Neurons in the lateral geniculate nucleus (LGN) of the thalamus are the major source of visual input to the cerebral cortex. Much of our understanding of the physiology of LGN neurons comes from data collected in anaesthetized animals. This study examines the visual responses of LGN neurons in alert animals and compares these responses to those from anaesthetized animals. Compared to the anaesthetized animal, LGN neurons in the alert animal respond to visual stimuli with stronger responses and follow stimuli drifting at higher spatial and temporal frequencies. Stronger responses are likely to translate into an increased coupling between the LGN and visual cortex, as firing rate and interspike interval are known to influence the dynamics of synaptic communication between the two structures.
Despite the increasing use of alert animals for studies aimed at understanding visual processing in the cerebral cortex, relatively little attention has been focused on quantifying the response properties of neurons that provide input to the cortex. Here, we examine the response properties of neurons in the lateral geniculate nucleus (LGN) of the thalamus in the alert macaque monkey and compare these responses to those in the anaesthetized animal. Compared to the anaesthetized animal, we show that magnocellular and parvocellular neurons in the alert animal respond to visual stimuli with significantly higher firing rates. This increase in responsiveness is not accompanied by a change in the shape of neuronal contrast response functions or the strength of centre–surround antagonism; however, it is accompanied by an increased ability of neurons to follow stimuli drifting at higher spatial and temporal frequencies.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Although it is generally assumed that the major differences between magnocellular and parvocellular neurons reported for anaesthetized animals should hold for alert animals, the details of these differences may be affected by anaesthesia. For instance, common agents used for anaesthesia, such as isoflurane, urethane, pentothal, nembutal, sufentanil and propofol, are known to have a variety of effects on synaptic transmission and/or neuronal excitability (Paddleford, 1999; Dilger, 2002; Campagna et al. 2003; Rudolph & Antkowiak, 2004; Evers & Crowder, 2005). Consequently, visual responses in the LGN of alert animals may be significantly stronger than those in the anaesthetized animal. Along these lines, an increase in neuronal excitability could be accompanied by shifts in tuning functions, response reliability, and/or the balance of centre–surround antagonism.
This study provides a quantitative examination of visual responses in the LGN of alert macaque monkeys. Data are also provided for comparison from macaque monkeys anaesthetized with a combination of sufentanil citrate (6 μg kg−1 h−1) and isoflurane (0.4%). For both magnocellular and parvocellular neurons, our results show that response reliability, centre–surround antagonism and the shape of neuronal contrast response functions are similar for neurons in the alert and anaesthetized animal. In contrast, magnocellular and parvocellular neurons in the alert animal respond to visual stimuli with significantly higher firing rates and follow stimuli drifting at significantly higher temporal frequencies. Magnocellular neurons in the alert animal also follow stimuli at significantly higher spatial frequencies.
Single-unit recordings were made from LGN neurons in two alert and 10 anaesthetized macaque monkeys (Macaca mulatta). For both alert and anaesthetized animals, receptive fields were parafoveal (>90% within the central 20 deg). Data from a subset of neurons also contributed to a recent study examining the temporal dynamics of extraclassical suppression in the primate (Alitto & Usrey, 2008). This study used contrast response functions to classify magnocellular and parvocellular neurons; it did not examine maximum or spontaneous firing rates, temporal frequency tuning, spatial frequency tuning, centre–surround interactions, response reliability, or burst activity in the alert or anaesthetized animal.
Surgery and preparation
All surgical and experimental procedures conformed to NIH guidelines, were in accordance with the policies and regulations described in this journal (Drummond, 2009), and were carried out with the approval of the Animal Care and Use Committee at the University of California, Davis.
For experiments in anaesthetized animals, anaesthesia was induced with ketamine (10 mg kg−1, i.m.) and maintained with sufentanil citrate (6 μg kg−1 h−1, i.v.) and isoflurane (0.4%). Animals were placed in a stereotaxic apparatus where the temperature, electrocardiogram, electroencephalogram and expired CO2 were monitored continuously. Proper anaesthetic depth was assessed by monitoring the EEG for changes in slow-wave or spindle activity and the ECG and expired CO2. If changes in any of these measures indicated a decreased level of anaesthesia, additional sufentanil was given and the rate of infusion increased. Pupils were dilated with 1% atropine sulfate and eyes were glued to posts attached to the stereotaxic frame. The eyes were fitted with contact lenses and focused on a tangent screen located 172 cm in front of the animal. A midline scalp incision was made and wound margins were infused with lidocaine. A small craniotomy was made above the LGN. Once all surgical procedures were complete, animals were subjected to neuromuscular block with vecuronium bromide (0.2 mg kg−1 h−1, i.v.) and mechanically respired. At the end of each experiment, animals were killed with an overdose of sodium pentobarbital (100 mg kg−1, i.v.).
Under full surgical anaesthesia (isoflurane 1.5–2.0%), the animals used for alert recordings were equipped with a scleral eye coil and a cranial implant containing a head restraint post and a recording cylinder.
Data acquisition and visual stimuli
Single-unit responses from LGN neurons were made using platinum-in-glass electrodes (Alpha Omega, Nazareth Illit, Israel). Responses were amplified, filtered and recorded to a PC with a Power 1401 data acquisition interface and Spike 2 software (Cambridge Electronic Design, Cambridge, UK). Visual stimuli were created with a VSG2/5 (Cambridge Research Systems, Rochester, UK) and presented on a gamma-calibrated Sony monitor running at 140 Hz. The mean luminance of the monitor was 38 candelas m−2.
Visual responses of LGN neurons were characterized quantitatively using drifting sinusoidal gratings (vertical orientation). For experiments with anaesthetized animals, drifting gratings were shown for 4 s, followed by 4 s of mean grey. For experiments with alert animals, drifting gratings were shown for 3 s while animals maintained fixation for a fluid reward. Proper fixation was the only behavioural task required for the reward. Trials were immediately aborted if eye position, measured with a scleral eye coil, deviated by more than 0.35 deg. The inter-stimulus interval was >3 s and the screen was mean grey. For both anaesthetized and alert recordings, the size of the visual stimulus was ∼3 times the size of the classical receptive field. Importantly, with the exception of stimulus duration, all comparisons of neuronal responses between alert and anaesthetized animals were made using identical stimuli (e.g. stimulus size, contrast and temporal frequency).
Spatial frequency tuning
Spatial frequency tuning curves were made both to determine the optimal spatial frequency for subsequent grating experiments and to assess the spatial frequency to evoke a response 50% of maximum (SF50), and the strength of the receptive field surround, relative to the centre. Responses to drifting sine-wave gratings (4 Hz, 100% contrast) presented at 10–16 different spatial frequencies (typical range: 0.1 to 10 cycles deg−1) were fitted to a frequency domain difference of Gaussians (DOGF) equation,
where R(f) is the first harmonic (f1) of the response evoked by spatial frequency f, rc is the radius of the centre subunit, and rs is the radius of the surround subunit. A constrained non-linear optimization procedure (MATLAB function: fmincon; The Mathworks Inc., Natick, MA, USA) was used to minimize the squared error (i.e. Σ (Data-Fit)2) when fitting the DOGF functions and all subsequent data sets.
To examine the relative strength of surround antagonism, we calculated an antagonism index:
With this equation, a neuron with equal response rates to the lowest spatial frequency examined and the preferred spatial frequency would have an antagonism index of zero, while a neuron with no response to the lowest spatial frequency would have an index equal to 1.
Contrast response functions
To determine the influence of contrast on neuronal activity, contrast response functions were made from responses (first harmonic, f1) to drifting sine-wave gratings (4 Hz, preferred spatial frequency) presented over a range of contrasts (0.1% to 100%). Neuronal responses were fitted to a hyperbolic ratio (Albrecht & Hamilton, 1982):
where C represents the contrast levels presented during the experiment, K represents the maximum response (f1), C50 is the contrast corresponding to 50% of the cell's maximum response, DC is the firing rate to a blank grey screen, and n is a variable reflecting the cell's sensitivity.
Temporal-frequency tuning curves were made from neuronal responses (first harmonic, f1) to drifting sinusoidal gratings (0.5 to 64 Hz; 100% contrast; preferred spatial frequency). Response curves were interpolated with a cubic spline to determine the preferred temporal frequency and the highest temporal frequency to evoke responses 50% of maximum (TF50).
Fano factor analysis
Response reliability was quantified by calculating the Fano factor from responses to >48 cycles of a drifting grating (100% contrast, 4 Hz, optimal spatial frequency).
Gratings were presented in blocks of 12 or 16 cycles. For each block, the first 3 cycles were excluded from analysis to avoid onset transients. Fano factors were also calculated using the full response and were not significantly different from those with the first 3 cycles removed. Fano factors were calculated using two different bin sizes, 250 ms (full stimulus cycle) and 16.6 ms (a value reflecting the integration time of postsynaptic cortical neurons; Usrey et al. 2000).
When sample averages (arithmetic mean) are provided, they are followed by the standard error of the mean (±s.e.m.).
We recorded single-unit, extracellular responses from LGN neurons in two alert macaque monkeys (n= 59 magnocellular neurons, 49 parvocellular neurons) and 10 macaque monkeys anaesthetized with a combination of sufentanil citrate (6 μg kg−1 h−1) and isoflurane (0.4%; n= 32 magnocellular neurons, 31 parvocellular neurons).
Contrast response functions
Contrast response functions were generated from neuronal responses to drifting sinusoidal gratings (4 Hz, preferred spatial frequency) that varied in contrast from 0.1 to 100%. Response functions from four representative neurons – a magnocellular and parvocellular neuron in the alert animal and a magnocellular and parvocellular neuron in the anaesthetized animal – are shown in Fig. 1A and B, respectively. Consistent with previous reports, the magnocellular neurons responded better to low contrast stimuli than the parvocellular neurons (Kaplan & Shapley, 1986). The magnocellular neurons also displayed response saturation to high contrast stimuli whereas the parvocellular neurons continued to increase their responses as contrast increased.
To compare quantitatively the response functions of LGN neurons in alert and anaesthetized animals, we calculated the contrast needed to evoke 50% of each neuron's maximum response (C50). As shown in Fig. 1C and D, the distributions of C50 values for magnocellular and parvocellular neurons were largely non-overlapping for both alert and anaesthetized animals. More importantly, C50 values were not significantly different for magnocellular neurons in alert and anaesthetized animals (Fig. 1E; 15.7 ± 0.9%vs. 14.6 ± 1.1%, respectively; P= 0.27, Wilcoxon's rank-sum test) nor were they significantly different for parvocellular neurons in the two groups of animals (Fig. 1F; 46.8 ± 1.3%vs. 47.6 ± 1.6%, respectively; P= 0.68, t test). Thus, anaesthesia does not have an appreciable influence on the overall shape of contrast response functions in the macaque LGN (Fig. 2A and B).
Although anaesthesia did not influence the stereotypical shapes of magnocellular and parvocellular contrast response functions, it did influence firing rate for both groups of neurons. To quantify this effect, we measured each neuron's peak firing rate from their contrast response function. As shown in Fig. 2C and D, peak firing rate was 1.3 times greater, on average, for magnocellular neurons and 2.1 times greater for parvocellular neurons in alert animals compared to anaesthetized animals (magnocellular neurons: alert = 45.4 ± 2.5 spikes s−1, anaesthetized = 34.5 ± 3.6 spikes s−1, P < 0.01, Wilcoxon's rank-sum test; parvocellular neurons: alert = 41.8 ± 2.5 spikes s−1, anaesthetized 19.7 ± 2.4 spikes s−1, P < 0.01, Wilcoxon's rank-sum test). Because anaesthesia did not affect C50 values (described above) or the normalized shape of contrast response functions, the reduction in peak firing rate indicates that anaesthesia acts to scale neuronal firing rate in a divisive fashion over a wide range of contrasts.
Temporal frequency tuning
Past studies have shown that neurons in the magnocellular layers of the LGN typically follow stimuli drifting at higher temporal frequencies than neurons in the parvocellular layers (Hicks et al. 1983; Derrington & Lennie, 1984; Levitt et al. 2001; Movshon et al. 2005; but see Spear et al. 1994; Hawken et al. 1996). To compare temporal frequency tuning among magnocellular and parvocellular neurons in the alert and anaesthetized animal, we determined each neuron's preferred temporal frequency (Fig. 3A and B, arrows) and highest temporal frequency to evoke responses 50% of maximum (TF50; Fig. 4A and B, crosses). For both magnocellular and parvocellular neurons, preferred temporal frequencies were significantly greater in alert animals compared to anaesthetized animals (Fig. 3C–F; magnocellular neurons: alert = 21.3 ± 1.3 Hz, anaesthetized = 10.0 ± 1.3 Hz, P < 0.01, Wilcoxon's rank-sum test; parvocellular neurons: alert = 14.9 ± 1.3 Hz, anaesthetized = 6.2 ± 0.4 Hz, P < 0.01, Wilcoxon's rank-sum test). Similarly, TF50 values were significantly greater in alert animals compared to anaesthetized animals (Fig. 4C–F; magnocellular neurons: alert = 45.2 ± 1.2 Hz, anaesthetized = 25.8 ± 2.2 Hz, P < 0.01, Wilcoxon's rank-sum test; parvocellular neurons: alert = 32.3 ± 1.9 Hz, anaesthetized = 11.4 ± 0.9 Hz, P < 0.01, Wilcoxon's rank-sum test). As somewhat of a surprise, the differences in temporal frequency tuning between alert and anaesthetized animals were so substantial that both the preferred temporal frequency and TF50 were greater, on average, for alert parvocellular neurons compared to anaesthetized magnocellular neurons (Fig. 3E and F).
An examination of peak firing rate using data from each neuron's temporal frequency tuning curve also revealed significantly higher response rates among magnocellular and parvocellular neurons in the alert animal compared to the anaesthetized animal (magnocellular neurons: alert = 50.1 ± 3.8 spikes s−1, anaesthetized = 22.0 ± 2.2 spikes s−1, P < 0.01, t test; parvocellular neurons: alert = 38.4 ± 2.6 spikes s−1, anaesthetized = 10.7 ± 1.1 spikes s−1; P < 0.01, t test).
Spatial frequency tuning and surround antagonism
Past studies report wide ranges and overlapping distributions of spatial frequency preferences for magnocellular and parvocellular neurons with an overall trend of parvocellular neurons preferring higher spatial frequencies than magnocellular neurons and a decrease in preferred spatial frequency with eccentricity (Derrington & Lennie, 1984; Spear et al. 1994; Usrey & Reid, 2000; Levitt et al. 2001). We found similar relationships for magnocellular and parvocellular neurons in alert and anaesthetized animals. To determine whether brain state has an influence on spatial frequency tuning, we identified each neuron's highest spatial frequency to evoke responses 50% of maximum (SF50; Fig. 5A–D). While SF50 values were not significantly different for parvocellular neurons in the alert and anaesthetized animal (Fig. 5; 2.95 ± 0.17 vs. 3.16 ± 0.70 cycles deg−1, respectively; P= 0.20, Wilcoxon's rank-sum test), they were significantly higher for magnocellular neurons in the alert animal compared to the anaesthetized animal (Fig. 5E; 2.59 ± 0.11 vs. 1.65 ± 0.25 cycles deg−1, respectively; P < 0.01, Wilcoxon's rank-sum test). To ensure this increase in SF50 was not influenced by differences in the distributions of receptive field locations, we repeated this analysis using a matched subset of cells from the alert and anaesthetized animal where cell pairs were randomly selected with the criterion that their receptive field eccentricities were less than 1 deg apart (n= 16 pairs; mean eccentricity: alert = 6.33 deg, anaesthetized = 6.27 deg). Results from this comparison also revealed significantly higher SF50 values for magnocellular neurons in the alert animal compared to the anaesthetized animal (2.44 ± 0.17 vs. 1.73 ± 0.28; P < 0.05, Wilcoxon's rank-sum test).
To evaluate surround antagonism in alert and anaesthetized animals, we compared neuronal responses to gratings of optimal spatial frequency (which modulate the centre, but not the surround) with responses to gratings of low spatial frequency (which modulate the centre and surround antagonistically; Fig. 6A and B). The strength of surround antagonism was quantified using an antagonism index (see Methods), where values near 1 represent cells with strong surround antagonism and values near 0 represent cells with weak antagonism. As shown in Fig. 6C and D, index values for magnocellular and parvocellular neurons in alert and anaesthetized animals were broadly distributed and overlapping. Consistent with previous results from the squirrel monkey and owl monkey, parvocellular neurons displayed slightly greater surround antagonism than magnocellular neurons (Usrey & Reid, 2000). More importantly, there was not a significance difference between antagonism index values among magnocellular neurons in the alert and anaesthetized animal (Fig. 6E; 0.44 ± 0.33 vs. 0.52 ± 0.06, respectively; P= 0.25, rank-sum test) or between parvocellular neurons in the alert and anaesthetized animal (Fig. 6F; 0.64 ± 0.03 vs. 0.52 ± 0.07, respectively; P= 0.33, rank-sum test).
To address the question of whether brain state affects the reliability of LGN responses, we calculated the Fano factor (response variance/mean response) for LGN neurons excited with a drifting 4 Hz sinusoidal grating of preferred spatial frequency and 100% contrast (Methods). Similar to previous descriptions of LGN responses in the anaesthetized cat and primate, both magnocellular and parvocellular LGN neurons in the alert and anaesthetized macaque monkey displayed sub-Poisson statistics with Fano factor values less than 1.0 (Fig. 7; Kara et al. 2000; Andolina et al. 2007; Victor et al. 2007). Although Fano factor values were similar (and lowest) among the four groups of neurons when this analysis was performed using a bin size corresponding to a full cycle of the sinusoidal stimulus (250 ms; Fig. 7A), some differences were apparent with smaller bin sizes. In particular, with a relatively small bin size (16.6 ms), Fano factor values were significantly less for magnocellular neurons than for parvocellular neurons (Fig. 7B, P < 0.01). This result indicates that magnocellular neurons display greater response precision and reliability during time epochs relevant for the postsynaptic summation of LGN inputs to cortex (Usrey et al. 2000). However, even with a bin size of 16.6 ms, there was little effect of brain state on response reliability.
Bursting and spontaneous activity
Similar to other thalamic neurons, LGN neurons produce spikes that can be classified as burst spikes or tonic spikes. Whether or not an LGN neuron generates burst or tonic spikes depends on the neuron's membrane potential history and the activation state of the T-type Ca2+ channel (Jahnsen & Llinás, 1984a,b; Zhou et al. 1997; Destexhe et al. 1998). Namely, if the cell's resting membrane potential is relatively depolarized, then T-type Ca2+ channels are in an inactivated state and suprathreshold inputs evoke tonic Na+ spikes. In contrast, if the cell's resting potential is more hyperpolarized, then T-type channels become de-inactivated and suprathreshold inputs evoke a low-threshold Ca2+ current that triggers a burst of Na+ spikes. Because T-type Ca2+ channels need to be hyperpolarized for >50 ms to become de-inactivated, this criterion (>50 ms of non-spiking, followed by 2 or more spikes with interspike intervals <4 ms) has been applied with success to identify low-threshold bursts with extracellular electrodes in vivo (Lo et al. 1991; Lu et al. 1992).
Similar to results from previous studies (Weyand et al. 2001; Bezdudnaya et al. 2006; Ruiz et al. 2006), there were significantly fewer bursts in the alert monkey compared to the anaesthetized monkey (Fig. 8A and B; P < 0.01, ANOVA). Indeed, for both magnocellular and parvocellular neurons, anaesthesia increased the percentage of spikes classified as burst spikes by ∼3 times. These findings indirectly indicate that LGN neurons in the macaque monkey are more hyperpolarized under anaesthesia, a view supported by the findings that anaesthesia decreased both the spontaneous activity of neurons (Fig. 8C and D; P < 0.01; ANOVA) and the maximum evoked activity of neurons (Fig. 2).
From the earliest days of microelectrode recordings, researchers have recognized that sensory responses differ between alert and anaesthetized animals. In particular, early studies indicated that single-unit responses in the alert animal are more robust and subject to influences from behaviour and attention (Hubel et al. 1959; Poggio & Mountcastle, 1963). For these reasons, along with ease of training and similarities with human vision, the alert macaque monkey is widely used in studies investigating visual processing, particularly in the cerebral cortex. Despite our increased understanding of visual responses in the cortex of alert animals, our understanding of the physiology of visual inputs to the cortex comes primarily from recordings in anaesthetized animals. The goal of this study was to assess visual responses from neurons in the LGN of alert monkeys and compare these responses to those from anaesthetized animals.
Results from numerous studies examining LGN responses in anaesthetized monkeys demonstrate significant differences between magnocellular and parvocellular neurons. In particular, magnocellular neurons have greater sensitivity to low contrast stimuli and follow stimuli at higher temporal frequencies than parvocellular neurons (Kaplan & Shapley, 1982, 1986; Derrington & Lennie, 1984; Norton et al. 1988; Benardete et al. 1992; O’Keefe et al. 1998; Solomon et al. 1999; Usrey & Reid, 2000; Levitt et al. 2001; Movshon et al. 2005; but see Spear et al. 1994; Hawken et al. 1996). Our results from the alert animal show a similar relationship between magnocellular and parvocellular neurons. However, a closer examination of neuronal response functions reveals some significant differences between alert and anaesthetized animals. Notably, neuronal firing rates were significantly higher in alert animals compared to anaesthetized animals. This increase in firing rate was not accompanied by a change in the overall shape of magnocellular and parvocellular contrast responses functions; however, it was accompanied by a significant increase in the ability of magnocellular and parvocellular neurons to follow stimuli at high temporal frequencies and by an increase in the ability of magnocellular neurons to follow high spatial frequencies. In the sections below, we compare our results to those of past studies and consider their functional implications.
Comparing responses in alert and anaesthetized animals
Despite an overall increase in the firing rate of magnocellular and parvocellular neurons in alert animals, contrast response functions from these neurons appeared similar in shape to those from anaesthetized animals. Similar results have been reported for awake rabbits as they shift between alert and non-alert states (Cano et al. 2006). These finding indicates that contrast response functions are scaled multiplicatively in alert and attentive animals. Consistent with this view, a recent modelling study that included data from Sanchez-Vives et al. (2000) demonstrates that alterations in the membrane potential of cortical neurons results in a multiplicative/divisive scaling of their contrast response functions (Murphy & Miller, 2003). Since many anaesthetic agents can influence the membrane potential of thalamic neurons (Ries & Puil, 1999), these three sets of results are consistent and complementary.
Our results demonstrate that magnocellular and parvocellular neurons in alert animals prefer stimuli drifting at higher temporal frequencies and have higher cutoff frequencies than in animals anaesthetized with a combination of sufentanil and isoflurane. Sufentanil is a powerful opiod analgesic with a relative potency 500–1000 times that of morphine and, when given at sufficient doses, can produce surgical levels of anaesthesia (Helmers et al. 1989; Miller et al. 2009). Sufentanil binds to opiod (G-protein coupled) receptors and acts, in part, on GABAergic transmission. Because sufentanil has less depressing actions on cortical activity than other anaesthetic agents, it has become a preferred anaesthetic agent for neurophysiological recordings in primates. Although the amount of isoflurane (0.4%) used in this study would not be sufficient for anaesthesia in the absence of sufentanil, it is important to be aware that most studies investigating visual processing in the macaque monkey avoid isoflurane due to its suppressive influence on cortical responses. With this in mind, it is worth noting that the preferred temporal frequencies and/or high cutoff frequencies reported in past studies using either sufentanil, fentanil, or nembutal exclusively for anaesthesia are typically less than those reported here for alert animals (Derrington & Lennie, 1984; Spear et al. 1994; Hawken et al. 1996; Levitt et al. 2001; Movshon et al. 2005; but see Hicks et al. 1983). Along these lines, recent results indicate that the influence of brain state on temporal frequency tuning extends beyond effects mediated by anaesthesia, as LGN neurons in awake rabbits follow higher temporal frequencies when animals are in an attentive state compared to a non-attentive state (Bezdudnaya et al. 2006). Likewise, spatial attention has been shown to modulate activity levels in the LGN of macaque monkeys and humans (O’Connor et al. 2002; Casagrande et al. 2005; McAlonan et al. 2008).
Compared to LGN neurons in the alert animal, LGN neurons in the anaesthetized animal had lower levels of spontaneous activity, lower peak firing rates and increased burstiness. Each of these differences is consistent with the view that anaesthesia increased the hyperpolarization of neurons. While this hyperpolarization affected the visual responsiveness of LGN neurons, it did not affect the relative strength of the centre and surround subunits in the classical receptive field (Fig. 6). Along these lines, the relative strength of the extraclassical surround has also been shown to be similar for LGN neurons in alert and anaesthetized animals (Alitto & Usrey, 2008). Increased hyperpolarization with anaesthesia could result from decreased excitation, increased inhibition, or both. It is therefore interesting to note that layer 6 corticogeniculate neurons, a major source of glutamateric synapses in the LGN (Guillery, 1969), are less responsive in anaesthetized animals (Briggs & Usrey, 2007, 2009). Perhaps more importantly, local interneurons providing GABAergic input to LGN relay cells have receptive fields with a centre–surround organization that is very similar to that of their target cells (Wang et al. 2007). Because these interneurons provide a ‘pull’ that compliments the ‘push’ supplied by the retina, changes in the strength of their inhibition should affect the centre and surround subunits similarly – a prediction supported by our finding that surround antagonism is similar for LGN neurons in the alert and anaesthetized animal.
Throughout this study, parvocellular neurons typically displayed greater differences in their response properties between alert and anaesthetized animals than magnocellular neurons. This was particularly noteworthy for measures of maximum firing rate and high temporal frequency cutoff. While the reasons for this differential influence of anaesthesia are unclear, two likely possibilities include (1) a cell size-dependent or membrane-specific influence of anaesthesia (Urban, 2008), whereby small cells are more affected by anaesthesia than large cells (mean cross-sectional area of parvocellular neurons is ∼40% less than that of magnocellular neurons; Weber et al. 2000; see also Montero & Zempel, 1985; Ahmad & Spear, 1993), and (2) differential effects of anaesthesia on the specific circuits and/or balance of excitation and inhibition that underlie magnocellular and parvocellular responses.
Across the four categories of neurons examined here, parvocellular and magnocellular neurons in the anaesthetized animal and parvocellular and magnocellular neurons in the alert animal, there was a positive relationship between maximum evoked firing rate and high temporal-frequency cutoff or TF50. Specifically, parvocellular neurons in the anaesthetized animal had the lowest mean firing rate and lowest mean TF50 (10.7 spikes s−1; 11.4 Hz), followed by magnocellular neurons in the anaesthetized animal (22.0 spikes s−1; 25.8 Hz), parvocellular neurons in the alert animal (38.4 spikes s−1; 32.3 Hz) and, lastly, magnocellular neurons in the alert animal (50.1 spikes s−1; 45.2 Hz). This finding suggests that mechanisms governing the firing rate of a neuron (e.g. membrane potential and/or intrinsic conductances) may also provide an upper bound or constraint on the temporal frequencies a neuron can follow. Consistent with this view, results from two species of New World monkeys reveal a similar relationship between firing rate and temporal frequency tuning, whereby LGN neurons in the diurnal squirrel monkey, which have higher maximum firing rates, follow higher temporal frequencies than neurons in the nocturnal owl monkey, which have lower maximum firing rates (Usrey & Reid, 2000). Given the relationship between firing rate and temporal frequency tuning, it would be interesting to know whether inputs to the LGN that serve to depolarize neurons, such as histamine and acetylcholine, can adjust the sensitivity of LGN neurons to stimuli presented at different temporal frequencies (Zhu & Uhlrich, 1997; Uhlrich et al. 2002).
This study provides a quantitative analysis of the response properties of LGN neurons in the alert macaque monkey and compares these values to those in monkeys under anaesthesia. Our results demonstrate that LGN neurons in the alert and anaesthetized animal do not differ in terms of the relative strength of their centre and surround subunits, the shape of their contrast response functions or their response reliability. In contrast, magnocellular and parvocellular neurons in the alert animal have increased peak and spontaneous levels of activity, decreased burstiness, and a greater ability to follow stimuli at high temporal frequencies. Magnocellular neurons in the alert animal also follow stimuli with higher spatial frequencies. Thus, a greater range of spatiotemporal features in the visual scene are conveyed to the cortex in alert animals than was previously appreciated from recordings in anaesthetized animals. In addition, the increased firing rate of LGN neurons in the alert animal is likely to underlie a greater functional coupling between the LGN and visual cortex, as firing rate and interspike interval are known to influence the dynamics of synaptic communication between pre- and postsynaptic neurons in the two structures (Usrey et al. 2000).
This work was conducted in the laboratory of W.M.U. at the University of California, Davis. H.J.A. and W.M.U. provided conception and design of the experiments, performed data collection, analysis and interpretation of the data, and wrote the article. B.D.M. and D.L.R. performed data collection and interpretation of the data.
This work was supported by NIH grants EY13588, EY12576, NSF grant 0727115 and the McKnight Foundation. Kelly Henning, Katie Neverkovec and Daniel Sperka provided expert technical assistance.
Authors’ present addresses
H. J. Alitto and B. D. Moore: Helen Wills Neuroscience Institute, University of California – Berkeley, Berkeley, CA 94720, USA.
H. J. Alitto: Howard Hughes Medical Institute, University of California – Berkeley, Berkeley, CA 94720, USA.
D. L. Rathbun: Centre for Integrative Neuroscience, University of Tuebingen, Tuebingen, Germany.