Next Generation Peptide Mapping with Ultra Performance Liquid Chromatography

Ultra performance liquid chromatography (UPLC) is a new category of liquid chromatography that researchers are using to increase resolution, speed, and sensitivity in a variety of applications. These benefits result from packing columns with 1.7 μm particles and using instruments that are optimized for such columns.

Jeff Mazzeo, Ph.D.

Peptide mapping is a complex process that uses samples requiring shallow gradients with long run times. This article describes chemical and operating considerations when adapting UPLC to peptide mapping. At optimal separation conditions, UPLC peptide maps show more resolution than other methods and increased sensitivity. UPLC can generate particularly good peak shapes for glycopeptides, and find good resolution for deamidated peptides. The UPLC columns show significant increases in electrospray ionization mass spectrometry (ESI-MS) sensitivity, with only small loss of retention and peak shape when formic acid is used in place of trifluoroacetic acid (TFA). We assert that the compelling benefits of UPLC for peptide mapping suggest that it will become the technique of choice in the near future.

MAKING PEPTIDE MAPS

Peptide mapping is a workhorse technique in biopharmaceutical characterization.¹ It is used to identify proteins based on the elution pattern of the peptide fragments, determine post-translational modifications, confirm genetic stability, and analyze protein sequence when interfaced with mass spectrometry.

To make a peptide map, it is necessary to separate every peptide into a single peak. Therefore, peptide mapping represents a significant chromatographic challenge, because of the complexity of peptide digests. In addition to the large number of peptides that are generated from the enzymatic digest of a protein, there can be a large number of alternative peptide structures, such as post-translational modifications.

There has been a long trend of reducing particle size in liquid chromatography. Modern reversed phase liquid chromatography (LC) began in the mid-1970s with the advent of irregular 10 μm particles, and within the last five years 2.5 μm particles have become available. The smaller particles have been used in short columns, which leads to fast analysis times but relatively modest gains in resolving power.

Column length has decreased with particle size because the system pressure required is inversely proportional to the cube of particle diameter. For example, reducing the particle size by a factor of two requires an increase in the operating pressure by a factor of eight. It is, therefore, necessary to use shorter columns at lower flow rates to remain within the capabilities of the instrument. Clearly, in order to take advantage of smaller particle sizes, both in terms of improved speed and improved resolution, we require instrumentation capable of high-pressure operation. In addition, system band broadening must be reduced to observe the narrow peaks generated with small particle packings.

In 2004, Schwartz and Murphy introduced the first LC system capable of operation up to 15,000 psi (1,000 bar).² The combination of a system capable of high-pressure operation and columns packed with sub-2 μm particles has been termed ultra performance liquid chromatography (UPLC) to differentiate it from high performance LC (HPLC).³

The benefits of UPLC vs. HPLC were originally demonstrated for small molecules (<500 Da) with reversed phase columns. Improvements in resolving power (1.7x), sensitivity (3x) and separation speed (9x) were demonstrated for many different applications.^3,4 All these benefits derive from the 1.7 μm particles used in UPLC columns. In addition to the high-pressure capabilities, the UPLC instrument also reduces system volume and detector-cell volume to preserve the high-efficiency separations. UPLC has gained rapid acceptance for small-molecule analytical separations, and we believe that it will displace HPLC for many applications. The capabilities of UPLC should make higher resolution peptide mapping possible.

Table 1. LC Conditions

EXPERIMENTAL SETUP

This article demonstrates the application of UPLC to peptide mapping in a series of experiments. Most of the instruments and chemicals were from Waters with three exceptions: acetonitrile (Optima Grade) was from Fisher, TFA was from Pierce, and formic acid was from EMD. Exact conditions are listed in Table 1.

The ACQUITY C18 chemistry is based on a bridged ethyl hybrid base particle, specifically designed for operation at higher pressure. It has an average pore diameter of 130 Å a pore volume of 0.7 mL/g, and a surface area of 185 m²/g. Its diameter is 1.7 μm.

PLATE HEIGHT AND VELOCITY

The chromatographic benefits of UPLC are largely derived from reduced band-broadening that is, in turn, a consequence of reduced diffusion distances in small particles. This process is quantitatively described in the van Deemter equation that relates height equivalent of a theoretical plate (H) to linear velocity. Figure 1 graphs this relationship for a peptide of 1,500 Da on 3.5 μm and on 1.7 μm packings. The minimum in the curve corresponds to the maximum efficiency, and greatest resolving power, for each particle size.

Figure 1. Van Deemter Plot for 1,500 Da peptide. The equation has the form, H = A + (B/u) + Cu, where H = Height equivalent of a theoretical plate (cm), u = average linear velocity (cm/s), and A, B, C are constants.

At linear velocities or flow rates above and below the optimum, resolving power declines. The smaller particles have higher resolving power at a higher linear velocity. In Figure 1, the 3.5 μm particles have a minimum plate height of 8.11 μm at a linear velocity of 0.17 mm/s. In contrast, a minimum plate height of 3.94 μm is observed at 0.33 mm/s with the 1.77 μm particles. In practical terms, these suggest that the small particles used in UPLC could increase the resolving power in a peptide mapping experiment, and should simultaneously reduce the separation time because the optimum is achieved at a higher linear velocity.

For the 3.5 μm particle, the optimum linear velocity corresponds to a flow rate of about 24 μL/min on a 2.1 mm i.d. column or about 5.5 μL/min on a 1 mm column. In practice, such flow rates would never be used for a peptide map because the separation times would be far too long. It is common practice to operate at higher flow rates, typically about 250 μL/min and 60 μL/min on 2.1 and 1 mm columns, respectively. A linear velocity of about 1.7 mm/s corresponds to a plate height of about 21 μm. This 2.6-fold loss of resolution with a 10-fold increase in separation speed becomes an accepted compromise in the analytical community. For 1.7 μm particles, however, the plate height at 250 μL/min only increases to 6.45 μm, a 1.6-fold sacrifice from the optimum.

This analysis suggests several ways to approach improving peptide maps using UPLC. First, columns with the smaller particles (1.7 μm) will improve both resolution and sensitivity by reducing diffusion-related band broadening. Second, reducing the plate height is consistent with obtaining the same or better resolution with shorter columns and higher flow rates. Third, the compromise between separation time and resolution is more favorable with the smaller particles.

PERFORMANCE TESTS

We investigated the influence of volumetric flowrate on peptide separation performance with 2.1 mm i.d. columns. A standard peptide mixture was separated on a UPLC column run at 100 μL/min and at 300 μL/min (Figure 2). Flowrate is the only variable because the gradient change per column volume is the same, ensuring that the chromatographic selectivity is constant. We ran a 75 min gradient at 100 μL/min and a 25 min gradient at 300 μL/min. Figure 2 shows a comparison of peak volumes, calculated by multiplying the flow rate by peak width at the base. Peak volumes at 100 μL/min average about half the volumes at higher flow.

Figure 2. Influence of flow rate in UPLC peptide mapping with a 2.1 mm column. The peaks are normalized versus the tallest peak, and detected by MS.

We expected peptide peak volumes, a measure of resolution and sensitivity, to be optimum at 100 μL/min. Most peptide maps done with 2.1 mm HPLC columns are run higher than 100 μL/min. The higher flow rates represent a compromise between resolution and run time. Other users of HPLC accept this compromise. Another justification are instrumental limits in reproducibly pumping liquid at flow rates less than 150 μL/min with accurate and precise gradients.

However, the ACQUITY UPLC instrument performs extremely well at a flow rate of 100 μL/min in gradient mode. Figure 3 demonstrates this performance by the overlay of six gradient runs of a peptide standard, with the average and standard deviations of retention time listed for each peak.

Figure 3. UPLC peptide mapping retention time reproducibility is good at 100μL/min. Mean and standard deviation are given for each normalized peak.

In some cases it would be desirable to reduce the run time of a peptide map. HPLC maps often require cycle times of 3 to 5 hours to separate all the peptides within the digest, especially for large proteins like antibodies. While faster peptide maps are desirable, it is critically important not to compromise resolution to ensure that the test results provide the same level of information.

The van Deemter equation predicts that plate height will be two to four fold less with 1.7 μm particles than with 3.5 μm particles. We expect to see the same resolving power with a shorter column.

A test demonstrates how UPLC can resolve the same number of peaks in a peptide map as HPLC but in less time. The separation of an enolase digest was done on a 50 mm long UPLC column with a 20 min gradient and on a 100 mm long HPLC column with a 40 min gradient; both with flow rates of 100 μL/minute. Figure 4 shows these chromatograms. The UPLC separation shows a comparable number of peaks and a similar overall elution pattern as the HPLC separation, but in half the time.

Figure 4. Using a shorter UPLC column to improve peptide-mapping throughput. Top plot is UPLC. Bottom plot is HPLC. The peaks are normalized versus the tallest peak.

UNDERSTANDING INTERACTIONS

Interaction experiments focus on the physical aspects of the separation as they relate to band broadening. Successful peptide mapping depends, of course, on the interaction among the peptides, the mobile phase, and the column surface chemistry. Figure 5 (top) shows the separation of a tryptic digest of enolase with a 3.5 mm C18 HPLC column whose particles have 300 Å pores. This is typical of the most common peptide separation columns. Figure 5 (bottom) is for a 1.7 μm UPLC column. Conditions are the same for both columns.

Figure 5. Suitability of UPLC for peptide mapping in digestion of enolase is shown. Top plot is HPLC. Bottom plot is UPLC. The peaks are normalized versus the tallest peak.

Note the greater number of peaks in the UPLC separation. The overall resolution and sensitivity are higher. In the UPLC map, there are several small peaks that are difficult to discern with the HPLC run. This demonstrates that UPLC offers higher resolution and sensitivity when compared with HPLC under the same gradient conditions. As is always observed when comparing two different column chemistries, the separations are not identical in every detail. The overall appearance of the chromatograms is, however, similar. This suggests that the selectivity of the UPLC column is suitable for peptide mapping.

The higher resolution and sensitivity with UPLC are particularly important when using the peptide map to detect modified peptides. Higher resolution ensures that modified peptides are resolved from the unmodified form, as well as from other peptides in the digest. Higher sensitivity means that modified peptides can be detected at lower levels. Figure 6 shows that UPLC is a good way to separate a deamidated peptide from its unmodified form. UPLC should be the technique of choice for detecting all the peptides in a sample.

Figure 6. Native and deamidated peptides can be separated. The upper graph is native peptide. The lower graph is of a sample intentionally degraded before analysis. The peaks are normalized versus the tallest peak.

Scientists frequently interface peptide mapping with Electroscopy Ionization (ESI-MS) to provide additional information about the eluting peptides, including molecular weight and sequence. MS can also identify modified peptides and glycosylation sites. Therefore, it is important that a peptide mapping technique work well under conditions that are favorable for ESI-MS.

Mobile phases make a difference. TFA is commonly used as an acidic additive for peptide maps with UV detection, but it can lead to suppression of ionization and reduced sensitivity in ESI-MS. Formic acid is more desirable for LC-MS, because it causes less ion suppression than TFA. However, many reversed phase columns used for peptide mapping show lower retention and broader peaks with formic acid than with TFA. Figure 7 compares the separation of several peptides with formic acid and with TFA on a UPLC column with MS detection. With formic acid the peak heights are about three times higher. At the same time, there is only a slight reduction in retention, corresponding to a difference of a few percent organic at the point of elution, and a slight increase in peak width. This result indicates that the UPLC columns are compatible with formic acid to produce good results for ESI-MS.

Figure 7. Plots show Influence of acidic additive in UPLC peptide mapping. Upper plot uses formic acid. Bottom plot uses TFA. The peaks are normalized versus the tallest peak. Initially column mobile phase is A and the carrier is B.

Glycosylation is an important post-translational modification that plays a critical role in determining the efficacy and safety of a therapeutic protein. Glycosylation can be analyzed on the intact protein by mass spectrometry, as released glycans by a variety of techniques, or as glycopeptides in LC-MS peptide maps. When glycosylation can be characterized with LC-MS of the glycopeptides, the site of attachment can be directly determined, and structural information can be obtained through MS-MS experiments. This approach is limited, however, by the poor chromatographic peak shape of glycopeptides and incomplete resolution of glycoforms with HPLC peptide mapping. The poor peak shape has been attributed to the large size of the glycopeptides and their heterogeneous structure.

Figure 8. UPLC of tryptic fragments of a-1 acid glycoprotein. Selected ion chromatogram at m/z 657 of sialyted glycopeptides shows many peaks. The peaks are normalized versus the tallest peak.

UPLC can do better. Figure 8 shows the UPLC-MS separation of a tryptic digest of α-1 acid glycoprotein. The MS detection was performed with a Q-Tof mass spectrometer, which is well suited for glycopeptides because of its extended mass range. Data are plotted as a selected-ion chromatogram for m/z 657, a signature ion for glycopeptides resulting from carbohydrate fragments. The glycopeptides are detected as sharp, symmetrical peaks. These characteristics are important for minimizing spectral overlap of different glycoforms of the same peptide. UPLC combined with ESI/TOF mass spectrometry will be a powerful tool for studying the glycosylation state of proteins.

CONCLUSIONS

UPLC provides better peptide maps than can be obtained with HPLC. Better resolution is available in combination with improved sensitivity. Multiple strategies are available for reducing run time without compromising resolution. Selectivity is comparable to that of common reversed phase HPLC peptide mapping columns and can be easily transferred to alternative modifiers that give better sensitivity in ESI-MS. The UPLC separations are proving highly suitable for separation of glycopeptides. With the extension of available columns to various alternative chemistries and even smaller particles, UPLC will represent the next generation tool for peptide mapping.

Jeff Mazzeo, Ph.D., is applied technology director at Waters Corp., 34 Maple Street, Milford, MA 01757, 508.482.3462, fax 508.482.4100, jeff_mazzeo@waters.com Tom Wheat, Ph.D., is life sciences application manager at Waters Corp., 508.482.8838, tom_wheat@waters.com Beth Gillece-Castro is principal scientist at Waters Corp., 508.482.3007, beth_gillece-castro@waters.com Ziling Lu is applications chemist at Waters Corp., 508.482.4663, ziling_lu@waters.com

REFERENCES

1. Hancock B, editor. New methods in peptide mapping for the characterization of proteins. Boca Raton: CRC Press; 1996.

2. Swartz M and Murphy B. Ultra performance liquid chromatography: Tomorrow's HPLC technology today. LabPlus International 2004 18(3):6-9.

3. Mazzeo J, Neue U, Kele M, and Plumb R, Advancing liquid chromatography performance with smaller particles and higher pressure. Analytical Chemistry 2005; 77(23):460A-467A.

4. Fischer T. Laboratory notes. Brandenburg Technical University at Cottbus. Available at URL: www.zal.tu-cottbus.de/zal/prakt/orgaanal.htm