y-mtPTM: Yeast mitochondrial posttranslational modification database

Corresponding authors: Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava, Mlynská dolina, Bratislava 842 48, Slovakia. Email: bronislava.brejova@uniba.sk; *Corresponding author: Department of Genetics, Faculty of Natural Sciences, Comenius University in Bratislava, Ilkovičova 6, Bratislava 842 15, Slovakia. Email: lubomir.tomaska@uniba.sk

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Department of Applied Informatics, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava

Corresponding authors: Department of Computer Science, Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava, Mlynská dolina, Bratislava 842 48, Slovakia. Email: bronislava.brejova@uniba.sk; *Corresponding author: Department of Genetics, Faculty of Natural Sciences, Comenius University in Bratislava, Ilkovičova 6, Bratislava 842 15, Slovakia. Email: lubomir.tomaska@uniba.sk

Conflicts of interest The author(s) declare no conflict of interest.

Cite

Bronislava Brejová, Veronika Vozáriková, Ivan Agarský, Hana Derková, Matej Fedor, Dominika Harmanová, Lukáš Kiss, Andrej Korman, Martin Pašen, Filip Brázdovič, Tomáš Vinař, Jozef Nosek, Ľubomír Tomáška, y-mtPTM: Yeast mitochondrial posttranslational modification database, Genetics, Volume 224, Issue 3, July 2023, iyad087, https://doi.org/10.1093/genetics/iyad087

Abstract

One powerful strategy of how to increase the complexity of cellular proteomes is through posttranslational modifications (PTMs) of proteins. Currently, there are ∼400 types of PTMs, the different combinations of which yield a large variety of protein isoforms with distinct biochemical properties. Although mitochondrial proteins undergoing PTMs were identified nearly 6 decades ago, studies on the roles and extent of PTMs on mitochondrial functions lagged behind the other cellular compartments. The application of mass spectrometry for the characterization of the mitochondrial proteome as well as for the detection of various PTMs resulted in the identification of thousands of amino acid positions that can be modified by different chemical groups. However, the data on mitochondrial PTMs are scattered in several data sets, and the available databases do not contain a complete list of modified residues. To integrate information on PTMs of the mitochondrial proteome of the yeast Saccharomyces cerevisiae, we built the yeast mitochondrial posttranslational modification (y-mtPTM) database (http://compbio.fmph.uniba.sk/y-mtptm/). It lists nearly 20,000 positions on mitochondrial proteins affected by ∼20 various PTMs, with phosphorylated, succinylated, acetylated, and ubiquitylated sites being the most abundant. A simple search of a protein of interest reveals the modified amino acid residues, their position within the primary sequence as well as on its 3D structure, and links to the source reference(s). The database will serve yeast mitochondrial researchers as a comprehensive platform to investigate the functional significance of the PTMs of mitochondrial proteins.

Introduction

One of the unexpected findings resulting from the determination of complete sequences of eukaryotic genomes was that compared to less complex prokaryotic organisms, the number of open reading frames is not dramatically increased. This raised a question of how proteomic complexity is achieved using a limited inventory of protein coding sequences. To produce various proteins from a single coding region, cells employ a variety of means including alternative starts of transcription or translation, alternative splicing, RNA editing, or noncanonical types of translation. However, even a population of proteins with the same amino acid sequence can exhibit distinct biochemical properties due to a particular combination of posttranslational modifications (PTMs).

Most likely, the best studied PTM is phosphorylation, a covalent modification of proteins by a phosphate group, discovered at the beginning of the 20th century ( Levene and Alsberg 1906). The importance of protein phosphorylation is underlined by the fact that 2–5% of protein coding genes are dedicated to protein kinases (PKs), enzymes catalyzing the transfer of gamma phosphate from ATP to a recipient amino acid. Phosphorylation is only 1 of ∼400 types of currently known PTMs that differ in the nature of a chemical group attached to a target protein ( Khoury et al. 2011). As a result, every polypeptide of a given amino acid sequence can, in principle, carry a distinct set of PTMs and thus possess unique properties. In addition, reversible nature of most PTMs makes them a perfect means for cellular signaling pathways employed for interpretation of changes in both intracellular and extracellular environments.

Signaling pathways in eukaryotic cells constitute an intricate network integrating a myriad of inputs. One of the central hubs of this network is formed by mitochondria, recently referred to as the mitochondrial (mt) information processing system, which participates in sensing and responding to signals, integration of information, and production of specific outputs, thus tuning the overall cellular behavior ( Picard and Shirihai 2022). PTMs play a central role in mt signal transduction. In fact, the first reported PK activity was isolated from rat liver mitochondria ( Burnett and Kennedy 1954). As early as the late 1960s, it was shown that the activity of pyruvate dehydrogenase (PDH) is inhibited by PDH-specific kinase in response to cellular energy demands ( Linn et al. 1969). Mitochondria are also primary sites for the production of various acyl-coenzymes A (acyl-CoA) that are donors of the corresponding acyl group, resulting in the modification of lysine residues on target proteins ( Ringel et al. 2018). However, studies on the roles of PTMs in the regulation of mt functions lag behind those involving other cellular compartments.

Advances in mass spectrometry technologies combined with methods yielding highly purified mitochondria have resulted in a rapid accumulation of data on PTMs of mt proteins (for a recent review on omic-based approaches for investigation of mitochondria, see Schäfer et al. 2023). Thousands of phosphorylated, acylated, or otherwise modified sites have been identified, but the functional importance of the majority of these PTMs is underexplored.

Saccharomyces cerevisiae is the traditional model organism for investigating the mechanisms of mt inheritance, biogenesis, metabolism, and signal transduction ( Schatz 2013; Friedman and Nunnari 2014). One of its many advantages is a wide repertoire of tools enabling functional studies aimed at understanding the physiological significance of given variables including PTMs. For example, several systematic studies have investigated the roles of protein phosphorylation in the regulation of mitochondrion-associated processes such as protein import, energy metabolism, iron–sulfur cluster biogenesis, retrograde signaling, mitochondrial DNA (mtDNA) transactions, or protein degradation (reviewed in Tomaska 2000; Opalińska and Meisinger 2014; Frankovsky, Vozáriková, et al. 2021). Even though the total number of phosphorylated sites (P-sites) on yeast mt proteins exceeds 13,500 (see Results and Discussion), only a few hundreds have been assigned a physiological role ( Frankovsky, Vozáriková, et al. 2021). In the case of acetylation and succinylation, the additional relatively abundant PTMs in yeast mitochondria ( Henriksen et al. 2012; Weinert et al. 2013; Frankovsky, Keresztesová, et al. 2021), information on their functional importance is mostly lacking.

One of the obstacles that can prevent more systematic analysis of the roles of mt PTMs is that the data sets listing the modified mt proteins are scattered in multiple sources. To address this issue, we built the yeast mitochondrial posttranslational modification (y-mtPTM) database, which integrates data on 22 PTMs and provides an integrated resource for those interested in understanding the molecular details of the mt information processing system.

Materials and methods

Database design

The PTMs collected from the literature were first manually organized into an Excel file. This file was then parsed by a custom Python script into a relational database containing 4 tables. The first table lists the basic properties of each protein in the S. cerevisiae mt proteome, such as its identifiers, length, description, and sequence. The second table lists the PTMs, each specified by the protein ID, the position within the sequence, and the type of modification. The third table lists the references used as the sources for the PTMs. Finally, the last table links individual PTMs and references. The second custom Python script then reads the information stored in the database and creates an HTML file for each protein and additional necessary HTML files. The basic information about the proteins is taken from UniProt ( The UniProt Consortium 2023; https://www.uniprot.org/) and the Saccharomyces Genome Database (SGD; Cherry et al. 2012; https://www.yeastgenome.org); the 3D structures are taken from the AlphaFold Protein Structure Database ( Varadi et al. 2022; https://alphafold.ebi.ac.uk/). The structures are visualized using a modified version of the GLmol viewer (https://github.com/biochem-fan/GLmol), and the PDB files containing protein structures are processed to a form suitable for the GLmol viewer using PyMOL (https://pymol.org/). The y-mtPTM website uses the jQuery (https://jquery.com/) and Bootstrap (https://getbootstrap.com/) libraries. The Python scripts used to build the website use several Python libraries, including Biopython ( Cock et al. 2009; https://biopython.org/), pandas ( McKinney 2010; https://pandas.pydata.org/), and Jinja templates (https://palletsprojects.com/p/jinja/). The relational database is stored in SQLite (https://sqlite.org/). The source code used to build the y-mtPTM website is available in a GitHub repository (https://github.com/fmfi-compbio/y-mtptm).

Mapping the modified positions on the F₁F_O-ATPase complex

The structure of yeast F₁F_O-ATP synthase dimer ( Guo et al. 2017) was downloaded from PDB database (6B8H), and the mt proteins PTMs were highlighted using PyMOL software (The PyMOL Molecular Graphics System, Version 2.3.0. Schrödinger, LLC).

Results and discussion

The y-mtPTM database lists nearly 20,000 PTMs of yeast mt proteins

To generate a reference mt proteome, we compiled the list of 1,393 S. cerevisiae proteins that have been identified in at least 1 of the proteomic studies on highly purified mitochondria ( Sickmann et al. 2003; Reinders et al. 2006; Morgenstern et al. 2017; Vögtle et al. 2017; di Bartolomeo et al. 2020). As noted on the landing webpage, we are aware that the list contains proteins that may not be localized to the organelle but rather copurify with mt fraction. We did not attempt to remove such candidate contaminants as it was shown that a large fraction of the yeast mt proteome exhibits dual (or multiple) intracellular localization ( Ben-Menachem et al. 2011). This possibility, however, was not tested by an alternative experimental strategy such as that described by Bader et al. (2020). Removal of such ambiguously localized proteins may have resulted in a skewed list; therefore, we decided to keep all 1,393 proteins as members of the reference mt proteome ( Supplementary File 24 ).

Next, we collected data on PTMs identified on yeast mt proteins by published studies. As a primer, we combined data sets from our recent studies on mt protein phosphorylation ( Frankovsky, Vozáriková, et al. 2021) and succinylation ( Frankovsky, Keresztesová, et al. 2021). We then expanded the list by the addition of PTMs according to ∼200 papers (for the complete list, see Supplementary File 1 ) and made further additions by comparing our data set with the information provided by The SGD ( Cherry et al. 2012; https://www.yeastgenome.org). The list contains 19,990 entries representing sites modified by 22 PTMs (1 ambiguity is caused by a pair of identical Tef1/Tef2 proteins [UniProt # P02994] encoded by 2 paralogs [YBR118W/YPR080W]; the reported PTMs [138 in total] are listed as if present on both Tef1p and Tef2p). Importantly, we found that ∼0.05% of the reported modified sites did not correspond to the amino acids appropriate for a particular PTM type, and we eliminated those from the list.

The resulting spreadsheet file of published PTM sites was used for building the y-mtPTM database (http://compbio.fmph.uniba.sk/y-mtptm/; Fig. 1; see also Materials and methods for more details). After checking for accuracy, the spreadsheet file together with additional inputs is converted into a SQLite database and exported as an HTML to the website. Here, a simple search of a protein of interest reveals the modified amino acid residues, their position within the primary sequence as well as on 3D structure, and links to the source reference(s). Alternatively, one can browse the entire reference proteome containing a list of all mt proteins with information about their length and number of the corresponding PTMs. To help the user to associate particular PTMs with the experimental conditions in which they were detected, each study is briefly annotated. In addition, for each reference, we provide a list of modifications described in the corresponding study. This allows the user to see all PTMs identified under the investigated conditions. Each protein page links to the pages of individual references as well. All PTMs for the corresponding protein can be downloaded in the tabular CSV format, which can be opened in spreadsheet programs or easily analyzed by custom scripts. Each protein page also contains links several databases (TheCellVision ( Masinas et al. 2020), Uniprot ( The UniProt Consortium 2023), SGD ( Cherry et al. 2012), and FungiDB ( Basenko et al. 2018)) allowing the user to extract additional information about the protein of interest. The Browse database page allows composite filtering to display only proteins containing a specified combination of modifications. Also, the table can be sorted by any column, including gene name, modification counts, and protein length. The table can be also downloaded in the CSV format to be analyzed as a spreadsheet.

A scheme of the data collection, cleaning, processing, and submission process to the y-mtPTM database (created with BioRender.com).

Below, we provide a general overview of the PTMs listed in the database ( Fig. 2) as well as examples of its utilization.

Overview of the PTMs of mt proteins included in the y-mtPTM database. Depicted are the most frequent PTMs, and their occurrence is quantified. Less frequent PTMs are listed as “others” (created with BioRender.com).

The mt phosphoproteome consists of 1,099 proteins phosphorylated at nearly 13,500 sites

Phosphorylation is thus far the most abundant PTM detected on mt proteins. We have updated the data set provided by ( Frankovsky, Vozáriková, et al. 2021) incorporating data from phosphoproteomic studies not covered in the original paper and generated a list of 13,463 P-sites on 1,099 mt proteins (i.e. 78.9% of mt proteins contain at least 1 P-site; Table 1 and Supplementary File 2 ). When compared with SGD, our list contains from several percent to up to several folds more P-sites, depending on the protein of interest. Examples include known phosphorylation targets such as components of the protein import machinery ( Opalińska and Meisinger 2014; Matic et al. 2018) Tom20p (5 P-sites in SGD/7 P-sites in y-mtPTM), Tom22p (6/9), Tom5 (0/1), Tom40p (1/7), Tom6 (0/1), Tom71 (3/8), Tim 50 (1/5), Tim44p (4/6), Pam16 (3/7), and Mas1 (3/7). The higher number of P-sites present in the y-mtPTM Database is due to the incorporation of data from additional high- and low-throughput phosphoproteomic studies into the database.

An overview of PTMs currently listed in the y-mtPTM.

An overview of PTMs currently listed in the y-mtPTM.

Although the number of bona fide mitochondrially localized PKs is relatively low ( Tomaska 2000; Rao et al. 2011; Opalińska and Meisinger 2014; Frankovsky, Vozáriková, et al. 2021), the relative abundance of P-sites on mt proteins (9.7 P-sites per protein) is higher than that for the entire yeast proteome (∼6.1 P-sites per protein; Supplementary File 27 ). This apparent discrepancy between a relative scarcity of mt PKs and a high level of phosphorylation may have various explanations. First, the number of mitochondrially localized PKs can be higher, as some enzymes normally present in other cellular compartments can be relocalized to mitochondria under distinct conditions. Second, PKs may be members of large protein complexes, and their proximity to their substrates may enhance the number of P-sites in the corresponding subcompartment. In such cases, the user can employ MitCOM, a high-resolution complexome profiling data set covering nearly the entire yeast mt proteome ( Schulte et al. 2023). Colocalization of a protein of interest with a PK/protein phosphatase in a particular complex can be an indication that serves as a substrate for the corresponding enzyme. Finally, many mt proteins may be phosphorylated prior to or during their import into the organelle, such as components of the protein import machinery ( Schmidt et al. 2011). This possibility is supported by the observation that there are only 10 P-sites reported for the proteins encoded by mtDNA ( Table 1 and Supplementary Files 1–30 ), which is almost 20-fold lower frequency than in the case of mitochondrially targeted proteins synthesized in the cytosol. As these proteins are produced in mitochondria, they must be modified inside the organelle, and the fact that their phosphorylation seems to be very rare may be an indicator of an overall lower level of intramitochondrial phosphorylation. A complementary reason for the low number of P-sites on mtDNA-encoded proteins may be their hydrophobic nature and localization within membrane and/or protein complexes, which are thus hidden from the action of PKs (see also below).

Lysine acetylation and succinylation affect nearly one-third of all yeast mt proteins

The reactivity of the epsilon amino group of lysine makes this amino acid 1 of the most frequent acceptors of various chemical moieties. This reactivity increases at higher pH when the amino group becomes deprotonated and more nucleophilic. The mt matrix, due to its increased pH, not only possesses a favorable environment for the reactivity of lysine residues but is also a compartment with high concentrations of acyl-CoA, donors of various acyl groups that can be, at least in some cases, covalently attached to lysine via nonenzymatic reactions driven by the concentration of the corresponding acyl-CoA ( Ringel et al. 2018). This is particularly important for succinylation and acetylation, because the concentration of acetyl-CoA and succinyl-CoA within the mitochondria depends on the activity of enzymes involved in the tricarboxylic acid (TCA) cycle and thus reflects the metabolic state of the cell. Lysine acetylation and succinylation may therefore represent powerful means of connecting metabolism and mt signaling. Indeed, approximately one-third of all yeast mt proteins were shown to carry at least 1 acetylated and/or succinylated lysine ( Table 1 and Supplementary Files 3 and 4 ). The number of succinylated (2,373 on 415 proteins) and acetylated (1,304 on 341 proteins) sites is more likely even higher as in contrast to phosphorylation, these PTMs were subjects of only 3 whole proteomic studies ( Henriksen et al. 2012; Weinert et al. 2013; Frankovsky, Keresztesová, et al. 2021), complemented by a handful of low-throughput analyses (for a complete list of references, see Supplementary File 1 ). When provided glucose, S. cerevisiae, as a Crabtree-positive yeast species, ferments the sugar to ethanol first and then, after a short metabolic adaptation called diauxic shift, enters a respiratory phase of growth and oxidizes ethanol. This transition is accompanied by reorganization of the entire mt proteome and changes in mt metabolism, including the activation of the TCA cycle ( di Bartolomeo et al. 2020). The concomitant changes in the concentration of acetyl-CoA and succinyl-CoA most likely result in a different pattern of acetylation and succinylation that may be of functional significance (see also below).

PTMs of mt proteins via short peptides

Proteins are often covalently modified by short peptides that affect their activity, localization, and/or susceptibility to proteolytic degradation. Whereas neddylation and urmylation of mt proteins are very rare, ubiquitylation affects approximately one-third (1,867 sites on 445 proteins), and SUMOylation affects >10% (628 sites on 239 proteins) of the mt proteome ( Table 1, Fig. 2, and Supplementary Files 20–23 ). On the one hand, the existence of a large fraction of ubiquitylated mt proteins is in line with the observation that some of the components of the ubiquitin–proteasome machinery reside in mitochondria ( Lehmann et al. 2016; Leo et al. 2018). The interplay between ubiquitylation and phosphorylation was shown to be an important means of regulation ( Swaney et al. 2013). However, it seems likely that ubiquitylation of mt proteins occurs in the cytosol, where it plays a role in regulating the levels of mt-targeted proteins by affecting their availability for protein import, as was found in the case of mt proteins residing in the intermembrane space ( Bragoszewski et al. 2013). Indeed, it was recently shown that ubiquitylation and deubiquitylation are important in the removal of preproteins from the entry gate at the TOM complex ( Schulte et al. 2023) supporting the conclusion that this PTM occurs in the cytosol.

Similar to ubiquitylation, SUMOylation of mt proteins most likely also occurs in the cytosol as it does not seem to rely on mt targeting. SUMOylation was shown to be enhanced upon import failure; therefore, it is considered to be a PTM marking nonfunctional mt proteins that accumulate upon defective mt import and impaired proteostasis ( Paasch et al. 2018). Since the SUMO modifications were detected on 5 mtDNA-encoded proteins (Var1p, Atp6p, Cox1p, Cox2p, and reverse transcriptase AI2), they may also affect polypeptides residing within mitochondria and thus may have additional role(s) in the regulation of mt functions.

Other PTMs of the yeast mt proteome

The other PTMs included in the y-mtPTM Database affect a relatively low number of mt proteins. Except for methylation (37 modified proteins), benzoylation (40 modified proteins), and glycosylation (60 modified proteins) sites, PTMs such as carbamoylation, glutathionylation, propionylation, acylations with short fatty acids (myristoylation, farnesylation, and palmitoylation), deamidations, oxidations, and N6-lipoylation were detected on less than 1% of mt proteins ( Supplementary Files 5–19 ). This may be due to a low number of proteomic studies aimed at the investigation of the corresponding PTM and/or by their rare occurrence.

y-mtPTM as a platform for investigating the functional significance of mt PTMs

To illustrate how y-mtPTM can be utilized as a platform for generating experimentally testable hypotheses, we provide an example of a mt protein, Rim1p/YCR028C-A, that was shown to be modified by multiple (18) PTMs ( Fig. 3). In addition to the basic information extracted from the SGD ( Cherry et al. 2012; www.yeastgenome.org), options for downloading the sequence in FASTA, and/or the modifications in CSV formats and links to other databases (UniProt, TheCellVision.org, FungiDB), the PTMs are highlighted on both the primary sequence and 3D structure. When the amino acid sequence of Rim1p is then aligned to its homologs, the conserved modified residues can be identified resulting in questions that may be useful for designing experiments aimed at functional analysis of the corresponding PTMs ( Fig. 3, part 7). For human homologs (SSBP in the case of Rim1p), this information may be of relevance to understanding human cellular pathologies caused by mutations in the corresponding mt protein.

Utilization of the y-mtPTM database as a platform for generating experimentally testable hypotheses. A search for the corresponding protein (Rim1p/YCR028C-A is used as an example) yields (1) basic information extracted from the SGD (www.yeastgenome.org) and (2) provides options for downloading the sequence in FASTA and/or the modifications in CSV formats. In addition, links to other databases (UniProt, TheCellVision.org, and FungiDB) enable exploration of the information about the protein. (3) A 3D structure visualized by GLmol depicts the positions of the modified residues. (4) Each PTM is linked to reference(s) where the modification was reported. (5) The modified residues are highlighted within the primary sequence. Information from the database may inspire the user to perform additional bioinformatics analyses, such as aligning homologs of the protein (6) that may result in identification of conserved modified residue and (7) formulation of questions that may be useful for designing experiments aimed at functional analysis of the corresponding PTMs.

Although the majority of the members of the yeast mt proteome contain multiple PTMs (such as Rim1p), there is a relatively large category of proteins with no reported PTM. This group (200 proteins) represents 14.35% of the entire mt proteome ( Table 1 and Supplementary Files 5–26 ). There may be several reasons why they lack a reported modified site(s); they can be present at very low concentrations, some very short proteins may be lost during protein extraction, or the modified peptides were not detected by mass spectrometry due to technical reasons. Alternatively, some of these proteins may be protected against PTMs by topological constraints; i.e. hydrophobic proteins may be embedded into the membrane or reside inside large complexes. Indeed, gene ontology (GO) analysis of the unmodified proteins revealed that nearly half of them are localized on the mt membrane corresponding to the enrichment of those participating in membrane transport or assembly of respiratory chain complexes. This is exemplified by the fact that mtDNA-encoded proteins that are usually highly hydrophobic and are part of membrane complexes carry in total only 20 PTMs (2 PTMs per protein) compared with 14.5 PTMs per mt protein encoded by nuclear DNA ( Supplementary File 30 ).

The idea that the low frequency of PTMs is in part due to inaccessibility of the corresponding positions is illustrated by the map of the F₁F_O-ATP synthase complex where the F_O part that is inserted into the inner mt membrane almost completely lacks modified residues ( Fig. 4). In contrast, the surface of the F₁ subcomplex is enriched for sites susceptible to PTMs. The illustration also highlights the usefulness of the y-mtPTM database in providing data for mapping modified sites on proteins and protein complexes that can be instrumental in designing experiments aimed at addressing their functional significance.

Mapping of detected PTMs on the 3D structure of the mt F₁F_O-ATP synthase complex of S. cerevisiae. Amino acids of the yeast mt F₁F_O-ATP synthase dimer ( Guo et al. 2017) that can bear PTMs are highlighted with the following colors: cyan, phosphorylation sites; red, K-succinylation sites; green, K-ubiquitylation sites; and magenta, sites that can be modified by various modifications (phosphorylation, succinylation, ubiquitylation, and SUMOylation). The mt inner membrane is indicated by a gray arc. Subunits of the F1 (α₃β₃ hexamer, γ, δ, ε, OSCP) and F_O (a, b, c₁₀-ring, d, e, f, g, h, i/j, k, 8) regions are shown.

In addition, we compared the frequency of all PTMs affecting mt proteins of S. cerevisiae (expressed as the number of PTMs per amino acid) ( Supplementary File 31 ). Although some positions can be modified by various moieties thus skewing the frequency, ∼7.5% of mt proteins can be modified at frequency of >0.1 PTM per amino acid. The GO enrichment analysis of this group of proteins revealed that they are enriched for abundant enzymes involved in glycolysis and gluconeogenesis (data not shown). In addition, it also contains subunits of large complexes (such as respiratory chain complexes [Atp7p, Atp15p, Atp5p, Atp17p, Inh1p, and Cyc1p] or mt-nucleoids [Abf2p, Rim1p, Ilv5p, and Lat1p]) that may represent primary targets of PTMs and thus serve as signaling hubs of the corresponding complex.

The data provided by the y-mtPTM database can also be useful for a systematic analysis of the roles of PTMs in the regulation of a particular mt process. For example, mtDNA is compacted into higher-order structures called mitochondrial nucleoids (mt-nucleoids) composed of 37 proteins that directly or indirectly regulate mtDNA stability, inheritance, replication, recombination, repair, transcription, and translation. mt-nucleoids are highly dynamic structures undergoing changes associated with cellular transitions such as meiosis, sporulation, or diauxic shift. Although the roles of PTMs of some mt-nucleoid components are known, their systematic investigation was not performed. The y-mtPTM database currently lists >1,100 PTMs on nucleoid-associated proteins with over 30 PTMs per protein, which is ∼2-fold higher frequency than is the case for the entire mt proteome ( Table 1 and Supplementary Files 28 and 29 ). Interestingly, whereas the average number of P-sites per mt-nucleoid protein compared with all mt proteins is almost 3-fold lower (3.4 vs 9.8), many components of mt-nucleoids are heavily succinylated (7.7-fold higher number of K-succinylation sites than in the case of the entire mt proteome) and/or acetylated (4-fold higher number of K-acetyl sites). In fact, >20% of all succinylated and 10% of acetylated lysines found on mt proteins are present in the components of mt-nucleoids ( Table 1 and Supplementary File 1 ). This overrepresentation may be caused by the fact that mt-nucleoid resides in metabolically active sites of mitochondria and some components of the mt-nucleoid (Pda1p, Kgd1p, and Kgd2p) are enzymes involved in the production or utilization of acetyl-CoA and succinyl-CoA, which are donors of the corresponding acyl group. In addition, it was shown that α-ketoglutarate dehydrogenase may act as a catalyst of the succinylation reaction ( Gibson et al. 2015; Wang et al. 2017), so it may directly facilitate the modification of those proteins in its vicinity and thus connect metabolism with mt-nucleoid dynamics. Namely, increased activity of the TCA cycle (e.g. during the diauxic shift from fermentative to respiratory metabolism) may result in a higher concentration of succinyl-CoA, leading to increased succinylation of proteins associated with mt-nucleoids. This may lead to changes in the compaction of the mt genome accompanied by a higher rate of transcription and an increased production of mtDNA-encoded subunits of the respiratory chain. Although changes in both the number and state of mt-nucleoids during diauxic shift have been observed ( Frankovsky, Keresztesová, et al. 2021), the patterns of succinylation (and other PTMs) were not investigated and remain one of the challenges in the field.

Future directions

The y-mtPTM database currently lists 22 PTMs representing ∼5% of all known modifications ( Supplementary File 25 ). It can be expected that the number of proteomic studies aimed at comprehensive identification of the modified sites will be increasing. While this will give us a more complete picture of the posttranslationally modified sites, it will become more difficult to assess their functional significance. In some proteins, the number of modified sites exceeds 100, and some sites (especially lysines) can be modified by various chemical moieties, making it technically difficult to test their role(s) in the regulation of the corresponding protein. Some of the positions with a reported PTM can be artifactual owing to their presence in a region containing tracks of the same amino acid, making it difficult to assign the exact position on a given peptide. Even when the assignment of a PTM to the particular site is straightforward, having several PTMs on the same protein can make their functional analysis nearly impossible. For example, Abf2p contains 13 lysine residues that can be succinylated in vivo. A traditional approach would be to replace these lysines with arginines that are refractory to succinylation. However, we have observed that substitutions of just 2 lysines with arginines almost completely abolishes the DNA-binding activity of the protein in vitro regardless of its succinylation status ( Frankovsky, Keresztesová, et al. 2021), thus complicating functional studies of the role of succinylation of Abf2p in vivo.

One possibility of how to decrease the number of relevant modified sites is to perform comparative proteomic analyses of mt proteomes in related yeast species and identify conserved modified positions ( Fig. 3). The utility of this approach is illustrated by the results of a comparative analysis of phosphoproteomes of 18 fungal species demonstrating that only a small fraction of phosphosites are conserved and that these “ancient” sites are enriched at protein interfaces and thus more likely to be functionally important ( Studer et al. 2016). In this respect, it would be beneficial to integrate information about PTMs in the mitochondria of other organisms into the y-mtPTM database. Although at present we do not have the resources for such an upgrade, we are considering extending the database in the future.

Additionally, comparison of PTMs between cells exposed to various growth conditions may reveal residues whose modification(s) are potentially significant for dealing with environmental and/or intracellular changes, especially those involving mitochondria (such studies were thus far limited to phosphorylation and succinylation, e.g. Ohlmeier et al. 2010; Weinert et al. 2013; Renvoisé et al. 2017). Finally, modified positions that were detected by multiple studies may represent the major modified sites that can be a priority for more detailed studies.

A population of multiple copies of the same protein coexisting in a cell may exhibit distinct combinations of PTMs each having different effects on its biochemical properties. To obtain a picture of this intermolecular heterogeneity and stoichiometry of modifications, new technologies enabling sequencing of individual polypeptide chains using novel technologies such as nanopore sequencing ( Brinkerhoff et al. 2021) will be of a high value especially when they will be able to distinguish individual PTMs.

Although the reactions resulting in PTMs (or their removal) are generally well described, the nature of the catalysts responsible for modifications of mt proteins is not always known. In the case of phosphorylation, there is a relatively low number of clearly defined pairs of PKs/protein phosphatases and their substrates (for review, see Tomaska 2000; Rao et al. 2011; Opalińska and Meisinger 2014; Frankovsky, Vozáriková, et al. 2021). Although lysine acylations can proceed nonenzymatically, some can be promoted by acyl transferases, the occurrence of which in yeast mitochondria has not been investigated. Nevertheless, most PTMs are reversible and require a protein catalyst for their removal. Either such enzymes do not localize to mitochondria, or their role in removal of the corresponding PTM was not investigated. One example of the latter category is Hst4p, an NAD + -dependent sirtuin deacetylase that was shown to localize to the mitochondria in response to biotin starvation ( Madsen et al. 2015). Mammalian sirtuin SIRT5, a homolog of Hst4p, also exhibits mt localization ( Nakagawa et al. 2009) and, in addition to weak deacetylase activity, was also shown to possess desuccinylase activity ( Du et al. 2011). However, the role of Hst4p in the desuccinylation of mt proteins is currently not known, and the situation for other enzymes involved in mt PTMs is even less clear. To address these and other questions related to PTMs of mt proteins, the y-mtPTM database may represent a valuable tool by providing a comprehensive source of data.

Data availability

The source spreadsheet file for the y-mtPTM is provided as Supplementary material . The source code used to build the y-mtPTM website is available in a GitHub repository (https://github.com/fmfi-compbio/y-mtptm).

Supplemental material available at GENETICS online.

Acknowledgments

The y-mtPTM database was in part implemented through a student research project during the Genomics course at the Faculty of Natural Sciences and Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava. We kindly ask the members of the yeast mitochondrial community to provide feedback and additional data on modified sites that are missing in the current version of y-mtPTM. The text was corrected according to the suggestions of editors of American Journal Experts (www.aje.com).

Funding

Funding was provided by the Slovak Research and Development Agency (APVV-19-0068 to L.T. and APVV-18-0239 to J.N.), the Scientific Grant Agency of the Ministry of Education, Science, Research and Sport of the Slovak republic (VEGA 1/0061/20 to L.T., 1/0463/20 to B.B., and 1/0538/22 to T.V.), and Operation Program of Integrated Infrastructure for the project, Advancing University Capacity and Competence in Research, Development and Innovation, ITMS2014+: 313021X329, cofinanced by the European Regional Development Fund.