Two-dimensional crystals of photosystem II: biochemical characterization, cryoelectron microscopy and localization of the D1 and cytochrome b559 polypeptides.

Photosystem II (PS II) is a photosynthetic reaction center found in higher plants which has the unique ability to evolve oxygen from water. Several groups have formed two-dimensional PS II crystals or have isolated PS II complexes and studied them by electron microscopy and image analysis. The majority of these specimens have not been well characterized biochemically and have yielded relatively low resolution two-dimensional projection maps with a variety of unit cell sizes. We report the characterization of the polypeptide and lipid content of tubular crystals of PS II. The crystals contain the reaction center core polypeptides D1, D2, cytochrome b559, as well as the chlorophyll-binding polypeptides (CP) CP47, CP43, CP29, CP26, CP24, and CP22. The lipid composition was similar to the lipids found in the stacked portion of thylakoids. We also report a 2.0-nm resolution projection map determined by electron microscopy and image analysis of frozen, hydrated PS II crystals. This projection map includes information on the portion of the complex buried in the lipid bilayer. The unit cell is a dimer with unit vectors of 17.0 and 11.4 nm separated by an angle of 106.6 degrees. In addition, Fab fragments against D1 and cytochrome b559 were used to localize those two polypeptides, and thus the reaction center, within the PS II complex. The results indicate that D1 and cytochrome b559 are found within one of the heaviest densities of the monomeric unit.

Interpolar spindle microtubules in PTK cells.

Spindle microtubules (MTs) in PtK1 cells, fixed at stages from metaphase to telophase, have been reconstructed using serial sections, electron microscopy, and computer image processing. We have studied the class of MTs that form an interdigitating system connecting the two spindle poles (interpolar MTs or ipMTs) and their relationship to the spindle MTs that attach to kinetochores (kMTs). Viewed in cross section, the ipMTs cluster with antiparallel near neighbors throughout mitosis; this bundling becomes much more pronounced as anaphase proceeds. While the minus ends of most kMTs are near the poles, those of the ipMTs are spread over half of the spindle length, with at least 50% lying > 1.5 microns from the poles. Longitudinal views of the ipMT bundles demonstrate a major rearrangement of their plus ends between mid- and late anaphase B. However, the minus ends of these MTs do not move appreciably farther from the spindle midplane, suggesting that sliding of these MTs contributes little to anaphase B. The minus ends of ipMTs are markedly clustered in the bundles of kMTs throughout anaphase A. These ends lie close to kMTs much more frequently than would be expected by chance, suggesting a specific interaction. As sister kinetochores separate and kMTs shorten, the minus ends of the kMTs remain associated with the spindle poles, but the minus ends of many ipMTs are released from the kMT bundles, allowing the spindle pole and the kMTs to move away from the ipMTs as the spindle elongates.

Kinetochore microtubules in PTK cells.

We have analyzed the fine structure of 10 chromosomal fibers from mitotic spindles of PtK1 cells in metaphase and anaphase, using electron microscopy of serial thin sections and computer image processing to follow the trajectories of the component microtubules (MTs) in three dimensions. Most of the kinetochore MTs ran from their kinetochore to the vicinity of the pole, retaining a clustered arrangement over their entire length. This MT bundle was invaded by large numbers of other MTs that were not associated with kinetochores. The invading MTs frequently came close to the kinetochore MTs, but a two-dimensional analysis of neighbor density failed to identify any characteristic spacing between the two MT classes. Unlike the results from neighbor density analyses of interzone MTs, the distributions of spacings between kinetochore MTs and other spindle MTs revealed no evidence for strong MT-MT interactions. A three-dimensional analysis of distances of closest approach between kinetochore MTs and other spindle MTs has, however, shown that the most common distances of closest approach were 30-50 nm, suggesting a weak interaction between kinetochore MTs and their neighbors. The data support the ideas that kinetochore MTs form a mechanical connection between the kinetochore and the pericentriolar material that defines the pole, but that the mechanical interactions between kinetochore MTs and other spindle MTs are weak.

Arrangement of inner dynein arms in wild-type and mutant flagella of Chlamydomonas.

We have used computer averaging of electron micrographs from longitudinal and cross-sections of wild-type and mutant axonemes to determine the arrangement of the inner dynein arms in Chlamydomonas reinhardtii. Based on biochemical and morphological data, the inner arms have previously been described as consisting of three distinct subspecies, I1, I2, and I3. Our longitudinal averages revealed 10 distinguishable lobes of density per 96-nm repeating unit in the inner row of dynein arms. These lobes occurred predominantly but not exclusively in two parallel rows. We have analyzed mutant strains that are missing I1 and I2 subspecies. Cross-sectional averages of pf9 axonemes, which are missing the I1 subspecies, showed a loss of density in both the inner and outer portions of the inner arm. Averages from longitudinal images showed that three distinct lobes were missing from a single region; two of the lobes were near the outer arms but one was more inward. Serial 24-nm cross-sections of pf9 axonemes showed a complete gap at the proximal end of the repeating unit, confirming that the I1 subunit spans both inner and outer portions of the inner arm region. Examination of pf23 axonemes, which are missing both I1 and I2 subspecies, showed an additional loss almost exclusively in the inner portion of the inner arm. In longitudinal view, this additional loss occurred in three separate locations and consisted of three inwardly placed lobes, one adjacent to each of the two radial spokes and the third at the distal end of the repeating unit. These same lobes were absent ida4 axonemes, which lack only the I2 subspecies. The I2 subspecies thus does not consist of a single dynein arm subunit in the middle of the repeating unit. The radial spoke suppressor mutation, pf2, is missing four polypeptides of previously unknown location. Averages of these axonemes were missing a portion of the structures remaining in pf23 axonemes. This result suggests that polypeptides of the radial spoke control system are close to the inner dynein arms.

Golgi structure in three dimensions: functional insights from the normal rat kidney cell.

Three-dimensional reconstructions of portions of the Golgi complex from cryofixed, freeze-substituted normal rat kidney cells have been made by dual-axis, high-voltage EM tomography at approximately 7-nm resolution. The reconstruction shown here ( approximately 1 x 1 x 4 microm3) contains two stacks of seven cisternae separated by a noncompact region across which bridges connect some cisternae at equivalent levels, but none at nonequivalent levels. The rest of the noncompact region is filled with both vesicles and polymorphic membranous elements. All cisternae are fenestrated and display coated buds. They all have about the same surface area, but they differ in volume by as much as 50%. The trans-most cisterna produces exclusively clathrin-coated buds, whereas the others display only nonclathrin coated buds. This finding challenges traditional views of where sorting occurs within the Golgi complex. Tubules with budding profiles extend from the margins of both cis and trans cisternae. They pass beyond neighboring cisternae, suggesting that these tubules contribute to traffic to and/or from the Golgi. Vesicle-filled "wells" open to both the cis and lateral sides of the stacks. The stacks of cisternae are positioned between two types of ER, cis and trans. The cis ER lies adjacent to the ER-Golgi intermediate compartment, which consists of discrete polymorphic membranous elements layered in front of the cis-most Golgi cisterna. The extensive trans ER forms close contacts with the two trans-most cisternae; this apposition may permit direct transfer of lipids between ER and Golgi membranes. Within 0.2 microm of the cisternae studied, there are 394 vesicles (8 clathrin coated, 190 nonclathrin coated, and 196 noncoated), indicating considerable vesicular traffic in this Golgi region. Our data place structural constraints on models of trafficking to, through, and from the Golgi complex.

Three-dimensional ultrastructural analysis of the Saccharomyces cerevisiae mitotic spindle.

The three dimensional organization of microtubules in mitotic spindles of the yeast Saccharomyces cerevisiae has been determined by computer-aided reconstruction from electron micrographs of serially cross-sectioned spindles. Fifteen spindles ranging in length from 0.6-9.4 microns have been analyzed. Ordered microtubule packing is absent in spindles up to 0.8 micron, but the total number of microtubules is sufficient to allow one microtubule per kinetochore with a few additional microtubules that may form an interpolar spindle. An obvious bundle of about eight interpolar microtubules was found in spindles 1.3-1.6 microns long, and we suggest that the approximately 32 remaining microtubules act as kinetochore fibers. The relative lengths of the microtubules in these spindles suggest that they may be in an early stage of anaphase, even though these spindles are all situated in the mother cell, not in the isthmus between mother and bud. None of the reconstructed spindles exhibited the uniform populations of kinetochore microtubules characteristic of metaphase. Long spindles (2.7-9.4 microns), presumably in anaphase B, contained short remnants of a few presumed kinetochore microtubules clustered near the poles and a few long microtubules extending from each pole toward the spindle midplane, where they interdigitated with their counterparts from the other pole. Interpretation of these reconstructed spindles offers some insights into the mechanisms of mitosis in this yeast.

Correction for non-perpendicularity of beam and tilt axis in tomographic reconstructions with the IMOD package.

A correction for non-perpendicularity between the beam axis and the tilt axis in electron tomographic tilt series has been implemented in the IMOD software package and its value and limitations have been explored. Correction for this effect can provide a significant improvement in the alignment error and the reconstruction quality in some cases. However, when the projection model being fit includes an anisotropic shrinkage (i.e. stretch) in the plane of the specimen, adding a variable for the beam tilt does not produce a lower alignment error; it is thus not possible to distinguish between the effects of stretch and beam tilt. Test reconstructions indicate that an alignment solution that includes stretch will adequately correct for the effects of a beam tilt. For specimens subject to deformation under the beam, an alignment solution that accounts for stretch is preferable to one that accounts for beam tilt instead, provided that the markers used for alignment are sufficiently well distributed. Otherwise, a correction for beam tilt should be used.

Nuclear pore complex number and distribution throughout the Saccharomyces cerevisiae cell cycle by three-dimensional reconstruction from electron micrographs of nuclear envelopes.

The number of nuclear pore complexes (NPCs) in individual nuclei of the yeast Saccharomyces cerevisiae was determined by computer-aided reconstruction of entire nuclei from electron micrographs of serially sectioned cells. Nuclei of 32 haploid cells at various points in the cell cycle were modeled and found to contain between 65 and 182 NPCs. Morphological markers, such as cell shape and nuclear shape, were used to determine the cell cycle stage of the cell being examined. NPC number was correlated with cell cycle stage to reveal that the number of NPCs increases steadily, beginning in G1-phase, suggesting that NPC assembly occurs continuously throughout the cell cycle. However, accumulation of nuclear envelope observed during the cell cycle, indicated by nuclear surface area, is not continuous at the same rate, such that the density of NPCs per unit area of nuclear envelope peaks in apparent S-phase cells. Analysis of the nuclear envelope reconstructions also revealed no preferred NPC-to-NPC distance. However, NPCs were found in large clusters over regions of the nuclear envelope. Interestingly, clusters of NPCs were most pronounced in early mitotic nuclei and were found to be associated with the spindle pole bodies, but the functional significance of this association is unknown.

Three-dimensional analysis and ultrastructural design of mitotic spindles from the cdc20 mutant of Saccharomyces cerevisiae.

The three-dimensional organization of mitotic microtubules in a mutant strain of Saccharomyces cerevisiae has been studied by computer-assisted serial reconstruction. At the nonpermissive temperature, cdc20 cells arrested with a spindle length of approximately 2.5 microns. These spindles contained a mean of 81 microtubules (range, 56-100) compared with 23 in wild-type spindles of comparable length. This increase in spindle microtubule number resulted in a total polymer length up to four times that of wild-type spindles. The spindle pole bodies in the cdc20 cells were approximately 2.3 times the size of wild-type, thereby accommodating the abnormally large number of spindle microtubules. The cdc20 spindles contained a large number of interpolar microtubules organized in a "core bundle." A neighbor density analysis of this bundle at the spindle midzone showed a preferred spacing of approximately 35 nm center-to-center between microtubules of opposite polarity. Although this is evidence of specific interaction between antiparallel microtubules, mutant spindles were less ordered than the spindle of wild-type cells. The number of noncore microtubules was significantly higher than that reported for wild-type, and these microtubules did not display a characteristic metaphase configuration. cdc20 spindles showed significantly more cross-bridges between spindle microtubules than were seen in the wild type. The cross-bridge density was highest between antiparallel microtubules. These data suggest that spindle microtubules are stabilized in cdc20 cells and that the CDC20 gene product may be involved in cell cycle processes that promote spindle microtubule disassembly.

Correlated firing of retinal ganglion cells.

Even in the absence of visual stimulation, retinal ganglion cells have a substantial maintained discharge. This maintained discharge is not generated independently within each ganglion cell, because the unstimulated activity of two neighboring ganglion cells can be remarkably correlated. These correlations show that two such cells respond together to strong, spontaneous signals from more distal retinal neurons and that, in some cases, ganglion cells even have effects on each other. Observations of correlated firing can give insights not only into the sources of maintained activity but also into retinal connections and signal processing. Correlating firing at the retinal level also has important implications for the use of correlation analysis to study connections between cells in higher visual centers. Much recent attention has focused on the role that correlated firing may play in forming appropriate, ordered connections to a target structure.

Organization of the cat's optic tract as assessed by single-axon recordings.

The organization of the cat's optic tract was assessed from recordings of single axons during vertical electrode penetrations. To analyse both the order and the scatter of axons in the optic tract, units were pooled from electrode tracks that had evidently passed through nearly equivalent parts of the tract. Receptive field elevation, the most strongly ordered parameter, was primarily organized horizontally, increasing from posteromedial to anterolateral tract. Azimuth tended to increase from dorsal to ventral, but was only half as well organized vertically as elevation was horizontally. Axon type was also organized vertically, at least among axons that were identified as X or Y (contralateral axons of less than 20 degrees eccentricity). Axons of ON-center X-type were located mostly in the dorsal quarter, OFF-center X-type in the dorsal two-thirds, and Y-type (ON- and OFF-center) in the ventral two-thirds of the tract. Another difference between the horizontal and vertical axes was revealed by estimates that axons of one type, from one retinal locus, spread about twice as far vertically as horizontally in the tract. Torrealba et al. ('82) have proposed that the sequential addition of ingrowing axons to the pial surface is a major source of the order in the tract. The differences between horizontal and vertical order seen here suggest an extension of this proposal: the sequential addition of axons to the ventral surface generates the vertical order, while some other process arranges axons horizontally, along the ventral surface, according to receptive field elevation (or some similar one-dimensional retinal coordinate).

Non-uniform postnatal growth of the cat retina.

The distributions of alpha-type ganglion cells in 3-week-old and adult cats were used to measure the increase in the distances between existing cells and thus the amount of growth in various regions of the retina. Growth shows two major non-uniformities. (1) The area centralis is at the point of minimum growth; its area increases by only about 3% while regions near the retinal margin increase in area by about 80%. (2) The retina grows about half as much in linear extent as does the radius of the eye and thus comes to occupy a smaller fraction of the globe. Measurements of retinal dimensions indicate that both non-uniformities also occur from birth to 3 weeks. These non-uniformities have the following implications. (1) They would tend to elongate dendritic fields radially, in the direction of the area centralis, in central retina but perpendicular to this direction in peripheral retina. However, these asymmetries are probably not the primary reason why ganglion cells throughout the retina tend to have radially oriented dendritic fields (Leventhal and Schall, '83). (2) Greater growth in the periphery could contribute to the gradient of increasing dendritic field size from central to peripheral retina if the dendritic fields of ganglion cells passively stretched as the retina expanded. Passive stretching is not the primary determinant of dendritic extent, however, because the dendritic fields of beta-type ganglion cells were found to grow 70% more from 3 weeks to adulthood than can be accounted for by passive stretching. (3) Greater peripheral growth steepens the central-to-peripheral gradient of decreasing ganglion cell density; if this trend also occurs prenatally, it could be the major factor in producing the final adult gradient.

Prior activity influences the velocity of impulses in frog and cat optic nerve fibers.

Under normal visual stimulation, simultaneous recording from ganglion cells in the retina and from the axons of these cells in the brain revealed activity-dependent differences in the velocity of impulse propagation. In frog ganglion cells, spikes initiated 5-500 ms after a previous impulse showed supernormal increases in conduction velocity of up to 17%; spikes initiated 500-2000 ms after a previous one traveled more slowly than a spike initiated after a long period of rest. Cat ganglion cell impulses showed much smaller supernormality (maximum 3%), but exhibited pronounced slowing due to refractoriness and long-term fatigue associated with their high levels of spontaneous activity.

Two types of cat retinal ganglion cells that are suppressed by contrast.

Despite some controversy in the literature, all available evidence indicates that cat retinal ganglion cells whose firing is suppressed by contrast are of two distinct types. The two types differ principally in whether they are suppressed tonically or transiently by contrast in the receptive field.

Two classes of single-input X-cells in cat lateral geniculate nucleus. I. Receptive-field properties and classification of cells.

Cells in the cat's dorsal lateral geniculate nucleus (LGN) were studied by presentation of visual stimuli and also by simultaneous recording of their ganglion cell inputs in the retina. This paper describes receptive-field properties and a new system of classification for LGN X-cells that appear to receive essentially only one excitatory retinal input. These X-cells were of two distinct classes. The visual responses of one class of cell (XS, single) replicated the basic form of the responses of a retinal X-cell. The other class of cell (XL, lagged) had responses with two remarkable features: their firing lagged 40-80 ms behind that of XS-cells or ganglion cells at response onset, and they fired anomalously at times when XS-cells or ganglion cells would not be firing. Thus, for a flashing spot, XL-cells were inhibited from firing after stimulus onset, during the time when XS-cells or retinal X-cells had an initial transient peak in firing; XL-cells generally had an anomalous peak in firing after stimulus offset, after XS-cells or retinal X-cells had stopped firing. For a moving bar, XS-cells or retinal X-cells responded primarily while the bar was in the receptive-field center, whereas most of a typical XL-cell's response occurred after the bar had left the receptive-field center. The latencies of various features in the visual responses were analyzed. For several visual response latencies, the distribution was clearly bimodal, thus objectively demonstrating the existence of two cell classes. Using only the latencies from spot and bar responses, over 90% of these single-input cells could be reliably identified as belonging to one of the two classes. The remaining cells (7 of 128) were intermediate between the two classes in some but not all respects; because they had some properties in common, these cells were kept in a separate group (XPL, partially lagged). The axons of both XS- and XL-cells could be antidromically activated from visual cortex. Cortical latencies were typically 0.7-2.0 ms for XS-cells but much longer, typically 2.4-5.0 ms, for XL-cells. It is possible that XL-cells have not previously been recognized as a separate class because cells with such long latencies have been recorded infrequently in the past. Responses to central flashing spots were more transient than those of retinal X-cells for most XS-cells and more sustained for most XL-cells.(ABSTRACT TRUNCATED AT 400 WORDS)

Two classes of single-input X-cells in cat lateral geniculate nucleus. II. Retinal inputs and the generation of receptive-field properties.

The retinal inputs to cells in the cat's lateral geniculate nucleus (LGN) were directly recorded to study the basis for the properties of two classes of LGN X-cells: Xs (single) and XL (lagged). The presence of excitatory or inhibitory input to an LGN cell from a particular simultaneously recorded ganglion cell was assessed with cross-correlograms during unstimulated activity. Because neighboring ganglion cells do not fire independently, features in a retinogeniculate correlogram can arise in two ways that must be distinguished by a direct effect of the ganglion cell on the LGN cell, or by correlated firing between that ganglion cell and some other ganglion cell that is an excitatory or inhibitory input to the LGN cell. It was possible to determine the origin of correlogram features because features indicating a retinogeniculate effect were distinctly different in timing and strength from features arising solely from correlated firing in the retina. The characteristic feature in a correlogram between an LGN cell and an excitatory retinal input was a sharp peak in LGN cell firing rate at the appropriate latency after the firing of the ganglion cell. The characteristic feature for an inhibitory input was a dip in LGN cell firing rate after the firing of the ganglion cell. Typically, this dip lasted 10-40 ms and was followed by a prolonged enhancement in LGN cell firing rate, which may reflect a postinhibitory rebound. XS-cells had a single retinal X input whose excitatory effect caused most of the LGN cell's spikes during stimulated and unstimulated activity. There was no conclusive evidence that any XS-cell received excitatory retinal input from either Y-cells or other X-cells of the same center sign. There was usually evidence for inhibition of XS-cells by retinal X-cells of opposite center sign with receptive fields highly overlapping that of the XS-cell, but rarely evidence for inhibition by Y-cells. XL-cells also had only a single excitatory input, but this X input had a relatively weak effect that caused only a minority of the LGN cell's spikes, typically 17% during maintained activity and 29% during visual stimulation. The input's excitatory effect was immediately followed by strong inhibition of the XL-cell. XL-cells were also inhibited by retinal X-cells of the same center sign that were adjacent (nearest neighbors) to the excitatory input. The strength and latency of both of these inhibitory effects indicate that the inhibition was disynaptic.(ABSTRACT TRUNCATED AT 400 WORDS)

Correlated firing of cat retinal ganglion cells. I. Spontaneously active inputs to X- and Y-cells.

1. The shared inputs to cat retinal ganglion cells have been investigated by studying correlations in the maintained firing of neighboring ganglion cells. The firing of one cell was recorded from its axon in the optic tract, while that of a neighboring cell was simultaneously recorded with a second electrode in the retina. The recorded cells were of the X- or Y-type and viewed a uniform screen having a luminance of 10 cd/m2. 2. Ganglion cells with overlapping receptive-field centers showed two basic forms of correlated firing: if they had the same center sign (both on-center or both off-center), then they tended to fire at the same time, as shown by a peak in their cross-correlogram; but if they had opposite center signs (an on- and and off-center cell), they tended not to fire at the same time, as shown by a well, or dip, in their cross-correlogram. 3. Both of these tendencies were strongest for cells that were close together and did not appear for cells with nonoverlapping receptive-field centers. The strongest correlations were between neighboring Y-cells, cells with large fields, and the weakest were between X-cells, cells with small fields. In general, the strength of the correlations depended primarily on the area of the overlap between fields. 4. These correlations in maintained firing appear to be principally or entirely caused by shared inputs to the ganglion cells from more distal retinal neurons. The signals from these distal neurons appear to have strong, brief (4-8 ms), well-defined effects on ganglion cells, which are observed even in the absence of a visual stimulus. The inputs responsible for the correlated firing are thus referred to as spontaneously active inputs or simply as active inputs. 5. An analysis of the features in the various types of cross-correlograms supports the following statements about these spontaneously active inputs. a) There are two types of active inputs: inputs excitatory to on-center cells and simultaneously inhibitory to off-center center cells and inputs excitatory to off-center cells and simultaneously inhibitory to on-center cells. b) The active inputs of each type provide excitation to both X- and Y-cells of one center sign and inhibition to both X- and Y-cells of the other center sign. There is no evidence for a special class of more selective inputs providing input only to X-cells or only to Y-cells. c) Active inputs account for the majority (about 80%) of the spikes in the maintained activity of Y-cells but only a small fraction (about 15%) of the spikes in the maintained activity of X-cells. 6. A likely source of the active input signals appears to be spiking amacrine cells with a low rate of spontaneous activity.

Dual-axis tomography: an approach with alignment methods that preserve resolution.

Tomographic reconstructions of biological specimens are now routinely being generated in our high voltage electron microscope by tilting the specimen around two orthogonal axes. Separate tomograms are computed from each tilt series. The two tomograms are aligned to each other with general 3-D linear transformations that can correct for distortions between the two tomograms, thus preserving the inherent resolution of the reconstruction throughout its volume. The 3-D Fourier transforms of the two tomograms are then selectively combined to achieve a single tomogram. Unlike a single-axis tomogram, a dual-axis tomogram shows good resolution for extended features at any orientation in the plane of the specimen; it also has improved resolution in the depth of the specimen. Calculations indicate that the improvements available from double tilting and from tilting to higher angles are largely additive. Actual and model data were used to assess whether varying the increment between tilted views in proportion to the cosine of the tilt angle would allow a reduction in the number of pictures required to achieve a given resolution of reconstruction. Analysis by Fourier sector correlation indicated that the variable tilt increment improved the reconstruction in some respects but degraded it in others. A varying tilt increment thus does not give an unqualified improvement, at least when using back-projection algorithms for the reconstruction.

Nonlagged relay cells and interneurons in the cat lateral geniculate nucleus: receptive-field properties and retinal inputs.

Simultaneous recording in the cat's retina and lateral geniculate nucleus (LGN) was used to find excitatory inputs to LGN cells. These recordings, correlated with measurements of LGN cell receptive-field properties, suggested new functional subdivisions of LGN cells. Distinctions between lagged and nonlagged cells were described before (Mastronarde, 1987a,b; Mastronarde et al., 1991), classification of nonlagged cells is examined here. The XS-type relay cells described before (Mastronarde, 1987a,b) each had detectable excitatory input from only one retinal X cell. Cells that received significant input from more than one retinal X cell were of three kinds: relay cells with pure X input (XM); relay cells with mixed X and Y input (X/Y); and cells that could not be antidromically activated from visual cortex (XI). In the series of relay cells, XS-XM-X/Y-Y, cells had progressively larger receptive-field centers, lower spatial resolution, and faster and more Y-like responses to various stimuli. XI cells resembled XM and X/Y cells in some respects but tended to have higher maintained firing rates, more sustained responses, and weaker surround suppression of the center response. The distinctness of XS, XM, X/Y, XI, and Y from each other was examined with a modification of discriminant analysis that allowed cells to lack measurements for some parameters. Any given pair of categories could be distinguished reliably with only three parameters, although less so for X/Y-Y. In particular, XI cells were distinguishable from relay cells by properties other than the results of cortical stimulation, thus supporting the identity of XI cells as a separate class of X interneurons. Two discontinuities in the behavior of retinal input suggest that XM cells are a separate class from XS and X/Y cells: (1) LGN X cells received either no detectable input from any of the retinal X cells adjacent to their main input, or an easily detectable amount from several such cells; and (2) cells received either no Y input or a certain minimum amount. No such discontinuity in input underlies the distinction between X/Y and Y cells. LGN Y cells were also heterogeneous. Those with substantial input from more than one retinal Y cell had larger receptive fields and a greater preference for fast-moving stimuli than did Y cells dominated by a single input. Three Y cells could not be antidromically activated. They tended to differ from Y relay cells and resemble X interneurons in several ways.(ABSTRACT TRUNCATED AT 400 WORDS)