tent of EBN digests, as well as the disappearance of high molecular weight protein, was decreased to a steady status after 60 min indi-cating a full digestion by SGF (Figure 1A and 1B). The EBN digests (Sample 1, 2, 3, and 4) at different time intervals were injected into HPLC for fingerprinting analysis. Absorption at 214 nm was chosen to obtain maximal peak height. Four distinguishable and dominant peaks (B–E) were selected as the reference points (Figure 2A). The areas of those peaks were analysed along the time course, and they reached maximum peak area after 60 min of digestion (Figure 1C). This digestion profile was revealed also from Yellow EBN (Sample 15, 16, 17, and 18) in Supplementary Figure S1A, and Red EBN (Sample 19, 20, 21, and 25) in Supplementary Figure S1B. Thus, 60 min was decided as the complete digestion time of EBN extract.

To identify major EBN proteins, the peaks of A–E were col-lected and sent to ion-trap LC–MS/MS analysis. Hundreds of pro-teins, including origins from avian, mammals, insects, and bacteria, were matched; and some of them had been reported in previous studies, including AMCase-like (Liu et al., 2012; You et al., 2015) and transferrin (Hou et al., 2015). None of the proteins identified from EBN extracts matched with Aerodramus, likely because the genome sequence of swiftlet was not completed. We then searched the protein sequences from C. pelagica, a species in the same family as A. fuciphagus. The common proteins were selected among three trials, and blank SGF was served as a control. Only the fragments of AMCase-like, lysyl oxidase homolog 3, Muc-5AC-like fragment, and AMCase-like were identified in peak B, C, D, and E, respec-tively, in all samples, which confirmed the identities of EBN proteins (Table 2). Although fragments from peaks B and E were identified as AMCase-like proteins, they represented unique AMCase-like frag-ments matching to accession numbers in the database.

Peptide fingerprint and its applicationThe HPLC peptide fingerprints were generated from 25 SGF-digested EBN samples, and typical fingerprints of White, Yellow, and Red EBN were given in Figure 2A. In contrast, the products commonly used to imitate EBN (e.g. agar, Tremella fungus, gelatin, pig skin, and carrageenan) did not show any similarity to the EBN fingerprint. The peptide fingerprints showed a close similarity among White, Yellow, and Red EBN, except that the peak heights were varied (Figure 2A).

To quantify the characteristic peaks of EBN fingerprints, apro-tinin, a small protein of ~6 kDa, was used as an external standard. Aprotinin was spiked into EBN samples, and it did not interfere with most of the characteristic peaks (Figure 2B). By referring to aprotinin peak, the peak retention time and peak height could be calibrated, and a calibration curve from 0.5 to 10 µg of aprotinin was estab-lished by absorption at 214 nm (Figure 2B insert). Using this stand-ard curve, the relative peptide content could be estimated. Figure 3A

and Supplementary Table S1 show the estimated peptide contents of all identified fragments of different types of EBN. In general, the peptide contents of Red EBN were lower than the others: this low amount could be an outcome of lower amount of extractable protein derived from Red EBN.

To verify the authenticating power of peptide fingerprint, the pep-tide fingerprints of three White EBNs from different sources and two adulterated EBN (being mixed with fake EBN) were generated. The chromatographic patterns of all samples were revealed, and similar peaks B–E were observed (Supplementary Figure S2). The relative amounts of peaks B–E were determined. As illustrated in Figure 3B, the amounts of peaks B–E for EBN (Indonesia), EBN (Malaysia), and EBN (Vietnam) were comparable to that of genuine EBN. The amounts of peaks of EBN mixed with fake EBN were significantly lower than genuine EBN (Figure 3B). Muc-5AC-like protein (peak D) was therefore being considered as a tentative marker with the highest discriminative power in authenticating EBN.

Principal component analysisPCA was performed on the peptide mapping of 25 EBN to detect anomalies and to study relationships in the data. Each sample num-ber was labelled in the model for a better visual illustration and interpretation. By using the relative peak amount of HPLC peptide fingerprints, the dendrogram, score plot, and loading plot based on different classifications of EBN were generated.

From the score plot (Figure 4A), Red EBN and White/Yellow EBN could be unambiguously identified utilizing the first two princi-pal components, PC1 (t1) and PC2 (t2), which accounted for 85.8% of the total variation. The values of fitness of data (R2) and predic-tive ability (Q2) were 85.6% and 92.3%, respectively, in the model. Analysis showed no samples being outside the Hotelling T2 95% confidence ellipse that could influence the analyses. Based on the loading plot (Figure 4B), Mucin-5AC-like (peak D) showed the high-est positive loading score of 0.70, accounting for the colour discrimi-nation of samples by PC1. In agreement with the results of PCA, application of PCA–HCA (Figure 4C) for the whole data set resulted in a distinct classification of the samples according to their colours. Twenty-five samples were correctly classified according to their col-our; nonetheless, White EBN and Yellow EBN were not distinguish-able according to the score plot and dendrogram.

Since colour of EBN shows significant variation in fingerprint and PCA, only White EBN samples were used for comparison of production site of EBN to hold other factors constant. House White EBN and cave White EBN could be distinctly classified into two clusters using the first two principal components (t1 and t2), which explained a total of 82.7% variation (Supplementary Figure S3). Analysis showed no samples being outside the Hotelling T2 95% confidence ellipse. Fitness of data (R2) of this model was 82.7%,

Table 2. Protein identification by liquid chromatography–tandem mass spectrometry (LC–MS/MS). AMCase-like, acidic mammalian chi-tinase-like.

Peak* Protein** Major matched fragment Score Accession***

A UnknownB AMCase-like LYEGPSDTGDLV 43 XP_010005363.1C Lysyl oxidase homolog 3 LKGGAKVGEGRVEVLR 71 XP_010006484.1D Muc-5AC-like MWDKKTSIF 41 XP_009994736E AMCase-like AIGGWNFGTAKF 39 XP_010006620.1

*Peaks A–E in peptide fingerprint of EBN were collected (Figure 2A) and subjected to LC–MS/MS for protein identification. n = 3.**The peptide sequences were matched with Chaetura pelagica (Chimney swift) protein database in NCBI.***Accession number was reference to C. pelagica in NCBI database.

Dow




and the model was of predictive ability (Q2) of 63.0%. In order to evaluate the predictive power of this model, the third and fourth principal components (t3 and t4) were also investigated. Although house White EBN and cave White EBN could be classified into two clusters using the first and third principal components (t1 and t3) in both score plot and dendrogram, as indicated in Supplementary Figure S4, the predictive ability (Q2) of this model dropped to 57.46%. The R2 and Q2 value distinctly increased to 98.0% and 88.9%, respectively, by using first and fourth principal components

(t1 and t4) in the model. Furthermore, house White EBN and cave White EBN could be clearly classified into two clusters using the

Figure 4. The differentiation of chromatographic fingerprints of various EBN colour. Twenty-five EBN peptide fingerprints from different colours were generated: White (1–14); Yellow (15–18); Red (19–25). (A) Score plot of EBN samples. The ellipse represents the Hotelling T2 with 95% confidence. (B) Loading plot for five characteristic peaks (A–E as shown in Figure 2A). (C) Dendrogram showing hierarchical clustering results of EBN samples. EBN, edible bird’s nest.

Figure 3. The relative amounts of peaks A–E in peptide fingerprint of EBN. The peptide contents of peaks A–E, as indicated in Figure 2A, were calibrated according to the spike aprotinin, as shown in Figure 2B. (A) White EBN contained the highest amount of the five distinguishable peaks, while Red EBN contained the least. Minimum and maximum peptide content of different groups are depicted by black caps, the box signifies the upper and lower quartiles, and the median is represented by a short black line within the box for each types of EBN. (B) The peptide fingerprints of three White EBNs from different sources (Indonesia, Malaysia, and Vietnam) and two adulterated EBN being mixed with fake EBN [Sample 3 + Tremella fungus (1:1 by weight); Sample 4 + fried porcine skin (1:1)] were generated as indicated in Supplementary Figure S2. The amounts of peaks B–E for EBN (Indonesia), EBN (Malaysia), and EBN (Vietnam) were comparable to that of genuine EBNs. The amounts of peaks of EBN mixed with fake EBN were significantly lower than genuine EBN. Values are presented in relative amount as compared to that of aprotinin. Mean ± SEM, n = 4. EBN, edible bird’s nest.

Dow




first and fourth principal components (t1 and t4) in score plot (Figure 5A). AMCase-like (peak E) showed the highest positive load-ing score at 0.60 accounting for the discrimination of samples into

two clusters based on their production site, as illustrated in loading plot (Figure 5B). In agreement with results of PCA-X, application of PCA–HCA (Figure 5C) for the whole data set resulted to good

Figure 5. The differentiation of chromatographic fingerprints of EBN from production sites. Fourteen White EBN peptide fingerprints from different production sites were generated according to the procedures: House (1, 4, 5, 6, 10, 11, and 13); cave (2, 3, 7, 8, 9, 12, and 14). The relative peak area from the chromatographic fingerprints was calculated and used for PCA. (A) Score plot of EBN samples. The ellipse represents the Hotelling T2 with 95% confidence. (B) Loading plot for five characteristic peaks (A–E as shown in Figure 2A). (C) Dendrogram showing hierarchical clustering results of EBN samples. EBN, edible bird’s nest; PCA, principal component analysis.

Figure 6. The differentiation of chromatographic fingerprints of EBN from country of origin. Fourteen White EBN peptide fingerprints from different country of origin were generated according to the procedures: Indonesia (1, 4, 5, and 11); Malaysia (6, 10, and 13); Vietnam (2, 8, and 12); Thailand (3, 7, 9, and 14). The relative peak area from the chromatographic fingerprints was calculated and used for PCA. (A) Score plot of EBN samples. The ellipse represents the Hotelling T2 with 95% confidence. (B) Loading plot for five characteristic peaks (A–E as shown in Figure 2A). (C) Dendrogram showing hierarchical clustering results of EBN samples. EBN, edible bird’s nest; PCA, principal component analysis.

Dow




separation of the samples; and all of them were correctly classified according to their production sites.

In order to eliminate the critical variation from EBN colour in fingerprint and PCA, only White EBN samples were used for com-parison of EBN from different countries of origin. As illustrated in the score plot (Figure 6A), White EBN from different countries could not be clearly classified into clusters by using the first two princi-pal components (t1 and t2), which explained 82.7% of the total variation. The values of fitness of data (R2) and predictive ability (Q2) were 82.7% and 63.0%, respectively, in this model. Analysis showed no samples being outside the Hotelling T2 95% confidence ellipse. PCA–HCA (Figure 6C) showed that the peptide fingerprint could not clearly determine country of origin, which was consistent with the results of PCA. Moreover, EBN samples from Indonesia and Malaysia clustered on the right of score plot, and EBN samples came from Vietnam and Thailand clustered on the left. From the loading plot (Figure 6B), Mucin-5AC (peak D) showed the highest positive loading score at 0.80: this could serve as a tentative maker in discriminating the samples between Indonesia/Malaysia EBN and Vietnam/Thailand EBN.

Method validationSensitivity, linearity, accuracy, and precision of the HPLC peptide fin-gerprint technique were evaluated. Sensitivity was evaluated by the limit of detection (LOD) and limit of quantification (LOQ). A series of decreasing concentrations of aprotinin was evaluated to deter-mine the LOD and LOQ. LOD was determined as the concentra-tion with a signal-to-noise ratio (S/N) of at least 3; while LOQ was the lowest concentration with a S/N of at least 10. Linearity was evaluated by five calibrators prepared by spiking the aprotinin in blank matric at concentrations of 0.5, 1, 2, 5, and 10 µg. Precision and accuracy were evaluated on the peak area of the characteristic peaks. Intra-batch precision was evaluated by six determinations per single concentration in a day. Inter-batch precision was evaluated by one determination per single concentration at six different days. Precision was expressed as the percent of relative standard deviation (RSD) calculated by using six determinations. The value of recov-ery was expressed as the ratio of determined concentration from six individual tests to corresponding known concentration of the spiked marker. A known amount of aprotinin was spiked into the EBN digest, and recovery was calculated based on the concentration of obtained from the chromatogram and the spiked amount. The results of method validation were listed on Table 3.

Discussion

Traditionally, EBN is stewed with water for 1–2 h before eating, but most of the EBN proteins still remain insoluble: this stewing method accounts only ~5% protein extraction rate. Before any kind of protein analysis, the EBN proteins need to be dissolved in

water. In order to enhance good extraction efficiency of EBN, an over-stewing method was developed. Only water was utilized in our extraction: because this should be the method used for preparation of EBN for human consumption. About 66~84% of EBN proteins were extracted, supporting the hypothesis that over-stewing method could be a suitable way to extract EBN proteins. Using the same extraction protocol, White and Yellow EBN showed better extrac-tion efficiency as compared to that of Red EBN. The difference in extraction rate of White EBN (~84%) and Red EBN (~66%) may be due to conformational or structural difference of the major proteins. Indeed, much longer time was required to extract Red EBN, which implied Red EBN possessed higher resistance to protein digestion. Therefore, the protein in Red EBN might not be fully extracted by stewing. Moreover, very limited research has been reported on Red EBN, especially the protein part. Thus, further studies have to be conducted on the proteomics of Red EBN.

The lack of a complete genome sequence of Aerodramus pre-sents a challenge for protein identification in EBN. The identified peptide sequences were matched with animal species in NCBI data-base; however, no useful result was achieved. The present identified protein is based on C. pelagica that belongs to the same family as Aerodramus. Thanks to the complete genome sequence of C. pelag-ica, a member of the Apodidae, the EBN proteins were searched with this database. Muc-5AC-like, AMCase-like, and lysyl oxidase pro-tein fragments were identified in the fingerprint. Mucins are a family of glycoprotein with high molecular weight produced by epithelial tissues in most organisms (Marin et al., 2007). The major character-istic of mucin is that it is able to form a gel, which therefore is a key component in most gel-like secretions, e.g. saliva. In addition, the mucin secreted in airways is Muc-5AC, and as a result it is found in the saliva. Chitinases have a family of 18 glycosyl hydrolases; however, their physiological functions are still not fully resolved. AMCase belongs to the glycosyl hydrolase family in mammals (Boot et al., 2001; Bussink et al., 2007). In the present work, AMCase-like protein was reported in EBN for the third time, and it seemed strange that ‘mammalian’ chitinase was found in birds. Until now, chitinase protein sequences have not been found in the databases of C. pelagica and M. gallopavo, and thus only the homologues of AMCase-like protein were matched. Insects are the common food of Aerodramus, and the existence of chitinase in swiftlets could facilitate the digestion of insect. Lysyl oxidase is a copper-dependent amine oxidase that plays a critical role in the biogenesis of connec-tive tissue matrices by cross-linking the extracellular matrix proteins, collagen, and elastin (Li and Kagan, 1998). The presence of lysyl oxidase could explain why a previous study found that EBN could facilitate the synthesis of collagen (Chua et al., 2013).

In a 2011 study, nitrite was detected in EBN available in the mar-ket, especially Red EBN. This incident revealed the insufficient qual-ity control of EBN products in the industry. A variety authentication methods based on the appearance (Liu and Zhang, 1995; Wu et al.,

Table 3. Method validation. EBN, edible bird’s nest; LOD, limit of detection; LOQ, limit of quantification.

LOD* LOQ* Linearity** Correlation coefficient** (R) Intra-batch precision*** Inter-batch precision*** Recovery****

0.2 µg 0.5 µg 0.5–10 µg >0.995 1.5% 3.5% 90.7%

*LOD and LOQ were calculated by aprotinin. n = 6.**Linearity and correlation coefficient (R) were calculated by calibration curve of aprotinin. n = 6.***Intra-batch precision and inter-batch precision were reference to peak D in EBN peptide fingerprint. n = 6.****A known amount of aprotinin was spiked into the EBN digest; and recovery was calculated on the concentration of obtained from the chromatogram

and the spiked amount. n = 6.

Dow




2007), total proteins (Hu and Lai, 1999; Liu et al., 2012; Zhang et al., 2012), N-acetylneuraminic acid (Chan et al., 2013a), sialic acid (Wang et al., 2006), DNA (Wu et al., 2010), and sodium nitrate (Chan et al., 2013b) of EBN have been established; but no official method could be found for quality control of EBN by specific protein analysis, regard-less of its abundance. Here, we utilized SGF digestion and HPLC to generate the peptide fingerprint, which could authenticate EBN on a qualitative and quantitative basis robustly and therefore provide a comprehensive picture in quality control of EBN. The peptide finger-print was able to clearly demonstrate the differences between genuine and fake EBN samples, and thus this method could serve as an authen-ticating tool for EBN. Muc-5AC-like fragment (peak D) was the major peptide in the fingerprint of EBN, accounting of ~20% total content in the fingerprinting. This indicated mucin-like protein could be the most abundant protein in EBN. In addition, this mucin peak could serve as an internal marker for the fingerprinting. The chromatographic pat-tern of peptide fingerprint was highly similar among EBN samples with different colours, suggesting that total extractable protein com-position does not vary widely based on colour.

From the score plot and dendrogram, significant differences were recorded between White EBN and Red EBN. Red EBN is known to dissolve poorly in water, and indeed the protein extraction from EBN is much lower than that of White or Yellow EBN, as shown in the present work. This low solubility of Red EBN protein could account for such a difference, as observed in our PCA. White EBN could change to Red EBN by treating with sodium nitrite in acidic medium, as reported (But et al., 2013; Chan et al., 2013b; Paydar et al., 2013). Herein, Paydar group suggested Red EBN could be a result of formation of aryl-C-N and NO2 side group in aromatic amino acids of White EBN by sodium nitrite (Paydar et al., 2013). Thus, we proposed that this structural change in Red EBN could be the reason of low water solubility and high resistance to enzymatic digestion. Moreover, further studies have to be conducted on the underlying mechanism of how EBN gets red.

No significant difference was found among the country of ori-gin of White EBN, thus implying the protein composition should be more or less the same regardless their production sources. However, EBN samples from Indonesia and Malaysia clustered on the right of score plot, and EBN samples from Vietnam and Thailand clustered on the left. The geographic location may explain the difference, i.e. Indonesia and Malaysia or Vietnam and Thailand are neighbouring countries.

The PCA results suggested that there was a difference between house and cave White EBN in both dendrogram and score plot. The protein composition is probably different due to habitat of the swift-lets. Most of the caves are located at the seafront and on islands, whereas bird nest houses often are built in suburban areas or forests. The diet of swiftlets could be different and thus affect the secretion of EBN. Besides, the environment of a bird nest house can be main-tained and regulated by the farmers, which may be different from the natural cave environment. These findings complement a recent study by Cheng’s group whose differentiated between house and cave EBN based on amino acid composition (Seow et al., 2016).

Conclusion

PCA suggested that the peptide fingerprint could serve as a promis-ing and useful tool to classify EBN samples by colour and production site. As proposed by the model developed in this study, Muc-5AC-like and AMCase-like proteins were the markers with the highest discriminative power. In addition, HPLC fingerprints were able to

clearly demonstrate peptide profile differences between genuine and adulterated EBN samples, which provide a comprehensive picture in quality assurance of EBN. The protein identification results sug-gested that Muc-5AC-like protein was the major protein in EBN. To our knowledge, the present work is the first report highlighting the use of protein, which is main component of EBN, in HPLC finger-printing studies.

Supplementary Material

Supplementary material is available at Food Quality and Safety online.

Funding

This research was supported by the grants of Hong Kong Research Grants Council TBS (T13-607/12R), GRF (661110, 662911, 660411, 663012, 662713), ITC (UIM/254), TUYF12SC02, TUYF12SC03, TUYF15SC01, The Hong Kong Jockey Club Charities Trust (HKJCCT12SC01) and Foundation of The Awareness of Nature (TAON12SC01), SRFDP and RGC ERG Joint Fund (2013009614001/M-HKUST604/13) to Karl Wah-Keung Tsim; and National Natural Science Foundation Item (81173498) to Xiao-Ping Lai.

AcknowledgementsChun-Fai Wong received a HK JEBN scholarship.

Conflict of interest statement. None declared.

ReferencesAbdi, H., William, L. J. (2010). Principal component analysis. Wiley Interdis-

ciplinary Reviews: Computational Statistics, 2: 433–459.Aswir, A. R., Wan Nazaimoon, W. M. (2011). Effect of edible bird’s nest on

cell proliferation and tumor necrosis factor-alpha (TNF-α) release in vitro. International Food Research, 18: 1123–1127.

Boot, R. G. et al. (2001). Identification of a novel acidic mammalian chitinase distinct from chitotriosidase. Journal of Biological Chemistry, 276: 6770–6778.

Bussink, A. P., Speijer, D., Aerts, J. M. F. G., Boot, R. G. (2007). Evolution of mammalian chitinase (-like) members of family 18 glycosyl hydrolases. Genetics, 177: 959–970.

But, P. H., Jiang, R. W., Shaw, P. C. (2013). Edible bird’s nests—how do the red ones get red? Journal of Ethnopharmacology, 145: 378–380.

Chan, G. K., Zheng, K. Y., Zhu, K. Y., Dong, T. T., Tsim, K. W. (2013a). Deter-mination of free N-acetylneuraminic acid in edible bird nest: a develop-ment of chemical marker for quality control. Journal of Ethnobiology and Traditional Medicine, 120: 620–628.

Chan, G. K., Zhu, K. Y., Chou, J. Y., Guo, J. Y., Dong, T. X., Tsim, W. K. (2013b). Surveillance of nitrite level in cubilose: evaluation of removal method and proposed origin of contamination. Food Control, 34: 637–644.

Chan, K. L. et al. (2015). Edible bird’s nest, an Asian health food supplement, possesses skin lightening activities: identification of N-acetylneuraminic acid as active ingredient. Journal of Cosmetics, Dermatological Sciences and Applications, 5: 262–274.

Chinese Pharmacopoeia Commission, China. (2015). Pharmacopoeia of the People’s Republic of China. China Medical Science Press.

Chua, K. H. et al. (2013). Edible bird’s nest extract as a chondro-protective agent for human chondrocytes isolated from osteoarthritic knee: in vitro study. BMC Complementary and Alternative Medicine, 13: 19.

Codex Alimentarius International Food Standards. (1999). Recommended Methods of Analysis and Sampling, CODEX STAN 234.

Dow




Deng, Y. E., Sun, S. Q., Zhou, Q., Li, A. (2006). Analysis and discrimination of Collocalia esculenta L. via FTIR spectroscopy. Spectroscopy and Spectral Analysis, 26: 1242–1245.

Department of Health, Hong Kong. (2010). 4.3 High-performance liquid chromatography fingerprinting. Hong Kong Chinese Materia Medica Standards, Vol. 3. pp. 13–20.

Guo, C.-T. et al. (2006). Edible bird’s nest inhibits influenza virus infection. Antiviral Research, 70: 140–146.

Hou, Z. et al. (2015). Lactoferrin and ovotransferrin contribute toward anti-oxidative effects of Edible Bird’s Nest against hydrogen peroxide-induced oxidative stress in human SH-SY5Y cells. Bioscience, Biotechnology, and Biochemistry, 79: 1570–1578.

Hu, S. M., Lai, D. M. (1999). SDS-PAGE identification of edible bird’s nest. Journal of Chinese Medicine, 24: 331–334.

Jiangsu New Medicine College (Former of Nanjing Medical University). (1977). Dictionary of Traditional Chinese Medicine, Vol. 2. Shanghai Peo-ple’s Publishing House, China, pp. 2653–2654.

Kang, N., Hails, C. J., Sigurdsson, J. B. (1991). Nest construction and egg-laying in edible-nest swiftlets Aerodramus spp. and the implications for harvesting. IBIS, 133: 170–177.

Kong, Y. C., Keung, W. M., Yip, T. T., Ko, K. M., Tsao, S. W., Ng, M. H. (1987). Evidence that epidermal growth factor is present in swiftlet’s (Collocalia) nest. Comparative Biochemistry and Physiology, Part B: Biochemistry & Molecular Biology, 87: 221–226.

Kong, Y. C., Tsao, S. W., Song, M. E., Ng, M. H., Lin, Z. F. (1989). Potentiation of mitogenic response by extracts of the swiftlet’s (Apus) nest collected from Huai-Ji. Acta Zoologica Sinica, 35: 429–435.

Koon, L. C., Cranbrook, E. (2002). Swiftlets of Borneo – Builders of Edible Nests. Natural History Publication (Borneo) SDN, B.H.D., Sabah, Malay-sia, pp. 1–171.

Li, S. M., Kagan, H. M. (1998). Lysyl oxidase: properties, regulation and mul-tiple functions in biology. Matrix Biology, 16: 387–398.

Liu, X. Q. et al. (2012). Proteomic profile of edible bird’s nest proteins. Journal of Agricultural and Food Chemistry, 60: 12477–12481.

Liu, X., Zhang, J. S. (1995). Research on the identification of edible bird’s nest and its products. Chinese Journal of Tour Medical Science, 1: 26–28.

Ma, F., Liu, D. (2012). Sketch of the edible bird’s nest and its important bioac-tivities. Food Research International, 48: 559–567.

Marin, F. D. R., Luquet, G., Marie, B, Medakovic, D. (2007). Molluscan shell proteins: primary structure, origin, and evolution. Current Topics in Developmental Biology, 80: 209.

Matsukawa, N. et al. (2011). Improvement of bone strength and dermal thick-ness due to dietary edible bird’s nest extract in ovariectomized rats. Biosci-ence, Biotechnology, and Biochemistry, 75: 590–592.

Paydar M., Wong Y. L., Wong W. F., Hamdi O. A. A., Kadir N. A., Looi C. Y., (2013). Prevalence of Nitrite and Nitrate Contents and Its Effect on Edible Bird Nest’s Color. Journal of Food Science, 78: 1940–1947.

Seow, E. K., Ibrahim, B., Muhammad, S. A., Lee, L. H., Cheng, L. H. (2016). Differentiation between house and cave edible bird’s nests by chemometric analysis of amino acid composition data. LWT - Food Science and Tech-nology, 65: 428–435.

Wang, H., Ni, K. Y., Wang, Y. (2006). Determination of sialic acid in edible bird’s nest. Chinese Journal of Pharmaceutical Analysis, 26: 1251–1253.

Wu, R. H., Chen, Y., Wu, Y. J., Zhao, J. Y., Ge, Y. Q. (2007). Review of EBN authentication method. Inspection and Quarantine Science, 17: 60–62.

Wu, Y. J. et al. (2010). Application of SYBR green PCR and 2DGE methods to authenticate edible bird’s nest food. Food Research International, 43: 2020–2026.

Yew, M. Y., Koh, R. Y., Chye, S. M., Othman, I., Ng, K. Y. (2014). Edible bird’s nest ameliorates oxidative stress-induced apoptosis in SH-SY5Y human neuroblastoma cells. BMC Complementary and Alternative Medicine, 14: 391.

You, Y. Y. et al. (2015). Purification and identification of α 2–3 linked sialo-glycoprotein. European Food Research and Technology, 240: 389–397.

Zhang, S. W. et al. (2012). Competitive enzyme-linked immunoassay for sialo-glycoprotein of edible bird’s nest in food and cosmetics. Journal of Agri-cultural and Food Chemistry, 60: 3580–3585.

Dow



Characterization of edible bird's nest by peptide fingerprinting ...

Documents