Post‐translational modification and phenotype

Post-translational modification (PTM) greatly expands the molecular diversity of a protein. The dynamic decoration of PTMs during the protein lifespan, normally in an amino acid site-specific manner, can induce protein structural and functional alterations, allowing cells to transduce signals and respond to perturbations. On the other hand, the abundance and activity alterations of certain proteins as a "proteotype" might render a human individual a specific "phenotype" in response to environmental or genetic perturbations [1, 2]. Therefore, the PTMs and phenotypes respectively generate molecular and phenotypic diversities for individual proteins and organisms. Proteomics has been a central tool in measuring both PTMs and phenotypes. Current mass spectrometry (MS) techniques enabled us to confidently and reproducibly measure peptidoforms and proteoforms [3] for a better understanding of both molecular and phenotypic diversities. For example, the search for "Phosphoproteomics" as a keyword in PubMed yields 596 hits in the year of 2022, but merely 54 hits in the year of 2005, demonstrating the dramatic increase in popularity and importance of proteomics in analyzing protein phosphorylation. Along with the numerous studies underway, many fundamental questions, both technical and biological, are still open in the field. For example, how far are we from the completed identification of most or all PTM sites in humans? How can we improve the measurement of PTMs as well as their spatial and temporal dynamics? How can we establish a structural or functional insight for each PTM type and site? Has the MS measurement on PTMs matured to the level that is ripe for clinical exploitation? While there is no doubt that all these questions would require continuous, community-wide effort for a long term, in this Special Issue, we bring together a series of original research and review articles (see below), with the hope to foster the discussion around the topics of "PTM and phenotype" and to appreciate both progress and challenges in the field. The first group of articles feature three Review articles. In the first Review article, Xiao et al overviewed a plethora of computational methods developed for phosphoproteomic data analysis [4]. These methods include those for basic data processing (such as data filtering, imputation, and normalization), functional analysis (such as kinase-substrate prediction, kinase activity inference, and signaling network reconstruction), phosphoproteome annotation, and multi-omics integration. Given the significant data amount and quality of modern phosphoproteomics, this review provides a roadmap for analyzing and interpreting phosphoproteomic data. The author further highlighted the importance of benchmarking relevant tools and resources. In the second Review article, Wang et al. summarized the available enrichment methods for analyzing protein terminome [5], anothervital molecular diversity associated with protein function and structure. Compared to phosphopeptide enrichment, the efficient separation of protein N- and C- terminal peptides from a complex mixture is technically more challenging. This Review comprehensively discussed the pros and cons of both positive and negative selection methods for enriching N- and C- terminal peptides respectively, as well as the rationale and application examples of these methods. The future improvement space of these methods was also discussed. In the third Review article, Duan et al. focused on a newly emerging field studying PTMs in microbiome [6]. After an introduction about unknown functional roles of microbial PTMs, they reviewed studies on phosphorylation, acetylation, succinylation, glycosylation, and other PTMs in prokaryotes, which might have distinctive biological implications compared to the same PTMs in humans or mice due to the unique biological context in bacterial organisms. The authors then discussed concrete findings on microbe-microbe and microbe-host interactions and how bioinformatic methods would impact relevant studies. Finally, they highlighted a few studies on microbiome-wide PTMomics. Together, the topics in the three Reviews above cover some of the most challenging and exciting research directions in the field of PTM proteomics. The second group of articles describe new techniques facilitating PTM measurement. In the first article, Phetsanthad et al. presented a new framework improving the detection of glycosylated neuropeptides in crustaceans, a model organism with a well-characterized neuroendocrine system [7]. They firstly modified the enrichment using hydrophilic interaction liquid chromatography (HILIC). Then they evaluated different MS fragmentation methods and found that product ions-triggered electron transfer/ higher-energy collision dissociation (pd-EThcD) outperformed stepped collision energy higher-energy collisional dissociation (sce-HCD) for characterizing both N- and O-linked glyconeuropeptides. With the establishment of glyconeuropeptide identities in neuronal tissues, this research may inform future MS-studies on signaling molecule glycosylation. In the second article, Li et al. developed a chemical proteomic approach for enriching lysine monomethylation (Kme1) [8]. They suggested that, by optimizing the ratio of glutaraldehyde to NaCNBH3, while primary amine groups of the peptides were converted to piperidine derivatives, the Kme1 was converted to 5-aldehyde-pentanyl derivative, which could facilitate the capture by hydrazide beads. As reported, their effort is totally antibody-free and presents an approach towards the enrichment of a challenging PTM, Kme1, which normally has a negligible effect on peptide charge and hydrophobicity. In the third article, Wang et al described a large-scale top-down proteomics (TDP) study of Arabidopsis thaliana leaf and chloroplast proteins [9]. Whereas most articles in this Issue are based on bottom-up proteomics, TDP is uniquely capable of measuring PTM-derived proteoforms directly. Despite being analytically less sensitive currently, PTM analysis using TDP is important, because, eventually, it is the unique chemical structure of the PTM-modified proteoform that exerts biological function. The authors demonstrated that electrophoretic mobility prediction of capillary zone electrophoresis coupled TDP not only achieved the identification of 4,700 unique proteoforms (including various types of PTMs), but also enabled the discovery of cleavage sites of sub-cellular localization signals and PTM-associated electrophoretic migration patterns. These three articles of the second group together represent examples of pioneering technical efforts in measuring new and complex PTMs in the large-scale. The third group of articles present bioinformatic advances in measuring PTMs and their dynamic functions. In the first article, Poudel et al developed JUMPptm, a computational pipeline identifying PTMs from unenriched whole proteome [10]. JUMPptm essentially integrates advantages of JUMP, MSFragger, and Comet search engines, and offers a pan-PTM search output. In an ultradeep (but unenriched) Alzheimer's disease (AD) dataset, JUMPptm impressively detected about 35,000 peptides carrying various types of PTMs, including deamidation (∼44%), phosphorylation (∼19%), and many others. The author further studied the dynamics of the PTM peptides during AD progression and, interestingly, found that Tau was the protein that contained the largest number of differentially expressed PTM peptides. Among all PTMs, lysine acylation is involved in many diseases such as cancers and diabetes but understudied in the brain. In the second article, Bons et al described an in-depth analysis of the Sirtuin 5-regulated mouse brain malonylome and succinylome [11]. They utilized a library-free, direct data-independent acquisition mass spectrometry (DIA-MS) analysis that integrates PTM identification, quantification, and site localization, which was able to detect 466 malonylated and 2211 succinylated sites. A Spectronaut-to-Skyline transferability of spectral library was adopted to validate the reported sites. The directDIA analysis was deemed powerful for the limited sample materials after the antibody-based PTM enrichment. In the final article of this group, Li et al. proposed a hypothesis-free computational strategy for analyzing the newly emerged peptide-protein turnover datasets [12]. This type of data is generated by the dynamic stable isotope labeling by amino acids in cell culture (i.e., dynamic SILAC) experiment integrated with PTM proteomics (such as phosphoproteomics) for the purpose of understanding the turnover diversity due to protein PTM. Instead of inferring protein lifetime, herein, the proposed strategy aims to identify differential turnover profiles using a time-series algorithm without any prior assumption. In a dynamic system of gradual cell starvation, temporal variabilities, in addition of the constant stabilizing or destabilizing effects, were discovered to fully delineate how phosphorylation impacts or is associated with SILAC turnover dynamics. The fourth group of articles are translational studies focusing on profiling PTMs or key PTM-impacted proteomes for cancer research and biomarker discovery. In the first article, Meng et al presented a Dataset Brief for clinically associated phosphosites in hepatocellular carcinoma (HCC) [13]. By applying the well-established statistical analysis, the authors mined the phosphoproteomic dataset in the previously published proteogeomic study on HCC to catalog those clinically important phosphosites strongly associated with overall survival, recurrence, and tumor stages in HCC. Both altered pathways and individual phosphosites are described as a rich resource for the liver cancer community. In the second article, Qi et al employed the parallel-reaction monitoring (PRM) assay for the measurement of reader, writer, and eraser (RWE) proteins which function in RNA modification process [14]. In particular, the abundances of 113 RWE proteins were measured and compared between metastatic and primary colorectal cancer (CRC) cell lines. Additional mapping of RWE proteins to CPTAC results unveiled particular protein markers for CRC initiation and metastasis. In the third article, Duda et al investigated the global proteome dysregulation induced by histone deacetylase inhibitors (HDACi) in high-grade serous ovarian cancer (HGSOC) cell lines [15]. HDAC removes acetyl group from lysine residues on histone and relevant proteins. And HDACi have been shown promising in sensitizing cells to chemotherapy. Interestingly, even all three HGSOC cell lines tested were BRCA-1/2 wildtype cells, they authors observed a heterogenous response to HDACi across the cell lines. Additionally, the different inhibitors seemed to function differently with respective off-target effects. Future investigations are therefore warranted. In the fourth article, using DIA-MS Zhong et al investigated both global proteome and phosphoproteome alterations in a lung cancer cell line following the treatment by tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) [16], a promising ligand in cancer therapeutics. A total of 1684 differential phosphosites together with 25 phosphorylation motifs were identified, indicating a role of Akt1 in TRAIL signaling. While the four articles above interrogated the proteomic and phosphoproteomic regulations in a defined clinical phenotype or a clarified drug response, the fifth article in this category by Wiskott et al [17], very interestingly, was set to identify blood biomarkers for the acute intracranial hemorrhage in infant victims of abusive head trauma (AHT), a status that is difficult to be effectively diagnosed in current hospital practice. The authors measured the matched postmortem and antemortem blood specimens of AHT cases, together with controls of sudden infant death syndrome using DIA-MS. Despite small number of samples, they discovered both global and a few of brain trauma-specific protein changes in AHT, such as the upregulation of Brain acid soluble protein 1. In summary, the 14 articles included in this Special Issue coherently addressed several aspects in such an exciting and challenging field of PTM and Phenotype, through the presentations of visionary directions, new experimental methods, new bioinformatic and computational solutions, and translational application examples. We hope this Special Issue offers useful resources and ideas and encourages the community to continue the great work on these topics from different angles. Finally, we would like to express our deepest gratitude to all of the authors and contributors of this Special Issue. The authors declare no conflict of interest.