Housekeeping genes are genes that stay switched on in virtually every cell of an organism, regardless of what tissue the cell belongs to or what developmental stage it is in. They encode the proteins and RNAs a cell needs just to exist: the machinery for generating energy, building and recycling proteins, maintaining the cell’s internal skeleton, and copying DNA before division. The concept sounds simple, but defining exactly which genes qualify has turned out to be surprisingly difficult, and the list keeps shifting as researchers get better tools to measure gene activity across dozens of tissues at once.
What Makes a Gene “Housekeeping”
The name is a metaphor borrowed from the domestic sense of the word. Just as a house needs basic upkeep regardless of what happens inside it on any given day, cells need a baseline set of genes running continuously to keep their fundamental processes going. A housekeeping gene is expressed broadly across tissues and conditions, rather than being turned on only in, say, liver cells or only during embryonic development. Large-scale gene-expression studies have given researchers a more precise way to draw this line, replacing older lists that were based on small numbers of tissues with genome-wide surveys spanning dozens of cell types.
One recent database, the HRT Atlas, mined over 11,000 high-quality RNA sequencing samples from 52 human tissue types and identified roughly 2,176 human housekeeping genes, along with about 3,277 in mice. Those numbers give a sense of scale: the human genome contains around 20,000 protein-coding genes, so housekeeping genes make up a meaningful minority, roughly one in ten.
The Jobs They Do
Housekeeping genes cluster around the metabolic and structural pathways every cell depends on. An analysis across human tissues found enrichment in functions like the cell cycle, regulation of messenger RNA stability, protein folding, protein stabilization, and protein transport. In simpler terms, these genes handle the basics: breaking down sugar for energy, assembling and quality-checking new proteins, shuttling molecules to where they are needed, and keeping the cell’s structural scaffolding intact. A separate study focusing on the most consistently expressed genes across tissues found enrichment in mitochondrial function and energy balance, which makes intuitive sense since mitochondria are the power plants that almost every human cell relies on.
Classic Examples
A handful of housekeeping genes are so well known that they show up in practically every molecular biology lab. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) is a central enzyme in glycolysis, the pathway that breaks glucose down for energy. β-actin is a cytoskeletal protein that helps cells maintain their shape and move. Because these proteins are needed in every cell type, the genes encoding them were long considered rock-steady benchmarks of constant expression. Researchers routinely used them as internal controls when measuring how active other genes were: if GAPDH or β-actin levels looked the same across your samples, you could be more confident that any differences you saw in your gene of interest were real and not just artifacts of how much material you loaded into the experiment.
Ribosomal protein genes are another large family of housekeeping genes. Ribosomes are the molecular machines that read messenger RNA and stitch amino acids into proteins, and every cell needs them in large quantities. Genes encoding ribosomal subunits, such as RPL13a or RPS26, are expressed broadly across tissues, which led researchers to consider them useful as reference genes as well.
Why “Always On” Does Not Mean “Always Identical”
The assumption that housekeeping genes produce the same amount of product everywhere has taken serious hits over the past two decades. GAPDH and β-actin, the two most popular reference genes in the world, turn out to be multifunctional proteins involved in processes well beyond their textbook roles. GAPDH participates in DNA repair, membrane transport, and even the regulation of cell death, and β-actin is involved in gene regulation in the nucleus. Because they moonlight in so many pathways, their expression levels can shift depending on what a cell is doing.
This variability is not just theoretical. A study comparing equine stem cells derived from fat tissue versus bone marrow found that GAPDH expression differed significantly between the two sources, and β-actin and β2-microglobulin expression differed even more, with fold changes as high as roughly 6.8 and 7.8 respectively. In cancer research, a survey of 21 classical housekeeping genes across colorectal, pancreatic, esophageal, and gastric cancers found distinct differences in expression variability between normal and cancerous tissue for most of those genes. Even in a simple cell-culture model, housekeeping gene expression fluctuated over time in cells that were not being treated with anything, just sitting in a dish.
The practical consequence is that blindly picking a single housekeeping gene as your experimental yardstick can produce misleading results. One study on rat retinal development demonstrated this directly: normalizing the expression of a cell-cycle gene using different reference genes gave different answers, some of which were outright wrong. β-actin, for instance, was downregulated in more mature retinal stages, so using it as a “stable” reference in a developmental time course would have distorted the data.
How Researchers Now Choose Reference Genes
Because no single housekeeping gene is truly universal, the field has moved toward using statistical algorithms that rank candidate reference genes by how stable their expression is in the specific experimental context at hand. Three widely used tools are geNorm, NormFinder, and BestKeeper, each of which evaluates a panel of candidates and identifies the least variable ones for a given set of samples. A study of postmortem human brain tissue illustrates how much the “best” reference gene depends on the situation: geNorm ranked IPO8 and POLR2A as the most stable, NormFinder favored IPO8 and POP4, and BestKeeper highlighted ABL1 and ELF1. Combining all three algorithms pointed to IPO8, POLR2A, and PES1 as the best choices for brain tissue, while commonly used references like HPRT1 and GAPDH ended up among the least stable.
The broader lesson is that the old habit of defaulting to GAPDH or β-actin for every experiment is outdated. Current best practice calls for validating a small panel of candidates in every new experimental system and using the most stable combination, often two or three genes together rather than just one. A newer study went further, integrating expression data from both normal tissue atlases and drug-perturbation databases to find genes that stay steady not only across tissues but also under chemical stress, aiming to identify references that hold up in pharmacological experiments where traditional housekeeping genes might not.
Compact Genomes for Busy Genes
Housekeeping genes have a distinctive physical structure in the genome. Their introns (the non-coding stretches that interrupt the gene’s protein-coding sequence) are shorter than those of tissue-specific genes, and their untranslated regions and coding sequences are more compact as well. This pattern, observed in both human-mouse comparisons and broader mammalian analyses, is thought to reflect selection for efficiency: a gene that needs to be transcribed rapidly and often benefits from being short, since a shorter gene takes less time and energy to copy into messenger RNA.
Their promoters, the stretches of DNA just upstream of the gene that control when and how strongly it is expressed, also have a characteristic signature. Housekeeping genes tend to rely on CpG-rich promoters, while tissue-specific genes more often use different regulatory architectures. CpG-island promoters are associated with a more “open” state of chromatin, the protein-DNA complex that packages the genome, which fits the idea that housekeeping genes need to be accessible to the transcription machinery at all times.
Where They Sit in the Genome’s 3D Structure
The genome is not just a linear string of genes; it folds into three-dimensional structures inside the nucleus. One level of this architecture involves topologically associating domains (TADs), regions of DNA that interact with themselves more than with neighboring regions. Research on fruit fly embryos found that housekeeping genes are strongly enriched at the boundaries between TADs, rather than sitting inside them. TAD boundaries tend to be more accessible and more actively transcribed, which aligns with the need for housekeeping genes to be expressed constitutively. The finding held up whether researchers identified housekeeping genes by their expression patterns or by the presence of housekeeping-associated enhancer elements, suggesting it is a robust feature of genome organization rather than a fluke of one classification method.
Slower Evolution, Stronger Conservation
Because housekeeping genes perform functions that every cell depends on, mutations in them are more likely to be harmful and more likely to be weeded out by natural selection. A comparison of human and mouse genes found that housekeeping genes accumulate protein-altering mutations at roughly half the rate of tissue-specific genes. The rate at which mutations change an amino acid (the nonsynonymous substitution rate) averaged about 0.046 for housekeeping genes versus 0.083 for tissue-specific ones. Even mutations that do not change the protein sequence were slightly slower in housekeeping genes, though the difference was smaller.
This pattern is not unique to mammals. In plants, a comparison of rice and Arabidopsis housekeeping gene pairs found that their synonymous substitution rates were significantly lower than those of tissue-specific pairs, indicating stronger selective constraint even at positions where the protein itself is not affected. Stronger constraint on synonymous sites suggests that codon usage in housekeeping genes is itself under selection, possibly to ensure efficient and accurate translation by the ribosome.
One counterintuitive finding is that while housekeeping genes’ coding sequences are highly conserved, their upstream regulatory regions are actually less conserved than those of tissue-specific genes, especially far from the transcription start site. This may be because tissue-specific genes need elaborate, precisely tuned regulatory sequences to ensure they turn on only in the right place and time, while housekeeping genes use a simpler, more generic “always on” promoter that does not require long-distance regulatory elements to be conserved.
Housekeeping Genes Are Not the Same as Essential Genes
It is tempting to equate “always on” with “can’t live without it,” but housekeeping genes and essential genes are overlapping but distinct categories. Essential genes are those whose loss kills the organism (or, in cell-culture experiments, kills the cell). Many housekeeping genes are essential, but not all essential genes are housekeeping: some genes are indispensable in a specific tissue or at a specific developmental stage without being expressed everywhere. A direct comparison found that housekeeping genes had lower evolutionary rates, higher protein sequence identity across species, fewer single-nucleotide polymorphisms per unit length, and higher linkage disequilibrium than essential genes. In other words, housekeeping genes are under even tighter evolutionary constraint than essential genes, which makes sense if you consider that a housekeeping gene has to work in every tissue simultaneously, while an essential gene may need to work only in one context.
When Housekeeping Genes Break
Given how fundamental these genes are, you might expect mutations in them to be uniformly lethal. Many are, which is why they are so conserved. But some mutations reduce function without eliminating it entirely, and the resulting diseases can be surprisingly tissue-specific. The most striking example involves ribosomal protein genes. Diamond-Blackfan anemia is caused by mutations in genes like RPS19 that encode components of the ribosome, a machine every cell needs. Yet the primary clinical feature is a failure to produce red blood cells, with relatively little impact on most other tissues. Other ribosomal gene mutations cause Shwachman-Diamond syndrome, dyskeratosis congenita, cartilage-hair hypoplasia, and Treacher Collins syndrome, each with a different pattern of affected tissues.
These conditions, collectively called ribosomopathies, reveal something unexpected about housekeeping genes: even though they are active everywhere, some tissues are more sensitive to partial loss of function than others. Red blood cell precursors, for instance, divide extremely rapidly and need enormous quantities of ribosomes, so even a modest reduction in ribosome production hits them harder than it hits a slower-dividing cell type. The 5q- syndrome, a subtype of myelodysplastic syndrome caused by loss of one copy of the ribosomal protein gene RPS14, produces a red-blood-cell phenotype strikingly similar to Diamond-Blackfan anemia, reinforcing the idea that the blood-forming system is especially vulnerable to ribosomal gene dosage.
Housekeeping Genes in Bacteria
The concept extends well beyond animals and plants. In bacteria, housekeeping genes encode the core machinery for DNA replication, transcription, translation, and basic metabolism. Efforts to define a minimal bacterial genome, the smallest set of genes a free-living bacterium would need, have arrived at a figure of about 206 protein-coding genes that cover self-maintenance and reproduction when all necessary nutrients are provided and environmental stress is absent. Most of these are housekeeping genes by any reasonable definition. In microbiology, housekeeping genes like rpoB (encoding part of RNA polymerase) and gyrB (encoding part of DNA gyrase) also serve as molecular barcodes for identifying and classifying bacterial species, because their sequences are conserved enough to be alignable across distant relatives yet variable enough to distinguish closely related species.
GAPDH Beyond the Textbook
GAPDH deserves special mention because it is both the most famous housekeeping gene and perhaps the best illustration of why the concept is more complicated than it first appears. Beyond its glycolytic role, GAPDH has been linked to the regulation of gene expression, cytoskeletal dynamics, membrane fusion, and programmed cell death. More recently, a pan-cancer analysis flagged GAPDH as having an oncogenic role, meaning its expression changes in ways that may actively contribute to tumor biology rather than simply reflecting it. If the gene you are using as your “normal baseline” is itself behaving abnormally in cancer, normalizing your experimental data to it can mask or distort the very changes you are trying to detect. This is why the field’s shift away from single-gene normalization has been especially urgent in cancer research, where the stakes of getting the measurement wrong are high.
The story of GAPDH encapsulates a broader theme in modern genomics: biology resists tidy categories. Housekeeping genes are real, in the sense that a core set of genes genuinely is expressed across most tissues most of the time. But the boundaries of that set depend on how strictly you define “most tissues” and “most of the time,” and even the genes firmly inside the set can behave in tissue-specific and condition-specific ways that matter for both basic science and clinical applications. The databases and algorithms developed over the past decade give researchers much better tools for navigating this complexity than the old approach of picking GAPDH and hoping for the best.