As climate change threatens crop yields, we need a second Green Revolution – one that, this time, is driven by genetic engineering.
Mexico, in the 1950s, was the birthplace of the first Green Revolution. An American agronomist named Norman Borlaug was its unlikely hero.
In his early twenties, Borlaug studied forestry and plant pathology at the University of Minnesota and then, for his PhD, studied a pest that infects maple trees. When World War II broke out, Borlaug was rejected from enlisting in the military and moved to Delaware, where he took a job at DuPont as a microbiologist. For one project, he tried to make a “glue that could withstand the warm salt water of the South Pacific.” Borlaug, up to this point in his career, had never worked with wheat, and had taken just one course in plant breeding.
But then, after the War, he moved to Mexico… to breed wheat. Over the next decade, his small team crossed thousands of wheat strains together each year, “while most breeders managed less than 200 per year,” according to a biography from the University of Minnesota. With funding from the Rockefeller and Ford Foundations, Borlaug helped to increase Mexico’s wheat yields six-fold between 1944 and 1963. The country went from a wheat importer to a net exporter in just a few years, and Borlaug won the Nobel Peace Prize in 1970.
This history is well-known, but what happened next is not (at least in the West).
In 1960, the Rockefeller and Ford Foundations took their money and established a research institute devoted entirely to rice, the main food for half the world’s population, in the Philippines. They dubbed the center IRRI: the International Rice Research Institute. Between 1960 and 2020, global rice yields increased by 250 percent, thanks in part to improved breeding techniques developed by IRRI scientists.
A Chinese agronomist, named Yuan Longping, also played an important role. Longping was the first person to breed hybrid rice from two very different parent strains. In 1970, he found a wild variety of rice near a railroad track, bred it with another variety (by painstakingly taking pollen from one and fertilizing the other), and eventually generated offspring with yields 20 percent higher than the rice grown in China at that time. Longping’s achievements are probably just as notable as Borlaug’s, but he did it with less recognition.
In the last two decades, IRRI has bred rice strains that fend off pests, survive floods, and withstand droughts. Most varieties are still made with traditional breeding: Cross two plants together, select offspring that look or ‘behave’ a certain way, and repeat. But this approach is slow and painful. To find a new rice crop with ‘ideal’ features, one must often breed many thousands of plants, and wait months to see the outcomes of each cross.
Now, shifts in temperature and rainfall, especially droughts, are expected to turn around the long trend in rice yield improvements. Ninety percent of all rice is grown in Asia, but studies suggest that rice yields could fall by up to 19 percent by the end of this century. Southeast Asia and sub-Saharan Africa — the two regions that rely on rice most — are already home to more than half of the world’s undernourished people, and so even a small drop in production is concerning.
Our existing strategies to boost food production, like water management and better fertilizers, probably can’t keep up with the planet’s pace of change while maintaining annual improvements in yield. Our best bet may be to use biotechnology instead. Today, it is possible to genetically manipulate rice to make them flood-resistant, drought-tolerant, or to produce larger (and more plentiful) grains.
Over the last decade, there has been astounding progress in gene-editing tools, but plant engineering is still an immature field. It’s tricky to precisely change DNA. A single experiment takes many months in plants because, after editing a gene, they must be regrown from a tiny cluster of cells. In rice, this process takes at least 3-4 months. A similar experiment, in bacteria, can be done in one day.
In other words, experiments are painfully slow at a time when progress must be swift.
The difficulties of gene-editing
Plant genetic engineering is an important tool to boost food production, but our basic understanding about how rice works has lagged behind the meteoric rise in gene-editing technologies. Drug companies invest many tens of billions of dollars, each year, into improving gene-editing tools to manufacture medicines, antibodies and vaccines from human cells. There are far fewer resources devoted to plants. In many cases, crop scientists don’t even know which genes to edit in the first place.
Most plants have never been sequenced (or even studied in the laboratory), and so we often have no idea which genes they contain. More than 6,000 varieties of rice are grown in India alone, and most of them have not been sequenced.
A shortage of DNA sequences may sound like a simple problem to solve — Just go collect some plants! Break ‘em open! Run their DNA on a fancy machine! — but some plants have outrageously large genomes and are thus expensive to study. Only about one in every 625 plants has a genome sequence available. Rice, fortunately, is relatively simple to sequence; the most widely cultivated varieties are diploid (meaning they have two copies of each chromosome, just like humans) and carry relatively small genomes. Rice was the first major crop to be sequenced, back in 2005.
In recent years, scientists have scoured the globe for additional rice cultivars, and sequenced many of them. By 2019, more than 3,000 rice varieties from 89 countries had been collected, sequenced, and uploaded to a public database.
After obtaining a DNA sequence, the next step is to select which genes should be altered (and in which ways) to make plants that have a desired trait, like large leaves or deep roots. This is, in many ways, the most difficult part of genetic engineering.
A genome sequence is merely a long string of letters on a computer — like ATGCTTGACTG — that stretches on and on. The rice genome contains 430 million of these “letters,” or nucleotides, which together encode 39,000 different genes. A large fraction of the genes in rice do not have a known function; we have no clue what they do or why they exist. We don’t even fully understand which genes confer resistance to droughts or floods. Biology is messy, and genetic material is difficult to decipher.
There are three basic ways to figure out a gene’s function. The first is to sequence a lot of different genes, and then use computational tools to infer their functions based on overlaps between sequences. If two sequences are similar, and one of them has a known function, then the second sequence might have a related function.
The second and third ways are more time-intensive: Mess with a gene (knock down or boost its levels) and then watch what happens to the plant afterward. Or, breed many varieties of rice together, sequence them, measure their traits, and then use statistics to match differences in traits to specific changes in the DNA sequence. This third technique is a bit like the studies we do on humans (called genome-wide association studies) to try and link genes to specific traits, such as height or a complex disease, like asthma.
Brook Moyers, a rice geneticist at UMass Boston, has identified dozens of genetic regions that are correlated with drought resistance, across multiple rice chromosomes, using similar approaches. Interestingly, her work suggests that the function of these genes changes depending on the environment in which plants are grown. This makes it even more difficult to study the myriad roles of genes, because biology is constantly in flux. Shifting genes make it difficult to breed plants that can weather both droughts and floods, for reasons we don’t fully understand. Even if a gene’s function is well known, we often don’t know how that gene is regulated – switched ‘on’ or ‘off’ – from one environment to the next.
Once a plant’s genome has been sequenced, and the genes responsible for desired traits are identified, it’s finally time to make genetic edits.
Editing a plant’s genome using CRISPR (there are other ways) sounds simple on paper, but is difficult in reality. Just two things are needed: A protein that cuts DNA, and a strand of RNA that tells the protein where to cut. The RNA latches onto the DNA-cutting protein and “guides” it to a matching sequence of DNA. When a match is found, the protein cleaves the DNA, and new genetic material can be inserted at the break (or, alternatively, one can simply wait for the cut to heal; the resulting “scar” often renders the gene non-functional, thus “knocking” it out of the genome.)
There are many DNA-editing proteins to choose from. Cas9 cuts DNA in two and is the most common CRISPR protein, but there are others: some make a small nick in DNA (without cleaving it), and others “block” a gene’s expression or can even swap individual letters within a sequence, such as an ‘A’ for a ‘G’. It’s also possible to edit many different locations within the genome in a single experiment, simply by adding more guide RNAs. In one experiment, researchers edited 46 DNA locations in rice in a single go.
In another experiment, a research group in China edited random combinations of seven different genes in rice and tested more than 150 offspring to see which ones grew largest. When a combination of genes 1, 4, and 6 were edited at the same time, the rice produced 31 percent more grain than wildtype. (When a paper reports dramatic improvements in crop yields, always look at the baseline values. Some gene-editing studies are performed in specific rice varieties that have naturally low yields; a 50 percent boost may not actually be all that useful.)
As well as boosting yields, CRISPR tools have been used to make rice plants that absorb more fertilizer and have accelerated flowering times, or that are more resistant to drought and salt, compared to unedited varieties. A gene-editing institute at UC Berkeley, founded by Nobel laureate Jennifer Doudna, recently received $11 million from Mark Zuckerberg’s nonprofit to engineer rice that can store excess carbon in soil.
CRISPR can even be used to identify gene functions. For one study, the gene-editing tool was used to eliminate 57 random genes, one at a time, from 30 different high-yielding rice varieties. The edited plants had stunted growth, turned brown, or produced empty seeds. By linking edited genes to physical outcomes, it was possible to infer the functions of DNA sequences.
All this stuff sounds great, but it’s not without challenges. Editing DNA may require just two components (a protein and the guide RNA), but each component must first enter a plant’s physical cells, and this is trickier than it seems.
For several decades, scientists have relied upon two basic techniques for DNA delivery. The first is to harness a microbe, called Agrobacterium, that naturally infects plants, and to basically “trick” it into delivering the gene-editing components. Agrobacterium contains a circular piece of DNA, called a plasmid, to which additional genes (such as those encoding the Cas9 protein, which cuts DNA for CRISPR gene-editing) can be added. The engineered bacteria are then incubated with rice calli, a mass of cells made by bathing seeds in a delicate concoction of hormones. The bacteria infect the cells and, within a few hours, the modified plasmids are carried inside. These genetically-engineered cells are then tended until they grow into a mature plant, which can take quite a while; typically at least 3 months.)
The second way to get DNA into plant cells is to coat gold nanoparticles with the genetic material, and then use a “gene gun” (which costs many times more than a real gun; between $10,000-$30,000) to bombard the plant. Some of the nanoparticles rip through the plant’s cells and leave behind trace amounts of DNA. A single copy of Cas9, and a single guide RNA, are enough to make a genetic edit.
In rice, transformation and editing efficiencies (how many cells take up the genetic material encoding the CRISPR tools and how often those tools actually work as expected) are quite low. It ranges from about 50 percent for the former, and between 5 and 80 percent for the latter, depending on the target genes.
The real killer in plant experiments is just the time required: It takes months to make just one gene-edited rice crop, and to evaluate whether the DNA changes did anything useful. Often, scientists have to repeat this process many times before hitting, finally, upon a crop that has the correct trait. At the end of all this work, scientists have made one cultivar that can be grown in one type of climate, and therefore often need to repeat the process many times over for different types of rice.
Unknown genes, poor DNA delivery, and painfully slow growth all make it difficult to engineer rice. But there are other things that throttle progress, too.
For one, it takes a long time to get approval for a gene-edited crop (more than a decade in some places), and anti-biotech backlash can be fierce. In 2013, a test site for Golden Rice – engineered to contain beta-carotene, which helps to prevent blindness – in the Philippines was razed by activists.
The United States, Argentina, Canada, and the Philippines all regulate plants at the final step — that is, regulators do not care which methods are used to make a crop, as long as the final product could have possibly been made without biotechnology methods. Australia and India regulate gene-edited crops that carry small changes in the genome as if they were normal plants. This is called product-based regulation.
But many countries, including New Zealand and nearly all of Europe, regulate plants based on the process used to create them. In these countries, if CRISPR is used to modify even one letter in DNA, but every other bit of a plant’s genetic material is indistinguishable from a natural variety, then the plant is still regulated as a GMO.
Engineered crops classified as GMOs are tested for nearly everything under the sun: Scientists carefully measure the various molecules within plant cells to ensure that their levels are similar to wildtype varieties, for instance. And if a plant carries a “new” gene that codes for a protein, researchers must prove that it does not cause allergies (often, by purifying the protein and incubating it with human antibodies in blood to see if the two react). Plants that are classified as non-GMO – including those that are gene-edited – are regulated in much the same way as conventional varieties; there is still some safety testing, but not as much.
Golden Rice was tested in field trials for more than a decade, and then eventually deemed ‘comparable’ to non-engineered counterparts. The project has been funded by the Syngenta Foundation (the humanitarian arm of an agricultural company that ‘has no commercial interest in Golden Rice’ in developing countries, but retained rights to its commercialization in places like the U.S. and Europe), USAID, and the Bill & Melinda Gates Foundation.
A few of the engineered Golden Rice strains were also developed by Syngenta, and the company conducted field trials in the summer of 2004. The Golden Rice team was lucky to find a commercial partner; most gene-edited plants never see the light of day because it’s too expensive to do the science, run field trials, file for approvals, and so on – especially in cases where the potential market is relatively small and would be unlikely to generate the large profits that would make it worthwhile.
Last year, the Philippines became the first country in southeast Asia to relax the approval process for some gene-edited crops, adopting policies similar to those in the U.S. and Argentina (where the end product, rather than process, reigns supreme.) The country also approved Golden Rice for cultivation in 2021. Last autumn, it was harvested from just 17 fields, so it’s too early to tell whether the crop will be adopted more widely.
It may have taken Golden Rice more than twenty years to get approval and reach a small number of farmers, but the tide in biotechnology is shifting. Progress in gene-editing appears to be accelerating. Problems with DNA delivery and tediously slow plant experiments will not remain challenges forever.
CRISPR tools have opened a Pandora’s box of opportunity, wherein crops can be modified to have broad-spectrum resistance to pests, higher yields, and drought resistance. In many cases, these traits cannot be achieved with traditional plant breeding – the required genome edits are too precise, or too many in number.
Still, we need more funding for basic plant research, must sequence more genomes, and will have to invent easier ways to experiment on plants in weeks, rather than months. We’ll also need revamped approval processes so that farmers can access the plants they need in a changing climate.
What began in Mexico in the 1950s is continuing today. The world will always need food, and will always rely upon great innovators to produce enough of it. Borlaug’s Revolution was painstakingly slow: Laborers delicately transferred grains of pollen from one plant to another to carefully cross thousands of plants each year. The second Green Revolution will look entirely different; it will be borne on the back of genetic engineering.
Biotechnology has gifted us with a generational opportunity to help the world’s poorest: Let’s not squander it.