Tech News
Epiregulin: The Growth Factor Redefining Human Brain Evolution
A new study uncovers that a growth factor, epiregulin, significantly contributes to the expansion of the human neocortex, enhancing our comprehension of what makes humans unique in cognitive functions.What makes us human? According to neurobiologists, it is our neocortex. This outer layer of the brain is rich in neurons and lets us do abstract thinking, create art, and speak complex languages. An international team led by Dr. Mareike Albert at the Center for Regenerative Therapies Dresden (CRTD) of TUD Dresden University of Technology has identified a new factor that might have contributed to neocortex expansion in humans. The results were published in the EMBO Journal.The neocortex is the characteristic folded outer layer of the brain that resembles a walnut. It is responsible for higher cognitive functions such as abstract thinking, art, and language. “The neocortex is the most recently evolved part of the brain,” says Dr. Mareike Albert, research group leader at the CRTD. “All mammals have a neocortex, but it varies in size and complexity. Human and primate neocortices have folds while, for example, mice have a completely smooth neocortex, without any creases.”The folds characteristic of the human brain increase the surface area of the neocortex. The human neocortex has a greater number of neurons that support complex cognitive functions. The molecular mechanisms driving neocortex evolution are still largely unknown. “Which genes are responsible for inter-species differences in neocortex size? What factors have contributed to brain expansion in humans? Answering these questions is crucial to understanding human brain development and potentially addressing mental health disorders,” explains Dr. Albert.The Power of Brain OrganoidsTo search for factors influencing brain expansion, the Albert group compared the developing brains of mice and humans. “Stem cells in mice don’t divide as much and don’t produce as many neurons compared to primates. Humans, on the other hand, have a large number of stem cells in the developing brain. This highly expanded pool of stem cells underlies the increase in the number of neurons and brain size,” explains Dr. Albert.A microscopy image of a human brain organoid. Credit: Janine HoffmannThe team found a factor that is present in humans but not in mice. Using 3D cell culture technology, the group tested if the newly identified factor could influence the expansion of the neocortex. “Thanks to the research awarded with the Nobel prize in 2012, it is possible to turn any cell into a stem cell. Such a stem cell can then be transformed into a three-dimensional tissue that resembles an organ, e.g., a brain. Human stem cells make it possible to study development and diseases directly in human tissues,” explains Dr. Albert.These 3D brain cultures, or brain organoids, may not resemble brains to an untrained eye, but they mimic the cellular complexity of developing brains. “Most of the cell types of the developing brain are present. They interact, signal, and are similarly arranged as in an actual human brain,” says Dr. Albert.Using 3D brain organoids, the group was able to show that a growth factor, known as epiregulin, indeed promotes the division and expansion of stem cells in the developing brain.All About the Amount“Knowing that epiregulin drives expansion of human neocortical stem cells, we looked back at the gene that codes for epiregulin and tried to trace it through the evolutionary tree,” says study lead author Paula Cubillos, a doctoral candidate at the CRTD. The gene is not unique to humans, but also present in other primates and even in mice. “Epiregulin is not produced in the developing mouse brain, however, because the gene is permanently shut off and not being used. We were intrigued to understand whether there are any differences in how epiregulin works in humans and other primates,” explains Paula Cubillos.The researchers turned again to the 3D culture technology. Using gorilla stem cells, the researchers generated gorilla brain organoids. “Gorillas are endangered species. We know very little about their brain development. Organoids made from stem cells offer a way to study their brain development without interacting with the species at all,” says Dr. Albert.Comparing the effect of epiregulin in human and gorilla brain organoids, the team found that adding epiregulin to gorilla brain organoids can further promote the expansion of stem cells. However, adding even more epiregulin to human brain organoids did not have the same effect. This might be because the human neocortex has already expanded to a very large extent.“Unlike previously identified factors, epiregulin as such seems not to be unique to humans. Instead, the amount of the growth factor seems to be the crucial regulator for the inter-species differences,” concludes Dr. Albert.This study not only advances our understanding of human uniqueness but also highlights the importance of new technologies that offer ethical and non-invasive complements to animal research.Reference: “The growth factor EPIREGULIN promotes basal progenitor cell proliferation in the developing neocortex” by Paula Cubillos, Nora Ditzer, Annika Kolodziejczyk, Gustav Schwenk, Janine Hoffmann, Theresa M Schütze, Razvan P Derihaci, Cahit Birdir, Johannes EM Köllner, Andreas Petzold, Mihail Sarov, Ulrich Martin, Katherine R Long, Pauline Wimberger and Mareike Albert, 21 March 2024, The EMBO Journal.DOI: 10.1038/s44318-024-00068-7The study was performed in collaboration with King’s College London, the Medical Faculty Carl Gustav Carus of TU Dresden, the Max Planck Institute of Molecular Cell Biology and Genetics, and Hannover Medical School.
Scientists Uncover Unique New 1D Superconducting State
New research has uncovered one-dimensional superconducting stripes at the EuO/KTO(110) interface, a result of the ferromagnetic proximity effect. This discovery highlights the significant influence of magnetism on superconducting states and provides a platform for further exploration of high-temperature superconductivity. This study not only advances our understanding of the intricate relationship between superconductivity and magnetism but also showcases the unique properties of superconducting oxide heterostructures.A team led by Chen Xianhui and Professor Xiang Ziji from the CAS Key Laboratory of Strongly-Coupled Quantum Matter Physics and the Department of Physics at the University of Science and Technology of China, uncovered a unique superconducting state characterized by one-dimensional superconducting stripes. This state is induced by the ferromagnetic proximity effect in an oxide heterostructure made up of ferromagnetic EuO and (110)-oriented KTaO3 (KTO). Their findings were published in Nature Physics.The academic community concurs that the emergence of unconventional superconducting pairings is intricately linked to magnetism, particularly in copper oxides and iron-based high-temperature superconductors. Magnetic fluctuations are deemed pivotal in the genesis of high-temperature superconductivity, where the interplay between superconductivity and magnetism gives rise to superconducting states exhibiting unique spatial modulation. Superconducting oxide heterostructures encompassing magnetic structural units emerge as an optimal platform for investigating such superconducting states.Building upon their prior achievements, the research team delved deeper into the superconductivity of this system and its relationship with the ferromagnetic proximity effect, meticulously adjusting the carrier concentration of the two-dimensional electron gas residing at the interface. They uncovered an intriguing in-plane anisotropy in superconductivity among samples with low carrier concentrations, which nevertheless vanished in samples exhibiting higher carrier concentrations. Observations of One-Dimensional Superconducting StripesThe superconductivity transition temperature related to the current direction at the heterojunction interface is caused by the formation of one-dimensional superconducting stripes due to the reduction of superconductivity dimension. Meanwhile, anomalous Hall effect and magnetoresistance hysteresis behavior indicate that the coupling between interfacial conduction electrons and ferromagnetism is affected by band filling. The hybridization of Eu and Ta atomic orbitals within a specific energy range leads to band spin splitting, which is consistent with the experimental results. Therefore, the emergence of one-dimensional superconducting stripes in EuO/KTO(110) heterojunction is confirmed to be caused by the coupling effect between superconductivity and magnetism.This study reveals the existence of a superconducting stripe phase at the EuO/KTO(110) interface, induced by the ferromagnetic proximity effect. It presents the first unambiguous experimental evidence of exotic superconducting states emerging from the intricate coupling between superconductivity and magnetism at oxide interfaces.Reference: “Superconducting stripes induced by ferromagnetic proximity in an oxide heterostructure” by Xiangyu Hua, Zimeng Zeng, Fanbao Meng, Hongxu Yao, Zongyao Huang, Xuanyu Long, Zhaohang Li, Youfang Wang, Zhenyu Wang, Tao Wu, Zhengyu Weng, Yihua Wang, Zheng Liu, Ziji Xiang and Xianhui Chen, 11 March 2024, Nature Physics.DOI: 10.1038/s41567-024-02443-x
Assassin Gene Discovery Shifts Cancer Treatment Paradigms – “Very Unexpected Finding”
Researchers at the Netherlands Cancer Institute have discovered a new mechanism of cancer cell death involving the Schlafen11 gene, which could revolutionize the understanding and treatment of cancer. This gene helps shut down protein production in response to DNA damage, offering a new target for cancer therapies, especially in cases where traditional p53 protein pathways are ineffective. Credit: SciTechDaily.comScientists have discovered that the way cancer cells die from chemotherapy appears to be different than previously understood. New research highlights the Schlafen11 gene’s role in cancer cell death, providing a fresh approach to chemotherapy and cancer treatment strategies.Chemotherapy kills cancer cells. But the way these cells die appears to be different than previously understood. Researchers from the Netherlands Cancer Institute, led by Thijn Brummelkamp, have uncovered a completely new way in which cancer cells die: due to the Schlafen11 gene. “This is a very unexpected finding. Cancer patients have been treated with chemotherapy for almost a century, but this route to cell death has never been observed before. Where and when this occurs in patients will need to be further investigated. This discovery could ultimately have implications for the treatment of cancer patients.” They published their findings in the journal Science.The Role of DNA Damage in Cancer TreatmentMany cancer treatments damage cell DNA. After too much irreparable damage, cells can initiate their own death. High school biology teaches us that the protein p53 takes charge of this process. p53 ensures repair of damaged DNA, but initiates cell suicide when the damage becomes too severe. This prevents uncontrolled cell division and cancer formation. Organoid of a patient with colon cancer, treated (lower) and not treated with the chemotherapy etoposide. The treatment causes DNA damage and a reduction in protein synthesis. This triggers a stress signal that causes the cells to die. Orange: marker for DNA damage. Green: marker for protein synthesis. Credit: Netherlands Cancer InstituteSurprise: Unanswered QuestionThat sounds like a foolproof system, but reality is more complex. “In more than half of tumors, p53 no longer functions,” says Thijn Brummelkamp. “The key player p53 plays no role there. So why do cancer cells without p53 still die when you damage their DNA with chemotherapy or radiation? To my surprise, that turned out to be an unanswered question.”His research group then discovered, together with the group of colleague Reuven Agami, a previously unknown way in which cells die after DNA damage. In the lab, they administered chemotherapy to cells in which they carefully modified the DNA. Thijn: “We were looking for a genetic change that would allow cells to survive chemotherapy. Our group has a lot of experience in selectively disabling genes, which we could perfectly apply here.”Turning Off Genes, One by OnePeople have thousands of genes, many of which have functions that are unclear to us. To determine the roles of our genes, researcher Thijn Brummelkamp developed a method using haploid cells. These cells contain only one copy of each gene, unlike the regular cells in our bodies that contain two copies. Handling two copies can be challenging in genetic experiments, because changes (mutations) often occur in just one of them. This makes it difficult to observe the effects of these mutations.Together with other researchers, Brummelkamp has been unraveling processes that are crucial in disease for years using this versatile method. For example, his group recently discovered that cells can make lipids in a different way than previously known. They uncovered how certain viruses, including the deadly Ebola virus, manage to enter human cells. They delved into cancer cell resistance against specific therapies and identified proteins that act as brakes on the immune system, which is relevant to cancer immunotherapy. Over the last years, his team discovered two enzymes that had remained elusive for four decades, and that turned out to be vital for muscle function and brain development.New Key Player in Cell DeathBy switching off genes, the research group found a new pathway to cell death headed by the gene Schlafen11 (SLFN11). Principle investigator Nicolaas Boon: “In the event of DNA damage, SLFN11 shuts down the protein factories of cells: the ribosomes. This causes immense stress in these cells, which leads to their death. The new route we discovered completely bypasses p53.”The SLFN11 gene is not unfamiliar in cancer research. It is often inactive in tumors of patients who do not respond to chemotherapy, says Thijn. “We can now explain this link. When cells lack SLFN11 they will not die in this manner in response to DNA damage. The cells will survive and the cancer persist.”Impact on Cancer Treatment“This discovery uncovers many new research questions, which is usually the case in fundamental research,” says Thijn. “We have demonstrated our discovery in lab-grown cancer cells, but many important questions remain: Where and when does this pathway occur in patients? How does it affect immunotherapy or chemotherapy? Does it affect the side effects of cancer therapy? If this form of cell death also proves to play a significant role in patients, this finding will have implications for cancer treatments. These are important questions to investigate further.”Reference: “DNA damage induces p53-independent apoptosis through ribosome stalling” 16 May 2024, Science.DOI: 10.1126/science.adh7950This research was financially supported by KWF Dutch Cancer Society, Oncode Institute, and Health Holland.
Twin X-Class Flares: NASA Captures Epic Solar Showdown
NASA’s Solar Dynamics Observatory captured this image of a solar flare – as seen in the bright flash on the right – on May 15, 2024. The image shows a subset of extreme ultraviolet light that highlights the extremely hot material in flares and which is colorized in teal. Credit: NASA/SDOOn May 15, 2024, NASA’s Solar Dynamics Observatory, which continually monitors the Sun, captured images of two powerful X-class solar flares. The first was classified as an X3.4 class flare and peaked at 4:37 a.m. ET. The second flare, classified as X2.9, peaked at 10:38 a.m. ET.Solar flares are intense bursts of radiation emanating from the release of magnetic energy associated with sunspots. These flares are visible across the electromagnetic spectrum, from radio waves to gamma rays, and are among the solar system’s most powerful phenomena. When directed towards Earth, solar flares can influence our planet in several ways, including disrupting satellite communications, affecting power grid operations, and enhancing auroral displays (Northern and Southern Lights).The impact on Earth primarily involves the interaction of solar radiation with Earth’s magnetic field, which can lead to geomagnetic storms. These storms can disrupt technology and infrastructure and expose astronauts and high-altitude flights to higher levels of radiation. NASA’s Solar Dynamics Observatory captured this image of a solar flare – as seen in the bright flash on the left – on May 15, 2024. The image shows a subset of extreme ultraviolet light that highlights the extremely hot material in flares and which is colorized in teal. Credit: NASA/SDOSolar flares are categorized by their intensity in the X-ray wavelengths, ranging from class A (the weakest) to class X (the strongest). Each class has a tenfold increase in energy output, and within each class, a finer scale from 1 to 9 further distinguishes the flare’s strength. X-class flares are the largest explosions in the solar system and can trigger planet-wide radio blackouts and long-lasting radiation storms.NASA’s Solar Dynamics Observatory (SDO) is a mission dedicated to understanding the Sun’s influence on Earth and Near-Earth space by studying the solar atmosphere on small scales of space and time and in many wavelengths simultaneously. Launched on February 11, 2010, SDO is part of NASA’s Living With a Star (LWS) program.Artist’s concept image of the SDO satellite orbiting Earth. Credit: NASAThe observatory is equipped with a suite of instruments that provide observations leading to a more complete understanding of the solar dynamics driving variability in the Earth’s environment. One of these instruments is the Atmospheric Imaging Assembly (AIA), which produces images of the solar disk in multiple wavelengths every 12 seconds, providing insights into the solar corona’s structure and dynamics. Another crucial instrument, the Helioseismic and Magnetic Imager (HMI), studies solar variability and characterizes the Sun’s interior and the various components of magnetic activity.SDO’s data are vital in helping scientists understand the Sun’s influence on Earth and the space environment by observing solar flares, coronal mass ejections (CMEs), and other solar phenomena. This information is crucial for improving the ability to forecast space weather events that can affect satellite operations, astronauts, and Earth-based systems.
How Game Theory Is Making AI Smarter
MIT CSAIL researchers developed a “consensus game” to improve AI text understanding and generation by treating the process as a game where one part generates sentences and another part evaluates them. This method, called equilibrium ranking, significantly enhances AI performance across tasks like reading comprehension, math problem-solving, and dialogue. Credit: SciTechDaily.comMIT CSAIL researchers have developed a new “consensus game” that elevates AI’s text comprehension and generation skills.MIT’s “consensus game” improves AI text generation using game theory. This method, equilibrium ranking, enhances AI performance and reliability but faces computational challenges. It could significantly advance language model decoding.AI Consensus Game: A New Approach to Language ModelsImagine you are playing a game with a friend where your goal is to communicate secret messages to each other using only cryptic sentences. Your friend’s job is to guess the secret message behind your sentences. Sometimes, you give clues directly, and other times, your friend has to guess the message by asking yes-or-no questions about the clues you’ve given. The challenge is that both of you want to make sure you’re understanding each other correctly and agreeing on the secret message. MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers have created a similar “game” to help improve how AI understands and generates text. It is known as a “consensus game” and it involves two parts of an AI system — one part tries to generate sentences (like giving clues), and the other part tries to understand and evaluate those sentences (like guessing the secret message).MIT researchers’ “consensus game” is a game-theoretic approach for language model decoding. The equilibrium-ranking algorithm harmonizes generative and discriminative querying to enhance prediction accuracy across various tasks, outperforming larger models and demonstrating the potential of game theory in improving language model consistency and truthfulness. Credit: Alex Shipps/MIT CSAILGame-Theoretic Approach to AIThe researchers discovered that by treating this interaction as a game, where both parts of the AI work together under specific rules to agree on the right message, they could significantly improve the AI’s ability to give correct and coherent answers to questions. They tested this new game-like approach on a variety of tasks, such as reading comprehension, solving math problems, and carrying on conversations, and found that it helped the AI perform better across the board.Traditionally, large language models answer one of two ways: generating answers directly from the model (generative querying) or using the model to score a set of predefined answers (discriminative querying), which can lead to differing and sometimes incompatible results. With the generative approach, “Who is the president of the United States?” might yield a straightforward answer like “Joe Biden.” However, a discriminative query could incorrectly dispute this fact when evaluating the same answer, such as “Barack Obama.”Balancing AI Responses With Equilibrium RankingSo, how do we reconcile mutually incompatible scoring procedures to achieve coherent, efficient predictions?“Imagine a new way to help language models understand and generate text, like a game. We’ve developed a training-free, game-theoretic method that treats the whole process as a complex game of clues and signals, where a generator tries to send the right message to a discriminator using natural language. Instead of chess pieces, they’re using words and sentences,” says Athul Jacob, an MIT PhD student in electrical engineering and computer science and CSAIL affiliate. “Our way to navigate this game is finding the ‘approximate equilibria,’ leading to a new decoding algorithm called ‘equilibrium ranking.’ It’s a pretty exciting demonstration of how bringing game-theoretic strategies into the mix can tackle some big challenges in making language models more reliable and consistent.”When tested across many tasks, like reading comprehension, commonsense reasoning, math problem-solving, and dialogue, the team’s algorithm consistently improved how well these models performed. Using the ER algorithm with the LLaMA-7B model even outshone the results from much larger models. “Given that they are already competitive, that people have been working on it for a while, but the level of improvements we saw being able to outperform a model that’s 10 times the size was a pleasant surprise,” says Jacob.Game On“Diplomacy,” a strategic board game set in pre-World War I Europe, where players negotiate alliances, betray friends, and conquer territories without the use of dice — relying purely on skill, strategy, and interpersonal manipulation — recently had a second coming. In November 2022, computer scientists, including Jacob, developed “Cicero,” an AI agent that achieves human-level capabilities in the mixed-motive seven-player game, which requires the same aforementioned skills, but with natural language. The math behind this partially inspired the Consensus Game.While the history of AI agents long predates when OpenAI’s software entered the chat in November 2022, it’s well documented that they can still cosplay as your well-meaning, yet pathological friend.The consensus game system reaches equilibrium as an agreement, ensuring accuracy and fidelity to the model’s original insights. To achieve this, the method iteratively adjusts the interactions between the generative and discriminative components until they reach a consensus on an answer that accurately reflects reality and aligns with their initial beliefs. This approach effectively bridges the gap between the two querying methods.Practical Applications and ChallengesIn practice, implementing the consensus game approach to language model querying, especially for question-answering tasks, does involve significant computational challenges. For example, when using datasets like MMLU, which have thousands of questions and multiple-choice answers, the model must apply the mechanism to each query. Then, it must reach a consensus between the generative and discriminative components for every question and its possible answers.The system did struggle with a grade school right of passage: math word problems. It couldn’t generate wrong answers, which is a critical component of understanding the process of coming up with the right one.Future Directions“The last few years have seen really impressive progress in both strategic decision-making and language generation from AI systems, but we’re just starting to figure out how to put the two together. Equilibrium ranking is a first step in this direction, but I think there’s a lot we’ll be able to do to scale this up to more complex problems,” says Jacob.An avenue of future work involves enhancing the base model by integrating the outputs of the current method. This is particularly promising since it can yield more factual and consistent answers across various tasks, including factuality and open-ended generation. The potential for such a method to significantly improve the base model’s performance is high, which could result in more reliable and factual outputs from ChatGPT and similar language models that people use daily.Expert Insights on AI Advancements“Even though modern language models, such as ChatGPT and Gemini, have led to solving various tasks through chat interfaces, the statistical decoding process that generates a response from such models has remained unchanged for decades,” says Google Research Scientist Ahmad Beirami, who was not involved in the work. “The proposal by the MIT researchers is an innovative game-theoretic framework for decoding from language models through solving the equilibrium of a consensus game. The significant performance gains reported in the research paper are promising, opening the door to a potential paradigm shift in language model decoding that may fuel a flurry of new applications.”Reference: “The Consensus Game: Language Model Generation via Equilibrium Search” by Athul Paul Jacob, Yikang Shen, Gabriele Farina and Jacob Andreas, 13 October 2023, Computer Science > Computer Science and Game Theory.arXiv:2310.09139Jacob wrote the paper with MIT-IBM Watson Lab researcher Yikang Shen and MIT Department of Electrical Engineering and Computer Science assistant professors Gabriele Farina and Jacob Andreas, who is also a CSAIL member. They presented their work at the International Conference on Learning Representations (ICLR) earlier this month, where it was highlighted as a “spotlight paper.” The research also received a “best paper award” at the NeurIPS R0-FoMo Workshop in December 2023.
