MetaUniDec: High-Throughput Deconvolution of Native Mass Spectra

Abstract

The expansion of native mass spectrometry (MS) methods for both academic and industrial applications has created a substantial need for analysis of large native MS datasets. Existing software tools are poorly suited for high-throughput deconvolution of native electrospray mass spectra from intact proteins and protein complexes. The UniDec Bayesian deconvolution algorithm is uniquely well suited for high-throughput analysis due to its speed and robustness but was previously tailored towards individual spectra. Here, we optimized UniDec for deconvolution, analysis, and visualization of large data sets. This new module, MetaUniDec, centers around a hierarchical data format 5 (HDF5) format for storing datasets that significantly improves speed, portability, and file size. It also includes code optimizations to improve speed and a new graphical user interface for visualization, interaction, and analysis of data. To demonstrate the utility of MetaUniDec, we applied the software to analyze automated collision voltage ramps with a small bacterial heme protein and large lipoprotein nanodiscs. Upon increasing collisional activation, bacterial heme-nitric oxide/oxygen binding (H-NOX) protein shows a discrete loss of bound heme, and nanodiscs show a continuous loss of lipids and charge. By using MetaUniDec to track changes in peak area or mass as a function of collision voltage, we explore the energetic profile of collisional activation in an ultra-high mass range Orbitrap mass spectrometer. Graphical abstract ᅟ.

Keywords: Collision-induced dissociation; Deconvolution; Heme proteins; Nanodiscs; Native mass spectrometry.

Figure 1

Schematic of MetaUniDec HDF5 file format organization (left) with example data from H-NOX (right). Examples are shown from the 50 V spectrum (index 10) and global analysis.

Figure 2

Schematic of MetaUniDec software. Raw data are parsed by the Python GUI and stored in the HDF5 with the configuration parameters. The UniDec.exe opens the HDF5 for data processing, deconvolution, and analysis. Results are written back into the HDF5 file, which is read by the GUI for visualization. Text colors on the HDF5 file match the outlined regions of the GUI they are plotted in or taken from.

Figure 3

MetaUniDec analysis of H-NOX. (A) A single data set with collision voltage ramping in 5V increments from 0 to 95V is deconvolved into (B) the corresponding mass distributions. Integration of the holo (green) and apo (black) forms is plotted in (C). Results from multiple data sets (D) are extracted and plotted in (E). Sigmoidal fits to the extracted peak areas reveal the VC₅₀ (F) and baseline dissociation (G). Vertical lines in E represent the midpoints of the sigmoidal fit.

Figure 4

MetaUniDec analysis of nanodisc data sets. A representative data set (A) is deconvolved to get the mass distribution (B), which is distilled into the average mass (C). Deconvolution of multiple data sets (D) provides average mass as a function of collision voltage (E). A linear fit (dashed lines) to the 100-200 V region shows no significant changes in the slope of the line (F) but shows a shift in the intercept of the linear fit with in-source trapping activation (G), which shows that a higher voltage is required in the IST region to achieve the same level of HCD activation.