2.1. AQP1
The first discovery of the existence of a water channel protein, later called aquaporin 1 (AQP1), was made from human erythrocyte cell membrane by Benga’s group in 1985 in Cluj-Napoca, Romania, and published in 1986
[69,134][18][19]. Later, Peter Agre purified the protein of AQP1 for which he won the 2003 Nobel Prize in Chemistry
[135][20]. AQP1 has been shown to mediate the transmembrane movement of water in several different organs, including the kidney, eye and brain
[53,81,136,137][21][22][23][24]. AQP1 expression in the human brain was first observed under pathological conditions after subarachnoid hemorrhage, and in peritumoral tissue, with the protein being located mainly on the processes of reactive astrocytes
[138][25]. AQP1 expression in the uninjured brains was absent or too low to be observed
[138,139][25][26]. Nevertheless, the labs of both Ben Barres and Baljit Khakh later confirmed using transcriptome studies that mRNA of AQP1 was indeed expressed by astrocytes
[140,141,142][27][28][29]. AQP1 transcripts and protein in cultured astrocytes were also identified, indicating localization of AQP1 on plasma membrane, in the cytoplasm compartment membrane and possibly also on the nuclear membrane
[53,133][21][30]. Another interesting finding of AQP1 in culture is that the protein was selectively lost when passaging primary astrocytes, but its mRNA was always detectable
[139][26]. The expression of AQP1 in the non-human primate brain was also found to be in the processes and perivascular end-feet of a subtype of astrocytes localized mainly in the white matter and at sites of the glia limitans
[48][31]. So far, no isoforms of AQP1 have been reported.
Unlike other AQP subtypes, it has been suggested that AQP1 might harbor the capability of transporting cations after stimulation with adenylate cyclase activator forskolin or protein kinase A in Xenopus egg
[72][32]. Further discoveries by the same group indicated that AQP1 is a cGMP-gated cation channel
[143,144][33][34] and that it functions as both a water channel and a gated ion-channel in the choroid plexus, contributing to the regulation of CSF production
[145][35]. Yool believed that the putative central pore in AQP1 is related to its cation transport functions
[1,81,146][22][36][37]. However, other studies reported that such cation transport could not be repeated in similar Xenopus egg models
[147,148,149,150,151,152][38][39][40][41][42][43]. In the molecular model presented by Murata and colleagues, the pore was only slightly larger than a typical water molecule, and the formation of hydrogen bonds between water and pore residues allowed only water molecules to pass through the channel
[153][44]. Similarly, the determination of the AQP1 crystal structure enabled the elucidation of the mechanism by which water molecules moved through the protein, but the passage of protons and other ions was blocked
[30][1]. The mechanisms for cation transport in different tissues and cell types thus still need further clarification. Nevertheless, a
recent paper clearly showed that AQP1 can mediate fast swelling kinetics and play a key role in triggering and accelerating regulatory volume decrease, a process where cells adjust their cellular volume in response to swelling
[154][45].
AQP1 was also one of the first AQP subtypes reported to facilitate the migration of endothelial cells
[155][46] and to promote reactive astrocyte migration
[139][26]. Cell migration is a fundamental process in multicellular organisms where tissue formation requires orchestrated movement of cells to specific locations. The migration of astrocytes is important under both, normal and pathological conditions, and the current knowledge of the mechanisms involved was summarized in
ourthe recent reviews
[43,156][15][47].
2.2. AQP3
AQP3 was first identified in kidney medulla and colon in 1994
[85][48]. It was found to be enriched in kidney collecting ducts and localized on the basolateral cell membrane of their principal cells playing a role in water transport
[85,157][48][49]. The existence and physiological function of AQP3 in astrocytes remains debated due to its relatively low abundance in the brain. In 2001, Yamamoto and colleagues showed that AQP3 is expressed in the brain and astrocytes at the mRNA level using reverse transcription–polymerase chain reaction
[51][50]. At the protein level, Yang et al. reported the induction of AQP3 expression in primary astrocyte cultures with hyperosmotic exposure using Western blot
[158][51]. Moreover, AQP3 is induced and co-localized with GFAP-positive astrocytes after permanent focal cerebral ischemia
[87][52]. It is worth mentioning that recent studies indicate there might be a continuum existing between GFAP-expressing astrocytes and other cell types such as oligodendrocytes
[159][53]. It warrants further studies to better elucidate the role of AQP3 in astrocytes.
2.3. AQP4
2.3.1. Isoforms of AQP4
AQP4 was identified by Peter Agre in 1994
[88][54] and is currently the most studied AQP in relation to its properties in astrocytes. The expression level of AQP4 in astrocytes is the highest among all AQP subtypes
[137,140,141,142][24][27][28][29]. So far, eight isoforms of AQP4 have been reported. These include AQP4a (M1), AQP4b, AQP4c (M23), AQP4d, AQP4e (AQP4 Mz), AQP4f, AQP4-△4 and AQP4ex
[160,161,162,163][55][56][57][58]. Most reports on AQP4 subcellular localizations and functions refer to the two major isoforms AQP4a (M1) and AQP4c (M23), which were the earliest found to be expressed and localized in the plasma membrane, particularly at the end-feet of rat astrocytes, and have water permeabilities
[164,165,166][59][60][61]. AQP4c is known to form higher-order structures called orthogonal arrays of particles (OAPs) in the plasma membrane, which are distinctive square arrays of particles originally seen and most densely concentrated in the plasma membrane of astrocytes
[167,168,169][62][63][64]. AQP4a can co-assemble with AQP4c to form these OAPs albeit of smaller size
[44,170,171][16][65][66]. Although there are as yet no reports regarding the protein of AQP4a and AQP4c in human astrocytes, studies of the human AQP4 genes in the brain did indicate two distinct mRNAs corresponding to these two functionally active isoforms found in the rat
[172][67]. AQP4b, AQP4d, AQP4e and AQP4f were identified in the rat brain in 2008
[160][55]. Of these four AQP4 isoforms, AQP4b and AQP4d are localized extensively in acidic compartments such as endosomes or lysosomes, and the Golgi apparatus, but they do not contribute to OAPs in the plasma membrane in transfected rat astrocytes
[173][68]. AQP4e was confirmed to have water permeability and is predominantly localized in the plasma membrane, with a minor amount in Golgi apparatus and degradation compartments
[130,160][55][69]. Similar to AQP4a, AQP4e is unable to form OAPs by itself but can associate with AQP4c to form OAPs
[161,170][56][65]. Moreover, AQP4e was found to have a novel effect on OAP reorganization and participated in the regulation of astrocyte cell volume
[174][70]. However, due to the presence of in-frame stop codons, AQP4e was only found in rat brains and not in humans or mice
[161][56]. AQP4f was believed to partially localize to the Golgi apparatus
[160][55]. The water transport capacity of the isoforms AQP4b, AQP4d and AQP4f are not clear yet
[137][24], and no information on these three isoforms in humans is available to date.
AQP4-△4 lacks exon 4 and is found in human skeletal muscle. There is no information on whether it is inherently expressed in astrocytes or not. AQP4-△4 transfected in HeLa cells resides mainly in the endoplasmic reticulum and does not show any water transport capacity
[162][57]. Moreover, AQP4-△4 transfected into astrocytes with stable expression of functional AQP4 could reduce the expression level of full-length AQP4 in the cell membrane
[162][57]. These findings suggest that AQP4-△4 may act as a pre-form of AQP4 in regulating water transport across cell membranes.
On the other hand, AQP4ex is characterized by a C-terminal extension that is generated by a programmed translational read-through and was recently identified in human, rat and mouse astrocytes
[163,175,176][58][71][72]. AQP4ex appears to be similar to AQP4a and AQP4c in terms of being present in OAPs, required to address OAPs to the perivascular astrocyte end-feet, and able to modulate OAP size and water transport
[163,175,176][58][71][72].
Since all these isoforms originated from the same pre-mRNA transcript after alternative splicing, each AQP4 isoform might play a role in influencing the total composition of AQP4 in a cell. However, it is unclear whether the proposed interactions among AQP4 isoforms would ultimately contribute to the exquisite water homeostasis function in astrocytes under both normal and pathological conditions.
2.3.2. AQP4 Polarization
AQP4 polarization occurs in astrocytes, referring to its ‘polarized’ expression primarily in the perivascular astrocytic end-feet domains
(Figure 1). Polarization is highly responsive to the extracellular microenvironment and has a preference for areas in contact with the fluid compartments in or around the central nervous system (CNS). This is best demonstrated in the specific occurrence of polarization in the boundary between parenchyma and major fluid compartments such as the brain–blood interface, the brain–subarachnoid CSF interface (glia limitans), and the brain–ventricular CSF interface (sub-ependymal astrocytes)
[177,178][73][74]. AQP4 is diffusely expressed in astrocyte end-feet covering blood vessels throughout the hippocampus
[179][75]. In the cerebral cortex, AQP4 is more enriched in the end-feet membranes adjacent to capillaries than in the parenchymal membranes and the degree of astrocytic polarization is species-dependent
[178,180,181][74][76][77]. In general, the predominant subcellular localization of AQP4 is in the astrocyte end-feet membranes directly in contact with the brain capillaries and the pia mater, and with a low but significant concentration in non-end-feet membranes, such as those astrocyte membranes that ensheath glutamatergic synapses
[10,178][74][78]. In the spinal cord, similar to the situation in the brain, AQP4 is primarily expressed in the astrocyte end-feet surrounding blood vessels. However, a strong AQP4 signal could also be found in astrocyte process wrapped around myelinated nerve fibers
[46][79]. AQP4 polarization is not found in cultured cells, where AQP4 was found to be evenly distributed throughout the cell membrane under normal conditions
[130[69][80],
182], even though these in vitro water channels were functional
[183][81].
2.4. AQP5
AQP5 was first identified in rat salivary glands in 1995
[96][82] and has been shown to convey a high degree of membrane water permeability in corneal, pancreatic, and bronchial epithelium, the secretory cells in salivary and lacrimal glands, airway submucosal glands and type I pneumocytes of the respiratory tract
[36,231,232][8][83][84].
AQP5 was first identified in primary cultures of rat astrocytes at the mRNA level, and its protein was observed in reactive astrocytes surrounding the lesion site of the hippocampus after middle cerebral artery occlusion (MCAO) induction in rats
[51][50]. As expected, only low-level expression of AQP5 was observed in human normal brain tissue
[233][85]. AQP5 protein was also detected in the piriform cortex, choroid plexus, dorsal thalamus and cerebellum Bergmann glia cells in rat brain
[52][86]. The expression of AQP5 in primary cultures of mouse cerebral cortical astrocytes at both mRNA and protein levels was further identified by
ourthe group
[53][21]. By using RNA-sequencing, Ben Barres’ group confirmed AQP5 expression in human astrocytes
[141][28]. No isoforms of AQP5 have as yet been identified.
AQP5 localizes in both the plasma membrane and the cytoplasmic compartment membranes in primary astrocyte cultures
[53][21]. The AQP5 expression was very low in 1-week primary astrocyte cultures but increased dramatically after 2 weeks
[53,127,128][21][87][88]. Similar to AQP1 and AQP4, AQP5 is also permeable to water
[7[89][90],
234], and able to activate ion channel protein TPRV4 and take part in regulatory volume decrease
[37][9]. Therefore, AQP5 in astrocytes may participate in water transportation and astrocytic regulatory volume decrease, regulating the osmotic pressure needed for neuronal functions in the brain. In addition, AQP5 can facilitate astrocyte migration. In the primary culture model
wresearche
rs built, AQP5 increased and polarized to the migrating processes and the leading edge cytoplasmic membrane of astrocytes, and its over-expression facilitated astrocyte process elongation
[53][21].
2.5. AQP8
AQP8 was initially described to be highly expressed in the testis and the liver
[235][91] and was later reported also to be expressed in neural tissues
(Table 2). Both AQP8 mRNA and protein expression have been determined in primary cultures of rat astrocytes
[51,158][50][51]. The protein expression of AQP8 was observed in astrocytes in the mouse spinal cord in 2004
[103][92]. Later, AQP8 expression was also demonstrated in astrocytes in the rat brain
[52,87][52][86]. Interestingly, considerable AQP8 mRNA was detected in adult NSCs, and its predominant expression was found in the mitochondria-enriched fraction indicating a possible role of AQP8 in regulating mitochondrial volume during NSC differentiation
[59][93]. So far, three isoforms of AQP8 have been found: the full-length AQP8 transcript and two other splice isoforms discovered in pig testis
[236][94]. Further investigations on AQP8 are required to clarify its expression and physiological roles in astrocytes and the NSCs.
2.6. AQP9
AQP9 was first identified in peripheral leukocytes in 1998
[116][95]. AQP9 is an aquaglyceroporin and functions in the transport of water, urea and glycerol
[116,117,237,238][95][96][97][98]. It was later found to also transport mannitol, purines and pyrimidines
(Table 2) [137,177,222,239,240,241,242] [24][73][99][100][101][102][103].
AQP9 expression in astrocytes was also confirmed in transcriptome studies, and its expression was higher in rodent brains than in human brains
[140,141,142][27][28][29]. AQP9 protein was detected in glia limitans, white matter and gray matter of the rodent and non-human primate brain
[60,61][104][105] To date, two isoforms of AQP9 have been identified in the brain
[239][100]: the long isoform, which appears to correspond to the liver isoform, is expressed throughout the astrocyte plasma membrane, whereas the short isoform is expressed in the mitochondria with higher enrichment in the inner mitochondrial membrane
[60,239][100][104] and a potential role in transporting lactic acid into the mitochondria
[239,243][100][106]. AQP9 is therefore very likely to be involved in brain metabolism.
In addition to AQP4 and AQP8, AQP9 was a key protein examined in NSC studies. AQP9 is expressed in adult murine subventricular zone- (SVZ-) derived NSCs
[41][13] but it has as yet not been found in human NSCs
[41,42][13][14]. It was reported that the levels of expression and cellular localization of AQP4 and AQP9 were differentially regulated upon differentiation of SVZ-derived NSCs into glial cells and neurons, indicating that AQP4 and AQP9 likely play different roles in these cells
[41][13].
Although AQP9 expression and subcellular localization have been identified in NSCs and astrocytes, interpretation of these data should be undertaken with caution. The inconsistencies between various studies may sometimes be due to the cross-reactivity with nonspecific epitopes for some AQP9 antibodies
[62,244][107][108]. Using AQP9 knockout mice, Rojek and colleagues demonstrated that AQP9 expression discovered in the mice brain could be due to artifacts related to the specificity of the anti-AQP9 antibodies used
[238][98]. However, Mylonakou et al. showed that AQP9 mRNA was present in the cells where the same AQP9 immunoreactivity was observed through in situ hybridization
[62][107]. In the same study, a lower level of AQP9 mRNA was found in mouse brain than in rat brain, which may well explain such discrepancy. Nevertheless, heterogeneity of AQP9 expression in different species and its existence of different splice variants need also to be considered. The development of novel tools with higher specificity would help to confirm the results of AQP9 expression and unravel the roles of AQP9 in astrocytes and CNS.