Three-dimensional structure of favin: saccharide binding-cyclic permutation in leguminous lectins.

The three-dimensional structure of favin, the glucose- and mannose-binding lectin of Vicia faba (vetch, broad bean), has been determined at a resolution of 2.8 angstroms by molecular replacement. The crystals contain specifically bound glucose and provide the first high-resolution view of specific saccharide binding in a leguminous lectin. The structure is similar to those of concanavalin A (Con A) and green pea lectin; differences from Con A show that minimal changes are needed to accommodate the cyclic permutation in amino acid sequence between the two molecules. The molecule is an ellipsoidal dimer dominated by extensive beta structures. Each protomer contains binding sites for two divalent metal ions (Mn2+ and Ca2+) and a specific saccharide. Glucose is bound by favin in a cleft in the molecular surface and has noncovalent contacts primarily with two peptide loops, one of which contains several metal ion ligands. The specific carbohydrate-binding site is similar to that of Con A in location and general peptide folding, despite several differences in specific amino acid residues.

Value-dependent selection in the brain: simulation in a synthetic neural model.

Many forms of learning depend on the ability of an organism to sense and react to the adaptive value of its behavior. Such value, if reflected in the activity of specific neural structures (neural value systems), can selectively increase the probability of adaptive behaviors by modulating synaptic changes in the circuits relevant to those behaviors. Neuromodulatory systems in the brain are well suited to carry out this process since they respond to evolutionarily important cues (innate value), broadcast their responses to widely distributed areas of the brain through diffuse projections, and release substances that can modulate changes in synaptic strength. The main aim of this paper is to show that, if value-dependent modulation is extended to the inputs of neural value systems themselves, initially neutral cues can acquire value. This process has important implications for the acquisition of behavioral sequences. We have used a synthetic neural model to illustrate value-dependent acquisition of a simple foveation response to a visual stimulus. We then examine the improvement that ensues when the connections to the value system are themselves plastic and thus become able to mediate acquired value. Using a second-order conditioning paradigm, we demonstrate that auditory discrimination can occur in the model in the absence of direct positive reinforcement and even in the presence of slight negative reinforcement. The discriminative responses are accompanied by value-dependent plasticity of receptive fields, as reflected in the selective augmentation of unit responses to valuable sensory cues. We then consider the time-course during learning of the responses of the value system and the transfer of these responses from one sensory modality to another. Finally, we discuss the relation of value-dependent learning to models of reinforcement learning. The results obtained from these simulations can be directly related to various reported experimental findings and provide additional support for the application of selectional principles to the analysis of brain and behavior.

Selective networks and recognition automata.

The results we have presented demonstrate that a network based on a selective principle can function in the absence of forced learning or an a priori program to give recognition, classification, generalization, and association. While Darwin II is not a model of any actual nervous system, it does set out to solve one of the same problems that evolution had to solve--the need to form categories in a bottom-up manner from information in the environment, without incorporating the assumptions of any particular observer. The key features of the model that make this possible are (1) Darwin II incorporates selective networks whose initial specificities enable them to respond without instruction to unfamiliar stimuli; (2) degeneracy provides multiple possibilities of response to any one stimulus, at the same time providing functional redundancy against component failure; (3) the output of Darwin II is a pattern of response, making use of the simultaneous responses of multiple degenerate groups to avoid the need for very high specificity and the combinatorial disaster that would imply; (4) reentry within individual networks vitiates the limitations described by Minsky and Papert for a class of perceptual automata lacking such connections; and (5) reentry between intercommunicating networks with different functions gives rise to new functions, such as association, that either one alone could not display. The two kinds of network are roughly analogous to the two kinds of category formation that people use: Darwin, corresponding to the exemplar description of categories, and Wallace, corresponding to the probabilistic matching description of categories. These principles lead to a new class of pattern-recognizing machine of which Darwin II is just an example. There are a number of obvious extensions to this work that we are pursuing. These include giving Darwin II the capability to deal with stimuli that are in motion, an ability that probably precedes the ability of biological organisms to deal with stationary stimuli, giving it the capability to deal with multiple stimulus objects through some form of attentional mechanism, and giving it a means to respond directly and to receive feedback from the world so that it can learn conventionally. Already, however, we have shown that a working pattern-recognition automaton can be built based on a selective principle. This development promises ultimately to show us how to build recognizing machines without programs and to provide a sound basis for the study of both natural and artificial intelligence.

Structure and function of concanavalin A.

Lectins have been extensively used to analyze a variety of fundamental processes in cell biology. In conjuntion with our studies on the cell surface and mitosis, we have determined the amino acid sequence and three-dimensional struction of concanavalin A (Con A), the mitogenic lectin from the jack bean. Knowledge of the structure has been helpful in interpreting experiments on lymphocyte mitogenesis and the effects of Con A on cell surface receptor mobility. Con A subunits for molecular weight 25,500 are folded into dome-like structures of maximum dimensions 42 times 40 times 39 A. The domes are related by 222 symmetry to form roughly tetrahedral tetramers. Each subunit contains two large antiparallel pleated sheets, and subunits are joined to form dimers and tetramers by interactions involving one of these pleated sheets. We have examined the binding of a variety of carbohydrates to Con A and have obtained preliminary data which suggest that there are differences in the saccharide-binding behavior of Con A in solution and in the crystalline state. Dimeric chemical derivatives of Con A have been prepared and shown to have biological activities different from those of the native tetrameric protein. Under different conditions, native Con A exhibits two antagonistic activities on the lymphoid cell surface: the induction of cap formation by its own receptors and the inhibition of the mobility of a variety of receptors, including its own receptors. The dimeric derivative, succinyl-Con A, is just as effective a mitogen as the native lectin, but it lacks the ability to modulate cell surface receptor mobility. The data suggest that neither extensive immobilization of cell surface receptors nor cap formation is required for cell stimulation. Further studies on modulation of receptor translocation suggest that hypothesis that there exists a connecting network of colchicine-sensitive proteins that links receptors of different kinds and mediates their rearrangement. The degree of connectivity of this postulated network appears to be altered by changes in the state of attachment of various surface receptors to the network. Thus the network might provide the cell with a means of transmitting signals such as the stimulus for mitosis by lectins or antigens.

Behaviorally based modeling and computational approaches to neuroscience.

The almost incredible advances that have recently occurred in the power of computers available to scientists in all disciplines have encouraged an explosion of neural network and behavioral models. Some of these have been constrained more by the imagination of the programmer than by rude biological facts. Their relevance for the experimental neuroscientist thus varies from case to case. Some models (e.g. Grillner's model of lamprey swimming movements) are so closely based on known neuroanatomy and neurophysiology that it becomes possible to generate and test precise experimental predictions. Other models (such as MURPHY and NOMAD) use neurobiological principles in their architectures, but do not portray any particular organism. Although it is harder to relate the study of these models of the study of real animals, they fulfill an important explanatory role. They make possible insights into how behavior is controlled by neuronal activity that would be unobtainable in real animals using present methods. Thus, even the excesses of neural modeling have provided a useful impetus to what is undoubtedly a most promising approach to integrating data from the various disciplines concerned with behavior and the mind. The problems have been pointed out by many authors (see citations in our introduction), and a phase of more critical evaluation appears to have begun. We hope that our brief survey of models based on widely different theoretical approaches, but all aimed at explaining behavior, will encourage critical comparisons to be made. As in more mature fields, such as thermodynamics, we can expect that more complete models will force an evaluation of theoretical hypotheses against the entire body of available evidence, rather than just a few pertinent test cases. Such evaluation will make possible a much more rigorous exclusion of invalid or inconsistent theoretical ideas. From such studies, a much smaller, but more robust, set of basic principles can be expected to emerge. From the perspective afforded by our own modeling studies, it appears essential that modeling be informed by a general theory of brain function. In this work, the theory of neuronal group selection provides a useful basis for further work by virtue of its consistency with basic evolutionary and physiological principles and the power of the selection paradigm to shape neural networks in behaviorally adaptive directions.

The composite neuron: a realistic one-compartment Purkinje cell model suitable for large-scale neuronal network simulations.

We present a simple method for the realistic description of neurons that is well suited to the development of large-scale neuronal network models where the interactions within and between neural circuits are the object of study rather than the details of dendritic signal propagation in individual cells. Referred to as the composite approach, it combines in a one-compartment model elements of both the leaky integrator cell and the conductance-based formalism of Hodgkin and Huxley (1952). Composite models treat the cell membrane as an equivalent circuit that contains ligand-gated synaptic, voltage-gated, and voltage- and concentration-dependent conductances. The time dependences of these various conductances are assumed to correlate with their spatial locations in the real cell. Thus, when viewed from the soma, ligand-gated synaptic and other dendritically located conductances can be modeled as either single alpha or double exponential functions of time, whereas, with the exception of discharge-related conductances, somatic and proximal dendritic conductances can be well approximated by simple current-voltage relationships. As an example of the composite approach to neuronal modeling we describe a composite model of a cerebellar Purkinje neuron.

Three-dimensional structure of beta 2-microglobulin.

The three-dimensional structure of beta 2-microglobulin, the light chain of the major histocompatibility complex class I antigens, has been determined by x-ray crystallography. An electron density map of the bovine protein was calculated at a nominal resolution of 2.9 A by using the methods of multiple isomorphous replacement and electron density modification refinement. The molecule is approximately 45 X 25 X 20 A in size. Almost half of the amino acid residues participate in two large beta structures, one of four strands and the other of three, linked by a central disulfide bond. The molecule thus strongly resembles Ig constant domains in polypeptide chain folding and overall tertiary structure. Amino acid residues that are the same in the sequences of beta 2-microglobulin and Ig constant domains are predominantly in the interior of the molecule, whereas residues conserved among beta 2-microglobulins from different species are both in the interior and on the molecular surface. In the crystals studied, the molecule is clearly monomeric, consistent with the observation that beta 2-microglobulin, unlike Ig constant domains, apparently does not form dimers in vivo but associates with the heavy chains of major histocompatibility complex antigens. Our results demonstrate that, at the level of detailed three-dimensional structure, the light chain of the major histocompatibility class I antigens belongs to a superfamily of structures related to the Ig constant domains.

Selective networks capable of representative transformations, limited generalizations, and associative memory.

Two parallel sets of selective networks composed of intercommunicating neuron-like elements have been connected to produce a new kind of automaton capable of limited recognition of two-dimensional patterns. Salient features of this automaton are (i) preestablished unchanging connectivity, (ii) preassigned connection strengths that are selectively altered according to experience, (iii) local feature detection in one network with simultaneous global feature correlation in the other, and (iv) reentrant interactions between the two networks to generate a new function, associative memory. No forced learning, explicit semantic rules, or a priori instructions are used.

Changes in the three-dimensional structure of concanavalin A upon demetallization.

When the Mn(2+) and Ca(2+) ions normally present in concanavalin A are removed, the protein becomes incapable of binding saccharides. To explore the structural differences between the native and demetallized forms and their effects on the saccharide-binding properties of the protein, we have refined and compared the crystal structures of both forms. Refinement, carried out by automated difference Fourier methods, has revealed a number of differences between the two structures as well as minor differences between the two crystallographically independent monomers in the demetallized structure. Significant differences between the holo- and apoproteins are confined to the region where the metals are bound. These differences include a reorganization and disordering of the loop, consisting of residues 7-25, that contains all of the direct metal ligands of the protein. In some molecules, the side chain of arginine-228 appears to move into the metal-binding region, possibly compensating in part for the absence of the metal's positive charge. The cis peptide observed in the native protein at alanine-207 is apparently not present in the demetallized protein. The conformational differences affect many of the residues currently thought to be involved in the specific binding of saccharides.

The covalent and three-dimensional structure of concanavalin A. IV. Atomic coordinates, hydrogen bonding, and quaternary structure.

The coordinates of the individual non-hydrogen atoms of the lectin concanavalin A have been determined from the molecular model at 2.0-A resolution and have been adjusted to make them consistent with the known stereochemistry of the constituent amino acid residues. From the coordinates, an analysis has been made of all intra- and intersubunit interactions in the molecule, yielding a description of the hydrogen-bonded structure of the monomer, including two extensive pleated sheet structures and other features of the folding of the polypeptide chain. The description of the noncovalent bonding is extended to include the interactions involved in stabilization of the dimeric and tetrameric structures of the molecule. The complete description of the molecular structure provides a basis for analysis of the biological activities of concanavalin A.

The structure of carboxypeptidase A. IX. The x-ray diffraction results in the light of the chemical sequence.

Several features of carboxypeptidase A (CPA) which were previously established by the X-ray diffraction structure studies have now been confirmed by the chemical sequence analysis. These results include the number (307) of amino acid residues in CPA(alpha), the identities of the residues (Arg 145, Glu 270, and Tyr 248) shown by the X-ray study to be involved in substrate binding and catalysis, and the existence of a disulfide bond. The Zn ligands, shown by the X-ray study to be residues 69, 72, and 196 and identified as His, Glx, and either Glx or Lys, are proved by the chemical sequence to be His, Glu, and His, respectively. No change is required in our previous mechanistic deductions, which are here extended to include a specific mechanism of activation of the substrate by a net charge on the metal ion, which suffers a change in local dielectric constant when it is covered by a substrate.

The NO hypothesis: possible effects of a short-lived, rapidly diffusible signal in the development and function of the nervous system.

Several observations suggest that the Ca2(+)-dependent postsynaptic release of nitric oxide (NO) may be important in the formation and function of the vertebrate nervous system. We explore here the hypothesis that the release of NO and its subsequent diffusion may be critically related to three aspects of nervous system function: (i) synaptic plasticity and long-term potentiation in certain regions of the adult nervous system, (ii) the control of cerebral blood flow in such regions, and (iii) the establishment and activity-dependent refinement of axonal projections during the later stages of development. In this paper, we detail and analyze the basic assumptions underlying this NO hypothesis and describe a computer simulation of a minimal version of the hypothesis. In the simulation, a 3-dimensional volume of neuropil is presented with patterned afferent input; NO is produced, diffuses, and is destroyed; and synaptic strengths are determined by a set of synaptic rules based on the correlation of synaptic depolarization and NO levels. According to the hypothesis, voltage-dependent postsynaptic release of this rapidly diffusing substance links the activities of neurons in a local volume of tissue, regardless of whether the neurons are directly connected by synapses. This property is demonstrated in the simulation, and it is this property that is exploited in the hypothesis to account for certain aspects of long-term potentiation and activity-dependent sharpening of axonal arbors.

Reentrant signaling among simulated neuronal groups leads to coherency in their oscillatory activity.

Recent experiments have revealed tightly synchronized oscillatory discharges in local assemblies of cortical neurons as well as phase coherency of oscillations at distant cortical sites. These findings are consistent with the theory of neuronal group selection, a population theory of brain function that is based on the properties of local groups of neurons. A set of computer simulations shows that cooperative interactions within and among neuronal groups can generate the observed phenomena. In the simulations, oscillations within neuronal groups are generated through local excitatory and inhibitory interactions. Different groups in general oscillate in an uncorrelated fashion. Coherency of the oscillatory activity of different neuronal groups depends crucially on reciprocal reentrant signaling and can reflect the spatial continuity of a stimulus. Separated or discontinuous features of a given stimulus can be transiently associated in a temporally coherent pattern through reentrant signaling between groups in networks responding to different aspects of that stimulus. A simulation of reentrant activity between arrays of neuronal groups selective for oriented lines and pattern motion displays cross-correlations between groups that are responsive to different parts of a stimulus contour if these parts move together. Such coherency among neuronal groups might be used in the discrimination of a stimulus from other stationary or differentially moving elements in a visual scene.

Synthetic neural modeling applied to a real-world artifact.

We describe the general design, operating principles, and performance of a neurally organized, multiply adaptive device (NOMAD) under control of a nervous system simulated in a computer. The complete system, Darwin IV, is the latest in a series of models based on the theory of neuronal group selection, which postulates that adaptive behavior is the result of selection in somatic time among synaptic populations. The simulated brain of Darwin IV includes visual and motor areas that are connected with NOMAD by telemetry. Under suitable conditions, Darwin IV can be trained to track a light moving in a random path. After such training, it can approach colored blocks and collect them to a home position. Following a series of contacts with such blocks, value signals received through a "snout" that senses conductivity allow it to sort these blocks on the basis of differences in color associated with differences in their conductivity. Darwin IV represents a new approach to synthetic neural modeling (SNM), a technique in which large-scale computer simulations are employed to analyze the interactions among the nervous system, the phenotype, and the environment of a designed organism as behavior develops. Darwin IV retains the advantages of SNM while avoiding the difficulties and pitfalls of attempting to simulate a rich environment in addition to a brain.

Structure of soncanavalin A at 4 A resolution.

Concanavalin A, a phytohemagglutinin isolated from the jack bean, crystallizes at pH 6.8 in the orthorhombic space group 1222 with a = 89.9, b = 87.2, and c = 63.1 A. We have analyzed x-ray diffraction intensity data to 4 A resolution on native concanavalin A and five heavy-metal derivatives: lead, mersalyl, chloroplatinate, uranyl, and o-mercuri-p-nitrophenol. Heavy-atom positions, occupancies, and isotropic thermal parameters have been refined by least-squares methods. The electron density maps clearly show the molecular shape and the packing of the concanavalin A molecules. The asymmetric unit (mol wt 27,000) forms an elliptical dome or "gumdrop" with a base of approximately 46 x 26 A and a height of 42 A. The subunits are paired across 2-fold axes parallel to the c-axis to form dimers. The dimers are in turn paired across points of D(2) symmetry to form tetramers of roughly tetrahedral shape. Each unit has a depression located on the surface which could be the site of saccharide binding. In many regions we have been able to trace the course of the polypeptide chain.

The covalent and three-dimensional structure of concanavalin A. III. Structure of the monomer and its interactions with metals and saccharides.

The three-dimensional structure of the lectin concanavalin A (Con A) has been determined at 2.0-A resolution by x-ray diffraction analysis. The protomers are ellipsoidal domes of dimensions 42 times 40 times 39 A. Folding of the polypeptide backbone is dominated by the presence of two antiparallel pleated sheets, a twisted sheet of seven strands passing through the center of the molecule and a bowed sheet of six strands which forms the back surface of the monomer. Manganese and calcium ions bind to the protein at adjoining sites to form a binuclear complex of two octahedra sharing a common edge. The ligands for each metal ion are four groups from the NH2-terminal region of the protein and 2 water molecules. The binding site for the inhibitor beta-(o-iodophenyl)-D-glucopyranoside is in a deep cavity which contains distinct hydrophobic and hydrophilic binding subsites. Studies of the binding of beta-(o-iodophenyl)-D-glucopyranoside to Con A in the crystalline state and in solution have indicated that the binding behavior of the protein is somewhat different in the two states.

The covalent and three-dimensional structure of concanavalin A.

The tentative amino-acid sequence and three-dimensional structure of the lectin concanavalin A have been determined. The amino-acid sequence, which was determined chemically, contains 238 residues. The sequences of three short stretches were assigned on the basis of x-ray crystallographic data. Interpretation of an electron density map at 2-A resolution indicates that the predominant structural element is extended polypeptide chain arranged in two anti-parallel pleated sheets or beta-structures. Residues not included in the beta-structures are arranged in regions of random coil. One of the pleated sheets contributes extensively to the interactions among the monomers to form both dimers and tetramers. The positions at which Mn(2+), Ca(2+), and saccharide are bound to the protein, and the point of cleavage for the formation of the naturally occurring fragments A(1) and A(2), have been tentatively assigned. Both metal-binding sites are at least 20-A removed from the position at which saccharides are bound. The saccharide-binding site is a deep pocket of approximately 6A x 7.5A x 18A, the inner portion of which is occupied by hydrophobic residues.