Skip to main content
NCBI home page
ACS Medicinal Chemistry Letters logoLink to ACS Medicinal Chemistry Letters
. 2020 May 1;11(6):1101–1110. doi: 10.1021/acsmedchemlett.0c00066

Ultra-High-Throughput Acoustic Droplet Ejection-Open Port Interface-Mass Spectrometry for Parallel Medicinal Chemistry

Kenneth J DiRico †, Wenyi Hua †, Chang Liu ‡, Joseph W Tucker †, Anokha S Ratnayake †, Mark E Flanagan †, Matthew D Troutman †, Mark C Noe †, Hui Zhang †,*
PMCID: PMC7294554  PMID: 32550988

Abstract

graphic file with name ml0c00066_0007.jpg

High-throughput experimentation (HTE) has emerged as an important tool in drug discovery, providing a platform for preparing large compound libraries and enabling swift reaction screening over wide-ranging conditions. Recent advances in automated high-density, material-sparing HTE have necessitated the development of rapid analytics with sensitivity and resolution sufficient to identify products and/or assess reaction performance in a timely and data-rich manner. Combination of an ultrathroughput (UT) reader platform with Acoustic Droplet Ejection-Open Port Interface-Mass Spectrometry (ADE-OPI-MS) provides the requisite speed and sensitivity. Herein, we report the application of ADE-OPI-MS to HTE in the areas of parallel medicinal chemistry and reaction screening.

Keywords: high-throughput experimentation, Acoustic Droplet Ejection-Open Port Interface-Mass Spectrometry, parallel medicinal chemistry

The concept of parallel synthesis is rooted in combinatorial chemistry advances that were made in the latter part of the 20th century.1 Many of these campaigns focused on preparing thousands of compounds to enrich corporate compound collections and provide leads for targets emerging from the human genome project. Indeed, combinatorial methods still form the backbone of contemporary file enrichment, as well as hit identification approaches rooted in DNA-encoded library technology.2 Parallel synthesis methods are widely used in lead optimization campaigns. The success of these approaches is highly dependent on the availability of diverse chemistry that can be subjected to automation and can be negatively impacted by some of the restrictions inherent to technologies such as DNA-encoded libraries. The utility of parallel synthesis is largely dependent on the yield of the desired product, selectivity for producing it relative to problematic side products, the ease of eliminating excess reagents, and the practicality of conducting reactions on small scale. Further, expanding the toolbox of library technologies requires efficiently examining diverse reaction conditions to test such parameters, with the goal of expanding the range of feasible substrates and virtual libraries.

Medicinal and process research chemists have long experimented with automated methods for rapidly screening reaction conditions. Many of these methods can be adapted to producing libraries of compounds, thereby serving a dual purpose in the synthesis laboratory. Recently, impressive advances in reaction screening technology have been made by the advancement of high-throughput experimentation (HTE) technologies.3,4 Importantly, these methods can employ microgram or even nanogram amounts per reaction, sparing precious templates and reducing the cost of the overall reaction optimization process. Many of these methods involve plate-based or segmented flow-based reactors and offer, for the first time, the ability to survey hundreds of reaction variables in a single experiment. The output of these experiments is a plot of reaction yields across a landscape of reaction conditions. Collecting the analytical data, however, is often a significant bottleneck.

Mass spectrometry (MS) coupled to HPLC (LC-MS) is a common method for analysis of such reactions. This approach requires a relatively small amount of material (typically ∼2 to 15 μL), allows simultaneous detection of starting material, product, and side products, and allows data to be collected in a matter of minutes for each reaction under study. Approaches to analyze reactions in situ are being pursued, but material requirements are typically much higher than what is used in microscale reactions for parallel medicinal chemistry (PMC).5−8 Although analysis times are relatively short for a single well, when aggregated over hundreds of compounds, several hours or even days can be required to analyze a high-density experiment.4 Therefore, techniques to accelerate analysis significantly increase the feasibility of HTE.

Advances in MS are often directed at improving the speed, sensitivity, and resolution of the instrument. Matrix-assisted laser desorption/ionization (MALDI) MS has been used as a high-throughput analytical platform for synthetic chemistry.9 Solid surface sampling methods such as desorption electrospray ionization (DESI) have been applied to PMC and reaction screening, but they require generating a sample array on a solid surface prior to analysis.10−13 We were therefore intrigued by the potential for direct sampling of chemical reaction mixtures via the recently described acoustic droplet ejection (ADE) open port interface (OPI) MS approach (ADE-OPI-MS).14 This technology relies on acoustic dispensing of droplets directly from the wells of the plate under analysis. The acoustically dispensed droplets, which are at nanoliter scale, with the precise control and independent of the sample solvent,14 are acoustically ejected from the reaction mixture and introduced to a vortex at the opening of the OPI and delivered directly to the electrospray ionization (ESI) source of the MS for detection. The extremely small samples required, coupled with the method’s resilience in handling unpurified samples, make this technology ideally suited for direct sampling of organic synthesis reaction mixtures. The ADE-OPI-MS method also offers significant speed advantages: with an average analysis time of 1–2 s per sample, such that a typical 384-well plate can be analyzed in under 15 min. Finally, the ADE-OPI is compatible with both nominal and high-resolution mass spectrometers, allowing rapid quantification with the former, and extensive analyte identification with the latter. We therefore decided to apply this technology to the analysis of synthetic reactions from HTE campaigns.

ADE-OPI-MS Instrumentation

The ADE-OPI-MS instrument used in this study is schematically shown in Figure 1. The ADE technology delivers nanoliter volumes of samples with high accuracy and precision via fine-tuned acoustic energy, and it has been widely used in drug discovery for compound and sample management applications that support various medicinal chemistry, pharmacology, and ADME screens. DMSO is a very common solvent that is compatible with ADE, and in this study all HTE samples were diluted/prepared in DMSO. Besides that, a few other common solvents (aqueous, methanol, acetonitrile, etc.) are also compatible with our current prototype ADE instrument.14 Other common solvents used for HTE can also be used with proper acoustic calibration as long as the final samples are homogeneous. The OPI is characterized by an elegant design in which the probe is a coaxial tube with one end open to the air (thus the name “open-port”) and the other end connected to the ESI source of the mass spectrometer. A high carrier flow of organic solvent (such as methanol) is continuously introduced through the side of the probe, with the flow rate optimized to balance the nebulizer gas flow, such that a stable vortex is created in the droplet capture region of the OPI. Once the nanoliter sample droplets enter the OPI carrier flow, they are significantly diluted (up to 1000-fold). This online dilution effectively minimizes ion suppression and eliminates the need for LC separation prior to MS detection. For example, samples containing phosphate ions or serum proteins have typically proved problematic in traditional LC approaches, due to ion suppression effects and/or fouling of the mass spectrometer source. Due to the high-fold dilution inherent to the ADE-OPI-MS approach, however, such “dirty” buffers and matrices are tolerated.14 Because of the continuous flow of the OPI and the very short time needed for MS analysis, the nanoliter sample droplets can be ejected continuously via ADE. Overall, the ADE-OPI-MS instrumentation allows for a very high sampling rate and enables rapid throughput that includes sample analysis. Importantly, the ADE-OPI-MS approach can be used very effectively with ESI, a robust and widely applicable method of ionization commonly employed for analysis of analytes with various molecular weights introduced via liquid. Additionally, we have shown ADE-OPI-MS to be compatible with both nominal and high-resolution MS analyzers thus further expanding the approaches utility.

Figure 1.

Figure 1

Open in a new tab

Schematic of an ADE-OPI-MS system. A pulse of acoustic energy ejects sample droplets (1–10 nL) upward into the inverted OPI sampling interface. A fluid pump delivers carrier solvent (180–300 μL/min) to a sample capture region equipped with a flow-stabilized vortex interface; sample is captured and diluted into a vortex of flowing carrier solvent. A high voltage (HV) supply and nebulizing gas (nitrogen) at the spray capillary drive conventional ESI.

ADE-OPI-MS in HTE

HTE enables discovery scientists to quickly scan multiple reaction variables.15 Modern techniques are highly material-sparing while enabling large numbers of reactions to be examined using small quantities of substrate.16 However, the enhanced production of reaction data over increasingly short time frames applies pressure to the downstream analytical experiments that assess reaction success; these results are key to the design of further rounds of experiments, and thus, analysis can become a bottleneck to the HTE workflow. Beyond basic reaction monitoring to assess degree of conversion, each reaction can yield further information relating to the appearance and disappearance of expected and unexpected species. High-resolution MS (HRMS) is particularly useful for synthetic chemistry because HRMS spectra provide high fidelity in identifying the chemical entities of interest, via both high-accuracy mass and isotope patterns.

Because HTE screening methods17,18 have become increasingly common in medicinal and process sciences, we were eager to assess the utility of the UT ADE-OPI-MS in this arena: given its material-sparing technology and rapid analytical capabilities, ADE-OPI-MS was envisioned to be highly complementary to modern HTE. We therefore devised three sets of experiments that mimic standard HTE/PMC screening processes. The first experiment consisted of a reaction optimization screen for a single transformation using a simple small-molecule model substrate with variable catalyst and base conditions. The second experiment examined the parallel synthesis of a library of drug-like small-molecules, involving reaction of common template structures with an array of reactant partner monomers. The third experiment examined a similar library synthesis but employed a larger template substrate (MW ∼5000 Da) with an array of small-molecule reactant partners.19

Results and Discussion

Experiment 1: Reaction Optimization with a Simple Model Substrate under Nanoscale Conditions20,21

For our first test, we chose to replicate an established reaction screen demonstrated by Cernak et al.,12 in which a nanoscale (100 nmol) palladium-catalyzed C–N coupling of 3-bromopyridine with 4-phenylpiperidine was examined, using a series of base and catalyst sets as variables in a 6 × 16 format, respectively. This report provided us with a test case and robust data for comparing standard ultraperformance LC single-quad low-resolution mass spectroscopy (UPLC-MS) with ADE-OPI-MS analysis results. The 96-well array described in the Cernak manuscript was run in a 384-well plate on 2 μL scale (200 nmol limiting reagent), using dimethyl sulfoxide (DMSO) as solvent. A more detailed description of the reaction procedure can be found in the Supporting Information. The two data sets were then visualized with Spotfire; the relative assessment of reaction performance is shown in Figures 2B and 2C, as total UV254 peak area results from the UPLC-MS method and peak areas for the extracted target product generated from ADE-OPI-MS analysis. These reaction plate profiles for formation of product and relative conversion rates across the reaction screen compare well between the two analytical modalities, demonstrating the potential for ADE-OPI-MS to capture both qualitative and semiquantitative outcomes across a series of reaction conditions. In addition, the ADE-OPI-MS data has the potential to capture low conversion rates not detected in our standard UPLC-MS method, possibly due to injection errors for the UPLC-MS or a higher detection sensitivity attained by MS compared to the UV detector.

Figure 2.

Figure 2

Open in a new tab

(A) Chemical reaction pathway for a simple coupling reaction (Experiment 1) with a targeted product C16H18N2.3 (B) Relative product conversion over 96 different reaction conditions using LC/UV (read at 254 nm). Peak areas of the target product were plotted against the sample numbers. Each injection consumed 1 μL of the reaction, and each sample took 3 min for analysis. Overall, it required 3 h to analyze all 96 reactions using conventional LC/UV detection. (C) The same set of samples was analyzed using ADE-OPI-MS, and peak areas of the target product (extracted at m/z 239.1) were plotted against the sample numbers. Sample consumption for each reaction was 5 nL. The 96 samples were injected continuously, and total analysis time was 5 min.

Experiment 2: Nanoscale PMC with a Small-Molecule Substrate

Another important HTE workflow involves PMC syntheses, which are typically driven by a project team’s desire to quickly explore design space around a novel template by reacting a core compound with a large set of monomers to introduce chemical diversity.1,21−24 For this experiment, we designed a prototypical PMC screen involving N-ethyl-N′-[3-(dimethylamino) propyl] carbodiimide hydrochloride (EDC-HCl) and 2-hydroxypyridine N-oxide (HOPO)-mediated amidation of two secondary amine templates, each with 96 carboxylic acids, for a total of 192 reactions.25 To continue the material-sparing approach, this experiment was conducted on nanoscale (200 nmol, 2 μL volume) in a 384-well format. A more detailed description of the reaction procedure can be found in the Supporting Information. Stock solutions of the 96 acids were prepared: wells A1–D24 were used for reactions with template 1, and wells E1–H24 for reactions with template 2. Plate mapping of the acid reactants (monomers) and the molecular weight and molecular formula for the expected products can be found in Supporting Information Table 3. Raw data for the LC area ratios for the 192-well plate is provided in Supporting Information Table 4.

As in Experiment 1, visualization of the 192 wells revealed a good agreement in results for the UPLC-MS and ADE-OPI-MS analyses (Figure 3): as shown in Figures 3D and 3E, the relative extracted ion current (EIC) UV254 peak areas generated by UPLC-MS compared reasonably well with the product peak areas derived from ADE-OPI-MS instrumentation of the reaction wells, even though absolute quantitation with MS peaks is not feasible due to different ionization efficiencies for different product ions. Reactions with average to maximum detected peak areas of targeted products in Figure 3D (pale red to dark red, 34 reactions) have been captured in Figure 3E with similar relative peak areas (28/34) while the nonmatched 6 reactions were near average. Additionally, both data sets indicate that template 2 was a poor reaction substrate for the conditions used in the screen and that column 20 had a clear dosing issue, as poor reaction performance is observed across both templates and multiple monomers for column 20. For the purpose of this manuscript we did not pursue the reason, but the visualized data provides the information needed to diagnose the issue so that it can be further investigated. Again, we found the total ADE-OPI-MS analysis time (∼7.5 min) to be significantly shorter than that of the 1.6 min UPLC-MS analyses (∼5 h).

Figure 3.

Figure 3

Open in a new tab

Experiment 2: Readout of 192 PMC library compounds with ADE-OPI-MS versus UPLC-MS. (A) Reaction scheme for PMC. (B) Total ion chronogram (TIC) of the 192 reactions from ADE-OPI-MS. Samples were injected continuously, and all 192 samples were analyzed within ∼7.5 min. (C) Zoom-in view of the first 25 wells showing the individual peaks representing the TIC of each reaction mixture. (D and E) Peak areas of the targeted product for each reaction are shown in heatmap mode for comparison. Data from the LC/MS method is depicted in part D, and ADE-OPI-MS results are given in part E.

ADE-OPI-HRMS

As demonstrated in Figure 4, the HRMS approach greatly simplified data acquisition by employing one generic mass-scanning method. Postacquisition data processing can easily extract data for different analytes from the same data file. For example, from the same PMC plate, unreacted template can be assessed for all the wells (Figures 4A and 4B). Naturally, the wells that had the largest amount of template remaining represent low yields of the targeted products. The expected products in several reactions (wells E8, F17, H6, and H18) share the same formula (C20H19N3O3). When the calculated m/z (350.149) for this formula was extracted, four peaks emerged (Figure 4C), corresponding to the exact well locations that were expected. As noted above, HRMS is particularly useful for synthetic chemistry, due to its high fidelity in identifying and confirming the chemical entities of interest, through both high-accuracy mass and isotope pattern matching. Different molecules with the same nominal mass can be easily distinguished using HRMS data. For example, there were several products with very similar molecular weights appearing in the matrix: C20H17N3O4 in wells F04 and H05, C22H22FN3O in well D14, and C21H21N3O3 in well F19, with expected m/z of 364.129, 364.182, and 364.165, respectively. These were tracked via their exact masses as shown in Figure 5: only two peaks showed up when an m/z of 364.1290 was extracted using a narrow window (0.01 Da), corresponding to products in wells F04 and H05. Wells D14 and F19 showed up when a wider window was extracted (0.1 Da) around m/z 364.13. In addition, a few other wells (E16 and F12) exhibited a signal using the 0.1 Da extraction window, corresponding to the M+1 isotopes of the products in those wells. Note that if a nominal mass instrument were used, all these masses would appear as m/z 364 and thus be indistinguishable.

Figure 4.

Figure 4

Open in a new tab

Selected data from Experiment 2: remaining template and product analysis from ADE-OPI-MS. (A) Extracted ion current (EIC) with targeted formula C15H28N6, or m/z 293.2448, for template 1. (B) EIC with targeted formula C19H28N6O, or m/z 357.2397, for template 2. (C) EIC with formula C20H19N3O3, or m/z 350.1499. Four peaks observed as expected with one of each corresponding to one of the four wells (E8, F17, H6, and H18) that with targeted product match with the formula.

Figure 5.

Figure 5

Open in a new tab

Selected data from Experiment 2; differentiation of products with similar molecular weights. Postacquisition processing of the ADE-OPI-MS data can allow similar products to be distinguished. Two peaks of the targeted product with formula C20H17N3O4 showed up in the expected wells (F04 and H05) when the accurate mass was extracted at a 0.01 Da window (A). Many additional peaks showed up when a wider extraction window (0.1 Da) was used (B). All the pertinent peaks were further evaluated by extracting the detailed mass spectral windows shown (C–H). Clearly, compounds in wells D14 (C), E16 (D), F12 (F), and F19 (G) had a different accurate mass than the C20H17N3O4 molecules; this was easily determined using the HRMS readout from ADE-OPI-HRMS.

Experiment 3: Nanoscale PMC with a Complex, Large-Molecule Substrate

With the gain in popularity of DNA-encoded chemical libraries (DECLs) as a lead identification platform,2 the need for rapid profiling of on-DNA reactions has also spurred widespread interest within the pharmaceutical industry.22,26,27 High yields and good purity of early on-DNA synthetic steps affect the performance of DECL technology by the presence of incomplete reaction products and residual damaged DNA leading to complicated decoding procedures. Recent reports of high throughput reaction screening have demonstrated increased reactant scope and expanded synthetic transformations of on-DNA compatible reactions leading to broader access of new building blocks for library construction and design.28,29 With that in mind, our third experiment focused on analysis of on-DNA reactions.

Comparison of the ADE-OPI-MS and LCMS approaches was performed in the following manner. For the on-DNA reaction, we chose to examine an amide formation between a commercially available DNA fragment possessing an amine reaction point and a set of 96 acids, as illustrated in Supporting Information Figure 6.25,30,31 Different from Experiments 1 and 2, the final samples were diluted with water and then analyzed by ADE-OPI-MS and UPLC-MS on a Waters time-of-flight (TOF) instrument. A more detailed description of the reaction procedure can be found in the Supporting Information.

One key advantage of ADE-OPI-MS over traditional LCMS for this application is the speed at which reactions can be profiled to give a quick read out of their components (e.g., quantitation and molecular weight characterization of products, byproducts, and impurities). In addition, the ability to obtain data using low sample volumes (≤20 nL/sample), and to analyze crude, complex reactions with only minimal sample preparation, makes ADE-OPI-MS an attractive analytical tool for monitoring nanomole-scale on-DNA reactions in an automated fashion. In contrast, LC-UV260-MS-based oligonucleotide analysis relies on an MS-compatible ion-pairing mobile phase (triethylamine/1,1,1,3,3,3-hexafluoro-2-propanol) to achieve optimal chromatographic resolution and typically requires longer run times (5–10 min/sample). On the upside, in addition to MS-based quantitation and characterization of oligonucleotide products, well-resolved LC peaks allow UV-AUC-based quantification (% yield) of separated oligonucleotide products in the reaction mixture. However, this can require laborious and time-consuming optimization of the LC-gradient elution conditions, as the on-column retention times of synthetically modified oligonucleotides can vary significantly based on the polarity of the attached ligands.

Unlike small molecules, where only singly charged ions are detected in ADE-OPI-MS, larger molecules such as the DNA headpiece can be detected as ions with multiple charges, as shown in Supporting Information Figure 2. Among all the peaks with different charge states, the −4 ions were of the highest abundance, followed by −5 and −3. In contrast, using the LC/MS method, −3 peaks are the most abundant; this change is presumably due to differences in the mobile phases and ESI source settings from different vendors. After deconvoluting all the differently charged ions, m/z 4934.88 was observed as the monoisotopic ion for the DNA headpiece, which matched well with the calculated value (m/z 4934.9) based on its molecular formula C154H215N52O101P17 (Supporting Information Figure 2H). Single-digit ppm mass errors were obtained for both the deconvoluted peak and the observed mass peaks with multiple charges, except for the −8 ions (due to the low ion intensities). The mass accuracy and detection of DNA peaks over multiple charge envelopes assured the high-fidelity detection of similar molecules in this high mass range (Supporting Information Table 1).

Similarly, compounds with the DNA headpiece attached were found to exhibit multiple charge envelopes (Supporting Information Figure 3). In terms of signal level, the same trend was observed as with the unmodified DNA headpiece: ions with −4 charges were found to have the highest signal, followed by −5, −3, and −6, etc. In addition, more than 10 Na+ adducts could also be observed, particularly for the −4, −3, and −2 ions (Supporting Information Figures 3A–C). These Na+ adducts, combined with the numerous natural isotopes of such large molecules, made the interpretation of spectra difficult. Interestingly, the ions with a −5 charge exhibited substantially fewer Na+ adducts (Supporting Information Figure 3D, where only one Na+ was detected) and very clean spectra were observed. Ions with a −6 charge seemed to be less prone to Na+ adducts as well, but the signal level was considerably lower than that for ions with a −4 or −5 charge. Ions with a –5 charge are good candidates for qualitative peak identification, offering a balance of sensitivity and spectral simplicity.

On the basis of these investigations, 96 on-DNA chemical transformations (acylations) were conducted, as described above, and samples were analyzed using both LC-UV-MS and ADE-OPI-MS platforms.32 With ADE-OPI-MS, all 96 samples were analyzed within 4 min; the total ion chronogram (TIC) of the 96 samples is shown in Figure 6A, in which each peak represents one reaction. As a typical example, the expanded mass spectrum for the reaction mixture in well #02 (corresponding to the peak at 5.35 min in Figure 6A) is given in Figure 6B; highlighted regions indicate the −5 charged species for both the DNA headpiece and the targeted product, at m/z 984–987 and m/z 1024–1026, respectively (Figures 6C and 6D). An unknown byproduct at m/z 1063 was also observed. LC/UV/MS was used to characterize the same sample; three major peaks were observed as shown in Supporting Information Figure 4, where the DNA headpiece, the expected product, and the same unknown byproduct (at retention times of 3.2, 5.0, and 5.7 min, respectively) all matched well with the findings from ADE-OPI-MS analysis (based on the mass match after deconvolution). Such comparisons between the analytical data derived from traditional LC-UV260-MS and from ADE-OPI-MS demonstrate that the ADE-OPI-MS platform is suitable not only for studying reactions between small-molecule partners but also for chemistry involving much larger, complex species such as synthetically modified oligonucleotides.

Figure 6.

Figure 6

Open in a new tab

On-DNA reaction monitoring with ADE-OPI-MS. 96 on-DNA chemical transformations (acylation) were analyzed within 4 min. (A) TIC of all 96 samples; each peak represents one reaction. (B) Expanded mass spectrum of well A02 (at 5.35 min in Figure 6A); headpiece DNA, the expected product, and an unknown byproduct were detected as [M – 5H]−5 ions. (C and D), Zoom-in view of the DNA headpiece and expected product from part B, overlaid with the calculated m/z and relative isotope intensities (solid dots). LC/UV/MS was used to characterize the same sample; the LC/UV trace is shown in Supporting Information Figure 4A, where the DNA headpiece, expected product, and unknown products were separated with retention times of 3.2, 5.0, and 5.7 min, respectively.

Conclusions

Comparison of ADE-OPI-MS with Other Mainstream High-Throughput MS Platforms

Currently, there are three common types of HT-MS approaches supporting PMC analysis, employing LC/MS, MALDI-MS, and DESI, respectively.3,9,13 Herein, we describe a fourth option, the ADE-OPI-MS platform, and we further note several major advantages of ADE-OPI-MS versus the other three approaches (Table 1). First, ADE-OPI-MS is very sample-sparing. Unlike the microliter injection volumes required for LC/MS, and the loading volumes of dozens of nanoliter to a few microliter for DESI13 or MALDI,9 only a few nanoliter of sample are consumed for ADE-OPI-MS analysis. This efficiency enables the bulk of the synthesized samples to be preserved for follow-up purification and/or further evaluation. Second, the sampling speed is more than an order of magnitude faster, which provides excellent complementarity to high-density microtiter plate (96-, 384-, or even 1536-well)-based PMC approaches.20,24,33 As demonstrated by the three experiments above, analysis typically requires only a few minutes per plate with the ADE-OPI-MS platform. Third, when equipped with an ESI source, the quantitative characteristics of ADE-OPI-MS are similar to LC/MS and more quantitative than MALDI. Furthermore, it has been reported that considerable portions (up to 50%) of chemical entities cannot be ionized effectively with MALDI; much higher coverage is attained using an ESI source (>80%).34 In addition, the raw MALDI or DESI signal must be calibrated with appropriate internal standards (structurally similar compounds), or the quantitation can be entirely inaccurate. For large and very diverse library syntheses, it is impractical to have multiple internal standards prepared for such quantitation. Use of an internal standard is unnecessary for ADE-OPI-MS, and the majority of chemical entities can be ionized/analyzed by ADE-OPI-MS directly.

Table 1. Comparison of High-Throughput Mass Spectrometry Platforms.

  LC-MS MALDI-MS DESI-MS ADE-OPI-MS
Sample Requirement 1–10 μL 100 nL–2 μL 50 nL 1–20 nL
Sample Handling No Need plating Need plating No
Analysis Time (384-well plate, min) 15–30 h 2–10 min 5–20 min 5–20 min
Compound Coverage High Low High High
Internal Standard Calibration No Yes Yes No
Open in a new tab

One major limitation of ADE-OPI-MS is the absolute sensitivity compared with LC/MS due to the significantly decreased sample loading; however, for chemistry samples this does not seem to be an issue as reactions need to be further diluted before analysis. In addition, there is no separation before MS just as DESI or MALDI approaches, so potentially there would be suppression or interferences with very dirty samples. Even though we did not see the necessity for our studies, additional sample cleanup or orthogonal separation techniques such as differential mobility spectrometry could be deployed for those scenarios.35

Sampling Volume

In these studies, reactions as small as 1 μL are analyzed after one simple step of dilution. Microtiter plates of different densities (such as 96-well and 1536-well formats) are routinely used in compound management laboratories with ADE technology. These plates can be directly used for ADE-OPI-MS detection (reference). In fact, provided the nanoliter samples can be ejected out, any device that is compatible with ADE will be compatible with ADE-OPI-MS. For example, it is foreseeable that special devices can be designed with even smaller reservoirs/higher density patterns to further decrease the volume of reaction, increase the number of parallel reactions or diverse reaction conditions, and lower the overall cost for medicinal chemistry efforts.

High Speed for Kinetic Reading

A relatively low sampling speed (2–3 s/sample) was used for these studies, but much faster (2 Hz) sampling capability has been demonstrated with the ADE-OPI-MS platform.14 All the studies described in this report were carried out offline after the reaction was quenched or stopped. Conceptually and ideally, reactions can be monitored online to study the real-time kinetics of reactions, and systems have been developed for that purpose.36 ADE-OPI-MS will be very attractive for such applications due to the minimal sample cleanup required, very rapid analysis, and nanoliter sample consumption per analysis, such that the volume of the bulk solution remains essentially unchanged. In fact, enzymatic kinetic readouts have been demonstrated with the ADE-OPI-MS platform in an earlier publication.14 Further improvements can be envisioned for the ADE-OPI-MS technology in its application to the monitoring of diverse chemical reactions. For instance, enhancements in sampling speed may be needed to monitor very rapid kinetics of certain chemical transformations; this acceleration would be achievable by minimizing the sample transit time and dispersion within the transfer line. Many chemical reactions are oxygen-sensitive, and these experiments would require proper enclosure. In addition, many reactions are carried out under conditions of heating or cooling. Such temperature variations must be addressed because the acoustic transducer energy needs to be fine-tuned when the physical conditions of the samples change. All these new development ideas are being investigated for future development of ADE-OPI-MS technologies.

In conclusion, the ADE-OPI-MS platform provides a very fast and thorough MS readout, consumes minimum sample, can quantitatively detect a broad range of chemical entities, and can be directly coupled to high-density microtiter plates.37 When equipped with HRMS (such as the TOF instrument used in this study), the instrument enables unambiguous identification of both small and larger (on-DNA compound) substrates. We believe the performance characteristics of the ADE-OPI-MS platform to be very well suited for sample-intensive, small-volume, and HTE workflows. In addition, this platform can be deployed for kinetic reaction monitoring, as demonstrated in an earlier report,14 taking advantage of direct sampling without sample cleanup as well as minimal sample consumption per analysis.

Acknowledgments

We thank K. Brighty for helpful discussions and for reviewing the manuscript, T. Covey for his support around ADE-OPI-MS instrumentation, C. Helal for his support and insights, and L. B. Burton, D. M. Cox, and J. Simpkins for their support around data processing.

Glossary

Abbreviations

HTE

high-throughput experimentation

UT

ultrathroughput

ADE-OPI-MS

acoustic droplet ejection-open-port interface mass spectrometry

PMC

parallel medicinal chemistry

DESI

desorption electrospray ionization

UPLC-MS

ultraperformance LC-MS

DECLs

DNA-encoded chemical libraries

DMSO

dimethyl sulfoxide

DMAc

dimethyl acetamide

EDC-HCl

N-ethyl-N′-[3-(dimethylamino)propyl]carbodiimide hydrochloride

HOPO

2-hydroxypyridine N-oxide

HATU

1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate

DIEA

N,N-diisopropylethylamine

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00066.

  • Experimental methods and figures, additional data analysis results, and plate mapping of parallel chemistry screens and expected product list (PDF)

Author Contributions

§ K.J.D. and W.H. contributed equally to this work.

The authors declare no competing financial interest.

Supplementary Material

ml0c00066_si_001.pdf (4.6MB, pdf)

References

  1. Bunin B. A.; Dener J. M.; Livingston D. A. Application of combinatorial and parallel synthesis to medicinal chemistry. Annu. Rep. Med. Chem. 1999, 34, 267–286. 10.1016/S0065-7743(08)60589-8. [DOI] [Google Scholar]
  2. Arico-Muendel C. C. From haystack to needle: finding value with DNA encoded library technology at GSK. MedChemComm 2016, 7 (10), 1898–1909. 10.1039/C6MD00341A. [DOI] [Google Scholar]
  3. Buitrago Santanilla A.; Regalado E. L.; Pereira T.; Shevlin M.; Bateman K.; Campeau L.-C.; Schneeweis J.; Berritt S.; Shi Z.-C.; Nantermet P.; Liu Y.; Helmy R.; Welch C. J.; Vachal P.; Davies I. W.; Cernak T.; Dreher S. D. Nanomole-scale high-throughput chemistry for the synthesis of complex molecules. Science (Washington, DC, U. S.) 2015, 347 (6217), 49–53. 10.1126/science.1259203. [DOI] [PubMed] [Google Scholar]
  4. Perera D.; Tucker J. W.; Brahmbhatt S.; Helal C. J.; Chong A.; Farrell W.; Richardson P.; Sach N. W. A platform for automated nanomole-scale reaction screening and micromole-scale synthesis in flow. Science (Washington, DC, U. S.) 2018, 359 (6374), 429–434. 10.1126/science.aap9112. [DOI] [PubMed] [Google Scholar]
  5. Schafer W. A.; Hobbs S.; Rehm J.; Rakestraw D. A.; Orella C.; McLaughlin M.; Ge Z.; Welch C. J. Mobile tool for HPLC reaction monitoring. Org. Process Res. Dev. 2007, 11 (5), 870–876. 10.1021/op7000854. [DOI] [Google Scholar]
  6. Khajeh M.; Bernstein M. A.; Morris G. A. A simple flowcell for reaction monitoring by NMR. Magn. Reson. Chem. 2010, 48 (7), 516–522. 10.1002/mrc.2610. [DOI] [PubMed] [Google Scholar]
  7. Foley D. A.; Bez E.; Codina A.; Colson K. L.; Fey M.; Krull R.; Piroli D.; Zell M. T.; Marquez B. L. NMR Flow Tube for Online NMR Reaction Monitoring. Anal. Chem. (Washington, DC, U. S.) 2014, 86 (24), 12008–12013. 10.1021/ac502300q. [DOI] [PubMed] [Google Scholar]
  8. McCullough B. J.; Bristow T.; O’Connor G.; Hopley C. On-line reaction monitoring by extractive electrospray ionisation. Rapid Commun. Mass Spectrom. 2011, 25 (10), 1445–1451. 10.1002/rcm.5016. [DOI] [PubMed] [Google Scholar]
  9. Lin S.; Dikler S.; Blincoe W. D.; Ferguson R. D.; Sheridan R. P.; Peng Z.; Conway D. V.; Zawatzky K.; Wang H.; Cernak T.; Davies I. W.; DiRocco D. A.; Sheng H.; Welch C. J.; Dreher S. D.. Mapping the dark space of chemical reactions with extended nanomole synthesis and MALDI-TOF MS. Science 2018, 361 ( (6402), ), eaar6236. 10.1126/science.aar6236 [DOI] [PubMed] [Google Scholar]
  10. Takáts Z.; Wiseman J. M.; Cooks R. G. Ambient mass spectrometry using desorption electrospray ionization (DESI): instrumentation, mechanisms and applications in forensics, chemistry, and biology. J. Mass Spectrom. 2005, 40 (10), 1261–1275. 10.1002/jms.922. [DOI] [PubMed] [Google Scholar]
  11. Ifa D. R.; Wu C.; Ouyang Z.; Cooks R. G. Desorption electrospray ionization and other ambient ionization methods: current progress and preview. Analyst 2010, 135 (4), 669–681. 10.1039/b925257f. [DOI] [PubMed] [Google Scholar]
  12. Sawicki J. W.; Bogdan A. R.; Searle P. A.; Talaty N.; Djuric S. W. Rapid analytical characterization of high-throughput chemistry screens utilizing desorption electrospray ionization mass spectrometry. Reaction Chemistry & Engineering 2019, 4 (9), 1589–1594. 10.1039/C9RE00054B. [DOI] [Google Scholar]
  13. Wleklinski M.; Loren B. P.; Ferreira C. R.; Jaman Z.; Avramova L.; Sobreira T. J. P.; Thompson D. H.; Cooks R. G. High throughput reaction screening using desorption electrospray ionization mass spectrometry. Chemical Science 2018, 9 (6), 1647–1653. 10.1039/C7SC04606E. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Zhang H.; Liu C.; Hua W.; Ghislain L. P.; Liu J.; Aschenbrenner L.; Noell S.; Dirico K.; Lanyon L. F.; Steppan C. M.; Arnold D. W.; Covey T. R.; Datwani S. S.; Troutman M. D.. Acoustic Ejection Mass Spectrometry for High-Throughput Analysis. bioRxiv 2020, 2020.01.28.923938. [DOI] [PubMed] [Google Scholar]
  15. Maier W. F. Early Years of High-Throughput Experimentation and Combinatorial Approaches in Catalysis and Materials Science. ACS Comb. Sci. 2019, 21 (6), 437–444. 10.1021/acscombsci.8b00189. [DOI] [PubMed] [Google Scholar]
  16. Shevlin M. Practical High-Throughput Experimentation for Chemists. ACS Med. Chem. Lett. 2017, 8 (6), 601–607. 10.1021/acsmedchemlett.7b00165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Williams T.; McCullough K.; Lauterbach J. A. Enabling Catalyst Discovery through Machine Learning and High-Throughput Experimentation. Chem. Mater. 2020, 32 (1), 157–165. 10.1021/acs.chemmater.9b03043. [DOI] [Google Scholar]
  18. Zhang J.; Bellomo A.; Creamer A. D.; Dreher S. D.; Walsh P. J. Palladium-Catalyzed C(sp3)–H Arylation of Diarylmethanes at Room Temperature: Synthesis of Triarylmethanes via Deprotonative-Cross-Coupling Processes. J. Am. Chem. Soc. 2012, 134 (33), 13765–13772. 10.1021/ja3047816. [DOI] [PubMed] [Google Scholar]
  19. Adlington N. K.; Agnew L. R.; Campbell A. D.; Cox R. J.; Dobson A.; Barrat C. F.; Gall M. A. Y.; Hicks W.; Howell G. P.; Jawor-Baczynska A.; Miller-Potucka L.; Pilling M.; Shepherd K.; Tassone R.; Taylor B. A.; Williams A. Process Design and Optimization in the Pharmaceutical Industry: A Suzuki-Miyaura Procedure for the Synthesis of Savolitinib. J. Org. Chem. 2019, 84 (8), 4735–4747. 10.1021/acs.joc.8b02351. [DOI] [PubMed] [Google Scholar]
  20. Sather A. C.; Martinot T. A. Data-Rich Experimentation Enables Palladium-Catalyzed Couplings of Piperidines and Five-Membered (Hetero)aromatic Electrophiles. Org. Process Res. Dev. 2019, 23 (8), 1725–1739. 10.1021/acs.oprd.9b00233. [DOI] [Google Scholar]
  21. Cernak T.; Gesmundo N. J.; Dykstra K.; Yu Y.; Wu Z.; Shi Z.-C.; Vachal P.; Sperbeck D.; He S.; Murphy B. A.; Sonatore L.; Williams S.; Madeira M.; Verras A.; Reiter M.; Lee C. H.; Cuff J.; Sherer E. C.; Kuethe J.; Goble S.; Perrotto N.; Pinto S.; Shen D.-M.; Nargund R.; Balkovec J.; DeVita R. J.; Dreher S. D. Microscale High-Throughput Experimentation as an Enabling Technology in Drug Discovery: Application in the Discovery of (Piperidinyl)pyridinyl-1H-benzimidazole Diacylglycerol Acyltransferase 1 Inhibitors. J. Med. Chem. 2017, 60 (9), 3594–3605. 10.1021/acs.jmedchem.6b01543. [DOI] [PubMed] [Google Scholar]
  22. Franzini R. M.; Samain F.; Abd Elrahman M.; Mikutis G.; Nauer A.; Zimmermann M.; Scheuermann J.; Hall J.; Neri D. Systematic Evaluation and Optimization of Modification Reactions of Oligonucleotides with Amines and Carboxylic Acids for the Synthesis of DNA-Encoded Chemical Libraries. Bioconjugate Chem. 2014, 25 (8), 1453–1461. 10.1021/bc500212n. [DOI] [PubMed] [Google Scholar]
  23. Krska S. W.; DiRocco D. A.; Dreher S. D.; Shevlin M. The Evolution of Chemical High-Throughput Experimentation To Address Challenging Problems in Pharmaceutical Synthesis. Acc. Chem. Res. 2017, 50 (12), 2976–2985. 10.1021/acs.accounts.7b00428. [DOI] [PubMed] [Google Scholar]
  24. Mennen S. M.; Alhambra C.; Allen C. L.; Barberis M.; Berritt S.; Brandt T. A.; Campbell A. D.; Castañón J.; Cherney A. H.; Christensen M.; Damon D. B.; Eugenio de Diego J.; García-Cerrada S.; García-Losada P.; Haro R.; Janey J.; Leitch D. C.; Li L.; Liu F.; Lobben P. C.; MacMillan D. W. C.; Magano J.; McInturff E.; Monfette S.; Post R. J.; Schultz D.; Sitter B. J.; Stevens J. M.; Strambeanu I. I.; Twilton J.; Wang K.; Zajac M. A. The Evolution of High-Throughput Experimentation in Pharmaceutical Development and Perspectives on the Future. Org. Process Res. Dev. 2019, 23 (6), 1213–1242. 10.1021/acs.oprd.9b00140. [DOI] [Google Scholar]
  25. Badland M.; Crook R.; Delayre B.; Fussell S. J.; Gladwell I.; Hawksworth M.; Howard R. M.; Walton R.; Weisenburger G. A. A comparative study of amide-bond forming reagents in aqueous media – Substrate scope and reagent compatibility. Tetrahedron Lett. 2017, 58 (46), 4391–4394. 10.1016/j.tetlet.2017.10.014. [DOI] [Google Scholar]
  26. Franzini R. M.; Neri D.; Scheuermann J. DNA-encoded chemical libraries: advancing beyond conventional small-molecule libraries. Acc. Chem. Res. 2014, 47 (4), 1247–55. 10.1021/ar400284t. [DOI] [PubMed] [Google Scholar]
  27. Malone M. L.; Paegel B. M. What is a ″DNA-Compatible″ Reaction?. ACS Comb. Sci. 2016, 18 (4), 182–7. 10.1021/acscombsci.5b00198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. de Pedro Beato E.; Priego J.; Gironda-Martínez A.; González F.; Benavides J.; Blas J.; Martín-Ortega M. D.; Toledo M. Á.; Ezquerra J.; Torrado A. Mild and Efficient Palladium-Mediated C–N Cross-Coupling Reaction between DNA-Conjugated Aryl Bromides and Aromatic Amines. ACS Comb. Sci. 2019, 21 (2), 69–74. 10.1021/acscombsci.8b00142. [DOI] [PubMed] [Google Scholar]
  29. Lu X.; Roberts S. E.; Franklin G. J.; Davie C. P. On-DNA Pd and Cu promoted C–N cross-coupling reactions. MedChemComm 2017, 8 (8), 1614–1617. 10.1039/C7MD00289K. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Li Y.; Gabriele E.; Samain F.; Favalli N.; Sladojevich F.; Scheuermann J.; Neri D. Optimized Reaction Conditions for Amide Bond Formation in DNA-Encoded Combinatorial Libraries. ACS Comb. Sci. 2016, 18 (8), 438–443. 10.1021/acscombsci.6b00058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Ratnayake A. S.; Flanagan M. E.; Foley T. L.; Smith J. D.; Johnson J. G.; Bellenger J.; Montgomery J. I.; Paegel B. M. A Solution Phase Platform to Characterize Chemical Reaction Compatibility with DNA-Encoded Chemical Library Synthesis. ACS Comb. Sci. 2019, 21 (10), 650–655. 10.1021/acscombsci.9b00113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Hargiss L. O.; Zipp G. G.; Jessop T. C.; Sun X.; Keyes P.; Rawlins D. B.; Liang Z.; Park K. J.; Gu H. Reaction profiling by ultra high-pressure liquid chromatography/time-of-flight mass spectrometry in support of the synthesis of DNA-encoded libraries. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2014, 971, 120–5. 10.1016/j.jchromb.2014.09.001. [DOI] [PubMed] [Google Scholar]
  33. Chow S. Y.; Nelson A. Embarking on a Chemical Space Odyssey. J. Med. Chem. 2017, 60 (9), 3591–3593. 10.1021/acs.jmedchem.7b00423. [DOI] [PubMed] [Google Scholar]
  34. Covey T. R.; Schneider B. B.; Javaheri H.; LeBlanc J. Y.; Ivosev G.; Corr J. J.; Kovarik P.. ESI, A., and MALDI a comparison of the central analytical figures of merit: sensitivity, reproducibility, and speed, ESI, APCI, and MALDI a comparison of the central analytical figures of merit: sensitivity, reproducibility, and speed. In Electrospray and MALDI Mass Spectrometry: Fundamentals, Instrumentation, Practicalities, and Biological Applications; Core R. B., Ed.; Wiley: 2010. [Google Scholar]
  35. Schneider B. B.; Nazarov E. G.; Londry F.; Vouros P.; Covey T. R. Differential mobility spectrometry/mass spectrometry history, theory, design optimization, simulations, and applications. Mass Spectrom. Rev. 2016, 35 (6), 687–737. 10.1002/mas.21453. [DOI] [PubMed] [Google Scholar]
  36. Muhammad Asif A.; Paul R.; Wei W.; Brandon P.; William K.. Method for Heterogeneous Sample Analysis on LCMS Without Sample Manipulation. chemRxiv 2019, 10.26434/chemrxiv.9722576.v2. [DOI] [Google Scholar]
  37. Cruz S. C.; Rothenberg G.; Westerhuis J. A.; Smilde A. K. Tackling Calibration Problems of Spectroscopic Analysis in High-Throughput Experimentation. Anal. Chem. 2005, 77 (7), 2227–2234. 10.1021/ac048421c. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ml0c00066_si_001.pdf (4.6MB, pdf)

Articles from ACS Medicinal Chemistry Letters are provided here courtesy of American Chemical Society

RESOURCES