Calcium Frequency Decoding

Cytosolic Ca²⁺ is a near-universal signalling currency in eukaryotic cells. It is used for muscle contraction, immune activation, gene transcription, programmed cell death and dozens of other pathways. If a cell's symbiosis decision responded to a Ca²⁺ level, any of these competing pathways could accidentally trigger it. The cell's solution is to encode the symbiosis signal in a specific frequency band of nuclear Ca²⁺ oscillations, and to build a decoder that responds only to that band — a biological matched filter.

What an oscillation looks like

Triggered by Myc-LCO, nuclear Ca²⁺ rises and falls with period 30–100 s, amplitude ~0.5–3 µM above a resting level of ~100 nM. Individual spikes last 10–30 s, with a fast rise and a slower decay. Each spike is localised to the nucleus and perinuclear ER — cytoplasmic Ca²⁺-binding proteins (calreticulin, calmodulin, calbindin) rapidly buffer it before it can spread far. This spatial confinement is what makes the signal a private channel from the LCO receptor to nuclear gene expression.

The decoder: CCaMK as a non-Markovian integrator

CCaMK (Calcium/Calmodulin-dependent protein Kinase) is the master decoder. Its structure is built for the job. At rest it is auto-inhibited; Ca²⁺ binding to its three EF-hand domains partially relieves the inhibition. Crucially, when CCaMK is active it auto-phosphorylates at Thr271, which increases its affinity for Ca²⁺/calmodulin by a factor of ~10. Each spike that activates CCaMK leaves a chemical memory of itself; fast-arriving subsequent spikes find the kinase already primed and push it further into the active state.

Mathematically: the activated fraction of CCaMK at time t is a convolution of the past Ca²⁺ trace with a memory kernel:

$$\text{CCaMK}(t) \;=\; \int_0^t K(t - \tau)\, [\text{Ca}^{2+}]_n(\tau)^2\, d\tau.$$

The square of Ca²⁺ reflects the cooperativity of the EF-hand binding (Hill coefficient ≈ 2–3 in the real system). The kernel K(τ) describes the memory: it is large for recent spikes (auto-phosphorylated state not yet erased) and decays as phosphatases reset the kinase. The simplest form that reproduces the observed behaviour is an underdamped oscillator:

$$K(\tau) = A\, e^{-\tau/\tau_d}\, \cos(\omega_0 \tau),$$

with decay time τ_d ≈ 1–3 min and resonant period T_res = 2π/ω₀ ≈ 30 s. An external Ca²⁺ signal with a periodic component near T_res is amplified by constructive convolution with K(τ); off-resonance inputs are filtered out.

The frequency response

Taking the Fourier transform of K(τ) gives the frequency response of the decoder:

$$|Q(f)|^2 \;\propto\; \frac{1}{(f^2 - f_0^2)^2 + (f/(\pi\tau_d))^2}.$$

This is a sharply peaked band-pass filter centred at the resonant frequency f₀ = ω₀/(2π). The width of the band is set by 1/(πτ_d): a long auto-phosphorylation memory yields a narrow band, a short memory yields a broad one. Evolution has tuned this band to the frequency the fungal partner actually emits.

Why frequency, and not amplitude or duration?

Frequency encoding has three advantages over amplitude or duration codes:

1. Robust to noise. Most cellular noise is broadband; a band-pass filter rejects most of it. A simple threshold detector on Ca²⁺ amplitude would be triggered by random spikes from unrelated pathways.
2. Information capacity. A single channel can encode multiple distinct signals by using different frequencies, just as multiple radio stations share the same broadcast spectrum.
3. Cross-talk avoidance. Other Ca²⁺-driven pathways (heart pacemaker, neuronal action potentials, GnRH pulses) operate at very different frequencies, ensuring no accidental triggering.

Comparison with other frequency-coded signals

What the simulation shows

In the live cell simulation (Module 2 page), watching the Ca²⁺ and CCaMK traces over time reveals the decoder in action: when LCO is sustained at a value that produces near-resonant oscillations, CCaMK climbs smoothly and crosses threshold. When LCO is too low, oscillations are sparse and CCaMK never accumulates enough phosphorylation. When LCO is unrealistically high (a pathological scenario), the oscillator behaviour breaks down and the cell may saturate or even oscillate erratically.

Open questions

• How does CCaMK distinguish the precise LCO-driven frequency from accidental in-band Ca²⁺ events triggered by other stimuli?
• Are there subtypes of CCaMK with different kernel parameters that decode different fungal partners?
• What is the role of stochastic resonance — could low-amplitude noise actually improve detection of weak fungal signals?
• What signal-encoding mechanisms operate after CCaMK activation to convert a binary go/no-go decision into a graded developmental programme?