Quantitative Global Proteomics of Yeast PBP1 Deletion Mutants and Their Stress Responses Identifies Glucose Metabolism, Mitochondrial, and Stress Granule Changes

Differential Protein Expression in Set5p-Mediated Acetic Acid Stress Response and Novel Targets for Engineering Yeast Stress Tolerance

Acetic acid is a prevalent inhibitor in lignocellulosic hydrolysate, which represses microbial growth and bioproduction. Histone modification and chromatin remodeling have been revealed to be critical for regulating eukaryotic metabolism. However, related studies in chronic acetic acid stress responses remain unclear. Our previous studies revealed that overexpression of the histone H4 methyltransferase Set5p enhanced acetic acid stress tolerance of the budding yeast Saccharomyces cerevisiae. In this study, we examined the role of Set5p in acetic acid stress by analyzing global protein expression. Significant activation of intracellular protein expression under the stress was discovered, and the functions of the differential proteins were mainly involved in chromatin modification, signal transduction, and carbohydrate metabolism. Notably, a substantial increase of Set5p expression was observed in response to acetic acid stress. Functional studies demonstrated that the restriction of the telomere capping protein Rtc3p, as well as Ies3p and Taf14p, which are related to chromatin regulation, was critical for yeast stress response. This study enriches the understanding of the epigenetic regulatory mechanisms underlying yeast stress response mediated by histone-modifying enzymes. The results also benefit the development of robust yeast strains for lignocellulosic bioconversion.

Bidirectional de novo peptide sequencing using a transformer model

In proteomics, a crucial aspect is to identify peptide sequences. De novo sequencing methods have been widely employed to identify peptide sequences, and numerous tools have been proposed over the past two decades. Recently, deep learning approaches have been introduced for de novo sequencing. Previous methods focused on encoding tandem mass spectra and predicting peptide sequences from the first amino acid onwards. However, when predicting peptides using tandem mass spectra, the peptide sequence can be predicted not only from the first amino acid but also from the last amino acid due to the coexistence of b-ion (or a- or c-ion) and y-ion (or x- or z-ion) fragments in the tandem mass spectra. Therefore, it is essential to predict peptide sequences bidirectionally. Our approach, called NovoB, utilizes a Transformer model to predict peptide sequences bidirectionally, starting with both the first and last amino acids. In comparison to Casanovo, our method achieved an improvement of the average peptide-level accuracy rate of approximately 9.8% across all species.

DbyDeep: Exploration of MS-Detectable Peptides via Deep Learning

Predicting peptide detectability is useful in a variety of mass spectrometry (MS)-based proteomics applications, particularly targeted proteomics. However, most machine learning-based computational methods have relied solely on information from the peptide itself, such as its amino acid sequences or physicochemical properties, despite the fact that peptides detected by MS are dependent on many factors, including protein sample preparation, digestion, separation, ionization, and precursor selection during MS experiments. DbyDeep (Detectability by Deep learning) is an innovative end-to-end LSTM network model for peptide detectability prediction that incorporates sequence contexts of peptides and their cleavage sites (by protease). Utilizing the cleavage site contexts could improve the performance of prediction, and DbyDeep outperformed existing methods in predicting peptides recognizable from multiple MS/MS data sets with diverse species and MS instruments. We argue for the necessity of a learning model that encompasses several contexts associated with peptide detection, as opposed to depending just on peptide sequences. There is a Python implementation of DbyDeep at https://github.com/BISCodeRepo/DbyDeep.

Critical evaluation of the use of artificial data for machine learning based de novo peptide identification

Proteins are essential components of all living cells and so the study of their in situ expression, proteomics, has wide reaching applications. Peptide identification in proteomics typically relies on matching high resolution tandem mass spectra to a protein database but can also be performed de novo. While artificial spectra have been successfully incorporated into database search pipelines to increase peptide identification rates, little work has been done to investigate the utility of artificial spectra in the context of de novo peptide identification. Here, we perform a critical analysis of the use of artificial data for the training and evaluation of de novo peptide identification algorithms. First, we classify the different fragment ion types present in real spectra and then estimate the number of spurious matches using random peptides. We then categorise the different types of noise present in real spectra. Finally, we transfer this knowledge to artificial data and test the performance of a state-of-the-art de novo peptide identification algorithm trained using artificial spectra with and without relevant noise addition. Noise supplementation increased artificial training data performance from 30% to 77% of real training data peptide recall. While real data performance was not fully replicated, this work provides the first steps towards an artificial spectrum framework for the training and evaluation of de novo peptide identification algorithms. Further enhanced artificial spectra may allow for more in depth analysis of de novo algorithms as well as alleviating the reliance on database searches for training data.

Application of a Novel Hybrid CNN-GNN for Peptide Ion Encoding

Almost all state-of-the-art de novo peptide sequencing algorithms now use machine learning models to encode fragment peaks and hence identify amino acids in mass spectrometry (MS) spectra. Previous work has highlighted how the inherent MS challenges of noise and missing peptide peaks detrimentally affect the performance of these models. In the present research we extracted and evaluated the encoding modules from 3 state-of-the-art de novo peptide sequencing algorithms. We also propose a convolutional neural network-graph neural network machine learning model for encoding peptide ions in tandem MS spectra. We compared the proposed encoding module to those used in the state-of-the-art de novo peptide sequencing algorithms by assessing their ability to identify b-ions and y-ions in MS spectra. This included a comprehensive evaluation in both real and artificial data across various levels of noise and missing peptide peaks. The proposed model performed best across all data sets using two different metrics (area under the receiver operating characteristic curve (AUC) and average precision). The work also highlighted the effect of including additional features such as intensity rank in these encoding modules as well as issues with using the AUC as a metric. This work is of significance to those designing future de novo peptide identification algorithms as it is the first step toward a new approach.

Synthetic negative genome screen of the GPN-loop GTPase NPA3 in Saccharomyces cerevisiae

The GPN-loop GTPase Npa3 is encoded by an essential gene in the yeast Saccharomyces cerevisiae. Npa3 plays a critical role in the assembly and nuclear accumulation of RNA polymerase II (RNAPII), a function that may explain its essentiality. Genetic interactions describe the extent to which a mutation in a particular gene affects a specific phenotype when co-occurring with an alteration in a second gene. Discovering synthetic negative genetic interactions has long been used as a tool to delineate the functional relatedness between pairs of genes participating in common or compensatory biological pathways. Previously, our group showed that nuclear targeting and transcriptional activity of RNAPII were unaffected in cells expressing exclusively a C-terminal truncated mutant version of Npa3 (npa3∆C) lacking the last 106 residues naturally absent from the single GPN protein in Archaea, but universally conserved in all Npa3 orthologs of eukaryotes. To gain insight into novel cellular functions for Npa3, we performed here a genome-wide Synthetic Genetic Array (SGA) study coupled to bulk fluorescence monitoring to identify negative genetic interactions of NPA3 by crossing an npa3∆C strain with a 4,389 nonessential gene-deletion collection. This genetic screen revealed previously unknown synthetic negative interactions between NPA3 and 15 genes. Our results revealed that the Npa3 C-terminal tail extension regulates the participation of this essential GTPase in previously unknown biological processes related to mitochondrial homeostasis and ribosome biogenesis.

Atxn2-CAG100-KnockIn mouse spinal cord shows progressive TDP43 pathology associated with cholesterol biosynthesis suppression

Large polyglutamine expansions in Ataxin-2 (ATXN2) cause multi-system nervous atrophy in Spinocerebellar Ataxia type 2 (SCA2). Intermediate size expansions carry a risk for selective motor neuron degeneration, known as Amyotrophic Lateral Sclerosis (ALS). Conversely, the depletion of ATXN2 prevents disease progression in ALS. Although ATXN2 interacts directly with RNA, and in ALS pathogenesis there is a crucial role of RNA toxicity, the affected functional pathways remain ill defined. Here, we examined an authentic SCA2 mouse model with Atxn2-CAG100-KnockIn for a first definition of molecular mechanisms in spinal cord pathology. Neurophysiology of lower limbs detected sensory neuropathy rather than motor denervation. Triple immunofluorescence demonstrated cytosolic ATXN2 aggregates sequestrating TDP43 and TIA1 from the nucleus. In immunoblots, this was accompanied by elevated CASP3, RIPK1 and PQBP1 abundance. RT-qPCR showed increase of Grn, Tlr7 and Rnaset2 mRNA versus Eif5a2, Dcp2, Uhmk1 and Kif5a decrease. These SCA2 findings overlap well with known ALS features. Similar to other ataxias and dystonias, decreased mRNA levels for Unc80, Tacr1, Gnal, Ano3, Kcna2, Elovl5 and Cdr1 contrasted with Gpnmb increase. Preterminal stage tissue showed strongly activated microglia containing ATXN2 aggregates, with parallel astrogliosis. Global transcriptome profiles from stages of incipient motor deficit versus preterminal age identified molecules with progressive downregulation, where a cluster of cholesterol biosynthesis enzymes including Dhcr24, Msmo1, Idi1 and Hmgcs1 was prominent. Gas chromatography demonstrated a massive loss of crucial cholesterol precursor metabolites. Overall, the ATXN2 protein aggregation process affects diverse subcellular compartments, in particular stress granules, endoplasmic reticulum and receptor tyrosine kinase signaling. These findings identify new targets and potential biomarkers for neuroprotective therapies.

Mouse Ataxin-2 Expansion Downregulates CamKII and Other Calcium Signaling Factors, Impairing Granule-Purkinje Neuron Synaptic Strength

Spinocerebellar ataxia type 2 (SCA2) is caused by polyglutamine expansion in Ataxin-2 (ATXN2). This factor binds RNA/proteins to modify metabolism after stress, and to control calcium (Ca2+) homeostasis after stimuli. Cerebellar ataxias and corticospinal motor neuron degeneration are determined by gain/loss in ATXN2 function, so we aimed to identify key molecules in this atrophic process, as potential disease progression markers. Our Atxn2-CAG100-Knock-In mouse faithfully models features observed in patients at pre-onset, early and terminal stages. Here, its cerebellar global RNA profiling revealed downregulation of signaling cascades to precede motor deficits. Validation work at mRNA/protein level defined alterations that were independent of constant physiological ATXN2 functions, but specific for RNA/aggregation toxicity, and progressive across the short lifespan. The earliest changes were detected at three months among Ca2+ channels/transporters (Itpr1, Ryr3, Atp2a2, Atp2a3, Trpc3), IP3 metabolism (Plcg1, Inpp5a, Itpka), and Ca2+-Calmodulin dependent kinases (Camk2a, Camk4). CaMKIV-Sam68 control over alternative splicing of Nrxn1, an adhesion component of glutamatergic synapses between granule and Purkinje neurons, was found to be affected. Systematic screening of pre/post-synapse components, with dendrite morphology assessment, suggested early impairment of CamKIIα abundance together with the weakening of parallel fiber connectivity. These data reveal molecular changes due to ATXN2 pathology, primarily impacting excitability and communication.

Mid-Gestation lethality of Atxn2l-Ablated Mice

Depletion of yeast/fly Ataxin-2 rescues TDP-43 overexpression toxicity. In mouse models of Amyotrophic Lateral Sclerosis via TDP-43 overexpression, depletion of its ortholog ATXN2 mitigated motor neuron degeneration and extended lifespan from 25 days to >300 days. There is another ortholog in mammals, named ATXN2L (Ataxin-2-like), which is almost uncharacterized but also functions in RNA surveillance at stress granules. We generated mice with Crispr/Cas9-mediated deletion of Atxn2l exons 5-8, studying homozygotes prenatally and heterozygotes during aging. Our novel findings indicate that ATXN2L absence triggers mid-gestational embryonic lethality, affecting female animals more strongly. Weight and development stages of homozygous mutants were reduced. Placenta phenotypes were not apparent, but brain histology showed lamination defects and apoptosis. Aged heterozygotes showed no locomotor deficits or weight loss over 12 months. Null mutants in vivo displayed compensatory efforts to maximize Atxn2l expression, which were prevented upon nutrient abundance in vitro. Mouse embryonal fibroblast cells revealed more multinucleated giant cells upon ATXN2L deficiency. In addition, in human neural cells, transcript levels of ATXN2L were induced upon starvation and glucose and amino acids exposure, but this induction was partially prevented by serum or low cholesterol administration. Neither ATXN2L depletion triggered dysregulation of ATXN2, nor a converse effect was observed. Overall, this essential role of ATXN2L for embryogenesis raises questions about its role in neurodegenerative diseases and neuroprotective therapies.

pNovo 3: precise de novo peptide sequencing using a learning-to-rank framework

MOTIVATION

De novo peptide sequencing based on tandem mass spectrometry data is the key technology of shotgun proteomics for identifying peptides without any database and assembling unknown proteins. However, owing to the low ion coverage in tandem mass spectra, the order of certain consecutive amino acids cannot be determined if all of their supporting fragment ions are missing, which results in the low precision of de novo sequencing.

RESULTS

In order to solve this problem, we developed pNovo 3, which used a learning-to-rank framework to distinguish similar peptide candidates for each spectrum. Three metrics for measuring the similarity between each experimental spectrum and its corresponding theoretical spectrum were used as important features, in which the theoretical spectra can be precisely predicted by the pDeep algorithm using deep learning. On seven benchmark datasets from six diverse species, pNovo 3 recalled 29-102% more correct spectra, and the precision was 11-89% higher than three other state-of-the-art de novo sequencing algorithms. Furthermore, compared with the newly developed DeepNovo, which also used the deep learning approach, pNovo 3 still identified 21-50% more spectra on the nine datasets used in the study of DeepNovo. In summary, the deep learning and learning-to-rank techniques implemented in pNovo 3 significantly improve the precision of de novo sequencing, and such machine learning framework is worth extending to other related research fields to distinguish the similar sequences.

AVAILABILITY AND IMPLEMENTATION

pNovo 3 can be freely downloaded from http://pfind.ict.ac.cn/software/pNovo/index.html.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

In Human and Mouse Spino-Cerebellar Tissue, Ataxin-2 Expansion Affects Ceramide-Sphingomyelin Metabolism

Ataxin-2 (human gene symbol ATXN2) acts during stress responses, modulating mRNA translation and nutrient metabolism. Ataxin-2 knockout mice exhibit progressive obesity, dyslipidemia, and insulin resistance. Conversely, the progressive ATXN2 gain of function due to the fact of polyglutamine (polyQ) expansions leads to a dominantly inherited neurodegenerative process named spinocerebellar ataxia type 2 (SCA2) with early adipose tissue loss and late muscle atrophy. We tried to understand lipid dysregulation in a SCA2 patient brain and in an authentic mouse model. Thin layer chromatography of a patient cerebellum was compared to the lipid metabolome of Atxn2-CAG100-Knockin (KIN) mouse spinocerebellar tissue. The human pathology caused deficits of sulfatide, galactosylceramide, cholesterol, C22/24-sphingomyelin, and gangliosides GM1a/GD1b despite quite normal levels of C18-sphingomyelin. Cerebellum and spinal cord from the KIN mouse showed a consistent decrease of various ceramides with a significant elevation of sphingosine in the more severely affected spinal cord. Deficiency of C24/26-sphingomyelins contrasted with excess C18/20-sphingomyelin. Spinocerebellar expression profiling revealed consistent reductions of CERS protein isoforms, Sptlc2 and Smpd3, but upregulation of Cers2 mRNA, as prominent anomalies in the ceramide-sphingosine metabolism. Reduction of Asah2 mRNA correlated to deficient S1P levels. In addition, downregulations for the elongase Elovl1, Elovl4, Elovl5 mRNAs and ELOVL4 protein explain the deficit of very long-chain sphingomyelin. Reduced ASMase protein levels correlated to the accumulation of long-chain sphingomyelin. Overall, a deficit of myelin lipids was prominent in SCA2 nervous tissue at prefinal stage and not compensated by transcriptional adaptation of several metabolic enzymes. Myelination is controlled by mTORC1 signals; thus, our human and murine observations are in agreement with the known role of ATXN2 yeast, nematode, and mouse orthologs as mTORC1 inhibitors and autophagy promoters.

Calcium-responsive transactivator (CREST) toxicity is rescued by loss of PBP1/ATXN2 function in a novel yeast proteinopathy model and in transgenic flies

Proteins associated with familial neurodegenerative disease often aggregate in patients’ neurons. Several such proteins, e.g. TDP-43, aggregate and are toxic when expressed in yeast. Deletion of the ATXN2 ortholog, PBP1, reduces yeast TDP-43 toxicity, which led to identification of ATXN2 as an amyotrophic lateral sclerosis (ALS) risk factor and therapeutic target. Likewise, new yeast neurodegenerative disease models could facilitate identification of other risk factors and targets. Mutations in SS18L1, encoding the calcium-responsive transactivator (CREST) chromatin-remodeling protein, are associated with ALS. We show that CREST is toxic in yeast and forms nuclear and occasionally cytoplasmic foci that stain with Thioflavin-T, a dye indicative of amyloid-like protein. Like the yeast chromatin-remodeling factor SWI1, CREST inhibits silencing of FLO genes. Toxicity of CREST is enhanced by the [PIN⁺] prion and reduced by deletion of the HSP104 chaperone required for the propagation of many yeast prions. Likewise, deletion of PBP1 reduced CREST toxicity and aggregation. In accord with the yeast data, we show that the Drosophila ortholog of human ATXN2, dAtx2, is a potent enhancer of CREST toxicity. Downregulation of dAtx2 in flies overexpressing CREST in retinal ganglion cells was sufficient to largely rescue the severe degenerative phenotype induced by human CREST. Overexpression caused considerable co-localization of CREST and PBP1/ATXN2 in cytoplasmic foci in both yeast and mammalian cells. Thus, co-aggregation of CREST and PBP1/ATXN2 may serve as one of the mechanisms of PBP1/ATXN2-mediated toxicity. These results extend the spectrum of ALS associated proteins whose toxicity is regulated by PBP1/ATXN2, suggesting that therapies targeting ATXN2 may be effective for a wide range of neurodegenerative diseases.

Mutations in the calcium-responsive transactivator (CREST) protein have been shown to cause amyotrophic lateral sclerosis (ALS). Here we show that the human CREST protein expressed in yeast forms largely nuclear aggregates and is toxic. We also show that the HSP104 chaperone required for propagation of yeast prions is likewise required for CREST toxicity. Furthermore deletion of HSP104 affects CREST aggregation. ATXN2, previously shown to modify ALS toxicity caused by mutations in the TDP-43 encoding gene, also modifies toxicity of CREST expressed in either yeast or flies. In addition, deletion of the yeast ATXN2 ortholog reduces CREST aggregation. These results extend the spectrum of ALS associated proteins whose toxicity is regulated by ATXN2, suggesting that therapies targeting ATXN2 may be effective for a wide range of neurodegenerative diseases.

Generation of an Atxn2-CAG100 knock-in mouse reveals N-acetylaspartate production deficit due to early Nat8l dysregulation

Spinocerebellar ataxia type 2 (SCA2) is an autosomal dominant neurodegenerative disorder caused by CAG-expansion mutations in the ATXN2 gene, mainly affecting motor neurons in the spinal cord and Purkinje neurons in the cerebellum. While the large expansions were shown to cause SCA2, the intermediate length expansions lead to increased risk for several atrophic processes including amyotrophic lateral sclerosis and Parkinson variants, e.g. progressive supranuclear palsy. Intense efforts to pioneer a neuroprotective therapy for SCA2 require longitudinal monitoring of patients and identification of crucial molecular pathways. The ataxin-2 (ATXN2) protein is mainly involved in RNA translation control and regulation of nutrient metabolism during stress periods. The preferential mRNA targets of ATXN2 are yet to be determined. In order to understand the molecular disease mechanism throughout different prognostic stages, we generated an Atxn2-CAG100-knock-in (KIN) mouse model of SCA2 with intact murine ATXN2 expression regulation. Its characterization revealed somatic mosaicism of the expansion, with shortened lifespan, a progressive spatio-temporal pattern of pathology with subsequent phenotypes, and anomalies of brain metabolites such as N-acetylaspartate (NAA), all of which mirror faithfully the findings in SCA2 patients. Novel molecular analyses from stages before the onset of motor deficits revealed a strong selective effect of ATXN2 on Nat8l mRNA which encodes the enzyme responsible for NAA synthesis. This metabolite is a prominent energy store of the brain and a well-established marker for neuronal health. Overall, we present a novel authentic rodent model of SCA2, where in vivo magnetic resonance imaging was feasible to monitor progression and where the definition of earliest transcriptional abnormalities was possible. We believe that this model will not only reveal crucial insights regarding the pathomechanism of SCA2 and other ATXN2-associated disorders, but will also aid in developing gene-targeted therapies and disease prevention.

Methodologies and Applications of Proteomics for Study of Yeast Strains: An Update

Yeasts are one of the mostly used microorganisms as models in several studies. A wide range of applications in different processes can be attributed to their intrinsic characteristics. They are eukaryotes and therefore valuable expression hosts that require elaborate post-translational modifications. Their arsenal of proteins has become a valuable biochemical tool for the catalysis of several reactions of great value to the food (beverages), pharmaceutical and energy industries. Currently, the main challenge in systemic yeast biology is the understanding of the expression, function and regulation of the protein pool encoded by such microorganisms. In this review, we will provide an overview of the proteomic methodologies used in the analysis of yeasts. This research focuses on the advantages and improvements in their most recent applications with an understanding of the functionality of the proteins of these microorganisms, as well as an update of the advances of methodologies employed in mass spectrometry.

Ataxin-2: A versatile posttranscriptional regulator and its implication in neural function

Ataxin-2 (ATXN2) is a eukaryotic RNA-binding protein that is conserved from yeast to human. Genetic expansion of a poly-glutamine tract in human ATXN2 has been implicated in several neurodegenerative diseases, likely acting through gain-of-function effects. Emerging evidence, however, suggests that ATXN2 plays more direct roles in neural function via specific molecular and cellular pathways. ATXN2 and its associated protein complex control distinct steps in posttranscriptional gene expression, including poly-A tailing, RNA stabilization, microRNA-dependent gene silencing, and translational activation. Specific RNA substrates have been identified for the functions of ATXN2 in aspects of neural physiology, such as circadian rhythms and olfactory habituation. Genetic models of ATXN2 loss-of-function have further revealed its significance in stress-induced cytoplasmic granules, mechanistic target of rapamycin signaling, and cellular metabolism, all of which are crucial for neural homeostasis. Accordingly, we propose that molecular evolution has been selecting the ATXN2 protein complex as an important trans-acting module for the posttranscriptional control of diverse neural functions. This explains how ATXN2 intimately interacts with various neurodegenerative disease genes, and suggests that loss-of-function effects of ATXN2 could be therapeutic targets for ATXN2-related neurological disorders. This article is categorized under: RNA in Disease and Development > RNA in Disease RNA Interactions with Proteins and Other Molecules > Protein-RNA Interactions: Functional Implications.

De novo peptide sequencing by deep learning

De novo peptide sequencing from tandem MS data is the key technology in proteomics for the characterization of proteins, especially for new sequences, such as mAbs. In this study, we propose a deep neural network model, DeepNovo, for de novo peptide sequencing. DeepNovo architecture combines recent advances in convolutional neural networks and recurrent neural networks to learn features of tandem mass spectra, fragment ions, and sequence patterns of peptides. The networks are further integrated with local dynamic programming to solve the complex optimization task of de novo sequencing. We evaluated the method on a wide variety of species and found that DeepNovo considerably outperformed state of the art methods, achieving 7.7-22.9% higher accuracy at the amino acid level and 38.1-64.0% higher accuracy at the peptide level. We further used DeepNovo to automatically reconstruct the complete sequences of antibody light and heavy chains of mouse, achieving 97.5-100% coverage and 97.2-99.5% accuracy, without assisting databases. Moreover, DeepNovo is retrainable to adapt to any sources of data and provides a complete end-to-end training and prediction solution to the de novo sequencing problem. Not only does our study extend the deep learning revolution to a new field, but it also shows an innovative approach in solving optimization problems by using deep learning and dynamic programming.