From the cells to the DNA

How to read it

• Scroll to descend. Each level is a new scale, and the view zooms from one into the next.
• The scale bar, bottom right, shows the true width of what is on screen.
• Labels name structures as they appear; notes flag anything modelled or schematic.
• Numbers link to the peer-reviewed sources listed at the end.

Scroll to begin

Every one of your cells carries the same genome

An adult body holds roughly 30 trillion human cells, about 90% of them in the blood-forming lineage, mostly red blood cells. Almost all of them carry the same DNA, about 3.1 billion base pairs per set of chromosomes, yet a neuron and a skin cell look and behave nothing alike. The difference is which genes each cell reads.

Tissue at cellular scale

Cells 10–20 µm across pack together beside a capillary. Inside each, the darker body is the nucleus. The discs flowing through the vessel are red blood cells, shaped from measured profiles: biconcave, about 7.8 µm wide.

Model: cell shapes and placement are illustrative; sizes are realistic.

The scale bar at the bottom right shows the true width of what is on screen at every moment. Before we enter a single cell: how different can cells be?

Same genome, different cells

The Tabula Sapiens atlas profiled nearly 500,000 single cells from 24 organs and identified more than 400 cell types. Shapes and sizes differ by orders of magnitude, so they are shown here at true relative size.

Six specialists

• Neuron. About 86 billion in the adult brain. They stop dividing, and some carry small DNA copy-number changes of their own.
• Epithelial cells. Polarised sheets that line surfaces; microvilli enlarge the apical face.
• Skeletal muscle fibre. A single cell with many nuclei, formed by the fusion of precursor cells.
• Red blood cells. Mature cells have expelled their nucleus, so they carry no nuclear genome at all.
• Lymphocytes. B and T cells cut and rejoin receptor gene segments (V(D)J recombination), so their DNA differs from every other cell.
• Fibroblasts. Build the extracellular matrix, including collagen fibrils.

What sets them apart is mostly not the DNA sequence but gene expression: which genes are switched on, and how strongly, in each tissue.

Inside one cell

A generic human cell, about 30 µm across. A phospholipid bilayer encloses a crowded cytoplasm where each organelle runs a specialised job.

• Rough ER and ribosomes make secreted and membrane proteins.
• Golgi apparatus modifies, sorts and ships them.
• Mitochondria make most of the cell's ATP and carry their own small circular genome: 16,569 base pairs, 37 genes.
• Lysosomes digest; peroxisomes break down fatty acids and peroxide; the centrosome organises microtubules.

Organelles approximately to scale; membranes and ribosomes thickened so they stay visible.

At the centre, the nucleus, about 10 µm across, holds the genome. We approach its envelope.

The nucleus

Two membranes, together called the nuclear envelope, separate the genome from the cytoplasm. Thousands of nuclear pore complexes perforate it.

Through a nuclear pore

Each pore is a large, eightfold-symmetric machine built from multiple copies of about 30 different proteins. Small molecules diffuse through; RNA is exported and proteins imported by dedicated carriers. A basket of filaments hangs on the nuclear side.

Pore shown schematically: correct symmetry and rings, simplified subunits.

Inside, each chromosome occupies its own territory. Gene-poor chromatin tends to line the nuclear lamina at the periphery, and gene-rich chromosomes sit nearer the centre. Chromosome 11, in gold, carries the β-globin genes we will follow down to single base pairs.

Model: territory sizes follow chromosome length; positions are simulated.

46 chromosomes

A human body cell holds 23 pairs: 22 pairs of autosomes and two sex chromosomes (XX in this cell). Together they carry about 6.2 billion base pairs, roughly 2 metres of DNA if laid end to end.

Chromosomes condense into these compact shapes only when a cell divides; in the nucleus we just left, they were loose territories. After DNA replication, each chromosome is two identical sister chromatids joined at the centromere. The two chromosomes of a pair are homologues, one from each parent.

The short arm is p, the long arm q, and stained bands give every region an address. Telomeres, long runs of TTAGGG repeats, cap the ends; centromeres are built from large arrays of repeated DNA. Our gene, HBB, lies in band 11p15.4.

G-band pattern from the NCBI GRCh38 550-band ideogram; chromosome shapes modelled.

The building block: a nucleotide

Each unit of DNA has three parts: a phosphate, a deoxyribose sugar and one of four bases, adenine (A), thymine (T), guanine (G) or cytosine (C). A and G are purines, with two rings; T and C are pyrimidines, with one.

A strand with a direction

Phosphodiester bonds join the 3′ carbon of one sugar to the 5′ carbon of the next, so every strand runs 5′ → 3′. The sequence here is the real start of the HBB coding sequence: ATG GTG CAT CTG.

Pairing

A second strand pairs in the opposite direction (antiparallel). A always pairs with T through two hydrogen bonds, and G with C through three.

Then the ladder twists into a right-handed helix, opening a wide major groove and a narrow minor groove.

Idealised heavy-atom model without hydrogens: ring topology and pairing are correct; geometry is regularised, not crystal coordinates.

The double helix

B-DNA, the form that dominates in cells, is about 2 nm wide. Each base pair rises 0.34 nm along the axis, and in solution the helix completes a turn every 10.5 base pairs, about 3.6 nm.

The grooves matter. The major groove exposes more of the chemically distinctive edges of each base pair, so most proteins that read a specific sequence make their contacts there.

Other forms

DNA is flexible. Dehydrated DNA adopts the wider, right-handed A-form, about 11 base pairs per turn. Some sequences, such as alternating G and C, can flip into left-handed Z-DNA, 12 base pairs per turn, with a zig-zag backbone.

B-form: space-filling idealised model. A and Z: simplified backbones built from Dickerson et al. (1982) parameters.

Spooled on histones

About 147 base pairs of DNA wrap around a core of eight histone proteins (two each of H2A, H2B, H3 and H4) in about 1.65 left-handed turns. This is the nucleosome, the repeating unit of chromatin.

Flexible histone tails reach out from the core. Enzymes add and remove chemical marks on them, such as acetyl and methyl groups, which help decide whether nearby DNA is open to be read or packed away.

Linker DNA joins one nucleosome to the next, giving a "beads on a string" fibre about 10 nm wide. Wrapping hides DNA; where nucleosomes are missing, regulatory proteins can reach it.

Wrapping geometry after PDB 1KX5; histone positions within the core are schematic.

How chromatin folds

Textbooks long showed nucleosomes coiling into a regular 30-nm fibre. Such fibres form from purified chromatin in the test tube, but imaging of intact human nuclei (ChromEMT) found mainly irregular, flexible chains 5–24 nm wide, packed at varying densities.

The 30-nm fibre is shown as a structural model, not as the universal form in cells.

Loops and domains

Contact maps from Hi-C show chromatin folded into loops, often anchored at pairs of CTCF sites facing each other and held by cohesin, and into topologically associating domains (TADs): neighbourhoods that contact mostly themselves. A leading model is loop extrusion: cohesin reels DNA into a growing loop until it meets CTCF.

Compartments and territories

Domains sort into two compartments: A, open euchromatin rich in active genes, and B, compact heterochromatin that is mostly silent and often near the nuclear periphery. Together they build each chromosome territory.

Structures modelled; the contact map shows what Hi-C data look like, not measured data.

Reading the sequence

Back on chromosome 11, in band 11p15.4, lies the β-globin cluster: genes used at different stages of life, from embryo (HBE1) through fetus (HBG2, HBG1) to adult (HBD, HBB), plus a pseudogene. Coordinates are from the GRCh38 reference.

HBB spans about 1.6 kb: three exons separated by two introns. Introns are spliced out of the RNA; the untranslated regions (UTRs) at each end stay in the mRNA but are not translated. Protein-coding exons are a small fraction of the genome: GENCODE annotates about 20,000 protein-coding genes.

One letter

Read three bases at a time, ATG GTG CAT CTG ACT CCT GAG spells Met-Val-His-Leu-Thr-Pro-Glu. A single change, GAG to GTG in codon 6 (rs334), swaps glutamate for valine and causes sickle-cell haemoglobin. Each person differs from the reference at 4–5 million sites; most of these differences are harmless.

Switching genes on

Whether HBB is read depends on regulatory DNA. RNA polymerase II assembles at the promoter, just upstream of the gene. Enhancers boost transcription from tens of kilobases away; silencers lower it; insulators mark boundaries. Transcription factors bind short motifs in these elements, mostly where chromatin is open.

A loop across the locus

In red-cell precursors, the locus control region, a cluster of enhancers upstream of the genes, physically loops to the active globin promoter. Experimentally forcing that loop can switch a silent globin gene on. CTCF-bound sites at both ends of the locus come together in their own loop, confining these contacts.

A linear sequence, folded in three dimensions, becomes gene expression: the same genome, read differently in every cell type.

Schematic: element order follows the real locus; distances and protein sizes are not to scale. The silencer is a generic example.