Development of a Microfluidic Open Interface with Flow Isolated

1 hour ago - (1-6) Given that SPME integrates the sampling and sample preparation steps of the analytical workflow, the main goal of such research has...
9 downloads 15 Views 2MB Size
Article pubs.acs.org/ac

Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

Development of a Microfluidic Open Interface with Flow Isolated Desorption Volume for the Direct Coupling of SPME Devices to Mass Spectrometry Marcos Tascon, Md. Nazmul Alam, Germán Augusto Gómez-Ríos, and Janusz Pawliszyn* Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada S Supporting Information *

ABSTRACT: Technologies that efficiently integrate the sampling and sample preparation steps with direct introduction to mass spectrometry (MS), providing simple and sensitive analytical workflows as well as capabilities for automation, can generate a great impact in a vast variety of fields, such as in clinical, environmental, and food-science applications. In this study, a novel approach that facilitates direct coupling of Bio-SPME devices to MS using a microfluidic design is presented. This technology, named microfluidic open interface (MOI), which operates under the concept of flow-isolated desorption volume, consists of an open-to-ambient desorption chamber (V ≤ 7 μL) connected to an ionization source. Subsequently, compounds of interest are transported to the ionization source by means of the self-aspiration process intrinsic of these interfaces. Thus, any ionization technology that provides a reliable and constant suction, such as electrospray ionization (ESI), atmospheric-pressure chemical ionization (APCI), or inductively coupled plasma ionization (ICP), can be hyphenated to MOI. Using this setup, the desorption chamber is used to release target compounds from the coating, while the isolation of the flow enables the ionization source to be continuously fed with solvent, all without the necessity of employment of additional valves. As a proof of concept, the design was applied to an ESI-MS/MS system for experimental validation. Furthermore, numerical simulations were undertaken to provide a detailed understanding of the fluid flow pattern inside the interface, then used to optimize the system for better efficiency. The analytical workflow of the developed Bio-SPME-MOI-MS setup consists of the direct immersion of SPME fibers into the matrix to extract/enrich analytes of interest within a short period of time, followed by a rinsing step with water to remove potentially adhering proteins, salts, and/or other interfering compounds. Next, the fiber is inserted into the MOI for desorption of compounds of interest. Finally, the volume contained in the chamber is drained and moved toward the electrospray needle for ionization and direct introduction to MS. Aiming to validate the technology, the fast determination of selected immunosuppressive drugs (e.g., tacrolimus, cyclosporine, sirolimus, and everolimus) from 100 μL of whole blood was assessed. Limits of quantitation in the subppb range were obtained for all studied compounds. Good linearity (r2 ≥ 0.99) and excellent precision, with (8%) and without (14%) internal standard correction, were attained.

A

interfacing with chromatographic systems requires the system to be continuously sealed so as to keep the pressure within the head of the column,10,11,14 a limitation of the method that undeniably diminishes its practicality and appeal for applications that require rapid turnaround times. Similarly to chromatography, electrophoretic systems need to be electrically isolated so as to avoid charge-leaking, which can lead to dramatic decreases in the reproducibility of the method and in the efficiency of separation of target compounds.4 Recently, with the development of more selective and sensitive detectors, such as mass spectrometers, direct hyphenations of SPME to such systems, which elude the separation step, have become widespread.2 While mass spectrometry (MS)

iming to increase analytical throughput and improve detection limits, the direct coupling of solid-phase microextraction (SPME) devices to a broad range of separation and detection systems has been extensively explored over the last 30 years.1−6 Given that SPME integrates the sampling and sample preparation steps of the analytical workflow, the main goal of such research has been to develop interfaces that also allow for automated processes, which in turn enhance the performance of such systems.7−9 Within this context, the history of SPME development includes the development of multiple interfaces that allow for its direct coupling to gas chromatography,10 liquid chromatography,1,11,12 and capillary electrophoresis.4,13 Arguably, one of the greatest advantages of direct-SPME couplings to chromatographic systems is the achieved sensitivity afforded by such methods, which is attained by minimizing analyte dilution and consequently, improving signal-to-noise ratios for target analytes. However, direct © XXXX American Chemical Society

Received: October 17, 2017 Accepted: January 20, 2018

A

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

sensitivity, reproducibility, and flexibility to different Bio-SPME geometries,9,31,32 would offer the reliably desired for any kind of fast routine analysis. In this work, a microfluidic open interface (MOI) that operates under the concept of a flow-isolated desorption volume was designed with aims to directly hyphenate Bio-SPME to different detection and ionization systems. Indeed, one of the highlights of this interface is that it is not limited to ESI-MS, but can be coupled to other ionization techniques such as APCI or ICP. The main requirements for this technique include the presence of a reliable flow dispenser, and the generation of constant aspiration toward the detector. Herein, as a proof of concept, the interface was directly coupled to a triple quadrupole system (MS/MS) via ESI. Fundamentally speaking, MOI allows for the desorption of Bio-SPME fibers in small volumes (7 μL) while it concurrently, and without the employment of a valve system, keeps feeding the ESI with a constant flow to minimize cross-talking. In addition, through application of the isolated flow concept, the developed system facilitates the prevention of electrospray instability while also evading the need for additional equilibration times between experiments, factors which can dramatically decrease experimental reproducibility and the throughput of the workflow. With aims of obtaining a better understanding of the dynamics of each step during the operation of the MOI, numerical simulations of the fluidic system were also generated in this work. Finally, to demonstrate the sensitivity, high speed of analysis, and robustness that can be attained with MOI, quantitation of a group of immunosuppressive drugs (tacrolimus, sirolimus, everolimus, and cyclosporine A) was carried out from 100 μL of whole blood. Due to their low partsper-billion therapeutic ranges33 and their high binding constants to red blood cells, such assays are very challenging via traditional SPME applications. Further, because chromatographic separation is circumvented through application of this method, turnaround times of analysis were significantly reduced.

would be an ideal detector in most cases due to its high selectivity, some hyphenations to spectroscopic detectors have also been reported in the literature.2,15−17 However, use of such techniques may also increase the chances of misidentification of compounds due to the coextraction of interferences with similar spectral characteristics. Conversely, the direct coupling of SPME to MS, owing to the diverse ionization techniques developed to date for this platform, has become a powerful tool for rapid quantitation and screening of a broad variety of compounds present in different matrices.2 For such applications, however, selection of optimal sample introduction parameters and detection methodology is highly dependent on the physicochemical properties of the target compounds, the geometry of the employed SPME device, and the required limits of quantitation.2,18−22 Certainly, the development of new ambient mass spectrometry (AMS)23 technologies has simplified the way SPME devices are interfaced with MS.3,9,24,25 Some of the AMS technologies previously used for the direct coupling of SPME to MS include direct analysis in real time (DART),6,26 dielectric barrier discharge ionization (DBDI),24 and substrate-electrospray ionization (ESI).3,5,9,25 Owing to their wide analyte ionization coverage and robustness, ESI-based technologies are undeniably the preferred ionization method employed by most laboratories in MS applications. In the case of SPME, the first interface of SPME to MS via liquid desorption and ESI was performed by Möder et al.27 with the use of a desorption chamber similar to the one designed by Chen et al.11 in 1995. Aiming to improve the sensitivity of this method, multiple improvements to the desorption chamber have since been carried out to decrease the volume of the elution/ionization solvent. Following this line of reasoning, Walles et al.3 proposed the use of nanoelectrospray ionization (nano-ESI) emitters in combination with SPME fibers for fast quantitation of compounds of biological interest. Likewise, Gomez-Rios et al.28 applied nano-ESI in combination with matrix-compatible SPME fibers (Bio-SPME) as a platform for sensitive analysis of small molecules in biological matrices. Both of the works combine the sample preconcentration and cleanup advantages of SPME with the ionization efficiency, low flow rates, and high tolerance to salts afforded by the nano-ESI platform. Aware of the limitations that could thwart the highthroughput implementation of SPME-nanoESI, such as the high cost per analysis (due to the nonreusability of the emitters), as well as the difficulties associated with automatization of the process, novel and cheaper alternatives began to be explored by our research group.28 One of these developments comprises the integration of Bio-SPME fibers and ESI through the open-port sampling interface25 originally designed by Van Berkel and Kertesz.29,30 Notwithstanding the speed and good sensitivity attained for quantitation of buprenorphine, fentanyl, and clenbuterol in urine samples, the intrinsic features of the open port interface, such as its large desorption volume (∼40 μL), the dynamic desorption step (i.e., continuous flow versus static desorption in nanoESI), and its wide entrance in relation to the size of the employed fiber, could lead to irreproducible analyte desorption, as well as loss of sensitivity due to significant analyte dilution. Although interfiber irreproducible desorption can be corrected through employment of adequate internal standards (IS), methods capable of offering comparable figures of merit that do not necessitate IS applications are always desired. In this context, a Bio-SPME-ESI interface that offers not only the benefits of previous designs, such as a simple open-to-ambient geometry, but also higher



EXPERIMENTAL SECTION Materials and Supplies. Formic acid (FA), ammonium acetate (both LC-MS grade), and polyacrylonitrile (PAN) were purchased from Sigma-Aldrich (Oakville, ON, Canada). Methanol (MeOH), acetonitrile (ACN), and water were LCMS grade, and purchased from Fisher Scientific. Tacrolimus, sirolimus, everolimus, and cyclosporine A were obtained from Millipore-Sigma (Milwaukee, U.S.A.). Deuterated analogues, tacrolimus-d2C1, sirolimus-d3, everolimus-d4, and cyclosporineA-d4 were purchased from TRC-Chemicals (Toronto, ON, Canada). Stock standard solutions were prepared in methanol at a concentration of 1000 μg·mL−1 and stored at −80 °C. Human whole blood (with K2-EDTA as anticoagulant) from different patients was purchased from BioreclamationIVT (Westbury, New York, U.S.A.). All blood samples were spiked and stored overnight at 4 °C prior to use so as to allow for drug−protein binding equilibrium to be established. Waters Corporation (Wilmslow, U.K.) kindly provided the hydrophilic lipophilic balance (HLB) particles (∼5 μm particle diameter) used to prepare the fibers employed in this study. The used fibers were manufactured following an in-house procedure9 that consisted of employing a slurry of HLB particles-PAN to coat nitinol wires (200 μm of diameter); all manufactured fibers were of a final coating thickness of approximately 20 μm and of a length of 10 mm (Figure S1). B

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. (a) Schematics of the hydrodynamics system of the MOI used for the coupling of Bio-SPME to ESI-MS/MS. (b) 3-D geometry used for the numerical simulation.

gas = 25; heated nebulizer temperature = 300 °C; and electrospray voltage = 5000 V. Computational Model. Three aspects of the MOI were modeled, aiming to elucidate the following: (1) The system must be able to transfer all desorbed compounds to the detector as fast as possible with minimum carry-over; (2) the desorption volume of the chamber should be completely isolated from the solvent that is continuously aspirated to the ESI; and (3) the refill of the chamber must be instantaneous so as to make the developed system compatible with highthroughput approaches. Previous work by our group introduced a computational simulation model based on Comsol Multiphysics for quantitative interpretation of SPME experimental data that cannot be otherwise easily experimentally measured.34,35 The model generated in COMSOL Multiphysics (v5.2) simulates the solvent flow that occurs as a function of the pressure drop at the ESI entrance, as well as the solvent flows driven by the pump. A 3-D geometry representing all the components of the MOI chamber was built on the COMSOL Multiphysics geometry interface, using simple shapes and parametric curves (Figure S2). The flow in the sample domain is governed by the Navier−Stokes equation, while the flow field is treated as steady. The steady-state of the system, assuming incompressible fluids, is described by partial differential equations and must be solved simultaneously (see Supporting Information). Sample Preparation, Desorption, and MS/MS Detection. Whole blood samples were spiked with concentrations of tacrolimus, sirolimus, and everolimus ranging between 1 and 50 ng·mL−1, while cyclosporine A concentrations were within the range of 10 and 500 ng·mL−1. All employed internal standards (see Table S1) were spiked at 10 ng·mL−1. An aliquot of 100 μL of whole blood was added to a vial, followed by addition of 900 μL of a mixture containing 30/60/10% v/v/v of ACN/ ZnSO4 0.1M/water.36 Immediately following, 10 mm HLB fibers were immersed for 90 min in the vial containing the mixture so as to allow for extraction of target compounds while blood proteins and cells denaturated. Fibers were quickly rinsed

Flow-Isolated Desorption Volume. As shown in Figure 1, the interface of the developed device consists of two sections. The top section, which functions as the SPME desorption chamber, consists of a Teflon cylinder with two holes; the upper hole has a diameter of 1 mm and a depth of 10 mm, while the lower hole has a diameter of 3.18 mm. Both holes are connected by a channel with a diameter of 0.5 mm and a length of 2 mm, which acts as a flow restrictor. In addition, the volume of the desorption chamber, without considering the fiber volume displacement, is approximately 7 μL. The connection between the open ambient desorption chamber and the electrospray needle employed in this device was inspired by the design of the open-port interface reported by Van Berkel et al.25,29 Succinctly, the procedure involves the employment of two coaxial tubes that allow for solvent delivery through the gap formed between these two tubes. Once the solvent reaches the top of the interface, it is aspirated by the inner tube toward the mass spectrometer through employment of the self-aspiration capabilities of the ESI.29 The outer tube is a 304 stainless steel, 1.75 mm i.d. × 3.18 mm o.d. × ∼ 9 cm long (Grainger, Lake Forest, IL, U.S.A.), and the inner tube is a capillary tube; 254 μm i.d. × 361 μm o.d. × ∼ 20 cm long (Upchurch Scientific, Oak Harbor, WA, U.S.A.). An LC pump (200 Series; PerkinElmer, Santa Clara, CA, U.S.A.) delivered the fluid, while the suction was generated by the Venturi effect produced by the TurboIon spray source. For this purpose, the standard ESI electrode (100 μm i.d.) was replaced with a 150 μm i.d. electrode, aiming to increase the aspiration flow performance of the system. The pump flow, the ESI line, and the fluidic system were secured with a PEEK Tee junction (Upchurch Scientific, Oak Harbor, WA, U.S.A.). As can be seen in Figure 1, the pumping solvent was bypassed with a valve (6-port valve) so as to rapidly switch the flow and allow for self-aspiration of the chamber toward the MS to occur. The conditions needed for the desorption step consisted of an equilibrium between the pump flow at (350 μL·min−1) and the ESI aspiration effect. ESI parameters were as follows: positive ion mode; nitrogen gases set at GS1 = 90, GS2 = 70; curtain C

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Experimental setup for Bio-SPME extraction from complex matrices and desorption−analysis via MOI.

(t ≤ 5 s, 2 rinsing steps) after extraction, then placed directly into the MOI for 5 s for desorption of compounds. Immediately following the removal of the fiber from the desorption chamber, the valve providing fresh solvent was switched so as to rapidly empty (in 2 s) the total volume of the chamber. The plug was then aspirated and ionized by the electrospray, with a dwell time of 3 s. The desorption and ionization solvent employed was methanol containing 0.1% FA and 12 mM ammonium acetate. All experiments herein described were carried out using an API 4000 triple quadrupole mass spectrometer (Applied Biosystems, California, U.S.A.) equipped with a TurboIon spray source. MS/MS analysis was performed in positive ionization mode, using single reaction monitoring (SRM) conditions (see Table S1). All compounds of interest were tuned by external infusion, which was carried out with the use of a syringe pump. Detailed values, including figures associatedwith declustering potential, collision energy, entrance potential, and exit potential, are available in Table S1. The analytical method was validated in terms of LOQ, linearity, and precision. Calibration curves were built with 7 points, using three independent triplicates. A linear regression was constructed as the ratio between the area of the compound and the internal standard as a function of the analyte concentration. Aiming to demonstrate the outstanding performance of the source, calibration curves without internal standard correction were also plotted.

MOI-ESI-MS workflows. Moreover, unlike other direct couplings of SPME to mass spectrometry, desorption time can be tuned from seconds to minutes, as the enclosed chamber can prevent rapid solvent evaporation.5,6,25 While a desorption time of 5 s was selected for this specific proof-of-concept experiment, desorption can be further accelerated through application of external vibration, either to the fiber or the chamber. (4) Once the fiber is removed from the chamber, the valve that controls the solvent is switched for 3 s so that the chamber can be quickly drained. Target compounds are detected in the MS with a full width at half-maximum (fwhm) of less than 2 s; as such, the whole sample plug is quantitatively eluted in about 6 s (Figure 2). Due to the small radius of the chamber, small volumes of solvent (≤7 μL) can be used during the desorption step without sacrificing the length of the fiber (Figure 1), thus allowing for the use of longer fibers (≤10 mm) in comparison to the length of the fiber employed in the Bio-SPME-OPP protocol. As 2.5 times more coating can be desorbed in the currently presented protocol in comparison to Bio-SPME-OPP, where fiber length is limited to a maximum of 4 mm,25 the sensitivity of the method is correspondingly increased. Furthermore, the reproducible desorption volume and the well-known boundaries of the chamber allow for more exhaustive and robust wetting and desorption of the whole coating. Finally, the dilution factor is reduced dramatically as the fibers are desorbed in 4 μL of solvent (3 μL of fiber volume displacement) versus the classical OPP interface, where desorption occurs in a volume of approximately 30−40 μL (total volume, V = F (flow, 350 μL/min) × t (desorption time, 5 s). These aspects, along with the fast flush of the chamber, make the currently presented MOI between 1 and 2 orders of magnitude more sensitive than the OPP previously explored by our group. Therefore, due to its high sensitivity and efficient band compression, Bio-SPME-MOI-ESI-MS emerges as a very attractive technology for clinical applications where (e.g., in vivo or ex vivo samplings) very low recoveries are attained (≤1%).38,39 The duty cycle of the MOI consists of four main stages (see Figure S3). First of all, once the MOI reaches the required steady state (i.e., the isolated volume is practically constant over time, and the solvent delivered by the pump along with the ESI aspiration rate are at equilibrium), the pumped solvent is aspirated at a constant rate by the ESI suction. This not only guarantees a stable ESI, without the need of further equilibration, but also a constant volume on the isolated chamber. In other words, no cross-talk occurs between the delivered fresh solvent and the solvent in the desorption chamber. Once these conditions are fulfilled, the interface is



RESULTS AND DISCUSSION Analytical Workflow. The analytical workflow of the devised SPME-MOI method, as depicted in Figure 2, consists of four main steps: (1) Extraction/enrichment, where the BioSPME coated surface is directly immersed in the sample matrix, and agitated at high speed. Owing to the biocompatibility of the fiber and the open bed nature of SPME, no further sample pretreatment is required, as the chances of protein attachment and clogging are negligible.5,37 (2) Rinsing with water (time ≤10 s); the main goal of the rinsing step is to remove salts and any nonspecific attachments present in the coating surface. (3) Introduction of the fiber into the isolated microfluidic desorption chamber. One of the main virtues of the MOI is the generation of an isolated volume (≤7 μL) while the ESI interface is constantly fed with fresh solvent. This aspect markedly reduces any kind of instabilities, such as bubble formation or ESI instabilities, for instance, that other interfaces may experience.3,28 This marked advantage afforded by the MOI, in turn, allows for easier development of fully automated D

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry considered ready to receive a sample (i.e., SPME device). The second step, the desorption step, is carried out by placing the fiber inside the chamber, where it remains for a fixed time to allow for desorption. Here, the extracted analytes are desorbed in the isolated volume without coming into contact with the flow lines that go toward the ESI system. Therefore, the desorption step is performed under static conditions without any dilution or extra band broadening more than diffusion. After desorption is completed, step 3, which is composed of rapidly flushing the chamber, is carried out. For this reason, the valve connected to the pump is switched to a recirculation position that quickly enables the flow of the pump to reach zero. As a result, the only flow acting on the chamber is the ESI suction, which can flush the chamber in less than 3 s. Theoretically, assuming that the suction flow at equilibrium is set at 350 μL·min−1, and that the desorption volume is 4 μL, approximately only 1 s would be required to empty the entire chamber. Undoubtedly, several practical factors, such as valve switching speed and slow movement of initially static fluids, delay the optimal flushing time toward values around 3 s. It is important to bear in mind that employment of longer switching times could cause air to be aspirated from the open port, which might lead to bubble generation and, as a result, decreases in workflow reproducibility due to ESI instabilities. Finally, in step 4, aiming to rapidly refill the desorption chamber, the system is overflowed at 1000 μL/min for about 3 s. Two main strategies can be used with similar success to accomplish this step; increasing the pump flow over the suction flow, or reducing the ESI gas nebulizer to reduce the aspiration rate of the source. Regardless of the chosen protocol, the idea is to first overflow the source so that the chamber is completely cleaned, then refill the chamber again for the following sample introduction. After rapidly cleaning the entire system, the pump is set to the previously described equilibrium conditions, thus closing the cycle. To summarize, the final source duty cycle, excluding the time needed for desorption, takes approximately 6 s in total. Numerical Modeling of the Isolated Volume. Aiming to validate experimental findings related to the operation of the MOI, COMSOL simulations of the three main stages of the interface duty cycle were performed. Essentially, these steps include the steady-state of the ready-to-sample interface, the fast drain of the volume after the desorption step, and the refilling of the chamber to restore initial conditions. Considering that all the relevant phenomena to describe and demonstrate the behavior of the interface takes place in the gap between the desorption chamber and the inlet of the ESI source, Figure 3 provides detailed depictions of this area as it pertains to the three modeled steps (the full modeled geometry can be seen in Figure S4), with the surface plot indicating the magnitude of the flow velocity, which is represented with different colors, while the arrows indicate the direction of the flow. Further, the overall size of the arrow represents the magnitude of the velocity field. Figure 3A provides confirmation of the isolation of the desorption volume from the flow that feeds the ESI (pump flow ≈ ESI aspiration ≈350 μL· min−1). Once the volume of the chamber is filled, all solvent supplied by the pump enters into the ESI inlet. Logically, this is evidenced by the absence of flow arrows in the desorption chamber. In addition, in order to keep the mass balance between the solvent being pumped and the electrospray, the flow velocity inside the inner tube (smaller cross section) increases dramatically in comparison to the velocity of the flow inside the outer tube (larger cross-section). As presented in

Figure 3. Numerical simulations of the MOI interface at three different stages of the duty cycle (general view). (A) Steady-state (source ready to sample); the flow of the pump is approximately the same as that of the aspiration flow. (B) Draining of the chamber; pump flow = 0 μL min−1, aspiration flow of 350 μL min−1. (C) Refilling of the chamber; pump flow 1750 μL min−1 and aspiration flow μL min−1.

Figure 3B, simulations of the chamber evacuation step (pump flow = 0; ESI flow ≈350 μL·min−1) exposed two critical factors that require special attention. First, the draining process keeps a constant aspiration rate toward the ESI, until it reaches the inlet tubing where, in order to keep the mass balance, the velocity increases approximately 5-fold, thus generating a really efficient plug. Second, during the aspiration step, the aspirated fluid creates a sheath flow toward the ESI that is not mixed with the gap fluid, minimizing any carry-over of the source between runs. This can be verified in Figure 3B, where all areas that surround the inlet appear arrowless and colored in a violet hue, indicating the absence of fluid flow. Finally, the overflow of the chamber, using flow velocities several times higher than the equilibrium condition estimated at 350 μL·min−1 (flow pump ≫ ESI aspiration), is demonstrated in Figure 3C; in this specific case, the flow set for calculations was 1000 μL·min−1. This overflow can refill the chamber in less than 1 s; thus, this procedure can work as a self-cleaning step, where three chamber volumes are circulated in that period of time. E

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. Quantitative analysis of 100 μL of human whole blood spiked with tacrolimus, sirolimus, everolimus (1 ng mL−1 to 50 ng mL−1) and cyclosporine A (10 ng mL−1 to 500 ng mL−1). The isotopologues tacrolimus-d2C1, sirolimus-d3, and everolimus-d4 were spiked at 10 ng mL−1, while cyclosporine-d4 was spiked at 100 ng mL−1. Analyses were performed using Bio-SPME-MOI-ESI-MS/MS. Bars represent the standard deviation of analyses for three replicates with independent fibers and sample.

in turn causes increases to the diffusion coefficients of the compounds. As can be seen in Figure 4, excellent calibration curves within the range of 1 and 50 ng·mL−1 were obtained for tacrolimus, sirolimus, and everolimus, while curves between 10 to 500 ng·mL−1 were attained for cyclosporine A. The LOQs were 1, 0.3, 0.3, and 0.6 ng·mL−1 for tacrolimus, sirolimus, everolimus and cyclosporine A, respectively. Moreover, the average precision of the method, expressed as RSD (%), was less than 8% (see Table S2), while the linearity was acceptable in all cases (r2 ≥ 0.99). As presented in Figure S5, ion chronograms for the compounds spiked in whole blood, tacrolimus, sirolimus, everolimus (spiked at 1 ng·mL−1) and cyclosporine (spiked at 10 ng·mL−1), showed adequate signalto-noise ratios when compared to blood blanks. The results obtained in this study are certainly outstanding in comparison to similar direct-to-MS technologies.5,24,25,28,43 Indeed, not only were exceptional LOQs achieved through the technique, the currently presented protocol allowed for the simultaneous quantitation of all four immunosuppressive drugs.36,44,45 Furthermore, unlike other direct-to-MS analysis protocols,46 while the use of internal standards is certainly useful to correct for intersample variabilities, their employment is not crucial to correct for errors associated with the ionization interface. As can be seen in Figure S6, calibration curves can be plotted based on the instrument response (peak area) as a function of the analyte concentration in the sample. The linearity of the method is as good as when correcting with the correspondent internal standard (r2 ≥ 0.99); while the precision, represented by error bars, has an average value of 14% (RSD, see Table S2). While the obtained figures of merit are slightly poorer in comparison to those obtained with IS, such values are still

Resultingly, the developed method offers minimum carry-over, while circumventing additional cleaning steps and waiting periods between cycles. Determination of Immunosuppressive Drugs from Whole Blood. As a proof of concept, the SPME-MOI-MS/MS workflow was carried out for determination of tacrolimus, everolimus, sirolimus, and cyclosporine A in human whole blood. From a clinical standpoint, owing to the low and narrow therapeutic ranges of these compounds, their accurate quantitation in biofluids such as whole blood is of paramount importance in clinical applications involving such compounds, which require individualized dosing and monitoring.40 From an analytical standpoint, quantitation of these compounds poses a challenge for SPME due to their high binding constants to erythrocytes (≥90%), and their high molecular masses (≥760 Da). As SPME only extracts unbound (free form) compounds, the poor availability of these analytes in their free form poses a significant challenge in terms of obtaining the sensitivity required for detection. Further, the high molecular masses of these compounds make the kinetics of extraction quite slow due to their low diffusion coefficients and low polarities.34,41Aiming to address these issues, we followed a protocol developed in our research group where, during the sample preparation step (see Experimental Section), a mixture of ACN/ZnSO4/H2O is added to whole blood.42 Although solvent addition can cause a significant reduction to the affinity constant between the extracting phase and the analyte, it markedly increases the free concentration of target analytes in the sample, thus enabling the acquirement of excellent limits of quantitation. Moreover, addition of solvent also speeds up the extraction process due to a reduction in sample viscosity, which F

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry acceptable considering that these values represent the total error associated with the method. Yet, one should keep in mind that in real case scenarios, where analysis involves samples with diverse characteristics (e.g., hematocrit levels), the use of internal standards is mandatory; as such, the attainment of such low interexperiment RSD values, even without the addition of internal standard, only demonstrates the robustness of the workflow. Comparison to Other Direct-to-MS Techniques. In order to point out the high sensitivity achieved by this interface, Figure S7 shows the results obtained through employment of the same sample preparation protocol, but using the Bio-SPME open-port probe coupled to tandem mass spectrometry (BioSPME-OPP-MS/MS)25 workflow previously published by our research group. In this comparison, the desorption time of the Bio-SPME-OPP-MS/MS25 workflow was set as the same as that used for the currently presented work (5 s), while the employed sorbent was identical in composition, but 4 mm in length (the maximum length that can be inserted in the open port) instead of the 10 mm used for the current study. Essentially, the attained results showed a significant increment in sensitivity for the four target compounds; more specifically, 3-, 30-, 15-, and 7-fold improvements in signal-to-noise ratios for tacrolimus, sirolimus, everolimus, and cyclosporine A, respectively. Nevertheless, the potential of Bio-SPME-MOIMS/MS not only lies in the high sensitivity it affords, but also in its open-to-ambient nature, as well as the simplicity of its coupling. In addition, these results can be also compare to those recently reported by our group using coated bladespray36 where the LOQs achieved were similar for tacrolimus but 3, 4, and 20 times better for everolimus, sirolimus, and cyclosporine A, respectively. It must be highlighted that in those experiments, 50% more sample was used when compared to the SPME-MOI work herein presented. Other technologies, such as paper spray, have been reported for the quantitation of tacrolimus44 and separately, cyclosporine A and sirolimus.45 Despite the sample amount used in PS was significantly lower (≤50 μL), the LOQs achieved in our work were 60, 8, and 1.5 times better for Cyclosporine A, sirolimus, and tacrolimus, respectively. Another advantage of SPME-MOI, as previously demonstrated with CBS, is the possibility of simultaneous quantitation of the four drugs. Regarding carryover, given that the operational mechanism of the MOI is the same as in the OPP,25 the solvent is continuously flushing allowing the selfcleaning of the source and chamber between experiments. As a matter of fact, and as shown in Figure S2, during the restitution step the chamber is overflowed for about 3 s in order to clean any potential remaining analyte. In contrast to other technologies, such as CBS and PS, where the collection/ extraction device also acts as the ionization source and the devices are disposable, the carry-over can be considered negligible. Besides, the solvent consumption for MOI and OPP is approximately 350 μL·ml−1 while the solvent use in CBS is typically under 20 μL per analysis. Although the pump flow used in MOI is relatively high, the time of analysis is extremely short (≤10s) making the amount of solvent used per experiment in the order of 70−90 μL. Certainly, this amount of solvent is lower than the 170 μL employed in the paper spray method reported for the same set of compounds.44 Another advantage of MOI is the small desorption volume employed (≈ 4 μL) against the 15 μL of blade spray,5,9,36 30 μL of the OPP,25 and 170 μL of paper spray,44 giving in this way, higher enrichment factors.

Article



CONCLUSIONS



ASSOCIATED CONTENT

In this work, an interface for online coupling of SPME devices to any kind of separation or detection system by means of an open-to-ambient desorption volume was presented. As a proof of concept, the developed MOI was adapted to ESI-MS/MS, taking advantage of the hydrodynamic technologies previously developed for the OPP.25 Furthermore, numerical simulations corroborated that, while in steady-state conditions (i.e., pump delivery flow equals to aspiration flow by the ESI source), an isolated volume is created on the open desorption chamber, which allows for a dramatic improvement in interexperimental reproducibility. In addition, the open-to-ambient nature of the interface facilitates a wide range of automation strategies. As a proof of concept, SPME-MOI-MS/MS was applied toward the determination of immunosuppressive drugs from 100 μL of whole blood. The method was shown to enable outstanding sensitivity for the analytes of interest (LOQs of 1, 0.25, 0.3, and 0.6 ng·mL−1 for tacrolimus, sirolimus, everolimus, and cyclosporine A, respectively), with figures of merit superior to those attained by the previously developed SPME-OPP approach (∼10). Method precision, assessed for 7 concentration levels, averaged 8% with internal standard correction, and 14% without employment of IS. Such results showcase the great interexperimental reproducibility of the method, which to-date has yet to be demonstrated by any other direct-to-ESIMS technology. Further, the design of MOI requires only minimal alterations to the front-end of the instrument. Hence, this interface, which offers high peak efficiency, sensitivity, simplicity, low cost of analysis, and robustness, is here presented as an extremely attractive alternative for fast analysis of a range of compounds from complex matrices, as required in clinical diagnosis or therapeutic drug monitoring. Further, in view of the ability of Bio-SPME to extract from complex biomatrices such as biofluids or tissue, future work will be directed toward fast and high-throughput therapeutic monitoring as well as bedside in vivo analytical determinations at the operating room.39,47 The size of the extraction devices is not limited to small fibers described in this work. It can be extended to larger surface area devices, such as coated blades39,47 resulting in higher sensitivity as more analytes are transferred to the mass spectrometer. This is possible as the desorption volume is isolated, so any design of the desorption chamber is possible.

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b04295. Details regarding MS parameters used to quantify each compound, and repeatability values using different calibration methods; details of the source and duty cycle are also provided, as well as equations, geometries, and full-scale images used in numerical simulations; and information regarding the quantitative determination of model immunosuppressive drugs and data associated with comparisons of the current method against the BioSPME-OPP-MS/MS25 protocol (PDF) G

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry



(25) Gomez-ríos, G. A.; Liu, C.; Tascon, M.; Reyes-Garces, N.; Arnold, D. W.; Covey, T. R.; Pawliszyn, J. Anal. Chem. 2017, 89 (7), 3805−3809. (26) Gomez-Rios, G. A.; Vasiljevic, T.; Gionfriddo, E.; Yu, M.; Pawliszyn, J. Analyst 2017, 142, 2928−2935. (27) Möder, M.; Löster, H.; Herzschuh, R.; Popp, P. J. Mass Spectrom. 1997, 32 (11), 1195−1204. (28) Gómez-Ríos, G. A.; Reyes-Garcés, N.; Bojko, B.; Pawliszyn, J. Anal. Chem. 2016, 88 (2), 1259−1265. (29) Van Berkel, G. J.; Kertesz, V. Rapid Commun. Mass Spectrom. 2015, 29 (19), 1749−1756. (30) Van Berkel, G. J.; Kertesz, V. Rapid Commun. Mass Spectrom. 2017, 31 (3), 281−291. (31) Poole, J. J.; Grandy, J. J.; Yu, M.; Boyaci, E.; Gómez-Ríos, G. A.; Reyes-Garcés, N.; Bojko, B.; Heide, H. V.; Pawliszyn, J. Anal. Chem. 2017, 89, 8021. (32) Piri-Moghadam, H.; Alam, M. N.; Pawliszyn, J. Anal. Chim. Acta 2017, 984, 42. (33) Shipkova, M.; Valbuena, H. TrAC, Trends Anal. Chem. 2016, 84, 23−33. (34) Alam, M. N.; Ricardez-Sandoval, L.; Pawliszyn, J. Anal. Chem. 2015, 87 (19), 9846−9854. (35) Alam, M. N.; Ricardez-Sandoval, L.; Pawliszyn, J. Ind. Eng. Chem. Res. 2017, 56 (13), 3679−3686. (36) Gómez-Ríos, G. A.; Tascon, M.; Reyes-Garcés, N.; Boyacı, E.; Poole, J. J.; Pawliszyn, J. Anal. Chim. Acta 2018, 999, 69−75. (37) Musteata, M. L.; Musteata, F. M.; Pawliszyn, J. Anal. Chem. 2007, 79 (18), 6903−6911. (38) Cudjoe, E.; Bojko, B.; de Lannoy, I.; Saldivia, V.; Pawliszyn, J. Angew. Chem., Int. Ed. 2013, 52, 12124−12126. (39) Bojko, B.; Gorynski, K.; Gomez-Rios, G. A.; Knaak, J. M.; Machuca, T.; Cudjoe, E.; Spetzler, V. N.; Hsin, M.; Cypel, M.; Selzner, M.; Liu, M.; Keshjavee, S.; Pawliszyn, J. Lab. Invest. 2014, 94 (5), 586− 594. (40) Koster, R. A.; Alffenaar, J. W. C.; Greijdanus, B.; Uges, D. R. A. Talanta 2013, 115, 47−54. (41) Pawliszyn, J. Handbook of Solid-Phase Microextraction; Industry Press: Beijing, 2009. (42) Gómez-Ríos, G. A.; Tascon, M.; Reyes-Garcés, N.; Boyacı, E.; Poole, J. J.; Pawliszyn, J. Sci. Rep. 2017, Article No. 16104. (43) Déglon, J.; Thomas, A.; Mangin, P.; Staub, C. Anal. Bioanal. Chem. 2012, 402 (8), 2485−2498. (44) Shi, R. Z.; Gierari, E. M. El; Manicke, N. E.; Faix, J. D. Clin. Chim. Acta 2015, 441, 99−104. (45) Shi, R. Z.; El Gierari, E. T. M.; Faix, J. D.; Manicke, N. E. Clin. Chem. 2016, 62 (1), 295−297. (46) Strittmatter, N.; Düring, R.-A.; Takáts, Z. Analyst 2012, 137 (17), 4037−4044. (47) Bessonneau, V.; Ings, J.; Mcmaster, M.; Smith, R.; Bragg, L.; Servos, M.; Pawliszyn, J. Sci. Rep. 2017, 7, 44038.

AUTHOR INFORMATION

Corresponding Author

*Phone: +1 (519) 888 4641. Fax: +1 (519) 746 0435. E-mail: [email protected]. ORCID

Marcos Tascon: 0000-0002-2417-0800 Janusz Pawliszyn: 0000-0002-9975-5811 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors are grateful to SCIEX and the Natural Sciences and Engineering Research Council (NSERC) of Canada for the financial support provided through the Industrial Research Chair program. Finally, we would like to express our sincere gratitude to Justen Poole and the Science Technical Services at the University of Waterloo for their technical support and collaboration during the construction and improvement of the MOI interface.



REFERENCES

(1) Zambonin, C. G. Anal. Bioanal. Chem. 2003, 375 (1), 73−80. (2) Deng, J.; Yang, Y.; Wang, X.; Luan, T. TrAC, Trends Anal. Chem. 2014, 55, 55−67. (3) Walles, M.; Gu, Y.; Dartiguenave, C.; Musteata, F. M.; Waldron, K.; Lubda, D.; Pawliszyn, J. J. Chromatogr. A 2005, 1067 (1−2), 197− 205. (4) Liu, Z.; Pawliszyn, J. J. Chromatogr. Sci. 2006, 44 (6), 366−374. (5) Gómez-Ríos, G. A.; Pawliszyn, J. Angew. Chem., Int. Ed. 2014, 53 (52), 14503−14507. (6) Gómez-Ríos, G. A.; Pawliszyn, J. Chem. Commun. 2014, 50 (85), 12937−12940. (7) Vuckovic, D.; Cudjoe, E.; Musteata, F. M.; Pawliszyn, J. Nat. Protoc. 2010, 5 (1), 140−161. (8) Reyes-Garcés, N.; Bojko, B.; Pawliszyn, J. J. Chromatogr. A 2014, 1374, 40−49. (9) Tascon, M.; Gómez-Ríos, G. A.; Reyes-Garcés, N.; Poole, J.; Boyacı, E.; Pawliszyn, J. Anal. Chem. 2017, 89 (16), 8421−8428. (10) Eisert, R.; Levsen, K. J. Chromatogr. A 1996, 733, 143−157. (11) Chen, J.; Pawliszyn, J. B. Anal. Chem. 1995, 67 (15), 2530− 2533. (12) Lord, H. L. J. Chromatogr. A 2007, 1152, 2−13. (13) Whang, C.; Pawliszyn, J. Anal. Commun. 1998, 35, 353−356. (14) Penalver, A.; Pocurull, E.; Borrull, F.; Marce, R. M. J. Chromatogr. A 2002, 953 (1−2), 79−87. (15) Odziemkowski, M.; Koziel, J. A.; Irish, D. E.; Pawliszyn, J. Anal. Chem. 2001, 73 (13), 3131−3139. (16) Nwaneshiudu, I. C.; Yu, Q.; Schwartz, D. T. Appl. Spectrosc. 2012, 66 (12), 1487−1491. (17) Fragueiro, S.; Lavilla, I.; Bendicho, C. J. Anal. At. Spectrom. 2004, 19 (2), 250−254. (18) Rahmi, D.; Takasaki, Y.; Zhu, Y.; Kobayashi, H.; Konagaya, S.; Haraguchi, H.; Umemura, T. Talanta 2010, 81 (4−5), 1438−1445. (19) Mester, Z.; Lam, J.; Sturgeon, R.; Pawliszyn, J. J. Anal. At. Spectrom. 2000, 15 (7), 837−842. (20) Guo, X.; Mester, Z.; Sturgeon, R. E. Anal. Bioanal. Chem. 2002, 373 (8), 849−855. (21) Chen, Y. C.; Sun, M. C. Rapid Commun. Mass Spectrom. 2002, 16 (12), 1243−1247. (22) Wang, Y.; Schneider, B. B.; Covey, T. R.; Pawliszyn, J. Anal. Chem. 2005, 77 (24), 8095−8101. (23) Monge, M. E.; Harris, G. A.; Dwivedi, P.; Fernández, F. M. Chem. Rev. 2013, 113 (4), 2269−2308. (24) Mirabelli, M. F.; Wolf, J.; Zenobi, R. Anal. Chem. 2016, 88 (14), 7252−7258. H

DOI: 10.1021/acs.analchem.7b04295 Anal. Chem. XXXX, XXX, XXX−XXX