What are proteins? What they are made of and what they do

Proteins are large molecules built as chains of smaller units called amino acids, of which living things use 20 main kinds. Each chain folds into a precise three-dimensional shape, and that shape decides what the protein does: speed up a chemical reaction, carry oxygen, hold a cell together, send a signal or recognize a germ. They do most of the work in every cell.

In short

• A protein is a chain of amino acids whose order is set by a gene; the order decides how the chain folds.
• The folded shape is the function: change one amino acid in the wrong place, and the protein can fail.
• Working out a protein's shape from its sequence was a 50-year problem until the AI system AlphaFold2 largely solved it in 2020.

What are proteins made of?

A protein is a molecule made of amino acids joined end to end. Every amino acid has the same backbone, a nitrogen-containing amino group at one end and an acid group at the other, plus a side chain that differs from one kind to the next. Some side chains are oily and avoid water, some carry an electric charge, some are small, some are bulky. The U.S. National Library of Medicine's MedlinePlus puts the count at 20 different types, combined into chains of hundreds or thousands. (A couple of rare extra amino acids appear in some proteins of some organisms.)

The order of amino acids in a protein is written in a gene. How the cell reads a gene, through an RNA copy and three-letter codons, is set out in what is DNA?. What matters here is the result: a chain with a fixed sequence, such as insulin's 51 amino acids or the 146 of one hemoglobin chain.

Biologists describe a protein's shape in four levels.

1. Primary structure: the sequence of amino acids, like letters in a sentence.
2. Secondary structure: short stretches that coil into spirals (alpha helices) or line up into flat, pleated sheets (beta sheets), held by weak bonds along the backbone.
3. Tertiary structure: the whole chain folded into a compact shape. Oily side chains bury themselves in the middle, away from water, and charged ones face outward.
4. Quaternary structure: two or more folded chains fitted together. Hemoglobin is four chains, two of one kind and two of another.

What do proteins do?

Almost everything a cell does is done by a protein. MedlinePlus sorts their jobs into five groups.

Job	What the protein does	Example
Enzyme	Speeds up a chemical reaction, often millions of times	Lactase, which breaks down milk sugar
Structure	Gives cells and tissues shape and strength, and lets muscles move	Actin; collagen in skin and bone
Transport and storage	Binds and carries atoms and small molecules	Hemoglobin carries oxygen; ferritin stores iron
Messenger	Carries signals between cells and organs	Insulin; growth hormone
Antibody	Binds a particular germ or foreign molecule so the immune system can destroy it	Immunoglobulin G

The largest group is the enzymes, which carry out almost all of the thousands of reactions in a cell. Others act as pumps and gates in the cell membrane, motors that walk along fibers, and switches that turn genes on and off. Some proteins do two of these jobs at once.

How does a sequence decide a shape?

The proof came from the American biochemist Christian Anfinsen in the 1950s and 1960s. He took an enzyme called ribonuclease, unfolded it with chemicals until it stopped working, then removed the chemicals. The chain folded back into its working shape by itself. Nothing but the sequence was needed. His Nobel Prize in Chemistry in 1972 was awarded, in the words of the Nobel committee, for work on "the connection between the amino acid sequence and the biologically active conformation."

How the chain finds that one shape is a puzzle the American biologist Cyrus Levinthal posed in 1969. Here is a version of his arithmetic.

1. Take a small protein of 100 amino acids, and suppose each link in the chain can sit in just 3 positions.
2. The number of possible shapes is 3 multiplied by itself 100 times, about 5 × 10⁴⁷.
3. Suppose the chain could try 10 trillion (10¹³) shapes a second. Checking them all would take about 5 × 10³⁴ seconds.
4. The universe is about 13.8 billion years old, about 4.4 × 10¹⁷ seconds. The search would take some 100 million billion times longer than that.

Yet real proteins fold in anything from millionths of a second to a few seconds. The chain doesn't search at random: each small fold that lowers its energy is kept, so it slides downhill toward the stable shape, like water finding the bottom of a valley.

One change, one disease

Hemoglobin shows how much rides on a single amino acid. Its beta chain has 146 amino acids. In sickle hemoglobin, the version behind the most common forms of sickle cell disease, the sixth is valine instead of glutamic acid. Glutamic acid carries a charge and sits happily in water; valine is oily. When hemoglobin lets go of its oxygen, that oily patch sticks to a neighboring hemoglobin molecule, and the molecules line up into long, stiff fibers. The fibers bend the red blood cell out of shape. Instead of flexible discs, the cells become crescent or "sickle" shaped, the National Heart, Lung, and Blood Institute explains, and they can block small blood vessels. One amino acid out of 146 changed the shape, and the shape changed the disease.

What did AlphaFold change?

For half a century, finding a protein's shape meant months or years of lab work, by methods such as X-ray crystallography. Predicting it from the sequence alone mostly failed. In 2020, Demis Hassabis and John Jumper of Google DeepMind presented AlphaFold2, an AI model that predicts protein structures with accuracy close to experiment in many cases. The 2024 Nobel Prize in Chemistry went half to the two of them "for protein structure prediction" and half to David Baker, who designs new proteins by computer. The Nobel committee says AlphaFold2 has been used to predict the structure of virtually all the 200 million proteins researchers have identified.

A prediction is not the whole story. AlphaFold2 gives a likely folded shape, but proteins move, many have floppy regions with no fixed shape, and their behavior depends on what they bind. Experiments are still needed to confirm how a protein actually works. The models behind it are related to the neural networks used elsewhere in AI.

Do you need protein in your diet?

Yes, because the body breaks the proteins in food back into amino acids and uses them to build its own. Of the 20 kinds, 9 can't be made by the human body and must come from food; these are called essential amino acids. Meat, fish, eggs, dairy, beans, lentils, soy and grains all supply protein. The Harvard T.H. Chan School of Public Health notes that the U.S. National Academy of Medicine recommends adults get a minimum of 0.8 grams of protein per kilogram of body weight a day, about 50 grams for a person weighing 140 pounds. Needs differ with age, pregnancy, illness and activity; this note explains the biology and isn't dietary advice.

A common misconception is that protein is mainly a muscle food. Every cell in the body is built and run by proteins, and the body replaces many of them continually, which is part of why it needs a steady supply.

Check yourself

• Without looking back, explain what decides a protein's shape and why the shape matters.
• Walk through what happens inside a red blood cell in sickle cell disease, starting from one amino acid.
• Why would a protein never fold if it had to try every possible shape?

Sources

• What are proteins and what do they do? — MedlinePlus, U.S. National Library of Medicine
• The Nobel Prize in Chemistry 2024, press release — Royal Swedish Academy of Sciences, 2024
• Christian B. Anfinsen, facts — Nobel Prize Outreach
• Sickle Cell Disease — National Heart, Lung, and Blood Institute
• Protein — The Nutrition Source, Harvard T.H. Chan School of Public Health