Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2010 Mar 1.
Published in final edited form as: Vision Res. 2009 Feb 13;49(5):505–513. doi: 10.1016/j.visres.2008.12.001

Adaptive changes of inner retina function in response to sustained pattern stimulation

Vittorio Porciatti 1, Lori M Ventura 1
PMCID: PMC2656435  NIHMSID: NIHMS83502  PMID: 19124035

Abstract

We have characterized adaptive changes of inner retina function in response to sustained pattern stimulation in 32 normal subjects with an age range 23 to 77 years by measuring changes of the pattern electroretinogram (PERG) as a function of time. Contrast-reversal stimuli had square-wave profile in space and time, with peak spatial and temporal frequency and high contrast to maximize response amplitude. The PERG signal was sampled over 5 minutes with a resolution of 15 s. PERG signals were non-stationary, resulting in either progressive amplitude decline or even enhancement to a plateau, with a time course that could be well described by an exponential function with a time constant of 1-2 minutes. Higher initial amplitudes were generally associated with amplitude decline, and lower initial amplitudes with enhancement. The delta amplitude (plateau minus initial) was a linear function of the initial amplitude. The magnitude of delta decreased with decreasing initial amplitude and inverted its sign for initial amplitudes about 1/3 lower than the maximum initial amplitude measured, but still about 3-4 times larger than the noise. Amplitude decline was generally associated with phase lag, whereas amplitude enhancement was associated with phase advance. Altogether, PERG generators appear to slowly adjust their gain in order to keep their sustained activity at an intermediate level that is rather independent of the level of activity at stimulus onset. This behavior is reminiscent of a buffering mechanism, where glial cells may play a primary role. An energy budget model of neural-vascular-glial interaction is provided together with an equivalent electrical circuit that accounts for the results.

Keywords: retinal ganglion cell function, pattern electroretinogram, energy metabolism

Introduction

When a neural ensemble is highly activated, there is a concurrent local increase in blood flow (Roy & Sherrington, 1890) (functional hyperemia), which represents one of the fundamental mechanisms exploited in brain imaging.(Logothetis & Wandell, 2004) While the cellular mechanisms underlying functional hyperemia are not fully understood, it is generally believed that neurons control in some way local changes in blood flow, and the glia plays an important role in neuro-vascular coupling. (Fellin & Carmignoto, 2004; Pellerin & Magistretti, 2004) A neural ensemble must have the ability to maintain activity within a range of conditions (dynamic equilibrium) through autoregulation of both neural activity and blood supply. The dynamic equilibrium depends on several factors including the metabolic demand of activated neurons, the available energy supply provided by local blood flow, and the ability of glial cells to deliver energy to active neurons in a chemically compatible form and in a timely manner.(Pellerin & Magistretti, 2004; Fields & Burnstock, 2006) In the primate retina, the RGC metabolic demand is predicted to be very high for ion pumping in unmyelinated RGC axons. (Wang et al., 2003) (Kageyama & Wong-Riley, 1984) Concurrently, the capillary network is denser in proximity to the optic nerve head, where the optic nerve fiber layer is thicker. (Snodderly et al., 1992) The metabolic demand of glial cells to sustain activated neurons as well as themselves may also be substantial.(Winkler et al., 2000) It is believed that functional magnetic resonance imaging signals may reflect changes in energy usage associated with neural activity.(Attwell & Laughlin, 2001; Lennie, 2003) Little is known about the temporal dynamics of neural activity in response to sustained visual stimulation in human. Neural activity is generally derived from Evoked Potentials or Evoked Magnetic fields through robust averaging in order to minimize the noise.(Regan, 1989) Robust averaging, however, assumes stationarity of signal with time and therefore precludes detailed analysis of the temporal dynamics. Here we show that the physiological activity of retinal ganglion cells (RGC) in response to sustained presentation of a high contrast pattern stimulus of peak spatial and temporal frequency is non-stationary. Rather, the temporal dynamics of RGC activity may display either progressive amplitude decline to a plateau) when the amplitude of signal at stimulus onset is relatively high, or progressive amplitude enhancement to a plateau when the amplitude of signal at stimulus onset is relatively low. A neuro-vascular-glial model is provided that accounts for the results.

Methods

Subjects

Subjects of this study were 32 normal individuals of both sexes and different age (range 23 to 77 years, mean age 41.7 ± 16 years). Subjects were free of systemic or ocular diseases as assessed by routine ophthalmologic examination, and had best corrected Snellen visual acuity of 20/20 or better. Subjects had refractive errors smaller than -3.0 spherical diopters and ± 1.5 cylindrical diopters. The methods applied in the study adhered to the tenets of the Declaration of Helsinki for the use of human subjects in biomedical research. Institutional Review Board/Ethics Committee approval was obtained for this study, and informed consent was obtained from each subject before recording.

The Pattern Electroretinogram

The physiological activity of RGC may be assessed by means of the electroretinogram in response to contrast-modulated patterns (Pattern ERG, PERG).(Maffei & Fiorentini, 1981; Zrenner, 1990; Porciatti, 2007) The PERG stimulus offers the opportunity to manipulate the dimensions of the stimulated area as well as the characteristics of pattern elements at constant mean luminance. This allows an approximate match between the size of pattern elements and the size of RGC receptive field at different retinal eccentricity. (Hess & Baker, 1984; Drasdo et al., 1990) Adjacent pattern elements can be temporally modulated either in luminance contrast or in pure chromatic (Red-green or Blue-yellow) contrast to generate luminance-contrast PERG or chromatic-contrast PERG that reflect the properties of different RGC subpopulations. (Morrone et al., 1994a; Morrone et al., 1994b; Sartucci et al., 2003) Luminance-contrast PERG and chromatic-contrast PERG may be differently altered in disease. (Porciatti & Sartucci, 1996; Porciatti et al., 1997; Sartucci et al., 2003)The PERG amplitude in response to luminance-contrast gratings is both spatially and temporally tuned, with a maximum at intermediate frequencies and attenuation for both higher and lower frequencies. (Hess & Baker, 1984; Porciatti et al., 1992) The PERG amplitude increases progressively with increasing contrast. (Zapf & Bach, 1999; Porciatti et al., 2005) The PERG amplitude also increases using gratings with square-wave in space and time compared to sine-wave. Therefore, choosing a pattern with square-wave in space and time, with highest contrast, and with peak spatial and temporal frequency allows maximizing response amplitude. (Hess & Baker, 1984)It has been previously shown that various stimulus conditions that maximize ERG response also elicit maximal, sustained increase of blood flow (∼ 40% compared to baseline) at the neuro-retinal rim of the optic disk. (Logean et al., 2002; Logean et al., 2005; Riva et al., 2005) Flicker and pattern-evoked increase of blood flow at the optic nerve head show no significant trend toward a decrease or increase over ∼ 1 hour of measurements. (Riva et al., 2005)

Technique of steady-state PERG recording

The technique of PERG recording has been described in detail elsewhere (Porciatti et al., 1992; Porciatti & Ventura, 2004) and has been adapted to evaluate non stationarities. The spatio-temporal characteristics of the stimulus have been optimized to yield the highest amplitude. The same spatio-temporal conditions that maximize PERG or focal ERG amplitude have been reported to also elicit maximal vascular response from capillaries overlying the optic nerve head as measured by Laser Doppler Flowmetry. (Logean et al., 2002; Riva et al., 2005) Based on previous systematic studies of the PERG properties, (e.g., Hess & Baker, 1984; Porciatti et al., 1992) the stimulus was set at the peak spatial and temporal frequency, and had maximum contrast. Specifically, the pattern stimulus consisted of horizontal gratings with square-wave profile (1.7 cycles/degree, 25 degree diameter circular field, 99% contrast, 40 cd/m2 mean luminance), reversed in counterphase at 8.14 Hz (16.28 reversals/s) and displayed on a TV monitor. The display was presented at the bottom of a Ganzfeld dome kept at a constant mean luminance of 4 cd/m2. In some experiments the contrast was set at different values (99%, 50%, 25%) at constant mean luminance of 40 cd/m2. In order to measure changes of signal with time, it is necessary that baseline conditions remain stable. We have chosen skin electrodes taped on the lower eyelids rather than traditional electrodes made of metallic or carbon fiber electrodes in contact with the cornea or inserted in the conjunctival sac.(Bach et al., 2000)) (Porciatti & Ventura, 2004) Inherent instability of corneal electrodes, (intrusion from eye movements, blinking) and/or decreased subject compliance with time may result in non-specific temporal changes of PERG signal and increased variability. Instability of corneal electrodes caused by eye movements and/or blinking is not expected to occur with skin electrodes. In addition, the quality of fixation and the level of comfort are not different from those typically achieved in psychophysical experiments, in which subjects can easily tolerate prolonged or repeated recording sessions. We have used gold plated Grass electrodes (10 mm diameter) taped over the lower eyelids and referenced to identical electrodes taped on the ipsilateral temples. The common ground was on the central forehead. (Porciatti & Ventura, 2004) The obvious disadvantage of skin electrodes compared to corneal electrodes is that the PERG signal is reduced by a factor of two-to-three.(McCulloch et al., 1997) However, the steady-state approach (relatively fast averaging, Fourier analysis) allows recording PERG from skin electrodes with signal-to-noise ratio comparable to that of traditional corneal electrodes.(Porciatti & Ventura, 2004; Bach & Hoffmann, 2008) The signal-to-noise ratio was high enough to record reliable responses with a minimum of averaging (50 sweeps), thereby allowing characterization of changes over time with sufficiently close sampling (∼15 s time bins).(Porciatti et al., 2005)

The PERG was recorded simultaneously from both eyes using a commercially available system (Lace Elettronica, Rome, Italy). The subjects fixated on a target at the center of the stimulus with the appropriate correction for a viewing distance of 30 cm. The presentation of the pattern stimulus was preceded by a presentation of a non modulated uniform field of the same mean luminance (40 cd/m2) as the pattern stimulus, during which the signal was also acquired to have a measure of the background noise. Subjects did not receive dilating drops, and were allowed to blink freely. None of the subjects reported visual strain or problems in maintaining fixation. This was also checked in 6 subjects by simultaneously recording the electrooculogram (EOG) from the temporal electrodes and checking that the EOG (measure of amplitude/frequency of horizontal eye movements) did not change with time over 20 minutes of continuous recording.

Signals were band pass filtered (1-30 Hz), amplified (100,000 fold), and averaged in synchrony with the stimulus frequency (122.8 ms period, corresponding to 2 contrast reversals) in successive averages of 50 sweeps each. The first 5 sweeps of the series (10 contrast reversals) were discarded from the analysis to avoid spurious transients. Sweeps contaminated by eye blinks or gross eye movements were automatically rejected over a threshold voltage of 25 μV. Typically, one to three rejections per average occurred. Due to intrinsic limitations of the commercial system, every other sweep was lost in the average. Therefore, the acquisition time for averaging 50 sweeps was longer than theoretically needed by a factor of 2 (122.8 ms *2 *50=12.28 s). Including occasional rejections, the acquisition time for averaging 50 sweeps was ∼ 15 s. Since the PERG was recorded in response to relatively fast alternating gratings, the response waveforms were sinusoidal-like with a frequency corresponding to the reversal rate (Fig. 1). Tiny oscillations superimposed on the main modulation are electro-magnetic artifacts at 65 Hz (the refresh rate of the CRT monitor) that have been incompletely removed by short averaging (50 sweeps only). Oscillations at 65 Hz do not interfere with measurement of PERG amplitude and phase of the contrast-reversal component, isolated by frequency domain analysis. Successive averages were automatically evaluated using Discrete Fourier Transform (DFT) to isolate the component at the reversal frequency (16.28 Hz), whose amplitude in μV (∼1/2 of the peak to trough amplitude) and phase in π rad have been measured. Phase values are bound within ± 1 π rad. To avoid inherent discontinuity of phase data around 0 and ± 1 π rad, phase readings were automatically unwrapped by subtracting actual readings from the value modulo of 2 π rad (2 minus actual reading). Phase values are thus bound between 1 and 3 π rad without discontinuities. At the reversal rate of 16.28 Hz, the value modulo of 2 π rad corresponds to 61.4 ms. Phase advances (shorter latencies) are associated with increasing values, and phase delays (longer latencies) are associated with decreasing values.

Figure 1.

Figure 1

Open in a new tab

Example of PERG waveforms (continuous tracings) recorded simultaneously from the two eyes by means of skin electrodes in response to a high-contrast grating of 1.6 cy/deg alternating 16.28 times/s. The dashed sinusoids superimposed on the PERG waveform represent the 16.28 Hz Fourier component automatically isolated by Discrete Fourier Transform.

Evaluating the temporal dynamics of PERG in response to steady-state stimuli

In order to have an objective evaluation of the dynamic changes of PERG amplitude, data recorded during stimulus presentation were fitted (Sigmaplot 9.01, Systat Software inc.) with an exponential decay function, [y = y0+ a ·exp (-τ/t)], where y is the PERG amplitude at a given time after stimulus onset, y0 is the initial amplitude, a is the delta between the plateau amplitude and the initial amplitude, exp is the exponential symbol, t is time after stimulus onset, and tau (τ) is the time constant of the decay function. Relevant outcome measures for the aims of this study are the initial amplitude y0 and the delta amplitude a. The variability of the response unrelated to the progressive changes with time (i.e., fluctuation around the decay function fit) has been measured as root-mean-square (RMS) of the residuals. Fitting data on log-linear scale did not improve the goodness of fit, as measured by R2. Tau data had a very broad distribution and are not reported. It should be taken into account that in a substantial proportion of subjects with small deltas, tau measurements are not expected to be precise. Another limiting factor for the precision of tau measurement is the low sampling rate (1/15 s). We have done several analyses allowing tau to fluctuate without constrains, constraining tau within the time window of the analysis, or even locking tau to a specific value within the time window. The measurement of deltas (the relevant outcome variable) was virtually unaffected.

Protocol

Initial amplitudes and delta were assessed in all subjects using a standardized protocol (one minute blank followed by 4 minutes or more of pattern stimulus). Prior testing, subjects were allowed at least 5 minutes exposure to the room background, approximately similar to the stimulus mean luminance. The set of successive PERG averages acquired during the first 4 minutes of stimulus presentation was fitted with an exponential decay function as described above to calculate the initial amplitude and delta parameters. The set of averages acquired during 1 minute blank was used to have a measure of noise level (mean: 0.15 ± 0.08 μV across subjects).

Results

An example of PERG progressive amplitude decline is shown (Fig. 2). In this 33 year old female subject a uniform, non modulated stimulus was presented for 1 min (blank), and then the pattern stimulus was presented for 5 min. During blank presentation, the PERG amplitude was at noise level, and the phase assumed random values within the modulo. The major feature of figure 2 is that during sustained pattern presentation, the PERG amplitude slowly decreases with time in both eyes and tends to level off. Data could be well fit with an exponential decay function with a time constant of about 2 minutes. The delta amplitude (plateau minus initial amplitude) was negative. The phase tended to decrease with time (latency increases). An example of positive delta is shown (Fig. 3). In this 68 years old female subject the initially low PERG amplitude progressively increased with time, tended to level off after about 2 minutes, and remained stable for at least 6 minutes. The response phase also tended to increase with time (latency shortened).

Figure 2.

Figure 2

Open in a new tab

Example of PERG decline. PERG amplitude (A) and phase (B) during sequential presentation of either a blank field or a patterned field of high contrast (grey bars over the x-axis). Data represent successive averages of 50 sweeps each (∼ 15 s sampling time). In this subject, during the pattern presentation PERG amplitude progressively declines to a plateau, whereas the phase tends to decrease. Amplitude and phase data have been fitted with an exponential decay function. During the blank presentation the amplitude represents the noise level, and the phase assumes random values.

Figure 3.

Figure 3

Open in a new tab

Example of PERG enhancement. PERG amplitude (A) and phase (B) during sequential presentation of either a blank field or a patterned field of high contrast (grey bars over the x-axis). Data represent successive averages of 50 sweeps each (∼ 15 s sampling time). In this subject, during the pattern presentation the PERG amplitude increases to a plateau; the phase also tends to increase. Amplitude and phase data have been fitted with an exponential decay function.

A relevant point is whether spontaneous fluctuations (RMS of residuals of fit) and noise impact the measurement of outcome variables (Initial amplitude and delta). We have previously shown that response amplitude fluctuation and noise are not related to each other; these measures are also not related to initial amplitude, delta, age, as well as length of recording. In general, deltas are considerably larger than response fluctuations (De Los Santos et al., 2006) We have further investigated the impact of response fluctuation on the goodness of fit (initial amplitude and delta parameters) by simulating conditions in which a given decay function was systematically added with random noise of different magnitude covering the range of fluctuations found in actual experimental conditions. As shown in Fig. 4, initial amplitude and delta amplitude parameters are relatively independent of the magnitude of fluctuations.

Figure 4.

Figure 4

Open in a new tab

Impact of simulated amplitude fluctuations on exponential decay fit. A typical habituating signal with no noise (A) has been added different amounts of random noise (B,C). Fitting the three sets of data with an exponential decay function yields curves with similar parameters.

Figure 5 summarizes the major finding of this study. The difference between final and initial amplitude (delta) depends on the level of initial amplitude (Fig. 5A). For relatively large initial amplitudes, deltas have negative sign, indicating amplitude decline. For both the right and left eyes, the magnitude of delta progressively decreases with decreasing initial amplitude. Delta amplitude assumes a zero value for initial amplitudes of 0.56 μV, which is about 1/3 of the maximum initial amplitudes and about 3.5 times larger than the average noise (∼0.15 μV). For initial amplitudes smaller than 0.56 μV, the sign of delta becomes positive, indicating amplitude enhancement. The magnitude of positive delta increases with decreasing initial amplitude, and this trend continues all the way through the noise level. For both right and left eyes, the correlation between delta and initial amplitude is very strong (R2 = 0.68, P<0.0001). Data can be well fit with a linear regression (delta = 0.29 - 0.52 · initial amplitude); the 95% confidence intervals of the intercept with x-axis are well above zero. Delta phases also display a trend that depends on the initial phase (Fig. 5B). For relatively high phase values, deltas have negative sign, indicating that the responses tend to become slower with time. For relatively slower initial phases, deltas have a positive value, indicating the responses tend to become faster with time. The correlation between delta phase and initial phase is less strong than that of amplitude (R2 = 0.35), but still highly significant (P<0.001). Data can be well fit with a linear regression (delta = 0.84 - 0.45 · initial phase); for delta = zero, the initial phase assumes a value of 1.85 π rad. Initial amplitude and initial phase are not related to each other (Fig. 5C, R2 = 0.09).

Figure 5.

Figure 5

Open in a new tab

Initial amplitude, phase, and delta parameters for all subjects (n=32; filled symbols, right eyes; open symbols, left eyes). A: relationship between delta amplitude and initial amplitude. Note the change in sign of delta amplitude for initial amplitudes about 0.6 μV. B: relationship between delta phase and initial phase. Note the change in sign of delta phase for initial phases about 1.8 π rad. C: relationship between initial amplitude and initial phase. Data in A and B have been fitted with a linear regression lines ± 95% confidence intervals. Note in A and B that the ± 95% confidence intervals do not include zero delta. Dashed lined in A and C represent the noise level.

Altogether, data presented in Fig. 5 indicate that the PERG in response to continuous presentation of optimized pattern stimuli undergoes substantial changes with time, until to reaching relatively steady-state values of amplitude and phase that may be substantially different from corresponding initial values. For initial amplitudes progressively larger that 0.56 μV and initial phases higher than 1.85 π rad, the steady-state response becomes progressively smaller and slower. For initial amplitudes progressively smaller that 0.56 μV and initial phases less than 1.85 π rad, the steady state response becomes progressively larger and faster.

As result of temporal dynamic changes, the range of plateau amplitudes (95 to 5 percentile) across individuals (Fig. 6) is considerably smaller than the corresponding range of initial amplitudes (OD: plateau/initial range = 0.55; OS: plateau/initial range= 0.56). The range of plateau phases is also smaller than the corresponding range of initial phase, however to a lesser extent amplitudes (OD: plateau/initial range = -0.93; OS: plateau/initial range= -0.83). Differences between the range of initial and plateau amplitude are not an artifact originating from fitting fluctuating signals with exponential decay function, since the effect could not be replicated by fitting random noise (not shown in figures).

Figure 6.

Figure 6

Open in a new tab

Ranges of initial and final amplitudes (A) and initial and final phases (B) measured in all subjects (n=32). Box plots represent the median and 25%-75% percentiles. Whiskers represent the 10%-90% percentiles, and filled symbols represent the 5%-95% percentiles. Amplitude data have been plotted on log scale to approximate normal distribution. The bottom of the scale represents the mean noise level.

We investigated whether the observed dependence of sign/magnitude of delta on initial amplitude could be replicated by manipulating the stimulus contrast. Since under the present spatial-temporal conditions the PERG amplitude is linearly related to contrast (Hess & Baker, 1984; Porciatti et al., 2005) it is possible to artificially reduce a initially high initial amplitude by reducing stimulus contrast. Initial amplitudes were ranked, and the subgroup of subjects with initial amplitudes within the first tercile (n=11) were also tested at 50% and 25% contrast (Fig. 7). The responses of the two eyes were averaged. By reducing stimulus contrast up to 25%, the initial amplitude is reduced to a level close to the noise level (Fig. 7A). The magnitude of delta also decreases with decreasing contrast; however the sign of deltas remains always negative (amplitude decline) except a couple of points slightly above zero at 25% contrast, at which the signal is very close to the noise level. There is no systematic change in phase with decreasing contrast (Fig. 7B). There is no relationship between initial amplitude and initial phase (Fig. 7C). The major difference with data displayed in Fig. 5 is that positive deltas cannot be replicated by artificially reducing initially high amplitudes. The phenomenon of positive deltas and phase advancement at high contrast seems therefore an intrinsic property of the PERG of some subjects but not of others. A possible predictor for positive delta might be the subjects’ age. Indeed, the PERG amplitude is know to be substantially reduced with age.(Porciatti & Ventura, 2004) In Fig. 8 initial amplitudes and deltas are displayed as a function of age. There is a weak but significant correlation for both eyes between age and initial amplitude (Fig. 8A, R2 =0.2, P<0.01) and between age and delta amplitude (Fig. 8B, R2 =0.15, P=0.03). However, multiple regression analysis using initial amplitude and age as explanatory variable of delta indicate a very small and not significant effect of age. As shown in Fig. 5A, positive deltas occur for initial amplitudes smaller than 0.56 μV; these low initial amplitudes occur more frequently in older subjects, but may also occur in some young subjects (Fig. 8A). Altogether, data analysis suggest that some subjects without clinically evident indication of eye or systemic disease, as established by routine ophthalmological examination, may have an initial PERG amplitude relatively lower than the majority of subjects in their age range, and this results in a positive delta.

Figure 7.

Figure 7

Open in a new tab

Initial amplitude, phase, and delta parameters as a function of decreasing contrast for a subset of subjects (n=11) whose initial amplitudes were in upper third of the population. Responses of the two eyes have been averaged (black symbols, contrast 99%; gray symbols, contrast 50%; open symbols, contrast 25%). A: relationship between delta amplitude and initial amplitude. B: relationship between delta phase and initial phase. C: relationship between initial amplitude and initial phase. Data in A and B have been fitted with a linear regression lines ± 95% confidence intervals. Note in A and B that the ± 95% confidence intervals include zero delta. Dashed lined in A and C represent the noise level.

Figure 8.

Figure 8

Open in a new tab

Initial amplitude, delta amplitude and delta phase as a function of age of subjects (n=32). Data in A and B have been fitted with a linear regression lines ± 95% confidence intervals.

An energy budget model

We have developed a simple energy-budget model that is consistent with the temporal dynamics of the PERG (Fig. 9). Although the model is entirely speculative, it is based on established notions of neural-vascular-glial interactions. The model assumes that visual stimulus-elicited increase in blood flow is virtually instantaneous at stimulus onset and remains stable with time. (Logean et al., 2002; Logean et al., 2005; Riva et al., 2005) The model also assumes that between the energy source (blood flow increase) and energy sink (highly active RGC), there is an intervening contribution of glial cells that act through energy-transforming, ion-buffering mechanisms. (Reichenbach et al., 1993; Pellerin, 2003; Kofuji & Newman, 2004; Bringmann et al., 2006; Newman, 2006; Giaume et al., 2007)

Figure 9.

Figure 9

Open in a new tab

Model of neural-glial-vascular interactions and energy flow during RGC activation. A: biological model. B: Equivalent electrical circuit. Photoresistor: activated neurons, Capacitor: glial energy store; Resistor: energy absorbed by glia; Current generator: vascular supply; Switch connected by dotted line: neurovascular coupling. The direction of arrows indicates the current flow.

Activation of RGCs by a metabolically challenging pattern stimulus results in an energy demand that causes a rapid vascular dilation in order to provide adequate energy supply; energy is eventually delivered to neurons via a relatively slower glial processing.(Ames, 2000; Logothetis, 2002) At any given time, the energy flow absorbed by active RGCs (ε n) depends on the energy flow produced by the blood flow (ε b) minus the energy flow dissipated by the glia (ε g). According to the principle of conservation of energy, the relative equation is: [Energy budget ΔE = (ε b - ε n - ε g) Δt]. An important variable in this model is the amount of energy stored in glial cells before the onset of pattern presentation. If at pattern onset the glial energy store is full, then activated neurons can fire maximally; their energy demand may be higher than the vascular-glial system can deliver, resulting in a progressive reduction of neural activity to reach a dynamic equilibrium compatible with the available energy budget and the time constant of the slower glial system. If, on the other hand, at pattern onset the glial energy store is empty (e.g., glial cells are unable to store energy or burn energy at a high rate for some reason), then activated neurons can only fire in proportion of the available energy budget at stimulus onset. However, they are still able to send a dilating signal to vessels. Subsequently, the energy flow provided from dilated vessels may be larger that the energy dissipated by glia; therefore the available energy flow for neurons will increase with time until to reaching a dynamic equilibrium compatible with the available energy budget at equilibrium and the time constant of the glial system. The model can be represented with a conceptually-equivalent electrical circuit, where the energy flowing per unit of time (flux) corresponds to an electric current (charge per unit of time, Q/t). The accumulated energy in the glia corresponds to the electric charge Q, which is represented by a capacitor. The flux of energy (current) absorbed by activated neurons is represented by a photoresistor; the flux of energy (current) dissipated by glia is represented by a resistor; the energy provided by the vascular supply is represented by a D.C. current generator, and the neurovascular coupling is represented by a switch (Fig. 9B).

Discussion

Our results show that continuous presentation of pattern-reversing gratings of peak spatial-temporal frequencies and high contrast, elicits non-stationary PERG signals whose time course can be well described by an exponential decay function with a time constant of approximately 2 minutes. Depending on the level of initial amplitude, non-stationarity may result in either amplitude decline or even enhancement, higher initial amplitudes being generally associated with decline and lower initial amplitudes with enhancement. We have previously shown that the plateau amplitude rapidly recovers initial values after a brief (15-60 s) interval with the pattern stimulus blanked to the mean luminance.(Porciatti et al., 2005) The effect can therefore be replicated many times with comparable results. Progressive amplitude decline and enhancement are unlikely to be result of regression toward the mean, since spontaneous fluctuations of response with time as well as the level of noise are unrelated to initial amplitude, delta, age, as well as length of recording. Deltas are generally much larger than spontaneous fluctuations. (De Los Santos et al., 2006)

The delta amplitude (final minus initial) is a linear function of the initial amplitude. For high initial amplitudes, delta has a negative sign. The magnitude of delta progressively decreases with decreasing initial amplitude and inverts its sign for initial amplitudes about 1/3 lower than the maximum initial amplitude measured, but still about 3-4 times larger than the noise. Amplitude decline is generally associated with phase lag, whereas amplitude enhancement is associated with phase advance. The range of inter-individual PERG amplitudes at plateau is considerably smaller than the range of inter-individual PERG amplitudes at stimulus onset. Altogether, the generators of the PERG signal appear to slowly adjust their gain in order to keep their sustained activity at an intermediate level that is rather independent of the level of activity at stimulus onset. This behavior is reminiscent of a buffering mechanism, where glial cells are expected to play a primary role.(Kofuji & Newman, 2004; Bringmann et al., 2006) Glial cells are a key element of neurovascular and neurometabolic coupling, where functional players are neuronal activity, metabolic rate in neurons and glia, and blood flow. (Bringmann et al., 2006; Metea & Newman, 2006) (Winkler et al., 2000) (Araque et al., 2001; Zonta et al., 2003) The very slow temporal dynamics of the PERG decline and enhancement we have described strongly suggests that these mechanisms reflect metabolic processes. Temporal changes in cortical Visual Evoked Potentials have been described. (Janz et al., 2001; Singh et al., 2003) Slow amplitude decline of VEPs is believed to represent an adaptive metabolic mechanism to account for changes in lactate levels. (Sappey-Marinier et al., 1992; Afra et al., 1998) Other VEP studies, however, report considerable inter-subject variability concerning all aspects of the time course.(Heinrich & Bach, 2001a) Reduction of neural activity with time also occurs under conditions that do not seem to depend on buffering mechanisms/metabolic processes. For example, the fast reduction in gain of visual neurons in response to high-contrast stimuli (contrast gain control) reflects rapid adaptive changes of the transfer function of visual neuron themselves at retinal (Shapley & Victor, 1979; Brigell et al., 1987; Chander & Chichilnisky, 2001; Heinrich & Bach, 2001b; Kaplan & Benardete, 2001; Baccus & Meister, 2002) and cortical (Ohzawa et al., 1982; Albrecht, 1995; Porciatti et al., 2000; Albrecht et al., 2002) levels. Contrast gain control mechanisms are unlikely to significantly contribute to the effects described in this study. We cannot exclude possibilities described in classic literature under different nomenclature such as adaptation, habituation, sensitization, neuronal fatigue and neuronal potentiation, some of which are inherent mechanisms in the cell and others being related to feedback circuits. However, the phenomenon we have described does not match those reported under the current nomenclature and associated compartments; the effects may be associated with levels of some message substance. While retinal ganglion cells are believed to be the main PERG generators,(Maffei & Fiorentini, 1981; Zrenner, 1990; Porciatti, 2007) we cannot exclude that the effects occur at least in part at pre-ganglionic level. Photoreceptors are unlikely to contribute since the level of light adaptation is constant.

We have developed a simple energy-budget model of neuron-glia-vascular interactions in the retina that is consistent with the temporal dynamics of the PERG we have described. While we realize that this model is entirely speculative, it may provide clues on mechanistic interactions among functional players. The main advantage is that the models can be represented with an equivalent electrical circuit that easily allows hypothesis testing. Second, the model supports a key role of glial function in neural-vascular interactions and homeostasis in health and disease. (Bringmann et al., 2006; Giaume et al., 2007; Grieshaber et al., 2007). There are no available tests of glial function in human. Under pathological conditions, the model will have an increased complexity, since the PERG dynamics will depend on both the number and function of surviving RGCs, the ability of the surviving vascular system to dilate in response to the increased metabolic demand of residual neurons, and the changed properties of the glial system (glial reactivity and gliosis).(Wang et al., 2002; Neufeld & Liu, 2003; Tezel & Wax, 2003; Bringmann et al., 2006) The model can be easily applied in the clinical environment, where the non-invasive PERG technique can be combined with Optical Coherence Tomography (OCT) to have surrogate measure of number of RGC (thickness of retinal nerve fiber layer and/or volume RGC layer in the macular area) and Laser Doppler Blood Flow (LDBF) to measure changes of blood flow at retinal and optic nerve head level. (Riva et al., 2005) Artificially reducing the initial amplitude by progressively decreasing stimulus contrast causes progressive reduction of magnitude of delta (delta become less negative) until to reaching a zero value but without inverting its sign even for initial amplitudes close to the noise level. The rationale for reducing stimulus contrast is that of submaximally stimulating RGC, thereby generating less metabolic demand that requires less energy supply from blood flow. Reduction in the amount of amplitude decline with decreasing contrast is therefore expected. (Porciatti et al., 2005) The phenomenon of amplitude enhancement at high contrast is not expected, and cannot be mimicked in subjects with initially high PERG amplitude by reducing contrast. However, it may occur in some subjects with initially low PERG amplitude. Since these individuals do not have clinically evident indication of eye or systemic disease, as established by routine ophthalmological examination, the phenomenon of enhancement appears to be a physiological property of some subjects whose initial PERG amplitude is relatively lower than that of the majority of subjects in their age range, and this is associated to positive deltas. Our energy budget model accounts for this phenomenon. Age-related changes in pupil size do not significantly change PERG amplitude (Porciatti et al., 1992) and are not expected to play a significant role in the effect. Fig. 8 also shows that age (and associated miosis) is not related to the delta amplitude and phase.

Preliminary results obtained by systematic application of the present PERG protocol in patients with early glaucoma indicate that the characteristics of PERG amplitude decline/enhancement may substantially depart from those described here, suggesting impairment of one or more functional players. This protocol may be usefully extended to a variety of retinal and optic nerve disease to disclose abnormalities of the PERG temporal dynamics and monitoring their changes with time without or with pharmacological treatments. Our model may also contribute a working hypothesis to develop a mechanistic model of neurovascular and neurometabolic coupling in the brain.

Acknowledgements

We wish to thank Antonello Fadda PhD, and William Buchser, PhD, for their generous contribution of ideas and criticism to the development of the energy-budget model.

Support: NIH-NEI RO1 EY014957, NIH center grant P30-EY014801, unrestricted grant from Research to Prevent Blindness.

Footnotes

Conflict of interest: None

References

  1. Afra J, Cecchini AP, De Pasqua V, Albert A, Schoenen J. Visual evoked potentials during long periods of pattern-reversal stimulation in migraine. Brain. 1998;121(Pt 2):233–241. doi: 10.1093/brain/121.2.233. [DOI] [PubMed] [Google Scholar]
  2. Albrecht DG. Visual cortex neurons in monkey and cat: effect of contrast on the spatial and temporal phase transfer functions. Vis Neurosci. 1995;12:1191–1210. doi: 10.1017/s0952523800006817. [DOI] [PubMed] [Google Scholar]
  3. Albrecht DG, Geisler WS, Frazor RA, Crane AM. Visual cortex neurons of monkeys and cats: temporal dynamics of the contrast response function. J Neurophysiol. 2002;88:888–913. doi: 10.1152/jn.2002.88.2.888. [DOI] [PubMed] [Google Scholar]
  4. Ames A., 3rd CNS energy metabolism as related to function. Brain Res Brain Res Rev. 2000;34:42–68. doi: 10.1016/s0165-0173(00)00038-2. [DOI] [PubMed] [Google Scholar]
  5. Araque A, Carmignoto G, Haydon PG. Dynamic signaling between astrocytes and neurons. Annu Rev Physiol. 2001;63:795–813. doi: 10.1146/annurev.physiol.63.1.795. [DOI] [PubMed] [Google Scholar]
  6. Attwell D, Laughlin SB. An energy budget for signaling in the grey matter of the brain. J Cereb Blood Flow Metab. 2001;21:1133–1145. doi: 10.1097/00004647-200110000-00001. [DOI] [PubMed] [Google Scholar]
  7. Baccus SA, Meister M. Fast and slow contrast adaptation in retinal circuitry. Neuron. 2002;36:909–919. doi: 10.1016/s0896-6273(02)01050-4. [DOI] [PubMed] [Google Scholar]
  8. Bach M, Hawlina M, Holder GE, Marmor MF, Meigen T, Vaegan, Miyake Y. Standard for pattern electroretinography. International Society for Clinical Electrophysiology of Vision. Doc Ophthalmol. 2000;101:11–18. doi: 10.1023/a:1002732114721. [DOI] [PubMed] [Google Scholar]
  9. Bach M, Hoffmann MB. Update on the pattern electroretinogram in glaucoma. Optom Vis Sci. 2008;85:386–395. doi: 10.1097/OPX.0b013e318177ebf3. [DOI] [PubMed] [Google Scholar]
  10. Brigell MG, Peachey NS, Seiple WH. Pattern electroretinogram threshold does not show contrast adaptation. Invest Ophthalmol Vis Sci. 1987;28:1614–1616. [PubMed] [Google Scholar]
  11. Bringmann A, Pannicke T, Grosche J, Francke M, Wiedemann P, Skatchkov SN, Osborne NN, Reichenbach A. Muller cells in the healthy and diseased retina. Prog Retin Eye Res. 2006;25:397–424. doi: 10.1016/j.preteyeres.2006.05.003. [DOI] [PubMed] [Google Scholar]
  12. Chander D, Chichilnisky EJ. Adaptation to temporal contrast in primate and salamander retina. J Neurosci. 2001;21:9904–9916. doi: 10.1523/JNEUROSCI.21-24-09904.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. De Los Santos R, Da Rocha Lima B, Ventura LM, Porciatti V. Physiological Fluctuations of Pattern Electroretinogram Associated With Presentation of Steady-State Pattern Stimuli. Invest Ophthalmol Vis Sci. 2006;47:4012. [Google Scholar]
  14. Drasdo N, Thompson DA, Arden GB. A comparison of pattern ERG amplitudes and nuclear layer thickness in different zones of the retina. Clin Vision Sciences. 1990;5:415–420. [Google Scholar]
  15. Fellin T, Carmignoto G. Neurone-to-astrocyte signalling in the brain represents a distinct multifunctional unit. J Physiol. 2004;559:3–15. doi: 10.1113/jphysiol.2004.063214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fields RD, Burnstock G. Purinergic signalling in neuron-glia interactions. Nat Rev Neurosci. 2006;7:423–436. doi: 10.1038/nrn1928. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Giaume C, Kirchhoff F, Matute C, Reichenbach A, Verkhratsky A. Glia: the fulcrum of brain diseases. Cell Death Differ. 2007;14:1324–1335. doi: 10.1038/sj.cdd.4402144. [DOI] [PubMed] [Google Scholar]
  18. Grieshaber MC, Orgul S, Schoetzau A, Flammer J. Relationship between retinal glial cell activation in glaucoma and vascular dysregulation. J Glaucoma. 2007;16:215–219. doi: 10.1097/IJG.0b013e31802d045a. [DOI] [PubMed] [Google Scholar]
  19. Heinrich SP, Bach M. Adaptation dynamics in pattern-reversal visual evoked potentials. Doc Ophthalmol. 2001a;102:141–156. doi: 10.1023/a:1017509717071. [DOI] [PubMed] [Google Scholar]
  20. Heinrich TS, Bach M. Contrast adaptation in human retina and cortex. Invest Ophthalmol Vis Sci. 2001b;42:2721–2727. [PubMed] [Google Scholar]
  21. Hess RF, Baker CL., Jr. Human pattern-evoked electroretinogram. J Neurophysiol. 1984;51:939–951. doi: 10.1152/jn.1984.51.5.939. [DOI] [PubMed] [Google Scholar]
  22. Janz C, Heinrich SP, Kornmayer J, Bach M, Hennig J. Coupling of neural activity and BOLD fMRI response: new insights by combination of fMRI and VEP experiments in transition from single events to continuous stimulation. Magn Reson Med. 2001;46:482–486. doi: 10.1002/mrm.1217. [DOI] [PubMed] [Google Scholar]
  23. Kageyama GH, Wong-Riley MT. The histochemical localization of cytochrome oxidase in the retina and lateral geniculate nucleus of the ferret, cat, and monkey, with particular reference to retinal mosaics and ON/OFF-center visual channels. J Neurosci. 1984;4:2445–2459. doi: 10.1523/JNEUROSCI.04-10-02445.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kaplan E, Benardete E. The dynamics of primate retinal ganglion cells. Prog Brain Res. 2001;134:17–34. doi: 10.1016/s0079-6123(01)34003-7. [DOI] [PubMed] [Google Scholar]
  25. Kofuji P, Newman EA. Potassium buffering in the central nervous system. Neuroscience. 2004;129:1045–1056. doi: 10.1016/j.neuroscience.2004.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lennie P. The cost of cortical computation. Curr Biol. 2003;13:493–497. doi: 10.1016/s0960-9822(03)00135-0. [DOI] [PubMed] [Google Scholar]
  27. Logean E, Falsini B, Riva CE. Optic nerve head blood flow responses elicited using pattern contrast reversal checkerboard stimuli. IOVS. 2002;43:3314. [Google Scholar]
  28. Logean E, Geiser MH, Riva CE. Laser Doppler instrument to investigate retinal neural activity-induced changes in optic nerve head blood flow. Optics and Lasers in Engineering. 2005;43:591–602. [Google Scholar]
  29. Logothetis NK. The neural basis of the blood-oxygen-level-dependent functional magnetic resonance imaging signal. Philos Trans R Soc Lond B Biol Sci. 2002;357:1003–1037. doi: 10.1098/rstb.2002.1114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Logothetis NK, Wandell BA. Interpreting the BOLD signal. Annu Rev Physiol. 2004;66:735–769. doi: 10.1146/annurev.physiol.66.082602.092845. [DOI] [PubMed] [Google Scholar]
  31. Maffei L, Fiorentini A. Electroretinographic responses to alternating gratings before and after section of the optic nerve. Science. 1981;211:953–955. doi: 10.1126/science.7466369. [DOI] [PubMed] [Google Scholar]
  32. McCulloch DL, Van Boemel GB, Borchert MS. Comparisons of contact lens, foil, fiber and skin electrodes for patterns electroretinograms. Doc Ophthalmol. 1997;94:327–340. doi: 10.1007/BF02580858. [DOI] [PubMed] [Google Scholar]
  33. Metea MR, Newman EA. Glial cells dilate and constrict blood vessels: a mechanism of neurovascular coupling. J Neurosci. 2006;26:2862–2870. doi: 10.1523/JNEUROSCI.4048-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Morrone C, Fiorentini A, Bisti S, Porciatti V, Burr DC. Pattern-reversal electroretinogram in response to chromatic stimuli: II. Monkey. Vis Neurosci. 1994a;11:873–884. doi: 10.1017/s0952523800003837. [DOI] [PubMed] [Google Scholar]
  35. Morrone C, Porciatti V, Fiorentini A, Burr DC. Pattern-reversal electroretinogram in response to chromatic stimuli: I. Humans. Vis Neurosci. 1994b;11:861–871. doi: 10.1017/s0952523800003825. [DOI] [PubMed] [Google Scholar]
  36. Neufeld AH, Liu B. Glaucomatous optic neuropathy: when glia misbehave. Neuroscientist. 2003;9:485–495. doi: 10.1177/1073858403253460. [DOI] [PubMed] [Google Scholar]
  37. Newman EA. A purinergic dialogue between glia and neurons in the retina. Novartis Found Symp. 2006;276:193–202. discussion 202-197, 233-197, 275-181. [PubMed] [Google Scholar]
  38. Ohzawa I, Sclar G, Freeman RD. Contrast gain control in the cat visual cortex. Nature. 1982;298:266–268. doi: 10.1038/298266a0. [DOI] [PubMed] [Google Scholar]
  39. Pellerin L. Lactate as a pivotal element in neuron-glia metabolic cooperation. Neurochem Int. 2003;43:331–338. doi: 10.1016/s0197-0186(03)00020-2. [DOI] [PubMed] [Google Scholar]
  40. Pellerin L, Magistretti PJ. Neuroenergetics: calling upon astrocytes to satisfy hungry neurons. Neuroscientist. 2004;10:53–62. doi: 10.1177/1073858403260159. [DOI] [PubMed] [Google Scholar]
  41. Porciatti V. The mouse pattern electroretinogram. Doc Ophthalmol. 2007;115:145–153. doi: 10.1007/s10633-007-9059-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Porciatti V, Bonanni P, Fiorentini A, Guerrini R. Lack of cortical contrast gain control in human photosensitive epilepsy. Nat Neurosci. 2000;3:259–263. doi: 10.1038/72972. [DOI] [PubMed] [Google Scholar]
  43. Porciatti V, Burr DC, Morrone MC, Fiorentini A. The effects of aging on the pattern electroretinogram and visual evoked potential in humans. Vision Res. 1992;32:1199–1209. doi: 10.1016/0042-6989(92)90214-4. [DOI] [PubMed] [Google Scholar]
  44. Porciatti V, Di Bartolo E, Nardi N, Fiorentini A. Responses to chromatic and luminance contrast in glaucoma: a psychophysical and electrophysiological study. Vision Res. 1997;37:1975–1987. doi: 10.1016/s0042-6989(97)00018-7. [DOI] [PubMed] [Google Scholar]
  45. Porciatti V, Sartucci F. Retinal and cortical evoked responses to chromatic contrast stimuli. Specific losses in both eyes of patients with multiple sclerosis and unilateral optic neuritis. Brain. 1996;119:723–740. doi: 10.1093/brain/119.3.723. [DOI] [PubMed] [Google Scholar]
  46. Porciatti V, Sorokac N, Buchser W. Habituation of retinal ganglion cell activity in response to steady state pattern visual stimuli in normal subjects. Invest Ophthalmol Vis Sci. 2005;46:1296–1302. doi: 10.1167/iovs.04-1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Porciatti V, Ventura LM. Normative data for a user-friendly paradigm for pattern electroretinogram recording. Ophthalmology. 2004;111:161–168. doi: 10.1016/j.ophtha.2003.04.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Regan D. Human brain electrophysiology. Evoked potentials and evoked magnetic fields in science and medicine. Elsevier; New York: 1989. [Google Scholar]
  49. Reichenbach A, Stolzenburg JU, Eberhardt W, Chao TI, Dettmer D, Hertz L. What do retinal muller (glial) cells do for their neuronal ‘small siblings’? J Chem Neuroanat. 1993;6:201–213. doi: 10.1016/0891-0618(93)90042-3. [DOI] [PubMed] [Google Scholar]
  50. Riva CE, Logean E, Falsini B. Visually evoked hemodynamical response and assessment of neurovascular coupling in the optic nerve and retina. Prog Retin Eye Res. 2005;24:183–215. doi: 10.1016/j.preteyeres.2004.07.002. [DOI] [PubMed] [Google Scholar]
  51. Roy CS, Sherrington CS. On the regulation of the blood supply of the brain. J Physiol. 1890;11:85–108. doi: 10.1113/jphysiol.1890.sp000321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sappey-Marinier D, Calabrese G, Fein G, Hugg JW, Biggins C, Weiner MW. Effect of photic stimulation on human visual cortex lactate and phosphates using 1H and 31P magnetic resonance spectroscopy. J Cereb Blood Flow Metab. 1992;12:584–592. doi: 10.1038/jcbfm.1992.82. [DOI] [PubMed] [Google Scholar]
  53. Sartucci F, Orlandi G, Lucetti C, Bonuccelli U, Murri L, Orsini C, Porciatti V. Changes in pattern electroretinograms to equiluminant red-green and blue-yellow gratings in patients with early Parkinson’s disease. J Clin Neurophysiol. 2003;20:375–381. doi: 10.1097/00004691-200309000-00010. [DOI] [PubMed] [Google Scholar]
  54. Shapley R, Victor JD. The contrast gain control of the cat retina. Vision Res. 1979;19:431–434. doi: 10.1016/0042-6989(79)90109-3. [DOI] [PubMed] [Google Scholar]
  55. Singh M, Kim S, Kim TS. Correlation between BOLD-fMRI and EEG signal changes in response to visual stimulus frequency in humans. Magn Reson Med. 2003;49:108–114. doi: 10.1002/mrm.10335. [DOI] [PubMed] [Google Scholar]
  56. Snodderly DM, Weinhaus RS, Choi JC. Neural-vascular relationships in central retina of macaque monkeys (Macaca fascicularis) J Neurosci. 1992;12:1169–1193. doi: 10.1523/JNEUROSCI.12-04-01169.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Tezel G, Wax MB. Glial modulation of retinal ganglion cell death in glaucoma. J Glaucoma. 2003;12:63–68. doi: 10.1097/00061198-200302000-00014. [DOI] [PubMed] [Google Scholar]
  58. Wang L, Cioffi GA, Cull G, Dong J, Fortune B. Immunohistologic evidence for retinal glial cell changes in human glaucoma. Invest Ophthalmol Vis Sci. 2002;43:1088–1094. [PubMed] [Google Scholar]
  59. Wang L, Dong J, Cull G, Fortune B, Cioffi GA. Varicosities of intraretinal ganglion cell axons in human and nonhuman primates. Invest Ophthalmol Vis Sci. 2003;44:2–9. doi: 10.1167/iovs.02-0333. [DOI] [PubMed] [Google Scholar]
  60. Winkler BS, Arnold MJ, Brassell MA, Puro DG. Energy metabolism in human retinal Muller cells. Invest Ophthalmol Vis Sci. 2000;41:3183–3190. [PMC free article] [PubMed] [Google Scholar]
  61. Zapf HR, Bach M. The contrast characteristic of the pattern electroretinogram depends on temporal frequency. Graefes Arch Clin Exp Ophthalmol. 1999;237:93–99. doi: 10.1007/s004170050201. [DOI] [PubMed] [Google Scholar]
  62. Zonta M, Angulo MC, Gobbo S, Rosengarten B, Hossmann KA, Pozzan T, Carmignoto G. Neuron-to-astrocyte signaling is central to the dynamic control of brain microcirculation. Nat Neurosci. 2003;6:43–50. doi: 10.1038/nn980. [DOI] [PubMed] [Google Scholar]
  63. Zrenner E. The physiological basis of the pattern electroretinogram. In: Osborne N, Chader G, editors. Progress in Retinal Research. Pergamon Press; Oxford: 1990. pp. 427–464. [Google Scholar]

RESOURCES