Loss of Scotopic Contrast Sensitivity in the Optomotor Response of Diabetic Mice

Abstract

Purpose.: Diabetes reduces retinal and visual sensitivity to dim light flashes. However, the impact of diabetes on contrast sensitivity in dim light is unknown. Based on the lowered visual sensitivity previously observed, we hypothesized that contrast sensitivity would similarly be reduced. We therefore examined scotopic contrast sensitivity of the optomotor response in the Ins2^Akita/+ mouse model of type 1 diabetes.

Methods.: A longitudinal study of spatial and temporal contrast sensitivity in Ins2^Akita/+ mice and wild-type Ins2^+/+ littermates was conducted. Contrast sensitivity of the optomotor response to rotating gratings of various spatial and temporal frequencies was measured at a dim luminance level (2.6 · 10⁻⁵ cd/m²) known to elicit rod- but not cone-driven responses.

Results.: An early, progressive loss in scotopic contrast sensitivity was observed in Ins2^Akita/+ mice that was absent from Ins2^+/+ littermate controls. The loss in contrast sensitivity developed over a 3- to 4-month period after the onset of hyperglycemia. Ins2^Akita/+ mice exhibited a nonselective 40% loss in sensitivity to all spatial frequencies and a selective loss in sensitivity to fast but not to slow varying gratings (temporal frequencies >0.1 Hz or, equivalently, speeds >3 deg/s). Such losses in sensitivity were prevented by glycemic control with insulin treatment.

Conclusions.: An association between a model of type 1 diabetes and scotopic contrast sensitivity of the optomotor response is indicated. Ins2^Akita/+ mice exhibit a uniform loss in optomotor contrast sensitivity to all spatial frequencies that, unexpectedly, can be explained as being secondary to a retinal or central loss in sensitivity to high temporal frequencies.

Introduction

Diabetic retinopathy (DR) is a microvascular disorder of diabetes and one of the major causes of visual impairment worldwide. ^1,2 Clinical expression of DR includes the presence of microaneurysms, intraretinal hemorrhages, neovascularization, edema, and loss of vision. ³ Several forms of retinal and visual dysfunction, however, are evident before the onset of clinical indicators. ⁴ Some of these early deficits include slow dark adaptation, ⁵ elevated rod and cone thresholds, ^6–8 diminished photopic contrast sensitivity, ⁹ and impaired electroretinographic (ERG) responses. ^6,10,11 The wide variety of retinal and visual deficits show as a potential consequence of diabetes acting at multiple stages of the visual pathway. ¹²  

Although any limitation in visual function is important, perturbations in contrast sensitivity can have particularly severe consequences. ¹³ Contrast sensitivity is a fundamental property of vision that provides a measure of the local difference(s) in luminance necessary to detect a target. ¹⁴ This sensitivity, which depends on both the spatiotemporal properties of the image and the levels of retinal illumination, ¹⁵ is closely related to a person's ability to perform common visual tasks. ¹⁶ A substantial number of diabetic subjects can experience a modest decline in contrast sensitivity before the onset of retinopathy, ^{5,9,12,14,17–22} a condition that seems to worsen progressively in relation to the degree of retinopathy. ^19,23 Importantly, the loss in contrast sensitivity occurs before diabetic subjects exhibit significant reductions in visual acuity, ^5,17–19 suggesting that changes in contrast sensitivity are an effective indicator of early visual dysfunction. 

To date, empirical investigations of the association between diabetes and contrast sensitivity have been restricted to photopic (bright light) conditions (i.e., when vision is driven largely by cone photoreceptors). Diabetes also impairs rod visual function, ⁶ however, and thus may potentially diminish contrast sensitivity during scotopic (dim light) conditions. As a first step toward addressing this issue, we examined scotopic contrast sensitivity of the optomotor response in the Ins2^Akita/+ mouse model of type 1 diabetes. This mouse line carries a dominant mutation in the Ins2 gene that induces, especially in males, spontaneous hallmarks of type 1 diabetes. ²⁴ Retinas exhibit vascular and cellular abnormalities that are characteristic of the early phase of the disease and are consistent with other models of diabetes. ^25,26 Ins2^Akita/+ mice also exhibit a progressive loss in the photopic contrast sensitivity of the optomotor response. ²⁷ Results reported here indicate that Ins2^Akita/+ mice experience an early, uniform reduction in scotopic contrast sensitivity to all spatial frequencies. Surprisingly, these mice also developed a significant deficit in the ability to detect fast-varying signals, with the ability to detect slow signals remaining normal. These observations suggest that diabetes significantly impairs rod-mediated visual mechanisms involved in temporal and/or motion processing of images. 

Methods

Animals

Male C57BL/6J Ins2^Akita/+ and Ins2^+/+ mice were purchased from Jackson Labs (Bar Harbor, ME) and maintained at State University of NY (SUNY) Upstate Medical University (Syracuse, NY). Ages of animals tested ranged from 2 to 8 months. Mice were fed Formulab diet (catalog no. 5008; Purina, St. Louis, MO) ad libitum and maintained on a 14-h light/10-h dark cycle. Animals were dark-adapted overnight prior to experiments and tested during the subjective day. Blood glucose levels were measured from tail vein samples with a glucose meter (One Touch Ultra; LifeScan, Milpitas, CA) at 3 to 5 h after lights were turned on. Protocols were approved by the Institutional Animal Care and Use Committee at SUNY Upstate Medical University and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

For most experiments in the study, blood glucose levels were not controlled and Ins2^Akita/+ mice were severely hyperglycemic (Fig. 1). In a subset of experiments designed to determine whether glycemic control could prevent loss of visual performance (Fig. 4), Ins2^Akita/+ mice were treated with sustained insulin release LinBit implants (LinShin; Toronto, Canada). Implants were inserted subcutaneously under dorsal skin. Because each insulin implant persists for approximately 30 days, mice received periodic implants to maintain their blood glucose near euglycemic levels during the 4-month study. 

Figure 1

 

Tracking blood glucose levels and scotopic visuomotor responses of Ins2^Akita/+ mice. (A) Average blood glucose levels from tail vein samples were plotted as a function of age in Ins2^Akita/+ mice (Akita) and their Ins2^+/+ siblings (CTL). Range of blood glucose measurements was limited by the glucometer to a maximum 600 mg/dL. Error bars indicate SEM; n = 16 and n = 8 for Ins2^Akita/+ and Ins2^+/+ mice, respectively. (B) Spatial resolution values of Ins2^Akita/+ and Ins2^+/+ mice did not change with age (P > 0.05, two-way RM ANOVA). Stimulus parameters were: contrast = 100%; speed of rotation = 12 cyc/deg; mean illumination = 2.6 · 10⁻⁵ cd · m⁻². (C) Contrast sensitivity of Ins2^Akita/+ mice declined with age (*P < 0.05, two-way RM ANOVA). Stimulus parameters were: spatial frequency = 0.064 cyc/deg; temporal frequency = 0.75 Hz. Mean illumination was the same as that shown in (B).

Visuomotor Response

Contrast visual sensitivity of mice was measured by observing optomotor behavior using a two-alternative forced-choice in combination with the OptoMotory apparatus ²⁸ (CerebralMechanics, Lethbridge, Canada) as described previously. ²⁹ Briefly, dark-adapted mice were placed on a pedestal located at the center of an enclosure formed by four video monitors that displayed the stimulus gratings. Head movements of mice were monitored by an observer using infrared illumination and a video camera positioned above the animal. The observer could observe only the animal and not the rotating pattern. The optomotor stimulus consisted of a vertically oriented sinusoidal pattern that rotated for 5-second periods. The direction of rotation was randomly selected by the computer-controlled protocol. The observer selected the direction of rotation based on the mouse movements and received auditory feedback indicating correct or incorrect response. Based on a two-alternative forced-choice method, the computer changed the contrast of the stimulus following a staircase paradigm ³⁰ that converged to a threshold value arbitrarily defined as 70% correct responses. ²⁸ Contrast sensitivity was defined as the reciprocal of the threshold value. Sensitivity for each mouse was estimated as the average of four independent trials. Results from trials differing by more than two standard deviations from the average were discarded. Spatial frequencies of the stimulus grating and speeds of rotation were varied independently. Luminance within the OptoMotory enclosure was attenuated with neutral density filters (Lee Filters, Burbank, CA) positioned between the video monitors and the mice. Light calibrations were performed as described previously. ^29,31  

To measure spatial contrast sensitivity functions, mice were presented with sinusoidal gratings of different spatial frequencies (0.031–0.236 cycles/degree of visual angle [cyc/deg]). Speed of grating rotation (0.5–48 deg/s) was adjusted according to the selected spatial frequency (f_s ) in order to keep the temporal frequency (f_t ) constant. The speed of grating rotation (s_p ) relates to the two independent variables f_s and f_t , as f_t = s_p f_s (Equation 1). We selected a temporal frequency of 0.75 Hz because it elicits maximal sensitivity during scotopic illumination. ²⁹ Analogously, temporal contrast sensitivity functions were measured by presenting mice with a sinusoidal grating at a constant spatial frequency (0.031 cyc/deg) and adjusting the speed of rotation to produce the desired temporal frequencies. 

To determine spatial frequency thresholds, we presented mice with sinusoidal gratings at 100% contrast and rotation speed of 12 deg/s. ²⁸ The spatial frequency of the stimulus was initially set at a relatively low value (0.2 cyc/deg) and systematically incremented in successive trials until the animals ceased to track the moving stimuli. Spatial resolution was defined as the highest spatial frequency to which the mice responded. 

Statistical Analysis

For the longitudinal experiments involving both Ins2^Akita/+ and Ins2^+/+ mice, two-way repeated measures analysis of variance (RM ANOVA) was used, with the nominal factors being genotype and age. Holm-Sidak's procedure for pairwise multiple comparisons was performed to test the hypotheses that (1) mean measurements obtained in Ins2^Akita/+ mice were not different from those in Ins2^+/+ mice and that (2) differences in measurements in Ins2^Akita/+ and Ins2^+/+ mice were independent of age. Logarithmic transformations of contrast sensitivity data were performed prior to statistical analysis to fulfill normality and equal variance requirements for ANOVA. In the case of the contrast sensitivity functions, two-way RM ANOVA was performed independently at each of the spatiotemporal frequency combinations. Data analysis was performed with SigmaStat software (Systat Software, San Jose, CA). Plot values are means ± standard error of the means (SEM). Numbers of mice and P values are indicated in the figure legends. 

Results

Elevated Blood Glucose Levels in Ins2^Akita/+ Mice

A longitudinal study of blood glucose levels and visual performance was conducted in Ins2^Akita/+ and control Ins2^+/+ mice. Because the goal was to examine early effects of diabetes on visual performance, we tested mice periodically over a 32-week period starting at 8 weeks of age. At this young age, mice were sufficiently mature for visuomotor testing and blood glucose levels for Ins2^Akita/+ mice had risen approximately 2.5-fold above normal (Fig. 1A). Over the next 24 weeks, Ins2^Akita/+ mice remained severely hyperglycemic, with their blood glucose levels consistently above 500 mg/dL. In contrast, blood glucose levels of control Ins2^+/+ mice remained at lower levels, approximately180 mg/dL, which is near euglycemic levels in C57BL/6J mice (the genetic background of the Ins2^Akita/+ line). ³² Consistent with the catabolic effects of diabetes, Ins2^Akita/+ mice weighed significantly less than Ins2^+/+ control mice at both 8 weeks (23.3 ± 0.2 vs. 24.5 ± 0.3 g; P < 0.05, n = 15) and 32 weeks of age (27.6 ± 1.2 vs. 31.1 ± 1.48, P < 0.05, n > 6). ²⁵ The expected blood glucose levels and weights of the experimental and control mice populations were thus confirmed. 

Age-Related Loss of Scotopic Contrast Sensitivity in Ins2^Akita/+ Mice

Visual performance of Ins2^Akita/+ mice was assessed by observing their visuomotor responses to moving sine-wave gratings. ²⁸ The independent parameters of the sine-wave stimuli were speed of motion and spatial frequency, the latter defined as the number of cycles in the sine-wave per degree of visual angle. We also controlled for two luminance-related aspects of the stimulus: (a) contrast, defined as the relative difference in luminance between the “peaks and valleys” of the sine-wave; and (b) mean luminance, estimated as the average luminance across one period of the grating. Because the goal was to study effects of diabetes on rod-driven vision, the mean luminance of the grating was set to relatively dim levels (2.6 · 10⁻⁵ cd · m⁻²), which, when accounting for attenuation by the pupillary reflex, produced fewer than 0.02 photoisomerizations/rod/s. ³¹ At such low luminance levels, the behavioral visual responses of mice are driven by rods, with no significant contribution from cone photoreceptors. ²⁹  

Spatial resolution thresholds and contrast sensitivity of Ins2^Akita/+ mice and Ins2^+/+ siblings was tracked during a 6-month period. We have previously used these tests to assess visual function in mouse models of retinal disease. ^33,34 Spatial resolution thresholds of Ins2^Akita/+ mice did not change significantly with age, closely matching those of control Ins2^+/+ mice at every time point (Fig. 1B). The absence of a significant phenotype for Ins2^Akita/+ in dim light was strikingly different from the loss in resolution that these same mice experienced in bright light, ²⁷ when the optomotor response is mediated by both rod and cone photoreceptors. ³¹ These results suggest that diabetes may differentially affect rod and rod/cone-driven aspects of vision. 

Ins2^Akita/+ mice experienced a significant (P < 0.05) age-related reduction in contrast sensitivity. For these tests, the spatial and temporal properties of the stimulus grating (f_s = 0.064 cyc/deg and f_t = 0.75 Hz) were chosen to elicit maximal sensitivity of the optomotor responses under scotopic conditions. ²⁹ Contrast sensitivities of Ins2^Akita/+ and Ins2^+/+ mice were similar at a young age (8–10 weeks). However, the sensitivity of Ins2^Akita/+ mice declined thereafter, stabilizing after 12 weeks to 50% to 60% of its initial value (Fig. 1C). Contrast sensitivity of control Ins2^+/+ mice did not change significantly over the same time period, confirming the fact that the losses in sensitivity exhibited by Ins2^Akita/+ mice were genotype-specific. These results indicate that the decline in contrast sensitivity occurred during a restricted period of development that began a few weeks after the onset of hyperglycemia (cf. Fig. 1A). 

Nonselective Loss in Spatial Contrast Sensitivity

The experiments described above revealed that Ins2^Akita/+ mice experienced age-related loss in scotopic sensitivity to moving gratings. The experiments did not establish, however, whether the loss in sensitivity arose from an impaired capacity to process the spatial and/or the temporal properties of the stimulus. To distinguish between these alternatives, we determined the spatial and temporal contrast sensitivity functions (CSF) for Ins2^Akita/+ mice (see Methods). In mice, these functions have a characteristic bell shape and are centered at the preferred (or optimal) frequency. ²⁹ Phenotypic changes in the shape of contrast CSFs are indicative of selective deficits in spatial or temporal vision. ³⁵  

To separate spatial from temporal influences, we determined the spatial CSF at a constant temporal frequency. Mice were presented with sinusoidal gratings of various spatial frequencies and the speed of the grating adjusted accordingly to maintain a constant temporal frequency (see Methods). In this context, temporal frequency is the number of cycles per second “viewed” by a single photoreceptor as the grating moves across the visual field. Stimulus presentation at 0.75 Hz was chosen because this temporal frequency elicits maximal sensitivity under scotopic illumination. ²⁹ Analogously, temporal CSFs were determined at a constant spatial frequency. The spatial frequency of the grating was sufficiently low (0.031 cyc/deg) to elicit strong, robust optomotor responses under dim light conditions. ²⁹  

CSFs were measured at two specific developmental time points: 8 weeks of age, which is before the mice experience a significant loss in contrast sensitivity; and 24 weeks of age, after the decline in contrast sensitivity has achieved a steady-state. At 8 weeks, the spatial CSFs of both Ins2^Akita/+ and Ins2^+/+ mice matched closely, peaking at 0.064 cyc/deg and exhibiting a gradual reduction in sensitivity at lower and higher spatial frequencies (Fig. 2A). The general properties of their CSFs were similar to those of wild-type C57BL6/J mice, their genetic background. ²⁹ Contrast sensitivity of Ins2^+/+ mice did not change at 24 weeks of age; however, contrast sensitivity of Ins2^Akita/+ mice was significantly reduced, particularly in response to low spatial frequencies (Fig. 3A). To further understand the effects of diabetes on contrast sensitivity, data were replotted using a logarithmic sensitivity axis. Following this transformation, the CSF measured at 24 weeks shifted vertically along the sensitivity axis, but its shape remained invariant with age. This analysis suggests that Ins2^Akita/+ mice experience a proportional reduction in sensitivity at all spatial frequencies. To illustrate this point, the ratios between contrast sensitivities at 24 and 8 weeks of age at several spatial frequencies were calculated. Each ratio remained close to 0.7, the proportionality or scaling factor (Fig. 3C). Results thus indicate that hyperglycemia but not age produces a uniform scaling in contrast sensitivity at all spatial frequencies tested. 

Figure 2

 

Normal contrast sensitivity of Ins2^Akita/+ mice at 8 weeks of age. (A) Spatial and (B) temporal contrast sensitivity functions of Ins2^Akita/+ mice (Akita) at 8 weeks of age (n = 8–12) and Ins2^+/+ mice (CTL) at 8 and 24 weeks of age (n = 4–8). Measurements of spatial contrast sensitivity were made at a constant temporal frequency of 0.75 Hz, while temporal contrast sensitivity measurements were made at a constant spatial frequency of 0.031 cyc/deg. Contrast sensitivity of Ins2^Akita/+ mice at 8 weeks of age is similar to that of Ins2^+/+ mice (P > 0.05, two-way RM ANOVA, pair-wise comparisons at each spatiotemporal frequency combination performed with Holm-Sidak's test). Contrast sensitivities of C57BL6/J mice were adapted from Umino et al. ²⁹

Figure 3

 

Age-related changes in contrast sensitivity in Ins2^Akita/+ mice (n = 8–12). Spatial (A) and temporal (B) contrast sensitivity functions of Ins2^Akita/+ mice at ages 8 and 24 weeks plotted with linear (top) and logarithmic (bottom) sensitivity axes. Measurements of spatial contrast sensitivity were made at a constant temporal frequency of 0.75 Hz; measurements of temporal contrast sensitivity were made at a constant spatial frequency of 0.031 cyc/deg. *P < 0.05, two-way RM ANOVA, pair-wise comparisons at each spatiotemporal frequency combination performed with the Holm-Sidak test. Spatial (C) and temporal (D) relative sensitivities were estimated as the ratios of the respective contrast sensitivities measured at 24 and 8 weeks. Data in (C) were fitted with a linear function (R ² = 0.9, n = 4) while data shown in (D) were well fitted by a decaying exponential function (R ² = 0.9, n = 5). The corresponding speed of rotation is shown on the second abscissa.

Selective Loss in Scotopic Contrast Sensitivity to High Temporal Frequencies

The temporal CSFs of Ins2^Akita/+ mice at 8 and 24 weeks of age were determined. Much like their spatial counterparts, the temporal CSFs were normal at 8 weeks (Fig. 2B). However, at 24 weeks, sensitivity to high temporal frequencies was selectively reduced, as is evident on plots with a linear or logarithmic sensitivity axis (Fig. 3B). The frequency-dependent reduction in contrast sensitivity ratios (0.9 at 0.1 Hz; approximately 0.7 at 1.5 Hz) confirms that the response to fast-varying signals was impaired (Fig. 3D) and is consistent with contrast sensitivity evaluated as a function of rotation speed (Fig. 3D, second abscissa in the same graph). Although temporal frequency was controlled in these experiments, Equation 1 (see Methods) can be applied to express the independent variable in terms of speed of rotation. From the graph, a reduction in contrast sensitivity to gratings rotating at speeds that exceed 3 deg/s was inferred. 

These results suggest that hyperglycemia associated with type 1 diabetes has dissimilar effects on the spatial and temporal (or speed) processing properties of the optomotor response. A proportional reduction in sensitivity to all spatial frequencies is produced while selectively limiting the sensitivity to faster temporal frequencies (or rotating speeds). Although the results cannot identify a proximate mechanistic cause of these visual deficits, a model that functionally relates the nonselective loss in spatial contrast sensitivity to deficits in temporal responses is suggested (see Discussion). 

Glycemic Control with Insulin Implants Prevents the Loss of Scotopic Contrast Sensitivity

Glycemic control in diabetic subjects is associated with a delay in the onset and reduction in severity of retinopathy and loss of visual acuity. ³⁶ We therefore investigated whether extrinsic control of blood glucose levels prevents the loss in contrast sensitivity observed in Ins2^Akita/+ mice at 24 weeks of age. Blood glucose levels in Ins2^Akita/+ mice were manipulated via subcutaneous, slow-release insulin implants (see Methods). Treatment began at age 10 weeks with the insertion of insulin implants and was discontinued at age 24 weeks, shortly after the assessment of contrast sensitivity (Fig. 4A). Dosage of the implants was adjusted to maintain blood glucose near normal levels (approximately 170 mg/dL). However, as the implants wore down (within 3–4 weeks after insertion), the blood glucose levels began to rise rapidly. In such instances, application of supplementary insulin implants promptly reduced blood glucose levels (Fig. 4A, arrowheads), although more often than not, the new inserts led to brief episodes of mild to moderate hypoglycemia. As a result, blood glucose levels transitioned from peaks of approximately 300 mg/dL down to values near 100 mg/dL. As a measure of treatment efficacy, the mean blood glucose levels were estimated during the treatment period (ratio of the area under the curve to the duration of the treatment) and compared to that of untreated Ins2^Akita/+ and Ins2^+/+ mice, computed over the same period. Our results indicate that treatment with insulin implants reduces mean blood glucose levels in Ins2^Akita/+ mice to nearly normal levels (Fig. 4B), although introducing strong glycemic oscillations with periods of 3 to 4 weeks. 

Effects of treatment with insulin implants upon scotopic contrast sensitivity of Ins2^Akita/+ mice at 24 weeks of age were determined. Stimulus parameter values (f_s = 0.031 cyc/deg, f_t = 0.75 Hz) were chosen to observe maximal reductions in sensitivity (Fig. 3). As in all experiments performed in this study, mean luminance of the stimulus was sufficiently dim (2.6 · 10⁻⁵ cd · m⁻²) to exclusively stimulate rod photoreceptors. Contrast sensitivity of insulin-treated Ins2^Akita/+ mice did not experience the reduction observed for untreated mice and was similar to that of control Ins2^+/+ mice (Fig. 4C). These results suggest that the developmental loss of scotopic contrast sensitivity in Ins2^Akita/+ mice is caused by a deficit in the production of “active” insulin and, despite periodic episodes of hypo- and hyperglycemia, can be prevented with glycemic control, using insulin implants. 

Figure 4

 

Glycemic control with insulin implants prevents loss of contrast sensitivity. (A) Tail vein blood glucose levels plotted as a function of age in an Ins2^Akita/+ mouse receiving a first insulin implant at 10 weeks (arrow), when the treatment began, and subsequently at the onset of each hyperglycemic event (arrowheads). Supplementary applications were usually followed by brief episodes of hypoglycemia, which gave rise to glycemic oscillations. Treatment duration was 14 weeks (gray bar). For reference, the average blood glucose levels of untreated Ins2^Akita/+ (Akita) and Ins2^+/+ (CTL) mice during that period of time are also shown (Fig. 1 A, dashed lines). (B) Average blood glucose levels of Ins2^+/+ (n = 4), Ins2^Akita/+ (n = 8), and insulin-treated Ins2^Akita/+ (n = 4) mice during the 14-week treatment. (C) Contrast sensitivities measured at age 24 weeks. Average blood glucose levels and contrast sensitivity of insulin-treated Ins2^Akita/+ mice are similar to those of Ins2^+/+ mice but are different from those of untreated Ins2^Akita/+ mice. *P < 0.05, one-way ANOVA, pair-wise comparisons using Holm-Sidak's test.

Discussion

In this study, the optomotor response of Ins2^Akita/+ mice was measured to investigate how hyperglycemic diabetes affects scotopic contrast sensitivity, a visual attribute driven by rod photoreceptors. In these mice, a significant loss in contrast sensitivity was observed 3 to 4 months after the onset of hyperglycemia. Examination of the optomotor response to vertical gratings of various spatial and temporal frequencies revealed that diabetes induces a selective loss in contrast sensitivity to fast- but not slow-varying gratings, independently of spatial frequency. Such losses in sensitivity can be prevented with application of insulin implants. Spatial frequency thresholds did not change significantly, suggesting that head movements per se were not compromised by diabetes. These results provide important new insights of early effects of diabetes on rod-driven vision, as detailed below. 

Hyperglycemic Loss of Scotopic Contrast Sensitivity Is Independent of Spatial Frequency

A key result of this investigation is that spatial contrast sensitivity is developmentally attenuated for Ins2^Akita/+ mice under conditions of dim illumination that favor rod-driven vision (photoisomerization rate of 0.02 R*/rod/s, where R* indicates number of photoisomerizations). Contrast sensitivity to all spatial frequencies was attenuated by approximately 40% relative to that of control measurements. This finding is consistent with studies of the optomotor response of Ins2^Akita/+ mice under photopic conditions. ²⁷ Together these results indicate that diabetic mice experience similar spatial vision deficits irrespective of luminance levels and, as a consequence, of the neural pathways activated in the inner retina. This conclusion is based upon the assumption that rod signals at mean illumination rates of 0.02 R*/rod/s are transmitted largely by rod bipolar cells along the primary rod pathway, ³⁷ whereas rod and cone signals at mesopic and photopic illumination levels are transmitted largely by cone bipolar cells and the secondary rod pathway. ^38,39 The observed spatial vision deficits in the optomotor contrast sensitivity may thus arise from impaired function at or proximal to sites where rod and cone pathways converge, which include the rod-cone gap junctions ⁴⁰ and the retinal inner plexiform layer. ⁴¹ In the inner plexiform layer, synaptic interactions between amacrine, bipolar, and ganglion cells shape the antagonistic center-surround receptive fields of ganglion cells that are critical to the processing of spatial contrast. ⁴² Electrophysiological support for this view in rodents is provided by the diabetes-induced abnormalities in two ERG components believed to reflect neural activity in the inner retina: the oscillatory potentials ⁴³ and the scotopic threshold response. ⁴⁴ Further evidence implicating a dysfunctional inner retina comes from documented reductions in the number of amacrine and ganglion cells, the trimming of their processes and general thinning of the inner retina. ^25,45–48 A shared retinal locus for diabetic impairments of photopic and scotopic spatial vision is thus suggested. 

Does the reduction in the number of ganglion and cholinergic amacrine cells from the peripheral retina ^45–47 explain the uniform loss in spatial contrast sensitivity? The receptive field size and response attributes of ganglion cells depend on retinal eccentricity, for example, ON-OFF direction-selective ganglion cells located in the periphery respond optimally to sinusoidal gratings with lower spatial frequency (wide bands, large objects), whereas cells located in central regions of the retina prefer high spatial frequencies (narrow bands, small objects). ⁴⁹ Therefore, a selective loss of cells in the periphery of the retinas raises the possibility that Ins2^Akita/+ mice experience a loss in contrast sensitivity that depends largely on the spatial properties of the stimulus: diabetes may preferentially reduce contrast sensitivity to gratings with low but not high spatial frequencies. Our results do not seem to support this possibility as they reveal a uniform loss in spatial contrast sensitivity of the optomotor response. One possible explanation is that the number of ganglion cells in the periphery of the retina mediating the optomotor response (most likely a class of ON-direction selective ganglion cells ^50,51 ) is not affected by diabetes; however, these results do not rule out a selective loss in spatial sensitivity for other visual tasks. 

Selective Loss in Scotopic Contrast Sensitivity to Fast-Varying Signals

Diabetic mice in this study exhibited diminished contrast sensitivity to high temporal frequencies (>0.1 Hz) or, equivalently, to fast moving gratings (speeds >3 deg/s). The mechanisms underlying this selective loss in temporal/speed contrast sensitivity could reside in the retina or central paths. One potential explanation is that diabetes selectively targets ganglion cells or visual pathways that transmit “brisk” temporal information. ⁵² Another possibility is that diabetes slows the response kinetics of one or more types of neurons along the visual pathway(s), compromising their ability to represent fast-varying signals. Consistent with this notion, studies with human diabetic subjects indicate a systematic increase in the implicit time of their ERG b-wave. ⁶ The slowed time course of the b-wave likely reflects slowed bipolar cell responses, which in turn can be explained in terms of abnormal rod photoreceptor and/or bipolar cell responses to light. Studies with streptozotocin-treated rats have produced inconsistent results, ^43,53,54 suggesting that, in rodents, additional neuronal sites with slowed responses could reside downstream of bipolar cells. 

Deficits in Spatial and Temporal Contrast Sensitivities May Be Interdependent

Rod-driven contrast sensitivity of the optomotor response is well described as the product of two functions, one function that depends on temporal frequency and another function that depends on spatial frequency of the grating. ²⁹ Based on this property, diabetes-induced changes in temporal contrast sensitivity are predicted to scale the spatial contrast sensitivity functions vertically along the sensitivity axis, as indicated in Fig. 3B. Analogously, a change in spatial contrast sensitivity should shift the temporal contrast sensitivity functions vertically, contrary to what is indicated in Figure 3D. Indeed, temporal CSFs are preferentially attenuated at the upper end of the frequency spectrum. This analysis suggests that the loss in spatial contrast sensitivity may be secondary to a diabetes-induced reduction in temporal contrast sensitivity to high temporal frequencies. 

Comparison between Optomotor Responses in Ins2^Akita/+ Mice and Visual Sensitivity in Diabetic Subjects

The optomotor response in mice shares features with human visual sensitivity, including Weber-like light adaptation, a light-dependent increase in temporal resolution, and band pass-shaped CSFs. ^28,29,55,56 It is therefore important that insights drawn from this study be referenced to early visual deficits experienced by diabetic subjects without retinopathy. Under photopic conditions, significant (30%–60%) losses in spatial contrast sensitivity have been observed in diabetic subjects with no signs of retinopathy. ⁹ However, while some studies with sinusoidal gratings indicate nonselective losses to both low and high spatial frequencies, ¹⁴ others report that the losses are specific to high spatial frequencies. ²¹ Changes in contrast sensitivity seem dependent on glycemic control as assessed by HbA1c concentration, ¹⁴ although a more recent study using the Pelli-Robson contrast sensitivity chart did not detect a similar correlation. ⁵ The small magnitude of the contrast sensitivity loss, the action over a broad frequency range, and the effects of glycemic control observed in diabetic humans are in qualitative agreement with the optomotor phenotype described in our study. Presently there are no available data describing spatial or temporal contrast sensitivity in diabetic subjects for conditions that selectively activate rod photoreceptors. 

In summary, the Ins2^Akita/+ mouse model of type 1 diabetes exhibits an early, progressive loss in rod-driven spatial and temporal contrast sensitivities. The loss begins shortly after the onset of hyperglycemia and grows in severity until mice are 4 to 5 months of age. Although the magnitude of the loss is modest (40%), characterizing its developmental progression provides useful information for investigating cellular and molecular mechanisms involved in the development of retinopathy and loss of rod-mediated vision in models of diabetes. Because Ins2^Akita/+ mice have variable retinal histology, ^25,48 which makes comparisons with the literature difficult, future studies will correlate the changes in retinal structure to the changes in optomotor sensitivity and investigate the generality of our findings in other models of diabetes. The goal is to identify the dysfunctional retinal and/or central neural mechanisms that impair temporal contrast sensitivity. 

Acknowledgments

We thank David Cameron and Gus Engbretson for critically reading the manuscript. 

Footnotes

 Supported by the Department of Ophthalmology, by an unrestricted grant from Research to Prevent Blindness to SUNY Upstate Medical University, and by the Lions of Central New York.

Footnotes

 Disclosure: Y. Umino, None; E. Solessio, None