Tackling Top-Down Protein Analysis

Kevin Schug discusses the advantages of using mobile phases with different pH values in the first and second dimension when using comprehensive two‑dimensional liquid chromatography (2D‑LC) methods for top‑down protein analysis.

Q. You recently developed a comprehensive two‑dimensional liquid chromatography (2D-LC) approach for “top-down” protein analysis. Why is “top-down” protein analysis important?
 

A: Protein analysis, proteomics, and other related -omics approaches have been major drivers of technological advances in separation science. These fields of research have driven instrument manufacturers to build more sensitive and capable analytical instruments, particularly for the different “flavours”of mass spectrometers that exist. Omics research has also prompted other suppliers to develop new sample preparation and chromatography tools to improve separations of complex biological mixtures. However, for the analysis of intact proteins, there is still plenty of room for more development.
 

The vast majority of protein analysis has been performed in a “bottom‑up” fashion, where proteins are digested into constituent peptides before instrumental analysis. The so‑called shotgun proteomics approach, developed initially by Yates et al. (1,2), has been perhaps the most popular approach for protein discovery. Additionally, a wide variety of methods exist to conduct quantitative protein analysis in a “bottom-up” fashion (3).
 

In contrast, “top‑down” methods have grown substantially recently. This approach refers to introducing intact proteins for instrumental analysis, and using instrumental methods, such as different gas phase dissociation methods, to interrogate the protein (4). Top-down methods have the advantage of preserving fractions of different proteoforms, or proteins that might possess different combinations of post-translational modifications. Top-down methods have been used much more in the context of qualitative discovery experiments, rather than for quantitative analysis.

The large growth in protein biopharmaceuticals has spurred the need for more and better methods for quality control in the pharmaceutical industry. What is striking to me is how this has prompted analysts to adopt, utilize, and improve both bottom-up and top-down methods, including a wide array of chromatographic and mass spectrometric (MS) approaches (5). Even “middle-down” approaches are useful, where partial digestion of proteins can provide a middle ground compromise between accessing more complete aspects of intact protein proteoforms, while taking advantage of analytical methods that are more attuned to peptide or short protein analysis (6).

Ultimately, in our laboratory, we are interested in two primary aspects of the protein analysis research. First, we would like to be able to develop quantitative methods that target multiple intact proteins. We believe these could be useful, both in clinical analysis to determine biomarkers, as well as for pharmacokinetic-type analyses associated with biopharmaceuticals. We have published several papers already along these lines, utilizing triple quadrupole mass spectrometry for multiple reaction monitoring (MRM) of intact proteins (7–10). Second, we are interested in advancing more complex methods, including the use of multidimensional separations, which can either handle more complex mixtures or reduce the need for manual analyst intervention (for example, on-line sample preparation) (11). This work on comprehensive 2D-LC using a reversed phase × reversed phase approach (12) certainly pushes the boundaries on the latter, and ultimately, we are interested in making a quantitative version of this method. It is important to note that mass spectrometry‑compatible mobile phase systems are inherently desirable in all of these efforts.

Q. You developed a novel 2D-LC approach for “top-down” proteomics combining a high pH mobile phase in the first dimension followed by low pH mobile phase in the second dimension. What is novel about this approach and what benefits does it offer the analyst?

A: If you took an instrumental analysis or analytical chemistry class in college, then you probably covered chromatography in some depth. The problem, in the context of this discussion, is that your instructor probably never touched upon protein separations. Everything was based on small molecules. Maybe some aspects of protein separations were covered in your biochemistry class, but those were probably not focused around the use of modern analytical instrumentation.

That said, there is quite a bit of previous research on protein separations to draw upon if you want to start using your LC–MS instrument for protein analysis. However, one problem is that much of the older literature did not use MS detection. A quick overview of literature from the end of the 20th century will show that most of the published methods relied on a fairly high concentration (0.1% and greater) of trifluoroacetic acid (TFA) in the mobile phase. This is not a good mobile phase for MS detection; therefore, some additional effort was needed to make intact protein separations compatible with MS detection. We initially achieved this in a one-dimensional LC–MS method using a combination of formic acid, and a lower concentration of TFA (0.05%) (8). Some TFA is necessary to act as an ion-pairing reagent to aid separations, while the formic acid provides for good ionization efficiency.

Suffice it to say that protein separations are more complicated than small molecule separations. Slower diffusion and interaction kinetics lead to the use of wide pore stationary phase supports and potentially elevated temperatures. Proteins also exhibit a steep adsorption isotherm when introduced to chromatographic stationary phases (13). This means that small changes in mobile phase composition can drastically affect protein retention. For example, in reversed-phase separations, this behaviour means that a protein may be essentially irreversibly retained (a very high capacity factor) at 30% organic in the mobile phase, but unretained with 35% organic. This amounts to an “on/off” retention mechanism. In previous work, Regnier and coworkers have discussed a stoichiometric displacement model, which is consistent with on/off retention (14,15). The premise is that a protein in a given conformation will remain adsorbed to the stationary phase until a sufficient mobile phase strength exists to desorb the protein. At that point, the protein is eluted. One should also be careful not to ramp the mobile phase gradient too quickly, because if the protein is desorbed and begins to elute, but then experiences a high organic mobile phase on its way out of the column, it may then be retained by a hydrophilic interaction retention mode. Thus, protein separations usually involve modest or slow mobile phase gradients.

A practising chromatographer will realize that the best way to affect selectivity in chromatography is to change the stationary phase. This does not work as well in protein separations. Because the proteins sit down and adsorb to the stationary phase surface, and are then desorbed and elute when they experience a strong enough mobile phase, selectivity is more significantly controlled by protein conformation (14,15). Changing the protein conformation changes the adsorption isotherm and will change the point at which the mobile phase can activate the on/off retention switch.

This thinking guided our choice of mobile phase conditions for developing reversed phase-based protein separation methods, which exhibited different selectivities. We already knew that comprehensive LC×LC in a reversed phase × reversed phase format could be very effective for peptide separations. Donato et al. showed that using a combination of a high pH mobile phase in one dimension with a low pH in the second dimension could achieve the highest degree of orthogonality in an LC×LC method (16). This means that such a configuration provides the best way to spread peptides across the two-dimensional separation space. Peptides that coelute in the first dimension separation can be well separated in the second dimension. We felt that this was a good approach to altering reversed‑phase selectivity for intact proteins. We investigated this first in one-dimensional separations (17), and then we extended this to LC×LC methodology (12).

Further considerations guiding this line of research included the fact that we could use a, perhaps, less MS-compatible mobile phase in the first dimension (high pH) and then a more MS-compatible mobile phase in the second dimension (low pH). It is worth noting that the high pH mobile phase, which comprises the addition of triethylammonium carbonate (modified to pH 10), is indeed MS‑compatible. However, it does not provide such sensitive detection as the low pH mobile phase, incorporating formic acid and TFA, to generate the typical multiply-charged (multiply-protonated) protein ion envelopes typically observed during electrospray ionization. Another reason for choosing a reversed phase × reversed phase approach is because there are more column chemistries and formats available, compared to other potential separation modes that might be considered.

Q. Could other characteristics of mobile phases be altered in the first and second dimension?

A: Perhaps they could, but they might not be as effective. Protein separations require the addition of some sort of ion-pairing reagent. TFA provides that role in the low pH separations that we have reported, as well as in a lot of studies reported in the past. The modifier provides multiple effects. The acidic nature suppresses ionization of residual unbonded silanols on the stationary phase support surface; this reduces their ability to generate secondary interactions with analytes as they traverse the column; secondary interactions are the primary culprit for poor peak shapes. Additionally, the trifluoroacetate can ion pair with basic residues on the protein surface. This further masks potential interactions between basic residues and silanols, and it enhances the hydrophobicity of the protein, which ought to support reversed-phase retention on alkyl-bonded stationary phases. If one only uses formic acid in the mobile phase, then there can be carryover issues and poor peak shapes. The addition of a little bit of TFA quite drastically ameliorates this issue, but as mentioned above, care should be given to keeping the concentration low enough to avoid ion suppression during the transfer of separated species into the mass spectrometer.

One very promising alternative modifier to use instead of a TFA and formic acid mixture is difluoroacetic acid (DFA). While little has been shown in the literature to date (18), we recently evaluated the use of DFA as a suitable modifier for LC–MS of proteins. Suffice it to say that DFA appears to provide the best of both worlds. It has sufficient solution-based ion-pairing capabilities to enable quality protein separations, and it also exhibits much less deleterious effects for protein ionization. We found as much as an order of magnitude increase in sensitivity with the use of DFA. The article reporting that work is currently going through the review process for publication.

With the desire to switch the mobile phase to a basic pH to alter the selectivity associated with protein separation, a suitable ion-pair reagent was needed. In the end, triethylammonium bicarbonate (adjusted to pH 10) was found to provide the best results. Here, one would expect the high pH conditions to induce more negative charge on the proteins; the triethylammonium cation can then effectively ion pair with these negatively charged sites to improve peak shape and reduce secondary interactions and carryover during the chromatographic separations. Of course, the use of pH 10 rules out the potential to use standard silica gel columns because these are not stable at high pH. This drastically reduces the number of column choices. In the end, it was possible to use a hybrid silica particle phase, which exhibits enhanced pH stability. Indeed, a very significant change in selectivity could be achieved using this basic mobile phase relative to the standard acidic mobile phase. This choice enabled us to also combine the basic (1st dimension) and the acidic (2nd dimension) mobile phases to create the reported reversed phase × reversed phase method. Solvent mismatch issues were partially solved through the use of partial loop filling during modulation.

Indeed, it might be possible to find some alternative modifiers for creating suitable basic pH mobile phases. Very little research has been performed to investigate this further.

Q. How does this technique benefit the analyst?

A: Unfortunately, for the routine analyst, this method may not yet be so useful. That is because so many challenges still exist for the widespread implementation of LC×LC methods in general, much less for separating intact proteins. Overall, incorporating the second dimension into LC separations increases peak capacity, or the number of compounds, which can be separated in one analytical run. The overall peak capacity of an LC×LC method is equal to the peak capacity of the first dimension multiplied by the peak capacity of the second dimension. Therefore, it is not unreasonable to gain an order of magnitude in peak capacity by adding a second dimension to the separation.

Development of such methods helps push the limits of what is currently available to researchers in a multitude of fields, particularly in omics research. Once various other hurdles to adoption and use are overcome (most notably the limited advances of LC×LC software, especially for quantitative applications), I believe we will see wider use of multidimensional LC separations. After all, there are scientists working on solving these problems (19). Currently, I think there is much more development and use of these techniques in Europe, compared to the United States. For this reason, in 2016, I took a sabbatical at the University of Messina (Italy) with Luigi Mondello and Paola Duga to learn comprehensive multidimensional techniques. I sense that the disparity in adoption of these techniques between Europe and the US stems from the greater emphasis on and monetary support for fundamental separation science in Europe. Such support in the United States is very focused on applications and short-term gains, and as such, we are lagging behind other parts of the world in this important area of development.

Q. What are the current challenges associated with quantitative top‑down protein analysis?

A: This is an area of research where there are no shortage of challenges. Compared to small molecules, there is much less fundamental understanding about the extent and nature of matrix effects that might exist during LC–MS of intact proteins. The majority of top-down work in the past has been discovery-based and has not focused on quantitative aspects. There are also fewer choices in terms of chromatographic columns and formats available for intact protein separations. This can limit flexibility and, ultimately, may limit the ability to solve some challenging separation problems. Non-immuno-based sample preparation technologies for intact proteins are virtually non-existent. I am still amazed that a commercial microplate‑based wide-pore solid-phase extraction format does not exist. Sample preparation often requires micro-formats as a result of the limited availability of samples and the often low amounts and high costs of proteins. What about internal standards? There are not many choices, and it could cost tens of thousands of dollars to create stable isotopically-labelled variants of a protein. The availability of intact protein standards is rapidly increasing, but there is a high probability that these standards could contain different proteoforms than the same protein targeted from an individual.

I do not think we will be considering LC–MS assays for the quantitative determination of intact proteins for routine analysis soon. That said, there can be some significant advantages to pursuing such a route. I will not expand on this now. Suffice it to say, this is a good area for research and development-to try to start to address and solve some of the existing challenges, and to add to the existing literature base with more information for others to drive further. After all, that is how science works and how things progress.

References

1. J.K. Eng, A.L. McCormack, and J.R. Yates, J. Amer. Soc. Mass Spectrom.5, 976–989 (1994).
2. D.A. Wolters, M.P. Washburn, and J.R. Yates III, Anal. Chem. 73, 5683–5690 (2001).
3. V. Vidova and Z. Spacil, Anal. Chim. Acta 964, 7–23 (2017).
4. N.L. Kelleher, Anal. Chem.76, 196A–203A (2004).
5. K. Sandra, I. Vandenheede, and P. Sandra, J. Chromatogr. A1335, 81–103 (2014).
6. F. Lermyte, Y.O. Tsybin, P.B. O’Connor, and J.A. Loo, J. Amer. Soc. Mass Spectrom. 30, 1149–1157 (2019).
7. E.H. Wang, P.C. Combe, and K.A. Schug, J. Amer. Soc. Mass Spectrom.27, 886–896 (2016).
8. E.H. Wang, Y. Nagarajan, F. Carroll, and K.A. Schug, J. Sep. Sci.39, 3716–3727 (2017).
9. E.H. Wang, D.K. Appulage, E.A. McAllister, and K.A. Schug, J. Amer. Soc. Mass Spectrom. 28, 1977–1986 (2017).
10. D.D. Khanal, Y.Z. Baghdady, B.J. Figard, and K.A. Schug, Rapid Commun. Mass Spectrom.33, 821–830 (2019).
11. D.K. Appulage, E.H. Wang, B.J. Figard, and K.A. Schug, J. Sep. Sci.41, 2702–2709 (2018).
12. Y.Z. Baghdady and K.A. Schug, Anal. Chem.91, 11085–11091 (2019).
13. M.W. Dong, J.R. Gant, and B.R. Larsen, BioChromatography4, 19–34 (1989)
14. F.E. Regnier, Science238, 319–323 (1987).
15. G. Xindu and F.E. Regnier, J. Chromatogr. 296, 15–30 (1984).
16. P. Donato, F. Cacciola, L. Mondello, and P. Dugo, J. Chromatogr. A1218, 8777–8790 (2011).
17. Y. Baghdady and K.A. Schug, J. Chromatogr. A1599, 108–114 (2019).
18. J.M. Nguyen, J. Smith, S. Rzewuski, C. Legido-Quigley, and M.A. Lauber, mAbs11, 1358–1366 (2019).
19. B.W.J. Pirok and P.J. Schoenmakers, LCGC Europe31(5), 242–249 (2018).

Kevin A. Schug  is a Full Professor and Shimadzu Distinguished Professor of Analytical Chemistry in the Department of Chemistry and Biochemistry at The University of Texas (UT) at Arlington. He joined the faculty at UT Arlington in 2005 after completing a Ph.D. in chemistry at Virginia Tech under the direction of Harold M. McNair and a postdoctoral fellowship at the University of Vienna under Wolfgang Lindner. Schug was named the LCGC Emerging Leader in Chromatography in 2009. He is a fellow of both the U.T. Arlington and U.T. System-Wide Academies of Distinguished Teachers.