Fig. 14. The schematic drawing of the organization of the retina (familiar from chapter 2.1.2).
Primary cellular origins to ERG components are illustrated. OPs: oscillatory potentials, PhNR:
photopic negative response, STR: scotopic negative response, PERG: pattern electroretinogram (Figure adapted and modified from (38)).
3.2 PATTERN ELECTRORETINOGRAPHY (PERG)
The pattern electroretinography (PERG) is a special technique used to record inner retina specific responses (174). Here, luminance is kept unaltered and the eye is stimulated by contrast reversing patterns on a computer monitor (Fig. 15). The cellular generators of flash ERG (FERG in this chapter) and PERG are substantially different (discussed next).
Fig. 15. Pattern electroretinography (PERG) in a C57BL mouse under ketamine & medetomidine anesthesia. A: The mouse eye is stimulated with a sinusoidal grating stimulus that reverses phase every 0.5 s. Two practically equal responses are generated (see text for details). For analysis, these responses are checked for consistency and then averaged. PERG amplitude is generally taken as the difference between the first positive peak and first major negative peak amplitudes. B: Waveforms at different pattern frequencies. C: As the pattern frequency increases, the amplitude of PERG decreases because the stimulation is subjectively weaker.
CPD: cycles per degree of visual angle. (Leinonen H. 2013, unpublished data).
3.2.1 Cellular origins and utility of the PERG
When Riggs et al. (1964) (175) were able to record real PERGs without luminance artefacts for the first time they did not know that its generators differ from the FERG, and thus they
did not call it PERG. Several years later Maffei and Fiorentini (1981) reported an absolute loss of PERG signal four months after optic nerve transection in cats while FERG remained intact (176). Sieving and Steinberg recorded extracellular responses to contrast reversing grating stimuli across the depth of cat retina and performed a current source density analysis to demonstrate the cellular origin of the PERG response (177). They found that flash and pattern stimulus evoked responses have different sources in the retina, the distal and proximal retina, respectively. Baker and coworkers reproduced a similar finding in primates (178). They concluded that FERG has a major generator in the distal half of the retina (from the receptor layer to the outer plexiform layer) whereas the PERG is generated in the proximal 20 % of the retina, spanning from the nerve fiber layer to the middle of the inner plexiform layer. The latter corresponds to synaptic and dendritic sites of ganglion,
amacrine and bipolar cells.
Two recent reports investigated the origins of PERG in C57BL mice and macaque monkeys in detail (158,179). Both studies used a similar experimental design to allow comparison. In both studies experimental glaucoma and intravitreal injections of pharmacological agents were used to dissect the PERG. The mouse and primate corneal PERG waveforms fairly well corrersponded to each other, showing a major positive wave (P1) peaking at around 50 ms in both species and a major negative wave (N2) peaking at around 95 ms in macaques and at around 130 ms in mice. Experimental glaucoma nearly abolished the PERG in both species leaving the FERG virtually intact. This is a consistent finding across species, including humans (10,158,180‐183). Suppression of Na+‐dependent spiking activity in RGCs and in a subset of amacrine cells by tetrodotoxin (TTX) decreased the P1 slightly over 50 % in macaques and even more, about 75 % in mice. The N2 was practically abolished in macaques but fairly well preserved in mice (158,179). Metabotropic glutamate receptor analoque 2‐amino‐4‐phosphonobutyric acid (APB) that blocks signal transduction in the cone ON bipolar cells, but not OFF cells (184), was used to dissect ON and OFF contributions. In macaques, APB removed about half of both P1 and N2, whereas in mice, P1 was completely eliminated and N2 largely intact (158,179). Finally, blockade of second‐order hyperpolarizing neurons (OFF bipolar cells, horizontal cells, some amacrines) and third‐order neurons (RGCs) by cis‐2,3‐piperidine dicarboxylic acid (PDA) (184) virtually eliminated N2 but increased P1 in macaques (179). In mice, PDA decreased both P1 and N2 substantially (158). To conclude, in mice the PERG P1 arises mainly from spiking activity of ON pathway RGCs. The N2 in mouse arises from spiking activity of OFF pathway but also has non‐spiking contributions from both pathways. In contrast, in macaque PERG the P1 reflects non‐spiking activity of RGCs (i.e. input to the RGCs) and N2 spiking activity of RGCs (within RGCs) (179). The PERG in macaques receives almost equal contributions from ON and OFF pathways. The obvious inter‐species differences in PERG generators are constraints in determining the exact cellular origins of the PERG. Likely because of that, they are still not known in depth (185).
To explain the PERG response in more practical terms: during PERG recording the radial currents are not seen on the corneal electrode because the luminance in pattern stimulus does not change. It should be noted, however, that the activation of radial currents is necessary for generating the PERG response, but they remain undetected because ON and OFF pathways are stimulated exactly to the same extent (for illustration see Fig. 16).
Instead, horizontally evoked nonlinear responses will constitute a small LFP in the corneal surface to be recorded.
Fig. 16. Schematic drawing of retinal activation upon pattern reversal (at constant luminance) (A) or uniform flicker (B), i.e. flash, stimuli. In both cases the stimulus elements change from time 0 to time 1 between two conditions of different luminance (black and white). A: Alternation of the patterned stimulus evokes local flash ERGs within the retina. These radially generated local ERGs (from vertical pathway) are 180 degrees out of phase and thus their summed activity is canceled at the remote corneal electrode. However, in addition to these linear local ERGs, a nonlinear response is generated at pattern reversal from mechanisms that are sensitive to changes in border contrast, i.e. lateral inhibition via horizontal pathways. Center-surround activation of the RGC dendritic field occurs differentially for both phases of the pattern reversal stimulation, but their magnitude is the same. These nonlinear responses are in-phase and thus the sum of their current is recordable at the cornea. B: Uniform flicker equals luminance change, flash stimulus. Here, radially generated responses are in-phase and thus generate a strong current at the corneal electrode. In contrast, the flash stimulation does not generate significant center-surround activation of the RGC receptive fields and thus nonlinear signals are absent. To simplify, a uniform flicker (flash) evokes LFPs at the corneal surface that originate mainly from outer retina activity (photoreceptors, bipolar cells), whereas corneal response for pattern reversal reflect inner retinal activity from lateral pathways. The PERG is said to occur at the second harmonic of the stimulation frequency. If the stimulus frequency is 1 Hz (i.e. the stimulus cycle restarts every 1 s), we get two responses in 1 s, i.e. at 2 Hz (Figure adapted from (186)).
PERG can be used to test the limits set by the retina for visual acuity, contrast threshold and temporal resolution. Porciatti, who is a pioneer of mouse PERG recordings, found that retinal acuity in C57BL mice correspond to cortical visual acuity as measured by VEPs or behavioral visual tests (186‐190). Furthermore, PERG has been shown to correlate well with cortical visual acuity in monkeys and humans (191,192). As PERG is assumed be generated at the level of RGC center‐surround structure, the maximum PERG response should be achieved with pattern stimulation having a spatial frequency that best corresponds to the average RGC receptive field size. Indeed, this has been shown in humans (193). Importantly, in disease state, such as in human glaucoma (194) or mouse optic neuropathy (195), the PERG impairment has been shown to be greater than expected from retinal nerve fiber layer thickness and RGC count. It is likely that some RGCs that are viable are not functional (194,196). Thus PERG holds promise to be able to detect RGC pathologies prior to cell death when the RGCs may be still curable. By current knowledge, PERG is the most sensitive method to test RGC function, exceeding the sensitivity of STR and PhNR.
3.2.2 Major sources of error in PERG
A quintessential requirement in pattern stimulation is the extremely careful calibration of the stimulation system. PERG responses are maximally around 10‐fold smaller than photopic flash ERG responses in C57BL mice (compare Figs. 12D‐F and 15). Thus, if an extra flash (luminance artefact) occurs at pattern reversal, the PERG will be contaminated or even overwhelmed by outer retinal activity. This is critical because normal outer retinal activity may mask inner retinal damage if pattern stimulation is not appropriate. In VEP experiments, luminance artefact may lead to overestimation of visual acuity among other problems (197). Typical consumer liquid‐crystal display (LCD) monitors produce an extra flash in pattern reversal, especially when using high contrast, due to slow response rate of the monitors. While this luminance artefact may be almost impossible to get rid of when using consumer LCD monitors, it can be minimized by using high‐speed LCD screens and by reducing the stimulus contrast (198). In addition to inappropriate stimulator calibration, also insufficient stimulus conformation, e.g. different number of black and white pixels, can also lead to luminance artefact as illustrated in Fig. 17.
Fig 17. The effect of luminance imbalance in the pattern stimulation. Upper: a fully symmetric pattern stimulus that reverses four times yields four similar PERG responses. Lower: creating asymmetrical stimulus by covering a part of the screen creates a flash response at every second reversal, rather than PERG (Figure adapted from (199)).
Because seeing pattern contrasts is quintessential for generation of valid PERG response, it is imperative that nothing, internal or external to the eye, is blocking the way between the retina and the stimulus. The eye should be as transparent as it is in its normal state. In mice, formation of transient opacity of ocular media under anesthesia is a big problem that may preclude PERG recording (200,201). The opacity under ketaminine &
xylazine anesthesia develops within minutes after induction of anesthesia without precautions. This happens mostly due to tear evaporation (201) but it has also been suggested that other factors play a role, such as cold, asphyxia and stress (200). The anesthesia induced transient opacity formation can be slowed or even completely blocked by application of artificial tears or inserting applicable contact lenses (201). It should be noted, however, that the artificial tears or contact lenses should not affect eye optics.
Although isoflurane gas anesthesia provokes less opacities of ocular media (Leinonen H., unpublished observations) the same precautions apply.