Daniel Tranchina

Stochastic mRNA Synthesis in Mammalian Cells

Individual cells in genetically homogeneous populations have been found to express different numbers of molecules of specific proteins. We investigated the origins of these variations in mammalian cells by counting individual molecules of mRNA produced from a reporter gene that was stably integrated into the cells genome. We found that there are massive variations in the number of mRNA molecules present in each cell. These variations occur because mRNAs are synthesized in short but intense bursts of transcription beginning when the gene transitions from an inactive to an active state and ending when they transition back to the inactive state. We show that these transitions are intrinsically random and not due to global, extrinsic factors such as the levels of transcriptional activators. Moreover, the gene activation causes burst-like expression of all genes within a wider genomic locus. We further found that bursts are also exhibited in the synthesis of natural genes. The bursts of mRNA expression can be buffered at the protein level by slow protein degradation rates. A stochastic model of gene activation and inactivation was developed to explain the statistical properties of the bursts. The model showed that increasing the level of transcription factors increases the average size of the bursts rather than their frequency. These results demonstrate that gene expression in mammalian cells is subject to large, intrinsically random fluctuations and raise questions about how cells are able to function in the face of such noise.

A population density approach that facilitates large-scale modeling of neural networks: analysis and an application to orientation tuning.

We explore a computationally efficient method of simulating realistic networks of neurons introduced by Knight, Manin, and Sirovich (1996) in which integrate-and-fire neurons are grouped into large populations of similar neurons. For each population, we form a probability density that represents the distribution of neurons over all possible states. The populations are coupled via stochastic synapses in which the conductance of a neuron is modulated according to the firing rates of its presynaptic populations. The evolution equation for each of these probability densities is a partial differential-integral equation, which we solve numerically. Results obtained for several example networks are tested against conventional computations for groups of individual neurons.We apply this approach to modeling orientation tuning in the visual cortex. Our population density model is based on the recurrent feedback model of a hypercolumn in cat visual cortex of Somers et al. (1995). We simulate the response to oriented flashed bars. As in the Somers model, a weak orientation bias provided by feed-forward lateral geniculate input is transformed by intracortical circuitry into sharper orientation tuning that is independent of stimulus contrast.The population density approach appears to be a viable method for simulating large neural networks. Its computational efficiency overcomes some of the restrictions imposed by computation time in individual neuron simulations, allowing one to build more complex networks and to explore parameter space more easily. The method produces smooth rate functions with one pass of the stimulus and does not require signal averaging. At the same time, this model captures the dynamics of single-neuron activity that are missed in simple firing-rate models.

A model for the polarization of neurons by extrinsically applied electric fields

A model is presented for the subthreshold polarization of a neuron by an applied electric field. It gives insight into how morphological features of a neuron affect its polarizability. The neuronal model consists of one or more extensively branched dendritic trees, a lumped somatic impedance, and a myelinated axon with nodes of Ranvier. The dendritic trees branch according to the 3/2-power rule of Rall, so that each tree has an equivalent cylinder representation. Equations for the membrane potential at the soma and at the nodes of Ranvier, given an arbitrary specified external potential, are derived. The solutions determine the contributions made by the dendritic tree and the axon to the net polarization at the soma. In the case of a spatially constant electric field, both the magnitude and sign of the polarization depend on simple combinations of parameters describing the neuron. One important combination is given by the ratio of internal resistances for longitudinal current spread along the dendritic tree trunk and along the axon. A second is given by the ratio between the DC space constant for the dendritic tree trunk and the distance between nodes of Ranvier in the axon. A third is given by the product of the electric field and the space constant for the trunk of the dendritic tree. When a neuron with a straight axon is subjected to a constant field, the membrane potential decays exponentially with distance from the soma. Thus, the soma seems to be a likely site for action potential initiation when the field is strong enough to elicit suprathreshold polarization. In a simple example, the way in which orientation of the various parts of the neuron affects its polarization is examined. When an axon with a bend is subjected to a spatially constant field, polarization is focused at the bend, and this is another likely site for action potential initiation.

Recovery of Sparse Translation-Invariant Signals With Continuous Basis Pursuit

We consider the problem of decomposing a signal into a linear combination of features, each a continuously translated version of one of a small set of elementary features. Although these constituents are drawn from a continuous family, most current signal decomposition methods rely on a finite dictionary of discrete examples selected from this family (e.g., shifted copies of a set of basic waveforms), and apply sparse optimization methods to select and solve for the relevant coefficients. Here, we generate a dictionary that includes auxiliary interpolation functions that approximate translates of features via adjustment of their coefficients. We formulate a constrained convex optimization problem, in which the full set of dictionary coefficients represents a linear approximation of the signal, the auxiliary coefficients are constrained so as to only represent translated features, and sparsity is imposed on the primary coefficients using an L1 penalty. The basis pursuit denoising (BP) method may be seen as a special case, in which the auxiliary interpolation functions are omitted, and we thus refer to our methodology as continuous basis pursuit (CBP). We develop two implementations of CBP for a one-dimensional translation-invariant source, one using a first-order Taylor approximation, and another using a form of trigonometric spline. We examine the tradeoff between sparsity and signal reconstruction accuracy in these methods, demonstrating empirically that trigonometric CBP substantially outperforms Taylor CBP, which, in turn, offers substantial gains over ordinary BP. In addition, the CBP bases can generally achieve equally good or better approximations with much coarser sampling than BP, leading to a reduction in dictionary dimensionality.

Hypermutation in Shark Immunoglobulin Light Chain Genes Results in Contiguous Substitutions

Among 631 substitutions present in 90 nurse shark immunoglobulin light chain somatic mutants, 338 constitute 2-4 bp stretches of adjacent changes. An absence of mutations in perinatal sequences and the bias for one mutating V gene in adults suggest that the diversification is antigen dependent. The substitutions shared no patterns, and the absence of donor sequences, including from family members, supports the idea that most changes arose from nontemplated mutation. The tandem mutations as a group are distinguished by consistently fewer transition changes and an A bias. We suggest this is one of several pathways of hypermutation diversifying shark antigen-receptor genes--point mutations, tandem mutations, and mutations with a G-C preference--that coevolved with or preceded gene rearrangement.

Light adaptation in turtle cones. Testing and analysis of a model for phototransduction

Light adaptation in cones was characterized by measuring the changes in temporal frequency responses to sinusoidal modulation of light around various mean levels spanning a range of four log units. We have shown previously that some aspects of cone adaptation behavior can be accounted for by a biochemical kinetic model for phototransduction in which adaptation is mediated largely by a sigmoidal dependence of guanylate cyclase activity on the concentration of free cytoplasmic Ca2+, ([Ca2+]i) (Sneyd and Tranchina, 1989). Here we extend the model by incorporating electrogenic Na+/K+ exchange, and the model is put to further tests by simulating experiments in the literature. It accounts for (a) speeding up of the impulse response, transition from monophasic to biphasic waveform, and improvement in contrast sensitivity with increasing background light level, I0; (b) linearity of the response to moderate modulations around I0; (c) shift of the intensity-response function (linear vs. log coordinates) with change in I0 (Normann and Perlman, 1979); the dark-adapted curve adheres closely to the Naka-Rushton equation; (d) steepening of the sensitivity vs. I0 function with [Ca2+]i fixed at its dark level, [Ca2+]i dark; (Matthews et al., 1988, 1990); (e) steepening of the steady-state intensity-response function when [Ca2+]i is held fixed at its dark level (Matthews et al., 1988; 1990); (f) shifting of a steep template saturation curve for normalized photocurrent vs. light-step intensity when the response is measured at fixed times and [Ca2+]i is held fixed at [Ca2+]i dark (Nakatani and Yau, 1988). Furthermore, the predicted dependence of guanylate cyclase activity on [Ca2+] closely matches a cooperative inhibition equation suggested by the experimental results of Koch and Stryer (1988) on cyclase activity in bovine rods. Finally, the model predicts that some changes in response kinetics with background light will still be present, even when [Ca2+]i is held fixed at [Ca2]i dark.

Phototransduction in cones: An inverse problem in enzyme kinetics

Phototransduction is a process which links the absorption of photons by a rod or cone to the modulation of voltage across the cell membrane. An important feature of many vertebrate photoreceptors is a mechanism that adjusts the sensitivity and dynamics of the response to light according to the level of illumination.We construct a system of ordinary differential equations that models what are currently thought to be the important molecule mechanisms involved in phototransduction: this includes consideration of both intracellular enzyme kinetics and the properties of light-insensitive and light-sensitive conductances in the cone membrane. The system contains negative feedback whose functional form is determined by constraining the steady-state behaviour of the system. Despite the highly nonlinear nature of the system of ordinary differential equations, our methods permit us to derive an analytic expression for the first-order frequency response parametric in the steady-state value of only one dynamic variable, the light input. Various unknown kinetic parameters are found by fitting the model to experimental data on the first-order frequency response of cones measured at several mean light levels spanning a range of four log units. Good fits are obtained to the data, and the computed shape of the feedback function agrees qualitatively with recent experiment. Moreover, the model accounts for the dramatic speeding up of the response kinetics and the decrease in response gain with increasing light level.

Multiple steps of phosphorylation of activated rhodopsin can account for the reproducibility of vertebrate rod Single-photon responses

Single-photon responses (SPRs) in vertebrate rods are considerably less variable than expected if isomerized rhodopsin (R*) inactivated in a single, memoryless step, and no other variability-reducing mechanisms were available. We present a new stochastic model, the core of which is the successive ratcheting down of R* activity, and a concomitant increase in the probability of quenching of R* by arrestin (Arr), with each phosphorylation of R* (Gibson, S.K., J.H. Parkes, and P.A. Liebman. 2000. Biochemistry. 39:5738–5749.). We evaluated the model by means of Monte-Carlo simulations of dim-flash responses, and compared the response statistics derived from them with those obtained from empirical dim-flash data (Whitlock, G.G., and T.D. Lamb. 1999. Neuron. 23:337–351.). The model accounts for four quantitative measures of SPR reproducibility. It also reproduces qualitative features of rod responses obtained with altered nucleotide levels, and thus contradicts the conclusion that such responses imply that phosphorylation cannot dominate R* inactivation (Rieke, F., and D.A. Baylor. 1998a. Biophys. J. 75:1836–1857; Field, G.D., and F. Rieke. 2002. Neuron. 35:733–747.). Moreover, the model is able to reproduce the salient qualitative features of SPRs obtained from mouse rods that had been genetically modified with specific pathways of R* inactivation or Ca2+ feedback disabled. We present a theoretical analysis showing that the variability of the area under the SPR estimates the variability of integrated R* activity, and can provide a valid gauge of the number of R* inactivation steps. We show that there is a heretofore unappreciated tradeoff between variability of SPR amplitude and SPR duration that depends critically on the kinetics of inactivation of R* relative to the net kinetics of the downstream reactions in the cascade. Because of this dependence, neither the variability of SPR amplitude nor duration provides a reliable estimate of the underlying variability of integrated R* activity, and cannot be used to estimate the minimum number of R* inactivation steps. We conclude that multiple phosphorylation-dependent decrements in R* activity (with Arr-quench) can confer the observed reproducibility of rod SPRs; there is no compelling need to invoke a long series of non-phosphorylation dependent state changes in R* (as in Rieke, F., and D.A. Baylor. 1998a. Biophys. J. 75:1836–1857; Field, G.D., and F. Rieke. 2002. Neuron. 35:733–747.). Our analyses, plus data and modeling of others (Rieke, F., and D.A. Baylor. 1998a. Biophys. J. 75:1836–1857; Field, G.D., and F. Rieke. 2002. Neuron. 35:733–747.), also argue strongly against either feedback (including Ca2+-feedback) or depletion of any molecular species downstream to R* as the dominant cause of SPR reproducibility.

Toward a unified model of vertebrate rod phototransduction.

Recently, we introduced a phototransduction model that was able to account for the reproducibility of vertebrate rod single-photon responses (SPRs) (Hamer et al., 2003). The model was able to reproduce SPR statistics by means of stochastic activation and inactivation of rhodopsin (R*), transducin (G alpha ), and phosphodiesterase (PDE). The features needed to capture the SPR statistics were (1) multiple steps of R* inactivation by means of multiple phosphorylations (followed by arrestin capping) and (2) phosphorylation dependence of the affinity between R* and the three molecules competing to bind with R* (G alpha, arrestin, and rhodopsin kinase). The model was also able to account for several other rod response features in the dim-flash regime, including SPRs obtained from rods in which various elements of the cascade have been genetically disabled or disrupted. However, the model was not tested under high light-level conditions. We sought to evaluate the extent to which the multiple phosphorylation model could simultaneously account for single-photon response behavior, as well as responses to high light levels causing complete response saturation and/or significant light adaptation (LA). To date no single model, with one set of parameters, has been able to do this. Dim-flash responses and statistics were simulated using a hybrid stochastic/deterministic model and Monte-Carlo methods as in Hamer et al. (2003). A dark-adapted flash series, and stimulus paradigms from the literature eliciting various degrees of light adaptation (LA), were simulated using a full differential equation version of the model that included the addition of Ca2+-feedback onto rhodopsin kinase via recoverin. With this model, using a single set of parameters, we attempted to account for (1) SPR waveforms and statistics (as in Hamer et al., 2003); (2) a full dark-adapted flash-response series, from dim flash to saturating, bright flash levels, from a toad rod; (3) steady-state LA responses, including LA circulating current (as in Koutalos et al., 1995) and LA flash sensitivity measured in rods from four species; (4) step responses from newt rods ( Forti et al., 1989) over a large dynamic range; (5) dynamic LA responses, such as the step-flash paradigm of Fain et al. (1989), and the two-flash paradigm of Murnick and Lamb (1996); and (6) the salient response features from four knockout rod preparations. The model was able to meet this stringent test, accounting for almost all the salient qualitative, and many quantitative features, of the responses across this broad array of stimulus conditions, including SPR reproducibility. The model promises to be useful in testing hypotheses regarding both normal and abnormal photoreceptor function, and is a good starting point for development of a full-range model of cone phototransduction. Informative limitations of the model are also discussed.

A comparison of release kinetics and glutamate receptor properties in shaping rod–cone differences in EPSC kinetics in the salamander retina

Synaptic transmission from cones is faster than transmission from rods. Using paired simultaneous recordings from photoreceptors and second‐order neurones in the salamander retina, we studied the contributions of rod–cone differences in glutamate receptor properties and synaptic release rates to shaping postsynaptic responses. Depolarizing steps evoked sustained calcium currents in rods and cones that in turn produced transient excitatory postsynaptic currents (EPSCs) in horizontal and OFF bipolar cells. Cone‐driven EPSCs rose and decayed faster than rod‐driven EPSCs, even when comparing inputs from a rod and cone onto the same postsynaptic neurone. Thus, rod–cone differences in EPSCs reflect properties of individual rod and cone synapses. Experiments with selective AMPA and KA agonists and antagonists showed that rods and cones both contact pharmacologically similar AMPA receptors. Spontaneous miniature EPSCs (mEPSCs) exhibited unimodal distributions of amplitude and half‐amplitude time width and there were no rod–cone differences in mEPSC properties. To examine how release kinetics shape the EPSC, we convolved mEPSC waveforms with empirically determined release rate functions for rods and cones. The predicted EPSC waveform closely matched the actual EPSC evoked by cones, supporting a quantal release model at the photoreceptor synapse. Convolution with the rod release function also produced a good match in rod‐driven cells, although the actual EPSC was often somewhat slower than the predicted EPSC, a discrepancy partly explained by rod–rod coupling. Rod–cone differences in the rates of exocytosis are thus a major factor in producing faster cone‐driven responses in second‐order retinal neurones.