A simple MALDI target plate with channel design to improve detection sensitivity and reproducibility for quantitative analysis of biomolecules

Overcoming the detrimental effects of sweet spots during crystallization is an impor-tant step to improve the quantitative abilities of matrix ‐ assisted laser desorption/ionization (MALDI) mass spectrometry. In this study, we introduce MALDI targets, which exhibit a channel design to reduce sweet spot phenomena and improve reproducibility. The size of the channels was 3.0 mm in length, 0.35 mm in depth, and 0.40 mm in width, adjusted to the width of the implemented laser beam. For sample deposition, the matrix/sample mixture was homogenously deposited into the channels using capillary action. To demonstrate the proof ‐ of ‐ principle, the novel plates were used for the quantification of acetyl ‐ L ‐ carnitine in human blood plasma using a combined standard addition and isotope dilution method. The results showed that the reproducibility of acetyl ‐ L ‐ carnitine detection was highly improved over a conventional MALDI ‐ MS assay, with RSD values of less than 5.9% in comparison with 15.6% using the regular MALDI method. The limits of quantification using the new plates were lowered approximately two ‐ fold in comparison with a standard rastering approach on a smooth stainless ‐ steel plate. Matrix effects were also assessed and shown to be negligible. The new assay was subsequently applied to the quantification of acetyl ‐ L ‐ carnitine in human plasma samples.


| INTRODUCTION
During the sample preparation for MALDI-MS, the analytes are cocrystalized with a chemical matrix such as α-cyano-4-hydroxycinnamic acid (CHCA) or 2,5-dihydroxybenozic acid (DHB). Because of commonly observed heterogenous co-crystallization of matrix and analytes, ion signals usually fluctuate strongly from laser shottoshot and from sample spottospot, as dried matrix/analyte mixtures form sweet spots on the sample targets. MALDI-MS is therefore often dismissed as a quantitative technique.
There are, however, multiple methods to improve the reproducibility of MALDI-MS to levels similar to LC-MS techniques. 1,2 For example, using an internal standard for the analytes that matches solution phase and ionization properties has been shown to strongly improve reproducibility. [3][4][5] The obvious choices for this purpose are stable isotope standards, but unfortunately, often these standards are not commercially available or very expensive. Alternatively, methods that improve the sample preparation procedures to force a homogenous distribution of matrix and analytes have been shown to significantly improve signal homogeneity. Methods that achieve these homogenous crystallizations include hydrophobic target coatings, 6,7 fast evaporation methods, 8 electrospraying techniques, 9,10 or seed layer approaches. 11,12 In addition, Zenobi and coworkers developed a selfaliquoting microarray plate that contains parallel lanes of hydrophilic reservoirs into which the samples were deposited using a metal sliding device. These arrays reduced sample requirements to just 10 μL per aliquot and exhibited excellent quantitative abilities, as demonstrated for angiotensin II and [Glu 1 ] fibrinopeptide B. 13 Finally, large-area graphene films were used as target surface to improve reproducibility of matrix/analytes preparation for quantification of biomolecules in MALDI-MS. 14 The downside of many of these techniques is that they require specialized hardware and that they are not as simple and convenient to use as conventional metal target plates. In this article, we describe a simple channel plate design to improve the reducibility of MALDI quantification, by forcing the matrix/analyte to dry within a narrowly confined cuboid of a few microliter internal volume on the steel target plates and subsequent scanning of the entire crystallized area using the MALDI laser beam. The quantitative abilities of this design were demonstrated for the analysis of acetyl-L-carnitine in human blood plasma matrices.

| Sample preparation
CHCA was prepared in methanol (TFA, 0.1% v/v) at 10 mg/mL. Acetyl-L-carnitine and d 3 -acetyl-L-carnitine were dissolved in water at 1mM as stock solutions. All stock solutions were stored at −20°C and diluted to the required concentration prior to use.
Samples were prepared from 100 μL of plasma and 100 μL of internal standard solution (20.00 μM). To this mixture, 5 mL of acetonitrile/methanol (3:1 v/v) were added, and the combined mixture vortexed for 10 seconds, followed by centrifugation at 13 000 rpm (5 min); the supernatant was transferred into a new plastic tube. Centrifugation was repeated once more, and the combined supernatants were evaporated under vacuum in a homemade system; 100 μL of water was added, followed by vortexing for 5 minutes at 35°C and centrifugation at 13 000 rpm (5 min).

| Calibration curves based on isotope dilution and standard addition
Acetyl-L-carnitine at different concentrations was added to plasma and vortexed for 30 seconds, followed by addition of CHCA solution.
A volume of 0.40 μL of this mixture was pipetted into the channels and dried under ambient environments.
For calibration, standard solutions of acetyl-L-carnitine at 0.50 μM, 1.00 μM, 2.00 μM, 5.00 μM, 10.00 μM, 20.00 μM, 37.50 μM, and 50.00 μM were used. Calibration curves were obtained by using a combined standard addition/isotope dilution method as described by Lee et al, to improve accuracy and precision 15 : In our experiments, c is,sol , V is,sol , and V sample were always invariant; therefore, both b and k were constants; c sample = concentration of native acetylcarnitine in the sample, IR sample = intensity ratio of

| Evaluation of matrix effects
To evaluate matrix effects, a calibration curve of acetylcarnitine from spiked pure solvent (water) was used, with identical concentrations as those used in human plasma. We assessed plasma matrix effects by comparing the signals of acetyl-L-carnitine in water with that in plasma. The relative ratio of the slopes b serum /b water was used as indicator for matrix effects.

| Trueness and precision
For trueness evaluation, 3 three quality controls (1.00 μM, 5.00 μM and 20.00 μM) were incorporated into each calibration curve and measured five times.
Precision was assessed by repeatability (intraday) and intermediate (interday) precision. Precision was determined from five same-day replicates and daily measurements over a period of 5 days. The results were expressed as % RSD.

| MALDI channel plates and sample preparation
Custom channel plates were made in the university workshop using commercial stainless steel plates ( Figure 1). Length, width, and depth of the channels were 3.0, 0.35, and 0.40 mm, respectively.

| SEM and light microscope images
The morphologies of CHCA/acylcarnitine co-crystals within the channels and on the surface of the conventional target plates were characterized by secondary electron microscopy (DEI Quanta 400, Hillsboro, Oregon) and Moticam 3.0 (Motic Deutschland, Wetzlar, Germany).

| RESULT AND DISCUSSION
In this study, we introduced a new target plate with customized channels to improve sample homogeneity and reproducibility in quantitative MALDI-MS. As shown in Figure 1, the channels were milled into the center of the sample spots of a regular MALDI plate, giving an 8 × 12 matrix of 96 channel wells. The length, width, and depth of each of the cuboid channels were 3.0, 0.35, and 0.40 mm, respectively. The width was marginally narrower than the diameter of the used laser beam, which was 0.40 mm, making sure that the beam covered the entire width of the channel, while the beam was moved along the length of the channel during data acquisition.
We used acetyl-L-carnitine in human plasma as analyte to demonstrate the abilities of the new plate. Acylcarnitine is an essential endogenous metabolite in mammals, usually at concentration levels in the range of 3 to 14 μmol/L in plasma. [16][17][18] As the metabolism of acetyl-L-carnitine is closely linked to a variety of metabolic problems, it is routinely monitored in clinical diagnostics, usually by using LC-MS techniques. As LC-MS methods typically include a time-consuming chromatography step, MALDI-MS was investigated here as a potentially faster alternative to LC-MS. 19 It has been previously demon-

| Comparison between channel and conventional target plates
The most desired feature of the new channel plate was to improve the quantitative reproducibility of MALDI analysis, by introducing constant volumes of sample and matrix solution into each of the welldefined channels and by illuminating the entire sampling surface with the laser light. In order to fill the channels with a well-defined, constant amount of sample solution, a 0.4 μL of the sample solution was pipetted into the channels (Figure 2A). The solutions spontaneously and evenly distributed within the channels by means of capillary action (a short video clip of this procedure is shown in the Supporting Matrix is added to channels using capillary action, and the analytes are added using a micropipette Information, with added methyl orange dye to improve visibility). As the solvent evaporates, co-crystals of analytes/matrix readily formed within the channels without the possibility for a coffee ring effect, similar to the forced co-crystallization of analyte/matrix on hydrophobic surfaces. 7,13 For comparison, the same solution was pipetted onto a conventional stainless-steel target plate, resulting in the well-known coffee ring effect ( Figure 2B).
The homogenous nature of the sample matrix crystals within channels, without any coffee ring formation, was further illustrated at higher magnifications using SEM (Figures 2C,D). According to Wong et. al, there is a minimal size for coffee ring structures for different materials. 23 The authors point out that for suspended particles of about 100nm size, the minimum diameter of the coffee ring structures is around 10 μm. In our channels, the surface is very rough, effectively dividing the channel surface into smaller subcompartments of few micrometers size or even smaller. We hypothesize that these small compartments are individually filled with CHCA and acetyl-L-carnitine crystals, preventing the formation of coffee ring structures in the channels such as those seen on the smooth regular steel surfaces.

| Reproducibility
Under optimized MALDI-MS/MS condition, we evaluated the reproducibility of the acetyl-L-carnitine signal produced from different channels across the plate. A plot of relative intensity (analyte/internal standard) of acetyl-L-carnitine from 10 different channels is shown in Figure 3. The relative standard deviation in these experiments was less  The improved signal reproducibility largely resulted from the complete consumption of the sample/matrix crystals within the channels, which effectively removed sweets spots issues. Complete laser ablation during quantitative MALDI has previously been shown to greatly improve precision. 2,13 In our experiments, the channels were 0.35 mm wide, while the laser beam's diameter was slightly larger, viz, 0.40 mm, thus enabling the entire width of the channel to be irradiated.

| Calibration curves
Standard addition method is commonly used to quantify endogenous compounds in complex biological samples, when no blank sample matrix is available. Here, we used this technique for the quantification of acylcarnitine in human plasma. The calibration curve was obtained by plotting the intensity ratio of acetyl-L-carnitine to d 3 -acetylcarnitne against the added concentration (Table 1). In comparison, a calibration curve of acetyl-L-carnitine from a conventional smooth target plate was used. Table 1 clearly shows, however, that all investigated analytical figures of merit obtained from the channel plate were superior to the conventional plate.

| Limits of quantification
LOQ was defined as the lowest concentration from the calibration curves that could be readily quantified with a precision of 20% RSD or better. From our experiments, the LOQ of acylcarnitine from the channel target plate was 0.50 μM, which was twofold lower than the LOQ from the conventional plate (Table 1).

| Matrix effects
We evaluated the effect of the plasma matrix by comparing the relative signal intensities in plasma and in pure solvent (water). The resulting calibration plots are illustrated in Figure 4. The slopes for acetyl-L-carnitine from plasma and water were 0.046 and 0.040, respectively. These values were used as indicators for matrix effects as described in experimental part. From these results, it could be seen that the matrix strongly influenced the acetyl-L-carnitine signal in plasma, requiring a combination of the standard addition method and isotope dilution analysis for correction.

| Human samples
Samples from two volunteers were analyzed using the developed method, and concentrations levels of 4.11 μM and 8.48 μM were determined (Table 2), which are comparable with levels typically seen in other studies. 16

| Trueness and precision
Trueness was assessed by analyzing quality control samples at low, medium, and high concentration levels in plasma ( Table 3). The