Aquaporin CHIP: the archetypal molecular water channel

Aquaporin
CHIP:
the archetypal
molecular
water channel
PETER AGRE, GREGORY
M. PRESTON,
BARBARA
L. SMITH,
JIN SUP JUNG,
SURABHI
RAINA, CHULSO MOON, WILLIAM
B. GUGGINO,
AND SBREN NIELSEN
Johns Hopkins
University
School of Medicine, Baltimore, Maryland 21205;
and University of Aarhus, Aarhus DK-8000, Denmark
WATER
IS THE MAJOR
COMPONENT
of all living cells as
well as the surrounding extracellular spaces. Transport
of water into and out of cells occurs during digestion,
respiration, circulation, regulation of body temperature,
elimination
of toxins, neural homeostasis, and during
many other essential body functions. Water crosses cellular plasma membranes by two fundamentally
distinct
pathways: 1) simple diffusion through the lipid bilayer;
and 2) channel-mediated
water transport, a process
which has been described physiologically
despite a lack
of molecular understanding
of the channel structure
(20, 26a, 55).
Diffusional Water Permeability
Diffusional water permeability
is a characteristic
of
the plasma membranes of all tissues. The coefficient of
diffusional permeability
(Pd) is variable among tissues
0363-6127/93
$2.00 Copyright
but is usually of relatively small magnitude. Diffusional
water permeability
can be measured by nuclear magnetic
resonance imaging or isotopic methods under isosmotic
conditions, and this is not pharmacologically
inhibitable.
Transport of water across pure lipid membranes occurs
solely by d iffusion, therefore the ratio of the coefficients
of osmotic water permeability to diffusional permeability
approximates unity (Pflpd = 1). Diffusional water permeability is strongly affected by the organization of membrane lipid molecules and is characterized by high Arrhenius activation energy (E, > 10 kcal/mol), indicating that
greater water movement will occur through the lipid bilayers at higher temperatures when the lipid mobilities
are increased (19).
Membrane Water Channels
Recognition of the existence of aqueous pores in the
plasma membranes of red blood cells (RBC) emerged
0 1993 the American
Physiological
Society
F463
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
Agre, Peter, Gregory M. Preston, Barbara L. Smith, Jin Sup
Jung, Surabhi Raina, Chulso Moon, William B. Guggino, and
Sgren Nielsen. Aquaporin CHIP: the archetypal molecularwater channel.
Am. J. Physiol. 265 (Renal Fluid Electrolyte Physiol. 34): F463-F476, 1993.
-Despite longstandinginterest by nephrologistsand physiologists,the molecular identities of membranewater channels remained elusive until recognition of CHIP, a 2%kDa channel-forming integral membraneprotein
from human red blood cells originally referred to as “CHIP28.” CHIP functions asan osmotically driven, water-selective pore; 1) expressionof CHIP
conferred Xenopus oocytes with markedly increasedosmotic water permeability but did not allow transmembrane passageof ions or other small
molecules;2) reconstitution of highly purified CHIP into proteoliposomes
permitted determination of the unit water permeability, i.e., 3.9 x log water
molecules.channelsubunit-l .s-l. Although CHIP exists asa homotetramer
in the native red blood cell membrane, site-directed mutagenesisstudies
suggestedthat each subunit contains an individually functional pore that
may be reversibly occluded by mercurial inhibitors reacting with cysteine189. CHIP is a major component of both apical and basolateralmembranes
of water-permeablesegmentsof the nephron, where it facilitates transcellular water flow during reabsorption of glomerular filtrate. CHIP is also
abundant in certain other absorptive or secretory epithelia, including choroid plexus, ciliary body of the eye, hepatobiliary ductules, gall bladder,
and capillary endothelia. Distinct patterns of CHIP expression occur at
these sites during fetal development and maturity. Similar proteins from
other mammalian tissuesand plants were later shown to transport water,
and the group is now referred to asthe “aquaporins.” Recognition of CHIP
has provided molecular insight into the biological phenomenonof osmotic
water movement, and it is hoped that pharmacological modulation of
CHIP function may provide novel treatments of renal failure and other
clinical problems.
channel-forming integral membrane proteins; water channels; fetal development; cysteine; channel-mediated water transport; osmotic water
permeability
F464
EDITORIAL
Search for Molecular
Identity
of a Water Channel
Candidate water channel proteins. Several attempts
to
isolate a molecular water channel have been made. These
were fraught with difficulty because of the confounding
existence of high baseline diffusional water permeability
and the lack of specific pharmacological
inhibitors
needed for unambiguous, covalent labeling of water chan-
nel proteins (14,26,62). Over the years several candidate
water channels were proposed.
Solomon et al. (54) postulated that the band 3 anion
exchanger (AEl) of RBC functions as an aqueous pore
permeated by water and urea. Evidence supporting this
theory included the large number of copies of band 3
(lo6 copies/RBC) and the existence of a labile sulfhydryl
group in band 3 that is covalently modified by pCMBS,
an organomercurial
inhibitor of water channels. Inconsistencies in the band 3-water pore hypothesis were subsequently noted, including discordant inhibition
of anion transport by stilbene analogues and water transport
by organomercurials
(36). More recent studies using the
Xenopus oocyte expression system also failed to demonstrate concordance of water transport and anion transport (69).
Certain known membrane
transporters
have been
shown to conduct some water in addition to their primary
substrates. The sodium-independent
glucose carrier
(GLUTl)
was demonstrated to permit water transport in
addition to glucose by Fischbarg et al. (21, 22); GLUT1
was therefore proposed as the membrane water channel.
These observations were confirmed by other laboratories, although it is now agreed that GLUT1 is not the
classically defined water channel with low E, and mercurial sensitivity (15, 67, 69). Expression of the cystic fibrosis transmembrane
regulator (CFTR) in Xenopus
oocytes was shown by Hasegawa et al. (27) to confer a
slight rise in membrane water permeability
in addition
to increased permeability
for other ions and molecules
when activated with CAMP. Estimates based on indirect
observations suggested the unit water permeability
may
be large; however, these studies await confirmation
with
direct measurement of CFTR-mediated
water transport.
Therefore,
the potential
biological
importance
of
GLUT1 - and CFTR-mediated
water movement remains
to be explained.
A more direct approach to the identification
of the
molecular water channel was taken by Harris et al. (26)
who isolated integral membrane proteins from intracellular vesicles of toad bladder epithelium
and from the
apical plasma membranes after treatment with vasopressin. These vesicles contain several integral membrane
proteins of which 55- and 53-kDa proteins are the most
abundant. The functions of these proteins are not yet
established, but roles in the membrane targeting or function of vasopressin-regulated
water channels have been
proposed.
Another direct approach was taken by Tsai et al. (56)
and Zhang et al. (71), who attempted to isolate water
channel cDNAs by expression cloning. mRNA was isolated from water-permeable
tissues and injected into
Xenopus laeuis oocytes for osmotic swelling measurements. When transcripts from reticulocytes and kidney
were expressed, the oocytes developed a significant increase in osmotic water permeability that was sensitive to
mercurial inhibition
(56,71). Osmotic water permeability
was not increased in control-injected
oocytes or oocytes
injected with mRNA from non-water-permeable
tissues.
Although logical, the most promising cDNA isolated
by this approach encoded a protein that conferred the
oocytes with only a modest increase in water permeability
and that lacked mercurial inhibition
(70).
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
from the biophysical studies of Paganelli and Solomon
(42) and Side1 and Solomon (51) in the 1950s. Channelmediated water transport permits movement of water in
the direction of greater osmolality, a recognized feature of
RBC and a subset of epithelia (20).
The Pf was measured spectroscopically under the driving force of osmotic gradients and was believed to represent channel-mediated
water transport (53). The magnitude of channel-mediated
water movement
is large;
therefore the ratio of PflPd >> 1 (11). The pace of water
transport is often so fast that the net fluid absorption
appears to be isosmotic (6). Membrane water channels
were first shown to be inhibited by mercurial sulfhydryl
reagents including HgC12 and p-chloromercuribenzenesulfonate (pCMBS) by Macey and Farmer (36). Unlike
diffusional
water permeability,
channel-mediated
osmotic water permeability
is characterized by a relatively
low Arrhenius activation energy that is close to the rate of
diffusion of water in bulk solution (E, < 5 kcal/mol). The
pathway is relatively unaffected by reductions of membrane lipid fluidity at subphysiological temperatures (63).
Most membrane water channels are thought to be constitutively activated; apparently no gating or other regulation is needed for water to permeate the channels in the
direction of an osmotic gradient. This includes the water
channels in membranes of RBC and in the apical and
basolateral membranes of renal proximal tubules and descending thin limbs of Henle’s loop where SO-90% of the
glomerular filtrate is reabsorbed, a volume approaching
200 liters/day for an adult human (33).
The water permeability
of renal collecting ducts and
amphibian urinary bladder is controlled by vasopressin
(antidiuretic hormone) which is released from the neurohypophysis in response to osmotic stimuli. Circulating
vasopressin associates with specific membrane receptors
in renal collecting duct epithelia and raises adenosine
3’,5’-cyclic monophosphate
(CAMP) levels. On the basis
of studies of toad urinary bladder and later of collecting
duct, vasopressin-mediated
regulation of apical membrane water permeability
was shown to involve endocytosis and exocytosis of water channels located in the apical membrane
and intracellular
compartments.
This
process, proposed by Wade et al. (64) as the “membraneshuttle mechanism,” was based on evidence generated by
their laboratory and others (10, 12).
The first known molecular water channel was recently
identified, the channel-forming
integral membrane protein CHIP, a 28-kDa integral membrane protein from
RBC and renal tubules (16, 45, 46, 52). At least two related proteins from mammalian
and plant tissues were
subsequently shown to function as water-selective pores,
and this group of molecular water channels is now referred to as the “aquaporins,”
abbreviated “AQP” (4).
This review will concentrate upon CHIP, since it is the
archetypal molecular water channel.
REVIEW
EDITORIAL
Meanwhile, van Hoek et al. used target analysis after
radiation inactivation
to estimate the molecular mass of
water channels. Renal brush-border membrane vesicles
(58) and RBC membranes (59) were subjected to electron
pulses with other membrane proteins serving as calibration standards. These studies further enhanced the prevailing view that water channels are membrane proteins
and provided the first evidence of a discrete functional
size of 30 t 3 kDa.
A Novel 28-kDa Integral Protein
F465
Table 1. Physical properties
of aquaporin CHIP protein
Subunits, apparent mol mass
28 kDa (SDS-PAGE)
CHIP
40-60 kDa (SDS-PAGE)
GlyCHIP (N-glycosylated)
-3:l
CHIP:glyCHIP
CHIP membrane oligomer
Stokes radius
61 A
Sedimentation coefficient
5.7 s
Partial specific volume
0.795 ml/g
Triton X-100 bound
0.41 mg/mg protein
Calculated mass of detergent-protein
190 kDa
complex
Calculated mass of protein
135 kDa
Fraction of total membrane protein
Human red blood cells
2.4%
Rat renal cortex
Total membranes
0.9%
BBMV
3.8%
Values were derived from Denker et al. (16), Smith and Agre (52),
and Nielsen et al. (40). BBMV, brush .-border membrane vesicles; SDSPAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
tion of the pure CHIP or detergent solubilized membranes in HgO or deuterated HZ0 revealed the protein to
be a noncovalently
associated homotetramer,
probably
comprised of three nonglycosylated
subunits and one
bearing a large polylactosaminoglycan
which may be removed by N-glycanase (Table 1; Fig. 1). The biological
explanation for this apparent nonstoichiometric
glycosylation of CHIP subunits remains to be established, but
this feature of the protein is also found in nonerythroid
tissues and in RBC from other mammalian
species (40).
Isolution of aquuporin CHIP cDNA. The original cloning of the CHIP cDNA was achieved by use of a polymerase chain reaction (PCR)-based strategy, since antibody screening of expression libraries failed to provide an
insert with recognized features of the CHIP protein (45).
The NHz-terminal
amino acid sequence obtained from
the purified CHIP protein permitted design of degenerate oligonucleotide
primers for amplification
of erythroid cDNA templates from a human fetal liver cDNA
Fig. 1. Schematic model representing CHIP integral membrane protein
within the membrane lipid bilayer. Notable features include 1) homotetrameric complex with 1 subunit bearing a polylactosaminoglycan, 2)
minimal polypeptide mass extending above or below the lipid bilayer ,
and 3) possible individual water pore within each subunit.
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
The water channel protein of RBC, renal proximal tubules, and other tissues was initially isolated as a byproduct of another purification.
A 28-kDa protein copurified
with the 32-kDa subunit of the human RBC Rh polypeptides (3, 49). Although first regarded as a proteolytic
breakdown product of the Rh polypeptide, antisera raised
to partially purified Rh polypeptides reacted on immunoblots with only a discrete protein of 28 kDa and a diffuse
protein of ~40-60 kDa. The 28-kDa protein failed to
stain with Coomassie blue and did not correspond to any
known membrane protein. Curiosity was enhanced by the
apparent abundance of the protein (-150, 000 copies of
28-kDa protein per RBC) and by development of a simple
purification
scheme that used extraction of membrane
vesicles with N-lauroylsarcosine
(16, 52).
Initial characterization
revealed that this is an integral
membrane protein, existing in membrane vesicles extracted of all peripheral proteins. The 40- to 60-kDa protein represented a similar or identical 28-kDa core protein with a complex carbohydrate attached by asparagine
linkage. A highly related or identical protein was also
isolated from renal proximal convoluted tubules and descending thin limbs of Henle’s loop. It was speculated
that the protein functions as a transporter or provides a
secondary linkage of the membrane skeleton to the lipid
bilayer (16).
Biochemical characterizations. Further studies of the
28-kDa protein suggested that it is a membrane channel
(52). The protein received the descriptive name of
CHIP28 for “channel-forming
integral protein of 28 kDa”
(45) and was rechristened “aquaporin CHIP” to conform
with a group of structurally and functionally related homologues (4).
CHIP was shown to be distinct from the Rh polypeptides, and the NHz-terminal
amino acid sequence was
determined for the first 35 residues of CHIP purified
from human RBC (52). The apparent linkage of CHIP to
the membrane skeleton was found to reflect an inherent
lack of solubility of the protein in Triton X-100, which
could be overcome at higher concentrations of detergent,
rather than a high-affinity
linkage with the membrane
skeleton (52). Selective proteolytic digestions and immunoblotting
with antibodies specific for the 4-kDa
COOH-terminal
domain of CHIP indicated that the
COOH terminus is located intracellularly.
Despite initial
confusion that resulted from a contaminating
antibody,
CHIP contains no known extracellular proteolytic cleavage sites.
CHIP exists as a tetramer within the native membrane
(52). Several studies including lectin affinity chromatography, glutaraldehyde
cross-linking,
and gel filtration
yielded evidence of oligomerization.
Velocity sedimenta-
REVIEW
F466
EDITORIAL
library. Size-selected amplification
products were subcloned, sequenced, and used to design nondegenerate oligonucleotide
primers that were used together with
lambda primers to amplify by PCR the 5’ and 3’ ends of
the cDNA. These products were then used to isolate a
nearly full-length 2.9-kb cDNA from a human bone marrow library. Northern blots of RNA from human bone
marrow, mouse spleen, and kidney contained single major
transcripts of 3 kb.
Analysis of the cDNAs confirmed their authenticity
(45). The sequences of the PCR products from the 5’ end
of the cDNA contained an initiating
methionine with a
strong translational
initiation
consensus prior to the de200
100
I111~1111
llrl~lrll
lrrl~r
3
.-52Ev2
-2
0
-.
+
J
> -2 1
68-1
Extracellular
1
0
-1
L
-2
duced amino acid sequence corresponding exactly to the
NH2-terminus
of the protein. An 8O%bp open reading
frame encoding a protein of 28.5 kDa was followed by -2
kb of 3’ untranslated sequence ending with a polyadenylation signal. Authenticity
of the 3’ end of the coding sequence was confirmed by expression in bacteria and immunoblot
analysis with an antibody specific for the
COOH-terminal4
kDa of CHIP. Hydropathy analysis of
the deduced amino acid sequence revealed six potential
bilayer-spanning
domains and five connecting loops.
Loops B and E are significantly hydrophobic, and loops A
and E contain potential N-glycosylation
sites (Fig. 2).
The lack of a hydrophobic leader peptide, distribution
of
charged amino acids, and immunoelectron
microscopic
studies (41) all indicate that the NH2 terminus is intracellular (Fig. 2, bottom). The cDNAs encoding the mouse
(34) and the rat homologues (13) were subsequently identified and are -94% identical to human CHIP.
Homology with the major intrinsic protein of mammalian lens, 1MIp26. The full-length coding sequence firmly
established the homology with the family of proteins related to MIP26, the major intrinsic protein of lens (24).
When their NH2 terminals are aligned, CHIP and MIP26
are 42% identical overall, with the greatest levels of homology existing in loops B and E, which are > 60% identical (Fig. 2, bottom). The cDNAs encoding several sequence-related proteins of 26-29 kDa had been isolated
from diverse organisms including
mammals,
insects,
plants, and microbes (43). Lesser homologies were found
between CHIP and the other MIP26 homologues. Although the functions of most of these proteins are unknown, it is thought that they form membrane channels.
An internal homology of the NH2-terminal
half and the
COOH-terminal
half of the MIP26 molecule was identified (43,66). CHIP and the other members of this group
of proteins have a three amino acid motif, Asn-Pro-Ala
(NPA) located at corresponding points in loops B and E
of the tandem repeats. The proposed orientation
of the
tandem repeats is unprecedented, since the first and second repeats would be oriented at 180’ angular to each
other (Fig. 2, middle). Initial topographical studies suggest that the proposed topology is correct.
The secondary structure of MIP26 was previously
characterized by near and far ultraviolet circular dichroism of octylglucoside-solubilized
MIP26 and detergentfree membrane fragments (31); it was concluded that the
a-helix content is -5O%, and the P-sheet content is
do%,
and this proportion
was not altered when the
COOH-terminal
5kDa domain was proteolytically
removed. Preliminary
studies of CHIP revealed a similar
composition (61). Although apparently not glycosylated,
Fig. 2. Predicted membrane topology of the deduced amino acid sequence of CHIP. Top: hydropathy analysis. Middle: internal homology
of the primary sequence of CHIP showing predicted 180” orientation of
tandem repeats. Solid circles, residues that are identical to corresponding residue of other tandem repeat; shaded circles, conservative substitutions; open circles, nonconserved residues. A-E correspond to loops
labeled at bottom. Bottom: predicted topology of primary sequence of
CHIP. Solid circles, residues that are identical to corresponding residue
in the major intrinsic protein of mammalian lens, MIP26 (24). * Potential N-glycosylation sites; Asn-Pro-Ala (NPA) motifs are bracketed;
arrow indicates cysteine-189. [Reprinted from Preston and Agre (45)
and Preston et al. (46).]
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
3 -
REVIEW
EDITORIAL
Aquaporin CHIP Transports Water
Several clues together strongly suggested that the
CHIP protein is a molecular water channel. The total
number of copies of glycosylated and nonglycosylated
subunits, -2 X lo5 copies/RBC (16,52), approaches the
predicted number of water channels, 2.7 X lo5 (54). The
28.5-kDa subunit size (45) is in close agreement with the
predicted size determined by radiation inactivation
of
native water channels (58, 59). The distribution
in RBC
and renal proximal tubular epithelia (16) are consistent
with the known distribution
of constitutively
active water channels but not the vasopressin-regulated
water
channels (33). Like water channels in native membranes
(8), CHIP resists proteolytic degradation while within
intact membranes. Most surprisingly, expression of one
homologue of the MIP26 protein family was induced in
the roots of pea plants grown under drought conditions
(25).
CHIP-mediated
water channel activity was first demonstrated with the X. lueuis oocyte expression system.
This system had previously been used in the search for
molecular water channels (21, 71), because of the low
diffusional water permeability _ exhibited
by the oocytes
.
that survive in freshwater ponds. An expression construct
was made by inserting the CHIP coding sequence between 5’ and 3’ untranslated sequences from X. laeuls
p-globin cDNA. In vitro transcribed RNA in 50 nl of
water was microinjected
into defolliculated oocytes; control oocytes were injected with 50 nl of water without
RNA. CHIP protein expression was assessed by immunoblot and found to be maximum after 72 h of incubation
(46). The Pf was determined by transferring the oocytes
from 200 to 70 mosmol/kgHsO
Barth’s buffer with measurement of oocyte swelling by video microscopy. Oocytes
expressing CHIP invariably ruptured after increasing in
volume by 30-50s (Fig. 3A). On-line computer integrations of surface tracings were generated throughout the
swelling experiments and permitted determination
of Pf
(Fig. 3B).
Several experiments indicated that CHIP expressed in
oocytes behaved in a manner characteristic of the water
channel in native membranes (46). When measured at
22”C, the CHIP oocytes exhibited an increase in Pf (~200
A
1.5
2.5
315
Minutes
B
.x CHIP+Hg+ME
ti
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
the hydrodynamic
behavior of the homologous protein
MIP26 had previously been characterized and was also
found to exist as a tetramer (2). It is assumed that the
membrane topologies and three-dimensional
structures of
most of the related proteins will be similar, and ultimate
crystallization
of one member of this group of proteins
may provide structural predictions for all.
The aquaporin CHIP structural gene. CHIP transcripts
of 3.1 kb are expressed in most tissues (13,34,45). Lung
and small intestine also contain transcripts of 4.2 kb,
whereas skeletal muscle contains only a 1.4-kb transcript
(39). Because of the possibility of multiple CHIP genes,
cDNAs were isolated from human kidney libraries by
nonamplification
methods, and the nucleotide sequences
were identical to that of erythroid CHIP.
The structural CHIP gene was isolated from a human
genomic library (39). The 17-kb DNA contains four exons
and a TATA consensus sequence in the 5’ untranslated
region. The 1st exon corresponds to the NHz-terminal
half of the molecule and is followed by a 9.6-kb intron.
The COOH-terminal
half of the molecule is encoded by
the 2nd-4th exons, which are separated by small introns.
Only a single CHIP gene was identified
by genomic
Southern analyses (39). The locations of the intron-exon
boundaries are identical to the MIP26 gene (44), confirming evolutionary divergence of these genes. The human
CHIP locus is on chromosome 7 at p14 (39). This location
in the human genome is not coincident with any likely
disease phenotypes. It remains to be established whether
homozygous expression of mutant forms of CHIP would
be lethal to embryos or whether tissue-specific mutations
or acquired defects may exist.
The existence of a single CHIP gene (39) and the published sequences of murine CHIP cDNAs (13,34) conflict
with the sequence reported for a putative rat kidney homologue referred to as “CHIP28k,”
which apparently corresponds to the rat CHIP homologue with human CHIP
cDNA at the 5’ end of the coding sequence (72).
F467
REVIEW
I /-‘“‘“-CHIP+Hg
YWater+Hg
-Water
1.0
rl...l...l...l...l...l
0
60
120
160
240
300
Time (set)
Fig. 3. Demonstration of water channel activity in X. Zueuis oocytes
expressing CHIP. A: swelling and rupture of oocyte injected with 10 ng
in vitro transcribed CHIP RNA (top) or control oocyte (bottom) at
indicated intervals after transfer from 200 to 70 mosmol/lsgHeO buffer.
B: oocyte volumes during swelling experiment. CHIP28, oocyte injected
with 1 ng CHIP RNA, water, control oocyte; Hg, 5-min incubation in
0.3 mM HgCl,; Hg + ME, 5-min incubation in 0.3 mM HgCle followed
by 15-min incubation in 5 mM j3-mercaptoethanol. [Reprinted from
Preston et al. (46,471.l
F468
EDITORIAL
Structure of Water Pathway
Site-directed mutagenesis. The X. laevis oocyte system
has been employed to probe the functional importance of
domains within the CHIP structure by site-directed mutagenesis. The known mercury sensitivity was evaluated
by individually
replacing each of the four cysteine residues with a serine, a residue with similar structure (47).
Each of the mutagenized proteins functioned like the
wild-type CHIP in the oocyte swelling assay (Fig. 4A).
When the assay was performed after incubation in HgCl,,
only C189S was insensitive to the inhibitor at all concentrations tested (0.1-3 mM). When other amino acid substitutions were made at residue 189, the Pf of the oocytes
coincided inversely with the size of the residue (Fig. 4B).
Replacement of residue 189 with an alanine (smaller than
cysteine) conferred wild-type Pf, whereas replacement
with a valine (larger than cysteine) caused a reduction in
Pf. Replacement of residue 189 with still larger residues,
methionine,
tryptophan,
or tyrosine, reduced the Pf to
the level of control oocytes. It was initially thought that
the greater size of the amino acids substituted at residue
189 may have produced simple occlusion of the channel
pore, but it was demonstrated that the functionally inactive CHIP mutants had incompletely formed high-mannose glycans, which may have led to altered membrane
trafficking (47).
It is unknown whether the individual subunits within
the CHIP tetramer each have an internal pore, analogous
to the bacterial porins, or whether the tetrameric complex
forms a single channel, analogous to the Shaker potassium channels (32). The radiation inactivation
data of
van Hoek et al. (5859) predicted that each CHIP subunit
will function as an individual water pore. Coexpression of
any of several distinct functionally inactive mutants with
wild-type CHIP demonstrated no diminution
of the Pfi
This lack of “squelching” suggested a lack of functional
cooperativity
among the subunits when coexisting in
mixed tetramers. The mercury-insensitive
mutant C 189s
was coexpressed in oocytes along with wild-type CHIP,
and their activities approached the anticipated sum of
A
Water
CHIP28
it-+
C87S
c102s
Cl 52s
I
I
Cl 89s
Water
w
CHIP28
C87S
m
m
c102s
c152s
m
e
C189S
I’
n + Mm2
0
50
Osmotic
100
150
Water Permeability
200
250
(Pf , x 10M4 cmkec)
B
1""1""1""1""l""l""l""
Water
CHIP28
C87S
C87M
C87W
C87Y
mb
.*..**.... '... ::. :.. ..:.: .*.. . :. . ..-. . . . . :. ...'*. ..: ,... :.. ..:... :. ..:.. . . :; .. :. . :. ..: . . : . ..:.. . :.. . . :.. : :.. ..: :.: , ..:.... :.:.:.. ..:. . :. ..:....:.. .. : . . . . : ,.... . . . :
. . . . . . :..:.. . . . . . . . . . . ..'J.'..'..'.......
.. ..
. . .:..: . ,.. . . . . . . . . :: . . . . . . . . . . . . . . . . . :..: . . . . . . . .:.::..:..:........
............ ............ .. . ................. .............................
..................... ......................
.... ::
................ ............
. : : .. .
... . .......
..
. .............
...............................
:..‘...:.
.....
. .
..
..:.
......................................................................................................................................................
C18gA
‘~
C18gS
~
0
......... ....
..:..:
:.
. ...... .. ...... .... ............
..:.
....
. ....
.....
..:.
.....
.
i
,
50
100
150
200
250
300
350
P, (x IO -4 cmkec)
Fig. 4. Measurement of the coefficient of osmotic water permeability
(P,) of oocytes expressing cysteine mutants of CHIP. A: bars are means
(error lines are &SD) from routine swelling assay (open bars) and swelling measured after incubation in 1 mM HgClz (solid bars). B: Pf determinations for indicated cysteine mutants of CHIP determined by routine swelling assay. [Reprinted from Preston et al. (47).]
these proteins when measured without mercury, but the
activity was reduced to the level of the mercury-insensitive mutant alone when measured after incubation in 3
mM HgC12 (47). These observations support the hypothesis that individual subunits each bear a water pore, raising the question of why the protein exists exclusively as a
tetrameric oligomer.
Reconstitution of purified aquuporin CHIP into proteoliposomes. Although the results were clear, the oocyte
studies could not rule out the existence of other water
channel components that might exist within the native
oocyte. Therefore highly purified CHIP protein was reconstituted into proteoliposomes
for direct demonstration of osmotic water permeability
(68). This was made
possible by the remarkably simple purification
system
that had already been developed (52), the abundance of
CHIP in human RBC that are available in liter quantities,
and the development of efficient reconstitution
systems
(5). Pure CHIP protein in octyl glucoside and purified
Escherichia coli phospholipid
(70% phosphatidylethanolamine, 15% phosphatidylglycerol,
15% cardiolipin) were
rapidlv diluted in the presence of carboxvfluorescein. The
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
x 10s4 cm/s), which was approximately
eightfold above
the pf of control oocytes. When measured at lO”C, the pf
of the CHIP oocytes was only slightly diminished,
whereas the Pf of control oocytes was substantially
reduced; their ratio corresponded to a >30-fold increase in
&. The Arrhenius activation energies of CHIP oocytes
are consistent with channel-mediated
water transport
(E, = 3 kcal/mol), whereas that of the control oocytes are
consistent with diffusional water permeability
(E, > 10
kcal/mol). Also, similar to native water channels, CHIPmediated Pf was reversibly inhibited by HgCIZ (Fig. 3B).
Increased osmotic water permeability was not observed in
oocytes microinjected with CHIP antisense RNA or with
RNA for the GABA channel. Several of these studies
were subsequently confirmed by other investigators (72).
CHIP permeability
is selective for water (46). When
placed in an osmotic gradient, oocytes expressing CHIP
absorbed 3HZ0, but failed to absorb urea or glucose (Preston, unpublished observations). Moreover, two-electrode
voltage-clamp experiments failed to detect an increase in
conductance in the CHIP oocytes until the point of membrane rupture during swelling experiments (46).
REVIEW
EDITORIAL
spontaneously forming proteoliposomes were purified by
ultracentrifugation
(Fig. 5A). Control liposomes were
prepared identically, except CHIP protein was omitted.
glyCHIP28
CHIP28
0.8
I
I
0.1
0.0
I
0.2
0.3
I
I
1
0.4
0.5
0.6
0.8
0.02
0.01
Time,
set
0.8
I
0.0
I
0.1
I
!
I
0.2
0.3
0.4
Time,
set
,
0.5
Measurements
of osmotic water permeability
of the
CHIP proteoliposomes
and control liposomes were performed by abruptly increasing the osmolality by 20%
while measuring the self-quenching of carboxyfluorescein
by stop-flow fluorometry
(68). CHIP proteoliposomes
and control liposomes both behaved as perfect osmometers, but the tracings demonstrated that equilibrium was
reached up to 50-fold faster by the CHIP proteoliposomes
(Fig. 5, B and C). The size of the proteoliposomes
(radius
= 70 nm) and average number of CHIP subunits (220
subunits/proteoliposome)
permitted
calculation of the
unit water permeability
(Table 2). Extrapolation
to the
CHIP-mediated
water permeability
of an individual
RBC, 0.017 cm/s, was close to the known RBC water
permeability
(20). Determination
of Pf over a range of
temperatures and in the presence of mercurial inhibitors
provided estimated Arrhenius activation energies that indicated water transport in the CHIP proteoliposomes was
channel mediated, whereas that of the control liposomes
represented diffusional water permeability
(Table 2).
The reconstitutions
revealed several other features of
CHIP. Specialized lipids are not required for channel
function (68). The CHIP-mediated
water permeability
was highly selective for water and did not transport protons or urea. The actual pore within the protein must be
both extremely narrow and contain structural mechanisms to prevent movement of protons that would otherwise be expected to jump along a continuous column of
water in the form of hydronium ions (HsO+). Other investigators have qualitatively
confirmed these observations by light scattering measurements with membrane
vesicles enriched in CHIP by N-lauroylsarcosine
extraction and with proteoliposomes
reconstituted with partially purified CHIP (60). Together, the osmotic water
permeability
of the CHIP proteoliposomes
studies reported by all investigators indicate that CHIP behaves
identically to the water channels in native membranes.
Other Aquaporins
Since the discovery that CHIP conferred oocytes with
increased osmotic water permeability,
at least two other
sequence-related proteins were also found to increase Pf
(Table 3). These proteins are a subset of the large family
of membrane proteins related to MIP26 found in diverse
organisms (43). To facilitate recognition of their function
as water-selective pores and promote better communication among scientists, the name “aquaporins” was selected (4). Preliminary
evidence from other laboratories
suggests that several other sequence-related proteins are
also functional water channels.
Collecting duct aquuporin (WCH-CD). A cDNA most
likely corresponding to the vasopressin-regulated
water
channel was isolated by Fushimi et al. (23) from a rat
kidney library by PCR amplification
using degenerate
oligonucleotide
primers corresponding to the cDNAs of
1.0
0.00
F469
I
0.6
Fig. 5. Osmotic water permeability of CHIP reconstituted into proteoliposomes. A: silver-stained sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) results showing 26kDa (CHIP28)
and glycosylated subunits (glyCHIP28) of highly purified CHIP after
reconstitution into proteoliposomes. Shrinking of control liposomes (B)
or CHIP proteoliposomes (C) after stop-flow transfer to buffer with
tonicity increased by 20%. [Reprinted in part with permission from
Zeidel et al. (68); 1992 Am. Chem. Sot.]
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
phospholipid
REVIEW
F470
EDITORIAL
Table 2. Physical properties
of reconstituted aquuporin CHIP
Pf9
X
lOA
Em
cm/s
kcalfmol
3.1
CHIP proteoliposomes
540
16
Control liposomes
97
Values were derived from Zeidel et al. (68). Coefficient of osmotic
water permeability (pf) was determined at 37°C. E,, Arrhenius activation energy. Unit permeability per subunit was 11.7 x lo-l4 ml/s (3.9 x
log Hz0 molecules/s).
Table 3. Aquaporins
Name
Mammalian
Known
Location(s)
Renal
(proximal tubule,
descending thin limb)
Brain
(choroid plexus)
Eye
(ciliary and lens epithelium,
cornea1 endothelium)
Gastrointestinal tract
(hepatic bile duct,
gall bladder epithelium)
Eccrine sweat glands
Nonfenestrated endothelium
Lymphatic endothelium
WCH-CD
Mammalian
Renal collecting duct
y-TIP
A. thulianu
Stem and leaf tonoplast
From Agre et al. (4), Nielsen et al. (40, 4l), Fushimi et al. (23), and
Maurel et al. (38).
CHIP and MIP26. Expression of the protein conferred
oocytes with increased osmotic water permeability,
and
the protein was localized to the apical membrane of renal
collecting duct cells (hence, “WCHCD”).
Moreover,
thirsting of rats enhanced expression of the WCH-CD
transcript (50).
Tonoplast intrinsic protein aquuporin (T-TIP). At least
eight proteins related to MIP26 have been identified in
plants, including four isoforms in Arabidopsis thuliana.
Two of these proteins were discovered in vacuolar membranes (tonoplast) by Hofte et al. (30) and are referred to
as “TIP.” The distinct distribution within roots, seeds, or
other plant tissues suggests that individual
TIPS may
have specialized physiological functions. When expressed
in oocytes, y-TIP conferred markedly increased osmotic
water permeability
without increased conductance to
protons (38). It is thought that y-TIP permits rapid exchange of water between the vacuole and cytoplasm in the
vegetative body of the plant.
Function of other MIP humologues. All proteins in the
MIP26 family are not aquaporins, since some do not selectively transport water. Although the native function of
MIP26 is not agreed upon, the protein conducts ions
when reconstituted into planar bilayers (18). GlpF facilitates glycerol uptake by E. coZi (29), and FPS may transport fermentable sugars into yeast (1). It is not yet established what physiological functions are provided by most
other homologues of MIP26. Demonstration
of function
may be difficult for some homologues, since water transport in the oocyte system requires a high level of protein
expression, folding of the protein in the native conformation, and targeting of the protein to the appropriate mem-
brane. Moreover, MIP26 does not confer significantly
increased Pf in the oocyte system (Preston, unpublished
observations). It therefore cannot presently be resolved
whether the failure of certain homologues to confer increased Pf results from problems in the expression system
or whether the proteins do indeed have different functions.
Phylogenetic comparisons of the deduced amino acid
sequences (Fig. 6) indicated that CHIP is intermediate
between the plant proteins and WCH-CD, which is restricted to the renal collecting duct (23). Moreover, CHIP
is widely distributed in many mammalian
tissues (see
below). These observations suggest that CHIP may be a
functionally less specialized water channel that may be
evolutionarily
more primitive.
Distribution
and Expression
Aquuporin CHIP in kidney. The original report describing CHIP in RBC also reported the identification
and isolation of a highly related protein from kidney
where it appeared in proximal tubules and descending
thin limbs (16). Renal CHIP was subsequently shown to
be the same gene product as erythroid CHIP (39). Moreover, injection of oocytes with mRNA from kidney is
known to increase the Pf (71). The increased Pf obtained
from cortical or medullary mRNA was neutralized when
mixed with CHIP antisense oligonucleotides;
the increased Pf from papillary mRNA was not neutralized (17,
72). These studies indicate that CHIP is a major water
channel in renal cortex and medulla but is not the predominant water channel in papilla.
CHIP was immunolocalized
to water-permeable
segments of rat nephron (Fig. 7). A polyclonal antibody was
raised in rabbits to a synthetic peptide corresponding to
the NH2 terminus, and another was raised to the whole
CHIP protein, although reactivity was restricted to the
4-kDa cytoplasmic domain (52). These antibodies were
affinity purified and used in detailed immunohistochemical and immunoelectron
microscopic
localizations
of
CHIP (41). Except for the initial portion of the Sl segment, abundant immunostaining
of CHIP was observed
over the apical brush border and basolateral membranes
of renal proximal convoluted and straight tubules (Figs. 7
and 8). The CHIP immunostaining
appeared less intense
over the lateral and basal membranes and more intense
over the apical brush border where CHIP comprises
-1
MIP26
many tissues
lens fiber cells
-
Mammalian
-
Drosophila
-
Plant
I
P
pcwq
BIB
-
.
L
renal collecting
neuronal
duct
precursors
IpcjFTiq
tonoplast
TUR
tonoplast,
NOD26
root nodules
water-stress
-
Yeast
-
E. coli
Fig. 6. Phylogenetic analysis of aquaporins (outlined with box) and
other proteins related to MIP26. BIB, big brain; TUR, turgor-responsive element; NOD26, soybean nodule protein of 26 kDa; FPSl, fdpl
suppressor; GLPF, glycerol facilitator. [Reprinted from Moon et al.
(39M
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
CHIP28
Species
REVIEW
EDITORIAL
Distal
Convoluted
Tubule
\
Glomerulus
REVIEW
F471
Connecting
Tubule
/
I
Cortex
Outer
stripe
Thick
Ascending
Limb
Inner
Stripe
Outer
Medulla
DTL2
Descending
Thin Limb
\
DTL3
Inner
Medulla
Collecting
Ducts
/
Fig. 7. Diagram of an individual long-looped nephron (A)
and summary of osmotic water permeabilities of indicated
segments in rodent nephrons from multiple physiological
studies (B). Sites of immunolocalization
of CHIP are
indicated. ADH, antidiuretic
hormone. [Reprinted
from Nielsen et al. (41) by copyright permission from the
Rockefeller Univ. Press.]
J
Final Urine
B
Osmotic
Water
Permeability
Pf, pm&c
10'
Proximal
Tubule
Descending
Thin
Ascending
Thick
Distal
PCT s1 ts2
PST S2,S3
Thin
Ascending
Limb
Limb
DTLl
-1-m
DTL2
2
DTL3
j
LOCATION
OF
CHIP28
ATL u
MTAL
Limb
1
CTAL
fj
fj
Tubule
DCT
Connecting
Tubule
CNT 1
-
Duct
1
3
5
Convoluted
Collecting
10”
CCDF
oMCD
liiib
IMCD
r
q
n
nearly 4% of the total membrane protein. The functional
significance of this apparent imbalance is uncertain.
Heavy, uninterrupted
CHIP immunostaining
was also
observed over apical and basolateral membranes of descending thin limbs of both short- and long-loops of
Henle (Figs. 7 and 8). Very little CHIP immunolabeling
was observed over intracellular
vesicles. The ascending
thin limbs, thick ascending limbs, and di stal tubules are
known to exhibit negligible transepithelial
water permeability and these structures were also not immunolabeled.
These studies were supported by those of Sabolic et al.
(48)) who performed immunofluorescence
localizations
with whole antiserum raised to CHIP protein. Although
of lower resolution, in situ hybridizations
with CHIP
antisense probes were interpreted similarly (9, 28, 72).
Paracellular water permeability
may also occur (20, 55,
65), but the abundance, distribution,
and unit water permeability of CHIP strongly suggest that it is the major or
activated water channel of renal
exclusive constitutively
proximal tubules and thin descending limbs of Henle’s
loop. Therefore, CHIP provides a molecular explanation
Basal
ADH
for transcellular water flow that is known to occur at
these locations.
Cortical, medullary, and papillary collecting ducts have
low basal water permeabilities,
but stimulation
with vasopressin produces a large increase in water permeability
(33). Collecting tubule epithelia from control and thirsted
rats were not immunolabeled
with affinity-purified
antibodies specific for CHIP (Fig. 8) (41). The cDNA encoding WCH-CD, the probable vasopressin-regulated
water
channel, was recently isolated and is 42% identical to
CHIP (23). The immunogenic
COOH-terminal
domains
of these proteins share only 6 of 36 residues (23, 45),
explaining the lack of cross-reactivity. Taken together,
these studies indicate that CHIP is neither the vasopressin-regulated water channel nor the basolateral water
channel of renal collecting ducts.
Aquuporin CHIP in other tissues. Water-permeable
epithelia exist in several tissues in addition to kidney.
Although kidneys function to conserve water by reabsorption of glomerular fluid, several other tissues function
to secrete water during generation of cerebrospinal fluid,
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
Ascending
Thin Limb
F472
EDITORIAL
REVIEW
..
b
,
’
:
e
Fig. 8. Immunohistochemical
and immunoelectron
microscopic
localization
of CHIP
in different
segments
of rat
nephron.
Sections were obtained
from deep cortex and outer stripe of outer medulla (a, a’, and b), inner stripe of outer
medulla (c and d), and inner medulla (e and f). Reactions
with anti-CHIP
and histochemical
staining
with peroxidase
or immunogold
were described
(41). a: S3 proximal
tubules exhibited
heavy immunolabeling
of apical and basolateral
membranes
(arrows),
as observed
in cross-sectioned
(a) and tangentionally
sectioned
tubules (a’); magnification,
x625.
b: Immunogold
labeling of cross-sectioned
microvilli
was confined
to cytoplasmic
leaflet of plasma membrane;
magnification,
~45,006.
c: Apical and basolateral
parts of descending
thin limbs were heavily
labeled (arrows),
whereas thick
ascending limbs (T) and collecting
ducts were unlabeled,
magnification,
x625. d: High-density
immunogold
labeling was
observed
over plasma membranes
of highly interdigitated
type II epithelium
of descending
thin limbs (arrows,
tight
junctions;
BM, basement
membrane);
magnification,
~45,000.
e: Inner medullary
descending
thin limbs (arrows)
were
labeled, but collecting
ducts (C) were unlabeled;
magnification,
x530. f: Apical and basolateral
membranes
(arrows)
of
descending
thin limbs from initial part of inner medulla were heavily
labeled; magnification,
~31.000.
[Modified
from
Nielsen et al. (41) by copyright
permission
from the Rockefeller
Univ. Press.]
aqueous humor, saliva, sweat, bile, semen, and pulmonary
secretions. The existence of water channels has been predieted, although the physiological
studies are less well
developed, in tissues other than kidney. It is not known
whether a single set of channels could explain these di-
verse processes, since water moves in one direction during
absorption and in the opposite direction during secretion.
The tandem repeats within CHIP (Fig. 2, middle) suggest
that it may function in either direction (46). Northern
analyses demonstrated that it is widely expressed (13,34,
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
.”
__
EDITORIAL
no fetal water channel isoform, since the water permeability of fetal RBC is low and the Arrhenius activation
energy is high, suggesting that only diffusional water permeability is present (52a). Rat kidneys express CHIP
protein in the first few weeks of postnatal life, coincident
with the appearance of the ability to concentrate. The
coincident expression of CHIP in RBC and kidneys suggests that the protein plays an adaptive role in RBC,
presumably by providing them with a mechanism to rapidly rehydrate after traversing the hyperosmotic
renal
medulla (37). The total lack of CHIP in rat RBC before
birth demonstrates that the protein is not needed for
oxygen transport or circulation throughout the vascular
bed (9, 52a).
Two other developmental patterns of expression were
also noted (9). CHIP mRNA is constitutively
highly expressed in fetal and adult rat choroid plexus, whereas the
transcript was abundant in certain other fetal tissues
where expression was downregulated after birth. The
mesenchyme surrounding developing bone and the developing endocardium have large amounts of CHIP mRNA
before birth (Fig. 9), but not in mature rats. Likewise,
fetal cornea1 endothelia have abundant CHIP transcripts
at the point in development when the tissue becomes
transparent, a desiccation process thought to be mediated
by CHIP. Adult corneas have lower levels of CHIP
mRNA (9) and CHIP protein (40). This transient expression is reminiscent of the expression noted by Lanahan et
al. (34), who identified CHIP among delayed early response genes that were transiently expressed by fibroblasts after stimulation
with selected growth factors, but
the physiological
importance
of transiently
expressing
CHIP protein remains unknown. All of the distributional
and developmental
studies underscore the hypothesis
that CHIP plays a major role in transcellular water movement at numerous sites within the body.
Future
Directions
Three aquaporins have been rigorously defined so far,
and it is likely that many more remain to be discovered.
Use of degenerate PCR oligonucleotide
primers corresponding to CHIP and MIP26 nucleotide sequences is an
approach that has already led to the isolation of the
cDNA for WCH-CD from the renal collecting duct (23).
This approach may yield other homologues, but demonstration of the function of newly identified cDNAs will be
essential, since it is likely that only a subset may be water
channels.
The unusual patterns of expression during development and after stimulation
with selected growth factors
remain to be explained at a molecular level. It is likely
that specific cis regulatory sequences will be identified
that lead to the expression of certain aquaporins. The
transcriptional
factors responsible for these events remain to be identified.
The structure of CHIP and related proteins is still
poorly understood. Water rapidly permeates the channel,
but protons are not conducted. This feature is consistent
with the physiological
need for the kidney to conserve
water while excreting acid, but the structural mechanisms
by which this is accomplished are entirely unknown. Creation of a mutant CHIP that is permeable to protons
could permit detailed electrophysiological
studies of the
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
45). Therefore CHIP may be a general water channel.
The affinity-purified
antibody specific for the COOHterminal cytoplasmic domain of CHIP was used to screen
several tissues (40). CHIP was detected by immunoblot
in
most tissues. Therefore immunohistochemical
and immunoelectron
microscopic
analyses of several tissues
were undertaken (Table 3). Only one structure in the
central nervous system was found to contain immunoreactive CHIP; the apical microvillar surface of the choroid
plexus contains abundant CHIP, implying a role in secretion of the aqueous component of cerebrospinal fluid.
Several structures in the anterior chamber of the eye
contained immunoreactive
CHIP in both apical and basolateral membranes. The ciliary epithelium
is rich in
CHIP and is the site of aqueous humor secretion. The
cornea1 endothelium is known to contain functional water
channels and also contains abundant CHIP protein.
CHIP is present in lens epithelium,
whereas lens fiber
cells lack CHIP but are rich in MIP26. Eccrine sweat
gland epithelia are also abundant in CHIP (G. Anhalt and
P. Agre, unpublished observations). The endothelia of
nonfenestrated
capillaries and venules at many sites
throughout the body are also rich in CHIP, including the
endothelium of descending vasa recta (S. Nielsen and A.
B. Maunsbach, unpublished observations). This suggests
that movement of water through these water channels
will balance the hydrostatic and oncotic pressures which
control interstitial
fluid and vascular volumes. In contrast, CHIP was not observed in fenestrated capillaries.
CHIP has an interesting distribution
within the gastrointestinal
tract and lung (40). CHIP is present in apical and basolateral membranes of epithelium
in hepatic
bile ductules and lining the gall bladder neck. There it
may play a role in secretion and concentration
of bile.
Although the existence of CHIP RNA was reported in
colonic crypts by in situ hybridization
(28), immunoreactive CHIP does not exist in intestinal mucosa (40). The
lack of CHIP in gastrointestinal
epithelium explains the
characteristic barrier to water absorption of these structures (57). It is probable that most intestinal water uptake is paracellular or diffusionally mediated (55). CHIP
in lymphatics and submucosal capillaries may promote
rapid movement of absorbed water into lymph and the
vascular bed. The presence of abundant CHIP in alveolar
capillary endothelia may be needed for humidification
of
airways; however, this may also explain the rapid and
direct uptake of water into the vascular space during a
freshwater drowning. The lack of CHIP in certain waterpermeable epithelia such as mammary, lacrimal, and salivary glands, as well as the initial segment (Sl) renal
proximal tubules, suggests that other water channels may
exist at those locations.
Aquaporin
CHIP during development. During in situ
hybridization
screening of diverse tissues, it was surprising to find a divergent pattern of expression in development (9). Fetal rats have no detectable CHIP transcripts
in liver or spleen, the fetal erythroid tissues (Fig. 9). Fetal
rat RBC did not contain detectable CHIP by immunoblot, although the presence of other known membrane
proteins was clear. Expression of CHIP is also very low in
fetal kidney (Fig. 9). These patterns were found to change
rapidly after birth when fetal RBC were replaced by RBC
containing immunoreactive
CHIP. There appears to be
F473
REVIEW
F474
EDITORIAL
REVIEW
channel that are not feasible with the oocyte swelling
assay. The existing evidence suggests that the aquaporins
have a novel structure with tandem repeats oriented 180”
to each other. The three-dimensional
structure remains
to be established, as does the structure of the water pathway within the CHIP protein. Such information
could
provide structural information
needed for the design of
inhibitors which may be tested in either the oocyte or
proteoliposome
water permeability
assays.
Several clinical settings exist in which therapeutic benefits may potentially
be achieved by modulation
of the
CHIP. I ) Inhibitors of proximal nephron water reabsorption or choroid plexus secretion may be useful in congestive heart failure, renal failure, or acute hydrocephalus. 2)
Topical inhibitors of aqueous humor secretion could be
helpful in managing glaucoma. 3) Inducers of eccrine
sweat gland water flow may permit enhanced heat exchange in threatening environmental
conditions. 4) Ge-
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
Fig. 9. Developmental tissue distribution
of CHIP gene expression. Sagittal section
from rat fetus 2 days before birth (E20).
A: stained with hematoxyl and eosin. B:
visualized by autoradiographic exposure
of in situ hybridizations with CHIP RNA
probe. Tissues symbols are as follows: co,
cornea; cp, choroid plexus; H, heart; K,
kidney; Li, liver; Lu, lung; P, forepaw; S,
stomach; and sp, spleen. Arrowheads,
periostium
surrounding
bone anlage.
Open arrow, aortic valve where leaflets
are intensely positive for CHIP mRNA.
C-F: spleens from mature, normal mice
(C-D) and anemic littermates (E-F). C
and E: stained tissue sections. D and F:
corresponding film autoradiographs that
reveal CHIP mRNA concentrated over
red pulp but not lymphoid nodules (arrows). G-H: CHIP gene expression in rat
kidneys at postnatal days P5 and P40.
Tissue symbols are as follows: cx, cortex;
mr, medullary ray; OS, outer stripe of
outer medulla. Bar = 2.7 mm in A and B,
0.8 mm in C-F, and 2 mm in G-H. [Reprinted from Bondy et al. (9).].
EDITORIAL
netic reconstitution
of damaged lacrimal . or salivary
glands may restore teari .ng or salivation in pati .ents suffering from Sjogren’s syndrome or after receiving local
irradiation.
Continued aquaporin research may provide
improved understanding and treatment of certain human
diseases in addition to contributing
molecular insight
into the fundamental biological phenomenon of osmotic
water movement.
L. V., S. Hohmann,
and J. M. Thevelein.
F. K.
Zimmermann,
A. W.
H.
A yeast homologue of the bovine
lens fibre MIP gene family complements the growth defect of a
Saccharomyces cereuisiae mutant on fermentable sugars but not its
defect in glucose-induced RAS-mediated CAMP signalling. ElMBO
J. 10: 2095-2104,
1991.
2. Aerts,
T., J.-Z.
Xia,
H.
Clauwaert.
Hydrodynamic
Slegers,
J.
de
Block,
and
J.
characterization of the major intrinsic protein from the bovine lens fiber membranes. J. Biol. Chem.
265: 8675-8680,
1990.
3. Agre,
4.
5.
6.
8.
P.,
A.
M.
Saboori,
A.
Asimos,
and
B.
L.
Smith.
Purification and partial characterization of the M, 30,000 integral
membrane protein associated with the erythrocyte Rh(D) antigen.
J. Biol. Chem. 262: 17497-17503, 1987.
Agre,
P., S. Sasaki,
and M. J. Chrispeels.
Aquaporins: a
family of water channel proteins. Am. J. Physiol. 265 (Renal Fluid
Electrolyte
Physiol.
34): F461, 1993.
Ambudkar,
S. V., and P. C. Maloney.
Anion exchange in
bacteria: reconstitution of phosphate: hexose 6-phosphate antiport from Streptococcus
lactis. Methods
Enzymol.
125: 558-563,
1986.
Andreoli,
T. E., and J. A. Schafer.
Effective luminal hypotonicity: the driving force for isotonic proximal tubular fluid absorption. Am. J. Physiol. 236 (Renal Fluid Electrolyte
Physiol.
5):
F89-F96, 1979.
Benga,
G., 0. Popescu,
and V. I. Pop.
Water exchange
through erythrocyte membranes. Cell Biot. Int. Rep. 7: 807-818,
1983.
9. Bondy,
Agre.
C.,
E. Chin,
B.
L.
Smith,
G. M.
Preston,
and
P.
Developmental gene expression and tissue distribution of
aquaporin-CHIP. Proc. Natl. Acud. Sci. USA 90: 4500-4504,1993.
10. Bourguet,
J., J. Chevalier,
and J. S. Hugon.
Alterations in
membrane-associated
particle distribution
during antidiuretic
challenge in frog urinary bladder epithelium. Biophys. J. 16: 627639, 1976.
11. Brahm,
J. Diffusional water permeability of human erythrocytes
and their ghosts. J. Gen. Physiol. 79: 791-819, 1983.
12. Brown,
D., and L. Orci.
Vasopressin stimulates formation of
coated pits in rat kidney collecting ducts. Nature Lond. 302: 253255, 1983.
13. Deen,
membrane protein from erythrocytes and renal
263: 15634-15642,
1988.
18.
Frindt,
G. M. Preston,
M. Snezana,
P.
and E. Windhager.
Expression of multiple water channel activities in Xenopus oocytes injected with
mRNA from rat kidney. J. Gen. Physiol. 101: 827-841, 1993.
Ehring,
G. R., G. Zampighi,
J. Horwitz,
D. Bok,
and J.
Hall.
Properties of channels reconstituted from the major intrinsic protein of lens fiber membranes. J. Gen. Physiol. 96: 631-664,
1990.
19. Fettiplace,
R., and D. A. Haydon.
Water permeability
membranes. Physiol. Reu. 60: 510-550,
1980.
20. Finkelstein,
and Plasma
A. Water
Membranes:
P. M. T., J. A. Dempster,
B. Wieringa,
and
C. H. Van
OS. Isolation of a cDNA for rat CHIP28 water channel: high
mRNA expression in kidney cortex and inner medulla. Biochem.
Biophys.
Res. Commun.
188: 1267-1273,
1992.
14. Dempster,
J. A., A. N. Van Hoek,
and C. H. Van OS. The
quest for water channels. News Physiol. Sci. 7: 172-176,
1992.
15. Dempster,
J. A., A. N. Van Hoek,
M. D. de Jong,
and C. H.
Van OS. Glucose transporters do not serve as water channels in
renal and intestinal epithelia. Pfluegers Arch. 419: 249-255, 1991.
16. Denker,
B. M., B. L. Smith,
F. P. Kuhajda,
and P. Agre.
Identification, purification, and characterization
of a novel M,
Movement
Theory
Through
Lipid Bilayers,
and Reality.
New York:
of lipid
Pores,
Wiley,
1987.
21. Fischbarg,
Rogozinski,
J., K. Kuang,
S. C. Silverstein,
J.
Hirsch,
S.
and J. Loike.
Lecuona,
L.
22. Fischbarg,
Silverstein,
J., K. Kuang,
J. C. Vera,
S. Arant,
S. C.
J. Loike,
and 0. M. Rosen.
Glucose transporters
channels. Proc. Nutl. Acad. Sci. USA 87: 3244-3247,
Evidence that
the glucose transporter serves as a water channel in 5774 macrophages. Proc. Nutl. Acud. Sci. USA 86: 8397-8401, 1989.
serve as water
1990.
23. Fushimi,
K.,
and S. Sasaki.
REFERENCES
1. Aelst,
Jans,
28,000 integral
~” tubules. J. Biol. Chem.
Echevarria,
M., G.
“‘* Agre,
J. Fischbarg,
S. Uchida,
Y. Hara,
Y. Hirata,
F. Marumo,
Cloning and expression of apical membrane water
channel of rat kidney collecting tubule. Nature Lond. 361: 549552, 1993.
24. Gorin,
Horwitz.
M.
B.,
S. B. Yancey,
J. Cline,
J. P. Revel,
and
J.
The major intrinsic protein (MIP) of the bovine lens
fiber membrane. Cell 39: 49-59, 1984.
25. Guerrero,
F. D., J. T. Jones,
and J. E. Mullet.
Turgorresponsive gene transcription and RNA levels increase rapidly
when pea shoots are wilted. Plant Mol. Biol. 15: 11-26, 1990.
26. Harris,
H. W., Jr., C. Hosselet,
L. Guay-Woodford,
L. Zeidel.
Purification and partial characterization
and
M.
of candidate
antidiuretic hormone water channel proteins of M, 55,000 and
53,000 from toad urinary bladder. J. Biol. Chem. 267: 2211522121, 1992.
26a.Harris,
H. W., K. Strange,
and M. L. Zeidel.
Current understanding of the cell biology and molecular structure of the antidiuretic hormone-stimulated
water transport pathway. J. Clin. Invest. 88: l-8, 1991.
27. Hasegawa,
Lingappa,
H.,
and
channel formed
1992.
28. Hasegawa,
W. Skach,
0. Baker,
M. C. Calayag,
V.
A. S. Verkman.
A multifunctional
aqueous
by CFTR. Science Wash. DC 258: 1477-1479,
H., R. Zhang,
A. Dohrman,
and
A. S. Verkman.
Tissue-specific expression of mRNA encoding rat kidney water
channel CHIP28k by in situ hybridization. Am. J. Physiol. 264
(Cell Physiol.
33): c237-C245,
1993.
29. Heller,
K. B., E. C. C. Lin,
and T. H. Wilson.
Substrate
specificity and transport properties of the glycerol facilitator of
Escherichia
coli. J. Bacterial.
144: 274-278,
1980.
30. Hofte,
Herman,
H.,
L. Hubbard,
J.
and M. J. Chrispeels.
Reizer,
D.
Ludevid,
E.
M.
Vegetative and seed-specific
forms of tonoplast intrinsic protein in the vacuolar membrane of
Arabidopsis
thaliana.
Plant Physiol.
99: 561-570,
1992.
31. Horwitz,
J., and D. Bok.
Conformational properties of the
main intrinsic polypeptide (MIP26) isolated from lens plasma
membranes. Biochemistry
26: 8092-8098,
1987.
32. Jan, L. Y., and Y. N. Jan. Voltage-sensitive ion channels. Cell
56: 13-25, 1989.
33. Knepper,
M. A., and F. C. Rector,
Jr. Urinary concentration
and dilution. In: The Kidney (4th ed.), edited by B. M. Brenner
and F. C. Rector, Jr. Philadelphia: Saunders, 1991, p. 445-482.
34. Lanahan,
A., J. B. Williams,
L. K. Sanders,
and
D.
Nathans.
Growth factor-induced delayed early response genes.
Mol. Cell. Biol. 12: 3919-3929, 1992.
35. Macey,
R. I. Transport of water and urea in red blood cells. Am.
J. Physiol.
246 (Cell Physiol.
15): C195-C203, 1984.
36. Macey,
R. I., and R. E. L. Farmer.
Inhibition of water and
solute permeability in human red cells. Biochim.
Biophys.
Acta
211: 104-106,
1970.
37. Macey,
R. I., and L. W. Yousef.
Osmotic stability of red cells
in renal circulation requires rapid urea transport. Am. J. Physiol.
254 (Cell Physiol.
23): C669-C674,
1988.
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
We are most grateful to John C. Parker, who first predicted that
CHIP would be the red cell water channel. We thank Mark A. Knepper
for suggesting this review and for suggesting the name “aquaporin.”
Financial support for these studies was provided by grants from the
National Institutes of Health, the Danish Medical Research Council,
and the Biomembrane Research Center of the University of Aarhus. G.
M. Preston received a postdoctoral fellowship from the Maryland Affiliate of the American Heart Association. J. S. Jung received a Fogarty
International Fellowship.
Address for reprint requests: P. Agre, Dept. of Biol. Chem., Johns
Hopkins University School of Medicine, 725 N. Wolfe St., Baltimore,
MD 21205.
F475
REVIEW
F476
38. Maurel,
eels.
EDITORIAL
C., J. Reizer,
J. I. Schroeder,
and
M.
J. Chrisp-
The vacular membrane protein g-TIP creates water specific
channels in Xenopus oocytes. ElMBO J. 12: 2241-2247,
1993.
39. Moon,
Agre.
C., G. M.
Preston,
C. A. Griffin,
E. W. Jabs,
and
P.
The aquaporin-CHIP
gene: structure, organization and
chromosomal localization. J. Biol. Chem. 268: 15772-15778,
1993.
40. Nielsen,
S., B. L. Smith,
E. I. Christensen,
and
P. Agre.
Distribution of the aquaporin CHIP in secretory and resorptive
epithelia and capillary endothelia. Proc. Nutl. Acud. Sci. USA. 90:
7275-7279,
1993.
41. Nielsen,
S., B. L. Smith,
E. I. Christensen,
M. Knepper,
and P. Agre.
CHIP28 water channels are localized in constitutively water-permeable segments of the nephron. J. Cell Biol. 120:
371-383,
1993.
42. Paganelli,
C. V., and A. K. Solomon.
The rate of exchange of
tritiated water across the human red cell membrane. J. Gen.
Physiol.
41: 259-277,
1957.
43. Pao,
G. M., L. F. Wu, K. D. Johnson,
H. Hofte,
M. J.
Chrispeels,
G. Sweet,
N. N. Sandal,
and M. H. Saier.
EvoMol. Microbial.
44. Pisano,
M.
5: 33-37, 1991.
and A. B. Chepelinsky.
Genomic cloning,
complete nucleotide sequence, and structure of the human gene
encoding the major intrinsic protein (MIP) of the lens. Genomics
M.,
11: 981-990,
1991.
45. Preston,
G. M., and
P. Agre.
Molecular cloning of the red cell
integral membrane protein of M, 28,000: a member of an ancient
channel family. Proc. Nutl. Acad. Sci. USA 88: 11110-l 1114, 1991.
46. Preston,
G. M.,
T. P. Carroll,
W. B. Guggino,
and
P. Agre.
Appearance of water channels in Xenopus oocytes expressing red
cell CHIP28 protein. Science Wash. DC 256: 385-387,
1992.
47. Preston,
G. M.,
J. S. Jung,
W. B. Guggino,
and
P. Agre.
The mercury-sensitive residue at cysteine-189 in the CHIP28 water channel. J. Biot. Chem. 268: 17-20, 1993.
48. Sabolic,
I., G. Valenti,
J. M.
A. S. Verkman,
D. A. Ausiello,
Verbavatz,
A. N. Van Hoek,
and D. Brown.
Localization
of the CHIP28 water channel in rat kidney. Am. J. Physiol. 263
(Cell Physiol.
32): C1225-C1233,
1992.
49. Saboori,
A. M., B. L. Smith,
and P. Agre.
Polymorphism in
the M, 32,000 Rh protein purified from Rh(D)-positive and -negative erythrocytes. Proc. Nutl. Acad. Sci. USA 85: 4042-4045,1988.
50. Sasaki,
S., K. Fushimi,
S. Uchida,
and F. Marumo.
Regu-
lation and localization of two types of water channels in the rat
kidney (Abstract). J. Am. Sot. Nephrol.
3: 798, 1992.
51. Sidel, V. W., and A. K. Solomon.
Entrance of water into human red cells under an osmotic pressure gradient. J. Gen. Physiol.
41: 243-257,
1957.
B. L., and
52. Smith,
Erythrocyte M, 28,000 transmembrane protein exists as a multi-subunit oligomer similar to channel
proteins. J. Biol. Chem. 266: 6407-6415,
1991.
52a.Smith,
Zeidel,
P. Agre.
B. L., R. Baumgarten,
S. Nielsen,
D. Raben,
M.
and P. Agre.
Concurrent expression of erythroid
L.
book of Physiology.
The Gastrointestinal
System.
Bethesda, MD:
Am. Physiol. Sot., 1991, sect. 6, vol. IV, p. 195-207.
S.-T.,
R. Zhang,
and A. S. Verkman.
High channel56. Tsai,
mediated water permeability in rabbit erythrocytes: characterization in native cells and expression in Xenopus oocytes. Biochemistry
57. Van
30: 2087-2092,
1991.
Heeswijk,
M. P. E.,
and C. H. Van OS. Osmotic water
permeability of brush border and basolateral membrane vesicles
from rat renal cortex and small intestine. J. Membr.
Biol. 92:
183-193,
1986.
58. Van Hoek,
A. N., M. L. Horn, L. H. Luthjens,
M. D. de Jong,
J. A. Dempster,
and C. H. van OS. Functional unit of 30 kDa
for proximal tubule water channels as revealed by radiation
tivation. J. Biol. Chem. 266: 16633-16635,
1991.
59. Van
Hoek,
A. N., L. H. Luthjens,
J. A. Dempster.
A 30 kDa
M.
L. Horn,
C. H. van
inacOS,
functional size for the erythrocyte water channel determined in situ by radiation inactivation.
Biochem.
Biophys.
Res. Commun.
184: 1331-1338, 1992.
60. Van Hoek,
A. N., and A. S. Verkman.
Functional reconstitution of the isolated erythrocyte water channel. J. Biol. Chem. 267:
and
18267-18269,
1992.
61. Van Hoek,
A. N., M. C. Wiener,
S. Bicknese,
Miercke,
R. M. Stroud,
and A. S. Verkman.
J. Biwersi,
L.
HPLC-purification of functional CHIP28 water channels for determination of
secondary structure by CD and FTIR spectroscopy (Abstract).
Biophys.
J. 64: 124, 1993.
A. S. Water channels
62. Verkman,
Physiol.
54: 97-108,
1992.
F. L., R. I. Sha’afi,
and
63. Viera,
in cell membranes. Annu.
Rev.
A. K. Solomon.
The state of
water in human and dog red cell membranes. J. Gen. Physiol. 55:
451-466,
64. Wade,
1970.
J. B., D. L.
Stetson,
and S. A. Lewis.
evidence for a membrane shuttle hypothesis. Ann.
372: 106-117,
65. Whittembury,
ADH action:
NY Acud.
Sci.
1981.
G.,
and L. Reuss.
Mechanisms of coupling of
solute and solvent transport in epithelia. In: The Kidney: Physiology and Puthophysiology,
edited by D. W. Seldin and G. Giebisch. New York: Raven, 1992, p. 317-360.
G. J., M. M. Pisano,
and A. B. Chepelinsky.
Tan66. Wistow,
dem sequence repeats in transmembrane channel proteins. Trends
Biol. Sci. 16: 170-171,
1991.
M. L., A. Abalak,
E. Grossman,
67. Zeidel,
and
A. Carruthers.
Role of glucose carrier in human erythrocyte water permeability.
Biochemistry
68. Zeidel,
M.
31: 589-596,
1992.
L., S. V. Ambudkar,
B. L. Smith,
and
P. Agre.
Reconstitution of functional water channels in liposomes containing purified red cell CHIP28 protein. Biochemistry
31: 7436-7440,
1992.
R., S. A. Alper,
B. Thorens,
and A. S. Verkman.
69. Zhang,
Evidence from oocyte expression that the erythrocyte water channel is distinct from band 3 and the glucose transporter. J. Clin.
Invest.
88: 1553-1558,
1991.
and
renal aquaporin CHIP and appearance of water channel activity in
perinatal rats. J. Clin. Invest. In press.
53. Solomon,
A. K. Water channels across the red blood cell and
other biological membranes. Methods
Enzymol.
173: 192-222,
70 . Zhang,
R., K. Fushimi,
D. Lei, and A. S. Verkman.
Expression cloning of water transporting proteins from rat kidney (Abstract). J. Am. Sot. Nephrol.
2: 730, 1991.
71. Zhang,
R., K. A. Logee,
and A. S. Verkman.
Expression of
mRNA coding for kidney and red cell water channels in Xenopus
1989.
54. Solomon,
A. K., B. Chasan,
J. A. Dix,
M. F. Lukacovic,
M.R.
Toon,
and A. S. Verkman.
The aqueous pore in the red
cell membrane. Ann. NY Acad. Sci. 414: 97-124, 1983.
55. Spring,
K. Mechanism of fluid transport by epithelia. In: Hund-
oocytes. J. Biol. Chem. 265: 15375-15378,
1990.
72. Zhang,
R., W. Skach,
H. Hasegawa,
A. N. Van Hoek,
and
A. S. Verkman.
Cloning, functional analysis and cell localization
of a kidney proximal tubule water transporter
CHIP28. J. Cell Biol. 120: 359-369.
1993.
homologous to
Downloaded from http://ajprenal.physiology.org/ by 10.220.32.246 on October 19, 2016
lution of the MIP family of integral membrane transport proteins.
REVIEW