Honey authenticity analysis: a proposed workflow using the SCIEX X500R QTOF System
KC Hyland, Diana Tran
SCIEX, USA
Introduction
Our global food supply chain is far reaching, complex, and profit-driven. A wide range of methods have been employed to screen food products and commodities to confirm authenticity and detect adulteration. Honey is one of the most commonly adulterated food commodities globally. One economically-motivated adulteration practice which has been observed in the global sales and trades of honey commodities is the dilution of the honey products with a cheaper sugar syrup, such as corn syrup. Dilution or fraudulent labelling can also occur in instances where honey products have disparate levels of value due to rarity or other unique properties. Techniques to screen for fraudulent or diluted products include various analytical techniques such as physical, chemical, or morphological assessments; for example, morphological pollen analysis for botanical origin of honey, or the C4 isotope test for the presence of corn-derived sugars. Mass spectrometry for testing food authenticity can be utilized for targeted, and/or nontargeted analytical methods. Targeted methods by nature rely on there being a known adulterant or residue for which products must be screened; examples include illegal dyes, or melamine, and are typically analyzed using multiple reaction monitoring (MRM) transitions on a triple quadrupole mass spectrometer.
Nontargeted mass spectrometry-based methods, common in the “Omics” disciplines (proteomics, metabolomics, lipidomics), can be employed in the identification and characterization of reliable marker compound(s) which would then be transferred to a targeted method for routine monitoring.1
This study shows the potential for the X500R QTOF System, SCIEX OS Software, MarkerView™ Software, and MS/MS libraries to be leveraged to develop and employ a nontargeted method for investigating honey chemical profiles and potential for adulterant screening.
Figure 1: Marker compounds for identifying and quantifying corn syrup adulteration in honey. A series of dilutions of a honey sample with corn syrup is used to demonstrate the ability to plot the response of these m/z features and thus illustrate a quantitative capacity to measure honey dilution with corn syrup. The MarkerView Software was used to pick out four mass/retention time features unique to the corn syrup relative to the honey samples. While the exact structure identification of these four compound features is as yet unknown, the workflow proposed in this study was employed to achieve this information and be able to show this potential for developing methods for accurate honey screening.
Advantages of the X500R QTOF System for honey authenticity testing
- SWATH® Acquisition allows for the collection of spectral data for all ionizable, detectable constituents in the honey sample. These data are then available for characterization and profiling of honey commodities and identifying unique chemical markers.
- SCIEX OS Software and SCIEX validated MS/MS library allow for the tentative identification of naturally occurring constituents in honey
- MarkerView Software allows for the rapid visualization of PCA and t-test statistical analyses in order to discern the most unique, and therefore most valuable, chemical features in the complex honey and corn syrup matrices
- The proposed workflow was employed to show that markers unique to corn syrup can be identified and used to screen and quantify dilution of honey products with corn syrup
Experimental
Sample preparation: Honey and corn syrup samples were purchased from local producers in order to best ensure that the samples were authentic in nature and also to capture different floral types of honey products. Eleven different honey samples were tested, and triplicate analyses were done for each. Honey was weighed out 1 gram and 10 milliliters of 50% methanol for LC-MS/MS analysis.
To mimic fraudulent honey products and assess the ability of the method to detect corn syrup adulateration, an experiment was performed in which honey samples were diluted with a series of increasing corn syrup concentrations, up to and including 100% corn syrup by mass, prior to sample extraction.
LC separation: A 10 µL volume of sample was injected using an ExionLC™ AD System coupled to the SCIEX X500R QTOF System. Separation was performed using a Phenomenex Luna Omega Polar C18 (150´4.6 mm, 3 um) analytical column. The LC mobile phases consisted of 0.1% formic acid in water (A) and 0.1% formic acid in methanol (B) at a flow-rate of 1 mL/min and column temperature of 25°C.
Mass spectrometry: Analysis was performed using the SCIEX X500R QTOF System, operated in both positive and negative polarity modes. The data independent acquisition strategy (SWATH Acquisition) was employed in order to collect high resolution precursor and product spectral information for all detectable constituents in the samples. The following MS source conditions were used: CUR=40 psi, CAD=11, IS =5500/-4500 V, TEM=500°C, GS1= 60 psi and GS2= 60 psi.
Variable window SWATH Acquisition was utilized to obtain high quality MS/MS spectra with 38 different precursor mass windows, the accumulation for the TOF MS is 0.1 sec and the accumuliation time for the TOF MS/MS is 0.025 sec.
Data processing: Data were processed using SCIEX OS-MQ Software 1.5 as well as MarkerView Software for statistical analyses. The SCIEX Natural Products 2.0 Library was used for searching database compound spectra for matches to experimentally derived spectra.
Table 1: Gradient conditions used for the LC separation. Flow rate of 0.3 mL/min was used.
Results
The SCIEX X500R QTOF System was used with SWATH Acquisition to collect high resolution mass spectral data on constituents present in the honey samples of different floral origins and honey diluted with corn syrup. The ability to collect high resolution MS1 data allows statiscal analyses to be applied to the different chemical profiles of the different sample sets. The additional collection of the comprehensive MS/MS spectral information afforded by the SWATH Acquisition allows for further exploration of those chemical profiles - including identifying candidate structural matches through use of spectral libraries.
Differentiating between honey variants with PCA
Acquired data were imported and processed with MarkerView Software. In this workflow, the software will first pick all the features present in the TOF MS data, each feature identified as an m/z and retention time pair. The feature profiles across the different samples can then be statistically compared to find differences between sample sets, and identify features which are uniquely present (up-regulated) or uniquely absent (down-regulated) in a particular sample. The Principal Component Analysis (PCA) shown in Figure 2 demonstrates the ability to use the SWATH Acquisition data acquired to statistically distinguish multiple honeys derived from three different floral origins.
Additionally, presenting the corn syrup data in the same PCA plot allows for a positive control; it is clearly observed that the corn syrup feature profile is highly differentiated from the honey samples in that it clusters far from the authentic honeys in the PCA plot. In practical applications, this type of analysis may serve two foreseeable purposes: to compare an unknown honey sample against a model built from data acquired for a large population of known authentic samples; or to profile honeys in order to investigate unique markers which may be present in different products of varying origins or processes.
Figure 2. Principal Component Analysis of honey variants. PCA output from MarkerView Software of honey variants shows visually that the chemical profiles between honey samples of different floral origins are distinct. Three different honeys of orange blossom origin, three different honeys of clover origin, and four different honeys of wildflower/mixed flower origin were used show this distinction. Additionally, corn syrup is observed to have unique characteristics that place it on the plot distinct from the honeys.
Investigating unique marker features using T-tests
The next step in the workflow is to identify specifically which m/z features represent unique markers for a particular sample type. This might be accomplished in one of several ways, but Figure 3 shows a direct comparison between two groups using a t-test, plotting a volcano plot of the m/z features in order to quickly and visually find those with the greater difference in signal (log fold change between the two compared sets) and greatest statistical significance (lowest p-value).
Figure 3. Volcano plot (p-value vs log fold change) constructed from T-test comparison of corn syrup sample set to combined set of all honeys. Green box shows features with the greatest upregulation in the honeys versus the corn syrup. Red box shows features with the greatest upregulation in corn syrup vs all honey.
Library searching for compound candidate ID
A primary advantage of acquiring data using SWATH Acquisition is the collection of MS/MS spectral information for every detectable precursor in the defined mass range. This allows product ion spectral information to be searched in a database for potential compound identification. For this study, the SCIEX Natural Products Library 2.0 was leveraged in order to attempt to identify some of the characteristic components of the different honey samples and see how these components’ occurrence varied between honey origins.
There were seven natural products which were all detected in at least one honey sample with a library match score of at least 75 and a corresponding mass error within 5ppm (most within 1ppm except when the level was very close to noise). These were Pinocembrin, Apigenin, Cardamomin, Luteolin, Quercetin, Chrysin, and Tectorigenin (Table 2). Five of these were identified by Cianciosi et al. as being among the most common polyphenols detected in honey.3
Figure 4 shows some example outputs from the SCIEX OS Software Analytics results table, in which can be seen the chromatographic peak, the TOF MS precursor data with empirical formula identification with FormulaFinder, and the MS/MS spectrum matched to a database entry and shown with a fit-based score representing how close the match is.
Table 2. Natural products identified in honeys using MS/MS. |
Figure 4. Screening the SWATH Acquisition data for the honey samples against the SCIEX Natural Products Library. Three identified natural products are shown as examples. From left to right is shown the chromatographic peak detected, the TOF MS spectrum and the corresponding empirical formula determination, and the acquired MS/MS spectrum mirrored with the matched MS/MS spectrum from the database with the compound name and fit score listed. A) Peak at 257.081 Da and 25.3 min. Formula match to C15H12O4 within 4ppm mass error. MS/MS match to Pinocembrin, a known metabolite in honey, with a 97.5 Purity score. B) Peak at 271.060 Da and 24.0 min. Formula match to C15H10O5 within 1ppm mass error. MS/MS match to Apigenin, a known metabolite in honey, with a 93.4 Purity score. C) Peak at 255.065 Da and 26.9 min. Formula match to C15H10O4 within 1ppm mass error. MS/MS match to Chrysin, a known metabolite in honey, with a 93.1 Fit score.
Relative quantitation of identified natural products
At this point in the study, the seven natural products identified in the honey samples could then be imported into SCIEX OS Software as a targeted components list. This targeted components list could then be applied to honey samples to achieve relative quantitation of these species in the different honey varietals. Figure 5 shows the amount of each of these as represented by the chromatographic peak area in the different honeys as well as in the corn syrup and extraction blank. It can be seen that while the profile of these flavonoids and other compounds varies between honey types, they do not appear to be present in either corn syrup or extraction blank.
Figure 5. Relative amounts of polyphenols detected in 11 different honey samples. Both the corn syrup and the extraction blank showed an absence of the natural products associated with the honeys. Different honey samples appear to have different polyphenol profiles. Error bars represent a standard deviation about the mean of triplicate analyses.
Dilutions with corn syrup
This experiment focused on one of the primary questions in testing honey for authenticity: whether or not the presence of a corn syrup diluent can be discerned. One representative honey product was selected, and made into dilutions with corn syrup at a range of concentrations from 0% corn syrup (pure honey), 25%, 50%, 90% and 100% corn syrup. The previous MarkerView Software statistical comparison showing the corn syrup t-test versus all combined honeys was used to pick out mass/retention time features unique to the corn syrup and not present in any honeys. Plotting the response of these features across the series of diluted honey samples thus illustrates capacity to measure honey dilution with corn syrup (Figure 1). While the exact structure identification of these compound features was not confirmed with structural elucidation or matching with an analytical stardard, the workflow proposed in this study was able to achieve this information and show potential for developing methods for accurate honey screening for the known adulterant corn syrup.
References
- Cajka, T., M.R. Showalter, K. Riddellova, O. Fiehn. (2016) Advances in Mass Spectrometry for Food Authenticity Testing: An Omics Perspective. In Woodhead Publishing Series in Food Science, Technology and Nutrition, 171-200.
- Lan X., Wang W., Li Q., J. Wang. (2016) The Natural Flavonoid Pinocembrin: Molecular Targets and Potential Therapeutic Applications. Mol Neurobiol. 53(3), 1794.
- Cianciosi D., Forbes-Hernández, T.Y., S. Afrin, et al. (2018) Phenolic Compounds in Honey and Their Associated Health Benefits: A Review. Molecules. 23(9), 2322.
- Noh, D., Choi, J., Huh, E., M. Oh. (2018) Tectorigenin, a Flavonoid-Based Compound of Leopard Lily Rhizome, Attenuates UV-B-Induced Apoptosis and Collagen Degradation by Inhibiting Oxidative Stress in Human Keratinocytes. Nutrients. 10(12), 1998.
- Kimura, Y., Takahashi, S.; Yoshida, I. (1968) Studies on the constituents of Alpinia. XII. On the constituents of the seeds of Alpinia katsumadai hayata. I. The structure of cardamomin. Yakugaku Zasshi: J. Pharm. Soc. Japan 88(2), 239.