John C. Gebler is a senior scientist at Biopharmaceutical Sciences, Waters Corporation, 34 Maple Street, Milford, MA 01757.

John C. Gebler

Poly(ethylene glycol) (PEG) is a polymer composed of repeating subunits of ethylene oxide. PEG and its functionalized derivatives can be produced in linear or branched forms with various molecular masses. Because PEG possesses many unique properties, for example, high water solubility and low toxicity, it is widely used in a range of biomedical applications (1). For instance, modification of protein-based therapeutics by chemical attachment of PEG chains to the proteins (i.e., PEGylation) has been demonstrated to be an effective approach for reducing immunogenicity, increasing circulating half-life and improving stability of the biopharmaceuticals (2). 

A major challenge faced in PEG applications is to ensure the quality and stability of PEG-based materials before the PEG-ylation process. In contrast to a protein that normally shows a single molecular weight, PEG is a polydisperse material with a molecular weight distribution (MWD). In addition, the polymer chains might have different end groups because of different initiation and termination processes, thereby creating a functionality type distribution (FTD). All of these have created complexity and diversity to the polymer, thus imposing great challenges to analytical characterizations of the materials. 

Desired information about PEG characterizations normally includes average molecular weight (MW), MWD or polydispersity, structural, compositional, and end-group identification. Many techniques are available for characterizing PEGs. However, no single technique could completely characterize all the attributes related to the material. Frequently, applications of multiple analytical techniques such as gel-permeation chromatography (GPC), nuclear magnetic resonance (NMR), Raman spectroscopy, and mass spectrometry (MS)  are required for thorough structure elucidation, with each technique addressing one aspect of the characterizations (3–6). 

Among the analytical techniques currently used in the characterization of synthetic polymers, mass spectrometry is of increasing importance (7). Mass spectrometry can be used to determine the molecular mass of synthetic polymers directly with unparalleled accuracy and precision. In addition, the accurate mass measurement attained by mass spectrometry yields much useful information for the repeat unit and end-group analysis, for providing evidence for the existence of copolymers and impurities, and as a quality control protocol to confirm synthetic pathways and batch-to-batch compositional variation (8–10). 

Both matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) have been useful for polymer analysis. However, for synthetic polymers with MW much greater than 5 kDa, MALDI is often a preferred ionization method for MS analysis because of the simplicity of the mass spectra, which show mainly singly charged molecular ions (11). The limited use of ESI in PEG analysis, especially in high molecular weight PEG, is largely restrained by the complex spectra resulted from an ESI process. One of the primary features of ESI is that it tends to generate multiply charged ions, and the propensity for generating multiply charged ions also increases with growing molecular weight for a given polymer (12). The convolution of a charge state distribution with a broad MWD of PEG can result in highly complex spectra from which the MWD is very difficult, if not impossible, to obtain. Multiple charging, therefore, limits the PEG oligomer size range amenable to molecular weight characterization by ESI. The upper limit is essentially defined by the resolving power of the mass analyzer. 

A majority of the work thus far using ESI for PEG analysis has focused on relatively low molecular weight PEGs to reduce the number of charges retained on the polymer ions, thus reducing the spectral complexity. For higher molecular mass polymers, a separation method is also helpful in reducing the MS spectra complexity by limiting the range of oligomers being introduced into the electrospray source at a given time (13). To overcome the hindrance in the mass spectrometric analysis of high molecular weight PEGs, a chemical deconvolution approach was also developed which uses ion–ion or ion–molecule reactions to greatly reduce the number of charges PEG ions carried to facilitate mass measurements (14–16). In one method, neutral crown ethers were successfully used to reduce charging of PEG cations in an ion trap mass spectrometer (14). Another interesting method presented a facile study using a base-mediated, gas-phase proton-transfer reaction in the source region of an ESI time-of-flight (TOF) mass spectrometer for the characterization of PEG and PEG-protein conjugates (15). Both methods have demonstrated to be effective at reducing the multiple charging of polymers in ESI analysis and producing accurate MW measurement for a range of PEGs. 

The work presented in this article describes an alternative approach to reducing the charge state and thus extending the mass range of PEG amenable for ESI analysis by use of a commercial ion-mobility TOF mass spectrometer. The charge reduction was achieved by conducting an ion–molecule reaction between highly charged PEG ions and a neutral gaseous base inside the mass spectrometer. The charge states of highly charged PEG ions were dramatically stripped down to a few narrowed charge-state distributions, thus greatly reducing the overlapping of PEG peaks from adjacent charge states and resulting in very simplified MS spectra. The chemical deconvolution process allows the molecular weight of PEG to be readily determined and the polydispersity more clearly identified. In contrast to the previous work (15) in which ion reactions were performed in the source region, this approach has the benefits of easy operation and less consumption of superbase materials for a given reaction. More important, the approach presents a simple and flexible method to modify the configurations of a commercial instrument for a tailored application. In addition, the ion-mobility capability of the instrument allowed differentiation of various chain lengths of the polymer, thereby providing enhanced specificity to the characterization of the materials. 

OH) and isopropanol (IPA) were purchased from 

 (Pittsburgh, PA). Ultrapure water (18.2 M) was obtained from a Milli-Q water purifier (

, Medford, MA). Ammonium acetate, cesium iodide and superbase, 1,8-diazabicyclo[5.4.0]-undec-7-ene (DBU), were purchased from 

  (St. Louis, MO). PEG 4450 was obtained from 

 (Milford, MA). PEG-aldehyde 20K was purchased from NOF Corporation (Tokyo, Japan). All materials were used without additional purification. 

 All polymer samples were prepared at analyte concentrations of 1–2 mg/mL (10

  M based on the nominal molecular mass) either in a 50/50 (v/v) solution of water and methanol or 10 mM ammonium acetate and methanol. The solutions were directly infused into the ESI source at 10 µL/min by a syringe pump (Harvard Apparatus, Holliston, MA). 

 All experiments were performed on a Synapt high-definition mass spectrometer, a hybrid quadrupole ion-mobility TOF mass spectrometer (Waters Corp.). Detailed descriptions of the instrument was found elsewhere (17, 18). A schematic diagram of the instrument is shown in Figure 1a. 

The instrument was operated in positive-ion mode with a capillary voltage of  3.0 kV. The sampling cone voltage was set at 40 V. The ion source block and nitrogen desolvation gas temperatures were set to 100 °C and 250 °C, respectively. Nitrogen was used as both the nebulizing gas and the desolvation gas. The desolvation gas was set to a flow rate of 250 L/h. The quadrupole was operated in nonresolving mode to transmit a wide 

The instrument can be operated in two modes: TOF mode and TOF-mobility mode. For TOF-mobility experiments, the ion mobility cell pressure was maintained at 0.5 mbar (N

) and separations were optimized by tuning the wave pulse height from 5 to 14 V. The transfer cell pressure was maintained at 0.02 mbar unless stated otherwise. Pressures in the trap and IMS cell were the direct readbacks from the active pirani gauges equipped with the instrument. The RF amplitude used on the trap, ion mobility cell, and transfer cell was 320 V peak-to-peak. Ion mobility cell traveling wave velocities of 240 m/s were used. 

The instrument was externally mass calibrated with cesium iodide (1.0 mg/mL in 50% IPA–50% water). Data acquisition and processing were controlled by MassLynx 4.1 software (Waters Corp). Mass spectra were acquired every 1 s in TOF mode and every 2 s in TOF-mobility mode from 

 10,000 for PEG-aldehyde 20K for 6 min. Additional data processing to yield the information on the average molecular weight and the polydispersity of the PEGs followed the methods described below. 

The method used to obtain the average MW of PEG was adapted from the method published by Bagal et al. (15). Briefly, the experimental data, after processing by MassLynx for smoothing and centering, were fit to a Gaussian function to determine the average MW and the polydispersity of PEGs, expressed as Equation 1 as follows: 

 is the center of the peak (i.e., average MW, and 

 is the width of the peak. The MWD of the polymer is calculated as 3

Because full isotopic resolution at baseline was achieved for PEG 4450, the experimental data for PEG 4450 were first smoothed using the Savitzky Golay filter from MassLynx and then centered. The most abundant isotopic cluster peaks (

C2) were selected and used to fit the Gaussian function. The final value that corresponds to the average molecular weight was subtracted off two 

C to compensate the extra mass account included in the curve-fitting process. For larger MW PEGs, the isotope peaks are either partially or not resolved. In this case, spectra for large PEGs are smoothed using a wavelet thresholding to generate only one centered mass value for each PEG oligomer at each charge state. These mass values were used for the curve-fitting process. The processed data from MassLynx were exported into the data analysis and graphing software OriginPro8 (

, Northampton, MA) for plotting and performing Gaussian curve-fitting. 

The ion–molecule reactions were performed in the trap cell of the Triwave in the mass spectrometer (see Figure 1a). To perform the ion–molecule reactions, some minor modifications to the gas pipelines of Synapt were made so that neutral reagent vapor could be introduced into the trap cell. However, all of the modifications are external to the mass spectrometer. A schematic diagram of the mechanical assembly surrounding the manifold is shown in Figure 1b. 

The superbase reagent (liquid) was placed in a custom-made glass tube (6.35 mm [0.25 in. o.d.] x 3.81 mm [i.d.] x 152.4 mm (6-in. length)], which was connected to a needle valve (Catalog No. 6060795, Waters Corp.) through a union connector (Catalog No. SS-400-6-2, 

 ). The needle valve was then connected with stainless steel tubing (0.125 in.) to a two-way switching valve (Catalog No. SS-41GS2, SV1, Swagelok), which was linked to a three-way union using the same type of stainless steel tubing. One arm of the three-way union was connected to the trap gas inlet of the analyzer housing, and the third arm was extended to the existing argon collision-gas pipeline of the instrument by means of the second two-way switching valve (SV2, see Figure 1b). The glass tube and the rest of the vapor flow path were heated to 100–150 °C using heating tape (Catalog No. 14-488-28, Fisher Scientific) to ensure rapid evaporation of DBU and to keep the superbase vapor from condensing back to liquid.  

The incorporation of the two two-way switching valves in the gas pipelines allows the instrument setup to be readily switched between the standard configuration and a configuration for introducing reagents of interest into the trap cell. An appropriate flow of superbase vapor into the trap was obtained by adjusting the needle valve at a chosen temperature. In a typical charge-reduction operation, both the superbase vapor and argon gas were simultaneously introduced into the trap cell, and the concentration of the reagent introduced into the trap was controlled by changing the argon gas flow rate through a software-controlled flow controller. Under the experimental conditions where only reagent vapor is desired in the trap cell, the two-way switching valve (SV2) was completely shut off. On the basis of the relative flow rates of the superbase vapor and argon gas, the typical pressure of trap cell during the experiments was maintained at 0.008–0.05 mbar (uncorrected, as measured at the vacuum housing ion pirani gauge). After experiments were completed each day, the Triwave section was isolated from the superbase gas line (by switching off the SV1 and turning on SV2) and placed back to the standard configuration. 

Analysis of PEG 4450 before charge reduction.

 An ESI-TOF analysis of PEG 4450 before the reaction with superbase DBU results in a complex ESI spectrum (see Figure 2a). As shown in the inset of Figure 2a, the instrument resolving power at the collected 

 window is sufficient to provide isotopic resolution for peaks in the spectrum. The isotopic resolution permits the determination of the number of charges that each oligomer holds as well as the charge states distribution in the spectrum. The data show that several charge states, from 2+ to 4+, are observed in Figure 2a, suggesting the attachment of multiple cations to each PEG oligomer. On the basis of these calculations, it is concluded that charge carriers (cation adduction) are not completely contributed by one type of metal ions (e.g., Na+). Some oligomeric ions are possibly generated by the attachment of protons (and also possibly K+), yielding many different ion series in the spectrum. As a result, the average molecular weight cannot be readily determined. 

 distributions for the PEG 4450 sample analyzed in the TOF-mobility mode (IMS–MS). In these experiments, the time required for IMS separations is less than 20 ms. As shown in Figure 2b, ions with different charge states, or different conformers of the same 

 value, were resolved by IMS. In comparison with the ESI-TOF spectrum of this sample (see Figure 2a), at least four resolvable charge envelopes (from 1+ to 4+) are observed in the IMS analysis. Individual oligomers can be clearly resolved not only in the 

 dimension but also by their mobility drift time. The separation greatly simplifies the complexity of the spectrum such that some of the minor components in the samples that can not be observed otherwise can be easily identified from the analyses. For example, without the mobility separation, the 1+ envelope from PEGs with a shorter chain length is not observable in the mass spectrum (see Figure 2a), demonstrating an increase in dynamic range when ion mobility is coupled with MS. Several higher charge-state envelopes are also present in the spectrum, but they are not well resolved from each other, thereby making complete characterization difficult. 

Analysis of PEG 4450 with charge reduction. 

To further simplify the mass spectrum, gaseous DBU was introduced into the trap cell. Upon the reaction, some of the protons are stripped from the multiply charged PEG ions, and the charge-state distribution is reduced to primarily a single charge state (2+) as illustrated in Figure 3. In this example, the multiply charged PEG ions were reacted with the DBU in the trap cell, and a symmetric distribution of doubly charged ions at a mass-to-charge interval of 22 

 was generated. Coexisting with the 2+ charge state ion species are ion series with the charge state of 1+. The MWs of these PEGs are estimated to be much lower than the one from the charge envelope of 2+, indicating that these peaks are possibly generated by the impurities in the sample. 

When the data from the series of doubly charged ions from the ion–molecule reaction between PEG 4450 with DBU (see Figure 2b) are fit to a Gaussian function, the mean (

) of the best fit Gaussian function are 2240.2 and 240, respectively. The dots in the inset of Figure 3 show the data points (peak maxima) that are used for the fitting process. As can be seen from the inset, the data fit the distribution well. The quality of fit is also measured by the correlation coefficient, which has a value of 0.9977. Because this is a doubly charged envelope, these values correspond to an average MW and MWD of 4480.4 and 720 Da, respectively, for the PEG 4450. In comparison, the average MW provided by the supplier, as was determined by GPC, is 4450. The value obtained by the current work is in good agreement with the value measured with an orthogonal method. 

Analysis of PEG-aldehyde 20K with and without charge reduction. 

 The ESI-TOF spectrum of PEG-aldehyde 20K (see Figure 4a) is more complex compared with that of PEG 4450. Individual oligomers are not discernible in the spectrum, and the spectrum comprises relatively broad "noisy" bands, which are attributed to the overlapping of the charge states, the polydispersity of the oligomers and the compression of the 

 range by the high-charge states. As noted previously, a PEG sample consists of a mixture of oligomers, each of which can have various charge states. The number of possible combinations of oligomer size and charge state increases rapidly with increasing nominal molecular weight. At a MW of 4450, there are as many as four or more discernible ions within an interval of one unit of 

 1000. It is believed that as many as 28 charges can be placed onto an oligomer chain with a mass of 20 kDa (6). Under the resolving power of a TOF analyzer, such complexity results in spectral congestion (see Figure 4a). Consequently, it is not possible to extract molecular weight information from the data. 

Figure 4b shows the mass spectrum resulting from reacting PEG 20 k with DBU in the trap cell. Two dominate charge state envelopes, 4+ and 5+, are observed. Detailed examinations of the spectrum also indicate that peaks corresponding to the individual oligomers of the polymer are baseline resolved. However, unlike the data for PEG 4450, isotopic resolutions for these PEG 20K ions are not observed.  

The appearance of the discernable PEG oligomers makes it feasible to determine the average MW of the PEG. To determine an average MW of the polymer by the curve-fitting method, an overall MWD represented by the total ion abundance of each oligomer is needed. Because 4+ and 5+ charge states dominate the charge-stripping spectrum, the summation of ion abundance from both charge states for each oligomer should naturally yield a close reflection of the original MWD of the polymer. For this purpose, the 

 value at each peak under the charge envelops of both 4+ and 5+ charge states was first converted to "zero-charge" distributions by multiplying with its respective charge states and subtracting the mass of sodium multiplied by the charge. This mathematical calculation is based on the assumption that the observed charge envelopes resulted from the sodium ion attachment to the PEG oligomers, as has been reported in the literature (15). The final assembly of the ion abundance was achieved by summing the abundances of the corresponding peaks of the same oligomer from these "zero-charge" distributions for both the 4+ and 5+ envelopes. The zero-charge distribution obtained by combining the distributions from the 4+ and 5+ charge envelopes was fitted to a Gaussian (dashed line, see Figure 4c), which provides an average MW of 21,552 Da and a MWD of 3228 Da. This MW value is in close agreement with the literature value measured by similar MS approach as well as an NMR method (21,920 Da) (15). 

It is noteworthy that MWs of polymers measured by mass spectrometric methods can potentially deviate from values obtained by other methods. Mass bias, ionization suppression, and instrument tuning conditions were cited as possible sources of errors in the mass measurement (6, 15). In addition, inhomogeneous ion–molecule reaction across many different PEG oligomer ions could bring in an additional source of deviation for the current approach. However, the good agreement between the measured values in current work and GPC values suggest that the potential impact of possible variables was minimal in our experimental setup, and accurate MW values for high-mass PEGs can be obtained. 

Several observations can be made from the ion–molecule reactions. First, the final charge states as well as the charge-state distribution depends on the partial pressure of DBU vapor in the cell. Because the transit time for PEG ions passing through the trap cell is limited by their collisions with the gas molecules, higher gas pressures imply more collisions taking place in the trap cell before ions exit out. Higher gas pressures obviously favor longer reaction time, and thus lower charge states could be obtained. Second, stronger bases are required to reduce the charge to the same terminal charge state for longer polymer chains. This observation is likely caused by a combination of two factors: First, the longer polymer chains have a higher initial charge state, requiring more efficient charge stripping reaction to reduce to the same charge state. In addition, longer polymer chains can retain an additional charge more effectively, which would make an additional charge more difficult to remove.  

This work provides a method for improving mass determination of high molecular weight poly(ethylene glycol) (PEG) by means of an ion–molecule reaction inside a Synapt high definition mass spectrometer. The improved performance results from reductions in the spectral complexity of polydisperse PEG materials. The reduction in spectral complexity yielded resolved oligomers and allowed the average molecular mass and polydispersities to be determined for the polymers. In addition, the combination of this ion–molecule reaction methodology with ion mobility separations and TOF mass spectrometry has provided enhanced sensitivity and specificity for characterizing PEG on the basis of chain lengths and for differentiating the low molecular weight impurities in the PEG materials. All the data demonstrate this method has great potential to rapidly characterize PEG materials to the extent unachievable by other methods.  

The work also introduces a simple and flexible method for modifying a commercial MS instrument with off-the-shelf components. The modifications have effectively coupled the high-performance tandem mass spectrometer with gas-phase ion–molecule reactions together without sacrificing any high performance attributes of the original instrumentation. With many unique capabilities of the instrument, such as ion–mobility separation, the ability to perform ion–molecule reaction inside Synapt has truly expanded the applicability of the instrument. It is conceivable that this configuration can be readily applied to many challenging analytical tasks in pharmaceutical industry. 

Introduction to Biotechnical and Biomedical Applications of Poly(Ethylene Glycol) in Poly(Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications (Topics in Applied Chemistry),

 J.M. Harris, Ed. (Plenum, New York, 1992). 

 2. A. Kozlowski and J. M. Harris, "Improvements in Protein PEGylation: Pegylated Interferons for Treatment of Hepatitis C," 

 3. D.J. Winzor. "Analytical Exclusion Chromatography," 

 4. J.Y. Shey and C.M. Sun, "Liquid-Phase Combinatorial Reaction Monitoring by Conventional 1H NMR Spectroscopy," 

 5. J.L. Koenig and A.C. Angood, "Raman Spectra of Poly(ethylene glycols) in Solution," 

 6. P.B. O'Connor and F.W. McLafferty, "Oligomer Characterization of 4-23 kDa Polymers by Electrospray Fourier Transform Mass Spectrometry," 

 7. S.D. Hanton, "Mass Spectrometry of Polymers and Polymer Surfaces," 

 8. C.G. de Koster et al. "Endgroup analysis of Polyethylene Glycol Polymers by Matrix-Assisted Laser Desorption/Ionization Fourier-Transform Ion Cyclotron Resonance Mass Spectrometry,"

Articles

The authors developed a method to accurately measure the average molecular weight of large poly(ethylene glycols) (PEGs) using ion-mobility time-of-flight mass spectrometry coupled with gas-phase ion–molecule reactions.

Improved Mass Determination of Poly(ethylene glycols) by Electrospray Ion-Mobility Time-of-Flight Mass Spectrometry Coupled with Ion–Molecule Reactions