Properties of the parylene matrix chip
MALDI-TOF MS has been commonly utilized for the detection of biomolecules through mild ionization using organic matrix molecules (Fig. 1a). Conventional organic matrices for MALDI-TOF mass spectrometry have been reported to produce unreproducible mass peaks in the low m/z range, which interfere with the detection of small molecules, especially those with a molecular weight of less than 1 kDa. When the conventional organic matrix (CHCA) was measured repeatedly at the same spot, unreproducible mass peaks from fragmented organic matrix molecules were observed in the low m/z range, as shown in Fig. 1b. The blue arrows of the mass peaks displayed different intensities, and the red arrows of the mass peaks sometimes disappeared in the mass spectra. Recently, MALDI-TOF mass spectrometry based on the parylene matrix chip has been developed for the quantitative analysis of small molecules, which could remove the interfering mass peaks from the fragmented organic matrices in the low m/z range of less than 500 (Kim et al. 2017b, 2014a, 2014b; Park et al. 2017). As previously reported (Kim et al. 2017b, 2014a, 2014b), a parylene matrix chip was produced via the deposition of a parylene film on a dried organic matrix, such as CHCA and DHB. Usually, the direct exposure of a dried matrix spot to a UV laser produces unreproducible mass peaks from fractions of matrix molecules. When a parylene film with a thickness of less than 50 nm was deposited on the dried matrix spot, the mass peaks from matrix molecules were observed to disappear upon absorption of the parylene film, as shown in Fig. 1c. Through addition of the sample, only the mass peaks from analytes were observed in the sample without any mass peaks from matrix molecules. The function of the parylene film was previously proposed to result from the mixing of the analyte between the dried matrix spot and the liquid sample within the porous parylene film. The porosity of the parylene film with a thickness of less than 50 nm, as well as the transfer of small ions through the porous parylene film, was demonstrated using cyclic voltammetry on a gold electrode, as shown in Additional file 1: Fig. S1 (Kim et al. 2014a). However, the mixing zone for analytes and matrix molecules has not been reported to be directly observed within the porous parylene film. To prove the existence of a mixing zone within the porous parylene film, an analog structure of the parylene matrix chip was developed, as shown in Fig. 2a. The fluorescence probe called Amplex Red (transparent with no fluorescence) was dried on the target plate prior to the deposition of the parylene film with a thickness of 50 nm. The fluorescence probe is oxidized into resorufin (with fluorescence, \(\lambda\)em = 590 nm) by HRP mixed with hydrogen peroxide (H2O2), as shown in Fig. 2b. Red fluorescence of resorufin was observed in the Amplex Red chip without parylene-N film deposition, and the morphology was the same as that of the non-homogeneous dried HRP/H2O2 mixture crystal (Fig. 2c). The parylene-N film was deposited on the Amplex Red spot after the HRP/H2O2 mixture solution was dried, and red fluorescence was also observed at the cross section of the parylene film. This result suggests that the HRP/H2O2 solution was mixed with the dried Amplex Red reagent through the parylene-N film. Therefore, a sample-matrix mixing zone of the parylene matrix chip was formed below the surface of the parylene-N film.
The inhomogeneous crystal formation of the matrix-analyte mixture caused difficulty in the quantitative analysis (Cohen and Gusev 2002). Fluorescence was detected by dispensing fluorescein dissolved in sodium hydroxide (0.1 M) on the bare plate, organic matrix, and parylene matrix chip. From the fluorescence image, as shown in Fig. 2d, the matrix in the parylene matrix chip showed a homogeneous distribution compared with the bare plate (stainless steel plate) and CHCA organic matrix spot. Accordingly, organic matrix molecules were considered to be homogeneously distributed on the surface of the parylene matrix chip, where the analyte molecules could be ionized with enhanced reproducibility. Homogeneous surfaces enhance the reproducibility of analytes in MALDI-TOF mass spectrometry (Ha et al. 2011; O'Rourke et al. 2018; Szaéjli et al. 2008).
Analysis of vitamin D
In this section, vitamin D metabolites were detected after chemical derivatization using MALDI-TOF mass spectrometry based on parylene matrix chip. The detection of vitamin D metabolites has been difficult due to their hydrophobic characteristics, lack of polar functional groups, and low concentrations in plasma. Among various vitamin D metabolites, 25(OH)D3 has been selected as a biomarker for various diseases (Ding et al. 2010; Zerwekh 2008).
For effective detection, derivatization methods of vitamin D have been studied, and BA was selectively reacted to the hydroxyl group of the 25(OH)D3 as well as vitamin D3 by nucleophilic addition to form a hemiacetal salt with a charge labeling effect, as shown in Fig. 3a. The BA-derivatized 25(OH)D3 and vitamin D3 could be detected using the conventional organic matrix called CHCA, and the mass peaks from the fragmented matrix molecules could be effectively removed using MALDI-TOF mass spectrometry based on the parylene matrix chip, as shown in Fig. 3b. When the mixture of BA-derivatized 25(OH)D3 and BA-derivatized vitamin D3 was analyzed using the parylene matrix chip, both target analytes could be simultaneously detected without the mass peaks from the fragmented matrix molecules.
Another chemical derivatization method was employed to prepare isotachysterol via the acid-catalyzed isomerization of vitamin D in ethanol (Agarwal 1990; Jin et al. 2004; Seamark et al. 1980). The dehydrated form of isotachysterol was detected under MALDI-TOF mass conditions as degraded ion peaks, as shown in Fig. 3c (Mahmoodani et al. 2017). The derivatized form of 25(OH)D3 was [25(OH)D3 + H-H2O]+ (m/z = 383.7), and the derivatized form of D3 was [D3 + H-H2O]+ (m/z = 367.6) based on conventional MALDI-TOF mass spectrometry using an organic matrix (CHCA) as well as a parylene matrix chip.
The Diels–Alder reaction of vitamin D with 4-phenyl-1,2,4-triazoline-3,5-dione (PTAD), including atoms with high proton affinity (such as oxygen and nitrogen), was also employed (Aronov et al. 2008; Ding et al. 2010; Higashi et al. 2001, 2008). However, PTAD was not effective for the modification of vitamin D molecules owing to the low solubility of PTAD in protic solvents, long reaction time, and difficult reaction with a low sample volume. The derivatized product of vitamin D molecules was observed to have a chemical structure without any charged functional groups. Further, no mass peaks were observed based on MALDI-TOF mass spectrometry using organic matrix molecules or a parylene matrix chip, as shown in Fig. 3d. Owing to these results, the vitamin D molecules were analyzed after derivatization of BA, and MALDI-TOF mass spectrometry based on the parylene matrix chip was used for effective detection at high S/N ratios without mass peaks from fragmented organic matrix molecules.
Quantitative analysis of vitamin D molecules was performed using a parylene matrix chip. As previously mentioned, the inhomogeneous crystal formation of the matrix-analyte mixture caused difficulty in the quantitative analysis, and the parylene matrix chip was demonstrated to be useful for quantitative analysis using MALDI-TOF mass spectrometry. The feasibility of the quantitative analysis of vitamin D molecules was investigated by estimating the reproducibility of the inter-spot and intra-spot measurements. To measure inter-spot reproducibility, five sample spots were independently prepared using BA-derivatized vitamin D molecules on the parylene matrix chip, and the MALDI-TOF mass spectra were obtained at different spots, as shown in Additional file 1: Fig. S2(a) and Fig. S2(b). Based on the analysis of the mass spectra, the inter-spot reproducibility was 4.2% for 25(OH)D3 (0.45 pmol/μL) as shown in Fig. 4a, and 3.1% for vitamin D3 (0.45 pmol/μL), as shown in Fig. 4b. To estimate intra-spot reproducibility, a sample spot was prepared using BA-derivatized vitamin D molecules on the parylene matrix chip, and the MALDI-TOF mass spectra were obtained at different positions, as shown in Additional file 1: Fig. S2c, d. Based on the analysis of the mass spectra, the intra-spot reproducibility was 3.0% for 25(OH)D3 (0.45 pmol/μL) as shown in Fig. 4c and 2.6% for vitamin D3 (0.45 pmol/μL), as shown in Fig. 4d. Therefore, the analysis of BA-derivatized vitamin D molecules could be conducted with high inter- and intra-spot reproducibility using MALDI-TOF mass spectrometry based on a parylene matrix chip.
For the quantitative analysis of vitamin D molecules, standard samples of BA-derivatized 25(OH)D3 and vitamin D3 in ethanol were prepared in the concentration range of 0.0056–1.35 pmol/μL. Further, mass spectra with the targeted mass peak of 25(OH)D3 ([25(OH)D3 + BA]+ = 502.5 and vitamin D3 ([D3 + BA]+ = 486.5 were obtained using MALDI-TOF mass spectrometry based on the parylene matrix chip, as shown in Fig. 5a and Additional file 1: Fig. S3(a). The standard curves for derivatized 25(OH)D3 (with a linearity factor of 0.994) and vitamin D3 (with a linearity factor of 0.999) were plotted, as shown in Fig. 5b, and the limit of detection (LOD) was estimated to be 0.0056 pmol/μL for 25(OH)D3 and 0.0056 pmol/μL for vitamin D3 with a cut-off value of 0.026 pmol/μL. The LOD was marked with an asterisk (*), and the cut-off value was marked with a double asterisk (**) in the standard curves of 25(OH)D3 and vitamin D3. The standard samples of 25(OH)D3 and vitamin D3 in serum were prepared by spiking an amount ranging from 0.0056 to 1.35 pmol/μL. Further, the vitamin D molecules were isolated from serum by liquid–liquid extraction with acetonitrile and hexane prior to derivatization with BA. The mass spectra with the targeted mass peak of 25(OH)D3 ([25(OH)D3 + BA]+ = 502.5, and vitamin D3 ([D3 + BA]+ = 486.5 were obtained using MALDI-TOF mass spectrometry based on the parylene matrix chip, as shown in Fig. 5c and Additional file 1: Fig. S3(b). The standard curves for derived 25(OH)D3 (with a linearity factor of 0.994) and vitamin D3 (with a linearity factor of 0.999) were plotted, as shown in Fig. 5d, and the limit of detection (LOD) was estimated to be 0.0056 pmol/μL for 25(OH)D3 and 0.0056 pmol/μL for vitamin D3, with a cut-off value of 0.026 pmol/μL. These results indicate that 25(OH)D3 and vitamin D3 in ethanol and serum could be quantitatively determined with high linearity using MALDI-TOF mass spectrometry based on a parylene matrix chip.
The standard techniques for the quantification of 25(OH)D3 in blood include enzyme-linked immunosorbent assay (ELISA) and liquid chromatography-tandem mass spectrometry (LC–MS/MS) (Hollis and Napoli 1985; Jenkinson et al. 2016; Kim et al. 2017a). In this study, Elecsys Vitamin D analysis kit from Roche (Mannheim, Germany) was used as a reference and its quantification results for 25(OH)D3 were compared to those from MALDI-TOF mass spectrometry based on a parylene matrix chip. The quantification results from MALDI-TOF mass spectrometry and ECLIA are plotted in the same graph in Fig. 6a, and these standard curves show a high linearity for both methods. To estimate the statistical co-incidence of the two methods for vitamin D3, the Bland–Altman plot and Passing–Bablok regression were conducted using MedCalc software (version 18.6). The Bland–Altman test revealed that a signal difference was distributed at a confidence level of 95% (\(\pm\) 1.96 \(\upsigma\)), as shown in Fig. 6b. This result demonstrated that the two methods were highly correlated and provided the same analysis result for the detection of 25(OH)D3. Passing–Bablok regression revealed that the analysis data from both methods were distributed at a confidence level of 95%, with a Spearman correlation coefficient (\(\uprho\)) of 1.000 (P < 0.0001), as shown in Fig. 6c. These results show that the two different methods were statistically highly coincident. This result also showed that the two methods were highly correlated and provided the same analysis results for the detection of 25(OH)D3.
The applicability of MALDI-TOF mass spectroscopy based on a parylene matrix chip for the quantification of vitamin D3 in real samples was presented using commercially available energy drinks and vitamin D3 tablets as real samples. The energy drink mainly contained vitamin C, vitamin D3, zinc, biotin, and sodium, according to the manufacturer’s label. An energy drink was added to the BA solution for BA derivatization and then loaded onto the parylene matrix chip for mass analysis. As shown in Fig. 7a, both vitamin C and vitamin D3 reacted with BA. The mass peak from vitamin C was observed to have a higher intensity than that from vitamin D3, indicating that the relative amount of vitamin C was markedly higher than that of vitamin D3, as indicated on the manufacturer’s label (vitamin C: 13.0 pmol/μL, vitamin D3: 0.65 pmol/μL). Vitamin D3 tablet as a dietary supplement was also analyzed using the same method, and the mass peak from BA-adducted vitamin D3 was measured without matrix interference (Fig. 7b). The amount of vitamin D3 was estimated to be 0.563 pmol/μL in energy drink (0.65 pmol/μL, according to the manufacturer’s label, recovery = 86.6%) and 0.942 pmol/μL for vitamin D3 tablet (1.0 pmol/μL in the manufacturer’s label, recovery = 94.2%) from the standard curve, as shown in Fig. 7c. Thus, the amount of vitamin D3 from the measurement was well matched with that on the manufacturer’s label, and the applicability of MALDI-TOF mass spectrometry based on the parylene matrix chip was estimated to be feasible for real sample analysis using energy drinks and vitamin D3 tablets.