αA-crystallin overexpression dampens the metabolic-stress induced pro-inflammatory transition of Müller Glial cells (MGCs).
We have previously reported that αA-crystallin is highly expressed by retinal MGCs under diabetic conditions. We also showed that rat retinal Müller cells (rMC-1) overexpressing aA-crystallin can protect R28 rat retinal neuron cells from metabolic stress (16). Since, MGCs play a vital role in retinal homeostasis, particularly regarding metabolism, neurotransmission, and inflammation, we further investigated the effect of the αA-crystallin expression on the inflammatory response of MGCs to metabolic stress. We observed that metabolic stress-induced by serum deprivation leads to elevated expression of pro-inflammatory cytokines including interleukin-6 (IL-6), IL-1beta (IL-1β), and monocyte chemoattractant proteins-1 (MCP-1) in rMC-1 (Figure 1). Interestingly, this serum deprivation-induced increase expression of pro-inflammatory mediators was significantly prevented by WT aA-crystallin overexpression as evidenced by 57%, 80%, and 82% reduction in levels of IL-6, IL-1β, and MCP-1 respectively, as compared to the empty vector.
Since aA-crystallin is induced in MGCs during diabetes, we also studied the effect of aA-crystallin overexpression on metabolic stress conditions more reminiscent of diabetes, that is high glucose and TNFα. Our data show that exposure of rMC-1 to a "diabetes-like condition" resulted in elevated levels of IL-1β and IL-6 comparable to serum starvation. Of note, a much more dramatic increase in MCP-1 expression was observed (352%) in "diabetes-like" conditions when compared to serum starvation. Consistent with a key role of aA-crystallin in the regulation of MGCs activation in metabolic stress, WT aA-crystallin overexpression significantly reduced the induction of IL-6 (61%), IL-1β (77%), and MCP-1 (63%) in “diabetes-like” condition as compared to the respective empty vector.
αA-crystallin effect on the metabolic-stress induced activation of Müller Glial cells is T148 phosphorylation-dependent.
We have previously reported that Thr148 phosphorylation of αA-crystallin was reduced dramatically in human donors with diabetic retinopathy. We further demonstrated that the phosphorylation of αA-crystallin on residue 148 controls its neuroprotective function in R28 rat retinal neuron cells (16). Thus, we next assessed the role of this phosphorylation on the dampening effect of aA-crystallin on the metabolic stress-induced pro-inflammatory response of MGCs. To do so, we overexpressed wild-type αA-crystallin (WT), the phosphomimetic (148D) αA-crystallin, or the non-phosphorylatable (148A) αA-crystallin mutant in rMC-1 cells (Figure 1D). Supporting a key role of this phosphorylation, the phosphomimetic (148D) αA-crystallin mutant had an even greater dampening effect than the WT protein on the expression of these pro-inflammatory cytokines, reducing their serum-deprivation induction by 64%, 86%, and 79% for IL-6, IL-1β and MCP-1 respectively. Conversely, the non-phosphorylatable (148A) αA-crystallin mutant was wildly ineffective at reducing the induction of any of these pro-inflammatory cytokines in serum deprivation stress. Similarly, in a "diabetes-like" condition, the phosphomimetic (148D) αA-crystallin mutant had an even greater dampening effect than that of the WT αA-crystallin as demonstrated by a greater decrease in the induction of IL-6 (78%), IL-1β (79%), and MCP-1 (83%). Conversely again, the levels of induction of IL-6, IL-1β, and MCP-1 by “diabetes-like” conditions in presence of the non-phosphorylatable (148A) mutant were comparable to those obtained in absence of aA-crystallin (EV).
Primary Müller cells isolated from αA-crystallin knockout mice
The data obtained from the rat retinal Müller cell lines (rMC-1) were consistent with a key role of αA-crystallin in Müller glial cells, especially in the modulation of their inflammatory response associated with metabolic stress. To further assess the function of aA-crystallin in a more physiologically relevant system, we continued our study in freshly isolated primary MGCs. Additionally, the primary MGCs were isolated from the retinas of αA-crystallin knockout mice to prevent any potential interference of the endogenously expressed αA-crystallin. The quality and stability of the primary MGCs were controlled by analysis of the expression of specific markers upon isolation. The analysis of expression of gene-specific to retinal MGCs, that is Prdx-6, GLUL, and Abc8a revealed that these cells maintain their specificities in our culture conditions until passage 6 (Figure 2B-D). The dramatic reduction of expression of the specific markers at passage 6 was associated with a change in morphology as the cells became flat and non-polarized (Figure 2A). Henceforth, all experiments were performed with primary MGCs before passage 6.
The metabolic-stress induced pro-inflammatory response of primary MGCs is dampened by αA-crystallin expression
Similar to what was observed in rMC-1, serum deprivation also induced increased expression of multiple pro-inflammatory cytokines by primary MGCs. Of note, the relative induction of IL-1β, IL-6, and MCP-1 seem even more dramatic in MGCs (≈10-15 fold) as compared to rMC-1 cells (≈5-8 fold), potentially due to the lack of endogenous aA-crystallin. Despite this dramatic increase, overexpression of WT αA-crystallin in primary MGCs nearly abolished the serum-deprivation induction of these pro-inflammatory cytokines as it resulted in 97%, 88%, 72%, and 89% reduction in expression of IL-6, IL-1β, IL-18, and MCP-1 respectively (Figure 3). Primary MGCs seem to be also very highly reactive to the "diabetes-like" conditions as compared to rMC-1 cells, with even greater induction of the different cytokines analyzed. Under "diabetes-like" conditions, WT αA-crystallin overexpression led to the significant reduction of the induction of IL-6 (55%), IL-1β (36%), IL-18 (93%), and MCP-1 (85%) by MGCs (Figure 3).
Phosphorylation of αA-crystallin on residue 148 controls the neuroinflammatory cascade in metabolically stressed Primary MGCs.
Next, we assessed if phosphorylation on residue 148 is necessary for the regulation by αA-crystallin of the metabolic-stress induced neuroinflammatory response of primary MGCs. Consistent with the key role of this phosphorylation, the metabolic-stress-induced inflammatory response of MGCs was significantly dampened by the phosphomimetic (148D) but not the non-phosphorylatable (148A) αA-crystallin mutant. The phosphomimetic (148D) mutant reduced IL-6, IL-1β, IL-18, and MCP-1 expression by 87%, 86%, 64%, and 28% respectively, while the expression of these inflammatory mediators remained comparable to the empty vector control in MGCs overexpressing the non-phosphorylatable (148A) αA-crystallin mutant. Under “diabetes-like” stress, the phosphomimetic (148D) αA-crystallin mutant was even more effective at reducing the pro-inflammatory mediator’s response of MGCs than the WT αA-crystallin protein. While the phosphomimetic (148D) αA-crystallin overexpressing MGCs showed a greater reduction of the “diabetes-like” stress-induced pro-inflammatory mediators evaluated, the non-phosphorylatable (148A) αA-crystallin mutant was completely devoid of dampening effect (Figure 3A-D).
αA-crystallin regulates MGCs inflammatory response through stress-specific inflammatory pathways
Previous studies have reported that diabetes and metabolic stress can cause the induction of pro-inflammatory cytokines through activation of nuclear factor kappa B (NF-κB) or the NLRP3 inflammasomes (37-39). In addition to demonstrating the key role of αA-crystallin in regulating the pro-inflammatory response of MGCs to metabolic stress, our data also demonstrated that it did so despite a cytokine profile varying based on the nature of the stress. This stress-specificity of the response prompted us to assess the role of aA-crystallin in the expression and activation of these respective pathways. This analysis showed that serum deprivation led to robust activation of the NLRP3 inflammasome (Figure 4A-C) while diabetes-like stress caused a specific and dramatic induction of NF-κB (Figure 4D). While aA-crystallin overexpression led to some degree of variability in the expression levels of various effectors of the inflammasome under diabetes-like conditions, none of them were significantly different from normal EV conditions. Conversely, serum deprivation had only a very moderate effect on NF-κB expression compared to the effect of diabetes-like stress. Consistent with its impact on pro-inflammatory cytokines expression, overexpression of WT αA-crystallin or the phosphomimetic mutant, lead to a significant reduction of Nlrp3 (85%), Asc (61%), and caspase-1 (68%) induction in serum-deprived MGCs (Figure 4A-C). Overexpression of phosphomimetic mutant significantly decreased the cytosolic Nlrp3 protein expression by ≈ 2-fold (Figure 5A-B). Similarly, overexpression of WT αA-crystallin or the phosphomimetic mutant nearly abolished the “diabetes-like” stress induction of NF-kB. Underlining the key role of T148 phosphorylation, and consistent with the effects on cytokine expression, overexpression of the non-phosphorylatable (148A) αA-crystallin mutant was virtually completely ineffective under both stress conditions (Fig 4A-D). At the protein level activation of NF-kB was significantly reduced by the phosphomimetic mutant by inhibiting NF-kB phosphorylation at Ser536 under the serum starvation and “diabetes-like” stress, where the presence of non-phosphorylatable (148A) αA-crystallin mutant had no effect (Figure 5C-F).