Influence of initial fixation protocol on the appearance of primary cilia on the rabbit corneal endothelial cell apical surface as assessed by scanning electron microscopy

The objective was to examine primary cilia at the apical surface of corneal endothelial cells after using different fixatives. Female albino rabbits (2 kg) were euthanised at 15:00 h and the corneas fixed immediately (usually with an isotonic 2% glutaraldehyde-cacodylate fixative) either after dissection, by application fixative at 4 degrees C, by immersion of the eyeball in fixative at room temperature (RT), or by application of an isotonic or a hypertonic (Karnovsky-type) fixative at RT. Images at 2000x were taken from the central corneal region, and number and length of primary cilia assessed. The length was the same regardless of method (overall average of 1.67+/-0.70 microm), but the incidence of primary cilia was hypertonic fixative (87% of cells) >cold drop fixation (71%), >whole globe immersion (68%) >dissect then fix methods (67%) >RT drop fixation (34%). The first four methods however yielded cells with unacceptable artefacts (especially distortion). More details should be provided of the primary fixation method used.     

Effects of theophylline on expression of the long cilia phenotype in sand dollar blastulae

Previously, increases in ciliary length have only been obtained through genetic mutation in Chlamydomonas or by incubation of swimming echinoderm blastulae in trypsin or elastase. We have found that the phenotypic switch from short to long cilia on sand dollar blastulae can also be effected by incubation in theophylline. Cilia detached from control blastulae have a mean length of 21 +/- 7 microns with 10% of the cilia being greater than 30 microns. Upon incubation in 10 mM theophylline additional long cilia appeared after 10 hours and by 24-32 hours 1/2-3/4 of the embryo was covered with long cilia. The percentage of long cilia increased to 65% with a mean length of 40.0 +/- 17.6 microns. Incubation in other methylxanthines, such as aminophylline, caffeine, or isobutylmethylxanthine, inhibited development but had no effect on ciliary length distribution. Dibutyryl cAMP, 8-bromoadenosine, and calcium ionophore also had no effect on ciliary length. Cyclic AMP levels were measured and showed only slight differences among controls and embryos incubated in trypsin, caffeine, or theophylline. These data suggest that theophylline may be altering ciliary length control through some mechanism other than elevations in cAMP.     

Quantitative ultrastructure of isolated Kupffer cells of rat liver

The average volume of isolated Kupffer cells of rat liver is 821 +/- 64 microns 3, the average surface being 423 +/- 24 microns 2 (599 microns 2, with cell processes included). The surface structure (pseudopodia, lamellipodia, filopodia, microvilli) of isolated cells is much less developed than that of Kupffer cells in situ. By morphometric characterization volume densities are 0.1264 +/- 0.0077 (SE) for mitochondria and 0.3591 +/- 0.0169 for lysosomal structures. The volume of mitochondria amount to 0.79 +/- 0.04 microns 3.     

Effects of total parenteral nutrition on rat bile canaliculi: a stereologic analysis

OBJECTIVE: To quantitate, in a stereologic manner, changes in bile canalicular morphology before and after choleretic infusion of total parenteral nutrition (TPN) and to determine whether TPN produces changes in localized regions within the hepatic lobule. STUDY DESIGN: Livers were obtained from sham-operated on normal adult male rats (control) and from rats that received intravenous TPN solution containing 20% glucose and 3.5% Molipron F. The tissues, obtained by a rigorous sampling procedure, were systematically subjected to stereologic analysis. Measurements were made on electron micrographs at two levels of magnification by point, intersection and profile counts, and then volume, surface area and length were estimated per unit parenchymal volume. RESULTS: The surface area of the canalicular wall per parenchymal volume increased significantly (from 5.33 x 10(-2) to 6.73 x 10(-2) microns 2/micron 3) after TPN treatment, as did the length of microvilli (from 0.241 to 0.267 microns/micron 3). However, the volume of bile canalicular lumina per parenchymal volume (0.306% and 0.320%), total length of bile canaliculi (1.05 x 10(-2) and 1.06 x 10(-2) microns/micron 3) and diameter of microvilli (8.73 x 10(-2) and 8.94 x 10(-2) microns) remained constant. CONCLUSION: These results indicate that changes in canalicular shape and microvillus hypertrophy may cause lowering efficiency of the bile flow rate.     

Primary cilia in vertebrate corneal endothelial cells

The presence of primary cilia in corneal endothelial cells of a range of species from six non-mammalian vertebrate classes (Agnatha, Elasmobranchii, Amphibia, Teleostei, Reptilia and Aves) is examined by scanning and transmission electron microscopy. Our aim is to assess whether these non-motile cilia protruding into the anterior chamber of the eye are a consistent phylogenetic feature of the corneal endothelium and if a quantitative comparison of their morphology is able to shed any new light on their function. The length (0.42-3.80 microm) and width (0.12-0.44 microm) of the primary cilia varied but were closely allied with previous studies in mammals. However, interspecific differences such as the presence of a terminal swelling in the Teleostei and Amphibia suggest there are functional differences. Approximately one-third of the endothelial cells possess cilia but the extent of protrusion above the cell surface varies greatly, supporting a dynamic process of retraction and elongation. The absence of primary cilia in primitive vertebrates (Agnatha and Elasmobranchii) that possess other mechanisms to control corneal hydration suggests an osmoregulatory and/or chemosensory function.     

An investigation of the presence of ultramicrocells in natural mineral water

The presence of 'ultramicrocells' in natural mineral water, capable of passing through a 0.2 micron filter, has been demonstrated. Filters allowing the greatest proportion of viable (culturable) cells to pass ranked in the order, 0.4 micron polycarbonate (5.02%) > 0.2 micron polycarbonate (0.02%) > or = 0.45 micron cellulose nitrate (0.02%) > 0.2 micron cellulose acetate (< 0.002%). Following incubation for 4 d at 22 degrees C, viable counts in filtered mineral water increased from < 2-8.7 x 10(2) cfu ml-1(-2).8 x 10(4)-1.9 x 10(6) cfu ml-1. Successive filtration/incubation cycles of mineral water increased the proportion of cells passing through a 0.2 micron cellulose acetate filter from < 0.003% to 0.11% and 0.69%, suggesting selection for 'ultramicrocells'. Cells isolated from this process and grown on liquid R2A medium were thin, Gram-negative rods, of 0.15-0.40 micron wide and 0.50-6.20 microns long. Membrane filtration techniques used for pathogen detection in mineral waters will not retain all the cells present. If pathogens are able to form ultramicrocells, these may go undetected.     

Ultrastructure of the kidney of a South American caecilian,Typhlonectes compressicaudus (Amphibia,Gymnophiona)

The ultrastructure of the renal corpuscle, the neck segment, the proximal tubule and the intermediate segment of the kidney of a South American caecilian, Typhlonectes compressicaudus (Amphibia, Gymnophiona) was examined by means of transmission electron microscopy (TEM), scanning electron microscopy (SEM) and freeze-fracture technique. The glomerular filter apparatus consists of the podocyte epithelium, a distinct basement membrane, a subendothelial space and the capillary endothelium. Emanating from the podocyte cell body, several long primary processes encircle neighboring capillaries. The short slender foot processes originating from the primary processes interdigitate with those from other primary processes, thereby forming the meandering filtration slit. Thick bundles of microfilaments are found in the primary processes, but absent in the foot processes. The basement membrane consists of a lamina rara externa and a rather thin lamina densa (50 nm thickness). The wide subendothelial space contains abundant microfibrils, a few collagen fibrils and many thin processes of mesangial cells. The endothelium is flat and fenestrated (compared to mammals displaying relatively few fenestrations); some of the fenestrations are bridged by a diaphragm. The glomerular mesangium is made up of the mesangial cells and a prominent mesangial matrix containing microfibrils and collagen fibrils. The cells of the neck and intermediate segments display numerous cilia with their microtubules arranged in the typical 9 + 2 pattern. The basal bodies of the cilia are attached to thick filaments with a clear crossbanding pattern of 65 nm periodicity. The proximal tubule is composed of cells typical for this segment (PT cells) and light cells lacking a brush border (bald-headed cells). The PT cells measure 10-25 micron in height and 15-30 micron in width and do not interdigitate at their lateral borders with each other. Their basolateral cell membrane is amplified by many folds projecting into lateral intercellular spaces and into basal recesses. The brush border is scarce and composed of loosely arranged short microvilli.     

The effect of size of polystyrene particles on their retention within the rat peritoneal compartment, and on their interaction with rat peritoneal macrophages in vitro

Polystyrene particles (size range 300 nm-3 microns diameter) were radioiodinated and their capture by rat peritoneal macrophages measured in vitro. For unmodified particles, most efficient accumulation was observed using a diameter of 600 nm (Endocytic Index (E.I.) = 16.4 +/- 2.9 microliters/10(6) cells/h). Particles (3 microns diameter) which had been modified to become more hydrophilic by hydroxymethylation showed an increased rate of capture (E.I. = 136.6 +/- 91.2 microliters/10(6) cells/h). Following intraperitoneal administration to rats, unmodified 3 micron particles showed selective accumulation in the omentum (18.4% injected dose/g), and this was increased for the hydroxymethylated bead (35.3% dose/g). The smaller (800 nm) particles were better able to leave the peritoneal compartment. Radiolabelled particles isolated from a peritoneal wash after 5 h were mostly cell-associated (72-86%, depending on the type of particle).     

Kinetic analysis of F-actin depolymerization in polymorphonuclear leukocyte lysates indicates that chemoattractant stimulation increases actin filament number without altering the filament length distribution

The rate of filamentous actin (F-actin) depolymerization is proportional to the number of filaments depolarizing and changes in the rate are proportional to changes in filament number. To determine the number and length of actin filaments in polymorphonuclear leukocytes and the change in filament number and length that occurs during the increase in F-actin upon chemoattractant stimulation, the time course of cellular F-actin depolymerization in lysates of control and peptide-stimulated cells was examined. F-actin was quantified by the TRITC-labeled phalloidin staining of pelletable actin. Lysis in 1.2 M KCl and 10 microM DNase I minimized the effects of F-actin binding proteins and G-actin, respectively, on the kinetics of depolymerization. To determine filament number and length from a depolymerization time course, depolymerization kinetics must be limited by the actin monomer dissociation rate. Comparison of time courses of depolymerization in the presence (pointed ends free) or absence (barbed and pointed ends free) of cytochalasin suggested depolymerization occurred from both ends of the filament and that monomer dissociation was rate limiting. Control cells had 1.7 +/- 0.4 x 10(5) filaments with an average length of 0.29 +/- 0.09 microns. Chemo-attractant stimulation for 90 s at room temperature with 0.02 microM N-formylnorleucylleucylphenylalanine caused a twofold increase in F-actin and about a two-fold increase in the total number of actin filaments to 4.0 +/- 0.5 x 10(5) filaments with an average length of 0.27 +/- 0.07 microns. In both cases, most (approximately 80%) of the filaments were quite short (less than or equal to 0.18 micron). The length distributions of actin filaments in stimulated and control cells were similar.     

Differential modulation of actin-severing activity of gelsolin by multiple isoforms of cultured rat cell tropomyosin. Potentiation of protective ability of tropomyosins by 83-kDa nonmuscle caldesmon

Multiple isoforms of tropomyosin (TM) of rat cultured cells show differential effects on actin-severing activity of gelsolin. Flow birefringence measurements have revealed that tropomyosin isoforms with high Mr values (high Mr TMs) partially protect actin filaments from fragmentation by gelsolin, while tropomyosins with low Mr values (low Mr TMs) have no significant protection even when the actin filaments have been fully saturated with low Mr TMs. We have also examined effect of nonmuscle caldesmon on the severing activity of gelsolin because 83-kDa nonmuscle caldesmon stimulates actin binding of rat cell TMs (Yamashiro-Matsumura, S., and Matsumura, F. (1988) J. Cell Biol. 106, 1973-1983). While nonmuscle caldesmon alone or low Mr TMs alone show no significant protection against fragmentation by gelsolin, the low Mr TMs coupled with 83-kDa protein are able to protect actin filaments. Further, high Mr TMs together with 83-kDa protein appear to block the severing activity completely. Electron microscopic analyses of length distribution of actin filaments have confirmed the results. The average length of control actin filaments is measured as 1.46 +/- microns, and gelsolin shortens the average length to 0.084 +/- 0.039 micron. Similar short average lengths are obtained when gelsolin severs actin complexed with low Mr TMs (0.080 +/- 0.045 micron) or with nonmuscle caldesmon (0.11 +/- 0.072 micron) while longer average length (0.22 +/- 0.18 micron) is measured in the presence of high Mr TMs. The simultaneous addition of nonmuscle caldesmon makes the average length considerably longer, i.e. 0.61 +/- 0.37 micron in the presence of low Mr TMs and 1.57 +/- 0.97 micron in the presence of high Mr TMs. Furthermore, the actin binding of gelsolin is strongly inhibited by co-addition of high Mr TMs and nonmuscle caldesmon. These results suggest that TM and gelsolin share the same binding site on actin molecules and that the differences in the actin affinities between TMs are related to their abilities of protection against gelsolin.     

Phenotypic switching to long cilia effected by various proteases: results with Dendraster excentricus and Stronglyocentrotus purpuratus blastulae

Incubation in trypsin effects a phenotypic switch from short to long cilia (greater than 30 micron) in hatched blastulae of the sea urchin, Arbacia punctulata. To determine how trypsin causes such a switch we tested whether the phenomenon was unique to the species, Arbacia, and to the protease, trypsin. With two other echinoderm species, the sand dollar, Dendraster excentricus, and the sea urchin, Strongylocentrotus purpuratus, trypsin incubation increased the percentage of long cilia. During incubation of D. excentricus in trypsin, the percentage of long cilia increased progressively from the normal 10% long cilia of the apical tuft to 45-50% long cilia covering 1/2-3/4 of the embryo. With S. purpuratus blastulae, however, the percentage of long cilia was lower (32-40%) and the results were more variable. Of the additional proteases tested with D. excentricus, elastase was more effective than trypsin in terms of the percentage of long cilia (58%), the mean length, and the broad distribution of lengths formed. Thermolysin was about as effective as trypsin but chymotrypsin was much less so. Thus, increases in ciliary length were not unique to a particular echinoderm species or to incubation in trypsin. The magnitude of the change in length distribution, however, was species- and enzyme-dependent. An extracellular or membrane component with differential susceptibility to various proteases may, therefore, be involved in altering ciliary length.     

Lysosomal interconnections and horseradish peroxidase compartmentalization in three-dimensionally reconstructed bovine alveolar macrophages

Following a 1-h incubation of bovine alveolar macrophages in 1 to 2 mg/ml exogenous horseradish peroxidase (HRP), ultrathin sections revealed vacuolar interconnections among both labeled and unlabeled vacuoles constituting the lysosomal compartment. Four entire cells and their vacuolar components were subsequently computer resconstructed from serial transmission electron micrographs and measured using a morphometric technique. HRP-labeled and unlabeled vacuoles ranged in size from 0.5 micron to greater than or equal to 4.0 microns in diameter and occupied up to 25% of the cytoplasmic volume. HRP-containing vacuoles were distributed throughout each cell in a clumped distribution (P less than 0.05) and occupied up to 75% of the total vacuole compartment. Up to 60% of all vacuoles were interconnected through a series of openings formed by membrane fusions (average pore diameter 0.42 micron), which resulted in a labyrinth of vacuoles comprising up to 55% of the total volume of the lysosomal compartment. The area of open interconnections resulting from vacuolar fusions represented less than 1% of the total surface area of the lysosomal membrane. Rotation of a three-dimensionally reconstructed macrophage about the Y-axis revealed an interconnected vacuolar network of 75 fused vacuoles in a chain up to 21 microns in length. We have demonstrated that HRP-labeled vacuoles interconnect with each other as well as with preexisting unlabeled vacuoles. As a result of such interconnections, individual vacuoles become contributing members of a large, continuous, lysosomal compartment in bovine alveolar macrophages.     

The corpus luteum of the pig. Scanning electron microscopic study of surface features at different times of incubation

The surface ultrastructure of porcine early corpus luteum cells (days 1-3 of the luteal phase) was studied in SEM and correlated with progesterone secretion. Luteal cells were divided into 2 groups: small cells (10-20 microns) and large cells (20-30 microns) and their surface features were observed after 1, 3, and 5 h of incubation in the control medium and in a medium supplemented with prolactin (PRL). The surface morphology of control cells was characterized by numerous smooth blebs and the presence or absence of thin microvilli. Small and large cells showed a tendency to adhere to the glass during the experiment, but on the large cells the number of thin adhesive filopodia was greater. After the 1st and 3rd h of incubation with PRL the number of microvilli and numerous filopodia on the small cells increased substantially. Nodular blebs were scattered and appeared to protrude from the cell surface. Many small cells adhered to the glass by thick, layered and thin thread-like cytoplasmic processes. After the 5th h distinct smoothing of the surface of the small cells was seen. The number of microvilli seen on the PRL stimulated surface of the large cells was smaller and in some cases even entirely absent. After the 1st and 3rd h of the experiment the large cell surface was ruffled with minute folds. Numerous nodular blebs protruded from the cell surface. The number of adhesive filopodia attaching the cells to the glass decreased or vanished during the experiment. After the 5 h of incubation most of the cells had smooth surface with smooth blebs. Progesterone secretion was measured by radioimmunoassay. The cells in the medium without exogenous hormone (control) secreted relatively low levels of progesterone throughout 1-5 h of the incubation period. After addition of PRL to the medium the amount of secreted progesterone increased.     

Isolation and characterization of single myocardial cells from the perch, Perca fluviatilis

In applying the enzymatic cell isolation technique to the fish heart about 40% of the dispersed myocytes maintained their spindle-shaped morphology, and about half of them tolerated physiological concentration of Ca2+ and excluded the vital dye, Evans blue. The length of spindle-shaped myocytes was on average 133 +/- 3 micron and the maximum width was 4.2 +/- 0.1 micron. The mean length of the sarcomeres was 2.1 +/- 0.1 micron. The sizes of the myocytes did not vary significantly with the weights of the fish. Electron microscopic examinations showed typical fish myocardial cell structure; absence of transverse tubule system, a sparse network of sarcoplasmic reticulum and from a few up to eight or more myofibrils. The cells were mononuclear. Most of the Ca2+-tolerant myocytes were quiescent, but the contraction in them could be induced by electric field stimulation. Both the spontaneous and electrically triggered contractions were of twitch type. The slowly propagating contraction waves, so-called phasic contractions common in isolated mammalian cardiac myocytes, could not be seen at all.     

Distribution of diaphragmatic lymphatic stomata

In seven anesthetized rabbits we measured the size, shape, and density of lymphatic stomata on the peritoneal and pleural sides of the diaphragm. The diaphragm was fixed in situ and processed for scanning electron microscopy. Results are from 2,902 peritoneal and 3,086 pleural fields (each 1,620 microns 2) randomly chosen from the various specimens. Stomata were seen in 9% of the fields examined, and in 30% of the cases they appeared grouped in clusters with 2-14 stomata/field. Stoma density was 250 +/- 242 and 72 +/- 57 (SD) stomata/mm2 on peritoneal and pleural sides, respectively, and it was similar over the muscular and tendinous portion of the two surfaces. The maximum diameter ranged from less than 1 to approximately 30 microns, with an average value of 1.2 +/- 3.1 micron. The ratio of the maximum to the minimum diameter and the surface area averaged 2 +/- 1.4 and 0.7 +/- 2.4 micron 2, respectively. The maximum and minimum diameter and surface area values followed a lognormal frequency distribution, suggesting that stomata geometry is affected by diaphragmatic tension.     

Ultrastructural localization of acid phosphatase activity in the small intestinal absorptive cells of adult rats.

Ultrastructural localization of acid phosphatase activity was investigated in ultrathin (0.05 micron) and semithin (0.5 and 0.75 micron) sections of the small intestinal epithelial cells of adult rats. The results showed that the enzyme activity was localized on the membrane of microvilli, lateral cell membranes, lysosomes, the Golgi complex, and the GERL of Novikoff (a part of the smooth-surfaced endoplasmic reticulum located in close proximity to the inner Golgi saccules) of duodenal absorptive cells. The lysosomes contained within the duodenal and jejunal absorptive cells appeared to be mainly heterolysosomes rather than autolysosomes. The enzyme activity of absorptive cells was lower in the jejunum than in the duodenum, and was barely detectable except in the GERL and lysosomes of the ileum. The average numbers of lysosomes having a diameter of 0.2 approximately 1.0 microns, per cell profile in sections of 214 duodenal, 226 jejunal and 318 ileal epithelial cells were 8.9 +/- 0.189, 6.4 +/- 0.155 and 3.5 +/- 0.027 (mean +/- SE), respectively. From these results, it was assumed that both the Golgi apparatus and GERL produce some lysosomes in the duodenal and jejunal absorptive cells, but only GERL does so in the ileum. It was considered also that because of an unexpectedly high number of lysosomes containes within the epithelial absorptive cells of the proximal intestine of adult rats, these cells may possess the strong heterophagic, as well as absorptive capacity.     

Observations on the solitary cilium of rabbit oviductal epithelium: its motility and ultrastructure

Solitary cilia have been observed on rabbit oviductal epithelial cells. In tissue cultures of fimbrial epithelium of 3- and 4-day-old animals observed by phase microscopy, most of these single cilia exhibited a vortical or funnel-type movement while others had the usual to-and-fro motility. Primary cilia are usually considered immotile. Transmission electron microscopy of specifically identified single cilia revealed differences between the ciliary shafts and basal bodies of the single cilia as compared to those of mature oviductal ciliated cells. The basal body of the solitary cilium often had at least two triangular, striated, basal foot processes, lacked electron-dense satellite material around its basal end, and occasionally had striated rootlets. In contrast, the cilia of mature ciliated cells had only one basal foot, exhibited much electron-dense satellite material, and lacked rootlets. Cross sections of the single cilia showed patterns of microtubules different from the usual 9 + 2 axonemal complexes of normal cilia and included 9 + 0, 10 + 2 singlets, 7 + 2 doublets, and 8 + 1 doublet and 2 singlets; one did have the usual 9 + 2 arrangement. We postulate that the presence of more than one basal foot process may be responsible for the vortical motility observed. The primary cilia are shorter than normal cilia; the longest one measured was 1.86 micron in length, 0.28 micron in width at its base, and 0.14 micron at its tip. Based on the light-microscopic, scanning-electron-microscopic and transmission-electron-microscopic observations, such solitary cilia were observed more frequently in the oviductal tissues of the 3- to 4-day postnatal rabbits grown in tissue culture and in ovariectomized and ovariectomized/progesterone-treated adult animals than in estrous, ovulatory, or ovariectomized/estradiol-treated rabbits.     

Evaluation of various methods of measurement of the microtubule length in the cytoplasm of cultured cells

It is generally assumed that microtubules in tissue culture cells extend from the centrosome to cell periphery, and the length of individual microtubules averages several dozens of microns. However, direct electron-microscopic measurements have cast some doubt on this assumption. In this study, the average length of microtubules in cultured Vero cells was estimated using a combined approach. The length of free cytoplasmic and centrosomal microtubules was determined by means of electron microscopy in serial sections; concurrently, the length of free microtubules in the lamella was measured in preparations stained with tubulin antibodies (an indirect immunofluorescent method), by tracing saltatory particle movements along the microtubules in living cells. According to the data of immunofluorescent microscopy, microtubule length in the lamella averaged 4.57 +/- 3.69 microns. However, since two or more microtubules can overlap, their length may be slightly overestimated by this method. On the other hand, saltatory movements are easy to monitor and measure fairly accurately, but their range may be shorter than the actual microtubule length because of a limited processiveness of motors (kinesin and dynein). On average, the trajectories of saltatory movements in living cells were 3.85 +/- 0.72 microns long. At the electron-microscopic level, microtubule length was analyzed using pseudo-three-dimensional reconstructions of the microtubule systems around the centrosome and in the lamella. The length of free microtubules in the lamella reached 18 microns, averaging 3.33 +/- 2.43 microns; the average length of centrosomal microtubules was 1.49 +/- 0.82 microns. Good correspondence between the data on microtubule length and arrangement obtained by different methods allows the conclusion that most of free microtubules in Vero cells actually have a length of 2-5 microns; i.e., they are much shorter than the cell radius (about 25 microns). Microtubules extending from the centrosome are shorter still and do not reach the cell periphery. Thus, most microtubules in the lamella of Vero cells are free and their ordered arrangement is not associated with their attachment to the centrosome.     

Aquaporin deletion in mice reduces corneal water permeability and delays restoration of transparency after swelling

Two aquaporin (AQP)-type water channels are expressed in mammalian cornea, AQP1 in endothelial cells and AQP5 in epithelial cells. To test whether these aquaporins are involved in corneal fluid transport and transparency, we compared corneal thickness, water permeability, and response to experimental swelling in wild type mice and transgenic null mice lacking AQP1 and AQP5. Corneal thickness in fixed sections was remarkably reduced in AQP1 null mice and increased in AQP5 null mice. By z-scanning confocal microscopy, corneal thickness in vivo was (in microm, mean +/- S.E., n = 5 mice) 123 +/- 1 (wild type), 101 +/- 2 (AQP1 null), and 144 +/- 2 (AQP5 null). After exposure of the external corneal surface to hypotonic saline (100 mosm), the rate of corneal swelling (5.0 +/- 0.3 microm/min, wild type) was reduced by AQP5 deletion (2.7 +/- 0.1 microm/min). After exposure of the endothelial surface to hypotonic saline by anterior chamber perfusion, the rate of corneal swelling (7.1 +/- 1.0 microm/min, wild type) was reduced by AQP1 deletion (1.6 +/- 0.4 microm/min). Base-line corneal transparency was not impaired by AQP1 or AQP5 deletion. However, the recovery of corneal transparency and thickness after hypotonic swelling (10-min exposure of corneal surface to hypotonic saline) was remarkably delayed in AQP1 null mice with approximately 75% recovery at 7 min in wild type mice compared with 5% recovery in AQP1 null mice. Our data indicate that AQP1 and AQP5 provide the principal routes for corneal water transport across the endothelial and epithelial barriers, respectively. The impaired recovery of corneal transparency in AQP1 null mice provides evidence for the involvement of AQP1 in active extrusion of fluid from the corneal stroma across the corneal endothelium. The up-regulation of AQP1 expression and/or function in corneal endothelium may reduce corneal swelling and opacification following injury.