Facilitate Drug Discovery Research Articles

Chemical Probes Review: Choosing the Right Path Towards Pharmacological Targets in Drug Discovery, Challenges and Future Perspectives.

Chemical probes are essential for academic research and target validation for disease identification. They facilitate drug discovery, target function investigation, and translation studies. A chemical probe provides starting material that can accelerate therapeutic values and safety measures for identifying any biological target in drug discovery. Essential read outs depend on their versatility in biochemical testing, proving the hypothesis, selectivity, specificity, affinity towards the target site, and valuable in new therapeutic approaches. Disease management will depend upon chemical probes as a primitive tool to ascertain the physicochemical stability for in vivo and in vitro studies useful for clinical trials and industrial application in the future. For cancer research, bacterial infection, and neurodegenerative disorders, chemical probes are integrated circuits which are on pipeline for the drug discovery process Furthermore, pharmacological modulators incorporate activators, crosslinkers, degraders, and inhibitors. Reports accessed depend on their structural, mechanical, biochemical, and pharmacological characterization in drug discovery research. The perspective for designing any chemical probes concludes with the utilization of drug discovery and identification of the potential target. It focuses mainly on evidence-based studies and produces promising results in successfully delivering novel therapeutics to treat cancers and other disorders at the target site. Moreover, natural product pharmacophores like rapamycin, cephalosporin, and α-lactamase are utilized for drug discovery. Chemical probes revolutionize computational-based study design depending on identifying novel targets within the database framework. Chemical probes are the clinical answers for drug development and goforward tools in solving other riddles for scientists and researchers working in this industries.

Multi-omics Data Integration Analysis Identified Therapeutic Targets and Potential Reuse Drugs for Osteoporosis.

To facilitate drug discovery and development for the treatment of osteoporosis. With global aging, osteoporosis has become a common problem threatening the health of the elderly. It is of important clinical value to explore new targets for drug intervention and develop promising drugs for the treatment of osteoporosis. To understand the major molecules that mediate the communication between the cell populations of bone marrow-derived mesenchymal stem cells (BM-MSCs) in osteoporosis and osteoarthritis patients and identify potential reusable drugs for the treatment of osteoporosis. Single-cell RNA sequencing (scRNA-seq) data of BM-MSCs in GSE147287 dataset were classified using the Seurat package. CellChat was devoted to analyzing the ligand-receptor pairs (LR pairs) contributing to the communication between BM-MSCs subsets. The LR pairs that were differentially expressed between osteoporosis samples and control samples and significantly correlated with immune score were screened in the GSE35959 dataset, and the differentially expressed gene in both GSE35959 and GSE13850 data sets were identified as targets from a single ligand or receptor. The therapeutic drugs for osteoporosis were screened by network proximity method, and the top-ranked drugs were selected for molecular docking and molecular dynamics simulation with the target targets. Twelve subsets of BM-MSCs were identified, of which CD45-BM-MSCS_4, CD45-BM- MSCS_5, and CD45+ BM-MSCs_5 subsets showed significantly different distributions between osteoporosis samples and osteoarthritis samples. Six LR pairs were identified in the bidirectional communication between these three BM-MSCs subsets and other BM-MSCs subsets. Among them, MIF-CD74 and ITGB2-ICAM2 were significantly correlated with the immune score. CD74 was identified as the target, and a total of 48 drugs targeting CD47 protein were identified. Among them, DB01940 had the lowest free energy binding score with CD74 protein and the binding state was very stable. This study provided a new network-based framework for drug reuse and identified initial insights into therapeutic agents targeting CD74 in osteoporosis, which may be meaningful for promoting the development of osteoporosis treatment.

Development of a humanized mouse model with functional human materno-fetal interface immunity.

Materno-fetal immunity possesses specialized characteristics to ensure pathogen clearance while maintaining tolerance to the semiallogeneic fetus. Most of our understanding on human materno-fetal immunity is based on conventional rodent models that may not precisely represent human immunological processes owing to the huge evolutionary divergence. Herein, we developed a pregnant human immune system (HIS) mouse model through busulfan preconditioning, which hosts multilineage human immune subset reconstitution at the materno-fetal interface. Human materno-fetal immunity exhibits a tolerogenic feature at the midgestation stage (embryonic day [E] 14.5), and human immune regulatory subsets were detected in the decidua. However, the immune system switches to an inflammatory profile at the late gestation stage (E19). A cell-cell interaction network contributing to the alternations in the human materno-fetal immune atmosphere was revealed based on single-cell RNA-Seq analysis, wherein human macrophages played crucial roles by secreting several immune regulatory mediators. Furthermore, depletion of Treg cells at E2.5 and E5.5 resulted in severe inflammation and fetus rejection. Collectively, these results demonstrate that the pregnant HIS mouse model permits the development of functional human materno-fetal immunity and offers a tool for human materno-fetal immunity investigation to facilitate drug discovery for reproductive disorders.

Improved Ensemble Model for Insecticide Recognition by Incorporating Insect Toxicity Data.

Pesticide molecules, such as insecticides, play a critical role in modern agricultural production. Traditional pesticide development methods are often inefficient and expensive, while data-driven artificial intelligence (AI) techniques have emerged as a useful tool to facilitate drug discovery. However, currently available commercial pesticide data is limited, which makes the trained models unsatisfactory in terms of performance and generalization. From a domain knowledge perspective, insect toxicity data were incorporated to improve the insecticide recognition of AI models. Compared to the models trained with the original data set, the new models performed better in the external validation, and their generalization was more desirable. In addition, by integrating different types of individual models, we obtained an ensemble model with better performance. Based on this, an online platform was developed to provide researchers with free access to insecticide screening (https://dpai.ccnu.edu.cn/InsectiVS/). Finally, two potential insecticide molecules with insecticidal activity against Plutella xylostella were successfully identified in a real-world scenario. In conclusion, this idea connects the fields of AI and agricultural chemistry and is expected to have wide application in pesticide research.

Generation of IPi002-A/B/C human induced pluripotent stem cell lines from MARCH amniotic fluid cells

Human induced pluripotent stem cells (hiPSCs) have become a revolutionary tool in biomedical research due to their unique in vitro properties and fate versatility. They offer insights into development or genetic disorders, facilitate drug discovery and hold promise for regenerative medicine. Here we generated three hiPSC cells – IPi002-A/B/C – from primary amniotic fluid cells (AFCs) obtained via amniocentesis for the prenatal diagnosis of MARCH syndrome: Multinucleated neurons, Anhydramnios, Renal dysplasia, Cerebellar hypoplasia, and Hydranencephaly. These AFCs underwent reprogramming through non-integrative viral transduction and the resulting hiPSCs exhibited normal karyotype and expressed typical pluripotency markers.

CHIMERA_NA: A Customizable Mutagenesis Tool for Structural Manipulations in Nucleic Acids and Their Complexes.

Studying the structure and dynamics of nucleic acids and their complexes is crucial for understanding fundamental biological processes and developing therapeutic interventions. However, the limited availability of experimentally characterized nucleic acid structures poses a challenge for exploring their properties comprehensively. To address this, we developed a customizable mutagenesis tool, CHIMERA_NA, to manipulate nucleic acid structures and their complexes. Utilizing the user-friendly CHIMERA_NA, researchers can perform mutations in nucleic acid structures, enabling the exploration of diverse structural configurations and dynamic behaviors. The tool offers the flexibility to generate all possible combinations of mutations or specific user-defined mutations based on research requirements. CHIMERA_NA leverages the capabilities of UCSF Chimera software, a widely used platform for molecular structure analysis, to facilitate the generation of mutations in nucleic acids. Our tool modifies the reference structure of nucleic acids or their complexes to generate initial coordinates of mutated structures/complexes within seconds for further computational exploration. This capability allows users to extend their investigations beyond structural repositories, enabling the study of DNA/RNA drug recognition, nucleic acid-protein interactions, and the intrinsic structural and dynamic properties of nucleic acids. By providing a user-friendly and customizable approach to nucleic acid mutagenesis, CHIMERA_NA contributes to advancing our understanding of nucleic acid biology and facilitating drug discovery efforts targeting nucleic acid-based mechanisms. CHIMERA_NA is freely available in the Supporting Information of this article.

DrugMAP 2.0: molecular atlas and pharma-information of all drugs.

The escalating costs and high failure rates have decelerated the pace of drug development, which amplifies the research interests in developing combinatorial/repurposed drugs and understanding off-target adverse drug reaction (ADR). In other words, it is demanded to delineate the molecular atlas and pharma-information for the combinatorial/repurposed drugs and off-target interactions. However, such invaluable data were inadequately covered by existing databases. In this study, a major update was thus conducted to the DrugMAP, which accumulated (a) 20831 combinatorial drugs and their interacting atlas involving 1583 pharmacologically important molecules; (b) 842 repurposed drugs and their interacting atlas with 795 molecules; (c) 3260 off-targets relevant to the ADRs of 2731 drugs and (d) various types of pharmaceutical information, including diverse ADMET properties, versatile diseases, and various ADRs/off-targets. With the growing demands for discovering combinatorial/repurposed therapies and the rapidly emerging interest in AI-based drug discovery, DrugMAP was highly expected to act as an indispensable supplement to existing databases facilitating drug discovery, which was accessible at: https://idrblab.org/drugmap/.

Predicting Small Molecule Binding Nucleotides in RNA Structures Using RNA Surface Topography.

RNA small molecule interactions play a crucial role in drug discovery and inhibitor design. Identifying RNA small molecule binding nucleotides is essential and requires methods that exhibit a high predictive ability to facilitate drug discovery and inhibitor design. Existing methods can predict the binding nucleotides of simple RNA structures, but it is hard to predict binding nucleotides in complex RNA structures with junctions. To address this limitation, we developed a new deep learning model based on spatial correlation, ZHmolReSTasite, which can accurately predict binding nucleotides of small and large RNA with junctions. We utilize RNA surface topography to consider the spatial correlation, characterizing nucleotides from sequence and tertiary structures to learn a high-level representation. Our method outperforms existing methods for benchmark test sets composed of simple RNA structures, achieving precision values of 72.9% on TE18 and 76.7% on RB9 test sets. For a challenging test set composed of RNA structures with junctions, our method outperforms the second best method by 11.6% in precision. Moreover, ZHmolReSTasite demonstrates robustness regarding the predicted RNA structures. In summary, ZHmolReSTasite successfully incorporates spatial correlation, outperforms previous methods on small and large RNA structures using RNA surface topography, and can provide valuable insights into RNA small molecule prediction and accelerate RNA inhibitor design.

BACE1 inhibitors: A promising therapeutic approach for the management of Alzheimer’s disease

Alzheimer’s disease is a neurological disorder marked by the accumulation of amyloid beta (Aβ) aggregates, resulting from mutations in the amyloid precursor protein. The enzyme β-secretase, also known as β-site amyloid precursor protein cleaving enzyme 1 (BACE1), plays a crucial role in generating Aβ peptides. With no targeted therapy available for Alzheimer’s disease, inhibiting BACE1 aspartic protease has emerged as a primary treatment target. Since 1999, compounds demonstrating potential binding to the BACE1 receptor have advanced to human trials. Structural optimization of synthetically derived compounds, coupled with computational approaches, has offered valuable insights for developing highly selective leads with drug-like properties. This review highlights pivotal studies on the design and development of BACE1 inhibitors as anti-Alzheimer’s disease agents. It summarizes computational methods employed in facilitating drug discovery for potential BACE1 inhibitors and provides an update on their clinical status, indicating future directions for novel BACE1 inhibitors. The promising clinical results of Elenbecestat (E-2609) catalyze the development of effective, selective BACE1 inhibitors in the future.

Allostery Illuminated: Harnessing AI and Machine Learning for Drug Discovery.

In the past several years there has been rapid adoption of artificial intelligence (AI) and machine learning (ML) tools for drug discovery. In this Microperspective, we comment on recent AI/ML applications to the discovery of allosteric modulators, focusing on breakthroughs with AlphaFold, structure-based drug discovery (SBDD), and medicinal chemistry applications. We discuss how these technologies are facilitating drug discovery and the remaining challenges to identify allosteric binding sites and ligands.

How drug repurposing can advance drug discovery: challenges and opportunities

Traditional de novo drug discovery, which typically presents an 11% approval rate from phase I trials and even higher failure rates in fields like neurodegeneration, often requires two to three billion dollars and 10–17 years per new drug. In contrast, drug repurposing can reduce risks and bring drugs to the market in 3–12 years, with an average of $300 million investment. In this article, we will outline how drug repurposing can accelerate the discovery of drugs derived from natural and synthetic products. The vast amount of chemical, biological, structural, and clinical data available in public repositories will greatly facilitate drug discovery, without the need to start a discovery campaign from scratch. In the big data era, data mining and artificial intelligence will play major roles in both drug repurposing and drug discovery. This article will provide valuable insights into how drug repurposing can support drug discovery and vice versa, emphasizing its impact in addressing unmet medical needs, achieving cost-effectiveness, and enabling faster market access. Despite legal and regulatory challenges, the cost-effectiveness, and the potential to give new life to compounds already in the pipeline make drug repurposing a crucial complement to traditional drug discovery in the era of precision medicine.

Are we ready to integrate 3D culture systems in acute myeloid leukemia and bone marrow microenvironment research?

Acute myeloid leukemia (AML) is the most aggressive adult leukemia and results in a dismal 5-year survival rate of less than 30%. While research has primarily focused on identifying intrinsic mutations driving leukemogenesis, the role of the bone marrow microenvironment (BMME) in disease progression remains poorly understood. For this purpose, conventional 2D cultures inadequately replicate the complex BMME interactions crucial for the maintenance of normal hematopoiesis and leukemia pathogenesis. In recent years, 3D cultures or microphysiological systems (MPS), have emerged as promising tools for in vitro modeling of the human BMME. These approaches provide a promise for a more physiologically relevant platform for investigating the mechanistic underpinnings of AML interactions with BMME components, as well as exploring chemoresistance mechanisms and facilitating drug discovery efforts. This review discusses the considerations in biomaterials, biophysical, and biochemical factors to develop the BMME in vitro for AML studies, the state-of-the-art 3D models of the BMME, and the challenges and prospects of adopting MPS for AML research.

Investigation of the synergistic effect mechanism underlying sequential use of palbociclib and cisplatin through integral proteomic and glycoproteomic analysis.

Chemoresistance largely hampers the clinical use of chemodrugs for cancer patients, combination or sequential drug treatment regimens have been designed to minimize chemotoxicity and resensitize chemoresistance. In this work, the cytotoxic effect of cisplatin was found to be enhanced by palbociclib pretreatment in HeLa cells. With the integration of liquid chromatography-mass spectrometry-based proteomic and N-glycoproteomic workflow, we found that palbociclib alone mainly enhanced the N-glycosylation alterations in HeLa cells, while cisplatin majorly increased the different expression proteins related to apoptosis pathways. As a result, the sequential use of two drugs induced a higher expression level of apoptosis proteins BAX and BAK. Those altered N-glycoproteins induced by palbociclib were implicated in pathways that were closely associated with cell membrane modification and drug sensitivity. Specifically, the top four frequently glycosylated proteins FOLR1, L1CAM, CD63, and LAMP1 were all associated with drug resistance or drug sensitivity. It is suspected that palbociclib-induced N-glycosylation on the membrane protein allowed the HeLa cell to become more vulnerable to cisplatin treatment. Our study provides new insights into the mechanisms underlying the sequential use of target drugs and chemotherapy drugs, meanwhile suggesting a high-efficiency approach that involves proteomic and N-glycoproteomic to facilitate drug discovery.

Reprogramming human urine cells into intestinal organoids with long-term expansion ability and barrier function

Generation of intestinal organoids from human somatic cells by reprogramming would enable intestinal regeneration, disease modeling, and drug screening in a personalized pattern. Here, we report a direct reprogramming protocol for the generation of human urine cells induced intestinal organoids (U-iIOs) under a defined medium. U-iIOs expressed multiple intestinal specific genes and showed resembling gene expression profiles to primary small intestines. U-iIOs can be stably long-term expanded and further differentiated into more mature intestinal lineage cells with high expression of metallothionein and cytochrome P450 (CYP450) genes. These specific molecular features of U-iIOs differ from human pluripotent stem cells derived intestinal organoids (P-iIOs) and intestinal immortalized cell lines. Furthermore, U-iIOs exhibit intestinal barriers indicated by blocking FITC-dextran permeation and uptaking of the specific substrate rhodamine 123. Our study provides a novel platform for patient-specific intestinal organoid generation, which may lead to precision treatment of intestinal diseases and facilitate drug discovery.

Machine Learning in Enhancing Protein Binding Sites Predictions - What Has Changed Since Then?

Accurate identification of protein binding sites is pivotal for understanding molecular interactions and facilitating drug discovery efforts. However, the dynamic nature of proteinligand interactions presents a formidable challenge, necessitating innovative approaches to bridge the gap between theoretical predictions and experimental realities. This review explores the challenges and recent advancements in protein binding site prediction. Specifically, we highlight the integration of molecular dynamics simulations, machine learning, and deep learning techniques to capture the dynamic and complex nature of protein-ligand interactions. Additionally, we discuss the importance of integrating experimental data, such as structural information and biochemical assays, to enhance prediction accuracy and reliability. By navigating the intersection of classical and the onset of machine learning and deep learning approaches, we aim to provide insights into current state-of-the-art techniques and chart a course for future protein binding site prediction advancements. Ultimately, these efforts could unravel the mysteries of protein-ligand interactions and accelerate drug discovery endeavors.

HTS384 NCI60: The Next Phase of the NCI60 Screen.

The NCI60 human tumor cell line screen has been in operation as a service to the cancer research community for more than 30 years. The screen operated with 96-well plates, a 2-day exposure period to test agents, and following cell fixation, a visible absorbance endpoint by the protein-staining dye sulforhodamine B. In this study, we describe the next phase of this important cancer research tool, the HTS384 NCI60 screen. Although the cell lines remain the same, the updated screen is performed with 384-well plates, a 3-day exposure period to test agents, and a luminescent endpoint to measure cell viability based upon cellular ATP content. In this study, a library of 1,003 FDA-approved and investigational small-molecule anticancer agents was screened by the two NCI60 assays. The datasets were compared with a focus on targeted agents with at least six representatives in the library. For many agents, including inhibitors of EGFR, BRAF, MEK, ERK, and PI3K, the patterns of GI50 values were very similar between the screens with strong correlations between those patterns within the dataset from each screen. However, for some groups of targeted agents, including mTOR, BET bromodomain, and NAMPRTase inhibitors, there were limited or no correlations between the two datasets, although the patterns of GI50 values and correlations between those patterns within each dataset were apparent. Beginning in January 2024, the HTS384 NCI60 screen became the free screening service of the NCI to facilitate drug discovery by the cancer research community. Significance: The new NCI60 cell line screen HTS384 shows robust patterns of response to oncology agents and substantial overlap with the classic screen, providing an updated tool for studying therapeutic agents. See related commentary by Colombo and Corsello, p. 2397.

Abstract B013: Deep learning-based prediction of synthetic essentialities in CTNNB1-mutated hepatocellular carcinoma

Abstract Background: Synthetic essentiality represents a promising therapeutic approach by identifying genes that are necessary for the proliferation and survival of tumors harboring hard-to-target gene alterations. Understanding and accurately predicting synthetic essential genes, through genetic dependencies, may reveal therapeutically effective drug targets in a specific molecular context. Deep learning, as exemplified by our published DeepDEP model, has the potential to capture intricate multi-omic profiles for such prediction tasks. However, the validity of such tools in specific biological contexts remains to be fully examined and presents a major obstacle to adoption by researchers. Materials and Methods: To address this gap, we conducted a case study in which we screened for synthetic essential genes for one of the most frequently mutated and yet undruggable genes, CTNNB1, in hepatocellular carcinoma (HCC). Specifically, we predicted the genetic dependencies of each HCC patient in The Cancer Genome Atlas (TCGA; n=346) by DeepDEP and identified potential dependencies that were intensified with the presence of CTNNB1 mutations. The top 10 genes, ranked by p-value of differential gene-effect scores for CTNNB1-mutated (n=92) versus CTNNB1-WT HCC (n=254), were reviewed in the literature to validate their essentiality in CTNNB1-mutated HCC as well as their potential for pharmacologic inhibition. Survival analysis was performed using published data from the IMBrave150 trial to validate one of the findings. Results: Experimental evidence in the literature supported the essentiality of many of the top 10 predicted genes for CTNNB1-mutated HCC, including one gene with mechanistic evidence of being a transcriptional co-activator of β-catenin target genes. Furthermore, several of these genes have known pharmacologic inhibitors which are either natural compounds or FDA-approved drugs. One example was PDGFB, which encodes a ligand activating the PDGF signaling pathway. PDGF signaling is targeted by sorafenib, an FDA-approved first line drug for HCC. Survival analysis of the sorafenib-treated arm of the IMBrave150 trial showed that patients with mutated CTNNB1 had improved progression-free survival compared to those with wild-type CTNNB1 (p = 0.044). Conclusions: Our study illustrates a potential application of deep learning to identify synthetic essential genes, including genes with readily available pharmacologic inhibitors, for targeting challenging gene alterations. Remarkably, our tool demonstrates the ability to predict cancer dependencies with molecular subtype specificity, suggesting a potential for in silico screening of gene dependencies to facilitate drug discovery and personalized medicine approaches. Our current efforts are focused on optimizing this computational pipeline and making it publicly available for cancer researchers. Citation Format: Tyler M. Yasaka, Michael Kasper, Li-Ju Wang, Michael Ning, Yufei Huang, Satdarshan P Monga, Yu-Chiao Chiu. Deep learning-based prediction of synthetic essentialities in CTNNB1-mutated hepatocellular carcinoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Expanding and Translating Cancer Synthetic Vulnerabilities; 2024 Jun 10-13; Montreal, Quebec, Canada. Philadelphia (PA): AACR; Mol Cancer Ther 2024;23(6 Suppl):Abstract nr B013.

SAF: Smart Aggregation Framework for Revealing Atoms Importance Rank and Improving Prediction Rates in Drug Discovery.

Machine learning, and representation learning in particular, has the potential to facilitate drug discovery by screening a large chemical space in silico. A successful approach for representing molecules is to treat them as graphs and utilize graph neural networks. One of the key limitations of such methods is the necessity to represent compounds with different numbers of atoms, which requires aggregating the atom's information. Common aggregation operators, such as averaging, result in a loss of information at the atom level. In this work, we propose a novel aggregating approach where each atom is weighted nonlinearly using the Boltzmann distribution with a hyperparameter analogous to temperature. We show that using this weighted aggregation improves the ability of the gold standard message-passing neural network to predict antibiotic activity. Moreover, by changing the temperature hyperparameter, our approach can reveal the atoms that are important for activity prediction in a smooth and consistent way, thus providing a novel regulated attention mechanism for graph neural networks. We further validate our method by showing that it recapitulates the functional group in β-lactam antibiotics. The ability of our approach to rank the atoms' importance for a desired function can be used within any graph neural network to provide interpretability of the results and predictions at the node level.

Improving protein-protein interaction prediction using protein language model and protein network features

Interactions between proteins are ubiquitous in a wide variety of biological processes. Accurately identifying the protein-protein interaction (PPI) is of significant importance for understanding the mechanisms of protein functions and facilitating drug discovery. Although the wet-lab technological methods are the best way to identify PPI, their major constraints are their time-consuming nature, high cost, and labor-intensiveness. Hence, lots of efforts have been made towards developing computational methods to improve the performance of PPI prediction. In this study, we propose a novel hybrid computational method (called KSGPPI) that aims at improving the prediction performance of PPI via extracting the discriminative information from protein sequences and interaction networks. The KSGPPI model comprises two feature extraction modules. In the first feature extraction module, a large protein language model, ESM-2, is employed to exploit the global complex patterns concealed within protein sequences. Subsequently, feature representations are further extracted through CKSAAP, and a two-dimensional convolutional neural network (CNN) is utilized to capture local information. In the second feature extraction module, the query protein acquires its similar protein from the STRING database via the sequence alignment tool NW-align and then captures the graph embedding feature for the query protein in the protein interaction network of the similar protein using the algorithm of Node2vec. Finally, the features of these two feature extraction modules are efficiently fused; the fused features are then fed into the multilayer perceptron to predict PPI. The results of five-fold cross-validation on the used benchmarked datasets demonstrate that KSGPPI achieves an average prediction accuracy of 88.96 %. Additionally, the average Matthews correlation coefficient value (0.781) of KSGPPI is significantly higher than that of those state-of-the-art PPI prediction methods. The standalone package of KSGPPI is freely downloaded at https://github.com/rickleezhe/KSGPPI.

MFTrans: A multi-feature transformer network for protein secondary structure prediction

In the rapidly evolving field of computational biology, accurate prediction of protein secondary structures is crucial for understanding protein functions, facilitating drug discovery, and advancing disease diagnostics. In this paper, we propose MFTrans, a deep learning-based multi-feature fusion network aimed at enhancing the precision and efficiency of Protein Secondary Structure Prediction (PSSP). This model employs a Multiple Sequence Alignment (MSA) Transformer in combination with a multi-view deep learning architecture to effectively capture both global and local features of protein sequences. MFTrans integrates diverse features generated by protein sequences, including MSA, sequence information, evolutionary information, and hidden state information, using a multi-feature fusion strategy. The MSA Transformer is utilized to interleave row and column attention across the input MSA, while a Transformer encoder and decoder are introduced to enhance the extracted high-level features. A hybrid network architecture, combining a convolutional neural network with a bidirectional Gated Recurrent Unit (BiGRU) network, is used to further extract high-level features after feature fusion. In independent tests, our experimental results show that MFTrans has superior generalization ability, outperforming other state-of-the-art PSSP models by 3 % on average on public benchmarks including CASP12, CASP13, CASP14, TEST2016, TEST2018, and CB513. Case studies further highlight its advanced performance in predicting mutation sites. MFTrans contributes significantly to the protein science field, opening new avenues for drug discovery, disease diagnosis, and protein.