V1 Complex Research Articles

Structure of yeast RAVE bound to a partial V1 complex

Vacuolar-type ATPases (V-ATPases) are membrane-embedded proton pumps that acidify intracellular compartments in almost all eukaryotic cells. Homologous with ATP synthases, these multisubunit enzymes consist of a soluble catalytic V1 subcomplex and a membrane-embedded proton-translocating VO subcomplex. The V1 and VO subcomplexes can undergo reversible dissociation to regulate proton pumping, with reassociation of V1 and VO requiring the protein complex known as RAVE (regulator of the ATPase of vacuoles and endosomes). In the yeast Saccharomyces cerevisiae, RAVE consists of subunits Rav1p, Rav2p, and Skp1p. We used electron cryomicroscopy (cryo-EM) to determine a structure of yeast RAVE bound to V1. In the structure, RAVE is an L-shaped complex with Rav2p pointing toward the membrane and Skp1p distant from both the membrane and V1. Only Rav1p interacts with V1, binding to a region of subunit A not found in the corresponding ATP synthase subunit. When bound to RAVE, V1 is in a rotational state suitable for binding the free VO complex, but in the structure, it is partially disrupted, missing five of its 16 subunits. Other than these missing subunits and the conformation of the inhibitory subunit H, the V1 complex with RAVE appears poised for reassembly with VO.

High-resolution electron cryomicroscopy of V-ATPase in native synaptic vesicles.

Intercellular communication in the nervous system occurs through the release of neurotransmitters into the synaptic cleft between neurons. In the presynaptic neuron, the proton pumping vesicular- or vacuolar-type ATPase (V-ATPase) powers neurotransmitter loading into synaptic vesicles (SVs), with the V1 complex dissociating from the membrane region of the enzyme before exocytosis. We isolated SVs from rat brain using SidK, a V-ATPase-binding bacterial effector protein. Single-particle electron cryomicroscopy allowed high-resolution structure determination of V-ATPase within the native SV membrane. In the structure, regularly spaced cholesterol molecules decorate the enzyme's rotor and the abundant SV protein synaptophysin binds the complex stoichiometrically. ATP hydrolysis during vesicle loading results in a loss of the V1 region of V-ATPase from the SV membrane, suggesting that loading is sufficient to induce dissociation of the enzyme.

Eukaryotic yeast V1-ATPase rotary mechanism insights revealed by high-resolution single-molecule studies.

Vacuolar ATP-dependent proton pumps (V-ATPases) belong to a super-family of rotary ATPases and ATP synthases. The V1 complex consumes ATP to drive rotation of a central rotor that pumps protons across membranes via the Vo complex. Eukaryotic V-ATPases are regulated by reversible disassembly of subunit C, V1 without C, and VO. ATP hydrolysis is thought to generate an unknown rotary state that initiates regulated disassembly. Dissociated V1 is inhibited by subunit H that traps it in a specific rotational position. Here, we report the first single-molecule studies with high resolution of time and rotational position of Saccharomyces cerevisiae V1-ATPase lacking subunits H and C (V1ΔHC), which resolves previously elusive dwells and angular velocity changes. Rotation occurred in 120° power strokes separated by dwells comparable to catalytic dwells observed in other rotary ATPases. However, unique V1ΔHC rotational features included: 1) faltering power stroke rotation during the first 60°; 2) a dwell often occurring ∼45° after the catalytic dwell, which did not increase in duration at limiting MgATP; 3) a second dwell, ∼2-fold longer occurring 112° that increased in duration and occurrence at limiting MgATP; 4) limiting MgATP-dependent decreases in power stroke angular velocity where dwells were not observed. The results presented here are consistent with MgATP binding to the empty catalytic site at 112° and MgADP released at ∼45°, and provide important new insight concerning the molecular basis for the differences in rotary positions of substrate binding and product release between V-type and F-type ATPases.

Color and luminance processing in V1 complex cells and artificial neural networks.

Object recognition by natural and artificial visual systems benefits from the identification of object boundaries. A useful cue for the detection of object boundaries is the superposition of luminance and color edges. To gain insight into the suitability of this cue for object recognition, we examined convolutional neural network models that had been trained to recognize objects in natural images. We focused specifically on units in the second convolutional layer whose activations are invariant to the spatial phase of a sinusoidal grating. Some of these units were tuned for a nonlinear combination of color and luminance, which is broadly consistent with a role in object boundary detection. Others were tuned for luminance alone, but very few were tuned for color alone. A literature review reveals that V1 complex cells have a similar distribution of tuning. We speculate that this pattern of sensitivity provides an efficient basis for object recognition, perhaps by mitigating the effects of lighting on luminance contrast polarity. The absence of a contrast polarity-invariant representation of chromaticity alone suggests that it is redundant with other representations.

Structural and functional understanding of disease-associated mutations in V-ATPase subunit a1 and other isoforms.

The vacuolar-type ATPase (V-ATPase) is a multisubunit protein composed of the cytosolic adenosine triphosphate (ATP) hydrolysis catalyzing V1 complex, and the integral membrane complex, Vo, responsible for proton translocation. The largest subunit of the Vo complex, subunit a, enables proton translocation upon ATP hydrolysis, mediated by the cytosolic V1 complex. Four known subunit a isoforms (a1-a4) are expressed in different cellular locations. Subunit a1 (also known as Voa1), the neural isoform, is strongly expressed in neurons and is encoded by the ATP6V0A1 gene. Global knockout of this gene in mice causes embryonic lethality, whereas pyramidal neuron-specific knockout resulted in neuronal cell death with impaired spatial and learning memory. Recently reported, de novo and biallelic mutations of the human ATP6V0A1 impair autophagic and lysosomal activities, contributing to neuronal cell death in developmental and epileptic encephalopathies (DEE) and early onset progressive myoclonus epilepsy (PME). The de novo heterozygous R740Q mutation is the most recurrent variant reported in cases of DEE. Homology studies suggest R740 deprotonates protons from specific glutamic acid residues in subunit c, highlighting its importance to the overall V-ATPase function. In this paper, we discuss the structure and mechanism of the V-ATPase, emphasizing how mutations in subunit a1 can lead to lysosomal and autophagic dysfunction in neurodevelopmental disorders, and how mutations to the non-neural isoforms, a2-a4, can also lead to various genetic diseases. Given the growing discovery of disease-causing variants of V-ATPase subunit a and its function as a pump-based regulator of intracellular organelle pH, this multiprotein complex warrants further investigation.

The interdependent transport of yeast vacuole Ca2+ and H+ and the role of phosphatidylinositol 3,5-bisphosphate

Yeast vacuoles are acidified by the v-type H+-ATPase (V-ATPase) that is comprised of the membrane embedded VO complex and the soluble cytoplasmic V1 complex. The assembly of the V1-VO holoenzyme on the vacuole is stabilized in part through interactions between the VO a-subunit ortholog Vph1 and the lipid phosphatidylinositol 3,5-bisphosphate (PI(3,5)P2). PI(3,5)P2 also affects vacuolar Ca2+ release through the channel Yvc1 and uptake through the Ca2+ pump Pmc1. Here, we asked if H+ and Ca2+ transport activities were connected through PI(3,5)P2. We found that overproduction of PI(3,5)P2 by the hyperactive fab1T2250A mutant augmented vacuole acidification, whereas the kinase-inactive fab1EEE mutant attenuated the formation of a H+ gradient. Separately, we tested the effects of excess Ca2+ on vacuole acidification. Adding micromolar Ca2+ blocked vacuole acidification, whereas chelating Ca2+ accelerated acidification. The effect of adding Ca2+ on acidification was eliminated when the Ca2+/H+ antiporter Vcx1 was absent, indicating that the vacuolar H+ gradient can collapse during Ca2+ stress through Vcx1 activity. This, however, was independent of PI(3,5)P2, suggesting that PI(3,5)P2 plays a role in submicromolar Ca2+ flux but not under Ca2+ shock. To see if the link between Ca2+ and H+ transport was bidirectional, we examined Ca2+ transport when vacuole acidification was inhibited. We found that Ca2+ transport was inhibited by halting V-ATPase activity with Bafilomycin or neutralizing vacuolar pH with chloroquine. Together, these data show that Ca2+ transport and V-ATPase efficacy are connected but not necessarily through PI(3,5)P2.

Coordinated conformational changes in the V1 complex during V-ATPase reversible dissociation

Vacuolar-type ATPases (V-ATPases) are rotary enzymes that acidify intracellular compartments in eukaryotic cells. These multi-subunit complexes consist of a cytoplasmic V1 region that hydrolyzes ATP and a membrane-embedded VO region that transports protons. V-ATPase activity is regulated by reversible dissociation of the two regions, with the isolated V1 and VO complexes becoming autoinhibited on disassembly and subunit C subsequently detaching from V1. In yeast, assembly of the V1 and VO regions is mediated by the regulator of the ATPase of vacuoles and endosomes (RAVE) complex through an unknown mechanism. We used cryogenic-electron microscopy of yeast V-ATPase to determine structures of the intact enzyme, the dissociated but complete V1 complex and the V1 complex lacking subunit C. On separation, V1 undergoes a dramatic conformational rearrangement, with its rotational state becoming incompatible for reassembly with VO. Loss of subunit C allows V1 to match the rotational state of VO, suggesting how RAVE could reassemble V1 and VO by recruiting subunit C.

Coding of chromatic spatial contrast by macaque V1 neurons.

Color perception relies on comparisons between adjacent lights, but how the brain performs these comparisons is poorly understood. To elucidate the underlying mechanisms, we recorded spiking responses of individual V1 neurons in macaque monkeys to pairs of stimuli within the classical receptive field (RF). We estimated the spatial-chromatic RF of each neuron and then presented customized colored edges using a closed-loop technique. We found that many double-opponent (DO) cells, which have spatially and chromatically opponent RFs, responded to chromatic contrast as a weighted sum, akin to how other V1 neurons responded to luminance contrast. Yet other neurons integrated chromatic signals nonlinearly, confirming that linear signal integration is not an obligate property of V1 neurons. The functional similarity of cone-opponent DO cells and cone non-opponent simple cells suggests that these two groups may share a common underlying circuitry, promotes the construction of image-computable models for full-color image representation, and sheds new light on V1 complex cells.

V-ATPase subunit a is required for survival and midgut development of Locusta migratoria.

The vacuolar-type H+ -ATPase (V-ATPase) is an ATP-dependent proton pump, which regulates various cellular processes. To date, most functional studies on V-ATPases of insects have focused on subunits of the V1 complex, and there is little information on the VO genes. In this study, two cDNA sequences of LmV-ATPase a were identified in Locusta migratoria. RT-qPCR analysis revealed that LmV-ATPase a1 and LmV-ATPase a2 are differentially expressed in various tissues and developmental stages. Injection of dsRNA for the common region of LmV-ATPase a1 and LmV-ATPase a2 into third-instar nymphs resulted in a significant suppression of LmV-ATPase a. The injected nymphs ceased feeding, lost body weight and finally died at a mortality of 98.6%. Furthermore, aberrations of midgut epithelial cells, the accumulation of electron-lucent vesicles in the cytoplasm, and a partially damaged brush border were observed in dsLmV-ATPase a-injected nymphs using transmission electron microscopy. Especially, the mRNA level of wingles, and notch genes were dramatically down-regulated in the dsLmV-ATPase a-injected nymphs. Taken together, our results suggest that LmV-ATPase a is required for survival and midgut development of L. migratoria. Hence, this gene could be a good target for RNAi-based control against locusts.

Learning receptive field properties of complex cells in V1.

There are two distinct classes of cells in the primary visual cortex (V1): simple cells and complex cells. One defining feature of complex cells is their spatial phase invariance; they respond strongly to oriented grating stimuli with a preferred orientation but with a wide range of spatial phases. A classical model of complete spatial phase invariance in complex cells is the energy model, in which the responses are the sum of the squared outputs of two linear spatially phase-shifted filters. However, recent experimental studies have shown that complex cells have a diverse range of spatial phase invariance and only a subset can be characterized by the energy model. While several models have been proposed to explain how complex cells could learn to be selective to orientation but invariant to spatial phase, most existing models overlook many biologically important details. We propose a biologically plausible model for complex cells that learns to pool inputs from simple cells based on the presentation of natural scene stimuli. The model is a three-layer network with rate-based neurons that describes the activities of LGN cells (layer 1), V1 simple cells (layer 2), and V1 complex cells (layer 3). The first two layers implement a recently proposed simple cell model that is biologically plausible and accounts for many experimental phenomena. The neural dynamics of the complex cells is modeled as the integration of simple cells inputs along with response normalization. Connections between LGN and simple cells are learned using Hebbian and anti-Hebbian plasticity. Connections between simple and complex cells are learned using a modified version of the Bienenstock, Cooper, and Munro (BCM) rule. Our results demonstrate that the learning rule can describe a diversity of complex cells, similar to those observed experimentally.

Structures of a Complete Human V-ATPase Reveal Mechanisms of Its Assembly

Vesicular- or vacuolar-type adenosine triphosphatases (V-ATPases) are ATP-driven proton pumps comprised of a cytoplasmic V1 complex for ATP hydrolysis and a membrane-embedded Vo complex for proton transfer. They play important roles in acidification of intracellular vesicles, organelles, and the extracellular milieu in eukaryotes. Here, we report cryoelectron microscopy structures of human V-ATPase in three rotational states at up to 2.9-Å resolution. Aided by mass spectrometry, we build all known protein subunits with associated N-linked glycans and identify glycolipids and phospholipids in the Vo complex. We define ATP6AP1 as a structural hub for Vo complex assembly because it connects to multiple Vo subunits and phospholipids in the c-ring. The glycolipids and the glycosylated Vo subunits form a luminal glycan coat critical for V-ATPase folding, localization, and stability. This study identifies mechanisms of V-ATPase assembly and biogenesis that rely on the integrated roles of ATP6AP1, glycans, and lipids.

Unsupervised experience with temporal continuity of the visual environment is causally involved in the development of V1 complex cells.

Unsupervised adaptation to the spatiotemporal statistics of visual experience is a key computational principle that has long been assumed to govern postnatal development of visual cortical tuning, including orientation selectivity of simple cells and position tolerance of complex cells in primary visual cortex (V1). Yet, causal empirical evidence supporting this hypothesis is scant. Here, we show that degrading the temporal continuity of visual experience during early postnatal life leads to a sizable reduction of the number of complex cells and to an impairment of their functional properties while fully sparing the development of simple cells. This causally implicates adaptation to the temporal structure of the visual input in the development of transformation tolerance but not of shape tuning, thus tightly constraining computational models of unsupervised cortical learning.

Heat shock factor 4 regulates lysosome activity by modulating the αB-crystallin-ATP6V1A-mTOR complex in ocular lens

BackgroundGermline mutations in heat shock factor 4 (HSF4) cause congenital cataracts. Previously, we have shown that HSF4 is involved in regulating lysosomal pH in mouse lens epithelial cell in vitro. However, the underlying mechanism remains unclear. MethodsHSF4-deficient mouse lens epithelial cell lines and zebrafish were used in this study. Immunoblotting and quantitative RT-PCR were used for expression analysis. The protein-protein interactions were tested with GST-pull downs. The lysosomes were fractioned by ultracentrifugation. ResultsHSF4 deficiency or knock down of αB-crystallin elevates lysosomal pH and increases the ubiquitination and degradation of ATP6V1A by the proteasome. αB-crystallin localizes partially in the lysosome and interacts solely with the ATP6V1A protein of the V1 complex of V-ATPase. Furthermore, αB-crystallin can co-precipitate with mTORC1 and ATP6V1A in GST pull down assays. Inhibition of mTORC1 by rapamycin or siRNA can lead to dissociation of αB-crystallin from the ATP6V1A and mTORC1complex, shortening the half-life of ATP6V1A and increasing the lysosomal pH. Mutation of ATP6V1A/S441A (the predicted mTOR phosphorylation site) reduces its association with αB-crystallin. In the zebrafish model, HSF4 deficiency reduces αB-crystallin expression and elevates the lysosomal pH in lens tissues. ConclusionHSF4 regulates lysosomal acidification by controlling the association of αB-crystallin with ATP6V1A and mTOR and regulating ATP6V1A protein stabilization. General significanceThis study uncovers a novel function of αB-crystallin, demonstrating that αB-crystallin can regulate lysosomal ATP6V1A protein stabilization by complexing to ATP6V1A and mTOR. This highlights a novel mechanism by which HSF4 regulates the proteolytic process of organelles during lens development.

Contrast-dependent phase sensitivity in area MT of macaque visual cortex.

In primate visual cortex (V1), about half the neurons are sensitive to the spatial phases of grating stimuli and generate highly modulated responses to drifting gratings (simple cells). The remaining cells show far less phase sensitivity and relatively unmodulated responses to moving gratings (complex cells). In the second visual area (V2) and the motion processing area MT (or V5), the majority of cells have unmodulated responses to drifting gratings - they are phase invariant. At just-detectable contrasts, 44% of V1 complex cells show highly modulated responses, but this contrast-dependent phase sensitivity is found in only 7% of V2 complex cells. We recorded from 149 cells in macaque MT - 142 classed as complex cells at high contrast. Approximately 14% (20/142) of MT complex cells showed significantly modulated responses to drifting gratings at just-detectable contrasts. A general feature of MT cells is that they can be divided into pattern and component selective types, but we found no correlation between this classification and contrast-dependent phase sensitivity. Phase sensitivity in MT is discussed in relation to MT's input structure.

A pH-sensitive luminal His-cluster promotes interaction of PAM with V-ATPase along the secretory and endocytic pathways of peptidergic cells.

The biosynthetic and endocytic pathways of secretory cells are characterized by progressive luminal acidification, a process which is crucial for posttranslational modifications and membrane trafficking. This progressive fall in luminal pH is mainly achieved by the vacuolar-type-H+ ATPase (V-ATPase). V-ATPases are large, evolutionarily ancient rotary proton pumps that consist of a peripheral V1 complex, which hydrolyzes ATP, and an integral membrane V0 complex, which transports protons from the cytosol into the lumen. Upon sensing the desired luminal pH, V-ATPase activity is regulated by reversible dissociation of the complex into its V1 and V0 components. Molecular details of how intraluminal pH is sensed and transmitted to the cytosol are not fully understood. Peptidylglycine α-amidating mono-oxygenase (PAM; EC 1.14.17.3), a secretory pathway membrane enzyme which shares similar topology with two V-ATPase accessory proteins (Ac45 and prorenin receptor), has a pH-sensitive luminal linker region. Immunofluorescence and sucrose gradient analysis of peptidergic cells (AtT-20) identified distinct subcellular compartments exhibiting spatial co-occurrence of PAM and V-ATPase. In vitro binding assays demonstrated direct binding of the cytosolic domain of PAM to V1H. Blue native PAGE identified heterogeneous high-molecular weight complexes of PAM and V-ATPase. A PAM-1 mutant (PAM-1/H3A) with altered pH sensitivity had diminished ability to form high-molecular weight complexes. In addition, V-ATPase assembly status was altered in PAM-1/H3A expressing cells. Our analysis of the secretory and endocytic pathways of peptidergic cells supports the hypothesis that PAM serves as a luminal pH-sensor, regulating V-ATPase action by altering its assembly status.

A biologically-based computational model of visual cortex that overcomes the X-junction illusion

The end-points of a moving bar (intrinsic terminators) contain unambiguous information that can be used to extract the bar’s correct direction of motion, regardless of the orientation of the bar. However, extrinsic terminators, formed at the intersection of two overlapping bars, can result in motion signals with conflicting directions compared to those of the intrinsic terminators. Using a computational model, we propose that interactions between form and motion information may assist neurons in the motion-specific regions of primate cortex to differentiate intrinsic from extrinsic terminators. The motion processing model has two stages. The first stage is a model of V1 complex neurons, including end-stopped neurons. The resulting first stage motion signals are transmitted to the second stage, which is a model of MT neurons. In the proposed model, MT neurons additionally receive form information from neurons in V1 that are not orientation or direction selective but respond strongly to the contrast of the stimulus. These neurons have polarity-dependent center–surround receptive fields, as found in layer 4 of V1 in primates. As the inhibitory surrounds of these neurons are less activated at the intrinsic terminators, the signals generated by the end-points of the objects are stronger than the signals from the extrinsic terminators, which are inhibited by strong suppression from the surround. Therefore, the excitatory inputs received by integration MT neurons from center–surround V1 neurons enhance the unambiguous motion signals at the intrinsic terminators, which therefore dominate over the local motion signals generated at X-junctions. The results show that, despite the inability of V1 end-stopped neurons to distinguish between the two different types of terminators, center–surround V1 neurons provide the capacity for the second stage of the model to preferentially respond to the intrinsic terminators and, therefore, predict the true directions of the crossing bars.

Functional Analyses of V-Atpase Mutations in Follicular Lymphoma

Introduction: Follicular lymphoma (FL) constitutes the second most common non-Hodgkin's lymphoma in the Western world. Recently, frequent mutations (RRAGC, V-ATPase) in the amino acid sensing pathway connected to MTOR signaling have been described in FL. The RRAGC mutations activate MTOR. The V-ATPase is a multi-protein complex primarily involved in proton pumping and acidification of cellular compartments. Here, we report initial results from functional analyses of the common hotspot mutations p.Y371C and p.R400Q in the V-ATPase subunit ATP6V1B2.Methods: We introduced the ATP6V1B2 hotspot mutations by site-directed mutagenesis. The cDNAs were cloned into lentiviral or retroviral vectors and used to generate stable lymphoma cells lines and HEK293T cells. We also generated recombinant S. cerevisiae expressing a mutation homologous to human ATP6V1B2 p.R400Q at the endogenous VMA2locus (R381Q). We modeled the location of the mutations onto the existing 3D structures.Results: Some of the stable cell lines lost the expression of the ATP6V1B2 mutants over time, while others were more permissive. Given the critical role of V-ATPase in lysosomal acidification and autophagy we measured LC3-II levels in various recombinant cell lines and detected very elevated LC3-II levels in the V-ATPase mutant lines. Highly elevated LC3-II levels at steady state could indicate impairments in LC3-II metabolism as seen with chemical V-ATPase inhibition (bafilomycin A1 treatment) or elevated autophagic initiation and flux. We resorted to multiple experimental approaches to resolve this question: i) chemical MTOR inhibition strongly activates autophagy. Using rapamycin and Torin 2, we induced LC3-II to levels comparable to those seen with the V-ATPase mutants; ii) using inhibitors of lysosomal LC3-II degradation, we further increased already elevated LC3-II levels in V-ATPase mutant HEK293T cell lines; iii) in S. cerevisiae Vma2R381Qcells we found elevated autophagy activity based on analysis of the LC3 homolog Atg8, vacuolar/lysosomal processing of a GFP-Atg8 chimera, and enzymatic analysis of an autophagy marker, Pho8Δ60. Together, these findings currently allow for the working hypothesis that V-ATPase mutants strongly increase autophagic flux.We performed studies of the effects of V-ATPase mutants on MTOR activity as measured through RPS6KB/S6-kinase phosphorylation. In transient transfections in HEK293T cells and in stable HEK293T cells lines achieving expression slightly above endogenous V-ATPase levels, performed under conditions of leucine starvation or sufficiency, we detected MTOR activation.We mapped the mutated p.Y371C and p.R400Q residues onto the published electron microscopy-derived structures of yeast V-ATPase. We found that both residues are closely spaced and are located at the interface of the V1B2 and V1A subunits; the latter with potential effects on subunit interactions, the conformational state of the V1 complex (open, loose or tight) and possibly effects on ATPase activity.To gain insights into the physiological consequences of these findings, we measured growth and viability of a recombinant lymphoma cell line (SUDHL4) expressing wild type and mutant V-ATPase in the presence of lowered leucine concentration. We found improved growth and survival of cells expressing the ATP6V1B2 mutations. In a parallel experiment, we induced cellular stress by intentionally overgrowing these lines and measured survival over time. We found improved growth and survival of cells expressing the ATP6V1B2 mutations. Complementary measurements of the biochemical and functional consequences of V-ATPase mutations in isolated primary FL B cells are ongoing and will be updated at the meeting.Conclusion: Our initial efforts at functional characterization of the common FL-associated hotspot mutations p.Y371C and p.R400Q in the V-ATPase subunit ATP6V1B2 uncovered substantially elevated autophagic flux. To our knowledge, this is the first report of mutational activation of autophagy in follicular lymphoma. Our current working model is that this elevated flux promotes lysosomal amino acid generation that subsequently activates MTOR through previously proposed inside-out signaling pathways. Combined, these changes may allow for improved survival of FL cells under conditions of nutrient stress. These findings may allow for future therapeutic targeting of this pathway in FL. DisclosuresNo relevant conflicts of interest to declare.

Learning Visual Spatial Pooling by Strong PCA Dimension Reduction.

In visual modeling, invariance properties of visual cells are often explained by a pooling mechanism, in which outputs of neurons with similar selectivities to some stimulus parameters are integrated so as to gain some extent of invariance to other parameters. For example, the classical energy model of phase-invariant V1 complex cells pools model simple cells preferring similar orientation but different phases. Prior studies, such as independent subspace analysis, have shown that phase-invariance properties of V1 complex cells can be learned from spatial statistics of natural inputs. However, those previous approaches assumed a squaring nonlinearity on the neural outputs to capture energy correlation; such nonlinearity is arguably unnatural from a neurobiological viewpoint but hard to change due to its tight integration into their formalisms. Moreover, they used somewhat complicated objective functions requiring expensive computations for optimization. In this study, we show that visual spatial pooling can be learned in a much simpler way using strong dimension reduction based on principal component analysis. This approach learns to ignore a large part of detailed spatial structure of the input and thereby estimates a linear pooling matrix. Using this framework, we demonstrate that pooling of model V1 simple cells learned in this way, even with nonlinearities other than squaring, can reproduce standard tuning properties of V1 complex cells. For further understanding, we analyze several variants of the pooling model and argue that a reasonable pooling can generally be obtained from any kind of linear transformation that retains several of the first principal components and suppresses the remaining ones. In particular, we show how the classic Wiener filtering theory leads to one such variant.

Autonomous learning of disparity–vergence behavior through distributed coding and population reward: Basic mechanisms and real-world conditioning on a robot stereo head

A robotic system implementation that exhibits autonomous learning capabilities of effective control for vergence eye movements is presented. The system, directly relying on a distributed (i.e. neural) representation of binocular disparity, shows a large tolerance to the inaccuracies of real stereo heads and to the changeable environment. The proposed approach combines early binocular vision mechanisms with basic learning processes, such as synaptic plasticity and reward modulation. The computational substrate consists of a network of modeled V1 complex cells that act as oriented binocular disparity detectors. The resulting population response, besides implicit binocular depth cues about the environment, also provides a global signal (i.e. the overall activity of the population itself) to describe the state of the system and thus its deviation from the desired vergence position. The proposed network, by taking into account the modification of its internal state as a consequence of the action performed, evolves following a differential Hebbian rule. The overall activity of the population is exploited to derive an intrinsic signal that drives the weights update. Exploiting this signal implies a maximization of the population activity itself, thus providing an highly effective reward for the developing of a stable and accurate vergence behavior. The role of the different orientations in the learning process is evaluated separately against the whole population, evidencing that the interplay among the differently oriented channels allows a faster learning capability and a more accurate control. The efficacy of the proposed intrinsic reward signal is thus comparatively assessed against the ground-truth signal (the actual disparity) providing equivalent results, and thus validating the approach. Trained in a simulated environment, the proposed network, is able to cope with vergent geometry and thus to learn effective vergence movements for static and moving visual targets. Experimental tests with real robot stereo pairs demonstrate the capability of the architecture not just to directly learn from the environment, but to adapt the control to the stimulus characteristics.

Primate area V1: largest response gain for receptive fields in the straight-ahead direction.

Although neuronal responses in behaving monkeys are typically studied while the monkey fixates straight ahead, it is known that eye position modulates responses of visual neurons. The modulation has been found to enhance neuronal responses when the receptive field is placed in the straight-ahead position for neurons receiving input from the peripheral but not the central retina. We studied the effect of eye position on the responses of V1 complex cells receiving input from the central retina (1.1-5.7° eccentricity) while minimizing the effect of fixational eye movements. Contrast response functions were obtained separately with drifting light and dark bars. Data were fit with the Naka-Rushton equation: r(c)=Rmax×c/(c+c50)+s, where r(c) is mean spike rate at contrast c, Rmax is the maximum response, c50 is the contrast that elicits half of Rmax, and s is the spontaneous activity. Contrast sensitivity as measured by c50 was not affected by eye position. For dark bars, there was a statistically significant decline in the normalized Rmax with increasing deviation from straight ahead. Data for bright bars showed a similar trend with a less rapid decline. Our results indicate that neurons representing the central retina show a bias for the straight-ahead position resulting from modulation of the response gain without an accompanying modulation of contrast sensitivity. The modulation is especially obvious for dark stimuli, which might be useful for directing attention to hazardous situations such as dark holes or shadows concealing important objects (Supplement 1: Video Abstract, Supplemental digital content 1, http://links.lww.com/WNR/A295).