Biology · Development, Differentiation & Stem Cells
Gene networks, pattern formation and cell identity
Draw two regulators that repress one another. It is tempting to conclude that one must win and permanently determine the cell's identity. But the arrows do not specify how fast either changes, how strong repression is, or how updates occur. A network becomes a predictive model only when its rules are stated.
Gene Regulation & Cell Signalling provides receptors, gene regulation and feedback; Cell Architecture, Growth & Division provides adhesion, cell shape, division and cell death. Gene Expression explains how cells can use a shared genome differently. Begin with those ideas rather than memorising a catalogue of embryonic stages. The numerical cases here need addition, multiplication and reading simple coordinates. Rules for the invented models are introduced beside them; no earlier knowledge of stem-cell assays or embryo anatomy is assumed.
Back to the Biology mapBy the end, you can
Connect regulatory interactions to cell-state patterns while distinguishing a proposed network from a tested mechanism.
A cell identity involves a coordinated programme
Regulatory proteins can activate or repress expression at particular DNA elements. Their products may alter other regulators, receptors, adhesion molecules or metabolic activities. A gene regulatory network represents these relationships. The relevant unit is a set of interactions in a particular cell and context; a gene's name alone is not a complete explanation of a cell type.
A stable cell state can be supported by feedback, chromatin organisation and continuing environmental inputs. Stability means that a state can persist or return after specified disturbances. It does not mean that every molecule stays still: proteins and RNAs can turn over while a regulatory pattern is maintained. The cell-regulation chapter's feedback model introduced this distinction between persistent activity and molecular replacement.
Evidence for an arrow needs a defined meaning. Two RNAs increasing together establishes association. Removing a regulator and changing a target supports a functional dependency, but the effect might pass through intermediate genes or altered cell survival. A direct regulatory interaction needs additional evidence appropriate to the claimed molecular mechanism. Keep 'candidate activates' distinct from 'directly binds and activates'.
Test the diagram with an explicit update rule
Our two-regulator toy uses 1 for active and 0 for inactive. At each step, A takes the opposite of B's previous value, while B takes the opposite of A's previous value. Both update together. Starting at (A,B) = (1,0) keeps (1,0); starting at (0,1) keeps (0,1). Each is a fixed state under this rule.
Starting at (0,0) gives (1,1), then (0,0), then (1,1). Both-inactive and both-active alternate rather than settling into one of the fixed states. This oscillation is a consequence of the assigned synchronous update. It is not a measurement of a developmental oscillator.
The toy makes a practical point: a mutually repressing pair can admit different outcomes, and the connection diagram alone does not prove that every starting condition chooses a stable identity. Graded expression, timing, external inputs and fluctuations can change the result. A biological claim of stable alternative states requires evidence beyond drawing two blunt-ended arrows.
A network's dynamics require more than arrows
Read the diagram in words
The (1,0) and (0,1) starts each remain fixed. The (0,0) start follows (0,0)→(1,1)→(0,0)→(1,1). Starting (1,1) would follow the same alternating cycle with the other phase.
| Regulator A ⊣ Regulator B | Represses |
|---|---|
| Regulator B ⊣ Regulator A | Represses |
A active starting state
| Update step | Regulator A | Regulator B |
|---|---|---|
| 0 | 1 | 0 |
| 1 | 1 | 0 |
| 2 | 1 | 0 |
| 3 | 1 | 0 |
B active starting state
| Update step | Regulator A | Regulator B |
|---|---|---|
| 0 | 0 | 1 |
| 1 | 0 | 1 |
| 2 | 0 | 1 |
| 3 | 0 | 1 |
Both inactive starting state
| Update step | Regulator A | Regulator B |
|---|---|---|
| 0 | 0 | 0 |
| 1 | 1 | 1 |
| 2 | 0 | 0 |
| 3 | 1 | 1 |
Two invented regulators A and B repress one another. Values 1 and 0 mean active and inactive. Both states update together once per step according to A(next) = 1 − B(now), B(next) = 1 − A(now). No noise, delay, graded expression or external input is added.
- A(next) = 1 − B(now)
- B(next) = 1 − A(now)
- Both regulators update together
- Update step
- 1: active
- 0: inactive
- State remains fixed
- States alternate
Local communication can create differences between neighbours
A tissue also connects regulatory states across cells. In a simple contact-inhibition sketch, a selected central cell sends a signal to its two immediate neighbours; they adopt an alternative scored state. The sketch helps explain how unlike neighbours can arise from local communication, but it deliberately leaves the central cell's initial selection unexplained.
In a classic Drosophila study, investigators analysed neural-precursor and epidermal fates using mutations and mosaics involving the Notch signalling system. The outcomes supported inhibitory communication between nearby cells. This is a particular developmental setting, rather than a rule that Notch always means 'become epidermis' in every tissue.
Signals can therefore provide an initial difference, and interactions can reinforce or distribute differences. The resulting pattern must be checked at several levels: regulator expression, differentiated cell behaviour and positions in the tissue. A striped expression photograph does not alone establish how the stripes formed or whether every stripe becomes a distinct organ.
Local inhibition illustrates a neighbour effect
Read the diagram in words
The model supplies a selected central cell. Its contact signals inhibit the selected-state outcome in each immediate neighbour, which adopts the alternative state. Initial selection and larger-scale spacing remain unexplained.
| Left neighbour | Alternative scored state |
|---|---|
| Selected central cell | Selection supplied by the model |
| Right neighbour | Alternative scored state |
| Selected central cell ⊣ Left neighbour | Contact inhibition |
|---|---|
| Selected central cell ⊣ Right neighbour | Contact inhibition |
A deliberately incomplete local inhibition model: the central cell is selected externally, sends an inhibitory contact signal to its two immediate neighbours, and those neighbours adopt the alternative scored state. It illustrates local communication; it does not explain selection of the central cell, repeat spacing or a whole tissue.
- Selection supplied by the model
- Alternative scored state
- How central selection arises is not represented
Worked example
Find the outcome that the arrows did not promise
Apply A(next) = 1 − B(now) and B(next) = 1 − A(now) to (0,0). Both previous values are zero, so both next values are one. On the next step, both previous values are one, so both next values are zero. Repeating the calculation produces an alternating path.
Now begin at (1,0): A(next) = 1 − 0 = 1 and B(next) = 1 − 1 = 0. That state remains fixed. The same connections and update rule support persistence for one starting state and alternation for another. To decide which path a real cell takes, specify or measure its initial state and the actual response dynamics.
Check your understanding
Starting from (A,B) = (1,1), what are the next two states in the toy?
Show reasoning
They are (0,0), then (1,1). Simultaneous updates make both regulators change together; the model does not arbitrarily choose one winner.
A regulator knockout reduces expression of a target RNA. Does it establish direct binding to the target's DNA?
Show reasoning
No. It supports a functional relationship in the tested setting. The change could be indirect, so a claim of direct binding and regulation needs further appropriate evidence.
What missing explanation prevents the three-cell inhibition sketch from predicting a repeated tissue-wide spacing pattern?
Show reasoning
It does not explain which cells become selected initially or how selections interact across a larger tissue. Its supplied central selection illustrates a local effect, not the full pattern-generating mechanism.
What to carry forward
Specify the interactions, their dynamics and their evidence before treating a network as a mechanism of cell identity.
References
Resources supporting this lesson’s explanations and examples.
- Morphogen rules: design principles of gradient-mediated embryo patterning (Development 142:3996–4009, 2015)James Briscoe and Stephen SmallResearch abstract
- The choice of cell fate in the epidermis of Drosophila (Cell 64:1083–1092, 1991)Pascal Heitzler and Pat SimpsonResearch abstract
- Eyeless initiates the expression of both sine oculis and eyes absent during Drosophila compound eye development (Development 125:2181–2191, 1998; article begins on PDF page 7)Georg Halder, Patrick Callaerts, Susanne Flister, Uwe Walldorf, Urs Kloter and Walter J. GehringOriginal research — selected article in university-hosted collection · PDF