Applying and improving AlphaFold at CASP14

We describe the operation and improvement of AlphaFold, the system that was entered by the team AlphaFold2 to the “human” category in the 14th Critical Assessment of Protein Structure Prediction (CASP14). The AlphaFold system entered in CASP14 is entirely different to the one entered in CASP13. It used a novel end‐to‐end deep neural network trained to produce protein structures from amino acid sequence, multiple sequence alignments, and homologous proteins. In the assessors' ranking by summed z scores (>2.0), AlphaFold scored 244.0 compared to 90.8 by the next best group. The predictions made by AlphaFold had a median domain GDT_TS of 92.4; this is the first time that this level of average accuracy has been achieved during CASP, especially on the more difficult Free Modeling targets, and represents a significant improvement in the state of the art in protein structure prediction. We reported how AlphaFold was run as a human team during CASP14 and improved such that it now achieves an equivalent level of performance without intervention, opening the door to highly accurate large‐scale structure prediction.

Highly accurate protein structure prediction for the human proteome

Protein structures can provide invaluable information, both for reasoning about biological processes and for enabling interventions such as structure-based drug development or targeted mutagenesis. After decades of effort, 17% of the total residues in human protein sequences are covered by an experimentally determined structure1. Here we markedly expand the structural coverage of the proteome by applying the state-of-the-art machine learning method, AlphaFold2, at a scale that covers almost the entire human proteome (98.5% of human proteins). The resulting dataset covers 58% of residues with a confident prediction, of which a subset (36% of all residues) have very high confidence. We introduce several metrics developed by building on the AlphaFold model and use them to interpret the dataset, identifying strong multi-domain predictions as well as regions that are likely to be disordered. Finally, we provide some case studies to illustrate how high-quality predictions could be used to generate biological hypotheses. We are making our predictions freely available to the community and anticipate that routine large-scale and high-accuracy structure prediction will become an important tool that will allow new questions to be addressed from a structural perspective.

Highly accurate protein structure prediction with AlphaFold

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1-4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence-the structure prediction component of the 'protein folding problem'8-has been an important open research problem for more than 50 years9. Despite recent progress10-14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm.

Association of cardiovascular risk factors with biological age

Funding Acknowledgements

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Academy of Medical Sciences of Ukraine

Background

Identifying reliable biomarkers of early ageing is an important goal for the prevention of cardiovascular disease (CVD).  Chronological age  (CA) is an essential risk factor for age-related disease including CVD. Biological age (BA) may be a more accurate metric for the risk of CVD and prediction of early vascular ageing, atherosclerosis. Available data allow us to judge the relationship of early ageing with all-cause mortality, healthspan and CVD. There is a hypothesis that BA can be used into models of risk prediction and stratification of CVD for personalized treatment.

Purpose

to determine the relationship between BA and factors of risk CVD in patients of different ages without clinical manifestations CVD.

Methods

We selected 102 patients aged 31–60 years with moderate cardiovascular risk without clinical symptoms of CVD. The patients did not receive regular drug therapy. Consenting subjects had a physical assessment, anthropometric measurements, electrocardiogram recording, blood sampling for laboratory analyses, including analysis of telomerase activity, telomere length in leukocytes and  buccal epithelium by a polymerase chain reaction in real-time. The SCORE scale was used for evaluation of the 10-year risk of fatal stroke and fatal myocardial infarction. In addition, 40 control individuals aged 31–60 years (20 men and 20 women), were also included in the study. BA was determined by 3 methods: PhenoAge, Voitenko’s method and Gorelkin-Pinhasov’s method. Multiple logistic regression analysis was used to develop a prediction of the early ageing model.

Results

The comparing patients of the control and main groups of the same CA shows, that patients with CVD risk have significantly higher BA, higher levels of proatherogenic lipids and shorter telomere length of the buccal epithelium. It was determined that body mass index, blood pressure, glucose levels are associated with an increase in BA in the main and the control groups. The increase of BA in the control group was associated with smoking, telomere length and telomerase activity and the level of antioxidant protection, in patients of the main group premature aging was associated with impaired lipid metabolism.

Conclusion

The biomarkers of biological aging can have benefits of the early identification of persons who age "faster" than others. The possibility of measuring biological aging can allow the assessment of health status at a time when there are still no symptoms, and interventions are more likely to be effective.

EFFECT THE CARDIOMETABOLIC RISK FACTORS ON VASCULAR AGING IN PATIENTS WITH NON-ALCOHOLIC FATTY LIVER DISEASE CONCOMITANT WITH SUBCLINICAL HYPOTHYROIDISM

The article presents the analysis of the relationship between thyroid function abnormality -subclinical hypothyroidism (SH) and non-alcoholic fatty liver disease (NAFLD), depending on age peculiarities (>50 years and <50 years), and the risk of cardiovascular complications in this category of patients. Research of early predictors of cardiovascular complications: dyslipidemia, insulin resistance, inflammatory marker- C-reactive protein, marker of vascular aging-telomerase activity and marker of endothelial dysfunction (ED) - CDECs and VEGF-A that have been analyzed are on the front burner. In this regard, the effect of the given values on the formation of cardiac risk in patients with NAFLD combined with SH was studied. 74 patients (29 men (39.2%) and 45 women (60.8%)), with verified NAFLD and SH have been examined. Patients were divided into two clinical groups: group 1 (n=31) - patients with NAFLD, with the mean age 47.2±2.6 years; group 2 (n=43) patients with NAFLD in combination with SH, with the mean age 56,8±6,5 years. Results of the performed tests have shown that patients with NAFLD combined with SH aged over 50 years have pro-atherogenic lipid profile and significantly more pronounced manifestations of endothelial dysfunction. The process of age-dependent shortening of telomere length predominantly in the buccal epithelium is an important point to be made. Consequently, the total effect of cardiometabolic risk factors in patients with NAFLD combined with SH probably is the determining factor of the rate of progression of vascular aging.

Meta-tetra(hydroxyphenyl)chlorin-sensitized photodynamic damage of cultured tumor and normal cells in the presence of high concentrations of alpha-tocopherol

Alpha-tocopherol at low concentrations protects biosubstrates from oxidative damage, while at high concentrations it may become toxic. Certain lines of tumor cells are reported to contain higher levels of vitamin E than normal cells. In our study alpha-tocopherol was successfully incorporated by cultured HT29 adenocarcinoma cells, but not by MRC-5 normal fibroblasts. At high concentrations (0.3-1 mM) alpha-tocopherol enhanced meta-tetra(hydroxyphenyl)chlorin (mTHPC)-sensitized photoinactivation of HT29 cells (415 nm), but not that of normal fibroblasts. At none of the concentrations used (0.001-1 mM) did alpha-tocopherol protect cells from photokilling, indicating that lipid peroxidation is of minor importance in mTHPC photoactivity. Our findings encourage the in vivo testing of phenolic antioxidants for selective enhancement of PDT-damage in tumors.

Characterization of photodegradation of meta-tetra(hydroxyphenyl)chlorin (mTHPC) in solution: biological consequences in human tumor cells

The photobleaching of meta-tetra(hydroxyphenyl)chlorin (mTHPC) (irradiation wavelength 413 nm) in protein-containing solution was evaluated by decay in absorbance in Soret band and in fluorescence (lambda exc = 423 nm, lambda em = 655 nm). Light exposure resulted in a decrease in absorption throughout the spectrum and simultaneous appearance of new absorption bands in the spectral region 325-450 nm. The rate of mTHPC photodegradation, followed by decay in absorbance, was 15-fold lower than that observed in fluorescence. This fact reflects the photobleaching of presumably monomeric, fluorescing species of mTHPC. In order to determine the consequences of photobleaching of fluorescing mTHPC material on cellular uptake and photocytotoxicity, human HT29 colon adenocarcinoma cells were incubated with photobleached mTHPC during 5 h with or without following irradiation with the fixed fluence. Surprisingly, but up to the time when the fluorescence decreased by 50%, only a slight decrease in photocytotoxicity was detected. Either aggregated forms that have been taken up undergo intracellular monomerization (but we did not observe increase in fluorescence in living cells) or the photodynamic activity is mostly due to aggregates. The discrepancy of mTHPC-photodynamic therapy (PDT) effect and fluorescence measurements may suggest that aggregated mTHPC plays an important role in mTHPC-PDT.

Spectroscopic and biological testing of photobleaching of porphyrins in solutions

The photobleaching of protoporhyrin IX (PP IX) and hematoporphyrin derivative (HpD) solutions was followed using three different methods: spectrophotometry, fluorometry and photodynamically induced cytotoxicity. The latter entails photoirradiation of HT29 human colon adenocarcinoma cells in the presence of preirradiated solutions of HpD and PP IX (lambda < or = 415 nm). The highest cytotoxicity was observed in the presence of unirradiated dye and decreased with the time of preirradiation. This decay in photocytotoxicity was further used to determine the porphyrin photobleaching kinetics in solution. For both sensitizers, quantum yields of photobleaching obtained by matching fluorescence were higher than that obtained from absorbance measurements (10 and 11 times for HpD and PP IX, respectively). This difference reflects preferential photobleaching of photolabile monomeric forms compared to aggregated. The highest quantum yield was obtained in the biological test (decay in cytotoxicity) which was 14 times higher for HpD and 30 times higher for PP IX than the quantum yield obtained from absorbance measurements. The absence of correlation between biological and fluorescence measurements has to be taken into account in the in vivo situation. Dark storage of preirradiated sensitizers (37 degrees C, 24 h) completely restored photocytotoxity for PP IX but only partially for HpD, whereas fluorescence patterns were partially restored for both sensitizers.