The Strongest Bond in Nature: Nitrogenase and How Bacteria Split Nitrogen at Room Temperature

The Hook: An Ocean of Food That Almost Nobody Can Eat

Take a breath. About 78 percent of what just entered your lungs is nitrogen — molecular nitrogen, N₂. Roughly eight tons of it press down on every square meter of the Earth's surface. Nitrogen is also one of the most indispensable building blocks of life: every amino acid carries it, every DNA base, every enzyme, every signaling molecule in your nervous system. Without nitrogen there are no proteins, and without proteins there is nothing we could call life.

And yet: you cannot use the nitrogen you just inhaled. No animal can. No plant can. No fungus. The overwhelming majority of living things on this planet stand in an ocean of nutrient and starve in it.

The reason is a single chemical bond. The two nitrogen atoms in N₂ are joined by a triple bond whose dissociation energy is about 941 kJ/mol — one of the strongest bonds chemistry knows at all. N₂ is therefore not a nutrient but a locked safe. It is unreactive, nonpolar, and offers no handle for the ordinary tools of biochemistry.

Humanity forced that safe open with brute strength. The Haber-Bosch process, running industrially since 1913, drives nitrogen and hydrogen over an iron catalyst at 400 to 500 °C and 150 to 300 bar to make ammonia. It works magnificently — and it costs: ammonia synthesis consumes on the order of 1.8 percent of global final energy consumption and accounts for roughly 1.5 percent of anthropogenic CO₂ emissions (about 420 Mt per year). Some 170 million tons of ammonia are produced annually, about 80 percent of it destined for fertilizer.

And then there is a bacterium in the root nodule of a pea plant that performs the same reaction — at ambient temperature, at atmospheric pressure, in aqueous solution, inside a cell that does not even possess a nucleus. It uses a single enzyme to do it: nitrogenase.

This article is about how that enzyme cracks the strongest bond in nature — and why, after more than a century of research, we still do not fully understand how it does so. It is a story about a metal cluster that hid a carbon atom in its heart, about an enzyme that wastes energy for reasons that long seemed inexplicable, and about one of the great open prizes in chemistry: fertilizer without natural gas.

Part 1: The Problem — Why N₂ Is So Hard to Crack

Thermodynamics Lies

The first surprise: converting nitrogen into ammonia is not thermodynamically unfavorable at all.

N₂ + 3 H₂ → 2 NH₃ has a standard reaction enthalpy of about −92 kJ/mol; it is exothermic. The reaction wants to proceed. It simply does not.

The obstacle is kinetic, not thermodynamic. To loosen the triple bond, the molecule must climb an enormous activation barrier. Breaking the first partial bond is the most expensive step — unlike most multiple bonds, the reaction does not get easier with each stage but begins with its highest hurdle. The early intermediates of reduction, diazene (HN=NH) in particular, are energetically far less favorable than either starting material or product. The road from N₂ to NH₃ runs uphill before it runs down.

A catalyst must level that hill. That is exactly what both the iron in the Haber-Bosch reactor and the iron in nitrogenase do — in two entirely different ways.

Two Strategies, One Goal

The decisive conceptual difference is in the second-to-last row. Haber-Bosch tears N₂ apart: on the hot iron surface the molecule dissociates into two adsorbed nitrogen atoms, which are then hydrogenated individually. That is what the brutal temperature is for. Nitrogenase takes the opposite route: it binds N₂ whole and weakens the triple bond step by step, alternately adding electrons and protons. The bond is not blown apart; it is unravelled.

I am of the opinion that this is where the real lesson of the comparison lies: where technology overwhelms barriers with energy, biology circumvents them with mechanism. Both approaches are legitimate — but only one of them scales on a planet with a climate budget.

Part 2: The Enzyme — Two Proteins, Three Metal Clusters

Nitrogenase is not a single protein but a two-component system, best pictured as engine and gearbox.

The Fe Protein: The ATP-Driven Electron Delivery Service

The smaller component, the Fe protein (also dinitrogenase reductase, gene product nifH), is a homodimer with a single [4Fe–4S] cluster wedged between its two subunits. Its job is narrow and uncompromising: it delivers electrons — one at a time — to the larger component.

The price is remarkable. The Fe protein binds two ATP, hydrolyzes them, and transfers exactly one electron. ATP hydrolysis triggers a conformational change that shifts the cluster's reduction potential from about −0.29 V to −0.40 V. That is the core of the trick: ATP is not used here to pay for a reaction energetically, but to charge an electron thermodynamically — to lift it to a potential at which it is capable of working on N₂ at all. The Fe protein is a molecular high-voltage pump.

After transfer, the Fe protein must undock, exchange ADP for ATP, and be re-reduced. This docking cycle is the rate-limiting step of the entire reaction. Nitrogenase is therefore a remarkably slow enzyme — one reason why nitrogen-fixing bacteria must invest up to 20 percent of their entire protein budget in this one system.

The MoFe Protein: The Actual Reactor

The larger component, the MoFe protein (gene products nifD and nifK), is an α₂β₂ heterotetramer of roughly 230 kDa. It contains two kinds of metal cluster, each present twice:

• The P-cluster ([8Fe–7S]) at the interface between α and β subunits. It is the waystation: it receives electrons from the Fe protein and passes them on. Think of it as a buffer capacitor.
• The FeMo cofactor (FeMoco) deep inside the α subunit. Here, and only here, N₂ is reduced.

The electrons therefore travel a defined path: [4Fe–4S] of the Fe protein → P-cluster → FeMoco → substrate. Roughly 14 angstroms separate the stations — short enough for electron tunneling, far enough to prevent uncontrolled short circuits.

Part 3: FeMoco — The Most Unusual Metal Cluster in Biology

A Cage of Iron and Sulfur

The FeMo cofactor has the composition [MoFe₇S₉C]-homocitrate. Structurally it consists of two cubes (cubanes) — a [MoFe₃S₃] and an [Fe₄S₃] portion — joined by three bridging sulfur atoms. The molybdenum at one end carries an additional organic ligand, homocitrate. The whole assembly is anchored in the protein by a cysteine (to an iron) and a histidine (to the molybdenum).

Six of the seven iron atoms form a trigonal prism — and at its center sits something that was for decades the greatest riddle in nitrogenase research.

The Atom Nobody Could See

When Oliver Einsle and colleagues published the structure of the MoFe protein at 1.16 Å resolution in Science in 2002, an unexpected additional light-atom ligand appeared at the center of the cofactor — an atom that had simply been invisible in all earlier, lower-resolution structures. The crucial question: what was it? Nitrogen, oxygen, or carbon? X-ray diffraction distinguishes these three neighbors in the periodic table only with difficulty, because they carry almost the same number of electrons.

The question carried enormous mechanistic weight. Had it been a nitrogen atom, one would have had to consider a component directly involved in the reaction — perhaps even an already-fixed N atom.

The answer came nine years later. In 2011, Spatzal et al. combined atomic-resolution X-ray data with ESEEM spectroscopy (electron spin echo envelope modulation) on ¹³C-labeled samples and demonstrated in Science: the central atom is a carbon — more precisely a µ₆-bridging carbide, a bare C⁴⁻ ion coordinated simultaneously by six iron atoms.

This was remarkable in two ways. First, an interstitial carbide in an enzyme is without precedent — the closest relatives of this bonding situation are found in metallurgy, in iron carbides such as cementite, which is to say in steel. Second, the carbide turned out to be astonishingly immobile: it stays put throughout catalysis. It is not a reaction partner but a structural anchor, allowing the cluster to deform under strain without falling apart. The carbide is the backbone, not the hand.

Installing It Is a Feat in Itself

FeMoco biosynthesis is a multistep process with its own machinery (NifB, NifEN, NifH). The carbide is inserted by NifB from a methyl group of S-adenosylmethionine — a radical SAM enzyme pushes a carbon into the center of a growing iron-sulfur cluster. The resulting precursor, NifB-co, is handed to the scaffold protein NifEN, where an apical iron is exchanged for molybdenum and homocitrate is attached. Only then is the finished cofactor inserted into the MoFe protein.

Structural work from 2025 and 2026 (Nature Chemical Biology) shows that the precursor first binds at a surface docking site before coordinated domain motions carry it into an internal cavity. The cell therefore does not assemble this cofactor like a workpiece; it moves it along a kind of conveyor belt with airlocks — presumably because every intermediate is extremely oxygen-sensitive and must never lie around freely in the cell.

Part 4: The Mechanism — Eight States and a God with Two Faces

The Lowe-Thorneley Ladder

The framework in which nitrogenase catalysis is still described today dates from the 1980s: the kinetic scheme of David Lowe and Roger Thorneley. It treats the enzyme as a counter. With every cycle of Fe-protein docking, ATP hydrolysis, and electron transfer, the cofactor advances one rung — from E₀ (resting state) through E₁, E₂, E₃ to E₈. The index records how many electrons (and protons) have accumulated.

The balanced equation for Mo nitrogenase reads:

N₂ + 8 H⁺ + 8 e⁻ + 16 ATP → 2 NH₃ + H₂ + 16 ADP + 16 Pᵢ

Two details in it demand explanation — and both lead into the heart of the mechanism.

Why Eight Electrons for a Reaction That Needs Six?

Stoichiometrically, reducing N₂ to two NH₃ requires the transfer of six electrons. Nitrogenase consumes eight and obligatorily produces one molecule of H₂. This obligatory hydrogen formation was long regarded as an irritating design flaw: two reducing equivalents and four ATP apparently thrown out the window with every single nitrogen molecule fixed.

It is not a flaw. It is the ignition key.

E₄(H₄): The Janus State

The key state on the ladder is E₄ — aptly named the "Janus state" by Brian Hoffman, Lance Seefeldt, and Dennis Dean, after the Roman god of transitions with two faces. One face looks back at the four electron and proton deliveries that built it; the other looks forward to the actual reduction of nitrogen. E₄ is the tipping point of catalysis.

In E₄ the FeMo cofactor carries two bridging iron hydrides ([Fe–H–Fe]) plus two protons on sulfur atoms — hence the notation E₄(H₄). And now the decisive maneuver occurs: the two hydrides are reductively eliminated and leave the cofactor as H₂.

The point of this reductive elimination is a piece of bookkeeping finesse. Formally the cofactor gives up two hydrogen atoms — but of the four electrons that resided in the two Fe–H bonds, most remain in the cluster. The result is a super-reduced state that could not be reached by the direct route of simply adding more electrons, because those extra electrons would long since have drained away as H₂. The hydrides are an intermediate store of reducing power, and H₂ release is the moment that store is discharged in a single stroke.

Only this super-reduced cofactor holds enough electron density to bind N₂ and weaken its triple bond through π backbonding. The process is reversibly coupled: N₂ binding and H₂ release are an exchange. That precisely explains an old and long-puzzling observation — H₂ is a specific inhibitor of N₂ reduction, and a competitive one at that. Raise the H₂ pressure and you push the equilibrium back, preventing nitrogen from binding. The observation had been known for decades; the mechanism behind it became comprehensible only through the characterization of E₄(H₄) in the 2010s.

The "waste," then, is the price of a thermodynamic trick: nature buys itself a reduction state it could not otherwise reach, and pays for it with one molecule of hydrogen.

And After That?

N₂ binding is followed by further protonations through states E₅ to E₈. The precise route — whether distal (one N atom is hydrogenated fully to NH₃ first, then the second) or alternating (both atoms are protonated in turn) — is not conclusively settled; experimental and computational evidence leans toward the alternating pathway. Equally contested is whether one of the bridging sulfur atoms (S2B) temporarily vacates its position during catalysis to open a binding site. Crystal structures with bound CO and selenide point in that direction, but the question is regarded as open.

After more than sixty years of intense research, the mechanism of nitrogenase is understood in outline and unresolved in detail. That is unusual for so central an enzyme — and it is because the cluster is spectroscopically extraordinary difficult: eight transition metals with coupled spins, whose electronic structure sits at the edge of what quantum-chemical calculation can manage. FeMoco has by now become one of the standard benchmark problems for future quantum computers — not as a marketing example, but because classical methods genuinely founder here on multireference effects.

Part 5: The Variants — Molybdenum, Vanadium, Iron

Mo nitrogenase is the most common but not the only one. Some bacteria and archaea possess alternative nitrogenases in which molybdenum is replaced by vanadium (V nitrogenase) or by another iron (Fe-only nitrogenase). Characteristically, these organisms virtually always possess the Mo version as well and switch over only under molybdenum scarcity.

For a long time the alternatives were dismissed as stopgaps — backup enzymes for molybdenum-poor soils. The more recent literature (for instance the 2020 Annual Review of Microbiology) argues more carefully: V nitrogenase probably plays a role of its own in the global nitrogen cycle, especially in cold habitats where its temperature advantage takes effect. And the Fe-only variant, with its CO₂- and CO-reducing capability, produces short-chain hydrocarbons that may help shape microbial communities.

That is a methodologically instructive correction: what looks like an inferior duplicate may serve a niche of its own. Anyone who looks only at the headline metric (here: N₂ turnover per ATP) will miss the side function that makes the system valuable in its actual context.

Part 6: The Oxygen Problem and the Root-Nodule Compromise

Nitrogenase has a lethal enemy: oxygen. O₂ destroys the Fe protein irreversibly within seconds, and the FeMo cofactor as well. This is no accident: a cluster electron-rich enough to attack N₂ is necessarily also electron-rich enough to be oxidized by O₂. The sensitivity is the flip side of the function.

From this follows a fundamental dilemma that evolution has solved along several routes:

• Spatial separation: cyanobacteria such as Anabaena differentiate specialized cells, heterocysts, which dismantle their oxygen-producing photosystem II and seal themselves against O₂ diffusion with a thick glycolipid envelope.
• Temporal separation: other cyanobacteria fix nitrogen at night, when photosynthesis is off.
• Respiratory protection: Azotobacter burns oxygen at an extremely high respiratory rate before it can reach the nitrogenase — a kind of biochemical bouncer.
• The root-nodule compromise: rhizobia in legume roots live in nodules with a diffusion barrier and — this is the elegant part — with leghemoglobin. This plant heme protein binds oxygen with very high affinity and holds the free O₂ concentration at nanomolar levels, while still delivering enough oxygen to the bacterial respiratory chain to produce the ATP that fixation demands. The nodule is simultaneously oxygen-poor and oxygen-supplied. If you have ever wondered why a sliced root nodule is pink: that is leghemoglobin.

It is worth comparing this with the hemoglobin described in Four That Act in Concert: Allostery, Hemoglobin, and the Logic of Cooperative Binding: the same protein family, but tuned in opposite directions — hemoglobin optimized for release in tissue, leghemoglobin for holding on at extremely low concentration.

Also remarkable is the case of Sierra Mixe, a traditional maize landrace from Mexico. It forms a sugar-rich mucilage on aerial roots in which diazotrophic bacteria live. The mucilage keeps the oxygen content below 5 percent at a depth of 8 mm — low enough for nitrogenase. Studies put the share of the plant's nitrogen requirement met this way at 29 to 82 percent. A cereal that organizes a substantial part of its own nitrogen — without any genetic modification at all.

Part 7: The Great Prize — Nitrogen Fixation for Cereals

Legumes (peas, beans, soy, clover) fix nitrogen through their symbionts. Cereals — wheat, maize, rice, that is to say the foundation of the world's food supply — do not. The consequence is synthetic fertilizer, and with it dependence on natural gas as well as the eutrophication of waterways.

Research pursues three main strategies:

1. Induce nodules in cereals. The symbiosis signaling pathways (Nod factors, the common symbiosis pathway shared with mycorrhiza) are partly present in cereals. The approach is conceptually elegant, but organ formation itself is developmentally complex.
2. Transfer nif genes directly into the plant. This is the most radical route — and the hardest. The minimal nif gene cluster comprises about 16 genes with finely tuned stoichiometry. All products are oxygen-sensitive. Candidates for hypoxic housing are mitochondria and chloroplasts, which could also supply ATP and reducing equivalents. In 2022, functional expression of the Fe protein in transgenic rice was demonstrated — a milestone, but one component among many that are needed.
3. Optimize associative diazotrophs. The most pragmatic route: bacteria that live on cereal roots are reprogrammed to fix nitrogen and release it even when ammonium is present. Normally ammonium shuts fixation down immediately (economically sensible for the bacterium, useless for the farmer). Transferring refactored nif clusters from Pseudomonas stutzeri into Pseudomonas protegens Pf-5 yielded strains with high nitrogenase activity that remained insensitive to external ammonia and to oxygen — the donor organism's regulatory logic did not take hold in the recipient.

In parallel, inorganic chemistry is working on the replica. In 2025, PNAS reported the synthesis of a FeMoco mimic bearing a trigonal-prismatic [Fe₆C] core with a µ₆-carbide — the first laboratory reproduction of the natural cofactor's characteristic geometry; an analogous [Fe₆N] model was also prepared. These models do not yet catalyze industrially useful ammonia synthesis, but they finally permit systematic comparison: what exactly does the carbide contribute, and what does the molybdenum?

I am of the opinion that a familiar maturation curve is taking shape here: first describe (structure), then understand (mechanism), then rebuild (model clusters), then apply. We stand between step three and step four — and historically that gap has rarely been short.

The Central Takeaway

Nitrogenase solves at room temperature a problem that technology can solve only at 500 °C and 300 bar. The price is not energy in the thermal sense but energy in the form of information: a precisely engineered metal cluster, an exactly timed electron counter, and a counterintuitive activation trick in which the enzyme appears to throw resources away (H₂) in order to reach a state otherwise beyond its grasp.

Three transferable patterns sit inside that.

First: apparent waste is often the mechanism. For decades the obligatory H₂ formation was regarded as an inefficiency of nitrogenase. It is the activation step. When you find a conspicuously "unnecessary" expenditure in a system, first check whether that expenditure serves a function you cannot yet see. That applies to retry logic, to redundancy, to buffers, and to process steps that look superfluous on a value-stream map.

Second: fragility is usually the flip side of capability. Nitrogenase is so oxygen-sensitive precisely because it must be so electron-rich. You cannot optimize that sensitivity away without losing the function — you must encapsulate it. The root nodule is not a repair; it is an architectural decision: a protected context around a necessarily fragile core component. It is the same decision you make when you isolate a sensitive service behind a trust boundary.

Third: speed is not always the target metric. Nitrogenase is an extremely slow enzyme. Evolution did not fix that; it compensated with quantity. That is a legitimate strategy — throughput through parallelism rather than through latency optimization.

A concrete next step: take a system you are responsible for and look deliberately for the one component that is necessarily fragile because it does the hardest thing. Then ask not "How do I make it more robust?" but "What nodule do I build around it?" — which isolation layer, which controlled environment, which supply discipline. In most architectures far too much energy goes into hardening fragile cores and far too little into giving them the right context.

A question to sit with: nitrogenase reaches its decisive state only by giving up something it has just laboriously built. Where in your own work do you cling to hard-won intermediate states — code, concepts, drafts — when the next leap in quality depends precisely on letting them go?

Cross-References in the Vault

• The Molecular Turbine: ATP Synthase and the Engine of Life – The 16 ATP that nitrogenase consumes per N₂ come from exactly this machine; engine and heavy consumer in direct comparison.
• The Machine That Splits Water: Photosystem II and the Origin of Oxygen – Photosystem II produces the oxygen that destroys nitrogenase, and does so with an equally extraordinary metal cluster; the two enzymes are the great metal-cluster masterpieces of bioenergetics.
• Four That Act in Concert: Allostery, Hemoglobin, and the Logic of Cooperative Binding – Leghemoglobin in the root nodule is a relative of hemoglobin with inverted tuning: maximum retention rather than controlled release.
• When Bacteria Take a Vote: Quorum Sensing and the Secret Language of Microbes – The symbiosis between rhizobia and legumes rests on chemical communication; here the grammar of that language is explained.
• The Enemy That Became a Power Plant: Endosymbiosis and the Bacterial Origin of Complex Life – The root nodule is a young, reversible symbiosis; that article covers the ancient, irreversible one.
• The Programmable Scissors: CRISPR and the Rewriting of Life – The tools with which nif clusters are today transferred into recipient organisms and refactored.
• The Machine Made of RNA: The Ribosome, the Ribozyme, and the Translation of the Genetic Code – The nitrogen that nitrogenase fixes ends up as an amino acid inside exactly this machine.

Sources and Further Reading

• Spatzal, T. et al. – Evidence for Interstitial Carbon in Nitrogenase FeMo Cofactor, Science 334 (2011), 940 – identification of the central carbide: https://www.science.org/doi/10.1126/science.1214025
• Hoffman, B. M.; Lukoyanov, D.; Yang, Z.-Y.; Dean, D. R.; Seefeldt, L. C. – Mechanism of Nitrogen Fixation by Nitrogenase: The Next Stage, Chemical Reviews 114 (2014), 4041 – the standard review of Lowe-Thorneley, E₄, and reductive elimination: https://pubs.acs.org/doi/10.1021/cr400641x
• Lukoyanov, D. et al. – Reductive Elimination of H₂ Activates Nitrogenase to Reduce the N≡N Triple Bond: Characterization of the E₄(4H) Janus Intermediate in Wild-Type Enzyme: https://www.osti.gov/pages/biblio/1466796-reductive-elimination-h2-activates-nitrogenase-reduce-nn-triple-bond-characterization-e4-janus-intermediate-wild-type-enzyme
• Seefeldt, L. C. et al. – The Critical E₄ State of Nitrogenase Catalysis, Biochemistry 57 (2018): https://pubs.acs.org/doi/10.1021/acs.biochem.8b00509
• Harris, D. F. et al. – Electron transfer in nitrogenase (review of the Fe protein, ATP coupling, and the electron path), PMC: https://pmc.ncbi.nlm.nih.gov/articles/PMC7466952/
• Rutledge, H. L.; Tezcan, F. A. et al. – Synthetic models of the nitrogenase FeMo cofactor, PNAS (2025) – the first [Fe₆C] mimic with a µ₆-carbide: https://www.pnas.org/doi/10.1073/pnas.2419655122
• Dynamics driving the precursor in NifEN scaffold during nitrogenase FeMo-cofactor assembly, Nature Chemical Biology (2025) – cofactor maturation: https://www.nature.com/articles/s41589-025-02070-4
• Addo, M. A.; Dos Santos, P. C. – Iron-Only and Vanadium Nitrogenases: Fail-Safe Enzymes or Something More?, Annual Review of Microbiology 74 (2020): https://www.annualreviews.org/content/journals/10.1146/annurev-micro-022620-014338
• Erisman, J. W.; Sutton, M. A.; Galloway, J.; Klimont, Z.; Winiwarter, W. – How a century of ammonia synthesis changed the world, Nature Geoscience 1 (2008), 636 – Haber-Bosch and world food supply: https://www.nature.com/articles/ngeo325
• Pankievicz, V. C. S. et al. – Biological nitrogen fixation in cereal crops: Progress, strategies, and perspectives, Plant Communications (2022) – transfer strategies, Sierra Mixe maize: https://www.cell.com/plant-communications/fulltext/S2590-3462(22)00346-7
• Engineering Nitrogenases for Synthetic Nitrogen Fixation: From Pathway Engineering to Directed Evolution, BioDesign Research: https://spj.science.org/doi/10.34133/bdr.0005
• Functional expression of the nitrogenase Fe protein in transgenic rice, PMC (2022): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9534833/

Created as part of the daily learning workflow. Field of interest: Biochemistry. Estimated reading time: ~30 minutes.

← All articles