Where in the Cell Does Glycolysis Occur?

Glycolysis takes place in the cytosol, the liquid portion of a cell’s interior that surrounds its organelles. That one-line answer has appeared in biology textbooks for decades, and it is correct as far as it goes. But calling the cytosol glycolysis’s “location” is a bit like saying a restaurant operates “in a building.” Recent research has revealed that glycolytic enzymes do not simply float around in cytoplasmic soup waiting to bump into their substrates. They cluster on membranes, ride along transport vesicles, attach to the cytoskeleton, migrate into the nucleus, and even condense into stress-induced granules. The story of where glycolysis happens turns out to be far richer than a single word on an exam.

Why the Cytosol, and What That Actually Means

Glycolysis is a sequence of ten enzyme-catalyzed reactions that break glucose down into two molecules of pyruvate, producing a small amount of ATP along the way. None of these ten enzymes are embedded in a membrane the way the machinery of oxidative phosphorylation sits inside mitochondria. They are soluble proteins, which is why glycolysis was historically mapped to the cytosol. In most animal, plant, and fungal cells, this remains true in the broadest sense: if you lyse a cell and spin down the organelles, the glycolytic enzymes stay in the supernatant.

But “soluble” does not mean “randomly distributed.” The cytosol is a crowded environment, packed with proteins, RNA, metabolites, and cytoskeletal filaments at concentrations high enough that free diffusion is sluggish. In that context, glycolytic enzymes have evolved ways to organize themselves spatially, and that organization turns out to matter for how fast and how efficiently the pathway runs.

Glycolytic Metabolons and Enzyme Clustering

One of the more striking findings in recent cell biology is that successive glycolytic enzymes physically associate with each other in short-lived complexes sometimes called metabolons. Computational modeling of the crowded cytoplasm has shown that three consecutive glycolytic enzymes, GAPDH, phosphoglycerate kinase, and phosphoglycerate mutase, form transient multi-enzyme complexes that persist on the scale of microseconds. That may sound brief, but it is long enough for a substrate molecule produced by one enzyme to be handed directly to the next without diffusing away into the surrounding fluid.¹ This “substrate channeling” keeps intermediates from being diluted or hijacked by competing reactions, effectively creating a private pipeline within the open cytosol.

A similar principle operates at the plasma membrane of insulin-secreting pancreatic beta cells, where an entire glycolytic metabolon has been shown to sit right next to potassium channels. Glucokinase and phosphofructokinase, which consume ATP in the early “investment” phase of glycolysis, generate ADP that opens these channels. Meanwhile, pyruvate kinase at the end of the pathway consumes that ADP and raises the local ATP-to-ADP ratio, closing the channels. The whole arrangement functions as a glucose-sensing switch because the glycolytic enzymes are physically anchored near the channels they control.² If these enzymes were drifting randomly through the cytosol, the local ATP and ADP concentrations near the channel would be meaningless noise. Position is what makes the signal work.

Tethered to the Cytoskeleton

Observations going back more than forty years have shown that glycolytic enzymes bind directly to actin filaments, the structural cables that give a cell its shape and allow it to move.³ This was long treated as a biochemical curiosity, but the functional logic is becoming clearer. Remodeling the actin cytoskeleton is energy-expensive. By parking glycolytic enzymes right on the filaments they fuel, the cell ensures that ATP is produced exactly where it is consumed. The same principle shows up in neurons, where glycolysis is increasingly understood not as a generic housekeeping process but as a spatially targeted energy source for specific structures like growing axons and retracting processes.

Riding Vesicles Down the Axon

Neurons present a dramatic example of glycolysis happening in unexpected places. A motor neuron can stretch a meter from spinal cord to toe, and the axon along that length is far from any mitochondrion for much of its run. Research published in Cell demonstrated that GAPDH, one of the central glycolytic enzymes, physically attaches to transport vesicles through a mechanism involving the protein huntingtin (the same protein mutated in Huntington’s disease). These vesicle-bound enzymes are enzymatically active: purified motile vesicles can generate ATP on their own. Blocking GAPDH on vesicles stalled fast axonal transport, while inhibiting mitochondrial ATP production did not.⁴

Follow-up work confirmed that the picture extends beyond GAPDH alone. Most of the enzymes in glycolysis’s ATP-producing second half have been found localized on synaptic vesicle precursors, secretory vesicles, and endosomes within both axons and dendrites.⁵ In other words, neurons package a miniature glycolytic engine onto the cargo itself, making each vesicle a self-propelling unit. Glycolysis here is not “in the cytosol” in any meaningful sense; it is on a moving platform inside the cell.

The Mitochondrial Surface

Perhaps the most well-characterized non-cytosolic location for a glycolytic enzyme is the outer mitochondrial membrane. Hexokinase, the enzyme that kicks off glycolysis by phosphorylating glucose, exists in multiple isoforms, and isoforms I and II bind to a channel protein called VDAC on the mitochondrial surface.⁶ Sitting there gives hexokinase privileged access to ATP streaming out of the mitochondrion, which it immediately uses to phosphorylate glucose. The arrangement neatly links the very first step of glycolysis to the organelle that handles the pathway’s downstream products.

This binding is not just a metabolic convenience. When hexokinase 2 detaches from VDAC, it can trigger inflammatory signaling. Specifically, hexokinase 2 dissociation from the mitochondrial outer membrane promotes assembly of the NLRP3 inflammasome, a molecular alarm system involved in immune defense and chronic inflammation.⁷ So where hexokinase sits inside the cell has consequences that go well beyond metabolism.

Membrane-Coupled Glycolysis in Neurons

Beyond vesicles, glycolysis in neurons is also tightly linked to the plasma membrane through a different mechanism: the sodium-potassium pump. In hippocampal neurons, experiments showed that glycolytic activation is predominantly a response to increased activity of this pump, which burns ATP to restore ion gradients after a nerve impulse. When sodium floods in through ion exchangers during calcium clearance, the pump ramps up, and glycolysis ramps up with it.⁸ The glycolytic enzymes supplying ATP to the pump need to be close to the membrane to keep up with demand. This is another case where glycolysis is functionally pinned to a specific cellular address, not diffusing through the bulk cytoplasm.

Inside the Nucleus

Some glycolytic enzymes do not just stay in the cytoplasm. They enter the nucleus, and when they get there, they take on entirely new jobs. The most studied example is pyruvate kinase M2 (PKM2), the enzyme that catalyzes the last step of glycolysis. In its normal cytoplasmic form, PKM2 exists as a tetramer and works as a standard metabolic enzyme. But when it is phosphorylated by certain signaling kinases, it converts to a monomer, slips into the nucleus, and acts as a histone kinase, modifying the proteins that package DNA. In that role, PKM2 drives expression of genes that promote cell growth and the metabolic shift known as the Warburg effect.⁹

This nuclear translocation is not a laboratory oddity. In liver cancer cells exposed to high glucose, increased PKM2 movement into the nucleus has been linked to suppression of immune-attracting signals, which helps tumors evade immune surveillance.¹⁰ A separate study found that a protein called JMJD5 interacts with PKM2 in a way that blocks its normal enzymatic activity while promoting its nuclear entry and activation of hypoxia-response genes.¹¹ PKM2 is not the only glycolytic enzyme that moonlights in the nucleus; a broader review of the field documents multiple metabolic enzymes translocating to the nucleus during processes like cell differentiation and immune cell activation, where they help regulate gene expression through epigenetic mechanisms.¹²

Stress Granules Made of Glycolytic Enzymes

When oxygen runs low, cells face a metabolic emergency: the mitochondrial route for ATP production shuts down, and glycolysis becomes the primary energy source. Under these conditions, something remarkable happens. In yeast subjected to hypoxia, glycolytic enzymes coalesce into a single, large, non-membrane-bound granule in the cytoplasm, dubbed a “G body” (glycolytic body). The rate-limiting enzyme phosphofructokinase is concentrated there, along with other pathway members. Critically, this is not just a storage dump: cells that form G bodies maintain higher glycolytic flux, and the same structures appear in human liver cancer cells under low-oxygen conditions.¹³

G bodies form through liquid-liquid phase separation, a process where proteins and RNA demix from the surrounding cytoplasm into a concentrated droplet, similar to oil beading in water. RNA plays a direct role in nucleating these structures. The resulting granules behave like hydrogels: they fuse when they touch, and they resist certain chemical disruptions that dissolve other phase-separated bodies.¹⁴ This adds yet another address for glycolysis: under stress, the enzymes can self-assemble into a temporary, purpose-built compartment right in the middle of the cytoplasm.

Plants Run Glycolysis in Two Places

If the picture in animal cells is more complicated than expected, plant cells add another layer. In plants, glycolysis occurs mainly in the cytoplasm, just as in animals. But plants also maintain a second, parallel set of glycolytic enzymes inside their plastids, the organelles best known for photosynthesis. This plastid-targeted lower glycolytic pathway is particularly active in non-photosynthetic tissues like roots and seeds, where plastids serve metabolic rather than light-harvesting roles.¹⁵ Plants express distinct pyruvate kinase isoforms for each compartment: a cytosolic version and a plastidic version, with different evolutionary histories, different structures, and different regulatory properties.¹⁶

Diatoms and other stramenopiles push this further still. In some oomycetes (the group that includes the organism behind potato blight), the second half of glycolysis has been found not just in plastids but inside mitochondria, linked to serine biosynthesis. Phylogenetic analysis reveals that these mitochondrial glycolytic enzymes have entirely different ancestries from their cytoplasmic counterparts, suggesting they were acquired independently rather than simply being copies of the same gene sent to a different address.¹⁷

Parasites That Lock Glycolysis in a Special Organelle

Trypanosomes, the single-celled parasites responsible for sleeping sickness and Chagas disease, offer the most extreme departure from the textbook answer. In these organisms, the bulk of glycolysis does not occur in the cytosol at all. Instead, it takes place inside a dedicated, membrane-bound organelle called the glycosome. This organelle is related to the peroxisome and sequesters the first seven glycolytic enzymes behind its own lipid bilayer.¹⁸

The reason for this quarantine appears to be safety. The early steps of glycolysis consume ATP before later steps repay it, and modeling has shown that without compartmentalization, the pathway can enter a lethal runaway state where the ATP-consuming reactions outpace the ATP-producing ones. The glycosome’s membrane keeps intermediates at controlled concentrations, preventing this imbalance. It is a striking case where the location of glycolysis is not incidental but essential to the organism’s survival.

Bacteria and the Inner Membrane

Bacteria lack membrane-bound organelles in the traditional sense, so their glycolysis has always been described as happening “in the cytoplasm.” But even in E. coli, a connection between glycolysis and membranes has emerged. A phosphatase called PgpA, located in the bacterial inner membrane, has been identified as a redox-sensitive switch that regulates glycolytic flux. PgpA dephosphorylates key glycolytic intermediates, linking the cell’s metabolic rate to its redox state and to the membrane itself.¹⁹ Even organisms without organelles, it turns out, organize glycolysis relative to their membranes.

Rewiring During Development

The spatial arrangement of glycolytic enzymes is not fixed over a cell’s lifetime. During embryonic stem cell differentiation into heart muscle cells, the entire glycolytic network undergoes a physical overhaul. Hexokinase isoforms switch, and the remaining enzymes relocate to sit between the newly forming myofibrils, mapping onto the expanding mitochondrial network. This repositioning is thought to wire glycolysis into a relay system that shuttles high-energy phosphates from mitochondria to the contractile machinery that needs them.²⁰

RNA adds another regulatory layer to this process. The glycolytic enzyme enolase 1 binds specific cellular RNAs, and those RNAs inhibit its enzymatic activity. In mouse embryonic stem cells, manipulating enolase 1’s ability to bind RNA altered both glycolytic capacity and the cells’ ability to differentiate into endoderm, one of the primary tissue layers of the developing embryo.²¹ The implication is that the cell can tune glycolysis not just by changing enzyme levels but by controlling where enzymes are and what they are interacting with at a given moment in development. Location, interaction partners, and enzymatic activity are all entangled.