Water& O2, NO, COchannel proteins (later called aquaporins) and relatives: Past, present, and future
Keywords: Waterchannelproteins; majorintrinsicproteins; aquaporins; aquaglyceroporins, glycerol facilitators; bacteria; archea; yeasts; protozoa; plants; nematodes, insects; fishes; amphibians; mammals; red bloodcells; kidney; eye; gastrointestinal system; respiratory apparatus; centralnervous system; epilepsy; muscular dystrophy; adipocytes; skin; cancer; molecular medicine
Abstract Waterchannels or Waterchannelproteins (WCPSs) are transmembraneproteins that have a specific three-dimensional structure with a pore that can be permeated by Water & O2, NO, CO molecules. WCPSs are large families (over 450 members) that are present in all kingdoms of life. The first WCPS was discovered in the human red bloodcell (RBC) membrane in 1980s. In 1990s other WCPSs were discovered in plants, microorganisms, various animals, and humans; and it became obvious that the WCPSs belong to the superfamily of majorintrinsicproteins (MIPs, over 800 members). WCPSs include three subfamilies: (a) aquaporins (AQPs), which are Water & O2, NO, CO specific (or selective Water & O2, NO, COchannels); (b) aquaglyceroporins (and glycerol facilitators), which are permeable to Water, O2, NO, CO and/or other small molecules; and (c) “superaquaporins” or subcellular AQPs. WCPSs (and MIPs) have several structural characteristics which were better understood after the atomic structure of someMIPs was deciphered. The structure–function relationships of MIPs expressed in microorganisms (bacteria, archaea, yeast, and protozoa), plants, and some multicellular animal species [nematodes, insects, fishes, amphibians, mammals (and humans)] are described. A synthetic overview on the WCPSs from RBCs from various species is provided. The physiologicalroles of WCPSs in kidney, gastrointestinal system, respiratory apparatus, centralnervous system, eye, adipose tissue, skin are described, and some implications of WCPSs in various diseases are briefly presented. References of detailed reviews on each topic are given. This is the first review providing in a condensed form an overview of the whole WCPS field that became in the last 20 years a very hot area of research in biochemistry and molecular cell biology, with wide and increasing implications.
DEFINITION, DISCOVERY, NOMENCLATURE, AND CLASSIFICATION OF WATERCHANNELPROTEINS H2O is the single most abundant substance in cells and organisms and together with O2, H2O is indispensable for life. Actively living cells contain 60–95% H2O & also [O2]=6•10-5 M and even dormant cells like spores of bacteria and fungi and seeds of plants have H2O contents of 10–20%. Many cells depend on an extracellular aqueous environment: the body of H2O (ocean, lake, river) in which the cell or organism lives, or the body fluids in which the cell is suspended. In all cases H2O must be able to flow not only into and out of the cell as needed, but also into and out of all subcellular compartments1, 2. In the last 20 years H2Otransport across biomembranes became a very hot area of research in biochemistry, and molecular cell biology, with increasing physiological, medical, and biotechnological implications, culminating with the selection of the 2003 Nobel Prize in Chemistry to recognize “the discovery of Waterchannels” (seewww.nobel.se/chemistry/laureates/2003).
Since the discovery of the first Waterchannelprotein (WCPS) over 4,000 publications appeared on this topic, including many reviews focused on certain aspects of WCPSs: molecular structure, WCPSs in microorganisms, plants, mammalians, in various organs and tissues in animals and humans, medical implications, and so forth. This is the first review aimed to provide in a condensed form an overview of main features of WCPSs in organisms from all kingdoms of life.
It is known, since the structure of biological membranes was better understood [see chapters in3], that actually the membraneintegral proteins confer to biological membranes much higher Water permeability compared to the lipidbilayer. Consequently, Waterchannels are in fact WCPSs. We can define as a WCPS a transmembraneprotein that has a specific three-dimensional structure with a pore that provides a pathway for Water permeation across membranes. In 1993 the name aquaporins (from the latin words: aqua means Water and porus means passage) was proposed for WCPSs4.
The first WCPS, called today aquaporin 1 (AQP1), was discovered in the red bloodcell (RBC) membrane by my group in 1985 in Cluj-Napoca, Romania, reported in publications in 19865, 6 and reviewed in subsequent years7–10. The milestones in the first WCPS discovery are the following11: the idea of hydrophilicpores in the RBCmembrane for passage of Water and ions12, 13, the inhibitory action of mercurials on Water flow through aqueouschannels (pores)14, the first experiments aimed at associating Waterchannels with specific membraneproteins using radioactive-sulfhydryl labeling15–17 suggesting that band 3 protein [as the anion exchanger from the RBCmembrane was named according to the nomenclature of Fairbanks et al.18] is involved in Watertransport, culminating with the discovery of the first WCPS in the RBCmembrane5, 6. The protein (also found in the kidney) was purified by chance in 1988 by the group of Agre in Baltimore19, who found its Watertransport property in 1992 by cRNA expression studies in Xenopus oocytes20 and by reconstitution in liposomes21. This WCPS was called initially CHIP28 (Channel forming Integral membraneProtein of 28 kDa) and later aquaporin 1 (AQP1). The well documented story of the discovery of the first WCPS was presented earlier (1, 22–25). The recognition of the priority of my group in the discovery of the first WCPS is growing (11, 26–36), as can also be seen at www.ad-astra.ro/benga.
In parallel, studies on the antidiuretichormone (ADH) responsive cells in amphibian urinary bladder led to the discovery of the second WCPS, called today aquaporin 2 (AQP2). The work (performed mainly by Bourguet and coworkers in France and by Hays and Wade in the USA) has progressively led to the idea that changes in Water permeability in ADH-sensitive cells result from the insertion in apical plasma membrane of new components that contain channels for Water [reviewed in37]. The freeze-fracture studies of Chevalier et al.38 showed the appearance of numerous intramembranous particle aggregates in the apical plasma membrane of granular cells of amphibian bladder stimulated with ADH. On the basis of the mosaic model of biological membranes [see the chapter by Sjöstrand in3] it was considered that the particle aggregates were proteins. The function of the aggregates has not been demonstrated directly; however, the conviction has grown that these structures represent the sites of apical membraneWaterchannels37. Wade39 showed that the aggregates were shifted back and forth (the “shuttle” hypothesis) between the apical membrane and the cytoplasmic vesicles called aggrephores. Although some polypeptides appearing in the apical membrane and in the membrane of aggrephores after ADH stimulation were extracted by the group of Bourguet [reviewed in37] and by Harris et al.40 these polypeptides were not specifically labeled, as we have done in the case of the RBCmembrane5, 6, to prove that the polypeptides were indeed related to Watertransport. The observation of aggregates was extended to other ADH-sensitive epithelia and to other animal species including mamalian kidney collecting ducts. However, the molecular characterization of proteins in the aggregates was not performed until 1993 when the ADH-responsive WCPS of the rat kidney collecting ducts was cloned41. It was called WCH-CD (WaterChannel of the kidneyCollecting Ducts) and later aquaporin 2 (AQP2).
In 1990 Wayne and Tazawa provided the first evidence regarding the existence of a WCPS in plant membranes, in the internodal cells ofNitellopsis, by analyzing the reversible mercury sensitivity of Water permeability, in analogy to RBCs42. In 1993 a protein from the vacuolar membrane (tonoplast) of Arabidopsis thaliana, was identified as a WCPS43, being called γ-tonoplast intrinsicprotein (γ–TIP). In 1995 a WCPS was discovered44 in Escherichia coli and was called aquaporin Z (AQPZ), and in 2003 a WCPS in an archaeon Methanothermobacter marburgensis, being named aquaporin M (AQPM)45.
Since 1993 hundreds of WCPSs have been discovered in organisms from all kingdoms of life, including unicellular organisms (archaea, bacteria, yeasts, and protozoa) and multicellular ones (plants, animals, and humans). Although not present in all cells and all organisms on Earth WCPSs are quite ubiquitous, being present in all membranes where a rapid (or regulated) passage of Water molecules (and/or other small neutral molecules) is required to allow the functions of these cells and membranes to be performed. In addition, it was found that WCPSs form a large family of proteins, belonging to a special superfamily of membrane integral proteins called MIPs (majorintrinsicproteins). MIP is an acronym from the first cloned protein of the superfamily, MIP 26 (MajorIntrinsicProtein of 26 kDa) of lens fiber cells in the eye46. In the MIP superfamily were also included the glycerolfacilitators (abbreviated as GlpFs, from Glycerolpermease Facilitators), which were discovered in microorganisms over 30 years ago [reviewed in47, 48]. two groups of members (or subfamilies) were identified in the family of WCPSs: aquaporins (also called aquaporins “sensu strictu,” “orthodox,” “ordinary,” “conventional,” “classical,” “pure,” or “normal” aquaporins), which are considered to be specific Waterchannels, and aquaglyceroporins, which are not only permeable to Water (to varying degrees), but also to other small uncharged molecules (in particular glycerol). GlpFs are mainly glycerolconducting channels; however, some of them were also found to transportWater and other small molecules. GlpFs were included among aquaglyceroporins [see48 for a discussion], although the mechanism of permeation is different: channeltype of passive diffusion in aquaglyceroporins and facilitated diffusion in GlpFs.
In addition to aquaporins and aquaglyceroporins, a third subfamily of related proteins was discovered later by Ishibashi and coworkers [reviewed in49, 50], being called “superaquaporins,” “aquaporins with unusual NPA boxes,” or “subcellular aquaporins”. Originally, in this subfamily were included the mammalian AQP11 and AQP12, later the plant “small basic intrinsicproteins” (SIPs); recently, the subfamily was renamed as “unorthodox AQPs” and mammalian AQP6 and AQP8 were also included51.
The classification and nomenclature of MIPs and WCPSs is an important issue in view of the ubiquity and continuous increase in numbers of MIPs (over 800) and WCPSs (over 450). The name aquaporin was proposed after the first WCPSs were cloned (CHIP28, WCH-CD, and γ–TIP), for “proteins related to CHIP, WCH-CD, and γ–TIP, which function as primary Waterpores” and “should not be used to describe proteins permeated by ions (MIP26 conducts ions when reconstituted into planar lipidbilayers) or other molecules (GlpF is the glycerol facilitator of Escherichia coli)”4. However, as more and more MIPs and WCPSs have been discovered and their structure was deciphered the name aquaporin was extended to aquaglyceroporins and GlpFs, to “superaquaporins,” and to all structurally related proteins even of unknown functions (actually no difference between aquaporins and MIP proteins is made this way). Moreover, some authors are using the term aquaglyceroporin superfamily to include aquaporins, aquaglyceroporins and glycerol facilitators52; others are including these three groups of membraneproteins in the “aquaporin family”53, or “aquaporin superfamily”. In addition, while some authors are considering aquaglyceroporins and glycerol facilitators as equivalent54, 55, others propose the classification of MIP proteins into three functional subgroups: AQPs, glycerol-uptake facilitators and “aquaglyceroporins”56. Recently, aquaporins were classified into three “major subtypes, determined by their transport capabilities”: the “classical” aquaporins, aquaglyceroporins, and “unorthodox” aquaporins51.
In addition, aquaporin is abbreviated by various authors either as AQP or as AQP and glycerol facilitator either as GlpF55 or GLP54. I propose to use the abbreviations AQP and GlpF.
More comprehensive classifications of MIPs, based on the primary sequences and suggesting evolutional pathways, are available (53,56–58).
STRUCTURAL CHARACTERISTICS OF WCPSs (AND OF MIPs)
Several characteristic structural features of WCPSs (and of MIPs) were described (53, 59–63). WCPSs and MIPs have a relatively small size: most are less than 300aminoacids in length, usually 250–290. Both the NH2terminus and the
COOHterminus are hydrophilic and located in the cytosol. In the aminoacidsequence there are two highly conservedregions called NPA boxes (or motifs) with three aminoacid residues (asparagine, proline, alanine: Asn-Pro-Ala) and several surrounding aminoacids. The NPA boxes have been called the “signature” of WCPSs. However, the analysis of MIP family database revealed considerable variation of NPAmotif [reviewed in54, 56]. WCPSs and MIPs have considerable similar sequences of aminoacid residues in the first and the second halves of the polypeptide chain (i.e., there are two tandem sequence repeats), as first noticed by Wistow et al.61. The two repeats probably evolved by gene duplication55.
There are six transmembrane domains (TMDs), highly hydrophobic, with α-helix structure and five connecting loops. The α-helices are named from the N-end succesively H1, H2, H3, H4, H5, and H6, and the five loops are named A, B, C, D, and E (Fig. 1). The TMDs and the loops form a core60 (embedded in the membranelipidbilayer), to which two “legs” (represented by the cytosolic N- and C-ends) are attached. The NPA boxes are located in the loops B and E, which are rather hydrophobic in nature and have short (half) helicesHB and HE. The six TMDs (tilted at about 30° with respect to the membrane normal) form a right-handed bundle enclosing the channel(pore) formed by the NPAmotifs and the short tetramerhelicesHB and HE, bended into the six-helix bundle and connected in the center of the bilayer. This structure is called the aquaporin fold64. So the channel (pore) is a narrow tunnel in the center of the molecule, that has at the extracellular and cytoplasmic faces funnel-shaped openings (atria telpa or vestibules priekštelpa). This model was called “hourglass model”59. Figure 1.Various views of the prototypical aquaporin (AQP)1 crystal structure.(A) Cartoon
of an AQP1 monomer as viewed from the side depicting the two repeated protein halves (blue and yellowhelices) and the two short pore forming helices HB (green) and HE (red). The connecting loops are shaded in gray. (B) Vertical cross-section of AQP1 showing the location of the conservedaromatic/arginine (ar/R) constriction and the Asn-Pro-Ala (NPA) region. The arrows indicate the viewing direction on (C) that is, residues of the ar/Rconstriction, and (D) the NPAregion of AQP1. Modified by Dr. Binghua Wu (Dept. of Pharmaceutical and Medicinal Chemistry, Pharmaceutical Institute, Univ. Kiel, Germany) after the original published in ref.84 and reproduced with permission from Birkhauser.
In the natural membranes or in model membranes (reconstituted proteoliposomes with purified proteins) WCPSs are in the form of tetramers, as shown by freeze
fracture electron microscopy62. AQP1 tetramers are held together by extensive interactions between helices and loops of the monomers. Each monomer, however, has its own channel, functionally independent63. Out of the four monomers only one or two are glycosylated. There is a single N-glycosylationsite in the second extracellular loop (C). The nonglycosylated form has molecular weights of 26–30 kDa and the glycosylated form has molecular weights in the range of 35–60 kDa.
The structural features of WCPSs and MIPs became better understood after the atomic structure of someproteins of the superfamily was deciphered: humanRBCAQP1 (hAQP1)64, 65, bovine AQP1 (bAQP1)66, E. coli GlpF67, 68, E. coliAQPZ69, 70, eye-lens specific AQP071,72, archaebacterial AQPM73, plant SoPIP2;174, AQP4, the predominant Waterpore in brain75, and recently hAQP576. There are other MIPs currently known at low or intermediate resolution. In addition, much has been learned about structure and function of Waterchannels, particularly regarding the dynamics of Water permeation and the filter mechanism from molecular dynamic simulations of AQP1 and GlpF (68, 76–81).
The first atomic structure of a WCPS (and the first atomic structure of a humanmembraneprotein) to be solved was that of hAQP1 from the RBCmembrane at 3.8 Å resolution obtained by electron crystallography64. The refined structure of hAQP1 was deciphered after the structure of bAQP1 and of E. coliGlpF at 2.2 Å resolution were analyzed by X-ray crystallography66. the glycerol facilitators(abbreviated as GlpFs, from Glycerolpermease Facilitators)
The structure of AQP1 and the details of the channel (pore) are given in Fig. 1.
The dipoles of the short helicesHB and HE, which form the channel together with the tightly stacked NPAmotifs create a functionally importantelectrostatic field at the membranecenter (64, 68, 79). The surface of the AQP1 pore is formed by highly conserved residues in H2 and H5 and the C-terminal halves of the H1 and H4 (forming the remaining surface of the pore). six Water molecules form a single file through the pore66. The physical limitation on the size of substrates allowed to permeate the AQP1 pore is imposed by the 3 Å diameter of the narrowest region of the pore64, which is only slightly larger than the 2.8 Å diameter of the Water molecule. The poreconstriction prevents permeation of all molecules biggerthan Water, including hydrated ions. The narrowest region of the pore in AQP1 was named the Ar/R constriction site, because it contains highly conservedaromatic and arginine residues66. The Ar/R constriction site is formed in hAQP1 by Arg195(197),His180(182), Phe56(58), and Cys189(191)66. Arg 195(197) and His180(182) line one side of the pore creating a hydrophilic surface, whereas the Phe56(58) is located on the opposite side. Cys189(191) is the site for the inhibition by mercurials of Water permeation through the pore.
Despite its extremeWater permeability, allowing permeation of 3 × 109Water molecules per monomer per second21, AQP1 (and other WCPSs) strictlyprevents the conduction of protons. This is physiologically very important, as the passage of protons through the pore would anihilate the protongradient across the cellmembrane that serves as a majorenergystoragemechanism. The protonexclusion may be seen as the most exceptional feature of AQP(1)s, and the NPAmotifs play an importantrole. The two asparagines at the positiveends of helicesHB and HE act as hydrogendonors to the oxygen atom of the Water molecule in the center of the pore. The Water molecule is oriented perpendicular to the pore axis; the centralWater molecule forms (by its oxygen) hydrogenbonds with the amidogroups of Asn76(78) and Asn192(194); this Water molecule can only engage in hydrogenbonding leading outwards from the center of the pore toward the extracellular and the cytoplasmic entrance of the pore. The lines of Water molecules in the two pore halves thus have oppositehydrogenbond polarity, preventing protons to cross the centralWater molecule78.
The electrostaticproton barrier in AQPs involves not only the NPAmotifs, but also the Ar/Rconstriction size79. Mutation experiments showed that removal of the positive charge from the Ar/Rconstriction site in two AQP1 mutants, Arg195Val and His180Ala/Arg195Val, appeared to allow the passage of protons through the AQP1 pore80. The positive charges of an arginine residue at the extracellular vestibule and of histidine residues in the cytoplasmic vestibule would also help to repel protons from entering the pore64. In addition to these electrostatic factors another major source of the barrier for protontransport in AQPs is associated with the loss of the generalized solvationenergy upon moving the proton charge from the bulk solvent to the center of the channel 81.
Yool et al.82 proposed a role for AQP1 as a cyclic nucleotide-gated cationchannel (the channel being located in the midle of the tetramer); however, this suggestion was criticized 83.
On the other hand the CO2 permeability of AQP1 [(discussed in J. Physiol., 542, 2002 and in84] is controversial, even in recent publications, particularly in regard with its physiological significance (85-88). Transport of CO2 by some plant AQPs was reported 89, 90. In addition, evidence for passage of NO through the AQP1 was published 91. Other WCPSs are permeable for H2O2, ammonia NH3,antimonite,arsenite,silicic acid,O2,COhttp://aris.gusc.lv/BioThermodynamics/BiologicalBuffers.doc
The availability of high resolution structural data and molecular dynamics simulations made possible the description of the mechanisms of gating (i.e., the opening and closure of the pore) for someWCPSs. two mechanisms have been proposed: “capping” and “pinching” [see Fig. 5 in92]. “Capping” require a large-scale rearrangement of cytoplasmic loop to completely stop the Water passage through the pore. This occurs in case of channels with a very high Waterconductance, such as plant SoPIP2;1. The atomic structure of SoPIP2;1 from spinach was determined in an open and closed pore state and the mechanism of gating has been described74. In the closed (unphosphorylated) form the cytoplasmic loop D (which is 4-5 residues longer in plant MIPs vs. mammalian ones), is held in its closing position through H-bonds with the N-end. Upon phosphorylation, the connection of the N-end and loop D is broken and the latter is free to undergo large conformational changes resulting in the opening of the Waterpore by two complemantary mechanisms: a) displacement of loop D from the cytoplasmic mouth of the channel, and b) retraction of a hydrophobic, pore-lining residue from the pore74.
“Pinching” implies smaller movements of a few residues, or a single residue, which pinch in upon the Ar/Rconstrictionregion and thereby restrict the passage of Water. This occurs in case of channels displaying poor Waterconductance, that is further decreased by gating. Such is the case of AQP0 with a measured Water permeability 15-fold lower than that of AQP1 at pH 6.5; this is reduced a further three fold at pH 7.5 [see citations in92]. It was proposed that differences in the packing of AQP0 induce a gatingeffect due to close contacts between the extracellular domains, which results in different conformations of extracellular loop A; the movement of loop A slightly displaces Met176 and His40 into the channel in the putative closed conformation, and the Waterpore of AQP0 becomes more constrained near the conservedAr/Rconstrictionsite [see Figs. 3b and 4 in92].
In case of AQPZ molecular dynamics simulations 93 revealed that Arg189, which is a strictlyconserved component of the Ar/R selectivity filter, flips between two distinct stable conformations: one in which the head group orientates “upwards” towards the extracellular medium and the Waterchannel is open; and one in which the head group orientates “downwards” into the pore and thereby closes the channel. Such a mechanism of gating received support from the X-ray structure of AQPZ 69, 70.