Abstract
Purpose:
We investigated whether cellular connectivity between Schlemm's canal (SC) inner wall (IW) endothelium, and juxtacanalicular connective tissue (JCT), and between IW endothelial cells, plays a role in giant vacuole (GV) and pore formation by comparing perfusion- and immersion-fixed eyes.
Methods:
Normal human donor eyes (n = 4) were either immersion-fixed (0 mm Hg) or perfusion-fixed (15 mm Hg). Trabecular meshwork near SC was imaged using serial block-face scanning electron microscopy. A total of 12 IW cells from each group were 3D-reconstructed from ∼7040 electron micrographs and compared. In each cell, connections between IW cells and JCT cells/matrix were quantified; IW/IW connectivity was measured by cell border overlap length. GV volume, density, shape, and intracellular and paracellular pores were analyzed.
Results:
The mean number of IW/JCT cell-cell connections per cell significantly decreased (P < 0.01) while the summed GV volume per cell significantly increased (P < 0.01) in perfusion-fixed eyes compared to immersion-fixed eyes. Intracellular pores were observed in 14.6% of GVs in perfusion-fixed eyes and not observed in immersion-fixed eyes. The mean IW/IW overlap length per cell decreased (P < 0.01), and paracellular pores were found only in regions where IW/IW connectivity was minimal (overlap length = 0 μm) in perfusion-fixed eyes and not observed in immersion-fixed eyes.
Conclusions:
Our data suggest that changes in IW/JCT connectivity may be an important factor in the formation of larger GVs, and decreased IW/IW connectivity may promote paracellular pore formation. Targeting the IW/JCT and IW/IW connectivity may therefore be a potential strategy to regulate outflow resistance and IOP.
Primary open-angle glaucoma (POAG) is a leading cause of blindness worldwide.
1 A primary risk factor for POAG is elevated intraocular pressure (IOP) from increased aqueous outflow resistance.
2–4 Most of the resistance is believed to be generated in the juxtacanalicular connective tissue (JCT) and modulated by the inner wall (IW) of Schlemm's canal (SC).
5–11 However, the mechanism through which outflow resistance is regulated remains unclear.
Rho-kinase inhibitors have been shown to increase outflow facility and to be associated with a physical separation between the IW and the underlying JCT in porcine,
12 bovine,
13 and monkey eyes
14 and with JCT expansion in human eyes.
15–17 JCT expansion and separation from the IW have also been observed in increasing outflow facility during washout in bovine eyes.
14,18 These studies suggest that IW/JCT connectivity may play a role in regulating outflow resistance. Furthermore, passage of aqueous humor into SC appears to require formation of giant vacuoles (GVs) and pores, and previous studies showed an increased pore density with increased accumulation of fluorescent tracers along the IW
19 and near the pores on both basal and apical sides of the IW endothelium of SC.
15 In previous studies of POAG eyes, there was an observed overall reduction in pore density compared to age-matched normal eyes.
10,20 Previous electron microscopy studies have investigated GV size with different perfusion pressures and described types of connections between IW and JCT cells in 1000 regions of contact between cells using random sections.
21 However, due to lack of methods, no previous study has been able to quantify accurately the degree of IW/JCT connectivity and GV volume in each individual IW cell. Therefore, the relationships between IW/JCT connectivity and GV and intracellular pore (I-pore) formation have not been fully elucidated. Additionally, a previous study reported that with increasing IOP, tight junctions connecting adjacent IW cells became simplified, and the overlap between IW cells decreased.
22 However, the relationship between decreased IW/IW connectivity and paracellular or border pore (B-pore) formation has not yet been studied also due to lack of an effective method.
Serial block-face scanning electron microscopy (SBF-SEM), in combination with 3D-reconstruction, was first introduced for examining large volumes of neuronal tissue.
23 In neuroscience, this technique allows researchers to study synaptic interactions of neurons in 3D through a substantial spatial volume of neuronal circuits.
23–27 In ophthalmology research, it was used to study the 3D aspects of matrix assembly by cells in the developing cornea
28 and to gain a structural understanding of retinal degeneration.
29 In this study, using this innovative method and large serial image sets, we investigated whether cellular connectivity between IW endothelial cells of SC and JCT cells/matrix and between adjacent IW cells plays a role in GV and pore formation by comparing 3D reconstructed IW cells from perfusion-fixed (15 mm Hg) eyes to immersion-fixed (0 mm Hg) eyes.
The goal of this study was to test our hypothesis that larger GVs in perfusion-fixed eyes are associated with a decrease in IW/JCT connectivity and increased I-pore formation and that a decrease in IW/IW connectivity may promote B-pore formation, and thus may contribute to the regulation of aqueous outflow resistance.
For immersion-fixation of eyes 01 and 02, a small cut (∼5 mm) was made along the equator of the eyes, followed by immediate immersion in modified Karnovsky's fixative (2.5% glutaraldehyde and 2% paraformaldehyde in a 0.1-M sodium phosphate-buffer, pH = 7.3) for 24 hours.
The perfusion procedure has previously been described in detail.
30 Briefly, eyes 03 and 04 were perfused with Dulbecco's phosphate-buffered saline (pH = 7.3; GIBCO, Carlsbad, CA, USA) containing 5.5 mM D-glucose (collectively referred to as GPBS) at 15 mm Hg to establish a stable baseline facility for 30 minutes. The eyes were then perfusion-fixed at 15 mm Hg for 30 minutes with the same fixative as immersion-fixation described above. Upon completion of perfusion, a small cut (∼5 mm) cut was gently made along the equator, and the eyes were immersed in the same fixative overnight.
After fixation, a total of five tissue wedges (2 × 2 × 2 mm3) of the regions of interests (ROIs), including trabecular meshwork, SC, and small portions of sclera, were dissected from the four eyes and delivered to Renovo Neural, Inc. (Cleveland, OH, USA) for imaging by SBF-SEM.
All of the GVs in every reconstructed cell were identified and manually outlined on every SBF-SEM image in which they appeared. The number of GVs per cell was counted. To differentiate vacuoles from pinocytotic vesicles, vesicular structures smaller than 1.0 um
3 were excluded from the vacuole count.
21,32,33 The GV volumes were autocalculated using the Volume tool in Reconstruct.
Three methods were used to evaluate the GV volume per cell. The values from these three methods were compared between cells from the immersion- and perfusion-fixed groups. In the first method, we calculated and compared the mean volume of all of the GVs between the two groups. In the second method, we identified the single largest GV in each cell, and compared the means of those largest GV volume between the two groups. In the third method, we summed the volumes of all of the GVs found in each cell to calculate summed GV volume per cell.
GVs were categorized into two general types based on their shape: rounded GVs and collapsed GVs. GVs that had a smooth, dome-shaped apical surface were categorized as rounded GVs. GVs that had a depressed surface or indentation on the apical surface were categorized as collapsed GVs.
The results from immersion- and perfusion-fixed eyes were compared to determine if there were changes in the IW/JCT connectivity, the IW/IW connectivity, GV density, GV volume and shape, and number of I- and B-pores. All statistical analyses were performed in R statistical computing package (v3.5.1; R Foundation for Statistical Computing, Vienna, Austria). Unpaired two-tailed t-tests were used to compare cellular dimensions with a required significance level of 0.05. We performed χ2 analysis to test the difference in distributions of rounded and collapsed GVs between groups. Cumulative distribution frequency analysis and a nonparametric Kolmogorov-Smirnov test were used to compare the distribution of GV volumes in each group. Nonparametric Wilcoxon-rank sum tests were used to compare the GV volume between groups.
In this study, using SBF-SEM and 3D reconstruction, we investigated IW/JCT and IW/IW connectivity and their potential roles in GV and pore formation by comparing the IW cells from perfusion-fixed eyes to those from immersion-fixed eyes. Our main findings were:
-
Mean number of IW/JCT cell-cell connections per cell decreased, while summed GV volume per cell increased in perfusion-fixed eyes compared to immersion-fixed eyes;
-
There was no difference in GV density per cell between immersion- and perfusion-fixed eyes;
-
No pores were observed in immersion-fixed eyes, while 14.6% of GVs had I-pores in perfusion-fixed eyes (I-pores only in GVs larger than 41.72 μm3);
-
In perfusion-fixed eyes, the mean IW/IW overlap length per cell decreased significantly, and B-pores were found only in regions where IW/IW overlap length became zero.
These results support our hypothesis that larger GVs in perfusion-fixed eyes are associated with a decrease in IW/JCT connectivity and increased I-pore formation and that a decrease in IW/IW connectivity may promote B-pore formation, and thus may contribute to the regulation of aqueous outflow resistance. Although previous studies suggest that the pores, on their own, may only generate 10% of outflow resistance,
36 a hydrodynamic interaction between the pores of the IW of SC and the underlying JCT may greatly increase outflow resistance in this region through a “funneling effect” by confining the flow to the JCT regions near the IW pores, forcing a funneling pattern of aqueous outflow; thus, pores may modulate the resistance in this region.
37
SBF-SEM provided serial, high-resolution images across a substantial volume of tissues. Combined with 3D reconstruction, this method overcomes the limitations of 2D imaging and enabled us for the first time to quantify accurately the IW/JCT and IW/IW connectivity, to determine GV volume, density, shape, and pores per IW cell, and to compare their differences between immersion- and perfusion-fixed tissues.
Grierson et al.
21 characterized 6 types of IW/JCT connections. They examined a total of 1000 randomly-selected IW/JCT connections in the IW endothelium of nine rhesus monkey eyes and described the IW/JCT cell-cell connections as desmosomes capable of sustaining cellular stress in the direction of outflow. However, the study was unable to document how extensive the IW/JCT connectivity was at the level of individual cells and how this connectivity changes under different pressure gradients. In the current study, we collected serial, high-resolution images through the entire tissue blocks (each block was 134 × 59 × 187 μm
3 on average; see
Table 1), which captured all of the 1663 IW/JCT connections in 22 entirely reconstructed IW cells. We were able to detect an additional type of IW/JCT cellular connection, IW cell body-JCT cell body connection. We demonstrated, for the first time to our knowledge, that the IW/JCT connectivity was extremely substantial in IW cells of immersion-fixed eyes, and the overall IW/JCT connectivity was significantly decreased in IW cells of perfusion-fixed eyes.
We found that along with decreased IW/JCT connectivity, the IW cells became shorter, significantly narrower in nonnuclear regions, and significantly thinner in perfusion-fixed eyes. IW cells are thought to respond to the pressure gradient in the form of mechanical stretch or distortion of the outflow resistance through changes in the ECM composition as well as cellular cytoskeletal organization.
38,39 Previous studies have also shown that many IW cells possess a large reservoir of excess membrane stored in ruffles, folds, or vesicles
40,41 that allow the cell to rapidly change shape without altering the total membrane surface area.
9,42 It is possible that with the physical detachment of IW cells from the underlying JCT, IW cells become more capable of deforming along the pressure gradient, forming larger outpouchings (GVs) into SC. The newly available cell membrane that initially made up the surface area of the original cellular connections may be recruited to help the cell body deform, and thereby form larger GVs. In this study, we tested the relationship between the number of IW/JCT cell-cell connections and the summed GV volume in each group, but did not find a linear relationship (perfusion-fixed:
R2 = 0.098;
n = 10;
P > 0.05); immersion-fixed:
R2 = 0.044;
n = 12;
P > 0.05). This suggests that in addition to IW/JCT cell-cell connections, other factors may influence GV volume, such as active and inactive flow regions, IOP/perfusion-pressure, cell and ECM stiffness, and these warrant further investigation.
We determined the exact number of GVs in each reconstructed cell and found that there were often multiple GVs per IW cell, and usually one large GV was accompanied by a few smaller GVs. We also found no significant difference in the total number of GVs per IW cell between the immersion- and perfusion-fixed groups, which was an interesting finding. Previous studies showed that to study GVs, perfusion-fixation is necessary, because GVs are pressure-dependent structures.
21,32,43–45 Another study also showed that after ocular perfusion was halted, 70% of GVs disappeared in 3 minutes.
45 In our study, although the GV density per cell was not significantly different between the two groups, the mean GV volume was significantly smaller, and there was a higher percentage of smaller GVs in immersion-fixed eyes. This suggests that the GVs in immersion-fixed eyes were deflating gradually, but did not disappear completely, during the 24-hour postmortem period as well as during a couple of minutes when a small cut was made in the eyes to be immersion-fixed immediately after. Additionally, since the GVs in immersion-fixed eyes were smaller and tended to span fewer serial images than larger GVs in perfusion-fixed eyes, these small GVs could have been easily missed if only one random cross-sectional image of SC was examined using traditional 2D imaging. This may have led to an underestimation of GV count/density in immersion-fixed tissues in previous studies.
33 In order to confirm this, we did additional investigation and comparison between GV density per cell in random TEM sections from eyes with immersion-fixed versus perfusion-fixed at 15 mm Hg from a previous study.
22 Four eyes were immersion-fixed, and four were perfusion-fixed at 15 mm Hg. The fixation methods were similar as those in the current paper. TEM images were taken randomly along the IW of SC of each eye. We analyzed 78 (at least 17 per eye) images containing 475 IW cells for immersion-fixed eyes and found 31 GVs. In perfusion-fixed eyes, we analyzed 88 images containing 803 IW cells and found 329 GVs. We found that the mean percentage of cells with GV in immersion-fixed eyes was much lower than that in 15 mm Hg (6.7% vs. 47.4%; 0.16 versus 1.46/μm of SC), compared to the similar density found between two groups in our current study with SBF-SEM and 3D reconstruction. Our data suggest that even when observing hundreds of cells at each pressure with similar fixation methods, 2D studies of random sections still can not provide a true picture of GV presence per cell.
However, GV density observed in immersion-fixed eyes could vary due to preparation methods. A previous study by Grierson and Lee reported that no GVs were found at 0 mm Hg when eyes were rapidly immersed in fixative without a cut along the equator.
33 Brilakis and Johnson
45 hemisected eyes and cut the anterior segment into 6 wedges before immersion in fixative and reported 70% decrease in GV counts after 3 minutes, 82% decrease after 15 minutes, and 88% decrease by 120 minutes after discontinuation of saline perfusion at 20 or 40 mm Hg when compared with perfusion-fixed fellow eyes at same pressure.
45 In our method, we made a small cut along the equator before immersion in fixative; this procedure may cause a pressure spike. However, the pressure influence in our preparation would be smaller than that in Brilakis and Johnson's method with hemisecting eyes. We observed an 89% decrease in GV counts in our immersion-fixed eyes compared to perfusion-fixed eyes at 15 mm Hg by analysis of the images from our previous study with same preparation method,
22 which is similar to the 88% decrease in GV counts after 120 minutes reported by Brilakis and Johnson.
45 Unlike Brilakis and Johnson's method, who placed tissue wedges in the fixative and thus fixation was nearly immediate, we placed the whole eye with a small cut along the equator into the fixative. It would take a longer time for fixative to diffuse to SC from the equator than across the sclera. The vacuoles would not be fixed until the fixative diffuses across the sclera and reaches the inner wall of SC. Given that the thickness of the sclera is about 1 mm, and using a diffusion coefficient of glutaraldehyde in polymer networks of 3 × 10
−6 cm
2/sec,
46 we estimate that it will take roughly 15 to 30 minutes before a significant concentration of glutaraldehyde reaches the inner wall of SC. Another difference was that the perfusion pressure was higher in Brilakis and Johnson's study, and thus GVs in their study might be larger before fixation than that in our study.
32,33 Our current study focuses on comparison of GVs between 2D and 3D methods using the same fixation preparation.
Grierson and Lee
33 calculated the theoretical GV volume (16.71 μm
3, 15 mm Hg, based on a spheroidal model of GV shape) from measurements from the light micrograph with the largest cross-sectional area of each GV (section thickness = 1.5 μm). Their GV volume in monkey eyes calculated from light micrographs was significantly smaller compared to our GV volume in human eyes observed via SBF-SEM and 3D reconstruction (144.98 ± 47.65 μm
3,
n = 41 GVs, 15 mm Hg; one-sample
t-test,
P = 0.01). We calculated the theoretical GV volume using Grierson and Lee's method and a small number of GVs (
n = 10) from our study by selecting the largest cross-sectional area of each GVs and several random cross-sectional areas away from the largest. Our analysis revealed that theoretical GV volume calculated from a random cross-sectional area was different than its true 3D-calculated GV volume by 29% to 98% depending on the distance from the largest cross-sectional area of that GV. Our results suggest that a single cross-sectional image of GV may not be representative of its true size, because in most cases the cross-sectional area is not imaged at the location where the GV is the largest, which may result in an underestimation or overestimation of its true size.
We found relatively more rounded GVs and fewer collapsed GVs in the perfusion-fixed group. Because the immersion-fixed eyes were without active outflow for less than 24 hours and a small cut was made at the equator of the eyes before immediate immersion-fixation upon their delivery to our laboratory, we speculated that many of the GVs observed were beginning to deflate due to the dissipation of IOP, resulting in fewer rounded GV. This may explain why more GVs appeared as if they were on their way to collapse or deflate in immersion-fixed samples, whereas more GVs appeared rounded in perfusion-fixed samples.
The technique of SBF-SEM followed by 3D reconstruction is advantageous in studying pores, because reconstructed cells could be freely rotated to any viewpoint so that even pores obstructed by surrounding structures could be observed. If traditional scanning or transmission electron microscopy were used, some pores could have been obstructed in en face imaging, or not imaged at all due to the lack of serial images. One previous study using random TEM imaging reported only 1.6% of GV had luminal pores at the same perfusion pressure.
47 In this study, a total of 6 I-pores (14.6%) were found in 12 inner wall cells from the perfusion-fixed eyes. This percentage is consistent with previous studies using SEM that found 13% to 29% of GVs to have pores,
48,49 but falls on the lower end of the range. This may be due to nonuniform distribution of pores along the inner wall endothelium,
20 and we might have used IW endothelial tissues from regions with lower pore density. This may be related to the segmental nature of aqueous humor outflow through the trabecular meshwork.
14,50–52 Future studies using 3D-EM to identify pores across a larger surface area of IW endothelium in active and inactive outflow regions are warranted.
All 6 of the I-pores observed in this study were associated with rounded GVs larger than 41.72 μm
3. Using 41.72 μm
3 as a threshold, we found that the perfusion-fixed group had relatively more GVs that were larger than this size compared to the immersion-fixed group (39% vs. 21%) (
Fig. 10;
Tables 2,
3). Since we only investigated a small number of cells in this study, further study using a larger sample size is needed to confirm our finding. In addition, it seems that GV volume is not the only contributing factor for pore formation, as shown by our finding that not all GVs larger than 41.72 μm
3 formed pores. In the perfusion-fixed group, only 37.5% (6 out of 16) of larger GVs (size >41.72 μm
3) formed pores; furthermore, in the immersion-fixed group, none of these larger GVs formed pores. This suggests that a large GV volume itself may not guarantee pore formation. Although mean volumes of the collapsed GVs were larger than rounded GVs, no I-pores were observed associated with collapsed GVs, suggesting that tension or strain of a GV may also play a role in pore formation. This finding is consistent with a previous study that reported increasing biomechanical strain acted as a trigger for pore formation in cultured SC cells.
53
SBF-SEM followed by 3D reconstruction enhanced our ability to quantify B-pores and study their surrounding environment. Two B-pores were found in the perfusion-fixed eyes, while none were observed in immersion-fixed eyes. Our observations in perfusion-fixed eyes of: 1) decreased IW/IW cell overlap length, and 2) B-pores were found only in regions where IW/IW overlap length became zero, suggest that decreased IW/IW connectivity may promote B-pore formation. Previous studies showed that pore formation may be influenced by biomechanical strain on the IW cells. IW cells have been thought to be able to regulate pore density based on biomechanical cues in the immediate environment through changing their own cell stiffness and gene expression.
53 Therefore, it is possible that tight junctions between adjacent IW cells are responsive to changing physiological conditions, which is a potential mechanism that regulates the reduced amount of cell-cell overlap, resulting in changes in local filtration characteristics of the IW endothelium. Our study on B-pore formation was limited by a small number of cells that we were able to reconstruct in a reasonable amount of time. Future studies are needed to investigate pores across larger surface areas of IW endothelium.
SBF-SEM followed by 3D reconstruction studies are generally time-consuming and labor-intensive. Our study examined over 7000 SBF-SEM images and completed thousands of tracings for cells and organelles, and we were only able to study a total of 24 cells in great detail. SBF-SEM is also limited to imaging of fixed tissues. Our study was unable to determine the stage at which each GV was observed during GV formation. There was also no way to determine whether the GVs without I-pores simply did not form pores, had formed pores moments before, or were on the way to develop pores. Future studies using live cell imaging and time-lapse microscopy may provide new insight into the dynamic process of GV and pore formation. Since there is no episcleral venous pressure in enucleated human donor eyes, our perfusion pressure at 15 mm Hg represents a higher pressure drop across the outflow pathway than the physiologic pressure in normal living eyes; therefore, the measured effects may be somewhat magnified compared to in vivo physiologic pressure.
In summary, while SBF-SEM and 3D-EM reconstruction were labor and time intensive, a major advantage of the technique was that the serial, high-resolution imaging allowed us to study cellular connectivity in a 3D space, while providing volumetric data. With this technique, we demonstrated that the IW/JCT cell-cell connectivity significantly reduced and the summed GV volume per cell significantly increased in perfusion-fixed eyes. The IW/IW connectivity (cell overlap) significantly decreased in perfusion-fixed eyes, which may increase the possibility of B-pore formation. Our findings suggest that IW/JCT and IW/IW connectivity may play a role in regulating aqueous outflow through modulating the size of GVs and number of pores. Therefore, targeting and modifying IW/JCT and IW/IW connectivity may be used as a strategy to develop potential therapeutics to regulate outflow resistance and IOP.
The authors thank Emily Benson and Grahame Kidd, PhD, of Renovo Neural, Inc. for the technical assistance and Mark Johnson, PhD, for discussion.
Supported by BrightFocus Foundation (2016099), NIH/NEI EY022634, EY019696, and The Massachusetts Lions Eye Research Fund.
Disclosure: J. Lai, None; Y. Su, None; D.L. Swain, None; D. Huang, None; D. Getchevski, None; H. Gong, None