夜上海论坛

Computers don’t have eyes, but they could revolutionize the way scientists visualize cells.

Researchers at the Allen Institute for Cell Science in Seattle have devised 3-D representations of cells, compiled by computers learning where thousands of real cells tuck their component parts.

Most drawings of cells in textbooks come from human interpretations gleaned by looking at just a few dead cells at a time. The new Allen Cell Explorer, which premiered online April 5, presents 3-D images of genetically identical stem cells grown in lab dishes (composite, above), revealing a huge variety of structural differences.
Each cell comes from a skin cell that was reprogrammed into a stem cell. Important proteins were tagged with fluorescent molecules so researchers could keep tabs on the cell membrane, DNA-containing nucleus, energy-generating mitochondria, microtubules and other cell parts. Using the 3-D images, computer programs learned where the cellular parts are in relation to each other. From those rules, the programs can generate predictive transparent models of a cell’s structure (below). The new views, which can cap­ture cells at different time points, may offer clues into their inner workings.
The project’s tools are available for other researchers to use on various types of cells. Insights gained from the explorations might lead to a better understanding of human development, cancer, health and diseases.

Researchers have already learned from the project that stem cells aren’t the shapeless blobs they might appear to be, says Susanne Rafelski, a quantitative cell biologist at the Allen Institute. Instead, the stem cells have a definite bottom and top, a proposed structure that’s now confirmed by the combined cell data, Rafelski says. A solid foundation of skeleton proteins forms at the bottom. The nucleus is usually found in the cell’s center. Microtubules bundle together into large fibers that tend to radiate from the top of the cell toward the bottom. During cell division, microtubules form structures called bipolar spindles that are necessary to divvy up DNA.
One surprise was that the membrane surrounding the nucleus gets ruffled, but never completely disappears, during cell division. Near the top of the cell, above the nucleus, stem cells store tubelike mitochondria much the way plumbing and electrical wires are tucked into ceilings. The tubular mitochondria were notable because some researchers thought that since stem cells don’t require much energy, the organelles might separate into small, individual units.

Old ways of observing cells were like trying to get to know a city by looking at a map, Rafelski says. The cell explorer is more like a documentary of the lives of the citizens.

Every year science offers a diverse menu of anniversaries to celebrate. Births (or deaths) of famous scientists, landmark discoveries or scientific papers — significant events of all sorts qualify for celebratory consideration, as long as the number of years gone by is some worthy number, like 25, 50, 75 or 100. Or simple multiples thereof with polysyllabic names.

2017 has more than enough such anniversaries for a Top 10 list, so some worthwhile events don’t even make the cut, such as the births of Stephen Hawking (1942) and Arthur C. Clarke (1917). The sesquicentennial of Michael Faraday’s death (1867) almost made the list, but was bumped at the last minute by a book. Namely:

10. On Growth and Form, centennial (1917)
    A true magnum opus, by the Scottish biologist D’Arcy Wentworth Thompson, On Growth and Form has inspired many biologists with its mathematical analysis of physical and structural forces underlying the diversity of shapes and forms in the biological world. Nobel laureate biologist Sir Peter Medawar praised Thompson’s book as “beyond comparison the finest work of literature in all the annals of science that have been recorded in the English tongue.”
11. Birth of Abraham de Moivre, semiseptcentennial (1667).
    Born in France on May 26, 1667, de Moivre moved as a young man to London where he did his best work, earning election to the Royal Society. Despite exceptional mathematical skill, though, he attained no academic position and earned a meager living as a tutor. He is most famous for his book The Doctrine of Chances, which was in essence an 18th century version of Gambling for Dummies. It contained major advances in probability theory and in later editions introduced the concept of the famous bell curve. Isaac Newton was impressed; the legend goes that when anyone asked him about probability, Newton said to go talk to de Moivre.
12. Exoplanets, quadranscentennial (1992)It seems like exoplanets have been around almost forever (and probably actually were), but the first confirmed by Earthbound astronomers were reported just a quarter century ago. Three planets showed up orbiting not an ordinary star, but a pulsar, a rapidly spinning neutron star left behind by a supernova.
    Astrophysicists Aleksander Wolszczan and Dale Frail found a sign of the planets, first detected with the Arecibo radio telescope, in irregularities in the radio pulses from the millisecond pulsar PSR1257+12. Some luck was involved. In 1990, the Arecibo telescope was being repaired and couldn’t pivot to point at a specific target; instead it constantly watched just one region of the sky. PSR1257+12 just happened to float by.
13. Birth of Marie Curie, sesquicentennial (1867)
    No doubt the most famous Polish-born scientist since Copernicus, Curie was born in Warsaw on November 7, 1867, as Maria Sklodowska. Challenged by poverty, family tragedies and poor health, she nevertheless excelled as a high school student. But she then worked as a governess, while continuing as much science education as possible, until her married sister invited her to Paris. There she completed her physics education with honors and met and married another young physicist, Pierre Curie.

Together they tackled the mystery of the newly discovered radioactivity, winning the physics Nobel in 1903 along with radioactivity’s discoverer, Henri Becquerel. Marie continued the work after her husband’s tragic death in 1906; she became the first person to win a second Nobel, awarded in chemistry in 1911 for her discovery of the new radioactive elements polonium and radium.

6. Laws of Robotics, semisesquicentennial (1942)
   One of science fiction’s greatest contributions to modern technological philosophy was Isaac Asimov’s Laws of Robotics, which first appeared in a short story in the March 1942 issue of Astounding Science Fiction. Later, those laws formed the motif of his many robot novels and appeared in his famous Foundation Trilogy (and subsequent sequels and prequels). They were:

A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Much later Asimov added a “zeroth law,” requiring robots to protect all of humankind even if that meant violating the other three laws. Artificial intelligence researchers all know about Asimov’s laws, but somehow have not managed to enforce them on social media. Incidentally, this year is also the quadranscentennial of Asimov’s death in 1992.

5. First sustained nuclear fission chain reaction, semisesquicentennial (1942)
   Enrico Fermi, the Italian Nobel laureate, escaped fascist Italy to come to the United States shortly after nuclear fission’s discovery in Germany. Fermi directed construction of the “atomic pile,” or nuclear reactor, on a squash court under the stands of the University of Chicago’s football stadium. Fermi and his collaborators showed that neutrons emitted from fissioning uranium nuclei could induce more fission, creating a chain reaction capable of releasing enormous amounts of energy. Which it later did.
6. Discovery of pulsars, semicentennial (1967)
   Science’s awareness of the existence of pulsars turns 50 this year, thanks to the diligence of Irish astrophysicist Jocelyn Bell Burnell. She spent many late-night hours examining the data recordings from the radio telescope she helped to build that first spotted a signal from a pulsar. She recognized that the signal was something special even though others thought it was just a glitch in the apparatus. But she was a graduate student so her supervisor got the Nobel Prize instead of her.
7. Einstein’s theory of lasers, centennial (1917)
   Albert Einstein did not actually invent the laser, but he developed the mathematical understanding that made lasers possible. By 1917, physicists knew that quantum physics played a part in the working of atoms, but the details were fuzzy. Niels Bohr had shown in 1913 that an atom’s electrons occupy different energy levels, and that falling from a high energy level to a lower one emits radiation.

Einstein worked out the math describing this process when many atoms have electrons in high-energy states and emit radiation. His analysis of matter-radiation interaction indicated that it would be possible to prepare many atoms in the same high-energy state and then stimulate them to emit radiation all at once. Properly done, all the atoms would emit radiation of identical wavelength with the waves in phase. A few decades later other physicists figured out how to build such a device for use as a powerful weapon or to read bar codes at grocery stores.

2. Qubits, quadranscentennial (1992)
   An even better quantum anniversary than lasers is the presentation to the world of the concept of quantum bits of information. Physicist Ben Schumacher of Kenyon College in Ohio unveiled the idea at a conference in Dallas in 1992 (I was there). A “quantum bit” of information, or qubit, represents the information contained in a quantum particle, which can exist in multiple states at once. A photon, for instance, might simultaneously be in a state of horizontal or vertical polarization. Or an electron’s spin could be up and down at the same time.

Such states differ from classical bits of information in a computer, recorded as either a 0 or 1; a quantum bit is both 0 and 1 at the same time. It becomes one or the other only when observed, much like a flipped coin is nether heads nor tails until somebody catches it, or it lands on the 50 yard line. Schumacher’s idea did not get a lot of attention at first, but it eventually became the foundational idea for quantum information theory, a field now booming with efforts to construct a quantum computer based on the manipulation of qubits.

1. Birth of modern cosmology, centennial (1917)
   It might seem unfair that Einstein gets two Top 10 anniversaries in 2017, but 1917 was a good year for him. Before publishing his laser paper, Einstein tweaked the equations of his brand-new general theory of relativity in order to better explain the universe (details in Part 1). Weirdly, Einstein didn’t understand the universe, and he later thought the term he added to his equations was a mistake. But it turns out that today’s understanding of the universe’s behavior — expanding at an accelerating rate — seems to require the term that Einstein thought he had added erroneously. But you can’t expect Einstein to have foreseen everything. He probably had no idea that lasers would revolutionize grocery shopping either.

A lizard’s intricately patterned skin follows rules like those used by a simple type of computer program.

As the ocellated lizard (Timon lepidus) grows, it transforms from a drab, polka-dotted youngster to an emerald-flecked adult. Its scales first morph from white and brown to green and black. Then, as the animal ages, individual scales flip from black to green, or vice versa.

Biophysicist Michel Milinkovitch of the University of Geneva realized that the scales weren’t changing their colors by chance. “You have chains of green and chains of black, and they form this labyrinthine pattern that very clearly is not random,” he says. That intricate ornamentation, he and colleagues report April 13 in Nature, can be explained by a cellular automaton, a concept developed by mathematicians in the 1940s and ’50s to simulate diverse complex systems.
A cellular automaton is composed of a grid of colored pixels. Using a set of rules, each pixel has a chance of switching its shade, based on the colors of surrounding pixels. By comparing photos of T. lepidus at different ages, the scientists showed that its scales obey such rules.
In the adult lizard, if a black scale is surrounded by other black scales, it is more likely to switch than a black one bounded by green, the researchers found. Eventually, the lizards’ scales settle down into a mostly stable state. Black scales wind up with around three green neighbors, and green scales have around four black ones. The researchers propose that interacting pigment cells could explain the color flips.

Computer scientists use cellular automata to simulate the real world, re-creating the turbulent motions of fluids or nerve cell activity in the brain, for example. But the new study is the first time the process has been seen with the naked eye in a real-life animal.
The scales on an ocellated lizard change color as the animal ages (more than three years of growth shown in first clip). Circles highlight four instances of color-flipping scales. Blue circles indicate a scale that switches from green to black, the green circle indicates a black to green transformation, and the light blue circle marks a scale that flip-flops from green to black to green. Researchers used a cellular automaton to simulate the adult lizard’s color-swapping scales (second clip), and re-create the labyrinthine patterns that develop on its skin.

A strain of wild Hawaiian worms has helped unmask long-studied genes as just plain selfish. The scammers beat the usual odds of inheritance and spread extra fast by making mother worms poison some of their offspring.

Biologists have for decades discussed how two genes in the familiar lab nematode Caenorhabditis elegans might help embryos build their organs. Working with a little-studied wild strain, however, caused a rethink of the genes’ supposedly beneficial role “that flipped it on its head,” says UCLA geneticist Leonid Kruglyak.
Instead of doing some body sculpting, the gene sup-35 doses the eggs with a toxin that will kill them after fertilization, two postdocs in the Kruglyak lab discovered. The toxin gene doesn’t poison itself out of the gene pool because it’s linked to a partner, pha-1, that lets embryos manufacture an antidote. Embryos die unless they inherit a copy of the antidote gene in either egg or sperm, and so the poison-antidote duo can spread unusually fast through populations.

Making a mom on occasion poison some of her offspring doesn’t benefit her but certainly helps the genes. Thus the long-known sup-35 and pha-1 form what’s called a selfish genetic element, Kruglyak’s team proposes May 11 in Science.

That analysis is “very clearly accurate,” says evolutionary geneticist Jack Werren of the University of Rochester in New York. The idea that a gene could behave selfishly, promoting its own spread regardless of its host’s interests, was once controversial (SN: 3/19/16, p. 12). But as molecular biology techniques have improved, researchers have found more and more examples. Many of the most dramatic forms of selfishness, the murderous cheats, come from bacteria, so Werren welcomes the C. elegans scam as a rare case discovered in animals.
Kruglyak’s lab has described an earlier example in C. elegans: a gene that doses sperm with a toxin that kills embryos unless an antidote gene rescues them. Finding a second selfish element in the nematode, he says, suggests that these may not be as rare in animals as people have thought.
The big community of researchers regularly studying C. elegans had missed discovering the selfish role for a simple reason: The main lab strain of nematodes carries the selfish element, explains study coauthor Eyal Ben-David. Whenever the standard strain mates or self-fertilizes (the species has both males and hermaphrodites), all the offspring inherit sup-35 and pha-1. Researchers see no weird die-offs.

In the Kruglyak lab, however, Ben-David and fellow postdoc Alejandro Burga were doing a project that required crossing the usual lab nematodes with the DL238 strain from Hawaii. In its natural state, this strain has somehow escaped invasion by the selfish sup-35/pha-1 pair.

A series of oddities in interbreeding the disparate strains pushed the researchers to reconsider the two genes. For instance, much higher percentages of offspring died in mixed-parent crosses than the routine few percent in same-strain pairings. And when Ben-David and Burga looked at the genes in the Hawaiian strain isolated from the wild, sup-35 and pha-1 just weren’t there.

That was a shock. Earlier experiments in the lab strain had shown that disabling pha-1 caused death in offspring — which it certainly does. The feeding tube of the dying embryos was not forming properly, so researchers at first speculated that the gene controlled tube development. Later work suggested a more nuanced role for it, Ben-David says, but the overall hypothesis remained that the genes helped regulate embryo development. The Hawaiian strain changed that thinking: “How is this wild isolate alive and happy without a gene that’s supposed to be essential for development?” Kruglyak wanted to know.

A better way of interpreting the old experiments, he and his colleagues suggest, is that the embryos died because pha-1 wasn’t providing the antidote to the sup-35 toxin. “No one had previously considered the possibility,” says David S. Fay of the University of Wyoming in Laramie, who has done much of the work exploring the role of these genes. “All the data, including a lot of our previous published and unpublished findings, seem to fit the [selfish gene] model perfectly,” he says. And perhaps the highest praise: “I wish we had somehow come up with the solution ourselves.”

The Zika virus probably arrived in the Western Hemisphere from somewhere in the Pacific more than a year before it was detected, a new genetic analysis of the epidemic shows. Researchers also found that as Zika fanned outward from Brazil, it entered neighboring countries and South Florida multiple times without being noticed.

Although Zika quietly took root in northeastern Brazil in late 2013 or early 2014, many months passed before Brazilian health authorities received reports of unexplained fever and skin rashes. Zika was finally confirmed as the culprit in May 2015.
The World Health Organization did not declare the epidemic a public health emergency until February 2016, after babies of Zika-infected mothers began to be born with severe neurological problems. Zika, which is carried by mosquitoes, infected an estimated 1 million people in Brazil alone in 2015, and is now thought to be transmitted in 84 countries, territories and regions.

Although Zika’s path was documented starting in 2015 through records of human cases, less was known about how the virus spread so silently before detection, or how outbreaks in different parts of Central and South America were connected. Now two groups working independently, reporting online May 24 in Nature, have compared samples from different times and locations to read the history recorded in random mutations of the virus’s 10 genes.

One team, led by scientists in the United Kingdom and Brazil, drove more than 1,200 miles across Brazil — “a Top Gear–style road trip,” one scientist quipped — with a portable device that could produce a complete catalog of the virus’s genes in less than a day. A second team, led by researchers at the Broad Institute of MIT and Harvard, analyzed more than 100 Zika genomes from infected patients and mosquitoes in nine countries and Puerto Rico. Based on where the cases originated, and the estimated rate at which genetic changes appear, the scientists re-created Zika’s evolutionary timeline.

Together, the studies revealed an epidemic that was silently churning long before anyone knew. “We found that in each of the regions we could analyze, Zika virus circulated undetected for many months, up to a year or longer, before the first locally transmitted cases were reported,” says Bronwyn MacInnis, an infectious disease geneticist at the Broad Institute, in Cambridge, Mass. “This means the outbreak in these regions was under way much earlier than previously thought.”

Although the epidemic exploded out of Brazil, the scientists also found a remote possibility of early settlement in the Caribbean. “It’s not immediately clear whether Zika stopped off somewhere else in the Americas before it got to northeast Brazil,” said Oliver Pybus, who studies evolution and infectious disease at the University of Oxford in England.
In a third study reported in Nature, researchers from more than two dozen institutions followed a trail of genetic clues to determine when and how Zika made its way to Florida. Those researchers concluded that Zika was introduced multiple times into the Miami area, most likely from the Caribbean, before local mosquitoes picked it up. The number of human cases increased in step with the rise in mosquito populations, said Kristian Andersen, an infectious disease researcher at the Scripps Research Institute in La Jolla, Calif. “Focusing on getting rid of mosquitoes is an effective way of preventing human cases,” he says.
Stealth spread
An analysis of more than 100 Zika genomes revealed that the virus showed up in nine countries 4.5 to 9 months earlier than the first confirmed cases of Zika virus infection. Colors indicate the distribution of groups of closely related strains of the virus.

Hover over/tap map to explore Zika’s spread in the Americas.
Previous studies have found traces of the virus’s footprints across the Americas, but none included so many different samples, says Young-Min Lee of Utah State University, who has also studied Zika’s genes. The current studies provide a higher-resolution look at the timing of the epidemic’s spread, he says, but in terms of Zika’s origins and progression from country to country, “overall the big picture is consistent with what we suspected.”

In addition to revealing Zika’s history, genetic studies are also valuable in fighting current and future disease outbreaks. Since diagnostic tests and even vaccine development are based on Zika’s genetics, it’s important to monitor mutations during an outbreak. Researchers developed quick-turnaround genomic analyses for Ebola in recent years, for example, that could aid a faster response during the next outbreak.

In the future, faster analysis of viral threats in the field might improve the odds of stopping the next epidemic, Lee says. It’s possible for a single infected traveler stepping off a plane to spark an epidemic long before doctors notice. “If one introduction [of a virus] can cause an outbreak, you have a very narrow window to try to contain it.”

Last year, Joan Peay slipped on her garage steps and smashed her knee on the welcome mat. Peay, 77, is no stranger to pain. The Tennessee retiree has had 17 surgeries in the last 35 years — knee replacements, hip replacements, back surgery. She even survived a 2012 fungal meningitis outbreak that sickened her and hundreds of others, and killed 64. This knee injury, though, “hurt like the dickens.”

When she asked her longtime doctor for something stronger than ibuprofen to manage the pain, he treated her like a criminal, Peay says. His response was frustrating: “He’s known me for nine years, and I’ve never asked him for pain medicine other than what’s needed after surgery,” she says. She received nothing stronger than over-the-counter remedies. A year after the fall, she still lives in constant pain.
Just five years ago, Peay might have been handed a bottle of opioid painkillers for her knee. After all, opioids — including codeine, morphine and oxycodone — are some of the most powerful tools available to stop pain.
But an opioid addiction epidemic spreading across the United States has soured some doctors on the drugs. Many are justifiably concerned that patients will get hooked or share their pain pills with friends and family. And even short-term users risk dangerous side effects: The drugs slow breathing and can cause constipation, nausea and vomiting.

A newfound restraint in prescribing opioids is in many cases warranted, but it’s putting people like Peay in a tough spot: Opioids have become harder to get. Even though the drugs are far from perfect, patients have few other options.
Many drugs that have been heralded as improvements over existing opioids are just old opioids repackaged in new ways, says Nora Volkow, director of the National Institute on Drug Abuse. Companies will formulate a pill that is harder to crush, for instance, or mix in another drug that prevents an opioid pill from working if it’s crushed up and snorted for a quick high. Addicts, however, can still sidestep these safeguards. And the newly packaged drugs have the same fundamental risks as the old ones.

The need for new pain medicines is “urgent,” says Volkow.

Scientists have been searching for effective alternatives for years without success. But a better understanding of the way the brain sends and receives specific chemical messages may finally boost progress.

Scientists are designing new, more targeted molecules that might kill pain as well as today’s opioids do — with fewer side effects. Others are exploring the potential of tweaking existing opioid molecules to skip the negative effects. And some researchers are steering clear of opioids entirely, testing molecules in marijuana to ease chronic pain.

Opioid action
Humans recognized the potential power of opioids long before they understood how to control it. Ancient Sumerians cultivated opium-containing poppy plants more than 5,000 years ago, calling their crop the “joy plant.” Other civilizations followed suit, using the plant to treat aches and pains. But the addictive power of opium-derived morphine wasn’t recognized until the 1800s, and scientists have only recently begun to piece together exactly how opioids get such a stronghold on the brain.

Opioids mimic the body’s natural painkillers — molecules like endorphins. Both endorphins and opioids latch on to proteins called opioid receptors on the surface of nerve cells. When an opioid binds to a receptor in the peripheral nervous system, the nerve cells outside the brain, the receptor changes shape and sets in motion a cellular game of telephone that stops pain messages from reaching the brain.

The danger comes because opioid receptors scattered throughout the body and in crucial parts of the brain can cause far-reaching side effects when drugs latch on. For starters, many opioid receptors are located near the base of the brain — the part that controls breathing and heart rate. When a drug like morphine binds to one of these receptors in the brain stem, breathing and heart rate slow down. At low doses, the drug just makes people feel relaxed. At high doses, though, it can be deadly — most opioid overdose deaths occur when a person stops breathing. And high numbers of opioid receptors in the gut — thanks in part to all the nerve endings there — can trigger constipation and sometimes nausea.
Plus, opioids are highly addictive. These drugs mess with the brain’s reward system, triggering release of dopamine at levels higher than what the brain is used to. Gradually, the opioid receptors in the brain become less sensitive to the drugs, so the body demands higher and higher doses to get the same feel-good benefit. Such tolerance can reset the system so the body’s natural opioids no longer have the same effect either. If a person tries to go without the drugs, withdrawal symptoms like intense sweating and muscle cramps kick in — the body is physically dependent on the drugs. Addiction is a more complex phenomenon than dependence, involving physical cravings so strong that a person will go to extreme lengths to get the next dose. Long-term users of prescription opioids might be dependent on the drugs, but not necessarily addicted. But dependence and addiction often go together.

Despite their risks, opioids are still widely used because they work so well, particularly for moderate to severe short-term pain.

“No matter how much I say I want to avoid opioids, half of my patients will get some kind of opioid. It’s just unavoidable,” says Christopher Wu, an anesthesiologist at Johns Hopkins Medicine.

In the late 1990s and early 2000s, more doctors began doling out the drugs for long-term pain, too. Aggressive marketing campaigns from Purdue Pharma, the maker of OxyContin, promised that the drug was safe — and doctors listened. Opioid overdoses nearly quadrupled between 2000 and 2015, with almost half of those deaths coming from opioids prescribed by a doctor, according to data from the U.S. Centers for Disease Control and Prevention.
Opioid prescriptions have dipped a bit since 2012, thanks in part to stricter prescription laws and prescription registration databases. U.S. doctors wrote about 30 million fewer opioid prescriptions in 2015 than in 2012, data from IMS Health show. But restricting access doesn’t make pain disappear or curb addiction. Some people have turned to more dangerous street alternatives like heroin. And those drugs are sometimes spiked with more potent opioids such as fentanyl (SN: 9/3/16, p. 14) or even carfentanil, a synthetic opioid that’s used to tranquilize elephants. Overdose deaths from fentanyl and heroin have both spiked since 2012, CDC data reveal.

A sharper target
Scientists have been searching for a drug that kills pain as successfully as opioids without the side effects for close to a hundred years, with no luck, says Sam Ananthan, a medicinal chemist at Southern Research in Birmingham, Ala. He is newly optimistic.

“Right now, we have more biological tools, more information regarding the biochemical pathways,” Ananthan says. “Even though prior efforts were not successful, we now have some rational hypotheses.”

Scientists used to think opioid receptors were simple switches: If a molecule latched on, the receptor fired off a specific message. But more recent studies suggest that the same receptor can send multiple missives to different recipients.

The quest for better opioids got a much-needed jolt in 1999, when researchers at Duke University showed that mice lacking a protein called beta-arrestin 2 got more pain relief from morphine than normal mice did. And in a follow-up study, negative effects were less likely. “If we took out beta-arrestin 2, we saw improved pain relief, but less tolerance development,” says Laura Bohn, now a pharmacologist at the Scripps Research Institute in Jupiter, Fla. Bohn and colleagues figured out that mu opioid receptors — the type of opioid receptor targeted by most drugs — send two different streams of messages. One stops pain. The other, which needs beta-arrestin 2, drives many of the negatives of opioids, including the need for more and more drug and the dangerous slowdown of breathing.

Since that work, Bohn’s lab and many others have been trying to create molecules that bind to mu opioid receptors without triggering beta-arrestin 2 activity. The approach, called biased agonism, “has been around some time, but now it’s bearing the fruit,” says Susruta Majumdar, a chemist at Memorial Sloan Kettering Cancer Center in New York City. Scientists have identified dozens of molecules that seem to avoid beta-arrestin 2 in mice. But only a few might make good drugs. One, called PZM21, was described in Nature last year.
Another one has shown promise in humans — a much higher bar. The pharmaceutical company Trevena, headquartered in King of Prussia, Pa., has been working its way through the U.S. Food and Drug Administration’s drug approval process with a molecule called oliceridine. In studies reported in April in San Francisco at the Annual Regional Anesthesiology and Acute Pain Medicine Meeting, oliceridine was as effective as morphine in patients recovering from bunion removal and others who had tummy tuck surgeries. Over the short term, people taking a moderate dose of the drug got pain relief comparable to that of morphine, but reported fewer side effects, such as vomiting and breathing problems.

Oliceridine is an intravenous opioid, not an oral one. That means it would be administered in the short term in hospitals, during and after surgeries. It’s not a replacement for the pills people can go home with, says Jonathan Violin, Trevena’s cofounder. And it’s not perfect: More side effects cropped up at higher doses. But it’s the first opioid using this targeted approach to get this far in human studies. The company hopes to submit an application for FDA approval by the end of 2017, Violin says.

Avoiding the beta-arrestin 2 pathway isn’t the only approach to targeted opioids — just one of the best studied. Ananthan’s lab is taking a different tack. His team showed that mice lacking a different opioid receptor, the delta receptor, tended not to show negative effects in response to the drugs. Now, the researchers are trying to find molecules that can activate mu opioid receptors while blocking delta receptors.

There may also be a way to direct pain-killing messages specifically to the parts of a person’s body that are feeling pain. In one recent study, scientists described a molecule that bound to opioid receptors only when the area around the receptors was more acidic than normal. Inflammation from pain and injury raises acidity, so this molecule could quash pain where necessary, but wouldn’t bind to receptors elsewhere in the body, reducing the likelihood of side effects. Rats in the study, published in the March 3 Science, didn’t find the new molecule as rewarding as fentanyl, so it may be less addictive. And they were less likely to have constipation and slowed breathing.

Drugs face a long uphill climb from even the most promising animal studies to FDA approval for use in humans. Very few make it that far. It’s too soon to tell whether PZM21 and other molecules being studied in mice will ever end up as treatments for patients.

Unwilling to wait, some people in pain are turning to substances that are already available — without a doctor’s order. And scientists are trying to catch up.

Kratom crackdown
In August 2016, the Drug Enforcement Administration announced that it was cracking down on a supplement called kratom. Officials wanted to put the herb in the same regulatory category as heroin and LSD, labeling it a dangerous substance with no medical value. Members of the public vehemently disagreed. More than 23,000 comments poured in from veterans, cancer survivors, factory workers, lawyers and teachers. Almost all of them said the same thing: Kratom freed them from pain.
Made from the leaves of the tropical plant Mitragyna speciosa , kratom is sold in corner convenience stores and through online retailers. Its pain-killing abilities come mainly from two different molecules in the plant’s leaves: mitragynine and the structurally similar 7-hydroxymitragynine. Both have a structure that’s very different from morphine, but they bind to opioid receptors. That technically makes them opioids, even though they don’t look like morphine or oxycodone, Majumdar says. And that’s what concerned the DEA.
But just like some of the new opioids that scientists are developing, kratom’s active ingredients appear — anecdotally, at least — to deliver pain relief with fewer problems and less risk of tolerance. Some chronic opioid users switch to kratom to wean themselves off of pain pills and ease withdrawal symptoms, says Oliver Grundmann, a medicinal chemist at the University of Florida in Gainesville. Other users have never habitually used opioids but are seeking relief from chronic pain or mental health problems, according to a survey he published online May 10 in Drug and Alcohol Dependence. Grundmann hopes the survey results will help guide research into the substance’s efficacy for specific medical concerns.

The safety and efficacy of kratom is still up for debate. There’s a lack of controlled clinical studies about the leaf’s impact on the body, Grundmann says. Plus, the way kratom is regulated — as a supplement — means that people buying it have no guarantee of what they’re actually getting.

While kratom has its fans, its active compounds aren’t very potent, says Majumdar. He thinks he could make a better drug by modifying these molecules.

Majumdar, Sloan Kettering collaborator András Váradi and colleagues tested a structural cousin of 7-hydroxymitragynine: mitragynine pseudoindoxyl. It binds to mu opioid receptors about 200 times as effectively as mitragynine in mice, the researchers reported in August in the Journal of Medicinal Chemistry. Just like Trevena’s oliceridine, the new molecule does not activate beta-arrestin 2. The pseudoindoxyl version also blocks the delta opioid receptor, further impeding nonpain-related activities.

Majumdar hopes a DEA ban on kratom won’t happen; it would severely restrict access, making research much harder to do. For now, there is no ban — but scientists are wary, he says.

Mix it up
Despite the potential for new, better opioids, other researchers are focused on an altogether different set of pain-killing drugs: the cannabinoids (made famous by marijuana, the dried leaves and other parts of the hemp plant, Cannabis sativa).

The active molecules in marijuana don’t have the same fast-acting pain-quenching abilities that opioids do. “If I go into an emergency room with acute pain, give me morphine,” says Yasmin Hurd, a pharmacologist at Mount Sinai in New York City. But with medical marijuana legal in 29 states plus the District of Columbia, the plant is getting more attention as a potential pain reliever, especially for chronic pain (SN: 6/14/14, p. 16).

Doctors in states where marijuana is legal write fewer prescriptions for opioid painkillers, a 2016 study in Health Affairs showed. Those states also had about a 25 percent lower rate of opioid overdose deaths compared with states that didn’t legalize marijuana, according to a 2014 study in JAMA Internal Medicine. When marijuana becomes legally available, some people might choose it instead of opioids.
There might be some merit to that choice. There are plenty of cannabinoid receptors in parts of the brain that process pain messages. But unlike opioid receptors, few exist in the brain stem. That means cannabinoids are far less likely to influence breathing than opioids, says Joseph Cheer, a neurobiologist at the University of Maryland School of Medicine in Baltimore. Fatal overdoses are nearly unheard of.

As with kratom, though, there’s a glut of anecdotal evidence suggesting marijuana’s power to cure everything from pain to anxiety to ulcers — but not many controlled clinical trials to back up the assertions (SN Online: 1/12/17). The knowledge gap is made even wider by the fact that marijuana has wildly different effects depending on how it’s ingested and the relative ratios of certain active molecules in each strain of the plant.There might be some merit to that choice. There are plenty of cannabinoid receptors in parts of the brain that process pain messages. But unlike opioid receptors, few exist in the brain stem. That means cannabinoids are far less likely to influence breathing than opioids, says Joseph Cheer, a neurobiologist at the University of Maryland School of Medicine in Baltimore. Fatal overdoses are nearly unheard of.

“People think they know how marijuana affects the brain,” Hurd says. In reality, “there’s been very little evidence-based structural scientific studies done with marijuana.”

Aron Lichtman, a pharmacologist at Virginia Commonwealth University in Richmond, agrees. “There’s definitely medicine in that plant — that’s been proven,” he says. “The challenge is that it may not work for everybody and every type of pain.”

Scientists who are serious about figuring out marijuana are breaking it down, looking at the plant’s active molecules — cannabinoids — one by one. Cannabidiol, or CBD, has garnered particular attention. Because of the way it indirectly interacts with cannabinoid receptors, it doesn’t give people the high that’s characteristic of tetrahydrocannabinol, or THC, the mind-altering chemical in marijuana. That makes CBD less rewarding and better suited to longer-term use. The molecule can influence signals sent by a number of other receptors in the brain, many involved in pain and inflammation.

But THC might have merit, too. It’s already used in a couple of FDA-approved drugs to treat nausea and vomiting from chemo-therapy. There’s some evidence that those medications might also help relieve pain, though Lichtman calls those studies a “mixed bag.”

Alone, cannabinoids might be fairly weak painkillers. But combined with opioids, he’s shown, they can amplify the pain relief and reduce the opioid dose needed in mice.

Drugs that might amp up the power of the body’s natural cannabinoids are another option. That’s what Ruth Ross of the University of Toronto is studying. A few years ago, her team identified a region on a cannabinoid receptor called CB1 that has an interesting property: Small molecules that bind to it act like volume knobs for the body’s natural cannabinoids, called endocannabinoids. When a molecule of the right shape locks on to CB1, it makes endocannabinoids naturally present in the body more likely to latch on. That boosts pain relief in a targeted way — when endocannabinoids are already being released by the body, such as after injury or stress.

“You magnify the already existing effects of the compound,” Ross says. Her team has identified and patented several of these volume-knob molecules, and is working on improving them.

“For various reasons they wouldn’t be good as drugs,” she says. They have too many effects on the body beyond their intended one. But she’s making slight tweaks to their chemical structures to try to reduce those off-target effects, with the hope that one day the molecules could be studied in patients.

Safer opioids or alternative painkillers would help people deal with their pain without risking addiction or death. Peay has gotten to know people — as a member of social media groups for those living with chronic pain — who are experiencing the crushing results of poorly managed pain. People lose their jobs, she says, or move to Colorado just to get access to legal marijuana. As for her? “I still have my sense of humor, and that helps me get through all the pain.” But she’s holding out for something better.

The woman in her 70s was in trouble. What started as a broken leg led to an infection in her hip that hung on for two years and several hospital stays. At a Nevada hospital, doctors gave the woman seven different antibiotics, one after the other. The drugs did little to help her. Lab results showed that none of the 14 antibiotics available at the hospital could fight the infection, caused by the bacterium Klebsiella pneumoniae.

Epidemiologist Lei Chen of the Washoe County Health District sent a bacterial sample to the U.S. Centers for Disease Control and Prevention. The bacteria, CDC scientists found, produced a nasty enzyme called New Delhi metallo-beta-lactamase, known for disabling many antibiotics. The enzyme was first seen in a patient from India, which is where the Nevada woman broke her leg and received treatment before returning to the United States.
The enzyme is worrisome because it arms bacteria against carbapenems, a group of last-resort antibiotics, says Alexander Kallen, a CDC medical epidemiologist based in Atlanta, who calls the drugs “our biggest guns for our sickest patients.”

The CDC’s final report revealed startling news: The bacteria raging in the woman’s body were resistant to all 26 antibiotics available in the United States. She died from septic shock; the infection shut down her organs.

Kallen estimates that there have been fewer than 10 cases of completely resistant bacterial infections in the United States. Such absolute resistance to all available drugs, though incredibly rare, was a “nightmare scenario,” says Daniel Kadouri, a micro-biologist at Rutgers School of Dental Medicine in Newark, N.J.

Antibiotic-resistant bacteria infect more than 2 million people in the United States every year, and at least 23,000 die, according to 2013 data, the most recent available from the CDC.

It’s time to flip the nightmare scenario and send a killer after the killer bacteria, say a handful of scientists with a new approach for fighting infection. The strategy, referred to as a “living antibiotic,” would pit one group of bacteria — given as a drug and dubbed “the predators” — against the bacteria that are wreaking havoc among humans.
The approach sounds extreme, but it might be necessary. Antimicrobial resistance “is something that we really, really have to take seriously,” says Elizabeth Tayler, senior technical officer for antimicrobial resistance at the World Health Organization in Geneva. “The ability of future generations to manage infection is at risk. It’s a global problem.”

The number of resistant strains has exploded, in part because doctors prescribe antibiotics too often. At least 30 percent of antibiotic prescriptions in the United States are not necessary, according to the CDC. When more people are exposed to more antibiotics, resistance is likely to build faster. And new alternatives are scarce, Kallen says, as the pace of developing novel antibiotics has slowed.

In search of new ideas, DARPA, a Department of Defense agency that invests in breakthrough technologies, is supporting work on predatory bacteria by Kadouri, as well as Robert Mitchell of Ulsan National Institute of Science and Technology in South Korea, Liz Sockett of the University of Nottingham in England and Edouard Jurkevitch of the Hebrew University of Jerusalem. This work, the agency says, represents “a significant departure from conventional antibiotic therapies.”

The approach is so unusual, people have called Kadouri and his lab crazy. “Probably, we are,” he jokes.

A movie-worthy killer
The notion of predatory bacteria sounds a bit scary, especially when Kadouri likens the most thoroughly studied of the predators, Bdellovibrio bacteriovorus, to the vicious space creatures in the Alien movies.

B. bacteriovorus, called gram-negative because of how they are stained for microscope viewing, dine on other gram-negative bacteria. All gram-negative bacteria have an inner membrane and outer cell wall. The predators don’t go after the other main type of bacteria, gram-positives, which have just one membrane.
When it encounters a gram-negative bacterium, the predator appears to latch on with grappling hook–like appendages. Then, like a classic cat burglar cutting a hole in glass, B. bacteriovorus forces its way through the outer membrane and seems to seal the hole behind it. Once within the space between the outer and inner membranes, the predator secretes enzymes — as damaging as the movie aliens’ acid spit — that chew its prey’s nutrients and DNA into bite-sized pieces.

B. bacteriovorus then uses the broken-down genetic building blocks to make its own DNA and begin replicating. The invader and its progeny eventually emerge from the shell of the prey in a way reminiscent of a cinematic chest-bursting scene.

“It’s a very efficient killing machine,” Kadouri says. That’s good news because many of the most dangerous pathogens that are resistant to antibiotics are gram-negative (SN: 6/10/17, p. 8), according to a list released by the WHO in February.

It’s the predator’s hunger for the bad-guy bacteria, the ones that current drugs have become useless against, that Kadouri and other researchers hope to harness.

Pitting predatory against pathogenic bacteria sounds risky. But, from what researchers can tell, these killer bacteria appear safe. “We know that [B. bacteriovorus] doesn’t target mammalian cells,” Kadouri says.

Saving the see-through fish
To find out whether enlisting predatory bacteria might be crazy good and not just plain crazy, Kadouri’s lab group tested B. bacteriovorus’ killing ability against an array of bacteria in lab dishes in 2010. The microbe significantly reduced levels of 68 of the 83 bacteria tested.

Since then, Kadouri and others have looked at the predator’s ability to devour dangerous pathogens in animals. In rats and chickens, B. bacteriovorus reduced the number of bad bacteria. But the animals were always given nonlethal doses of pathogens, leaving open the question of whether the predator could save the animals’ lives.

Sockett needed to see evidence of survival improvement. “If we’re going to have Bdellovibrio as a medicine, we have to cure something,” she says. “We can count changes in numbers of bacteria, but if that doesn’t change the outcome of the infection — change the number of [animals] that die — it’s not worth it.”

So she teamed up with cell biologist Serge Mostowy of Imperial College London for a study in zebrafish. The aim was to see how many animals predatory bacteria could save from a deadly infection. The team also tested how the host’s immune system interacted with the predators.

The researchers gave zebra-fish larvae fatal doses of an antibiotic-resistant strain of Shigella flexneri, which causes dysentery in humans. Before infecting the fish, the researchers divided them into four groups. Two groups had their immune systems altered to produce fewer macrophages, the white blood cells that attack pathogens. Immune systems in the other two groups remained intact. B. bacteriovorus was injected into an unchanged group and a macrophage-deficient group, while two groups received no treatment.

All of the untreated fish with fewer macrophages died within 72 hours of receiving S. flexneri, the researchers reported in December in Current Biology. Of the fish with a normal immune system, 65 percent that received predator treatment survived compared with 35 percent with no predator treatment. Even in the fish with impaired immune systems, the predators saved about a quarter of the lot.
“This is the first time that Bdellovibrio has ever been used as an injected therapy in live organisms,” Sockett says. “And the important thing is the injection improved the survival of the zebrafish.”

The study also pulled off another first. In previous work, researchers had been unable to see predation as it happened within an animal. Because zebra-fish larvae are transparent, study coauthor Alexandra Willis captured images of B. bacteriovorus gobbling up S. flexneri.

“We were literally having to run to the microscope because the process was just happening so fast,” says Willis, a graduate student in Mostowy’s lab. After the predator invades, its rod-shaped prey become round. Willis saw Bdellovibrio “rounding” its prey within 15 minutes. From start to finish, the predatory cycle took about three to four hours.

The predator’s speed may be what gave it the edge over the infection, Mostowy says. B. bacteriovorus attacks fast, chipping away at the pathogens until the infection is reduced to a level that the immune system can handle. “Otherwise there are too many bacteria and the immune system would be overwhelmed,” he says. “We’re putting a shocking amount of Shigella, 50,000 bacteria, into the fish.”

Within 48 hours, S. flexneri levels dropped 98 percent in the surviving fish, from 50,000 to 1,000.

The immune cells also cleared nearly all the B. bacteriovorus predators from the fish. The predators had enough time to attack the infection before being targeted by the immune system themselves, creating an ideal treatment window. Even if the host’s immune system hadn’t attacked the predators, once the bacteria are gone, Willis says, the predators are out of food. Unable to replicate, they eventually die off.

A clean sweep
Predatory bacteria are efficient in more ways than one. They’re not just good killers — they eliminate the evidence too.

Typical antibiotic treatments don’t target a bacterium’s DNA, so they are likely to leave pieces of the bacterial body behind. That’s like killing a few bandits, but leaving their weapons so the next invaders can easily arm themselves for a new attack. This could be one way that multidrug resistance evolves, Mitchell says. For example, penicillin will kill all bacteria that aren’t resistant to the drug. The surviving bacteria can swim through the aftermath of the antibiotic attack and grab genes from their fallen comrades to incorporate into their own genomes. The destroyed bacteria may have had a resistance gene to a different antibiotic, say, vancomycin. Now you have bacteria that are resistant to both penicillin and vancomycin. Not good.

Predatory bacteria, on the other hand, “decimate the genome” of their prey, Mitchell says. They don’t just kill the bandit, they melt down all the DNA weapons so no pathogens can use them. In one experiment that has yet to be published, B. bacteriovorus almost completely ate up the genetic material of a bacterial colony within two hours — showing itself as a fast-acting predator that could prevent bacterial genes from falling into the wrong hands.

On top of that, even if pathogenic bacteria mutate, a common way they pick up new forms of resistance, they aren’t protected from predation. Resistance to predation hasn’t been reported in lab experiments since B. bacteriovorus was discovered in 1962, Mitchell says. Researchers don’t think there’s a single pathway or gene in a prey bacterium that the predator targets. Instead, B. bacteriovorus seem to use sheer force to break in. “It’s kind of like cracking an egg with a hammer,” Kadouri says. That’s not exactly something bacteria can mutate to protect themselves against.

Some bacteria manage to band together and cover themselves with a kind of built-in biological shield, which offers protection against antibiotics. But for predatory bacteria, the shield is more of a welcome mat.

Going after the gram-positives
When bacteria cluster together on a surface, whether in your body, on a countertop or on a medical instrument, they can form a biofilm. The thick, slimy shield helps microbes withstand antibiotic attacks because the drugs have difficulty penetrating the slime. Antibiotics usually act on fast-growing bacteria, but within a biofilm, bacteria are sluggish and dormant, making antibiotics less effective, Kadouri says.
But to predatory bacteria, a biofilm is like Jell-O — a tasty snack that’s easy to swallow. Once inside, B. bacteriovorus spreads like wildfire because its prey are now huddled together as confined targets. “It’s like putting zebras and a lion in a restaurant and closing the door and seeing what happens,” Kadouri says. For the zebras, “it can’t end well.”

Kadouri’s lab has shown repeatedly that predatory bacteria effectively eat away biofilms that protect gram-negative bacteria, and are in fact more efficient at killing bacteria within those biofilms.

Gram-positive bacteria cloak themselves in biofilms too. In 2014 in Scientific Reports, Mitchell and his team reported finding a way to use Bdellovibrio to weaken gram-positive bacteria, turning their protective shield against them and perhaps helping antibiotics do their job.

The discovery comes from studies of one naturally occurring B. bacteriovorus mutant with extra-scary spit. The mutant isn’t predatory. Instead of eating a prey’s DNA to make its own, it can grow and replicate like a normal bacterial colony. As it grows, it produces especially destructive enzymes. Among the mix of enzymes are proteases, which break down proteins.

Mitchell and his team tested the strength of the mutant’s secretions against the gram-positive Staphylococcus aureus. A cocktail of the enzymes applied to an S. aureus biofilm degraded the slime shield and reduced the bacterium’s virulence. Biofilms can make bacteria up to 1,000 times more resistant to antibiotics, Mitchell says. The next step, he adds, is to see if degrading a biofilm resensitizes a gram-positive bacterium to antibiotics.

Mitchell and his team also treated S. aureus cells that didn’t have a biofilm with the mutant’s enzyme mix and then exposed them to human cells. Eighty percent of the bacteria were no longer able to invade human cells, Mitchell says. The “acid spit” chewed up surface proteins that the pathogen uses to attach to and invade human cells. The enzymes didn’t kill the bacteria but did make them less virile.

No downsides yet
Predatory bacteria can efficiently eat other gram-negative bacteria, munch through biofilms and even save zebrafish from the jaws of an infectious death. But are they safe? Kadouri and the other researchers have done many studies, though none in humans yet, to try to answer that question.
In a 2016 study published in Scientific Reports, Kadouri and colleagues applied B. bacteriovorus to the eyes of rabbits and compared the effect with that of a common antibiotic eye drop, vancomycin. The vancomycin visibly inflamed the eyes, while the predatory bacteria had little to no effect. The eyes treated with predatory bacteria were indistinguishable from eyes treated with a saline solution, used as the control treatment. Other studies looking for potential toxic effects of B. bacteriovorus have so far found none.

In 2011, Sockett’s team gave chickens an oral dose of predatory bacteria. At 28 days, the researchers saw no difference in health between treated and untreated chickens. The makeup of the birds’ gut bacteria was altered, but not in a way that was harmful, she and her team reported in Applied and Environmental Microbiology.

Kadouri analyzed rats’ gut microbes after a treatment of predatory bacteria, reporting the results in a study published March 6 in Scientific Reports. Here too, the rodents’ guts showed little to no inflammation. When they sequenced the bacterial contents of the rats’ feces, the researchers saw small differences between the treated and untreated rats. But none of the changes appeared harmful, and the animals grew and acted normally.

If the rats had taken common antibiotics, it would have been a different story, Kadouri points out. Those drugs would have given the animals diarrhea, reduced their appetites and altered their gut flora in a big way. “When you take antibiotics, you’re basically t hrowing an atomic bomb” into your gut, Kadouri says. “You’re wiping everything out.”
Both Mitchell and Kadouri tested B. bacteriovorus on human cells and found that the predatory bacteria didn’t harm the cells or prompt an immune response. The researchers separately reported their findings in late 2016 in Scientific Reports and PLOS ONE .
Microbiologist Elizabeth Emmert of Salisbury University in Maryland studies B. bacterio-vorus as a means to protect crops — carrots and potatoes — from bacterial soft rot diseases. For humans, she calls the microbes a “promising” therapy for bacterial infections. “It seems most feasible as a topical treatment for wounds, since it would not have to survive passage through the digestive tract.”

There are plenty of questions that need answering first. Mitchell guesses that there will probably be 10 more years of rigorous testing in animals before moving on to human clinical studies. But pursuing these alternatives is worth the effort.

“The drugs that we’re taking are not benign and cuddly and nice,” Kadouri says. “We need them, but they don’t come without side effects.” Even though a living antibiotic sounds a bit crazy, it might be the best option in this dangerous era of antibiotic resistance.

Although the term “quantum computer” might suggest a miniature, sleek device, the latest incarnations are a far cry from anything available in the Apple Store. In a laboratory just 60 kilometers north of New York City, scientists are running a fledgling quantum computer through its paces — and the whole package looks like something that might be found in a dark corner of a basement. The cooling system that envelops the computer is about the size and shape of a household water heater.

Beneath that clunky exterior sits the heart of the computer, the quantum processor, a tiny, precisely engineered chip about a centimeter on each side. Chilled to temperatures just above absolute zero, the computer — made by IBM and housed at the company’s Thomas J. Watson Research Center in Yorktown Heights, N.Y. — comprises 16 quantum bits, or qubits, enough for only simple calculations.

If this computer can be scaled up, though, it could transcend current limits of computation. Computers based on the physics of the super­small can solve puzzles no other computer can — at least in theory — because quantum entities behave unlike anything in a larger realm.

Quantum computers aren’t putting standard computers to shame just yet. The most advanced computers are working with fewer than two dozen qubits. But teams from industry and academia are working on expanding their own versions of quantum computers to 50 or 100 qubits, enough to perform certain calculations that the most powerful supercomputers can’t pull off.
The race is on to reach that milestone, known as “quantum supremacy.” Scientists should meet this goal within a couple of years, says quantum physicist David Schuster of the University of Chicago. “There’s no reason that I see that it won’t work.”
But supremacy is only an initial step, a symbolic marker akin to sticking a flagpole into the ground of an unexplored landscape. The first tasks where quantum computers prevail will be contrived problems set up to be difficult for a standard computer but easy for a quantum one. Eventually, the hope is, the computers will become prized tools of scientists and businesses.

Attention-getting ideas
Some of the first useful problems quantum computers will probably tackle will be to simulate small molecules or chemical reactions. From there, the computers could go on to speed the search for new drugs or kick-start the development of energy-saving catalysts to accelerate chemical reactions. To find the best material for a particular job, quantum computers could search through millions of possibilities to pinpoint the ideal choice, for example, ultrastrong polymers for use in airplane wings. Advertisers could use a quantum algorithm to improve their product recommendations — dishing out an ad for that new cell phone just when you’re on the verge of purchasing one.

Quantum computers could provide a boost to machine learning, too, allowing for nearly flawless handwriting recognition or helping self-driving cars assess the flood of data pouring in from their sensors to swerve away from a child running into the street. And scientists might use quantum computers to explore exotic realms of physics, simulating what might happen deep inside a black hole, for example.

But quantum computers won’t reach their real potential — which will require harnessing the power of millions of qubits — for more than a decade. Exactly what possibilities exist for the long-term future of quantum computers is still up in the air.

The outlook is similar to the patchy vision that surrounded the development of standard computers — which quantum scientists refer to as “classical” computers — in the middle of the 20th century. When they began to tinker with electronic computers, scientists couldn’t fathom all of the eventual applications; they just knew the machines possessed great power. From that initial promise, classical computers have become indispensable in science and business, dominating daily life, with handheld smartphones becoming constant companions (SN: 4/1/17, p. 18).
Since the 1980s, when the idea of a quantum computer first attracted interest, progress has come in fits and starts. Without the ability to create real quantum computers, the work remained theoretical, and it wasn’t clear when — or if — quantum computations would be achievable. Now, with the small quantum computers at hand, and new developments coming swiftly, scientists and corporations are preparing for a new technology that finally seems within reach.

“Companies are really paying attention,” Microsoft’s Krysta Svore said March 13 in New Orleans during a packed session at a meeting of the American Physical Society. Enthusiastic physicists filled the room and huddled at the doorways, straining to hear as she spoke. Svore and her team are exploring what these nascent quantum computers might eventually be capable of. “We’re very excited about the potential to really revolutionize … what we can compute.”

Anatomy of a qubit
Quantum computing’s promise is rooted in quantum mechanics, the counterintuitive physics that governs tiny entities such as atoms, electrons and molecules. The basic element of a quantum computer is the qubit (pronounced “CUE-bit”). Unlike a standard computer bit, which can take on a value of 0 or 1, a qubit can be 0, 1 or a combination of the two — a sort of purgatory between 0 and 1 known as a quantum super­position. When a qubit is measured, there’s some chance of getting 0 and some chance of getting 1. But before it’s measured, it’s both 0 and 1.

Because qubits can represent 0 and 1 simultaneously, they can encode a wealth of information. In computations, both possibilities — 0 and 1 — are operated on at the same time, allowing for a sort of parallel computation that speeds up solutions.

Another qubit quirk: Their properties can be intertwined through the quantum phenomenon of entanglement (SN: 4/29/17, p. 8). A measurement of one qubit in an entangled pair instantly reveals the value of its partner, even if they are far apart — what Albert Einstein called “spooky action at a distance.”
Such weird quantum properties can make for superefficient calculations. But the approach won’t speed up solutions for every problem thrown at it. Quantum calculators are particularly suited to certain types of puzzles, the kind for which correct answers can be selected by a process called quantum interference. Through quantum interference, the correct answer is amplified while others are canceled out, like sets of ripples meeting one another in a lake, causing some peaks to become larger and others to disappear.

One of the most famous potential uses for quantum computers is breaking up large integers into their prime factors. For classical computers, this task is so difficult that credit card data and other sensitive information are secured via encryption based on factoring numbers. Eventually, a large enough quantum computer could break this type of encryption, factoring numbers that would take millions of years for a classical computer to crack.

Quantum computers also promise to speed up searches, using qubits to more efficiently pick out an information needle in a data haystack.

Qubits can be made using a variety of materials, including ions, silicon or superconductors, which conduct electricity without resistance. Unfortunately, none of these technologies allow for a computer that will fit easily on a desktop. Though the computer chips themselves are tiny, they depend on large cooling systems, vacuum chambers or other bulky equipment to maintain the delicate quantum properties of the qubits. Quantum computers will probably be confined to specialized laboratories for the foreseeable future, to be accessed remotely via the internet.

Going supreme
That vision of Web-connected quantum computers has already begun to Quantum computing is exciting. It’s coming, and we want a lot more people to be well-versed in itmaterialize. In 2016, IBM unveiled the Quantum Experience, a quantum computer that anyone around the world can access online for free.
With only five qubits, the Quantum Experience is “limited in what you can do,” says Jerry Chow, who manages IBM’s experimental quantum computing group. (IBM’s 16-qubit computer is in beta testing, so Quantum Experience users are just beginning to get their hands on it.) Despite its limitations, the Quantum Experience has allowed scientists, computer programmers and the public to become familiar with programming quantum computers — which follow different rules than standard computers and therefore require new ways of thinking about problems. “Quantum computing is exciting. It’s coming, and we want a lot more people to be well-versed in it,” Chow says. “That’ll make the development and the advancement even faster.”

But to fully jump-start quantum computing, scientists will need to prove that their machines can outperform the best standard computers. “This step is important to convince the community that you’re building an actual quantum computer,” says quantum physicist Simon Devitt of Macquarie University in Sydney. A demonstration of such quantum supremacy could come by the end of the year or in 2018, Devitt predicts.

Researchers from Google set out a strategy to demonstrate quantum supremacy, posted online at arXiv.org in 2016. They proposed an algorithm that, if run on a large enough quantum computer, would produce results that couldn’t be replicated by the world’s most powerful supercomputers.

The method involves performing random operations on the qubits, and measuring the distribution of answers that are spit out. Getting the same distribution on a classical supercomputer would require simulating the complex inner workings of a quantum computer. Simulating a quantum computer with more than about 45 qubits becomes unmanageable. Supercomputers haven’t been able to reach these quantum wilds.

To enter this hinterland, Google, which has a nine-qubit computer, has aggressive plans to scale up to 49 qubits. “We’re pretty optimistic,” says Google’s John Martinis, also a physicist at the University of California, Santa Barbara.

Martinis and colleagues plan to proceed in stages, working out the kinks along the way. “You build something, and then if it’s not working exquisitely well, then you don’t do the next one — you fix what’s going on,” he says. The researchers are currently developing quantum computers of 15 and 22 qubits.

IBM, like Google, also plans to go big. In March, the company announced it would build a 50-qubit computer in the next few years and make it available to businesses eager to be among the first adopters of the burgeoning technology. Just two months later, in May, IBM announced that its scientists had created the 16-qubit quantum computer, as well as a 17-qubit prototype that will be a technological jumping-off point for the company’s future line of commercial computers.
But a quantum computer is much more than the sum of its qubits. “One of the real key aspects about scaling up is not simply … qubit number, but really improving the device performance,” Chow says. So IBM researchers are focusing on a standard they call “quantum volume,” which takes into account several factors. These include the number of qubits, how each qubit is connected to its neighbors, how quickly errors slip into calculations and how many operations can be performed at once. “These are all factors that really give your quantum processor its power,” Chow says.

Errors are a major obstacle to boosting quantum volume. With their delicate quantum properties, qubits can accumulate glitches with each operation. Qubits must resist these errors or calculations quickly become unreliable. Eventually, quantum computers with many qubits will be able to fix errors that crop up, through a procedure known as error correction. Still, to boost the complexity of calculations quantum computers can take on, qubit reliability will need to keep improving.

Different technologies for forming qubits have various strengths and weaknesses, which affect quantum volume. IBM and Google build their qubits out of superconducting materials, as do many academic scientists. In superconductors cooled to extremely low temperatures, electrons flow unimpeded. To fashion superconducting qubits, scientists form circuits in which current flows inside a loop of wire made of aluminum or another superconducting material.

Several teams of academic researchers create qubits from single ions, trapped in place and probed with lasers. Intel and others are working with qubits fabricated from tiny bits of silicon known as quantum dots (SN: 7/11/15, p. 22). Microsoft is studying what are known as topological qubits, which would be extra-resistant to errors creeping into calculations. Qubits can even be forged from diamond, using defects in the crystal that isolate a single electron. Photonic quantum computers, meanwhile, make calculations using particles of light. A Chinese-led team demonstrated in a paper published May 1 in Nature Photonics that a light-based quantum computer could outperform the earliest electronic computers on a particular problem.

One company, D-Wave, claims to have a quantum computer that can perform serious calculations, albeit using a more limited strategy than other quantum computers (SN: 7/26/14, p. 6). But many scientists are skeptical about the approach. “The general consensus at the moment is that something quantum is happening, but it’s still very unclear what it is,” says Devitt.

Identical ions
While superconducting qubits have received the most attention from giants like IBM and Google, underdogs taking different approaches could eventually pass these companies by. One potential upstart is Chris Monroe, who crafts ion-based quantum computers.
On a walkway near his office on the University of Maryland campus in College Park, a banner featuring a larger-than-life portrait of Monroe adorns a fence. The message: Monroe’s quantum computers are a “fearless idea.” The banner is part of an advertising campaign featuring several of the university’s researchers, but Monroe seems an apt choice, because his research bucks the trend of working with superconducting qubits.

Monroe and his small army of researchers arrange ions in neat lines, manipulating them with lasers. In a paper published in Nature in 2016, Monroe and colleagues debuted a five-qubit quantum computer, made of ytterbium ions, allowing scientists to carry out various quantum computations. A 32-ion computer is in the works, he says.

Monroe’s labs — he has half a dozen of them on campus — don’t resemble anything normally associated with computers. Tables hold an indecipherable mess of lenses and mirrors, surrounding a vacuum chamber that houses the ions. As with IBM’s computer, although the full package is bulky, the quantum part is minuscule: The chain of ions spans just hundredths of a millimeter.

Scientists in laser goggles tend to the whole setup. The foreign nature of the equipment explains why ion technology for quantum computing hasn’t taken off yet, Monroe says. So he and colleagues took matters into their own hands, creating a start-up called IonQ, which plans to refine ion computers to make them easier to work with.

Monroe points out a few advantages of his technology. In particular, ions of the same type are identical. In other systems, tiny differences between qubits can muck up a quantum computer’s operations. As quantum computers scale up, Monroe says, there will be a big price to pay for those small differences. “Having qubits that are identical, over millions of them, is going to be really important.”

In a paper published in March in Proceedings of the National Academy of Sciences, Monroe and colleagues compared their quantum computer with IBM’s Quantum Experience. The ion computer performed operations more slowly than IBM’s superconducting one, but it benefited from being more interconnected — each ion can be entangled with any other ion, whereas IBM’s qubits can be entangled only with adjacent qubits. That interconnectedness means that calculations can be performed in fewer steps, helping to make up for the slower operation speed, and minimizing the opportunity for errors.
Early applications
Computers like Monroe’s are still far from unlocking the full power of quantum computing. To perform increasingly complex tasks, scientists will have to correct the errors that slip into calculations, fixing problems on the fly by spreading information out among many qubits. Unfortunately, such error correction multiplies the number of qubits required by a factor of 10, 100 or even thousands, depending on the quality of the qubits. Fully error-corrected quantum computers will require millions of qubits. That’s still a long way off.

So scientists are sketching out some simple problems that quantum computers could dig into without error correction. One of the most important early applications will be to study the chemistry of small molecules or simple reactions, by using quantum computers to simulate the quantum mechanics of chemical systems. In 2016, scientists from Google, Harvard University and other institutions performed such a quantum simulation of a hydrogen molecule. Hydrogen has already been simulated with classical computers with similar results, but more complex molecules could follow as quantum computers scale up.

Once error-corrected quantum computers appear, many quantum physicists have their eye on one chemistry problem in particular: making fertilizer. Though it seems an unlikely mission for quantum physicists, the task illustrates the game-changing potential of quantum computers.

The Haber-Bosch process, which is used to create nitrogen-rich fertilizers, is hugely energy intensive, demanding high temperatures and pressures. The process, essential for modern farming, consumes around 1 percent of the world’s energy supply. There may be a better way. Nitrogen-fixing bacteria easily extract nitrogen from the air, thanks to the enzyme nitrogenase. Quantum computers could help simulate this enzyme and reveal its properties, perhaps allowing scientists “to design a catalyst to improve the nitrogen fixation reaction, make it more efficient, and save on the world’s energy,” says Microsoft’s Svore. “That’s the kind of thing we want to do on a quantum computer. And for that problem it looks like we’ll need error correction.”

Pinpointing applications that don’t require error correction is difficult, and the possibilities are not fully mapped out. “It’s not because they don’t exist; I think it’s because physicists are not the right people to be finding them,” says Devitt, of Macquarie. Once the hardware is available, the thinking goes, computer scientists will come up with new ideas.

That’s why companies like IBM are pushing their quantum computers to users via the Web. “A lot of these companies are realizing that they need people to start playing around with these things,” Devitt says.

Quantum scientists are trekking into a new, uncharted realm of computation, bringing computer programmers along for the ride. The capabilities of these fledgling systems could reshape the way society uses computers.

Eventually, quantum computers may become part of the fabric of our technological society. Quantum computers could become integrated into a quantum internet, for example, which would be more secure than what exists today (SN: 10/15/16, p. 13).

“Quantum computers and quantum communication effectively allow you to do things in a much more private way,” says physicist Seth Lloyd of MIT, who envisions Web searches that not even the search engine can spy on.

There are probably plenty more uses for quantum computers that nobody has thought up yet.

“We’re not sure exactly what these are going to be used for. That makes it a little weird,” Monroe says. But, he maintains, the computers will find their niches. “Build it and they will come.”

Tsutomu Miyasaka was on a mission to build a better solar cell. It was the early 2000s, and the Japanese scientist wanted to replace the delicate molecules that he was using to capture sunlight with a sturdier, more effective option.

So when a student told him about an unfamiliar material with unusual properties, Miyasaka had to try it. The material was “very strange,” he says, but he was always keen on testing anything that might respond to light.
Other scientists were running electricity through the material, called a perovskite, to generate light. Miyasaka, at Toin University of Yokohama in Japan, wanted to know if the material could also do the opposite: soak up sunlight and convert it into electricity. To his surprise, the idea worked. When he and his team replaced the light-sensitive components of a solar cell with a very thin layer of the perovskite, the illuminated cell pumped out a little bit of electric current.

The result, reported in 2009 in the Journal of the American Chemical Society, piqued the interest of other scientists, too. The perovskite’s properties made it (and others in the perovskite family) well-suited to efficiently generate energy from sunlight. Perhaps, some scientists thought, this perovskite might someday be able to outperform silicon, the light-absorbing material used in more than 90 percent of solar cells around the world.
Initial excitement quickly translated into promising early results. An important metric for any solar cell is how efficient it is — that is, how much of the sunlight that strikes its surface actually gets converted to electricity. By that standard, perovskite solar cells have shone, increasing in efficiency faster than any previous solar cell material in history. The meager 3.8 percent efficiency reported by Miyasaka’s team in 2009 is up to 22 percent this year. Today, the material is almost on par with silicon, which scientists have been tinkering with for more than 60 years to bring to a similar efficiency level.
“People are very excited because [perovskite’s] efficiency number has climbed so fast. It really feels like this is the thing to be working on right now,” says Jao van de Lagemaat, a chemist at the National Renewable Energy Laboratory in Golden, Colo.

Now, perovskite solar cells are at something of a crossroads. Lab studies have proved their potential: They are cheaper and easier to fabricate than time-tested silicon solar cells. Though perovskites are unlikely to completely replace silicon, the newer materials could piggyback onto existing silicon cells to create extra-effective cells. Perovskites could also harness solar energy in new applications where traditional silicon cells fall flat — as light-absorbing coatings on windows, for instance, or as solar panels that work on cloudy days or even absorb ambient sunlight indoors.

Whether perovskites can make that leap, though, depends on current research efforts to fix some drawbacks. Their tendency to degrade under heat and humidity, for example, is not a great characteristic for a product meant to spend hours in the sun. So scientists are trying to boost stability without killing efficiency.

“There are challenges, but I think we’re well on our way to getting this stuff stable enough,” says Henry Snaith, a physicist at the University of Oxford. Finding a niche for perovskites in an industry so dominated by silicon, however, requires thinking about solar energy in creative ways.

Leaping electrons
Perovskites flew under the radar for years before becoming solar stars. The first known perovskite was a mineral, calcium titanate, or CaTiO3, discovered in the 19th century. In more recent years, perovskites have expanded to a class of compounds with a similar structure and chemical recipe — a 1:1:3 ingredient ratio — that can be tweaked with different elements to make different “flavors.”

But the perovskites being studied for the light-absorbing layer of solar cells are mostly lab creations. Many are lead halide perovskites, which combine a lead ion and three ions of iodine or a related element, such as bromine, with a third type of ion (usually something like methylammonium). Those ingredients link together to form perovskites’ hallmark cagelike pyramid-on-pyramid structure. Swapping out different ingredients (replacing lead with tin, for instance) can yield many kinds of perovskites, all with slightly different chemical properties but the same basic crystal structure.

Perovskites owe their solar skills to the way their electrons interact with light. When sunlight shines on a solar panel, photons — tiny packets of light energy — bombard the panel’s surface like a barrage of bullets and get absorbed. When a photon is absorbed into the solar cell, it can share some of its energy with a negatively charged electron. Electrons are attracted to the positively charged nucleus of an atom. But a photon can give an electron enough energy to escape that pull, much like a video game character getting a power-up to jump a motorbike across a ravine. As the energized electron leaps away, it leaves behind a positively charged hole. A separate layer of the solar cell collects the electrons, ferrying them off as electric current.

The amount of energy needed to kick an electron over the ravine is different for every material. And not all photon power-ups are created equal. Sunlight contains low-energy photons (infrared light) and high-energy photons (sunburn-causing ultraviolet radiation), as well as all of the visible light in between.

Photons with too little energy “will just sail right on through” the light-catching layer and never get absorbed, says Daniel Friedman, a photovoltaic researcher at the National Renewable Energy Lab. Only a photon that comes in with energy higher than the amount needed to power up an electron will get absorbed. But any excess energy a photon carries beyond what’s needed to boost up an electron gets lost as heat. The more heat lost, the more inefficient the cell.
Because the photons in sunlight vary so much in energy, no solar cell will ever be able to capture and optimally use every photon that comes its way. So you pick a material, like silicon, that’s a good compromise — one that catches a decent number of photons but doesn’t waste too much energy as heat, Friedman says.

Although it has dominated the solar cell industry, silicon can’t fully use the energy from higher-energy photons; the material’s solar conversion efficiency tops out at around 30 percent in theory and has hit 20-some percent in practice. Perovskites could do better. The electrons inside perovskite crystals require a bit more energy to dislodge. So when higher-energy photons come into the solar cell, they devote more of their energy to dislodging electrons and generating electric current, and waste less as heat. Plus, by changing the ingredients and their ratios in a perovskite, scientists can adjust the photons it catches. Using different types of perovskites across multiple layers could allow solar cells to more effectively absorb a broader range of photons.

Perovskites have a second efficiency perk. When a photon excites an electron inside a material and leaves behind a positively charged hole, there’s a tendency for the electron to slide right back into a hole. This recombination, as it’s known, is inefficient — an electron that could have fed an electric current instead just stays put.

In perovskites, though, excited electrons usually migrate quite far from their holes, Snaith and others have found by testing many varieties of the material. That boosts the chances the electrons will make it out of the perovskite layer without landing back in a hole.

“It’s a very rare property,” Miyasaka says. It makes for an efficient sunlight absorber.

Some properties of perovskites also make them easier than silicon to turn into solar cells. Making a conventional silicon solar cell requires many steps, all done in just the right order at just the right temperature — something like baking a fragile soufflé. The crystals of silicon have to be perfect, because even small defects in the material can hurt its efficiency. The need for such precision makes silicon solar cells more expensive to produce.

Perovskites are more like brownies from a box — simpler, less finicky. “You can make it in an office, basically,” says materials scientist Robert Chang of Northwestern University in Evanston, Ill. He’s exaggerating, but only a little. Perovskites are made by essentially mixing a bunch of ingredients together and depositing them on a surface in a thin, even film. And while making crystalline silicon requires temperatures up to 2000° Celsius, perovskite crystals form at easier-to-reach temperatures — lower than 200°.

Seeking stability
In many ways, perovskites have become even more promising solar cell materials over time, as scientists have uncovered exciting new properties and finessed the materials’ use. But no material is perfect. So now, scientists are searching for ways to overcome perovskites’ real-world limitations. The most pressing issue is their instability, van de Lagemaat says. The high efficiency levels reported from labs often last only days or hours before the materials break down.

Tackling stability is a less flashy problem than chasing efficiency records, van de Lagemaat points out, which is perhaps why it’s only now getting attention. Stability isn’t a single number that you can flaunt, like an efficiency value. It’s also a bit harder to define, especially since how long a solar cell lasts depends on environmental conditions like humidity and precipitation levels, which vary by location.

Encapsulating the cell with water-resistant coatings is one strategy, but some scientists want to bake stability into the material itself. To do that, they’re experimenting with different perovskite designs. For instance, solar cells containing stacks of flat, graphenelike sheets of perovskites seem to hold up better than solar cells with the standard three-dimensional crystal and its interwoven layers.

In these 2-D perovskites, some of the methylammonium ions are replaced by something larger, like butylammonium. Swapping in the bigger ion forces the crystal to form in sheets just nanometers thick, which stack on top of each other like pages in a book, says chemist Aditya Mohite of Los Alamos National Laboratory in New Mexico. The butylammonium ion, which naturally repels water, forms spacer layers between the 2-D sheets and stops water from permeating into the crystal.
Getting the 2-D layers to line up just right has proved tricky, Mohite says. But by precisely controlling the way the layers form, he and colleagues created a solar cell that runs at 12.5 percent efficiency while standing up to light and humidity longer than a similar 3-D model, the team reported in 2016 in Nature. Although it was protected with a layer of glass, the 3-D perovskite solar cell lost performance rapidly, within a few days, while the 2-D perovskite withered only slightly. (After three months, the 2-D version was still working almost as well as it had been at the beginning.)

Despite the seemingly complex structure of the 2-D perovskites, they are no more complicated to make than their 3-D counterparts, says Mercouri Kanatzidis, a chemist at Northwestern and a collaborator on the 2-D perovskite project. With the right ingredients, he says, “they form on their own.”

His goal now is to boost the efficiency of 2-D perovskite cells, which don’t yet match up to their 3-D counterparts. And he’s testing different water-repelling ions to reach an ideal stability without sacrificing efficiency.

Other scientists have mixed 2-D and 3-D perovskites to create an ultra-long-lasting cell — at least by perovskite standards. A solar panel made of these cells ran at only 11 percent efficiency, but held up for 10,000 hours of illumination, or more than a year, according to research published in June in Nature Communications. And, importantly, that efficiency was maintained over an area of about 50 square centimeters, more on par with real-world conditions than the teeny-tiny cells made in most research labs.

A place for perovskites?
With boosts to their stability, perovskite solar cells are getting closer to commercial reality. And scientists are assessing where the light-capturing material might actually make its mark.

Some fans have pitted perovskites head-to-head with silicon, suggesting the newbie could one day replace the time-tested material. But a total takeover probably isn’t a realistic goal, says Sarah Kurtz, codirector of the National Center for Photovoltaics at the National Renewable Energy Lab.

“People have been saying for decades that silicon can’t get lower in cost to meet our needs,” Kurtz says. But, she points out, the price of solar energy from silicon-based panels has dropped far lower than people originally expected. There are a lot of silicon solar panels out there, and a lot of commercial manufacturing plants already set up to deal with silicon. That’s a barrier to a new technology, no matter how great it is. Other silicon alternatives face the same limitation. “Historically, silicon has always been dominant,” Kurtz says.
For Snaith, that’s not a problem. He cofounded Oxford Photo-voltaics Limited, one of the first companies trying to commercialize perovskite solar cells. His team is developing a solar cell with a perovskite layer over a standard silicon cell to make a super-efficient double-decker cell. That way, Snaith says, the team can capitalize on the massive amount of machinery already set up to build commercial silicon solar cells.
A perovskite layer on top of silicon would absorb higher-energy photons and turn them into electricity. Lower-energy photons that couldn’t excite the perovskite’s electrons would pass through to the silicon layer, where they could still generate current. By combining multiple materials in this way, it’s possible to catch more photons, making a more efficient cell.

That idea isn’t new, Snaith points out: For years, scientists have been layering various solar cell materials in this way. But these double-decker cells have traditionally been expensive and complicated to make, limiting their applications. Perovskites’ ease of fabrication could change the game. Snaith’s team is seeing some improvement already, bumping the efficiency of a silicon solar cell from 10 to 23.6 percent by adding a perovskite layer, for example. The team reported that result online in February in Nature Energy.

Rather than compete with silicon solar panels for space on sunny rooftops and in open fields, perovskites could also bring solar energy to totally new venues.

“I don’t think it’s smart for perovskites to compete with silicon,” Miyasaka says. Perovskites excel in other areas. “There’s a whole world of applications where silicon can’t be applied.”

Silicon solar cells don’t work as well on rainy or cloudy days, or indoors, where light is less direct, he says. Perovskites shine in these situations. And while traditional silicon solar cells are opaque, very thin films of perovskites could be printed onto glass to make sunlight-capturing windows. That could be a way to bring solar power to new places, turning glassy skyscrapers into serious power sources, for example. Perovskites could even be printed on flexible plastics to make solar-powered coatings that charge cell phones.

That printing process is getting closer to reality: Scientists at the University of Toronto recently reported a way to make all layers of a perovskite solar cell at temperatures below 150° — including the light-absorbing perovskite layer, but also the background workhorse layers that carry the electrons away and funnel them into current. That could streamline and simplify the production process, making mass newspaper-style printing of perovskite solar cells more doable.

Printing perovskite solar cells on glass is also an area of interest for Oxford Photovoltaics, Snaith says. The company’s ultimate target is to build a perovskite cell that will last 25 years, as long as a traditional silicon cell.

As far as last meals go, squid isn’t a bad choice. Cephalopod remains appear to dominate the stomach contents of a newly analyzed ichthyosaur fossil from nearly 200 million years ago.

The ancient marine reptiles once roamed Jurassic seas and commonly pop up in England’s fossil-rich coast near Lyme Regis. But a lot of ichthyosaur museum specimens lack records of where they came from, making their age difficult to place.

Dean Lomax of the University of Manchester and his colleagues reexamined one such fossil. Based on its skull, they identified the creature as a newborn Ichthyosaurus communis. Microfossils of shrimp and amoeba species around the ichthyosaur put the specimen at 199 million to 196 million years old, the researchers estimate.

Tiny hook structures stand out in the newborn’s ribs — most likely the remnants of prehistoric black squid arms. Another baby ichthyosaur fossil that lived more recently had a stomach full of fish scales. So the new find suggests a shift in the menu for young ichthyosaurs at some point in their evolutionary history, the researchers write October 3 in Historical Biology.