Reducing Akt2 in retinal pigment epithelial cells causes a compensatory increase in Akt1 and attenuates diabetic retinopathy

Abstract

The retinal pigment epithelium (RPE) plays an important role in the development of diabetic retinopathy (DR), a leading cause of blindness worldwide. Here we set out to explore the role of Akt2 signaling-integral to both RPE homeostasis and glucose metabolism-to DR. Using human tissue and genetically manipulated mice (including RPE-specific conditional knockout (cKO) and knock-in (KI) mice), we investigate whether Akts in the RPE influences DR in models of diabetic eye disease. We found that Akt1 and Akt2 activities were reciprocally regulated in the RPE of DR donor tissue and diabetic mice. Akt2 cKO attenuated diabetes-induced retinal abnormalities through a compensatory upregulation of phospho-Akt1 leading to an inhibition of vascular injury, inflammatory cytokine release, and infiltration of immune cells mediated by the GSK3β/NF-κB signaling pathway; overexpression of Akt2 has no effect. We propose that targeting Akt1 activity in the RPE may be a novel therapy for treating DR.

Fig. 1. Electroretinography (ERG) suggests that RPE-specific Akt2 cKO partially rescues diabetes-induced disruption of retinal function (4 month duration of diabetes).

a Representative scotopic ERG a- and b-waveforms, showing response to a 0 log₁₀ cd·s/m² stimulus luminance after overnight dark adaptation. Scotopic (b) a-wave and (c) b-wave amplitudes were decreased in diabetic Akt2^fl/fl mice compared to nondiabetic controls. These changes were partially rescued in RPE-specific Akt2 cKO diabetic mice. d Representative photopic ERG waveforms response to a 1 log₁₀ cd·s/m² stimulus luminance after light adaptation. e Induction of diabetes decreased the photopic b-wave amplitude in diabetic Akt2^fl/fl mice, which was rescued in RPE-specific Akt2 cKO diabetic mice. f Representative ERG c-waveforms. (g) Diabetes decreased the c-wave amplitude in Akt2^fl/fl mice compared to nondiabetic animals, which was partially mitigated in Akt2 cKO diabetic mice. In (b, c, e, g), n = 6 mice for each group, the data are expressed as mean ± SD. *p < 0.05; **p < 0.01; ****p < 0.0001 shows changes between diabetic Akt2^fl/fl and Akt2^fl/fl nondiabetic control. ^†p < 0.05, ^††p < 0.01 represents changes between diabetic Akt2 cKO and diabetic Akt2^fl/fl mice. ^#p < 0.05 represents changes of diabetic Akt2 cKO versus nondiabetic Akt2 cKO mice. Statistical test used in (b, c, e) is Two-way ANOVA followed by Tukey’s multiple comparisons test, and in (g) is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b At −1 log₁₀ cd*s/m², p = 0.0116 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N); At 0 log₁₀ cd*s/m², p = 0.0017 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0357 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0244 (Akt2 cKO D vs. Akt2 cKO N). c p = 0.0065 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.037 (Akt2 cKO D vs. Akt2 ^fl/fl D). e p = 0.0247 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0046 (Akt2 cKO D vs. Akt2 ^fl/fl D). g p < 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0039 (Akt2 cKO D vs. Akt2 ^fl/fl D). N nondiabetic, D diabetic, cKO conditional knockout.

Fig. 2. RPE-specific Akt2 cKO inhibits the diabetes-induced increases in inflammatory proteins, leukostasis, and generation of reactive oxygen species in the mouse retina after 2 months of diabetes.

a Representative immunoblots and densitometry graphs demonstrating the diabetes-induced increases of (b) ICAM-1, (c) iNOS, and (d) the ratio of pIκB/IκB were inhibited in the retina of RPE-specific Akt2 cKO diabetic mice. e Representative images and (f) quantification of attached leukocytes in the retina. Arrows indicate leukocytes adherent to the retinal blood vessels. Diabetes increased the number of adherent leukocytes in the Akt2^fl/fl diabetic retina compared to nondiabetic Akt2^fl/fl mice; this number was markedly reduced in Akt2 cKO diabetic mice, Scale bar: 100 µm. g Retinal superoxide was measured using lucigenin; diabetes increased retinal production of superoxide in Akt2^fl/fl mice. These levels were attenuated in Akt2 cKO diabetic mice. h Representative images and (i) quantification of DHE-stained (red) retinal sections showing the levels of ROS in each group. The DHE stain was primarily localized in the nuclear layers, Scale bar: 100 µm. The intensity of red fluorescence was quantified at the INL and ONL as they represent the majority of red staining in the retina. The diabetes-induced increase of ROS in Akt2^fl/fl retina was not observed in diabetic Akt2 cKO mice. j Representative images and (k) quantification of DCF-stained retinal sections showing ROS accumulated in the inner and outer segments of photoreceptors. The blue stain is DAPI. The diabetic Akt2 cKO mice displayed significantly low retinal ROS levels compared to the diabetic Akt2^fl/fl mice. Scale bar: 100 µm. Data are expressed as mean ± SD. **p < 0.01; ***p < 0.001; ****p < 0.0001 show changes versus Akt2^fl/fl nondiabetic (N) controls. ^††p < 0.01, and ^††††p < 0.0001 show changes versus Akt2^fl/fl diabetic (D) mice. Statistical test used in (b, c, d, f, g, i, k) is One-way ANOVA followed by a Tukey’s post hoc test. n = 6 mice for each group. Exact p values are: b p = 0.0024 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0081 (Akt2 cKO D vs. Akt2 ^fl/fl D). c p = 0.0004 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0024 (Akt2 cKO D vs. Akt2 ^fl/fl D). d p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0036 (Akt2 cKO D vs. Akt2 ^fl/fl D). f p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0019 (Akt2 cKO D vs. Akt2 ^fl/fl D). g p = 0.0003 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0043 (Akt2 cKO D vs. Akt2 ^fl/fl D). i p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p < 0.0001 (Akt2 cKO D vs. Akt2 ^fl/fl D). k p < 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N, Akt2 cKO D vs. Akt2 ^fl/fl D). cKO conditional knockout, GCL ganglion cell layer, INL inner nuclear layer, ONL outer nuclear layer, RIS/ROS rod inner/outer segment, DHE dihydroethidium, DCF dichlorofluorescein. Source Data is provided in the Source Data file.

Fig. 3. Flow cytometry to identify the population of leukocytes infiltrating the RPE and retina isolated from diabetic (2 months of diabetes) and age-matched nondiabetic Akt2^fl/fl and Akt2 cKO mice.

a FSC-A versus SSC-A dot plots were gated to eliminate debris. b Single cells were selected on FSC-A versus FSC-H. c Representative dot plots were gated on CD11b⁺ and CD45⁺ cells. d Ly6C, Ly6G and (e) CCR2 antibodies were used to define the neutrophils and monocytes gated on CD11b⁺CD45^high cells. f The absolute number of infiltrating leukocytes (CD11b⁺CD45^high) was increased in Akt2^fl/fl diabetic mice. However, this was inhibited in RPE-specific Akt2 cKO diabetic mice. g The neutrophil and (h) monocyte populations were increased in the retinas of Akt2^fl/fl diabetic mice compared to nondiabetic controls. Interestingly, Akt2 cKO mice demonstrated noticeable decline in the diabetes-induced increase in retinal neutrophil and monocyte infiltration. In (f–h), data are shown as Mean ± SD. N = 4 animals per group. **p < 0.01 and ***p < 0.001 denotes changes versus Akt2^fl/fl nondiabetic group. ^†p < 0.05, and ^††p < 0.01, shows changes versus Akt2^fl/fl diabetic mice. Statistical test used in (f–h) is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: f p = 0.0026 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.025 (Akt2 cKO D vs. Akt2 ^fl/fl D) g p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0031 (Akt2 cKO D vs. Akt2 ^fl/fl D). h p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.025 (Akt2 cKO D vs. Akt2 ^fl/fl D).

Fig. 4. RPE-specific Akt2 cKO inhibits the development of diabetes-induced retinal vascular lesions.

a Representative micrographs of retinal vessels from diabetic mice (8 months of diabetes) and age-matched nondiabetic mice. Arrows indicate degenerated capillaries and arrowheads indicate capillary pericytes. Scale bar: 100 µm. b Diabetes increased the number of degenerated capillaries and (c) decreased the number of retinal capillary pericytes in diabetic Akt2^fl/fl mice compared to nondiabetic animals. RPE-specific Akt2 cKO partially rescued this degeneration. d Representative micrographs of retinal sections after mice were intravenously injected with FITC-albumin. Scale bar: 100 µm. e Average fluorescence intensity was quantified from a large area of INL, IPL, and OPL, excluding obvious microvessels. RPE-specific Akt2 cKO reduced the diabetes-induced accumulation of FITC-BSA in the OPL, INL, and IPL of mouse retina compared to diabetic Akt2^fl/fl mice. In (b, c, e), n = 6 mice for each group, the data are expressed as mean ± SD. **p < 0.01; ***p < 0.001; ****p < 0.0001 versus Akt2^fl/fl nondiabetic control (N). ^†p < 0.05; ^††p < 0.01; ^†††p < 0.0001 versus Akt2^fl/fl diabetic mice (D). Statistical test used in (b, c, e) is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b p < 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0096 (Akt2 cKO D vs. Akt2 ^fl/fl D). c p = 0.0008 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0187 (Akt2 cKO D vs. Akt2 ^fl/fl D). e OPL, p = 0.0046 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0388 (Akt2 cKO D vs. Akt2 ^fl/fl D); INL, p = 0.0002 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0087 (Akt2 cKO D vs. Akt2 ^fl/fl D). IPL, p < 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0009 (Akt2 cKO D vs. Akt2 ^fl/fl D). cKO conditional knockout, IPL inner plexiform layer, INL inner nuclear layer, OPL outer plexiform layer.

Fig. 5. ERG suggests that RPE-specific Akt2 KI does not impact the diabetes-induced disruption of retinal function (4 month duration of diabetes).

a Representative scotopic ERG a- and b- waveforms, showing response to a 0 log₁₀ cd·s/m² stimulus luminance after overnight dark adaptation. Scotopic (b) a-wave and (c) b-wave amplitudes were decreased in both WT and Akt2 KI diabetic mice compared to appropriate nondiabetic controls. Akt2 overexpression in the RPE (Akt2 KI) did not exhibit any protective effect on diabetic-induced ERG abnormalities. d Representative photopic ERG b-waveforms response to a 1 log₁₀ cd·s/m² stimulus luminance after light adaptation. e RPE-specific Akt2 KI had no protective effect on the diabetes-induced detrimental changes in the photopic b-wave. f Representative ERG c-waveforms. g c-wave amplitude is decreased in WT and Akt2 KI diabetic mice compared to appropriate nondiabetic controls. There was no significant change in the c-wave between diabetic WT and Akt2 KI groups. In (b, c, e, g), n = 6 mice for each group, the data are expressed as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 denotes changes with respect to nondiabetic WT control. ^†p < 0.05; ^†††p < 0.001; ^††††p < 0.0001 denotes changes versus nondiabetic Akt2 KI mice. Statistical test used in (b, c, e) is Two-way ANOVA followed by Tukey’s multiple comparisons test. Test used in (g) is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b p = 0.0030 (WT D vs. WT N), p = 0.0003 (Akt2 KI D vs. Akt2 KI N). c At 0 log₁₀ cd*s/m² p = 0.0255 (WT D vs. WT N), p = 0.0147 (Akt2 KI D vs. Akt2 KI N); At −1 log₁₀ cd*s/m² p = 0.0020 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. Akt2 KI N). e p < 0.0001 (WT D vs. WT N, Akt2 KI D vs. Akt2 KI N). g p = 0.0095 (WT D vs. WT N), p = 0.0006 (Akt2 KI D vs. WT N), p = 0.0007 (Akt2 KI D vs. Akt2 KI N). N nondiabetic, D diabetic, WT wild type, KI knock-in.

Fig. 6. RPE-specific Akt2 KI does not influence the diabetes-induced increase in pro-inflammatory protein level, leukostasis, or production of reactive oxygen species in mice after 2 months of diabetes.

a Representative immunoblots and densitometry graphs showing that the diabetes-induced increase in the levels of pro-inflammatory markers (b) ICAM-1, (c) iNOS, and (d) the ratio of pIκB/ IκB were not significantly decreased in Akt2 KI diabetic mice compared to WT diabetic mice. This result is in contrast to that found with Akt2 cKO mice (e) Representative images and (f) quantification of retinal leukostasis from each group. Arrows indicate leukocytes adherent to the retinal blood vessels. Diabetes increased the number of adherent leukocytes in the retina of both WT and RPE-specific Akt2 KI mice; Scale bar: 100 µm. g Retinal superoxide was measured using the lucigenin method. Diabetes increased retinal superoxide production in both diabetic WT and Akt2 KI mice. h Representative images and (i) quantification of dihydroethidium (DHE)-stained (red) ROS. Scale bar: 100 µm. Red fluorescence intensity (DHE stain) was quantified at the INL and ONL together as they represent the majority of staining in the retina. Diabetes increased the production of retinal ROS in both WT and Akt2 KI mice. j Representative images and (k) quantification of dichlorofluorescein (DCF) stained ROS. The DCF stain is localized primarily in the inner and outer segments of photoreceptors. The blue (nuclear) stain is DAPI. Diabetes induced an increase of ROS in the retina of both diabetic WT and Akt2 KI mice compared to appropriate nondiabetic controls. Scale bars: 100 µm. In (b–d, f, g, i, k), n = 6 mice for each group, data are represented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 represent changes versus WT N. ^†p < 0.05, ^††p < 0.01, ^†††p < 0.001, and ^††††p < 0.0001 denotes changes versus Akt2 KI N. Statistical test used in this study is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b p = 0.004 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N, Akt2 KI D vs. Akt2 KI N). c p = 0.0019 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N, Akt2 KI D vs. Akt2 KI N). d p = 0.0231 (WT D vs. WT N), p = 0.0004 (Akt2 KI D vs. WT N), p = 0.0004 (Akt2 KI D vs. Akt2 KI N). f p = 0.0033 (WT D vs. WT N), p = 0.001 (Akt2 KI D vs. WT N), p = 0.0003 (Akt2 KI D vs. Akt2 KI N). g p = 0.001 (WT D vs. WT N), p = 0.0002 (Akt2 KI D vs. WT N), p = 0.0015 (Akt2 KI D vs. Akt2 KI N). i p < 0.0001 (WT D vs. WT N), p = 0.0003 (Akt2 KI D vs. WT N), p = 0.0012 (Akt2 KI D vs. Akt2 KI N). k p < 0.0001 (WT D vs. WT N), p = 0.0011 (Akt2 KI D vs. WT N), p = 0.017 (Akt2 KI D vs. Akt2 KI N). WT wild type, N nondiabetic, D diabetic, KI knock-in, GCL ganglion cell layer, INL inner nuclear layer, ONL outer nuclear layer, RIS/ROS rod inner/outer segment. Source Data is provided in the Source Data file.

Fig. 7. RPE-specific Akt2 KI does not influence the development of diabetes-induced retinal vascular lesions.

a Representative micrographs of retinal vessels from diabetic mice (8 months of diabetes) and age-matched nondiabetic mice. Scale bar: 100 µm. Arrows indicate degenerated capillaries and arrowheads indicate capillary pericytes. b Diabetes increased the number of degenerated capillaries and (c) decreased the number of retinal capillary pericytes in diabetic WT mice compared to nondiabetic animals. There was no significant difference in the numbers of acellular capillaries and pericytes in the retina between diabetic WT and diabetic Akt2 KI mice. d Representative micrographs of retinal sections from each group after mice were intravenously injected with FITC-albumin. Scale bar: 100 µm. e Average fluorescence intensity was quantified from a large area of INL, IPL, and OPL, excluding obvious microvessels. Diabetes-induced accumulation of FITC-BSA in these retinal layers was higher in WT and Akt2 KI diabetic mice (8 months of diabetes) compared to age-matched nondiabetic controls. There was no difference in diabetes-induced retinal vascular leakage between diabetic WT and diabetic Akt2 KI mice. In (b, c, e), n = 6 mice for each group, the data are expressed as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 shows changes versus WT nondiabetic control. ^†p < 0.05 and ^††††p < 0.0001 shows changes versus Akt2 KI nondiabetic mice. Statistical test used in this study is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b p = 0.0003 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N and Akt2 KI D vs. Akt2 KI N). c p = 0.0001 (WT D vs. WT N), p = 0.0005 (Akt2 KI D vs. WT N), p = 0.0005 (Akt2 KI D vs. Akt2 KI N). e OPL, p = 0.0029 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N), p < 0.0001 (Akt2 KI D vs. Akt2 KI N); INL, p = 0.0009 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N, Akt2 KI D vs. Akt2 KI N); IPL, p = 0.0024 (WT D vs. WT N), p < 0.0001 (Akt2 KI D vs. WT N, Akt2 KI D vs. Akt2 KI N). N nondiabetic, D diabetic, WT wild type, KI knock-in, IPL inner plexiform layer, INL inner nuclear layer, OPL outer plexiform layer.

Fig. 8. Akt2 cKO promotes upregulation of Akt1 and inhibits the diabetes-induced inflammatory response through GSK3β/NF-κB signaling in the RPE cells (2 months of diabetes).

a Representative immunoblots and quantification of (b) phospho-Akt2, (c) total-Akt2, (d) phospho-Akt1 and (e) total Akt1 in RPE lysates obtained from diabetic and nondiabetic Akt2^fl/fl and cKO mice. Diabetes increased the level of phospho-Akt2 but did not affect the levels of total Akt2 in Akt2^fl/fl RPE cells compared to the nondiabetic control. RPE-specific Akt2 cKO reduced the levels of both phospho- and total-Akt2 in the RPE; induction of diabetes did not further affect the levels of either phospho- or total Akt2 in Akt2 cKO RPE. Induction of diabetes in Akt2^fl/fl mice decreased the level of phospho-Akt1 but did not affect the levels of total Akt1 in the RPE cells compared to nondiabetic control. Akt2 cKO increased the levels of phospho-Akt1 in diabetic Akt2 cKO mice, but not in nondiabetic cKO mice compared to diabetic or nondiabetic Akt2^fl/fl mice, respectively. The total Akt1 in the RPE was upregulated in both nondiabetic and diabetic Akt2 cKO mice compared to Akt2^fl/fl mice. f Representative immunoblots. g The ratio of phospho-GSK3β/total GSK3β was lower in diabetic Akt2^fl/fl mice, relative to nondiabetic control mice. However, it was increased in diabetic Akt2 cKO mice. h In addition, Akt2 cKO rescued the diabetes-induced increase in the ratio of p-NF-κB p65/total NF-κB p65. i Representative immunoblots. Diabetes induction increased the protein level of (j) ICAM-1 and (k) iNOS in Akt2^fl/fl mice, whereas in diabetic Akt2 cKO mice this diabetes-induced increase in inflammatory protein levels was significantly rescued. In (b–e, g, h, j, k), n = 6 mice per group, the data are expressed as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 denotes changes versus Akt2^fl/fl nondiabetic (N) control. ^†p < 0.05, ^††p < 0.01, ^†††p < 0.001, and ^††††p < 0.0001 denotes changes with respect to Akt2^fl/fl diabetic (D) mice. ^###p < 0.001 and ^####p < 0.0001 shows changes versus nondiabetic Akt2 cKO group. Statistical test used in this study is One-way ANOVA followed by a Tukey’s post hoc test. Exact p values are: b p = 0.025 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p < 0.0001 (Akt2 cKO N vs. Akt2 ^fl/fl N, Akt2 cKO D vs. Akt2 ^fl/fl D). c p < 0.0001 (Akt2 cKO N vs. Akt2 ^fl/fl N, Akt2 cKO D vs. Akt2 ^fl/fl D). d p = 0.0109 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0009 (Akt2 cKO D vs. Akt2 ^fl/fl N), p < 0.0001 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0007 (Akt2 cKO D vs. Akt2 cKO N). e p = 0.0005 (Akt2 cKO N vs. Akt2 ^fl/fl N), p < 0.0001 (Akt2 cKO D vs. Akt2 ^fl/fl N). p < 0.0001 (Akt2 cKO D vs. Akt2 ^fl/fl D). g p = 0.0326 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0001 (Akt2 cKO D vs. Akt2 cKO N), p < 0.0001 for the rest. h p = 0.0003 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0283 (Akt2 cKO D vs. Akt2 ^fl/fl D). j p = 0.0004 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.011 (Akt2 cKO D vs. Akt2 ^fl/fl D). k p < 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0076 (Akt2 cKO D vs. Akt2 ^fl/fl D). Source Data is provided in the Source Data file.

Fig. 9. Inflammatory cytokines are elevated in the retina/RPE of Akt2^fl/fl diabetic mice, an effect inhibited in Akt2 cKO mice; however inhibition of Akt1 reverses this protective effect.

a Diabetes-induced increase of retinal VEGF was inhibited in Akt2 cKO mice. b Diabetes increased the RPE/retinal expression of inflammatory cytokines, including IL-1β, IL-6, IL-17A, IFN-γ and TNF-α, in Akt2^fl/fl mice compared to Akt2^fl/fl nondiabetic animals. Akt2 cKO inhibits diabetes-induced elevation of these inflammatory cytokines. c RPE explants were obtained from Akt2^fl/fl and Akt2 cKO nondiabetic and diabetic mice and cultured in 5 mM and 25 mM glucose medium, respectively, for 48 h. The spent medium was collected and ELISA assays performed. As expected, Akt2 cKO inhibited the release of inflammatory cytokines caused by diabetes, including IL-1β, IL-6, IL-12, IL-17A, IFN-γ and TNF-α. Such inhibitory effects were abolished if an Akt1 inhibitor was added to the RPE explants (isolated from Akt2 cKO diabetic mice) culture. d RPE explant culture was collected for western blot analysis. e–j Akt2 cKO increased the level of phospho-Akt1 in the RPE from diabetic mice and inhibited the diabetes-induced reduction in the ratio of p-GSK3β/total GSK3β, as well as the diabetes-induced elevation in the ratio of p-NF-κB p65/total NF-κB p65. The presence of the Akt1 inhibitor in the Akt2 cKO diabetic mouse RPE explants significantly reversed this protective effect for the levels of p-Akt1, p-GSK3β/total GSK3β and p-NF-κB p65/total NF-κB p65. In (a–c, e–j), data are shown as Mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, and ***p < 0.0001 versus the Akt2^fl/fl nondiabetic mice. ^†p < 0.05, ^††p < 0.01, ^†††p < 0.001, and ^††††p < 0.0001 versus Akt2^fl/fl diabetic mice. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001, and ^###p < 0.0001 versus Akt2 cKO diabetic group. Statistical tests used in (a, e–j) is One-way ANOVA followed by a Tukey’s post hoc test, n = 6 mice for each group. Statistical test used in (b, c) is two tailed, unpaired t-test, n = 6 animals; 3 samples in each group; each sample was composed of 2 animals. Exact p values are: a p = 0.0001 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0049 (Akt2 cKO D vs. Akt2 ^fl/fl D), b IL-1β: p = 0.0342, 95% CI 0.4993 to 2.146, R² = 0.8326 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0128, 95% CI −2.095 to −0.4477, R^2. = 0.8212 (Akt2 cKO D vs. Akt2 ^fl/fl D); IL-6: p = 0.0352, 95% CI 0.026 to 0.4347, R² = 0.7101 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0483, 95% CI −0.4111 to −0.002517, R² = 0.6639 (Akt2 cKO D vs. Akt2 ^fl/fl D); IL-17A: p = 0.0388, 95% CI 0.01557 to 0.3558, R² = 0.6966 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0374, 95% CI −0.3671 to −0.01823, R² = 0.706 (Akt2 cKO D vs. Akt2 ^fl/fl D); IFN-γ: p = 0.0028, 95% CI 0.1819 to 0.4488, R² = 0.915 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0026, 95% CI −0.4227 to −0.1746, R² = 0.9178 (Akt2 cKO D vs. Akt2 ^fl/fl D); TNF-α: p = 0.0291, 95% CI 0.04679 to 0.5145, R² = 0.7351 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0418, 95% CI −4694 to −0.01459, R² = 0.6858 (Akt2 cKO D vs. Akt2 ^fl/fl D). c IL-1β: p = 0.0154, 95% CI 0.06971 to 0.373, R² = 0.8042 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.048, 95% CI −0.3297 to −0.002349, R² = 0.6684 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0274, 95% CI 0.2393 to 0.2387, R² = 0.7424 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D), p = 0.0042, 95% CI 0.09867 to 0.2747, R² = 0.8966 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N); IL-6: p = 0.0154, 95% CI 0.06357 to 0.3391, R² = 0.8045 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.044, 95% CI −0.2824 to −0.006241, R² = 0.678 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.041, 95% CI 0.05295 to 0.1464, R² = 0.8977 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D), p = 0.007, 95% CI 0.1109 to 0.2024, R² = 0.9576 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N); IL-12: p = 0.0122, 95% CI 0.02614 to 0.1185, R² = 0.8254 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N), p = 0.0012, 95% CI 0.05197 to 0.1054, R² = 0.9436 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D); IL-17A: p = 0.0097, 95% CI 0.03451 to 0.1368, R² = 0.8439 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0358, 95% CI −0.1173 to −0.006687, R² = 0.7077 (Akt2 cKO D vs. Akt2 ^fl/fl D); IFN-γ: p = 0.0114, 95% CI 0.0195 to 0.08512, R² = 0.8308 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0227, 95% CI −0.08379 to −0.01087, R² = 0.7646 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0055, 95% CI 0.03739 to 0.1153, R² = 0.881 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D), p = 0.0031, 95% CI 0.0458 to 0.1169, R² = 0.9099 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N); TNF-α: p = 0.0404, 95% CI 0.00717 to 0.1955, R ² = 0.6909 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0392, 95% CI −0.1894 to −0.007913, R² = 0.6949 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0038, 95% CI 0.03767 to 0.1017, R² = 0.9014 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D), p = 0.0078, 95% CI 0.03166 to 0.113, R² = 0.8591 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N). e p = 0.0003 (Akt2 cKO N vs. Akt2 ^fl/fl N), p < 0.0001 for the rest. f p = 0.0018 (Akt2 cKO N vs. Akt2 ^fl/fl N), p = 0.0016 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0009 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl D). g p = 0.0124 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0019 (Akt2 cKO D vs. Akt2 ^fl/fl N), p = 0.0043 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N), p = 0.018 (Akt2 cKO N vs. Akt2 ^fl/fl D), p < 0.0001 for the rest. h p = 0.0007 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N), p < 0.0001 for the rest. i p = 0.003 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p = 0.0004 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N, Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0002 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D). j p = 0.0002 (Akt2 ^fl/fl D vs. Akt2 ^fl/fl N), p < 0.0001 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 ^fl/fl N), p = 0.0012 (Akt2 cKO D vs. Akt2 ^fl/fl D), p = 0.0006 (Akt2 cKO D + Akt1 inhibitor vs. Akt2 cKO D). Source Data is provided in the Source Data file.

Fig. 10. Schematic.

RPE-specific Akt2 cKO reduced the diabetes-mediated induction of the inflammatory protein level, infiltration of leukocytes and production of ROS in the retina after a 2 month duration of diabetes. Akt2 cKO also showed reduction in retinal capillary degeneration and vascular leakage even after an 8 month duration of diabetes. Such beneficial effects were indicative of the compensatory upregulation of Akt1 in the diabetic Akt2 cKO RPE cells, which downregulated GSK3β/NF-κB mediated inflammatory molecules like ICAM-1, iNOS, VEGF, TNF-α, IFN-γ and IL-1β in cKO RPE cells even after diabetes induction.