Research Article Received: 10 July 2013
Revised: 3 September 2013
Accepted: 4 September 2013
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664 (wileyonlinelibrary.com) DOI: 10.1002/rcm.6733
Systematic optimization of ion-pairing agents and hexafluoroisopropanol for enhanced electrospray ionization mass spectrometry of oligonucleotides A. Cary McGinnis, E. Claire Grubb and Michael G. Bartlett* Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, The University of Georgia, Athens, GA 30602-2352, USA RATIONALE: New methods to enhance the electrospray ionization (ESI) signals are essential for low-level analysis of oligonucleotides. We report a systematic evaluation comparing 13 ion-pairing agents with and without hexafluoroisopropanol to understand their effect on the ion abundance of hetero-oligonucleotides. METHODS: A Waters Synapt G2 HDMS quadrupole time-of-flight instrument was used to compare oligonucleotide signal intensity with 13 alkylamine ion-pairing agents at varying concentrations. The alkylamines that yielded the highest signal intensity were further evaluated with hexafluoroisopropanol at concentrations between 5 and 100 mM. The chemical properties of the solution components and analytes were evaluated to identify key factors in predicting optimal mobile phase conditions for different classes of oligonucleotides. RESULTS: We identified a series of optimized mobile phase systems using diisopropylamine, tripropylamine, dimethylbutylamine, methyldibutylamine, and dimethylhexylamine along with 25 to 50 mM hexafluoroisopropanol that yielded significantly higher MS signal intensity for both siRNA and DNA compared with the traditionally used triethlyamine/hexafluoroisopropanol system. We explored charge state reduction, adduct formation and ESI mechanisms and identify the Henry’s Law constant k aq/g as a key chemical property in predicting alkylamines that will increase oligonucleotide ion intensity. We also find that the hydrophobicity of the oligonucleotide plays a major role in choosing ion-pairing agents that will increase ion abundance. CONCLUSIONS: This comprehensive and systematic optimization finds that the hydrophobicity of the oligonucleotide was a key factor in choosing alkylamine ion-pairing agents to increase ESI abundance. We identified that diisopropylamine and tripropylamine combined with lower concentrations of hexafluoroisopropanol yielded the highest signal intensity for these oligonucleotides. Copyright © 2013 John Wiley & Sons, Ltd.
Electrospray ionization mass spectrometry (ESI-MS) has become the gold standard for the analysis of macromolecules. It allows for quantification and characterization of a wide range of proteins and oligonucleotides. Great strides have been made in using ESI-MS to increase detection limits for proteins and peptides.[1] Still oligonucleotides can benefit from improvements to the ESI conditions so that these methods can be more effectively used for pharmacokinetic and toxicokinetic evaluations. Chromatographic separations of oligonucleotides initially used triethlyamine (TEA) with acetic acid as the counter ion.[2–4] This mobile phase provided good separation efficiency but lacked sensitivity with ESI-MS. Acetic acid has a higher boiling point than TEA. In the electrospray droplet, the TEA evaporated more rapidly. The acetate counter ion would then compete for ionization with the oligonucleotide thus lowering ionization efficiency. Dimethylbutylamine
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Copyright © 2013 John Wiley & Sons, Ltd.
2655
* Correspondence to: M. G. Bartlett, Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, The University of Georgia, Athens, GA 30602–2352, USA. E-mail: [email protected]
(DMBA) with a bicarbonate counter ion was shown to yield higher sensitivity than TEA with acetic acid for DNA.[5] In addition, DMBA was shown to have greater sensitivity than TEA with the bicarbonate counter ion. Because DMBA is more hydrophobic than TEA, it required a higher organic concentration to elute the oligonucleotides. The higher organic concentration reduced surface tension in the electrospray droplet providing up to 10 times greater sensitivity when compared to TEA with bicarbonate.[5] A number of studies have evaluated TEA with other counter ions.[6–11] Huber and Krajete evaluated acetate, bicarbonate, formate, and chloride counter ions.[11] In this comparison, the volatility of the counter ion did not correlate with increased signal intensity. Instead they found that acetate yielded the best sensitivity and argued that it was the lower conductivity of the counter ion that caused less suppression of oligonucleotide ionization. Other studies have focused on improving signal intensity through charge state reduction.[12–14] Charge state reduction involves shifting the charge state envelope to lower m/z values and increasing the intensity of the remaining charge states. Cheng et al. found that charge state reduction increased signal intensity due to the compression of the
A. C. McGinnis, E. C. Grubb and M. G. Bartlett charge states.[13] Muddiman et al. proposed a mechanism for charge state reduction in which the hydrogen-bound proton is shared as a dimer between the phosphodiester backbone and the counter ion.[14] Because of the higher proton affinity of the phosphodiester, the counter ion is lost as a neutral on entry, as well as, within the gas phase. Adduction of metal cations also lowers signal intensity for oligonucleotides.[14] Gaus et al. evaluated the signal intensity of a psDNA using seven alkylamines with no counter ion. This study focused on reducing adduct formation.[15] They found that tripropylamine (TPA) yielded the highest signal intensity. The most widely used mobile phase in the literature is based on the work of Apffel et al. in 1997.[16,17] This work introduced the use of triethlyamine (TEA) as an ion-pairing (IP) agent and hexafluoroisopropanol (HFIP) as a counter ion for ESI-MS of oligonucleotides. They proposed that the low boiling point of HFIP (58 °C) allowed it to evaporate from the electrosprayed droplet, which raised the pH of the droplet, allowing for enhanced ionization of oligonucleotides. Since that time, liquid chromatography (LC)/MS analysis of siRNAs and DNA has almost exclusively used different compositions of the TEA/HFIP mobile phase.[18–20] Recently, Chen et al. published a study on the effect of seven IP agents on the desorption and ionization of a phosphorothioate DNA.[21] They reported that diisopropylethylamine (DIEA) combined with HFIP yielded the highest sensitivity for psDNA. They also reported the bioanalysis of a psDNA in rat plasma with a lower limit of quantitation of 2.5 ng/mL.[22] We follow these studies with the evaluation of 13 IP agents and their effect on ion abundance of different classes of hetero-oligonucleotides. In this study, the alkylamines are evaluated at different concentrations alone and in the presence of HFIP. Mobile phase compositions are optimized for maximum ESI signal intensity for both siRNA and DNA. The implications of chemical properties and structure of both the alkylamines and the oligonucleotides are evaluated in order to provide a predictive model of mobile phase composition for modified oligonucleotides.
EXPERIMENTAL
2656
A 21mer siRNA duplex modified with DNA overhangs (sense 5’-GGA UCU CAU GCA CAU CUC AdTdT-3’, antisense 5’-UGA GAU GUG CAU GAG AUC CdTdG-3’) was purchased from Invitrogen (Life Technologies Carlsbad, CA, USA). The 24mer DNA (5’-TCGTGCTTTTGTTGTTTTC GCGTT-3’) was purchased from Integrated DNA Technologies (Coralville, IA, USA). LC/MS Chromasolv grade methanol and water were obtained from Fluka (St. Louis, MO, USA). Diisopropylamine (DIPA), diisopropylethylamine (DIEA), N’N’-dimethylbutylamine (DMBA), dimethylcyclohexylamine (DMCHA), dimethylhexylamine (DMHA), hexylamine (HA), N-methyldibutylamine (MDBA), octylamine (OA), propylamine (PA), tetramethylethylenediamine (TMEDA), triethylamine (TEA), tributylamine (TBA), tripropylamine (TPA), and 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) were purchased from Sigma–Aldrich Inc. (St. Louis, MO, USA). DNA Lobind centrifuge tubes were purchased from Eppendorf (Hauppauge, NY, USA).
wileyonlinelibrary.com/journal/rcm
Preparation of working solutions for infusion experiments Each alkylamine was prepared in 50:50 methanol/water at concentrations between 5 and 30 mM. Solutions containing HFIP were prepared between 5 and 100 mM in the varying concentrations of the IP agent (5–30 mM). The siRNA or DNA was prepared in each solution at a concentration of 10 μg/mL. Instrumental conditions Infusion experiments were performed on a Synapt G2 HDMS quadrupole time-of-flight hybrid mass spectrometer with an electrospray ionization (ESI) source (Waters, Milford, MA, USA). Chromatography was performed on an Acquity ultrahigh-performance liquid chromatography (uHPLC) system coupled with a Synapt G2 HDMS quadrupole timeof-flight hybrid mass spectrometer with an ESI source (Waters, Milford, MA, USA). Chromatographic separations were performed at a flow rate of 0.3 mL/min on a 1.7 μm Acquity BEH C18 column (Waters, Milford, MA, USA) 2.1 x 50 mm column. Mobile phase A consisted of 10 mM DIPA and 25 mM HFIP in water, and mobile phase B consisted of the same concentration of DIPA and HFIP in water/methanol (50:50). A 20 μL injection of each sample was loaded onto the column and separated using the following gradient conditions [time (min), % mobile phase B]: (0, 18) (2, 18) (8, 25). Gradient conditions to separate siRNA, DNA, and psDNA were [time (min), % mobile phase B]: (0, 18) (2, 18) (8, 32). The column temperature was maintained at 75 °C. The column eluent from 0–1 min was diverted to waste. The mass spectrometer was operated in negative-ion MS mode with a 1 s scan time. The mass spectrometer tuning was optimized for each different solution using the same source temperature, desolvation temperature and gas flows. For 10 mM DIPA/25 mM HFIP in 50% methanol, the tuning was as follows: capillary voltage 2.0 kV, cone voltage 25 V, extraction cone voltage 2 V, source temperature 125 °C, desolvation temperature 450 °C, cone gas 0 L/h and desolvation gas 1000 L/h. The data were collected in full-scan mode over the mass range 500–3000 m/z. All measurements were performed in triplicate. Base peak signal intensities were measured by combining 20 scans, and all data is represented as the mean ± the standard deviation.
RESULTS AND DISCUSSION Optimization of siRNA ESI-MS signal intensity Figures 1(a) and 1(b) show the base peak ion intensity of the sense and antisense strands with each alkylamine (N = 3). The ion-pairing (IP) agents are listed in the legends in order of decreasing signal intensity. As demonstrated previously, the siRNA denatures in the gas phase to its sense and antisense strands.[23] The first thing to note is the difference in maximum signal for the sense and antisense strands. The signal intensity for the antisense strand is more than 400,000 counts above that for the sense strand. Secondly, the alkylamines that provide the highest signal intensities are different for the sense and antisense strands. Only DIPA and DMBA are found in common among the top five IP agents for both strands. The difference in signal intensity is
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides
Figure 1. Comparison of the base peak ion intensity of siRNA in the presence of 13 alkylamines at various concentrations: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
alkylamines except for TPA. The optimized maximum signal intensities for the sense and antisense strands, respectively, are 10 mM DIPA, 25 mM HFIP – 1.2 × 106, 2.1 × 106, 10 mM DMBA, 25 mM HFIP – 1.0 × 106, 1.8 × 106, 10 mM TPA, 10 mM HFIP – 1.0 × 106, 1.7 × 106, followed by 15 mM TEA, 25 mM HFIP – 9.5 × 105, 1.6 × 106. The differences in signal are from 200,000 counts to as high as 500,000 counts. The optimized TEA/HFIP combination is as much as 500,000 counts lower than DIPA/ HFIP. The signal intensity differences of these optimized mobile phases would impact the low-level quantitation of siRNAs. Maximizing signal intensity must also be combined with adequate chromatographic separation. Each of the top alkylamines was combined with HFIP to find the best separation of siRNA and its chain-shortened metabolites from both the sense and antisense siRNA strands. The best separation, in addition to the maximum signal intensity, was provided with 10 mM DIPA, 25 mM HFIP (Fig. 4).[24] Chen et al. found that DIEA provided the highest signal intensity for a mixed base phosphorothioate DNA.[21] Interestingly, DIEA provided a much lower signal intensity for siRNA when compared with other alkylamines. The addition of sulfur to the phosphate backbone increases the hydrophobicity of a psDNA.[25] The siRNA used in these experiments would be significantly more hydrophilic. To further explore the effect of analyte hydrophobicity on ion abundance, we used a 24mer unmodified DNA with the same sequence as that used by Chen et al. and performed the same systematic evaluation of the effect of alkylamines and HFIP on signal intensity.
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2657
likely due to differences in the hydrophobicity of the single strands. These differences also seem to effect which alkylamine is most effective at increasing the signal. The role of analyte hydrophobicity is discussed further in a later section. Five of the top-performing IP agents for the sense and antisense strands were evaluated with different concentrations of HFIP. Each alkylamine was tested at concentrations between 5 and 25 mM with concentrations of HFIP at 10, 25, 50 and 100 mM. Figure 2 shows the base peak signal intensity for each alkylamine with varying concentrations of HFIP. Only the concentration of alkylamine that provided the best signal intensity when combined with HFIP is shown in the graph. The addition of HFIP doubles the signal intensity for both the sense and antisense strands for all of the alkylamines except octylamine. Interestingly, the signal intensity is suppressed by almost one-half by the addition of HFIP to octylamine. Figure 3 compares the base peak ion intensity of siRNA with 25 mM HFIP to show the optimization of alkylamine concentration with a fixed concentration of HFIP. Most of the alkylamines show an optimum concentration where the signal intensity is the highest. DMBA, however, shows a close performance between the 10 and 15 mM concentration in 25 mM HFIP. This would allow optimization of the DMBA concentration to enhance chromatographic separations while maintaining high signal intensity. Figure 2 provides further information about the remaining alkylamines and the effect of HFIP combination. A concentration of 25 mM HFIP provided the maximum signal intensity for the
Figure 2. Comparison of the base peak ion intensity of siRNA with alkylamines and various concentrations of HFIP: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
A. C. McGinnis, E. C. Grubb and M. G. Bartlett Optimization of DNA ESI-MS signal intensity Figure 5 shows the base peak ion intensity for each alkylamine for a 10 μg/mL solution of unmodified DNA. MDBA, TBA, DMHA, and TPA yielded the highest signal intensities. Interestingly, these are different than those optimized for siRNA except for TPA. Signal intensities for DIPA, TEA, and DIEA that yielded higher intensities for siRNA are essentially equivalent for DNA and are approximately 600,000 counts below the highest signal for MDBA. Optimization continued by evaluating the alkylamines that yielded the highest signal intensity with varying concentrations of HFIP following the same protocol as with siRNA. Figure 6 shows the base peak signal intensity for the alkylamines at the concentration that provided the highest signal intensity when combined with HFIP. The highest signal intensity was provided with 15 mM TPA with 50 mM HFIP; 15 mM MDBA, 25 mM HFIP followed at 200,000 counts lower, while 10 mM TBA, 5 mM HFIP is 400,000 counts lower than the TPA/HFIP combination. Once again, these differences in signal intensities for the alkylamine/HFIP combinations would favorably impact low-level analysis of unmodified DNAs. Supplementary Fig. S1 (see Supporting Information) shows the optimization of the alkylamine concentration with fixed concentrations of HFIP (25 and 50 mM). We see an optimal combination of alkylamine concentration with either 25 or 50 mM with both TPA and MDBA. The ion intensity signals for both TBA and DMHA are far lower and the differences in signal intensity are not as great. Figure 3. Comparison of the base peak ion intensity of siRNA in 25 mM HFIP to optimize the concentration of alkylamines. Note that no signal was obtained with 5 mM TEA in 25 mM HFIP: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
2658
Figure 4. Ion-pairing reversed-phase chromatography with 1.7 μm Acquity BEH C18 column (2.1 × 50 mm; Waters, Milford, MA, USA). After a 2 min hold, a linear gradient from 18–25% B was applied over 6 min. Mobile phase A consisted of 10 mM DIPA and 25 mM HFIP in water, and mobile phase B consisted of the same concentration of DIPA and HFIP in water/methanol (50:50).
wileyonlinelibrary.com/journal/rcm
Effects of alkylamines on adduction and charge state distribution The charge state distribution was the same for siRNA in the presence of the top four alkylamines alone (data not shown). All have base peaks at the 12 charge state for both the sense and antisense strands. Figure 7 shows the charge state envelope for siRNA with DIPA, TEA, DMBA, and TPA combined with HFIP. The charge state is increased (moved to higher m/z) in the presence of HFIP to 10 and 11 for the antisense and sense strand, respectively. Cheng et al. found that charge state reduction increased signal intensity due to the compression of the charge states.[13] The presence
Figure 5. Comparison of the base peak ion intensity of DNA in the presence of 13 alkylamines at various concentrations. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides or neutrals rather than remaining with the oligonucleotide as a metal cation would. This lowers adduction, simplifies the mass spectrum, and significantly increases the signal.[26] The inserts on Fig. 7 show the expanded antisense base peak and associated adduction. While differences in adduction were seen with the IP agents alone (data not shown), this difference did not appear to significantly change the spectra for these alkylamines when combined with HFIP. This trend was slightly different for DNA (Fig. 8). Increased adduction was seen with TPA and MDBA compared with TBA and DMHA. Despite a slight increase in adduction, TPA and MDBA still provided higher signal intensity for the DNA. Figure 6. Comparison of the base peak ion intensity of DNA with alkylamines and various concentrations of HFIP. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Structural and chemical properties of alkylamines and their relation to electrospray ionization
of HFIP does not reduce the number of charge states, but it does compress the charge state envelope for siRNA, by providing more intense signals for the sense and antisense strands at the 10 and 11 charge states. The addition of HFIP lowers the pH slightly and provides more protons in the solution. The more acidic solution allows more protons to be donated to the oligonucleotide anions thereby altering the charge state. Interestingly, a shift in the charge state was not observed with the addition of HFIP to alkylamines for DNA (Fig. 8). Adduction of metal cations may also lower signal intensity for oligonucleotides.[14] Ion-pairing agents act to reduce adduction by pairing with the negatively charged backbone. The IP agents are evaporated as positively charged molecules
Figure 9 shows the structures for the top alkylamines without HFIP for the siRNA antisense and sense strands, DNA and psDNA.[21] The top alkylamines for the RNA strands include primary, secondary, and tertiary amines. Commonalities between the oligonucleotides include DIPA and DMBA, which are secondary and tertiary amines, respectively. The structures that provided the highest signal intensity for both DNA and psDNA were all tertiary amines and have no commonalities, although the evaluation for psDNA by Chen et al. did not include DMHA or DMCHA. Figure 10 shows the structures of the top alkylamines combined with HFIP. These structures were all tertiary amines except for DIPA. Commonalities between siRNA, DNA and psDNA included TPA and TEA. Observations of the chain length for top groups appear to favor longer chain length alkylamines for the DNA where TPA, MDBA, TBA
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2659
Figure 7. Mass spectra of siRNA with the alkylamines with HFIP that yielded the highest signal intensity. The base peaks for the sense (S) and antisense (AS) strands are highlighted. The insets display the base peak charge states to illustrate adduction.
A. C. McGinnis, E. C. Grubb and M. G. Bartlett
Figure 8. Mass spectra of DNA with the alkylamines and HFIP that yielded the highest signal intensity. The insets display the base peak charge states to illustrate adduction.
Figure 9. Structures of the ion-pairing agents that yielded the highest signal intensity for siRNA single strands, DNA, and psDNA.
2660
and DMHA provide higher intensities. The chemical properties of each alkylamine were also evaluated for their impact on ionization efficiency. Table 1 provides many of the chemical properties for the alkylamines used in the optimization procedures.[27,28] The production of gas-phase ions by ESI depends heavily on the chemical properties of the solution components including the analyte.[29] The first steps are production of charged droplets from the capillary tip, then shrinkage of the charged droplets by solvent evaporation and droplet disintegration. The final step is the production of ions from the charged droplets.[30–32] The surface tension and conductivity effect the initial droplet production and
wileyonlinelibrary.com/journal/rcm
droplet disintegration. The organic portion of the mobile phase (50% methanol in these experiments) largely governs the surface tension.[33,34] The concentration of the solutes can affect the conductivity, which can be measured by pH. In these experiments, the pH was between 9 and 9.5 for all of the top performing alkylamines (data not shown). As the pKa data shown in Table 1 indicate, there is little difference between these solutes. When HFIP is added, the pH drops to between 8.0 and 9.0 (data not shown). The addition of the weak acid, HFIP, increases the conductivity and therefore reduces the droplet size emitted from the capillary.[32] This reduction in the ESI droplet size would increase ionization and this would
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides
Figure 10. Structures of the ion-pairing agents that yielded the highest signal intensity when combined with HFIP for siRNA single strands, DNA, and psDNA
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
five alkylamines to 428.54 kHCC for hexylamine up to 6238.37 kHCC for TMEDA. A lower Henry’s Law constant would indicate that the IP agent was volatile enough to evaporate from the droplet as the oligonucleotide was ionized, and did not enrich in the droplet, which would increase the alkylamine concentration and suppress ionization. But how do the other chemical properties of these IP agents affect the ionization efficiency? The chemical properties for Log P, proton affinity and gas-phase basicity are all too similar to explain the differences in ionization enhancement. Therefore, the differences appear to be related to the hydrophobicity of the analyte along with the ion-pairing efficiency of the alkylamines. Impact of analyte hydrophobicity on IP agents for maximizing signal intensity The hydrophobicity of different DNA bases has been investigated by others.[3,7,10] The hydrophobicity is lower for C < G < A < T based on the chromatographic retention of oligonucleotides on a reversed-phase column. Null et al. investigated oligonucleotides and found that more hydrophobic DNA strands were ionized more efficiently based on the calculated hydrophobicity of the bases.[12] Table 2 shows the calculated hydrophobicity for siRNA, DNA and psDNA. For each oligonucleotide, the number of each base is multiplied by its hydrophobicity as measured by ΔGtr(w→chx) (kcal/mol) where a higher number indicates lower hydrophobicity.[37] If we compare the hydrophobicity of the siRNA strands, it appears that the sense strand is the least hydrophobic. Recall that the signal intensity of the sense strand is far lower than the antisense strand for all of the alkylamines with and without HFIP (Figs. 1 and 2). To investigate this further, we used reversed-phase chromatography to evaluate the hydrophobicity of the siRNA along with the DNA and psDNA that have identical sequences (Fig. 11). The chromatography indicates that the sense strand is less hydrophobic than the antisense strand based on its earlier retention time. We can explore this further by looking at the charge states for the sense and antisense strands in the presence of HFIP (Fig. 6). The base peak for the
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2661
partially explain the increase in signal intensity when HFIP is added to the alkylamines.[32] In addition, HFIP has a low boiling point (58.20 °C) and a high Henry’s Law constant kH,invCC (1.85 × 10–1 g/aq representing volatility). This allows it to be removed from the droplet at a higher rate than other solutes, further enhancing ionization.[21] However, this still does not resolve the difference in ionization efficiency between the different alkylamines. The difference in signal intensity between the alkylamines may fall at the transition of molecules into the gas phase. Although there has been much debate in the literature about the mechanism of ESI of biomolecules, the work by Nguyen and Fenn provides a compelling argument for the Ion Evaporation Model (IEM) model as the more convincing explanation for the desorption of large molecules.[26] As the droplets evaporate, the field strength of the ions on the droplet surface overcomes solvation forces allowing the ions to be ejected into the gas phase. In the case of negatively charged oligonucleotides, we must also consider the environment in the droplet that facilitates their transport to the surface. In general, analytes that are more hydrophobic will ionize more readily.[29,31] This is due to their lower solubility which pushes the analyte to the surface making it more available for desorption and ionization. The alkylamines pair with the charged backbone of the oligonucleotide making it more hydrophobic. This enhances the ability of the analyte to reach the droplet surface. The oligonucleotide can then desorb while the alkylamine is either evaporated from the surface or dissociated from the oligonucleotide in the gas phase. One interesting finding is that the Henry’s Law constant kHCC aq/g (Table 1) is lowest for the top eight alkylamines that provide the highest signal intensity for siRNA, DNA, and psDNA (Fig. 10). The Henry’s Law constant kHCC represents the solubility of a compound in aqueous solution. In the ESI droplet a compound with a kHCC (aq/g) >1 will enrich in solution while a compound with a kHCC (aq/g)
Revised: 3 September 2013
Accepted: 4 September 2013
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664 (wileyonlinelibrary.com) DOI: 10.1002/rcm.6733
Systematic optimization of ion-pairing agents and hexafluoroisopropanol for enhanced electrospray ionization mass spectrometry of oligonucleotides A. Cary McGinnis, E. Claire Grubb and Michael G. Bartlett* Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, The University of Georgia, Athens, GA 30602-2352, USA RATIONALE: New methods to enhance the electrospray ionization (ESI) signals are essential for low-level analysis of oligonucleotides. We report a systematic evaluation comparing 13 ion-pairing agents with and without hexafluoroisopropanol to understand their effect on the ion abundance of hetero-oligonucleotides. METHODS: A Waters Synapt G2 HDMS quadrupole time-of-flight instrument was used to compare oligonucleotide signal intensity with 13 alkylamine ion-pairing agents at varying concentrations. The alkylamines that yielded the highest signal intensity were further evaluated with hexafluoroisopropanol at concentrations between 5 and 100 mM. The chemical properties of the solution components and analytes were evaluated to identify key factors in predicting optimal mobile phase conditions for different classes of oligonucleotides. RESULTS: We identified a series of optimized mobile phase systems using diisopropylamine, tripropylamine, dimethylbutylamine, methyldibutylamine, and dimethylhexylamine along with 25 to 50 mM hexafluoroisopropanol that yielded significantly higher MS signal intensity for both siRNA and DNA compared with the traditionally used triethlyamine/hexafluoroisopropanol system. We explored charge state reduction, adduct formation and ESI mechanisms and identify the Henry’s Law constant k aq/g as a key chemical property in predicting alkylamines that will increase oligonucleotide ion intensity. We also find that the hydrophobicity of the oligonucleotide plays a major role in choosing ion-pairing agents that will increase ion abundance. CONCLUSIONS: This comprehensive and systematic optimization finds that the hydrophobicity of the oligonucleotide was a key factor in choosing alkylamine ion-pairing agents to increase ESI abundance. We identified that diisopropylamine and tripropylamine combined with lower concentrations of hexafluoroisopropanol yielded the highest signal intensity for these oligonucleotides. Copyright © 2013 John Wiley & Sons, Ltd.
Electrospray ionization mass spectrometry (ESI-MS) has become the gold standard for the analysis of macromolecules. It allows for quantification and characterization of a wide range of proteins and oligonucleotides. Great strides have been made in using ESI-MS to increase detection limits for proteins and peptides.[1] Still oligonucleotides can benefit from improvements to the ESI conditions so that these methods can be more effectively used for pharmacokinetic and toxicokinetic evaluations. Chromatographic separations of oligonucleotides initially used triethlyamine (TEA) with acetic acid as the counter ion.[2–4] This mobile phase provided good separation efficiency but lacked sensitivity with ESI-MS. Acetic acid has a higher boiling point than TEA. In the electrospray droplet, the TEA evaporated more rapidly. The acetate counter ion would then compete for ionization with the oligonucleotide thus lowering ionization efficiency. Dimethylbutylamine
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Copyright © 2013 John Wiley & Sons, Ltd.
2655
* Correspondence to: M. G. Bartlett, Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, The University of Georgia, Athens, GA 30602–2352, USA. E-mail: [email protected]
(DMBA) with a bicarbonate counter ion was shown to yield higher sensitivity than TEA with acetic acid for DNA.[5] In addition, DMBA was shown to have greater sensitivity than TEA with the bicarbonate counter ion. Because DMBA is more hydrophobic than TEA, it required a higher organic concentration to elute the oligonucleotides. The higher organic concentration reduced surface tension in the electrospray droplet providing up to 10 times greater sensitivity when compared to TEA with bicarbonate.[5] A number of studies have evaluated TEA with other counter ions.[6–11] Huber and Krajete evaluated acetate, bicarbonate, formate, and chloride counter ions.[11] In this comparison, the volatility of the counter ion did not correlate with increased signal intensity. Instead they found that acetate yielded the best sensitivity and argued that it was the lower conductivity of the counter ion that caused less suppression of oligonucleotide ionization. Other studies have focused on improving signal intensity through charge state reduction.[12–14] Charge state reduction involves shifting the charge state envelope to lower m/z values and increasing the intensity of the remaining charge states. Cheng et al. found that charge state reduction increased signal intensity due to the compression of the
A. C. McGinnis, E. C. Grubb and M. G. Bartlett charge states.[13] Muddiman et al. proposed a mechanism for charge state reduction in which the hydrogen-bound proton is shared as a dimer between the phosphodiester backbone and the counter ion.[14] Because of the higher proton affinity of the phosphodiester, the counter ion is lost as a neutral on entry, as well as, within the gas phase. Adduction of metal cations also lowers signal intensity for oligonucleotides.[14] Gaus et al. evaluated the signal intensity of a psDNA using seven alkylamines with no counter ion. This study focused on reducing adduct formation.[15] They found that tripropylamine (TPA) yielded the highest signal intensity. The most widely used mobile phase in the literature is based on the work of Apffel et al. in 1997.[16,17] This work introduced the use of triethlyamine (TEA) as an ion-pairing (IP) agent and hexafluoroisopropanol (HFIP) as a counter ion for ESI-MS of oligonucleotides. They proposed that the low boiling point of HFIP (58 °C) allowed it to evaporate from the electrosprayed droplet, which raised the pH of the droplet, allowing for enhanced ionization of oligonucleotides. Since that time, liquid chromatography (LC)/MS analysis of siRNAs and DNA has almost exclusively used different compositions of the TEA/HFIP mobile phase.[18–20] Recently, Chen et al. published a study on the effect of seven IP agents on the desorption and ionization of a phosphorothioate DNA.[21] They reported that diisopropylethylamine (DIEA) combined with HFIP yielded the highest sensitivity for psDNA. They also reported the bioanalysis of a psDNA in rat plasma with a lower limit of quantitation of 2.5 ng/mL.[22] We follow these studies with the evaluation of 13 IP agents and their effect on ion abundance of different classes of hetero-oligonucleotides. In this study, the alkylamines are evaluated at different concentrations alone and in the presence of HFIP. Mobile phase compositions are optimized for maximum ESI signal intensity for both siRNA and DNA. The implications of chemical properties and structure of both the alkylamines and the oligonucleotides are evaluated in order to provide a predictive model of mobile phase composition for modified oligonucleotides.
EXPERIMENTAL
2656
A 21mer siRNA duplex modified with DNA overhangs (sense 5’-GGA UCU CAU GCA CAU CUC AdTdT-3’, antisense 5’-UGA GAU GUG CAU GAG AUC CdTdG-3’) was purchased from Invitrogen (Life Technologies Carlsbad, CA, USA). The 24mer DNA (5’-TCGTGCTTTTGTTGTTTTC GCGTT-3’) was purchased from Integrated DNA Technologies (Coralville, IA, USA). LC/MS Chromasolv grade methanol and water were obtained from Fluka (St. Louis, MO, USA). Diisopropylamine (DIPA), diisopropylethylamine (DIEA), N’N’-dimethylbutylamine (DMBA), dimethylcyclohexylamine (DMCHA), dimethylhexylamine (DMHA), hexylamine (HA), N-methyldibutylamine (MDBA), octylamine (OA), propylamine (PA), tetramethylethylenediamine (TMEDA), triethylamine (TEA), tributylamine (TBA), tripropylamine (TPA), and 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) were purchased from Sigma–Aldrich Inc. (St. Louis, MO, USA). DNA Lobind centrifuge tubes were purchased from Eppendorf (Hauppauge, NY, USA).
wileyonlinelibrary.com/journal/rcm
Preparation of working solutions for infusion experiments Each alkylamine was prepared in 50:50 methanol/water at concentrations between 5 and 30 mM. Solutions containing HFIP were prepared between 5 and 100 mM in the varying concentrations of the IP agent (5–30 mM). The siRNA or DNA was prepared in each solution at a concentration of 10 μg/mL. Instrumental conditions Infusion experiments were performed on a Synapt G2 HDMS quadrupole time-of-flight hybrid mass spectrometer with an electrospray ionization (ESI) source (Waters, Milford, MA, USA). Chromatography was performed on an Acquity ultrahigh-performance liquid chromatography (uHPLC) system coupled with a Synapt G2 HDMS quadrupole timeof-flight hybrid mass spectrometer with an ESI source (Waters, Milford, MA, USA). Chromatographic separations were performed at a flow rate of 0.3 mL/min on a 1.7 μm Acquity BEH C18 column (Waters, Milford, MA, USA) 2.1 x 50 mm column. Mobile phase A consisted of 10 mM DIPA and 25 mM HFIP in water, and mobile phase B consisted of the same concentration of DIPA and HFIP in water/methanol (50:50). A 20 μL injection of each sample was loaded onto the column and separated using the following gradient conditions [time (min), % mobile phase B]: (0, 18) (2, 18) (8, 25). Gradient conditions to separate siRNA, DNA, and psDNA were [time (min), % mobile phase B]: (0, 18) (2, 18) (8, 32). The column temperature was maintained at 75 °C. The column eluent from 0–1 min was diverted to waste. The mass spectrometer was operated in negative-ion MS mode with a 1 s scan time. The mass spectrometer tuning was optimized for each different solution using the same source temperature, desolvation temperature and gas flows. For 10 mM DIPA/25 mM HFIP in 50% methanol, the tuning was as follows: capillary voltage 2.0 kV, cone voltage 25 V, extraction cone voltage 2 V, source temperature 125 °C, desolvation temperature 450 °C, cone gas 0 L/h and desolvation gas 1000 L/h. The data were collected in full-scan mode over the mass range 500–3000 m/z. All measurements were performed in triplicate. Base peak signal intensities were measured by combining 20 scans, and all data is represented as the mean ± the standard deviation.
RESULTS AND DISCUSSION Optimization of siRNA ESI-MS signal intensity Figures 1(a) and 1(b) show the base peak ion intensity of the sense and antisense strands with each alkylamine (N = 3). The ion-pairing (IP) agents are listed in the legends in order of decreasing signal intensity. As demonstrated previously, the siRNA denatures in the gas phase to its sense and antisense strands.[23] The first thing to note is the difference in maximum signal for the sense and antisense strands. The signal intensity for the antisense strand is more than 400,000 counts above that for the sense strand. Secondly, the alkylamines that provide the highest signal intensities are different for the sense and antisense strands. Only DIPA and DMBA are found in common among the top five IP agents for both strands. The difference in signal intensity is
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides
Figure 1. Comparison of the base peak ion intensity of siRNA in the presence of 13 alkylamines at various concentrations: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
alkylamines except for TPA. The optimized maximum signal intensities for the sense and antisense strands, respectively, are 10 mM DIPA, 25 mM HFIP – 1.2 × 106, 2.1 × 106, 10 mM DMBA, 25 mM HFIP – 1.0 × 106, 1.8 × 106, 10 mM TPA, 10 mM HFIP – 1.0 × 106, 1.7 × 106, followed by 15 mM TEA, 25 mM HFIP – 9.5 × 105, 1.6 × 106. The differences in signal are from 200,000 counts to as high as 500,000 counts. The optimized TEA/HFIP combination is as much as 500,000 counts lower than DIPA/ HFIP. The signal intensity differences of these optimized mobile phases would impact the low-level quantitation of siRNAs. Maximizing signal intensity must also be combined with adequate chromatographic separation. Each of the top alkylamines was combined with HFIP to find the best separation of siRNA and its chain-shortened metabolites from both the sense and antisense siRNA strands. The best separation, in addition to the maximum signal intensity, was provided with 10 mM DIPA, 25 mM HFIP (Fig. 4).[24] Chen et al. found that DIEA provided the highest signal intensity for a mixed base phosphorothioate DNA.[21] Interestingly, DIEA provided a much lower signal intensity for siRNA when compared with other alkylamines. The addition of sulfur to the phosphate backbone increases the hydrophobicity of a psDNA.[25] The siRNA used in these experiments would be significantly more hydrophilic. To further explore the effect of analyte hydrophobicity on ion abundance, we used a 24mer unmodified DNA with the same sequence as that used by Chen et al. and performed the same systematic evaluation of the effect of alkylamines and HFIP on signal intensity.
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2657
likely due to differences in the hydrophobicity of the single strands. These differences also seem to effect which alkylamine is most effective at increasing the signal. The role of analyte hydrophobicity is discussed further in a later section. Five of the top-performing IP agents for the sense and antisense strands were evaluated with different concentrations of HFIP. Each alkylamine was tested at concentrations between 5 and 25 mM with concentrations of HFIP at 10, 25, 50 and 100 mM. Figure 2 shows the base peak signal intensity for each alkylamine with varying concentrations of HFIP. Only the concentration of alkylamine that provided the best signal intensity when combined with HFIP is shown in the graph. The addition of HFIP doubles the signal intensity for both the sense and antisense strands for all of the alkylamines except octylamine. Interestingly, the signal intensity is suppressed by almost one-half by the addition of HFIP to octylamine. Figure 3 compares the base peak ion intensity of siRNA with 25 mM HFIP to show the optimization of alkylamine concentration with a fixed concentration of HFIP. Most of the alkylamines show an optimum concentration where the signal intensity is the highest. DMBA, however, shows a close performance between the 10 and 15 mM concentration in 25 mM HFIP. This would allow optimization of the DMBA concentration to enhance chromatographic separations while maintaining high signal intensity. Figure 2 provides further information about the remaining alkylamines and the effect of HFIP combination. A concentration of 25 mM HFIP provided the maximum signal intensity for the
Figure 2. Comparison of the base peak ion intensity of siRNA with alkylamines and various concentrations of HFIP: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
A. C. McGinnis, E. C. Grubb and M. G. Bartlett Optimization of DNA ESI-MS signal intensity Figure 5 shows the base peak ion intensity for each alkylamine for a 10 μg/mL solution of unmodified DNA. MDBA, TBA, DMHA, and TPA yielded the highest signal intensities. Interestingly, these are different than those optimized for siRNA except for TPA. Signal intensities for DIPA, TEA, and DIEA that yielded higher intensities for siRNA are essentially equivalent for DNA and are approximately 600,000 counts below the highest signal for MDBA. Optimization continued by evaluating the alkylamines that yielded the highest signal intensity with varying concentrations of HFIP following the same protocol as with siRNA. Figure 6 shows the base peak signal intensity for the alkylamines at the concentration that provided the highest signal intensity when combined with HFIP. The highest signal intensity was provided with 15 mM TPA with 50 mM HFIP; 15 mM MDBA, 25 mM HFIP followed at 200,000 counts lower, while 10 mM TBA, 5 mM HFIP is 400,000 counts lower than the TPA/HFIP combination. Once again, these differences in signal intensities for the alkylamine/HFIP combinations would favorably impact low-level analysis of unmodified DNAs. Supplementary Fig. S1 (see Supporting Information) shows the optimization of the alkylamine concentration with fixed concentrations of HFIP (25 and 50 mM). We see an optimal combination of alkylamine concentration with either 25 or 50 mM with both TPA and MDBA. The ion intensity signals for both TBA and DMHA are far lower and the differences in signal intensity are not as great. Figure 3. Comparison of the base peak ion intensity of siRNA in 25 mM HFIP to optimize the concentration of alkylamines. Note that no signal was obtained with 5 mM TEA in 25 mM HFIP: (a) antisense strand and (b) sense strand. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
2658
Figure 4. Ion-pairing reversed-phase chromatography with 1.7 μm Acquity BEH C18 column (2.1 × 50 mm; Waters, Milford, MA, USA). After a 2 min hold, a linear gradient from 18–25% B was applied over 6 min. Mobile phase A consisted of 10 mM DIPA and 25 mM HFIP in water, and mobile phase B consisted of the same concentration of DIPA and HFIP in water/methanol (50:50).
wileyonlinelibrary.com/journal/rcm
Effects of alkylamines on adduction and charge state distribution The charge state distribution was the same for siRNA in the presence of the top four alkylamines alone (data not shown). All have base peaks at the 12 charge state for both the sense and antisense strands. Figure 7 shows the charge state envelope for siRNA with DIPA, TEA, DMBA, and TPA combined with HFIP. The charge state is increased (moved to higher m/z) in the presence of HFIP to 10 and 11 for the antisense and sense strand, respectively. Cheng et al. found that charge state reduction increased signal intensity due to the compression of the charge states.[13] The presence
Figure 5. Comparison of the base peak ion intensity of DNA in the presence of 13 alkylamines at various concentrations. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides or neutrals rather than remaining with the oligonucleotide as a metal cation would. This lowers adduction, simplifies the mass spectrum, and significantly increases the signal.[26] The inserts on Fig. 7 show the expanded antisense base peak and associated adduction. While differences in adduction were seen with the IP agents alone (data not shown), this difference did not appear to significantly change the spectra for these alkylamines when combined with HFIP. This trend was slightly different for DNA (Fig. 8). Increased adduction was seen with TPA and MDBA compared with TBA and DMHA. Despite a slight increase in adduction, TPA and MDBA still provided higher signal intensity for the DNA. Figure 6. Comparison of the base peak ion intensity of DNA with alkylamines and various concentrations of HFIP. Experiments were performed in triplicate, and data are presented as the mean ± the standard deviation.
Structural and chemical properties of alkylamines and their relation to electrospray ionization
of HFIP does not reduce the number of charge states, but it does compress the charge state envelope for siRNA, by providing more intense signals for the sense and antisense strands at the 10 and 11 charge states. The addition of HFIP lowers the pH slightly and provides more protons in the solution. The more acidic solution allows more protons to be donated to the oligonucleotide anions thereby altering the charge state. Interestingly, a shift in the charge state was not observed with the addition of HFIP to alkylamines for DNA (Fig. 8). Adduction of metal cations may also lower signal intensity for oligonucleotides.[14] Ion-pairing agents act to reduce adduction by pairing with the negatively charged backbone. The IP agents are evaporated as positively charged molecules
Figure 9 shows the structures for the top alkylamines without HFIP for the siRNA antisense and sense strands, DNA and psDNA.[21] The top alkylamines for the RNA strands include primary, secondary, and tertiary amines. Commonalities between the oligonucleotides include DIPA and DMBA, which are secondary and tertiary amines, respectively. The structures that provided the highest signal intensity for both DNA and psDNA were all tertiary amines and have no commonalities, although the evaluation for psDNA by Chen et al. did not include DMHA or DMCHA. Figure 10 shows the structures of the top alkylamines combined with HFIP. These structures were all tertiary amines except for DIPA. Commonalities between siRNA, DNA and psDNA included TPA and TEA. Observations of the chain length for top groups appear to favor longer chain length alkylamines for the DNA where TPA, MDBA, TBA
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2659
Figure 7. Mass spectra of siRNA with the alkylamines with HFIP that yielded the highest signal intensity. The base peaks for the sense (S) and antisense (AS) strands are highlighted. The insets display the base peak charge states to illustrate adduction.
A. C. McGinnis, E. C. Grubb and M. G. Bartlett
Figure 8. Mass spectra of DNA with the alkylamines and HFIP that yielded the highest signal intensity. The insets display the base peak charge states to illustrate adduction.
Figure 9. Structures of the ion-pairing agents that yielded the highest signal intensity for siRNA single strands, DNA, and psDNA.
2660
and DMHA provide higher intensities. The chemical properties of each alkylamine were also evaluated for their impact on ionization efficiency. Table 1 provides many of the chemical properties for the alkylamines used in the optimization procedures.[27,28] The production of gas-phase ions by ESI depends heavily on the chemical properties of the solution components including the analyte.[29] The first steps are production of charged droplets from the capillary tip, then shrinkage of the charged droplets by solvent evaporation and droplet disintegration. The final step is the production of ions from the charged droplets.[30–32] The surface tension and conductivity effect the initial droplet production and
wileyonlinelibrary.com/journal/rcm
droplet disintegration. The organic portion of the mobile phase (50% methanol in these experiments) largely governs the surface tension.[33,34] The concentration of the solutes can affect the conductivity, which can be measured by pH. In these experiments, the pH was between 9 and 9.5 for all of the top performing alkylamines (data not shown). As the pKa data shown in Table 1 indicate, there is little difference between these solutes. When HFIP is added, the pH drops to between 8.0 and 9.0 (data not shown). The addition of the weak acid, HFIP, increases the conductivity and therefore reduces the droplet size emitted from the capillary.[32] This reduction in the ESI droplet size would increase ionization and this would
Copyright © 2013 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
Effect of ion-pairing agents on ESI of oligonucleotides
Figure 10. Structures of the ion-pairing agents that yielded the highest signal intensity when combined with HFIP for siRNA single strands, DNA, and psDNA
Rapid Commun. Mass Spectrom. 2013, 27, 2655–2664
five alkylamines to 428.54 kHCC for hexylamine up to 6238.37 kHCC for TMEDA. A lower Henry’s Law constant would indicate that the IP agent was volatile enough to evaporate from the droplet as the oligonucleotide was ionized, and did not enrich in the droplet, which would increase the alkylamine concentration and suppress ionization. But how do the other chemical properties of these IP agents affect the ionization efficiency? The chemical properties for Log P, proton affinity and gas-phase basicity are all too similar to explain the differences in ionization enhancement. Therefore, the differences appear to be related to the hydrophobicity of the analyte along with the ion-pairing efficiency of the alkylamines. Impact of analyte hydrophobicity on IP agents for maximizing signal intensity The hydrophobicity of different DNA bases has been investigated by others.[3,7,10] The hydrophobicity is lower for C < G < A < T based on the chromatographic retention of oligonucleotides on a reversed-phase column. Null et al. investigated oligonucleotides and found that more hydrophobic DNA strands were ionized more efficiently based on the calculated hydrophobicity of the bases.[12] Table 2 shows the calculated hydrophobicity for siRNA, DNA and psDNA. For each oligonucleotide, the number of each base is multiplied by its hydrophobicity as measured by ΔGtr(w→chx) (kcal/mol) where a higher number indicates lower hydrophobicity.[37] If we compare the hydrophobicity of the siRNA strands, it appears that the sense strand is the least hydrophobic. Recall that the signal intensity of the sense strand is far lower than the antisense strand for all of the alkylamines with and without HFIP (Figs. 1 and 2). To investigate this further, we used reversed-phase chromatography to evaluate the hydrophobicity of the siRNA along with the DNA and psDNA that have identical sequences (Fig. 11). The chromatography indicates that the sense strand is less hydrophobic than the antisense strand based on its earlier retention time. We can explore this further by looking at the charge states for the sense and antisense strands in the presence of HFIP (Fig. 6). The base peak for the
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2661
partially explain the increase in signal intensity when HFIP is added to the alkylamines.[32] In addition, HFIP has a low boiling point (58.20 °C) and a high Henry’s Law constant kH,invCC (1.85 × 10–1 g/aq representing volatility). This allows it to be removed from the droplet at a higher rate than other solutes, further enhancing ionization.[21] However, this still does not resolve the difference in ionization efficiency between the different alkylamines. The difference in signal intensity between the alkylamines may fall at the transition of molecules into the gas phase. Although there has been much debate in the literature about the mechanism of ESI of biomolecules, the work by Nguyen and Fenn provides a compelling argument for the Ion Evaporation Model (IEM) model as the more convincing explanation for the desorption of large molecules.[26] As the droplets evaporate, the field strength of the ions on the droplet surface overcomes solvation forces allowing the ions to be ejected into the gas phase. In the case of negatively charged oligonucleotides, we must also consider the environment in the droplet that facilitates their transport to the surface. In general, analytes that are more hydrophobic will ionize more readily.[29,31] This is due to their lower solubility which pushes the analyte to the surface making it more available for desorption and ionization. The alkylamines pair with the charged backbone of the oligonucleotide making it more hydrophobic. This enhances the ability of the analyte to reach the droplet surface. The oligonucleotide can then desorb while the alkylamine is either evaporated from the surface or dissociated from the oligonucleotide in the gas phase. One interesting finding is that the Henry’s Law constant kHCC aq/g (Table 1) is lowest for the top eight alkylamines that provide the highest signal intensity for siRNA, DNA, and psDNA (Fig. 10). The Henry’s Law constant kHCC represents the solubility of a compound in aqueous solution. In the ESI droplet a compound with a kHCC (aq/g) >1 will enrich in solution while a compound with a kHCC (aq/g)
