Native Mass Spectrometry Imaging of Proteins and Protein Complexes by Nano-DESI

Abstract

Previously, we have demonstrated native mass spectrometry imaging (native MSI) in which the spatial distribution of proteins maintained in their native-like, folded conformations was determined using liquid extraction surface analysis (LESA). While providing an excellent testbed for proof of principle, the spatial resolution of LESA is currently limited for imaging primarily by the physical size of the sampling pipette tip. Here, we report the adoption of nanospray-desorption electrospray ionization (nano-DESI) for native MSI, delivering substantial improvements in resolution versus native LESA MSI. In addition, native nano-DESI may be used for location-targeted top-down proteomics analysis directly from tissue. Proteins, including a homodimeric complex not previously detected by native MSI, were identified through a combination of collisional activation, high-resolution MS and proton transfer charge reduction.

Figure 1

(a) Photograph of the rat kidney during sectioning with labeled regions corresponding to the following mass spectra. (b–f) Native mass spectra from distinct regions of the kidney. Some proteins exhibited strong signals in one region, e.g., H-FABP in the medulla (b), regucalcin in the cortex (c), and adipocyte lipid-binding protein in the renal pelvis (f). Spectra are the result of averaging multiple orbitrap scans ((b): 48, ∼0.096 mm² of tissue sampled, (c): 148, ∼0.3 mm², (d): 18, ∼0.036 mm², (e): 22, ∼0.044 mm², (f): 188, ∼0.376 mm²) from raster line scans acquired at 20 μm/s in the respective region of tissue to improve the signal-to-noise ratio.

Figure 2

Native nano-DESI mass spectrometry imaging: (a) photograph of the kidney during the production of the section for analysis. (b) Optical image of the kidney section after native nano-DESI analysis. Ion images show the distribution of intact proteins throughout the tissue section; (c) MUP (outer cortex), (d) K-FABP (outer cortex), (e) RidA trimer (cortex), (f) regucalcin (inner cortex), (g) H-FABP (medulla), (h) S100-A6 dimer (medulla, renal pelvis) (i) ALBP (renal pelvis adipose tissue), (j) PEBP1 (bulk tissue), (k) α^H (vasculature), and (l) αβ^2H (vasculature). Images showing additional charge states are found in Figures S13–S18, S20, S22, S24, and S25.

Figure 3

(a) Pattern of four blood vessels was observable on the kidney. The blood vessels (∼200 μm across) are resolved in ion images for m/z 3163.3 (holo-α globin) with pixel dimensions of (b) 200 μm × 200 μm. (c) Optical image of the tissue section post-analysis; blood vessels are highlighted with the red circles. (d, e) Extracted ion chronograms for m/z 3040.3 (apo-α-globin) allow approximation of spatial resolution. Tailing of the signal suggests the resolving power is closer to 200 μm.

Figure 4

Identification of S100-A6 protein dimer. (a) The precursor ion signal showed overlap of two charge states. (b) PTCR-enabled determination of an 8⁺ charge state obscured by the 4⁺ signal. (c) HCD MS² (NCE = 42%) of m/z 2487.3 ± 2.5 (average of 148 scans from a raster line scan through the renal pelvis) dissociated the 8⁺ dimer to 5⁺ (d) and 4⁺ and 3⁺ (e) subunits, and produced sequence ions noted in (f).

Figure 5

PTCR-enabled charge state determination for low-abundance proteins, which cannot be isotopically resolved by high-resolution MS. (a) m/z 3315.94¹¹⁺ (36.5 kDa), the 5⁺ charge state of S100-A6 is also detected due to overlap of the (unknown) 11⁺ precursor with the (S100-A6 dimer) 6⁺ precursor. (b) m/z 3247.07¹¹⁺ (35.7 kDa), (c) m/z 3268.25^9+/11+ (29.4/35.9 kDa) indicated the presence of two overlapping signals within the isolation window. (d) m/z 3632.97¹⁰⁺ (36.3 kDa).