Phototransduction: rod and cone physiology, the cGMP cascade, and the molecular basis of vision

Intuition Beginner

At the back of your eye, a thin sheet called the retina catches light. About 120 million rod cells give you night vision, and about 6 million cone cells give you daylight and color vision. Each cell is packed with stacks of folded membrane loaded with a light-sensitive protein called rhodopsin. When a single particle of light, a photon, strikes rhodopsin, the protein changes shape like a molecular switch. That one tiny shape change is the first step in everything you see.

That shape change fires an amplifier cascade inside the cell. One rhodopsin activates 800 transducin proteins, each of which switches on a PDE6 enzyme. Together the 800 enzymes destroy about 100,000 molecules of a messenger called cyclic GMP. With the messenger gone, sodium channels in the membrane snap shut. The cell's voltage drops, and it releases less neurotransmitter. The brain reads the change as: a photon arrived here.

Rods are so sensitive they can register individual photons, the dimmest signal physics allows. Cones are less sensitive but come in three flavors — S, M, and L, tuned to blue, green, and red light. By comparing the three cone outputs, your brain recovers color. Phototransduction is how a single quantum of light becomes a nerve signal.

Visual Beginner

The picture shows a rod outer segment with its stack of folded disk membranes at the top, the amplifier cascade in the middle, and the cell's glutamate output at the bottom. In the dark, cyclic GMP holds cation channels open, sodium flows in, and the cell steadily releases glutamate. A photon hitting rhodopsin triggers transducin, then PDE6, which destroys cyclic GMP. Channels close, the cell hyperpolarizes, and glutamate release drops.

The dark current is the key surprise. Unlike most neurons, which fire when they are activated, a photoreceptor is most active in the dark. Light turns it off. The brain reads the silence.

Worked example Beginner

In 1979, Denis Baylor, Trevor Lamb, and King-Wai Yau placed a single toad rod cell inside a glass pipette so fine that it sucked the outer segment into its tip. This suction-pipette technique let them measure the tiny electrical current flowing across the rod's membrane. They delivered very dim flashes of light, so dim that each flash was expected to deliver either 0 or 1 photon to the rod.

Step 1: with no flash, the rod held a steady "dark current" of about 20 picoamperes (a picoampere is one trillionth of an ampere). The cell sat depolarised, quietly releasing glutamate.

Step 2: a dim flash produced one of two outcomes. Either the current dropped by a consistent 1 picoampere for about 1 second, or it did not change at all. The two outcomes appeared in proportion to the predicted photon delivery: roughly 0 photons gave no response, roughly 1 photon gave the 1 picoampere response.

Step 3: repeating the dim flash hundreds of times showed that each single-photon response was the same size. One photon, one stereotyped electrical signature. No averaging required.

What this tells us: a rod cell can detect a single quantum of light. The biophysical amplification is so large that one absorbed photon produces a measurable, repeatable electrical response in a single cell — confirming at the cellular level what Hecht, Shlaer, and Pirenne had inferred in 1942 from human psychophysics alone.

Check your understanding Beginner

Formal definition Intermediate+

Vertebrate phototransduction is the canonical G-protein-coupled receptor (GPCR) cascade of neurobiology. The receptors are the rods and cones of the retina; the second messenger is 3',5'-cyclic GMP (cGMP); and the G-protein is transducin (${\mathrm{G}}_t$). The defining peculiarity — that the receptor potential is hyperpolarising, not depolarising — was established by Tomita and co-workers in 1965 and explained at the molecular level by Fesenko, Yau, and colleagues in 1985 ^{[YauNakatani1985]}.

Definition (Rod photoreceptor). A rod is a ciliated photoreceptor with an outer segment containing a stack of several hundred flattened membrane disks densely packed with the visual pigment rhodopsin (roughly $10^8$ copies per rod, density $\sim 30,000\mu {\mathrm{m}}^{-2}$). Rods mediate scotopic (dim-light) vision; the human retina contains about $1.2\times 10^8$ rods. Their integration time is long ($\sim 200\mathrm{ms}$), which suits them to photon counting at the cost of temporal resolution.

Definition (Cone photoreceptor). A cone is a ciliated photoreceptor whose outer segment contains one of three cone opsins — S (short-wavelength-sensitive, ${\lambda }_{\max }\approx 420\mathrm{nm}$), M (medium, ${\lambda }_{\max }\approx 530\mathrm{nm}$), or L (long, ${\lambda }_{\max }\approx 560\mathrm{nm}$) — conjugated to the same 11-cis-retinal chromophore as rhodopsin. Cones mediate photopic (daylight) vision and color discrimination; the human retina contains about $6\times 10^6$ cones, concentrated in the fovea. Cones are roughly 100-fold less sensitive than rods and recover much faster.

Definition (Rhodopsin). Rhodopsin is a 39 kDa, seven-transmembrane GPCR consisting of the apoprotein opsin covalently bound to the chromophore 11-cis-retinal via a Schiff-base linkage to lysine-296. Absorption of a photon isomerises the chromophore to all-trans-retinal within $\sim 200\mathrm{fs}$, producing the activated intermediate metarhodopsin II (${\mathrm{R}}^{\ast }$ or $\mathrm{MII}$), the catalytically active species that binds and activates transducin ^[Wald1968].

Definition (Phototransduction cascade). The phototransduction cascade is the sequence of biochemical events by which absorption of a photon by rhodopsin produces a hyperpolarising receptor potential. The five molecular players are rhodopsin ($\mathrm{R}$), the heterotrimeric G-protein transducin (${\mathrm{G}}_t\alpha \beta \gamma$, with ${\mathrm{G}}_t\alpha$ carrying the catalytic GDP/GTP site), the effector enzyme cGMP phosphodiesterase-6 (PDE6), the intracellular messenger cGMP, and the cGMP-gated cation channel (CNG) in the plasma membrane. The cascade is

$${\mathrm{R}}^{\ast }\stackrel{\text{catalytic}}{\to }{\mathrm{G}}_t\alpha \text{-GTP}\stackrel{\text{catalytic}}{\to }{PDE6}^{\ast }\stackrel{\text{hydrolysis}}{\to }[\mathrm{cGMP}]\downarrow \stackrel{\text{binding}}{\to }\mathrm{CNG}\text{closes}\stackrel{\text{conductance}}{\to }{V}_m\downarrow .$$

Definition (Dark current). In darkness, cGMP concentration is high ($\sim 4\mu \mathrm{M}$), a fraction of CNG channels are open, and a steady inward current of ${\mathrm{Na}}^{+}$ and ${\mathrm{Ca}}^{2+}$ — the dark current $J_{\mathrm{dark}}\approx 20$ to $60\mathrm{pA}$ in mammalian rods — flows through the outer-segment membrane. The rod is therefore depolarised to $V_{\mathrm{dark}}\approx -40\mathrm{mV}$ and continuously releases glutamate at its synaptic terminal. Light closes CNG channels, the dark current collapses, the membrane hyperpolarises toward $-65\mathrm{mV}$, and glutamate release falls.

Counterexamples to common slips

• Photoreceptors depolarise to light, like most sensory receptors. No. Vertebrate rods and cones hyperpolarise to light, because the light-activated cascade closes cation channels rather than opening them. (Invertebrate photoreceptors, such as those of Drosophila, use a different cascade involving PLC and TRP channels and do depolarise — a convergent solution to the same physical problem via different molecular machinery.)

• Rhodopsin and cone opsins are different chromophores. No. All four vertebrate visual pigments (rhodopsin and the S/M/L cone opsins) share the same chromophore, 11-cis-retinal. The spectral tuning comes from amino-acid residues in the opsin binding pocket that shift the chromophore's absorption maximum; the photochemistry (11-cis to all-trans isomerisation) is identical.

• Light adaptation is just pupil constriction. No. Pupil constriction accounts for roughly one order of magnitude of light-range compensation. The rod itself adapts over roughly five orders of magnitude via the ${\mathrm{Ca}}^{2+}$-dependent GCAP/guanylate-cyclase feedback loop described in the Key mechanism below. Most adaptation happens at the molecular level inside the photoreceptor, before any cortical processing.

• The single-photon response is negligible because a rod contains $10^8$ rhodopsins. No. Baylor, Lamb, and Yau (1979) ^[Baylor1979] showed that a single absorbed photon produces a $\sim 1\mathrm{pA}$, $\sim 1\mathrm{s}$ current response in a toad rod — about $5\%$ of the dark current and easily distinguishable from noise. The amplification makes the response to one rhodopsin out of $10^8$ large enough to measure on a single cell.

Key mechanism: the rhodopsin-transducin-PDE6-cGMP cascade Intermediate+

Mechanism (the amplifying cascade). Absorption of one photon by rhodopsin activates one ${\mathrm{R}}^{\ast }$ molecule. The ${\mathrm{R}}^{\ast }$ catalyses GDP-to-GTP exchange on transducin at a rate ${\nu }_{\mathrm{R}}\sim 100$ to $1000{\mathrm{s}}^{-1}$ for an active lifetime ${\tau }_{\mathrm{R}}\approx 40\mathrm{ms}$, producing $N_{\mathrm{G}}\approx 800$ activated transducins (${\mathrm{G}}_t\alpha$-GTP). Each ${\mathrm{G}}_t\alpha$-GTP binds and activates one PDE6 catalytic subunit. Each activated PDE6 hydrolyses cGMP at $k_{\mathrm{cat}}\sim 10^3$ to $10^4{\mathrm{s}}^{-1}$. The total cGMP hydrolysed over the cascade time ${\tau }_{\mathrm{casc}}\approx 100\mathrm{ms}$ is

$$N_{\mathrm{cGMP}}\approx N_{\mathrm{G}}\cdot k_{\mathrm{cat}}\cdot {\tau }_{\mathrm{casc}}\sim 800\times 10^3\times 0.1\sim 10^5,$$

i.e. approximately $10^5$ cGMP molecules destroyed per photon. The drop in cGMP closes several hundred CNG channels, the dark current drops by about $1\mathrm{pA}$, and the membrane hyperpolarises by a few millivolts for roughly one second. Three of the four steps in the cascade are catalytic, which is the structural reason the single-photon response is large enough to measure.

Derivation. (i) Photoisomerisation. A 500-nm photon carries energy $E=hc/\lambda \approx 4\times 10^{-19}\mathrm{J}$. Absorption by rhodopsin isomerises the 11-cis-retinal chromophore to all-trans-retinal through a sequence of spectroscopic intermediates (bathorhodopsin $\to$ lumirhodopsin $\to$ metarhodopsin I $\to$ metarhodopsin II), reaching the catalytically active $\mathrm{MII}$ state in $\sim 1\mathrm{ms}$. The quantum yield is $\Phi \approx 0.67$: two out of three absorbed photons produce an active ${\mathrm{R}}^{\ast }$.

(ii) Rhodopsin to transducin: the first amplification. ${\mathrm{R}}^{\ast }$ is a GPCR in its active conformation. It diffuses in the disk membrane and collides with transducin heterotrimers at the disk surface. Each productive collision induces GDP release and GTP binding on ${\mathrm{G}}_t\alpha$, after which ${\mathrm{G}}_t\alpha$-GTP dissociates from ${\mathrm{G}}_t\beta \gamma$. The catalytic rate is ${\nu }_{\mathrm{R}}\approx 120$ to $1000{\mathrm{s}}^{-1}$ in vertebrate rods at $37{}^{\circ }\mathrm{C}$. The active lifetime of a single ${\mathrm{R}}^{\ast }$ is set by rhodopsin kinase phosphorylation followed by arrestin binding and is ${\tau }_{\mathrm{R}}\approx 40$ to $80\mathrm{ms}$. The number of transducins activated by one ${\mathrm{R}}^{\ast }$ is therefore

$$N_{\mathrm{G}}={\nu }_{\mathrm{R}}\cdot {\tau }_{\mathrm{R}}\sim 10^3\times 0.04\sim 600\text{ to }1000\approx 800.$$

(iii) Transducin to PDE6: the second amplification. Each ${\mathrm{G}}_t\alpha$-GTP binds the $\gamma$-inhibitory subunit of PDE6, relieving inhibition and producing an active PDE6 dimer. The coupling is one-to-one and stoichiometric (each activated transducin activates one PDE6). The number of active PDE6 molecules is therefore also $N_{\mathrm{G}}\approx 800$.

(iv) PDE6 to cGMP: the third and dominant amplification. Activated PDE6 hydrolyses cGMP to 5'-GMP at a catalytic rate $k_{\mathrm{cat}}\approx 1000{\mathrm{s}}^{-1}$ (per $\alpha \beta$ catalytic dimer, with $K_M\approx 100\mu \mathrm{M}$). With $N_{\mathrm{G}}$ active PDE6 dimers acting over ${\tau }_{\mathrm{casc}}\approx 100\mathrm{ms}$, the total cGMP destroyed is

$$N_{\mathrm{cGMP}}\approx N_{\mathrm{G}}\cdot k_{\mathrm{cat}}\cdot {\tau }_{\mathrm{casc}}\sim 800\times 10^3\times 0.1\sim 8\times 10^4\text{ to }10^5,$$

i.e. roughly $10^5$ cGMP molecules hydrolysed per absorbed photon. This is the dominant amplification step of the cascade: a single molecule of ${\mathrm{R}}^{\ast }$ leads to the destruction of about $10^5$ molecules of cGMP, an amplification of $10^5$-fold.

(v) cGMP to channel closure to voltage. The CNG channel opens cooperatively with cGMP (Hill coefficient $n\approx 3$), so the open probability is

$$P_{\mathrm{open}}=\frac{[\mathrm{cGMP}{]}^n}{K_{1/2}^n+[\mathrm{cGMP}{]}^n},K_{1/2}\approx 50\mu \mathrm{M}.$$

The dark cGMP level is $\sim 4\mu \mathrm{M}$ in the outer segment cytosol; a single-photon response drops it locally by several percent, closing several hundred of the $\sim 10^4$ CNG channels in the outer-segment plasma membrane. The dark current falls by approximately $1\mathrm{pA}$ out of $J_{\mathrm{dark}}\approx 20\mathrm{pA}$, the membrane input resistance ($\sim 500\mathrm{M\Omega }$) translates this into a hyperpolarisation of $\Delta V\approx -1\times 500\approx -0.5\mathrm{mV}$ at the soma, and the change propagates passively to the synaptic terminal where it reduces the calcium influx that drives glutamate release.

(vi) Recovery and adaptation: the ${\mathrm{Ca}}^{2+}$ feedback loop. In the dark, ${\mathrm{Ca}}^{2+}$ enters through CNG channels and is extruded by a ${\mathrm{Na}}^{+}/{\mathrm{Ca}}^{2+}\text{-}{\mathrm{K}}^{+}$ exchanger. When CNG channels close, ${\mathrm{Ca}}^{2+}$ influx stops but extrusion continues, so intracellular $[{\mathrm{Ca}}^{2+}]$ falls. Low ${\mathrm{Ca}}^{2+}$ liberates GCAP (guanylate-cyclase-activating protein) from its ${\mathrm{Ca}}^{2+}$-bound form; the apo-form of GCAP stimulates guanylate cyclase (RetGC) to resynthesise cGMP from GTP. This negative-feedback loop is the principal mechanism of light adaptation: it restores cGMP after a flash, speeds recovery, and allows the photoreceptor to operate across roughly five orders of magnitude of background light. Rhodopsin kinase and arrestin provide a parallel shutoff on the ${\mathrm{R}}^{\ast }$ side; the RGS9 complex accelerates ${\mathrm{G}}_t\alpha$-GTP hydrolysis on the transducin side. Recovery of all three catalytic steps is required for the rod to return to its dark state within $\sim 200\mathrm{ms}$. $\square$

Bridge. The three catalytic stages — ${\mathrm{R}}^{\ast }$ activating $\sim 800$ transducins, each activating one PDE6, each hydrolysing $\sim 10^2$ cGMP molecules — is exactly the structure that turns a single-molecule event into a measurable cellular signal, and this is the foundational reason the Baylor-Lamb-Yau single-photon response is reproducible at the $\sim 1\mathrm{pA}$ scale. The cascade generalises across every GPCR in biology: the same G-protein-cycle logic reappears in olfaction (odorant receptor $\to$ ${\mathrm{G}}_{\mathrm{olf}}$ $\to$ adenylyl cyclase $\to$ cAMP-gated channel, 18.13.03), and the cGMP second messenger reappears in smooth-muscle relaxation (nitric oxide $\to$ guanylate cyclase $\to$ cGMP $\to$ PKG). The bridge is from a retinal GPCR to the universal G-protein-coupled signalling grammar of eukaryotic cell biology, and the pattern appears again in 29.03.04 as the input stage whose output — glutamate release from photoreceptors — is the raw material for the cortical feature extraction of Hubel and Wiesel.

Exercises Intermediate+

Advanced results Master

Result 1 (Wald 1933-1968: rhodopsin and the retinal cycle). George Wald, working at Harvard from the 1930s, identified the visual pigments and established that vitamin A is the dietary precursor of the visual chromophore ^[Wald1935]. Over three decades his laboratory traced the photochemistry: rhodopsin is a complex of opsin with 11-cis-retinal, absorption isomerises the chromophore to all-trans-retinal, and a metabolic cycle in the retinal pigment epithelium (the retinoid cycle) regenerates 11-cis-retinal in the dark. The 1967 Nobel Prize in Physiology or Medicine (shared with Granit and Hartline) cited "their discoveries concerning the primary physiological and chemical visual processes in the eye"; Wald's Nobel Lecture ^[Wald1968] is the canonical summary.

Result 2 (Hecht-Shlaer-Pirenne 1942: the psychophysical threshold). Before any single-cell recording existed, Hecht, Shlaer, and Pirenne ^[Hecht1942] inferred that human rod vision operates at the quantum limit. Subjects reliably detected flashes delivering an average of 5 to 14 photons at the cornea (with optical losses and disk-arrival corrections, this is consistent with 1 to 2 photons absorbed per rod at threshold). Their statistical argument — that a Poisson-distributed number of photon absorptions must cross a threshold of roughly 5 to 7 independent rod events — predicted that an individual rod must respond to a single photon, three decades before this was confirmed electrophysiologically.

Result 3 (Baylor-Lamb-Yau 1979: single-photon detection at the cellular level). Baylor, Lamb, and Yau, working at Yale, used the suction-pipette technique on single toad rod outer segments ^[Baylor1979]. At flash intensities expected to deliver 0 or 1 photon, responses fell into two populations: a stereotyped $\sim 1\mathrm{pA}$ current suppression lasting $\sim 1\mathrm{s}$, or no response. The single-photon response was highly reproducible in shape and amplitude, with a coefficient of variation of $\sim 0.2$ — far smaller than the Poisson variation that free-running rhodopsin lifetimes would produce. This narrow distribution is the empirical signature of the all-or-none amplifier cascade, and it confirmed the Hecht-Shlaer-Pirenne prediction at the level of individual cells.

Result 4 (Yau-Nakatani 1985: the cGMP-gated channel). Yau and Nakatani, working at the Marines Biological Laboratory, used a truncated rod outer segment to demonstrate that the light-sensitive conductance of the plasma membrane is gated directly by cGMP ^{[YauNakatani1985]}. This settled a long debate between the "${\mathrm{Ca}}^{2+}$ hypothesis" (Hagins, in which light released a calcium buffer) and the "cGMP hypothesis" (Fesenko and Yau) by showing that cGMP, not calcium, is the internal ligand that directly opens the channel. The result generalised to all vertebrate photoreceptors and established cGMP as a bona fide second messenger in sensory transduction, on a par with cAMP.

Result 5 (Stryer 1985-1986: the amplification mechanism). Lubert Stryer and colleagues showed that the rhodopsin-transducin-PDE6 cascade produces the massive signal amplification that Baylor's single-photon responses had implied ^[Stryer1986]. The 1985 Biochemistry paper measured the catalytic rates in reconstituted systems and the 1986 Annual Review of Neuroscience synthesis framed the cascade as an amplifier with three gain stages: rhodopsin activating $\sim 800$ transducins per photon, each activating one PDE6, each hydrolysing $\sim 10^2$ cGMP molecules. The product is the $\sim 10^5$-fold single-photon amplification, which Stryer explicitly compared to a photomultiplier tube and identified as the structural reason the rod is single-photon sensitive.

Result 6 (Nathans 1986: the cone opsin genes and trichromacy). Nathans, Thomas, and Hogness cloned the human S, M, and L cone opsin genes from a genomic library ^{[Nathans1986]}. The S opsin sits on chromosome 7 and is only $\sim 40\%$ identical to the others; the M and L opsins sit head-to-tail on the X chromosome and share $\sim 98\%$ amino-acid identity, reflecting a recent primate-specific duplication. Three amino-acid substitutions at the retinal-binding pocket tune ${\lambda }_{\max }$ by $\sim 30\mathrm{nm}$ between M and L. The work explained the molecular genetics of normal and anomalous trichromacy, the high incidence of red-green color-vision deficiency in males (unequal crossing over on the M/L tandem array), and established that the cone opsins are GPCRs homologous to rhodopsin.

Result 7 (Palczewski 2000: rhodopsin as the GPCR prototype). Palczewski and colleagues solved the crystal structure of bovine rhodopsin at $2.8\mathring{\mathrm{A}}$ resolution ^{[Palczewski2000]}, the first high-resolution structure of any GPCR. The structure revealed the seven-transmembrane helix bundle, the 11-cis-retinal chromophore in its Schiff-base linkage to Lys-296, and the conformational constraints that keep rhodopsin silent in the dark. Because rhodopsin is the only GPCR that is naturally locked in a single well-defined inactive conformation by its covalently bound inverse-agonist chromophore, the structure served as the template for homology modelling of hundreds of other GPCRs, including the ${\beta }_2$-adrenergic receptor (solved in 2007 by Kobilka and Stevens, 2012 Nobel Prize in Chemistry).

Result 8 (Boyden-Deisseroth 2005: optogenetics and the microbial rhodopsins). The discovery that microbial rhodopsins (channelrhodopsin-2 from Chlamydomonas, a type I microbial rhodopsin unrelated to animal type II rhodopsins) could be expressed in mammalian neurons and used to drive action potentials with millisecond precision ^[Boyden2005] opened the field of optogenetics. The same physical chemistry — retinal isomerisation driving a conformational change in a membrane protein — is repurposed: instead of closing a cGMP-gated channel via a G-protein cascade, channelrhodopsin is itself the channel, opening directly upon illumination. The convergence is physical (the retinal chromophore) and conceptual (light-driven membrane-protein conformational change as a generic actuator), even though the molecular implementations diverged over two billion years of evolution.

Synthesis. The foundational reason phototransduction is the canonical GPCR cascade of biology is that rhodopsin is the GPCR solved first — by Palczewski structurally in 2000 — and the one whose kinetics Stryer measured quantitatively in 1985. This is exactly the structural template that was then exported to hundreds of hormone and neurotransmitter GPCRs, and putting these together with the Baylor 1979 single-photon measurement identifies the rhodopsin-transducin-PDE6 cascade with the universal amplifying grammar of G-protein signalling. The central insight — that three catalytic stages in series produce $10^5$-fold signal amplification — generalises to olfaction (18.13.03, where the same G-protein cycle drives a cAMP-gated channel via ${\mathrm{G}}_{\mathrm{olf}}$) and to smooth-muscle relaxation (NO $\to$ guanylate cyclase $\to$ cGMP). The bridge is from a retinal GPCR to the entire GPCR superfamily, and the pattern recurs when channelrhodopsin, a convergent microbial rhodopsin, becomes the optogenetic actuator that drives modern circuit neuroscience 29.03.04. The dual use of retinal photochemistry — once as a sensor (animal vision) and once as an actuator (microbial ion pumps and channels) — closes a loop between sensory transduction and neural control that the field is still exploring.

Full proof set Master

Proposition (Steady-state cGMP under light step and the adaptation locus). Let the cytosolic cGMP concentration in a rod outer segment obey $\frac{d}{dt}[\mathrm{cGMP}]=\alpha ([{\mathrm{Ca}}^{2+}])-({\beta }_{\mathrm{dark}}+{\beta }_{\mathrm{light}})[\mathrm{cGMP}]$, where $\alpha$ is the GCAP-dependent guanylate-cyclase rate and ${\beta }_{\mathrm{dark}}+{\beta }_{\mathrm{light}}$ is the total PDE6 activity. Assume GCAP feedback acts on a slower timescale than the cGMP hydrolysis. Then (i) the early response to a step increase in ${\beta }_{\mathrm{light}}$ is a fall in $[\mathrm{cGMP}]$ toward the intermediate steady state $[\mathrm{cGMP}{]}_{\mathrm{fast}}=\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{dark}})/({\beta }_{\mathrm{dark}}+{\beta }_{\mathrm{light}})$, and (ii) the late, adapted steady state has $[\mathrm{cGMP}{]}_{\mathrm{slow}}=\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{adapted}})/({\beta }_{\mathrm{dark}}+{\beta }_{\mathrm{light}})$, where $\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{adapted}})>\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{dark}})$ because lowered ${\mathrm{Ca}}^{2+}$ activates GCAP. The adapted $[\mathrm{cGMP}]$ partially recovers toward the dark value, the basis of Weber-law light adaptation over roughly five orders of magnitude of background intensity.

Proof. (i) Treat $\alpha$ as a slow variable and hold it at its dark value ${\alpha }_0=\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{dark}})$ over the fast timescale. The rate equation becomes linear in $[\mathrm{cGMP}]$ with constant coefficients:

$$\frac{d[\mathrm{cGMP}]}{dt}={\alpha }_0-{\beta }_{\mathrm{tot}}[\mathrm{cGMP}],{\beta }_{\mathrm{tot}}={\beta }_{\mathrm{dark}}+{\beta }_{\mathrm{light}}.$$

The solution with initial condition $[\mathrm{cGMP}](0)=[\mathrm{cGMP}{]}_{\mathrm{dark}}={\alpha }_0/{\beta }_{\mathrm{dark}}$ is

$$[\mathrm{cGMP}](t)=\frac{{\alpha }_0}{{\beta }_{\mathrm{tot}}}+\left(\frac{{\alpha }_0}{{\beta }_{\mathrm{dark}}}-\frac{{\alpha }_0}{{\beta }_{\mathrm{tot}}}\right)e^{-{\beta }_{\mathrm{tot}}t}.$$

The fast transient relaxes to $[\mathrm{cGMP}{]}_{\mathrm{fast}}={\alpha }_0/{\beta }_{\mathrm{tot}}$ on the timescale $1/{\beta }_{\mathrm{tot}}\sim 100\mathrm{ms}$, i.e. on the cascade timescale observed in single-photon responses. For a step doubling total PDE6 activity (${\beta }_{\mathrm{light}}={\beta }_{\mathrm{dark}}$), $[\mathrm{cGMP}{]}_{\mathrm{fast}}={\alpha }_0/(2{\beta }_{\mathrm{dark}})=[\mathrm{cGMP}{]}_{\mathrm{dark}}/2$.

(ii) On the slower adaptation timescale, the closure of CNG channels reduces ${\mathrm{Ca}}^{2+}$ influx while the ${\mathrm{Na}}^{+}/{\mathrm{Ca}}^{2+}\text{-}{\mathrm{K}}^{+}$ exchanger continues to extrude ${\mathrm{Ca}}^{2+}$, so cytosolic $[{\mathrm{Ca}}^{2+}]$ falls over $\sim 500\mathrm{ms}$ to seconds. As $[{\mathrm{Ca}}^{2+}]$ falls, ${\mathrm{Ca}}^{2+}$ dissociates from GCAP1 and GCAP2; apo-GCAP is a potent activator of retinal guanylate cyclases RetGC1 and RetGC2, increasing $\alpha$. The system migrates from $[\mathrm{cGMP}{]}_{\mathrm{fast}}$ to the adapted steady state

$$[\mathrm{cGMP}{]}_{\mathrm{slow}}=\frac{\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{adapted}})}{{\beta }_{\mathrm{tot}}},$$

with $\alpha ([{\mathrm{Ca}}^{2+}{]}_{\mathrm{adapted}})>{\alpha }_0$. Because CNG channel open probability depends on $[\mathrm{cGMP}{]}^n$ with Hill coefficient $n\approx 3$, even a partial recovery of $[\mathrm{cGMP}]$ produces a substantial restoration of the dark current. This partial restoration is exactly the phenomenon of light adaptation: a steady background that initially halved the dark current will, after adaptation, suppress it by much less than half, restoring sensitivity to incremental flashes.

The Weber-Fechner law for the rod, $1/{\tau }_{\mathrm{sens}}\propto I_{\mathrm{background}}$, emerges when the feedback loop is fast enough to hold $[\mathrm{cGMP}]$ within a narrow range across background intensities spanning five orders of magnitude. Rhodopsin kinase phosphorylation of ${\mathrm{R}}^{\ast }$ and the RGS9-catalysed GTP hydrolysis on ${\mathrm{G}}_t\alpha$ provide parallel recovery routes on the catalytic side; together with GCAP they constitute the molecular machinery that returns the rod to its ready state in roughly $200\mathrm{ms}$, fast enough to support fusion flicker at $\sim 15\mathrm{Hz}$ in rods and $\sim 60\mathrm{Hz}$ in cones. $\square$

Proposition (Total amplification of the single-photon response). A single activated rhodopsin ${\mathrm{R}}^{\ast }$ in a mammalian rod activates $N_{\mathrm{G}}\approx {\nu }_{\mathrm{R}}{\tau }_{\mathrm{R}}$ transducins, each activating one PDE6 dimer, each hydrolysing cGMP at rate $k_{\mathrm{cat}}$ over the cascade time ${\tau }_{\mathrm{casc}}$. The total number of cGMP molecules hydrolysed per photon is

$$N_{\mathrm{cGMP}}={\nu }_{\mathrm{R}}{\tau }_{\mathrm{R}}{k}_{\mathrm{cat}}{\tau }_{\mathrm{casc}}.$$

With ${\nu }_{\mathrm{R}}\approx 10^3{\mathrm{s}}^{-1}$, ${\tau }_{\mathrm{R}}\approx 4\times 10^{-2}\mathrm{s}$, $k_{\mathrm{cat}}\approx 10^3{\mathrm{s}}^{-1}$, ${\tau }_{\mathrm{casc}}\approx 10^{-1}\mathrm{s}$, this gives $N_{\mathrm{cGMP}}\approx 4\times 10^3\times 10^2=4\times 10^5$, in order-of-magnitude agreement with the empirically measured $\sim 10^5$ cGMP per photon.

Proof. Each activated transducin (${\mathrm{G}}_t\alpha$-GTP) binds the $\gamma$-inhibitory subunit of PDE6 and relieves inhibition stoichiometrically, so the number of active PDE6 dimers equals the number of active transducins $N_{\mathrm{G}}$. Each active PDE6 dimer hydrolyses cGMP at rate $k_{\mathrm{cat}}$ until the transducin hydrolyses its bound GTP (catalysed by RGS9) and dissociates; the effective time over which a typical PDE6 is active is ${\tau }_{\mathrm{casc}}$, of the same order as ${\tau }_{\mathrm{R}}$. The total cGMP hydrolysis is therefore the integral

$$N_{\mathrm{cGMP}}={\int }_0^{\infty }{N}_{PDE6}(t)k_{\mathrm{cat}}dt\approx N_{\mathrm{G}}{k}_{\mathrm{cat}}{\tau }_{\mathrm{casc}}={\nu }_{\mathrm{R}}{\tau }_{\mathrm{R}}{k}_{\mathrm{cat}}{\tau }_{\mathrm{casc}}.$$

Substituting the empirical rate constants yields the order-of-magnitude estimate. Three of the four multiplicative factors are catalytic turnover numbers (${\nu }_{\mathrm{R}}$, $k_{\mathrm{cat}}$, and the implicit one-to-one transducin-to-PDE6 stoichiometry), and the two lifetimes (${\tau }_{\mathrm{R}}$, ${\tau }_{\mathrm{casc}}$) are controlled by separate shutoff machineries (rhodopsin kinase/arrestin and RGS9, respectively). The narrow distribution of single-photon response amplitudes observed by Baylor, Lamb, and Yau (1979) follows: because ${\tau }_{\mathrm{R}}$ is set by the stochastic arrival of arrestin at the phosphorylated rhodopsin, the variance of $N_{\mathrm{G}}$ would naively be Poisson, but the dominant noise source is in fact the ${\mathrm{Ca}}^{2+}$-feedback modulation of the CNG channel gain, which is shared across the response and does not contribute multiplicative noise. This is the structural reason the single-photon response has a coefficient of variation of only $\sim 0.2$ rather than the $\sim 1$ that a strictly Poisson cascade would predict. $\square$

Connections Master

• Sensory systems survey 18.13.01 is the immediate prerequisite and parent unit, whose general framework — sensory receptor cells converting a physical stimulus into a graded receptor potential via an ion-channel mechanism — this unit instantiates in full molecular detail for vision. Phototransduction is the most thoroughly characterised of the receptor mechanisms sketched in 18.13.01, and the cGMP-cascade grammar established here reappears, with the substitution of cAMP for cGMP, in the olfactory transduction of 18.13.03.

• Hubel-Wiesel visual cortex architecture 29.03.04 is the downstream cortical partner of this unit, consuming the glutamate signal that rods and cones produce at the photoreceptor-to-bipolar synapse and re-encoding it as orientation-, ocular-dominance-, and direction-tuned spikes in V1. The entire cortical feature-extraction machinery of 29.03.04 is downstream of, and bounded in dynamic range by, the phototransduction events described here: the cortical hypercolumn cannot represent features the photoreceptors did not encode.

• Hair cell mechanotransduction and cochlear tuning 18.13.02 provides the chapter-closing sensory peer, illustrating the other major class of vertebrate transduction mechanism — direct mechanical gating of an ion channel, rather than G-protein-coupled cascade gating. The contrast is instructive: phototransduction trades speed for sensitivity (cascades amplify but are slow, $\sim 100\mathrm{ms}$ integration time), whereas hair-cell transduction is fast ($\sim 10\mu \mathrm{s}$) but unsuitable for single-quantum sensitivity. The two together bracket the design space of vertebrate sensory receptors.

• Chemosensation: taste, smell, and the Buck-Axel olfactory receptor revolution 18.13.03 shares the GPCR-cascade grammar with this unit, with the substitution of odorant receptors and ${\mathrm{G}}_{\mathrm{olf}}$ for rhodopsin and transducin, and cAMP for cGMP. The same catalytic amplification logic — receptor activates G-protein activates effector enzyme produces second messenger opens cyclic-nucleotide-gated channel — appears in both, and the two cyclic-nucleotide-gated channel families (CNG for vision, CNG for olfaction) are homologous. Olfaction is vision's biochemical sibling.

Historical & philosophical context Master

George Wald, working at Harvard from 1933, traced rhodopsin to its molecular basis over three decades ^[Wald1935]. His laboratory identified 11-cis-retinal as the light-absorbing chromophore in 1938, demonstrated that vitamin A deficiency abolishes dark adaptation (the molecular basis of night blindness), and elaborated the retinoid cycle by which the retinal pigment epithelium regenerates 11-cis-retinal in the dark. The 1967 Nobel Prize in Physiology or Medicine, shared with Ragnar Granit and Haldan Keffer Hartline, cited the three laboratories' complementary discoveries of the primary visual processes. Wald's Nobel Lecture ^[Wald1968] framed the entire field: "the act of vision begins with the absorption of a photon by retinal."

The single-photon threshold, predicted psychophysically by Hecht, Shlaer, and Pirenne in 1942 ^[Hecht1942], was confirmed at the cellular level by Baylor, Lamb, and Yau in 1979 ^[Baylor1979] using the suction-pipette technique they had developed. The molecular cascade underlying that single-photon response was decoded in the mid-1980s: Yau and Nakatani (1985) identified the cGMP-gated channel ^{[YauNakatani1985]}; Stryer's group (1985-1986) measured the catalytic amplification ^[Stryer1986] and framed the cascade as a three-stage photomultiplier; Fesenko and colleagues demonstrated directly that cGMP gates the channel without requiring calcium. Nathans, Thomas, and Hogness (1986) cloned the cone opsin genes ^{[Nathans1986]}, placing color vision on a secure molecular-genetic footing and explaining the high incidence of red-green color-vision deficiency by the recent, unstable duplication of the M/L gene array on the X chromosome.

The structure of rhodopsin at $2.8\mathring{\mathrm{A}}$ resolution by Palczewski and colleagues in 2000 ^{[Palczewski2000]} made rhodopsin the first solved GPCR and the template for the entire GPCR superfamily, which now includes more than 800 human receptors and is the target of roughly one third of all FDA-approved drugs. The optogenetic revolution of 2005 ^[Boyden2005] repurposed the unrelated microbial type I rhodopsins (channelrhodopsins) as genetically targetable neural actuators, exploiting the same retinal photochemistry in a new role. The two threads — sensory and actuator — converge on the same physical chemistry of retinal isomerisation, even though the animal type II and microbial type I rhodopsins diverged before the eukaryote-prokaryote split roughly two billion years ago.

Bibliography Master

@article{Wald1935,
  author = {Wald, George},
  title = {Carotenoids and the visual cycle},
  journal = {Journal of General Physiology},
  volume = {19},
  year = {1935},
  pages = {351--371},
}

@article{Wald1968,
  author = {Wald, George},
  title = {The molecular basis of visual excitation},
  journal = {Nature},
  volume = {219},
  year = {1968},
  pages = {800--807},
}

@article{Hecht1942,
  author = {Hecht, Selig and Shlaer, Simon and Pirenne, Maurice H.},
  title = {Energy, quanta, and vision},
  journal = {Journal of General Physiology},
  volume = {25},
  year = {1942},
  pages = {819--840},
}

@article{Baylor1979,
  author = {Baylor, Denis A. and Lamb, Trevor D. and Yau, King-Wai},
  title = {Responses of retinal rods to single photons},
  journal = {Journal of Physiology},
  volume = {288},
  year = {1979},
  pages = {613--634},
}

@article{YauNakatani1985,
  author = {Yau, King-Wai and Nakatani, Keiji},
  title = {Light-suppressible, cyclic {GMP}-sensitive conductance in the plasma membrane of a truncated rod outer segment},
  journal = {Nature},
  volume = {317},
  year = {1985},
  pages = {252--255},
}

@article{Stryer1986,
  author = {Stryer, Lubert},
  title = {Cyclic {GMP} cascade of vision},
  journal = {Annual Review of Neuroscience},
  volume = {9},
  year = {1986},
  pages = {87--119},
}

@article{Nathans1986,
  author = {Nathans, Jeremy and Thomas, David and Hogness, David S.},
  title = {Molecular genetics of human color vision: the genes encoding blue, green, and red pigments},
  journal = {Science},
  volume = {232},
  year = {1986},
  pages = {193--202},
}

@article{Palczewski2000,
  author = {Palczewski, Krzysztof and Kumasaka, Tetsuya and Hori, Takashi and Behnke, Craig A. and Motoshima, Hiroyuki and Fox, Brian A. and Le Trong, Isolde and Teller, David C. and Okada, Tetsuji and Stenkamp, Ronald E. and Yamamoto, Masashi and Miyano, Mitsue},
  title = {Crystal structure of rhodopsin: a {G} protein-coupled receptor},
  journal = {Science},
  volume = {289},
  year = {2000},
  pages = {739--745},
}

@article{Boyden2005,
  author = {Boyden, Edward S. and Zhang, Feng and Bamberg, Ernst and Nagel, Georg and Deisseroth, Karl},
  title = {Millisecond-timescale, genetically targeted optical control of neural activity},
  journal = {Nature Neuroscience},
  volume = {8},
  year = {2005},
  pages = {1263--1268},
}

@book{Lodish2008,
  author = {Lodish, Harvey and Berk, Arnold and Kaiser, Chris A. and Krieger, Monty and Scott, Matthew P. and Bretscher, Anthony and Ploegh, Hidde and Matsudaira, Paul},
  title = {Molecular Cell Biology},
  edition = {6th},
  publisher = {W. H. Freeman},
  address = {New York},
  year = {2008},
}

@book{Kandel2013,
  author = {Kandel, Eric R. and Schwartz, James H. and Jessell, Thomas M. and Siegelbaum, Steven A. and Hudspeth, A. J.},
  title = {Principles of Neural Science},
  edition = {5th},
  publisher = {McGraw-Hill},
  address = {New York},
  year = {2013},
}