A Study of Calibrant Selection in Measurement of Carbohydrate and

Oct 20, 2014 - Department of Chemistry, University of Nebraska Lincoln, Lincoln, ... carbohydrate versus peptide) or charge state as the unknowns are ...
9 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/ac

A Study of Calibrant Selection in Measurement of Carbohydrate and Peptide Ion-Neutral Collision Cross Sections by Traveling Wave Ion Mobility Spectrometry Abby S. Gelb, Rebecca E. Jarratt, Yuting Huang, and Eric D. Dodds* Department of Chemistry, University of NebraskaLincoln, Lincoln, Nebraska 68588-0304, United States S Supporting Information *

ABSTRACT: While ion-neutral collision cross sections (CCSs) can be directly calculated from drift tube ion mobility spectrometry (DTIMS) data, measurements made using the more recently introduced traveling wave ion mobility spectrometry (TWIMS) technique are usually calibrated using standards with known CCS. Presently, there remains some question regarding how selection of calibrants influences TWIMS CCS measurements. This is of particular concern when calibrants of the same molecular class (e.g., carbohydrate versus peptide) or charge state as the unknowns are unavailable. This report presents a study of calibrant ion influence on CCS determination via TWIMS. Drift times from TWIMS were calibrated to CCSs using either carbohydrates or peptides as standards. These calibrations were then applied to other carbohydrates and peptides with known CCSs, and the errors of the measurements were assessed. In addition, calibrations with standards having charge states either matched or mismatched with those of the target analytes were applied and evaluated for accuracy. The use of carbohydrates to calibrate peptide CCSs and vice versa was found to introduce errors only modestly larger than the inherent uncertainties of the measurements (on average, 1.0%). Charge state mismatching while the same molecular class of calibrant and analyte was maintained yielded larger errors (on average, 3.5%). Mismatching of both calibrant molecular class and charge state resulted in the largest errors (on average, 4.7%). These results suggest that matching of both molecular class and charge state is recommended when possible, while matching at least the charge state is strongly advisable.

T

difficult to relate to CCSs.11 This is in contrast to classical static field drift tube ion mobility spectrometry (DTIMS) instruments, which are capable of measuring drift times directly relatable to CCSs through first principles.12 While an approach for direct measurement of CCSs from TWIMS data has been suggested,13 TWIMS drift times are far more commonly calibrated to CCSs with the aid of standard ions having known CCSs established by DTIMS.14−16 Although calibration of TWIMS drift times to estimate CCSs has now been established for several years, there remains some uncertainty with respect to how the choice of calibrant ions influences the accuracy of CCSs calculated from TWIMS drift times. For instance, different classes of compounds exhibit distinct mass/mobility correlations and inherent conformational ordering.17−19 Such observations have led to the suggestion that CCS calibrants of the same class of compounds as the analytes should be most suitable for measurements of analyte CCSs. This matter becomes particularly vexing when no DTIMS CCSs are known for ions of the same molecular class as the target analytes. For instance, a search of the literature

he application of ion mobility spectrometry (IMS) to the separation and characterization of biomolecules has seen rapid growth over the past several years, particularly in conjunction with mass spectrometry (MS) analysis. 1−6 Acceleration of research in this area has been catalyzed in part by the relatively recent advent of commercial IMS instruments based on a variety of operating principles. One of the most appealing potential outcomes of such measurements is the determination of orientationally averaged ionneutral collision cross sections (CCSs). In concert with CCSs calculated for theoretical candidate structures and conformations, experimentally observed CCS values can be used to make inferences regarding the covalent and noncovalent structures of analytes.7 Unfortunately, these endeavors are complicated in part by the fact that CCSs cannot be directly obtained from IMS drift times observed using certain IMS methods.8 Traveling wave ion mobility spectrometry (TWIMS), a relatively new IMS technique, is of significant interest from the standpoint of CCS determination due to the wide commercial availability of such instrumentation. TWIMS makes use of radio frequency (rf) and direct current (dc) electric fields, which are varied both temporally and spatially.9,10 The interaction of analyte ions with these electric fields and the resultant ion motion is highly complex, rendering TWIMS drift times © 2014 American Chemical Society

Received: September 5, 2014 Accepted: October 20, 2014 Published: October 20, 2014 11396

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

prepared in water. A 10 μL aliquot of this thyroglobulin solution was combined with 25 μL of a 25 mM sodium phosphate buffer solution (pH 7.5−8.0) and 2.5 μL of a 200 mM aqueous dithiothreitol solution. The resulting mixture was incubated at 60 °C for 45 min (for reduction of disulfide linkages). The solution was then treated with 10 μL of a 200 mM aqueous iodoacetamide solution and subsequently incubated in the dark at ambient temperature for 1 h (for cysteine alkylation). Additional aliquots of the dithiothreitol solution (2.5 μL), sodium phosphate buffer solution (100 μL), and water (300 μL) were added prior to boiling the mixture for 10 min. The sample was cooled to room temperature, treated with 15 μL of a 250 units/mL of PNGase F solution in water, and incubated for 18−22 h at 37 °C (for N-glycan release). The sample was then briefly boiled to deactivate PNGase F. The protein content of the solution was precipitated by addition of 585 μL of 100% ethanol (−20 °C) followed by chilling at −20 °C for 30 min. The sample was centrifuged and the supernatant containing released N-glycans was recovered. The recovered supernatant was brought to dryness by vacuum centrifugation (Speed Vac SC110, Thermo Savant, Holbrook, NY), and the glycans were then resuspended in 200 μL of water prior to purification using solid-phase extraction (SPE) with a graphitized carbon stationary phase. Each SPE micropipette tip (Glygen, Columbia, MD) was first washed with 100 μL of 80% (v/v) ACN containing 0.1% (v/v) trifluoroacetic acid (TFA) and then equilibrated with 100 μL of deionized water. A 20 μL portion of the reconstituted N-glycan solution was next loaded onto the tip, which was then washed with 100 μL of deionized water. The N-glycans were finally eluted into 20 μL of 50% (v/v) ACN containing 0.1% TFA. Ion Mobility Spectrometry and Mass Spectrometry. All TWIMS and MS experiments were performed using a Waters Synapt G2-S HDMS quadrupole time-of-flight hybrid mass spectrometer (Q-TOF-MS) equipped with a TWIMS separator (Waters, Manchester, UK). The instrument was fitted with a nanoelectrospray ionization (nESI) source which had been custom modified to make use of borosilicate emitters pulled in-house. The emitters were formed from 100 mm × 1.5−1.8 mm Pyrex melting point capillaries (Corning, NY) using a vertical micropipette puller (David Kopf Instruments, Tujunga, CA). For nESI analysis, approximately 10 μL of solution was added to an emitter, which was then placed onto the nESI source stage with the aid of an electrode holder (Warner Instruments, Holliston, MA). This allowed the solution within the emitter to be put into contact with a platinum wire, which was used to apply the nESI capillary potential. The nESI capillary voltage and sampling cone voltage were optimized for each individual solution and emitter and ranged between 1.0−1.7 kV and 10.0−30.0 V, respectively. The ion source temperature was set at 80 °C. For tandem mass spectrometry (MS/MS), precursor ions of interest were quadrupole-selected and subjected to collision-induced dissociation (CID) using argon as the collision gas prior to entering the TWIMS separator. For all TWIMS experiments, the helium cell gas flow was held at 180.0 mL/min and the mobility cell gas flow was held at 60.0 mL/min (N2). The trap dc bias [essentially, the potential difference between the trap region stacked ring ion guide (SRIG) and the TWIMS separator] was optimized for each analyte and ranged from 44.2 to 80.0 V. The TWIMS traveling dc wave height and velocity were held constant at 40.0 V and 650.0 m/s, respectively. The rf amplitudes applied to the TWIMS cell,

reveals no DTIMS CCSs have been published for glycosylated peptide ions at this writing. Nevertheless, these analytes are becoming a focal point of interest for analysis by IMS methods.20−22 Ion charge state constitutes another calibrant characteristic with implications for CCS accuracy, as charge state can affect the extent of ion-neutral interactions with the drift gas. This is a particularly important consideration in TWIMS experiments, which routinely make use of nitrogen as the drift gas. The polarizability of nitrogen is substantially greater than that of helium, which has been most commonly (but not exclusively) used in traditional DTIMS experiments.23,24 Use of a drift gas with higher polarizability would be expected to increase ionneutral interaction between the analyte and drift gas, thus potentially rendering an IMS experiment even more sensitive to the charge state and charge localization of analyte ions.25,26 Overall, then, the TWIMS drift times of compounds with different molecular packing, polarity, absolute charge, and dispersion of charge may interact differently with the drift gas and thus have drift times that scale differently with CCS. This is of particular concern given that the availability of known CCSs from DTIMS measurements is somewhat limited in terms of the range of CCSs, the drift gases for which CCSs have been measured, the diversity of molecular classes, and the charge states of the ions. Therefore, matching the molecular class and charge state of standards and unknowns for the purpose of calibrating TWIMS drift times to CCS may not be possible or practical for a given class of analytes. Here, we present a study of the implications of calibrant ion selection for CCS measurement using TWIMS. The TWIMS CCSs of several model carbohydrate and peptide analytes having known DTIMS CCSs were obtained using both other peptides and other carbohydrates with known CCSs as calibrants. TWIMS CCSs were also determined with the use of calibrant ions having charge states either matched or mismatched with those of the model analytes. For the range of analyte and calibrant classes, charge states, and CCSs considered here, the matching of both molecular class and charge state was found to be advisable in order to obtain the most accurate CCS measurements. Moreover, mismatching calibrant/analyte charge state appears to contribute significantly greater error than mismatching calibrant/analyte compound class. These findings advocate the careful consideration of both calibrant ion selection and the associated implications for experimental error when measuring CCSs via calibrated TWIMS drift times.

■

MATERIALS AND METHODS Solution Preparation. Polyalanine, leucine enkephalin (leu enk), bradykinin, substance P, maltotriose, and lacto-Ndifucohexaose II (LNDFH II) were purchased from SigmaAldrich (St. Louis, MO). Lacto-N-fucopentaose I (LNFP I) was purchased from Accurate Chemical & Scientific (Westbury, NY). A solution of polyalanine was prepared at a concentration of 12.5 μg/mL in 50% (v/v) aqueous acetonitrile (ACN). Other peptide and carbohydrate solutions were prepared at concentrations of 10 μM in 50% (v/v) aqueous ACN. N-Glycan Release. Thyroglobulin from Sus scrofa and PNGase F from Elizabethkingia meningoseptica were obtained from Sigma-Aldrich. N-Linked glycans were released from thyroglobulin using a procedure similar to those reported elsewhere,27,28 with some minor modifications as summarized below. Briefly, a solution of 30 mg/mL thyroglobulin was 11397

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

TOF interface region prior to reaching the pusher stack. Thus, it was necessary to calculate a corrected drift time (td′) by subtracting the mobility dead time (tm), the transfer region transit time (tt), and TOF interface region transit time (ti) such that only the relevant ion transit time was included:

the trap region SRIG, and the transfer region SRIG were held constant for all experiments (250, 350, and 380 V, respectively). TWIMS data was analyzed and processed using MassLynx 4.1 and DriftScope 2.7 (Waters). The data was further evaluated and plotted using SigmaPlot 10.1 (Systat, Chicago, IL). Ion-Neutral Collision Cross Section Calibration. Drift times from TWIMS experiments were calibrated to CCSs through application of established protocols14−16 with some adaptations as noted below. Under conditions typical of a DTIMS experiment, the Mason−Schamp equation relates the ion mobility (K) to the ion-neutral CCS (Ω): K=

3ze 16N Ω

2π kBTμ

td′ = td − tm − t t − t i

The td′ values for all ions of interest were calculated according to eq 7; td′ served as the dependent variable for TWIMS drift time to CCS calibration. Calculation of td′ required that tm, tt, and ti each be computed. First, tm was determined as the time required for ions to transit the TWIMS cell in the absence of ion-neutral collisions. This is analogous to the void time of an unretained species in chromatography and is equal to the time required for a traveling dc wave of velocity vm to traverse the distance of the mobility cell (dm, 25.4 cm):

(1)

Here, z is the integer charge, e is the elementary charge, N is the drift gas number density, kB is the Boltzmann constant, T is the absolute temperature, and μ is the ion-neutral reduced mass, defined for two collision partners having masses m1 and m2 as m1m2 μ= m1 + m2 (2)

tm =

tt =

Ω μ 1 ∝ K z

(5)

The Ω′ values for all calibrant ions were calculated according to eq 5; Ω′ served as the independent variable for TWIMS drift time to CCS calibration. For each ion of interest, the experimentally observed TWIMS arrival time distribution (ATD) was collected as a trace of ion intensity versus analog-to-digital converter (ADC) bin. In Synapt instruments, the nth ADC bin corresponds to the nth TOF pulse occurring since an ion packet was last gated into the mobility cell (and 1 ≤ n ≤ 200 for each TWIMS separation). The TWIMS ATDs were centroided in order to obtain the corresponding ADC drift bin value (bd), which was then converted to td using the TOF pusher period (Δtp): td = bdΔt p

(9)

m z

(10)

where kEDC is an empirically determined proportionality constant referred to as the “enhanced duty cycle” constant. The value of kEDC is determined for each individual Synapt instrument at installation and is normally used to allow operators to synchronize TOF pulses with the m/z-dependent arrival times of ions at the pusher stack from the transfer region SRIG. It should be noted that kEDC contains units of microseconds; thus, eq 10 provides ti with units of microseconds. Calibration curves were generated by plotting log(td′) (experimentally observed; dependent variable) as a function of log(Ω′) (accepted literature values; independent variable) and fitting a linear function of the form log(td′) = m log(Ω′) + b. These relations were then used to calculate unknown Ω′ values (mass and charge normalized CCSs) from experimentally determined td′ values. The corresponding Ω values (absolute CCSs) were then calculated according to eq 5. In all cases, calibration and analysis were conducted using identical TWIMS parameters.

(4)

Ω μ z

dt vt

t i = kEDC

Equation 4 provides a useful proportionality between td and Ω, assuming constant N, T, and E and under conditions in which eq 1 is valid (i.e., at relatively low values of E/N and in the presence of a uniform and unchanging electric field, as is typical of DTIMS). This is also useful in relating TWIMS td values to DTIMS Ω values from the literature. Conveniently, a reduced mass and charge state normalized CCS (Ω′) can be defined, with which TWIMS td values scale positively: Ω′ =

(8)

Third, the drift times were corrected for the time (ti) required for ions to transit the TOF interface region of the instrument under the influence of several static dc fields. In contrast to tm and tt, ti depends on m/z according to

(3)

Equation 3 illustrates the inverse proportionality relating td to K; in turn, K depends on Ω, μ, and z in accord with eq 1. This can be summarized as td ∝

dm vm

Second, the drift times were corrected for the time (tt) required to transit the distance of the transfer region SRIG (dt, 13.5 cm) as dictated by the velocity (vt) of the transfer cell traveling dc wave:

The drift velocity (vd, at which the drift distance dd is traversed in the drift time td) is directly proportional to K and to the magnitude of the electric field (E) according to

d vd = d = KE td

(7)

■

RESULTS AND DISCUSSION Overview. TWIMS drift times of various model analytes were calibrated to CCSs in multiple ways: (1) using singly protonated polyalanine ions with CCSs previously measured by Clemmer and co-workers using DTIMS and helium as the drift gas,29 (2) using multiply protonated polyalanine with CCSs previously measured by Bush et al. using DTIMS and helium as the drift gas,30 and (3) using singly sodiated N-glycans released from thyroglobulin and CID MS/MS fragments thereof with

(6)

This apparent drift time was comprised of the time required to traverse the TWIMS cell, the transfer region SRIG, and the 11398

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

CCSs previously measured by Pagel and Harvey using DTIMS and helium as the drift gas.31 Therefore, all CCS values reported here were measured on the basis of calibrations established using helium values. In order to evaluate the accuracy of various calibration approaches, experimentally determined CCSs were compared to the established CCSs of these calibrants as well as several model analytes including three carbohydrates (maltotriose, LNFP I, and LNDFH II) and three peptides (leu enk, bradykinin, and substance P). The structures of the model analytes are shown in Scheme 1, and their previously established CCSs were obtained from several sources.32−35 Biomolecule Class Matched and Mismatched Calibrations. Representative CCS calibration lines established using peptide standards (singly protonated polyalanines) and carbohydrate standards (singly sodiated N-glycans released from thyroglobulin and CID fragments thereof) are presented in Figure 1. Both log(td′) versus log(Ω′) plots were found to be well-fit by linear functions of the form log(td′) = m log(Ω′) + b. Moreover, the calibrations were highly reproducible, with drift times of the calibrant ions varying less than 0.5% from day to day. We note here that these calibration plots differ from some of those appearing elsewhere in the literature in two regards: the selection of independent and dependent variable and the use of a log−log plot. Although it has become commonplace to plot TWIMS calibrations with CCS values on the y axis and drift times on the x axis, here the reverse is used. This representation is in accord with the usual convention of plotting known values on the x axis [independent variable; here, log(Ω′) values for the calibrant ions derived from the literature], and the resulting analytical values on the y axis [dependent variable; here, log(td′) values measured for the calibrant ions]. In addition, some TWIMS calibrations found elsewhere in the literature plot Ω′ versus td′ and fit a power function of the form Ω′ = m(td′)b. Here, a log−log plot is adopted, as this allows for the possibility of linear extrapolation beyond the available range of CCSs from the literature.14 While some differences between the peptide-based and carbohydratebased calibrations can be noted [particularly at low log(Ω′) values], the calibrations exhibit strong concordance over a significant range of log(Ω′). In order to assess the errors on CCSs as determined by these calibrations, the CCSs of the thyroglobulin N-glycans were calculated using the polyalanine peptide based calibration, and alternately, the CCSs of the polyalanine peptides were calculated using the thyroglobulin N-glycan-based calibration. In both cases, the calculated CCSs were then compared to their known values from the literature. These results are summarized in Table S1 of the Supporting Information. When the t-test was used to compare the experimental and accepted CCS values, 12 of the 22 calculated values (9 of the 11 carbohydrate CCSs and 3 of the 11 peptide CCSs) were found to be distinguishable from literature values at the 95% confidence level. The average absolute percent error of the determinations was found to be 2.8% (ranging from 0.2% to 7.8%). Considering that the uncertainties of CCS values measured by TWIMS are typically on the order of a few percent, many of these values are consistent with the current state-of-the-art in accuracy, despite having been measured with biomolecule class mismatched calibrations. These results also serve to illustrate the range of uncertainties associated with TWIMS CCS calibrations under the conditions of these experiments.

Scheme 1. Structures of Model Carbohydrates (purple) and Peptides (green) Studied Herein

The CCS measurement accuracies resulting from biomolecule class matched and mismatched calibrations were further investigated by applying the thyroglobulin N-glycan and polyalanine calibrations shown in Figure 1 to the model carbohydrate and peptide analytes presented in Scheme 1. In this comparison, none of the model analytes were involved in the calibration, although literature CCS values have been previously reported for these model analytes, as noted above. Moreover, the model analytes were studied as singly charged ions (singly sodiated in the case of the carbohydrates; singly protonated in the case of the peptides) and thus matched the 11399

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

Figure 1. Comparison of log(td′) versus log(Ω′) calibration plots for singly protonated peptide ions (polyalanine) and singly sodiated carbohydrate ions (N-glycans released from thyroglobulin and CID fragments thereof). Literature values of Ω′ were derived from Henderson et al.29 and Pagel and Harvey.31

charge states of the calibrants. Each calibration (both carbohydrate-based and peptide-based) was applied to each model analyte (both carbohydrates and peptides), and the resulting measurements were compared. The percent difference between CCSs measured using the two calibrations ranged from 0.05% to 4.69% (2.43% difference on average), although in four of the six cases considered the two CCS values determined for a given model analyte were found to significantly differ by the t-test (Table S2, Supporting Information). Interestingly, the absolute percent errors of the carbohydrate-calibrated and peptide-calibrated CCSs were quite similar in most cases. This suggests that, while CCS values calculated on the basis of calibrant biomolecule class matching can yield results that are statistically distinguishable from each other, the errors of the determinations were not significantly influenced by which calibration was applied (on average, 1.0% error versus 1.7% error for mismatched and matched calibrations, respectively). This is further illustrated in Figure 2, which underscores the general agreement between the calibrant biomolecule matched and mismatched CCS values, both to one another and with respect to literature values. Charge State Matched and Mismatched Calibrations. The effects of relative calibrant and analyte charge state were next studied using calibrations based on singly protonated, doubly protonated, and triply protonated polyalanines to calculate the CCSs of model peptide analytes as singly protonated ions. The calibrations and resulting analyte CCSs are provided in Figure 3. The three calibration plots shown in Figure 3a for z = 1+, 2+, and 3+ polyalanines were constructed under identical TWIMS conditions; in fact, the parameters employed for these experiments yielded polyalanine ions of all three charge states simultaneously. Again, the calibrations were found to be well-represented by linear functions. Furthermore, while regions of significant overlap were found to exist among the calibrations for the different charge states, the calibrations based on z = 1+ polyalanine peptides were consistently quite distinct from those based on z = 2+ and z = 3+ polyalanine species. This observation is consistent with those of other researchers who have measured CCS values of differently charged analytes using nitrogen as the drift gas30,31 and is also

Figure 2. CCSs determined for model carbohydrate analytes (a) and model peptide analytes (b) based on calibrations using peptide calibrants (polyalanine) or carbohydrate calibrants (N-glycans from thyroglobulin and CID fragments thereof). Literature values of Ω were obtained from several sources.32−35 Error bars, where visible, indicate the standard deviation of replicate measurements (n = 5 in part a; n = 4 in part b).

consonant with the general prediction that charge state effects become more pronounced as drift gas polarizability increases.25,26 Interestingly, the z = 2+ and z = 3+ calibrations exhibit a particularly high degree of overlap. When these calibrations were applied to model peptide analytes as their z = 1+ (singly protonated) ions, some slight differences were noted among the resulting CCSs (Figure 3b). As summarized in Table S3 of the Supporting Information, the errors associated with charge state mismatched CCS values were slightly higher than those of charge state matched values (3.5% versus 2.5%, respectively), although all were within 6% of literature values. As noted in Table S3 (Supporting Information), some of the calculated CCS values were based on drift times that fell outside of the range furnished by calibrants of a given charge state. In these cases, the CCS values were calculated on the basis of extrapolation of the linear calibration, as suggested elsewhere.14 Although not ideal, the errors associated with the measured CCS values resulting from both extrapolated calibrations using charge state mismatched analytes resulted in a 5.7% error, at most, relative to literature values. This issue also highlights one of the challenges facing the field at large; namely, the difficulty in finding convenient groups of CCS standards that can be used to construct a calibration or 11400

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

Figure 4. CCSs calculated for singly metalated model carbohydrate analytes on the basis of calibrations using peptide calibrants (polyalanine) with charge states of z = 1+ and 2+. Literature values of Ω were obtained from Huang and Dodds.35 Error bars, where visible, indicate the standard deviation of replicate measurements (n = 4).

resulting from charge state matched calibration were systematically smaller than those measured on the basis of charge state mismatched calibrations. In addition, Table S4 of the Supporting Information illustrates that the errors associated with charge state matched calibration (on average, 1.2%) were significantly less than those associated with charge state mismatched calibration (on average, 4.7%). Therefore, in this case involving model carbohydrate analytes, charge state mismatching led to values substantially less accurate than those obtained when charge state was matched. This was true regardless of whether biomolecule class was matched or mismatched. Meanwhile, the errors of CCS determinations were similar for charge state matched calibrations with either matched or mismatched biomolecule classes (cf. Tables S2 and S4, Supporting Information).

Figure 3. Comparison of log(td′) versus log(Ω′) calibration plots for protonated peptide ions (polyalanine) with charge states of z = 1+, 2+, and 3+ (a) and CCSs calculated for singly protonated model peptide analytes using each of the three calibrations (b). Literature values of Ω′ and Ω were gathered from several sources.30,32−34 Error bars, where visible, indicate the standard deviation of replicate measurements (n = 4).

■

calibrations that encompass a realistic range of drift times and charge states while the biomolecule class is also matched. Biomolecule Class and Charge State Mismatched Calibrations. In an even less ideal scenario, a CCS calibration might be mismatched with analytes in both biomolecule class and charge state. In order to assess the errors in CCS values that would result from such an approach, calibrations were established using multiple charge states (z = 1+ and 2+) of standard polyalanine peptides with known CCSs as measured by Bush et al.30 These calibrations were then used to calculate the CCSs of singly charged, monovalent cationized carbohydrates. Figure 4 depicts the resulting CCS values and their relation to literature values. We note here that the literature values used for comparison in this case were obtained from our previous TWIMS study of carbohydrate/group I metal adducts.35 Nevertheless, where comparisons were possible, the values obtained in that study were found to be in close accord with DTIMS measurements as well as other calibrated TWIMS results.4,36,37 Prominent differences were noted between the CCSs of carbohydrates calculated according to biomolecule class mismatched but charge state matched (i.e., z = 1+ calibrants) and both biomolecule class and charge state mismatched (i.e., z = 2+ calibrants) calibrations. The CCSs

CONCLUSIONS For the range of carbohydrate and peptide ions studied here, some general conclusions can be drawn regarding the consequences of calibrant ion selection when CCS measurements are sought using TWIMS. Calibrations either matched or mismatched with analytes with respect to biomolecule class resulted in CCS measurements that were statistically distinguishable; however, the average absolute errors in CCS resulting from biomolecule class matched and mismatched calibrations were only 1.7% and 1.0%, respectively. These errors are only modestly larger than the inherent uncertainties of the measurements. Therefore, it may be concluded that matching of calibrant and analyte biomolecule class is desirable but not essential for the range of carbohydrates and peptides considered here. Although previous works have implied a need for matching calibrant ion and analyte ion class in TWIMS, those studies focused on calibrations and CCS measurements involving denatured intact proteins, nativelike intact proteins and protein complexes, and salt clusters.38,39 Given these observations, it is highly likely that TWIMS CCS measurements involving each specific type of calibrant/analyte mismatch (e.g., calibrating with salt clusters to measure the 11401

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402

Analytical Chemistry

Article

(9) Giles, K.; Pringle, S. D.; Worthington, K. R.; Little, D.; Wildgoose, J. L.; Bateman, R. H. Rapid Commun. Mass Spectrom. 2004, 18, 2401−2014. (10) Giles, K.; Williams, J. P.; Campuzano, I. Rapid Commun. Mass Spectrom. 2011, 25, 1559−1566. (11) Shvartsburg, A. A.; Smith, R. D. Anal. Chem. 2008, 80, 9689− 9699. (12) Verbeck, G. F.; Ruotolo, B. T.; Sawyer, H. A.; Gillig, K. J.; Russell, D. H. J. Biomol. Tech. 2002, 13, 56−61. (13) Giles, K.; Wildgoose, J. L.; Langridge, D. J.; Campuzano, I. Int. J. Mass Spectrom. 2010, 298, 10−16. (14) Ruotolo, B. T.; Benesch, J. L. P.; Sandercock, A. M.; Hyung, S.J.; Robinson, C. V. Nat. Protoc. 2008, 3, 1139−1152. (15) Thalassinos, K.; Grabenauer, M.; Slade, S. E.; Hilton, G. R.; Bowers, M. T.; Scrivens, J. H. Anal. Chem. 2008, 81, 248−254. (16) Smith, D.; Knapman, T.; Campuzano, I.; Malham, R.; Berryman, J.; Radford, S.; Ashcroft, A. Eur. J. Mass Spectrom. 2009, 15, 113−130. (17) Fenn, L. S.; Kliman, M.; Mahsut, A.; Zhao, S. R.; McLean, J. A. Anal. Bioanal. Chem. 2009, 394, 235−244. (18) McLean, J. J. Am. Soc. Mass Spectrom. 2009, 20, 1775−1781. (19) May, J. C.; Goodwin, C. R.; Lareau, N. M.; Leaptrot, K. L.; Morris, C. B.; Kurulugama, R. T.; Mordehai, A.; Klein, C.; Barry, W.; Darland, E.; Overney, G.; Imatani, K.; Stafford, G. C.; Fjeldsted, J. C.; McLean, J. A. Anal. Chem. 2014, 86, 2107−2116. (20) Creese, A. J.; Cooper, H. J. Anal. Chem. 2012, 84, 2597−2601. (21) Li, H.; Bendiak, B.; Siems, W. F.; Gang, D. R.; Hill, H. H., Jr. Int. J. Ion Mobility Spectrom. 2013, 16, 105−115. (22) Both, P.; Green, A. P.; Gray, C. J.; Sardzik, R.; Voglmeir, J.; Fontana, C.; Austeri, M.; Rejzek, M.; Richardson, D.; Field, R. A.; Widmalm, G.; Flitsch, S. L.; Eyers, C. E. Nat. Chem. 2014, 6, 65−74. (23) Asbury, G. R.; Hill, H. H. Anal. Chem. 2000, 72, 580−584. (24) Ruotolo, B. T.; McLean, J. A.; Gillig, K. J.; Russell, D. H. J. Mass Spectrom. 2004, 39, 361−367. (25) Beegle, L. W.; Kanik, I.; Matz, L.; Hill, H. H., Jr Int. J. Mass Spectrom. 2002, 216, 257−268. (26) Matz, L. M.; Hill, H. H., Jr; Beegle, L. W.; Kanik, I. J. Am. Soc. Mass Spectrom. 2002, 13, 300−307. (27) Kyselova, Z.; Mechref, Y.; Al Bataineh, M. M.; Dobrolecki, L. E.; Hickey, R. J.; Vinson, J.; Sweeney, C. J.; Novotny, M. V. J. Proteome Res. 2007, 6, 1822−1832. (28) Kang, P.; Mechref, Y.; Novotny, M. V. Rapid Commun. Mass Spectrom. 2008, 22, 721−734. (29) Henderson, S. C.; Li, J.; Counterman, A. E.; Clemmer, D. E. J. Phys. Chem. B 1999, 103, 8780−8785. (30) Bush, M. F.; Campuzano, I. D. G.; Robinson, C. V. Anal. Chem. 2012, 84, 7124−7130. (31) Pagel, K.; Harvey, D. J. Anal. Chem. 2013, 85, 5138−5145. (32) Counterman, A. E.; Valentine, S. J.; Srebalus, C. A.; Henderson, S. C.; Hoaglund, C. S.; Clemmer, D. E. J. Am. Soc. Mass Spectrom. 1998, 9, 743−759. (33) Ruotolo, B. T.; Tate, C. C.; Russell, D. H. J. Am. Soc. Mass Spectrom. 2004, 15, 870−878. (34) Polfer, N. C.; Bohrer, B. C.; Plasencia, M. D.; Paizs, B.; Clemmer, D. E. J. Phys. Chem. A 2008, 112, 1286−1293. (35) Huang, Y.; Dodds, E. D. Anal. Chem. 2013, 85, 9728−9735. (36) Williams, J. P.; Grabenauer, M.; Holland, R. J.; Carpenter, C. J.; Wormald, M. R.; Giles, K.; Harvey, D. J.; Bateman, R. H.; Scrivens, J. H.; Bowers, M. T. Int. J. Mass Spectrom. 2010, 298, 119−127. (37) Fenn, L. S.; McLean, J. A. Phys. Chem. Chem. Phys. 2011, 13, 2196−2205. (38) Bush, M. F.; Hall, Z.; Giles, K.; Hoyes, J.; Robinson, C. V.; Ruotolo, B. T. Anal. Chem. 2010, 82, 9557−9565. (39) Salbo, R.; Bush, M. F.; Naver, H.; Campuzano, I.; Robinson, C. V.; Pettersson, I.; Jørgensen, T. J. D.; Haselmann, K. F. Rapid Commun. Mass Spectrom. 2012, 26, 1181−1193.

CCSs of lipids) will require an independent assessment of accuracy. When considering the use of calibrants and analytes with matched molecular class but mismatched charge states, the average absolute error in measured CCSs increased to 3.5%. This suggests a greater dependence of CCS measurement accuracy upon analyte/calibrant charge state matching as compared to molecular class matching. Lastly, when both the molecular classes and charge states of the calibrants and analytes were mismatched, the average error of CCS determination was found to be 4.7%. Hence, it can be concluded that measurement of carbohydrate CCSs based on peptide calibrants (and vice versa) is likely suitable for many applications. By contrast, charge state matching of calibrants and analytes of the types considered here is strongly recommended owing to its substantially greater impact on CCS measurement accuracy. Overall, these results underscore the need for conscientious selection of calibrant ions for calibrating TWIMS drift times to CCSs and the need to characterize the uncertainty imposed when ideally suitable calibrants are not available for use.

■

ASSOCIATED CONTENT

S Supporting Information *

Tabulated CCS values, their associated absolute percent errors of determination, and the results of relevant statistical tests, as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Telephone: 1-402-472-3592. Fax: 1-402-472-9402. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS

■

REFERENCES

Funding from the University of Nebraska, which was provided in part by the Nebraska Tobacco Settlement Biomedical Research Development Fund, is gratefully acknowledged. R.E.J. also acknowledges support from the University of Nebraska Lincoln Undergraduate Creative Activities and Research Experiences (UCARE) program. Finally, the authors thank Venkata Kolli and Deepali Rathore for constructive comments on a draft of the manuscript.

(1) Bohrer, B. C.; Merenbloom, S. I.; Koeniger, S. L.; Hilderbrand, A. E.; Clemmer, D. E. Annu. Rev. Anal. Chem. 2008, 1, 293−327. (2) Fenn, L. S.; McLean, J. A. Anal. Bioanal. Chem. 2008, 391, 905− 909. (3) Uetrecht, C.; Rose, R. J.; van Duijn, E.; Lorenzen, K.; Heck, A. J. Chem. Soc. Rev. 2010, 39, 1633−1655. (4) Huang, Y.; Gelb, A. S.; Dodds, E. D. Curr. Metabolomics 2013, 1, 291−305. (5) Niu, S.; Rabuck, J. N.; Ruotolo, B. T. Curr. Opin. Chem. Biol. 2013, 17, 809−817. (6) Lanucara, F.; Holman, S. W.; Gray, C. J.; Eyers, C. E. Nat. Chem. 2014, 6, 281−294. (7) Lapthorn, C.; Pullen, F.; Chowdhry, B. Z. Mass Spectrom. Rev. 2013, 32, 43−71. (8) Kanu, A. B.; Dwivedi, P.; Tam, M.; Matz, L.; Hill, H. H. J. Mass Spectrom. 2008, 43, 1−22. 11402

dx.doi.org/10.1021/ac503379e | Anal. Chem. 2014, 86, 11396−11402