In addition to potassium (Kir 2.1 and 4.1) and water (AQP4 and 5) channel proteins found differentially expressed in retinal membrane proteins of horses with ERU by us [
12], we detected AQP11 as the most significantly downregulated candidate in pathological condition of ERU. AQP11 is a very interesting water channel protein because it is quite different to most of the other members of aquaporin family proteins [
13]. An exception is AQP12 that shares similarities with AQP11, at least regarding the unusual N-terminal sequence with an NPA (asparagine-proline-alanine) motif and low homology to the other aquaporins [
13,
16]. AQP11 was primarily described as not only most abundantly expressed in testis and to a lower extent, but also prominent in the kidney, liver, and brain [
29]. AQP11 knockout mice die early because of renal failure and retarded growth [
30]. An ocular phenotype was not described in these mice so far [
30]. Nevertheless, all aquaporins were reported to be expressed in the eyes of several mammals analyzed [
31]. AQP11 is expressed at Müller glia endfeet in the eyes of humans [
26]. In this study, we could clearly confirm AQP11 expression for Müller glia endfeet, but additionally, AQP11 is expressed throughout entire Müller glia plasma cell membrane in horse retina (Fig.
1a, c, d). After we identified AQP11 as the most pronounced downregulated channel protein in retinal membrane protein fraction (Table
1) in a differential proteomics experiment analyzing physiological retina versus autoimmune diseased retina in a spontaneous horse model [
12], we were quite interested in the exact cellular localization and function of this less good characterized member of the aquaporin family [
32]. Therefore, we developed respective antibody to horse AQP11 that allows detection of AQP11 in paraffin-embedded sections (Figs.
1 and
3) and western blots (Fig.
2) and to block AQP11 function in native Müller glial cells (Figs.
4 and
5). This was important because tissue localization and functional studies about unorthodox aquaporins in general are currently hampered because of poor-quality antibodies available [
17,
33‐
35]. Our novel monoclonal antibody is well suited to analyze AQP11 expression in immunohistochemistry and will probably aid other researchers interested in characterization of AQP11 in other animals or organs, since there is a reported lack of a good antibody suited for immunohistochemistry so far (positive stainings with antibody 8H9 in other species/tissues: Fig.
1h: mouse retina, i: rat retina, j: mouse kidney). With our novel tool, we could confirm marked reduction of AQP11 expression in a spontaneous model of retinal inflammation (ERU) to one third of physiological amount in retinas (Fig.
2). AQP11 was selectively expressed in retinal Müller glial cells in the retina of horses (Figs.
1a, c and
3g) and especially declined (Fig.
3h) in the areas of Müller cell gliosis (Fig.
3d, f) characterized in situ through upregulation of GFAP and concomitant lower expression of glutamine synthase in ERU (Fig.
3d, f). Reduction of AQP11 expression in ERU (Fig.
3h) closely followed glutamine synthase expression in respective areas (Fig.
3f, overlay j). Since Müller glial cells are still present in these regions, AQP11 reduction is not secondary to loss of expression sites, but seems to be related to inflammation-associated gliosis (Fig.
3). To our knowledge, this is the first link of AQP11 to a spontaneous autoimmune disease and in neuroinflammation. Because AQP11 function is still debated and ranges from it have no water permeability ability at all, as shown with Xenopus oocytes expressing AQP11 [
15] or has low but normal water channel activity with mercurial sensitivity, we were interested to test the influence of AQP11 to water permeability in Müller cells. Our results clearly show that AQP11 channel is especially important during osmotic stress conditions and blockage of respective channel protein results in the rescue from cell shrinking in extracellular hypertonic conditions (Figs.
4 and
5). Inhibition of AQP4 showed the same results as application of anti-AQP11 antibody, a selective saving from shrinking in hypertonic milieu, whereas blockage of AQP5 positively influenced cellular behavior in hypotonic environment (Fig.
4). Further, we could clearly show that AQP11 is expressed at Müller glial plasma cell membrane (we additionally confirmed the expression of AQP11 at Müller glia endfeet in mouse and rat retina). AQP11 expression was detected intracellularly [
31,
32] in many organs of other species, especially at rough ER [
36,
37]. Interestingly, we could not discover AQP11 at Müller glial cell ERs, as tested with ER markers calreticulin and calnexin in double stainings with AQP11, nor with various other cell organelle markers tested (e.g., for mitochondria, ribosomes, Golgi a. o., analyzed with apotome) or other cell types like endothelial cells (costaining with specific plant lectins). Therefore, we conclude that AQP11 is preliminary expressed at the cell surface of Müller glial cells. This would indicate an interesting difference of AQP11 in various cell types. But so far, several discrepancies in AQP distribution have been reported for different cells and tissues, including its location within lung and airways, gut, and exocrine glands [
38], underscoring the need for further analyses regarding cellular and subcellular distribution of AQP11. For example, localization of AQP11 at the plasma cell membrane remains controversial for other cells. Immunocytochemistry was used to show the expression of AQP11 on the cell surface of transfected Chinese hamster ovary cells [
15], whereas other data indicated that AQP11 is not targeted to the plasma membrane and stays intracellularly [
36]. In liver-specific AQP11 knockout mice, immunoblotting of membrane proteins from control mouse liver indicated AQP11 in this fraction, but exact cellular localization could not be investigated in this study because their antibody was not suited for immunohistochemistry [
17]. Probably, this information can now be created with our novel antibody and therefore shed more light on the unclear biological function of this interesting membrane channel protein.
Most studies focused on the role of AQP11 in ER function so far. It was shown that disruption of AQP11 in knockout mice leads to apoptosis in kidney cells associated with upregulation of ER stress genes [
39]. In Müller glial cells, AQP11 is expressed at cell surface and its significant reduction in ERU will therefore result in a change in transport function through plasma cell membrane. In our study with primary Müller glial cells, we could clearly show a function in cell volume regulation after osmotic stress to the cells. Blockage of AQP11 totally abolished the shrinking of treated cells (Figs.
4 and
5) like AQP4 did, indicating a similar function for both AQPs in outward water transport of retina. Redundant expression of different aquaporins was also shown for other tissues, e.g., of AQP3 and AQP4 in the basolateral plasma membrane of collecting duct principal cells, where they colocalize [
38]. The functional meaning of such coexpression is not fully understood to date. Since these channels also transport some other compounds besides water, there could be some other functional differences. The data concerning the ability of AQP11 to transport water are conflicting, and there is a lack of knowledge about the permeability of AQP11 to other solutes [
15,
40]. Therefore, the meaning of AQP4 and AQP11 coexpression und probable cofunction in the retina needs further investigation in our opinion.
Previously, we already identified a loss of potassium (Kir 2.1 and 4.1) and water channels (AQP4 and 5) in gliotic Müller glial cells in ERU [
10]. Retinal edema is a severe complication of late-stage ERU [
20] and can be caused by swollen Müller glial cells [
21]. Severe cases of gliosis induce the de-differentiation of Müller glial cells, followed by impairment of the ion and water homeostasis after downregulation of K+ channels [
21]. Downregulation and changed expression of Kir channels already takes place in early stages of ERU [
10]. Downregulation of these K+ channels can subsequently inhibit K+ release into the blood and interrupt the dehydrating K+ currents through Müller cells, which can increase the osmotical pressure within the cells. This results in an osmotic difference at the gliovascular interface which favors water inflow into the cells via still expressed aquaporin (especially AQP4) water channels [
21]. Osmotic stress can then activate phospholipase A2 and arachidonic acid (or inflammation during ERU) and can further increase theinflux of Na+ ions and therefore create an osmotic driving force for water inflow into the cells, resulting in Müller cell swelling [
21]. These mechanisms lead to significant dysregulation of retinal dehydration by Müller cells and to exacerbation of intra- and extracellular fluid accumulation [
21]. The loss of AQP11 channels at Müller glia plasma cell membranes (Fig.
3) probably reduces the ability of the cells to reduce cell volume through outflow of water and therefore leads to cell swelling and subsequent fatal retinal edema not only in ERU but also other retinal diseases, e.g., diabetic macular edema.
The role of AQP11 could additionally be tested by administering recombinant AQP11 to the cells in the in vitro studies. Although we have recently established virus-based transduction protocols with the hope to overcome the minimal transfection efficiencies observed for primary Müller cells (from horse and pig), we so far only achieved low transduction rates with lentivirus (maximal 5 %) and consequently at this time, those experiments are not feasible. But in future studies, we plan to study the effect of adding AQP11 to the cells under different environmental conditions in vitro.