An Electrophysiological Study of Retinal Function in the Diabetic Female Rat

Abstract

purpose. To examine retinal function in female Long-Evans rats with streptozotocin (STZ)-induced diabetes.

methods. Hyperglycemia was induced by IV injection of STZ, and ERG responses were recorded at 4-week intervals over 12 weeks. Oscillatory potentials (OPs) and responses to intermittent stimulation were analyzed with a custom computer program. The GABA-induced responses of individual rod bipolar cells were obtained from patch-clamp recordings, and immunohistochemistry was used to illustrate the retinal distribution of GABA (γ-aminobutyric acid) and GABA_C receptors.

results. Hyperglycemia developed in rats 2 to 3 days after injection of STZ. Compared with previous reports of the effects of diabetes in male rats, visual function abnormalities appeared to be milder in STZ-treated female rats. No significant differences were observed in the sensitivity or amplitude of the a- or b-wave components of the ERG between diabetic and control animals, and both animal groups exhibited a similar time course of neuronal dark adaptation. In contrast, diabetic animals showed significant differences in the pattern of OPs and in the amplitudes of their responses to flicker. The accumulation of GABA in the inner retina of diabetic rats, combined with the results of patch-clamp recordings from individual bipolar cells, revealed that the circuitry underlying the GABA signal of the proximal retina is affected by hyperglycemia.

conclusions. The results suggest that changes in the GABA-signaling pathway may be the underlying cellular mechanism for altered ERG responses in STZ-induced diabetes in rats. Recognition of these early neurosensory defects would enable a better understanding of the pathophysiological basis of diabetic retinopathy.

Diabetic retinopathy, a common complication of diabetes, is a leading cause of blindness in the United States. ¹ Although diabetic retinopathy is diagnosed clinically by the appearance of abnormalities in the microvasculature, neural deficits often occur early in the disease. For example, psychophysical tests have shown color vision defects and a reduction in contrast sensitivity, ² and neurosensory changes have been detected before the onset of observable retinopathy by means of the flash electroretinogram (ERG) ³ and multifocal electroretinography. ^{4 5} These neurosensory defects can occur in the absence of retinopathy, but tend to be predictive of proliferative disease that ultimately leads to degenerative changes and significant visual impairment. ^{6 7 8 9 10} Studies in animal models of diabetes have also revealed early functional changes in the retinas of streptozotocin (STZ)-treated rodents, ^{11 12 13 14} alloxan-treated rabbits, ¹⁵ and pancreatectomized cats. ¹⁶  

The prevalence of diabetes is approximately equal among men and women, ¹⁷ but most experimental investigations of the changes in visual function have focused on STZ-induced diabetes in male rats. Although some assays comparing male and female rats show similarities in the severity of diabetes after administration of STZ (e.g., nearly equivalent serum glucose and insulin levels), ^{18 19} others have found gender differences in response to multiple low-dose injections of the drug. ^{20 21} Moreover, male and female rats display different physiological responses to STZ-induced diabetes in weight loss, ¹⁹ cardiovascular function, ^{22 23} antioxidative enzyme production, ²⁴ and neuroendocrine responses, ²⁵ as well as in response to islet transplantation. ¹⁸ In almost every instance, the dysfunction has been shown to be more severe in male animals. 

A further point of interest concerns the fact that with few exceptions, ^{11 26} most studies of visual function have used male albino rats. This is rather surprising considering that the visual system of albino rats is known to exhibit structural and functional differences when compared with their pigmented counterparts, ^{27 28} and their retinas have been shown to be more susceptible to photic damage, ²⁹ degeneration resulting from genetic abnormalities, ^{30 31} and ischemia-induced injury. ³²  

The present study was undertaken to evaluate the effects of diabetes on retinal function in STZ-treated female pigmented (Long-Evans) rats. Specifically, we used ERG recordings to examine the neurosensory changes in early diabetes, patch-clamp recordings from bipolar cells isolated from the retinas of normal and diabetic rats to determine whether there were differences in their responses to GABA (γ-aminobutyric acid), and immunochemistry to visualize the cellular distribution of GABA and the GABA_C receptor. The ERG has been used for decades to study the functional integrity of normal and diseased retina, ^{33 34 35} and it affords a quantitative, objective, noninvasive method to examine the neurosensory changes in early diabetes. Although there are conflicting reports on whether there are changes in the a- and b-wave potentials of the flash ERG, a consistent observation has been that the oscillatory potentials (OPs) are diminished and/or delayed in diabetes. ^{3 36 37} The OPs comprise four to six wavelets that are present on the rising phase of the ERG b-wave. OPs have been shown to arise within the proximal retina, ³⁸ and the individual peaks have different retinal depth profiles, suggestive of distinct cellular origins. ³⁹ Indeed, the specific retinal cells that are responsible for the OPs are still being debated, but there is strong evidence that GABAergic neurons, and their synaptic interactions, are key elements in the generation of the response. ³⁹  

We also measured the time course of neuronal dark adaptation, as well as the ERG response to a flickering-light stimulus of various intensities. The recovery of sensitivity after a brief flash that bleaches an insignificant fraction of the available rhodopsin has been shown to reflect mainly changes in neural activity during the adaptive process, ^{40 41} whereas the responses to intermittent illumination provide an index of the temporal resolution of the retina. In our results, the effects of hyperglycemia on visual function appeared to be milder in female rats than in their male counterparts. Nevertheless, the changes observed in OPs were consistent with an alteration of the GABA signaling pathway in the diabetic retina, as were enhancement of the GABA response from individual bipolar cells and the abundant accumulation of GABA in the inner retina. 

Methods

Animals

Female Long-Evans rats (80 ± 3 days old; 218 ± 6 g), anesthetized by intraperitoneal (IP) injection of ketamine hydrochloride (100 mg/kg) and xylazine (5 mg/kg), were given 60 mg/kg STZ (Sigma-Aldrich, St. Louis, MO) dissolved in 0.9% Na citrate buffer (pH 4.5) by intravenous administration into the femoral vein. Age-matched animals served as control subjects. Subsequent measurements of blood glucose levels were obtained by expressing a small amount of blood from the lateral tail vein after puncture with a 22-gauge needle and analyzing the sample with a blood glucose meter (Ascensia Elite; Bayer, Leverkusen, Germany). A plasma glucose level more than 350 mg/dL was considered diabetic. Animals were housed in pairs in the Biological Resources Laboratory (BRL) of the University of Illinois at Chicago in a 14/10-hour light–dark cycle, the standard lighting regimen of the BRL. Testing was performed at various time points over an extended period after the STZ injection. It is unlikely that the longer period of photic exposure provided some measure of protection by suppressing the oxygen demand of photoreceptors in our hyperglycemic rats. ⁴² Nevertheless, to obviate this potential factor, normal and hyperglycemic rats were raised in a variable light cycle in which they were exposed to 300 lux for 12 hours (7 AM to 7 PM) and to a reduced luminance of 100 lux for the 12 evening hours. Although dark-adapted b-wave responses were mildly enhanced in both groups of animals after being reared under these conditions, the hyperglycemia-induced pattern of OP responses were still observed in our experimental animals (data not shown). Food and water were provided ad libitum, and all animals were treated in accordance with institutional guidelines and the principles of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Analysis of Lens and Vitreous Humor

After the ERG recordings were completed, animals were killed, and their lenses were excised according to an established protocol. ⁴³ Briefly, a small incision was made just posterior to the ora serrata, and a full-thickness incision was extended circumferentially to separate the anterior segment from the posterior eyecup. The lens was then carefully separated from its capsule and photographed under dark-field illumination with a dissecting microscope. Vitreous humor was collected from the posterior eyecup, and its osmolarity was measured with a vapor pressure osmometer (Vapro Model 5520; Wescor, Logan, UT). 

Electroretinography

Recordings were obtained from eight control rats and seven diabetic animals after overnight dark adaptation. Under dim red light illumination, animals were anesthetized by IP injection of ketamine hydrochloride and xylazine. Before the recording was begun, the pupil was fully dilated with topical phenylephrine HCl (2.5%) and tropicamide (1%), and a topical anesthesia (proparacaine, 0.5%) was administered. The electrical responses were recorded with a silver wire corneal electrode, forehead reference electrode, and ground electrode in the tail. Ganzfeld stimuli were delivered (UTAS-E 3000 ERG system; LKC Technologies, Gaithersburg, MD). Body temperature was maintained at ∼37°C with a heating pad. All the animals tested recovered from anesthesia after the ERG recording sessions. 

ERG responses were recorded in response to brief (<1 ms) white light flashes from the Xenon arc delivered at 10-second intervals for dim stimuli and for intervals up to 60 seconds for high-intensity flashes. The range of stimulus intensities extended from 1.6 × 10⁻⁵ to 11 cd-s/m², and each data point was derived from the average of three recordings. The ERG waveforms were recorded with a band width of 0.3 to 500 Hz and sampled at 2 kHz by a digital acquisition system (LKC Technologies, Gaithersburg, MD) and were analyzed in a custom-built program (MatLab; The MathWorks, Natick, MA). Plots of the intensity–response functions for both the a- and b-waves were fit to a hyperbolic (Michaelis-Menten) function of the form  

\[R/R_{\mathrm{max}}{=}I^{n}/(I^{n}{+}K^{n})\]

where R is the response amplitude at flash intensity I, R _max is the amplitude of the maximum response that can be elicited, K is the intensity that evokes a half-maximum response, and n represents a dimensionless variable that describes the slope of the intensity–response function. 

OPs occur as a stereotypical series of high-frequency oscillations superimposed in time on the rising phase of the b-wave in response to a high-intensity light flash. A digital band-pass filter (40–200 Hz) was used to isolate the OP components from the original traces, ⁴⁴ and the peak amplitudes of the first five wavelets were measured from baseline. A low-pass filter (50 Hz) removed the OPs and enabled direct amplitude measurements of the a-wave (baseline to negative peak) and b-wave (a-wave trough to highest positive peak) potentials. Flicker responses, tested at frequencies of 4, 8, and 16 Hz, with flash intensities of 8.1 × 10⁻⁴, 3.5 × 10⁻², and 0.35 cd-s/m², were recorded after 2 seconds of preadaptation to each stimulus and analyzed by fast Fourier transform (conducted in MatLab; The Math Works). The flicker responses were recorded at the end of each session to avoid light-adapting the retina. 

Cell Isolation and Patch-Clamp Recording

Solitary bipolar cells were isolated from the rat retina according to published protocols. ⁴⁵ Briefly, eyes were hemisected, and the retinas gently removed from the posterior eyecup and immersed for 40 minutes in a modified Ames medium (supplemented with 0.88 g/L NaCl, 2.36 g/L HEPES, and 10,000 U/L penicillin/streptomycin [pH 7.4]) containing 2 mg/mL papain (EMD Biosciences, San Diego, CA) and 1 mg/mL of l-cysteine (Sigma-Aldrich). After several brief washes, the tissue was triturated through a sterile pipette, and aliquots of the supernatant containing dissociated cells were placed in culture dishes with modified Ames medium. The cells were maintained at room temperature for up to 8 hours, and before recording, the culture medium was replaced with an extracellular solution consisting of (in mM) NaCl 120, KCl 5, NaHCO₃ 25, CaCl₂ 2, MgCl₂ 1, HEPES 10, and dextrose 10 (pH 7.4). Whole-cell membrane currents were recorded with a patch pipette filled with an intracellular solution containing (in mM) CsCl 130, KCl 4, CaCl₂ 1, MgCl₂ 2, EGTA 11, HEPES 10, MgATP 1, and NaGTP 0.2 (pH 7.4). The signals were fed to the input stage of an amplifier (Axopatch 200B), and all protocols were controlled by the system software (pCLAMP; Axon Instruments, Union City, CA). The GABA-elicited responses were recorded with the membrane potential held at −60 mV. Data were plotted with commercial software (Origin; Microcal Software, Northampton, MA). 

Immunohistochemistry

The posterior eyecups were fixed in 4% paraformaldehyde in 0.1 M Sorenson’s phosphate buffer (PB; pH 7.4) for 20 minutes at room temperature, washed four times in PB, and cryoprotected by stepping through 10%, 20%, and 30% sucrose overnight at 4°C. The tissue was then embedded in optimal cutting temperature compound (OCT), sectioned at 10 μm, and mounted (Superfrost Plus Slides; Fisher, Pittsburgh, PA). Retinal sections were stained with primary antibodies, a polyclonal antibody directed against the GABA_C receptor (1:100; kindly provided by Ralf Enz), and a mouse monoclonal anti-GABA antibody (1:200, Sigma-Aldrich). Exposure was overnight at room temperature, followed by a 40-minute incubation with the secondary antibodies (TRITC-tagged donkey anti-mouse, 1:200, and Cy5-tagged donkey anti-rabbit, 1:200; Jackson ImmunoResearch Laboratory, West Grove, PA). Confocal and differential interference contrast (DIC) image acquisition was conducted with a confocal microscope (SB2-AOBS; Leica, Solms, Germany). 

Results

Characterization of Diabetic Rats

Blood Glucose.

In this study, animals with a plasma glucose level greater than 350 mg/dL were considered diabetic. Those in the control group maintained blood glucose levels of approximately 104 mg/dL over the course of the study. As shown in Figure 1A , injection of STZ induced severe hyperglycemia, resulting in an average blood glucose level in excess of 400 mg/dL (mean 446 ± 80; range, 352–598). Most animals became diabetic within 48 hours, but in some the drug took 72 hours to produce its effect. One animal failed to achieve a blood glucose level of more than 350 mg/dL after 72 hours and was excluded from the study. Diabetic animals also manifested polydipsia, polyuria, and glucosuria. Abdominal organomegaly was also noted, but no attempt was made to quantitate organ weights. Diabetic animals also had impaired growth relative to control animals. In the control group, animals exhibited a weight gain of approximately 4 g/wk and achieved an average body weight of 274 g at the 12-week time point. In contrast, age-matched diabetic animals showed no significant increase in body weight over the course of 12 weeks (Fig. 1B) . The fact that female rats show little or no weight gain after the onset of hyperglycemia has been reported by other investigators. ^{18 19 22}  

Lens Opacities.

Control animals showed no signs of lens opacification at any time during the study, whereas mild cataracts were observed in the diabetic animals during the later stages of the study. Figure 1Cshows examples of dark-field images of the lenses obtained from a control animal (i) and from rats after 15 and 25 weeks of diabetes (ii and iii, respectively). Evaluation of the lenses isolated from rats after 12 to 15 weeks of diabetes revealed cortical cataracts (grade II to II+), characterized by the presence of visible posterior sutures and isolated vacuoles in the lens cortex (ii). By 20 to 25 weeks, lenses isolated from a separate group of diabetic animals had grade III cataracts with coalescing vacuoles that extended to the posterior pole (iii). In all cases, the effects were bilateral—that is, lenses from the same animal were of the same grade. Osmolarity measurements of the vitreous humor at the time of lens extraction indicated a significant (P < 0.001) elevation in diabetic rats (336 ± 7 mOsM; n = 5) compared with control subjects (314 ± 5 mOsM; n = 7), a possible cause of cataract formation in the diabetic rat. ⁴⁶  

Electrophysiology

The Flash ERG.

Figure 2Aillustrates typical ERG waveforms elicited by a brief Ganzfeld flash at various intensities from a control rat (first column) and a diabetic rat (second column) at the 12-week time point. The averaged intensity–response functions for a- and b-waves are shown in Figure 2B , where the data are presented as the mean ± SEM. Although there was a slight enhancement of the a-wave amplitude in diabetic animals, the difference was not statistically significant, nor was there a significant difference in the b-wave intensity–response function between diabetic and control animals. ERGs recorded at other time points (4, 8, and 20 weeks) gave similar results (data not shown). The a- and b-wave data were well fit by the Michaelis-Menten equation, with K = 0.21 and 3.5 × 10⁻³ cd-s/m² for a- and b-waves, respectively, in control animals, and 0.18 and 3.8 × 10⁻³ cd-s/m² for the a- and b-waves in diabetic rats. The slopes (n) of the corresponding intensity–response functions were also nearly identical. 

Neuronal Dark Adaptation.

The b-wave, generated primarily by the light-evoked radial currents of ON bipolar cells, ^{47 48} was used to determine whether diabetes alters the kinetics of neural adaptation in the distal retina. ERG responses to a weak test stimulus (8.1 × 10⁻⁴ cd-s/m²) were recorded at 5-second intervals after a brief conditioning flash of 11 cd-s/m². This value converts to a retinal illuminance of ∼10 scotopic Troland seconds, estimated to bleach less than 1% of the rod photopigment. ^{49 50} Under these conditions, the temporal course of response recovery reflects mainly neuronal activity. ^{40 41} Examples of typical ERG responses obtained from control and diabetic rats (12 weeks after injection) are shown in Figure 2Cfor various times after exposure to the adapting flash. The two sets of responses showed nearly equivalent peak amplitudesduring the course of dark adaptation, and plotting the normalized amplitude values of the probe response recorded during 90 seconds of dark adaptation (Fig. 2D)revealed the similarity in the kinetics of neuronal adaptation. In sum, there is no significant difference in the amplitudes of the responses from the two groups and the time course of response recovery for control and diabetic animals can be fitted by a single exponential function with equivalent time constants (τ) of 14.5 and 14.2 seconds, respectively. 

Oscillatory Potentials.

It is well documented that diabetes leads to alterations in the OPs of the flash ERG in humans and animal models. ^{7 8 11} Typically, these changes involve reduced amplitudes and delays in the times-to-peak of various wavelets. Figure 3Ashows the OP responses of the filtered ERG recordings from a control and a diabetic animal at a light intensity of 1.3 cd-s/m²; the first five OPs are enumerated on each trace. At this 12-week time point, it is apparent that the OPs from the diabetic rat differed from those of its normal counterpart. The first three wavelets appeared to be smaller, whereas the fourth and fifth tended to be greater than those in the control animal. Of interest, the times-to-peak of each wavelet were unchanged in the diabetic animals, suggesting that the amplitudes, but not the latencies of the responses were affected by the hyperglycemic condition in these female rats. The effect of diabetes on the pattern of the OP wavelets is shown in Figure 3B , which plots the averaged amplitudes of the individual OPs obtained at several light intensities for the eight control and seven diabetic rats, respectively. In both groups of animals, the OP amplitudes increased as the light intensity was increased. However, the maximum OP amplitude was seen in the third OP in control animals, whereas OP4 had the highest amplitude in diabetic rats at all light intensities tested. The small negativity seen in OP2 when tested at 0.11 cd-s/m² arises from the fact that at this low intensity the peak potential was below the baseline from which response amplitudes were measured. An example of the changes in OP amplitude observed during a 12-week time period is illustrated in the bar graph of Figure 3Cfor the fourth wavelet of the response. At the beginning of the study (week 0), there was no statistical difference in the amplitude of OP4 between the control group and the group of animals designated for STZ injection. At the 4-week time point, the amplitude of the fourth OP had increased significantly in the diabetic animals, and the increase continued at 8 and 12 weeks after injection. These differences persisted in a pair of diabetic and control animals that we observed for nearly 6 months (data not shown). The reduction in the amplitude of the third OP in diabetic rats followed a similar time course (data not shown). 

Flicker ERG.

Responses to intermittent stimuli were recorded at frequencies of 4, 8, and 16 Hz with flash intensities of 8.1 × 10⁻⁴, 3.5 × 10⁻², and 0.35 cd-s/m². Typical flicker ERG waveforms from a control and a diabetic rat are shown in Figure 4A . The ERG responses to a flickering light contain multiple components (note the small second peak that emerges in the 4-Hz recordings at 3.5 × 10⁻² cd-s/m²), and do not lend themselves to a simple observational analysis. The fast Fourier transform has proven valuable in this regard. For a pulse train consisting of a series of flashes with intensity V, duration t ₀, and period T, the Fourier transform in the frequency domain is given by the expression  

\[x(t){=}V\left[k{+}\frac{2}{{\pi}}\left(\mathrm{sin}\ k{\pi}\ \mathrm{cos}\ {\omega}t{+}\ \frac{1}{2}\mathrm{sin}\ 2k{\pi}\ \mathrm{cos}\ 2{\omega}t{+}{\ldots}{+}\frac{1}{n}\mathrm{sin}\ nk{\pi}\ \mathrm{cos}\ n{\omega}t{+}{\ldots}\right)\right],\]

where k is t ₀/T and ω is 2π /T. When the duration of each flash is very brief (i.e., t ₀ is small compared with T, as is the case with a xenon flash), k is negligible and the Fourier series reduces to  

\[x(t){=}V{[}k{+}2k(\mathrm{cos}\ {\omega}t{+}\mathrm{cos}\ 2{\omega}t{+}{\ldots}{+}\mathrm{cos}\ n{\omega}t{+}{\ldots}){]}.\]

In other words, for a flickering stimulus comprising a series of such flashes, the frequency domain consists of a series of harmonics, each with equal energy. 

To compare the temporal resolution of the two groups of animals, we analyzed the flicker ERG responses in the frequency domain, and the averaged results are illustrated in Figure 4B . The three groups of graphs show the mean amplitudes of the fundamental and the second harmonic of the flicker responses at three light intensities. There were no significant changes in the implicit times for the flicker response at any stimulus intensity. It is interesting that the amplitudes of the 8-Hz component were similar, irrespective of whether they were derived as the second harmonic from a 4-Hz stimulus or as the fundamental of an 8-Hz stimulus. Despite the fact that the 8-Hz stimulus contained double the light intensity of the 4-Hz stimulus, the small difference in 8-Hz responses suggests that rat flicker ERG responses are essentially linear. The same is true for the amplitude of the 16-Hz response component. Linearity in the temporal response has also been observed in the mouse flicker ERG. ⁵¹  

At the lowest stimulus intensity, there was clearly no difference in the amplitudes of either the first or second harmonic between control and diabetic rats. At higher intensities, however, significant differences in both harmonics emerged. Specifically, a reduction in the first harmonic of the 8-Hz flicker ERG was seen in diabetic rats at the intermediate intensity of 3.5 × 10⁻² cd-s/m², and in the responses to both 4- and 8-Hz flicker at the highest intensity (0.35 cd-s/m²). 

GABA-Mediated Responses of Bipolar Cells.

The altered OPs in diabetic rats suggest that the GABA signaling pathway in the retina may be compromised. Accordingly, we compared the GABA-induced currents of individual bipolar cells enzymatically dissociated from normal and diabetic retinas. The majority of bipolar cells isolated from the rat retina are rod bipolar cells, ⁵² and their morphology has been described previously. ⁵³ A typical example (Fig. 5A)shows that these cells have a rounded cell body, extend several dendrites, and possess a long axon that ends with a prominent synaptic terminal. We found that more than 80% of cells morphologically identified as bipolar cells stained for PKCα, a specific marker for rod bipolar cells (data not shown). An example of the GABA (3 μM)-elicited membrane current recorded from these cells is shown in Figure 5A . With the cell held at −60 mV, the application of GABA generated a large, sustained inward current that deactivated slowly when the drug was withdrawn. Note, however, that with the addition of 100 μM bicuculline, the response was reduced by only 20% and deactivation was further retarded, indicating that the response is generated primarily by activation of GABA_C receptors, with only a small contribution from GABA_A receptors. The predominance of GABA_C receptors on rod-driven rat bipolar cells was reported previously in several studies. ^{54 55} When we recorded the dose–response relation of rod bipolar cells from control and diabetic rats (Fig. 5B) , it was apparent that both the maximum current response and the sensitivity to GABA were greater in cells derived from the diabetic animals. These results provide direct evidence that the GABA signaling pathway is altered in the diabetic retina. 

Immunohistochemistry

The striking differences in the GABA-induced responses of bipolar cells from diabetic animals prompted us to examine the GABA content and GABA_C receptor expression in the retina by immunohistochemical techniques. An example of GABA immunostaining from control and diabetic retina is shown in Figures 6B and 6E . Under our experimental conditions, the normal retina was very weakly labeled by the GABA antibody (Fig. 6B) . In contrast, antibody labeling of the hyperglycemic retina revealed a significant increase in the GABA content of the inner retina (Fig. 6E) , most notably staining a subgroup of cells in the inner nuclear layer (INL) and their processes in the inner plexiform layer (IPL). From the DIC image shown in Figure 6D , it is likely that the labeled cells are GABAergic amacrine cells and/or reflect GABA uptake into Müller cells. ⁵⁶  

We also studied expression of the GABA_C receptor in the normal and diabetic rat retina (Figs. 6C 6F) . In agreement with earlier studies, ⁵⁷ we found that the majority of GABA_C receptors were expressed in the IPL, presumably on the bipolar cell terminals. In addition, faint staining was detected in the outer plexiform layer (OPL), and on a subset of cell bodies in the INL. A similar GABA_C receptor expression pattern was observed in the diabetic retina (Fig. 6F) . 

Discussion

To our knowledge, this is the first study conducted to examine the effects of STZ-induced hyperglycemia on retinal function in a female animal model of diabetes. Although ERG responses in female diabetic animals have been reported, ⁵⁸ the animals that were studied possessed other severe ocular abnormalities or were shown subsequently to develop spontaneous retinal degeneration. ^{59 60} It would be difficult to ascribe the changes in retinal function to any specific diabetes-induced abnormality in the presence of such advanced ocular disease. Nevertheless, the results of the present study show that in many respects, the STZ-induced hyperglycemic response of female Long-Evans (pigmented) rats differs from that of male rats described in earlier studies. This is particularly evident in the ERG recordings. Although several studies have reported STZ-induced reductions in the a- and b-waves of the ERG, ^{11 12 56} we find these response components to be entirely comparable to those recorded from normal control animals (Figs. 2A 2B) . Moreover, the time course of neuronal dark adaptation, as seen in the recovery of the b-wave after a brief light flash, was unaffected in the diabetic retina (Figs. 2C 2D) . Thus, our findings indicate that the functional integrity of the distal retina is not affected by hyperglycemia, at least over the 12-week period of the study. 

It is possible that hormonal differences between male and female rats may be one of the factors responsible for the distinct gender difference in the ERG responses of diabetic rats. Several studies have suggested that estrogen is protective of the kidney in diabetes, ^{61 62} and it has been shown to exert an effect on the physiology of the retinal microvasculature. ⁶³ Although it is highly speculative, the relatively mild functional deficits that we observed in the female rat model of diabetes suggest that estrogen could alleviate or retard the development of diabetic retinopathy. This notion is consistent with experimental studies showing that testosterone enhances the hyperglycemic response in castrated and noncastrated female rats, ²¹ and that replacement of 17-β estradiol to STZ-induced diabetic ovariectomized female rats attenuated the hyperglycemia and preserved renal transport functions. ⁶⁴  

A clear difference between male and female diabetic rats was also seen in the OP recordings. Typical findings in STZ-treated male rats, as well as in diabetic humans, include significant delays in the onset of the various OP components, and reductions in the amplitudes of many of the wavelets. ^{3 11 56} In contrast, the female diabetic rats did not exhibit any significant changes in the timing of the individual OPs. However, there was a marked change in the overall pattern of the OP components, resulting from significant reductions in the amplitudes of the early OPs and enhancement of later OPs (Fig. 3) . This pattern change in the OP response is similar to that observed in diabetic humans. ⁶⁵ It has been postulated that early and late OPs originate from different populations of inner retinal neurons. ^{37 66} Thus, the shift in the OP pattern in the hyperglycemic rat suggests a redistribution of the synaptic strength of various inhibitory circuits in the inner retina. We also observed a progressive, age-related decline in OP amplitude in the normal (control) Long-Evans rats (Fig. 3C) . 

We are not aware that the flicker ERG has been studied in animal models of diabetes, but there is evidence of a delayed implicit time and reduced amplitude flicker response in diabetic patients. ^{67 68 69} In our study, the diabetic female rats did not show significant changes in latency, but exhibited reduced amplitude flicker responses (Fig. 4) . The reduction was significant with both the 4- and 8-Hz stimuli at the highest flash intensity (0.35 cd-s/m²), whereas at intermediate intensity (3.5 × 10⁻² cd-s/m²), the reduced flicker response for diabetic animals was observed in both the fundamental harmonic response to the 8-Hz flicker, as well as in the second harmonic response to the 4-Hz stimuli. No significant differences were seen at the lowest stimulus intensity (8.1 × 10⁻⁴ cd-s/m²). The fact that the diabetes-induced changes in the responses to intermittent stimulation occurred only at the higher (photopic) stimulus intensities suggests that the cone pathway may be more susceptible to hyperglycemia in the rat model of STZ-induced diabetes. It is noteworthy that increased blood flow to the inner retina induced by intermittent photic stimulation is compromised in diabetes. ⁷⁰ There is the possibility that the reduction we observed in the neural responses to intermittent stimuli is responsible for the decline in the flicker-induced vascular reaction noted in the retina of diabetic subjects. 

It has been shown that the GABAergic mechanisms of the inner retina play an important role in the generation of OPs, ^{38 39} although neither the cellular origins nor the specific mechanisms that generate the potentials have been identified. It was not surprising that we found that the GABA responses of GABA-sensitive bipolar cells and the GABA distribution within the retina were significantly altered in diabetic rats. Membrane current recordings from isolated bipolar cells indicated that the GABA_C receptor-mediated responses of diabetic animals exhibited a greater sensitivity and larger maximum currents to the application of GABA than did those derived from control rats (Fig. 5) . In addition, the enhanced accumulation of GABA we observed in the diabetic retina (Fig. 6)is consistent with reports that diabetes enhances GABA synthesis and reduces GABA degradation. ⁵⁶  

We noted that the expression pattern of the GABA_C receptor was not significantly altered by the hyperglycemic condition (Fig. 6) . Nevertheless, there was an ∼20% increase in the maximum GABA_C receptor mediated currents from bipolar cells of the diabetic retina. If the increased current responses were mediated solely by the enhanced expression of GABA_C receptor protein, it is unlikely that the difference would be seen by immunohistochemistry. The approximately two- to threefold increase in the EC₅₀ and the significant differences in the slopes of the dose–response relation of the GABA_C receptor-mediated response suggest that the GABA_C receptor itself is modified by hyperglycemia (Fig. 6) . Although the precise mechanism for this difference is unknown, it is interesting to note that GABA_C receptors on rat bipolar cells are heteromeric complexes of ρ1 and ρ2 subunits, ⁷¹ formed most likely by random assembly of the GABA ρ subunits. ^{72 73} Clearly, the observed differences in GABA activity in diabetic rats may result from alterations in the transcription/translation efficacy of GABA ρ subunit genes. 

Although diabetic retinopathy in humans is often diagnosed clinically by the detection of significant vascular abnormalities, and frank retinopathy can typically be observed ophthalmoscopically or by fluorescein angiography, these pathognomonic signs are not seen in STZ-induced hyperglycemic rats. However, by selectively staining leukocytes with intravenous acridine orange, changes were observed in their velocity within the retinal capillaries of STZ-injected rats, ⁷⁴ and alterations in neuronal function have been well documented in both human patients and animal models. ^{3 36 37} The observed alterations in the flicker ERG and OP recordings and related changes in the GABAergic pathway of the inner retina may represent fundamental changes induced by the onset of diabetes. Recognition of these early neurosensory defects will enhance our understanding of the pathophysiological basis of diabetic retinopathy and will also provide a foundation for designing novel screening tests and developing new therapeutic approaches to treat diabetic patients. 

 Supported by Grants EY12028, EY06516, and EY-01792 from the National Eye Institute; Research To Prevent Blindness (RPB); the Illinois Society to Prevent Blindness; the Midwest Eye Banks; and a UIC Provost Award (DJR).

 Submitted for publication April 3, 2006; revised June 6 and July 24, 2006; accepted September 22, 2006.

 Disclosure: D.J. Ramsey, None; H. Ripps, None; H. Qian, None

 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) Blood glucose levels of diabetic and control rats as a function of time after the induction of diabetes. Control and diabetic animals had similar blood glucose levels before the injection of STZ. Rats in the control group maintained their blood glucose levels at ∼104 mg/dL over the course of the study, whereas blood glucose levels in the STZ-treated rats increased by more than fourfold, to a mean of 446 mg/dL (*** P < 0.001). (B) Control rats gained weight throughout the experiment, with an average growth rate of approximately 4 g/wk. Diabetic rats did not show significant weight gain over the course of the 12-week study. The differences in body weight between control and diabetic animals were statistically significant at and after 4 weeks of STZ induced diabetes (* P < 0.05; ** P < 0.01). (C) Rat lenses isolated and photographed from control (i) and diabetic rats at 15 (ii) and 25 (iii) weeks after the administration of STZ. Note the appearance of the posterior lens sutures and the development of cortical vacuoles in the periphery of the diabetic lenses. Scale bar, 500 μm.

Figure 2.

 

(A) Representative ERG waveforms from dark-adapted control and diabetic rats at the 12-week time point. The flash intensity used to elicit the responses is given to the left of each pair of responses in cd-s/m², and the inverted triangles indicate the onset of the light stimuli. (B) The intensity–response functions for the two groups of animals after low-pass filtering of the potentials. The continuous curves are a- and b-wave data fit to the Michaelis-Menten equation with K values of 0.21 and 3.5 × 10⁻³ cd-s/m² and an n value of 1.1 and 0.64 for a- and b-waves, respectively, in control animals, and 0.18 and 3.8 × 10⁻³ cd-s/m² and an n value of 1.2 and 0.62 for the a- and b-waves in diabetic rats. (C) Examples of the ERG responses from control and diabetic rats in response to a fixed stimulus intensity (8.1 × 10⁻⁴ cd-s/m²) delivered at various times after a brief conditioning flash (11 cd-s/m²). (D) b-Wave amplitudes (normalized to the response maximum) recorded at 5-second intervals during 90 seconds of dark adaptation. For both groups of animals, the time course of the recovery of sensitivity can be fit by a single exponential function with a time constant (τ) of ∼14 seconds.

Figure 3.

 

(A) OPs, isolated by the band-pass filter, recorded from control and diabetic rats in response to a flash of 1.3 cd-s/m². The first five OPs have been enumerated on each trace. (B) Averaged OP amplitudes at various light intensities. Note that the maximum amplitudes were seen in the third OP for control animals at all the light intensities tested; in diabetic rats the maximum shifted to the fourth OP. (C) The amplitude of the fourth OP from control and diabetic rats during the course of the study. Whereas no significant difference in OP4 amplitude was seen at the beginning of the study, the responses from diabetic rats grew rapidly with time (* P < 0.05; ** P < 0.01).

Figure 4.

 

(A) Representative ERG responses to intermittent stimulation in control and diabetic rats at the 12-week time point. The flicker responses were recorded at three light intensities for frequencies of 4, 8, and 16 Hz. (B) The amplitude of the fundamental and the second harmonic responses of the flicker ERG are plotted for three light intensities. Note that no significant differences were observed between control and diabetic rats when tested with low-intensity flicker, but a statistically significant (* P < 0.05) reduction of the flicker response was evident at medium and high light intensities.

Figure 5.

 

(A) Phase-contrast image of a rod bipolar cell isolated from the rat retina (inset). The membrane current response elicited by GABA (3 μM) was reduced by only ∼20% when coapplied with 100 μM bicuculline. It is apparent that the major component of the GABA-mediated response results from activation of GABA_C receptors. (B) Averaged dose–response data of 10 cells from diabetic animals and 11 bipolar cells from age-matched control retinas. The curves were fit by the hyperbolic equation with K of 1.65 and 4.2 μM and slopes (n) of 2.2 and 1.2 for diabetic and control cells, respectively.

Figure 6.

 

DIC images of transverse sections from control (A) and diabetic (D) retinas are shown together with immunohistochemical results after reacting the tissue with a monoclonal GABA antibody (B, E) and a polyclonal antibody to the GABA_C receptor (C, F) after a 12-week induction of hyperglycemia. OS, photoreceptor outer segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar, 50 μm.

The authors thank David Geenen, Manuela Zampino, and Milana Yuzhakova for help with the development of the STZ model; James Artwohl and staff at the BRL for veterinary support; Marek Mori and Gerald Giovannelli for instrument fabrication; Gleen Aduana for help with software development; Ralph Enz for the generous gift of the GABA_C receptor antibody; and Kenneth Alexander for providing the algorithm used for filtering the OP responses and many helpful discussions related to this project.