1288
Reports
TOYS, May 1999, Vol. 40, No. 6
11. Sievv EL, Bettelheim FA. Light scattering parameters of calcium
cataracts. Exp Eye Res. 1996;62:265-270.
12. Bettelheim FA, Churchill AC, Zigler JS Jr. On the nature of hereditary
cataract in strain 13/N guinea pigs. Curr Eye Res. 1997; 16:667-674.
13- Bettelheim FA. Syneretic response to pressure in ocular l e n s . /
Theor Biol. 1999;197:277-280.
14. Kamiya T, Zigler JS Jr. Long term maintenance of monkey lenses in
organ culture: a potential model system for the study of human
cataractogenesis. Exp Eye Res. 1996;6l:425-431.
15- Loewenstein MA, Bettelheim FA. Pressure induced turbidity in fish
lenses. Exp Eye Res. 198O;3O:315-317.
16. Koretz JF, Handelman GH, Brown NP. Analysis of human crystalline lens curvature as a function of accommodative state and age.
Vision Res. 1984;24:1141-1151.
17. Koretz JF, Handelman G.H. Modeling age-related accommodative
loss in the human eye. Math Modeling. 1986;7:1003-10l4.
18. Farnsworth PN, Shyne SE. Anterior zonular shifts with age. Exp Eye
Res. 1979;28:291-297.
19- Koretz JF, Handelman GH. How the human eye focuses. Sci Am.
1988;259:92-99.
20. FisherRF. The force of contraction of the human ciliary muscle
during accommodation. J Physiol. 1971;270:51-74.
Molecular Identification and
Immunolocalization of the Water
Channel Protein Aquaporin 1 in
CBCECs
1
1
with the anti-AQPl antibody, and the labeling was selectively localized to the plasma membrane by light
microscopy. Subcellular localization by EM revealed immunoreactivity with the inner leaflets of the plasma
membrane.
The identity of the aquaporin, its abundance,
and its membrane location suggest that it is a major pathway forfluidflowacross endothelial cell membranes. This
is consistent with transcellular endothelial fluid transport.
(Invest Ophthalmol Vis Sci. 1999;40:1288-1292)
CONCLUSIONS.
2
Jun Li, Kunyan Kuang, S0ren Nielsen, and
Jorge Fiscbbarg,1'5
PURPOSE. Water
channel proteins are important pathways
for water movements across cell membranes, including
those in the corneal endothelium that contribute to the
fluid transport mechanism essential in maintaining corneal
transparency. This study was conducted to identify and
locate the water channel protein(s) in cultured bovine
corneal endothelial cells (CBCECs).
1
METHODS. PolyCA)" " RNA was isolated from CBCECs, and
MMLV reverse transcriptase and random hexamer primers were used to generate a cDNA pool by reverse
transcription-polymerase chain reaction (RT-PCR).
Two specific degenerate primers were synthesized
based on consensus sequences from the major intrinsic
lens protein superfamily; a "touchdown" PCR protocol
accommodated the degeneracy. Immunolocalization
was performed by incubating sections of CBCECs with
an antibody against human aquaporin 1 (AQP1). Cryosections (0.85 ju,m) of CBCECs were used for light microscopy, and 800-A ultrathin cryosections were used
for electron microscopy (EM).
A 372-bp fragment was isolated. Its encoded
amino acid sequence was 100% identical with that of
bovine AQP1 (AQPl_bovin). CBCECs reacted strongly
RESULTS.
From the Departments of 'Physiology and 'Ophthalmology, College of Physicians and Surgeons, Columbia University, New York, New
York; and the Department of 2Cell Biology, Institute of Anatomy,
University of Aarhus, Denmark.
Supported by Grant EY06178 from the National Institutes of
Health, Bethesda, Maryland; and Research to Prevent Blindness, New
York, New York.
Presented in preliminary form at the 12th International Congress
of Eye Research, Yokohama, Japan, 1996.
Submitted for publication September 1, 1998; revised January 4,
1999; accepted January 25, 1999.
Proprietary' interest category: N.
Reprint requests: Jorge Fischbarg, Department of Ophthalmology,
College of Physicians and Surgeons, Columbia University, 630 West
168th Street, New York, NY 10032.
he corneal endothelium is essential in maintaining the
normal level of dehydration and transparency of the cornea through its fluid pump. Yet, although fluid transport has
been shown in many other epithelia including cultured corneal
endothelial cells, the inner workings of the fluid pump remain
unsolved. Some mechanisms presumably involved in it have
been reviewed,1 but cellular details such as the identity and
location of the membrane transporters and channels involved
and the signaling and regulatory pathways for them remain to
be elucidated.
The discovery of the water channel protein, channelforming integral membrane protein of 28 kDa (CHIP28), in
human red blood cells2 brought new impetus to studies in this
field. The sequence of this protein turned out to be highly
homologous to that of the major intrinsic lens protein (MIP),
and to date, at least nine water channel isoforms (later renamed
aquaporins, or AQP1-9) have been identified. The distribution
of water channels is characteristic: For instance, within a
nephron, all epithelia involved in fluid movement (and only
those epithelia) have water channels.3 This suggests that water
channels are the major pathway for epithelial fluid transport
and that such fluid movement is transcellular.
AQP1 has been located in rat corneal endothelium by in
situ hybridization,4 and we have reported the existence of
mRNA encoding functional water channels in CBCECs, the
expression of which is abolished by AQP1 antisense RNA.5
From that study, we concluded that CBCECs express a CHIP28type protein. In the present study, we identified a segment
encoding nearly half of the coding region of AQP1 and localized it to the inner leaflet of the plasma membrane.
T
METHODS
Cell Culture and mRNA Preparation
Bovine eyes were obtained from a local abattoir. Corneas with
a 2- to\3$kim-wide annulus of surrounding sclera were excised
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933213/ on 07/31/2017
IOVS, May 1999, Vol. 40, No. 6
under sterile conditions. The endothelial layer was then incubated at 37°C for 10 minutes with 1 ml trypsin containing
Dulbecco's modified Eagle's medium (Gibco, Gaithersburg,
MD), the endothelial surface was gently scraped off, and detached cells were harvested into a centrifuge tube containing 5
ml Dulbecco's modified Eagle's medium with 10% fetal bovine
serum (Gibco) and antibiotics (100 units/ml penicillin G and
100 ng/ml streptomycin). After centrifugation at 1000 q?m
(model GLC-1 with HL4 rotor; Sorvall, Newtown, CT), the
supernatant was discarded, and the pellet was resuspended in
2 ml of the same medium. Cells were then apportioned into
75-cm2 tissue culture flasks. The primary cultures reached
confluence within 1 week. Cell morphology was monitored
with an inverted phase-contrast microscope. Subculture was
performed to three to four generations.
Passages 1 through 3 of CBCECs were used for RNA
preparation. Total RNA was isolated by a single-step procedure
(RNAzol; Biotecx Laboratories, Houston, TX). Cells were harvested from a flask, washed as described earlier, and homogenized with RNAzol in a hand-driven glass teflon homogenizer
(5-10 strokes). Protein and DNA in the homogenate were
precipitated by phenol-chloroform, and RNA in the supernatant was then precipitated with isopropanol. Diethylpyrocarbonate-treated RNase-free solution was used to resuspend
RNA. Poly(A)"f RNA (mRNA) was purified by affinity chromatography on oligo(dT) cellulose columns (Cat. No. 808229;
Boehringer Mannheim, Indianapolis, IN) and was subsequently
divided in batches and kept frozen at — 20°C until used. The
integrity of the RNA batches was routinely checked by polyacrylamide denaturing gel electrophoresis.
Determination of Gene-Specific Primers
According to a previously used method, 15 proteins from the
MIP superfamily were used for the multiple sequence alignment, and two highly conserved regions were selected: sense
primer: HL(V)NPAVT, 5'-CA(C/T)(G/C/T)T(G/A/T/C)AA(C/T>
CC(G/A/T/C)GC(G/A/T/C)GT(G/A/r/)AC-3'; and antisense primer: NPARSF, 5'-AA(G/A/T/C)(G/CXA/T)(G/AAVC)C(G/T)(G/
A/T/C)GC(G/A/T/C)GG(G/A)TT-3'.
The two conserved regions included NPA boxes characteristic of the MIP superfamily. In these regions, most of the
amino acids are identical in the superfamily except residue 75,
which is valine in human AQP1 and leucine in bovine MIP
proteins. At the time of the study the bovine AQP1 sequence
was unknown, and we therefore designed primers with valine
or leucine at this position. Primers therefore consisted of 17 or
20 nucleotides. The span between two primers is 124 amino
acids in human AQP1 (total sequence is 269 amino acids) or
372 bp. The amino acid sequence was transformed into encoding nucleotide sequences by a computer program (Backtranslate; GCG, Madison, WI). Primers had a 3072- and 4096-fold
degeneracy, respectively. Oligonucleotides were synthesized
(ABI 373A) in our campus DNA facility. A column (NAP-5;
Pharmacia Biotech, Uppsala, Sweden) was used to purify the
synthesized primers, which were then dissolved at 20 picomoles/ju,l (stock solution).
Reverse Transcription-Polymerase Chain
Reaction
Reverse transcription-polymerase chain reaction (RT-PCR)
was performed according to known protocols. CBCEC (1.5 /ttg)
Reports
1289
poly(A)+ RNA (1 )ag//Ltl) was incubated with 1 /al of 1.6 ju,g//ixl
random hexamer primer reagent (Cat. No 1034731; Boehringer
Mannheim, Indianapolis, IN) in a bath at 70°C for 2 minutes.
The solution was then mixed with 1 U reverse transcriptase
(Superscript II RNase H~ Cat. No. 18064014; Gibco) with
buffer, 0.1 M dithiothreitol, 2.5 mM deoxyribonocleoside
triphosphate (dNTP, Cat. No. 27203501; Pharmacia, Pi seataway, NJ), and 1 /ml RNasin (Cat. No. N2111, Promega, Madison, WI). The RT reaction was conducted at 37°C for 1 hour
and then at 70°C for 5 minutes. Reverse transcriptase-free
solution was used as a control. Different volumes (1.5, 2, 2.5,
and 3 ju,l) of RT reaction solution were subsequently added into
the PCR solution. For each PCR, 0.5 U Taq enzyme (Perkin
Elmer, Norwalk, CT) was used. Other PCR compounds were
used as a standard.
Because of the high degeneracy of the primers, a "touchdown " PCR protocol was used. PCR without a DNA template
was used as a control. The PCR products were separated with
1% agarose minigel electrophoresis. DNA was then stained
with ethidium bromide and visualized under a UV transilluminator (Fischer Scientific, Springfield, NJ). The size of the PCR
products was assessed using DNA markers. DNA bands within
the expected size range were extracted and purified from the
gel using gel extraction kits (QIAEX II; Qiagen, Chatsworth,
CA) and used for TA cloning.
TA Cloning Technique and Plasmid Amplification
Freshly purified PCR products (containing 3' A overhangs)
were mixed with ligation solution that included 50 ng vector
(pCRII; Invitrogen, San Diego, CA) with 3' T overhangs at its
insertion site, 5 U T4 DNA ligase, and buffer. The ligation
solution was incubated at 14.5°C overnight. After ligation,
Escherichia coli-competent cells (Epicutian Coli; Stratagene,
La Jolla, CA) were used for the plasmid transformation. The
transformed bacteria were plated on agar culture media containing ampicillin (100 /Ltg/ml) and incubated at 37°C overnight. The sizes of the inserts in selected clones were assessed
by PCR using two flanking primers (SP6 and T7). The vectors
with predicted inserts were isolated using a kit (Plasmid Miniprep; Qiagen) for sequencing.
Immunohistochemistry and Immunoelectron
Microscopy
Anti-CHIP28 antibody purified from sera of rabbits immunized
with nonglycosylated CHIP28 was used. The preparation and
characterization have been described in detail previously.^6
The antibody and the labeling conditions were strictly as described.3 CBCECs grown in treated plastic flasks (passage 1)
were trypsinated, fixed with 4% paraformaldehyde in 0.1 M
sodium cacodylate buffer, and embedded in gelatin. Blocks
were postfixed in the samefixativefor 2 hours, infiltrated with
2.3 M sucrose containing 2% paraformaldehyde for 30 minutes,
mounted on holders, and rapidly frozen in liquid nitrogen. For
light microscopic observations, 0.85 /xm cryosections were
incubated with anti-CHIP28 (0.03-0.07 /xg/ml) and the labeling was visualized by use of horseradish peroxidase-conjugated secondary antibody.
For EM the immunolabeling was performed on 800-A
ultrathin cryosections, and the labeling was visualized by use of
protein A-gold. Specificity of immunolabeling was confirmed
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933213/ on 07/31/2017
1290
IOVS, May 1999, Vol. 40, No. 6
Reports
12
3 4
5 6 7 8 9
RESULTS
Isolation of a Partial Gene Encoding AQP1
by RT-PCR
.506/517
404
307
FIGURE 1. The PCR products were inserted into the vector (pCRII;
Invitrogen) and isolated using the EcoRI restriction enzyme. The size of
the inserts was assessed by 1% agarose gel electrophoresis. Lanes 1 and
9: DNA markers pBR322-Ms/Jl digest (New England Biolaboratories,
Beverly, MA) and I KB (Gibco, Gakhersburg, MD), respectively. Lanes
2 through 8: runs of the inserts isolated from the vector. Only the DNA
fragments in lanes 2, 3, 5, and 7 were close to the expected size range
(350-400 bp). These were sequenced in the local DNA facility and the
length of the fragments determined: lane 2, 352 bp; lane 3, 400 bp;
lanes 5 and 7, 372 bp. Sequences in lanes 5 and 7 were identical with
that of"AQPI_bovin.
by substituting anti-CHTP28 with purified nonimmune rabbit
IgG, or by incubation without primary antibody, or by incubation without primary or secondary antibody.
In an initial attempt, a A-phage cDNA library constructed from
CBCECs was used as a template. Using the gene-specific primers
shown earlier, 16 PCR rims were performed, and four clones with
the expected size were isolated. Limes 2 and 3 in Figure 1 show
bands for two of them (352 bp and 400 bp, respectively); however, their sequences turned out to be from proteins from the E.
colt cells and the A-phage used. Therefore, for the next attempt,
we used as a template cDNA reverse transcribed from CBCEC
poly(A)+ mRNA. PCR runs were performed with this template
using the same gene-specific primers shown earlier. Seven runs of
PCR were conducted using three concentrations of primers (0.2
JLIM, 1 /LLM, and 5 /xM, final [primerl in total PCR solution) to
optimize yield. The products were separated by electrophoresis,
and seven DNA fragments of a size close to that expected were
isolated and inserted into a TA cloning vector. The determination
of insert size was refined by resorting to PCR using two flanking
primers (SP6 and 17).
Five clones were isolated (Fig. 1, lanes 4 - 8). Of these, the
fragments in lanes 5 and 7 were between 352 bp and 400 bp.
Sequencing showed that these two fragments were 372 bp in
length and were identical without a single exception to AQP1
from bovine ciliary body.7 Other DNA bands (Fig, 1; lanes 4, 6,
8) were out of the expected range and were not sequenced.
A,
A
A
A
A
A
C
C
C
C
C
AQPI_B0V(N
AQP1_SHEEP
AQPI_RAT
AQPI_MOUSE
AQPI_HUMAN
V
V
V
V
V
G
G
G
G
G
QSV
Q SV
QSV
QSV
QSV
L P O N 5 L
L P D N S L
L L E N S L
L V D N SL
LFIG N S L
A
A
A
A
A
extracellular
TTYJP I K S N Q T T G A V Q D
PI K S N Q T T G A V Q D
L E R N Q T
L V QD
L E R N Q T
L V QD
A V QD
P V G Nl
B, Intracellular
T L G L L L S C Q
S V L R
T L G L L L S C Q
S I
T L G L L L SCQ
S I
T L G L L L SCQ
S I
S I F R
T L G L L L S C Q
C, extracellular
G L N A L A P G V N S
G L M A L A P G V N S
G R N D L A R G V N S
G R N D L A H G V N S
G R N D L A D G V N S
D, Intracelluler
G
G
G
G
G
O G L G
Q G L G
QG L G
Q G L G
Q G L G
L
L
L
L
L
C
C
C
C
C
V L
VL
VL
VL
VL
H
H
H
H
H
L
L
L
L
L
L
L
L
L
L
R
R
R
R
R
R R D L
R R D L
RRDL
RRDL
RRDL
201
AQPI_BOVIN
AQPI_SHEEP
AQP1_RAT
AQPI_MOUSE
AQPI_HUMAN
GGS6P
5DS8P
G SA P
G G SA P
G G SA P
A
A
A
A
A
B, oxtracellular
S F
G S S V I
T H N F g D H W
S F
G S S V I
T H N F g D H W
G S A V L T R N F S N H W
G S A V L T R N F S N H W
GS
A 0 l
250
AQPI_BOVIN
AQP1_SHEEP
AQPI_RAT
AQPI_MOUSE
AQPI_HUMAN
G
G
G
G
G
AAL
AAL
SAL
OAL
G AL
S D L T D R V K V W T S
S D L T D R V K V W T S
T H N F S N H W
C terminal,
Intracellular
G Q V E E Y D L D A D D I
G Q V E E Y D L D A D D I
S D F T D R
M K V W T S
G Q V E E Y D L D A D D I
S D F T D R
M K V W T S
G Q V E E Y D L D A D D I
S D L T D R V K V W T S
G Q V E E Y D L D A D D I
N
N
N
N
N
S
S
S
S
S
272
R V E M K P K
R V E M K P K
R V E M K P K
R V E M K P K
R V E MIKIP K
FIGURI- 2. Multiple sequence alignment of live AQP1 from different species. Rectangular areas with double line border represent the six
predicted transmembrane helical segments. 12 The segments predicted as loops are marked as predicted intracellular (cytophismic) or extracellular.
Boxed individual residues denote a location with sidedness confirmed by mutagenesis. 1012 Residues that differ from human AQP1 are in bold and
underlined. Rectangles with gray background mark the NPA boxes.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933213/ on 07/31/2017
IOVS, May 1999, Vol. 40, No. 6
Reports
.
.
•
1291
*
o
K<
i
FIGURE 3. Cryosections were incubated with anti-CH!P28, and immunoreuctions were visualized by horseradish peroxidase- conjugated secondary antibody (A, B, C) or by protein A-gold (D, E). (A) CHIP28 was localized to the plasma membrane. Most cells exhibited strong immunoreactions
and labeled cells were uniformly labeled on the entire surface. (B) Negative control with nonimnume IgG instead of anti-CHIP28. (C) Higher
magnification showing the selective labeling of the plasma membrane on the entire surface (arrows). No intracellular compartments exhibit
anti-CHIP28 reactivity. (D, E) immiinogold labeling was restricted to the plasma membrane (arrows). Magnification, (A, B) X440; (C) XI100; (D,
E) X 45,000.
Given the complete identity between our 372-bp segments and those of bovine AQP1, the entire published sequence of AQPl_bovin was used for multiple sequence analysis. This sequence was aligned with those of AQP1 proteins
from other species. Figure 2 presents the resultant alignment,
arranged so that the location of sequence differences can be
readily compared with predicted structural segments of AQP1.
(not shown), consistent with findings in studies using this
antibody preparation.3 Subcellular localization by immunoelectron microscopy with the same antiserum and protein A-gold
revealed immunoreactivity with the inner leaflets of; the plasma
membrane but not with membranes in intracellular compartments.
Immunolocalization of AQP1 in CBCECs
DISCUSSION
The cellular location of AQP1 was examined directly by using
anti-human_AQPl antiserum. CBCECs exhibited strong inimiinoreaction (Fig. 3). All labeling was confined to plasma membrane domains, and there was virtually no background labeling. Specificity of the staining was shown with nonimmune IgG
that failed to react (Fig. 3). Absorption controls using preabsorbed antibody with purified AQP1 were completely negative
Sequencing of AQP1
These results identify for the first time the sequence of a water
channel in bovine corneal endothelium as that of bovine AQP1.
This finding is in line with those in our previous physiological
study5 in which we found evidence of the presence of CH1P28type water channels in CBCECs based on inhibition of their
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933213/ on 07/31/2017
1292
IOVS, May 1999, Vol. 40, No. 6
Reports
expression by a 25-base antisense oligonucleotide. In the
present study, we found that the ammo acid sequence encoded
by the 372-bp mRNA fragment identified was 100% identical
with that of AQP1 from bovine ocular ciliary epithelium.7
Of the water channel isoforms found so far, the present
study reinforces the notion that AQP1 is the major one in
corneal endothelium; whether it is the only one is unclear. The
osmotic water permeability of CBCECs did not disappear completely when mercurial reagents were applied (75% reduction
in water permeability5), which leaves open the possibility that
a mercurial insensitive AQP may also be present in these cells.
AQP1 exists in apical and basolateral membranes of rat corneal
endothelium/' In addition to expression in corneal endothelium, AQP1 is expressed in ciliary nonpigmented epithelium,
scleral nbroblasts, keratocytes, endothelium covering the trabecular meshwork and Schlemm's canal, posterior and anterior
epithelium of the iris, and anterior lens epithelium.8 The complete 3'-untranslated region encoded by the human AQP1 gene
has been reported in the retinal pigment epithelium,9 but the
presence of this isoform was not confirmed by immunolabeling.*
So far, by comparison,8 AQP1 seems to be the dominant
isoform in the eye. Whether that reflects functional homology
among the different tissues that express it is less clear. In the
case offluid-transportingepithelia, AQP1 could be an obvious
transcellular route for water. However, its presence in keratocytes or scleral fibroblasts has not been explained.
Multiple Sequence Comparison
Bovine AQP1 is closest to sheep AQP1 (Fig. 2). In addition,
aside from minor conservative substitutions, it is clear that
most variations in these two and the other family members
appear only in loops A and C (both probably extracellular).
Therefore, presumably these regions are less crucial than
the remainder of the protein for AQP functions. Conversely,
loops B and E (both related to the water pore formation in
the hourglass model of Agre and colleagues10) are well
conserved, which is consistent with the alleged role. In
addition, it may be noted that the C-terminal region is also
well conserved. This may be related to a possible protein
kinase A phosphorylation site at residues R234-second235second236 in human AQP1 (J. W. Regan, personal communication, January 1997), or perhaps to sequences that modulate expression or translocation.
Immunolocalization
The use of the anti-CHlP28 antibody is particularly revealing.
Our results show that immunostaining was selectively localized
to the plasma membrane of CBCECs cells (Figs. 3A, 3C) and
subcellularly localized to the inner leaflets of the plasma membrane but not to the membranes of intracellular compartments
(Figs. 3D, 3E). The labeling is restricted to the inner leaflet,
consistent with the presence of the COOH-terminal at this
location. Several studies by Nielsen et al. (e.g., Ref. 4) have
documented that this antibody exclusively recognizes die
COOH-terminal end of AQP 1.
The strong immunoreactions of CBCECs toward the antiCHIP28 antibody indicate the abundance and ubiquity of
CHIP28 proteins in these cells. In addition, all plasma membrane domains were labeled, consistent with the labeling pattern described earlier in vivo. The membrane-bound distribution strongly suggests that these proteins constitute a major
pathway for fluid movement across corneal endothelial cell
membranes. The presence of abundant corneal endothelial
water channels that are functional5 and membrane-bound (the
current report) further suggests that, as has been hypothesized, ' ' most if not all the fluid transport by this layer could
well travel a transcellular route.
References
1. Fischbarg J. Mechanism offluidtransport across corneal endothelium and other epithelial layers: a possible explanation based on
cyclic cell volume regulatory changes. BrJ Ophthalmol. 1997;81:
85-89.
2. Preston GM, Carroll WB, Guggino WB, Agre P. Appearance of
water channels in Xenopus oocytes expressing red cell CHIP28
protein. Science. 1992;256:385-387.
3. Nielsen S, Smith BL, Christensen El, Knepper MA, Agre P. CHIP28
water channels are localized in constitutively water-permeable
segments of the nephron. / Cell Biol. 1993;120:371-3834. Nielsen S, Smith BL, Christensen El, Agre P. Distribution of the
aquaporin CHIP in secretory and resorptive epithelia and capillary
endothelia. Proc Natl Acad Sci USA. 1993:90:7275-7279.
5. Echevarria M, Kuang K, Iserovich P, et al. Cultured bovine corneal
endothelial cells express CHIP28 water channels. Am J Physiol.
1993;265:C1349-C1355.
6. Smith BL, Agre P. Erythrocyte Mr 28,000 transmembrane protein
exists as a multisubunit oligomer similar to channel proteins./Z?»o/
Chem. 1991;266:6407-64l5.
7. Patil RV, Yang X, Saito I, Coca-Prados M, Wax MB. Cloning of a
novel cDNA homologous to CHIP28 water channel from ocular
ciliary epithelium. Biochem Biophys Res Commun. 1994;204:
861-866.
8. Hamann S, Zeuthen T, La Cour M, et al. Aquaporins in complex
tissues: distribution of aquaporins 1-5 in human and rat eye. AmJ
Physiol. 1998;274:C1332-C1345.
9. Ruiz A, Bok D. Characterization of the 3' UTR sequence encoded
by the AQP-1 gene in human retinal pigment epithelium. Biocbim
Biophys Acta. 1996;1282:174 -178.
10. Jung JS, Preston GM, Smith BL, Guggino WB, Agre P. Molecular
structure of the water channel through aquaporin CHIP: the hourglass model. J Biol Chem. 1994;269:l4648-l4654.
11. Fischbarg J, Warshavsky CR, Lim JJ. Pathways for hydraulically and
osmotically-induced water flows across epithelia. Nature. 1977;
266:71-74.
12. Preston GM, Jung JS, Guggino WB, Agre P. Membrane topology of
aquaporin CHIP. Analysis of functional epitope-scanning mutants
by vectorial proteolysis./ Biol Chem. 1994;269:1668-1673.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933213/ on 07/31/2017