Proving the proteoform hypothesis via spatial top-down proteomics of the human cell

Proteins exist in their biological context as molecularly defined "protein forms", or proteoforms, carrying specific sets of genetic and chemical modifications. The "proteoform hypothesis" posits that the proteome-to-phenotype correlation is better explained through the analysis of endogenous proteoforms via top-down proteomics rather than by the analysis of protein surrogates (i.e., proteolytic peptides) investigated through bottom-up proteomic approaches.

Here we present the results of the study of the intact 0-70 kDa proteome of HEK293 cells performed at the subcellular level. Subcellular fractions were obtained using the dynamic organellar mapping (DOM) protocol. Intact proteoforms extracted from each of the seven subcellular fractions (nucleus, mitochondria, ER, endosomes, Golgi apparatus, larger protein complexes, cytosol) were analyzed using a specific approach, depending on their molecular weight (MW). Proteoforms <30 kDa were fractionated using an in-house cast, high-resolution 15% T GELFrEE, with each MW fraction (~5 kDa mass bins) interrogated via liquid chromatography-field asymmetric ion mobility spectrometry-tandem mass spectrometry (LC-FAIMS-MS²) on an Orbitrap Eclipse Tribrid (Thermo Scientific). Larger proteoforms 30-70 kDa were fractionated by 10% T GELFrEE and each fraction was analyzed using the targeted proton transfer charge reduction data acquisition method (tPTCR) on an Orbitrap Ascend Tribrid (Thermo Scientific). In tPTCR experiments, 4 partially overlapping m/z windows (1.5 Th width) are surveyed per LC-MS2 run. Cations from each narrow isolation window are first subjected to PTCR, to obtain a new charge state envelop from which the average mass of the proteoform can be derived; and then, they are fragmented using higher-energy collisional dissociation for amino acid sequencing. Five LC-MS2 tPTCR measurements were run for each fraction, sampling a total of ~24 m/z units, to obtain information on "all" proteoforms included in a given GELFrEE fraction.

Preliminary results from four of the seven subcellular fractions show three to four-fold increase in the number of identified UniProt accessions (>3,000) and proteoforms (>28,000) compared to a similar top-down analysis performed on HEK293 whole cell lysate. Additionally, while each characterized DOM fraction contains a large portion of unique UniProt accessions, the variability at the proteoform level is substantially higher than at the protein family level. For instance, 36 kDa Y-box-binding protein 1 (P67809), known to be present both in the nucleus and in the cytoplasm, was identified as full-length proteoforms in the nuclear fraction and as C-terminally truncated proteoforms (20-29 kDa range) in the cytosolic fraction.

Taken together, these results demonstrate the validity of the proteoform hypothesis and represent the most comprehensive catalog of proteoform for a single cell line ever produced.