Segmentation Clock

The big idea: turning a rhythm into a ruler

Your spine is a stack of nearly identical, regularly spaced units — about 33 vertebrae, each with its own pair of ribs (in the thoracic region), its own pair of spinal nerves, and its own block of muscle. That repeated, periodic pattern is not carved from a pre-existing template. It is counted out in real time by a clock running inside the embryo. The segmentation clock is that timer: a chemical oscillation that ticks at a steady tempo while the embryo grows tailward, and every tick stamps down one more body segment.

The trick is converting time into space. A clock by itself just oscillates in place — it would make a whole field of cells flash on and off in unison, producing nothing. To turn ticks into a ruled grid, the embryo combines the clock with a moving boundary. As the tail extends, a "decision line" called the determination wavefront retreats tailward at a constant rate. Cells in front of the line are still negotiating with the clock; cells behind it have committed. Each time the clock reaches its permissive phase at the line, one fresh slab of cells gets locked into a segment. Steady clock × steady-moving line = a row of equal-sized blocks. This is the famous clock-and-wavefront idea, and the segmentation clock is the molecular machine that makes it real.

How the clock works, step by step

The body-building tissue is the presomitic mesoderm (PSM), two long rods of unsegmented cells flanking the neural tube in the embryo's posterior. New cells are continuously added to the tail end of the PSM from a progenitor pool, while somites are chiseled off the anterior (head-facing) end. Here is what happens in one cycle:

1. The oscillator runs. The heart of the clock is a delayed negative-feedback loop. A bHLH transcription factor — Hes7 in mouse, her1/her7 in zebrafish, HES7 in human — represses its own gene. Protein accumulates, shuts off transcription, then is rapidly degraded; transcription restarts, and the cycle repeats. Because there are built-in delays (time to transcribe the gene, splice the introns, export the mRNA, translate it), the system overshoots and undershoots instead of settling — a textbook biochemical oscillator (Lewis, 2003).
2. Waves sweep tail-to-head. Cells are slightly out of phase along the PSM, so a snapshot of cyclic gene expression (Hes7, Lfng, Axin2, Spry2) looks like a wave that travels from the tail toward the head, slowing and compressing as it approaches the front. These are kinematic (phase) waves, not moving material — like the illusion of motion in a stadium "wave."
3. Gradients set the wavefront. A high FGF8 and Wnt3a signal at the posterior keeps cells immature and oscillating. Retinoic acid diffusing from the already-formed somites antagonizes FGF at the anterior. Where the falling FGF8 gradient crosses a threshold, cells stop responding to the clock and become competent to segment. As the embryo elongates, this threshold point — the wavefront — regresses tailward.
4. A boundary is fixed. At the wavefront, the clock phase plus the gradient state switch on the segment-boundary gene Mesp2 in a one-somite-wide stripe. Mesp2 sets up the rostral–caudal polarity of the next segment and triggers a sharp boundary: cells undergo a mesenchymal-to-epithelial transition, round up, and pinch off as an epithelial somite. Eph/ephrin signaling sculpts the physical cleft.
5. Neighbors stay in sync. Throughout, Notch signaling (DLL1/DLL3 ligands, modulated cyclically by the glycosyltransferase LFNG) couples adjacent cells so their individual oscillators keep the same phase. Without this coupling the cells drift apart and boundaries become ragged.

Then it repeats. One tick, one somite pair. In a mouse, that is roughly every two hours, about 65 times, building ~65 somite pairs from head to tail-tip.

The molecular players

• Hes/Her repressors — the pendulum. Hes7 (mouse/human), her1 and her7 (zebrafish) are basic helix-loop-helix repressors that bind their own promoters. Their short half-lives (Hes7 protein and mRNA each turn over in roughly 20 minutes in mouse) make fast oscillation possible. Changing the built-in delay — for example by adding or removing introns from the Hes7 gene — measurably alters the period.
• The Notch pathway — the synchronizer. Membrane ligands DLL1 and DLL3 and the receptor Notch, with cyclic LFNG glycosylation tuning sensitivity, transmit phase information between touching cells. Notch target genes (Hes7, Lfng) are themselves cyclic, weaving Notch into the oscillator.
• FGF8 and Wnt3a — the posterior "stay young" gradient. Transcribed only in the tailbud, FGF8 mRNA decays as cells are left behind by the growing tail, forming a smooth posterior-to-anterior protein gradient that defines the wavefront. Wnt3a sits upstream and drives Axin2 oscillations.
• Retinoic acid — the anterior antagonist. Made by RALDH2 in the formed somites, it opposes FGF and sharpens the wavefront, and helps keep left and right sides symmetric.
• Mesp2 — the boundary stamp. A bHLH factor switched on in a single prospective-somite-width stripe; it defines where the cleft forms and the segment's front-back identity, then activates the Eph/ephrin and Ripply machinery that physically separates the somite.
• Tbx6 — the competence factor. Required in the PSM for Mesp2 to be expressed in the right place; its anterior boundary helps position each new stripe.

Segmentation clock vs circadian clock

The word "clock" appears in two very different biological oscillators. They share the logic of delayed negative feedback but almost nothing else.

Property	Segmentation clock	Circadian clock
Period	~25 min (fish) to ~5 h (human)	~24 hours
Core genes	Hes7 / her1 / her7 (bHLH repressors)	BMAL1/CLOCK ↔ PER/CRY feedback
Output	Body segments (somites)	Daily physiology, sleep, metabolism
Where	Presomitic mesoderm, embryo only	Nearly every adult cell; master clock in SCN
Lifetime	Runs only during somitogenesis (days)	Runs for the whole organism's life
Cell coupling	Notch (juxtacrine, cell contact)	VIP/neuropeptides (paracrine, in SCN)
Entrained by	FGF/Wnt/RA gradients (spatial)	Light, temperature, feeding
Spatial readout?	Yes — converts time into a ruled axis	No — purely temporal

The numbers: tempo, size, and counts

Quantity	Value	Notes
Zebrafish period	~25–30 min (28 °C)	Fastest common model; temperature-dependent
Chick period	~90 min	Classic Hamburger–Hamilton staging by somite count
Mouse period	~120 min	Set by Hes7 loop; ~65 somite pairs total
Human period	~5 h	2–3× slower than mouse even in vitro
Human somite pairs	~42–44	Some caudal pairs later regress
Hes7 protein/mRNA half-life	~20 min (mouse)	Short half-life enables fast cycling
PSM length	several hundred µm	Holds ~1.5–2 wavelengths of the phase wave
Somite formation window	~days 20–35 (human)	The clock runs once, then stops forever

Where it shows up: organisms, evolution, and disease

• Snakes ran the clock fast. A corn snake has ~315 vertebrae versus ~33 in a human. Comparative work (Gomez et al., 2008) showed snakes accelerate the segmentation clock relative to axis elongation — more ticks fit into the growing tail, so far more, smaller somites form. The clock's period-to-growth ratio is a tuning knob evolution uses to set body plan.
• Inter-species tempo difference is intrinsic. When mouse and human PSM-like cells are differentiated from stem cells under identical culture conditions (Matsuda et al., 2020; Diaz-Cuadros et al., 2020), the human clock still ticks 2–3× slower. The cause is slower core biochemistry — protein degradation, mRNA processing — not the environment. This made the human segmentation clock the first developmental oscillator reconstituted in a dish from iPS cells.
• Congenital scoliosis & spondylocostal dysostosis. Human mutations in DLL3 (SCD type 1), MESP2 (type 2), LFNG (type 3), HES7 (type 4), and TBX6 produce hemivertebrae, fused and wedge vertebrae, and disorganized ribs — a short, curved trunk. These are living proof that clock tempo and boundary placement build the skeleton.
• Left–right symmetry. The clock on each side of the embryo must stay in phase so left and right somites form together; retinoic acid buffers the sides against the asymmetric Nodal signals that set up the heart and gut, preventing scoliosis-like axial twisting.
• The repeated body plan of vertebrates. Metameric (segment-by-segment) patterning of vertebrae, ribs, dorsal muscles, and the segmental exit of spinal nerves all trace to how many times the clock ticked and where each Mesp2 stripe landed.

Common misconceptions

• "The waves are physically moving signals." No. The tail-to-head stripes of cyclic gene expression are kinematic (phase) waves. Each cell oscillates in place; neighbors are slightly out of phase, so the pattern appears to travel. Nothing material flows head-ward.
• "Notch generates the rhythm." Notch synchronizes neighboring oscillators; it does not create the oscillation. Single isolated PSM cells can still cycle. Block Notch and the first somites still form (cells start in phase) — only later ones become irregular as cells drift apart.
• "The clock measures absolute size." It measures time, not length. Somite size is set by how far the wavefront moves during one period. Speed up elongation or slow the clock and you get bigger, fewer somites; the reverse gives smaller, more numerous ones.
• "Somites are the vertebrae." Somites are transient epithelial balls. Each one re-splits into sclerotome (bone), myotome (muscle), dermomyotome (dermis), and syndetome (tendon). And each vertebra is built from the posterior half of one somite plus the anterior half of the next — a re-segmentation called resegmentation, which is why your muscles span the joints between bones.
• "It's the same clock as your daily rhythm." The segmentation clock and the ~24 h circadian clock share only the abstract logic of delayed feedback. Different genes, periods differing by tens of fold (~25 min vs ~24 h), different job, and the segmentation clock runs only for a few days in the embryo and then stops forever.
• "All vertebrates tick at the same rate." Period spans from ~25 minutes (zebrafish) to ~5 hours (human) — a 12-fold range — using essentially the same gene network. Tempo is a species-specific, tunable parameter.