High throughput screening of complex biological samples with mass spectrometry – from bulk measurements to single cell analysis

Emily E. Kempa a, Katherine A. Hollywood b, Clive A. Smith c and Perdita E. Barran *ab
aMichael Barber Centre for Collaborative Mass Spectrometry, Manchester Institute of Biotechnology, The University of Manchester, Manchester, M1 7DN, UK. E-mail: Perdita.Barran@Manchester.ac.uk
bManchester Centre for Synthetic Biology of Fine and Speciality Chemicals (SYNBIOCHEM), Manchester Institute of Biotechnology, The University of Manchester, Manchester M1 7DN, UK
cSphere Fluidics Limited, The Jonas-Webb Building, Babraham Research Campus, Babraham, Cambridge, CB22 3AT, UK

Received 30th July 2018 , Accepted 8th December 2018

First published on 10th December 2018


High throughput screening (HTS) of molecular analytes is in high demand from and implemented in many areas of chemistry, medicine and industrial biotechnology including the discovery of biomarkers and the development of new chemical entities. Despite its prevalence, technical challenges remain in many of the new application areas of HTS which require rapid results from complex mixtures, for example in: screening biotransformations; targeted metabolomics; and in locating drugs and/or metabolites in biological matrices. Common to all of these are lengthy and costly sample preparation stages, involving recovery from cell cultures, extractions followed by low throughput LC-MS/MS methods or specific fluorescence measurements. In the latter the target molecules need to be inherently fluorescent or to include a fluorescent label or tag which can adversely influence a cellular system. Direct infusion mass spectrometry coupled with robotic sample infusion is a viable contender for information rich HTS with sub-second analysis times, and recent developments in ambient ionisation have heralded a new era where screening can be performed on crude cell lysates or even from live cells. Besides commercially available technologies such as RapidFire, Acoustic Mist Ionisation, and the TriVersa ChipMate there are promising new developments from academic groups. Novel applications using desorption electrospray ionisation, microfluidics, rapid LC-separation and ‘one cell’ direct infusion methods offer much potential for increasing throughput from ‘messy’ complex samples and for significantly reducing the amount of material that needs to be analysed. Here we review recent advances in HTS coupled with MS with an emphasis on methods that reduce or remove all sample preparation and will facilitate single cell screening approaches.


1. High throughput screening (HTS) and ultra-high throughput screening (uHTS)

High throughput screening (HTS) and ultra-high throughput screening (uHTS) techniques are described as methods able to analyse ten to one hundred thousand samples per day.1–6 The opportunity to screen at this speed is in high demand from the pharmaceutical and biotechnological industries as well as for disease and health monitoring, to develop new drug leads and to develop high value chemicals.5,7,8 Accompanying such an aspiration to screen complex mixtures ever more rapidly are regulatory requirements, which necessitate more sustainable processes, and require reproducible methods. Industrial biotechnology aims to use bacterial cells or other small organisms to produce compounds of interest in a manner that produces less waste and consumes less energy than the equivalent synthetic chemistry route. This is an aspect of synthetic biology, where new biological components such as enzymes and genetic circuits are reprogramed into bacterial cells to alter cellular behaviour or properties9,10 in order to manufacture specified products such as biofuels and biomaterials.11–13 Synthetic biology can be used to produce new compounds via a process known as directed evolution where genetic sequences are randomly altered to improve the efficiency of a given cellular process,10,13 with a concomitant requirement for the analysis of up to and over 100[thin space (1/6-em)]000 cellular variations for a given target.13–15

Traditionally, the starting components for a screen are provided in 96, 384 or even 1536 well-plate formats16,17 for their ease of flexibility between plate handling robotics4,5 and multiple analytical techniques; particularly fluorescent or colorimetric well plate readers18–20 and chromatography and mass spectrometry injection systems.21,22 Fluorescent well plate screening is widely utilised,6,23 notably in the analysis of drug interactions to produce dose–response curves.18,24–27 These systems often offer the highest throughput, with a fluorescent well plate reader achieving below 1 s per sample,28 and additionally do not require great amounts of data analysis to determine ‘hit’ rates, allowing for a simple positive or negative response for each well in the plate. However, these assays rely on the presence of an active chromophore within the system of interest, or to be coupled to secondary chemical reactions29 or fluorescent tags as an indirect readout of activity. Both cases can suffer from interference and require high system suitability for successful and accurate assays to take place.

Decreasing the reliance on fluorescent and colorimetric assays for monitoring chemical flux in cells increases the number of detectable species that define the process, or indeed are the target applicable to many areas including disease screening and synthetic biology. Mass Spectrometry (MS) is heavily utilised in these fields30–33 due to its ability to detect a characteristic mass to charge ratio (m/z) of analytes in the sample, providing the analyte can form a gas phase ion or ions.34 The range of platforms available for MS and the diversity of ionisation sources mean that different samples are amenable to MS analysis. For many analyses the most common approach is to include a separation step prior to MS – most notably gas or liquid chromatography.34 Most of the apparatus for these systems utilise autosamplers to increase the throughput; however, the separation of a complex biological sample, for example, plasma or liquid bacterial extracts, can take in excess of 30 minutes per sample – i.e. not ideal for studies which require analysis of greater than 500 samples. Not only would the time required be unattainable for one analyst, the large amount of both sample material and solvents required (LC), data processing and waste generated becomes undesirable. As a consequence, there has been considerable effort to remove lengthy chromatography separations, and instead infuse the sample directly into the mass spectrometer, allowing higher throughput analysis.

Commercialised instruments that are coupled to mass spectrometers for direct infusion discussed herein include the TriVersa NanoMate, LESA and the Agilent RapidFire, with the applications, advantages and disadvantages discussed for each within the relevant sections. Instruments yet to be brought to the market, such as microfluidics and Labcyte Echo coupled MS (or Acoustic Mist Ionisation MS) systems, are also detailed as well as MS imaging technologies which have shown the potential to be adapted for HTS applications. Fig. 1 summarises the throughput and the smallest amount of sample required for the aforementioned techniques. The latter half of this review will concentrate on MS technologies utilised in the realm of single cell analysis, with particular focus on their current throughput and the potential for this to be advanced. It is noted that many of these single cell MS technologies are yet to come to the market and have been developed by academic groups; hence, an assessment of the future marrying of HTS, commercialised instrumentation and single cell analysis will be made, alongside recommendations regarding the technological improvements needed to achieve such a goal.


image file: c8an01448e-f1.tif
Fig. 1 Selected commercialised mass spectrometry HTS instrumentation discussed here depicted low to ultra-high sample throughput. The reported rate of sample analysis for each has been converted into the number of samples possible to analyse per 24 hour period (day) should the instrumentation be performed continuously at this level. The minimum sample volume has also been quoted for each technique. All schematics and photography of the instrumentation summarised have been taken from the websites of the respective manufacturers (https://www.agilent.com/en/products/mass-spectrometry/rapidfire-high-throughput-ms-systems/rapidfire-365, https://advion.com/products/triversa-nanomate/, http://www.spherefluidics.com/product-category/microfluidic-biochips/, https://prosolia.com/products/desi-2d/, https://www.labcyte.com/echo-technology/acoustic-mass-spec).

2. TriVersa NanoMate and LESA

2.1. NanoMate

The TriVersa NanoMate (Advion, Ithaca, NY) is a fully automated nanoelectrospray (nESI) sample introduction method for the mass spectrometry analysis of multiple samples, including those directly obtained from microtiter plates.35–38 This front end system consists of a liquid handling robot coupled with a silicon microchip containing an array of up to 400 nanoelectrospray nozzles to deliver the sample to the mass spectrometer inlet via a pipette tip into which the sample has been aspirated from a conventional well plate.35,36 Each nozzle and pipette is used only once to minimise sample carryover between experiments and, as with conventional nESI, only requires small sample volumes (1–10 μL).35,36,39,40 Upon delivery of the sample from the pipette tip, nitrogen gas and a spray voltage are applied to the nozzle to generate a nESI plume directed towards the MS inlet (see Fig. 2).35,40 MS data acquisition can then take place for a pre-set time for each sample, a parameter which can be shortened to increase sample throughput or lengthened to improve data quality (increased number of data points per sample).35 Van Pelt et al. reported that MS data acquisition of 5 seconds, and a NanoMate cycle time of 40 seconds, allowed for a 1 μL sample to be analysed every 45 seconds.35 This equates to just under 2000 samples in 24 hours, positioning the NanoMate at the lower end of a scale of samples per time unit (Fig. 1). Early inceptions of this instrument struggled with analysis in negative ionisation mode, and this problem has not been entirely surmounted. In addition, the advantages gained by reducing carryover provided by using single use nozzles and chip orifices mean that each analysis is costly. Chip orifices do clog and some solvents can strip the conductive coating from the sampling chip. Despite these disadvantages, the NanoMate is versatile and has been interfaced with numerous different mass spectrometers across multiple vendors.39,41–43 The low flow rates and lower applied voltage of nESI bring increased sensitivity, milder ionisation conditions allowing non-covalent complexes to be examined38,40 and lower sample volumes in comparison with electrospray ionisation (ESI) mass spectrometry, as well as no carryover.36,38 The lack of an LC separation stage can lead to ion suppression of some analytes and oxidation of certain compounds due to the conductive pipette tips utilised to apply the nESI voltage to the solution.35 Due to the one time use of the chip and electrospray nozzles, sustained procurement of consumables is needed for this apparatus to run continuously in a high throughput manner. This may become unsustainable for laboratories in comparison with LC-MS approaches or even direct infusion from an autosampler.
image file: c8an01448e-f2.tif
Fig. 2 (A) Photograph of the TriVersa NanoMate mounted on a mass spectrometer; (B) photography of the NanoMate nESI chip containing 10 nozzles with a US quarter dollar for scale (24.3 mm); (C) a schematic illustration of the Avidon TriVersa NanoMate action, in which the sample containing a pipette tip is pressed onto the nESI chip, where a voltage is applied to generate an electrospray plume. Reproduced with permission from Van Pelt et al., Rapid Commun. Mass Spectrom., 2003, 17, 1576.35

The TriVersa NanoMate has also been used to facilitate LESA (Liquid Extraction Surface Analysis) approaches (discussed in section 2.2), which have enabled ganglioside38 and ceramide lipidomic analysis,42 top down40 and bottom up proteomics37 as well as selected39 and multi-reaction monitoring.37

Omitting LC separation can remove in excess of 30 minutes of run time per sample, leading to greatly increased throughput; however, the lack of separation can lead to ion suppression during the electrospray process due to both co-elution and matrix effects, which may affect the accuracy of quantitative results. Van Pelt et al. undertook a study in 2003 where they compared a fully automated NanoMate approach with an LC-MS/MS method for the analysis of Caco-2 cell samples. Here the LC method consisted of a gradient elution time of 5 minutes per sample, whereas the NanoMate method achieved sufficient sample data in 1 minute and 45 seconds per sample, an increase in throughput of almost 2.9 times. However, it is important to note that for the NanoMate analysis an additional off-line desalting step was required, but again this can be undertaken through the use of C18 ZipTips and a fully automated liquid handling robot to minimise loss of throughput. In this case, two different internal standards were analysed within the sample sets and the results were in good agreement, leading to the conclusion that the NanoMate method was robust and holds great potential as a high throughput alternative to current LC-MS/MS methods.35 More than 15 years since its launch, the NanoMate continues to be a useful method for HTS in many laboratories.44,45

Gangliosides are complex glycosphingolipids composed of a ceramide lipid unit and an oligosaccharide chain. These species are prominent within the cells of the human nervous system and are known to play key roles in cell signalling processes and are often the focus of research into biomarkers of brain diseases including cancer.38,46 In the past (and still to this day), chromatographic methods (both LC and TLC) were key to the characterisation of these species due to the need to separate complex mixtures prior to mass spectrometry analysis;38,43,46,47 however, many groups have now begun to omit this chromatography step due to lack of sensitivity and instead adopt the high throughput capabilities of a NanoMate MS/MS workflow. Zamfir and Serb et al. have demonstrated their ability to identify different gangliosides in both healthy and tumorous nervous system tissues by mass assignments obtained using a NanoMate workflow.43,46,47 In 2011, mass spectra were acquired under identical conditions at a rate of 1 minute per sample, with the healthy tissues found to contain a larger variety of singly, doubly and triply charged ganglioside structures in comparison with diseased samples, which were instead found to contain only singly charged moieties with shorter oligosaccharide chain lengths and reduced sialic acid content.46 The rapid characterisation of such structures has potential for biomarker discovery, and enables the identification of biological pathways which can eventually lead to the development of a therapeutic intervention. Park et al. also adopted tandem MS analysis for metabolomics and biomarker identification, in which the MS profiles of ceramide species were studied in mouse skin following UVB irradiation.42

Flangea et al. developed high throughput top-down sequencing with samples infused by a NanoMate employing ETD (electron transfer dissociation) and CID (collision induced dissociation) using a high-capacity ion trap MS.40 Top-down analysis of intact proteins allows analysis of protein complexes and non-covalent interactions, and quantification of post-translational modifications and can result in higher sequence coverage in comparison with bottom-up methods.48 Flangea et al. achieved 80% sequence coverage for the 16+ charge state of apomyoglobin (16.95 kDa) using a NanoMate and performed ETD with high reproducibility during only 30 seconds of fragmentation time within the mass spectrometer.40 Hence, this information rich, high throughput approach gives hope that larger proteins may be sequenced this way, along with detailed studies of their post-translational modifications.

2.2. Liquid extraction surface analysis (LESA)

Liquid extraction surface analysis (LESA) is a technique first described by van Berkel and co-workers and was commercialised by Advion allowing surfaces to be sampled under ambient conditions for subsequent automated MS analysis.37,49–51 Through the use of the conductive pipette tips mentioned above, a solvent is aspirated from a trough or well and deposited onto the surface for a defined period of time. During this time the pipette tip remains in contact with the liquid spot as the solvent extracts analytes from the surface into the droplet, before being re-aspirated and inserted into the inlet side of the nESI chip.49–53 At this stage, a voltage is applied to the chip along with nitrogen gas pressure to directly electrospray the extracted analytes from the solvent droplet. This extraction procedure can then continue, under the control of the on-board camera,52,54 or pre-programmed XYZ positions across the sample to achieve high throughput analysis of different sections of the solid surface. Surfaces that have been successfully analysed in this way include bloodspots previously dropped onto paper,37,52 biological tissue sections,51,55,56 agar plates50,53 or even the surface of a fruit or vegetable.54 As with conventional NanoMate, the nozzles and pipettes are single use, thus minimizing carryover between each sample, and adding to its advantages is its ability to minimise sample preparation steps such as extraction prior to analysis.

The LESA-NanoMate coupled to an LTQ-OrbitrapXL MS (Thermo Fisher Scientific, Bremen, Germany) has been used to monitor antibiotic product formation directly from bacterial colonies on an agar plate.53 Optimisation of solvent composition and procedure minimised the extraction of media components from the agar. Additionally, only small modifications to the Petri dish were required to accommodate the NanoMate platform size.53 This application to whole colony screening (and similar approaches involving other ambient ionisation methods; see later in this review57) has great potential for synthetic biology applications to rapidly identify and quantify products from genomic variants.58 In 2014, Randall and co-workers took a similar approach to the direct analysis of E. coli strains using LESA to observe a range of intact proteins, confirming that this approach to analysis is feasible for a range of bacterial strains and compound classes. This method is still at an early stage, and more work must be done to facilitate the transmission and analysis of lower abundance proteins. The authors show that the appearance of some proteins and subsequent top-down analysis can help to determine the health of a given colony, which may have future use in analysing contaminated surfaces.50

3. The use of solid phase extraction coupled to mass spectrometry

Solid phase extraction (SPE) is a sample preparation technique first introduced in the 1970s, which selectively isolates compounds of interest from a liquid sample through the use of a cartridge or syringe packed with a porous solid phase chosen for the retention of the analyte class in question.59–61 The sample is loaded onto a stationary phase through manual pipetting or an automated injection system. Washing the stationary phase with varying solvent compositions removes the sample matrix (as illustrated in Fig. 4[thin space (1/6-em)]62) which may interfere with downstream processing, for example in MS analysis.61 Post clean-up, the sample is then re-eluted for further processing or analysis before re-equilibration of the porous phase to remove any strongly retained species.59 Cartridges can be disposable or reusable, situated online, coupled with chromatography, or offline within the workflow.61 SPE has been and is still a fundamental sample preparation technique applied in many areas including food science,63,64 pesticide analysis,65,66 drug discovery and evaluation,67–69 forensics70 and bioanalysis.71–73 In the past, higher throughput sample preparation has also been achieved by employing automated SPE approaches such as the Zymark Rapid Trace which can perform extractions of up to ten samples in tandem.74
image file: c8an01448e-f3.tif
Fig. 3 Illustration detailing the individual steps undertaken during LESA surface analysis. (A) A clean conductive pipette tip is taken up by the robotic arm; (B) extraction solvent is aspirated from a well; (C) extraction solvent is deposited onto the sample position; (D) extraction solvent and dissolved sample analytes are re-aspirated into the tip; (E) the pipette tip is pressed onto the nESI chip for MS analysis. Image reproduced with permission from Kertesz et al., J. Mass Spectrom., 2010, 45, 255.51

image file: c8an01448e-f4.tif
Fig. 4 Diagrammatic representation of a solid phase extraction procedure (SPE) highlighting the retention of the sample analytes, rinsing of the sorbent to remove and discard unwanted analytes, and elution steps to remove and collect the molecules of interest. Reproduced with permission from Solid-Phase Extraction: Principles, Techniques and Applications.62

3.1. Agilent RapidFire

The Agilent RapidFire is a front-end sample introduction system for mass spectrometry, which is based on a solid phase extraction procedure in the place of a LC column to achieve salt removal and sample clean-up prior to analysis.75–77 This online preparation system can deliver a cycle time as small as seven seconds per sample.77–79 The procedure first involves sample aspiration (or ‘sipping’) from a well plate, before adsorption (or ‘loading’) of the sample onto a removable SPE cartridge of a user defined stationary phase. A washing step then removes buffer salts and other non-retained analytes from the cartridge for a set period of time, before a solvent of differing composition is then used to co-elute the remaining sample components of interest from the cartridge and into the mass spectrometer inlet for analysis. The cartridge is then re-equilibrated to the clean-up method starting conditions before injection of the next sample.80–84 Each wash cycle, sipping time (i.e. sample volume loaded onto the cartridge), elution time and MS acquisition time contributes to the total sample analysis time. Timings need optimisation for differing sample types, and similarly, both wash and elution solvent compositions and the cartridge stationary phase will also need to be considered, with each cartridge capable of enduring in excess of 3000 injections before decline.75,79,81 The instrument also incorporates a robot such that up to 63 well plates (96 or 384 wells) can be handled at a time.85,86 The whole system facilitates the analysis of a maximum of 24[thin space (1/6-em)]192 samples before plate intervention, and if the minimum reported cycle time of ∼7 seconds per sample is possible,87 12[thin space (1/6-em)]300 samples can be screened in one 24 hour period. This quoted time of 7 s is rarely achieved, and ∼30 s are more commonly presented.

Similar to other ‘chromatography-free’ approaches, ion suppression and matrix effects during the analysis of complex samples can be problematic due to the lack of a separation step and, equally, insufficient washing of the SPE cartridge leads to carryover between samples.88,89 In comparison with LC-MS, the time taken to analyse a blank sample by the RapidFire system is negligible; hence the addition of blank runs at suitable intervals to check for carryover need not significantly hinder the throughput of the method.75 Elution and re-equilibration steps during analysis must also be optimised to minimise carryover from the SPE cartridge and retain high throughput. For complex samples, some clean-up prior to analysis is common, the fluidic lines can start to clog and the sensitivity of the ionisation is therefore reduced. Despite these disadvantages, the Agilent RapidFire is a robust platform which is ubiquitously employed for HTS in the pharmaceutical industry. Noteworthy applications undertaken by the RapidFire system include lead molecule identification,90 removal of false positive results from prior fluorescence assays,77 and analysis of blood,91 plasma,79 honey,92 and in particular, P450 binding interactions.76,93

Lim et al. and Wu et al. have demonstrated the use of the Agilent RapidFire system in place of fluorescence and/or LC-MS/MS approaches to quantitatively probe reactions between P450 enzymes and drug candidates.76,93 Drug–drug interaction (DDI) between existing and new drug candidates has become a major concern when considering a new drug for use, and cytochrome P450 enzymes have been used as exemplar systems to develop new targeting approaches due to previous false positives.76,94 So far, recombinant P450 enzymes have been used in fluorescence based assays with up to 1536 well plates to increase throughput, although these may not best represent endogenous interactions. Alternatively, LC-MS monitoring of reaction products between enzymes and substrates has been used to quantitatively probe drug–drug interactions; however, the lengthy nature of the LC separation step required hampers this approach as a high throughput technique.76 SPE approaches permit the extraction of the desired components within a matter of seconds rather than minutes. Both of these studies76,93 utilised liquid handling robots to prepare multiple concentrations of each substrate with various P450 enzymes within standard well plates, and dose–response curves of each substrate allowed for the determination of the binding constants and inhibition concentrations, KM and IC50, between enzyme and drug. Both SPE-MS/MS methods employed a cycle time of 9 s per well, and in each case they compared RapidFire results to those generated using an LC-MS method. Wu et al. concluded that the SPE-MS/MS approach might be used to initially screen drug libraries and determine lead compounds, which would then in turn have their inhibition levels confirmed by LC-MS/MS.76 Lim et al. found that the SPE-MS/MS method for probing their P450 systems gave a 15-fold increase in throughput in comparison with the traditional LC-MS/MS approach, and the method was both sensitive and reproducible.93

To improve the throughput further and to obtain rates similar to those of fluorescence assays, Leveridge et al. developed a strategy in which sets of 4 samples were pooled (or ‘multiplexed’) during well plate preparation prior to RapidFire analysis.90 Approaches such as these do necessitate more complex data analysis and more intelligent compound management, for example, the requirement to combine compounds which will not react with one another; however, this can be a small price to pay for a 4-fold decrease in analysis time in comparison with the conventional method of 1 sample per well. Given the acquisition speed of many modern mass spectrometers, multi-reaction monitoring is still possible and has, for example, been employed for the analysis of peptides. Using two RapidFire instruments and 4-fold sample multiplexing, it has been stated that up to 280[thin space (1/6-em)]000 compounds could be screened in 1 week.90 Potentially more samples could be incubated within a single well to increase throughput further; however this would be sample dependent and require optimisation.

4. Acoustic droplet mass spectrometry

4.1. Labcyte Echo liquid handling technology

Acoustic droplet ejection technology is a promising recently developed alternative to traditional tip-based liquid handling robotics, and with the commercialisation of the Labcyte Echo, this has allowed for the decrease in sample preparation time, in particular aiding in the generation of well plate assays for subsequent high-throughput screening methods.95–97 The focussing of sound waves through the use of transducers allows nanolitre volumes of liquids to be transferred from one reservoir to another without the need for a manual or automated pipette. In the case of the Labcyte Echo, 2.5 nL droplets are moved from a source plate into a destination plate well that is suspended upside down to achieve a desired transfer volume.98 To avoid samples ‘falling out’ of the well, the volumes transferred are optimised to ensure that surface tension overcomes gravity.99,100 This ‘tip-less’ approach not only removes the risk of carryover between samples, but additionally allows for increased accuracy and precision during dispensing due to the removal of forces such as surface tension and viscosity between the liquid sample and a disposable tip.99–101 The removal of tip-based transfers is also advantageous for the avoidance of shear force upon cell cultures which can in turn cause damage and affect cellular viability.100 As with traditional automated liquid handling approaches, here operational speeds are increased and the miniaturizations of sample volumes also benefit the throughput of subsequent screening approaches.101 Once prepared, the destination well plate can then proceed to optical analysis such as fluorescence or even to prepare plates for automated injection for direct infusion mass spectrometry analysis (Fig. 5).
image file: c8an01448e-f5.tif
Fig. 5 Schematic diagram of the Labcyte Echo highlighting the destination and source plate. The acoustic transducer causes ejection of a droplet from the reservoir and is captured within an empty well held upside down above the source plate. Adapted with permission from Seidel et al., Angew. Chem., Int. Ed., 2010, 53, 7948–7951.98

4.2. Acoustic mist ionisation mass spectrometry

The use of acoustic technologies to aid throughput within mass spectrometry sample preparation has been possible for over 10 years, most notably during the deposition of MALDI matrices directly onto sample plates.102,103 More recently, manipulation of liquids by acoustic waves or pulses has allowed the transfer of samples from well plates or other reservoirs directly into the mass spectrometer. This has been achieved both directly by Sinclair et al.104via an adaptation of the Echo liquid handling system termed Acoustic Mist Ionisation (AMI) and through the coupling of SAWN (surface acoustic wave nebulisation) to mass spectrometers,105 examples of each will be described below.

A novel advancement in acoustic droplet ejection which holds great promise in dramatically improving the rates of HTS was demonstrated in 2016 by Sinclair et al. as part of a collaboration between AstraZeneca (Macclesfield, UK), Waters Corporation (Wilmslow, UK), and Labcyte Inc. (Sunnyvale, CA, USA).104 In this method, ultrasonic pulses eject a mist of femtolitre volume droplets from well plates directly into the mass spectrometer inlet via a heated transfer tube (see Fig. 6). The fine droplets become charged during this process, due to a voltage gradient applied between the well plate and the sampling orifice, desolvation of these ultralow volume droplets and subsequent transfer to the mass spectrometer is rapid. An xyz stage accommodating a 384 well plate has automated the sample analysis with cited analysis times of 250 ms per well, meaning that an entire 384 well plate can be analysed in less than two minutes. This throughput of approximately three samples per second means that this technology surpasses, in terms of frequency of each sample delivery, any commercially available mass spectrometry introduction technique. The reduction of the droplet diameter from the acoustic wave device drastically improves ionisation efficiency as has been shown in nESI tips by Williams and co-workers.106,107 As with all direct ionisation MS approaches this platform omits chromatographic separation of complex mixtures with commensurate ion suppression and matrix effects; however, it is noted that the addition of ion mobility separation again has the potential to add another dimension to structure identification should it be incorporated into the current assembly.104


image file: c8an01448e-f6.tif
Fig. 6 Schematic diagram of the acoustic misting process undertaken from a well plate during the coupling of a Labcyte Echo with a MS. As the ejected droplets enter the ion guide capillary they become charged due to the application of a voltage before undergoing the desolvation process within the source. Adapted with permission from Sinclair et al., J. Lab. Autom., 2016, 21, 19–26.104

AMI-MS instrumentation is still in an early stage of development and at the time of writing not commercially available. It is likely that some finessing is still needed in order to bring this technique into mainstream usage; however, it has been made available for pre-competitive projects by AstraZeneca in the UK. To date AMI has been limited to one make (Waters Corp.) and model (Q-Tof Xevo) of instrument; it also remains to be seen if this system can be flexible between other mass spectrometry platforms, although clearly it is most suited to analysers with rapid scan rates (Fig. 6).

Although not directly coupled with mass spectrometry and similar to MALDI plate depositions, an approach of Haarhoff et al. utilises acoustic transfers for the deposition of assay mixtures including drug–enzyme inhibition studies from traditional well plates on to a 384 LazWell plate for laser diode thermal desorption (LDTD) MS/MS analysis.28 Using a Labcyte Echo 550, the transfer of 100 nL portions of material for 384 wells can be undertaken in three minutes and with a LDTD-MS/MS analysis time of ∼2 s per sample, the incorporation of this sample deposition technique has greatly improved the throughput in comparison with manual spotting of 1–2 microliter samples onto LazWell plates. It is also noted that the minimised sample volume effectively eliminates associated sample drying time, something that also needs to be taken into consideration during manual deposition of larger amounts. IC50 and dose–response curves for the assays undertaken were compared and yielded good agreement with the results obtained from RapidFire assay analysis; however, since the analysis time for RapidFire averaged ∼10 s per well, the Echo coupled LDTD approach offered greater throughput and miniaturised volumes. The lack of any SPE cartridge and mobile phase in LDTD also removes the possibility of sample carryover between wells.28

5. Mass spectrometry imaging for HTS

MS imaging (or MSI) refers to a number of techniques which utilise mass spectrometry to build a 2D (and sometimes 3D) map of the location of chemical species within a sample, and is of particular use for solid samples and surface analysis.108 The major techniques include desorption electrospray ionisation (DESI), matrix assisted laser desorption ionisation (MALDI) and secondary ionisation mass spectrometry (SIMS) each of which has its own advantages and disadvantages.109,110 These MSI approaches each offer different spatial resolutions, with DESI reaching ∼40 μm,51,111 MALDI ∼5 μm112–114 and SIMS ∼100 nm.115 SIMS in this case offers the greatest potential for application in single cell analysis; however; as the throughput and sensitivity of this technique are limited it will not be discussed here.

5.1. MALDI-MS

Matrix assisted laser desorption ionisation (or MALDI-MS) is an approach in which a laser is used alongside a chemical charge donor (or ‘matrix’) to promote ionisation and transfer of non-volatile analytes into the gas phase for subsequent mass analysis.116 This technique is deemed as ‘soft’ ionisation and is often used for labile molecules and is particularly useful in biochemistry for the analysis of peptides, proteins, lipids and metabolites without fragmentation of the molecules of interest.116,117 Although MALDI-MS has been utilised in imaging applications such as the profiling of tissue sections,118–120 the instrumentation also lends itself well to the microarray format (sometimes referred to as MAMS – microarrays for mass spectrometry) which is often utilised to increase sample throughput.121,122 These arrays often take a similar format and size to a multi-well plate and are mounted on a translatable stage in order for each sample to be ‘struck’ or irradiated by the laser in turn. Additionally, to increase throughout further prior to analysis, automated or spray based matrix deposition can be employed, for example, the use of an acoustic multi-spotter.123

MAMS coupled with MALDI allowed for high throughput monitoring of cocaine in hair samples.121 This procedure utilises a custom made MAMS slide which can contain 600 hydrophilic spots, with each MALDI target plate able to hold 3 of these custom made slides (i.e. 1800 spots per target plate). Besides the fabrication of the target plate, an aliquoting procedure via a metal slider was developed to generate 60 replicate deposits of a sample from a reservoir also incorporated into the MAMS slide design. The same sliding technique facilitated reproducible spotting of the matrix onto all of the sample spots, with one pass of the slider device. Samples in the reservoir consist of a ‘solvent-based’ liquid extraction from hair samples followed by MALDI-TOF-MS analysis involving continuous sample plate stage motion.

Microarray formats can also be of use for tissue and organ section analysis as the grid like application generates defined co-ordinates for systematic sampling across a surface.124,125 This is particularly useful when the tissue sections are large in area, and for screening multiple sections. Arraying the matrix allows the analysis time to be reduced as the instrumentation can be targeted to quickly raster across a larger area, but this limits sample collection to predefined spots. The advantage of this is that the entirety of the tissue section is not destroyed or perturbed by the sample preparation procedure, or by the impact of the laser beam, allowing the sample to be revisited for further analysis including histology if required. Groseclose et al. employed tissue microarrays and MALDI microarray formats in a high throughput proteomic study. Each tissue biopsy was analysed in a grid formation following automated spotting of trypsin solution and matrix upon the tissue (i.e. the entirety of the tissue section was not coated with trypsin or matrix solution).125 Only the grid coordinates which contained the trypsin digest/matrix spots were then subjected to MALDI-MS analysis (Fig. 7). The proteolytic peptide mass spectra acquired along with histology staining produced a training set then applied in the identification of cancerous and non-cancerous regions. Although this approach holds promise with regard to the throughput and the breadth of the ensuing data produced, the authors caution that more robust data analysis tools and larger sample sets of differing types would be required before translation into the clinic.125


image file: c8an01448e-f7.tif
Fig. 7 Example of mass spectrometry imaging utilising tissue microarrays. The TMA used in this study contained various types of biopsies including squamous cell carcinoma, adenocarcinoma, and bronchioloalveolar carcinoma as well as non-cancer tissue from matched individuals. (a) Cylindrical tissue biopsies are first taken from individual tumours with cancerous and normal cell regions indicated by the inset. (b) Trypsin and matrix are spotted in selected positions across the array with the inset demonstrating on average 6 × 6 sample spots available per tissue microarray. Adapted with permission from Groseclose et al.125

5.2. DESI-MS

Desorption electrospray ionisation (DESI) mass spectrometry is a relatively new technique in comparison with MALDI, and was first described in 2004 by Z. Takáts and R. G. Cooks.126 Utilising a solvent stream, gas flow and applied voltage, an electrospray plume is generated and directed at the surface of interest under ambient conditions. As the charged droplets impact the surface, sample analytes are dissolved within the electrospray stream, and secondary droplets transferred towards the mass spectrometry interface, into which desolvation occurs and the charges within the droplet come to reside upon the analyte, as seen in conventional ESI mechanisms.127–130 An xyz stage mounted below the DESI spray facilitates imaging analysis to be conducted at varying speeds and resolution as desired by the user. Since its inception DESI-MS has been applied to many sample types, namely tissue sections,131,132 bacterial and fungi analysis57,133,134 and even for foodstuffs;135,136 however, for the purpose of this review two high throughput applications (direct bacterial colony analysis57 and reaction substrate screening137) will be briefly discussed.

A compatible xyz stage (such as the one manufactured by Prosolia, Inc.) facilitates screening in microarray format to be undertaken with DESI, as demonstrated by Wleklinski et al., who performed reaction screening upon DESI plates containing 6144 (50 nL) sample spots. Under the speeds and spot densities analysed, sample throughput approached ∼10[thin space (1/6-em)]000 spots per hour, a throughput that rivals many other techniques.137 Reaction screening undertaken allowed the assessment of product formation from a number of different substrates tested within an n-alkylation reaction, and also the effect of different bases on a Suzuki cross-coupling reaction, both of which have important applications within medicinal chemistry. Each reaction was first prepared within 384 well plates before transfer to the DESI plate with both of these steps performed by a liquid handling robot.137

Yan and co-workers have used DESI to analyse bacterial colonies in order to determine the production of particular biotransformation products – terming it DiBT-MS (Direct BioTransformation-Mass Spectrometry).57 This was achieved by culturing bacterial colonies upon agar plates containing a nylon membrane support, which allowed intact colonies to be subjected to a range of different reaction conditions and substrates and subsequently analysed using DESI (see Fig. 8). Not only does DiBT-MS permit different reaction conditions to be screened, but also post sampling, colonies of interest can be recovered and subjected to subsequent DNA analysis to identify the mutations which may be of interest.57 This approach has promise for industrial biotechnology and synthetic biology as the number of different genetic variations of interest often surpasses the throughput of currently available label-free technologies. Yan et al. show that DiBT-MS can quantify the substrate–product conversion rates and also perform some semi-quantitative limit of detection analysis; however, absolute quantification (readily achieved by LCMS) remains a challenge. Limit of detection levels are required to be quoted in units of area, which does not translate well to colonies spread unevenly upon the membrane. Additionally, colonies are variable and behave differently with different substrates, which indicates that reaction conversion rates will be challenging to evaluate.


image file: c8an01448e-f8.tif
Fig. 8 Workflow utilised for the direct monitoring of biotransformations within live bacterial colonies by DESI-IM-MS. E. coli cells containing the plasmid of interest are first spread onto agar plates containing a nylon membrane support, on which the cells actively grow and are transferred between agar plates containing a protein inducer. After growth the membranes are removed and placed upon filter paper soaked in the reaction substrate. The membrane containing colonies are then analysed by DESI-IM-MS. Adapted with permission from Yan et al., J. Am. Chem. Soc., 2017, 139, 1408–1411.57

As the above examples show, methods based on DESI-MS hold much promise in the area of HTS, and it is noted that this technology has recently become available as a standard ‘inclusive instrument package’ via Waters Corp. (Wilmslow, UK), which may herald an increased uptake. Wleklinski et al. and Yan et al. have both demonstrated the potential of DESI-MS for HTS, but in our experience such experiments still require significant time to optimise. In particular the sprayer position relative to the sample stage requires ‘tuning’ for each membrane for optimum results, indicating the need for specialised sample stages. Another obstacle that must be overcome in future applications of MS imaging techniques to HTS is that of the software that controls the scan rate across surfaces. For biomedical imaging applications, the DESI sprayer will be controlled by software to carefully (and slowly) raster across the surface, whereas for HTS a randomised data directed search which may sacrifice resolution for speed would be more appropriate.

6. Microfluidics coupled with mass spectrometry

6.1. Microfluidics overview

The transition towards lab-on-a-chip devices from traditional glass and plasticware has already brought significant benefits to the analysis of chemical and biological samples. Integration of robotic methods for chemical synthesis or robotic handling systems for cell-based chemical assays followed by high-throughput analytical screening into one device which incorporates only small fluid volumes will have substantial advantages.138 Microfluidics is central to lab-on-a-chip approaches and allows compartmentalised picolitre to nanolitre volumes of liquid to be manipulated for experiments and assays equivalent to those undertaken in microtitre well plates.139–141 Such small volumes mean reduced reagent cost and consumption, as well as decreased labour requirements. Emerging technologies allow several analytical processes to be incorporated within the same chip-based device, including reagent mixing, reaction termination and analysis.142 Many analytical techniques involve preparation steps that result in sample losses, and whilst working with ultra-small volumes can be challenging for the human analyst, microfluidics lab-on-a-chip approaches can bridge these gaps through full automation of sample processing and analysis without losses.143

The most common microfluidic approaches can be categorised into two subdivisions, digital microfluidics and droplet microfluidics. Digital microfluidics manipulates a liquid into sub-microliter volumes using an array of electrodes which alter the interfacial tension between the liquid and the surface;141,144 whereas droplet microfluidics utilises the flow of two immiscible liquid phases and pre-formulated channels to form droplets of the required dimensions with high mono-dispersity.145,146 This review will largely consider the applications and technologies of the latter approach due to its higher throughput capabilities,23 and more specifically it will focus on the coupling of droplet microfluidics with electrospray ionisation mass spectrometry.

6.2. Microfluidics coupled with electrospray ionisation (ESI) mass spectrometry

The coupling of MS to microfluidics has attracted increasing attention in recent years, due to its ability to offer a sensitive, label-free detection technique for both on and off chip analysis.146,147 ESI and MALDI have been the most commonly exploited ionisation techniques, and have been utilised in many applications such as protein identification,148 metabolomics-based studies of E. coli149 and in situ reaction monitoring.150 ESI and closely related coupling techniques are discussed below, with MALDI-based microfluidic techniques discussed elsewhere in the literature.147,150,151

The most common approach to on-line microfluidics to mass spectrometry coupling is ESI, in particular via nanoflow (nESI), due to its ability to generate a stable spray of the dissolved analyte into the mass spectrometer inlet with flow rates in the range of microliters per hour.152 Other advantages of nESI over conventional electrospray configurations include increased sensitivity, higher ionisation efficiency and the application of lower electrospray voltages at the emitter tip due to the decreased tip aperture.153 To couple microfluidic chips to a mass spectrometer, different approaches have been used including spraying directly from the edge of a chip,152,154 the addition of an electrospray tip to the channel outlet post chip fabrication (external emitter),155 and the integration of a sprayer tip in the initial chip design and fabrication stages (integrated emitter).144,153,156 Successful chip–MS coupling utilising an integrated emitter and borosilicate microfluidic chip has been demonstrated by Belder et al.156 with several applications, including on-chip organocatalysis,157 and micro-free-flow electrophoresis separation.158 Although fabrication of chips assembled from glass often requires intricate micro-milling or HF wet etching,159 the integration of emitters within these chips avoids many of the disadvantages associated with external emitters, especially in their assembly.156 As a number of other materials become amenable to integrated emitter fabrication,160,161 this may become the favoured method of chip based MS coupling to ensure reliable use. However, one such example yielding promise for high throughput applications, in which an external emitter (in the form of an inserted gold coated nESI capillary) has been utilised for droplet MS analysis, has been described by Smith et al.162 This allowed for highly sensitive (sub-femtomole) label-free identification of individual droplets filled with proteins ranging from 12 kDa to 148 kDa in molecular weight. In these experiments, a high voltage in the range of 2.5–3.6 kV was applied to the gold nESI capillary via a copper wire, with a MS scan rate optimised to the flow rate of droplets travelling towards the emitter. Single scans allowed each protein to be identified by accurate mass analysis, although it was reported that there were small amounts of surfactant and protein cross-contamination in some scans, possibly due to incomplete spraying of the previous droplet as the next droplet reaches the emitter. There is said to be no indication that droplet fusion or protein cross-talk occurs during the droplet mixing or re-injection stage, and this re-injection occurs at a rate of up to 2.6 droplets per second.162 This setup has the potential to be improved further through chip design, and will likely become applicable for whole cell analysis, as the droplets would have the ability to be utilised as discrete bioreactors (Fig. 9).


image file: c8an01448e-f9.tif
Fig. 9 Integration of a microfluidic chip with ESI-MS for the analysis of ∼500 pL droplet filled with proteins. (i) Droplets of the sample of interest are generated; (ii) these droplets are collected and then stored overnight; (iii) droplets are re-injected into the mass spectrometer chip with the aid of a spacing oil. Droplets flow through the chip towards a gold coated electrospray emitter, at the end of which a voltage is applied to generate an electrospray of the droplet and the surrounding spacing oil. Adapted with permission from Smith et al., Anal. Chem., 2013, 85, 3812–3816.162

In the external ESI coupling approach of Smith et al. described above,162 the continuous oil phase (along with the pico-droplet) enters the mass spectrometer inlet freely, as the oil does not substantially interfere with the mass spectrum obtained for each protein. Perfluorous components (both oil and surfactant) available for droplet microfluidic flows are now specifically designed to minimise positive mode ESI interferences. However, in instances where interference does become a problem, it may be necessary to transfer or extract the droplet or encapsulated components into an alternative continuous phase.143,163 Additionally, since publication in 2013, a similar droplet–MS coupling approach of Belder and co-workers utilised Teflon tubing inserted below the electrospray emitter as an ‘oil drain’, less so due to oil interference, but instead to ensure a stable electrospray process.164–166 Although this oil drain fulfils its purpose to reduce the interference, it introduces some added complexity during fabrication. Another approach to oil removal has been demonstrated by Kelly et al., where a purposely designed chip allowed for contact between adjacent oil and aqueous streams through cylindrical baffles spaced approximately 3 μm apart. Differences in interfacial tension between the two adjacent liquids prevented bulk mixing; however, as the aqueous droplet or plug (generated prior to this junction) enters the baffle region, rapid coalescence of the droplet into the aqueous stream occurs through the gaps between the baffles. The droplet then travels with the aqueous stream to the ESI emitter, and the oil stream is diverted to waste.143 Differences in the length of the carrier channel from the coalescence junction to the MS inlet were also investigated to identify the extent of droplet dilution prior to the MS analysis, as minimal analyte dilution, particularly in sample-limited analysis, is desired. Shorter distances resulted in superior analyte MS signal with less dilution and band broadening, whereas larger distances were found to be better suited to analyte separations such as those that transpire in capillary electrophoresis.143

With its low sample volumes allowing for miniaturised assays, microfluidics does indeed seem to be a viable approach for future generations of HTS, particularly in the realm of single cell analysis (see later sections). A number of downfalls, however, can be identified – namely in device manufacture and testing. Often, multiple iterations of device designs and fabrication approaches are required before a final (or even usable) chip is acquired. Fabrication requires specialist equipment, training and materials, and often for the smallest of design features (sub-100 μm) a low dust environment such as a class 1000 or better laboratory is recommended, something that increases the cost of undertaking projects such as these exponentially. Repeated usage of devices can also cause difficulties as once a microfluidic channel has become blocked the device is often rendered unusable, requiring repeated fabrication of the same design. Often the microfluidics industry and research groups do not explicitly showcase these failures within their published work, often leading others to believe that chip–MS coupling is easily achievable with their current mass spectrometer. Hence, such an undertaking should be approached with caution, as often a number of analysts will be required to work on such a project at any one given time if results are required promptly.

7. Towards single cell mass spectrometry analysis in omics

The analysis of heterogeneity between cells of the same phenotype can allow for a greater understanding of biological systems, in particular, the study of intracellular processes, and how these infer differentiation of healthy and disease states and cause certain remedial resistances in both mammalian and bacterial cells.167–170 One such advantage of studying these processes is to facilitate earlier diagnosis of disease and to aid therapeutic intervention. Current and established techniques in single cell analysis are focussed on microscopic and cytometric-based methods coupled with fluorescence;167,171 however, these approaches limit the applicability of the cell types and molecules to be analysed. If the sample of interest is not inherently fluorescent, then the analyst must incorporate a fluorescent marker either chemically or genetically.172 Other competing detection technologies for single cell analysis include FTIR (Fourier-transform infrared spectroscopy),173 Raman spectroscopy,174,175 and mass spectrometry,176–178 each with its advantages and disadvantages. Mass spectrometry single cell analysis is often coupled with other techniques and to date most approaches are not reliable or sensitive enough for high throughput application.

As mentioned previously, secondary ion mass spectrometry (SIMS) can map the spatial distribution of small molecular biomarkers in cells,179–182 for example, metabolites183 and lipids,184 but SIMS remains time consuming, and the instrumentation currently available is not able to perform high throughput analysis.

7.1. Proteomics from a few to single cells

One such example of combining a number of differing techniques for single cell mass spectrometry analysis was presented by Mellors et al. in 2010, in which an automated microfluidic device was integrated with capillary electrophoresis and ESI-MS.185 Their microchip was able to incorporate a number of operations including cell delivery, cell lysis, separation of lysed products via capillary electrophoresis and ionisation for ESI-MS analysis, all with a throughput of 5 s per single cell. Using fluorescence imagery, single cells were visualised passing through the microchip to the mass spectrometer and were found to have random spacing between cells. This was also visualised in the corresponding total ion chromatogram in which it was believed that the majority of peaks were due to single cells passing through the microchip and the mass spectra for the hemoglobin α and β units within these cells could be obtained for that event.185 This work provides great promise for the coupling of microfluidics with MS for high throughput single cell analysis of overexpressed proteins and metabolites within cells. Refinement is needed, however, as only one hour of good performance was attained before build-up of lysed cellular components occurred, and flushing steps were required to restore the performance.185 As the manufacture of these microfluidic devices becomes more routine (something that will require additional development investment from industry and research groups), it is likely that such issues will become less of an occurrence.

Although not explicitly a single cell analysis, an approach termed nanoPOTS (nanodroplet processing in one pot for trace samples) detailed by Zhu et al. shows great potential wherein sufficient data are obtained for whole proteome analysis to be performed on very few cells (∼10 to ∼140 cells).186 This chip-based system coupled with a robotic platform capable of dispensing picolitre volumes allows for proteomic sample preparation, including reduction, alkylation, and digestion, before peptide collection within a capillary, which can be stored prior to analysis by ultra-sensitive LC-MS. This chip-based approach also facilitates cell counting via microscopic imaging, and the miniaturised volumes (<200 nL) and surface area seen by the sample reduces protein and peptide losses during preparation. Using this method the authors were able to identify ∼1500 to ∼3000 proteins within ∼10–140 cells respectively, a coverage that has only previously been achieved from cultures containing several thousands of cells.186 The liquid handling system used here has been estimated to dispense 350 droplets in less than 30 minutes, offering the possibility for scaling up towards high throughput analysis although this would necessitate a redesign of the chip to accommodate a larger number of droplets. This workflow does, however, include a number of incubation steps (1 of which is overnight) to achieve efficient digestion for bottom-up proteomic analysis, and although these do limit the throughput prior to analysis, this research may pave the way for similar chip systems that could be coupled with top-down proteomics, to reduce sample pre-processing and increase sample throughput (Fig. 10).


image file: c8an01448e-f10.tif
Fig. 10 (A) Expansion of the components of the nanoPOTS chip; (B) relative size of the nanoPOTS chip in comparison with a US quarter dollar; (C) schematic diagram detailing the procedure employed during nanoPOTS sample preparation including analysis of the number of cells present within the well, reduction of proteins, alkylation of free cysteine residues, 2 protein digestion steps, surfactant cleavage and collection of the resulting peptides within a capillary. Reproduced with permission from Zhu et al., Nat. Commun., 2018, 9, 1–10.186

7.2. Metabolomics from a few to single cells

A relatively new approach in single cell capture is the use of fluid force microscopy coupled with MALDI-MS for the metabolomics analysis of single cells. This method has recently been published by Guillaume-Gentil et al., where HeLa cells are selected underneath a microscope, and the fluid force microscopy probe is positioned above the cell to allow for in situ collection of the cytoplasm (1–3 pL). The probe was then retracted and the extract deposited on to the MALDI target plate and MALDI matrix applied.187 This again utilises the MALDI microarray format employed within many single cell analyses and allowed for the retrieval of a number of metabolite types with minimal perturbation of the cell within its cultured environment. The cells remained viable after collection allowing for other analysis types to be subsequently performed, and it is also noted that this approach could allow time course experiments upon the same single cell to be carried out.187 However, the extraction and deposition times of each sample take ∼3 and ∼4 minutes respectively, and there are a number of manual steps such as finding the cells under the microscope and depositing the MALDI extract which all add to the total analysis time per single cell. Again some refinement is needed to increase throughput before this promising approach could become routine for single cell analysis (Fig. 11).
image file: c8an01448e-f11.tif
Fig. 11 Fluid force microscopy coupled with MALDI-MS workflow for single cell metabolomics analysis. (A) The fluid force microscopy probe is positioned above a single cell and the cytoplasm extracted; (B) the extracted cytoplasm is deposited onto a MALDI plate; (C) the MALDI plate is coated in a MALDI matric solution; (D) MALDI-MS analysis of deposited samples. Reproduced with permission from Gillaume-Gentil et al., Anal. Chem., 2017, 89, 5017–5023.187

Although not strictly metabolomics, cellular bioavailability of elements has been probed through the use of ICP-MS in a high throughput and quantitative manner by Meyer et al.188 with a number of other ICP-MS single cell applications reported by Mueller et al.189 Single mammalian cells exposed to arsenite by Meyer et al. were analysed to determine arsenic uptake over time, with both sulphur and phosphorus content examined as well. This study utilised a commercial ICP-MS for analysis and typical cultivation procedures allowing for simple yet effective single cell analysis, with 330 cell events detected within a 90 s period (estimated throughput of 1 cell every 3.67 s). The sensitivity of this method was also explored, with the limit of detection for arsenic stated to be 0.35 fg per cell.188 These advantages aside, as ICP coupled techniques are restricted to elemental analysis alone, their applicability to molecular single cell studies will be limited.

7.3. Lipidomics from a few to single cells

A microfluidics coupled MS approach reported by Xie et al. has applied MALDI imaging to the analysis of phospholipids from single cells. Using a micro-well format, cells were efficiently captured within a well-organised grid and MALDI matrix deposited upon the cell positions. This micro-well based device was fabricated using standard photolithography and soft lithography techniques, and cell capture was optimised by altering both the well diameters and cell suspension concentrations, resulting in >90% of the wells being filled with 1–4 lung cancer cells. The resultant capture efficiency, defined as wells from which phospholipids could be identified via MALDI-MS, was ∼25%, although it is noted that the single cell phospholipid concentration was not sufficient for tandem MS analysis and hence lipid extracts from a denser cell population were utilised to perform more in depth structural identification.190 Ultimately, this suggests that for MALDI, and possibly other single cell MS approaches, sensitivity is a limiting factor that will need to be addressed before the field can progress in other aspects such as throughput. MALDI analysis time for the microarray is not explicitly stated within the work; however, the high density of cells able to adhere to the device (10[thin space (1/6-em)]000 wells per cm2) is a step forward for uncomplicated single cell isolation techniques. In a number of other publications single cell analysis has been performed using MALDI-MS although not all are high-throughput.191–194

An alternative method for generating single cell microarrays has been described by Ellis et al., in which inject printing is used to obtain high spatial control of cells to assist in the subsequent analysis of lipids by LESA-MS.195 Inject technology (or bioprinting) allows cell suspensions to be deposited onto glass slides in a 10 by 10 grid, resulting in single cell deposition according to a Poisson distribution. In this case approximately 37% of the wells contained single cells. The spots were dried and microscopy image analysis was used to determine positions upon the slide which correctly contained a single cell. LESA-MS sampled these positions in a similar manner to the procedure depicted in Fig. 3. Using this technology the authors acquired 8.3 minutes of MS (and MS/MS) data per spot containing 1–5 cells, which is at the lower end of the desired throughput scale; however, it is noted that this approach yields good reproducibility even at the single cell level as well as sufficient sensitivity for a number of different cell types investigated. Comparisons of spots containing up to 100 cells with those of single occupancy found a number of lipids to be present in both positions with similar abundances, and the resulting lipid fingerprints allowed for cell identification between different cell types by multivariate analysis. It was noted that there was little heterogeneity between the single cells analysed, which may have been a consequence of the controlled nature in which the cells were cultured.195 The use of microarrays allows for accountability of a sample position within a grid similar to that of microtitre well plates, a system that is favoured by many analysts.

Nanomanipulation of cells is another technique that has been successfully coupled with differing types of MS for the analysis of lipids by Verbeck and co-authors.196–200 This has been achieved through the use of a nanomanipulator workstation (DCG Systems Inc., Fremont, CA, USA) which utilises either quartz rods, micropipettes or nESI capillary emitters197 to extract single cells or even organelles from a culture dish or tissue. This procedure has been termed ‘One-Cell’ analysis, and as the cell or organelle of interest has been located under a microscope prior to analysis, its spatial information relative to the tissue (which in some applications is carcinogenic) is retained.197 Additionally, nearby cells and structures also remain viable for future analysis due to the ambient nature of the technique. In regard to MS coupling, the two ionisation techniques here are nESI and MALDI. For nESI the material of interest is collected directly into a metal coated emitter tip and transferred to the MS source for direct ionisation and analysis of the extracted components. This has been achieved for lipidomic determinations within carcinogenic and normal breast tissues, in which lipid droplets were extracted from the inside of single adipocytes, and the triglyceride and fatty acid contents probed by MS and MS/MS experiments.197 Coupling nanomanipulation with MALDI-MS for adipocyte analysis follows a similar approach; however, instead of the extraction emitter transferring directly to nESI, the emitter is positioned on a glass slide and the extracted components ejected. Matrix deposition on the ejected components then takes place using a second emitter system. In the presence of an appropriate matrix, triacylglyceride species were determined via MALDI ionisation and accurate mass measurements.200 In regard to throughput, this extraction procedure is lengthy (<30 min) in comparison with some of the other techniques discussed in this review; however, for single cell lipidomic analysis this time frame is relatively short in comparison with traditional chromatographic separations and purification steps,200 and does include the additional spatial information of cellular location within a tissue section. Coupling to MALDI within an array and other hardware improvements may have the potential to advance this technique in the future.

8. Conclusions and outlook

To conclude, recent developments in the coupling of automated and chip based sample inlet systems to mass spectrometers are enabling the analysis of >10[thin space (1/6-em)]000 samples in a 24 hour time period at increased sensitivity. Ambient ionisation, SPE methods and MSI techniques have facilitated the analysis of material from untreated biofluids, tissues, organs and intact live cells, substantially lowering the time required to prepare samples prior to the acquisition of useful MS data. Whilst many of the techniques described herein are commercially available, still more are emerging from the adaptation of technologies by academic groups often in close collaboration with industry. Challenges surrounding contamination, ion suppression, carryover and device fouling due to excessive cellular components can limit the sensitivity of these techniques, although ion mobility has gone some way to alleviate the first two of these. Future microfluidics developments will address the latter two. Many HTS-MS approaches are aimed towards targeted analysis where yes/no answers are required and ultra-high sensitivity is not the ultimate aim, but it must be noted that many methods described such as ‘nanoPOTS’ and the ‘Echo–MS’ have indicated impressive sensitivity along with substantially reduced amounts of starting material. This increased sensitivity at higher throughput from decreasing amounts of sample (truly ‘less is more’) indicates an increased efficiency of ESI when there is less to analyse, and is also attributed to many developments in the transmission and detection of ions in modern mass spectrometers.

Conflicts of interest

Clive Smith is employed by Sphere Fluidics who are a company that develops microfluidics that enables droplet delivery for cell screening applications.

Acknowledgements

This review was supported by the BBSRC through the award of a DTP studentship to EK (BB/M011208/1) and by the funding from our SYNBIOCHEM centre (BB/M017702/1), and we acknowledge the entire SYNBIOCHEM team for their on-going contributions in support of our work in microfluidics and intact cell mass spectrometry. Sphere Fluidics Ltd have also supported this work in the form of a BBSRC CASE award to EK.

References

  1. J. Inglese, R. L. Johnson, A. Simeonov, M. Xia, W. Zheng, C. P. Austin and D. S. Auld, Nat. Chem. Biol., 2007, 3, 466–479 CrossRef CAS PubMed.
  2. R. P. Hertzberg and A. J. Pope, Curr. Opin. Chem. Biol., 2000, 4, 445–451 CrossRef CAS PubMed.
  3. M. de Raad, C. R. Fischer and T. R. Northen, Curr. Opin. Chem. Biol., 2016, 30, 7–13 CrossRef CAS PubMed.
  4. J. W. Armstrong, Am. Biotechnol. Lab., 1999, 17, 26–28 CAS.
  5. S. A. Sundberg, Curr. Opin. Biotechnol., 2000, 11, 47–53 CrossRef CAS PubMed.
  6. P. Szymański, M. Markowicz and E. Mikiciuk-Olasik, Int. J. Mol. Sci., 2012, 13, 427–452 CrossRef PubMed.
  7. A. A. Hajare, S. S. Salunkhe, S. S. Mali, S. S. Gorde, S. J. Nadaf and S. A. Pishawikar, Am. J. PharmTech Res., 2014, 4, 112–129 Search PubMed.
  8. J. P. Hughes, S. S. Rees, S. B. Kalindjian and K. L. Philpott, Br. J. Pharmacol., 2011, 162, 1239–1249 CrossRef CAS PubMed.
  9. D. E. Cameron, C. J. Bashor and J. J. Collins, Nat. Rev. Microbiol., 2014, 12, 381–390 CrossRef CAS PubMed.
  10. P. C. Gach, K. Iwai, P. Kim, N. J. Hillson and A. K. Singh, Lab Chip, 2017, 17, 3388–3400 RSC.
  11. D. F. Savage, J. Way and P. A. Silver, ACS Chem. Biol., 2008, 3, 13–16 CrossRef CAS PubMed.
  12. A. S. Khalil and J. J. Collins, Nat. Rev. Genet., 2010, 11, 367–379 CrossRef CAS PubMed.
  13. J. J. Agresti, E. Antipov, A. R. Abate, K. Ahn, A. C. Rowat, J.-C. Baret, M. Marquez, A. M. Klibanov, A. D. Griffiths and D. A. Weitz, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 6550–6550 CrossRef CAS PubMed.
  14. M. T. Reetz, D. Kahakeaw and R. Lohmer, ChemBioChem, 2008, 9, 1797–1804 CrossRef CAS PubMed.
  15. F. H. Arnold, Acc. Chem. Res., 1998, 31, 125–131 CrossRef CAS.
  16. P. Jacques, M. Béchet, M. Bigan, D. Caly, G. Chataigné, F. Coutte, C. Flahaut, E. Heuson, V. Leclère, D. Lecouturier, V. Phalip, R. Ravallec, P. Dhulster and R. Froidevaux, Bioprocess Biosyst. Eng., 2017, 40, 161–180 CrossRef CAS PubMed.
  17. S. Bershtein and D. S. Tawfik, Curr. Opin. Chem. Biol., 2008, 12, 151–158 CrossRef CAS PubMed.
  18. Y.-C. Tung, A. Y. Hsiao, S. G. Allen, Y. Torisawa, M. Ho and S. Takayama, Analyst, 2011, 136, 473–478 RSC.
  19. E. E. Swartzman, S. J. Miraglia, J. Mellentin-Michelotti, L. Evangelista and P. M. Yuan, Anal. Biochem., 1999, 271, 143–151 CrossRef CAS PubMed.
  20. P. Hodder, R. Mull, J. Cassaday, K. Berry and B. Strulovici, J. Biomol. Screening, 2004, 9, 417–426 CrossRef CAS PubMed.
  21. M. Yadav, P. Contractor, V. Upadhyay, A. Gupta, S. Guttikar, P. Singhal, S. Goswami and P. S. Shrivastav, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2008, 872, 167–171 CrossRef CAS PubMed.
  22. S. X. Peng, T. M. Branch and S. L. King, Anal. Chem., 2001, 73, 708–714 CrossRef CAS PubMed.
  23. D. Hümmer, F. Kurth, N. Naredi-Rainer and P. S. Dittrich, Lab Chip, 2015, 16, 447–458 RSC.
  24. O. V. Trubetskoy, J. R. Gibson and B. D. Marks, J. Biomol. Screening, 2005, 10, 56–66 CrossRef CAS PubMed.
  25. M. Smilkstein, N. Sriwilaijaroen, J. X. Kelly, P. Wilairat and M. Riscoe, Antimicrob. Agents Chemother., 2004, 48, 1803–1806 CrossRef CAS PubMed.
  26. M. Wodnicka, R. D. Guarino, J. J. Hemperly, M. R. Timmins, D. Stitt and J. B. Pitner, J. Biomol. Screening, 2000, 5, 141–152 CrossRef CAS PubMed.
  27. S. M. Rodems, B. D. Hamman, C. Lin, J. Zhao, S. Shah, D. Heidary, L. Makings, J. H. Stack and B. A. Pollok, Assay Drug Dev. Technol., 2002, 1, 9–19 CrossRef CAS PubMed.
  28. Z. Haarhoff, A. Wagner, P. Picard, D. M. Drexler, T. Zvyaga and W. Shou, J. Biomol. Screening, 2016, 21, 165–175 CrossRef CAS PubMed.
  29. L. Wu and X. Qu, Chem. Soc. Rev., 2015, 44, 2963–2997 RSC.
  30. M. R. Wenk, Nat. Rev. Drug Discovery, 2005, 4, 594–610 CrossRef CAS PubMed.
  31. R. Aebersold and M. Mann, Nature, 2003, 422, 198–207 CrossRef CAS PubMed.
  32. K. Dettmer, P. A. Aronov and B. D. Hammock, Mass Spectrom. Rev., 2007, 26, 51–78 CrossRef CAS PubMed.
  33. M. T. Henke and N. L. Kelleher, Nat. Prod. Rep., 2016, 33, 942–950 RSC.
  34. E. de Hoffmann and V. Stroobant, Mass Spectrometry: Principles and Applications, John Wiley & Sons, Chichester, 3rd edn, 2007 Search PubMed.
  35. C. K. Van Pelt, S. Zhang, E. Fung, I. Chu, T. Liu, C. Li, W. A. Korfmacher and J. Henion, Rapid Commun. Mass Spectrom., 2003, 17, 1573–1578 CrossRef CAS PubMed.
  36. C. Flangea, C. Mosoarca, C. Cozma, M. Galusca, M. Przybylski and A. D. Zamfir, Electrophoresis, 2013, 34, 1572–1580 CrossRef CAS PubMed.
  37. N. J. Martin, J. Bunch and H. J. Cooper, J. Am. Soc. Mass Spectrom., 2013, 24, 1242–1249 CrossRef CAS PubMed.
  38. R. Almeida, C. Mosoarca, M. Chirita, V. Udrescu, N. Dinca, Ž. Vukelić, M. Allen and A. D. Zamfir, Anal. Biochem., 2008, 378, 43–52 CrossRef CAS PubMed.
  39. K. Geddes, G. Adamson, N. Dube, S. Crathern and R. C. King, Rapid Commun. Mass Spectrom., 2009, 23, 1303–1312 CrossRef CAS PubMed.
  40. C. Flangea, C. Schiopu, F. Capitan, C. Mosoarca, M. Manea, E. Sisu and A. D. Zamfir, Cent. Eur. J. Chem., 2012, 11, 25–34 CrossRef.
  41. A. D. Zamfir, N. Lion, Z. Vukelic, L. Bindila, J. Rossier, H. H. Girault and J. Peter-Katalinic, Lab Chip, 2005, 5, 298–307 RSC.
  42. H. M. Park, J. H. Shin, J. K. Kim, S. J. Lee, G. S. Hwang, K. H. Liu and C. H. Lee, Metabolomics, 2014, 10, 663–676 CrossRef CAS.
  43. A. Serb, C. Schiopu, C. Flangea, Z. Vukelić, E. Sisu, L. Zagrean and A. D. Zamfir, Eur. J. Mass Spectrom., 2009, 15, 541–553 CrossRef CAS PubMed.
  44. C. Y. Chen, B. L. Lam and S. K. Bhattacharya, Mol. Vision, 2014, 20, 1605–1611 CrossRef.
  45. H. Ye, J. Hill, A. C. Gucinski, M. T. Boyne and L. F. Buhse, AAPS J., 2015, 17, 405–415 CrossRef CAS PubMed.
  46. A. D. Zamfir, A. Serb, Ž. Vukelić, C. Flangea, C. Schiopu, D. Fabris, S. Kalanj-Bognar, F. Capitan and E. Sisu, J. Am. Soc. Mass Spectrom., 2011, 22, 2145–2159 CrossRef CAS PubMed.
  47. A. F. Serb, E. Sisu, Ž. Vukelić and A. D. Zamfir, J. Mass Spectrom., 2012, 47, 1561–1570 CrossRef CAS PubMed.
  48. T. K. Toby, L. Fornelli and N. L. Kelleher, Annu. Rev. Anal. Chem., 2016, 9, 499–519 CrossRef CAS PubMed.
  49. M. Bailey, E. C. Randall, C. Costa, T. Salter, A. Race, M. de Puit, M. Koeberg, M. Baumert and J. Bunch, Anal. Methods, 2016, 8, 3373–3382 RSC.
  50. E. C. Randall, J. Bunch and H. J. Cooper, Anal. Chem., 2014, 86, 10504–10510 CrossRef CAS PubMed.
  51. V. Kertesz and G. J. Van Berkel, J. Mass Spectrom., 2010, 45, 252–260 CrossRef CAS PubMed.
  52. R. L. Griffiths, A. Dexter, A. J. Creese and H. J. Cooper, Analyst, 2015, 140, 6879–6885 RSC.
  53. M. Kai, I. González, O. Genilloud, S. B. Singh and A. Svatoš, Rapid Commun. Mass Spectrom., 2012, 26, 2477–2482 CrossRef CAS PubMed.
  54. D. Eikel and J. Henion, Rapid Commun. Mass Spectrom., 2011, 25, 2345–2354 CrossRef CAS PubMed.
  55. W. B. Parson, S. L. Koeniger, R. W. Johnson, J. Erickson, Y. Tian, C. Stedman, A. Schwartz, E. Tarcsa, R. Cole and G. J. Van Berkel, J. Mass Spectrom., 2012, 47, 1420–1428 CrossRef CAS PubMed.
  56. M. Wisztorski, A. Desmons, J. Quanico, B. Fatou, J. P. Gimeno, J. Franck, M. Salzet and I. Fournier, Proteomics, 2016, 16, 1622–1632 CrossRef CAS PubMed.
  57. C. Yan, F. Parmeggiani, E. A. Jones, E. Claude, S. A. Hussain, N. J. Turner, S. L. Flitsch and P. E. Barran, J. Am. Chem. Soc., 2017, 139, 1408–1411 CrossRef CAS PubMed.
  58. P. Carbonell, A. J. Jervis, C. J. Robinson, C. Yan, M. Dunstan, N. Swainston, M. Vinaixa, K. A. Hollywood, A. Currin, N. J. W. Rattray, S. Taylor, R. Spiess, R. Sung, A. R. Williams, D. Fellows, N. J. Stanford, P. Mulherin, R. Le Feuvre, P. Barran, R. Goodacre, N. J. Turner, C. Goble, G. G. Chen, D. B. Kell, J. Micklefield, R. Breitling, E. Takano, J.-L. Faulon and N. S. Scrutton, Commun. Biol., 2018, 1, 66 CrossRef PubMed.
  59. C. F. Poole, A. D. Gunatilleka and R. Sethuraman, J. Chromatogr., A, 2000, 885, 17–39 CrossRef CAS PubMed.
  60. M. J. Telepchak, Forensic Sci. Med., 2004, 122, 41–53 Search PubMed.
  61. M.-C. Hennio, J. Chromatogr., A, 1999, 856, 3–54 CrossRef.
  62. N. J. K. Simpson and M. J. M. Wells, in Solid-Phase Extraction: Principles, Techniques, and Applications, Marcel Dekker Inc., New York, 2000, pp. 1–2 Search PubMed.
  63. A. Becalski, B. P. Y. Lau, D. Lewis, S. W. Seaman, S. Hayward, M. Sahagian, M. Ramesh and Y. Leclerc, J. Agric. Food Chem., 2004, 52, 3801–3806 CrossRef CAS PubMed.
  64. A. Molinelli, R. Weiss and B. Mizaikoff, J. Agric. Food Chem., 2002, 50, 1804–1808 CrossRef CAS PubMed.
  65. D. Štajnbaher and L. Zupančič-Kralj, J. Chromatogr., A, 2003, 1015, 185–198 CrossRef.
  66. M. J. M. Wells and J. L. Michael, Anal. Chem., 1987, 59, 1739–1742 CrossRef CAS PubMed.
  67. R. J. Lewis, R. D. Johnson, M. K. Angier and R. M. Ritter, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2004, 806, 141–150 CrossRef CAS PubMed.
  68. E. Kim, J. Lee, S. Choi, M. Lim and H. Chung, Forensic Sci. Int., 2008, 174, 197–202 CrossRef CAS PubMed.
  69. R. J. Lewis, R. D. Johnson and R. A. Hattrup, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2005, 822, 137–145 CrossRef CAS PubMed.
  70. H. Choi, S. Baeck, M. Jang, S. Lee, H. Choi and H. Chung, Forensic Sci. Int., 2012, 215, 81–87 CrossRef CAS PubMed.
  71. Z. Kuklenyik, J. A. Reich, J. S. Tully, L. L. Needham and A. M. Calafat, Environ. Sci. Technol., 2004, 38, 3698–3704 CrossRef CAS PubMed.
  72. F. Lachâtre, P. Marquet, S. Ragot, J. M. Gaulier, P. Cardot and J. L. Dupuy, J. Chromatogr. B: Biomed. Sci. Appl., 2000, 738, 281–291 CrossRef.
  73. P. M. Kabra, J. H. Wall and P. Dimson, Clin. Chem., 1987, 33, 2272–2274 CAS.
  74. D. T. Rossi and N. Zhang, J. Chromatogr., A, 2000, 885, 97–113 CrossRef CAS.
  75. P. J. Jannetto and L. J. Langman, Clin. Biochem., 2016, 49, 1032–1034 CrossRef CAS PubMed.
  76. X. Wu, J. Wang, L. Tan, J. Bui, E. Gjerstad, K. McMillan and W. Zhang, J. Biomol. Screening, 2012, 17, 761–772 CrossRef CAS PubMed.
  77. G. C. Adam, J. Meng, J. M. Rizzo, A. Amoss, J. W. Lusen, A. Patel, D. Riley, R. Hunt, P. Zuck, E. N. Johnson, V. N. Uebele and J. D. Hermes, J. Biomol. Screening, 2014, 20, 212–222 CrossRef PubMed.
  78. C. Haslam, J. Hellicar, A. Dunn, A. Fuetterer, N. Hardy, P. Marshall, R. Paape, M. Pemberton, A. Resemannand and M. Leveridge, J. Biomol. Screening, 2016, 21, 176–186 CrossRef CAS PubMed.
  79. I. Vrobel, H. Janeckova, E. Faber, K. Bouchalova, K. Micova, D. Friedecky and T. Adam, Ther. Drug Monit., 2016, 38, 516–524 CrossRef CAS PubMed.
  80. H. Lu, L. Kopcho, K. Ghosh, M. Witmer, M. Parker, S. Gupta, M. Paul, P. Krishnamurthy, B. Laksmaiah, D. Xie, J. Tredup, L. Zhang and L. M. Abell, Anal. Biochem., 2016, 501, 56–65 CrossRef CAS PubMed.
  81. D. Grote-Koska, S. Czajkowski and K. Brand, Ther. Drug Monit., 2015, 37, 400–404 CrossRef CAS PubMed.
  82. M. Takamiya, M. Sakurai, F. Teranishi, T. Ikeda, T. Kamiyama and A. Asai, Biochem. Biophys. Res. Commun., 2016, 480, 721–726 CrossRef CAS PubMed.
  83. L. J. Gordon, M. Allen, P. Artursson, M. M. Hann, B. J. Leavens, A. Mateus, S. Readshaw, K. Valko, G. J. Wayne and A. West, J. Biomol. Screening, 2016, 21, 156–164 CrossRef CAS PubMed.
  84. A. H. Luippold, T. Arnhold, W. Jörg and R. D. Süssmuth, Int. J. Mass Spectrom., 2010, 296, 1–9 CrossRef CAS.
  85. J. R. Neifeld, L. E. Regester, J. M. Holler, S. P. Vorce, J. Magluilo, G. Ramos and T. Z. Bosy, J. Anal. Toxicol., 2016, 40, 379–387 CrossRef CAS PubMed.
  86. Agilent Technologies, Agilent G9530A RapidFire 365 High-throughput Mass Spectrometry System User Guide, 2014 Search PubMed.
  87. M. Razavi, L. E. Frick, W. A. Lamarr, M. E. Pope, C. A. Miller, L. N. Anderson and T. W. Pearson, J. Proteome Res., 2012, 11, 5642–5649 CrossRef CAS PubMed.
  88. A. Amaral, C. Saran, J. Amin and P. Hatsis, J. Biomol. Screening, 2016, 21, 620–625 CrossRef CAS PubMed.
  89. D. Danso, P. J. Jannetto, R. Enger and L. J. Langman, Ther. Drug Monit., 2015, 37, 319–324 CrossRef CAS PubMed.
  90. M. Leveridge, R. Buxton, A. Argyrou, P. Francis, B. Leavens, A. West, M. Rees, P. Hardwicke, A. Bridges, S. Ratcliffe and C. W. Chung, J. Biomol. Screening, 2014, 19, 278–286 CrossRef PubMed.
  91. M. K. Highkin, M. P. Yates, O. V. Nemirovskiy, W. A. Lamarr, G. E. Munie, J. W. Rains, J. L. Masferrer and M. M. Nagiec, J. Biomol. Screening, 2011, 16, 272–277 CrossRef CAS PubMed.
  92. B. T. Veach, T. K. Mudalige and P. Rye, Anal. Chem., 2017, 89, 3256–3260 CrossRef CAS PubMed.
  93. K. B. Lim, C. C. Ozbal and D. B. Kassel, J. Biomol. Screening, 2010, 15, 447–452 CrossRef CAS PubMed.
  94. C. K. Imamura, K. Furihata, S. Okamoto and Y. Tanigawara, J. Clin. Pharmacol., 2016, 56, 408–413 CrossRef CAS PubMed.
  95. A. Zimmerlin, M. Trunzer and B. Faller, Drug Metab. Dispos., 2011, 39, 1039–1046 CrossRef CAS PubMed.
  96. Q. K. Huynh, S. J. Wise, K. A. Koch, L. A. Castonguay, B. G. Reid, E. E. Pagratis, D. Koditek, C. B. Glascock, K. R. Pitts, B. A. Turner, X. Liu, M. Hung, B. Han and N. Pagratis, J. Biomol. Screening, 2011, 16, 724–733 CrossRef CAS PubMed.
  97. C.-W. Shu, C. Madiraju, D. Zhai, K. Welsh, P. Diaz, E. Sergienko, R. Sano and J. C. Reed, J. Biomol. Screening, 2011, 16, 174–182 CrossRef CAS PubMed.
  98. S. A. I. Seidel, N. A. Markwardt, S. A. Lanzmich and D. Braun, Angew. Chem., Int. Ed., 2014, 53, 7948–7951 CrossRef CAS PubMed.
  99. R. J. Grant, K. Roberts, C. Pointon, C. Hodgson, L. Womersley, D. C. Jones and E. Tang, J. Biomol. Screening, 2009, 14, 452–459 CrossRef CAS PubMed.
  100. M. A. Liebert and R. Ellson, Assay Drug Dev. Technol., 2005, 3, 607–612 CrossRef PubMed.
  101. P. Kanigowska, Y. Shen, Y. Zheng, S. Rosser and Y. Cai, J. Lab. Autom., 2016, 21, 49–56 CrossRef CAS PubMed.
  102. H. Aerni, D. S. Cornett and R. M. Caprioli, Anal. Chem., 2006, 78, 827–834 CrossRef CAS PubMed.
  103. J. Chin, E. Wood, G. S. Peters and D. M. Drexler, J. Lab. Autom., 2016, 21, 204–207 CrossRef PubMed.
  104. I. Sinclair, R. Stearns, S. Pringle, J. Wingfield, S. Datwani, E. Hall, L. Ghislain, L. Majlof and M. Bachman, J. Lab. Autom., 2016, 21, 19–26 CrossRef PubMed.
  105. S. H. Yoon, T. Liang, T. Schneider, B. L. Oyler, C. E. Chandler, R. K. Ernst, G. S. Yen, Y. Huang, E. Nilsson and D. R. Goodlett, Rapid Commun. Mass Spectrom., 2016, 30, 2555–2560 CrossRef CAS PubMed.
  106. A. C. Susa, Z. Xia and E. R. Williams, Anal. Chem., 2017, 89, 3116–3122 CrossRef CAS PubMed.
  107. D. N. Mortensen and E. R. Williams, Anal. Chem., 2014, 86, 9315–9321 CrossRef CAS PubMed.
  108. C. Wu, A. L. Dill and L. S. Eberlin, Mass Spectrom. Rev., 2013, 32, 218–243 CrossRef CAS PubMed.
  109. E. R. Amstalden van Hove, D. F. Smith and R. M. A. Heeren, J. Chromatogr., A, 2010, 1217, 3946–3954 CrossRef CAS PubMed.
  110. Imaging Mass Spectromety, ed. M. Setou, Springer, Japan, Tokyo, 2nd edn, 2010 Search PubMed.
  111. D. I. Campbell, C. R. Ferreira, L. S. Eberlin and R. G. Cooks, Anal. Bioanal. Chem., 2012, 404, 389–398 CrossRef CAS PubMed.
  112. A. D. Feenstra, M. E. Dueñas and Y. J. Lee, J. Am. Soc. Mass Spectrom., 2017, 28, 434–442 CrossRef CAS PubMed.
  113. A. Römpp and B. Spengler, Histochem. Cell Biol., 2013, 139, 759–783 CrossRef PubMed.
  114. D. R. Bhandari, Q. Wang, W. Friedt, B. Spengler, S. Gottwald and A. Römpp, Analyst, 2015, 140, 7696–7709 RSC.
  115. M. Kubicek, G. Holzlechner, A. K. Opitz, S. Larisegger, H. Hutter and J. Fleig, Appl. Surf. Sci., 2014, 289, 407–416 CrossRef CAS PubMed.
  116. F. Hillenkamp and J. Peter-Katalinic, MALDI MS: A Practical Guide to Instrumentation, Methods and Applications, Wiley Blackwell, Weinheim, 2nd edn, 2013 Search PubMed.
  117. N. Singhal, M. Kumar, P. K. Kanaujia and J. S. Virdi, Front. Microbiol., 2015, 6, 1–16 Search PubMed.
  118. R. M. Caprioli, T. B. Farmer and J. Gile, Anal. Chem., 1997, 69, 4751–4760 CrossRef CAS PubMed.
  119. T. C. Rohner, D. Staab and M. Stoeckli, Mech. Ageing Dev., 2005, 126, 177–185 CrossRef CAS PubMed.
  120. L. H. Cazares, D. A. Troyer, B. Wang, R. R. Drake and O. John Semmes, Anal. Bioanal. Chem., 2011, 401, 17–27 CrossRef CAS PubMed.
  121. A. Kernalléguen, R. Steinhoff, S. Bachler, P. S. Dittrich, F. Saint-Marcoux, S. El Bakhi, F. Vorspan, G. Léonetti, D. Lafitte, A. L. Pélissier-Alicot and R. Zenobi, Anal. Chem., 2018, 90, 2302–2309 CrossRef PubMed.
  122. A. Svatoš, Anal. Chem., 2011, 83, 5037–5044 CrossRef PubMed.
  123. H. R. Aerni, D. S. Cornett and R. M. Caprioli, Anal. Chem., 2006, 78, 827–834 CrossRef CAS PubMed.
  124. J. Kononen, L. Bubendorf, A. Kallioniemi, M. Bärlund, P. Schraml, S. Leighton, J. Torhorst, M. J. Mihatsch, G. Sauter and O. P. Kallioniemi, Nat. Med., 1998, 4, 844–847 CrossRef CAS PubMed.
  125. M. R. Groseclose, P. P. Massion, P. Chaurand and R. M. Caprioli, Proteomics, 2008, 8, 3715–3724 CrossRef CAS PubMed.
  126. Z. Takáts, J. M. Wiseman, B. Gologan and R. G. Cooks, Science, 2004, 306, 471–473 CrossRef PubMed.
  127. Z. Takáts, J. M. Wiseman and R. G. Cooks, J. Mass Spectrom., 2005, 40, 1261–1275 CrossRef PubMed.
  128. A. Venter, M. Nefliu and R. Graham Cooks, Trends Anal. Chem., 2008, 27, 284–290 CrossRef CAS.
  129. Z. Takáts, J. M. Wiseman, B. Gologan and R. G. Cooks, Anal. Chem., 2004, 76, 4050–4058 CrossRef PubMed.
  130. C. Wu, A. L. Dill, L. S. Eberlin, R. G. Cooks and D. R. Ifa, Mass Spectrom. Rev., 2013, 32, 218–243 CrossRef CAS PubMed.
  131. J. M. Wiseman, D. R. Ifa, Y. Zhu, C. B. Kissinger, N. E. Manicke, P. T. Kissinger and R. G. Cooks, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 18120–18125 CrossRef PubMed.
  132. J. Tillner, V. Wu, E. A. Jones, S. D. Pringle, T. Karancsi, A. Dannhorn, K. Veselkov, J. S. McKenzie and Z. Takats, J. Am. Soc. Mass Spectrom., 2017, 28, 2090–2098 CrossRef CAS PubMed.
  133. A. Tata, C. J. Perez, M. O. Ore, D. Lostun, A. Passas, S. Morin and D. R. Ifa, RSC Adv., 2015, 5, 75458–75464 RSC.
  134. C. F. F. Angolini, P. H. Vendramini, F. D. S. Araújo, W. L. Araújo, R. Augusti, M. N. Eberlin and L. G. De Oliveira, Anal. Chem., 2015, 87, 6925–6930 CrossRef CAS PubMed.
  135. J. F. Garcia-Reyes, A. U. Jackson, A. Molina-Diaz and R. G. Cooks, Anal. Chem., 2009, 81, 820–829 CrossRef CAS PubMed.
  136. S. Gerbig, G. Stern, H. E. Brunn, R. A. Düring, B. Spengler and S. Schulz, Anal. Bioanal. Chem., 2017, 409, 2107–2117 CrossRef CAS PubMed.
  137. M. Wleklinski, B. P. Loren, C. R. Ferreira, Z. Jaman, L. Avramova, T. J. P. Sobreira, D. H. Thompson and R. G. Cooks, Chem. Sci., 2018, 9, 1647–1653 RSC.
  138. H. A. Stone, A. D. Stroock and A. Ajdari, Annu. Rev. Fluid Mech., 2004, 36, 381–411 CrossRef.
  139. L. Mazutis, J. Gilbert, L. Ung, D. A. Weitz, A. D. Griffiths and J. A. Heyman, Nat. Protoc., 2013, 8, 870–891 CrossRef CAS PubMed.
  140. A. K. Price and B. M. Paegel, Anal. Chem., 2016, 88, 339–353 CrossRef CAS PubMed.
  141. W. L. Chou, P. Y. Lee, C. L. Yang, W. Y. Huang and Y. S. Lin, Micromachines, 2015, 6, 1249–1271 CrossRef.
  142. L. Mazutis, J.-C. Baret, P. Treacy, Y. Skhiri, A. F. Araghi, M. Ryckelynck, V. Taly and A. D. Griffiths, Lab Chip, 2009, 9, 2902–2908 RSC.
  143. R. T. Kelly, J. S. Page, I. Marginean, K. Tang and R. D. Smith, Angew. Chem., Int. Ed., 2009, 48, 6832–6835 CrossRef CAS PubMed.
  144. K. Choi, A. H. C. Ng, R. Fobel and A. R. Wheeler, Annu. Rev. Anal. Chem., 2012, 5, 413–440 CrossRef CAS PubMed.
  145. A. Zinchenko, S. R. A. Devenish, B. Kintses, P. Y. Colin, M. Fischlechner and F. Hollfelder, Anal. Chem., 2014, 86, 2526–2533 CrossRef CAS PubMed.
  146. Y. Zhu and Q. Fang, Anal. Chim. Acta, 2013, 787, 24–35 CrossRef CAS PubMed.
  147. D. Gao, H. Liu, Y. Jiang and J.-M. Lin, Lab Chip, 2013, 13, 3309–3322 RSC.
  148. J. Ji, L. Nie, L. Qiao, Y. Li, L. Guo, B. Liu, P. Yang and H. H. Girault, Lab Chip, 2012, 12, 2625–2629 RSC.
  149. J. Heinemann, B. Noon, M. J. Mohigmi, A. Mazurie, D. L. Dickensheets and B. Bothner, J. Am. Soc. Mass Spectrom., 2014, 25, 1755–1762 CrossRef CAS PubMed.
  150. M. Brivio, R. H. Fokkens, W. Verboom, D. N. Reinhoudt, N. R. Tas, M. Goedbloed and A. Van Den Berg, Anal. Chem., 2002, 74, 3972–3976 CrossRef CAS PubMed.
  151. S. K. Küster, M. Pabst, K. Jefimovs, R. Zenobi and P. S. Dittrich, Anal. Chem., 2014, 86, 4848–4855 CrossRef PubMed.
  152. Q. Xue, F. Foret, Y. M. Dunayevskiy, P. M. Zavracky, N. E. McGruer and B. L. Karger, Anal. Chem., 1997, 69, 426–430 CrossRef CAS PubMed.
  153. K. Sander and E. Verpoorte, Lab Chip, 2007, 7, 1394–1412 RSC.
  154. R. S. Ramsey and J. M. Ramsey, Anal. Chem., 1997, 69, 1174–1178 CrossRef CAS.
  155. N. H. Bings, C. Wang, C. D. Skinner, C. L. Colyer, P. Thibault and D. J. Harrison, Anal. Chem., 1999, 71, 3292–3296 CrossRef CAS PubMed.
  156. P. Hoffmann, U. Häusig, P. Schulze and D. Belder, Angew. Chem., Int. Ed., 2007, 46, 4913–4916 CrossRef CAS PubMed.
  157. S. Fritzsche, S. Ohla, P. Glaser, D. S. Giera, M. Sickert, C. Schneider and D. Belder, Angew. Chem., Int. Ed., 2011, 50, 9467–9470 CrossRef CAS PubMed.
  158. C. Benz, M. Boomhoff, J. Appun, C. Schneider and D. Belder, Angew. Chem., Int. Ed., 2015, 54, 2766–2770 CrossRef CAS PubMed.
  159. K. Ren, J. Zhou and H. Wu, Acc. Chem. Res., 2013, 46, 2396–2406 CrossRef CAS PubMed.
  160. E. Mery, F. Ricoul, N. Sarrut, O. Constantin, G. Delapierre, J. Garin and F. Vinet, Sens. Actuators, B, 2008, 134, 438–446 CrossRef CAS.
  161. A. V. Forzano, V. Becirovic, R. S. Martin and J. L. Edwards, Anal. Methods, 2016, 8, 5152–5157 RSC.
  162. C. A. Smith, X. Li, T. H. Mize, T. D. Sharpe, E. I. Graziani, C. Abell and W. T. S. Huck, Anal. Chem., 2013, 85, 3812–3816 CrossRef CAS PubMed.
  163. S. Mashaghi, A. Abbaspourrad, D. A. Weitz and A. M. van Oijen, Trends Anal. Chem., 2016, 82, 118–125 CrossRef CAS.
  164. L. Mahler, K. Wink, R. J. Beulig, K. Scherlach, M. Tovar, E. Zang, K. Martin, C. Hertweck, D. Belder and M. Roth, Sci. Rep., 2018, 8, 13087 CrossRef PubMed.
  165. K. Wink, L. Mahler, J. R. Beulig, S. K. Piendl, M. Roth and D. Belder, Anal. Bioanal. Chem., 2018, 410, 7679–7687 CrossRef CAS PubMed.
  166. R. J. Beulig, R. Warias, J. J. Heiland, S. Ohla, K. Zeitler and D. Belder, Lab Chip, 2017, 17, 1996–2002 RSC.
  167. T. W. Murphy, Q. Zhang, L. B. Naler, S. Ma and C. Lu, Analyst, 2018, 143, 60–80 RSC.
  168. W. M. Weaver, P. Tseng, A. Kunze, M. Masaeli, A. J. Chung, J. S. Dudani, H. Kittur, R. P. Kulkarni and D. Di Carlo, Curr. Opin. Biotechnol., 2014, 25, 114–123 CrossRef CAS PubMed.
  169. B. L. Khoo, P. K. Chaudhuri, N. Ramalingam, D. S. W. Tan, C. T. Lim and M. E. Warkiani, Int. J. Cancer, 2016, 139, 243–255 CrossRef CAS PubMed.
  170. F. Lyu, M. Pan, S. Patil, J. H. Wang, A. C. Matin, J. R. Andrews and S. K. Y. Tang, Sens. Actuators, B, 2018, 270, 396–404 CrossRef CAS.
  171. K. Galler, K. Bräutigam, C. Große, J. Popp and U. Neugebauer, Analyst, 2014, 139, 1237–1273 RSC.
  172. R. Trouillon, M. K. Passarelli, J. Wang, M. E. Kurczy and A. G. Ewing, Anal. Chem., 2013, 85, 522–542 CrossRef CAS PubMed.
  173. J. Doherty, G. Cinque and P. Gardner, Appl. Spectrosc. Rev., 2017, 52, 560–587 CrossRef CAS.
  174. R. Smith, K. L. Wright and L. Ashton, Analyst, 2016, 141, 3590–3600 RSC.
  175. M. Li, J. Xu, M. Romero-Gonzalez, S. A. Banwart and W. E. Huang, Curr. Opin. Biotechnol., 2012, 23, 56–63 CrossRef CAS PubMed.
  176. L. Zhang and A. Vertes, Angew. Chem., Int. Ed., 2018, 57, 4466–4477 CrossRef CAS PubMed.
  177. T. J. Comi, T. D. Do, S. S. Rubakhin and J. V. Sweedler, J. Am. Chem. Soc., 2017, 139, 3920–3929 CrossRef CAS PubMed.
  178. T. Masujima, Anal. Sci., 2009, 25, 953–960 CrossRef CAS PubMed.
  179. T. L. Colliver, C. L. Brummel, M. L. Pacholski, F. D. Swanek, A. G. Ewing and N. Winograd, Anal. Chem., 1997, 69, 2225–2231 CrossRef CAS PubMed.
  180. J. S. Fletcher, N. P. Lockyer, S. Vaidyanathan and J. C. Vickerman, Anal. Chem., 2007, 79, 2199–2206 CrossRef CAS PubMed.
  181. S. Parry and N. Winograd, Anal. Chem., 2005, 77, 7950–7957 CrossRef CAS PubMed.
  182. P. Sjövall, J. Lausmaa, H. Nygren, L. Carlsson and P. Malmberg, Anal. Chem., 2003, 75, 3429–3434 CrossRef.
  183. N. Musat, R. Foster, T. Vagner, B. Adam and M. M. M. Kuypers, FEMS Microbiol. Rev., 2012, 36, 486–511 CrossRef CAS PubMed.
  184. H. W. LI, X. HUA and Y. T. LONG, Chin. J. Anal. Chem., 2018, 46, 61–66 CAS.
  185. J. S. Mellors, K. Jorabchi, L. M. Smith and J. M. Ramsey, Anal. Chem., 2010, 82, 967–973 CrossRef CAS PubMed.
  186. Y. Zhu, P. D. Piehowski, R. Zhao, J. Chen, Y. Shen, R. J. Moore, A. K. Shukla, V. A. Petyuk, M. Campbell-Thompson, C. E. Mathews, R. D. Smith, W. J. Qian and R. T. Kelly, Nat. Commun., 2018, 9, 1–10 CrossRef PubMed.
  187. O. Guillaume-Gentil, T. Rey, P. Kiefer, A. J. Ibáñez, R. Steinhoff, R. Brönnimann, L. Dorwling-Carter, T. Zambelli, R. Zenobi and J. A. Vorholt, Anal. Chem., 2017, 89, 5017–5023 CrossRef CAS PubMed.
  188. S. Meyer, A. López-Serrano, H. Mitze, N. Jakubowski and T. Schwerdtle, Metallomics, 2018, 10, 73–76 RSC.
  189. L. Mueller, H. Traub, N. Jakubowski, D. Drescher, V. I. Baranov and J. Kneipp, Anal. Bioanal. Chem., 2014, 406, 6963–6977 CrossRef CAS PubMed.
  190. W. Xie, D. Gao, F. Jin, Y. Jiang and H. Liu, Anal. Chem., 2015, 87, 7052–7059 CrossRef CAS PubMed.
  191. Y. Schober, S. Guenther, B. Spengler and A. Römpp, Anal. Chem., 2012, 84, 6293–6297 CrossRef CAS PubMed.
  192. M. E. Dueñas, J. J. Essner and Y. J. Lee, Sci. Rep., 2017, 7, 1–10 CrossRef PubMed.
  193. A. Amantonico, P. L. Urban, S. R. Fagerer, R. M. Balabin and R. Zenobi, Anal. Chem., 2010, 82, 7394–7400 CrossRef CAS PubMed.
  194. A. J. Ibanez, S. R. Fagerer, A. M. Schmidt, P. L. Urban, K. Jefimovs, P. Geiger, R. Dechant, M. Heinemann and R. Zenobi, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 8790–8794 CrossRef CAS PubMed.
  195. S. R. Ellis, C. J. Ferris, K. J. Gilmore, T. W. Mitchell, S. J. Blanksby and M. In Het Panhuis, Anal. Chem., 2012, 84, 9679–9683 CrossRef CAS PubMed.
  196. M. Phelps, J. Hamilton and G. F. Verbeck, Rev. Sci. Instrum., 2014, 85, 124101 CrossRef PubMed.
  197. J. S. Hamilton and G. F. Verbeck, J. Anal. Oncol., 2016, 5, 47–54 CrossRef CAS.
  198. J. S. Hamilton, R. Aguilar, R. A. Petros and G. F. Verbeck, J. Am. Soc. Mass Spectrom., 2017, 28, 918–928 CrossRef CAS PubMed.
  199. P. J. Horn, N. R. Ledbetter, C. N. James, W. D. Hoffman, C. R. Case, G. F. Verbeck and K. D. Chapman, J. Biol. Chem., 2011, 286, 3298–3306 CrossRef CAS PubMed.
  200. M. S. Phelps, D. Sturtevant, K. D. Chapman and G. F. Verbeck, J. Am. Soc. Mass Spectrom., 2016, 27, 187–193 CrossRef CAS PubMed.

This journal is © The Royal Society of Chemistry 2019