Visual Receptive Field Properties of Neurons in the Mouse Lateral Geniculate Nucleus

Abstract

The lateral geniculate nucleus (LGN) is increasingly regarded as a "smart-gating" operator for processing visual information. Therefore, characterizing the response properties of LGN neurons will enable us to better understand how neurons encode and transfer visual signals. Efforts have been devoted to study its anatomical and functional features, and recent advances have highlighted the existence in rodents of complex features such as direction/orientation selectivity. However, unlike well-researched higher-order mammals such as primates, the full array of response characteristics vis-à-vis its morphological features have remained relatively unexplored in the mouse LGN. To address the issue, we recorded from mouse LGN neurons using multisite-electrode-arrays (MEAs) and analysed their discharge patterns in relation to their location under a series of visual stimulation paradigms. Several response properties paralleled results from earlier studies in the field and these include centre-surround organization, size of receptive field, spontaneous firing rate and linearity of spatial summation. However, our results also revealed "high-pass" and "low-pass" features in the temporal frequency tuning of some cells, and greater average contrast gain than reported by earlier studies. In addition, a small proportion of cells had direction/orientation selectivity. Both "high-pass" and "low-pass" cells, as well as direction and orientation selective cells, were found only in small numbers, supporting the notion that these properties emerge in the cortex. ON- and OFF-cells showed distinct contrast sensitivity and temporal frequency tuning properties, suggesting parallel projections from the retina. Incorporating a novel histological technique, we created a 3-D LGN volume model explicitly capturing the morphological features of mouse LGN and localising individual cells into anterior/middle/posterior LGN. Based on this categorization, we show that the ON/OFF, DS/OS and linear response properties are not regionally restricted. Our study confirms earlier findings of spatial pattern selectivity in the LGN, and builds on it to demonstrate that relatively elaborate features are computed early in the visual pathway.

Fig 1. Spatial receptive field profiles of mouse LGN neurons.

A) Confocal image of a coronal section of a brain slice (200 μm thickness). The electrode was dyed with DiI/DMSO solution (red track), and the slice stained with DAPI (Diamidino-2-phenylindole, displayed in green). The LGN boundary and electrode track within the LGN are delineated in white and purple respectively. Electrode sites corresponding to seven recorded neurons are labelled. B) 2D profiles of receptive fields from seven units reconstructed with the STA technique. The colour map in grey-scale shown at the bottom right indicates the intensity of the response of each unit to the contrast-noise movie (see Methods). MEA channel labels correspond to those in the histological image (A). C) Spike waveforms for the neurons shown in (B). Vertical scale bars indicate 25 mV. D) Receptive field centre location for all units in the dataset. The main panel displays the receptive field location of each cell within the visual display, with histograms at left and bottom showing the distributions of receptive field locations on the vertical and horizontal axis respectively. E) Correlation between receptive field radius vertical and horizontal components. Three single-cell representatives show cells whose receptive field fell into the three categories respectively: larger size on the vertical axis (1), circular (2) and larger size on the horizontal axis (3).

Fig 2. Spontaneous and evoked firing rates of mouse LGN neurons.

A) Distributions of spontaneous (triangle) and evoked (square) firing rates of single cells. Solid lines show medians with interquartile ranges; the medians were 2.0 spikes/sec and 7.3 spikes/sec for spontaneous and evoked firing rates respectively. The horizontal position of data points is for clarity in showing the distribution, and conveys no further meaning. B) Correlation of spontaneous (x) and evoked (y) responses. The red line marked with stars represent y = x. Solid line shows the least squares best-fit line, y = 2.25x, and dashed lines the 95% confidence interval 2.08–2.42. R² = 0.31.

Fig 3. Spatial frequency tuning properties of mouse LGN cells.

A) Examples of spatial frequency tuning of three single cells. Open circles indicate mean firing rates and error bars indicate SEM across six repeated presentations of the drifting gratings. Black curves show the best fits of these raw data to a DoG function. The grey area indicates the SEM of spontaneous activity, with thinner lines indicating mean values. Note: logarithmic scale of x-axis. Top left, a typical cell showing a “low-pass” tuning response. Middle panel: a “band-pass” cell. Bottom left, a “bandpass” cell preferring higher spatial frequencies. B) Distribution of preferred spatial frequency for bandpass cells (45 cells of 92). Median (arrow) = 0.035 c/deg. C) Distribution of cut-off spatial frequency for bandpass cells. Median (arrow) = 0.16 c/deg.

Fig 4. Band-pass temporal frequency tuning properties of neurons in mouse LGN.

A) An example of commonly encountered temporal frequency tuning profile (“band-pass”). Open circles indicate mean firing rates, and error bars the SEM across four drifting grating repeats (see Table 1). The black curve shows the best fit of a two-half-Gaussian function. The grey area indicates the SEM of spontaneous activity, with the thinner dark line indicating the mean value. Panels B-E present additional parameters measured from the band-pass subtype that represent 84.5% (49/58) of neurons in the data set. In all cases, the arrows show the median of the distribution. B) Distribution of high₅₀ cut-off, with median 6.0 (5.3, 7.0) Hz. C) Distribution of low₅₀ cut-off, with median 1.40 (1.07, 1.60) Hz. D) Distribution of tuning bandwidth, calculated as the difference between high₅₀ and low₅₀ (range illustrated in red in the inset) with median 4.70 (4.07, 5.60) Hz. E) Distribution of preferred temporal frequency with median 3.2 (2.9, 3.6) Hz.

Fig 5. Low-pass and high-pass temporal frequency tuning.

A) An example of a cell showing low-pass temporal frequency tuning. Open circles indicate mean firing rates across four repeated presentations of drifting gratings (high₅₀ 1.1 Hz). Black curves show the best fits of a two-half-Gaussian function. Grey areas represent SEM of spontaneous activity, with thinner lines indicating mean. Inset: high₅₀ of all low-pass cells. B) An example of a cell showing high-pass temporal frequency tuning (low₅₀ 1.4 Hz). Inset: low₅₀ of all high-pass cells.

Fig 6. Linearity of LGN neuronal responses.

Activity of a single linear (A) and non-linear (B) classified cell across a 7 second presentation of a sinusoidal grating at the preferred spatial frequency of the cell with 1 Hz temporal frequency (average of 6 trials). The top trace represents the time course of the stimulus. The Linearity Index for the linear cell was 1.03 and for the non-linear cell was 0.05. C) Distribution of the Linearity Index across the population of cells. Note logarithmic scale of X-axis. The red dotted line shows the threshold for demarcating linear and non-linear responses. Four cells responded in a non-linear fashion and the remaining 88 were classified as linear in their responses.

Fig 7. Contrast sensitivity of neurons in mouse LGN.

Examples of single cell response tuning to drifting gratings of varying contrasts are shown in A-D. Open circles indicate mean firing rates (average of 10 repetitions). Black curves show the best fits to a hyperbolic function. Red rhomboid indicates C₅₀ value. Note logarithmic scale of X-axis. Four types of responses were observed. A) A cell whose response amplitude increased sharply from very low contrast and began to saturate at relatively low contrast. B) A cell showing a sigmoid response curve: response amplitude began to increase rapidly after a short period of slow increase at low contrasts. C) A cell showing an almost linear increase in response amplitude with increasing contrast. D) A cell displaying a linear increase only at higher contrast. E) Distribution of contrast gain across the population. Mean (arrow) ± SEM = 0.98 ± 0.04 spikes/sec. F) Distribution of C₅₀ across mouse LGN cells. Mean (arrow) ± SEM = 0.50 ± 0.02.

Fig 8. Direction/orientation selectivity in mouse LGN.

Panels A and B show examples of single DS/OS neurons that responded preferentially to direction (A), and to anterior-posterior orientation (B). Radial units are spikes/sec. The red dashed line indicates the spontaneous activity level. C) Distribution of Direction-Selectivity Index values. The red dashed lines indicate the threshold for classification (0.33) and the arrow marks the median of the distribution 0.071 (0.038, 0.12). D. Distribution of Orientation-Selectivity Index values. The red dashed lines indicate the threshold for classification (0.6) and the arrow marks the median of the distribution 0.13 (0.082, 0.20).

Fig 9. Transient/sustained responses of mouse LGN neurons.

A) An example of a cell that responded transiently to a flicker stimulus. The top bar represents stimulus onset, with full-field white stimulation starting at time 0 msec and changing to a black stimulation starting from 600 msec and persisted for another 600 msec. The middle panel shows the raster plot of the cell’s response for all trials represented in the Y-axis against time in the X-axis. The bottom panel shows the PSTH of the same response in 50 msec. bins. B) A cell that responded in a sustained manner to the same stimulus. All parameters as in A. C) Distribution of the transient/sustained index across the population of cells. The red dashed line indicates the threshold for classifying cells as either sustained (<1) and transient (>1). The majority of the cells in our dataset (121 out of 127, 95.3%) responded transiently.

Fig 10. Different response properties of ON- and OFF-centre cells.

A) ON-centre cells show significantly higher mean contrast gain than OFF-centre cells (t-test, P<0.01). ON-centre cells: 1.10 ± 0.05 spikes/sec; OFF-centre cells: 0.87 ± 0.06 spikes/sec. B) ON-centre cells show significantly lower mean C₅₀ values than OFF-centre cells (t-test, P<0.05). ON-centre cells: 0.47 ± 0.03; OFF-centre cells: 0.56 ± 0.03. C) ON-centre cells show significantly lower mean preferred temporal frequencies than OFF-centre cells (t-test, P<0.01). ON-centre cells: 3.1 ± 0.1 Hz; OFF-centre cells: 3.6 ± 0.2 Hz. D) ON-centre cells have significantly lower mean temporal frequency bandwidths than OFF-centre cells (t-test, P<0.001). ON-centre cells: 4.9 ± 0.3 Hz; OFF-centre cells: 6.5 ± 0.3 Hz. E) Mean temporal frequency high₅₀ values are comparable between ON- and OFF-centre cells (t-test, P>0.1). ON-centre cells: 7.2 ± 0.4 Hz; OFF-centre cells: 7.6 ± 0.3 Hz. F) ON-centre cells have significantly higher mean temporal frequency low₅₀ values than OFF-centre cells (t-test, P<0.01). ON-centre cells: 1.0 ± 0.1 Hz; OFF-centre cells: 0.6 ± 0.1 Hz. The value for each single ON- or OFF-centre cell is presented as a triangle or a square respectively. The horizontal dark line represents the mean value and the vertical error bar represents ± SEM in each plot.

Fig 11. Histological reconstruction of mouse LGN in 3D, with the locations of individual cells.

A) The front view of the 3D LGN model. B) The back view of the 3D LGN model. C) The side view of the LGN volume. The model was evenly divided into three sub-regions along the anterior-posterior axis and named as the anterior (blue), the middle (green) and the posterior (magenta) LGN respectively. A representative slice is taken out from each of the subdivision and is shown on the right. D) Mapping of confocal (green) localisation (back view) onto the 3D LGN volume model (outlined in white) along with the electrode tracks in red from one subject to delineate recording sites as belonging to anterior, middle or posterior areas of LGN. E) Side-view of the same mapping. The LGN volume was made transparent to visualize the electrode track. The side view (E) indicates that in this specific recording, the electrode was located in the middle LGN. F) Front view of location of each cell within the LGN volume. G) Side view of panel F. Cell location was determined by mapping confocal image of electrode track and electrode site with the LGN volume. 41, 129 and 15 cells were categorized as located in the anterior (red stars), middle (black dots) and posterior (blue crosses) sub-regions. Scales represent the actual dimensions in μm. H) The three panels show 3D representation within the LGN of ON and OFF cells (first panel), DS/OS cells (second panel), and linear cells (third panel). The exact numbers of these cells are given in Results.