Discovery of micrornas and their role in post-transcriptional gene regulation

On October 7, the Karolinska Institute in Sweden announced that the 2024 Nobel Prize in Physiology or Medicine will be awarded to American scientists Victor Ambros and Gary Ruvkun for their discovery of microRNA (miRNA) and its role in gene regulation.

For a long time, it was believed that genetic information is transcribed from DNA to mRNA, which is then translated into proteins. However, the findings of Victor Ambros and Gary Ruvkun demonstrate that RNA, in addition to encoding proteins, also includes non-coding RNA (ncRNA) that plays a regulatory role in gene function.

Discovery of miRNA

In 1993, Victor Ambros discovered a gene called lin-4 while studying Caenorhabditis elegans, which expresses a very short RNA molecule and does not code for a protein. This small RNA from lin-4 appeared to negatively regulate the expression levels of the lin-14 protein. Subsequently, Victor Ambros and Gary Ruvkun further found that lin-4 miRNA blocks the expression of lin-14 protein by binding to complementary sequences in lin-14 mRNA^[2].

Although this discovery was groundbreaking and revealed a new mechanism of gene regulation, it did not garner much attention in the scientific community at the time.

It wasn't until 2000 that Gary Ruvkun discovered another gene, let-7, which is widely present in humans and many other species^[4]. This finding demonstrated that the existence of miRNA extends far beyond C. elegans. Researchers began to realize the universality and importance of the miRNA regulatory mechanism in gene expression across various organisms.

The discovery of the let-7 gene sparked great interest among researchers, motivating scientists worldwide to conduct related studies. Over the following years, hundreds of different miRNAs were identified. Due to the rapid discovery and functional elucidation of miRNAs, scientists from the Sanger Institute established the microRNA Registry in 2002, which was later renamed miRBase. The creation of this database provided a standardized and convenient resource for miRNA research, enabling researchers to easily access comprehensive information about known miRNAs, including their sequences, origins, and functions.

miRNA: Synthesis and Application

In addition to identifying new miRNAs, researchers have elucidated the mechanisms by which miRNAs are generated and bind to mRNA, leading to the inhibition of protein synthesis or the degradation of mRNA. This will not be elaborated on here.

Interestingly, due to the low complementarity of miRNAs, they often do not perfectly match their target genes, allowing them to regulate the expression of multiple target genes simultaneously. Increasing evidence suggests that miRNA dysregulation is associated with various human diseases, including cancer, diabetes, and cardiovascular diseases. For example, the loss of let-7 plays a pathogenic role in several cancers. During the process of muscle cell fibrosis, miR-21 is significantly upregulated, leading to cardiac hypertrophy^[5][6][7].

Currently, efforts are underway to develop oligonucleotide drugs targeting miRNAs. Examples include RGLS4326 (which inhibits the function of miR-17) for the treatment of autosomal dominant polycystic kidney disease (ADPKD), Miravirsen (which inhibits the function of miR-122) for HCV treatment, and Cobomarsen (which inhibits the function of miR-155) for the treatment of B-cell lymphoma.

miRNA Related Products

MedChemExpress (MCE) has independently designed and developed a series of tools for studying miRNA functions based on the mature miRNA sequences of humans, mice, and rats from the miRBase database. These tools include miRNA mimics, miRNA inhibitors, miRNA agomirs, and miRNA antagomirs. Currently, these miRNA-related products are available at a 35% discount promotion. We welcome all customers to inquire about our custom services related to miRNA.

For promoting the fundamental discovery and invention of machine learning using artificial neural networks

On October 8, the Karolinska Institute in Sweden announced that the 2024 Nobel Prize in Physics will be awarded to American scientist John J. Hopfield and Canadian scientist Geoffrey E. Hinton for their foundational discoveries and inventions that advanced the use of artificial neural networks in machine learning.

The awardees utilized tools from physics to develop various methods that laid the groundwork for today's powerful machine learning. John Hopfield created a structure capable of storing and reconstructing information. Geoffrey Hinton invented a method that can independently discover data properties, which is crucial for the large artificial neural networks currently in use.

The development of machine learning has employed a structure known as artificial neural networks. Today, when we talk about "hot" AI, we are typically referring to this technology.

In recent years, this technology has also begun to be used to compute and predict the properties of molecules and materials, such as calculating the molecular structure of proteins that determine their function or identifying which new materials may have the best characteristics for more efficient solar cells.

Machine Learning

As early as 1950, the Turing Test was proposed, igniting a heated discussion about whether machines can "think."

Machine learning differs from traditional software in its operation. Traditional software works by receiving data, processing it according to clear instructions, and producing results. In contrast, in machine learning, computers learn through examples, enabling them to solve problems.

Yes, computers cannot think, but machines can now mimic functions such as memory and learning. This year's Nobel Prize winners contributed to making this possible, laying the groundwork for today's powerful machine learning. They utilized fundamental concepts and methods from physics to develop techniques for processing information using network structures, specifically through artificial neural networks.

The advancements we are witnessing today have been made possible by the acquisition of vast amounts of data available for training networks and a significant increase in computational power.

AI and Drug Development

Today’s artificial neural networks are often very large and consist of multiple layers. These are known as deep neural networks, and their training method is referred to as deep learning.

Deep learning is a branch of artificial intelligence (AI) that utilizes neural networks for learning. This technology has made significant advances in the field of biomedicine.

Research has shown that deep learning techniques have advantages in optimizing chemical synthesis pathways, predicting the pharmacokinetic properties of drugs, forecasting drug targets, and generating novel molecules.

Currently, researchers have developed a range of application strategies based on deep learning for disease diagnosis, protein design, and medical image recognition. The pharmaceutical industry is also beginning to recognize the value of deep learning technology, hoping to leverage it to accelerate drug development and reduce costs.

MCE One-Stop Drug Screening Platform

MCE's one-stop drug screening platform focuses on the drug discovery phase and utilizes generative artificial intelligence technology to create an ultra-large compound library called MegaUni, which combines novelty, drug-like properties, diversity, and synthetic feasibility. This library is suitable for AI drug screening and large-scale virtual screening.

The MegaUni library contains a vast number of previously unreported novel structural molecules, providing limitless possibilities for scientific innovation. Additionally, AI algorithms are applied to the construction of various types of Mini compound libraries, helping clients more efficiently obtain compound libraries that meet their specific needs.

Furthermore, the drug screening platform includes both computer-based virtual screening and physical drug screening. In the realm of virtual screening, it combines AI active learning with molecular docking to support larger-scale virtual screenings.

"Computational Protein Design" + "Protein Structure Prediction"
On October 9, 2024, the Royal Swedish Academy of Sciences decided to award the 2024 Nobel Prize in Chemistry, with half going to David Baker for his contributions to the field of "computational protein design." The other half is jointly awarded to Demis Hassabis and John M. Jumper for their outstanding achievements in "protein structure prediction."

David Baker has created entirely new types of proteins, successfully accomplishing feats that were previously thought to be nearly impossible. Demis Hassabis and John M. Jumper developed an artificial intelligence model to tackle a problem that has persisted for 50 years: predicting the complex structures of proteins. These two distinct discoveries are closely linked and hold tremendous potential.

“Computational Protein Design”

The field of protein design began in the late 1990s, with researchers designing custom proteins with new functions. In many cases, researchers adjusted existing proteins so that they could break down harmful substances or serve as tools in chemical manufacturing.

In 2003, David Baker successfully designed a new protein that was distinct from other proteins. Since then, his research team has continued to create one imaginative protein after another, including proteins that can be used as drugs, vaccines, nanomaterials, and miniaturized sensors.

“Protein Structure Prediction”

Proteins are typically composed of 20 different amino acids, which can be combined in countless ways. Using the information stored in DNA as a template, amino acids are linked together in our cells to form long chains. These chains of amino acids twist and fold into unique (and sometimes one-of-a-kind) three-dimensional structures. This structure imparts functionality to the proteins.

Since the 1970s, researchers have been trying to predict protein structures based on amino acid sequences, but this has proven to be very challenging.

It was not until 2020 that Demis Hassabis and John Jumper introduced an AI model called AlphaFold2. With this model, they were able to predict the structures of nearly all 200 million proteins discovered by researchers.

Statistics show that about two-thirds of the structures predicted by AlphaFold 2 achieved a prediction accuracy that matches the measurement precision of structural biology experiments.

Since this breakthrough, more than 2 million users from 190 countries have utilized AlphaFold 2.

Additionally, AlphaFold 3, a new revolutionary artificial intelligence (AI) model, has been developed through significant advancements in the architecture and training process of AlphaFold 2. It adapts to more general chemical structures and improves the efficiency of learning from data. It predicts the structures of a wider range of biomolecules with unprecedented accuracy, including complexes involving ligands, ions, nucleic acids, and modified residues.

Compared to existing prediction methods, AlphaFold 3 shows at least a 50% improvement in the accuracy of predicting protein interactions with other biomolecules, with prediction accuracy even doubling for certain important categories of interactions.

Without proteins, life cannot exist. The ability to predict protein structures and design our own proteins has a profound impact on humanity.