Inside a cell

Most cells in your body hold two copies of your genome, one from each parent, each about 3.1 billion base pairs long. That DNA is not left loose: it is wound around histone proteins into nucleosomes, which compact it and limit which parts the cell's machinery can reach.

This human nucleosome was solved by X-ray crystallography. Its 146 base pairs of DNA wrap about 1.65 times (a little under two turns) around a core of eight histones, two each of H2A, H2B, H3 and H4. The grip is a balance: tight enough to keep the DNA packed and protected, loose enough for the enzymes that read genes to get in.

To read a gene, RNA polymerase II separates the two DNA strands and reads one of them, the template strand, building an RNA that base-pairs with it. The RNA therefore has the same sequence as the other (coding) strand, with U in place of T. It is built one nucleotide at a time at its growing end.

Researchers pulled these polymerases out of human cell nuclei while they were transcribing the cell's own DNA, then imaged them by cryo-electron microscopy. In this model, 11 nucleotides of new RNA are resolved, and 10 of them are still paired with the template strand inside the enzyme.

A human gene is usually interrupted: stretches that are kept in the finished messenger RNA (exons), which between them carry the protein-coding sequence, alternate with stretches that are not (introns). The first RNA transcript contains both, so the introns have to be cut out. The spliceosome, a large machine made of proteins and small nuclear RNAs, removes each intron and joins the exons on either side.

This human spliceosome was caught just after it joined two exons. The cut-out intron is still held, as a loop called a lariat. The RNA being spliced, called MINX, is listed in the PDB entry as coming from human adenovirus 2.

Before it leaves the nucleus, a finished mRNA is packaged with proteins and recognised by TREX, the transcription-export complex. The TREX subunit ALYREF binds the exon junction complexes that splicing leaves on the mRNA, so TREX picks out messages that have been spliced.

The mRNA is thought to be handed on to the export factor NXF1, which takes it out through the nuclear pores into the cytoplasm. This structure is TREX purified from human cells: four copies of the THO complex, each with the RNA helicase UAP56 (also called DDX39B).

In the cytoplasm, ribosomes translate the mRNA into protein. The small subunit holds the mRNA and matches each three-letter codon with the anticodon of a transfer RNA (tRNA), which carries the matching amino acid. The large subunit joins the amino acids into a chain, a reaction carried out by the ribosome's RNA rather than its proteins.

This human 80S ribosome was frozen after a new tRNA had been accepted and before the ribosome moved on to the next codon. Two tRNAs sit side by side, each paired with its codon on the mRNA.

It is a test-tube complex, not one taken from a working cell. The message is a synthetic run of U (UUU codons, read as phenylalanine), and the tRNAs are modelled with the sequence of a bacterial tRNA for phenylalanine. The drug cycloheximide, bound in the large subunit, holds the ribosome in this state by blocking the move to the next codon.

As the chain grows, it leaves the ribosome through a tunnel in the large subunit, lined by ribosomal RNA and proteins. Here, 39 amino acids of a new chain can be traced down the tunnel, reaching about 8 nm from the tRNA that holds it.

The chain is a stretch of E-cadherin (CDH1), a human cell-adhesion protein. Researchers could capture it because a drug-like molecule, PF-06446846, binds in the tunnel beside the chain and stops the ribosome part way through making a few particular proteins, this one among them.

Most new chains fold into their working shape by themselves, but some need help. The chaperonin TRiC, also called CCT, is built from two stacked rings of eight different subunits. It closes a chamber around a chain so that it can fold, driven by ATP.

Tubulin, the building block of microtubules, is one of the proteins that need TRiC to fold. Researchers solved four structures of human β-tubulin folding inside TRiC. As folding goes on, more of the chain is ordered enough to model: 170, then 269, 351 and finally 439 of its 444 amino acids. This is the last of the four, and the model contains the TRiC ring that encloses the tubulin.

A folded protein can now do its job. Inside red blood cells, haemoglobin collects oxygen as blood passes through the lungs and hands it over to the tissues. Each molecule is built from two α and two β chains. Every chain holds a haem, a flat ring with an iron atom at its centre, and each iron can bind one oxygen molecule.

The four sites cooperate. When one binds oxygen the protein shifts shape and the others bind more easily, so haemoglobin loads up fully in the lungs and gives its oxygen up where oxygen is scarce. This model is human oxyhaemoglobin at 1.25 Å resolution, with oxygen on all four haems.