Predicting the dynamics of protein abundance

Predicting the S. cerevisiae Gene Expression Score by a Machine Learning Classifier

The topic of this work is gene expression and its score according to various factors analyzed globally using machine learning techniques. The expression score (ES) of genes characterizes their activity and, thus, their importance for cellular processes. This may depend on many different factors (attributes). To find the most important classifier, a machine learning classifier (random forest) was selected, trained, and optimized on the Waikato Environment for Knowledge Analysis WEKA platform, resulting in the most accurate attribute-dependent prediction of the ES of Saccharomyces cerevisiae genes. In this way, data from the Saccharomyces Genome Database (SGD), presenting ES values corresponding to a wide spectrum of attributes, were used, revised, classified, and balanced, and the significance of the considered attributes was evaluated. In this way, the novel random forest model indicates the most important attributes determining classes of low, moderate, and high ES. They cover both the experimental conditions and the genetic, physical, statistical, and logistic features. During validation, the obtained model could classify the instances of a primary unknown test set with a correctness of 84.1%.

Unveiling novel features and phylogenomic assessment of indigenous Priestia megaterium AB-S79 using comparative genomics

Priestia megaterium strain AB-S79 isolated from active gold mine soil previously expressed in vitro heavy metal resistance and has a 5.7 Mb genome useful for biotechnological exploitation. This study used web-based bioinformatic resources to analyze P. megaterium AB-S79 genomic relatedness, decipher its secondary metabolite biosynthetic gene clusters (BGCs), and better comprehend its taxa. Genes were highly conserved across the 14 P. megaterium genomes examined here. The pangenome reflected a total of 61,397 protein-coding genes, 59,745 homolog protein family hits, and 1,652 singleton protein family hits. There were also 7,735 protein families, including 1,653 singleton families and 6,082 homolog families. OrthoVenn3 comparison of AB-S79 protein sequences with 13 other P. megaterium strains, 7 other Priestia spp., and 6 other Bacillus spp. highlighted AB-S79's unique genomic and evolutionary trait. antiSMASH identified two key transcription factor binding site regulators in AB-S79's genome: zinc-responsive repressor (Zur) and antibiotic production activator (AbrC3), plus putative enzymes for the biosynthesis of terpenes and ranthipeptides. AB-S79 also harbors BGCs for two unique siderophores (synechobactins and schizokinens), phosphonate, dienelactone hydrolase family protein, and phenazine biosynthesis protein (phzF), which is significant for this study. Phosphonate particularly showed specificity for the P. megaterium sp. validating the effect of gene family expansion and contraction. P. megaterium AB-S79 looks to be a viable source for value-added compounds. Thus, this study contributes to the theoretical framework for the systematic metabolic and genetic exploitation of the P. megaterium sp., particularly the value-yielding strains.

IMPORTANCE

This study explores microbial natural product discovery using genome mining, focusing on Priestia megaterium. Key findings highlight the potential of P. megaterium, particularly strain AB-S79, for biotechnological applications. The research shows a limited output of P. megaterium genome sequences from Africa, emphasizing the importance of the native strain AB-S79. Additionally, the study underlines the strain's diverse metabolic capabilities, reinforcing its suitability as a model for microbial cell factories and its foundational role in future biotechnological exploitation.

Protein constraints in genome-scale metabolic models: Data integration, parameter estimation, and prediction of metabolic phenotypes

Genome-scale metabolic models provide a valuable resource to study metabolism and cell physiology. These models are employed with approaches from the constraint-based modeling framework to predict metabolic and physiological phenotypes. The prediction performance of genome-scale metabolic models can be improved by including protein constraints. The resulting protein-constrained models consider data on turnover numbers (kcat ) and facilitate the integration of protein abundances. In this systematic review, we present and discuss the current state-of-the-art regarding the estimation of kinetic parameters used in protein-constrained models. We also highlight how data-driven and constraint-based approaches can aid the estimation of turnover numbers and their usage in improving predictions of cellular phenotypes. Finally, we identify standing challenges in protein-constrained metabolic models and provide a perspective regarding future approaches to improve the predictive performance.

Accurate prediction of in vivo protein abundances by coupling constraint-based modelling and machine learning

Quantification of how different environmental cues affect protein allocation can provide important insights for understanding cell physiology. While absolute quantification of proteins can be obtained by resource-intensive mass-spectrometry-based technologies, prediction of protein abundances offers another way to obtain insights into protein allocation. Here we present CAMEL, a framework that couples constraint-based modelling with machine learning to predict protein abundance for any environmental condition. This is achieved by building machine learning models that leverage static features, derived from protein sequences, and condition-dependent features predicted from protein-constrained metabolic models. Our findings demonstrate that CAMEL results in excellent prediction of protein allocation in E. coli (average Pearson correlation of at least 0.9), and moderate performance in S. cerevisiae (average Pearson correlation of at least 0.5). Therefore, CAMEL outperformed contending approaches without using molecular read-outs from unseen conditions and provides a valuable tool for using protein allocation in biotechnological applications.

Synonymous mutations in the phosphoglycerate kinase 1 gene induce an altered response to protein misfolding in Schizosaccharomyces pombe

Background

Proteostasis refers to the processes that regulate the biogenesis, folding, trafficking, and degradation of proteins. Any alteration in these processes can lead to cell malfunction. Protein synthesis, a key proteostatic process, is highly-regulated at multiple levels to ensure adequate adaptation to environmental and physiological challenges such as different stressors, proteotoxic conditions and aging, among other factors. Because alterations in protein translation can lead to protein misfolding, examining how protein translation is regulated may also help to elucidate in part how proteostasis is controlled. Codon usage bias has been implicated in the fine-tuning of translation rate, as more-frequent codons might be read faster than their less-frequent counterparts. Thus, alterations in codon usage due to synonymous mutations may alter translation kinetics and thereby affect the folding of the nascent polypeptide, without altering its primary structure. To date, it has been difficult to predict the effect of synonymous mutations on protein folding and cellular fitness due to a scarcity of relevant data. Thus, the purpose of this work was to assess the effect of synonymous mutations in discrete regions of the gene that encodes the highly-expressed enzyme 3-phosphoglycerate kinase 1 (pgk1) in the fission yeast Schizosaccharomyces pombe.

Results

By means of systematic replacement of synonymous codons along pgk1, we found slightly-altered protein folding and activity in a region-specific manner. However, alterations in protein aggregation, heat stress as well as changes in proteasome activity occurred independently of the mutated region. Concomitantly, reduced mRNA levels of the chaperones Hsp9 and Hsp16 were observed.

Conclusion

Taken together, these data suggest that codon usage bias of the gene encoding this highly-expressed protein is an important regulator of protein function and proteostasis.

Bioinformatic Assessment of Factors Affecting the Correlation between Protein Abundance and Elongation Efficiency in Prokaryotes

Protein abundance is crucial for the majority of genetically regulated cell functions to act properly in prokaryotic organisms. Therefore, developing bioinformatic methods for assessing the efficiency of different stages of gene expression is of great importance for predicting the actual protein abundance. One of these steps is the evaluation of translation elongation efficiency based on mRNA sequence features, such as codon usage bias and mRNA secondary structure properties. In this study, we have evaluated correlation coefficients between experimentally measured protein abundance and predicted elongation efficiency characteristics for 26 prokaryotes, including non-model organisms, belonging to diverse taxonomic groups The algorithm for assessing elongation efficiency takes into account not only codon bias, but also number and energy of secondary structures in mRNA if those demonstrate an impact on predicted elongation efficiency of the ribosomal protein genes. The results show that, for a number of organisms, secondary structures are a better predictor of protein abundance than codon usage bias. The bioinformatic analysis has revealed several factors associated with the value of the correlation coefficient. The first factor is the elongation efficiency optimization type-the organisms whose genomes are optimized for codon usage only have significantly higher correlation coefficients. The second factor is taxonomical identity-bacteria that belong to the class Bacilli tend to have higher correlation coefficients among the analyzed set. The third is growth rate, which is shown to be higher for the organisms with higher correlation coefficients between protein abundance and predicted translation elongation efficiency. The obtained results can be useful for further improvement of methods for protein abundance prediction.

Protein Abundance Prediction Through Machine Learning Methods

Proteins are responsible for most physiological processes, and their abundance provides crucial information for systems biology research. However, absolute protein quantification, as determined by mass spectrometry, still has limitations in capturing the protein pool. Protein abundance is impacted by translation kinetics, which rely on features of codons. In this study, we evaluated the effect of codon usage bias of genes on protein abundance. Notably, we observed differences regarding codon usage patterns between genes coding for highly abundant proteins and genes coding for less abundant proteins. Analysis of synonymous codon usage and evolutionary selection showed a clear split between the two groups. Our machine learning models predicted protein abundances from codon usage metrics with remarkable accuracy, achieving strong correlation with experimental data. Upon integration of the predicted protein abundance in enzyme-constrained genome-scale metabolic models, the simulated phenotypes closely matched experimental data, which demonstrates that our predictive models are valuable tools for systems metabolic engineering approaches.

PIKE-R2P: Protein-protein interaction network-based knowledge embedding with graph neural network for single-cell RNA to protein prediction

Background

Recent advances in simultaneous measurement of RNA and protein abundances at single-cell level provide a unique opportunity to predict protein abundance from scRNA-seq data using machine learning models. However, existing machine learning methods have not considered relationship among the proteins sufficiently.

Results

We formulate this task in a multi-label prediction framework where multiple proteins are linked to each other at the single-cell level. Then, we propose a novel method for single-cell RNA to protein prediction named PIKE-R2P, which incorporates protein–protein interactions (PPI) and prior knowledge embedding into a graph neural network. Compared with existing methods, PIKE-R2P could significantly improve prediction performance in terms of smaller errors and higher correlations with the gold standard measurements.

Conclusion

The superior performance of PIKE-R2P indicates that adding the prior knowledge of PPI to graph neural networks can be a powerful strategy for cross-modality prediction of protein abundances at the single-cell level.

Proteogenomic interrogation of cancer cell lines: an overview of the field

Introduction: Cancer cell lines (CCLs) have been a major resource for cancer research. Over the past couple of decades, they have been instrumental in omic profiling method development and as model systems to generate new knowledge in cell and cancer biology. More recently, with the increasing amount of genomic, transcriptomic and proteomic data being generated in hundreds of CCLs, there is growing potential for integrative proteogenomic data analyses to be performed.Areas covered: In this review, we first describe the most commonly used proteome profiling methods in CCLs. We then discuss how these proteomics data can be integrated with genomics data for proteogenomics analyses. Finally, we highlight some of the recent biological discoveries that have arisen from proteogenomics analyses of CCLs.Expert opinion: Protegeonomics analyses of CCLs have so far enabled the discovery of novel proteins and proteoforms. It has also improved our understanding of biological processes including post-transcriptional regulation of protein abundance and the presentation of antigens by major histocompatibility complex alleles. With proteomics data to be generated in hundreds to thousands of CCLs in coming years, there will be further potential for large-scale proteogenomics analyses and data integration with the phenotypically well-characterized CCLs.

Comparison of protein and peptide fractionation approaches in protein identification and quantification from Saccharomyces cerevisiae

Shotgun proteomics is a high-throughput technology which has been developed with the aim of investigating the maximum number of proteins in cells in a given experiment. However, protein discovery and data generation vary in depth and coverage when different technical strategies are selected. In this study, three different sample preparation approaches, and peptide or protein fractionation methods, were applied to identify and quantify proteins from log-phase yeast lysate: sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), filter-aided sample preparation coupled with gas phase fractionation (FASP-GPF), and FASP - high pH reversed phase fractionation (HpH). Fractions were initially analyzed and compared using nanoflow liquid chromatography - tandem mass spectrometry (nanoLC-MS/MS) employing data dependent acquisition on a linear ion trap instrument. The number of fractions and analytical replicates was adjusted so that each experiment used a similar amount of mass spectrometric instrument time. A second set of experiments was performed, comparing FASP-GPF, SDS-PAGE and FASP-HpH using a Q Exactive Orbitrap mass spectrometer. Compared with results from the linear ion trap mass spectrometer, the use of a Q Exactive Orbitrap mass spectrometer enabled a substantial increase in protein identifications, and an even greater increase in peptide identifications. This shows that the main advantage of using the higher resolution instrument is in increased proteome coverage. A total of 1035, 1357 and 2134 proteins were separately identified by FASP-GPF, SDS-PAGE and FASP-HpH. Combining results from the Orbitrap experiments, there were a total of 2269 proteins found, with 94% of them identified using the FASP-HpH method. Therefore, the FASP-HpH method is the optimal choice among these approaches, when applied to this type of sample.

Challenges in proteogenomics: a comparison of analysis methods with the case study of the DREAM proteogenomics sub-challenge

Background

Proteomic measurements, which closely reflect phenotypes, provide insights into gene expression regulations and mechanisms underlying altered phenotypes. Further, integration of data on proteome and transcriptome levels can validate gene signatures associated with a phenotype. However, proteomic data is not as abundant as genomic data, and it is thus beneficial to use genomic features to predict protein abundances when matching proteomic samples or measurements within samples are lacking.

Results

We evaluate and compare four data-driven models for prediction of proteomic data from mRNA measured in breast and ovarian cancers using the 2017 DREAM Proteogenomics Challenge data. Our results show that Bayesian network, random forests, LASSO, and fuzzy logic approaches can predict protein abundance levels with median ground truth-predicted correlation values between 0.2 and 0.5. However, the most accurately predicted proteins differ considerably between approaches.

Conclusions

In addition to benchmarking aforementioned machine learning approaches for predicting protein levels from transcript levels, we discuss challenges and potential solutions in state-of-the-art proteogenomic analyses.

Innovating the Concept and Practice of Two-Dimensional Gel Electrophoresis in the Analysis of Proteomes at the Proteoform Level

Two-dimensional gel electrophoresis (2DE) is an important and well-established technical platform enabling extensive top-down proteomic analysis. However, the long-held but now largely outdated conventional concepts of 2DE have clearly impacted its application to in-depth investigations of proteomes at the level of protein species/proteoforms. It is time to popularize a new concept of 2DE for proteomics. With the development and enrichment of the proteome concept, any given “protein” is now recognized to consist of a series of proteoforms. Thus, it is the proteoform, rather than the canonical protein, that is the basic unit of a proteome, and each proteoform has a specific isoelectric point (pI) and relative mass (M_r). Accordingly, using 2DE, each proteoform can routinely be resolved and arrayed according to its different pI and M_r. Each detectable spot contains multiple proteoforms derived from the same gene, as well as from different genes. Proteoforms derived from the same gene are distributed into different spots in a 2DE pattern. High-resolution 2DE is thus actually an initial level of separation to address proteome complexity and is effectively a pre-fractionation method prior to analysis using mass spectrometry (MS). Furthermore, stable isotope-labeled 2DE coupled with high-sensitivity liquid chromatography-tandem MS (LC-MS/MS) has tremendous potential for the large-scale detection, identification, and quantification of the proteoforms that constitute proteomes.

Reduced changes in protein compared to mRNA levels across non-proliferating tissues

BACKGROUND

The quantitative relations between RNA and protein are fundamental to biology and are still not fully understood. Across taxa, it was demonstrated that the protein-to-mRNA ratio in steady state varies in a direction that lessens the change in protein levels as a result of changes in the transcript abundance. Evidence for this behavior in tissues is sparse. We tested this phenomenon in new data that we produced for the mouse auditory system, and in previously published tissue datasets. A joint analysis of the transcriptome and proteome was performed across four datasets: inner-ear mouse tissues, mouse organ tissues, lymphoblastoid primate samples and human cancer cell lines.

RESULTS

We show that the protein levels are more conserved than the mRNA levels in all datasets, and that changes in transcription are associated with translational changes that exert opposite effects on the final protein level, in all tissues except cancer. Finally, we observe that some functions are enriched in the inner ear on the mRNA level but not in protein.

CONCLUSIONS

We suggest that partial buffering between transcription and translation ensures that proteins can be made rapidly in response to a stimulus. Accounting for the buffering can improve the prediction of protein levels from mRNA levels.

Structural hot spots for the solubility of globular proteins

Natural selection shapes protein solubility to physiological requirements and recombinant applications that require higher protein concentrations are often problematic. This raises the question whether the solubility of natural protein sequences can be improved. We here show an anti-correlation between the number of aggregation prone regions (APRs) in a protein sequence and its solubility, suggesting that mutational suppression of APRs provides a simple strategy to increase protein solubility. We show that mutations at specific positions within a protein structure can act as APR suppressors without affecting protein stability. These hot spots for protein solubility are both structure and sequence dependent but can be computationally predicted. We demonstrate this by reducing the aggregation of human α-galactosidase and protective antigen of Bacillus anthracis through mutation. Our results indicate that many proteins possess hot spots allowing to adapt protein solubility independently of structure and function.

Mutations in aggregation prone regions of recombinant proteins often improve their solubility, although they might cause negative effects on their structure and function. Here, the authors identify proteins hot spots that can be exploited to optimize solubility without compromising stability.

Direct and Absolute Quantification of over 1800 Yeast Proteins via Selected Reaction Monitoring

Defining intracellular protein concentration is critical in molecular systems biology. Although strategies for determining relative protein changes are available, defining robust absolute values in copies per cell has proven significantly more challenging. Here we present a reference data set quantifying over 1800 Saccharomyces cerevisiae proteins by direct means using protein-specific stable-isotope labeled internal standards and selected reaction monitoring (SRM) mass spectrometry, far exceeding any previous study. This was achieved by careful design of over 100 QconCAT recombinant proteins as standards, defining 1167 proteins in terms of copies per cell and upper limits on a further 668, with robust CVs routinely less than 20%. The selected reaction monitoring-derived proteome is compared with existing quantitative data sets, highlighting the disparities between methodologies. Coupled with a quantification of the transcriptome by RNA-seq taken from the same cells, these data support revised estimates of several fundamental molecular parameters: a total protein count of ∼100 million molecules-per-cell, a median of ∼1000 proteins-per-transcript, and a linear model of protein translation explaining 70% of the variance in translation rate. This work contributes a “gold-standard” reference yeast proteome (including 532 values based on high quality, dual peptide quantification) that can be widely used in systems models and for other comparative studies.

Qualitative and Quantitative Protein Complex Prediction Through Proteome-Wide Simulations

Despite recent progress in proteomics most protein complexes are still unknown. Identification of these complexes will help us understand cellular regulatory mechanisms and support development of new drugs. Therefore it is really important to establish detailed information about the composition and the abundance of protein complexes but existing algorithms can only give qualitative predictions. Herein, we propose a new approach based on stochastic simulations of protein complex formation that integrates multi-source data--such as protein abundances, domain-domain interactions and functional annotations--to predict alternative forms of protein complexes together with their abundances. This method, called SiComPre (Simulation based Complex Prediction), achieves better qualitative prediction of yeast and human protein complexes than existing methods and is the first to predict protein complex abundances. Furthermore, we show that SiComPre can be used to predict complexome changes upon drug treatment with the example of bortezomib. SiComPre is the first method to produce quantitative predictions on the abundance of molecular complexes while performing the best qualitative predictions. With new data on tissue specific protein complexes becoming available SiComPre will be able to predict qualitative and quantitative differences in the complexome in various tissue types and under various conditions.