High-Throughput Reactive Desorption Electrospray Ionization Mass Spectrometry for Targeted Derivatization and Analysis of Poorly Ionized Small Molecules

Abstract

Mass spectrometry (MS) analysis of some biologically relevant compounds is limited due to their inherently poor ionization efficiency. This challenge can often be overcome by incorporating easily ionized moieties via functional group-targeted chemical derivatization. However, the wide variety of derivatization agents and conditions can make method development cumbersome. In this study we employ an automated high-throughput (HT) platform based on desorption electrospray ionization (DESI) MS to rapidly screen (1 sample per second) and select appropriate derivatization strategies for poorly ionized analytes leveraging accelerated on-the-fly microdroplet reactions. This approach allowed the rapid identification of efficient derivatization strategies that were then readily applied to the qualitative and quantitative analysis of molecules of biological importance. Specifically, hydroxysteroids (e.g. cholesterol, testosterone, cholecalciferol) were imaged in tissue sections using 4-formyl-1-methylpyridinium benzenesulfonate without the loss of the intrinsic spatial resolution of DESI, while 4-borono-N,N,N-trimethylbenzenaminium allowed sensitive and HT quantitation of urinary 3-methoxy-4-hydroxyphenylglycol, a metabolite whose urine levels have been correlated with neurological disorders.

Graphical Abstract

INTRODUCTION

Many biologically relevant compounds, such as neurotransmitters,^1–3 hormones^4–6, vitamins^7–9, and their metabolites, are inherently difficult to detect by mass spectrometry (MS) due to their low concentrations and poor ionization efficiency. A widely adopted strategy to overcome these limitations is chemical derivatization, which introduces functional groups that either carry a permanent charge or that otherwise enhance ionization efficiency.^10–12 Targeting functional groups that are broadly represented across classes of biological molecules offers a rational approach to expedite these derivatization strategies, since reactions optimized for one compound can often be extended to structurally related molecules.^13–15 However, identifying broadly applicable functional group-specific reactions requires systematic evaluation of a large and chemically diverse set of biologically relevant compounds—a task ideally suited for high-throughput (HT) experimentation. Furthermore, HT approaches could also facilitate the optimization of derivatization reaction hits, speeding up their translation to analytical applications.

Here we describe the implementation of an automated HT (1 sample per second) platform based on desorption electrospray ionization (DESI)-MS for the discovery of selective and fast chemical derivatization strategies. DESI, an ambient ionization technique, operates by directing a charged solvent spray onto a surface, forming a thin liquid film in which analytes are extracted and subsequently desorbed into secondary splashed microdroplets that are carried to a mass spectrometer for analysis.^16–18 This mechanism allows use of DESI for direct (i.e. preparation-free), contactless, and spatially resolved analysis of complex samples at high rates, paralleling its many demonstrated applications as an analytical tool in molecular imaging^19–21 as well as HT bioanalysis^22–24. Importantly, DESI can also function as a synthetic tool by exploiting the reaction acceleration phenomenon that takes place in the secondary microdroplets generated at each desorption event.^25–27 In these small droplets, with large surface-to-volume ratios, organic reactions are accelerated by up to six orders of magnitude compared to bulk solution.²⁸ This acceleration is attributed to the unique properties of the droplet surface (i.e. air-solution interface)^28,29, including partial solvation of reagents³⁰ and the strong interfacial electric field^31,32 which can influence molecular orientation, the generation of highly reactive species, and extreme values of local pH^33,34. The combination of accelerated synthesis and in operando high-speed analysis, both inherent capabilities of DESI, makes this technology a unique strategy for rapid chemical space exploration via droplet-based HT reaction screening.^17,35–37

EXPERIMENTAL METHODS

Chemicals and Materials

The biological substrates were purchased from Cayman Chemical (Ann Arbor, MI), Sigma Aldrich (St. Louis, MO), and MedChemExpress (Monmouth Junction, NJ). The functionalization reagents, 2-chloro-1-methylpyridinium iodide (CMPI), N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), and 4-formyl-1-methylpyridinium benzenesulfonate (FMPB) were acquired from Sigma Aldrich, while 4-borono-N,N,N-trimethylbenzenaminium (BTBA) was purchased from AstaTech (Bristol, PA). All solvents for the reagent solutions, including water, dimethyl sulfoxide (DMSO), acetonitrile (ACN), and chloroform were LC-MS grade (Fisher Scientific, Waltham, MA). DESI slides are prepared in house using custom-sized soda-lime glass slides (Abrisa Technologies, Santa Paula, CA) coated with a porous PFTE membrane (Zitex G-115, Saint Gobain, Malvern, PA) using a light layer of low-VOC spray adhesive (3M, St. Paul, MN).

Reaction screening

Biological substrate stocks (5 mM) were prepared from solid reagents using an appropriate solvent based on the substrate solubility (see Table S1). Final dilutions of all the substrates were made in ACN (100 μM) and transferred to a 384-well plate (V-bottom, polypropylene; Eppendorf, Framingham, MA). All functionalization reagents were similarly prepared as 5 mM stock solutions, diluted to 100 μM with ACN and transferred into separate 384-well plates.

Reaction screening was carried out using a new-generation HT-DESI-MS platform comprised of a Vantage 2.0 fluid handling workstation (Hamilton Company, Reno, NV) and a Xevo^™ G2-XS quadrupole time-of-flight mass spectrometer coupled to a prototype AutoDESI source equipped with a DESI XS sprayer (Waters Corporation, Milford, MA) and a homebuilt heated transfer capillary. The system is a fully integrated and automated platform operated via custom software and a seamless driver. Briefly, the reaction screening workflow (Figure 1) involved pinning of the biological substrates on DESI plates using a 50 nL floating slotted 384-pin tool (V&P Scientific, San Diego, CA), followed by pinning of the functionalization reagents onto the same positions. Spotted plates were directly transferred to the AutoDESI stage and subjected to HT-DESI-MS screening. Note that three dye reference spots were pinned in predefined positions near three corners of the plate for position calibration after DESI slide transfer to the stage. Derivatization reactions occurred during the flight time of the microdroplets generated by the DESI source to the mass spectrometer inlet (millisecond timescale). Four independent replicates were prepared and screened for each unique reaction condition, with each one being pinned four times (i.e. technical replicates), yielding a total of 16 replicates per reaction.

Figure 1.

HT-DESI-MS/MS workflow for online functionalization. The compounds to be screened are first spotted onto a PTFE-coated plate generating high-density arrays with up to 6,144 spots per plate. Functionalization reagents are then spotted in the same positions (i.e. on top of the compounds) generating reaction mixtures that are then subjected to automated HT-DESI-MS screening. Throughout the DESI process, secondary microdroplets containing the desorbed mixtures of substrate and derivatization reactant accelerate the reactions so that products are generated on-the-fly (millisecond time scale) and then detected via single-stage MS at high speed (~1 Hz). Hits based on signal-to-noise ratios are finally confirmed by reanalysis of successful reaction mixture spots via online MS/MS experiments at lower throughputs (~0.3 Hz).

The DESI spray was generated using LC-MS grade methanol (Chromasolv Riedel-de Haën, Honeywell, Charlotte, NC) at a constant flow rate of 2 μL/min supplied via a syringe pump (Harvard Apparatus, Holliston, MA) using a gas-tight syringe (Hamilton Company, Reno, NV), and nitrogen as nebulizing gas (18 psi regulated on the stage). All experiments were carried out in the positive ion mode using a spray voltage of 0.7 kV and the “Sensitivity” setting of the instrument. The transfer capillary and the source temperature were set to 450 °C and 150 °C, respectively. All other ion optical parameters were kept at their default values. The scan time was fixed at 100 ms in all cases. The rastering motion of the AutoDESI stage (i.e. pattern followed within each sample spot) was set as a zigzag with a movement rate of 3500 μm/s, providing a throughput of ~1 second per sample. Raw spectral data was processed and analyzed using a custom MATLAB (MathWorks, Natick, MA) app. MS/MS experiments were carried out at a slower movement rate (500 μm/s) to completely desorb the reactants with a throughput of 3 seconds per sample. The low-mass quadruple resolution parameter was set to a nominal value of 18 to isolate a single m/z unit, and the collision energy utilized varied between 20 to 35 nominal value using argon as collision gas.

Droplet collection experiments for further product characterization

For estimation of derivatization efficiencies, an array-to-array DESI-based synthesis device³⁸ was used to recreate the reactive DESI conditions for derivatization of estriol (compound 15 in Table S1) using CMPI, selected as model reaction. Four replicates of a 5-μL aliquot of a 5-mM reaction mixture containing 1:1 molar ratio of estriol (in ethanol) and CMPI (in ACN) were subjected to collection. Using a 1.25-minute rastering, reaction products were collected on chromatography paper in a 96-well format. Each collected crude was redissolved in 150 μL ethanol and via liquid chromatography (LC) MS.

Two different LC- MS methods were developed for estriol and its CMPI-functionalized product on an Agilent 6460 triple quadrupole mass spectrometer coupled to an Agilent 1290 Infinity HPLC system. Reverse-phase separation was carried out using an Agilent InfinityLab Poroshell 120 EC-C18 column (2.1 × 50 mm, 1.9 μm). In the negative ion mode, precursor ion scanning was used to target estriol (precursor m/z 287.1; 500 ms dwell time). The LC method used water as mobile phase A and ACN as mobile phase B, with an isocratic elution gradient of 35% B (0.35 mL/min, 3 min). In the positive ion mode, precursor ion scanning was used to target CMPI-functionalized estriol (precursor m/z 380.1; 500 ms dwell time). The LC method used 0.1% formic acid in water as mobile phase A and 0.1% formic acid in ACN as mobile B, with an elution gradient from 5% to 60% B (0.30 mL/min, 5 min) followed by 65% B (8 min). The general MS source parameters include gas temperature at 350°C, gas flow at 8 L/min, nebulizer at 45 psi, sheath gas temperature at 250 °C, sheath gas flow at 11 L/min, and capillary voltage at ±3.5 kV. Using each method, we built calibration curves and quantified the collected amount of functionalized product and leftover substrate in the collected material. Note that the calibration curve of the product was created using the material collected, purified, and subjected to NMR analysis (vide infra) accounting for its purity as estimated via NMR.

For further structural confirmation experiments, the S_NAr reaction product of estriol was scaled up using electrosonic spray ionization (ESSI) operated at a positive voltage of 5 kV, a flow rate of 300 μL/min, and a gas pressure of ~170 psi. A reaction mixture containing 1 mL estriol (25 mM) in DCM, 1 mL 2-chloro-1-methylpyridinium iodide (25 mM) in ACN, 0.2 mL potassium carbonate (160 mM) in water, and 0.5 mL 2-propanol was sprayed into a 20-mL vial, forming the reaction product. The collected crude was redissolved in DCM and centrifuged to partially remove the methylpyridinium reagent. After drying under nitrogen, the crude was then redissolved in ACN and performed a passthrough clean-up using Waters OASIS HLB SPE (3 cc, 60 mg). The passthrough extract was dried under nitrogen and redissolved in deuterated methanol for ¹H NMR characterization using a Bruker Advance-III-800 equipped with a 5-mm QCI Z-gradient cryoprobe.

Qualitative application: mass spectrometry imaging (MSI)

Healthy male rat testis tissues were obtained from an animal carcass used in unrelated studies and provided by the Transgenic and Genome Editing Facility. Testis were stored at −80 °C until cryosectioning. Longitudinal sections (25 μm thick) were prepared using a Shandon Cryotome FE (Thermo Fisher Scientific, Waltham, MA) and mounted onto standard microscope glass slides (precleaned Plain Microscope Slides, Fisherbrand). For on-tissue derivatization, the functionalization reagent FMPB was prepared at 5 mM in acetonitrile and uniformly applied to tissue sections using a TM-Sprayer (HTX Technologies, Carrboro, NC) at a flow rate of 100 μL min⁻¹, a stage velocity of 1000 mm min⁻¹, a track spacing of 3 mm, and 30 passes in a crisscross pattern. The spray nozzle temperature was set to 30 °C, with a nitrogen nebulizing gas pressure of 10 psi. Tissue sections were analyzed by DESI mass spectrometry imaging (DESI-MSI) using a SYNAPT^™ XS mass spectrometer equipped with a DESI XS stage (Waters Corporation, Milford, MA). Data were acquired in positive ion mode over an m/z range of 200–1,200 in TOF resolution mode, with a spray voltage of 0.5 kV and the instrument resolution setting enabled. Methanol was used as the DESI spray solvent at 1.5 μL min⁻¹, with nitrogen as the nebulizing gas at 15 psi. The transfer capillary and source temperatures were both maintained at 150 °C. Instrument control, MSI acquisition, and data analysis were performed using Waters HDI software (version 1.5). Imaging was conducted at a spatial resolution of 120 μm with a scan time of 235 ms per pixel.

Quantitative application: metabolite determination in urine

The analyte chosen was 4-hydroxy-3-methoxyphenylglycol (MHPG), which was dissolved in DMSO at 5 mM and subsequently diluted using an artificial urine matrix (Sigmatrix urine diluent, Sigma Aldrich, St. Louis, MO) supplemented with 0.1% (w/v) bovine serum albumin to a final concentration of 5 μM. This stock was utilized to prepare the calibration curves for metabolite analysis. The functionalization reagent, BTBA was prepared as a 10 mM stock solution in methanol and diluted with water to 0.5 mM. One microliter of this solution was spiked into 100 μL sample and mixed. Finally, 1 μL of 0.25 mM 2,4,5-triphenylpyrylium (TPP) solution was added as internal standard. Samples were then directly subjected to HT-DESI-MS(/MS) analysis. The ion ratio between functionalized MHPG (m/z 328) and TPP (m/z 309) was used for quantitation. A real urine sample from a healthy male rat specimen was obtained from the Bindley Bioscience Center at Purdue University. A BTBA stock solution (10 mM) was mixed with the urine at 1:1 (v/v) ratio followed by the addition of IS. For the medium and high MHPG-spiked urine samples, 1 μL of 25 μM and 100 μM MHPG solution in water were added to 100 μL of rat urine, respectively. The mixtures were directly pinned onto a DESI slide for analysis. HT-DESI-MS conditions were identical to those described for the reaction screening.

The quantitative results obtained for the rat urine sample via HT-DESI-MS were validated using LC-MS/MS. An Agilent 6460 triple quadrupole mass spectrometer coupled to an Agilent 1290 Infinity HPLC system equipped with a Waters Atlantis T3 column (2.1 × 75 mm; 3 μm particle size) was used. Chromatographic separation was achieved using a binary mobile phase system consisting of 0.01% acetic acid in water (A) and 0.01% acetic acid in methanol (B). The gradient elution implemented was as follows: 5% B (1 min), 20% B (2 min), 90% B (3 min), 90% B (4 min), 5% B (5 min), followed by re-equilibration at 5% B (5–13 min). The flow rate was maintained at 0.3 mL/min, and the injection volume was 3 μL. MS analysis proceeded via multiple reaction monitoring (MRM) in the negative ion mode using kV as spray voltage and 80 V as fragmentor voltage. The quantifier MRM transition (50 ms dwell time; collision energy: 10 V; cell accelerator voltage: 4 V) was set on m/z 183 as precursor ion and m/z 150 as product ion, both with unit resolution. Quantitation was carried out using a calibration curve generated with MHPG standard solutions prepared in the same Sigmatrix artificial urine matrix used in the HT-DESI-MS experiments.

RESULTS AND DISCUSSION

In this study, we demonstrate the capabilities of HT-DESI-MS for the identification of successful derivatization strategies aimed at increasing the ionization efficiency of otherwise poorly ionized small molecules. Here, as a proof-of-concept, we focus on using this HT screening platform to evaluate a variety of derivatization reactions against a panel of biologically relevant molecules representing diverse functional groups. However, note that the same technology could also be utilized to optimize the conditions of such reactions, as we have reported.^39–42 Derivatization was enabled by accelerated reactions occurring in secondary DESI microdroplets, an approach known as reactive DESI, following the desorption of sequentially spotted analyte and functionalization reagents. Automated deposition of minimal sample aliquots (50 nL) onto PTFE-coated glass slides using a pin tool is part of the standard operating procedure of the HT-DESI-MS system. This workflow enables the generation of high-density two-dimensional arrays (up to 6,144 sample spots per standard microplate-sized array) with high precision and supports automated spot-to-spot analysis at throughputs exceeding 1 Hz. Notably, this sequential spotting strategy differs from previous reactive DESI implementations, in which functionalization reagents were introduced directly into the spray solvent,^43–46 and offers key advantages by eliminating solvent line switching for each new reagent and preventing cross-contamination from residual reagents in the solvent lines.

The functionalization screening campaign employed four reagents: CMPI, FMPB, BTBA, and EDC, across a library of 32 small biological compounds (Table S1). The first three reagents were selected for their potential to enhance the ionization efficiency of poorly detected analytes by targeting a range of nucleophilic moieties, whereas EDC, a well-established reagent for carboxylic acid activation in peptide coupling chemistry^47,48, was included to derivatize carboxylic acid groups and reverse their favored ionization polarity (Figure 2A). Carboxylic acids are generally detected as deprotonated ions in negative ion mode; however, reaction with EDC forms an O-acylisourea intermediate bearing a tertiary amine, which enables detection in positive ion mode. This polarity switch can be advantageous for distinguishing isomeric species and for confirming the presence of carboxylic acid functionalities. The choice of CMPI, FMPB, and BTBA, followed the widespread presence of hydroxy moieties in the set of poorly ionizable substrates of interest (see Table S1). In particular, CMPI, a pyridinium-based electrophile, was selected as it has been widely used in nucleophilic aromatic substitution (S_NAr) chemistry.^49,50 Previous reports have shown that CMPI readily reacts with phenol groups, while exhibiting limited reactivity toward carboxylic acids in the absence of base⁵¹, thus making it an interesting alternative for selective derivatizations. We also included FMPB, another highly electrophilic pyridinium derivative which our group has recently utilized for water quantification via ambient MS, a study in which we also observed it shows high reactivity towards alkyl alcohols.⁵² Finally, BTBA was included to specifically target diol-containing compounds, as boronic acid derivatives have been widely reported to selectively form cyclic esters with diols^43,46,53, including via accelerated microdroplet reactions.⁵⁴

Figure 2.

HT-DESI-MS/MS screening results for the derivatization of 32 biologically relevant small molecules (1-32) with four reagents (CMPI, FMPB, BTBA, EDC). (A) Single-stage MS screening results (color log scale) are included as signal-to-noise (SNR) ratios using each unreacted functionalization reagent sample as blank for noise calculation. Heatmap circle outlines indicate whether the hits identified via full MS screening were confirmed or discarded via diagnostic fragment identification in online MS/MS experiments. *: SNR values were normalized across each reaction for the sake of visualization as EDC products are not formally charged, which also led to different hit cutoff selections (log₁₀(SNR*) = 0.9 for all reactions except EDC, whereas 0.5 was chosen for EDC derivatization). (B-E) Schemes of the four reactions evaluated as well as corresponding representative single-stage and tandem mass spectra. Proposed diagnostic cleavages are shown in product structures without indicating whether or not they are associated with hydrogen atom transfer. Note the m/z 147 fragment in the BTBA reaction (D), which is assigned to a radical methyl loss from the m/z 162 fragment based on MSⁿ and exact mass measurements (Table S3). **: In B, the presence of a chlorine-to-iodide substitution is observed in unreacted CMPI, likely to a Finklestein-like reaction favored by activation of iodine (counterion of the chloropyridinium ion in commercial CMPI salt) at the microdroplet surface.^55,56

The HT-DESI-MS screening results are summarized in Figure 2A. Overall, the evaluated reactants were found to be successful for on-the-fly derivatization of the biomolecule panel via accelerated microdroplet reactions yielding detectable products in a millisecond timescale despite the modest conversions estimated offline via LC-MS analysis of collected DESI droplets (Figure S1). CMPI showed a broad substrate scope, reacting with diverse nucleophilic groups including amines, alkyl hydroxyls and phenols, although with different and substrate-dependent degrees of efficiency. In total, CMPI yielded 18 hits, 58% of the 31 nucleophile-containing compounds (note that compound 20 does not include suitable nucleophilic moieties to be targeted via this S_NAr derivatization strategy). The other three functionalization reagents showed significantly higher selectivity. For instance, FMPB was identified as being useful for the selective derivatization of alkyl hydroxy groups across a variety of structures, while showing no reactivity towards phenols. On the other hand, BTBA was successful at functionalizing 1,2- and 1,3-diols generating five- or six-member heterocyclic structures, respectively, including catechols but not trans 1,2-diols (e.g. estriol, 15). Finally, EDC was found to react with all carboxylic acids screened except those on an amino acid, generating products detected as positively charged ions. Note that sequential reactions were not detected to any significant extent (> 2% relative intensity compared to the singly functionalized product; Figure S2) in the cases where multiple potential reacting sites exist in the substrate (e.g. multiple alkyl hydroxy sites).

It is important to highlight that online HT-DESI-MS/MS analysis (i.e. automatically revisiting an already screened spot to obtain MS/MS information on the residual product) was used to confirm the identity of the derivatization products from all the reaction hits (Figure 2B–E and Figure S3), an important requirement for simple addition reactions (e.g. FMPC, EDC) where covalent reaction products and ion-molecule complexes between the reagents are isobaric (Table S2). It is worth noting that each specific reaction was found to produce a distinctive fragmentation pattern characterized by diagnostic product ions that correspond to the charged moiety introduced (see proposed fragmentation in Figure 2B–E). These diagnostic ions can then become a useful tool for functional group identification within complex samples, for instance via precursor ion scans. Note that products of interest identified via HT-DESI-MS screening can be readily scaled up using microdroplet methods and further characterized using orthogonal analytical strategies, such as NMR (Figure S4).

To demonstrate the applicability of this approach, we carried out relevant qualitative and quantitative experiments using analyte-reagent pairs identified in the HT derivatization screening. Application to qualitative analysis was evaluated via MS imaging (MSI) of the poorly ionized compound cholesterol (29) in rat testes tissue sections. Due to the role of cholesterol and other metabolically related compounds in spermatogenesis, its spatial distribution within the testis is interesting as it can serve as an indicator of overall reproductive health, as well as provide insights into the testicular functional organization via correlation of its accumulation with specific cell types and morphological features.

Since HT screening had identified FMPB as an effective derivatization reagent for cholesterol, we carried out online derivatization-analysis via DESI-MSI after uniformly spraying FMPB onto a mouse testicle cryosection. Following this approach, the distribution of functionalized cholesterol as well as other related hydroxysteroid metabolites was readily detected, with confirmation of their identity via exact mass measurements and MS/MS experiments (Figure 3B and Figure S5). As shown in Figure 3A, cholesterol (m/z 508.4155, Δ = 1.18 ppm) was found to accumulate largely in the testis while the ion species assigned to testosterone (m/z 410.2698, Δ = 1.95 ppm) was concentrated in the caput epididymis. The spatial distribution observed with this reactive DESI-MSI approach aligns with the known localization of cholesterol-producing cells in the testis. Cholesterol is primarily synthesized and stored in Leydig cells, which are located within the interstitial spaces between seminiferous tubules and serve as the main source of steroid precursors for testosterone biosynthesis. By contrast, Sertoli cells, situated inside the seminiferous tubules, contribute to local cholesterol metabolism by facilitating lipid exchange to developing germ cells but are not major cholesterol producers. Accordingly, the strong cholesterol signal detected in the interstitial regions of the testicular section corresponds to Leydig cell–rich areas, while the higher testosterone abundance observed in the caput epididymis reflects downstream sites of steroid hormone accumulation and secretion, in agreement with its reported localization across seminiferous tubules rather than the interstitium.^57,58 This spatial concordance supports the biological relevance of the derivatization-based imaging approach and highlights its potential for mapping lipid metabolism within functionally distinct cell populations.^59,60 Note that an ion assigned to vitamin D₃ (cholecalciferol, m/z 506.4002, Δ = 1.78 ppm) was also detected in the epididymal region of the rat testicular section, suggesting its localized enrichment or uptake. This finding aligns with prior reports of vitamin D₃ metabolic activity within the male reproductive tract and supports its role in testicular development and spermatogenesis.^61–63 It is also worth noting that this approach for the detection of steroids in biological samples can be readily applied to other biological samples, for instance discrete tissue extracts (Figure S6), not necessarily for imaging purposes.

Figure 3.

Results of the application of two of the identified successful functionalization strategies for DESI-based MSI of tissue sections and metabolite quantitation in urine. (A) FMPB was utilized to functionalize cholesterol (29) and related steroids in a rat testicle section after spraying the derivatization reagent onto the tissue. The overlaid imaging results for cholesterol (m/z 508), testosterone (m/z 410), and cholecalciferol (m/z 506) are shown together with the two main anatomical regions observed on the tissue. The image follows a linear single-color scale for each ion after base-peak normalization. (B) The observed ions were confirmed via MS/MS experiments upon reaction of standards. Note the presence of the diagnostic product ion m/z 140. (C) BTBA was utilized for the functionalization and HT quantitation of MHPG (8) directly from urine. A calibration curve was created using commercial artificial urine as matrix (TPP is used as internal standard for ion ratio determination) and applied for the quantitation of the metabolite in a rat urine sample (both with and without MHPG spiking). The data point at 0 μM MHPG concentration in the rat urine comes from the actual sample comes from a sample not treated with BTBA. Also note that rat urine samples were processed by 1:1 (v/v) addition of a BTBA solution, thus the concentrations obtained with the calibration curve are halved. (D) The identity of MHPG in the rat urine sample was confirmed via MS/MS analysis of the reaction product in comparison to a standard.

Application of the derivatization strategies identified in the screening to quantitative analysis was carried out by directly monitoring a disease biomarker in biofluids. For this demonstration we focused on the HT determination of 3-methoxy-4-hydroxyphenylglycol (MHPG, 8) in rat urine, as this norepinephrine metabolite has been proposed as a potential biomarker for diagnosis of Alzheimer’s disease based on rat models.^64–66 As derivatization of MHPG with BTBA was shown to be effective in HT screening (Figure 2), we coupled this approach with rapid HT-DESI-MS(/MS) analysis directly from the biofluid matrix for quantitation (Figure 3C) and online confirmation (Figure 3D) of this metabolite in spiked commercial artificial urine down to low nM levels (LOD = 31 nM) with high quantitative performance (R² = 0.996, RSDs < 14%; see Table S4), no sample preparation, and minimal sample consumption (150 nL). This methodology was applied to a urine sample from a healthy rat (Figure 3C–D, Table S5), which yielded a MHPG concentration of 0.13 ± 0.02 μM, in close agreement with reported literature values⁶⁴ as well as LC-MS/MS-based analysis (Figure S7 and Tables S5–S7). Moreover, upon spiking this urine sample with medium (+0.25 μM) and high (+1 μM) MHPG levels, accuracies of 88% and 97% were obtained, respectively, further demonstrating the performance of this online derivatization strategy coupled with HT-DESI-MS-based quantitation (1 second per sample).

Note that in both applications demonstrated here, DESI-based MSI and HT quantitation, analyses took place immediately after reagent addition without incubation. This is because the derivatization reactions largely take place in the DESI microdroplets rather than in bulk solution or the tissue section, as evidenced by the direct effects of the spray distance (Figure S8) and the DESI solvent (Figure S9) on the product yields. Consequently, integration of this derivatization strategy to other MS-based approaches (i.e. besides DESI) would have to proceed via offline microdroplet collection (as we showed here using an array-to-array synthesis device; see Figure S1) or online integration of microdroplet reactions to, for instance, liquid chromatography (LC)-MS systems in a post-column fashion.^54,67,68

CONCLUSIONS

In this work, we demonstrate the applicability of HT-DESI-MS for the rapid exploration of functionalization strategies aimed at improving the ionization efficiency of multiple classes of biologically relevant compounds and/or by altering their typical ionic species (e.g., converting carboxylic acid anions into positively charged tertiary amines) to enable diagnostic identification of functional groups. With a throughput of approximately one sample per second, HT-DESI-MS provides a fast platform to identify efficient transformations and evaluate their specificity and substrate scope by leveraging online, accelerated microdroplet reactions. Using this approach, we identified a general reaction of nucleophilic functional groups with a relatively broad substrate scope, as well as three derivatization strategies that selectively target 1,3- and cis-1,2-diols, alkyl hydroxyl groups, and carboxylic acids. We anticipate that the reactions discovered using HT-DESI-MS will be readily transferable to other analytical workflows (e.g., LC-MS) or directly coupled with DESI, as demonstrated here, for imaging of derivatized compounds in tissue sections at ~50 μm or lower spatial resolution and for direct high-throughput quantitation of derivatized metabolites in biofluids.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information. PDF file containing identities and structures of the compounds screened, numerical summary of screening results, further representative spectra, and additional information on exact masses and quantitation.