Retinal Origins of the Primate Multifocal ERG: Implications for the Human Response

Abstract

purpose. To better understand the cellular contributions to the human multifocal ERG (mfERG), rhesus monkey and human mfERGs were recorded using the same stimulus conditions. The monkey mfERGs were recorded before and after injections of pharmacologic agents known to selectively block activity of particular cells and circuits in the retina.

methods. Photopic mfERGs were recorded with Dawson-Trick-Litzkow (DTL) fiber electrodes from 16 eyes of 10 anesthetized adult rhesus monkeys (Macaca mulatta) and from 4 normal humans. The display consisted of 103 equal-sized hexagons within 17° of the fovea. Monkey mfERGs were obtained before and after inner retinal responses were suppressed with intravitreal injections of tetrodotoxin (TTX), TTX+N-methyl-d-aspartic acid (NMDA), TTX+NMDA with the γ-aminobutyric acid (GABA_A&C) antagonist picrotoxin (PTX), or the inhibitory amino acid GABA and after l-2 amino-4-phosphonobutyric acid (APB) to block signal transmission to ON-bipolar cells. Finally, a combination of APB and cis-2,3 piperidine dicarboxylic acid (PDA) was used to isolate the contributions from the cone photoreceptors.

results. TTX, which blocks sodium-based action potentials, removes a large contribution from the monkey’s mfERG, but it does not remove all inner retinal influences. After administration of TTX, the mfERG is further modified by the addition of NMDA. TTX+NMDA, TTX+NMDA+PTX, or GABA alone have similar effects, suggesting that, at the concentrations used, they are largely removing the inner retinal contributions. After removing the inner retinal influences, the monkey’s mfERG is mainly composed of ON- and OFF-bipolar contributions, as revealed after APB and PDA were injected. The leading edge of the first negative potential (N1) is largely shaped by the initial hyperpolarization of the OFF-bipolar cells. The photoreceptors also contribute to the leading edge of N1, but this contribution is small, except in the central 6°. The depolarization of the ON-bipolars and the recovery of the OFF-bipolars contribute to the leading edge of the major positive component (P1), with the recovery of the ON-bipolars being the dominant influence on the trailing edge. The waveform of the human mfERG most closely resembles the rhesus monkey’s mfERG after administration of TTX.

conclusions. The monkey’s mfERG is shaped by large contributions from ON- and OFF-bipolar cells, combined with both spiking and nonspiking inner retinal contributions, and a small contribution from the photoreceptors. In comparison, the human mfERG resembles the monkey’s mfERG after reduction of inner retinal contributions. Based on the pharmacologic dissection of the monkey’s mfERG, a model of the waveform of the human mfERG is proposed. This model suggests that the waveform can be understood as a combination of overlapping ON- and OFF-bipolar cell contributions combined with smaller contributions from inner retina and photoreceptors.

The multifocal electroretinogram (mfERG) technique, developed by Sutter and Tran, ^{1 2} allows for the simultaneous recording of many focal retinal responses in a relatively brief recording period. For example, in the present study 103 human mfERG responses were recorded from the central 33° in 7 minutes. The typical parameters used assure that the responses are photopic (i.e., cone-driven). Thus, with this technique, the health of the foveal, parafoveal, and near peripheral photopic retina can be evaluated simultaneously. As a consequence, this technique is gaining wide acceptance as a tool for both diagnosing and studying diseases of the human retina under photopic conditions (see Ref. ³ for a review). How useful this technique will be in the future depends, in part, on our ability to relate changes in the waveform of the mfERG responses to particular layers and/or cells of the retina. 

Our understanding of the photopic, full-field ERG in primates has been greatly enhanced by studies using pharmacologic agents to block the activity of particular cell types. ^{4 5 6 7 8} Less is known about the cellular contributions to the mfERG. ^{9 10 11 12} In the present study, relatively standard pharmacologic dissection techniques were used to better understand the cellular contributions to mfERG of the rhesus monkey (Macaca mulatta). Recording from rabbits, Horiguchi et al. ⁹ were the first to show that both ON- and OFF-pathways contribute to the ERG response to stimuli (full-field) modulated by the temporal m-sequence of the mfERG. Recently, Hare and Ton ¹² showed that the cynomolgus monkey’s mfERG is shaped by overlapping contributions from ON- and OFF-pathways, as is the case for the full-field flash ERG. ⁵  

One objective of the present study was to describe in detail how the contributions from ON- and OFF-bipolar cells shape the waveform of the human mfERG. The inner retina makes a larger contribution to the mfERG of the rhesus monkey than it does to the human mfERG, ^{13 14} and unless the inner retinal influences are removed, the rhesus monkey’s mfERG does not show a close resemblance to the human mfERG. ^{10 11 13 15} Thus, we examined the effects of pharmacologically blocking the ON- and OFF-pathways after removing the influences of the inner retina. Further, because the shape of the human mfERG waveform is influenced by parameters of the stimulus, mfERGs were recorded from humans and monkeys under similar stimulus conditions, and the waveform of the human mfERG was compared with the monkey’s before and after injections of pharmacologic agents. Based on these comparisons, a model of the cellular contributions to the human mfERG is proposed. 

Methods

Preparation

Recordings were made from 17 eyes of 11 adult rhesus monkeys (Macaca mulatta). Animals were anesthetized intramuscularly with ketamine (20–25 mg/kg · h) and xylazine (0.8–0.9 mg/kg · h) and were treated with atropine sulfate (0.04 mg/kg, injected subcutaneously). Pupils were fully dilated to approximately 9 mm in diameter with topical tropicamide (1%) or atropine (0.5%) and phenylephrine (2.5%), and the eye to be studied was refracted retinoscopically and fitted with appropriate contact lenses. Heart rate and blood oxygen were monitored with a pulse oximeter (model 4402L; Heska Corp., Fort Collins, CO) and temperature maintained at 36.5°C to 38°C. Three animals were intubated, immobilized with intravenous pancuronium bromide (1.5–2.0 mg/kg · h), and artificially respirated during the last few hours of a terminal experiment on the second eye. Ketamine anesthesia was continued, but xylazine was no longer used. mfERG recordings obtained before and after paralysis were indistinguishable. Experimental and animal care procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were reviewed by the Institutional Animal Care Committee of the University of Houston. 

Intravitreal Injections

Intravitreal injections of 40 to 50 μL were made nasally and temporally in the globe behind the limbus with a sterile 30-gauge needle inserted through the pars plana into the vitreal cavity. Intravitreal concentrations of the pharmacologic agents were estimated by assuming that the vitreal volume is 2.1 mL. The following drugs and concentrations, all in sterile balanced salt solution, were used: Tetrodotoxin citrate (TTX: 4.8–8.4 μM), N-methyl-d-aspartic acid (NMDA: 1.4–6.4 mM), γ-aminobutyric acid (GABA: 37–55 mM) and glycine (44 mM). Picrotoxin (PTX: 0.1–0.4 mM), l-2 amino-4-phosphonobutyric acid (APB: 0.83–3.8 mM) and cis-2,3 piperidine dicarboxylic acid (PDA: 3.3–3.8 mM) also were used. These concentrations of pharmacologic agents were sufficient in other experiments in rabbits and primates to have the desired effects on the full-field flash ERG. ^{5 6 7 8 9 10} In our experiments, these effects on the full-field flash ERG were verified. Recordings were made before and at least 1 hour after injections when effects had stabilized. In a few cases (e.g., animal D), recordings were made a few weeks after injections of TTX and NMDA, and responses had returned essentially to control waveforms. As described in the results, eyes were often injected with a sequence of pharmacologic agents, with same agent or agents, in at least two and sometimes as many as six eyes. For other details, see Viswanathan et al. ⁷ and Hood et al. ¹⁰  

Stimulation

The stimulus display consisted of 103 equal-sized hexagons, each approximately 3.3° wide, in a field of approximately 35° by 33° (Fig. 1A) . The display was centered on the monkey’s fovea using an ophthalmoscopic technique. The ONH is approximately 16.5° from the fovea in the monkey ¹⁶ ; its center is approximately at the location marked “x” in Figures 1A and 1B . Recordings were made with the room lights on, and the luminance of the white and black hexagons were 200 and 15 cd/m², respectively. The surround was set to 100 cd/m². An experimental run consisted of an m-sequence with 2¹⁵ − 1 elements. Thus, each hexagon was presented approximately 32,767 times. The elements of this sequence were 13.3 ms in duration (corresponding to a screen frame rate of 75 Hz). However, the actual duration of the incremental light producing a white hexagon was shorter, decaying to 90% in approximately 2 ms. Each run required approximately 7 minutes’ total recording time, recorded as a single segment or, in some cases, broken into two equal segments. The analyses were based on the average of between one and three runs. First-order responses were analyzed by computer (VERIS [Visual Evoked Response Imaging System] software; Electro-Diagnostic Imaging [EDI]; San Mateo, CA). (For more detailed information about the multifocal technique see Refs. ^{1 2 3 17} .) 

Full-field, photopic flash ERGs ⁷ and slowed-sequence mfERGs, which show standard oscillatory potentials, ¹⁸ also were monitored in all experiments to assess drug effects and to confirm that previous findings where applicable (e.g., effects of APB and PDA ^{5 6} ). 

Recordings in Monkeys

ERGs were recorded between Dawson-Trick-Litzkow (DTL) electrodes that were placed across the center of the cornea and under a corneal contact lens of both eyes (see Ref. ¹⁹ for details); one eye was covered. The low- and high-frequency cutoffs were set at 1 and 300 Hz, and no additional notch filtering was done. When nasotemporal comparisons are made, all records are presented as if the left eye had been stimulated. 

Recordings in Humans

Recordings were also obtained in four human control subjects (age range, 22–57 years) with no known abnormalities of the visual system. DTL electrodes were used as previously described. ²⁰ The conditions were as close to those for the monkeys as was feasible. The stimulation and recording parameters were essentially identical. The 7-minute runs were broken into 16, rather than 1 or 2, equal segments to allow for blinking. The humans were seated rather than prone, without anesthetic, of course, and wore their own glasses if correction was needed. In all subjects, the recordings were from the right eye. However, for ease of comparison to the records from the monkey the human records in Figure 6 are presented as if the left eye had been stimulated. For the averaged records shown in Figure 6B , the records of the four humans were combined in the software (VERIS; EDI). Informed consent was obtained before their participation. Procedures adhered to the tenets of the Declaration of Helsinki, and the protocol was approved by the University of Houston Committee for the Protection of Human Subjects. 

Results

Control Recordings

The array of multifocal responses from one of the monkeys (VA) can be seen in Figure 1B . Figure 2A shows these responses and responses from one other monkey (M), averaged for rings around the fovea (Fig. 1C) and expressed as response per unit area (nanovolts per square degree). Only the first 60 ms of the mfERG responses are shown in this, as well as in subsequent figures, so that the positive and negative waves in the early part of the response are clearly visible. The first 120 ms of the responses from ring 1 and the average of the responses from rings 3, 4, and 5 for monkey VA are shown in Figure 2B . A similar analysis for monkey V before and after TTX+NMDA can be found in Figure 3B . Very little of interest to the analyses in this study occurred in the response beyond 60 ms (vertical dashed line). 

The mfERG waveform varies with eccentricity (Fig. 2A) . ^{2 10 11 15} This is best appreciated by comparing the response from ring 1 to the responses from rings 3 to 5. The waveforms in rings 3, 4, and 5 were very similar but differed from the waveforms in ring 1. Because this was true in all animals, for conciseness of presentation, only the responses from ring 1 and the average of the responses from rings 3 to 5 are shown in subsequent figures. The results for animal M are shown in Figure 2C (first column). Although there was considerable variability in waveforms across animals (see dashed curves in Fig. 3A ), the results to be described will show that this variability was greatly reduced by pharmacologic agents that suppress inner retinal activity. 

As previously reported, the monkey also shows striking nasotemporal variations in waveform, ^{10 11 13 15} especially prominent when mfERGs are recorded referenced to the other eye as in this study. ^{13 21} These variations can be seen in the response array in Figure 1B . The records averaged within quadrants of the field (Fig. 1D) are shown in Figure 2D for two control eyes. The quadrants are labeled based on the visual field stimulated: UTF (upper temporal field), UNF (upper nasal field), LNF (lower nasal field), and LTF (lower temporal field). Responses from the nasal fields differed in waveform from those from the temporal fields, whereas the responses from the quadrants within the nasal or temporal fields tended to produce similar waveforms. In general, in recordings from control eyes, the responses from the temporal field (nasal retina) appeared to have a more prominent secondary peak (Fig. 2D , arrows). 

TTX to Block Sodium-Based Action Potentials

TTX blocks voltage-gated sodium channels and prevents the generation of sodium-based action potentials. In the monkey, ganglion cells and some amacrine cells are known to generate action potentials. In previous work, Hood et al. ¹⁰ have shown that after treatment with TTX, the responses from different retinal regions become far more similar in shape and have a single positive peak. Figure 2C (second column) and Figure 2D (third column) show the responses from monkey M, of the previous study, after TTX. After injection, the responses were larger, and the responses from ring 1 were more similar in waveform to the responses from rings 3 to 5 (Fig. 2C) . Similarly, the responses from the four quadrants were far more similar in waveform than they were in the control records (Fig. 2D) . TTX removes nearly all the nasotemporal variation in waveform. ^{10 11} We observed essentially the same effects in five eyes of five monkeys treated with TTX. 

Attempts to Suppress All Inner Retinal Activity

After injection of TTX, inner retinal cells (i.e., amacrine and ganglion cells) still generate local potentials. Figure 3A shows the responses after attempts to suppress all inner retinal activity with TTX+NMDA, TTX+NMDA+PTX, or GABA. NMDA, a glutamate agonist, depolarizes cells that have functional NMDA receptors (i.e., ganglion cells and at least some types of amacrine cells; see Ref. ²² for a review). As in Figure 2 and all subsequent figures, only the first 60 ms of the records are shown. Figure 3B shows the first 120 ms of the responses for V(OD) to illustrate that very little response occurs beyond 60 ms, particularly in an eye in which the inner retinal activity has been removed. 

We thought that TTX+NMDA would suppress nearly all inner retinal activity. ²³ However, some inner retinal activity may have remained after TTX+NMDA, because NMDA receptors are not present on all amacrine cells. Further, we examined the effects of TTX+NMDA in six eyes in six animals, and in our hands, the NMDA was more effective in some than others in removing inner retinal influences, as indicated by the oscillatory potentials (OPs) on the leading edge of the flash response (not shown). These OPs are believed to originate from the inner retina. ²⁴ To assure that the inner retinal influences were suppressed, we attempted to block all inner retinal activity in two other ways. In three animals, including Z (in Fig. 3A ), picrotoxin (PTX) was added after TTX+NMDA. PTX blocks GABA_A and GABA_C receptors. In animal VA (in Fig. 3A ) and one other animal, a control eye was treated only with GABA. GABA, an inhibitory neurotransmitter, suppresses inner retinal contributions to the ERG. ²⁵ In two additional eyes, we observed the effects of glycine, another inhibitory neurotransmitter known to suppress inner retinal contributions to the ERG. ²⁵ Results again were similar, but responses were slowed (data not shown). Although GABA or the combination of TTX+NMDA+PTX should suppress all inner retinal activity, GABA and PTX could also affect the outer retina. ^{26 27 28} Despite the differences in the mode of action of these drugs, the effects of TTX+NMDA+PTX and GABA were similar to the effects produced by TTX+NMDA when TTX+NMDA had its largest effect (V in Fig. 3A and 3D in Fig. 4 ). It should be noted that we did not attempt to describe separately the additional effects for PTX, because of the variability of TTX+NMDA results. 

After the inner retina was suppressed, all the responses in Figure 3A had a triphasic negative-positive-negative waveform that was quite similar in all three monkeys. The similarity is best seen in Figure 3C (first column) where the responses from rings 3 to 5, normalized to have the same trough-to-peak amplitude, are shown for all three animals. The three waveforms are similar and differ in similar characteristic ways from the waveform after only TTX. The middle column of Figure 3C contains the normalized post-TTX records from Figure 2C and from another animal whose post-TTX data were published previously. ¹¹ In the third column in Figure 3C , these TTX records are shown together with those in the first column. Inner retinal contributions that remained after TTX (i.e., those sensitive to NMDA, PTX, or GABA) contributed to the shape of both the leading and trailing edges (Fig 3C , arrows) of the prominent positive potential (P1) of the mfERG. Removing these inner retinal influences removed a negative potential that produced a delay in the leading edge of P1 and a positive potential that produced a “shelf” on the trailing edge of P1. 

Application of APB to Block the ON-Bipolar Contribution

The similarity of the waveforms in Figure 3A suggests that TTX+NMDA removed most, if not all, the influence of the inner retina on the monkey’s mfERG. Thus, after TTX+NMDA, the mfERG that remained was mainly a combination of the contributions from the cones and the ON- and OFF-bipolar cells (see Fig. 4A ). To separate these influences, responses were recorded after injection of APB. APB is a glutamate analogue that blocks transmission from the photoreceptors to the ON-bipolar cells ²⁹ and eliminates the photopic b-wave in monkeys. ^{4 5 30} Figure 4B summarizes the results from monkey D, which received an injection of APB after TTX+NMDA. The records in red are the responses after APB and should be a combination of contributions from the photoreceptors and the OFF-bipolar cells (see Fig. 4A ), probably shaped by horizontal cell feedback. Although the horizontal cells may also contribute a negative potential directly to the post-APB response, there is little evidence for a direct horizontal cell contribution. ⁶ The component removed by APB, shown in blue in Figure 4 , should be the contribution of the ON-bipolar cells, without the influence of the inner retina, but probably also shaped by horizontal cell feedback. As indicated in Figure 4A , it was obtained by subtracting the responses after injection of APB (shown in red) from the responses for the TTX+NMDA condition (shown in green). Figure 4C shows the results in the animal (VA in Figs. 1B 2A 2D 3A ) in which inner retinal activity was first suppressed with GABA. 

The general effects of APB shown in the current study were confirmed in three additional animals. In one animal, the effects of NMDA were not complete in most of the field but after APB the responses were similar to D in Figure 4B , except for some oscillatory activity after the peak of the prominent positive component. In another (M), APB was injected after TTX. The response after APB contained residual inner retinal activity, as indicated by oscillatory potentials. In the third animal, GABA and glycine were injected before APB. The responses were somewhat delayed, but the effects were similar. In addition, APB was injected in three animals before the inner retina was blocked, and, as in the control eyes, the response removed by the drug was too oscillatory to see clearly the waveform of the underlying inner nuclear layer. 

Injection of APB and PDA, to Isolate the Cone Photoreceptor Contribution

To isolate the contributions from the photoreceptors, three monkeys were treated with both APB and PDA. PDA is a glutamate analogue that blocks transmission to OFF-bipolar and horizontal cells as well as to inner retinal neurons. ^{5 25 29} In one animal (monkey V), PDA was injected into a control eye and followed by injection of APB. Figure 4D shows the results for the control condition (gray), after PDA (blue), and after PDA and APB (black). The thin black records in the third column of Figure 4B show the PDA+APB results for animal V (from Fig. 4D ), and the bold black records show the results for monkey D when PDA was injected after TTX+NMDA and APB. The results for a third animal, treated first with APB and then PDA, fell between these two sets of curves. Although there was some variability among animals because of the overall amplitude of the signal and/or the relative completeness of the action of the drugs, the conclusion is clear. The photoreceptors contributed relatively little to the mfERG of the outer rings, as might be expected from earlier results. ^{5 12} In the center ring, however, the photoreceptor contribution helped shape the leading edge of the mfERG response (see also the recent paper by Hare and Ton ¹² ). 

Estimates of the ON- and OFF-Bipolar Cell Contributions

The contributions removed by APB after TTX+NMDA (Fig. 4B) or after GABA (Fig. 4C) provide an estimate of the waveform of the ON-bipolar contribution to the monkey’s mfERG. These APB-sensitive contributions, normalized to have the same peak amplitude, are shown in Figure 5A for the two animals as bold curves (gray: monkey D; black: monkey VA). The agreement was reasonably good, and these records provided an estimate of the waveform of the contribution of the ON-bipolar cells. The summed contribution from the ON-bipolar cells began approximately 9 ms after light onset and peaked at approximately 23 to 25 ms. 

After PDA, only the receptor-to-ON-bipolar synapse functions. ⁵ Thus, the record in blue in Figure 4D should be a combination of the contributions from the ON-bipolar cells and the cone photoreceptors. By subtracting the photoreceptor response isolated with PDA+APB (shown in black in Fig. 4D ), an estimate of the isolated ON-bipolar contributions can be obtained. The thin black lines in Figure 5A show these responses normalized to have the same peak amplitude as the APB-sensitive responses. The leading edge of these responses agreed with those from the APB-sensitive responses, but the trailing edges of the positive component were broader. This is consistent with the results of similar full-field ERG experiments showing a discrepancy between the ON-bipolar contributions isolated with APB and PDA. ⁵ Because PDA removes the lateral interactions and adaptive influences of the horizontal cells, it is not surprising that the trailing edge of the responses differed. Although other explanations for these differences are possible, ⁵ the key point is that the APB-sensitive component was the better estimate of the ON-bipolar contribution to the mfERG, because APB affects only the receptor–ON-bipolar synapse. 

The records after APB (red curves) in Figure 4 provide an estimate of the combined contribution of the OFF-bipolar cells plus the photoreceptors. The normalized responses for the two animals in Figure 4 are shown together in Figure 5B . The agreement was good. To obtain an estimate of the waveform of the contribution of the OFF-bipolar cells, the receptor contribution must be subtracted. This subtraction was performed for the responses from rings 3 to 5 from monkey D (see Fig. 4B , third column). The first column of Figure 5C shows the isolated OFF-bipolar response (black) from monkey D compared with the ON-bipolar contribution from the same animal (from Fig. 5A ). (A similar analysis for ring 1 was not performed, because of the range of the estimated receptor contributions to this ring. See rightmost column in Figure 4B .) The contribution of the OFF-bipolar cells started approximately 2 ms earlier and reached its trough approximately 8 ms sooner than the contribution of the ON-bipolars. This can be seen more easily in the second column of Figure 5C , in which the OFF-bipolar contribution was reversed in polarity and normalized to have the same peak amplitude as the ON-bipolar response. Nearly identical timing differences can be seen in the case of animal VA in Figure 4C . 

As can be seen in Figures 4B and 4C , both the OFF- and ON-bipolar contributions decreased in amplitude with eccentricity, as should be expected on the basis of change in cell densities. However, there is little support for the recent suggestion based on the b- and d-waves of the human mfERG that the ratio of ON- to OFF-bipolar contribution decreases with eccentricity. ^{31 32} In general, the peak hyperpolarization of the OFF-bipolar cells (plus a small receptor contribution) was equal to, or slightly smaller than, the peak depolarization of the ON-bipolar cells at all eccentricities. 

The Second-Order Kernel

Although a full treatment of the second-order kernel is beyond the scope of this study, some general observations about inner and outer retinal contributions are possible. Briefly, the second-order kernel is a measure of how the mfERG response is influenced by proceeding flashes. ^{3 17} The records in Figure 6A , from the same recordings that were illustrated in Figure 3A , are the second-order kernels before (dashed) and after (solid) the removal of inner retinal influences. These are the so-called first slice of the second order and show the effects of the immediately preceding flash. Hare and Ton ¹² reported previously that TTX removes much, but not all, of the second-order kernel. It is clear in Figure 6A that removing inner retinal activity, spiking and otherwise, greatly modified, but by no means eliminated, the second-order kernel. 

Further, our recordings after removal of inner retinal activity indicate that the second-order kernel had contributions from both ON- and OFF-bipolar cells. The gray records in Figure 6B are the records from Figure 6A (rightmost column) after injection of GABA. The black records show the second-order kernel after APB (left column) and for the response removed by APB (right column). It is important to keep in mind that the second-order responses are thought to be the consequence of temporal adaptation processes in the retina. Presumably, these records show, in the absence of inner retinal effects, the effects of temporal adaptation on the responses of the OFF- (left column) and ON- (right column) bipolar cells. 

The mfERG from Human Control Subjects

The waveforms of the mfERGs from the control animals (Figs. 1B 2) appeared to be different from those previously published for humans. The differences were especially obvious in the case of ring 1. Because of the marked difference in waveform, mfERGs were obtained from four human control subjects under conditions that were as similar as possible to those for the monkeys (see the Methods section). The responses from one of the subjects and the average response of the four subjects are shown in Figures 7A and 7B , respectively. These responses can be directly compared with the array in Figure 1B from monkey VA. In general, the human records were noisier. However, the most striking differences were to be found in a comparison of waveforms. To make this comparison, the responses were grouped as in Figure 1C for the monkey. The responses from Figures 7A and 7B for rings 1 and 3 to 5 are shown in Figure 7C for subject S1 and for the average of the four subjects. 

The waveforms of the human responses from ring 1 in Figure 7C differed markedly from those of control rhesus monkey eyes (Fig. 2C) . The responses from rings 3 to 5 also differed in waveform, although the difference was not as great. However, the human mfERG responses more closely resembled the monkey’s mfERG responses in eyes treated with TTX (e.g., Fig. 2C ). To appreciate these similarities, in the first two columns of Figure 8 the responses averaged from the four subjects (Fig. 7C , rightmost column) are superimposed on the TTX records from monkey M from Figure 2 and from a second monkey (E) whose post-TTX data were published previously. ¹¹ The records from the humans (black curves) were scaled in amplitude so that the trough-to-peak amplitude was approximately the same as for the records after TTX (gray curves). Although the match was not perfect, the waveforms of the human and TTX-treated mfERG responses were dominated by N1 and P1 waves, and shared most of the key features for responses from rings 1 and 3 to 5. (See the figure caption for the scaling factors and the figure for the definition of N1 and P1.) By contrast, there are marked differences between the human mfERGs and the control records from these same animals (compare Figs. 1 and 2 with Fig. 7 ). The differences between the waveforms from human and monkey control subjects, in contrast to the relative similarity after TTX, reinforces the point that the TTX-sensitive contributions are largely responsible for the difference in appearance between the human and rhesus monkey mfERGs. 

However, the waveform of the human mfERG was not exactly like the waveform of the mfERG from the TTX-treated monkey. After TTX the nasotemporal variation in waveform disappeared in the monkey (Fig. 2D) . A similar analysis is shown in Figure 7D for the records from the human control subjects. The nasotemporal differences were more subtle than those in the records from the monkeys (Fig. 2D) , but nasotemporal differences are present in the human mfERG ^{11 33 34 35} and focal macular ERG. ^{36 37} In the case of the mfERG, these nasotemporal differences bear a qualitative similarity to those seen in the monkey. ^{11 38} The similarity is best seen in the records averaged across the four subjects and presented in the rightmost column in Figure 7D . As in the monkeys’ records (Fig. 2D) , the nasal and temporal fields showed different waveforms. As suggested by Sutter and Bearse, ¹⁴ the nasotemporal variations in the human mfERG were probably due to a small optic nerve head component (ONHC) that influences the shape of both the leading and trailing edges of P1. However, we cannot rule out the possibility that other inner retinal contributions were responsible. The species differences observed in this study are consistent with other evidence, discussed later, for a relatively smaller inner retinal contribution, including a smaller ONHC, to the human mfERG. ^{13 14}  

Further evidence of an inner retinal contribution to the human mfERG can be found in the comparison to the monkey’s records after TTX+NMDA. These comparisons can be seen in the rightmost column of Figure 8 . Note that the human records, similar to the TTX-treated records, have a shelf after the first prominent peak P1 (Fig. 8 , arrow in third column). This shelf was removed by NMDA, as well as by the other ways in which inner retinal activity was suppressed (Fig. 3) . It is interesting that it is diminished in some patients with glaucoma and diabetes as well. ^{11 38 39}  

Discussion

The mfERG, like the traditional full-field ERG, reflects the sum of the contributions from various retinal cell types. The objective in this study was to better understand the cellular basis of the monkey’s photopic mfERG and to use this information to better understand the cellular basis of the human mfERG. 

Cellular Contributions to the Monkey’s mfERG

Contributions from the Inner Retina.

The waveform of the monkey’s mfERG was influenced by the inner retina (i.e., the amacrine cells, the ganglion cells, and perhaps the inner plexiform cells, and their connections). The extent of inner retinal influences on the monkey’s mfERG depends to a great extent on the anesthetic and recording conditions. Bipolar recording, ^{13 21} 60-Hz filtering, ⁴⁰ and anesthetic agents ²¹ can reduce the amount of inner retinal activity recorded. Bipolar recording also removes some outer retinal activity. ¹³ The conditions used in the present study (i.e., monopolar recording referenced to the other eye, recording without line filtering, and anesthesia with ketamine) should minimize the removal of signals from the mfERG, especially signals from the inner retina. For example, the nasotemporal variations and the prominent OPs that are associated with inner retinal activity were clearly visible in our recordings, compared with bipolar recording and/or recording under isoflurane anesthesia. ^{12 13 21}  

TTX blocks the action potentials produced by the ganglion cells and some amacrine cells. It removes a large inner retinal contribution to the monkey’s mfERG ^{10 11 41} (Fig. 2) and the full-field ERG as well. ^{7 8} Further, this inner retinal contribution removed by TTX is larger in the central retina. ^{10 12} Hare and Ton ¹² also reported that TTX affects the monkey’s mfERG, although the effects they observed were more subtle. Although species differences cannot be ruled out (Hare and Ton studied cynomolgus monkeys) it is likely that their control recordings had smaller inner retinal contributions due, at least in part, to the use of Burian-Allen bipolar electrodes and line filtering (as discussed earlier). 

TTX essentially eliminated the nasotemporal variation seen in the control records. ^{10 11 41} However, the large contribution removed by TTX has at least two components. ^{13 41} The first component has the characteristics of the ONHC extracted from the human mfERG by Sutter and Bearse ¹⁴ and is largely responsible for the nasotemporal variation in waveform. ¹³ The second is a high-frequency local component or components (HFC) ⁴¹ that can include a contribution from the adaptive effects of previous flashes. (See a discussion of the “induced version of the second-order response in the first-order response,” Refs. ^{3 17} . Note also that the ONHC may include high-frequency or oscillatory activity ³⁵ that contributes to the nasotemporal variations in the OPs of the human full-field ERG, Ref ⁴¹ and Fortune B, submitted for publication, 2002.) After TTX, an inner retinal contribution to the monkey’s mfERG remained that was largely, if not entirely, removed by NMDA. After injection of TTX+NMDA, relatively little was changed by adding PTX or by simply suppressing all inner retinal activity with GABA. That is, the waveforms after TTX+NMDA, TTX+NMDA+PTX, or GABA were quite similar (see Fig. 3B ). This is not to say that a study undertaken specifically to describe the differences would not show them or perhaps demonstrate other retinal effects of GABA and PTX. For example, the origin of the slightly larger amplitude of the waveform after GABA remains to be investigated. However, whatever effects these drugs might have had, beyond eliminating inner retinal activity, appeared to be relatively minor in this study. 

In summary, our findings are consistent with at least three components of the monkey mfERG that depend on the integrity of the inner retina, a TTX-sensitive ONHC, a TTX-sensitive HFC, and an NMDA- and GABA-sensitive component. It is reasonable to assume that in all three components, both the ON and OFF pathways are involved. APB alone removed oscillatory responses at least up to the peak of P1. 

Contributions from the Outer Retina.

After the inner retinal influences were removed, the influence of the outer retina, the photoreceptors and the ON- and OFF-bipolar cells, could be seen more easily. It is clear from Figure 4 that the N1 and P1 components of the primate mfERG, ¹² similar to the a- and b-waves of the full-field ERG, ⁵ were influenced by both ON- and OFF-bipolar activity. Figure 9A supplies a summary of these influences for the responses from ring 1 (left) and rings 3 to 5 (right). The responses after the addition of TTX+NMDA (black) and after the addition of APB (red) and the contribution removed by APB (blue) are shown superimposed (from monkey D in Fig. 4B ). They represent an estimate of the contributions from the ON-bipolar cells (blue) and the contributions from the OFF-bipolar cells plus the cone photoreceptors (red) to the response after injection of TTX+NMDA (black). (For ease of presentation, all records in Fig. 9A are scaled so that the responses after TTX+NMDA in ring 1 and rings 3 to 5 have the same peak-to trough-amplitude.) 

Consider the response from rings 3 to 5 first. The initial portion of the leading edge of N1 (black) was largely due to the hyperpolarization of the OFF-bipolar cells that occurred in response to an increase in light. Figure 4B demonstrates that the receptors made a very small contribution to the responses from the outer rings. The onset of the ON-bipolar contribution (blue) occurred just slightly before the peak of N1 was reached and in control subjects the initial portion of the leading edge of P1. Well before the peak of P1 was reached, the OFF-bipolar cells started to recover (depolarize), and this recovery also contributed to the leading edge of P1. (In fact, in rings 3 to 5 an inflection could be seen on the leading edge of P1, caused by the depolarization of the OFF-bipolars.) The peak of P1 occurred between the time that the ON-bipolar contribution reached its peak and the time at which the OFF-bipolar contribution reached the positive peak of its recovery. Thus, the peak time of P1 was influenced by both ON- and OFF-bipolar cells. Although the trailing edge of P1 also was influenced by both types of bipolar cells, it was defined mainly by the recovery of the ON-bipolar contribution. The story for ring 1 was essentially the same, with one exception. The initial portion of N1 was more heavily influenced by the hyperpolarization of the photoreceptors than it was in the outer rings. 

Cellular Contributions to the Human mfERG

Contributions of the Inner Retina.

As described, we propose that the rhesus monkey’s mfERG is shaped by at least three inner retinal influences, the ONHC, the HFC, and a NMDA (or GABA)-sensitive (TTX-insensitive) component. To the extent that the spatiotemporal variations in waveform are due to the ONHC, the less marked variations in the human’s mfERG suggest a relatively smaller ONHC. This is consistent with the relative sizes of the ONHC extracted from human and monkey mfERGs. ^{13 14} The HFC also appeared to be extremely small in the human mfERG. In fact, compared with the monkey, the HFC was essentially missing in ring 1. Thus, compared with the monkey mfERG we propose that the human mfERG has a small ONHC and even smaller HFC. This explains, at least in part, the difference between the mfERG results from patients with glaucoma and monkeys with experimental glaucoma. Although changes in the mfERGs from patients with glaucoma can be detected, and these changes can even be consistent with those seen in the TTX-treated monkey, ^{3 11 38 39 41} the changes are usually subtle ^{3 11 33 34 42 43 44 45 46 47 48 49 50} (Fortune B, submitted for publication, 2002) compared with those in monkeys, where the ONHC and HFC are essentially eliminated. ^{15 21 40 51 52}  

The third influence of the inner retina identified in the rhesus monkey is the NMDA-sensitive component. A similar component is probably present in the human mfERG as well. The evidence comes from the appearance of a shelf on the trailing edge of P1 (monkey: Fig. 3 arrows; human: Fig. 8 ). This shelf is removed by NMDA (or GABA) in the monkey and appears to be absent in some patients with diseases of the inner retina. ^{11 38} (Note that in the region of the shelf there was a nasotemporal difference in waveform in the human mfERG [Fig. 7C ] but not in the monkey treated with TTX [Fig. 2D ]. The variation in the human mfERG is due to the algebraic summation of a shelf that does not show a nasotemporal variation with an ONHC that does.) 

In sum, most of the human mfERG can probably be captured by a waveform rather like that from the animal treated with TTX+NMDA combined with small inner retinal contributions that have subtle influences on the leading and trailing edges of P1, as well as N1. In short, the overall shape of the human mfERG is dominated by the cells of the outer retina (i.e., photoreceptors and bipolar cells). Consequently, we feel justified in turning to the results after TTX+NMDA summarized in Figure 9A for a better understanding of the influences shaping the human mfERG. 

A Working Model of the Human mfERG.

Figure 9B presents our working model for the human mfERG. It is essentially the model of the monkey mfERG after administration of TTX+NMDA in Figure 9A , with the recognition that the inner retina helps shape the human mfERG. The mfERGs summed for ring 1 and rings 3 to 5 are shown by the black line. We assume that the leading edge of N1 is largely the onset of the OFF-bipolar cell contribution with smaller contributions from the receptors and perhaps from the inner retina as well. Before the trough of N1 is reached, the shape of N1 is altered by the onset of the ON-bipolar contribution (see lower blue arrows in Fig. 9B ). The initial portion of the leading edge of P1 is a combination of the depolarization of the ON-bipolar and the hyperpolarization of the OFF-bipolar. However, after the OFF-bipolar contributions have reached their trough, the leading edge of P1 is a combination of the depolarization of both the ON- and OFF-bipolars, as shown by the lower red arrows in the figure. In fact, in rings 3 to 5, an inflection can be seen on the leading edge of P1 where, according to the model, the OFF-bipolars begin to depolarize. Finally, the trailing edge of P1 is defined largely by the recovery of the ON-bipolars with some contribution from the OFF-bipolar recovery, and inner retina. 

It is important to note that in the human model the peak of P1 occurs between the time of the peak of the ON-bipolar contributions (Fig. 9B , upper blue arrows) and the peak of the OFF-bipolar contributions (Fig. 9B , upper red arrows). In the monkey’s records from ring 1 in Figure 9A , the peak of the mfERG was at 26.7 ms, whereas the ON- and OFF-bipolar contributions peaked at 23.3 and 29.2 ms, respectively. Thus, if the ON-bipolar contribution were absent, the latency of the peak of P1 would increase for ring 1 by 2.5 ms. The increase for rings 3 to 5 would be approximately 1.2 ms. In keeping with this small latency increase, Kondo et al. ⁵³ found that patients with complete congenital stationery night blindness (CSNB), thought to be missing an ON-bipolar response, have P1 peak times that are, on average, 3.2 ms longer than control values. 

Of course, care should be taken when comparing monkey and human ERGs, especially in the macula, where cell densities may differ. ^{54 55} However, the similarity in the waveforms of the mfERG after the removal of the inner retina (Fig. 9) suggests that the model in Figure 9B captures the key aspects of the cellular contributions to the human mfERG. 

Authors’ Notes

Note: The response to a presentation at location x at time t affects the response to a flash that occurs on the next frame change (t + 13.3 ms) at the same location. This affects the waveform of the response at time t in addition to the response at t + 13 ms, because these mfERGs are not flash responses in the traditional sense but are first-order serial correlations, or first-order kernels. An explanation of this phenomenon is beyond the scope of the present paper but can be found in several studies ^{3 17 56} where this “induced version of the second-order kernel in the first-order kernel” is explained. As Figure 26 in Hood ³ shows, the effects here are subtle in the case of the human mfERG. (They are much more dramatic in the case of the monkey’s control records.) In particular, for the human mfERG, the response (first-order kernel) is modified slightly, starting just before the peak of P1 with the modification reaching its maximum on the trailing edge of P1 and the subsequent negative trough N2. We find (unpublished data, 1999) that removing the inner retinal influences with the drugs represented in Figure 3 eliminates most but not all the effect of adaptation. Thus the “inner retinal influence” should be thought of as both “direct” contributions and influences of adaptation (“the induced version” of the second-order kernel). 

 Supported in part by National Eye Institute Grants R01-EY09076, R01-EY06671, P30-EY07751, and T32-EY07024, and a William C. Ezell fellowship (SV).

 Submitted for publication September 7, 2001; revised December 19, 2001; accepted January 29, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) The stimulus array used in the multifocal recordings. (B) Multifocal records are shown for the control condition of the left eye of monkey VA. The calibration markers indicate 200 nV and 60 ms. (C) Response groups for rings. Ring 1 includes the central 7 responses. (D) Response groups for quadrants.

Figure 2.

 

(A) Responses summed by rings (see Fig. 1C ) for monkeys VA (from Fig. 1B ) and M. (B) The first 120 ms of the responses from rings 1 and 3 to 5 from monkey VA. (C) Responses from rings 1 and 3 to 5 from monkey M before and after injection of TTX (6.3 μM vitreal concentration). (D) Responses summed by quadrants (see Fig. 1D ) for monkeys VA (from Fig. 1B ) and M. The vertical calibration line in the lower right-hand corner applies to all the records.

Figure 3.

 

(A) Responses from rings 1 and 3 to 5 for three monkeys with inner retinal activity blocked with TTX (4.8 μM)+NMDA (1.6 mM), TTX (7.1 μM)+NMDA (1.6 mM)+PTX (0.14 mM), or GABA (55 mM). Dashed curves: control records. (B) The first 120 ms of the responses from rings 1 and 3 to 5 from monkey V(OD) before (dashed line) and after TTX+NMDA (solid line). (C) The first column contains the normalized responses from (A) after administration of the drugs to block the inner retina. The second column contains the normalized responses after TTX from Figure 2C and another animal E (TTX, 6.3 μM). The third column contains the records from the first two columns, superimposed for comparison.

Figure 4.

 

(A) Illustration of the logic behind the isolation of the contributions from ON-bipolar cells after administration of APB, which blocks the photoreceptor–ON-bipolar synapse. Application of APB to a retina in which the inner retinal responses have been suppressed produced responses (shown in red) containing only photoreceptor and OFF-bipolar contributions. Subtracting these responses from the control condition yields an estimate of the ON-bipolar contribution (shown in blue). (B) Responses by rings from monkey D after TTX (6.3 μM)+NMDA (5.2 mM; green), after application of APB (2.1 mM; red), and after PDA (7.5 mM; bold black). The blue records in the center column are the isolated ON-bipolar contributions obtained by subtracting the red records from the green records. The thin black records in the last column are the results from another monkey (V) after injection of PDA and APB. (C) Same for the responses from rings 1 and 3 to 5 of monkey VA. (D) The responses from rings 1 and 3 to 5 from monkey V before (gray) and after (blue) PDA (3.8 mM). The records in black show the responses after APB (3 mM) had been added after injection of PDA. The calibration marker in the lower right corner applies to all records in the figure. A preliminary analysis of the results from monkey D appears in Ref. ³ .

Figure 5.

 

(A) Estimates of ON-bipolar contributions isolated with APB, bold gray, monkey D; bold black, monkey VA, or APB after PDA, thin black, monkey V. Responses were scaled to have the same peak amplitude. (B) Estimates of OFF-bipolar+photoreceptor contributions isolated with APB, gray, monkey D; black, monkey VA. Responses were scaled to have the same peak negative (trough) amplitude. (C, left) Estimated OFF-bipolar contribution (black) isolated by subtracting the responses in Figure 4B after APB+PDA from those after APB for monkey D. The gray curve is the contribution (from A) of the ON-bipolars isolated with APB. (C, right) The same gray curve (ON-bipolars) is shown with the black curve (OFF-bipolars) from the left panel inverted in polarity and scaled to have the same peak amplitude.

Figure 6.

 

(A) Second-order kernels (solid curves) extracted by computer from rings 1 and 3 to 5 for the three monkeys in Figure 3A with inner retinal activity blocked with TTX+NMDA, TTX+NMDA+PTX, or GABA. Dashed curves: second-order kernels of the control records. (B) Dark curves: second-order kernels for monkey VA treated with APB after GABA. Gray curves: are the same as those in the rightmost column of (A). The vertical calibration bar in the lower right hand corner applies for all records in the figure.

Figure 7.

 

(A) Multifocal records for one of the human subjects (S1; first column) and for the average of the four human subjects (second column). The calibration markers indicate 200 nV and 60 ms. (B, C) Responses grouped for subject S1 (first column) and the average of the four subjects (second column) by rings (B) and by quadrants (C) as in Figure 2 for the monkeys. The calibration marker in (C) applies to all records in (B) and (C).

Figure 8.

 

A comparison of the average response for rings 1 and 3 to 5 from the four human subjects (black) to those from monkeys (gray) treated with TTX or TTX+NMDA. The records from the humans were scaled to have the same peak-to-trough amplitude as the corresponding records from the monkeys. The scaling for ring 1 was 2.1, 1.3, and 1.6 and for rings 3 to 5 was 3, 2.6, and 3 to match monkeys M, E, and D, respectively.

Figure 9.

 

(A) Records from monkey D (Fig. 4B) after TTX+NMDA to remove the contributions from the inner retina are shown with the estimated contributions from the ON-bipolars (blue) and the post-APB records (red), which should be the combined contributions from the OFF-bipolars and photoreceptors. All records were scaled so that the responses after TTX+NMDA in ring 1 and rings 3 to 5 had the same peak-to-trough amplitude. (B) A model of the contributions to the human mfERG based on the results from the monkey in (A). All records were scaled so that the control responses in ring 1 and rings 3 to 5 had the same peak-to-trough amplitude.

The authors thank John G. Robson for help and advice and Nalini Rangaswamy for recording the human mfERGs.