Documenta Ophthalmologica

, 122:149

Post-receptoral contributions to the rat scotopic electroretinogram a-wave


  • Trung M. Dang
    • Department of Optometry and Vision SciencesUniversity of Melbourne
  • Tina I. Tsai
    • Department of Optometry and Vision SciencesUniversity of Melbourne
  • Algis J. Vingrys
    • Department of Optometry and Vision SciencesUniversity of Melbourne
    • Department of Optometry and Vision SciencesUniversity of Melbourne
Original Research Article

DOI: 10.1007/s10633-011-9269-y

Cite this article as:
Dang, T.M., Tsai, T.I., Vingrys, A.J. et al. Doc Ophthalmol (2011) 122: 149. doi:10.1007/s10633-011-9269-y


The electroretinogram is a widely used objective measure of visual function. The best characterised feature of the full-field dark-adapted flash ERG, is the earliest corneal negativity, the a-wave, which primarily reflects photoreceptoral responses. However, recent studies in humans and primates show that there are post-receptoral contributions to the a-wave. It is not clear if such contributions exist in the rat a-wave. We consider this issue in the rat a-wave, using intravitreal application of pharmacological agents that isolate post-receptoral ON-pathways and OFF-pathways. In anaesthetised adult long Evans rats, we show that the ON-pathway (2-amino-4-phosphonobutyric acid, APB sensitive) makes negligible contribution to the a-wave. In contrast, CNQX (6-cyano-7-nitroquinoxaline-2,3-dione) or PDA (cis-piperidine-2,3-dicarboxylic acid) sensitive mechanisms modify the a-wave in two ways. First, for bright luminous energies, OFF-pathway inhibition (CNQX or PDA) results in a 22% reduction to the early phase of the leading edge of the a-wave up to 14 ms. Second, OFF-pathway inhibition removed a corneal negativity that resides between the a-wave trough and the b-wave onset.







2-amino-4-phosphonobutyric acid




cis-piperidine-2,3-dicarboxylic acid


The a-wave of the full-field dark-adapted flash electroretinogram is thought to arise largely from photoreceptor activity. However, studies in humans, non-human primates [1, 2] and cats [3] show that the dark-adapted a-wave also receives contributions from post-receptoral responses. It is not clear if similar post-receptoral contributions exist in the rodent scotopic a-wave.

In past work, the post-receptoral contribution to the dark-adapted a-wave has been characterised using brief flashes [1, 2]. However, the balance of activity induced in ON-pathway and OFF-pathway cells is known to depend on flash duration, so it is appropriate to probe with both brief and long (longer than time to a-wave peak) flashes as has been employed in previous studies of scotopic ERG responses [4]. The waveforms are measured with and without drugs that selectively inhibit glutamatergic neurotransmission via group III metabotropic glutamate receptors (2-amino-4-phosphonobutyric acid, APB) and AMPA/KA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid and kainite sensitive) ionotropic glutamate receptors (6-cyano-7-nitroquinoxaline-2,3-dione, CNQX, cis-piperidine-2,3-dicarboxylic acid, PDA). APB removes contributions from the ON-pathway, whereas CNQX and PDA will remove contributions from the OFF-pathway, as well as AMPA/KA sensitive responses arising from the outer retina (horizontal cells), and inner retina (amacrines and ganglion cells).


All animal experimental procedures were conducted with approval from our institutional Animal Experimentation Ethics Committee (A04001) and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Subjects were adult long Evans rats between 10 and 12 weeks of age (180–270 g, Monash Animal Services, Clayton, VIC, Aust). The rats were maintained at 22°C in a 30–70 lux environment with a 12-h light/dark cycle (on at 8 am). Water and rat chow were provided ad libitum.


Prior to experimentation, animals were dark adapted overnight (>12 h). All animal preparation was performed in the dark using a dim red light, which has very little output below 580 nm (λmax~650 nm). Anaesthesia was induced with intramuscular injections of 60 mg/kg ketamine (ketamil 100 mg/ml, Troy laboratories, Smithfield, Aust) and 5 mg/kg xylazine (xylazil 100 mg/ml, Troy laboratories). Supplemental injections of half the original dose were given every hour after the initial injection. Corneal anaesthesia was obtained by instillation of one drop of proxymetacaine (Opthetic 0.5%, Allergan, Irvine, CA). Mydriasis was induced with one drop of 0.5% tropicamide (Mydriacyl, Alcon Laboratories, Frenchs Forest, NSW, Aust). Following anaesthesia, core temperature was maintained by covering the animal with an insulating blanket and placing the rat over a circulating-water heat-pad (MGW pump Lauda, Lauda-Königshoffen, Ger). The animal and heating pad were mounted, and stabilised on a custom-built platform with the rat’s neck and abdomen lightly secured to the stage with Velcro to minimise movement during recordings.

Single full-field ERGs were recorded using a custom-built Ganzfeld sphere (Photometric Solutions International, Huntingdale, VIC, Aust) using methods common to our laboratory [5]. Recordings were taken simultaneously from both eyes using chlorided silver electrodes. The recording electrode was placed at the centre of the cornea. The reference was a ring electrode placed around the equator of the eye. The ground was a stainless steel needle (F-E2-30, Grass Telefactor, West Warwick, RI) inserted subcutaneously into the tail. Electrode contact was improved by applying a small amount of 1.0% carboxymethylcellulose sodium (Celluvisc, Allergan, Irvine, CA, USA). A readaptation period of 10 min was allowed following electrode placement prior to data collection. The signals were amplified and band-pass filtered (0.3 Hz–1 kHz) over a 2,560 (640 ms) epoch with a 4 kHz sampling rate. A 50-Hz notch filter was applied post hoc to eliminate line noise.

Stimuli were either relatively short (1–4 ms) or long (256 ms) white flashes (5 Watt LEDs, Luxeon, Philips Lumileds Lighting Co., CA, USA) delivered via a Ganzfeld integrating sphere. Stimulus energy was calibrated using an IL-1700 Research Radiometer (International light, Newbury Port, MA) with a scotopic filter in place (Z CIE, International Light). Values for short flashes are specified by their luminous energies (scotopic cd.s.m−2), whereas those for long flashes are specified as luminances (scotopic cd.m−2). Modification of Lyubarsky et al.’s [6] formulæ to account for the different ocular dimensions in rats means that 1 scot cd.s.m−2 produces 828 Φ/rod [5]. ERG responses were obtained for luminous energies ranging from −1.15 to 1.55 log cd.s.m−2 for short flashes and for luminances between 0.76 and 3.11 log cd.m−2 (0.76, 0.97, 1.27, 1.68, 2.18, 2.53, 2.82 and 3.11) for 256 ms flashes. For flash durations from 1 to 4 ms, time 0 is specified from the middle of the stimulus. The interstimulus interval was varied between 45 and 240 s depending on luminous energy, to allow for complete recovery of a-wave amplitude between successive flashes.

Intravitreal injections

As described previously [7], pharmacological agents were delivered into the vitreal chamber via a 30-gauge needle attached with fine polyethelyne tubing (inner diameter = 0.38 mm) to a Hamilton syringe (SGE syringes, Ringwood, Aust). The needle was introduced through the pars plana, approximately 1 mm posterior to the superior limbus and at an angle of 45 degrees to avoid the lens. Animals were excluded from analysis if opacification of the lens occurred (<10%). Aliquots of 2–5 μl were used with the upper limit chosen to minimise intraocular pressure changes [8]. Contributions from the ON-pathway are removed using APB (400 μM, n = 4 for long flashes and n = 5 for short flashes), whereas the OFF-pathway is removed using either CNQX (250 μM, n = 4 for long flashes) or PDA (5 mM, n = 6 for short flashes). All agents were obtained from Sigma–Aldrich (Sigma–Aldrich Pty. Ltd., Castle Hill, NSW, Aust). APB and PDA were diluted using distilled H2O, whereas CNQX was mixed in the minimal volume of dimethyl sulphoxide (DMSO, <0.1%) that gave full dilution. This concentration of DMSO has been shown to have no effect on the rat ERG [9]. These concentrations represent the estimated vitreal concentrations assuming complete dilution and a rat vitreous volume of 50 μl [10].

Control data represent a group of control eyes (n = 22 for long flashes, n = 11 sham injected for short flashes). Following drug injection, single ERG waveforms (−0.74 log cd s.m−2) were collected at 5-min intervals, to track drug onset for 40 min, ensuring stabilisation before the ERG protocol began.

Data analysis

Response amplitudes were baseline corrected to the average level for 5 ms prior to stimulation. Amplitudes were measured at fixed times after stimulus onset (5–15 ms) in order to define the amplitude–energy functions of the putative mechanisms; an approach employed by Saszik et al. [11]. Amplitudes earlier than 5 ms were within noise levels (2–4 μV RMS) and were not analysed. An artefact likely arising from the electronics used to drive the LEDs occurs within the first 1–2 ms and does not influence our analysis (see Fig. 1).
Fig. 1

Representative waveforms for an animal treated with APB (thick line) and its fellow control eye (thin line) for short (A and B), and long flashes (Panel D). The numbers along the right hand side represent the luminous energy in log cd.s.m−2. Waveforms are overlayed in order to better visualise the a-wave (Panels B and D). Panels C and E compares the average amplitude (±SEM) of APB-treated eyes (filled) to that of control eyes (unfilled) for amplitude collected at fixed times of 6 (circles) and 15 ms (diamonds). The dashed lines show the best-fit models to the control eyes at each criterion time. The data points have been displaced horizontally (APB shifted by 0.02 log units right and control 0.02 log units to the left) to allow better visualisation

A hyperbolic function was used to fit the amplitude–energy curves for various criterion times following stimulus onset [12]. The function is given by a maximal amplitude (Vmax, μV) and semi-saturation constant (K, log cd.m2).


Statistical analysis was performed using Prism® software (v4.0, GraphPad Software Inc., 2003, CA, USA). Two-way ANOVA was conducted to consider interactions between the main effects of treatment (control vs APB, CNQX or PDA) and luminous energy or luminance nested within treatment. A Bonferroni post hoc test was applied in the presence of significant interactions. Throughout this study, group data are reported as means ± SEM.


Contributions from APB sensitive elements

Figure 1 shows representative control and APB-treated waveforms in response to short (Panels A and B) and long duration flashes (256 ms, Panels C and D) of varying luminous energy and luminances, respectively. Figure 1 shows that APB (thick black) completely abolishes the positive b-wave, leaving a corneal negative waveform. The leading edge was unaffected by APB (Fig. 1A and 1B) for short or long duration flashes (Fig. 1D).

Figures 1C and 1E compare amplitudes returned at fixed times of 6 and 15 ms. An analysis of variance showed that there was no significant interaction effect and no significant difference between the control group and the APB-treated group for either short or long flashes. This was the case for all fixed times between 5 and 15 ms (F1,5 = 0.04–0.61, P > 0.05 for short flashes, F1,7 = 0.07–0.32, P > 0.05 for long flashes). Table 1 also shows that there was no significant treatment effect between control and APB for short or long flashes. Thus, in rats, contributions from post-receptoral elements in the ON-pathway do not become manifest in the dark-adapted ERG a-wave until at least 15 ms after stimulus onset for the brightest luminous energy employed in this study.
Table 1

Treatment effects for the different criterion times (Ax) for controls and eyes treated with APB (short and long flashes), CNQX or PDA (short flash, n = 6)


Short flashes

Long flashes

Criterion time (s)




























































Italics indicate those P values that are significantly different from the control group (P < 0.05)

Contributions from CNQX and PDA sensitive elements

Contributions to the rat a-wave from post-receptoral responses arising from the OFF-pathway were considered by examining the effect of PDA and CNQX. Both treatments resulted in a reduction in scotopic b-wave amplitude as has been previously reported for high concentrations of PDA in rats [7]. Figures 2B and 2D also show that both PDA and CNQX causes a change in the shape of the a-wave at later times (>14 ms). Figures 2C and 2E shows amplitude at fixed times of 6 and 15 ms for control and drug-treated groups. Analysis of variance did not find a significant interaction between control and treatment across intensity for either PDA (for short flashes F1,7 = 0.15–0.50, P > 0.05 for all fixed times from 5–15 ms) or CNQX (for long flashes F1,7 = 0.36–1.38, P > 0.05 for all fixed times from 5–15 ms). However, both PDA and CNQX treatments significantly reduced amplitudes. For shorter flashes, PDA significantly reduced amplitude from 7 ms onwards, whereas for long flashes CNQX had significant effects for 7 ms and from 10 ms and onwards (Table 1).
Fig. 2

Representative waveforms for an animal treated with PDA (thick line) and its fellow control eye (thin line) for short flashes (Panels A and B), and for an animal treated with CNQX and its fellow control eye for long flashes (Panel D). The numbers along the right hand side represent the luminous energy in log cd.s.m−2. Waveforms are overlayed in order to better visualise the a-wave (Panels B and D). Panels C and E compares the average amplitude (±SEM) of treated eyes (filled) compared with control eyes (unfilled) for amplitude collected at fixed times of 6 (circles) and 15 ms (diamonds). The dashed lines show the best-fit models to the control eyes at each criterion time. The data points have been displaced horizontally (treated shifted by 0.02 log units right and control 0.02 log units to the left) to allow better visualisation

The possibility that a scaling process can explain the a-wave reduction caused by PDA and CNQX was evaluated by normalising treated waveforms to those of controls. The scaling factor needed to achieve this alignment for a fixed amplitude at 12 ms was 22 ± 13% for PDA-treated eyes with short flashes and 23 ± 14% for CNQX-treated eyes with long flashes.

Although the leading edge of the a-wave can be scaled accurately, treated waveforms still show a change in shape at later times, which becomes more evident with brighter luminous energies, as shown in Fig. 3 for long flashes. In particular, control waveforms have a secondary negativity at a time when CNQX-treated waveforms show an onset of a corneal positive response, one that cannot be aligned with scaling. APB-treated waveforms follow the control waveform up to 18 ms.
Fig. 3

Average APB (n = 4, grey line), CNQX (n = 4, scaled, thick line) treated waveforms collected using long duration flashes are normalised to the control amplitude at a fixed time of 12 ms (n = 22; thin line) in Panel A. Once normalised treated waveforms are subtracted form the control waveform (control–treatment, APB long flash-grey line, CNQX long flash-thick line) are shown in Panel B. The numbers along the left show the luminance in log cd.m−2 for long flashes. Also, shown in Panel B are the difference traces for PDA subtracted from control for short flashes. The luminous energies for the short flashes are (from top down) 1.55, 0.63, 0.08 and −0.48 log cd.s.m−2

In order to better visualise this feature, treated waveforms were subtracted from controls (Fig. 3B) following scaling. For the brightest stimulus, the departure between control and APB-treated waveforms begins after 18 ms. This departure for CNQX treatment in long duration flashes begins at 14 ms. Also, plotted on Fig. 3B is the difference for PDA treatment assessed using short flashes. This shows that CNQX and PDA modify the descending limb of the a-wave in similar ways.


We show that the pigmented rat scotopic a-wave like non-human primates [1, 2] and cats [3] also contains post-receptoral contributions. Green and Kapousta–Bruneau [13] using trans-retinal microelectrode recordings in the isolated albino rat retina report that the scotopic a-wave is entirely due to photoreceptor-derived currents. The lack of post-receptoral involvement in these rat a-waves is surprising, and may represent differences in the albino strain of rats [14, 15] and/or the isolated retinal approach employed by Green and Kapousta–Bruneau [13].

Rat dark-adapted a-wave receives no ON-pathway contributions

The rat scotopic a-wave, contains little contribution from the ON-pathway. This is consistent with previous literature showing that it is unchanged following APB application in rats [13]. APB is an agonist to group III metabotropic receptors [16], and since no change was observed with APB, it is unlikely that any of the Group III mGluRs make any contribution to the rat dark-adapted a-wave. Interestingly, Jamison et al. [1] find a 15–30% scotopic a-wave increase after APB application in non-human primates. This discrepancy may reflect species differences.

Rat dark-adapted a-wave is scaled by an ionotropic sensitive mechanism

To explore the contributions of the hyperpolarising bipolar cells and horizontal cells to the rat dark-adapted a-wave, CNQX and PDA were used to inhibit AMPA/KA receptors. These drugs had two effects on the scotopic a-wave. The first was an amplitude reduction. Indeed, the treated a-waves could be scaled to match control waveforms along their entire leading edge (Fig. 3). This ‘scaling’ of approximately 22% might reflect a decreased receptor output or the loss of an ionotropic sensitive corneal negative response that shares a similar time course to the ERG a-wave.

Jamison et al. [1] have reported that PDA caused an amplitude increase ‘scaling’ without affecting phototransduction sensitivity in primates. Similarly, modelling our CNQX data using the Hood and Birch [17] formulation of the delayed Gaussian model [18], produces significantly smaller phototransduction amplitudes (control −467 ± 23 vs CNQX, −346 ± 38 μV, 26% loss, unpaired t-test, P < 0.01), with a similar sensitivity (control, 2.56 ± 0.04 vs. CNQX 2.46 ± 0.06 log−1.s−3, unpaired t-test, P = 0.32).

Alternatively, AMPA/KA sensitive elements in the outer retina may also be involved. Shiells and Falk [19] show in dogfish retina that the ERG a-wave is reduced in the absence of horizontal cell responses, and that horizontal cells produced a negativity with a similar time course to the photoreceptors. Hanitzsch and colleagues [20], simultaneously recorded intracellular potentials from horizontal cells and ERG responses, and showed that the negative horizontal cell response overlaps in time and has a similar intensity response function to the PIII in rabbits. Despite this evidence, it is not clear how currents from the lateral oriented horizontal cell can manifest in the a-wave.

Rat a-wave trough modulated by an ionotropic sensitive mechanism

The second effect of ionotropic receptor blockade is to remove a contribution that modulates the a-wave leading edge after 14 ms for the brightest stimulus energies employed here (Fig. 3). This is consistent with findings following blockade of AMPA/KA receptors in primate scotopic a-waves [1]. The timing of the hyperpolarising bipolar cell intrusion in rat a-waves (>14 ms at 3.11 log cd.m−2) is similar to that reported for primate rod responses [1]. Although the exact origins are unclear, it is likely to arise from inner retinal elements [3]. Robson and Frishman [3] have shown that NMDA, results in a reduction of the cat dark-adapted a-waves, recorded using vitreal electrodes. Moreover, these authors find that such contributions are removed by a dim light background suggesting that the elements responsible arise from the rod pathway. These outcomes they interpret as contributions from NMDA sensitive elements in inner retinal rod pathway. Consistent with this idea, responses collected using a short interstimulus interval (500 ms) are unaffected by PDA treatment (data not shown). Thus, it is likely that the differences noted here arise from the rod pathway.


The findings from this experiment show that the rat dark-adapted a-wave is unaffected by application of APB. CNQX and PDA however, result in changes that appear to arise from two different mechanisms. First, a scaling of the rat a-wave along the entire leading edge, and second a modification of the a-wave trough.


NHMRC CJ Martin Postdoctoral Fellowship (BVB), NHMRC project grant 400127 (BVB).

Copyright information

© Springer-Verlag 2011