Abstract
Recently microdroplet reactions have aroused much interest because the microdroplet provides a unique medium where organic reactions could be accelerated by a factor of 103 or more. However, microdroplet reactions of proteins have been rarely studied. We report the occurrence of multiple-step reactions of a large protein, specifically, the digestion, reduction, and deglycosylation of an intact antibody, which can take place in microseconds with high reaction yields in aqueous microdroplets at room temperature. As a result, fast structural characterization of a monoclonal antibody, essential for assessing its quality as a therapeutic drug, can be enabled. We found that the IgG1 antibody can be digested completely by the IdeS protease in aqueous microdroplets in 250 microseconds, a 7.5 million-fold improvement in speed in comparison to traditional digestion in bulk solution (>30 min). Strikingly, inclusion of the reductant tris(2-carboxyethyl)phosphine (TCEP) in the spray solution caused simultaneous antibody digestion and disulfide bond reduction. Digested and reduced antibody fragments were either collected or analyzed online by mass spectrometry. Further addition of PNGase F glycosylase into the spray solution led to antibody deglycosylation, thereby producing reduced and deglycosylated fragments of analytical importance. In addition, glycated fragments of IgG1 derived from a glucose modification were identified quickly with this ultrafast digestion/reduction technique. We suggest that microdroplets can serve as powerful microreactors for both exploring large molecule reactions and speeding their structural analyses.
Keywords: mass spectrometry, microdroplet reaction, antibody characterization, digestion, deglycosylation
Graphical Abstract

Introduction
Microdroplets have been recently found to be a unique reaction media in which reaction acceleration can occur.1 It has aroused much attention in the field of chemistry and has been extensively investigated.2-22 Various reactions of organic molecules are markedly promoted in sprayed micron-sized droplets (microdroplets) compared with the same reactions in bulk-phase solution.23 Numerous explanations have been offered for why reaction rate acceleration occurs in aqueous microdroplets. These explanations include evaporation, increased autoionization, partial desolvation, the presence of an intrinsic strong electric field at the interface, the enhanced concentration of solutes on the surface of the microdroplet, and restricted orientations.24-33 The factors that are dominant in any given situation still need to be identified. However, most previous work focused on one-step reactions of small molecules. Biochemical reactions involving proteins have been rarely investigated, except a recent report of trypsin digestion of proteins in microdroplets.34
Therapeutic monoclonal antibodies (mAbs) are one of the fastest growing classes of drugs. More than one hundred mAbs for treatment of many pathologies such as cancer and autoimmune diseases have been approved or are in regulatory review in the US and EU.35,36 This ever-growing abundance has created a need for rapid technologies to characterize mAbs to secure drug product safety, quality, and efficacy.37-39 Traditional mass spectrometry (MS) mAb characterization methods include intact and subunit mass analysis and peptide mapping.40-44 To characterize mAbs in a bottom-up or middle-down proteomics approach, mAbs must be subjected to enzymatic digestion into peptides or polypeptides before peptide mapping analysis by MS. However, digestion is usually a time-consuming step that can take from 30 min to overnight incubation.45,46 In addition, commonly used digestion methods often include additional steps of protein denaturation, reduction, and alkylation to unfold mAb structure and facilitate digestion; these additional steps may also lengthen process time and reduce analysis speed and throughput.47 To accelerate digestion of mAb (or other proteins), a variety of methods have been investigated, including increasing the digestion temperature, adding organic solvents, applying microwave energy, using high intensity focused ultrasound or employing microchip reactor.48-56 Nevertheless, an alternative method that is very fast for mAb digestion would be highly valuable.
In this study, we present unprecedented and fast microdroplet reactions involving a large protein substrate (i.e., an antibody), which would have high impact in proteomics for rapid characterization of antibodies. In this work, ultrafast digestion of the NIST IgG1 antibody in 250 microseconds in microdroplets was achieved. The IgG1 was selectively cleaved by IdeS protease into antigen-binding fragment F(ab′)2 and single-chain, crystallizable fragment scFc (Figure 1a). When we included the reductant tris(2-carboxyethyl)phosphine (TCEP) in the spray solution, simultaneous disulfide bond reduction occurred, leading to digested and reduced IgG1 fragments that included light chain LC, N-terminal half of heavy chain Fd′ and scFc subunits (Figure 2a). These fragments could be either collected for further analysis or detected online by MS. In addition, we achieved ultrafast deglycosylation of IgG1 by including PNGase F glycosylase in the microdroplets (Figure 4a), which constituted an alternative method to rapidly remove glycans from antibodies. In addition, following incubation with glucose, we successfully digested and characterized glycated IgG1 fragments in microdroplets, which suggested a fast way to detect antibody modifications in general.
Figure 1.
Microdroplet digestion of IgG1 (the NIST IgG1 monoclonal antibody reference material 8671) by IdeS enzyme: (a) Schematic drawing of intact IgG1 and IdeS-cleaved IgG1 fragments; The light chain, heavy chain, and hinge region are highlighted in green, blue, and grey, respectively. Black solid lines indicate disulfide bonds connecting heavy and light chains. The IdeS cleavage site is indicated with scissors and dash line. (b) Workflow of microdroplet digestion of IgG1 by IdeS; nESI-MS spectra of (c) antibody fragments obtained from microdroplet digestion of IgG1 by IdeS at room temperature and (d) in-solution digested IgG1 by IdeS for 5 min at 37°C.
Figure 2.
Simultaneous microdroplet digestion and reduction of IgG1. (a) Schematic representation of intact IgG1 and fragment structures of IgG1 after IdeS digestion and TCEP reduction. (b) Schematic drawing of the microdroplet digestion and reduction workflow. (c) The expanded MS spectrum with LC, Fd′ and glycosylated scFc fragments detected and annotated. (d) The corresponding deconvoluted MS spectrum of (c). (e) Schematic drawing of simultaneous microdroplet digestion and reduction coupled with online EESI-MS detection. (f) Deconvoluted MS spectrum of digested and reduced IgG1 from online EESI-MS detection. Note that asterisk (*) in f) denotes a harmonic peak of Fab (equivalent to half of F(ab′)2).
Figure 4.
“One-pot” microdroplet reactions with IgG1, PNGase F, IdeS, and TCEP. (a) Schematic illustrating the PNGase F deglycoylation; (b) The expanded MS spectrum with LC, Fd′ and scFc (both deglycosylated and glycosylated) fragments detected and annotated; (c) The deconvoluted MS spectrum of (b).
Experimental Section
Chemicals
NIST monoclonal antibody reference material 8671 (NIST mAb, humanized IgG1κ monoclonal antibody) was obtained from the National Institute of Standards and Technology (Gaithersburg, MD). The IdeS enzyme was purchased from Genovis Inc (Cambridge, MA). Tris(2-carboxyethyl)phosphine hydrochloride (TCEP, bioultra grade), ammonium bicarbonate (bioultra grade), sodium phosphate dibasic (bioultra grade), and sodium phosphate monobasic (bioultra grade) were purchased from Sigma Aldrich (St. Louis, MO). D-(+)-Glucose (98% purity) was obtained from TCI America (Montgomeryville, PA). Acetonitrile (HPLC grade), formic acid (FA, LCMS grade) and trifluoroacetic acid (TFA, LCMS grade) were purchased from Fisher Scientific (Fair Lawn, NJ). A Millipore Direct-Q5 purification system (Burlington, MA) was used to obtain purified water for sample preparation.
Microdroplet Generation
Microdroplets were generated by spraying an aqueous sample solution (NIST IgG1 mixed with IdeS, TCEP, PNGase F, or all reagents together) through a home-made sprayer with the assistance of nitrogen gas as nebulizer at 120 psi. The home-made sprayer was built exactly as previously described for an electrosonic spray source.57 Briefly, the aqueous sample solution was delivered through a fused-silica capillary (100 μm i.d. and 200 μm o.d., Polymicro Technologies, Phoenix, AZ). Another coaxial fused-silica capillary (250 μm i.d. and 300 μm o.d.) was used and the small size difference between two capillaries was capable of providing nebulizing gas at high velocity.14 Based on the previous reports,14,34 such an sprayer with the use of 120 psi nitrogen nebulization gas pressure produces small microdroplets of 13±6 μm in diameter which is critical for achieving reaction acceleration effect in microdroplets. The sprayer does not need an applied voltage.
Microdroplet Digestion with nESI-MS Analysis
For the experiment shown in Figure 1b, 50 μL of 1 mg/mL IgG1 in 5 mM NH4HCO3 buffer (pH 8) was loaded in one syringe. Then 50 μL of 2 units/μL IdeS enzyme in 5 mM NH4HCO3 buffer (pH 8) was loaded in the other syringe. The flow rate of both syringes were 5 μL/min and the reactants were mixed by a Tee. The length of capillary between mixing Tee and sprayer was 2 cm. The microdroplets were generated through the sprayer and collected in a vial containing a quenching solvent of 50 μL of H2O and 1% FA. The distance between the sprayer tip and the surface of quenching solution was 20 mm. After 5 min collection, desalting and reagent removal was performed with a C4 Ziptip (MilliporeSigma, Burlington, MA). The desalted sample was analyzed by nESI in front of a QExactive Orbitrap mass spectrometer (Thermo Scientific, San Jose, CA). The injection flow rate was 2 μL/min and +3 kV was applied for ionization. The temperature of the MS inlet was 250°C. The mass spectrometer resolution was set to 17,500 for all the MS analysis. All sample mixing and injecting steps were performed rapidly without delay time between each step to avoid any reaction occurring in the bulk solution under ambient temperature. For the experiment in the Figure 2b, 50 μL of 1 mg/mL IgG1 in 5 mM NH4HCO3 buffer containing 5 mM TCEP (pH 8) was loaded in one syringe. Then 50 μL of 2 units/μL IdeS enzyme in 5 mM NH4HCO3 buffer (pH 8) was loaded in the other syringe. The remaining steps were the same as the steps in the aforesaid Figure 1b experiment.
Microdroplet Digestion with Online EESI-MS Analysis
In the online workflow, the EESI-MS method reported by Chen and coworkers58 was adopted with minor modifications. The schematic drawing is shown in Figure 2e. Briefly, one sprayer was used to generate microdroplets from a one-pot mixture of NIST IgG1, IdeS and TCEP solution without applying any voltage. The mixed sample solution consisted of 5 μL of 10 mg/mL NIST IgG1, 2 μL of 50 units/μL IdeS enzyme and 93 μL of 5 mM NH4HCO3 buffer containing 5 mM TCEP. The other sprayer emitted ACN/H2O/FA solvent (50:50:0.5%), and +3 kV voltage was applied to the solvent through a metal alligator clip attached to the stainless steel tip of the syringe used for solvent infusion. The flow rates of both sprayers were 10 μL/min. The pressure of nitrogen gas for both sprayers was 120 psi. The distance between the tip of solvent sprayer to the MS inlet was 15 mm, and the distance between the tip of mixed sample sprayer to the MS inlet was 20 mm. The EESI source was aligned carefully to the MS inlet to achieve the highest sensitivity. The temperature of MS inlet was 250°C.
Glycated IgG1 Characterization
Twenty microliters of 10 mg/mL NIST IgG1 was added to 180 μL of 200 mM NH4HCO3 buffer containing 200 mM glucose (pH 8). The final concentration of IgG1 was 1 mg/mL. Then, sample solution was incubated at 37°C for 0, 2 and 5 days. After incubation, 10 μL of 1 mg/mL glycated IgG1 was drawn out and mixed with 2 μL of 50 units/μL IdeS in 88 μL of 5 mM NH4HCO3 buffer containing 5 mM TCEP (pH 8). The final concentration of glycated IgG1 was 0.1 mg/mL in the resulting mixture which was sprayed to trigger microdroplet reactions as a flow rate of 10 μL/min for 5 min. Then 50 μL of H2O containing 0.1% TFA was added to reconstitute the collected microdroplets and terminate enzymatic reaction. Desalting and reagent removal was performed with a C4 Ziptip. The desalted sample was analyzed by nano-electrospray ionization (nESI) using the QExactive Orbitrap mass spectrometer. The injection flow rate was 2 μL/min and +3 kV was applied for ionization. The temperature of the MS inlet was 250°C.
Deglycosylation of IgG1 by PNGase F
Fifty micrograms of NIST IgG1 in 5 mM NH4HCO3 buffer (pH 8) was loaded in one syringe. The final concentration of IgG1 was 1 mg/mL. Then 100 μL of 1 unit/μL mM PNGase F in 5 mM NH4HCO3 buffer (pH 8) was loaded in the other syringe. The remaining steps were the same as the steps in the aforesaid Figure 1b experiment.
Results and Discussion
We first investigated the direct microdroplet digestion of IgG1 by IdeS enzyme. The IdeS enzyme, cloned from Streptococcus pyogenes and expressed in Escherichia coli, specifically digests IgG1 below the hinge region and generates F(ab')2 and scFc fragments (Figure 1a). In our experiment, IgG1 and IdeS were pre-loaded in two syringes, respectively (Figure 1b). Both syringes were pumped at a flow rate of 5 μL/min and reactants were brought together in a mixing Tee connected to a sprayer via a piece of 2 cm fused silica capillary. The mixed sample was sprayed for 5 min and the microdroplets were collected into a vial containing 1% formic acid in water as a quenching solvent (IdeS is inactivated at pH<5). The vial lid was closed and a hole was drilled on the lid to allow the insertion of the sprayer tip into the vial for spray. The distance between sprayer and the quenching solution in the vial was 2 cm, and the speed of the sprayed microdroplets was 84±18 m/s as previously measured.14 Thus, the flight time of microdroplet between sprayer and quenching solution was only 250 μs [2 cm/(84 m/s)≅250 μs]. Even in such a short spray time, we achieved optimal digestion efficiency of antibody in microdroplets. The collected sample was desalted and analyzed by nano-electrospray ionization mass spectrometry (nESI-MS, see the workflow shown in Figure 1b). Figure 1c shows that all the IdeS-cleaved subunits (i.e., F(ab')2 and scFc) were clearly observed in the MS spectra. In addition, we efficiently detected and resolved different glycoforms (G0F, G1F, and G2F) of scFc fragments (sequence coverage 100%). Furthermore, we did not observe any remaining intact IgG1 after microdroplet digestion, which indicated 100% digestion efficiency in microdroplets. In addition, 100% digestion efficiency was obtained during 9 individual runs, suggesting the high reproducibility of IdeS digestion of IgG1 in microdroplets (see deconvoluted MS spectra in Figure S1, Supporting Information). The time of 250 μs in microdroplets for completely digesting an intact antibody at 25°C represented a 7.5 million-fold speed improvement, in comparison with traditional digestion in bulk solution that requires at least 30 minutes to 1 hour at 37°C.45 To prove that IgG1 was truly digested in microdroplets during the spray process (Figure 1b), we also conducted a control experiment in which 1 mg/mL IgG1 was mixed and incubated with 2 units/μL IdeS (the same concentrations as used in the microdroplet digestion) in solution for digestion at 37°C for 5 min and no notable digestion was observed (Figure 1d). Clearly, the microdroplets did markedly accelerate enzymatic digestion of the intact antibody. Note that, in Figure 1b of the microdroplet experiment, the residual time of the IgG1 and IdeS in the 2-cm connection capillary after Tee mixing was only 4.4 seconds (see calculation in Supporting Information), which we did not expect to lead to noticeable digestion inside the capillary. Furthermore, we halved the amount of IdeS in the microdroplet digestion of IgG1 (1 unit IdeS for 1 μg of IgG1) and 100% digestion efficiency was still obtained in 9 individual runs (Figure S2, Supporting Information).
Strikingly, we found that adding another reagent into the spray solvent enabled another ultrafast reaction step. For example, addition of the reducing reagent TCEP in the spray solution containing IdeS and IgG1 accelerated disulfide bond reduction of antibody, leading to simultaneous digestion and reduction of antibody in the microdroplets, which is illustrated in Figure 2a. Following the workflow shown in Figure 2b, IgG1 and TCEP were pre-mixed in one syringe, and the IdeS was pre-loaded in the other syringe. Both syringes were pumped at a flow rate of 5 μL/min. Reactants were mixed via a Tee and transferred to the sprayer for generating microdroplets, which were again directed to a collection vial containing an acidic quenching solution (Figure 2b). After digestion in the microdroplets, we clearly observed in the MS spectrum all the digested and reduced fragments, LC, Fd′, and glycosylated scFcs including scFc+G0F, scFc+G1F, and scFc+G2F (Figure 2c) and the deconvoluted MS spectrum (Figure 2d), with mass of each fragment within the measurement accuracy (Table S1, Supporting Information). In addition, in the full deconvoluted MS spectrum (Figure S3, Supporting Information), neither intact antibody nor F(ab')2 peaks were seen, suggesting that both digestion and reduction were complete. In antibody characterization, reduced fragments are valuable because they are amenable to MS/MS techniques that provide high-sequence coverage and localization of structural modifications.
Besides conducting the microdroplet reaction in an “one-pot reaction” manner as described above, the dual reaction of digestion and reduction could be performed stepwise. According to the stepwise workflow shown in Scheme S1a (Supporting Information), we mixed IgG1 and IdeS via a Tee mixer and sprayed the mix as microdroplets to digest IgG1. The collected microdroplets were then mixed with TCEP in a vial and sprayed again to accelerate reduction in microdroplets. The F(ab′)2 fragment was further reduced to light chain (LC) and Fd′ fragment. The collected microdroplets were reconstituted and desalted for nESI-MS analysis. Figure S4 (Supporting Information) shows that all the IdeS-cleaved and TCEP-reduced subunits (i.e., LC, Fd′, and scFcs) were clearly observed in the MS spectra. Different glycoforms (G0F, G1F, and G2F) of scFc fragments were also detected and resolved efficiently. In addition, this workflow successfully digested and reduced IgG1 in phosphate buffer (see Figure S5, Supporting Information), which indicated that microdroplet digestion could be performed in a nonvolatile buffer solution. Likewise, for the stepwise workflow, we could mix IgG1 and TCEP first for microdroplet reduction and then further mixed with IdeS for digestion (Scheme S1b, Supporting Information). All the IdeS-cleaved and TCEP-reduced fragments (LC, Fd′ and scFcs) were also observed (Figure S6, Supporting Information), suggesting that the reduction and digestion reaction order can be switched.
To accelerate further the analysis process, one can use online MS detection in combination with simultaneous microdroplet digestion and reduction. It would be worthwhile to conduct microdroplet digestion and MS detection simultaneously as a real-time analysis because simultaneous operation would condense the digestion and reduction time to a minimum.14 For the online workflow, extractive electrospray ionization mass spectrometry (EESI-MS)58-61 was used, in which two sprayers were aligned and used together in the front of the MS inlet (Figure 2e). One sprayer served as the sample sprayer to generate microdroplets containing a mix of IgG1, IdeS and TCEP without applying any voltage. The other sprayer worked as a solvent sprayer to generate microdroplets of ACN/H2O/FA (50:50:0.5%) with +3 kV voltage applied to produce charged solvent microdroplets to assist the ionization of resulting fragments contained in the sample spray microdroplets. The IgG1 was cleaved with IdeS and reduced with TCEP into smaller fragments at the same time in the sample microdroplets in air at room temperature. Charged microdroplets of solvent mixed with and contacted the sample microdroplets and enabled sample ionization; the LC, Fd′, scFc+G0F, scFc+G1F, and scFc+G2F fragments were all detected by online mass spectrometric analysis (Figure 2f). This approach allowed us to online detect digested and reduced IgG1 fragments directly from the sprayed microdroplets, without any sample pretreatment or preparation. We noted a peak at 24,402 Da (asterisked in Figure 2f) corresponding to a harmonic peak of Fab (see the deconvoluted MS spectrum in Figure S7a, Supporting Information), and a small F(ab′)2 peak (Figure S7b, Supporting Information) as well as a small intact antibody peak (Figure S7c, Supporting Information). It is likely that the IdeS digestion of antibody and TCEP reduction for F(ab′)2 were not complete, due to the shortened travel distance of the sample microdroplets in the EESI experiment (intercepted by the EESI solvent spray before traveling 20 mm to reach MS inlet), in comparison the offline experiment described in Figure 2b. Note that the ion intensity of antibody fragments in the online EESI experiment was lower (Figure 2f) in comparison to the offline experiment (Figure 2d). One possible reason is that more water and less organic solvents were sprayed into the MS inlet when using an online EESI-MS approach. Another possible reason is that the excess amount of TCEP and ammonium bicarbonate could reduce the ion signal of antibody fragments as there was no sample purification step in this online process. One possible solution for this issue might be to perform online desalting62 by increasing organic solvent composition in the solvent spray of EESI so that salts from the sample droplets could be excluded in the newly fused secondary droplets for ion generation.
Glycation is a nonenzymatic protein modification that occurs between reducing sugars (e.g., glucose, fructose, and galactose) and proteins.63 The primary amine groups of lysine residues and the N-termini of proteins react with aldehyde groups from reducing sugars to form Schiff base intermediates, which are further converted via Amadori rearrangement into more stable ketoamines. Glycation has been commonly observed in therapeutic antibodies during manufacturing and storage, and glycation could affect drug product stability, safety, and efficacy.64 Thus, it is of major importance to characterize and determine the effect of glycation on mAb structure. However, conventional methods, such as boronate affinity chromatography, capillary isoelectric focusing (cIEF), and liquid chromatography-mass spectrometry, suffer from low throughput because of the long separation times required. Fit-for-purpose assays are still lagging to meet the high-throughput demands of the pharmaceutical industry. It has been stated that a top priority of the US Food and Drug Administration is development of new tools for high-throughput therapeutic protein characterization.65
Consequently, we were motivated to attempt a rapid characterization method for glycated IgG1 using microdroplet digestion and reduction. After incubation with glucose, the glycated NIST IgG1 underwent microdroplet digestion and reduction in the “one-pot reaction” workflow (i.e., mixing with both IdeS and TCEP) and Figure 3a-c presents the MS spectra of IgG1 obtained after 0, 2 and 5 days of incubation with glucose. Figure 3a shows the digested and reduced IgG1 fragments from IgG1 sample without incubation with glucose. After 2-days incubation, mono-glycated LC and mono-glycated Fd′ fragments were seen (Figure 3b). With 5-days continued incubation, di-glycated LC and di-glycated Fd′ fragments were also detected and resolved in the MS spectra (Figure 3c). In addition, after 5-days incubation, both mono-glycated LC and Fd′ fragments appeared to be more abundant (Figure 3c), in comparison to the sample after 2-days incubation (Figure 3b). With a longer incubation time, more glycated fragments were produced and detected (data not shown). All the antibody fragments including mono- and di-glycated LC and Fd′ fragments were detected and resolved in the deconvoluted MS spectrum (Figure 3d) and 2D spectrum (Figure 3e). The microdroplet-based digestion facilitates rapid glycation analysis of mAb subunits subjected to glycation stress. Microdroplet digestion of glycated IgG constitutes a proof-of-concept of a method to analyze other modifications of mAbs, such as oxidation, small molecule or peptide conjugation, and sequence variations, truncations, and extensions. Thus, the mAb subunits generated via microdroplets add to the toolbox of methods available for antibody characterization and product quality assessment during development and manufacturing.
Figure 3.
Microdroplet digestion and reduction for analysis of glycated IgG1. In this “one-pot reaction” approach, glycated IgG1, IdeS and TCEP were mixed and reacted together. The subunits of LC (and glycated-LC), Fd′ (and glycated-Fd′) and scFc fragments were produced simultaneously. (a) The expanded view of MS spectrum of non-glycated IgG1 (at day 0); (b) The MS spectrum of glycated IgG1 (incubated with glucose for 2 days); (c) The MS spectrum of glycated IgG1 (incubated with glucose for 5 days); (d) The deconvoluted MS spectrum of glycated IgG1 after 5-days incubation and microdroplet reactions; and (e) The 2D spectrum of glycated IgG1 after 5-days incubation and microdroplet reactions, with m/z along the x-axis and charge number along the y-axis.
Antibody glycosylation is heterogeneous, and variables in cell culture can increase glycan diversity. A conserved N-glycan at Asn297 of the scFc region of IgG1 is critical for stability, conformation, aggregation, and effector function of therapeutic antibodies.66 Removing glycosylation during pharmaceutical production analysis would be beneficial to ease the characterization of antibodies and to obtain the correct N-linked oligosaccharide (N-glycan) profile. N-Glycosidase F (PNGase F), a recombinant glycosidase from Elizabethkingia miricola, is one of the most effective enzymes to cleave N-glycans from proteins (illustrated in Figure 4a). In our experiment, first we pre-loaded IgG1 and PNGase F in two syringes, respectively (see the workflow in Figure S8, Supporting Information). Both syringes were pumped at a flow rate of 5 μL/min and reactants were mixed by a Tee and then sprayed to generate microdroplets. The microdroplets were quenched with H2O containing 1% formic acid in the collection vial and collected for 5 min. The distance between sprayer and quenching solution was kept as 2 cm. The collected sample was desalted and analyzed by nESI-MS. Figure S8c (Supporting Information) shows the MS spectrum of deglycosylation of IgG1 by PNGase F in microdroplets at ambient temperature. In comparison to the intact IgG1 MS spectrum (Figure S8d, Supporting Information), new antibody peaks appeared (Figure S8c, Supporting Information). The deconvoluted MS spectrum (Figure S8b, Supporting Information) clearly presents deglycosylated IgG1 peaks and a small amount of IgG1 with one glycan attached. This analysis indicated a high microdroplet reaction efficiency of deglycosylation. The deglycosylation yield for producing fully deglycosylated IgG1 was estimated to be 96.1%±3.2% (calculated using the ratio of fully deglycosylated IgG1 peak intensity and the sum of the fully deglycosylated IgG1 and the partially deglycosylated IgG1 peak intensities in Figure S8b), based on 5 individual runs.
We further tested a stepwise workflow in which the IgG1 was first deglycosylated by PNGase F and then reduced and digested by addition of IdeS and TCEP at a second step (Figure S9a). In this workflow, the IgG1 and PNGase F were mixed via a Tee and sprayed as microdroplets to remove N-glycans from IgG1. Then, the collected microdroplets were mixed with IdeS and TCEP and sprayed again to accelerate reduction and digestion in the microdroplets. The deglycosylated IgG1 was reduced and digested into LC, Fd′, and deglycosylated scFc fragments. The collected microdroplets were reconstituted and desalted for nESI-MS analysis. Figure S9b shows that all the fragments were observed clearly in the MS spectra. In addition, we did not observe any glycosylated (G0F, G1F, and G2F) scFc fragments; thus, deglycosylation in the microdroplets appeared to be 100% efficient, a fact that was confirmed in the deconvoluted MS spectrum (Figure S9c). Therefore, by conducting deglycosylation in the microdroplets, non-glycosylated antibody fragments can be obtained for rapid structure characterization, thereby avoiding the influence of the complex profiles of the N-glycans.
We went on to perform an “one-pot reaction” test, in which IgG1, PNGase F, IdeS and TCEP were mixed and sprayed for triggering simultaneously deglycosylation, digestion, and reduction in microdroplets. We observed the LC, Fd′, and scFc (both deglycosylated and glycosylated) fragments by nESI-MS analysis after microdroplet reaction and sample collection (Figures 4b and 4c). Although deglycosylation of scFc was not complete in this trial (38.0%±2.6% deglycosylation efficiency in 8 individual runs), the results indicated that antibody digestion, deglycosylation, and reduction can occur simultaneously in the microdroplets in a short amount of time (250 μs).
Conclusions
Currently, microdroplet reactions have been applied to many small molecule substrates, for various purposes including 1) synthesis of organic products and nanoparticles17,67, 2) kinetic measurements14, 3) rapid derivatization1,68 or degradation69, and 4) chemical behavior and fundamental studies18. Typically, only one-step reaction was involved in these reported microdroplet reactions. Biochemical reactions in microdroplets were rarely reported so far, with exception of the syntheses of sugar phosphates and uridine ribonucleoside via condensation19 and trypsin digestion of peptides and small proteins in microdroplets.34 Our work represents the first report of multiple-step reactions involving a large protein of intact antibody in microdroplets.
In summary, we demonstrated ultrafast IdeS digestion of intact antibody in microdroplets. We achieved and detected digestion in 250 microseconds, a vastly improved speed compared with conventional in-solution digestion. Further ultrafast reduction is feasible by doping the spray solvent with TCEP, leading to the IdeS-cleaved and TCEP-reduced fragments (LC, Fd′ and scFc with different glycoforms). Moreover, we have shown that microdroplet digestion can be achieved in both volatile and nonvolatile aqueous buffers; thus, other biochemical reactions are likely to occur in microdroplets. In addition, glycated IgG1 after incubation with glucose was digested successfully in microdroplets and analyzed by MS. Furthermore, microdroplet reaction coupled with online EESI-MS analysis would be an alternative approach having the least time for digestion, reduction and detection without sample pre-treatment. We also achieved rapid deglycosylation of IgG1 by PNGase F enzyme in the microdroplets. Although only NIST IgG1 was selected as a test sample, we believe that all these workflows and approaches are general and will greatly aid efforts to reduce analysis time and improve throughput of mAb characterization.
Supplementary Material
Acknowledgements
We thank the National Science Foundation (CHE-1915878) and National Institute of Health (1R15GM137311-01) for funding. The authors also acknowledge support from Janssen Research & Development.
Footnotes
Conflicts of interest
There are no conflicts to declare.
Supporting Information: additional MS spectra, experimental setup and experimental description.
References
- (1).Girod M; Moyano E; Campbell DI; Cooks RG Accelerated Bimolecular Reactions in Microdroplets Studied by Desorption Electrospray Ionization Mass Spectrometry. Chem. Sci 2011, 2 (3), 501–510. [Google Scholar]
- (2).Lee JK; Walker KL; Han HS; Kang J; Prinz FB; Waymouth RM; Nam HG; Zare RN Spontaneous Generation of Hydrogen Peroxide from Aqueous Microdroplets. Proc. Natl. Acad. Sci. U. S. A 2019, 116 (39), 19294–19298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (3).Lee JK; Samanta D; Nam HG; Zare RN Micrometer-Sized Water Droplets Induce Spontaneous Reduction. J. Am. Chem. Soc 2019, 141 (27), 10585–10589. [DOI] [PubMed] [Google Scholar]
- (4).Wei Z; Li Y; Cooks RG; Yan X Accelerated Reaction Kinetics in Microdroplets: Overview and Recent Developments. Annu. Rev. Phys. Chem 2020, 71, 31–51. [DOI] [PubMed] [Google Scholar]
- (5).Yan X; Bain RM; Cooks RG Organic Reactions in Microdroplets: Reaction Acceleration Revealed by Mass Spectrometry. Angewandte Chemie - International Edition. Wiley-VCH Verlag; October 10, 2016, pp 12960–12972. [DOI] [PubMed] [Google Scholar]
- (6).Nie H; Wei Z; Qiu L; Chen X; Holden DT; Cooks RG High-Yield Gram-Scale Organic Synthesis Using Accelerated Microdroplet/Thin Film Reactions with Solvent Recycling. Chem. Sci 2020, 11 (9), 2356–2361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (7).Basuri P; Gonzalez LE; Morato NM; Pradeep T; Cooks RG Accelerated Microdroplet Synthesis of Benzimidazoles by Nucleophilic Addition to Protonated Carboxylic Acids. Chem. Sci 2020, 11, 12686–12694 [DOI] [PMC free article] [PubMed] [Google Scholar]
- (8).Li Y; Mehari TF; Wei Z; Liu Y; Cooks RG Reaction Acceleration at Air-Solution Interfaces: Anisotropic Rate Constants for Katritzky Transamination. J. Mass Spectrom 2020, e4585. [DOI] [PubMed] [Google Scholar]
- (9).Li Y; Yan X; Cooks RG The Role of the Interface in Thin Film and Droplet Accelerated Reactions Studied by Competitive Substituent Effects. Angew. Chemie Int. Ed 2016, 55 (10), 3433–3437. [DOI] [PubMed] [Google Scholar]
- (10).Sahraeian T; Kulyk DS; Badu-Tawiah AK Droplet Imbibition Enables Nonequilibrium Interfacial Reactions in Charged Microdroplets. Langmuir 2019, 35 (45), 14451–14457. [DOI] [PubMed] [Google Scholar]
- (11).Tang S; Cheng H; Yan X On-Demand Electrochemical Epoxidation in Nano-Electrospray Ionization Mass Spectrometry to Locate Carbon–Carbon Double Bonds. Angew. Chemie Int. Ed 2020, 59 (1), 209–214. [DOI] [PubMed] [Google Scholar]
- (12).Vaitheeswaran S; Thirumalai D Hydrophobic and Ionic Interactions in Nanosized Water Droplets. J. Am. Chem. Soc 2006, 128 (41), 13490–13496. [DOI] [PubMed] [Google Scholar]
- (13).Sahota N; Abusalim DI; Wang ML; Brown CJ; Zhang Z; El-Baba TJ; Cook SP; Clemmer DE A Microdroplet-Accelerated Biginelli Reaction: Mechanisms and Separation of Isomers Using IMS-MS. Chem. Sci 2019, 10 (18), 4822–4827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (14).Lee JK; Kim S; Nam HG; Zare RN Microdroplet Fusion Mass Spectrometry for Fast Reaction Kinetics. Proc. Natl. Acad. Sci. U. S. A 2015, 112 (13), 3898–3903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (15).Cheng H; Tang S; Yang T; Xu S; Yan X Accelerating Electrochemical Reactions in a Voltage-Controlled Interfacial Microreactor. Angew. Chemie Int. Ed 2020, 59 (45), 19862–19867. [DOI] [PubMed] [Google Scholar]
- (16).Bain RM; Sathyamoorthi S; Zare RN “On-Droplet” Chemistry: The Cycloaddition of Diethyl Azodicarboxylate and Quadricyclane. Angew. Chemie 2017, 129 (47), 15279–15283. [DOI] [PubMed] [Google Scholar]
- (17).Lee JK; Samanta D; Nam HG; Zare RN Spontaneous Formation of Gold Nanostructures in Aqueous Microdroplets. Nat. Commun 2018, 9, 1562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (18).Gnanamani E; Yan X; Zare RN Chemoselective N-Alkylation of Indoles in Aqueous Microdroplets. Angew. Chemie Int. Ed 2020, 59 (8), 3069–3072. [DOI] [PubMed] [Google Scholar]
- (19).Nam I; Lee JK; Nam HG; Zare RN Abiotic Production of Sugar Phosphates and Uridine Ribonucleoside in Aqueous Microdroplets. Proc. Natl. Acad. Sci. U. S. A 2017, 114 (47), 12396–12400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (20).Banerjee S; Zare RN Syntheses of Isoquinoline and Substituted Quinolines in Charged Microdroplets. Angew. Chemie - Int. Ed 2015, 54 (49), 14795–14799. [DOI] [PubMed] [Google Scholar]
- (21).Gao D; Jin F; Lee JK; Zare RN Aqueous Microdroplets Containing Only Ketones or Aldehydes Undergo Dakin and Baeyer-Villiger Reactions. Chem. Sci 2019, 10 (48), 10974–10978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (22).Luo K; Li J; Cao Y; Liu C; Ge J; Chen H; Zare RN Reaction of Chloroauric Acid with Histidine in Microdroplets Yields a Catalytic Au-(His)2 Complex. Chem. Sci 2020, 11 (9), 2558–2565. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (23).Banerjee S; Gnanamani E; Yan X; Zare RN Can All Bulk-Phase Reactions Be Accelerated in Microdroplets? Analyst. Royal Society of Chemistry; May 7, 2017, pp 1399–1402. [DOI] [PubMed] [Google Scholar]
- (24).Marsh BM; Iyer K; Cooks RG Reaction Acceleration in Electrospray Droplets: Size, Distance, and Surfactant Effects. J. Am. Soc. Mass Spectrom 2019, 30 (10), 2022–2030. [DOI] [PubMed] [Google Scholar]
- (25).Mondal S; Acharya S; Biswas R; Bagchi B; Zare RN Enhancement of Reaction Rate in Small-Sized Droplets: A Combined Analytical and Simulation Study. J. Chem. Phys 2018, 148 (24), 244704. [DOI] [PubMed] [Google Scholar]
- (26).Xiong H; Lee JK; Zare RN; Min W Strong Electric Field Observed at the Interface of Aqueous Microdroplets. J. Phys. Chem. Lett 2020, 11, 7423–7428. [DOI] [PubMed] [Google Scholar]
- (27).Chamberlayne CF; Zare RN Simple Model for the Electric Field and Spatial Distribution of Ions in a Microdroplet. J. Chem. Phys 2020, 152 (18), 184702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (28).Tuck AF Gibbs Free Energy and Reaction Rate Acceleration in and on Microdroplets. Entropy 2019, 21 (11), 1044. [Google Scholar]
- (29).Xiong H; Lee JK; Zare RN; Min W Strong Concentration Enhancement of Molecules at the Interface of Aqueous Microdroplets. J. Phys. Chem. B 2020, 124 (44), 9938–9944. [DOI] [PubMed] [Google Scholar]
- (30).Chamberlayne CF; Zare RN; Santiago JG Effects of Weak Electrolytes on Electric Double Layer Ion Distributions. J. Phys. Chem. Lett 2020, 11 (19), 8302–8306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (31).Narendra N; Chen X; Wang J; Charles J; Cooks RG; Kubis T Quantum Mechanical Modeling of Reaction Rate Acceleration in Microdroplets. J. Phys. Chem. A 2020, 124 (24), 4984–4989. [DOI] [PubMed] [Google Scholar]
- (32).Zhou Z; Yan X; Lai YH; Zare RN Fluorescence Polarization Anisotropy in Microdroplets. J. Phys. Chem. Lett 2018, 9 (11), 2928–2932. [DOI] [PubMed] [Google Scholar]
- (33).Rovelli G; Jacobs MI; Willis MD; Rapf RJ; Prophet AM; Wilson KR A Critical Analysis of Electrospray Techniques for the Determination of Accelerated Rates and Mechanisms of Chemical Reactions in Droplets. Chem. Sci 2020, 11, 13026–13043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- (34).Zhong X; Chen H; Zare RN Ultrafast Enzymatic Digestion of Proteins by Microdroplet Mass Spectrometry. Nat. Commun 2020, 11, 1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (35).Tsumoto K; Isozaki Y; Yagami H; Tomita M Future Perspectives of Therapeutic Monoclonal Antibodies. Immunotherapy 2019, 11 (2), 119–127. [DOI] [PubMed] [Google Scholar]
- (36).Kaplon H; Muralidharan M; Schneider Z; Reichert JM Antibodies to Watch in 2020. MAbs 2020, 12 (1), 1703531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (37).Carter PJ; Lazar GA Next Generation Antibody Drugs: Pursuit of the “High-Hanging Fruit.” Nature Reviews Drug Discovery. Nature Publishing Group; February 28, 2018, pp 197–223. [DOI] [PubMed] [Google Scholar]
- (38).Fisher AC; Lee SL; Harris DP; Buhse L; Kozlowski S; Yu L; Kopcha M; Woodcock J Advancing Pharmaceutical Quality: An Overview of Science and Research in the U.S. FDA’s Office of Pharmaceutical Quality. International Journal of Pharmaceutics. Elsevier B.V. December 30, 2016, pp 390–402. [DOI] [PubMed] [Google Scholar]
- (39).Nagy G; Attah IK; Conant CR; Liu W; Garimella SVB; Gunawardena HP; Shaw JB; Smith RD; Ibrahim YM Rapid and Simultaneous Characterization of Drug Conjugation in Heavy and Light Chains of a Monoclonal Antibody Revealed by High-Resolution Ion Mobility Separations in SLIM. Anal. Chem 2020, 92 (7), 5004–5012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (40).Beck A; Wagner-Rousset E; Ayoub D; Van Dorsselaer A; Sanglier-Cianférani S Characterization of Therapeutic Antibodies and Related Products. Analytical Chemistry. American Chemical Society; January 15, 2013, pp 715–736. [DOI] [PubMed] [Google Scholar]
- (41).Kang L; Weng N; Jian W LC–MS Bioanalysis of Intact Proteins and Peptides. Biomed. Chromatogr 2020, 34 (1), e4633. [DOI] [PubMed] [Google Scholar]
- (42).Jin Y; Lin Z; Xu Q; Fu C; Zhang Z; Zhang Q; Pritts WA; Ge Y Comprehensive Characterization of Monoclonal Antibody by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. MAbs 2019, 11 (1), 106–115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (43).Donnelly DP; Rawlins CM; DeHart CJ; Fornelli L; Schachner LF; Lin Z; Lippens JL; Aluri KC; Sarin R; Chen B; Lantz C; Jung W; Johnson KR; Koller A; Wolff JJ; Campuzano IDG; Auclair JR; Ivanov AR; Whitelegge JP; Paša-Tolić L; Chamot-Rooke J; Danis PO; Smith LM; Tsybin YO; Loo JA; Ge Y; Kelleher NL; Agar JN Best Practices and Benchmarks for Intact Protein Analysis for Top-down Mass Spectrometry. Nat. Methods 2019, 16 (7), 587–594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (44).Lodge JM; Schauer KL; Brademan DR; Riley NM; Shishkova E; Westphall MS; Coon JJ Top-Down Characterization of an Intact Monoclonal Antibody Using Activated Ion Electron Transfer Dissociation. Anal. Chem 2020, 92 (15), 10246–10251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (45).An Y; Zhang Y; Mueller H-M; Shameem M; Chen X A New Tool for Monoclonal Antibody Analysis. MAbs 2014, 6 (4), 879–893. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (46).Huang L; Wang N; Mitchell CE; Brownlee T; Maple SR; De Felippis MR A Novel Sample Preparation for Shotgun Proteomics Characterization of HCPs in Antibodies. Anal. Chem 2017, 89 (10), 5436–5444. [DOI] [PubMed] [Google Scholar]
- (47).Fekete S; Guillarme D; Sandra P; Sandra K Chromatographic, Electrophoretic, and Mass Spectrometric Methods for the Analytical Characterization of Protein Biopharmaceuticals. Analytical Chemistry. American Chemical Society; January 5, 2016, pp 480–507. [DOI] [PubMed] [Google Scholar]
- (48).Mouchahoir T; Schiel JE Development of an LC-MS/MS Peptide Mapping Protocol for the NISTmAb. Anal. Bioanal. Chem 2018, 410 (8), 2111–2126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (49).Carvalho LB; Capelo-Martínez J-L; Lodeiro C; Wiśniewski JR; Santos HM Ultrasonic-Based Filter Aided Sample Preparation as the General Method to Sample Preparation in Proteomics. Anal. Chem 2020, 92 (13), 9164–9171. [DOI] [PubMed] [Google Scholar]
- (50).Rivera-Albarran ME; Ray SJ A Novel Combined Microstrip Resonator/Nanospray Ionization Source for Microwave-Assisted Trypsin Digestion of Proteins. J. Am. Soc. Mass Spectrom 2020, 31 (8), 1684–1696. [DOI] [PubMed] [Google Scholar]
- (51).Basile F; Hauser N Rapid Online Nonenzymatic Protein Digestion Combining Microwave Heating Acid Hydrolysis and Electrochemical Oxidation. Anal. Chem 2011, 83 (1), 359–367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (52).López-Ferrer D; Capelo JL; Vázquez J Ultra Fast Trypsin Digestion of Proteins by High Intensity Focused Ultrasound. J. Proteome Res 2005, 4 (5), 1569–1574. [DOI] [PubMed] [Google Scholar]
- (53).Russell WK; Park ZY; Russell DH Proteolysis in Mixed Organic-Aqueous Solvent Systems: Applications for Peptide Mass Mapping Using Mass Spectrometry. Anal. Chem 2001, 73 (11), 2682–2685. [DOI] [PubMed] [Google Scholar]
- (54).Ji J; Zhang Y; Zhou X; Kong J; Tang Y; Liu B Enhanced Protein Digestion through the Confinement of Nanozeolite-Assembled Microchip Reactors. Anal. Chem 2008, 80 (7), 2457–2463. [DOI] [PubMed] [Google Scholar]
- (55).Bark SJ; Muster N; Yates JR; Siuzdak G High-Temperature Protein Mass Mapping Using a Thermophilic Protease. J. Am. Chem. Soc 2001, 123 (8), 1774–1775. [DOI] [PubMed] [Google Scholar]
- (56).Wang Y; Zhang W; Ouyang Z Fast Protein Analysis Enabled by High-Temperature Hydrolysis. Chem. Sci 2020, 11 (38), 10506–10516. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (57).Takáts Z; Wiseman JM; Gologan B; Cooks RG Electrosonic Spray Ionization. A Gentle Technique for Generating Folded Proteins and Protein Complexes in the Gas Phase and for Studying Ion-Molecule Reactions at Atmospheric Pressure. Anal. Chem 2004, 76 (14), 4050–4058. [DOI] [PubMed] [Google Scholar]
- (58).Chen H; Yang S; Li M; Hu B; Li J; Wang J Sensitive Detection of Native Proteins Using Extractive Electrospray Ionization Mass Spectrometry. Angew. Chemie Int. Ed 2010, 49 (17), 3053–3056. [DOI] [PubMed] [Google Scholar]
- (59).Chen H; Venter A; Cooks RG Extractive Electrospray Ionization for Direct Analysis of Undiluted Urine, Milk and Other Complex Mixtures without Sample Preparation. Chem. Commun 2006, No. 19, 2042–2044. [DOI] [PubMed] [Google Scholar]
- (60).Chen H; Gamez G; Zenobi R What Can We Learn from Ambient Ionization Techniques? J. Am. Soc. Mass Spectrom 2009, 20 (11), 1947–1963. [DOI] [PubMed] [Google Scholar]
- (61).Chen H; Yang S; Wortmann A; Zenobi R Neutral Desorption Sampling of Living Objects for Rapid Analysis by Extractive Electrospray Ionization Mass Spectrometry. Angew. Chemie - Int. Ed 2007, 46 (40), 7591–7594. [DOI] [PubMed] [Google Scholar]
- (62).Miao Z; Chen H Direct Analysis of Liquid Samples by Desorption Electrospray Ionization-Mass Spectrometry (DESI-MS). J. Am. Soc. Mass Spectrom 2009, 20 (1), 10–19. [DOI] [PubMed] [Google Scholar]
- (63).Quan C; Alcala E; Petkovska I; Matthews D; Canova-Davis E; Taticek R; Ma S A Study in Glycation of a Therapeutic Recombinant Humanized Monoclonal Antibody: Where It Is, How It Got There, and How It Affects Charge-Based Behavior. Anal. Biochem 2008, 373 (2), 179–191. [DOI] [PubMed] [Google Scholar]
- (64).Wei B; Berning K; Quan C; Zhang YT Glycation of Antibodies: Modification, Methods and Potential Effects on Biological Functions. mAbs. Taylor and Francis Inc. May 19, 2017, pp 586–594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (65).Woodcock J; Woosley R The FDA Critical Path Initiative and Its Influence on New Drug Development. Annu. Rev. Med 2008, 59 (1), 1–12. [DOI] [PubMed] [Google Scholar]
- (66).Jefferis R Glycosylation as a Strategy to Improve Antibody-Based Therapeutics. Nature Reviews Drug Discovery. Nature Publishing Group; 2009, pp 226–234. [DOI] [PubMed] [Google Scholar]
- (67).Espy RD; Wleklinski M; Yan X; Cooks RG Beyond the Flask: Reactions on the Fly in Ambient Mass Spectrometry. TrAC - Trends in Analytical Chemistry. Elsevier B.V. May 1, 2014, pp 135–146. [Google Scholar]
- (68).Chen H; Cotte-Rodríguez I; Cooks RG Cis-Diol Functional Group Recognition by Reactive Desorption Electrospray Ionization (DESI). Chem. Commun 2006, 6 (6), 597–599. [DOI] [PubMed] [Google Scholar]
- (69).Li Y; Hu Y; Logsdon DL; Liu Y; Zhao Y; Cooks RG Accelerated Forced Degradation of Therapeutic Peptides in Levitated Microdroplets. Pharm. Res 2020, 37 (7), 138. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.




