Will Plants Ever Fertilize Themselves?

Biologists aim to engineer crops that can eat nitrogen straight from the air.

By Matthew Hutson

February 6, 2024. New Yorker, New Yorker


Here’s the thing about nitrogen. It’s essential for life—a key ingredient in both DNA and proteins. It also makes up seventy-eight per cent of the air we breathe. It would be useful for us if we could pull nitrogen out of the air and make use of it inside our bodies. But nitrogen atoms typically come in pairs—N2 molecules—that our cells can’t easily pry apart. Instead, we get our nitrogen by eating plants, or by eating animals that eat plants (or animals that eat animals that eat plants).

Unfortunately, plants are in the same boat. They can’t make direct use of atmospheric nitrogen, either. In fact, the only cells on Earth that can render nitrogen palatable for plants and animals are certain kinds of microbes. These microbes, known as diazotrophs, “fix” nitrogen, by using N2 to make NH3, also known as ammonia. The nitrogen in this ammonia is ready to eat. The survival of every plant and animal on Earth depends on the work of diazotrophs, which must fix enough nitrogen to keep the biosphere’s machinery running.

For most of human history, the world’s diazotrophs fixed enough nitrogen to keep up with the human appetite. But that started to change about a hundred and twenty years ago. In 1898, William Crookes, the president of the British Association for the Advancement of Science, gave an alarming inaugural address. “England and all civilized nations stand in deadly peril of not having enough to eat,” he said. “Our wheat-producing soil is totally unequal to the strain put upon it.” Ordinarily, agricultural soil is bolstered with fertilizer, which supplies nitrogen and other nutrients to crops. But Crookes noted that sodium-nitrate deposits in Chile—a major source of usable nitrogen for plants—would soon dwindle. He ran through several untapped sources of ammonia, including coal distillation and sewage, but none were up to the task.

“There is a gleam of light amid this darkness of despondency,” Crookes told his audience. “In its free state nitrogen is one of the most abundant and pervading bodies on the face of the earth.” Scientists had tried for years to fix atmospheric nitrogen, he said, including by passing current through the air. Lightning fixes millions of tons of nitrogen each year. But putting lightning in a bottle had turned out to be expensive and difficult. “The fixation of atmospheric nitrogen, therefore, is one of the great discoveries awaiting the ingenuity of chemists,” he said.

Crookes didn’t need to wait long. In 1909, the German chemist Fritz Haber demonstrated a nascent but scalable method for turning N2 into ammonia. Carl Bosch, at the chemical-and-dye company B.A.S.F., industrialized Haber’s method, and they each earned a Nobel Prize. Today, the Haber-Bosch process produces roughly two hundred million tons of ammonia a year, and has allowed the human population to reach eight billion. Without it, crops would require four times the area that they do now, covering half of Earth’s ice-free landmass.

About half of the nitrogen in your body comes from the Haber-Bosch process. But its costs are enormous. The reaction happens at approximately a thousand degrees Fahrenheit and three hundred times atmospheric pressure, using between one and two per cent of the world’s energy. Meanwhile, fertilizer runoff pollutes the environment. And yet, all the while, humble bacteria in the dirt are fixing nitrogen all day long. Recent developments in biotechnology, unimaginable in Crookes’s time, suggest a new possibility: we might be able to extract these bacteria’s mechanisms and place them inside plants. Some crops, like legumes, act as hosts for diazotrophs, which fix nitrogen from within the plant. But cereals—including wheat and rice, staple crops for many people around the world—are dependent on eating nitrogen in the surrounding soil that has already been broken down by diazotrophs or by the Haber-Bosch process. Researchers are hoping to transfer genes from diazotrophs into cereals, giving them the power to fix nitrogen. We may someday have plants that can fertilize themselves.

Self-fertilizing plants have been a scientific goal since the nineteen-seventies. In 1972, two British scientists published a paper in Nature reporting that they’d induced E. coli, a bacterial species that does not normally fix nitrogen, to do so, by importing genes from another bacterial species that does. Quick progress to plants seemed imminent. A 1975 Science paper, citing the Nature paper among others, carried the title “Nitrogen Fixation Research: A Key to World Food?”

“That was a long time ago, and we haven’t got very far, have we,” Ray Dixon, a molecular biologist at the John Innes Centre, in England, and one of the two Nature authors, told me, laughing. “But, in my defense, or the defense of the field, we didn’t have the techniques to do this kind of stuff. It’s only in the last ten years when any of this has really been feasible.”

Diazotrophs fix nitrogen using an enzyme complex called nitrogenase, which is made up of several proteins and helper molecules. The system is like a little assembly line. Essentially, one component uses ATP, an energy-carrying molecule, to funnel electrons into a second component. This component splits N2 in half, binding each atom to hydrogen taken from water and forming two molecules of ammonia. Other proteins supply these two components with metal clusters, containing iron and sometimes molybdenum. Two clusters collect and feed electrons to the third cluster, which splits the N2. The whole system, which has been likened to an anvil for splitting N2, requires at least ten to twenty genes (no one is quite sure of the minimum), though some bacteria use fifty or more.

The largest hurdle is assembling and inserting the metal clusters. N2 is floating around in the air, but metal is harder to come by. “That has to come from other pathways that we’re basically begging and borrowing from,” Craig Wood, a plant synthetic biologist at C.S.I.R.O., Australia’s science agency, told me. The cluster molecules “are being made and dissolved and used all the time. It’s like an economy, and it’s tightly regulated.” Once you obtain the metal clusters, you need to find the right holes to slip them into. “This is the trickiest metalloenzyme known in nature,” Wood said.

Getting nitrogenase up and running outside of its bacterial home is a bit like taking a finely tuned Rube Goldberg machine and reassembling it on Mars. You need the right genes to build proteins at the right times and in the right quantities, and the proteins—strings of amino acids—need to fold up into the right shapes and interact in the right ways. “The whole thing can just crumble if you don’t have the correct expression of all the genes,” Luis Rubio, a biochemist at the Spanish National Research Council, said. “I mean, just one gene is off and the whole system collapses. It’s very, very delicate.” Dennis Dean, a biochemist at Virginia Tech, echoed the sentiment: “I have a colleague I’ve collaborated with for many years, and he said that nitrogenase is good evidence that if God exists, he’s very devious.”

Researchers aren’t even sure where in a plant to insert the proteins that would create the nitrogenase. Two possibilities stand out: chloroplasts and mitochondria, both organelles that float around inside plant cells. Neither is perfect; it’s like choosing whether to build your Rube Goldberg machine on Mars or the moon. Chloroplasts have several advantages. A crucial one is that they already possess some of the machinery required to assemble nitrogenases. But photosynthesis, which occurs in chloroplasts, produces oxygen, and oxygen destroys nitrogenase. Scientists, therefore, are working on ways to generate nitrogenase only at night, when photosynthesis won’t occur.

Most of the researchers I spoke with leaned toward mitochondria. These structures contain less oxygen, and, as the cell’s little engines, they provide an ample supply of ATP. One complication is that it’s almost impossible to edit mitochondrial DNA; this makes it difficult to program the mitochondria to manufacture the molecular parts required to build nitrogenase. But there’s a way around the problem: according to Wood, it’s possible to engineer genes so that, when they’re expressed in the nucleus, they create proteins which bear “a little postage stamp”—a marker that delivers them to specific locales in the cell. By this means, genes in the cell’s core could produce the component parts of nitrogenase, and tag them so that they’ll ship out to the mitochondria. Wood and Rubio are both working to create such proteins, in plants and yeast.

Finding a building site is one thing; getting the parts delivered is another. The parts themselves also have to be well manufactured. The proteins involved in nitrogenase aren’t so easy to make. Dixon, Rubio, and Wood have worked out how to build a number of candidate parts—proteins with names like NifH, NifDK, NifU, NifB, and NifEN. All of them, or their equivalents, must “fold” correctly, taking on the proper shapes after they’ve been created by means of genes.

There are two ways to make proteins that are usually built in microbes more likely to fold correctly in the mitochondria of yeasts or plants. One way is to gently reëngineer them, making slight modifications. The other is to capitalize on nature’s biodiversity by surveying proteins from a range of bacteria and finding those that adapt best to the new environment. Often, these proteins come from thermophilic microbes—those that tolerate hot temperatures. “The funny thing is that we will end up with a pathway that is formed by proteins from five, six, or seven different organisms,” Rubio said. Adapting proteins and engineering them aren’t mutually exclusive; Rubio’s lab is currently exploring how machine learning might be used to optimize the proteins that they’ve found.

There are different kinds of nitrogenases, and they depend on different metals. The one with iron and molybdenum is a bit like a Ferrari, Wood told me: it’s the most well known but also the most complex. Another version uses vanadium instead of molybdenum; it’s also complex and not much studied. “But there’s a third one, which is a little bit like your VW Beetle from the nineteen-sixties,” Wood said. This nitrogenase uses only iron, without molybdenum or vanadium. It’s slower, but also simpler, and iron is easier to find in nature than the other metals. Last June, Wood’s lab reported that they’d got plant cells to express four protein ingredients of the iron-only nitrogenase in their nuclei and insert them into mitochondria. Some of the proteins also folded up correctly. “From a public point of view, it probably doesn’t feel like much is happening,” Wood said. “But, from our point of view, we’re basically assembling a Rubik’s Cube a step at a time. And people have got a fairly clear path forward.”

Dennis Dean, who works in the U.S., has been gratified watching and advising Dixon, Rubio, and Wood, in England, Spain, and Australia, respectively; meanwhile, his lab has discovered that some plant genes perform a similar function as bacterial nitrogenase genes. Researchers now think that they’ll need to insert only five to seven bacterial genes into plants. “We have absolutely spectacular scientists who are also spectacularly good people,” Dean said. “They’re not buddies who are, you know, giving away their secrets. But they work together in a very collegial and collaborative way, in spite of being after the same prize. It’s one of those things that’s rare in science.” Dean, like Dixon, is nearing the end of his career. “This has really been a great ride,” he said.

The ride’s been great, but how close are we? When I first spoke to researchers, two or three years ago, they expected nitrogenase to begin functioning in transgenic yeast in maybe a couple of years, and in model plants a couple of years after that. In more recent conversations, they’ve also said that the yeast target is two or three years away. Getting from yeast to real crops could take more than a decade. The plan, broadly speaking, is to try self-fertilization in tobacco—the crop equivalent of a lab rat—and then in rice. But there are inherent challenges. “The enzyme that fixes nitrogen works really, really slowly, and that’s why it’s so energy-intensive,” Dean said. “What’s going to be really important is optimizing it so the plant cells are making ammonia like gangbusters. And they’ve got to be able to do that without poisoning the cells and making them sick from too much ammonia. In bacteria, this is a very carefully orchestrated process.”

“I worked as a student in a lab trying to do this many years ago,” Sieg Snapp, an agriculturalist and director of the Sustainable Agrifood Systems program at cimmyt (the International Maize and Wheat Improvement Center), near Mexico City, said. “And it proved to involve so many more genes than originally thought.” In addition to the genes for nitrogenase, plants need genes to transport and store fixed nitrogen throughout themselves, for example. Researchers hope that, once scientists have the basic mechanism in place, they can put such details in nature’s hands by selectively breeding successful plants.

Vipula Shukla, a plant biologist and senior program officer for agriculture at the Bill & Melinda Gates Foundation, told me that success will require more than engineering plants. She pointed to the complementary roles of genetics, environment, and management—the so-called G x E x M paradigm. “I think it’ll probably be a while before we see that translation at scale, in a diversity of crops in the field where farmers are cultivating food,” she said. “But the first time it happens will be exciting. The second and third and tenth or hundredth time it happens is when we really get to societal impact.” Along the way, scientists will also have to prove that the technology is safe, durable, and cost-effective.

A day or two into my reading about nitrogen fixation, a possibly obvious question occurred to me: If the nitrogen-fixation process is so vital to plants, why didn’t they evolve to do it themselves? (Perhaps it wasn’t so obvious; given our need for energy and nutrients, one could also ask why humans didn’t evolve to photosynthesize or to eat dirt, questions I don’t regularly contemplate.) The experts I queried had no definite answers, but two plausible ones.

“That worries us,” Rubio said. “You know, this hasn’t happened in two thousand million years. There could be a reason for it.” He went on, “Fixing nitrogen is a very heavy commitment, metabolically speaking, O.K.? Very, very heavy. And it may be incompatible with other things”—such as generating oxygen through photosynthesis. Maybe self-fertilization is too big an ask. The other answer leaves room for optimism: perhaps plants simply haven’t faced enough evolutionary pressure. Why fix nitrogen when diazotrophs do it for you? Only since the advent of industrial agriculture has the productivity of plants fallen short of anyone’s expectations.

There are other approaches to possibly increasing nitrogen fixation on cropland. One is to take a cue from legumes, which, unlike cereals, can fix nitrogen, in a way. They form a symbiosis with bacteria, which infect their roots. The root builds a structure around the infection called a nodule. “So it’s a little nitrogen-fixing factory there on the plant root,” Dixon said. “It’s beautiful.” The plant supplies carbon and nutrients in exchange for ammonia. But genetically programming cereals to grow nodules could be even more complicated than programming them to generate nitrogenase. “It’s like a whole lot of legal teams trying to work out, you know, how are they going to merge?” Wood said, of the plant-microbe symbiosis. “They’re still trying to work out terms and conditions.” (The Gates Foundation has funded both nitrogenase, through Rubio, and nodule engineering.)

Some cereals build looser associations with bacteria, allowing the organisms to form biofilms on their roots. “It’s not as intricate,” Wood said. “But there’s always this trade-off. The plant is not quite sure if the bacteria is actually a pathogen, so there’s a limiting effect. They don’t really want to hand out sugar and carbon to bacteria without being sure what they’re going to get back.” He noted an ingenious maize in Mexico that drips sugar onto the soil from aerial roots; this allows the plant to feed nitrogen-fixing bacteria without risking invasion. But that plant grows slowly. Meanwhile, Bill Gates has invested in Pivot Bio, a company that aims to genetically engineer microbes to more efficiently fix nitrogen in crop soil. Such a solution would make diazotrophs better; it wouldn’t be as profound a transformation as truly self-fertilizing plants.

Although Crookes warned, in 1898, that we stood “in deadly peril of not having enough to eat,” today there’s no global food shortage; we waste about a third of what we produce. But there are local food crises created in part by problems with food distribution and storage. The fertilizer-supply chain is also a factor. Shocks including the pandemic, the war in Ukraine, and inflation have quickly more than doubled the price of fertilizer in some countries, leading to widespread hunger. Even in the best of times, fertilizer production uses vast amounts of energy and resources, and its application poisons the environment. Snapp told me that self-fertilizing crops could reduce agriculture’s greenhouse impact by a third, both by reducing the production and transport of fertilizer and by allowing more crops to be grown locally. Shukla suggested that these engineered plants could help alleviate poverty in regions of the world where crops are a major economic driver and farmers can’t afford synthetic fertilizer. Self-fertilizing plants aren’t an existential need for humanity, but they would be of great benefit.

I came to this topic a few years ago, via an unpublished paper titled “Endowing Plants with the Capacity for Autogenic Nitrogen Fixation.” The abstract boasted, “This study represents a milestone toward realizing the goal of endowing plants with the capacity for self-fertilization.” Comments on the Web page from Rubio, Dixon, and others questioned the results, pointing to, at minimum, sloppy methods and descriptions. (The paper’s authors didn’t reply to my e-mails.) “I’ve been in the field for forty-plus years,” Dean told me. “There’s been so many promises about biological nitrogen fixation. It’s one of the grand challenges. And, quite frankly, this is biotechnology’s biggest failure. We’re fifty years down the road, and we’re not there yet. And the last thing we need is sensational claims.”

“This will happen more often now,” Rubio told me. “I mean, we’re getting closer. We will see more and more of these reports, like, ‘We got it! It’s a miracle! We just put the genes in and it all worked.’ ” The nitrogenase system is complex; it keeps turning out to be more delicate than it looks. But that doesn’t mean we aren’t making progress. Sometime in the next decade or two, we could finally have Rube Goldberg machines on Mars. ♦