OPEN ACCESS DOCUMENT
Information of the Journal in which the present paper is published:
· Elsevier, Microchemical Journal, 2014 117 255–261
· DOI http://dx.doi.org/10.1016/j.microc.2014.07.
A non-target chemometric strategy applied to UPLC-MS sphingolipid analysis of a cell line exposed to chlorpyrifos pesticide: a feasibility study
Kássio M. G. Limaa,b*, Carmen Bediab, Romá Taulerb
aUFRN-IQ, Biological Chemistry and Chemometrics, 59072-970, Natal, Brazil
bIDAEA-CSIC, Jordi Girona 18, 08028 Barcelona, Spain
* Corresponding author: Kássio M. G. Lima, UFRN-IQ, Biological Chemistry and Chemometrics, 59072-970 Natal, Brazil. Tel.:+55 84 3342 2323; fax: +55 83 3211 9224.
A non-target chemometric strategy applied to UPLC-MS sphingolipid analysis of a cell line exposed to chlorpyrifos pesticide: a feasibility study
Kássio M. G. Limaa,b*, Carmen Bediab, Romá Taulerb
aUFRN-IQ, Biological Chemistry and Chemometrics, 59072-970, Natal, Brazil
bIDAEA-CSIC, Jordi Girona 18, 08028 Barcelona, Spain
Abstract: A non-target chemometrics study based on the application of Multivariate Curve Resolution Alternating Least Squares (MCR-ALS) method to a data set obtained by ultra-performance liquid chromatographic coupled to mass spectrometry (UPLC-MS) has been applied to the study of human prostate cancer (DU145) cell line samples treated with the organophosphate pesticide chlorpyrifos (CPF). Full scan UPLC-MS data sets were segmented in 17 different chromatographic windows and submitted to a non-target detailed study. Every one of these chromatographic windows of the different analyzed samples (treated and non-treated with CPF) was column-wise augmented in a new data matrix with their m/z values in the common column mode to preserve the fulfillment of the assumed spectral bilinear model. MCR-ALS was used to recover the elution and mass spectral profiles of the pure components present in each of the analyzed chromatographic windows. ANOVA (p £ 0.05) was then applied to compare the areas under the concentration profiles of the MCR-ALS resolved components in the CPS treated and control samples. This analysis allowed the detection of those sphingolipids having their concentration in cells modified by the presence of CPS compared to control samples where this contaminant was absent. Positively identified sphingolipids included sphingomyelins, dihydrosphingomyelin and C16 ceramide. The strategy described in this work is proposed for a general non-target UPLC-MS MCR-ALS analysis of the effect of environmental contaminants in cells in lipidomic and metabonomic studies.
Keywords: lipidomics; Chlorpyriphos; sphingolipids; cancer cells; MCR-ALS; UPLC-MS
* Corresponding author. Tel.: +55 84 3342 2323; fax: +55 83 3211 9224.
1. Introduction
Sphingolipids are a highly diverse family of lipids that serve not only as critical components of biological membranes but also as regulators of a vast number of cellular processes such as regulation of cell cycle, apoptosis, migration, inflammation, proliferation and recognition among others[1]. Two of the most studied sphingolipids, ceramide and sphingosine-1-phosphate (S1P), which are metabolically interconnected by two enzymatic steps, have opposite functions in cell signaling. Whereas ceramide mediates many cell-stress responses, like apoptosis and cell senescence, S1P has crucial roles in cell survival, migration and inflammation[2]. Some other bioactive sphingolipids include the sphingoid base sphingosine, ceramide-1-phosphate, glucosylceramide or dihydroceramide. Many of the bioactive sphingolipids in biological systems are often closely related structurally and metabolically forming an interconnected network of bioactive mediators whose relevance in homeostasis and disease is gaining scientific appreciation.
Different analytical strategies for sample preparation, ionization modes and instrumental designs have been proposed for the analysis of sphingolipids by mass spectrometry technology[3]. Design for this methodology has been provided structure specific, quantitative analysis of the “signaling” backbone species, Cer and Cer 1-phosphates, sphingoid base, sphingoid base 1-phosphates, N-acyl chaims, polar headgroups and others[4–8]. In these approaches for sphingolipids, some of the advantages that mass spectrometry provide are: (a) an in-depth profile of small samples (e.g., ~106 cells or even fewer); (b) a signal response which can be correlated to analyte concentration provided there are suitably matched internal standards to normalize for differences in ionization and fragmentation of individual molecular species; (c) a broad dynamic range which enables analysis of most of the compounds presents in biological samples.
Liquid chromatography – electrospray ionization – tandem mass spectroscopy (LC-ESI – MS/MS) is often employed for sphingolipid studies because it can lead to the development of fast and sensitive analytical protocols with high-throughput potential [9,10]. Nowadays, emerging development in analytical technologies such as fast high-resolution separation systems (e.g., ultraperformance liquid chromatography, UPLC) coupled with high-mass accuracy such as time-of-flight (TOF)[11], quadrupole-time-of-flight (Q-TOF)[12] also can provide more information from the sphingolipid experimental data generated.
In addition to the development of analytical technologies for sphingolipids, another key contributing factor to the rise of this field are the advances in data processing and bioinformatics[13–15]. The analytical platform in lipidomic experiments generates large amounts of data from a single sample of two-dimensional nature (chromatogram/mass spectra). For example, a typical data set obtained from a quadrupole instrument, scanning in the mass range of 100–1000 m/z, with 0.5 amu resolution sampled at 2.5 Hz for 30 minutes results in approximately 8 million data values. Some shortcomings can be usually overcome by chemometrics approaches such as denoising[16], compression of the data matrices[17] and models using the second order advantage[18]. Multivariate Curve Resolution (MCR) methods can be applied to the complete resolution of elution/concentration and mass spectra profiles for the different components present in very complex samples, such as those coming from metabonomic and lipidomic studies, analyzed by chromatographic methods,. Among multivariate curve resolution methods, the Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS) method has become a very popular chemometric tool which has been applied successfully to resolve multiple component responses from unknown unresolved mixtures[19–21].
Chlorpyrifos (CPF) is an important organophosphate endocrine disruptor pesticide, which has raised considerable concern in recent decades because it damages epithelial cells and acts mainly against the central nervous system [22,23]. In this work, a chemometric strategy based on MCR-ALS is applied to UPLC-MS three-way data arrays to perform a sphingolipid study in prostate cancer cell line samples (DU145) following treatment with chlorpyrifos. Informative UPLC-MS fingerprint sphingolipids data sets were segmented in 17 chromatographic windows for their non-target study. MCR-ALS was then applied on the augmented data matrices obtained from treated and non-treated samples. To test for statistically significative differences on resolved component areas upon CPS treatment, ANOVA was applied at every chromatographic window of different cell samples (control and CPF treated).
Results of the analysis of the sphingolipidome from cell extracts can contribute to a better understanding of the role of sphingolipids in the investigated context[24], in this case, in their involvement in the cytotoxicity of chlorpyrifos, on the prostate cancer cell line DU145. The goal of this study is therefore to increase the understanding of the biological toxic effects of CPF as endocrine disruptor pesticide on a prostate cancer cell line DU145. This study is based on a non-target chemometric analysis of the data sets obtained from UPLC-MS analysis of sphingolipid extracts of CPS treated and no-treated prostate cancer cell samples.
2. Experimental
2.1 Materials
Chlorpyrifos, cell culture media and reagents were obtained from Sigma. Analytical grade methanol and chloroform were purchased from Merck and Carlo Erba respectively. HPLC Gradient Grade acetonitrile was from Fischer Chemicals. Sphingolipid standards were obtained from Avanti Polar Lipids.
2.2 Cell Culture
DU145 prostate cancer cells were obtained from the American Type Culture Collection. This cell line was cultured in RPMI 1640 medium supplemented with 10% heat inactivated fetal bovine serum, 100U/mL penicillin and 100 mg mL-1 streptomycin, at 37ºC in a humidified atmosphere containing 5% of CO2. The experiments were carried out at low passage of cells.
2.3 Treatment of cells
Two million of DU145 cells were seeded in 10 cm diameter Petri dishes in 10 mL of RPMI media. After 24 hours, cells were treated with 25 mmol L-1 of chlorpyrifos or vehicle (DMSO) in triplicate. The DMSO concentration was 0.008% (v/v) and was without effect on cell viability (data not shown). After 24 hours of treatment, cells were harvested using a rubber scrapper into 2 mL of ice-cold PBS and counted. Cells were centrifuged at 1300 rpm for 3 minutes at 4ºC and cell pellets were washed twice with cold PBS.
2.4 Extraction procedure for sphingolipid analysis by UPLC-TOF
Sphingolipid extracts were prepared as described[10]. Briefly, 100 mL of deionized water were added to the cell pellets and the suspension was transferred to borosilicate glass test tubes with Teflon caps. Then, 500 mL of methanol and 250 mL chloroform were subsequently added. This mixture was fortified with internal standards of sphingolipids (N-dodecanoylsphingosine, N-dodecanoylglucosyl-sphingosine, D-erythro-dihydrosphingosine and N-dodecanoylsphingosylphosphorylcholine), 200 pmol each. Samples were sonicated until they appeared dispersed, then incubated overnight at 48ºC in a heating water bath. The tubes were then cooled and 75 mL of 1 mol L-1 KOH in methanol were added. After 2h incubation at 37ºC, KOH was neutralized with 75 mL of 1 mol L-1 acetic acid. The samples were then evaporated under N2 stream and transferred to 1.5 mL eppendorf tubes after addition of 500 mL of methanol. Samples were evaporated again and resuspended in 150 mL of methanol. The tubes were centrifugated at 10000 rpm for 3 minutes and 130 mL of the supernatants were transferred to UPLC vials for injection.
2.5 Liquid chromatography and mass spectrometry
The LC/MS analysis consisted of a Waters Aquity UPLC system connected to a Waters LCT Premier orthogonal accelerated time of flight mass spectrometer (Waters), operated in positive electrospray ionization mode. Full scan spectra from 50 to 1500 Da were acquired, and individual spectra were summed to produce data points each of 0.2s. Mass accuracy and reproducibility were maintained by using an independent reference spray via the LockSpray interference. The analytical column was a 100 X 2.1-mm inner diameter, 1.7 mm C8 Acquity UPLC bridged ethylene hybrid (Waters). The two mobile phases were phase A: MeOH/H2O/HCOOH (74:25:1, v/v) and phase B: MeOH/ HCOOH (99:1, v/v); both contained 5 mmol L-1 ammonium formate. The column was held at 30 °C.
2.6 Peak assignment and identification of sphingolipids
Positive identification of compounds was based on the accurate mass measurement with an error of <5 ppm and its LC retention time, compared to the data of a previously elaborated homemade database of sphingolipid standards injected under the same chromatographic and spectrometric conditions[25].
2.7 Data analysis
Figures 1 and 2 show a detailed scheme of the different steps involved in this study.
[Insert Figure 1 here]
After the extraction procedure for sphingolipid analysis, each UPLC-LCMS chromatographic run (see Experimental section) recorded for every sample (three treated and three nontreated samples) give a data set which was stored in ASCII format by the Databridge function of MassLynxTM V 4.1 software. This data set obtained in the analysis of every sample was then imported in MATLAB 7.10.0 (R2010a) computational environment using an in-house program specially designed for this purpose. The size of the two-way data matrix generated by this program for every sample was 1275 x 9001 (retention time and m/z values, respectively) and 0.1 resolution. Data from every chromatogram were normalized by the added internal sphingolipid standard area and by the number of cells present in each sample (see Experimental section) m/z data values were binned to 4501 using level one Daubechies simpler wavelet[26], without losing relevant chemical information and filtering noise. Every reduced data matrix was then subdivided in seventeen chromatographic time windows (as shown in Table 1 and Figure 1) to simplify the overall complexity of the data sets.
[Insert Table 1 here]
As with any other instrumental signal, chromatograms contain three major components: signal, noise and background baseline. Elimination of the chromatogram baseline is a critical step for reducing the complexity of the measured chromatograms and facilitates their analysis. With this goal in mind, the methodology of asymmetric least-squares[27], was applied to the different chromatographic windows of every analyzed sample.
A short description of the Multivariate Curve Resolution – Alternating Least Squares (MCR-ALS) method used in this work is given here. For a more detailed description of the method, see refs [21,28]. This algorithm is based on a bilinear model (Equation 1) that decomposes the data matrix D, containing the raw information about all the components present in a data set, into the product of two matrices C and ST, containing the pure response profiles associated with the variation of each contribution in the row (matrix C) and the column directions (matrix ST).
D = CST + E (1)
In the case of hyphenated chromatographic spectroscopic detection data, every analyzed sample gives a data matrix D of dimensions (I,J) which contains the MS spectra (rows) at all retention times (i=1,…I) in its rows, and the chromatograms at all spectra m/z channels (j=1…J) in its columns. C matrix of dimensions (I,N) has the elution or concentration profiles of each resolved component (n=1,N) and matrix ST of dimensions (N,J) has the pure spectra of these components, respectively. E is the error matrix of dimensions (I,J), i.e., the residual variation of the data set that is not related to any of the resolved components. Decomposition of data matrix D is achieved by an alternating iterative least-squares minimization of E under suitable constraining conditions such as nonnegativity in spectral and concentration profiles. The matrix C contains the N elution profiles (column-wise) and the matrix ST contains the MS spectra (row-wise) of the N resolved components.
In this work, Multivariate Curve Resolution Alternating Least Squares (MCR-ALS) was applied to the analysis of the whole set of MS chromatographic runs obtained in the analysis of the investigated prostate cancer cell line (CPS treated and control) samples. Although MCR-ALS could have been applied individually to every chromatographic window of the data matrix obtained in the analysis of every cell sample, simultaneous analysis of data matrices obtained in the analysis of a particular chromatographic window for the six different analyzed samples, three control ones and three CPS treated ones, was preferred, both to improve the resolution of the different coeluted components and to allow their relative quantitative estimation and comparison.
[Insert Figure 2 here]
In Equation 2 and Figure 2, MCR-ALS bilinear data decomposition analysis is shown for the simultaneous analysis of the data matrices from the same chromatographic window for both treated and control samples, setting one on top of the other and keeping their column vector space in common (augmented data matrices) as it is shown in next equation: