|Home | About | Journals | Submit | Contact Us | Français|
The use of poly(2-ethyl-2-oxazoline) with an Mn of 500,000 g/mole in a sieving matrix for SDS capillary electrophoresis size separation of proteins is investigated using polymer solutions with concentrations between 6% and 12% w/v. Optimal separation efficiency is obtained using 10% w/v, with an average separation efficiency of 150,000 (±12,000) theoretical plates observed for myoglobin (MW = 18 kDa) and 8,800,000 (±2,400,000) theoretical plates for carbonic anhydrase (MW = 30 kDa) for fourteen capillaries with an 80-cm effective length.
Some adsorbed polymers were found to be effective in reducing EOF when applied to a clean capillary; however, a covalently-attached polyacrylamide coating was found to be much less troublesome after initial wall treatment.
Two different buffer systems, (1) 25 mM HEPES and (2) 20 mM Tris with 40 mM or 60 mM tricene, were tried with 0.05% or 0.10% w/v SDS added. Tris-tricene gave generally better current stability than HEPES but with no observed improvement in separation efficiency.
This polymer has all the desired characteristics for an SDS-CE sieving matrix using LIF detection, including hydrolytic stability, optical clarity, low viscosity, acceptable hydrophilicity, and commercial availability.
SDS-PAGE is a common method for performing size-based separation of proteins. This method has been adapted to capillary gel electrophoresis, SDS-CGE, using a replaceable sieving matrix. In the past, linear and branched polymers (non-cross-linked) such as polyacrylamide, poly(ethylene glycol), dextran, pullulan and poly(ethylene oxide) have been used for sieving matrixes. Generally, a coated capillary is required to avoid displacement of the sieving matrix by endo-osmotic flow, which was found to occur even for the more viscous, linear polyacrylamide at concentrations less than 4% w/v when using SDS. More recently, hydroxypropylcellulose was used as a sieving matrix using LIF detection. LIF detection has the advantage of several orders of magnitude higher sensitivity for low-abundance proteins and wider dynamic range for protein profiling compared to UV detection. Using LIF detection, proteins from a single cell were detected, first using pullulan  and later using PEO with an Mv of 100,000 g/mole  because their low viscosities allowed hydrodynamic injection of single cells into a coated capillary. In addition, different fluorescent labels can be used for test samples and standards so as to facilitate a comparison. PEO is commercially available with lower average molecular weights and typically requires a polymer concentration of only 2.0% w/v to 2.5% w/v to achieve the optimal average mesh size for protein separation. However, PEO suffers from rather poor hydrolytic stability even when stored at 4°C, as can readily be confirmed by viscometry, and marginal solubility in water. Also, PEO is somewhat hydrophobic for protein separations, which may limit separation efficiency and resolution.
Other polymers also have disadvantages. The high viscosity of linear polyacrylamide makes it less attractive for use, especially with micro-fluidic chips and for hydrodynamic injection of sample. Cellulosic-based polymers, including pullulan and dextran, may be better alternatives. They are, however, derived from naturally occurring polymers, rather than synthetic, so variability is a concern. Polydimethylacrylamide and polyvinylpyrrolidone are generally considered to be too hydrophobic for efficient protein separations.
As an alternative to these, we have investigated poly(2-ethyl-2-oxazoline), PEOX, with an Mw of 500,000 g/mole, as a replaceable sieving matrix for SDS-CGE. This polymer has more hydrophilic character than PEO for increased separation efficiency of proteins. PEOX is a homolog of N,N-dimethylpropionamide with chemical formula:
PEOX has good hydrolytic stability in aqueous solutions because the amide is so highly substituted. The use of cross-linked oxazoline polymers for bio-separations is the subject of a 1994 patent, and poly(2-methyl-2-oxazoline) was reported as an alternative to polyacrylamide gel support for iso-electric focusing of proteins with higher protein solubility. Although covered in the patent, adaptation of poly(oxazolines) using the linear polymer for capillary electrophoresis has never been reported. At temperatures below the cloud point of about 60°C, aqueous solutions of PEOX are optically clear, ideal for LIF detection.
PEOX is commercially available with Mw ranging from 50,000 g/mole to 500,000 g/mole with low polydispersity and is synthesized by cationic ring-opening polymerization of 2-ethyl-2-oxazoline. We obtained PEOX with Mw=500,000 g/mole (polydispersity = Mw/Mn =3-4), from Sigma-Aldrich without further purification or fractionation to make matrixes with buffer and SDS.
SDS, HEPES, Tris, tricene, allyl glycidyl ether, and lyophilized proteins were also obtained from Sigma-Aldrich. Acrylamide (99.9 + % pure) used to make the capillary coatings was obtained from MP Biomedicals, and FQ dye was obtained from Invitrogen. Solutions were made using 18.2 MΩ-cm Nanopure water (Barnstead). 25 mM HEPES with 0.1% w/v SDS (pH = 7.4) and 20 mM Tris and 40 mM or 60 mM tricene with either 0.05% or 0.1% w/v SDS (pH = 7.6) were made from the salts and 18.2 MΩ-cm water and filtered before use. These buffers were used to make sieving matrixes with 6%, 8%, 10%, or 12% w/v PEOX, which were out-gassed before use to reduce bubble formation.
The viscosity of a 10% w/v solution of poly (2-ethyl-2-oxazoline) with a Mw of 500,000 g/mole in 20 mM Tris, 40 mM tricine, and 0.1% w/v SDS was found to be 84 cP at 30.0°C using a Gilmont falling-ball viscometer with a glass ball. This low viscosity allows easy pumping into capillaries.
A standard protocol was used to prepare protein samples for SDS-CGE with fluorescence detection. A 1 mg/mL stock solution of each protein standard was prepared by dissolving 1.0 mg of lyophilized protein sample in 1.00 mL of phosphate-buffered saline solution (pH 7.4). 10 μL aliquots of the stock solution were stored in 200 μL micro-centrifuge tubes at −20°C. One 200-μL micro-centrifuge tube was thawed and stored for up to one week at 4°C for use. To prepare SDS-CGE samples, a 1.0 μL aliquot of each protein standard to be included in the sample mixture was mixed and diluted to 100 μL by first adding 50 μL of a buffer solution containing 100 mM Tris, 2% w/v SDS, and 20% v/v ethylene glycol (or glycerol) adjusted to a pH of 9.2 with HCl and then adding 18 MΩ-cm water to a total volume of 100 μL. This pH was used to facilitate the reaction with FQ. This gave 10 μg/mL of each protein in the sample. This solution was sealed and heated to 95°C for 5 min. to denature the protein with SDS.
The resulting SDS-protein sample was conjugated with FQ dye using the manufacturer’s recommended protocol. For optimal separation efficiency, denaturing with SDS was done before labeling with FQ. First, 10 μL of 20 mM KCN in water was mixed with 5 μL of the SDS-protein solution. Then, 10 μL of 10 mM FQ in methanol was added, and the mixture was vortexed, centrifuged, and then reacted at 65°C for 3 to 5 minutes. This order of addition was used to reduce a side reaction between the FQ and the KCN alone resulting in the formation of cyanohydrins as seen for the similar reaction with CBQ. A color change to tan could sometimes be observed. FQ reacts reproducibly with the ε–amine of lysine residues and the terminal amino group of proteins and is fluorogenic (i.e., the unconjugated dye is not fluorescent).
Capillary gel electrophoresis was performed using a Spectrumedix capillary-array electrophoresis instrument using 96 capillaries with a 200-micron O.D, 50-micron or 75-micron I.D., and 100 cm length with 80 cm effective length. Capillary inner walls were coated with covalently attached polyacrylamide using the method of Hjerten.
Samples were electro-kinetically injected directly from micro-titer plates containing identical samples. Electro-kinetic injection was consistently performed using 10 kV for 30 sec, preceded by a 30-sec water injection at 10 kV, followed by electrophoresis at 100 V/cm. The non-viscous nature of the PEOX sieving matrix required that the SpectruMedix instrument be slightly modified by adjusting the height of the draining anodic buffer to match the height of the cathodic buffer to avoid siphoning. A pumping speed of 0.2 mL/min was used to fill the capillaries with the SDS-containing matrix.
Good separation results were obtained using 8%, 10%, and 12% w/v PEOX solutions with HEPES or Tris/tricene buffers and 0.05% or 0.10% w/v SDS at 30°C, although the optimum was a 10% w/v polymer solution with 20 mM Tris, 40 mM tricene and 0.1% w/v SDS.
The separation results using 10% w/v PEOX and a covalently attached polyacrylamide wall coating are shown in Figure 1. Twelve-microliter samples containing FQ-labeled equine myoglobin and carbonic anhydrase (from bovine erythrocytes) were pipetted into columns “1” and “2” of a 96-well plate, with water in the other ten sample columns. Both protein bands are narrow and well-defined. Small satellite bands are seen, which may be a result of a variable extent of reaction between the proteins and FQ or SDS. The width of the bands compared to the migration time was used to determine the number of theoretical plates, for the fourteen capillaries for which data was obtained. Using this method for the two bands as seen in Figure 1, an efficiency of 152,000 ± 12,000 theoretical plates is obtained for the first band and an efficiency of 8,800,000 ± 2,400,000 theoretical plates is obtained for the second band. The migration times and standard deviations for the two bands are 2245 ± 45 sec and 4268 ± 73 sec, respectively.
Several dynamic wall coatings were tried as an alternative to the covalently attached polyacrylamide coating. Dynamic coatings were not expected to work well with SDS present in the run buffer or sieving matrix, since SDS interferes with hydrophobic interactions between the coating polymer and the glass surface, reducing wall interactions. However, we found two dynamic coatings that gave acceptable performance.
Good results were obtained using 0.1% w/v poly(acrylamide-co-allyl glycidyl ether) in water, which was polymerized using a 35:1 molar feed ratio of acrylamide to allyl glycidyl ether using the previously described method.  The coating was easily applied after pretreatment of a clean capillary with 0.1 N NaOH and water, and it lasted several runs. In order to reapply the coating, the old coating was first removed by washing the capillary with 0.1 N NaOH for 10 min, followed by 5 min with 18 MΩ-cm water as recommended by M. Chiari et al. Next, the 0.1% w/v solution of the poly(acrylamide-co-allyl glycidyl ether) was pumped in and allowed to sit in the capillaries for at least 10 min before use. Inconsistent results were obtained if this washing of the capillaries and reapplication of the coating were not done properly.
When the wall coating was inadequate, the PEOX sieving matrix was observed to be pulled out of the capillary toward the cathode during the run. Since the sieving matrix has higher background fluorescence or scattered light than buffer, the loss of matrix caused the background fluorescence or light scattering to drop, as shown in Figure 2. This occurred 3000 sec to 7000 sec after electrophoresis was begun. The time at which the background signal changed can be used to estimate the EOF using the distance from the anode end of the capillary array to the detection window, which is 20 centimeters. The EOF obtained using this method was 2.5 × 10−9 m2/V-sec. This is considerably less than the value of EOF in bare glass with only buffer, which is in the range of 5 to 8 × 10 −8 m2/V-sec, but still unacceptable. This residual EOF resulting from an inadequate dyanamic coating was about a factor of two greater when using HEPES compared to that obtained when using the Tris /tricene buffer because of its lower ionic strength. In Figure 1, where no change in background from loss of sieving matrix is seen, the EOF is less than 1.2 × 10−9 m2/V-sec.
Poly(N-hydroxyethylacrylamide), was also tried as a wall coating, which is reported in the literature as an effective wall coating when the walls are pretreated with 1 N HCl with an EOF that is reduced to 2 × 10−10 m2/V-sec, which is comparable to the lowest reported EOF for a covalently coated capillary. The results were similar to those obtained using poly(acrylamide-co-allyl glycidyl ether) shown in Figure 2.
PEOX sieving matrix has superior separation efficiency for SDS-CE size separation of proteins because of (1) its hydrophilic nature, (2) its hydrolytic stability, and (3) its low viscosity, which is an advantage for filling capillaries and should allow hydrodynamic injection of sample. In addition, the PEOX provides a useful antifoaming property that makes it easier to handle a matrix containing SDS. With the presence of SDS in the matrix, a good covalent wall coating is preferred PEOX is not suitable for UV detection at wavelengths shorter than 250 nm. However, PEOX has all the desired characteristics for SDS-protein separations when using LIF detection.
Grant support from the National Institutes of Health, R43CA96059 and R43RR021803, is gratefully acknowledged.
Conflict of Interest
The authors have no conflict of interests in this work or its publication as open for public dissemination.