Axial

Axial partners with great founders and inventors. We invest in early-stage life sciences companies such as Appia Bio, Seranova Bio, Delix Therapeutics, Simcha Therapeutics, among others often when they are no more than an idea. We are fanatical about helping the rare inventor who is compelled to build their own enduring business. If you or someone you know has a great idea or company in life sciences, Axial would be excited to get to know you and possibly invest in your vision and company. We are excited to be in business with you — email us at info@axialvc.com

Inventors are consistently at the forefront of business but historically have trouble creating value from their work. Transformative business models often use a new technology to solve a problem that others could not. And enduring businesses are often open to new inventions and show a willingness to adopt new models. In this setting, inventors and their inventions have a important role to play in not only creating businesses but keeping them relevant.

Taking an invention to create a business is a complex task but history would show it is incredibly worthwhile and lucrative. It is a matching problem. Finding the right problem for an invention. Versus inventing objects then searching for a problem. This combination of invention & business models can generate a perpetual emergence of new problems to solve. The coal plant has given us electricity but polluted the air. The Internet has connected the world but reduced trust. The corn fields have fed the world but dried up the ground. From Henry David Thoreau: “[All our inventions are] but improved means to an unimproved end.”

As a result, any business model needs explicit design to maximize the benefits of an invention while doing everything to mitigate downsides. For the good of its own business and for society. Inventors are the starting point to create new wealth and have more influence than they think they have on deciding how to distribute this wealthy fairly. Shaping a society with more invention, thereby generating even more prosperity for more people.

Patent rates versus parent income. Source: EoO

There is accelerating change but the problems are no more complex to solve than they were centuries ago. The invention of steel brought on new products unimagined but created the machines of war that WWI and WWII were built upon. The personal computer and Internet themselves brought high levels of progress for companies and individuals while at the same time being used to intrude into the lives of entire populations and making people equivalent to numerical objects. And new inventions and models are created to solve these problems. In particular, the US is a haven for inventors, acting as a driving force for the world’s progress.

First-generation immigrants make up 12% of the US population and 16% of the US workforce, but 35% of inventors. Map of the source of inventors to the US. Source: ITIF

Countries with the most immigrant inventors (2000–2010). Source: Atlas

Inventors come in many forms — scientists (i.e. Herbert Boyer), entrepreneurs (i.e. Earl Bakken), engineers (i.e. James Watt). For most of history, parts of society have met these inventors with skepticism. Maybe it is the same response ancient people had toward shamans. Where new technologies are often perceived as magic. Parts of this response are justified: technology often destroys more than businesses and jobs, it also breaks down culture and values. Inventions are usually radical discontinuities that bring together seemingly disparate ideas (i.e. wide epistemic base). Some inventions are breakthroughs, others are really a bundle of tweaks from one inventor, or a series of them (i.e. establishment of genetics between Mendel and Morgan). Just like zooming in constantly at a coastline, there is no end only scale.

Many of the important companies around the world started with a single invention. Sony made low-cost radios. Goodyear’s name comes from it’s eponymous founder Charles Goodyear who invented vulcanized rubber after years of trial and error. Ultimately, great inventors make something people want. But a great company cannot survive off one invention alone. A unique design must emerge. A business model can be referred as “a term of art.” Like art, a great business model is simple to recognize but hard to define. A business model represents the logic of a company — the structure, governance, and history of transactions creating value for customers and stakeholders. Core structural advantages such as exclusivity and distribution emerge from business model design. When these advantages become embedded into a business model, competitors may know a company’s secret and still cannot replicate it.

This is where business models can help. Inventors create a lot of value but fail to accumulate it most of the time. This really is a business problem rather than a technology one. Conventional wisdom suggests a business ought to come up with new products to make older versions obsolete. Rather, successful businesses identify new needs and build products to meet them. Instead of focusing on disruption, the most important companies focus on need/problems.

Eastman Kodak had a great business around photographs. The company failed to make the transition toward Internet-centered products. Likely, the business accepted the fallacy of innovation — funding initiatives around disruption rather than studying real customer needs. In contrast, a great business, Netflix, was built around DVDs. Overtime, the company transitioned toward streaming. The business was focused on what customers actually wanted. Users want video on demand and are agnostic toward the distribution method preferring the one with the most convenience. A business centered around electroplating would only exist today if it focused on customer needs beyond trinkets. Against an emerging threat from Nest, Honeywell used its advantages around customer relationships to understand what products needed to be built. Honeywell improved its HVAC software and so far has successfully defended itself. Honeywell could have engaged in superficial efforts completely separate from core work. Instead, Honeywell invented a new way to sell products and did a good job replicating Nest into its HVAC product line.

A business model is perceived as a static design almost like an economic model. A countering perspective is to describe a business model like a model organism. Biologists study a small set of organisms (i.e. C. elegans, Drosophila melanogaster, yeast, zebrafish) to generalize them to other organisms. The elements that helped Facebook or Southwest succeed can be used to create new models. In biology, model organisms help address a lack of knowledge. Just as there is more to be discovered in biology than known the same is probably true in business. As a result, the taxonomy of business models has yet to be fully characterized.

For example, the McDonalds’ business model is the canonical case study of franchising. In the genre, a business creates a format to deliver a product and sells this knowledge and support for a recurring fee. This model moved beyond restaurants and became generalized to convenience stores, gyms, and hotels. McDonalds still remains the benchmark when studying franchising.

A great invention doesn’t necessarily create a great business. Taking franchising to a new industry doesn’t guarantee success. Historically, some of the most successful businesses had advantages around design, marketing, and production. The parts built around an invention are the important features of a business. Edison didn’t really invent the lightbulb but he, along with Westinghouse, designed a business around electricity generation to make the lightbulb usable. Often, great inventors synthesize existing ideas, building the infrastructure to bring their ideas to society. Ultimately, inventions succeed when they complete useful tasks that people recognize as important.

Trial and error

History’s great inventors often lacked formal training and created their inventions through trial and error. The inventor of microscopy, Antonie van Leeuwenhoek, was self-taught and over decades perfected his microscope to see what no others could see at the time. In London, the Royal Society caught wind of his work where he was accepted into but actually never attended the ceremony.

Evolution

Some inventions just evolve — both gradual and discrete — to finally emerge into a final form. There really is no single inventor. The iPhone is a great invention but one dependent on government-funded research for the processor, chip, and screen.

Breakthrough

Inventions are usually equated to breakthroughs. There are many types of inventions, but a breakthrough advances human knowledge providing an opportunity for commercialization. The discovery of the structure of DNA enabled sequencing and cloning to emerge. Great companies like Illumina, IDT, Biogen, and Genentech emerged from this work. The pH meter was a breakthrough in measurement and was the centerpiece for Beckman Instruments. The radio emerged from breakthroughs around magnetism during the 19th century from Faraday, Maxwell, and Hertz.

Accidental inventions

Great inventions like viagra and the birth control pill have unexpected applications. Roy Plunkett a scientist at DuPont accidently created a white goo and after some tests realized the incredible non-stick properties it had ultimately leading to Teflon. In 1957, the FDA approved the pill first for only menstrual disorders and infertility. Each box of pills actually came with a warning of contraceptive activity as a side effect. Women began using the pill for contraception and the next year an oral contraceptive was approved.

Inventors have created tremendous amounts of change for ourselves. Their history suggests that inventions are hard to predict and figuring out who the creators will be is even more difficult. Great inventors are similar to artists. Where each inventor has a unique persona. Their eclectic personalities create the initial conditions for their work to impact society.

Flight

Wright Brothers making their first flight. Source: Library of Congress

Flight was developed from directed trial and error. It was only until the late 19th century that enough experiments had been conducted with hot-air balloons, gliders, and kites that heavier-than-air flight was even a remote possibility. The progression led to Orville and Wilbur Wright to collect this information and implement the first airplane. Using research Wilbur was able to receive from the Smithsonian Institute, the brothers conducted and replicated past research. On December 17, 1903, at 10:35 a.m., after eight years of work, Orville Wright took flight for a historic twelve seconds. He covered 37 meters of ground and 152 meters of air space. This first flight got little attention where only a few field experts initially appreciated the work. After two years of improving designs, in 1905, the Wright Flyer III was inadvertently seen by the public. News spread slowly over the years with heavy skepticism in Europe where the brothers were called bluffers. The brothers focused on obtaining a patent before building a business. This likely was their critical mistake. Customers were unwilling to pay for a plane without a demonstration. The Wright brothers were unwilling to give a demonstration without a signed contract and a patent. The new Wright Company spent most of its efforts in patent litigation rather than commercialization. Individuals like Glenn Curtiss focused on selling products where his company actually still exists to this day.

Assembly line

Ford Motor Company’s Model T assembly line. Source: Library of Congress

The assembly line emerged from evolution of various ideas in production culminating into Henry Ford’s work to make motorized cars accessible to the masses in the early 20th century. As an amalgamation of ideas, the assembly line was no breakthrough, but was operated on by various selective forces in a wide set of industries. The evolutionary nature of these inventions emerges from learnings aggregated over time. Henry Ford’s borrowed ideas from the meat-packing houses of Chicago of the late 1860s and an automated grain mill being run by Oliver Evans to create a production process with continuous flow. Just like a slaughterhouse, a car was moved across a line where each worker conducted a specific task toward completion. This division of labor reduced a complex job into a series of simple tasks. The evolution from custom production to mass production.

In 1913, the first moving assembly line was implemented to make Model Ts. This invention gave the Ford Motor Company an incredible advantage that was soon replicated by others. Unlike the Wright Brothers, Ford used his invention to make products but began building various moats around them such as a brand of reliability and a company culture providing double the minimum wage. These structural advantages has allowed Ford to survive multiple downturns. Without the assembly line, Ford likely is a much smaller and less important company. Without the assembly line, the US is likely a less influential country. The assembly line helped the US massively improve productivity.

Penicillin

Sir Alexander Fleming. Source: Getty Images

Antibiotics starting with penicillin is one of the most important breakthroughs of the modern era. During the early 20th century, scientists knew antibacterial substances existed but could specifically isolate the factor. In 1928, Sir Alexander Fleming discovered penicillin. It took another decade for penicillin to be purified and clinically tested. Countless lives were saved by it and subsequent antibiotics discovered. Businesses emerged to commercialize these drugs and help make infectious disease relatively irrelevant in the developed world. Penicillin was the beginning of massive economic growth where an essential need for humans, survival was met.

Gunpowder

Proof-of-concept eruptor cannon from the Huolongjing. Source: Wikimedia

In 1044, Chinese alchemists accidentally invented gunpowder by mixing potassium nitrate, charcoal, and sulfur. Rather than discovering an elixir for immortality, they invented a potion for war. Slowly, gunpowder moved to Europe forever changing the structure of society. With a gun, the lowliest person had similar power as a general. War became industrialized accelerating colonization of the world by the West.

Ultrasound

Paul Langévin used ideas from the Curie lab in 1880 to create an ultrasonic system to detect submarines. The system used beams of focused, high frequency sound waves that evolved into sonar. The invention helped drive victory for Allied forces in WWII. Another inventor, Ian Donald, began exploring how to apply sonar to medicine particularly in gynecology and obstetrics, his expertise. While serving in the British Royal Air Force, Donald was exposed to the use of sonar to detect U-boats. In the 1950s, Donald moved to Glasgow and began collaborating with medical researchers. In 1957, the group used a ceiling mounted ultrasound scanner to examine a woman previously diagnosed with gastric cancer. Their initial test actually reclassified the woman’s cancer to an ovarian cyst. In 1958, The Lancet published the team’s work of the first ultrasound images of a fetus at 14 weeks gestation. Several companies were funded subsequently leading to the development of the Automatic Scanner that was able to move in 3 dimensions with a joystick. In 1963, the research team built a smaller box attached to a hinged wall allowing the device to swing over a patient called the Diasonograph. By the 1970s, the Diasonograph and related techniques became adopted as the clinical standard to measure fetal developments during pregnancy. The ultrasound made the fetus visible and became an important part of the medical practice.

Velcro

Georges de Mestral — inventor of Velcro. Source: Pinterest

Between 1948 and 1957, Swiss inventor Georges de Mestral noticed how frequently cockleburs would stick to his clothing. Using a microscope, he discovered that cockleburs had thousands of ends tipped with tiny hooks. This observation led Mestral to invent velcro, a contraction of the French phrase velvet crochet translating to velvet hook, ultimately patenting it in 1955 — one strip with millions of tiny hooks and another strip with millions of loops. This property made velcro a fast-acting fastener. Mestral opened several shops across Europe and the US. He did not have as much success as the companies coming after him — public perception of quality was low and only until the use of velcro for the space missions, did the product begin to gain traction. By then, Mestral’s patent had expired and other businesses began commercializing the technology.

Electron microscope

The electron microscope was first invented by Ernst Ruska and Max Knoll in 1931. The device provided exquisite view of small objects using electron beams instead of light rays. Ruska’s and Knoll’s invention was only made possible by the theoretical work of George Airy who discovered that light is limited by its wavelength. By 1935, the electron microscope had surpassed the resolution of optical microscopes. Work was done to reduce the damaging effects of electron beams on specimens and companies such as the Radio Corporation of America began commercializing the invention.

Transistor

One of the first transistors in development. Source: CED

The transistor is the key driver of success for the personal computer and the Internet. The device is made up of a semiconductor with multiple electrical contacts. The transistor is used as an amplifier, detector, and a switch forming the basis for modern electronics.

Western Electric became interested in electronic amplification especially as a way to strengthen telephone signals. As a result, one of Western Electric’s subsidiaries, Bell Laboratories, began research on transistors. Led by three individuals, John Bardeen, Walter Brattain, and William Shockley, experiments were conducted to create a device that produces a signal with a greater power than its input. Ultimately, the team publicized the first point-contact transistor in 1947 and subsequently were awarded a Nobel Prize in Physics in 1956.

William Shockley formed a company in Mountain View, CA to be close to his mother. Shockley found support from his old mentor, Arnold Beckman, where Shockley Semiconductor Laboratory was a subsidiary of Beckman Instruments. Eight individuals left Shockley’s business to form Fairchild Semiconductor spawning businesses such as Intel, Kleiner Perkins, and Sequoia Capital. In Japan, two inventors from Sony, Ibuka and Morita licensed the transistor to introduce the first Sony transistor radio (TR-55) in 1955. From the transistor, businesses like Intel and Sony built incredible products that transformed society and create enduring companies.

Cyclotron

The cyclotron allows physicists to study the structures of atoms. Invented in 1929/1930 by Ernest Orlando Lawrence at the University of California, the cyclotron was based on Ernest Rutherford’s work on the nucleus of the atom. During the 1920s, other projects had unsuccessful attempts to accelerate alpha particles and electrons. A consensus emerged that a million volts would be required to accelerate these particles with high-enough energies to penetrate an atomic nucleus. At the time no insulator existed to deal with voltages at these levels. However, in 1929, Lawrence read a German article describing a linear accelerator that passed ions through 2 sets of electrodes with the same voltage and increased the ions’ energy concomitantly. With this, Lawrence discovered that spacing electrodes and using alternating electrical fields would accelerate subatomic particles and bring them to sufficient energies. To increase the energies, the spacing between the electrodes would need to be increased leading to successive development of larger and larger cyclotrons starting at 25 centimeters in 1931 that produced one million electronvolt particles. With support from the Research Corporation and the Rockefeller Foundation, Lawrence’s team created larger cyclotrons and by 1939, built a 152 centimeter cyclotron that had applications in medicine, war, and fundamental research.

Computer-aided design

Computer-aided design (CAD) transformed manufacturing. CAD simulates geometric objects on a computer allowing a design to realistically transform an object. CAD evolved from the 1960s IBM Drafting system. Overtime, these systems empowered designers to automate calculations and drafting converging these multiple tasks into one role. During the 1970s and 1980s, the power and accessibility of personal computers began to grow driven by the development of the integrated circuit. This drove the use of CAD systems across multiple industries in particular the US automotive industry, which had been struggling with competition from Japan and high manufacturing costs. By the 1990s, American car companies recovered and reported record breaking profitability numbers. Companies such as Leap and Autodesk drove the commercialization of this invention building incredible businesses.

Steel

Bessemer converter — one of the first devices for the mass production of steel. Source: Wikimedia

The modern world is built on steel. In 1857, Henry Bessemer invented the first large-scale method to produce steel — the Bessemer converter. Up until that time, steel had a few specialty uses but by the 1880s Bessemer work led to mass production of steel to help build skyscrapers and container ships. Bessemer’s invention infused ordinary air, then later pure oxygen after 1929 when the Linde-Frankl process was invented to separate oxygen in air from other elements inexpensively, into molten metal burning off excess carbon and creating steel. By releasing previous constraints, Bessemer allowed new structures to be built and existing ones to become safer. With a patent, Bessemer began commercializing the invention. After visiting Bessemer, Andrew Carnegie invested in a new mill, Edgar Thomson Steel Works, in 1875. With this invention, Carnegie Steel brought steel prices down more than 10x and drove production growth about 100x.

Tuberculosis vaccine

Doctor working on UNICEF’s tuberculosis campaign in Ceylon (now Sri Lanka) in 1950. Source: The Historical Medical Library of The College of Physicians of Philadelphia

In 1921, two French microbiologists, Albert Calmette and Camille Guérin, invented the first vaccine to prevent tuberculosis using an avirulent strain of bovine tuberculosis bacilli. Tuberculosis was discovered to be caused by Mycobacterium tuberculosis by Robert Koch in 1882. Previously known as consumption, active tuberculosis (most infected people have latent tuberculosis) inflames and damages a person’s lungs and spreading to other tissues leading to a slow death. Koch discovered a factor he called tuberculin that was released by the bacterial strain eliciting a reaction in people exposed to or suffering from tuberculosis. With this work, around 1906, researchers discovered that people exposed to tuberculosis were acquiring resistance. Along with work on declining virulence of the strain and the role of the immune system in resistance, Calmette and Guérin cultured avirulent bacteria from cattle and stimulated immunity and antibody production without causing tuberculosis in cows. By 1921, the two inventors developed an avirulent strain that was harmless in humans but provided protection, which they called Bacillus Calmette-Guérin (BCG). In 1922, newborns from the Charité Hospital in Paris began receiving the BCG vaccine gaining widespread use across the world by the 1930s. However, use in the US and England only started in the 1950s due to concerns of the vaccine being composed of live bacteria unlike other vaccines. The inventors’ work was incredibly valuable in reducing the spread of tuberculosis and like most vaccines and public health inventions, the tuberculosis vaccine drove high levels of economic growth over the last century.

Radio

The invention of radio was based on the discovery of Hertzian waves. In the US, Reginald Fessenden, a Canadian immigrant, was building a radio system to broadcast weather forecasts for the US Weather Bureau. In Europe, Guglielmo Marconi was working on wireless transmission as well. Marconi was granted a patent in 1896 and began development in London after little interest in his home country of Italy. Unlike Marconi, Fessenden had conviction that continuous waves would provide the power to handle voice communication — he was right. Working out of Rock Point, Maryland, Fessenden submitted his first patent in 1902, had General Electric produce an alternator for his radio transmitter, and raised $1M to create the National Electric Signaling Company. On Christmas Eve of 1906, Fessenden and his team announced their first radio broadcast — alerting ships across the Atlantic coast, playing phonograph recordings, and wishing listeners a Merry Christmas. Both Fessenden’s and Marconi’s commercial work ended up not reaching their potential and being wounded by the financial panic of 1907. Regular radio broadcasts were hard to implement due the lack of stations and infrastructure. It took WWI to develop radio for military telecommunications allowing Radio Corporation of America to pool patents together and bring radio to the masses.

Plastic

Blue plaque for Alexander Parkes — the first inventor of a man-made plastic. Source: Wikimedia

Like radio, the invention of plastic was being worked on concurrently transforming almost every industry and society. The first human-made plastic led the way for the modern materials revolution of the 20th century. In the 1860’s, the billiard company Phelan and Collender offered a prize of $10K to anyone who could produce a substance that would be an inexpensive substitute for ivory. Building on top of previous research, John Wesley Hyatt discovered that adding camphor to nitrocellulose created white materials he named celluloid that could mimic ivory. Hyatt won the prize and began to commercialize his invention. Hyatt realized that celluloid was too easily deformable to be useful to make billiard balls. He ultimately sold his company to GM. In England, another inventor, Alexander Parkes was working at a company called Elkington & Company where he actually was the first to invent the first human-made plastic. In 1856, Parkes patented his work calling it Parkesine. His colleague, Daniel Spill, later improved the material in the form of Xylonite.

The invention of plastic led to more and more work in synthetic polymers and materials science in general. Companies such as DuPont were formed based on these inventions creating new materials beyond plastics. These new materials allowed new products previously unattainable to be developed such as aircraft and high-performance implants.

Nuclear reactor

In 1943, the construction of a nuclear reactor at Oak Ridge National Laboratory was essential for the efforts during WWII. The reactor was required to produce significant amounts of plutonium initially to create an atomic bomb and later generate usable amounts of energy. The invention was formed on the basis of a 1942 discovery by Enrico Fermi showing the first self-sustaining, controlled atomic chain reaction. Within a nuclear reactor, the chain reaction of fission is controlled to make the reactions proceed relatively slowly. The heat generated is used to boil water and create steam that turns turbine generators thereby producing electricity.

The initial use case of the reactor at Oak Ridge was to separate plutonium from uranium. In a little under a year, in March 1944, the Oak Ridge proof-of-concept produced several grams of plutonium. The laboratory still produces radioisotopes, now for research and medical applications. And the plutonium used in the first atomic bombs came from reactors at the Hanford Engineer Works. Nuclear reactors led the the creation of nuclear warfare but also produced new forms of electric power generation and medicine.

Engineered insulin

Founders of Genentech — Herb Boyer and Bob Swanson — the company who first cloned insulin. Source: Kleiner Perkins

The first engineered and artificially manufactured human insulin sold was Humulin. The product was used as a treatment for people with diabetes. Until the 1920s, the standard of care for diabetes was a diet with low carbohydrates until the work of Frederick Banting and Charles Best around purifying insulin from animal pancreases (winning the 1923 Nobel Prize in Physiology or Medicine) gave diabetics another treatment option. Diabetics began receiving regular doses of insulin sourced from large facilities isolating insulin from cattle and pig pancreases; however, this form of insulin induced an allergic reaction in 5% of diabetics and supply was constrained by the number of animals raised.

With the invention of molecular cloning, Robert Swanson and Herbert Boyer founded Genentech in 1976. In 1978, the company in particular David Goeddel cloned the first synthetic human insulin in bacteria. Partnering with Eli Lilly, the FDA approved the first human insulin product called Humulin on May 14, 1982.

The invention used two strains of the bacterium, Escherichia coli, to produce A and B chains separately. After purifying each chain, they are combined to produce insulin. During the research, synthetic human insulin was shown to be biologically equivalent and most importantly not produce adverse reactions and no longer being constrained by the number of animals to slaughter. This work not only helped diabetics around the world but showed the power of molecular biology to produce high value drugs leading to the biotechnology industry.

Internal combustion engine

The four cycles of a standard internal combustion engine. Source: Getty

The most common engine in automobiles today is the internal combustion engine (ICE). For centuries the engine was under development and the ICE emerged from an evolution of designs. Beyond the invention, fuel was consumed externally limiting the potential settings an engine could be used in. However, ICEs burned fuel internally enabling the invention of cars and airplanes. The first engines were extremely heavy — 10 to 15 lbs per output horsepower, as opposed to 1 to 2 lbs today, slow — 1K or fewer revolutions per minute or less versus around 5K today, and inefficient. WWI drove improvements in these engines.

An inventor at the center of the development of internal combustion engines was Harry Ricardo. Spending a lifetime developing engines first with a coal-fired steam engine for his bicycle as a child to motorcycles during his studies at Cambridge University to his two-cycle engine called the Dolphin used in automobiles and boats. In 1916 during the war, the British government requested a tank engine that would not billow smoke exhaust from the business Ricardo’s was working for. The company came up with a design that used air circulation around the carburetor and within the engine to cool the fuel. A year later, Ricardo founded his own engine company and came up with a series of breakthroughs such as the the toluene number to measure fuel detonation eventually replaced by the octane number and new piston designs to make engines more efficient. The internal combustion engine created new products and enduring businesses such as Mercedes-Benz and Aston Martin.

Battery

The first usable battery was invented by Gaston Planté in 1859. The battery used electrodes made of lead and lead oxide within a sulfuric acid electrolyte. This work ultimately led to the nickel-iron alkaline battery, the first portable power source that enabled new types of motors and devices. In the early 1900s, most cabs in New York City were actually powered by electric motors. The first alkaline battery was invented by Waldemar Jungner in 1899 with Thomas Edison inventing one in 1901. Edison had more resources and was able win a patent war & commercialize the invention at a larger scale. The components, iron and nickel, for Edison’s battery were not available in high enough purities for battery use; as a result, new factories were set up to prepare the materials at the proper purities. In 1904, the Edison batteries entered the market but were recalled in 1905 due to a defect in an electrode and released again in 1910. The battery lost out to the internal combustion engine to become the standard in automobiles but ended up changing the energy industry, consumer devices, and communications.

Inventors are a small group of individuals who solve important problems. Whether or not they build important businesses is up to them.

Special thanks to Bob Langer, Adam Cohen, Omri Gottesman as well as others kind enough to review and provide feedback on this piece.

Tectonic Therapeutic: Protein Engineering and Fc-relaxin for Hypertension

Joshua Elkington — Fri, 31 Jan 2025 01:16:23 GMT

Axial: https://linktr.ee/axialxyz

Centessa Pharmaceuticals: Asset-Centric Business Model

Joshua Elkington — Fri, 31 Jan 2025 01:15:58 GMT

Axial: https://linktr.ee/axialxyz

Synthetic integrin antibodies discovered by yeast display reveal αV subunit pairing preferences with β subunits

Fri, 31 Jan 2025 01:11:55 GMT

Axial: https://linktr.ee/axialxyz

The research paper uses yeast display to generate a panel of highly specific and functional antibodies against key integrin heterodimers. Beyond the technical advancements in antibody discovery, the paper also provides new insights into the subunit pairing preferences of the integrin αV subunit, a topic that is crucial to understanding the roles and functionalities of this receptor family.

Integrins are a family of cell surface receptors that mediate crucial cellular processes, including cell adhesion, migration, and intracellular signaling. They are heterodimeric glycoproteins, consisting of α and β subunits, with diverse combinations allowing for specific interactions with various extracellular ligands. Eight members of this large family of 24 integrins recognize the tripeptide Arg-Gly-Asp (RGD) motif, which is present in a wide array of extracellular matrix proteins and cell surface proteins. These so-called RGD-binding integrins, including αVβ3, αVβ5, αVβ6, αVβ8, and α5β1, are implicated in numerous physiological and pathological conditions, making them attractive targets for therapeutic interventions. Despite the similar RGD-binding motif, each integrin demonstrates distinct ligand specificities and cellular functions, highlighting the need for selective targeting.

The authors begin by addressing a persistent challenge in the field of integrin research: the lack of highly specific antibodies that can directly block the interaction of integrins with small RGD-containing ligands. Many antibodies to date have been shown to interact outside the ligand-binding pocket, impacting only the binding of large macromolecules. In contrast, this study successfully engineered antibodies that compete with small RGD peptides for integrin binding, demonstrating their specificity and potential for therapeutic development. They achieve this through the use of a synthetic yeast-displayed Fab antibody library, a technology that offers several advantages over traditional immunization-based methods. Namely, the library is capable of generating antibodies against both human and mouse antigens, and is less subject to the self-tolerance mechanisms of immune systems, expanding the library's sequence space and increasing the likelihood of discovering high-affinity antibodies.

The research team employed a combination of magnetic-activated cell sorting (MACS) and fluorescence-activated cell sorting (FACS) to select yeast clones displaying Fabs that specifically bound to the target integrin heterodimers. They targeted five RGD-binding integrins: αVβ3, αVβ5, αVβ6, αVβ8, and α5β1. Their careful screening strategy included positive selection with purified integrin ectodomains and negative selection with untargeted integrins and a poly-specificity reagent to avoid cross-reactive antibodies. This meticulous approach resulted in the identification of 11 antibodies, all exhibiting high specificity and affinity for their respective target integrins. Of these 11, six had a ligand-mimetic R(G/T/L)D motif in their complementarity-determining region (CDR3), a structural feature that allows for more efficient binding competition with RGD-containing ligands.

The next stage of the work is focused on the characterization of these antibodies. Initially, the team performed immunofluorescence staining on cells expressing intact human and mouse integrins, assessing antibody specificity and cross-reactivity. The data showed that the vast majority of these antibodies were highly selective for the integrins that were used in the selection process, while also displaying significant binding to their mouse orthologs. The binding affinity was further characterized through surface plasmon resonance (SPR). Here, purified integrin ectodomains were exposed to immobilized antibodies to determine association and dissociation rates, and dissociation constants (Kd) ranging from sub-nanomolar to double-digit nanomolar affinities were determined. These results highlighted the high-affinity binding of all 11 antibodies toward their targeted integrin, as well as their remarkable selectivity.

To further examine the function of these antibodies, the team conducted a fluorescence polarization (FP) competition assay with fluorescent RGD-mimetic peptide ligands. This analysis confirmed that all six RGD-motif-containing antibodies indeed compete with peptide ligands for integrin binding. It also revealed significant cross-reactivity of some of the antibodies towards untargeted integrins, albeit with several magnitudes lower affinity compared to the target integrins. This particular result confirms that many of these antibodies are capable of directly engaging the RGD-binding pocket of the integrins and effectively block the binding of small-molecule antagonists. Importantly, these results support the hypothesis that incorporation of a ligand-mimetic motif in the antibody’s CDR3 loop can facilitate this mechanism.

Furthermore, the authors examined the apparent affinity of these antibodies on cell surfaces, an assessment that reflects avidity and is most pertinent to biological systems. Using flow cytometry without washing, the group measured antibody binding to cell surface integrins and determined dissociation constants for each of the RGD-mimetic antibodies. The results were notable for a strong avidity effect; antibodies in the IgG format bound with a significantly higher effective affinity than Fab fragments of the same antibody, demonstrating that IgG’s two antigen-binding sites lead to enhanced binding to cell surfaces. They also demonstrated the efficacy of the blocking activity of these antibodies, especially with the use of a cell adhesion assay. In a similar vein, this analysis showed that the RGD-motif-containing antibodies were effective at blocking cell attachment to fibronectin.

One of the most important discoveries described by the authors is related to the binding preferences of the αV subunit to the five different β subunits with which it can associate. This is highly relevant to the biology of αV integrins given that it determines the specific roles of each heterodimer in cellular signaling, development, and disease. By transiently expressing αV with different ratios of each of the five β-subunits, the authors used their panel of antibodies to study the preferences of αV for the different β-subunits. By titrating the concentration of a given β subunit relative to the αV subunit, the authors could precisely map the preference of each of these heterodimers. Through these experiments, the authors revealed a consistent preference hierarchy for αVβ interactions: β6=β8>β3>β1=β5. This finding provides important new insights into how these heterodimers are formed in cells and has significant implications for understanding their distinct roles in biology. The authors also showed that this hierarchy was observed not just in transfected cells, but also in endogenous integrins expressed on tumor cell lines.

Finally, the authors compare these antibodies to previously reported antibodies. They note that most antibodies that have been reported to date do not directly bind to the ligand-binding pocket, and are therefore incapable of competing with small molecules that block integrin activity. Moreover, the only antibodies known to block small molecule binding to integrin are those that target αIIbβ3. The antibodies described in this manuscript represent novel approaches, showing that antibodies that directly compete with small molecule ligands are obtainable via engineered Fab libraries. These results establish the potential of these antibodies for integrin targeting.

In summary, this paper provides significant advances in several important areas of integrin research. The generation of a new panel of antibodies, targeted against specific RGD-binding integrins and capable of blocking the activity of small molecule RGD-mimetic antagonists, represents a key advancement that will facilitate further research in this space. Furthermore, by mapping the subunit pairing preferences of the integrin αV subunit, this work gives insight into the biological activity of these critical receptors. The methods described in this paper are rigorous and thorough, using a combination of in vitro, cell-based, and in vivo assays to characterize their newly developed antibodies. The conclusions are well supported by the data and contribute significantly to our understanding of integrin function.

Moving forward, this study has created several opportunities for the use of these antibodies in biological studies and for potential therapeutic development. The authors acknowledge the therapeutic potential of integrin inhibition, and the data presented here suggest that their antibodies are well suited for this application. Their successful development of antibodies that block the binding of small molecules and proteins within the integrin ligand-binding pocket opens new avenues for developing more effective and specific therapeutics. The authors, recognizing the value of these reagents, are now distributing these antibodies in collaboration with the IPI and Addgene, further bolstering their impact on the research community.

This work on synthetic integrin antibodies presents a valuable resource for the research community, providing tools that directly compete with small-molecule antagonists and elucidate subunit pairing preferences within the integrin family. This research pushes forward our comprehension of the intricacies of integrin biology and unlocks new possibilities for therapeutic interventions.

https://pubmed.ncbi.nlm.nih.gov/38328192/

Real-Time Multistep Asymmetrical Disassembly of Nucleosomes and Chromatosomes Visualized by High-Speed Atomic Force Microscopy

Fri, 31 Jan 2025 01:10:52 GMT

Axial: https://linktr.ee/axialxyz

This research paper presents a groundbreaking investigation into the real-time dynamics of nucleosome and chromatosome disassembly using high-speed atomic force microscopy (HS-AFM). The authors, led by Bibiana Onoa and Carlos Bustamante, provide a detailed, frame-by-frame visualization of these fundamental chromatin structures as they unwind and dissociate, revealing a multistep asymmetrical process previously only inferred through ensemble and indirect measurements. Their work significantly advances our understanding of how DNA access is regulated through these dynamic processes, and how these processes are altered by the presence of linker histones like H1.

The study's central innovation lies in its application of HS-AFM to visualize the step-by-step changes in nucleosome and chromatosome morphology with sub-second time resolution. This is critical because these structures are highly dynamic, and their disassembly is thought to be a crucial regulator of genome expression and replication. Using this approach, the team is able to track individual nucleosomes and chromatosomes deposited on a mica surface and monitor their transformation from intact complexes to their subnucleosomal products (hexasomes, tetrasomes, and disomes). They carefully controlled the experimental conditions by using magnesium-free buffers supplemented with polyamines to maintain the molecules’ physiological integrity and prevent interactions with the charged surface of the mica. By tracking the changes in the nucleosome core particle (NCP) volume and DNA arm angles in real time, the authors have established a highly detailed picture of their dynamic structural transformations.

One of the key findings of this study is the revelation of the sequential and asymmetrical nature of nucleosome disassembly. The team's work clearly demonstrates that the removal of histone dimers during nucleosome unwrapping is not a symmetrical process. Rather, they show that the H2A-H2B heterodimer located on the rigid side of the 601 nucleosome positioning sequence (NPS) dissociates first, causing a significant change in the short DNA arm angle and a slight decrease in the histone core volume. This initial dimer dissociation is followed by the ejection of the second H2A-H2B heterodimer on the flexible side of the NPS, which prompts further changes in the long DNA arm angle and additional volume loss. This finding strongly corroborates models based on earlier studies by sm-FRET, time-resolved SAXS, and molecular dynamics.

Further analysis of the real-time data revealed that the formation of a tetrasome, after the dissociation of the two H2A-H2B heterodimers, marks a significant change in the disassembly pathway, slowing the rate of disassembly considerably. The tetrasome, consisting of the (H3-H4)2 complex, was found to be particularly stable. This observation provides a compelling reason for the longevity of tetrasomes in vivo, as has been previously documented in other studies involving RNA polymerase activity. The team also reports a dynamic, constrained diffusion of the tetramer along the dyad of the NPS, suggesting that it can slide along the DNA, potentially allowing it to find an optimal position for subsequent nucleosome assembly and function. This dynamic tetramer behavior may also be critical to how nucleosomes are assembled and positioned on the genome.

The presence of linker histone H1 was also found to dramatically affect the disassembly of the nucleosome. When H1 is present, it transforms the nucleosome into a more compact structure, a chromatosome, by stabilizing the linker DNA regions at the entry-exit site. The binding of H1 dramatically increased the lifetime of the complex before the first heterodimer dissociation, effectively slowing down disassembly by a factor of two. This stabilization of DNA also constrained DNA unwrapping and, thus, altered the pathway of heterodimer release. Interestingly, when the first dimer is released from a chromatosome, the linker histone frequently docks reversibly back into the core before a second dissociation step occurs. This rebinding of H1 is not observed in nucleosomes and further demonstrates the impact of the linker histone. Finally, the dissociation of chromatosomes, like that of nucleosomes, is ultimately a sequential and asymmetrical process that leaves a tetrasome as its most stable intermediate.

To analyze the dynamic data, the researchers implemented a neural network for automated segmentation of the molecules. By training this network with manually segmented nucleosomes and PANS, the team was able to differentiate between the protein and DNA components, thus tracking the NCP volume changes and DNA arm movements automatically. Their approach was crucial for the high-throughput analysis required by the dynamic data obtained by HS-AFM and represents a powerful tool for the analysis of such data. It demonstrates how a combination of machine learning and innovative microscopy can be leveraged to advance our understanding of biological structures.

The methodology employed by Onoa and colleagues is noteworthy for its use of purified, well-defined nucleosomes and chromatin components. They carefully reconstituted these molecules and characterized their purity through gel electrophoresis, reducing the possibility of artifacts that can arise from the use of less defined samples. They also performed control experiments, like crosslinking the molecules to prevent unwrapping, and demonstrated that the observed unwrapping is driven by intrinsic dynamics, rather than induced by interactions with the AFM tip or surface. This careful control and characterization of the experimental system is vital for making robust and reproducible observations.

The implications of this research are far-reaching for our understanding of genome regulation. The work underscores the dynamic and asymmetrical nature of nucleosome disassembly and how this is affected by the linker histone H1. These insights suggest that the dynamics of nucleosome disassembly could play a key role in regulating gene expression. The authors suggest that the polarity of the unwrapping as well as the timing of the dimer release are important determinants of transcriptional access to the DNA. The finding that the tetrasome is a particularly stable intermediate that can diffuse along the DNA adds a new layer of complexity and has implications for both nucleosome assembly and its role in transcription elongation.

Moreover, the study's use of HS-AFM opens new avenues for research in this field. The ability to directly visualize the structural dynamics of nucleosomes and their disassembled products in real time provides a powerful tool for investigating the effects of different protein interactions, epigenetic modifications, and enzyme activities on nucleosome structure and stability. This paper provides a firm foundation for further exploration into the complex and dynamic world of chromatin.

In conclusion, this study provides a significant contribution to the field of chromatin biology by leveraging the strengths of HS-AFM, an innovative data processing pipeline, and well-defined nucleosome and chromatin components. The team's findings demonstrate the sequential, asymmetrical nature of nucleosome disassembly, the importance of tetrasomes in the chromatin landscape, and the profound impact of linker histone H1. The methodologies, analysis, and interpretations of this research are a significant advance in our understanding of chromatin structure and function and will be invaluable for exploring the dynamic nature of this fundamental biological structure in the future.

https://pubs.acs.org/doi/10.1021/acscentsci.3c00735

Rapid proteome-wide prediction of lipid-interacting proteins through ligand-guided structural genomics

Sun, 05 Jan 2025 02:50:26 GMT

Axial: https://linktr.ee/axialxyz

The research paper, "Rapid proteome-wide prediction of lipid-interacting proteins through ligand-guided structural genomics" by Chou et al., details the development and implementation of a novel bioinformatic tool called SLiPP (Structure-based Lipid-interacting Pocket Predictor). This tool is designed to rapidly and accurately identify proteins that interact with lipids within a given proteome, offering a significant advancement in the field of lipid biology.

Lipids, crucial components of cellular membranes and signaling pathways, are implicated in various diseases, including cancer and infectious diseases. Understanding the interactions between lipids and proteins is crucial for unraveling their roles in cellular processes and for identifying potential therapeutic targets. However, identifying lipid-binding proteins has been challenging due to the lack of conserved amino acid motifs and the limitations of current experimental techniques.

SLiPP addresses this challenge by leveraging the growing availability of protein structures, both experimentally determined and computationally predicted using AlphaFold. The tool employs a machine learning algorithm trained on a dataset of lipid-binding and non-lipid binding pockets extracted from experimentally determined protein structures in the PDB database. This training enables SLiPP to distinguish between pockets that are likely to bind lipids and those that are not, based on their physiochemical properties.

The development of SLiPP involved several key steps. First, the researchers curated a dataset of protein structures with known ligands, including lipids and non-lipids. They then used the fpocket software to identify potential ligand-binding pockets within these structures, which were further categorized as lipid-binding pockets (LBPs), non-lipid binding pockets (nLBPs), and pseudo-pockets (PPs), representing unliganded pockets identified by fpocket.

The researchers performed principal component analysis (PCA) on these pockets using 17 physiochemical descriptors provided by fpocket. The PCA revealed a clear separation between LBPs and nLBPs, particularly along the second principal component, which was dominated by hydrophobicity-related properties. This separation suggested that hydrophobicity plays a significant role in lipid binding, as anticipated due to the hydrophobic nature of lipid tails.

Next, the researchers evaluated six different machine learning algorithms to build a classifier capable of accurately distinguishing LBPs from other types of pockets. They found that the random forest (RF) algorithm exhibited the best performance, achieving a high F1 score and accuracy. However, the initial model suffered from low sensitivity due to the highly imbalanced nature of the dataset, with a much larger number of PPs compared to LBPs and nLBPs.

To improve the sensitivity of the classifier, the researchers explored two strategies. First, they tested different ratios of PPs to LBPs in the training dataset, finding that a dataset with 20-fold more PPs than LBPs performed optimally. Second, they fine-tuned the hyperparameters of the RF algorithm to further enhance its performance, although the improvements were marginal. Ultimately, the researchers prioritized sensitivity over precision, opting for a model that produced more potential hits for further validation.

The performance of the optimized SLiPP model was rigorously assessed using various test datasets, including an independent test dataset, a dataset of apo (ligand-free) structures, and a dataset of AlphaFold-predicted models. The model performed remarkably well on the independent test dataset, achieving an accuracy of 96.8% and an F1 score of 0.869. However, its performance on the apo and AlphaFold datasets was slightly lower, likely due to the inherent limitations of fpocket in accurately identifying ligand-binding pockets in the absence of ligands. Despite this limitation, the high precision of SLiPP makes it a powerful tool for discovering novel lipid-binding proteins.

To demonstrate the utility of SLiPP in real-world applications, the researchers applied it to predict lipid-binding proteins in the proteomes of three well-annotated organisms: Escherichia coli (E. coli), Saccharomyces cerevisiae (yeast), and Homo sapiens (human). They used AlphaFold-predicted structures for these proteomes, removing signal peptides and filtering out proteins with low confidence scores or fewer than 100 amino acids.

In the E. coli proteome, SLiPP identified 159 putative lipid-binding proteins, representing 3.6% of the proteome. Notably, many of the top hits were already annotated LBPs, providing confidence in the accuracy of the predictions. Additionally, SLiPP highlighted several uncharacterized proteins as potential lipid binders, suggesting new avenues for research.

Similarly, in the yeast proteome, SLiPP predicted 273 hits (4.5% of the proteome). Gene ontology (GO) enrichment analysis of these hits revealed a significant overrepresentation of lipid-related processes, further supporting the validity of the predictions. Interestingly, the analysis also pointed to a potential role of lipids in cation homeostasis and ion transport, suggesting an interplay between these processes.

In the human proteome, SLiPP identified 935 putative lipid-binding proteins (4.6% of the proteome). GO enrichment analysis again confirmed the enrichment of lipid-related processes among the hits, with a focus on transport functions. The researchers also noted that several hits were associated with diseases, highlighting the potential of SLiPP for identifying drug targets.

One compelling example of SLiPP's potential lies in its prediction of GDAP2, a protein implicated in spinocerebellar ataxia, as a lipid-binding protein. SLiPP identified a large hydrophobic pocket within the CRAL-TRIO domain of GDAP2, suggesting a potential lipid-binding site. Interestingly, pathogenic mutations in GDAP2 co-localize with this predicted pocket, suggesting a link between lipid binding and disease pathogenesis.

To understand the underlying features driving SLiPP's predictions, the researchers analyzed the importance of the 17 pocket descriptors used by the classifier model. They found that hydrophobicity-related features, particularly the hydrophobicity score and mean local hydrophobicity density, were the most important factors in distinguishing LBPs from other pockets. This finding reinforces the importance of hydrophobicity in lipid binding and explains the tendency of SLiPP to misidentify heme-binding pockets as LBPs due to their shared hydrophobic characteristics.

In conclusion, SLiPP represents a significant advancement in the field of lipid biology by providing a rapid and accurate method for identifying lipid-binding proteins on a proteome-wide scale. Its reliance on physiochemical properties rather than sequence homology allows for the discovery of novel lipid-binding domains and expands our understanding of lipid-protein interactions. While SLiPP has certain limitations, particularly its inability to detect lipid binding sites formed upon protein oligomerization, its high precision and ability to predict lipid-binding proteins in poorly studied organisms make it a valuable tool for advancing research in this field. The researchers anticipate that SLiPP will be particularly useful in identifying new drug targets for bacterial infections by revealing lipid-binding proteins in pathogenic bacteria that rely on host lipids for survival.

https://www.biorxiv.org/content/10.1101/2024.01.26.577452v1

The role of Claudin-1 in fibrotic disease

Sun, 05 Jan 2025 02:46:46 GMT

Axial: https://linktr.ee/axialxyz

Fibrosis, a debilitating condition characterized by the excessive accumulation of extracellular matrix (ECM) components, poses a significant global health challenge, contributing to a substantial proportion of deaths, particularly in developed nations. It affects various organs, including the kidneys, liver, and lungs, leading to organ dysfunction and potentially culminating in life-threatening complications like cancer. Despite its widespread impact, current therapeutic options for fibrosis remain limited, emphasizing the urgent need for novel treatment strategies. Recent research has unveiled a promising therapeutic target for fibrotic diseases: Claudin-1 (CLDN1), a protein typically residing within tight junctions and crucial for regulating cell-to-cell adhesion.

In healthy tissues, CLDN1 is sequestered within tight junctions, effectively concealed from the extracellular environment. However, in fibrotic tissues, CLDN1 is overexpressed and aberrantly exposed outside of tight junctions, a phenomenon termed non-junctional CLDN1 (njCLDN1). This exposed njCLDN1 acts as a potent trigger for pro-fibrotic signaling pathways, driving the excessive deposition of collagen and other ECM components, ultimately leading to organ failure.

Numerous studies have established a compelling correlation between elevated CLDN1 expression and the severity of fibrosis across various organs. For instance, in individuals with idiopathic pulmonary fibrosis (IPF), CLDN1 expression escalates with disease progression, signifying its potential role in disease pathogenesis. Similarly, in liver tissues from patients with chronic liver diseases like HBV, HCV, or NASH, CLDN1 expression is markedly elevated, with njCLDN1 levels being particularly pronounced in advanced liver fibrosis. This observation underscores the potential of njCLDN1 as a valuable biomarker for fibrosis progression and a target for therapeutic intervention.

Advanced techniques like scRNA-seq analyses have provided a more nuanced understanding of CLDN1 expression patterns within the intricate cellular landscape of fibrotic tissues. In the context of liver fibrosis, CLDN1 exhibits high expression in a variety of cell types implicated in the fibrotic process, including hepatocytes, cholangiocytes, and activated hepatic stellate cells (aHSCs), the key players in hepatic collagen production. Notably, njCLDN1 is preferentially enriched in a distinct subpopulation of diseased hepatocytes situated at the junction between epithelial and stromal tissues, hinting at its involvement in the cellular transformations that drive fibrosis.

While the precise mechanisms by which njCLDN1 orchestrates the fibrotic cascade are still under investigation, several pivotal pathways have been brought to light. One prominent mechanism involves the activation of pro-fibrotic signaling pathways. njCLDN1 interacts with various cell membrane proteins, including EGFR and ITGA5, triggering downstream signaling cascades like the SRC and MAPK pathways, well-established drivers of fibrosis. Moreover, njCLDN1 appears to exert a modulatory influence on the TGF-β signaling pathway, a central regulator of fibrosis. Although it doesn't significantly impact the canonical SMAD2/3 pathway of TGF-β signaling, njCLDN1 effectively inhibits non-canonical pathways such as AKT, p38, and ERK signaling, known to be involved in fibrosis.

Adding another layer of complexity, njCLDN1 is upregulated by the TNF-α signaling pathway, a critical mediator of inflammation and fibrosis. In vitro experiments have convincingly demonstrated that exposing aHSCs and hepatocytes to TNF-α results in increased njCLDN1 expression, a process mediated by the NF-κB pathway, a downstream effector of TNF-α signaling.

Beyond its role in signaling pathways, njCLDN1 also plays a crucial role in modulating cell-matrix interactions and cell plasticity. Its interaction with ECM components like laminin suggests its involvement in the excessive deposition of ECM components, a hallmark of fibrosis. Furthermore, the association between njCLDN1 expression and alterations in cell plasticity, particularly in hepatocytes and liver progenitor cells, implies its contribution to the de-differentiation of hepatocytes and the proliferation of progenitor cells, processes intimately linked to fibrosis progression.

The therapeutic potential of targeting njCLDN1 has been strongly supported by a series of compelling preclinical studies. One notable study demonstrated that in vivo knockdown of CLDN1 using siRNA technology effectively mitigated liver fibrosis in a mouse model of NASH. Mice treated with CLDN1-targeting siRNAs exhibited significantly reduced collagen accumulation and a lower number of liver nodules compared to their control counterparts.

Further bolstering the therapeutic promise of targeting njCLDN1, monoclonal antibodies specifically designed to neutralize njCLDN1 have shown remarkable anti-fibrotic effects in various preclinical models. For example, in a humanized mouse model mimicking NASH-associated liver fibrosis, administration of a CLDN1 mAb effectively reduced liver fibrosis and suppressed the expression of fibrosis markers. Similar anti-fibrotic benefits were observed in a mouse model of biliary fibrosis treated with the same CLDN1 mAb.

Encouragingly, the therapeutic efficacy of CLDN1 mAbs extends beyond the liver, underscoring its broad applicability. Studies employing a mouse model of kidney fibrosis induced by unilateral ureteral obstruction and a mouse model of lung fibrosis induced by bleomycin have unequivocally shown that CLDN1 mAb treatment significantly attenuates fibrosis in these organs, highlighting its potential as a pan-fibrotic therapy.

These preclinical findings converge to paint a compelling picture: targeting njCLDN1 with precisely engineered mAbs represents a potentially safe and effective strategy for combating fibrosis across diverse organs. Importantly, toxicity studies conducted in non-human primates have revealed that CLDN1 mAbs are well-tolerated and do not disrupt the integrity of tight junctions, reinforcing their safety profile.

In conclusion, CLDN1 has emerged as a highly promising target for the development of innovative anti-fibrotic therapies. Its overexpression and aberrant exposure outside of tight junctions in fibrotic tissues, coupled with its pivotal roles in pro-fibrotic signaling pathways and cellular processes, render it an attractive candidate for therapeutic intervention. The robust anti-fibrotic effects observed in various animal models treated with CLDN1-targeting siRNAs and mAbs strongly support the translational potential of this approach, paving the way for clinical development.

Looking ahead, future research endeavors should prioritize several key areas:

Delineating the precise molecular mechanisms by which njCLDN1 contributes to fibrosis in different organs. A deeper understanding of these mechanisms will enable the development of more targeted and effective therapies.
Optimizing the delivery and efficacy of CLDN1-targeting therapies. This may involve exploring novel drug delivery systems or developing next-generation CLDN1 mAbs with enhanced potency and target specificity.
Initiating clinical trials to rigorously evaluate the safety and effectiveness of CLDN1-targeting therapies in patients with fibrotic diseases. This crucial step will determine the clinical translatability of this promising therapeutic approach.

The successful development of safe and effective therapies targeting njCLDN1 holds immense promise for addressing the substantial unmet medical need in the management of fibrotic diseases. By disrupting the fibrotic cascade and restoring tissue homeostasis, CLDN1-targeted therapies have the potential to revolutionize the treatment landscape for fibrosis, offering hope for improved patient outcomes and a brighter future for those battling this debilitating condition.

Single molecule imaging of the central dogma reveals myosin-2A gene expression is regulated by contextual translational buffering

Tue, 10 Dec 2024 19:59:35 GMT

Axial: https://linktr.ee/axialxyz

This paper by O’Neil Wiggan and Timothy Stasevich details the creation and application of a novel human cell line (eMyo2AGFP) to study the regulation of the MYH9 gene, which encodes myosin-2A, with unprecedented single-molecule precision. The authors achieve this by employing CRISPR-Cas9 gene editing to introduce multiple fluorescent tags into the endogenous MYH9 locus, allowing simultaneous visualization of transcription, translation, mRNA, and mature protein. This innovative approach provides a powerful tool to dissect the central dogma in living cells, moving beyond the limitations of traditional methods.

The introduction effectively establishes the significance of understanding gene expression homeostasis, particularly focusing on the challenges in determining regulatory mechanisms at different levels of the central dogma. The authors highlight the ambiguity surrounding MYH9 regulation, given its dual categorization as both a housekeeping gene (suggesting stable expression) and a gene implicated in cancer (suggesting dynamic regulation). This sets the stage for their innovative approach, emphasizing the need for a system capable of directly imaging all aspects of gene expression with single-molecule resolution.

The results section systematically demonstrates the capabilities of the eMyo2AGFP cell line. The authors meticulously validate the functionality of their tagged MYH9 gene, confirming its proper localization and expression levels comparable to untagged controls. They then demonstrate the ability to resolve individual transcription events, quantify mRNA numbers and localization, and precisely identify and characterize translation sites using both fixed-cell immunostaining and live-cell imaging with anti-FLAG intrabodies. This detailed characterization establishes the cell line's robustness and sensitivity for studying MYH9 expression dynamics.

The study proceeds to investigate MYH9 expression regulation across different cellular contexts. Analysis of MYH9 expression throughout the cell cycle reveals a dynamic relationship between transcription and translation, with a marked decrease in translation efficiency during mitosis, independent of mRNA levels. This observation suggests a cell cycle-dependent regulatory mechanism fine-tuning myosin-2A expression to avoid potential disruptions to cell division. The authors carefully consider and rule out alternative explanations, like changes in cell size, supporting the conclusion of cell cycle-dependent translational regulation.

Further exploration of MYH9 regulation focuses on the impact of actin depolymerizing protein knockdown. The authors observe a dramatic increase in MYH9 transcription upon silencing cofilin and ADF, driven by increased transcription burst frequency and amplitude. They demonstrate that this transcriptional upregulation is largely dependent on Serum Response Factor (SRF), a key regulator of cytoskeletal gene expression. This finding establishes a direct link between disrupted actin dynamics, increased MYH9 transcription, and the potential for detrimental cellular consequences.

Interestingly, the authors uncover a mechanism of translational buffering in response to this transcriptional upregulation. While cofilin/ADF knockdown leads to a significant increase in MYH9 mRNA, the increase in myosin-2A protein is considerably less pronounced. They demonstrate that this buffering effect is achieved by a reduction in the fraction of cytoplasmic mRNA undergoing translation, rather than changes in individual translation site intensity. This suggests a mechanism switching translation on or off for individual mRNA molecules rather than gradually modulating translation rate. The authors rule out this buffering as a general effect, confirming the specificity of the MYH9 regulation.

In contrast to the translational buffering observed with cofilin/ADF knockdown, the response to serum stimulation shows a different regulatory pattern. Upon serum stimulation, MYH9 transcription and translation increase in tandem, enabling rapid actomyosin cytoskeletal reorganization. The authors provide evidence that both SRF-dependent transcription and new protein synthesis are essential for the formation of stress fibers, but not for initial peripheral actomyosin bundling in response to serum, This finding underscores the context-dependent nature of MYH9 regulation, highlighting the adaptability of the system in response to different cellular stimuli.

The discussion section synthesizes the key findings, emphasizing the multi-layered regulatory network governing MYH9 expression. The authors effectively portray MYH9 translational regulation as a crucial homeostatic mechanism, analogous to a thermostat, ensuring appropriate myosin-2A levels. They propose a model where translational buffering acts as a safeguard against excessive myosin activity, preventing detrimental effects. The contrast between the responses to cofilin/ADF knockdown and serum stimulation highlights the context-dependent nature of this regulatory mechanism.

The authors acknowledge limitations of their approach, such as the challenge in detecting dimmer translation sites and the incomplete tagging of MYH9 alleles. They also discuss the potential for future research, including the identification of specific molecular players involved in translational buffering and further exploration of the generalizability of their approach to other cell types and genes. Despite these limitations, they convincingly argue that the eMyo2AGFP cell line represents a significant advance, offering a valuable tool for investigating the complexities of gene regulation. The concluding paragraphs effectively emphasize the broad applicability of this innovative system, highlighting its potential for future studies in diverse cellular contexts. Overall, the paper presents a well-executed and impactful study that significantly advances our understanding of gene expression regulation.

https://www.biorxiv.org/content/10.1101/2024.02.11.579797v1

DNA foundation modeling

Tue, 10 Dec 2024 19:35:43 GMT

Axial: https://linktr.ee/axialxyz

The development of foundation models for DNA sequences represents a significant leap forward in computational biology. These models, trained on massive datasets of genomic information, are beginning to reveal intricate patterns and relationships within the genetic code, promising breakthroughs across diverse biological fields. The ability to analyze and even generate DNA sequences at scale offers unprecedented opportunities for understanding and manipulating life itself. However, the complexity of genomic data presents unique challenges, demanding innovative approaches to model architecture and training strategies.

One of the most significant hurdles in building DNA foundation models lies in the sheer length of genomic sequences. Human genomes, for example, contain billions of base pairs, posing a substantial challenge for traditional deep learning architectures designed for shorter text sequences. The computational resources required to process such extensive datasets are immense, demanding the development of highly efficient algorithms and specialized hardware. Furthermore, the sensitivity required to capture subtle variations in DNA sequences is crucial for accurately predicting biological functions and evolutionary trajectories. A single nucleotide change can have profound consequences, making it essential for models to possess a fine-grained understanding of the genetic code. Early attempts at DNA sequence modeling often focused on limited contexts and specific tasks, neglecting the rich, interconnected nature of genomic information.

Addressing these challenges has led to the development of novel architectures specifically tailored for processing long DNA sequences. These architectures often deviate from the standard transformer models prevalent in natural language processing, employing alternative strategies to handle the vast length and complexity of genomic data. Some approaches incorporate mechanisms to selectively focus on relevant regions of the sequence, avoiding the computational overhead of processing irrelevant information. Others utilize specialized attention mechanisms designed to capture long-range dependencies efficiently. These architectural innovations are critical for enabling the analysis of entire genomes and extracting meaningful biological insights.

The success of DNA foundation models is inextricably linked to the availability of large-scale, high-quality training datasets. The sheer volume of genomic data available publicly, coupled with ongoing sequencing efforts, provides an abundance of information for training these models. However, curating and preprocessing this data is a non-trivial task. Careful consideration must be given to data quality, biases, and representation. The inclusion of metadata alongside raw sequence data, such as gene annotations, epigenetic modifications, and expression levels, can significantly enhance the model's ability to learn complex relationships. The development of standardized formats and pipelines for handling and processing genomic data is crucial for ensuring the reproducibility and reliability of these models.

Beyond the technical challenges of architecture and data, a significant limitation in current DNA foundation models is their limited understanding of regulatory DNA. While these models can effectively analyze protein-coding sequences, predicting the complex interplay of regulatory elements remains a significant challenge. Regulatory regions in genomes, often scattered throughout non-coding sequences, play a critical role in gene expression and other cellular processes. They are significantly more complex than protein-coding sequences, employing diverse mechanisms and exhibiting highly context-dependent behavior. The inherent ambiguity and variability within these sequences make it challenging for models to accurately predict their regulatory functions based on sequence alone. Future advancements will require integrating additional data sources, such as chromatin accessibility profiles, transcription factor binding data, and gene expression patterns, to improve the accuracy of regulatory sequence analysis.

Moreover, the challenge extends to the cross-species generalizability of these models. The evolutionary plasticity of regulatory sequences across different species often results in variations in the regulatory logic, making it difficult to create models that effectively generalize across multiple organisms. Developing models capable of handling this cross-species diversity remains a key objective, necessitating the incorporation of phylogenetic information and sophisticated comparative genomics approaches. Currently, most models are trained on specific subsets of organisms, leading to limitations in their broader applicability.

The potential impact of DNA foundation models on biology is truly transformative. These models offer the potential to revolutionize several key areas of biological research. For example, they could significantly accelerate drug discovery efforts by enabling the rapid identification of potential drug targets and the design of novel therapeutics. By analyzing the sequence and structure of proteins, these models could predict protein function and interactions, paving the way for personalized medicine and more effective treatments for diseases. In agricultural sciences, they could help design crops with enhanced yields and improved resistance to pests and diseases, offering solutions to global food security challenges.

In the realm of synthetic biology, DNA foundation models open exciting new possibilities for engineering new biological systems and creating novel functionalities. These models could allow researchers to design custom genes and genomes with specific properties, creating synthetic organisms for diverse applications. The ability to generate sequences at the whole-genome scale allows for the creation of entirely novel organisms with pre-defined characteristics, though such applications raise complex ethical considerations that require careful consideration.

However, it is crucial to acknowledge that these models are not without limitations. Their predictions and generated sequences require rigorous experimental validation to confirm their accuracy and functionality. It is also important to address potential biases in training data, which can lead to inaccurate or misleading predictions. Furthermore, the responsible development and deployment of these powerful technologies must be prioritized to prevent unintended consequences. The ethical considerations surrounding genome editing and synthetic biology require careful attention, necessitating transparent dialogue and established guidelines to ensure the safe and beneficial application of these advancements.

In conclusion, DNA foundation models represent a pivotal moment in computational biology. While significant challenges remain in accurately modeling the complexities of regulatory regions and achieving cross-species generalizability, their potential to reshape biological research, accelerate drug discovery, and advance synthetic biology is immense. Continued advancements in model architectures, training datasets, and validation techniques are essential for fully realizing the transformative potential of these models, ensuring that their development and application are both scientifically rigorous and ethically responsible. The field is evolving rapidly, and future breakthroughs promise to further unlock the secrets encoded within the genome, leading to profound advancements in our understanding and manipulation of life itself.

Functional protein mining with conformal guarantees

Thu, 28 Nov 2024 14:42:38 GMT

Axial: https://linktr.ee/axialxyz

The paper develops an approach to protein function prediction and homology search, leveraging conformal prediction to provide statistically rigorous guarantees and calibrated probabilities. The authors address a critical limitation in current methods: the lack of reliable statistical assurances, hindering the efficient selection of proteins for further experimental or computational investigation. Their method offers a framework for robustly pre-filtering candidate proteins, annotating genes of unknown function, and improving enzyme classification performance, all while offering calibrated probabilistic predictions rather than relying on arbitrary thresholds.

The core innovation lies in the application of conformal prediction principles. Unlike traditional machine learning approaches that rely on often unrealistic model assumptions (e.g., linearity), conformal prediction offers a model-agnostic framework. This means the method's validity holds regardless of the underlying predictive model's complexity, accommodating the prevalence of deep learning "black box" models in bioinformatics. By employing conformal prediction, the authors transform raw similarity scores (from various protein search models, including Protein-Vec, TM-Vec, Foldseek, and TOPH) into calibrated probabilities and well-defined retrieval sets. These sets are constructed to guarantee user-specified risk levels concerning biologically relevant loss metrics (e.g., false discovery rate, false negative rate), allowing researchers to select proteins with controlled error probabilities. This addresses the problem of arbitrary threshold selection often encountered in current protein homology search methods, leading to more reliable and interpretable results.

The paper demonstrates the efficacy of their approach across several key applications. First, they tackle the challenge of annotating genes of unknown function, focusing on the minimal viable genome of Mycoplasma mycoides (JCVI Syn3.0). Using their conformal prediction framework, they effectively control the false discovery rate while assigning calibrated probabilities to predicted functional matches. This results in a substantial increase in the confidence of functional annotations compared to traditional methods. The method is shown to be robust and reliable, identifying a significant portion (39.6%) of the coding genes with functional matches. This application highlights the practical utility of the approach for exploring poorly understood genomes and identifying potential candidates for further experimental studies.

Secondly, the paper addresses the problem of enzyme function prediction. The authors demonstrate the power of their conformal framework by applying it to an existing state-of-the-art enzyme classification model (CLEAN). They improve upon CLEAN's existing selection procedures (max-separation and p-value selection) by incorporating conformal prediction. This leads to more accurate and statistically robust classification performance across different datasets (New and Price), exceeding the results obtained by the original CLEAN methods. Notably, their approach achieves this improved performance without requiring the training of new models, showcasing the versatility and efficiency of the conformal prediction framework. This improvement is particularly valuable in the context of high-throughput enzyme annotation, where speed and accuracy are critical. The application to the challenging Price dataset underscores the robustness of their method, overcoming potential data biases or inconsistencies.

Finally, the authors introduce a novel strategy for pre-filtering proteins for computationally intensive structural alignment algorithms (like DALI). They use a faster, embedding-based model (Protein-Vec) to pre-filter candidate proteins before applying DALI. By controlling the false negative rate via conformal prediction, they demonstrate a significant reduction in the number of proteins needing full DALI analysis, while retaining a substantial proportion (82.8%) of high-scoring DALI matches. This pre-filtering approach drastically speeds up the overall workflow without compromising accuracy, making large-scale structural homology searches significantly more feasible. This is especially relevant given the rapid increase in the number of predicted protein structures and the computational expense associated with high-resolution structural alignment methods.

The paper also addresses potential limitations and future directions. The authors acknowledge that the assumption of exchangeability, required for conformal prediction, might not always perfectly hold in biological data due to potential sampling biases or data heterogeneity. They suggest that future work could explore techniques designed to address distribution shifts, potentially enhancing the robustness and applicability of their approach. Furthermore, they highlight the need for more comprehensive and accurate datasets, as the reliability of conformal prediction relies on the representativeness of the calibration data.

In conclusion, this paper presents a significant advancement in the field of protein functional annotation and homology search. The use of conformal prediction provides a unique framework for generating statistically rigorous predictions, eliminating the reliance on arbitrary thresholds and enhancing interpretability. The authors effectively demonstrate the utility of their approach across diverse applications, showcasing its potential to accelerate biological discovery and accelerate research workflows. The emphasis on calibrated probabilities and user-specified risk levels makes their method exceptionally practical and suitable for high-throughput analyses in bioinformatics and computational biology. The model-agnostic nature of the conformal approach ensures widespread applicability, facilitating its integration with various protein prediction and homology search algorithms. The future integration of this framework with advancements in protein language models and techniques for handling distribution shifts holds great promise for further enhancing the accuracy and reliability of protein function prediction and homology detection.

https://www.biorxiv.org/content/10.1101/2024.06.27.601042v3

Transannular C–H functionalization of cycloalkane carboxylic acids

Thu, 28 Nov 2024 14:28:47 GMT

Axial: https://linktr.ee/axialxyz

This paper details the development of novel catalytic methods for the site- and diastereoselective synthesis of functionalized carbocycles through transannular C-H functionalization of cycloalkane carboxylic acids. Carbocycles, cyclic structures containing only carbon atoms, are prevalent structural motifs in both natural products and pharmaceuticals. Their inherent rigidity and control over molecular shape contribute significantly to the biological activity and bioavailability of drug candidates. Traditional synthetic approaches to carbocycles often involve multi-step processes, including cycloadditions or intramolecular cyclizations from acyclic precursors, which can be challenging and inefficient. This study proposes C-H activation as a more direct and potentially efficient strategy for the synthesis of these important compounds.

The central challenge in achieving transannular C-H functionalization lies in the inherent strain associated with forming the required palladacycles, especially with larger rings. Existing methods often rely on pre-installed directing groups or are limited to specific ring sizes and functional groups. This research overcomes these limitations by introducing two novel classes of ligands: quinuclidine-pyridones (QuinNuPyridones) and sulfonamide-pyridones (SulfonaPyridones). These ligands facilitate the selective transannular γ-C-H arylation of a wide range of cycloalkane carboxylic acids, encompassing ring sizes from cyclobutane to cyclooctane.

The researchers demonstrate the effectiveness of QuinNuPyridones, specifically L1 and L2, in promoting the transannular γ-arylation of cyclopentane to cyclooctane carboxylic acids. Remarkably, excellent γ-regioselectivity is observed even in the presence of multiple competing β-C-H bonds. This high level of regiocontrol is a significant advancement in C-H activation methodologies. The reaction conditions were optimized using a combination of palladium catalyst, ligand, oxidant, and solvent. The scope of the reaction was explored by varying both the substituents on the cycloalkane ring and the aryl iodide coupling partner. A diverse range of substituents, including electron-donating and electron-withdrawing groups, are well tolerated, showcasing the broad applicability of the method. The utility of this methodology is further highlighted by the successful synthesis of several patented biologically active molecules in just two steps. Previous syntheses of these compounds often required considerably more steps (up to eleven in one case).

The successful transannular γ-arylation of cyclobutane carboxylic acids proved to be more challenging due to the greater strain and preference for β-functionalization. However, the researchers addressed this challenge by employing a different ligand system: the SulfonaPyridone ligand L3 in combination with a monodentate pyridone ligand L4. This combination enabled the highly selective transannular γ-arylation of cyclobutane carboxylic acids via a double C-H activation pathway. This strategy highlights the critical role of ligand design in controlling both reactivity and selectivity in C-H activation reactions. The precise mechanism by which L3 and L4 achieve this high level of selectivity is not fully elucidated, but the authors propose that the combination of these ligands may promote a sequential process involving non-directed C-H activation of the arene and then a subsequent C-H activation/cross-coupling on the cyclobutane ring.

The authors thoroughly investigate the reaction scope for both the cyclopentane-cyclooctane and cyclobutane systems. For the larger rings (cyclopentane to cyclooctane), a wide range of substrates with various α-substituents were successfully arylated, demonstrating the tolerance of the reaction for steric hindrance. Both electron-rich and electron-poor aryl iodides and even heteroaryl iodides reacted efficiently. For cyclobutane carboxylic acids, a similar scope of arene coupling partners is demonstrated although the yield varied depending on the electronics of the arene. The authors further demonstrated that the methodology could be applied to complex molecules, including the late-stage functionalization of the natural product isosteviol.

Beyond the synthesis of novel compounds, the research contributes significantly to the understanding of C-H activation catalysis. The observed regioselectivity and the impact of ligand structure on the outcome of the reaction provide valuable insights into the mechanistic details of transannular C-H activation. The observation that different ligand systems are required for optimal results with different ring sizes further underscores the importance of ligand design in overcoming the challenges of strain and selectivity in these transformations.

In summary, this work represents a significant advancement in the field of C-H activation catalysis. The development of new ligand systems that enable the highly selective transannular γ-arylation of cycloalkane carboxylic acids, particularly the challenging cyclobutane system, opens up new possibilities for the synthesis of complex carbocycles. The methodology is remarkably versatile, tolerating a wide range of substrates and coupling partners, and offers a simplified and efficient route to biologically relevant molecules. The efficient synthesis of known biologically active compounds in fewer steps than previous approaches showcases the practical value and broad applicability of this novel catalytic system. This research not only provides new tools for organic synthesis but also deepens our understanding of the factors that govern the selectivity and reactivity of C-H activation reactions. The insights gained could inspire the design of future catalyst systems for a wider range of C-H functionalization transformations. Future research could focus on expanding the scope of the reaction to other functional group transformations, exploring the catalytic mechanism in more detail, and applying the methodology to the synthesis of even more complex and biologically important molecules.

https://www.nature.com/articles/s41586-023-06000-z