Conservation of kinetic stability, but not the unfolding mechanism, between human transthyretin and a transthyretin-related enzyme.

Kinetic stability is thought to be an attribute of proteins that require a long lifetime, such as the transporter of thyroxine and holo retinol-binding protein or transthyretin (TTR) functioning in the bloodstream, cerebrospinal fluid, and vitreous humor. TTR evolved from ancestral enzymes known as TTR-related proteins (TRPs). Here, we develop a rate-expansion approach that allows unfolding rates to be measured directly at low denaturant concentration, revealing that kinetic stability exists in the Escherichia coli TRP (EcTRP), even though the enzyme structure is more energetically frustrated and has a more mutation-sensitive folding mechanism than human TTR. Thus, the ancient tetrameric enzyme may already have been poised to mutate into a kinetically stable human transporter. An extensive mutational study that exchanges residues at key sites within the TTR and EcTRP dimer-dimer interface shows that tyrosine 111, replaced by a threonine in TTR, is the gatekeeper of frustration in EcTRP because it is critical for function. Frustration, virtually absent in TTR, occurs at multiple sites in EcTRP and even cooperatively for certain pairs of mutations. We present evidence that evolution at the C terminus of TTR was a compensatory event to maintain the preexisting kinetic stability while reducing frustration and sensitivity to mutation. We propose an "overcompensation" pathway from EcTRPs to functional hybrids to modern TTRs that is consistent with the biophysics discussed here. An alternative plausible pathway is also presented.

Comprehensive classification of TP53 somatic missense variants based on their impact on p53 structural stability.

Somatic variation is a major type of genetic variation contributing to human diseases including cancer. Of the vast quantities of somatic variants identified, the functional impact of many somatic variants, in particular the missense variants, remains unclear. Lack of the functional information prevents the translation of rich variation data into clinical applications. We previously developed a method named Ramachandran Plot-Molecular Dynamics Simulations (RP-MDS), aiming to predict the function of germline missense variants based on their effects on protein structure stability, and successfully applied to predict the deleteriousness of unclassified germline missense variants in multiple cancer genes. We hypothesized that regardless of their different genetic origins, somatic missense variants and germline missense variants could have similar effects on the stability of their affected protein structure. As such, the RP-MDS method designed for germline missense variants should also be applicable to predict the function of somatic missense variants. In the current study, we tested our hypothesis by using the somatic missense variants in TP53 as a model. Of the 397 somatic missense variants analyzed, RP-MDS predicted that 195 (49.1%) variants were deleterious as they significantly disturbed p53 structure. The results were largely validated by using a p53-p21 promoter-green fluorescent protein (GFP) reporter gene assay. Our study demonstrated that deleterious somatic missense variants can be identified by referring to their effects on protein structural stability.

Circadian period is compensated for repressor protein turnover rates in single cells.

Most mammalian cells have molecular circadian clocks that generate widespread rhythms in transcript and protein abundance. While circadian clocks are robust to fluctuations in the cellular environment, little is known about the mechanisms by which the circadian period compensates for fluctuating metabolic states. Here, we exploit the heterogeneity of single cells both in circadian period and a metabolic parameter-protein stability-to study their interdependence without the need for genetic manipulation. We generated cells expressing key circadian proteins (CRYPTOCHROME1/2 (CRY1/2) and PERIOD1/2 (PER1/2)) as endogenous fusions with fluorescent proteins and simultaneously monitored circadian rhythms and degradation in thousands of single cells. We found that the circadian period compensates for fluctuations in the turnover rates of circadian repressor proteins and uncovered possible mechanisms using a mathematical model. In addition, the stabilities of the repressor proteins are circadian phase dependent and correlate with the circadian period in a phase-dependent manner, in contrast to the prevailing model.

Systematic computer-aided disulfide design as a general strategy to stabilize prefusion class I fusion proteins.

Numerous enveloped viruses, such as coronaviruses, influenza, and respiratory syncytial virus (RSV), utilize class I fusion proteins for cell entry. During this process, the proteins transition from a prefusion to a postfusion state, undergoing substantial and irreversible conformational changes. The prefusion conformation has repeatedly shown significant potential in vaccine development. However, the instability of this state poses challenges for its practical application in vaccines. While non-native disulfides have been effective in maintaining the prefusion structure, identifying stabilizing disulfide bonds remains an intricate task. Here, we present a general computational approach to systematically identify prefusion-stabilizing disulfides. Our method assesses the geometric constraints of disulfide bonds and introduces a ranking system to estimate their potential in stabilizing the prefusion conformation. We hypothesized that disulfides restricting the initial stages of the conformational switch could offer higher stability to the prefusion state than those preventing unfolding at a later stage. The implementation of our algorithm on the RSV F protein led to the discovery of prefusion-stabilizing disulfides that supported our hypothesis. Furthermore, the evaluation of our top design as a vaccine candidate in a cotton rat model demonstrated robust protection against RSV infection, highlighting the potential of our approach for vaccine development.

Redesigning amino/carboxyl ends of DARPin G3 for high thermostability and production in tobacco transplastomic plants.

KEY MESSAGE: Redesigning the N- and C-capping repeats of the native DARPin G3 significantly improved its stability, and may facilitate its purification from the total soluble proteins of high-temperature dried leaf materials of transplastomic plants. Designed ankyrin repeat proteins (DARPins) constitute a promising class of binding molecules that can overcome the limitations of monoclonal antibodies and enable the development of novel therapeutic approaches. Despite their inherent stability, detailed studies have revealed that the original capping repeats derived from natural ankyrin repeat proteins impair the stability of the initial DARPin design. Consequently, the development of thermodynamically stabilized antibody mimetics may facilitate the development of innovative drugs in the future. In this study, we replaced the original N- and C-capping repeats with improved caps to enhance the thermostability of native DARPin G3. Computational analyses suggested that the redesigned thermostable DARPin G3 structure possessed optimal quality and stability. Molecular dynamics simulations verified the stability of the redesigned thermostable DARPin G3 at high temperatures. The redesigned thermostable DARPin G3 was expressed at high levels in tobacco transplastomic plants and subsequently purified from high-temperature dried leaf materials. Thermal denaturation results revealed that the redesigned thermostable DARPin G3 had a higher Tm value than the native DARPin G3, with a Tm of 35.51 °C greater than that of native DARPin G3. The results of the in vitro bioassays confirmed that the purified thermostable DARPin G3 from high-temperature dried leaf materials maintained its binding activity without any loss of affinity and specifically bound to the HER2 receptor on the cell surface. These findings demonstrate the successful improvement in the thermostability of DARPin G3 without compromising its biological activity.

p16-dependent increase of PD-L1 stability regulates immunosurveillance of senescent cells.

The accumulation of senescent cells promotes ageing and age-related diseases, but molecular mechanisms that senescent cells use to evade immune clearance and accumulate in tissues remain to be elucidated. Here we report that p16-positive senescent cells upregulate the immune checkpoint protein programmed death-ligand 1 (PD-L1) to accumulate in ageing and chronic inflammation. We show that p16-mediated inhibition of cell cycle kinases CDK4/6 induces PD-L1 stability in senescent cells via downregulation of its ubiquitin-dependent degradation. p16-expressing senescent alveolar macrophages elevate PD-L1 to promote an immunosuppressive environment that can contribute to an increased burden of senescent cells. Treatment with activating anti-PD-L1 antibodies engaging Fcγ receptors on effector cells leads to the elimination of PD-L1 and p16-positive cells. Our study uncovers a molecular mechanism of p16-dependent regulation of PD-L1 protein stability in senescent cells and reveals the potential of targeting PD-L1 to improve immunosurveillance of senescent cells and ameliorate senescence-associated inflammation.

PON-Tm: A Sequence-Based Method for Prediction of Missense Mutation Effects on Protein Thermal Stability Changes.

Proteins, as crucial macromolecules performing diverse biological roles, are central to numerous biological processes. The ability to predict changes in protein thermal stability due to mutations is vital for both biomedical research and industrial applications. However, existing experimental methods are often costly and labor-intensive, while structure-based prediction methods demand significant computational resources. In this study, we introduce PON-Tm, a novel sequence-based method for predicting mutation-induced thermal stability variations in proteins. PON-Tm not only incorporates features predicted by a protein language model from protein sequences but also considers environmental factors such as pH and the thermostability of the wild-type protein. To evaluate the effectiveness of PON-Tm, we compared its performance to four well-established methods, and PON-Tm exhibited superior predictive capabilities. Furthermore, to facilitate easy access and utilization, we have developed a web server.

Ubiquitin-specific protease UBP14 stabilizes HY5 by deubiquitination to promote photomorphogenesis in Arabidopsis thaliana.

Transcription factor ELONGATED HYPOCOTYL5 (HY5) is the central hub for seedling photomorphogenesis. E3 ubiquitin (Ub) ligase CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) inhibits HY5 protein accumulation through ubiquitination. However, the process of HY5 deubiquitination, which antagonizes E3 ligase-mediated ubiquitination to maintain HY5 homeostasis has never been studied. Here, we identified that Arabidopsis thaliana deubiquitinating enzyme, Ub-SPECIFIC PROTEASE 14 (UBP14) physically interacts with HY5 and enhances its protein stability by deubiquitination. The da3-1 mutant lacking UBP14 function exhibited a long hypocotyl phenotype, and UBP14 deficiency led to the failure of rapid accumulation of HY5 during dark to light. In addition, UBP14 preferred to stabilize nonphosphorylated form of HY5 which is more readily bound to downstream target genes. HY5 promoted the expression and protein accumulation of UBP14 for positive feedback to facilitate photomorphogenesis. Our findings thus established a mechanism by which UBP14 stabilizes HY5 protein by deubiquitination to promote photomorphogenesis in A. thaliana.

Biochemical characterization of naturally occurring mutations in SARS-CoV-2 RNA-dependent RNA polymerase.

Since the emergence of SARS-CoV-2, mutations in all subunits of the RNA-dependent RNA polymerase (RdRp) of the virus have been repeatedly reported. Although RdRp represents a primary target for antiviral drugs, experimental studies exploring the phenotypic effect of these mutations have been limited. This study focuses on the phenotypic effects of substitutions in the three RdRp subunits: nsp7, nsp8, and nsp12, selected based on their occurrence rate and potential impact. We employed nano-differential scanning fluorimetry and microscale thermophoresis to examine the impact of these mutations on protein stability and RdRp complex assembly. We observed diverse impacts; notably, a single mutation in nsp8 significantly increased its stability as evidenced by a 13°C increase in melting temperature, whereas certain mutations in nsp7 and nsp8 reduced their binding affinity to nsp12 during RdRp complex formation. Using a fluorometric enzymatic assay, we assessed the overall effect on RNA polymerase activity. We found that most of the examined mutations altered the polymerase activity, often as a direct result of changes in stability or affinity to the other components of the RdRp complex. Intriguingly, a combination of nsp8 A21V and nsp12 P323L mutations resulted in a 50% increase in polymerase activity. To our knowledge, this is the first biochemical study to demonstrate the impact of amino acid mutations across all components constituting the RdRp complex in emerging SARS-CoV-2 subvariants.

Stable and reusable calcium-responsive biopolymer for affinity precipitation of therapeutic antibodies.

Affinity precipitation is an attractive method for protein purification due to its many advantages, including the rapid capture of target proteins, simple processing, high specificity, and ease of scale-up. We previously reported a robust antibody purification method using Ca2+-dependent precipitation of ZZ-hCSQ2, a fusion protein of human calsequestrin 2, and the antibody-binding protein ZZ. However, the stability of this fusion protein was not sufficiently high for industrial use because the antibody recovery yield decreased to 60% after being reused 10 times. To identify a more stable calsequestrin (CSQ), we calculated Rosetta energy values for the folding stabilities of various CSQ homologs and selected human CSQ1 (hCSQ1) with lowest energy value (-992.6) as the new CSQ platform. We also identified that the linker sequence between ZZ and CSQ was vulnerable to proteases and alkaline pH by N-terminal protein sequencing. Therefore, we changed the linker to four asparagine (4N) sequences, which were shorter and less flexible than the previous glycine-rich linker. The new version of ZZ-CSQ, ZZ-4N-hCSQ1, was stable in a protease-containing conditioned medium obtained from the cultured Chinese hamster ovary cell or high pH condition (0.1M sodium hydroxide) for more than 5 days and could be reused at least 25 times for antibody purification without loss of recovery yield. The antibodies purified by ZZ-4N-hCSQ1 precipitation also showed greater purity (~33.6-fold lower host cell DNA and ~6.4-fold lower host cell protein) than those purified by protein A chromatography. These data suggest that ZZ-4N-hCSQ1 precipitation is more efficient and can achieve cost-effectiveness of up to 12.5-fold cheaper than previous antibody purification methods and can lower the production costs of therapeutic antibodies.

Molecular dynamics simulations reveal differences in the conformational stability of FtsZs derived from Staphylococcus aureus and Bacillus subtilis.

FtsZ is highly conserved among bacteria and plays an essential role in bacterial cell division. The tense conformation of FtsZ bound to GTP assembles into a straight filament via head-to-tail associations, and then the upper subunit of FtsZ hydrolyzes GTP bound to the lower FtsZ subunit. The subunit with GDP bound disassembles accompanied by a conformational change in the subunit from the tense to relaxed conformation. Although crystal structures of FtsZ derived from several bacterial species have been determined, the conformational change from the relaxed to tense conformation has only been observed in Staphylococcus aureus FtsZ (SaFtsZ). Recent cryo-electron microscopy analyses revealed the three-dimensional reconstruction of the protofilament, in which tense molecules assemble via head-to-tail associations. However, the lower resolution of the protofilament suggested that the flexibility of the FtsZ protomers between the relaxed and tense conformations caused them to form in less-strict alignments. Furthermore, this flexibility may also prevent FtsZs other than SaFtsZ from crystalizing in the tense conformation, suggesting that the flexibility of bacterial FtsZs differs. In this study, molecular dynamics simulations were performed using SaFtsZ and Bacillus subtilis FtsZ in several situations, which suggested that different features of the FtsZs affect their conformational stability.

The ubiquitin-like protein UBTD1 promotes colorectal cancer progression by stabilizing c-Myc to upregulate glycolysis.

Dysfunction of the ubiquitin-proteasome system (UPS) is involved in the pathogenesis of various malignancies including colorectal cancer (CRC). Ubiquitin domain containing 1 (UBTD1), a ubiquitin-like protein, regulates UPS-mediated protein degradation and tumor progression in some cancer types. However, the biological function and mechanism of UBTD1 are far from being well elucidated, and its role in CRC has not been explored yet. In our study, we analyzed CRC patients' clinical information and UBTD1 expression data, and found that the expression of UBTD1 in cancer tissue was significantly higher than that in adjacent normal tissue. Higher UBTD1 expression was significantly associated with poorer survival and more lymph node metastasis. Overexpression of UBTD1 could facilitate, while knockdown could inhibit CRC cell proliferation and migration, respectively. RNA-seq and proteomics indicated that c-Myc is an important downstream target of UBTD1. Metabolomics showed the products of the glycolysis pathway were significantly increased in UBTD1 overexpression cells. In vitro, we verified UBTD1 upregulating c-Myc protein and promoting CRC cell proliferation and migration via regulating c-Myc. UBTD1 promoted CRC cells' glycolysis, evidenced by the increased lactate production and glucose uptake following UBTD1 overexpression. Mechanistically, UBTD1 prolonged the half-life of the c-Myc protein by binding to E3 ligase ß-transducin repeat-containing protein (ß-TrCP), thereby upregulated the expression of glycolysis rate-limiting enzyme hexokinase II (HK2), and enhanced glycolysis and promoted CRC progression. In conclusion, our study revealed that UBTD1 promotes CRC progression by upregulating glycolysis via the ß-TrCP/c-Myc/HK2 pathway, suggesting its potential as a prognostic biomarker and therapeutic target in CRC.

Co-chaperonin GroES subunit exchange as dependent on time, pH, protein concentration, and urea.

The discovery of a subunit exchange in some oligomeric proteins, implying short-term dissociation of their oligomeric structure, requires new insights into the role of the quaternary structure in oligomeric protein stability and function. Here we demonstrate the effect of pH, protein concentration, and urea on the efficiency of GroES heptamer (GroES7) subunit exchange. A mixture of equimolar amounts of wild-type (WT) GroES7 and its Ala97Cys mutant modified with iodoacetic acid (97-carboxymethyl cysteine or CMC-GroES7) was incubated in various conditions and subjected to isoelectric focusing (IEF) in polyacrylamide gel. For each sample, there are eight Coomassie-stained electrophoretic bands showing different charges that result from a different number of included mutant subunits, each carrying an additional negative charge. The intensities of these bands serve to analyze the protein subunit exchange. The protein stability is evaluated using the transverse urea gradient gel electrophoresis (TUGGE). At pH 8.0, the intensities of the initial bands corresponding to WT-GroES7 and CMC-GroES7 are decreased with a half-time of (23 ± 2) min. The exchange decreases with decreasing pH and seems to be strongly hindered at pH 5.2 due to the protonation of groups with pK âˆ¼ 6.3, which stabilizes the protein quaternary structure. The destabilization of the protein quaternary structure caused by increased pH, decreased protein concentration, or urea accelerates the GroES subunit exchange. This study allows visualizing the subunit exchange in oligomeric proteins and confirms its direct connection with the stability of the protein quaternary structure.

Impact of Missense Mutations on Spike Protein Stability and Binding Affinity in the Omicron Variant.

The global effort to combat the COVID-19 pandemic faces ongoing uncertainty with the emergence of Variants of Concern featuring numerous mutations on the Spike (S) protein. In particular, the Omicron Variant is distinguished by 32 mutations, including 10 within its receptor-binding domain (RBD). These mutations significantly impact viral infectivity and the efficacy of vaccines and antibodies currently in use for therapeutic purposes. In our study, we employed structure-based computational saturation mutagenesis approaches to predict the effects of Omicron missense mutations on RBD stability and binding affinity, comparing them to the original Wuhan-Hu-1 strain. Our results predict that mutations such as G431W and P507W induce the most substantial destabilizations in the Wuhan-Hu-1-S/Omicron-S RBD. Notably, we postulate that mutations in the Omicron-S exhibit a higher percentage of enhancing binding affinity compared to Wuhan-S. We found that the mutations at residue positions G447, Y449, F456, F486, and S496 led to significant changes in binding affinity. In summary, our findings may shed light on the widespread prevalence of Omicron mutations in human populations. The Omicron mutations that potentially enhance their affinity for human receptors may facilitate increased viral binding and internalization in infected cells, thereby enhancing infectivity. This informs the development of new neutralizing antibodies capable of targeting Omicron's immune-evading mutations, potentially aiding in the ongoing battle against the COVID-19 pandemic.

Protein Thermal Stability Changes Induced by the Global Methylation Inhibitor 3-Deazaneplanocin A (DZNep).

DZNep (3-deazaneplanocin A) is commonly used to reduce lysine methylation. DZNep inhibits S-adenosyl-l-homocysteine hydrolase (AHCY), preventing the conversion of S-adenosyl-l-homocysteine (SAH) into L-homocysteine. As a result, the SAM-to-SAH ratio decreases, an indicator of the methylation potential within a cell. Many studies have characterized the impact of DZNep on histone lysine methylation or in specific cell or disease contexts, but there has yet to be a study looking at the potential downstream impact of DZNep treatment on proteins other than histones. Recently, protein thermal stability has provided a new dimension for studying the mechanism of action of small-molecule inhibitors. In addition to ligand binding, post-translational modifications and protein-protein interactions impact thermal stability. Here, we sought to characterize the protein thermal stability changes induced by DZNep treatment in HEK293T cells using the Protein Integral Solubility Alteration (PISA) assay. DZNep treatment altered the thermal stability of 135 proteins, with over half previously reported to be methylated at lysine residues. In addition to thermal stability, we identify changes in transcript and protein abundance after DZNep treatment to distinguish between direct and indirect impacts on thermal stability. Nearly one-third of the proteins with altered thermal stability had no changes at the transcript or protein level. Of these thermally altered proteins, CDK6 had a stabilized methylated peptide, while its unmethylated counterpart was unaltered. Multiple methyltransferases were among the proteins with thermal stability alteration, including DNMT1, potentially due to changes in the SAM/SAH levels. This study systematically evaluates DZNep's impact on the transcriptome, the proteome, and the thermal stability of proteins.

The regulation of the thermal stability and affinity of the HSPA5 (Grp78/BiP) by clients and nucleotides is modulated by domains coupling.

The HSPA5 protein (BiP/Grp78) serves as a pivotal chaperone in maintaining cellular protein quality control. As a member of the human HSP70 family, HSPA5 comprises two distinct domains: a nucleotide-binding domain (NBD) and a peptide-binding domain (PBD). In this study, we investigated the interdomain interactions of HSPA5, aiming to elucidate how these domains regulate its function as a chaperone. Our findings revealed that HSPA5-FL, HSPA5-T, and HSPA5-N exhibit varying affinities for ATP and ADP, with a noticeable dependency on Mg2+ for optimal interactions. Interestingly, in ADP assays, the presence of the metal ion seems to enhance NBD binding only for HSPA5-FL and HSPA5-T. Moreover, while the truncation of the C-terminus does not significantly impact the thermal stability of HSPA5, experiments involving MgATP underscore its essential role in mediating interactions and nucleotide hydrolysis. Thermal stability assays further suggested that the NBD-PBD interface enhances the stability of the NBD, more pronounced for HSPA5 than for the orthologous HSPA1A, and prevents self-aggregation through interdomain coupling. Enzymatic analyses indicated that the presence of PBD enhances NBD ATPase activity and augments its nucleotide affinity. Notably, the intrinsic chaperone activity of the PBD is dependent on the presence of the NBD, potentially due to the propensity of the PBD for self-oligomerization. Collectively, our data highlight the pivotal role of allosteric mechanisms in modulating thermal stability, nucleotide interaction, and ATPase activity of HSPA5, underscoring its significance in protein quality control within cellular environments.

Relationship between protein conformational stability and its immunogenicity when administering antigens to mice using adjuvants-Analysis employed the CH2 domain in human antibodies.

Antigen-presenting cells (APCs) play a crucial role in the immune system by breaking down antigens into peptide fragments that subsequently bind to major histocompatibility complex (MHC) molecules. Previous studies indicate that stable proteins can impede CD4+ T cell stimulation by hindering antigen processing and presentation. Conversely, certain proteins require stabilization in order to activate the immune response. Several factors, including the characteristics of the protein and the utilization of different adjuvants in animal experiments, may contribute to this disparity. In this study, we investigated the impact of adjuvants on antigen administration in mice, specifically focusing on the stability of the CH2 domain. Consequently, the CH2 domain induced a stronger IgG response in comparison to the stabilized one when using Alum and PBS (without adjuvant). On the other hand, animal experiment using Freund's adjuvant showed the opposite results. These findings indicate the significance of considering the intrinsic conformational stability of a protein when eliciting its immunogenicity, particularly within the context of vaccine development.

PRMT6 facilitates EZH2 protein stability by inhibiting TRAF6-mediated ubiquitination degradation to promote glioblastoma cell invasion and migration.

Invasion and migration are the key hallmarks of cancer, and aggressive growth is a major factor contributing to treatment failure and poor prognosis in glioblastoma. Protein arginine methyltransferase 6 (PRMT6), as an epigenetic regulator, has been confirmed to promote the malignant proliferation of glioblastoma cells in previous studies. However, the effects of PRMT6 on glioblastoma cell invasion and migration and its underlying mechanisms remain elusive. Here, we report that PRMT6 functions as a driver element for tumor cell invasion and migration in glioblastoma. Bioinformatics analysis and glioma sample detection results demonstrated that PRMT6 is highly expressed in mesenchymal subtype or invasive gliomas, and is significantly negatively correlated with their prognosis. Inhibition of PRMT6 (using PRMT6 shRNA or inhibitor EPZ020411) reduces glioblastoma cell invasion and migration in vitro, whereas overexpression of PRMT6 produces opposite effects. Then, we identified that PRMT6 maintains the protein stability of EZH2 by inhibiting the degradation of EZH2 protein, thereby mediating the invasion and migration of glioblastoma cells. Further mechanistic investigations found that PRMT6 inhibits the transcription of TRAF6 by activating the histone methylation mark (H3R2me2a), and reducing the interaction between TRAF6 and EZH2 to enhance the protein stability of EZH2 in glioblastoma cells. Xenograft tumor assay and HE staining results showed that the expression of PRMT6 could promote the invasion of glioblastoma cells in vivo, the immunohistochemical staining results of mouse brain tissue tumor sections also confirmed the regulatory relationship between PRMT6, TRAF6, and EZH2. Our findings illustrate that PRMT6 suppresses TRAF6 transcription via H3R2me2a to enhance the protein stability of EZH2 to facilitate glioblastoma cell invasion and migration. Blocking the PRMT6-TRAF6-EZH2 axis is a promising strategy for inhibiting glioblastoma cell invasion and migration.

Stability Oracle: a structure-based graph-transformer framework for identifying stabilizing mutations.

Engineering stabilized proteins is a fundamental challenge in the development of industrial and pharmaceutical biotechnologies. We present Stability Oracle: a structure-based graph-transformer framework that achieves SOTA performance on accurately identifying thermodynamically stabilizing mutations. Our framework introduces several innovations to overcome well-known challenges in data scarcity and bias, generalization, and computation time, such as: Thermodynamic Permutations for data augmentation, structural amino acid embeddings to model a mutation with a single structure, a protein structure-specific attention-bias mechanism that makes transformers a viable alternative to graph neural networks. We provide training/test splits that mitigate data leakage and ensure proper model evaluation. Furthermore, to examine our data engineering contributions, we fine-tune ESM2 representations (Prostata-IFML) and achieve SOTA for sequence-based models. Notably, Stability Oracle outperforms Prostata-IFML even though it was pretrained on 2000X less proteins and has 548X less parameters. Our framework establishes a path for fine-tuning structure-based transformers to virtually any phenotype, a necessary task for accelerating the development of protein-based biotechnologies.