Abstract
Abstract: :
Purpose: An enduring problem in rod physiology has been to explain SPR reproducibility in terms of inherently variable underlying biochemistry. We developed a biochemical model with stochastic "front-end" reactions (through PDE* inactivation). SPR reproducibility is achieved via graded inactivation of R* by multiple steps of phosphorylation (Pn) with final quench by arrestin (Arr). We test the model against a broad array of data, including genetic knockout (KO) data. We present a new theoretical framework for evaluating SPR variability, and use simulations and existing data to rule out entire classes of models. Methods: Monte-Carlo simulations of dim-flash responses, assuming: Poisson arrival of photons; G protein (G), R* kinase (RK) & Arr compete to bind with R*, with Pn-dependent binding affinities; R* activity is quenched upon Arr-binding; Ca++-modulation of guanylate cyclase is the only feedback. Results: (1) The MPn model accounts for 4 measures of reproducibility: low coefficients of variation of SPR amplitude (CVampl=0.2-0.25) and SPR area (CVarea=0.3-0.4); ~proportionality between mean2 and variance of dim-flash responses; low SPR variance that peaks much later, and is broader than the mean2 SPR. (2) The MPn model reproduces transduction gain experiments of Rieke & Baylor (1998). When [GTP] is lowered, response amplitude decreases with almost no change in kinetics unless [ATP] is also lowered (to slow Pn). (3) The MPn model reproduces KO data from Arr-/-, RK-/-, GCAPs-/- and RGS9-/- rods. (4) Theory shows that CVarea is a preferred gauge of SPR variability. Classical measures (CVampl or CVduration) overestimate the number of underlying R* inactivation steps. There is a tradeoff between CVampl and CVdur that depends on the relative inactivation kinetics of R* vs downstream reactions. (5) Simulations and KO data rule out local depletion of G or PDE as the main mechanism for SPR reproducibility, and any model in which R* activity is fixed until quenching. Conclusions: (I) Result (2) argues against Rieke & Baylor’s idea that Pn can only account for a small fraction of R* inactivation. (II) A 1-step R* shutoff with Ca++-feedback onto R* lifetime (Whitlock & Lamb, 1999; Pugh & Lamb, 2001) cannot account for SPR behavior (Result 5). (III) Experimental manipulations that selectively slow downstream reactions will increase CVampl, decrease CVdur, with no effect on CVarea. Slowing R* inactivation will shift the balance towards duration variability, but may increase CVarea (Result 4).
Keywords: retina: distal(photoreceptors, horizontal cell • phosphorylation • computational modeling