Christopher Southan SmithKline Beecham Pharmaceuticals
Introduction
The many advantages of reducing the internal diameter of HPLC columns for protein and peptide chromatography have been described in a recent ABRF News article on capillary HPLC (1). However, with high-performance packings, these columns, from 0.5 to 0.1 mm internal diameter (ID), require specialized or adapted equipment capable of running gradients in the range of 20 ml/min down to 2 ml/min. A useful compromise between standard HPLC and dedicated microbore or capillary instrumentation has been made possible by two developments. The first is the availability of chemically inert plastic HPLC fittings and tubing that can provide a complete set of components for easy assembly of small-diameter homemade columns. The second is the development of new chromatographic materials specifically designed for protein and peptide separations at high linear flow velocities with low operating pressures (2, 3). These materials are well suited for homemade small-ID columns not only because they can be packed and operated at moderate pressures but also because the high flow rates allow these to be operated in the 100 ml/min to 500 ml/min range. This provides the option of using conventional HPLC pumps and detectors as well as microbore instrumentation. This article describes the construction and some applications of a range of small-ID columns that can be easily packed and assembled in the laboratory from standard HPLC tubing, fittings, and parts.
Column Components
For most of us who have become adept in manipulating HPLC fittings (possibly more from necessity than choice) the column end-fitting design outlined in Figure 1 is very familiar. Within the universal nut-tube-ferrule-frit-union assembly concept there is now a wide choice of both stainless steel and plastic components (PEEK, Teflon, Tefzel, Vespel, Kel-F, etc.) based on 1/16" and 1/8" outer diameter (OD) tubing. An outline of the properties associated with the various types of tubing that can be easily secured into 1/16" and/or 1/8" fittings for homemade column construction are given in Table 1.
Figure 1: Diagram of a 1/16" HPLC end fitting adapted as a packing bed support for a glass capillary column (adapted from reference 4). This was assembled from a Kel-F union from Alltech (part number 3240) and a 1/16" 2 mm frit with a PEEK annular ring from Upchurch (part number C-407). The 1/8" fittings and tubing have essentially the same construction as the 1/16" assembly shown here.
Some practical caveats associated with some listings in Table 1 are as follows. The manufacturers' pressure ratings are solvent dependent and for 1/8" tubing are always below those for 1/16". With a little practice the 1/16" OD thick-walled glass tubing can be cut cleanly, but cutting 1/8" OD glass tubing is more difficult.
The choice of tubing is likely to be dictated by the practical experience of the chromatographer and what may be sitting in the
lab drawer. PEEK combines the most advantages, but the easiest material with which to acquire home-packing experience is probably Tefzel, which is flexible, easy to cut, and allows packing to observed. The major manufacturers of plastic HPLC tubing and fittings are Upchurch, Jour, and Optimax. All these suppliers offer a full range of 1/16" fittings, but to date apparently only Upchurch and Optimax supply the 1/8"-1/16" plastic union adapters appropriately threaded for the assembly of 1/8" OD columns. Glass tubing is not usually included in HPLC supply catalogues. However, glass capillary tubes can be obtained with an OD that will fit into the 1/16" fittings and an ID of less than 1 mm in order to ensure adequate wall thickness. In the work described in this article, two sizes of pre-calibrated glass capillary tubes from Vitrex (125 mm in length) that fulfill these criteria were used (part numbers DMP 1266, double-white band, calibrated for 25 ml with 0.65 mm ID; and DMP 1270, green band, calibrated for 50 ml with 0.96 mm ID).
Almost any combination of tubing + plastic low-dead -volume unions + plastic ferrules + plastic or metal frits + plastic male nuts can be used, but there are some practical limitations on compatibility. An example is the importance of using a softer ferrule material such as Kel-F or Teflon, to tighten onto glass or Tefzel tubing. Ferrules of PEEK, although very effective with PEEK tubing, are hard enough to crack glass and may "pinch" Tefzel. The double-sided ferrule design will secure tubes with less torque but the single or integral ferrule designs are also adequate for PEEK tubing. Both hand-tight and hexagonal -headed nuts can be used, but care has to be taken not to overtighten the headed nuts. The particular choice of frits for the lower bed support is also broad. These are available in steel or more recently PAT (PEEK alloyed with Teflon) for better biocompatability. It is also easy to make simple frits from wide-pore PVDF or from pads removed from solid-phase extraction cartridges.
Table 1: Properties of Tubing for Microbore Columns
Properties Steel Glass PEEK Teflon Tefzel Wide range of IDs available + + + Flexible + + + Can withstand > 2k p.s.i. + (+) + Transparent + + + Not pinched by ferrules + + + Easily and cleanly cut (+) + + + Low protein adsorption + + +
Packings
In principal any type or source of chromatographic material can be packed into a homemade column as described in this article. However, the practical convenience of using plastic fittings and simple packing methods make some materials inappropriate for small-particle (less than 10 mm) silica packings or long columns (greater than 100 mm) needing high packing
pressures and theoretical plates. Essentially anything else can be tried, including loose preparative material, pellicular guard material, and packing extracted from guard cartridges or discarded columns. Standard 10 mm reversed-phase silicas are quite adequate for protein separations (4), but these do not provide the advantages of the high-velocity materials for any of the common adsorptive modes of elution mentioned in the introduction, which include Poros (PerSeptive Biosystems) (2), Resource (Pharmacia), PLRPS (Polymer Laboratories), and HyperD (Beckman/Biosepra). For size-exclusion chromatography any robust packing should be effective, but my experience so far has been limited to Sephadex (medium) and Superdex (3), both from Pharmacia.
Packing Methods
The simplest packing method is to dry-pack the empty tubing by shaking particles down with the aid of vibration from a sonicator bath or an engraving tool (4). A cut-back pipette tip serves as a reservoir at the top, and the tube is plugged with parafilm or a tube cap at the bottom. Because of the particle dust generated, this is best performed in a fume hood.
The dry-packed tube is then secured into the union with a ferrule, frit, and male nut, as illustrated in Figure 1. The same fittings, without a frit, are then used to assemble the top of the column. Packing is completed by running the appropriate starting solvent for that packing at approximately 50-100% higher than the expected operating flow rate. With reversed-phase packings some acetonitrile or methanol should be included in the packing solvent. This causes the dry packing to wet and compress by approximately 20-40%.
When bed compacting has ceased and running pressure has stabilized, typically within 30 minutes, the tubing is cut down to the bed surface and re-assembled ready for use. Slurry-packing is easily performed with the Poros Self-Pack reservoir (PerSeptive Biosystems). The empty column is attached to the packing reservoir before being filled with the appropriate amount of dilute slurry. The end of the reservoir column is then screwed on firmly with the aid of a wrench. For constant-pressure packing any instrument or single HPLC pump unit can be used provided that they ramp-down flow rate when the maximum pressure is reached rather than shutting down. For the PEEK tubing, packing can be performed with the maximum pressure set at 2000 p.s.i. Because the opacity of PEEK prevents the observation of the packing bed, it is expedient to pack two to five column lengths of tubing (e.g., 10-60 cm) that can be equilibrated to 2000 p.s.i. over several hours or overnight. By cutting the appropriate lengths of packed tubing from the bottom, several full columns can be obtained until the cut exposes empty tubing. A 50 mm length is sufficient for analytical protein separations. To increase peptide resolution or for micro-preparative capacity, columns could be cut to 100 mm. For packing the softer size-exclusion gels into Tefzel tubing, the maximum packing pressures were reduced to 500 p.s.i. With this translucent plastic the gel surface is visible so the progress of packing can be observed. The tube could then be cut back to remove any dead volume and re-assembled.
The same method was used for glass tubing, but packing pressures were held below 1000 p.s.i. To avoid the expense of the Poros Packer, slurry packing can also be carried out either by using an empty commercial 4.6 mm PEEK column (Alltech or Upchurch) as a reservoir or by using 1/8" tubing to construct an empty column.
During the development of these columns the advantages of omitting the top frit became apparent, particularly as these are a notorious cause of blockage in commercial microbore columns. A great convenience with the plastic end-fitting is the ability, when necessary (e.g., following completed bedding down, discoloration, or pressure rises) to disassemble and cut back the topmost few mm of tubing. This removes incremental dead volume and any blockage or contamination. This restorative trimming can be repeated on longer columns with only a minimal effect on protein separation parameters. A consequent advantage is that scarce samples do not have to be filtered before injection.
Tuning the HPLC Instrumentation
The columns described in this article can be operated on any instrument, even with dead volumes in excess of 1 ml, provided extended run times are acceptable. However, it is usually possible to reduce dead volumes by minor modifications such as those described here for a standard Beckman System Gold with a model 126 solvent module and a model 167 dual wavelength detector. These were:
·attaching a Rheodyne 8125 injection valve and a 20 or 50 ml PEEK injection loop
·attaching a Beckman microbore dynamic mixer
·bypassing the purge valve by direct connection of the mixer outlet to the injector inlet by 0.01" ID PEEK tubing, and
·reducing the lengths of both inlet and outlet tubes from the UV detector to 2.5 cm.
These modifications reduced the volume delay between the electronic initiation of a gradient and the detection of B solvent in the flow cell (excluding the column or injector loop volume
contributions) from 1.5 to 0.5 ml. Another tip to reduce gradient initiation delay is to maintain flow and pressure on both pumps by using a small amount of solvent B (2-5%) during column equilibration and initialization.
Results and Discussion
With practice it is possible to prepare home-packed columns of the type described ready for evaluation within 1 hour. The sensitivity that can be achieved is demonstrated in Figure 2 where sub-picomole amounts of BSA were detected. Absorption at 205 nm was used to give twice the sensitivity of detection at 215 nm. An additional gain in sensitivity is derived from the long -pathlength flow cell in the standard UV detector.
Figure 2: High-sensitivity quantitation of a BSA standard. Column: 0.25 x 50 mm Tefzel packed with 10 mm Poros RI. Solvent A: 0.08% TFA, solvent B: 0.06% TFA in 80% acetonitrile. Gradient: 10-100% B in 9 minutes at 100 ml/min. Pressure: 850 p.s.i. Sample: 1 ml of BSA at 50 mg/ml (50 ng protein).
The essence of speed is demonstrated by the separation in Figure 3. This could even be run faster but may be limited by detector response times. A good application would be rapid reversed-phase HPLC desalting.
Figure 3: High-speed separation. Column: 1/8" OD PEEK, 1.5 mm ID x 50 mm, packed with 10 mm Poros RII . Solvent A: 0.08% TFA, solvent B: 0.08% TFA in 80% acetonitrile. Gradient: 20 -70% B in 30 seconds at 1.5 ml/min. Pressure: 1200 p.s.i. Sample: 5 ml of a myoglobin/RNAse mixture, 5 mg/ml each. The y-axis shows absorption at 215 nm.
In Figure 4 a short column is used to follow the time course of a tryptic digest. This application shows a compromise between speed and sensitivity. The peptide resolution using the Poros packing is limited but adequate to monitor the conversion of protein into peptide products.
Figure 4: Monitoring proteolytic digestion. Column: 0.65 mm x 50 mm glass capillary packed with 10 mm Poros RII . Solvent A: 0.08% TFA, solvent B: 0.08% TFA in 80% acetonitrile. Gradient: 5-70%B in 5 minutes at 0.5 ml/min. Sample: the equivalent of 2 mg of protein was injected for each run. Prior to digestion the protein concentration (1 mg/ml) and the enzyme:substrate ratio of 1:20 were determined by HPLC peak height measurements. Chromatogram A shows undigested sample; chromatogram B was obtained 15 minutes after trypsin addition, and chromatogram C after 180 minutes. Chromatogram D shows a trypsin solution incubated as a control for autodigestion, at 160 minutes.
All the chromatograms in Figures 2-4 also show the potential of this high-speed chromatography for direct and sensitive protein assay simply by measuring the area under the peak at 205 or 215 nm and comparing with any standard protein run under the same conditions. The extension of this technology to an anion-exchanger is shown in Figure 5. This demonstrates another useful characteristic of high-speed packings, a very short re-equilibration time between injections. This can be monitored by the rapid return of the UV baseline to pre-injection levels.
Figure 5: Consecutive anion-exchange chromatography of ovalbumin. Column: 0.96 x 60 mm glass capillary packed with Poros IIQ material. Solvent A: 25 mM Tris, pH 8.0, solvent B: 25 mM Tris, pH 8.0, 0.5 M NaCl. Gradient: 0-100% B in 5 minutes at 500 ml/min. Sample: 10 ml of ovalbumin (5 mg/ml). The point of injection of a second sample is indicated on the time axis.
A third separation mode is shown by the application of the size-exclusion column for protein desalting and solvent exchange in Figure 6. In terms of absolute amounts of protein injected, high detection sensitivities can be achieved compared with conventional size-exclusion chromatography column diameters. If Tris buffer concentrations are low, it is possible to use 215 nm for detection, which gives an approximate 20-fold gain in detection sensitivity over 280 nm.
Figure 6: Size-exclusion chromatography. Column: Sephadex G25 Superfine packed into 2 x 800 mm Teflon tubing secured into 1/8" fittings. Solvent: 25 mM Tris, pH 8.0. Flow rate: 200 ml/min. Pressure: 70 p.s.i. Sample: 5 ml of BSA (1 mg/ml) with sodium azide, which forms a salt peak detected at the column's total volume at 13 minutes.
The results above demonstrate that simple homemade column technology, combined with high linear flow rate materials, is clearly of great utility for small-scale separations where some compromise in sensitivity and resolution can be accepted in favor of speed. They also have the advantage of being usable in standard HPLC equipment. The main limitation is that they are unsuitable for high packing or operating pressures. They may not match the performance of commercial columns where a close equivalent is available, but they can be an effective low-cost substitute. The design is inherently flexible for a wide range of packing/tube length combinations to be prepared, and they are easily scalable in capacity from 1.5 mm ID (for 1/8" OD) to 0.5 mm ID (for 1/16" OD). They, therefore, cover a useful intermediate niche in the nanomole to low-picomole sample range, i.e., somewhere between commercial 2 mm ID columns and the more technically demanding capillary columns (1). The small amounts of packing used not only allow new or unusual packings to be
quickly compared in the same column configuration with minimal amounts of protein but also make the columns inherently cheap and therefore disposable. This broadens the scope of their application to the use of harsh solvents and crude samples that could compromise the lifetime of commercial microbore columns.
References
1. Swiderek, K.M., Ducret, A., and Kasel, D.B. (1996) Methods for capillary isolation and collection of peptides. ABRF News 7(2), 17-19 (and references therein).
2. Afeyan, N.B., Gordon, N.F., Mazsaroff, I., Varady, L., Fulton, S.P., Yang, Y.B. and Regnier, F.E. (1990) Flow -through particles for the HPLC separation of biomolecules: perfusion chromatography. J. of Chromatog., 519, 1-29.
3. Kagedal, L., Engstrom, B., Ellegren, H., Liebel, A.K., Lundstrom, H., Skold, A. and Schenning, M. (1991) Chemical, physical and chromatographic properties of Superdex 75 prep grade and Superdex 200 prep grade gel filtration media. J. of Chromatog. 537, 17-32.
4. Southan, C. (1989) The use of glass capillary tubes as disposable microbore columns for RP-HPLC of proteins and peptides. In Techniques in Protein Chemistry (Hugli, T., ed.) New York: Academic Press, pp. 392-398.
The author may be contacted at SmithKline Beecham Pharmaceuticals, Dept. of Molecular Recognition, New Frontiers Science Park, Third Ave., Harlow, Essex CM19 5AW, U.K. Tel: (44) 1279-622184, Fax: (44) 1279-622555, E-mail: Christo -pher_D_Southan@sbphrd.com.
Return to the The ABRF Home Page