High-resolution ultrahigh-pressure long column reversed-phase liquid chromatography for top-down proteomics

Highlights

• Column length was found as an important factor for top-down proteomic RPLC separation.

• Long (≥1 m) columns can provide peak capacities of >400 for resolving proteoforms.

• Both porous and superficially porous particles were effective to separate proteins. Particles with 200–450 Å pores enabled chromatographing >100 kDa proteoforms.

• C1-C18-bonded phases had their own limits for eluting various sizes of proteoforms.

Abstract

Separation of proteoforms for global intact protein analysis (i.e. top-down proteomics) has lagged well behind what is achievable for peptides in traditional bottom-up proteomic approach and is becoming a true bottle neck for top-down proteomics. Herein, we report use of long (≥1 M) columns containing short alkyl (C1-C4) bonded phases to achieve high-resolution RPLC for separation of proteoforms. At a specific operation pressure limit (i.e., 96.5 MPa or 14 K psi used in this work), column length was found to be the most important factor for achieving maximal resolution separation of proteins when 1.5–5 μm particles were used as packings and long columns provided peak capacities greater than 400 for proteoforms derived from a global cell lysate with molecular weights below 50 kDa. Larger proteoforms (50–110 kDa) were chromatographed on long RPLC columns and detected by MS; however, they cannot be identified yet by tandem mass spectrometry. Our experimental data further demonstrated that long alkyl (e.g., C8 and C18) bonded particles provided high-resolution RPLC for <10 kDa proteoforms, not efficient for separation of global proteoforms. Reversed-phase particles with porous, nonporous, and superficially porous surfaces were systematically investigated for high-resolution RPLC. Pore size (200–400 Å) and the surface structure (porous and superficially porous) of particles was found to have minor influences on high-resolution RPLC of proteoforms. RPLC presented herein enabled confident identification of ∼900 proteoforms (1% FDR) for a low-microgram quantity of proteomic samples using a single RPLC–MS/MS analysis. The level of RPLC performance attained in this work is close to that typically realized in bottom-up proteomics, and broadly useful when applying e.g., the single-stage MS accurate mass tag approach, but less effective when combined with current tandem MS. Our initial data indicate that MS detection and fragmentation inefficiencies provided by current high-resolution mass spectrometers are key challenges for characterization of larger proteoforms.

Introduction

Proteomics is now advancing from measurements of protein expressions (e.g. using shotgun bottom-up approaches for analysis of enzyme digests [1]) to characterizing actual proteoforms more directly tied to specific biological processes, based on intact (or top-down) protein characterization approaches. The functional states of proteins include proteoforms arising from hydrolysis [2], signal peptide cleavages, proteolytic processing, site mutations, etc., and broad range of (combinatorial) post-translational modifications (PTMs), which largely go undetected in traditional bottom-up proteomic analyses. But proteoform characterization is technically challenging, and bottom-up approaches still dominate the field. For proteome degradomic activities, which typically involve smaller protein fragments (e.g., <15 kDa) [3], high-resolution RPLC–MS/MS approach has been successfully demonstrated [4], [5]. However, the quality of separations greatly degrades as the protein mass increases. Even though proteins with mass above 100 kDa have been successfully analyzed [6], top-down proteomics remain far inferior to bottom-up approaches in terms of coverage, sensitivity, and throughput. As a recent example, assignment of ∼3000 proteoforms used a complex four-dimensional separation platform resulting in low sensitivity (mg sample amounts required) and low throughput (long analysis times) [6]. These requirements, together with lack of robust computational tools for proteoform characterization, greatly limit the use and applicability of top-down proteomics.

Quality separation improves proteome coverage, enhances sensitivity and simplifies analysis. While no single suitable separation is yet available, the last stage of separation applied on-line with MS analysis, typically RPLC, is of critical importance for these aspects. A major obstacle for development of high-resolution RPLC of proteins arises from the traditional concept that the resolution of gradient RPLC of proteins is unrelated to column length [7], [8]. Gradient RPLC of proteins is considered as a single adsorption/desorption event near the top of the column, rather than protein partitioning along the column, which leads to a claim that protein separation efficiency is irrelevant to column length [7], [8] and the inability to predict RPLC protein resolution using classical chromatographic parameters such as peak capacity expression [9]. This is also the reason why majority of efforts for improving gradient RPLC resolution are focused on optimization of separation media using large pore particles [6], [10], nonporous particles [11], [12], core-shell particles [13], [14], and monoliths [15], [16] in conjunction with short (e.g., 2–20 cm) columns. It is thus not surprising that strongly hydrophobic reversed phases (e.g. C8, C8-similar polystyrene-divinylbenzene, and C18 phases) have been traditionally selected for protein separation and top-down proteomic applications [6], [10], [11], [12], [13], [14], [15], [16], [17], [18]. Such approaches provide moderate quality resolution for proteins, evidenced by absence of reports of complex chromatograms, regardless of the magnitude of separation power (e.g., peak capacity) estimated [6], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18].

Herein, we report a new approach for achieving high-resolution RPLC of proteins. In this approach, we utilized C1-C4 reversed phases to minimize the effect of hydrophobic adsorption on RPLC resolution observed for C8 and C18 for proteins [7], [8] and thus enhance the role of column length in separation resolution. This approach is hereby referred as weak interaction RPLC (WI-RPLC) in contrast to conventional strong interaction RPLC (SI-RPLC) employing C8 and C18 phases. With use of long columns containing such weak reversed phases, complex chromatograms displaying over 200 countable peaks and separation peak capacities of ∼400 have been achieved. We systematically investigated various reversed phase particles with size of 1.5–5 μm and porous, nonporous, and superficially porous surfaces for high-resolution RPLC, and demonstrated that WI-RPLC coupled with FT-MS/MS enabled identifying ∼900 proteoforms (at 1% FDR) from a single analysis of a few micrograms of a simple microbial lysate.